Workability and Process Design PDF

1119 FM 7/29/03 3:39 PM Page i
2003 ASM International. All Rights Reserved. www.asminternational.org

Handbook of Workability and Process Design (#06701G)
Handbook
of
Workability
and
Process Design
Edited by
George E. Dieter
Howard A. Kuhn
S. Lee Semiatin
INTERNATIONAL
The Materials
Information Society
www.asminternational.org
1119 FM 7/29/03 3:39 PM Page ii

Copyright 2003
by
ASM International
All rights reserved
No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means,
electronic, mechanical, photocopying, recording, or otherwise, without the written permission of the copyright owner.
First printing, September 2003
Great care is taken in the compilation and production of this book, but it should be made clear that NO
WARRANTIES, EXPRESS OR IMPLIED, INCLUDING, WITHOUT LIMITATION, WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE, ARE GIVEN IN CONNECTION WITH
THIS PUBLICATION. Although this information is believed to be accurate by ASM, ASM cannot guarantee that
favorable results will be obtained from the use of this publication alone. This publication is intended for use by persons
having technical skill, at their sole discretion and risk. Since the conditions of product or material use are outside of
ASMs control, ASM assumes no liability or obligation in connection with any use of this information. No claim of any
kind, whether as to products or information in this publication, and whether or not based on negligence, shall be greater
in amount than the purchase price of this product or publication in respect of which damages are claimed. THE
REMEDY HEREBY PROVIDED SHALL BE THE EXCLUSIVE AND SOLE REMEDY OF BUYER, AND IN NO
EVENT SHALL EITHER PARTY BE LIABLE FOR SPECIAL, INDIRECT OR CONSEQUENTIAL DAMAGES
WHETHER OR NOT CAUSED BY OR RESULTING FROM THE NEGLIGENCE OF SUCH PARTY. As with any
material, evaluation of the material under end-use conditions prior to specification is essential. Therefore, specific
testing under actual conditions is recommended.
Nothing contained in this book shall be construed as a grant of any right of manufacture, sale, use, or reproduction, in
connection with any method, process, apparatus, product, composition, or system, whether or not covered by letters
patent, copyright, or trademark, and nothing contained in this book shall be construed as a defense against any alleged
infringement of letters patent, copyright, or trademark, or as a defense against liability for such infringement.
Comments, criticisms, and suggestions are invited, and should be forwarded to ASM International.
ASM International staff who worked on this project include Steve Lampman, Acquisitions Editor; Bonnie Sanders,
Manager of Production; Carol Terman, Jill Kinson, and Nancy Hrivnak, Production Editors; and Scott Henry,
Assistant Director of Reference Publications.
Library of Congress Cataloging-in-Publication Data
Handbook of workability and process design / edited by George E. Dieter, Howard A.

Kuhn, Lee Semiatin.
p. cm.
Updated and expanded ed. of: Workability testing techniques. 1984.
ISBN 0-87170-778-0
1. MetalsFormabilityTesting. I. Dieter, George Ellwood. II. Kuhn, Howard A. III.
Semiatin, S. L. IV. Workability testing techniques.
TA460.H3195 2003
671.3dc21
2003052197
ISBN: 0-87170-778-0
SAN: 204-7586
ASM International
Materials Park, OH 44073-0002
Printed in the United States of America
Multiple copy reprints of individual articles are available from ASM International, Technical Department.
1119 FM 7/29/03 3:39 PM Page iii

Contents
Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi Chapter 7 Hot-Tension Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

P.D. Nicolaou, R.E. Bailey, and S.L. Semiatin
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Equipment and Testing Procedures . . . . . . . . . . . . . . . . . . . . . . . 68
Hot Ductility and Strength Data from the Gleeble Test . . . . . . . . 72
Isothermal Hot-Tension Test Data . . . . . . . . . . . . . . . . . . . . . . . . 74
Chapter 1 Workability and Process DesignAn Introduction . . . . . . . 3 Modeling of the Isothermal Hot-Tension Test . . . . . . . . . . . . . . . 78
Howard A. Kuhn, George E. Dieter, and S.L. Semiatin Cavitation During Hot-Tension Testing . . . . . . . . . . . . . . . . . . . . 80
Workability Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Process Design and Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Chapter 8 Torsion Testing to Assess Bulk Workability . . . . . . . . . . . . 86
Approximate and Closed Form Solutions . . . . . . . . . . . . . . . . . . 14 S.L. Semiatin and J.J. Jonas
Numerical Techniques in Process Modeling . . . . . . . . . . . . . . . . 14 Material Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Microstructural Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Specimen Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
Processing Maps. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Torsion Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Flow-Stress Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
Interpretation of Torsion Fracture Data . . . . . . . . . . . . . . . . . . . 103
Chapter 2 Bulk Workability of Metals . . . . . . . . . . . . . . . . . . . . . . . . 22 Measuring Flow-Localization-Controlled Workability . . . . . . . . 108
George E. Dieter Microstructure Development During Deformation
Stress, Strain, and Stress-Strain Curves. . . . . . . . . . . . . . . . . . . . 22 Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Multiaxial Stress States. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Processing History Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
Material Factors Affecting Workability . . . . . . . . . . . . . . . . . . . . 27
Process Variables Determining Workability. . . . . . . . . . . . . . . . . 30 Chapter 9 Hot Working Simulation by Hot
Workability Fracture Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 Torsion Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
Process Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 Joseph R. Pickens
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Types of Hot Working Simulation Tests . . . . . . . . . . . . . . . . . . 122
Hot Torsion Testing Practice . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Chapter 3 Evolution of Microstructure during Hot Working. . . . . . . . 35 Hot Torsion Application Examples . . . . . . . . . . . . . . . . . . . . . . 124
S.L. Semiatin Chapter 10 Thermomechanical Testing . . . . . . . . . . . . . . . . . . . . . . 128
Mechanisms of Microstructure Evolution . . . . . . . . . . . . . . . . . . 35 Stphane Guillard and Koushik Ray
Phenomenology of Plastic Flow and Microstructure Typical Types of Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
Evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Thermophysical Properties Tests . . . . . . . . . . . . . . . . . . . . . . . . 129
Mechanistic Models for Microstructure Evolution. . . . . . . . . . . . 39 Designing Thermomechanical Tests . . . . . . . . . . . . . . . . . . . . . 131
Obtaining, Analyzing, and Using Thermomechanical
Workability Testing Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Testing Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
Examples of Thermomechanical Testing: Design,
Chapter 4 Bulk Workability Testing . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Experiment, and Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 132
George E. Dieter
Primary Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Process Design and Workability. . . . . . . . . . . . . . . . . . . . . . . . . . . 137
Specialized Tests. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Workability Analysis Using the Fracture Limit Line . . . . . . . . . . 54 Chapter 11 Design for Deformation Processes . . . . . . . . . . . . . . . . . 139
Why Use Deformation Processes?. . . . . . . . . . . . . . . . . . . . . . . 139
Chapter 5 Cold Upset Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 Characteristics of Manufacturing Processes. . . . . . . . . . . . . . . . 140
Howard A. Kuhn Categories of Deformation Processes . . . . . . . . . . . . . . . . . . . . 144
Upset Test Technique . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 Cold Working . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
Test Characteristics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 Hot Working . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
Fracture Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 Forgeability of Alloys . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
Workability Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Chapter 12 Workability Theory and Application in
Chapter 6 Hot Compression Testing. . . . . . . . . . . . . . . . . . . . . . . . . . 61 Bulk Forming Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
George E. Dieter Howard A. Kuhn
Cylindrical Compression Test . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 Stress and Strain States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
Ring Compression Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 Empirical Criterion of Fracture . . . . . . . . . . . . . . . . . . . . . . . . . 173
Plane-Strain Compression Test . . . . . . . . . . . . . . . . . . . . . . . . . . 65 Theoretical Fracture Models and Criteria . . . . . . . . . . . . . . . . . 177
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
iii
1119 FM 7/29/03 3:39 PM Page iv

Chapter 13 Workability in Forging. . . . . . . . . . . . . . . . . . . . . . . . . . 188 Forward (Direct) Extrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299

S.L. Semiatin Backward (Indirect) Extrusion . . . . . . . . . . . . . . . . . . . . . . . . . 302
Workability Tests for Open-Die Forging of Cast Structures . . . . 189 Cold Extrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 303
Workability Tests for Hot/Warm Open-Die Forging of Hot Extrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
Recrystallized Structures. . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 Hydrostatic Extrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310
Workability Tests for Cold Open-Die Forging of Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 314
Recrystallized Structures. . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Workability in Closed-Die Forging . . . . . . . . . . . . . . . . . . . . . . 201 Chapter 21 Workability and Process Design in
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205 Extrusion and Drawing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
Roger N. Wright
Multipass Workability versus Single-Pass Workability . . . . . . . 316
Chapter 14 Process Design in Impression Die Forging. . . . . . . . . . . 208 General Observations on Fracture and Flaw Development. . . . . 316
T. Altan and M. Shirgaokar Shapes and Tubes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 319
Forging Process Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209 Some Comments on Fracture Mechanisms . . . . . . . . . . . . . . . . 319
Design of Finisher Dies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211 Projecting Workability from Mechanical Tests . . . . . . . . . . . . . 319
Preform (Blocker) Design in Impression-Die Forging . . . . . . . . 212 Process Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
Prediction of Forging Stresses and Loads . . . . . . . . . . . . . . . . . 213
Process Simulation to Predict Metal Flow and Multidisciplinary Process Design and Optimization . . . . . . . . . . . 323
Forging Stresses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
Cold and Warm Forging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218 Chapter 22 Multidisciplinary Process Design and
Optimization: An Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
H.L. Gegel, R. Grandhi, J.S. Gunasekera, and Charles Gure
Chapter 15 Modeling Techniques in Forming Processes. . . . . . . . . . 220
Why Use MPDO?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
W.T. Wu, J.T. Jinn, and C.E. Fischer
Process Design Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
Modeling Techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Computer Aided Optimization . . . . . . . . . . . . . . . . . . . . . . . . . 329
Finite Element Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
Manual Design Process. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329
Heat Transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225
Example Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226 Chapter 23 Optimal Design of Thermomechanical
Current and Future Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230 Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 337
J.C. Malas, W.G. Frazier, H.L. Gegel, and V. Srinivasan
Chapter 16 Rolling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232 Concepts of Dynamic Modeling in Optimal Design . . . . . . . . . 338
Flat Rolling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 Material Trajectory Optimization . . . . . . . . . . . . . . . . . . . . . . . 341
Modeling of Strip Rolling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
Mechanics of Plate Rolling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241 Chapter 24 Application of Multidisciplinary
Shape Rolling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242 Optimization (MDO) Techniques to the
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246 Manufacturing of Aircraft Engine Components. . . . . . . . . . . . . . . 346
Shesh K. Srivatsa
Chapter 17 Thermomechanical Processing by Turbine Disk Design and Manufacturing. . . . . . . . . . . . . . . . . . 346
Controlled Rolling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248 Optimization Techniques. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348
Fundamentals of Controlled Rolling . . . . . . . . . . . . . . . . . . . . . 248 Collaborative Optimization Environment . . . . . . . . . . . . . . . . . 349
Restoration Processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 Formulation of Forging Shape Optimization . . . . . . . . . . . . . . . 351
Conventional Controlled Rolling. . . . . . . . . . . . . . . . . . . . . . . . 252 Alternative Forging Optimization Formulations . . . . . . . . . . . . 357
Recrystallization Controlled Rolling . . . . . . . . . . . . . . . . . . . . . 254 Heat Treatment Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . 359
Thermomechanical Processing of Microalloyed Bar . . . . . . . . . 254 Summary of Optimization Techniques . . . . . . . . . . . . . . . . . . . 363
Summary and Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . 256 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 364
Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 365
Chapter 18 Workability and Process Design in Rolling . . . . . . . . . . 258 Chapter 25 Computer-Aided Optimization for Improved
John G. Lenard Process Engineering Productivity of Complex Forgings . . . . . . . . 368
The Rolling Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258 Ramana V. Grandhi
Workability in Rolling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264 Computer-Aided Optimization (CAO). . . . . . . . . . . . . . . . . . . . 369
Mathematical Models for Workability Prediction . . . . . . . . . . . 270 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
Process Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274 Reference Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
Chapter 19 Drawing of Wire, Rod, and Tube . . . . . . . . . . . . . . . . . . 278 Tabular Summaries of Typical Flow Stress Data . . . . . . . . . . . . . . . . 379
Drawing of Bar and Wire . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278 Table 1 K and n values for flow stress-strain relation,
K(e )n, of various steels . . . . . . . . . . . . . . . . . . . . . . . . . 379
s
Tube Drawing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
Lubrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284 Table 2 C and m values for the flow stress-strain rate
relation, s C(e )m, of steels at various temperatures . . . . . . 380
Wear . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289 Table 3 K and n values for the flow stress-strain
relation, s K(e )n, of various aluminum alloys . . . . . . . . . . 382
Chapter 20 Extrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291 Table 4 C and m values for the flow stress-strain rate relation,
Extrusion Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 291 C( e )m, of aluminum alloys at various temperatures . . . . 383
s
The Extrusion Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 294 Table 5 K and n values for the flow stress-strain
Friction and Lubrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297 relation, s K(e )n, of various copper alloys . . . . . . . . . . . . 384
iv
1119 FM 7/29/03 3:39 PM Page v

Table 6 C and m values for the flow stress-strain rate Table 8 Average flow stress values obtained from ring
C( e )m, of various nonferrous alloys . . . . . . . . . 384
relation, s compression tests suggested for use in practical applications. 385
Table 7 Average flow stress values determined in the
Abbreviations and Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 386
uniform compression test that might be used in
practical load-predicting applications . . . . . . . . . . . . . . . . . . 385 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
v
1119 FM 7/29/03 3:39 PM Page vi

Preface
Workability is a vital aspect of the processing of materials, having roots At the time of the formulation of Workability Testing Techniques, the
in both material behavior and process design. Whether a part can be pro- use of finite element methods (FEMs) for the modeling and simulation of
duced by plastic deformation without cracking or the generation of other metal deformation processes was in its infancy. In the ensuing 20 years,
defects is of important economic consequence. Because of the complex na- the use of FEM analysis for process design has become rather common-
ture of the workability of metals, there is no single test that can be used to place. Therefore, in contemplating this revision and update, the editors
evaluate it. Several laboratory tests have been developed that are useful in decided to expand the scope to incorporate process design, especially as
screening materials for workability, but in other instances, very specialized influenced by FEM analysis. By doing this, the Handbook of Workability
tests that are specific to the process are commonly used. and Process Design takes on a more mathematical flavor than its prede-
The Handbook of Workability and Process Design is an update and ex- cessor while still retaining a balance with its original intent. Thus, the
pansion in scope of Workability Testing Techniques that was published by chapters that describe the various workability tests continue true to the
the American Society for Metals in 1984. This original work was devel- original intent of providing practical workability testing techniques that
oped by the Metal Working Group of ASM to provide a readily available can be used by the inexperienced practitioner.
description and interpretation of the most common workability tests in the We appreciate the contributions from the many experts who have con-
deformation processing of metals. Prior to its introduction, this informa- tributed to this Handbook. Also, special thanks go to Steve Lampman, of
tion was widely scattered in the literature. The nearly 20 year life of this the ASM staff, who not only provided editorial guidance throughout this
book bears witness to the value and acceptance of the concept behind this project but also expertly provided the chapters that describe the basics of
project. forging, rolling, extrusion, and wiredrawing.
George E. Dieter Howard A. Kuhn S. Lee Semiatin

College Park, MD Johnstown, PA Dayton, OH
May 1, 2003 May 1, 2003 May 1, 2003
vi
ASM International is the society for materials
engineers and scientists, a worldwide network
dedicated to advancing industry, technology, and
applications of metals and materials.
ASM International, Materials Park, Ohio, USA

This publication is copyright ASM International. All rights reserved.
Publication title Product code
Handbook of Workability and Process Design #06701G
To order products from ASM International:

Online Visit www.asminternational.org/bookstore
Telephone 1-800-336-5152 (US) or 1-440-338-5151 (Outside US)

Fax 1-440-338-4634
Customer Service, ASM International
Mail
9639 Kinsman Rd, Materials Park, Ohio 44073-0002, USA
Email CustomerService@asminternational.org
American Technical Publishers Ltd.
27-29 Knowl Piece, Wilbury Way, Hitchin Hertfordshire SG4 0SX,
In Europe United Kingdom
Telephone: 01462 437933 (account holders), 01462 431525 (credit card)
www.ameritech.co.uk
Neutrino Inc.
In Japan Takahashi Bldg., 44-3 Fuda 1-chome, Chofu-Shi, Tokyo 182 Japan
Telephone: 81 (0) 424 84 5550
Terms of Use. This publication is being made available in PDF format as a benefit to members and
customers of ASM International. You may download and print a copy of this publication for your
personal use only. Other use and distribution is prohibited without the express written permission of
ASM International.
No warranties, express or implied, including, without limitation, warranties of merchantability or
fitness for a particular purpose, are given in connection with this publication. Although this
information is believed to be accurate by ASM, ASM cannot guarantee that favorable results will be
obtained from the use of this publication alone. This publication is intended for use by persons having
technical skill, at their sole discretion and risk. Since the conditions of product or material use are
outside of ASM's control, ASM assumes no liability or obligation in connection with any use of this
information. As with any material, evaluation of the material under end-use conditions prior to
specification is essential. Therefore, specific testing under actual conditions is recommended.
Nothing contained in this publication shall be construed as a grant of any right of manufacture, sale,
use, or reproduction, in connection with any method, process, apparatus, product, composition, or
system, whether or not covered by letters patent, copyright, or trademark, and nothing contained in this
publication shall be construed as a defense against any alleged infringement of letters patent,
copyright, or trademark, or as a defense against liability for such infringement.
Handbook of Workability and Process Design Copyright 2003 ASM International
G.E. Dieter, H.A. Kuhn, and S.L. Semiatin, editors, p3-21 All rights reserved.
DOI:10.1361/hwpd2003p003 www.asminternational.org
Chapter 1
Workability and Process Design

An Introduction
WROUGHT FORMS are produced by a wide basic objective of process design and control. and product modeling and simulation, and sen-
variety of metalworking operations that can be Material control is also important, because many sors and model based process control. Effective
classified either as bulk working or sheet form- problems can be traced back to the process of application of such methods and tools leads to
ing. Bulk working operations (Fig. 1) (Ref 1) in- melting and solidification. For example, poros- process design and controls that prevent defects
clude (a) primary operations where cast ingots, ity and shrinkage cavities (pipe) in an ingot can and reach the full advantages offered by bulk-
continuously cast bars, or consolidated powder be passed on to the wrought form (Fig. 4). forming operations in the production of high-
billets are worked into mill shapes (such as bar, The demands of high-performance products performance components.
plate, tube, sheet, wire) and (b) secondary oper- and rapid product development also can be This Handbook focuses on bulk-forming
ations where mill products are further formed major concerns in the manufacturing of new processes, the defects that frequently occur in
into finished products by hot forging, cold forg- products. For example, metallurgical features such processes, and the roles of materials test-
ing, drawing, extrusion, straightening, sizing, (such as dispersoids and reinforcing particles) ing, process design, and process control in
and so forth. From a geometric viewpoint, bulk that lead to desirable properties in modern high- avoiding defects. The types of workability prob-
forming operations are distinguished by large performance materials may also render them lems that may occur are introduced first, and the
changes in cross-sectional area (e.g., round bar hard-to-work. At the same time, rapid product general concepts of process modeling for de-
extrusion or flat rolling) and may be accompa- and process development have achieved a high signing and controlling bulk working processes
nied by large changes in shape (e.g., impression level of sophistication through the use of modern are described. This introductory chapter also in-
die forging or shape rolling). In contrast, sheet design methods and tools, enhanced by comput- cludes a brief overview on modeling of bulk
forming operations (Fig. 2) (Ref 1) typically in- erization. These innovations can be classified as forming processes by numerical techniques such
volve large changes in shape (e.g., cup forming materials testing and data acquisition, process as finite element analysis (FEA). Finite element
from a flat blank) without a significant change in
the sheet thickness. Sheet forming has several
characteristics that distinguish it from bulk
working; for example, sheet formability in-
cludes different criteria such as springback and
the resistance of a sheet material to thinning.
Sheet formability and process design are not ad-
dressed in this Handbook and are left as topics
for another publication. Nevertheless, many of
the same concepts and methods described in the
present Handbook can be applied to sheet form-
ing processes.
While the major role of bulk forming opera-
tions is to produce the desired shape, in doing so
they also modify the material structure and sur-
face. Generally, the effects of bulk working
processes are beneficial, leading to improved in-
ternal quality (closure of casting porosity, re-
finement of grain structure, and grain alignment)
as well as improved surface quality (burnished
surfaces and worked surface material). However,
the large amount of metal movement during bulk
forming operations also can introduce material
discontinuities that are potential defects (i.e.,
imperfections that degrade intended function).
Some general types of surface and internal
discontinuities of wrought products are illus-
trated in Fig. 3 for the example of rolled bar.
Imperfections such as segregation, porosity, or
seams can influence the potential or likelihood
of a defective part, and their prevention is one Fig. 1 General classification of bulk deformation processes. Source: Ref 1
4 / Introduction
Fig. 4 Longitudinal sections of two types of ingots

showing typical pipe and porosity. Pipe is a con-
dition that develops in the nominal top centerline of the
ingot because of metal shrinkage during solidification.
Fig. 2 General classification of sheet forming processes. Source: Ref 1 When the ingots are rolled into bars, pipe and porosity be-
come elongated throughout the center of the bars.
Fig. 3 Ten different types of imperfections that might be found in rolled bar
analysis is an effective method for solving a part produced by the bulk working process is un- that the force that must be applied by the pro-
wide variety of engineering problems and has acceptable and must be scrapped or reworked. cessing equipment and the stresses on the dies
been useful in the design and analysis of both From this practical point of view, workability are lower. On the other hand, a poorly designed
bulk and sheet forming processes (Ref 2). also may be defined by other factors such as the or controlled process can produce defects in an
generation of a rough surface finish or the in- easy-to-work material, leading to a scrap part.
ability to achieve a required tolerance on a criti- By way of example, lead is a very ductile and
Workability Problems cal dimension. workable material that can be formed readily by
Workability also is a complex technological compressive operations such as forging and
Workability refers to the relative ease with concept that is related to both material and rolling; yet, lead fails to form easily in tensile
which a material can be shaped through plastic process characteristics. Workability depends not operations such as drawing.
deformation. Workability is usually thought of only on the fracture resistance (ductility) of the A hard-to-work alloy presents other chal-
as being limited by the onset of fracture. Greater material but also on the specific details (stress lenges to provide a deformation process envi-
workability of a material allows greater defor- state) of the process as influenced by die geom- ronment that prevents defect formation. The
mation and/or a more complex shape that can be etry, workpiece geometry, and lubrication condi- evaluation of a material by the various workabil-
produced before fracture occurs. In general, tions. Ease of manufacture is aided when the ity testing and analysis methods described in this
however, a workability problem occurs when the material has a low flow stress (yield strength) so Handbook provide a framework for intelligently
Chapter 1: Workability and Process DesignAn Introduction / 5
choosing materials for best workability or for such as triple-point cracking/cavitation at hot The fracture locus concept has been used to pre-
changing the design of the process to enhance working temperatures or inhomogeneous defor- vent free surface cracking in forging and to pre-
the workability of a given material. As a first mation that cause internal defects such as central vent edge cracking in rolling. With modifica-
step toward devising a solution to a workability bursts or chevron cracking. Internal fracture can tions, the fracture locus approach has also
problem, it is useful to categorize workability be an extremely dangerous type of defect be- provided insight into such failure modes as cen-
problems in terms of their general source: cause it cannot be detected visually. ter bursting in extrusion and forging and die-
Free Surface Cracking (Adapted from Ref workpiece contact fractures in forging. These
Fracture-related problems: for example, in-
3). A free surface, by definition, has neither a limits change with chemistry, grain size, temper-
ternal bursts or chevron cracks; cracks on
stress normal to it nor a shear stress on it. Free ature, second-phase content, and possibly with
free surfaces; cracks on die-contacted sur-
surface fractures occur on surfaces undergoing strain rate.
faces
free expansion due to compressive loads on con- The concept of a working limit for free sur-
Metal-flow-related problems: for example,
tact surfaces between the tools (rolls or dies) and face fracture is important because the workabil-
end grain and poor surface performance; in-
the workpiece. The tensile stresses leading to ity of a metal may be characterized for a partic-
homogeneous grain size; shear bands and
free surface fracture are often referred to as sec- ular set of process conditions. In general, as the
locally weakened structures; cold shuts,
ondary stresses since they are not applied di- working temperature is increased, the location
folds, and laps; flow-through defects
rectly by metalworking equipment. Edge crack- of the fracture line will move upward, indicating
Control, material selection, and utilization
ing in rolling of plates, strip, or rings is a that higher deformation can be accommodated
problems: for example, underfill, part dis-
primary example of free surface cracking. before fracture. However, higher temperatures
tortion, and poor dimensional control; tool
Another is the surface cracking occurring on ex- are not always beneficial, as in the case of IN718
overload and breakage; excessive tool wear;
posed expanding surfaces during upsetting or on nickel-base superalloy, which has a temperature
high initial investment due to equipment
the leading edges of localized areas of extrusion limit of about 1120 C (2050 F) for hot work-
cost; poor material utilization and high
in forgings. ing (Ref 5 and Chapter 12, Workability Theory
scrap loss
One of the most successful and useful design and Application in Bulk Forming Processes in
These types of problems are introduced briefly tools to come from bulk workability research is this Handbook). Experimental observations have
in the following sections of this chapter. Each the workability diagram for free surface fracture shown that the slope of the line increases with
type of problem may involve different kinds of during the cold working of wrought and powder strain rate for some metals, most notably some
methods or solutions in the design of the process metals. An example of a workability diagram of brasses and austenitic stainless steels (Ref 6).
and/or product. For example, workability can be this type is shown in Fig. 5 from Chapter 3 of The position of the line drops (lower workabil-
improved by changes in die geometry, work- Ref 4, Cold Upset Testing. The graph indicates ity) as the second-phase content increases, much
piece geometry, lubrication conditions, or pro- the locus of free surface normal strains (one ten- like the tensile ductility decreases with second
cessing temperature. Much plastic deformation sile and one compressive) that cause fracture. phase. The fracture line location is also sensitive
of metals is carried out at elevated temperature The workability diagrams are used during to the microstructure. For example, a spher-
(hot working) because flow stress decreases process design by plotting calculated or esti- oidized structure for a high-carbon steel has a
with increasing temperature. mated surface strain paths that are to be imposed higher fracture-line position than a pearlitic mi-
The same general approach applies to sheet during forming on the fracture locus diagram crostructure has. More details and references on
forming operations in that the major emphasis of (Fig. 5). If the final strains lie above the locus, workability in terms of free surface cracking are
formability evaluation also is on measuring and part failure is likely, and changes are necessary in Chapter 11, Design for Deformation
predicting the limits of deformation before frac- in preform design, lubrication, and/or material. Processes.
ture. Sometimes the term formability is used in
conjunction with either the sheet formability or
bulk workability. However, the term formability
is limited more properly to sheet forming opera-
tions because there are major distinctions in the
conditions of sheet forming and bulk forming
processes. In sheet forming, metal is deformed
plastically by tensile loads, often without signif-
icant changes in sheet thickness or surface char-
acteristics and with the possibility of significant
elastic recovery or springback because the mag-
nitudes of plastic and elastic deformation may
be similar. In contrast, metal is deformed plasti-
cally by compressive loads during bulk forming,
and the plastic deformation is proportionally
much more prevalent than elastic deformation.
Fracture-Related Problems
The general types of fracture in bulk working
are:
Free surface fracture
Die contact surface fracture
Internal fracture
The most common types of fracture failures in
bulk working are free surface fracture (at hot or
cold processing temperatures) and internal frac- Fig. 5 Schematic workability diagrams for bulk forming processes. Strain path a would lead to failure for material A.
ture. Internal fracture occurs by mechanisms Both strain paths (a and b) can be used for the successful forming of material B. Source: Ref 4
6 / Introduction
Die-Contact Surface Cracking. Cracking caused by high hydrostatic tension in combina- ciently high to tear the material apart internally,
on surfaces in contact with a die is a common tion with internal material weaknesses. Chevrons particularly if the forging temperature is too
problem. Frequently, cracks occur during forg- are internal flaws named for their shape (Fig. 3k). high (Ref 8). Similarly, if the metal contains
ing on surfaces that are in contact with the dies. A central burst, or chevron crack, is associated low-melting phases resulting from segregation,
One common location of such defects is in the most commonly with extrusion and drawing op- these phases may rupture during forging. Ingot
vicinity of a die or punch corner. A combination erations, although it can be generated by forging pipe, unhealed center conditions, or voids asso-
of shear deformation with tension or low values and rolling processes as well. Internal bursts in ciated with melt-related discontinuities may also
of hydrostatic pressure in the vicinity of a rolled and forged metals result from the use of induce center bursts if reduction rates are too se-
die corner are responsible for surface cracking equipment that has insufficient capacity to work vere or temperatures are incorrect during work-
(Ref 7). the metal throughout its cross section. If the ing. The conversion practice to bar or billet must
Observation of a variety of such defects working force is not sufficient, the outer layers of impart sufficient homogenization or healing to
shows that an apparently common characteristic the metal will be deformed more than the inside produce a product with sound center conditions.
is an abrupt change in frictional shear traction metal, sometimes causing wholly internal, inter- An example of an unsound condition that did not
distribution in the region of the crack. High fric- granular fissures that can act as initiation sites for heal is shown in Fig. 6.
tion to retard metal flow in advance of the crack further crack propagation during service loads. In It also is useful to point out that, if the stress
location is one method for preventing such de- forward cold extrusion, the occurrence of central state is compressive in areas where material
fects. These cracks usually do not propagate bursts or chevrons is nearly always restricted to weaknesses occur, the compressive stress state
deeply into the workpiece but instead result in isolated lots of material and usually to only a will close up any porosity or pipe and will retard
unacceptable surface quality or unacceptable small percentage of the pieces extruded in any any inclusions or segregation from initiating
machining depths if that surface is to be finish particular production run. cracks. The stress state, as controlled by the
machined. As the forging community has moved A change in deformation zone geometry is process parameters, thus has an important role in
closer to net-shape forming, this type of defect usually sufficient to eliminate the problem. The the degree of soundness in the worked material.
has become an increasing problem. Causes of conservative design approach is to ensure that no The classic work by Lou Coffin and Harry
this problem include nonuniform lubrication, hydrostatic tension develops. Often, however, Rogers (Ref 9) showed very clear relationships
flow around die corners, and an improper start- the part or tooling design cannot be changed suf- between damage generation and tensile stress
ing workpiece surface. ficiently to eliminate hydrostatic tension. If the states (as well as damage healing and compres-
Central or Internal Bursts (Chevron level of hydrostatic tension can be kept below a sive stress states) during deformation processing.
Cracking). Central bursts are internal fractures critical level, bursting can likely be avoided. Macroetching and ultrasonic inspection meth-
This may be accomplished by a change in lubri- ods are the most widely used for identifying re-
cant, die profile, temperature, deformation level, gions of unsoundness. Bursts usually display a
or process rate. distinct pattern of cracks and do not show
The probability of internal burst is enhanced spongy areas, thus distinguishing them from
in areas of material weakness if they are acted on pipes. Bursts are readily detected by macroetch-
by hydrostatic tension stress states during the ing. Figure 7 shows a large burst that occurred
deformation process. For example, with ingot during the forging of an electroslag-remelted
imperfections (such as pipe, porosity, segrega- (ESR) ingot. The cause was traced to a weak so-
tion, or inclusions), tensile stresses can be suffi- lidification plane near the bottom of the ingot
Fig. 6 Section through a heat resistant alloy forging

showing a central discontinuity that resulted
from insufficient homogenization during conversion. Step
machining was used to reveal the location of the rupture; Fig. 7 Cross section of a forged bar showing a forging burst. The burst is located approximately at the centerline of
original diameter is at right. the workpiece. Arrow indicates the direction of working.
combined with higher than normal forging tem-

peratures.
Flow-Related Defects
The defects described in this section are re-
lated to the distribution of metal. They can be
avoided by proper die design, preform design,
and choice of lubrication system. Strictly speak-
ing, these defects are not fundamental to the
workability of the material but are related to the
process details. However, knowledge of these
common forging defects is necessary for a prac-
tical understanding of workability. These are the
defects that commonly limit deformation in sec- Fig. 8 Micrograph of a forging lap. Note the included
ondary operations. oxide material in the lap. 20
Most types of flow-related defects occur in
hot forging, which is described in more detail in
Chapter 13, Workability in Forging. However, fillets are initially large and progressively be-
the following provides a general introduction to come smaller as the forging steps are completed.
typical types of defects that may occur from Figure 9 (Ref 10) shows schematically a lap
bulk working. They are common for impression- forming as metal flows around a die corner.
die forging, in which case defect formation may Extrusion-Type Defects. The tail of an extru-
also involve entrapment of oxides and lubricant. sion is unusable because of nonuniform flow
When this occurs, the metal is incapable of through the extrusion die. This results in a cen-
rewelding under high forging pressures; the term ter-to-surface velocity gradient, with metal from
cold shut is frequently applied in conjunction the workpiece interior moving through the die at
with laps, flow-through defects, and so on to de- a slightly higher velocity than the outer material.
scribe the flaws generated. The result shows up at the tail of the extrusion as
Underfill may not seem like a flow-related a suck-in or pipe, and, for extrusions, the tail is
defect, but aside from simple insufficient start- simply cut off and discarded. Alternatively, a
ing mass, the reasons for underfill are flow re- follower block of cheaper material may be
lated. These include improper fill sequence, in- added so that most of the defect falls in the
sufficient forging pressure, insufficient preheat cheaper material, and less length of the extruded
temperature, lubricant buildup in die corners, workpiece is lost.
poor or uneven lubrication, and excessive die For forgings that involve forward or back-
chill. An improper fill sequence may result in ward extrusion to fill a part section, the same
excessive flash loss, or it may be the result of ex- situation can develop. Metal flow into a rib or
traordinary pressure requirements to fill a partic- hub can result in a suck-in defect, which in a
ular section. Sometimes, venting may eliminate worst-case scenario would show up as a fold on
the problem; more often than not, a change in the face opposite to the rib. A best case would
the incoming workpiece shape or a change in the be a depression on what otherwise should be a
deformation sequence is required. flat surface. One method of eliminating this
Laps and Folds. Laps are surface irregularities type of defect is to position more material on Fig. 9 Lap formation in a rib-web forging caused by
improper radius in the preform die. Source:
that appear as linear defects and are caused by the the back face initially. Another method is to Ref 10
folding over of hot metal at the surface. These change the rib geometry (aspect ratio and/or an-
folds are worked into the surface but are not met- gles). If neither of these changes can be accom-
allurgically bonded (welded) because of the plished, an extra forging step may be needed to
oxide present between the surfaces (Fig. 8). Thus, limit the amount of extrusion that is done in any Fig. 11 (Ref 7) for a trapped-die forging that has
a discontinuity with a sharp notch is created. one step. a rib on the top surface. The rib denoted by 2
In rolling, laps are most often caused by ex- Extrusion-type defects are formed when cen- is filled early in the forging sequence, and sig-
cessive material in a given hot roll pass being trally located ribs formed by extrusion-type flow nificant mass must flow past the rib in order to
squeezed out into the area of the roll collar. draw too much metal from the main body or web fill the inner hub, zone 4. The result can be a
When turned for the following pass, the material of the forging. A defect similar to a pipe cavity complete shearing-off of the rib in the worst
is rolled back into the bar and appears as a lap on is thus formed (Fig. 10). Methods of minimizing case, with a lesser case being the formation of a
the surface (Fig. 3). A lap or fold occurs in die the occurrence of these defects include increas- shear-type crack.
forging from an improper progression in fill se- ing the thickness of the web or designing the Similar to laps in appearance, flow-through
quence. Normally, a lap or fold is associated forging with a small rib opposite the larger rib, defects can be shallow, but they are indicative of
with flow around a die corner, as in the case of as shown in Fig. 10. an undesirable grain flow pattern or shear band
an upper rib or lower rib, or with a reversal in Shear-Related or Flow-Through Defects. that extends much deeper into the forging. An
metal-flow direction. Shearing defects are also known as flow-through example is shown in (Fig. 12) (Ref 12). Flow-
In die forging, a general rule of thumb is to defects because they result from excessive metal through defects can also occur when trapped lu-
keep metal moving in the same direction. The flow past a filled detail of the part. Flow-through bricant forces metal to flow past an impression.
die corner radius is a critical tool dimension, and defects are formed when metal is forced to flow Seams are crevices in the surface of the metal
it should be as generous as possible. In pro- past a recess after the recess has filled or when that have been closed, but not welded, by work-
gressing through a forging sequence, the die cor- material in the recess has ceased to deform being the metal. Seams result from elongated
ners should become tighter so that the workpiece cause of chilling. An example of this is shown in trapped-gas pockets or from cracks. Even
8 / Introduction
Seams may be continuous or intermittent, as

indicated in Fig. 3(h). Depth of seams varies
widely, and surface areas sometimes may be
welded together in spots. Seams seldom pene-
trate to the core of bar stock. Seams can be
difficult to detect because they may appear as
scratches or because a machining process
may obliterate them. Seams are normally closed
tight enough that no actual opening can be de-
tected visually without some nondestructive in-
spection techniques such as magnetic particle
inspection. Figure 14 is an example of a seam
detected by routine magnetic particle inspection
of a hot-rolled 4130 steel bar. No stringer type
inclusions were observed in the region of the
flaw, but it did contain a substantial amount of
Fig. 11 Schematic of a flow-through crack at the base
oxide (Fig. 14b).
of a rib in a trapped-die forging. Excessive
metal flow past region 2 causes a shear crack to form at A Seams may not become evident until the part
and propagate toward B. Source: Ref 7 has been subjected to installation and service
stresses. For example, seams are sometimes dif-
also result from highly stressed planes in cold- ficult to detect in an unused fastener, but they are
drawn bars or from improper quenching during readily apparent after a fastener has been sub-
heat treatment. Cracks created from these latter jected to installation and service stresses. Seams
Fig. 10 Extrusion-type defect (a) in centrally located
two causes show no evidence of oxidized sur- also may not become evident until the constraint
rib and (b) die-design modification used to exerted by the bulk of material is removed from
avoid defect. Source: Ref 11 faces. Seams are generally heavily oxidized and
decarburized in steels (Fig. 13). the neighborhood of a seam. The incomplete re-
Seams have a large number of possible ori- moval of seams from forging stock can cause ad-
though seams can occur from cracks, the two gins, some mechanical and some metallurgical. ditional cracking in hot forging and quench
can be distinguished from the presence of ox- Seams can develop from cracks caused by work- cracking during heat treatment.
ides. In mill processing, for example, cracks ing or from an imperfection in the ingot surface,
with little or no oxide present on their edges may such as a hole, that becomes oxidized and is pre- Microstructure and Properties
occur when the metal cools in the mold, setting vented from healing during working. In this
up highly stressed areas. Seams develop from case, the hole simply stretches out during forg- A product with poor properties is another po-
these cracks during rolling as the reheated outer ing or rolling, producing a linear seam in the tential problem, and it usually arises from an in-
skin of the billet becomes heavily oxidized, workpiece surface. Seams also result from adequate microstructure such as grain flow and
transforms into scale, and flakes off the part dur- trapped-gas pockets, cracks, a heavy cluster of grain size. A major advantage of shaping metal
ing further rolling operations. In contrast, cracks nonmetallic inclusions, or a deep lap. parts by rolling, forging, or extrusion stems from
the opportunities such processes offer the de-
signer with respect to the control of grain flow.
The strength of these and similar wrought prod-
Fig. 13 Micrograph of a seam in a cross section of a 19

mm (3/4 in.) diameter medium-carbon steel bar
Fig. 12 Flow-through defect in Ti-6Al-4V rib-web structural part. Source: Ref 12 showing oxide and decarburization in the seam. 350
ucts is almost always greatest in the longitudinal

(or equivalent) direction of grain flow. The max-
imum load-carrying ability in the finished part is
attained by providing a grain flow pattern paral-
lel to the direction of the major applied service
loads when, in addition, sound, dense, good-
quality metal of satisfactorily fine grain size has
been produced throughout.
Grain Flow and Anisotropy. Metal that is
rolled, forged, or extruded develops and retains
a fiberlike grain structure that is aligned in the
principal direction of working. This characteris-
tic becomes visible on external and sectional
surfaces of wrought products when the surfaces
are suitably prepared and etched (Fig 15). The Fig. 15 Section through a forged head on a threaded
fibers are the result of elongation of the mi- fastener showing uniform grain flow that min-
crostructural constituents of the metal in the di- imizes stress raiser and unfavorable shear planes
rection of working. Thus the phrase direction of
grain flow is commonly used to describe the
dominant direction of these fibers within
wrought metal products from the crystallo- size for good workability) and an optimal work-
graphic reorientation of the grains during defor- ing temperature for adequate workability with-
mation and/or the alignment of nonmetallic in- out excessive growth of grain size.
clusions, voids, and chemical segregation. The Control of grain size during thermomechani-
occurrence and severity of fibering varies with cal processing is a topic of significant practical
such factors as composition, extent of chemical importance, and extensive efforts have been em-
segregation, and the amount of work or reduc- ployed in this regard. For example, the concept
tion the workpiece receives. of grain size control has been used for many
In wrought metal, the direction of grain flow years in the production of flat-rolled products.
is also evidenced by measurements of mechani- Small niobium additions increase the austenite
cal properties. Strength and ductility are almost recrystallization temperature, and controlled
always greater in the direction parallel to that of rolling is used to refine the relatively coarse
working. The characteristic of exhibiting differ- austenite structure by a series of high-tempera-
ent strength and ductility values with respect to ture rolling and recrystallization steps. It com-
the direction of working is referred to as me- bines repeated deformation and recrystallization
chanical anisotropy and is exploited in the de- steps with the addition of austenite grain-growth
sign of wrought products. Although best proper- inhibitors such as titanium nitride to refine the
ties in wrought metals are most frequently the starting austenitic grain size and to restrict grain
longitudinal (or equivalent), properties in other Fig. 14 Seam in rolled 4130 steel bar. (a) Close-up of
growth after recrystallization. This is a topic of
seam. Note the linear characteristics of this
directions may yet be superior to those in prod- flaw. (b) Micrograph showing cross section of the bar. ongoing interest, as described in more detail in
ucts not wroughtthat is, in cast ingots or in Seam is normal to the surface and filled with oxide. 30 the section Microstructural Modeling in this
forging stock taken from ingot only lightly chapter.
worked. Similar metallurgical effects on grain size
Although all wrought metals are mechani- momechanical processing on grain size are crit- apply to other materials and hot working opera-
cally anisotropic, the effects of anisotropy on ical in obtaining products with satisfactory prop- tions. Minor variations in composition may also
mechanical properties vary among different erties. This requires a more extensive evaluation cause large variations in workability, grain size,
metals and alloys. For example, a vacuum- of the dynamic and thermophysical conditions and final properties. In one case, for example,
melted steel of a given composition is generally that influence metallurgical processes at a mi- wide heat-to-heat variations in grain size oc-
less mechanically anisotropic than a conven- croscopic level. Thus, the effects of thermome- curred in parts forged from nickelbase alloy
tionally killed, air-melted steel of the same chanical processing on grain size require more 901 (UNS N09901) in the same sets of dies. For
composition. Response to etching to reveal the complex models based on the principles of phys- some parts, optimal forging temperatures had to
grain flow characteristic of anisotropy also ical metallurgy and the continuum mechanics of be determined for each incoming heat of mate-
varies. Metals with poor corrosion resistance mechanical flow. rial by making sample forgings and examining
are readily etched, whereas those with good Grain size and grain structure also influence them after heat treatment for variations in grain
corrosion resistance require more corrosive workability. Workability typically decreases size and other properties. This illustrates the
etchants and extended etching times to reveal with increasing grain size, because cracks may complexity of process design and modeling of
grain flow. Because grain flow can be a vital initiate and propagate easily along the grain thermomechanical forming.
factor in the service performance of a part, it is boundaries. When the grain size is large relative
useful to use arrows on forging drawings to to the overall size of the workpiece, as in con-
show optimum grain flow direction that corre- ventionally cast ingot structures, hot working is Material Control, Selection, and
sponds to the direction of principal service required because of the low workability of the Equipment Utilization
stress. Effects of grain flow are described in material. In general, the workability of metals
more detail in Chapter 11 Design for increases with temperature. However, as temper- The efficiency and productivity of a forming
Deformation Processes. ature increases, grain growth also occurs. Thus, operation are influenced by a number of factors
Grain Size. The influence of grain size on the design of thermomechanical processes may such as material selection and control, equipment
properties such as strength and ductility is gen- involve a complex set of factors such as materi- cost, tool wear, part distortion, poor dimensional
erally well understood, and the effects of ther- als control (preforms with suitably fine grain control, poor material utilization, and high scrap
10 / Introduction
loss. Material control is important, as material characteristics (such as energy consumption models besides just continuum mechanics.
variation can have significant effects on proper- and maximum productivity) and/or product Microstructural modeling also requires quanti-
ties and characteristics of the product such as properties (such as microstructural homo- fying the thermal field in the materials and the
grain size and mechanical properties. The re- geneity, grain flow characteristics, and grain associated metallurgical phenomena. The practi-
sponsibility of assuring and verifying the proper- size) cal application of such physical metallurgy-
ties and characteristics of the product is vested in based models continues to advance further as
material control, which controls all processes Each objective requires different types of analyt- numerical techniques such as finite element
employed in production, from selection of raw ical tools or models. For example, models based analysis (FEA) become more sophisticated with
material to final inspection. It establishes manu- on continuum mechanics can address flow- improvements in computer hardware. The abil-
facturing standards to ensure reproducibility in related problems such as insufficient die fill, ity to accurately predict thermomechanical his-
processing and product uniformity. Material con- poor shape control, and fracture conditions. tories by finite element modeling is being used
trol depends on the proper application of draw- The four major design considerations in a to predict the evolution of microstructure,
ings, specifications, manufacturing process con- bulk forming process are material flow, worka- thereby replacing data-intensive empirical meth-
trols, and quality-assurance programs to satisfy bility, resultant properties (microstructure) of ods with more knowledge-based analytical/nu-
all requirements for metallurgical integrity, me- the product, and utilization factors (economics, merical methods using the fundamental princi-
chanical properties, and dimensional accuracy. It efficiency, productivity, etc.). The first consider- ples of transport phenomena, continuum
also provides for identification and certification, ation is the workpiece material and its flow mechanics, and physical metallurgy.
so that a product history can be traced. stress behavior. Flow stress is the stress needed The fourth consideration involves added con-
Effective utilization of materials and equip- to cause plastic deformation and is affected by straints of available equipment and economics in
ment also depends on the type of workpiece ma- temperature, rate of deformation, and amount of addition to flow stress, forming, and part per-
terial and forming operation. Other chapters de- previous plastic deformation. Flow stress behav- formance considerations. The fourth considera-
scribe briefly equipment and workability for ior is based on mechanical testing and can be tion usually dominates the other considerations,
specific types of bulk-working operations. modeled by constitutive equations that describe sometimes to the detriment of the material being
Chapter 11 Design for Deformation Processes mathematically the relationship between stress worked. Material utilization may also include
also describes briefly tool materials. A wide and strain during plastic deformation. factors of economic productivity, efficiency, tool
range of materials for tools and dies is available The second consideration is the fracture be- wear, and scrap loss.
to designers. Among the important attributes are havior of the material and the effects of temper-
hardenability; machinability; and resistance to ature, stress state, and strain rate on fracture; this Concepts of Process Modeling
wear, plastic deformation, shock loading, and combined view of ductility and stress state is
heat checking. The needed levels of resistance to termed workability for bulk forming processes. Engineering models are often used to deter-
wear, plastic deformation, and so forth, are de- Even if the desired shape is obtained, cracking mine the response of structure, component,
termined by factors such as type of equipment or some other cracklike discontinuity may occur. process, or system to a set of conditions. The de-
used, workpiece temperature, expected die tem- For example, edge cracking may occur during velopment of any model requires the definition
perature, and number of parts to be fabricated. rolling, or central bursts may occur during ex- of appropriate boundary conditions and the ap-
Low-alloy steels and hot-work die steels are trusion. Cracking of the tool itself may even plication of appropriate mathematical equations
often suitable for conventional metalworking. occur. Many methods have been developed to (which are typically differential equations for
On the other hand, high-temperature die materi- evaluate workability, as described in more detail many physical situations). The complexity of the
als are required for special applications such as in subsequent chapters, with damage criteria de- boundary conditions and the governing differen-
isothermal forging of titanium and nickel-base tailed in Chapter 12 Workability Theory and tial equations determine the possible methods of
alloys. These die materials include various su- Application in Bulk Forming Processes. solution. In some cases, solutions may be ade-
peralloys and TZM molybdenum (Mo-0.5Ti- The third major consideration is a determina- quately achieved by analytical (closed-form)
0.1Zr). Recommendations on the selection of tion of the desired final microstructure needed to equations, while in other cases numerical tech-
these materials are made in Chapter 11 Design produce an acceptable product. Microstructural niques (such as FEA) may be required.
for Deformation Processes. The approach used optimization or prediction requires additional In the case of deformation processes like a forg-
in making these recommendations and the tool
materials are discussed in detail for hot forging
tooling.
Process Design and Modeling
A considerable array of mechanical tests and

modeling techniques has been developed to pre-
vent defects and/or optimize results in bulk
forming operations such as forging, extrusion,
and rolling. In general, the tools and techniques
for process design have one or more of the fol-
lowing objectives:
Prevent improper part shape or final geom-
etry that differs from expectations
Prevent defects or cracking that occur dur-
ing shaping
Prevent poor properties from the develop-
ment of inadequate microstructure
Optimize results in terms of desired process Fig. 16 Typical physical phenomena occurring during a forging operation
ing operation (Fig. 16), the major process vari-

ables and interactions are shown in Fig. 17. It can
be seen in Fig. 17 that for a metal forming analy-
sis, one needs to satisfy the equilibrium condi-
tions, compatibility equations/strain-displacement
relations, constitutive equations, and, in some in-
stances, the heat balance equation. Modeling the
microstructural effects of these variables could, in
principle, require a more robust set of appropriate
mathematical equations for the process phenom-
ena (such as convection, radiation, chemical reac-
tion, and diffusion, in addition to deformation).
Historically, deformation processing has
fallen in the gap between the traditional disci-
plines of metallurgy and mechanics, and, as a re-
sult, this area has often been neglected in an ac-
ademic sense. A main reason for this is the
difference in length scales by which mechanics
and materials science view a material. The
length scale of deformation is at the atomic
level, and it is at this level that materials science
addresses deformation. In manufacturing, defor-
mation effects are related or measured at a
macroscopic level, and continuum mechanics is Fig. 17 Interaction among major process variables during forming
applied to analyze and explain plasticity quanti-
tatively. However, to understand the macro-
scopic response of materials to temperatures and plastic deformation of material during bulk work- 2 u 2 u 2 u g( x, y, z )
rates of deformation, consideration must be at a ing processes, can be described by differential + + + =0 (Poisson equation)
x 2 y 2 z 2 k
lower length scale level, at least to qualitatively equations. Partial differential equations are re-
explain material behavior. The empirical equa- quired when the behavior is a function of time
tions used to address plasticity do not generally and space or of more than one space variable. where the Possion equation includes a heat-
capture the microscopic aspects of deformation. However, if the problem can be simplified to one source function, g(x, y, z), with k as the thermal
Fortunately, the metals that are commonly independent variable (time or one space variable), conductivity of the material.
processed by bulk deformation methods have ordinary differential equations can be used in un- Continuum Mechanics Equations. Con-
many grains per unit volume, and microscopic derstanding the effect of certain parameters on the tinuum mechanic models are based on steady-
events are suitably averaged at the macroscopic response or behavior of the process or object. state equations of mechanical equilibrium and
level. The macroscopic or continuum mechan- Partial differential equations can be divided constitutive equations for mechanical flow. The
ics approach begins to break down when the into three categories: hyperbolic, parabolic, and complete set of equations serves as the founda-
grain size approaches the physical size of the elliptic. Standard hyperbolic equations include tion for continuum mechanics models of bulk
workpiece or when a dominant crystallographic the wave equation: working processes. A model for each case is de-
texture is present in a workpiece. Examples of veloped by imposing appropriate boundary con-
the former include fine wire drawing, bending 1 2u 2u 2u 2u
ditions and initial conditions (tool and work-
of fine wire, and sheet forming processes. = + + piece geometry, temperature, heat flow, etc.) on
c 2 t 2 x 2 y 2 z 2
Primary examples of the latter are sheet-metal the solutions for the set of equations. In addition,
forming processes or bending of heavily drawn models of thermomechanical processes may also
where c is the wave speed. Parabolic partial require description of thermophysical behavior
wire where crystallographic texture plays a
differential equations include the diffusion and the contact interface between the tool and
dominant role. An additional complication is
equation: worked material.
the fact that most metals have more than one
phase present in their microstructure. The sec- Equilibrium and Compatibility (Strain-
ond phase may be present due to alloying (e.g., u 2 u 2 u 2 u Displacement) Equations. Steady-state equa-
D = + + tions include the forces acting on an element and
cementite in iron), or it may be an unwanted t x 2 y 2 z 2
phase (e.g., sulfide or silicate inclusions in the compatibility of strain displacements. The
steel). The effects of these second phases are steady-state equations in describing the various
again averaged at the macroscopic level, and the where D is the diffusivity. forces acting on an element in mechanical equi-
material has not been adequately described as Elliptic equations are usually used to model librium are:
having separate phases from a mathematical steady-state phenomena. When hyperbolic or
sense. The materials science community and the parabolic equations are assumed to be invariant s x t xy t xz
+ + = - Fx
mechanics community are trying to bridge this with time, then they reduce to elliptic equations. x y z
length-scale problem, but for now the most use- For example, when the time dependence is re- t xy s y t yz
ful analysis tools are combinations of contin- moved from the wave equation or the diffusion + + = - Fy
equation, they reduce to the Laplace or Poisson x y z
uum mechanics and empirical results. This is
changing, however, as computer-based numeri- equations for steady-state heat conduction in t xz t yz s z
+ + = - Fz
cal techniques allow more effective modeling solids with constant properties: x y z (Eq 1)
from first principles.
Types of Differential Equations. The physical where s is the normal stress component, is the
response or behavior of a system, such as the 2u 2u 2u shear stress component, and F is the body
+ + =0 (Laplace equation)
x 2 y 2 z 2
12 / Introduction
force/unit volume component. Similarly, the proposed by Sellars and Tegart (Ref 13) by as- where 0 is the uniaxial yield (flow) stress
strain-displacement relationships are given as: suming materials flow during deformation as a measured in tension or compression, and 1, 2,
thermally activated process: 3 are the three principal stresses.
u u v
ex = g xy = + Fracture Criteria. Continuum mechanics can
x y x Q also be used in conjunction with multiaxial frac-
e = A [sinh( as )]n exp -
v v w RT ture criteria to gain insights and solutions on the
ey = g yz = +
y z y prevention of flow related cracks. Various fracture
w w u where A, , and n are constants determined by (or microstructural damage) criteria have been de-
ez = g zx = + fitting empirical data, and Q is the apparent acti- veloped to evaluate workability (e.g., Ref 1721),
z x z (Eq 2) vation energy. At low stresses (
0.8), the but not all are easily incorporated into continuum
where e is the normal strain, is the shear strain, equation reduces to a power law: mechanics models of processes. One criterion that
and u, v, and w are the displacements in the x, y, gives a very accurate description of workability
and z directions, respectively. In addition to Q and is easily implemented in the models is that due
e = A1s n exp -
boundary conditions, the solution of these equa- RT to Crockcroft and Latham (Ref 17):
tions may be complicated further by time de- ef
pendence of the force functions, the nonlinear
stress-strain behavior of plastic deformation
1.2), the equation re-
At high stresses (
duces to an exponential form:
0
s1 d e C
(i.e., constitutive equations), and the flow rules where 1 is the maximum principal stress, ef is
of plastic deformation (such as the Von Mises Q
e = A2 exp(bs )exp - the equivalent strain at fracture, and C is a con-
yield citerion) under combined stresses. RT stant representing the workability of the material.
Constitutive Equations. For a given material, If 1
0, then there are only compressive
the relations between stress components in Eq 1 where n . stresses and no fracture occurs. The Crockcroft-
and strain components in Eq 2 are given by the Other constitutive relations have been pro- Latham criterion is phenomenological based
constitutive equations representing the behavior posed to describe dynamic recovery and dy- rather than mechanistic, but it captures the phys-
of that material. The simplest example of a con- namic recrystallization, such as the Laasoui- ical concept and intuitive understanding that
stitutive equation is the well-known Hookes law Jonas model (Ref 14) and the internal variable damage accumulation, or workability, is depend-
in the elastic regime: model (Ref 15). In all cases, constitutive equa- ent both on the degree of plastic deformation
tions are empirical-based relations derived from (represented by the integral over effective strain)
eE
the reduction of test data. and tensile stress (represented by the maximum
where E is the elastic modulus of the material, Yield Criteria (Flow Rules). The theory of principal stress). Furthermore, both of these
which is measured in a simple tension test or by continuum plasticity involves the definition of physical quantities are readily available outputs
ultrasonic means. yield criterion for when a material yields plasti- of continuum mechanics models and can be eval-
During plastic deformation of most metallic cally or flows. In structural analysis, yield crite- uated at every localized region or point through-
materials, the stress-strain curve becomes non- rion may be characterized as a failure theory, out the material to determine potential sites of
linear, because hardening (or less frequently because plastic deformation is an undesired out- crack initiation. As with any constitutive relation,
softening) of the material can occur when con- come in structural design. In bulk deformation, the value of incorporating this workability crite-
tinuing plastic strain is built up. The strain rate yielding of the workpiece is intended, and yield rion into process analyses depends on accurate
can also influence the hardening or softening of criteria are used in the modeling of flow under representation of the actual material behavior,
a material. The general form of the constitutive combined stresses. represented by the coefficient C in this case.
equation for deformation processing is: The continuum mechanics of metallic materi- A more recent fracture criterion is theorized
als includes several theories for yielding, as as (Ref 20):

f (e,e,T)
described in a historical sketch of continuum
where is the equivalent (or effective) com- plasticity theory with an introduction on com- e
s1f C
bined stress, e is the equivalent true strain, e is putational methods in solid mechanics (Ref 16). s
the equivalent true strain rate, and T is the pro- The first yield criterion for metals was proposed where 1f is the maximum principal stress at
cessing temperature. Most software packages by Henri Tresca in the 1860s. The Tresca crite- fracture, and C is a workability constant. This
for bulk forming modeling have options to input rion is based on the premise that yielding is de- criteria can be related directly to bulk workabil-
the testing data in a tabular form or as a consti- pendent on just shear stresses, whereby plastic ity tests. For incremental loading, Bandstra (Ref
tutive equation. The tabular form is easy to use flow begins when the shear stresses exceed the 21) proved that the criterion can be expressed as:
but is not based on metallurgical principles, as shear yield strength of the metallic material.
with some constitutive equations. Although the Tresca yield criterion is adequate, ef s1
The most frequently used constitutive equa- it neglects the intermediate principal stress, 2. 0 s
d e C
tion is: The Levy-von Mises yield criterion is consid-
ered to be a more complete and generally appli- A review of fracture criteria can be found in Ref

K en em Y 20 and some examples are given in Chapter 12,
cable yielding criteria. It is based on the theory of
where n is the strain hardening exponent, m is Richard von Mises that incorporated a proposal Workability Theory and Application in Bulk
the strain-rate sensitivity, and Y and K are coef- by M. Levy, which stated that the tensor compo- Forming Processes.
ficients. Strain-rate sensitivity is important at el- nents of plastic-strain increments are in propor-
evated temperatures, while it has little influence tion to each other just as are the tensor compo-
Methods of Solution
at room temperature for most metallic materials. nents for deviatoric stress. It is based on the The general methods of solving the underly-
In contrast, the importance of the strain-harden- second tensor invariant of the deviatoric stresses ing equations of a model depend on the com-
ing exponent becomes more significant with de- (that is, of the total stresses minus those of a plexity of boundary conditions and process vari-
creasing temperature. hydrostatic state with pressure equal to the aver- ables such as time, space, and internal states.
The most immediately preceding equation age normal stress over all planes) (Ref 16). The Developing an analytical or a closed-form solu-
does not reflect the influence of temperature. For von Mises yield criterion is thus expressed as: tion model may be advantageous in many in-
each temperature, there is a set of equations. A stances, but numerical techniques may be re-
more fundamentally sound equation has been 2s 20 = (s1 - s 2 )2 + (s 2 - s 3 )2 + (s 3 - s1 )2 quired. In some circumstances analytic solutions
of partial differential equations can be obtained. Contains different phases and grains cally indeterminate (i.e., there is no longer a
However, this is only true of simple forms of the Contains discontinuous behavior such as a closed-formed solution that specifies deflection
equations and in simple geometric regions. For phase change over the length of the beam). In this case, nu-
most practical problems, computational or nu- Has a nonlinear physical process such as merical techniques (such as finite element analy-
merical solutions are needed. when the heat transfer coefficient is a non- sis) are required to find approximate solutions of
There are several techniques for solving dif- linear function of the temperature deflection and bending stresses. Likewise, the
ferential equations. Numerical algorithms to solution of dynamic problems may also require
solve differential equations consist of lumped- In many instances, meshed models are supple- numerical techniques, depending on the com-
parameter methods or the so-called meshed-solu- mented by some non-meshed symbolic or analyt- plexity of the equations in the models and the
tion methods (such as finite element analysis). ical modeling. This is done in order to decide on appropriate boundary conditions.
If the problem can be simplified for the use of appropriate boundary conditions for the meshed In analytical solutions models, boundary con-
ordinary differential equations, then a lumped- part of the problem, because it is the boundary ditions need to be set at a very early stage. In
parameter model might be used. A lumped- conditions that effectively model the physical meshed-solution models, boundary conditions
parameter model may help in understanding the problem and control the form of the final solu- are typically represented separate from the main
effect of certain parameters on the process. tion. Analytic models are always useful for dis- equations and decoupled to some extent from
However, these methods do not model spatial tinguishing between mechanisms that have to be the model itself. Therefore, sensitivity analysis
variation directly, and the parameters do not nec- modeled separately or modeled as a coupled set. can be done much easier using meshed methods.
essarily always have direct physical significance. Boundary Conditions. Application of appro- Material Properties. All models of bulk-
Typically, partial differential equations are re- priate boundary conditions is a major part of the working processes require input of accurate ma-
quired to describe the process in terms of time, activity of process modeling. Boundary and initial terial properties so that the fundamental materi-
space, field variables, and internal states. Several conditions represent geometric factors (e.g., sym- als behavior can be represented faithfully by the
methods have been developed for the numerical metry, tool shape, and workpiece shape), thermal constitutive equations. Acquiring these proper-
solution of partial differential equations. This factors (e.g., heat flux and temperature) and loads ties can be difficult and expensive. Sensitivity
typically involves meshed-solution methods (e.g., tool pressure and tool friction) pertinent to analysis of the model with respect to variations
such as the finite element method, the finite dif- the particular problem being modeled. These con- in property data should be done. In many in-
ference method (FDM), and the boundary ele- ditions or constraints influence the complexity of stances, it may be possible to use models with
ment method (BEM). Each has different suitabil- the problem and the most appropriate method or inexact material property information in order to
ity for different types of problems. For example, algorithm for solving the equations of the model. predict trends, as opposed to determining actual
the finite difference method is often adopted in For example, consider the simple example of values. Problems may arise if the material prop-
fluid mechanics but is seldom used in solid and a uniformly loaded cantilever beam (Fig. 18a). erties are improperly extrapolated beyond their
structural mechanics. Finite element methods In this case (assuming uniform loading and a range of validity. In deformation modeling,
are the most common for linear and nonlinear rigid connection), then closed-formed equations Youngs modulus, Poissons ratio, anisotropic
continuum mechanics, although the boundary can provide analytical solutions for the deflec- behavior, and flow stress (as functions of stress,
element method (BEM) has advantages in some tion and bending stress over the length of the strain, strain rate, and temperature) are needed.
applications of continuum mechanics. beam (Fig. 18a). However, if the boundary con- Closed-Form and Numerical Methods of
These numerical methods provide approxi- ditions are altered by the addition of simple sup- Solution. Several types of theoretical methods
mate solutions by converting a complex contin- ports (Fig. 18b), then the system becomes stati- are available for metal forming analysis, as
uum model into a discrete set of smaller prob-
lems with a finite number of degrees of freedom.
The meshes are created by using structured ele-
ments like rectilinear blocks or unstructured
meshes with variable-shaped elements (e.g.,
tetrahedra, bricks, hexahedral, prisms, and so
forth) for better fidelity to the macroscopic con-
ditions or boundaries. Once a discrete-element
model has been created, mathematical tech-
niques are used to obtain a set of equilibrium
equations for each element and the entire model.
By applying various boundary conditions and
loads to the model, the solution of the simultane-
ous set of equations provides the resulting re-
sponse anywhere in the model while still provid-
ing continuity and equilibrium. The process of
solution is essentially a computer-based numeri-
cal method, where interpolation functions (poly-
nomials) are used to reduce the behavior at an in-
finite field of points to a finite number of points.
Meshed-solution models have advantages
over the typical closed-form solutions, as they
more readily give solutions to irregular shapes,
variable material properties, and irregular
boundary conditions. Construction of a discrete
meshed model for numerical solution may be
necessary if the modeled volume:
Fig. 18 Effect of boundary conditions on the solution of a cantilever problem. (a) The beam deflection and bending
stresses for a uniformly loaded cantilever can be solved by a closed-form equation as shown. (b) A supported
Has a complex shape (as is common in bulk cantilever beam is statically indeterminate, and numerical methods are required to approximate defection and bending-
forming) stress conditions that are consistent with the boundary conditions established by the additional supports.
14 / Introduction
described in Ref 22 and 23. Generally, the meth- large length of contact zone compared to the sections leads to progressively smaller estimates
ods of deformation modeling fall into one of two small strip thickness, or wire drawing, which has of the power dissipated, hence the term upper
categories: a large length of contact in the die compared to bound. The equilibrium equations are not
the wire diameter. Under these conditions, the solved in this case, but as the upper limit is min-
Closed form analytical approximations
normal stresses can be assumed to vary only in imized, satisfaction of equilibrium is ap-
Numerical methods using discrete elements
the direction of the large dimension. This re- proached. With the upper bound method, again,
However, this division is not entirely distinct, as duces the complex set of continuum equations to closed form solutions can be developed for sim-
closed-form methods are often solved by dis- one ordinary differential equation of equilibrium ple geometries, while numerical methods are
crete-element numerical methods, as well. The and one strain equation. The ordinary differen- used for complex geometries.
general distinction between the two sets of ana- tial equation can be solved in closed form, or, for The upper bound method is based on the limit
lytical methods is that the first set is based on complex geometries, it can be solved by numer- theorem stating that the power dissipated by the
simplifying assumptions that permit closed form ical methods. boundary forces at their prescribed velocities is
solution as well as rapid numerical and graphi- In this approach, the deformation is assumed always less than or equal to the power dissipated
cal methods. Solving the partial differential to be homogeneous, and the force equilibrium by the same forces under any other kinemati-
equations by numerical methods, in contrast, re- equations are set up and solved using an appro- cally admissible velocity field. A kinematically
duces the need for simplifying assumptions. priate yield criterion. The slab method is a quick admissible velocity field is one that satisfies the
Regardless of the solution method, continuum way of obtaining approximate load and strain velocity boundary conditions and material in-
mechanics is the starting point. Each problem estimates in axisymmetric and plane-strain prob- compressibility. This method allows kinemati-
starts with the continuum equations and then, lems and is therefore widely used. This method cally admissible velocity fields to be set up as a
through various simplifying assumptions, leads does not account for the contribution of redun- function of an unknown parameter. Power dissi-
to a solution method that may be closed form or dant work, which is the additional work ex- pation is then minimized with respect to the un-
numerical. The next two sections of this chapter pended due to inhomogeneous deformation. In known parameter to yield a reasonable estimate
briefly introduce some typical closed formed this method, stress and strain gradients are ac- of load.
methods and numerical techniques. counted for in only one direction and are The main disadvantage of this method is that
assumed to be uniform in the perpendicular the choice of velocity field is rather arbitrary,
direction. and the poorer the selection, the more the esti-
Approximate and Closed Slip Line Method. The slip line field method mated load will exceed the true load. Another
Form Solutions (Ref 31) utilizes the wave nature of the hyper- limitation is that no local stress field is com-
bolic differential equations that describe plastic pared. The upper bound method does offer a rel-
The equations of continuum mechanics are deformation to produce a graphical method of atively simple way of calculating the major force
the basic starting point of a model, and then var- solution of the stress field in a two-dimensional requirements (Ref. 27).
ious simplifying assumptions lead to a solution plane strain or axisymmetric field. These con- The lower bound method is not of great
method that may be closed form or numerical. In structions can be carried manually for problems practical significance, because forming loads are
general, the boundary conditions in metal form- having simple geometry or by numerical tech- underestimated. However, it does provide an in-
ing operations are too complicated for analytical niques for complex geometries. (Interestingly, dication of how conservative the upper bound
solution of plasticity in the forming of parts with the equations describing hypersonic fluid flow solution is if the lower bound solution is known.
even relatively simple geometric features. are also hyperbolic, and methods similar to slip The lower bound approach is based on the limit
However, the need to obtain at least approximate line fields are used to design rocket nozzles and theorem stating that the power dissipation of the
solutions may be satisfied by simplifying as- expanders. The equivalent of slip lines in plastic actual surface forces at their prescribed veloci-
sumptions, but each of the analytical approaches flow are the shock waves occurring in hyper- ties is always greater than the power dissipation
has its limitations. The most commonly used ap- sonic fluid flow.) of the surface tractions corresponding to any
proximate methods include: The slip line field approach was developed for other statically admissible stress field. A stati-
plane-strain problems. It assumes that the mate- cally admissible stress field must satisfy force
The slab method, which restricts the change
rial is rigid and ideally plastic (that is, the mate- equilibrium and not violate the yield criterion.
of stress to only one direction
rial does not strain harden). The theory is based
The uniform deformation energy method,
on the fact that any state of stress in plane strain
which neglects redundant work involved in
can be represented as the sum of a hydrostatic Numerical Techniques in
internal shearing due to nonuniform defor-
stress and a pure shear stress. Given the force Process Modeling
mation
and velocity boundary conditions, this slip line
The slip-line field solution, which is limited
field is constructed. The main advantage of this
to rigid-plastic materials under plane-strain Numerical solution of the continuum equa-
method over the slab method is that it can pro-
conditions tions is required when the simplifying assump-
vide local stress calculations even when the de-
The bounding methods, which can provide tions of methods described in the preceding are
formation is not homogeneous. The major limi-
fairly good estimates of upper and lower not justified. The general types of numerical
tations of the slip-line field approach are:
limits of the deformation force but cannot techniques used in the solution of partial differ-
provide details of local stress and strain dis- It is usable only for plane-strain problems. ential equations includes:
tributions It assumes rigid and ideally plastic materials.
The method is tedious, and solutions are The finite element method
These methods are discussed only briefly with- The boundary element method
difficult to verify.
out attempting a rigorous description of the The finite difference method
equations that are solved. More details are avail- This technique has been applied to forging, ex-
able in several other sources (Ref 2430). These trusion, and other processes. With the continued improvement in computer
methods are described in detail with examples in The upper bound method breaks up the de- capabilities, these numerical techniques have be-
the cited references and are presented here in information volume into a number of simple trian- come very effective in the engineering analysis
creasing order of complexity. gular or rectangular sections and applies the of static problems, dynamic conditions (where
The Sachs (slab) method is applicable to principle of compatibility of deformation along the calculation of inertial and/or damping forces
problems in which one dimension is greater than with the limit theorem on power dissipation. involves derivatives with respect to time), or
the others, such as strip rolling, which has a Rearrangement or refinement of the deformation quasi-static conditions where rate-dependent
plasticity may require a realistic estimation of ble mainly to linear elliptic partial differential Table 1 Engineering problems addressed
time (but where inertial forces may still be neg- equations. It also requires reformulation of the by the finite element method
lected.). The general theory and practical use of governing partial differential equations into a Possible
the methods are described in Ref 32. Fredholm integral equation, which applies to a Discipline Typical unknown boundary conditions
In applications involving continuum mechan- range of physical problems. For example, ellip-
Structural Displacement Stress or
ics, the finite element method is the most com- tic partial differential equations like the Laplace displacement
mon technique, although the boundary element or Helmholtz equations can be reformulated as Thermal Temperature Heat flux or
method (BEM) has advantages in some applica- Fredholm integral equations and then be solved convective term
tions of continuum mechanics. The finite differ- by the BEM. The advantage is that the mesh or radiative term
Electrical Voltage Current source
ence method has proven to be useful in fluid and need only cover the boundaries of the domain. Magnetic Electromotive Magnetic field
thermal problems but is seldom used in solid and The BEM can be an effective tool in the force source or intensity
structural mechanics. Nonetheless, the finite dif- analysis of various metal forming problems in Fluid flow Pressure, velocity Velocity
ference method is a simple and efficient method rolling and extrusion. For example, Ref 33 Diffusion Mass concentration Flux of species
(Fickian)
for solving ordinary differential equations in demonstrates that the BEM can be used to effi- Diffusion Flow velocity Boundary flow
problem regions with simple boundaries. For ciently and accurately analyze planar and ax- (porous
each node of the mesh, the unknown function isymmetric forming problems involving both media)
values are found, replacing the relevant differen- material and geometric nonlinearities, along Corrosion Anode consumption Electropotential
rate
tial equation, (i.e., dy f [x,y]dx) by a difference with complicated interface conditions. Like Crack Strain energy Stress
equation: FEM, the use of BEM in the modeling of metal propagation release rate
forming operations may require consideration of Acoustic Sound pressure Velocity
elastic-plastic behavior and elastic-viscoplastic noise level
y f(x + x/2,y + y/2) x
problems involving large strains. Elastic strains
where x y are steps in an iterative procedure. are assumed to be small, while nonelastic strains
Hyperbolic and parabolic partial differential (plastic or viscoplastic) are presumed to be tailed temperature, strain, and strain-rate histo-
equations are often solved using a hybrid of the large. When strains become large, the original ries at each individual material point in a work-
FEM and FDM. The spatial variables are mod- mesh may become so distorted that the interpo- piece are also available from FEM simulations.
elled using the FEM, and their variation with lation polynomials are incapable of modeling Simulation of thermomechanical processes can
time is modelled by the FDM. the geometry of the elements and their relevant be based on models of continuum mechanics,
Other numerical techniques include: state variables. This requires a process of transport phenomena (heat flow), and metallur-
remeshing. gical phenomena (e.g., grain growth and recrys-
Finite-volume methods, which are impor-
tant in highly nonlinear problems of fluid tallization).
mechanics Finite-Element Analysis The range of FEA applications in the area of
Spectral methods, which are based on trans- material processing is extremely wide, and a
Finite element analysis (FEA) is a computer- brief review of the finite element techniques ap-
forms that map space and/or time dimen-
based analytical technique that allows numerical plied to metal forming, nonmetal forming, and
sions to spaces where the problem is easier
solutions to be obtained for complex mathemat- powder metallurgy are briefly discussed in Ref
to solve
ical and engineering problems by creating a dis- 35 with an encyclopedic view of the different
Mesh-free methods, which are a recent de-
crete or finite number of individual nodes and el- possibilities in these various fields of applica-
velopment of finite difference methods with
ements. Discrete elements fill the appropriate tion. Many texts (e.g., Ref 2, 23, 34) also de-
arbitrary grids
geometry, and the method enables the system- scribe finite element theory and how it is used in
These techniques and the finite difference atic solution of equilibrium equations for each forming analyses. Models may be based on con-
method are only mentioned for reference with- element and the entire model with as much fi- tinuum mechanics in the evaluation of flow
out further discussion. The finite element and delity to geometry as needed. and/or fracture problems supplemented by more
boundary element methods are more common in Finite element analysis is a powerful analysis sophisticated thermomechanical/thermophysical
the modeling of bulk deformation processes, as tool that is flexible for solving problems with ir- models for the simulation of microstructural
described later in this chapter in more detail. regular shapes, variable material properties, and evolution.
Application of these computer models has be- irregular boundary conditions. With advance- Finite element models of flow during defor-
come an essential tool in meeting product re- ments in computer technology, the use of nu- mation processes are based on the equations of
quirements of dimensional tolerances, surface merical methods such as FEA has grown. The mechanical equilibrium and flow behavior, as
finish, and consistency of properties. method was originally developed for structural previously described in the section Continuum
problems (stress displacement of complex Mechanics Equations in this chapter. As noted,
Boundary Element Method geometries), but the same concepts and princi- constitutive equations are mathematical expres-
ples apply to other kinds of engineering prob- sions that describe stress-strain curve in the re-
The boundary element method (BEM) is a lems such as those listed in Table 1. In deforma- gion of nonelastic (plastic or viscoplastic) flow.
technique for representing a complex structure tion processes, FEA is a useful tool in die design Constitutive equations for plastic deformation
or component as a computer model in order to and process analysis. Common problems solved are typically nonlinear, as plastic deformation of
determine its response to a set of given condi- by FEA include insufficient die filling, poor materials is generally a function of the strain
tions. Like the finite element method (FEM), the shape control, poor flow of material, cracks and hardening/softening and the strain-rate harden-
model is formed by subdividing the structure voids that lead to fracture, and inadequate prop- ing/softening response of a material for different
into small elements to form the overall model. erties from microstructural variations (grain conditions of temperature, stress, and mi-
However, unlike the FEM, only the surface (or size). crostructure. Constitutive equations are required
boundary) of the problem requires subdivision, The application of finite element modeling for realistic modeling, and they are unique for
thereby reducing the dimensionality of the prob- (FEM) in metal forming (e.g., Ref 2, 23, 34) has each material under each processing condition.
lem and thus dramatically reducing the compu- brought great changes to design methodologies Constitutive equations may be developed from
tational effort in obtaining a solution. that were formerly based on trial and error data obtained under simplified experimental
The BEM has a more restricted range of ap- approaches. For example, shape changes during conditions or from thermomechanical testing.
plication than FEM has. In general it is applica- forging are easily predicted by FEM. The de- Constitutive equations are then extended to more
16 / Introduction
complex situations of combined stresses by formation. Another example is a rigid-viscoplas- had apparently developed during a slow cooling
well-known hypotheses of flow rules (e.g., von tic model used to predict surface defects from de- cycle (such as during casting of the ingot stock).
Mises yield criterion) from the theory of contin- formation processes (Ref 44) where defect initia- Cracks were observed in both longitudinal and
uum plasticity. tion and development are predicted and some transverse directions and arrested just short of
Finite element methods, as they apply to critical processing parameters are identified. the outer bar diameter. Finite element analysis
metal forming analysis, can be classified into ei- Other application examples of rigid-viscoplastic was used to determine the cause and remedy for
ther elastic-plastic or rigid-viscoplastic methods, models are described in Ref 4549. this cracking. In addition, new processing pa-
depending on the assumptions made with regard Regardless of the flow model, the modeling of rameters were determined to eliminate cracking.
to the material flow behavior. The elastic-plastic deformation processes by FEM (and BEM) typ- A 2D model of the bar was developed and
method assumes that the material deformation ically requires techniques for the remeshing of discretized into a finite element mesh. A public-
includes a small, recoverable elastic part and elements. When the original mesh becomes domain, nonlinear FEA code from Lawrence
a much larger, nonrecoverable plastic part that highly distorted, the interpolation functions are Livermore National Laboratory (Ref 56),
is time independent. In contrast, the rigid- incapable of modeling the geometry and state NIKE2D, was used to calculate the stress plots
viscoplastic method assumes time-dependent variables. Large strains in finite element simula- under forging conditions. The stress/strain histo-
deformation behavior (i.e., creep or viscous tions of metal forming require several remeshing ries resulting from forging were required for
behavior). procedures during the complete simulation ductile fracture analyses. In this particular case,
Elastic-plastic FEA can give details regarding cycle. These remeshings may be performed au- the Oyane et al. ductile fracture criteria (Ref 18)
deformation loads, stresses and strains, and tomatically both for saving computational time were used (Ref 57). These criteria attempt to
residual stresses. Reference 23 describes in de- and for convenience. The methods of remeshing predict ductile fracture based on a porous plas-
tail elastic-plastic FEA and how it is used in involve the generation of new elements by using ticity model.
forming analyses. The elastic-plastic FEM has the interpolation functions, as briefly described Figure 19 (Ref 55) is a plot of in-process hy-
been applied to a large variety of problems, in- in Ref 23. Automatic remeshing algorithms have drostatic stress history at the center of a bar dur-
cluding upsetting (Ref 36 and 37), indentation been well established for two-dimensional (2D) ing forging with two rigid, parallel flat dies.
(Ref 38), rolling (Ref 11, 39), extrusion (Ref applications but have also been established for (Hydrostatic stress is the mean of any three mu-
40), and expansion of a hole in a plate (Ref 41). some three-dimensional (3D) cases (e.g., Ref tually perpendicular normal stresses at a given
However, because of the large change in the ma- 50). Other examples of automatic remeshing point. It is invariant with direction.) Various
terial flow behavior between elastic and plastic techniques are given in Ref 5153. strain rate and temperature conditions are shown
deformation and the need to check the status of Remeshing algorithms, whether in 2D or 3D, having similar trends. At very low reductions
each element, the deformation steps must be offer the possibility of using a finer grid in cer- (1.25%), the center of the bar is still within the
small, and this makes the method uneconomical. tain areas of the finite element mesh. These elastic range and, therefore, does not experience
The rigid-viscoplastic FEA method assumes remeshing/refinement algorithms can play an large stresses. At 2.5% reduction, the outward
that the deformation stresses are primarily de- important role in the simulation of microstruc- movement of the free sides of the bar produces a
pendent on deformation (strain) rates. An early ture. In general, it is preferable to define loads high tensile hydrostatic stress at the center of the
example of a rigid-viscoplastic FEM is the and boundary conditions relative to the geome- bar. As reduction continues, the compressive
ALPID (Analysis of Large Plastic Incremental try rather than to the finite element mesh. This stresses acting in the vertical direction begin to
Deformation computer program) (Ref 42). This permits remeshing of the geometry without the grow. This acts to reduce the hydrostatic stress at
program uses the approach of Kobayashi et al. need to redefine loads and boundary conditions. the center. When the bar has been reduced by
(Ref 43) and includes the incorporation of con- It is also desirable to specify hard points, lines, 11%, the hydrostatic stress is compressive in all
venient features including capabilities for han- or surfaces where parts will interact with other cases. Beyond this reduction, the hydrostatic
dling arbitrary die geometries and remeshing. parts, so that the interaction can be properly stress at the center of the bar continues to grow
Although predictions regarding residual represented. in a compressive manner.
stresses cannot be made with the rigid-viscoplas- Example 1: Use of FEA to Study Ductile Figure 20(b) (Ref 55) shows a map of ductile
tic FEM, the larger steps that can be used in mod- Fracture during Forging (Ref 54). Large fracture accumulation after one forging pass for
eling metal forming procedures make the method nickel-copper alloy K-500 (K-Monel) shafts air-melted material at 930 C (1706 F) and a
very economical, especially for modeling hot de- sometimes revealed fractures after forging and strain rate of 10 s1. Notice that the maximum
heat treating (Ref 55). Inspection of failed frac- damage is located at the center of the bar. After
ture surfaces revealed that the cracks were inter- this pass, the bar would be rotated and further re-
granular and occurred along a carbon film that ductions would be taken. Damage would accu-
Fringe levels
0.130
0.077
0.024
0.029
0.082
0.135
Value at fracture = 0.623
(a) (b)
Fig. 19 Hydrostatic stress at the bar center during forg- Fig. 20 Ductile fracture map of a nickel-copper alloy K-500 bar at 10% reduction, 930 C, and a strain rate of 10.0
ing. Source: Ref 55 s1 using the Oyane et al. ductile fracture criteria. (a) Three forging dies. (b) Two forging dies. Source: Ref 55
mulate during these subsequent passes. Ductile sequences of different strain paths, thermal his-
fracture was predicted to occur at the bar center tories, and strain rates remains a major challenge
from the FEA results after six reducing passes for physical metallurgy if the full predictive
under the specified conditions. One possible way capabilities of finite element models are to be
of eliminating this ductile fracture is to use a V- realized.
shaped die assembly as shown in Fig. 20(a). In
this case, the fracture does not occur at the center
but at the edges and gets an opportunity to heal Processing Maps
as the bar is rotated during subsequent passes.
Workability is also evaluated by the use of pro-
cessing maps that show the occurrence of dam-
Microstructural Modeling age as a function of processing variables such as
temperature and strain-rate sensitivity. This ap-
Microstructural modeling is an active area of proach, however, involves extensive testing, and
practical importance in the design and analysis it is difficult to account for all the variable factors
of thermomechanical processing. One early ex- in mechanistic models. For example, the location
ample is the pioneering work of Sellars and of the fracture boundaries in processing maps is
coworkers, who developed methods for the mi- very sensitive to microstructure, prior thermome- Fig. 21 Composite processing map for aluminum
crostructural modeling of carbon-manganese chanical history, and numerous material parame- showing the safe region for forming. Bound-
ters such as diffusivity. Nonetheless, processing aries shift with microstructure. Instabilities due to purely
steels in hot rolling (Ref 5861). Similar meth- continuum effects, such as shear localization in sheet
ods of microstructural modeling also have been maps, such as that shown in Fig. 21 (Ref 65), are metal forming, are not considered. Source: Ref 65
developed for hot rolling operations by others a very useful guide for the selection of deforma-
(Ref 62). Though the details of the models differ tion processing conditions.
from group to group, these models emphasize the need for a macroscopic description of flow,
the importance of quantifying the thermal and Dynamic Material Modeling fracture, and workability of complex engineer-
metallurgical phenomena. Each basically in- ing materials under hot-working conditions. As
volves the discrete characterization of dynamic Processing maps are also used in conjunction noted, this approach makes use of the constitu-
structural changes (e.g., dynamic recrystalliza- with a method known as dynamic material mod- tive equations for plastic flow determined from
tion, static recrystallization, and grain growth) eling (DMM), which is a top-down approach hot deformation tests performed at different
by using empirical equations based on funda- that begins with macroscopic determination of temperatures and strain rates. Parameters such as
mental physical metallurgy principles. Such flow stress as a function of temperature, strain the strain-rate sensitivity and temperature sensi-
physical metallurgy-based models grasp the rate, and strain and ends with a microscopic tivity of the flow stress are related to the manner
essence of the microstructural changes. evaluation of the microstructure and the final in which the workpiece dissipates energy instan-
Models for process design continue to advance properties of the forging (Ref 66). This method taneously during hot deformation. The rate of
from data-intensive empirical methods to more maps the power efficiency of the deformation of change of stress with strain rate at constant lev-
knowledge-based analytical/numerical methods the material in a strain rate/temperature space. els of strain (e) and temperature (T) is known as
using the fundamental principles of transport The DMM methodology describes the dynamic the strain-rate sensitivity parameter (m), which
phenomena, continuum mechanics, and physical path in response to an instantaneous change in is defined as follows:
metallurgy. By this means, expensive trials can strain rate at a given temperature and strain. As
be minimized, and robust processes to make such, it describes how workpiece material dissi- m [(log s)/(log e)]e,T (Eq 3)
high-integrity parts within tight tolerances can be pates applied power by various metallurgical
readily designed. The ability to accurately pre- processes, depending on the instantaneous The temperature sensitivity of the flow stress is
dict thermomechanical histories is thus being change in strain rate at a given temperature and analyzed in terms of a parameter (s):
used to predict the evolution of microstructure by strain. By measuring flow stress as a function of
coupling microstructure models to FEM results. strain, strain rate, and temperature and by calcu- 1
For example, Ref 63 describes carefully designed lating strain rate sensitivity at each value of tem- s=
T
[ ln s / (1/ T )]e ,e (Eq 4)
workability tests and FEM simulations to de- perature (T) and strain rate (e), it is then possible
velop a general model for microstructural evolu- to identify boundaries between safe and unsafe
tion during thermomechanical processing of a regions on a processing map (Fig. 22) (Ref 67). Furthermore, an apparent activation energy (Q)
nickel-base superalloy. This process-modeling The safe regions on the DMM processing map can be determined from computations of m and
tool was then applied to predict the percentage of correspond to conditions of DMM stability (see s using:
recrystallization and grain size in actual the following paragraphs), which is determined
Waspaloy forgings. by Liapunov stability analysis. sRT
Q= (Eq 5)
Quantification of the evolution of microstruc- The DMM method allows modeling for when m
ture during industrial processing (as described stable deformation and microstructure evolution
in more detail in Chapter 3, Evolution of can be expected. A collection of DMM process- where R is the universal gas constant.
Microstructure during Hot Working) is increasing maps is contained in Ref 68, and the appli- Metallurgical Interpretation of DMM
ingly recognized as an essential tool for metal cation of DMM concepts is discussed in some Stability (Ref 65). The regions of DMM stability
producers and fabricators. One major challenge chapters of the section Multidisciplinary correspond to areas on a DMM map where the
is the need to understand the microstructural Process Design and Optimization in this energy dispersion processes of the material are in
consequences of processing variables in differ- Handbook. In its general form, the dynamic ma- a steady state. The DMM stability criteria are:
ent elements of a mesh model. This is discussed terial model is sometimes considered to be a
in Ref 64 for different types of models applied to generalized concept of workability that plays a 0
m1 (Eq 6)
hot rolling, where the success and limitation of role in unifying the relationships among consti- m/(log e)
0 (Eq 7)
predicting variations in local microstructures are tutive behavior, hot workability, and microstruc-
described for microstructural models based on ture development (Fig. 23) (Ref 69). Dynamic s1 (Eq 8)
FEMs. Understanding the microstructural con- material modeling was developed in response to s/(log e)
0 (Eq 9)
18 / Introduction
Fig. 22 Processing map for Ti-6242 microstructure with stable regions identified. Source: Ref 67
ing or newly formed microcracks in the work-

piece, which eventually lead to fracture of the
workpiece. As the value of m increases, the ten-
dency for localized deformation decreases, en-
abling extensive elongation of specimens under
tensile loading without necking. Similarly, the
occurrence of shear-band formation can be in-
hibited very effectively at high m values. For
metals and alloys, m 1 represents ideal super-
plastic behavior resulting from the Newtonian
flow typical of a glass material.
The stability criterion relating to the variation
of m with log e (Eq 7) stems from the theoretical
requirement for the material system to continu-
ously lower its total energy. If fracture stress is
Fig. 23 Models of material behavior in hot deformation studies. (a) Discrete approach where flow and microstruc- assumed to be independent of strain rate, then an
ture development are separate in the evaluation of workability. (b) The DMM approach that is linked with increasing m(e) will probably lead to catastrophic
mechanical flow and the activation energy of microstructural development. Source: Ref 69
failure at high strain rates. In contrast, a decreas-
ing m(e) has a lower probability of inducing frac-
The range of strain-rate-sensitivity (m) values conditions promoting dynamic strain aging (a ture in the workpiece. Moreover, this stability
for stable material flow (Eq 6) is derived from product of the interaction of mobile dislocations criterion leads to more uniform stress fields
theoretical considerations of the maximum rate and solute atoms). However, the imposed strain across the workpiece and a decreased tendency
of power dissipation by material systems and rates for hot working operations are, in general, for strain localization, which is clearly desirable.
experimental observations. Metals and alloys sufficiently high that such interactions are The lower limit of s values (Eq 8) is derived
generally satisfy this criterion under hot working avoided. from the premise that the net entropy-production
conditions. Negative values of the strain-rate- Another mechanism that can lead to negative rate associated with irreversible processes must
sensitivity parameter are obtained only under m values is the dynamic propagation of preexist- be positive for stable conditions. It is well estab-
lished that when dynamic recovery alone oper- contact surface fracture, and internal fracture REFERENCES
ates as a softening mechanism, the temperature such as center bursts. The mechanism by which
dependence of flow stress is relatively weak. In the fracture originates depends on the tempera- 1. J. Schey, Manufacturing Processes and
contrast, when both dynamic recrystallization ture and strain rate of the deformation. Their Selection, Materials Selection and
and dynamic recovery are in operation, the flow Other workability related defects are not due Design, Vol 20, ASM Handbook, ASM
stress varies markedly with temperature. Thus, to fracture, but are related to the distribution of International, 1997, p 687704
low values of s are indicative of the dynamic re- the metal in the deformation process. These are 2. R.H. Wagoner and J.-L. Chenot, Metal
covery processes, while high values of s are usu- described by such terms as die underfill, laps Forming Analysis, Cambridge University
ally associated with dynamic recrystallization and folds, extrusion-type defects, flow-through Press, 2001
processes. defects, and seams. 3. L. Ferguson, Design for Deformation
The stability criterion relating to the variation Deformation processing not only changes the Process, Materials Selection and Design,
of s with log e (Eq 9) arises from the necessity shape of the material, but it also changes its me- Vol 20, ASM Handbook, ASM International,
for the material system to continuously lower its chanical properties. Ideally, these changes result 1997, p 736737
total energy. As the strain rate increases, the ex- in property improvement. Improvement results 4. P.W. Lee and H.A. Kuhn, Cold Upset
tent of adiabatic heating in local regions in- from the closure of pores and voids and the re- Testing, Workability Testing Techniques,
creases significantly. Hence, in any particular finement of grain size. Metal that is rolled, G.E. Dieter, Ed., American Society for
region of the workpiece, if the local strain rate forged, or extruded, however, develops and re- Metals, 1984, p 3750
increases above the nominal value, then it is ac- tains a fiberlike structure that is aligned in the 5. S.L. Semiatin, Workability in Forging,
companied by a reduction in the local flow principal direction of working. There also may Workability Testing Techniques, G.E. Dieter,
stress. However, if s(e) increases with e, then a be crystallographic reorientation of grains. Ed., American Society for Metals, 1984, p
very significant thermal softening will be en- These structural changes result in properties that 197247
countered in the regime of high strain rates, vary with orientation to the fiber structure. 6. H.A. Kuhn and G.E. Dieter, Workability in
which will produce severe strain localization Proper design of process tooling and optimum Bulk Forming Processes, Fracture 1977,
and adiabatic shear bands. This autocatalytic selection of material and material process condi- Fourth International Conf. on Fracture, Vol
process is likely to lead to severe cracking of the tions can do much to minimize workability de- 1, (Waterloo, Canada), 1977, p 307324
workpiece if the material has poor resistance to fects and optimize the properties resulting from 7. H.A. Kuhn and B.L. Ferguson, Powder
nucleation and growth of cracks. On the other the deformation. The merging of metallurgy with Forging, Metal Powder Industries Federa-
hand, compliance with this stability criterion has mechanics and the availability of powerful com- tion, 1990
a mitigating influence on this tendency for flow puter-based models has greatly enhanced our 8. V.N. Whittacker, Nondestructive Testing,
localization. ability to achieve favorable workability, even for Oct 1971, p 320
Thus, the four stability criteria of the DMM hard-to-work materials. This chapter has laid the 9. L.F. Coffin Jr. and H.C. Rogers, Influence
approach have a sound metallurgical basis. foundation for understanding the concepts and of Pressure on the Structural Damage in
Since the material behavior is both dynamic and mathematics of deformation modeling that will Metal Forming Processes, ASM Trans.
stochastic, these criteria should be regarded as be useful as subsequent chapters discuss specific Quart., Vol 60 (No. 4), Dec 1967, p
probabilistic indicators of the behavioral trends processes and examples. The development fol- 672686
exhibited by a material during hot working. The lows a historical order, starting with the slab 10. A. Chamouard, Closed Die Forging: Part 1,
effects of these instabilities on the workability of method, advancing to the slip-line field theory, Dunod, Paris, 1964
a material can be alleviated substantially by the upper and lower bound methods, the bound- 11. Z.-C. Lin and C.-C. Shen, Rolling Process
proper control of the die design. For instance, ary element method, and finally to the widely Two-Dimensional Finite Element Model
die designs that result in a high ratio of mean-to- used finite element method. It also reviews cur- Analysis, Finite Elem. Anal. Des., Vol 26
effective stress can, in part, offset the adverse rent work on the use of computer modeling to (No. 2), June 1997, p 143160
effect of positive values of m/(log e). This ob- predict the microstructures produced from ther- 12. F.N. Lake and D.J. Moracz, Comparison of
servation is exemplified by the remarkable suc- momechanical processing and to predict the re- Major Forging Systems, Technical Report
cesses associated with processes such as hydro- gions of temperature and strain rate that are most AFML-TR-71-112, TRW, Inc., May 1971
static extrusion, pack rolling, and closed-die favorable for achieving good workability. 13. C.M. Sellars and W.J. McG. Tegart, Mem.
forging. Sci. Rev. Metall., Vol 63, 1966, p 731
14. A. Laasroui and J.J. Jonas, Prediction of
ACKNOWLEDGMENTS Steel Flow Stresses at High Temperatures
Summary and Strain Rates, Metall. Trans. A, Vol 22,
Portions of this chapter were adapted from the 1991, p 15451558
Workability refers to the relative ease with following ASM Handbook articles: 15. S.B. Brown, K.H. Kim, and L.A. Anand, An
which a material can be shaped through plastic J. Schey, Manufacturing Processes and Internal Variable Constitutive Model for
deformation. It is a complex technological con- Their Selection, Materials Selection and Hot Working of Metals, Int. J. Plast., Vol 5,
cept that depends not only on the fracture resist- Design, Vol 20, ASM Handbook, ASM 1989, p 95130
ance (ductility) of the material but also on the International, 1997, p 687704 16. J.R. Rice, Foundations of Solid Mechanics,
specific details of the stress state generated by L. Ferguson, Design for Deformation Mechanics and Materials: Fundamentals
the deformation process. The chief process vari- Process, Materials Selection and Design, and Linkages, M.A. Meyers, R.W.
ables that determine the stress state are the Vol 20, ASM Handbook, ASM International, Armstrong, and H. Kichner, Ed., John
workpiece geometry, the die geometry, and lu- 1997, p 736737 Wiley & Sons, 1999
brication conditions. The material fracture re- H.L. Gegel et al., Modeling Techniques 17. M.G. Cockcroft and D.J. Latham, Ductility
sistance is determined by its composition, Used in Forging Process Design, Forming and Workability of Metals, J. Inst. Met., Vol
microstructure, and freedom from inclusions, and Forging, Vol 14, Metals Handbook, 96, 1968, p 3339
as well as important process parameters such as 1988, p 425 18. M. Oyane, Criteria of Ductile Fracture
the working temperature and the speed of de- Strain, Bull Jpn. Soc. Mech. Eng., Vol 15,
formation. The editors also thank Jim Malas for providing 1972, p 15071513
Fractures found in deformation processing the text from Ref 65 on Metallurgical 19. A.M. Freudenthal, The Inelastic Behaviour
can be classified as free surface fracture, die Interpretation of DMM Stability. of Solids, John Wiley & Sons, 1950
20 / Introduction
20. D. Zhao, J.P. Bandstra, and H.A. Kuhn, A Closed-Die Upsetting, Int. J. Mach. Tools 50. L.H. Tack, R. Schneiders, J. Debye, R.
New Fracture Criterion for Fracture Manuf., Vol 36 (No. 9), Sept 1996, p Kopp, and W. Oberschelp, Two- and Three-
Prediction in Metalworking Processes, 10211032 Dimensional Remeshing, Mesh Refinement
Concurrent Engineering Approach to 38. C.H. Lee and S. Kobayashi, Elastoplastic and Application to Simulation of
Materials Processing, S.N. Dwivedi, A.J. Analysis of Plane-Strain and Axisymmetric Micromechanical Processes, Conference:
Paul, and F.R Dax, Ed., TMS, 1992, Flat Punch Indentation by the Finite Computational Modeling of the Mechanical
p 107119 Element Method, Int. J. Mech. Sci., Vol 12, Behavior of Materials, 1516 Nov 1993,
21. J.P. Bandstra, 3D Extension of Kuhn Sur- 1970, p 349 published in Comput. Mater. Sci., Vol 3 (No.
face Fracture Criterion, CTC Memoran- 39. Z.-C. Lin and C.-C. Shen, Three-Dimen- 2), Nov 1994, p 241246
dum, CTC/JPB-M0469-95, Concurrent sional Asymmetrical Rolling of an Alu- 51. J. Zhu, M. Gotoh, J. Shang, and Z.G. Sun,
Technologies Corporation, 1995 minium Flat Strip Using an Elastic-Plastic An Integrated FEM System Facilitated with
22. H.L. Gegel et al., Modeling Techniques Finite Element Method, Int. J. Comput. Automatic Remeshing, Metals and Mater.
Used in Forging Process Design, Forming Appl. Technol., Vol 9 (No. 56), 1996, Vol 4 (No. 4), 1998, p 657661
and Forging, Vol 14, Metals Handbook, p 281297 52. R. Krishna Kumar and Ch. Pavana,
1988, p 425 40. A. Mihelic and B. Stok, Tool Design Remeshing Issues in the Finite Element
23. G.W. Rowe, C.E.N. Sturgess, P. Hartley, Optimization in Extrusion Processes, Analysis of Metal Forming Problems, J.
and I. Pillingre, FiniteElement Plasticity Comput. Struct., Vol 68 (No. 13), Mater. Process. Technol., Vol 75 (No. 13),
and Metalforming Analysis, Cambridge JulyAug 1998, p 283293 March 1998, p 6374
University Press, 1991 41. J.S. Gunasekera and J.M. Alexander, 53. Y.Y. Zhu, T. Zacharia, and S. Cescotto,
24. T. Altan, S. Oh, and H. Gegel, Metal Form- Analysis of the Large Deformation of an Application of Fully Automatic Remeshing
ing: Fundamentals and Applications, Elastic-Plastic Axially-Symmetric Contin- to Complex Metal-Forming Analyses,
American Society for Metals, 1983 uum, International Symposium on Founda- Comput. Struct., Vol 62 (No. 3), Feb 1997, p
25. O. Hoffman and G. Sachs, Introduction to tions of Plasticity, Noordhoff International, 417427
the Theory of Plasticity for Engineers, 1972, p 25146 54. A.J. Paul, Modeling of Manufacturing
McGraw-Hill, 1953 42. S.I. Oh, Finite Element Analysis of Processes Materials Selection and Design,
26. S. Kobayashi and E.G. Thomson, Approxi- Metal Forming Problems with Arbi- Vol 20, ASM Handbook, 1997, p 708709
mate Solutions to a Problem of Press trarily Shaped Dies, Int. J. Mech. Sci., 55. M.L. Tims, J.D. Ryan, W.L. Otto, and M.E.
Forging, Trans. ASME, Series B, Vol 81, Vol 17, 1982, p 293 Natishan, Crack Susceptibility of Nickel-
1959, p 217227 43. S. Kobayashi, Rigid Plastic Finite-Element Copper Alloy K-500 Bars During Forging
27. B. Avitzur, Metal Forming: Processes and Analysis of Axisymmetrical Metal Forming and Quenching, Proc. First International
Analysis, McGraw-Hill, 1968 Processes, Numerical Modeling of Manu- Conf. on Quenching and Control of Dis-
28. W. Johnson and H. Kudo, The Mechanics of facturing Processes ASME, PVP-PB-025, tortion, ASM International, 1992, p 243
the Metal Extrusion, Manchester University 1977, p 4968 250
Press, 1962 44. Y. Peng, X. Ruan, and T. Zuo, Defect 56. J.O. Hallquist, NIKE2DA Vectorized
29. K. Lange, Handbook of Metal Forming, Prediction during Conform Process by Implicit, Finite Deformation Finite Element
McGraw-Hill, 1985 FEM, J. Mater. Process. Technol., Vol 45 Code for Analyzing the Static and Dynamic
30. R. Hill and S.J. Tupper, A New Theory (No. 14), Sept 1994, p 539543 Response of 2-D Solids with Interactive
of the Plastic Deformation in Wire Drawing, 45. M.S. Joun, J.H. Chung, and R. Shivpuri, An Rezoning and Graphics, Users Manual,
J. Iron Steel Inst., Vol 159, 1948, p 353 Axisymmetric Forging Approach to UCID-19677, Rev. 1, Lawrence Livermore
359 Preform Design in Ring Rolling Using a National Laboratory, 1986
31. W. Johnson, R. Sowerby, and R.D. Venter, Rigid-Viscoplastic Finite Element Method, 57. S.E. Clift, P. Hartley, C.E.N. Sturgess, and
Plane Strain Slip Line Fields for Metal Int. J. Mach. Tools Manuf., Vol 38 (No. G.W. Rowe, Fracture Prediction in Plastic
Deformation Processes: A Source Book and 1011), OctNov 1998, p 11831191 Deformation Processes, Int. J. Mech. Sci.,
Bibliography, Pergamon Press, 1982 46. S. G. Xu, K. J. Weinmann, D. Y. Yang, and Vol 32 (No. 1), p 117
32. J. Raamachandran, Boundary and Finite J. C. Lian, Simulation of the Hot Ring 58. C.M. Sellars, Hot Working and Forming
Elements: Theory and Problems, CRC Rolling Process by Using a Thermo- Processes, C.M. Sellars and G.J. Davies,
Press, FL, 2000 Coupled Three-Dimensional Rigid- Ed., TMS, London, 1979, p 315
33. A Chandra, Analyses of Metal Forming Viscoplastic Finite Element Method, J. 59. C.M. Sellars, Mater. Sci. Technol., 1990, Vol
Problems by the Boundary Element Manuf. Sci. Eng., Vol 119 (No. 4A), Nov 6, p 10721081
Method, Int. J. Solids Struct, Vol 31 (No. 1997, p 542549 60. C.M. Sellars, International Conf. on
1213), JuneJuly 1994, p 16951736 47. D. Y. Yang, N. K. Lee, and J. H. Yoon, A Physical Metallurgy of Thermomechanical
34. S. Kobayashi, S.I. Oh, and T. Altan, Metal Three-Dimensional Simulation of Processing of Steels and Other Metals,
Forming and the Finite Element Method, Isothermal Turbine Blade Forging by the THERMEC 88, 610 June 1988, (Tokyo), I.
Oxford University Press, New York, 1989, Rigid-Viscoplastic Finite-Element Method, Tamura, Ed., Iron and Steel Institute of
p 222243 J. Mater. Eng. Perform., Vol 2 (No. 1), Feb Japan, Tokyo, 1988, p 448457
35. N. Brannberg and J. Mackerle, Finite 1993, p 119124 61. J.H. Beynon, P.R. Brown. S.I. Mizban,
Element Methods and Material Processing 48. S. M. Hwang and M. S. Joun, Analysis of A.R.S. Ponter, and C.M. Sellars, Proc. of
Technology, Eng. Comput. (WALES), Vol 11 Hot-Strip Rolling by a Penalty Rigid- NUMIFORM Conf., Aug 2529 1986,
(No. 5), Oct 1994, p 413455 Viscoplastic Finite Element Method, Int. J. (Gothenburg, Sweden), K. Mattiasson, A
36. C.H. Lee and S. Kobayashi, Analysis of Mech. Sci., Vol 34 (No. 12), Dec 1992, Samuelsson, R.D. Wood, and O.C.
Axisymmetrical Upsetting and Plane-Strain p 971984 Zienkiewicz, Ed., A.A. Balkerna,
Side-Pressing of Solid Cylinders by Finite 49. M. Fu and Z. Luo, The Prediction of Macro- Rotterdam, Holland, p 213218
Element Method, J. Eng. Ind. (Trans. Defects During the Isothermal Forging 62. C. Devadas, I.V. Samarasekera, and E.B.
ASME), Vol 93, 1971, p 445 Process by the Rigid-Viscoplastic Finite- Hawbolt, The Thermal and Metallurgical
37. I. Pillinger, M. Bakhshi-Jooybari, P. Element Method, J. Mater. Process. State of Steel Strip during Hot Rolling, Part
Hartley, and T. A. Dean, Finite Element Technol., Vol 32 (No. 3), Aug 1992, III: Miscrostructural Evolution, Metall.
Simulation and Experimental Study of Hot p 599608 Trans. A, 1991, Vol 22, p 335349
63. G. Shen, S.L. Semiatin, and R. Shivpuri, 65. R. Raj, Development of a Processing Map Continuum Modeling to Describe Micro-
Modeling Microstructural Development for Use in Warm-Forming and Hot-Forming structure, Computer Simulation in Materials
during the Forging of Waspaloy, Metall. and Processes, Metall. Trans. A, Vol 12, 1981, p Processing, Proc. of Materials Science Sem-
Materials Trans., Vol 26A, July 1995, p 10891097 inar, ASM International, 1987
17951803 66. Y.V.R.K. Prasad, H.L. Gegel, S.M. 68. Y.V.R. K. Prasad and S. Sasidhara, Hot
64. C. M. Sellars, The Role of Computer Doraivelu, J.C. Malas, J.T. Morgan, K.A. Working Guide: A Compendium of Pro-
Modelling in Deformation Processing., Lark, and D.R. Barker, Modeling of cessing Maps, ASM International, 1997
Conference: Fifty Years of Metallurgy: Dynamic Material Behavior in Hot 69. J.C. Malas and V. Seetharaman, Using
Retrospect and Prospect, New Delhi, India, Deformation: Forging of Ti-6242, Metall. Material Behavior Models to Develop
Nov 1996, Trans. Indian Inst. Met., Vol 50 Trans. A, Vol 15, 1984, p 18831892 Process Control Strategies, JOM, Vol 44
(No. 6), Dec 1997, p 563571 67. H.L. Gegel, Synthesis of Atomistics and (No. 6), June 1992, p 813
Chapter 2
Bulk Workability of Metals
George E. Dieter, University of Maryland
WORKABILITY refers to the relative ease tests and on workability tests specific to a par- slip system, and the shear stress is that which ac-
with which a metal can be shaped through plas- ticular bulk forming process, such as forging, are tivates plastic deformation, then Eq 2(b) is
tic deformation. In this chapter, consideration is given in subsequent chapters of this Handbook. known as Schmids law.
restricted to the shaping of materials by such The state of stress at a point is defined by
bulk deformation processes as forging, extru- three normal stresses and six shear stresses. To
sion, and rolling. The evaluation of workability
Stress, Strain, and Stress-Strain simplify, it is common practice to work with
of a material involves both the measurement of Curves three principal stresses. The principal stresses lie
the resistance to deformation (strength) and de- parallel to three mutually perpendicular princi-
termination of the extent of possible plastic de- Stress and Strain. This section provides the pal axes and are normal to the three principal
formation before fracture (ductility). Therefore, definitions of stress and strain for readers with- planes. The defining feature is that no shear
a complete description of the workability of a out a strong background in mechanics. This stress lies in any of the principal planes. Any
material is specified by its flow stress depend- knowledge is needed to understand coverage in state of stress can be resolved into the equivalent
ence on processing variables (for example, subsequent chapters. three principal stresses (Ref 1). It is important to
strain, strain rate, preheat temperature, and die In simplest terms, stress is force per unit area. note that both maximum and minimum values of
temperature), its failure behavior, and the metal- Two fundamentally different types of stress can normal stresses occur along the principal axes.
lurgical transformations that characterize the be readily identified (Fig. 1). The normal stress The usual convention is to denote the principal
alloy system to which it belongs. (s) acting on plane ab (Fig. 1) is the force per- stresses as s1, s2, s3, where algebraically s1 >
However, the major emphasis in workability pendicular to the plane divided by its area, A. s2 > s3.
is on measurement and prediction of the limits Alternatively, the normal stress is the axial force Strain is defined in terms of changes in di-
of deformation before fracture. The emphasis in (P) divided by the cross-sectional area of the mensions. If two gage marks were placed a dis-
this chapter is on understanding the factors that plane perpendicular to the axis of the rod: tance L0 apart on the specimen in Fig. 1, and if
determine the extent of deformation a metal can those marks were L1 apart after a tensile load P1
withstand before cracking or fracture occurs. It P cos q P was applied, then the engineering normal strain
s= =
is important, however, to allow for a more gen- Aab A (Eq 1) (e) would be:
eral definition in which workability is defined as
L1 - L0 DL 1 L1
the degree of deformation that can be achieved
in a particular metalworking process without
where A = Aab/cosq.
Shear stresses are those stresses applied along
e=
L0
=
L0
=
L0 dL
L0 (Eq 3)
creating an undesirable condition. Generally, the a specific direction in a specific plane. To deter-
undesirable condition is cracking or fracture, but mine the shear stress, the applied load P is re- If the bar had been compressed from an initial
it may be another condition, such as poor surface solved along the prescribed shearing direction length L0 to a shorter length L1, then the normal
finish, buckling, or the formation of laps, which (for example, in the direction along ab lying at strain calculated from Eq 3 would have a negative
are defects created when metal folds over itself an angle l in Fig. 1): sign, indicating that the strain was compressive.
during forging. In addition, in the most general Deformation of a body can result not only in
definition of workability, the creation by defor- Pshear = P cos l (Eq 2a) the change of length of a line in the body, but it
mation of a metallurgical structure that results in may also produce a change in the angle between
unsatisfactory mechanical properties, such as Hence, the shear stress, t, is given by: two lines. The resulting angular change is
poor fracture toughness or fatigue resistance, known as shear strain, g. Figure 2 illustrates the
can be considered to be a limit on workability. Pshear P strain produced by the simple shear on one face
t= = cos l cos q
Generally, workability depends on the local Aab A (Eq 2b)
conditions of stress, strain, strain rate, and tem-
perature in combination with material factors, When the plane and direction for which the
such as the resistance of a metal to ductile frac- shear stress is determined is a crystallographic
ture. In addition to a review of the many process
variables that influence the degree of workabil-
ity, the mathematical relationships that describe
the occurrence of room-temperature ductile frac-
ture under workability conditions are summa-
rized in this chapter. The most common testing
techniques for workability prediction are dis-
cussed in Chapter 4, Bulk Workability Testing
in this Handbook. Much greater detail on these Fig. 1 Normal and shear stresses Fig. 2 Shear strain
Chapter 2: Bulk Workability of Metals / 23
of a cube. The angle at O, which originally was of a testing machine, and the force and extension strain curve increases continuously up to frac-
90, is decreased by a small amount, q, by the are recorded until the specimen fractures. The ture (Fig 4). Then, if the strain is expressed as
application of a shear stress. The shear strain is load is converted into engineering normal stress true strain, the true-stress/true-strain curve re-
equal to the displacement, a, divided by the dis- with Eq 8, and the extension between gage sults. This often is called the flow curve because
tance between the shear planes, h: marks is converted to engineering strain with Eq it gives the stress required to plastically deform
3. This results in the engineering stress-strain (flow) the metal under conditions of no geomet-
a curve with a typical shape, as shown in Fig 3. rical or frictional constraint.
g= = tan q @ q (at small values of q)
h (Eq 4) In the elastic region of the curve, stress is lin- The formation of a neck in the tensile speci-
early related to strain, s = E e, where E is the men introduces a complex triaxial state of stress
As with stress, the complete description of the elastic (Youngs ) modulus. When the load ex- in that region. The necked region is, in effect, a
strain at a point requires the determination of ceeds a value corresponding to the yield stress, the mild notch. A notch under tension produces ra-
three normal strains and six shear strains. specimen undergoes gross plastic deformation. If dial stress (sr) and transverse stress (st) that
Engineering normal strain is based on a defini- the specimen were loaded part way and then un- raise the value of longitudinal stress required to
tion of strain with reference to an original unit loaded, it would be found to have been perma- cause the plastic flow. Therefore, the average
length, L0. This definition of strain is satisfactory nently deformed after the load returned to 0. The true stress at the neck, which is determined by
for elastic strains where DL is very small. stress to produce permanent (plastic) deformation dividing the axial tensile load by the minimum
However, in metalworking processes where the increases with increasing strain; the metal strain cross-sectional area of the specimen at the neck,
strains often are large, it is no longer logical to hardens. To a good engineering approximation, is higher than the stress that would be required
base strain on the original gage length. In this case, the volume remains constant as the specimen de- to cause flow if simple tension prevailed.
the calculation of strain is based on the instanta- forms plastically, AL = A0L0. Initially, the strain Figure 5 illustrates the geometry at the necked
neous gage length. This is called the true strain, e: hardening more than compensates for the decrease region and the stresses developed by this local-
in cross-sectional area of the specimen, and the ized deformation. The radius of curvature of the
L1 - L0 L2 - L1 L3 - L2 engineering stress continues to rise with increas- neck (R) can be measured either by projecting
e= L0
+
L1
+
L2
+ L
(Eq 5) ing strain. Eventually, a point is reached where the the contour of the necked region on a screen or
decrease in area is greater than the increase in by using a tapered, conical radius gage. A math-
load-carrying capacity arising from strain harden- ematical analysis has been made that provides a
or ing. This condition is reached first at some point in correction to the average axial stress to compen-
the specimen that is weaker than the rest. All fur-
L dL L
e= L0 L
= ln
L0 (Eq 6)
ther plastic deformation is concentrated in this re-
gion, and the specimen begins to neck or thin
down locally. Because the cross-sectional area is
True strain and engineering normal strain are now decreasing far more rapidly than strain hard-
easily related. From Eq 3: ening is increasing the deformation load, the engi-
neering stress continues to decrease until fracture
L - L0 L occurs. The maximum in the engineering stress-
e= = -1
L0 L0 strain curve is known as the ultimate tensile
strength, su. The strain at maximum load, up to
and from Eq 6: which point the cross-sectional area decreases uni-
formly along the gage length as the specimen
L elongates, is known as the uniform elongation, eu.
e = ln = ln(e + 1) The necking instability that occurs in the ten-
L0 (Eq 7)
sion test makes interpretation of the curve be-
yond maximum load more difficult. Because this
Comparison values of true strain and engi- is the region that is often of importance in met-
neering strain are given as follows: alworking processes, a better interpretation
needs to be used. The falloff in stress beyond Fig. 4 True-stress/true-strain curve (flow curve)
True strain, e Engineering strain, e
Pmax is artificial and occurs only because the
0.01 0.01 stress is calculated on the basis of the original
0.10 0.105 cross-sectional area, A0, when in fact the area at
0.20 0.22
0.50 0.65 the necked region is now much smaller than A0.
1.0 1.72 If the true stress based on the actual cross-sec-
4.0 53.6 tional area of the specimen is used, the stress-
Note that the normal stress given by Eq 1 is

based on the actual area over which the force (P)
acts. Therefore, it is a true normal stress, s. If
the normal stress were based on the original area
(A0) before the application of load, as is the case
in the usual engineering stress-strain curve, then
it would be defined as an engineering normal
stress, s:
s = P/A0 (Eq 8) Fig. 5 Stress distribution at the neck of a tensile spec-

imen. (a) Geometry of necked region. R is the
Stress-Strain Curves. In the conventional radius of curvature of the neck; a is the minimum radius at
the neck. (b) Stresses acting on element at point O. sx is
engineering tensile test, a test specimen is the stress in the axial direction; sr is the radial stress; st is
gripped at opposite ends within the load frame Fig. 3 Engineering stress-strain curve the transverse stress.
24 / Introduction
sate for the introduction of transverse stresses ness in workability studies is somewhat limited, A0
(Ref 2). According to this analysis, known as the due to the test being limited to relatively small e f = ln
Af (Eq 17)
Bridgman correction, the uniaxial flow stress strains, because fracture soon follows the onset
corresponding to that which would exist in the of necking. Thus, it is not generally possible to This parameter represents the maximum true
tension test if necking had not introduced triax- achieve strains of the same magnitude as those strain that the material can withstand before
ial stresses is: found in the metal deformation process. fracture and is analogous to the total strain to
As discussed previously, true stress in tension fracture of the engineering stress-strain curve.
(s x ) avg is the deformation load divided by the actual Because Eq 12 is not valid beyond the onset of
s=
1 + 2R a cross-sectional area: necking, it is not possible to calculate ef from
ln 1 +
a 2 R (Eq 9) measured values of engineering fracture strain,
P ef. However, for cylindrical tensile specimens,
s= the reduction in area, q, is related to the true
where (sx)avg is the measured stress in the axial A (Eq 11)
direction (load divided by minimum cross sec- fracture strain by:
tion), and a is the minimum radius at the neck. The true strain, e, may be determined from the 1
Figure 4 shows how the application of the engineering or conventional strain, e, by: e f = ln
Bridgman correction changes the true-stress/ 1- q (Eq 18)
true-strain curve. The values of a/R needed for L
the analysis can be obtained either by straining a e = ln(e + 1) = ln where
L0 (Eq 12)
specimen a given amount beyond necking and
unloading to measure a and R directly, or by A0 - Af
This equation is applicable only to the onset q=
measuring these parameters continuously past A0 (Eq 19)
necking, using photography or videography. of necking, because it assumes homogeneity of
To avoid these tedious measurements, deformation along the specimen gage length.
Ductility in the Tension Test. In addition
Bridgman presented an empirical relation be- Beyond maximum load, the true strain should be
to evaluating the strength of a material, the ten-
tween a/R and the true strain in the neck. Figure based on actual area or diameter, D, measure-
sion test also gives an indication of material duc-
6 shows that this gives close agreement for steel ments:
tility. Ductility is the ability of a material to de-
specimens but not for other metals with widely form without fracture. This is very similar to the
different necking strains. It has been shown that p 2 concept of workability. The common distinction
D
the Bridgman correction factor, B, can be esti- A0 4 0 D
e = ln = ln = 2 ln 0 is that workability refers to situations of large de-
mated from (Ref 3): A p D2 D formation, as in metalworking processes, while
4 (Eq 13) the term ductility is usually used in the context of
B = 0.83 - 0.186 log e (0.15 e 3) (Eq 10) machines and structures, where it is thought of as
The true stress at maximum load corresponds the ability to accommodate to a small amount of
where B = s/(sx)avg. to the true tensile strength. For most materials, plastic deformation without fracture.
True-Stress/ True-Strain Curve. The true- necking begins at maximum load at a value of The conventional measures of ductility that
stress/true-strain curve obviously has many ad- strain where the true stress equals the slope of are obtained from the tension test are the engi-
vantages over the engineering stress-strain curve the flow curve. Let su and eu denote the true neering strain to fracture and the reduction of
for determining the flow and fracture character- stress and true strain at maximum load when the area at fracture. The percentage elongation, %e,
istics of a material. This section develops many cross-sectional area of the specimen is Au. The is the elongation between the gage marks of the
of the properties that can be obtained from this ultimate tensile strength can be defined as: specimen divided by the original gage length,
test. While the test is considered a valuable basic expressed as a percentage:
test of material mechanical behavior, its useful- Pmax
su =
A0 (Eq 14) L - L0
%e = f 100
L0 (Eq 20)
and
Because an appreciable fraction of the defor-
Pmax mation is concentrated in the necked region of
su =
Au (Eq 15) the tensile specimen, the percentage elongation
for a material is higher if the gage length is short
Eliminating Pmax yields the true stress at maxi- than if the gage length was longer. Thus, it is al-
mum load: ways important to use a standard gage length
and to report the gage length over which the per-
A0 centage elongation was measured.
s u = su Empirical studies have shown that elongation
Au (Eq 16)
at fracture, ef, correlates with specimen geome-
try, according to the Bertella-Oliver equation
The true fracture stress is the load at fracture (Ref 4):
divided by the cross-sectional area at fracture.
This stress should be corrected for the triaxial -a
state of stress existing in the tensile specimen at L
ef = e0 = e0 K - a
fracture. Because the data required for this cor- A (Eq 21)
rection frequently are not available, true fracture
stress values are frequently in error. where K = L/A is the slimness ratio of the spec-
The true fracture strain, ef, is the true strain imen. If elongation is plotted against K on a log
Fig. 6 Relationship between Bridgman correction fac- based on the original area, A0, and the area after scale, a straight line results. The value of elon-
tor, s/(sx)avg, and true tensile strain fracture, Af: gation at K = 1 is e0, and a is the slope of the
line. Equation 21 predicts that a given elonga- where s0 is the yield stress, and K and n are the
tion will be produced in a material if same constants as in Eq 22. This equation may
be more satisfying than Eq 22, because the latter
K = L/ A implies that at zero true strain, the stress is zero.
It has been shown that s0 can be obtained from
is maintained constant. Thus, at a constant per- the intercept of the strain-hardening portion of
centage elongation: the stress-strain curve and the elastic modulus
line (Ref 6):
L1 / A1 = L2 / A2
1
K 1- n
s0 =
where 1 and 2 are two different specimens. To En (Eq 28)
provide comparable test results using a gage Fig. 7 Log-log plot of true-stress/true-strain curve. n is
the strain-hardening exponent; K is the strength
length L2 on a specimen with area A2 by means coefficient. Effect of Strain Rate and Temper-
of measurements in a specimen with area A1, it ature. The rate at which strain is applied to the
would only be necessary to adjust the gage tension specimen has an important influence on
length to the stress-strain curve. Strain rate is defined as
L1 = L2 A1 / A2 e = de/dt. It is expressed in units of s-1.
Increasing strain rate increases the flow stress.
Moreover, the strain-rate dependence of strength
Reduction of area, q, as measured by Eq 19, is increases with increasing temperature. The yield
the other ductility measure provided by the ten- stress and the flow stress at lower values of plas-
sion test. The occurrence of necking in the ten- tic strain are more affected by strain rate than the
sion test, however, makes any quantitative con- tensile strength.
version between elongation and reduction in If the crosshead velocity, v, of the testing ma-
area impossible. Although elongation and reduc- chine is v = dL/dt, then the strain rate expressed
tion in area usually vary in the same wayfor in terms of conventional engineering strain, e, is:
example, as a function of test temperature, tem-
pering temperature, or alloy contentthis is not de d ( L - L0 ) / L0 1 dL v
always the case. Generally, elongation and re- Fig. 8 Various forms of power curve s = Ken e = = = =
dt dt L0 dt L0 (Eq 29)
duction in area measure different types of mate-
rial behavior. Provided the gage length is not too
short, percent elongation is primarily influenced is not identical to the strain-hardening exponent. The engineering strain rate is proportional to
by uniform elongation, and thus it is dependent From the definition of n: the crosshead velocity. In a modern testing ma-
on the strain-hardening capacity of the material. chine in which the crosshead velocity can be set
Reduction in area is more a measure of the de- d (log s ) d (ln s ) e ds accurately and controlled, it is a simple matter to
n= = =
formation required to produce fracture, and its d (log e ) d (ln e ) s de (Eq 23) carry out tension tests at a constant engineering
chief contribution comes from the necking strain rate. The true strain rate, e, is given by:
process. Because of the complicated stress state
ds s
in the neck, values of reduction in area are de- =n de d[ln( L / L0 )] 1 dL v
de e (Eq 24) e = = = =
pendent on specimen geometry and deformation dt dt L dt L (Eq 30)
behavior, and they should not be taken as true
material properties. However, reduction in area It can be readily shown that the strain-harden-
is the most structure-sensitive ductility parame- ing exponent is equal to the true uniform strain, Equation 30 shows that for a constant
ter, and as such, it is very useful in detecting the eu (Ref 1): crosshead velocity, the true strain rate decreases
influence of structural changes in the material, as the specimen elongates or cross-sectional area
such as second-phase particles, inclusions, and A0 shrinks. To run tension tests at a constant true
n = e u = ln strain rate requires monitoring the instantaneous
porosity. Au (Eq 25)
Mathematical Expressions for the Flow cross section of the deforming region with
Curve. A simple power curve relation can ex- closed-loop control feedback to increase the
where Au is the area at maximum load. crosshead velocity as the area decreases (Ref 7).
press the flow curve of many metals in the re- Deviations from Eq 22 frequently are ob- The strain-rate dependence of flow stress at
gion of uniform plastic deformation, that is, served, often at low strains (10-3) or high strains constant strain and temperature is given by:
from yielding up to maximum load: (e 1.0). One common type of deviation is for a
log-log plot of Eq 22 to result in two straight s = C(e ) m (Eq 31)
e ,T
s = Ken (Eq 22) lines with different slopes. Sometimes data that
do not plot according to Eq 22 yield a straight
where n is the strain-hardening exponent, and K line according to the relationship: The exponent m in Eq 31 is known as the strain-
is the strength coefficient. A log-log plot of true rate sensitivity, and C is the strain-hardening co-
stress and true strain up to maximum load results s = K(e0 + e)n (Eq 26) efficient. The exponent m can be obtained from
in a straight line if Eq 22 is satisfied by the data the slope of a plot of log s versus log e. However,
(Fig. 7). The linear slope of this line is n, and K where e0 can be considered to be the amount of a more sensitive way to determine m is with a
is the true stress at e = 1.0 (corresponds to q = strain that the material received prior to the ten- rate-change test (Fig. 9). A tensile test is carried
0.63). As shown in Fig. 8, the strain-hardening sion test (Ref 5). out at strain rate e1, and at a certain flow stress,
exponent may have values from n = 0 (perfectly Another common variation on Eq 22 is the s1, the strain rate is suddenly increased to e2. The
plastic solid) to n = 1 (elastic solid). For most Ludwik equation: flow stress quickly increases to s2. The strain-
metals, n has values between 0.10 and 0.50. rate sensitivity, at constant strain and tempera-
Note that the rate of strain hardening, ds/de, s = s0 + Ken (Eq 27) ture, can be determined from Eq 32:
26 / Introduction
In this equation, K, C, m, and n are material pa- a deviator stress that represents the shear
rameters, and e0 and T0 are evaluated at refer- stresses in the total state of stress. Experience
ence conditions (Ref 9). shows that the greater the compressive mean
stress, the better the workability of a material.
Multiaxial Stress States For example, it is known that most materials can
be deformed more easily in an extrusion press
and even more so with hydrostatic extrusion
In design and metal processing situations, one (Ref 10). The mean or hydrostatic component of
expects to find stresses acting in more than just a the stress state is given by sm:
single direction. This section discusses some of
the most common situations that are encountered.
Plane Stress and Plane Strain. Because of s1 + s 2 + s 3
sm =
different geometries of the tooling and the work- 3 (Eq 38)
piece and the different ways that forces are ap-
Fig. 9 Strain-rate change test, used to determine plied in metal deformation processes, different In general, the greater the level of tensile
strain-rate sensitivity, m states of stress can be developed in the work- stress, the more severe the stress system is with
piece. It has been seen that the general state of regard to workability. For a given material, tem-
ln s e s D log s stress can be defined by three principal stresses, perature, and strain rate of deformation, the
m= = =
ln e e ,T s e D log e s1 > s2 > s3. Two states of stress that frequently workability is much improved if the stress state
occur in metalworking applicationsplane is highly compressive. A general workability pa-
log s 2 - log s1 log(s 2 /s1 )
= = stress and plane straindeserve special men- rameter, b, has been proposed that allows for the
log e 2 - log e 1 log(e 2 /e 1 ) (Eq. 32)
stress state (Ref 11):
tion. The use of the adjective plane implies that
the stresses are confined to a two-dimensional
The strain-rate sensitivity of metals is quite (plane) situation. 3s m
low (<0.1) at room temperature, but m increases b=
Plane stress occurs when the stress state lies in s (Eq 39)
with temperature. At hot working temperatures the plane of the member. This typically occurs
(T/TM > 0.5), m values of 0.1 to 0.2 are common when one dimension of the member is very small
in metals. Polymers have much higher values of Figure 10 shows the workability parameter
compared to the other two, and the member is plotted for various mechanical tests and metal-
m and may approach m = 1 in room-temperature loaded by a force lying in the plane of symmetry
tests for some polymers. working processes. The strain to fracture is the
of the body so that the stress normal to the plane ordinate. The curve is evaluated with three basic
The temperature dependence of flow stress surface is zero. Examples are thin, plate-type
can be represented by: tests: tension, torsion, and compression. Other
structures, such as thin-wall pressure vessels. common metal deformation processes are super-
Plane strain occurs when the strain in one of imposed. This figure emphasizes the critical role
s = C2eQ/RTe, e (Eq. 33) the three principal directions is zero (for exam- in workability that is played by the state of stress
ple, e3 = 0), as in the rolling of a wide sheet. In developed in the workpiece.
where Q is an activation energy for plastic flow, this case, the sheet does not change in width Yielding Criteria. The ease with which a
cal/gmol; R is the universal gas constant, 1.987 while it is reduced in thickness and increases in metal yields plastically or flows is an important
cal/ Kmol; and T is testing temperature in de- length. Although e3 = 0 for plane strain, there is factor in workability. If a metal can be deformed
gree Kelvin. From Eq 33, a plot of ln s versus a stress acting in that direction. For plastic de- at low stress, as in superplastic deformation,
1/T gives a straight line with a slope Q/R. formation, the stress in the principal direction then the stress levels throughout the deforming
Constitutive equations are mathematical for which strain is zero is the average of the workpiece are low, and fracture is less likely.
expressions that relate the stress in terms of the other two principal stresses, that is: The dominant metallurgical conditions and tem-
variable strain, strain rate, and temperature. For
perature are important variables, as is the stress
example, Eq 22, 31, and 33 are simple constitu- s1 + s 2
s3 = state. Plastic flow is produced by slip within the
tive equations. Such expressions are necessary 2 individual grains, and slip is induced by a high
for the computer modeling of deformation of
resolved shear stress. Therefore, the beginning
materials, but it must be noted that no univer- Effective Stress and Strain. Stress states in of plastic flow can be predicted by a maximum
sally accepted equations have been developed. metalworking processes are often complex. It is shear stress, or Tresca, criterion:
One of the oldest and most useful equations of convenient to be able to express this situation by
this type is: a single expression, the effective stress, s:
1
Z = A[sinh (as)]1/m (Eq 34) 1
s= [(s1 - s 2 )2 + (s 2 - s 3 )2 + (s 3 - s1 )2 ] 2
2
where Z, the Zener-Holloman parameter, is: (Eq 36)
Q Tensile stresses are positive, and compressive

Z = e exp
RT stresses are negative. Note that for uniaxial ten-
sion, the effective stress reduces to s = s1.
The evaluation of the parameter a is given in The effective strain, e, in terms of total plastic
Ref 8. strain, is:
The Johnson-Cook equation is widely used in 1
computer codes that handle large plastic defor- 2 2
e = (e12 + e 22 + e 32 )
mations: 3 (Eq 37)
e T - T0
m

s = (s 0 + Ke n )1 + C ln 1 - Hydrostatic Stress. In the theory of plas-
e 0 TM - T0 ticity, it can be shown that the state of stress can
Fig. 10 Influence of the stress state on the strain to
(Eq 35) be divided into a hydrostatic or mean stress and fracture, ef
1 s occurs because a die element keeps one dimen-

t max = ( s1 - s 3 ) = 0 sion constant, or only one part of the workpiece
2 2 (Eq 40)
is deformed, and adjacent nondeforming por-
where tmax is the maximum shear stress, and s0 tions exert a restraining influence. In either case,
is the yield (flow) stress measured in either a the restraint creates a stress in that principal di-
uniaxial tension or uniaxial compression test. rection.The stress is the average of the two other
Although adequate, this yield criterion neglects principal stresses (corresponding to point 4).
the intermediate principal stress, s2. The stress required for deformation is still s0 ac-
A more complete and generally applicable cording to Tresca but is 1.15 s0 according to von
yielding criterion is that proposed by von Mises: Mises. The latter is usually regarded as the
plane-strain flow stress of the material. It is
sometimes called the constrained flow stress.
2s20 = (s1 - s2)2 + (s2 - s3)2 + (s3 - s1)2 (Eq 41)
An important stress state is pure shear, in
which two principal stresses are of equal magni-
where s1 > s2 > s3 are the three principal
tude but opposite sign (point 5, Fig. 11), t = s1
stresses, and s0 is the uniaxial flow stress of the
= -s2 and s3 = 0. When these stresses are sub-
material. This is the same equation given earlier
for effective stress, Eq 36.
stituted into the von Mises yield criterion, Eq 41, Fig. 12 Tensile fracture modes as a function of tem-
it is seen that plastic deformation occurs at the perature (measured on the absolute tempera-
The significance of yield criteria is best illus- ture scale) and strain rate. TM, melting temperature
shear flow stress:
trated by examining a simplified stress state in
which s3 = 0 (plane stress). The Tresca yield cri- t0 = s0 / 3 dimpled rupture type of ductile fracture usually
terion then defines a hexagon, and the von Mises occurs. At very high temperatures, a rupture type
criterion an ellipse (Fig. 11). of fracture occurs in which the material recrys-
Figure 11 illustrates how the stress required to In metalworking theory, the shear flow stress is tallizes rapidly and pulls down to a point, with
produce plastic deformation varies significantly often denoted by the symbol k. nearly 100% reduction in area. A transgranular
with the stress state and how it can be related to creep type of failure occurs at temperatures less
the basic uniaxial flow stress of the material than those causing rupture. Intragranular voids
through a yield criterion. Yielding (plastic flow) Material Factors Affecting form, grow, and coalesce into internal cavities,
can be initiated in several modes. In pure ten- Workability which result in a fracture with a finite reduction
sion, flow occurs at the flow stress, s0 (point 1 in area.
in Fig. 11). In pure compression, the material Fracture Mechanisms. Fracture in bulk A more commonly found representation of
yields at the compressive flow stress, which, in deformation processing usually occurs as ductile possible fracture mechanisms is the fracture
ductile materials, is usually equal to the tensile fracture, rarely as brittle fracture. However, de- mechanism (Ashby) map (Fig. 13). Such a map
flow stress (point 2). When a sheet is bulged by pending on temperature and strain rate, the de- shows the area of dominance in terms of nor-
a punch or a pressurized medium, the two prin- tails of the ductile fracture mechanism vary. malized stress versus normalized temperature
cipal stresses in the surface of the sheet are equal Figure 12 illustrates in a schematic way the dif- for the dominant fracture mechanisms. The
(balanced biaxial tension) and must reach s0 ferent modes of ductile fracture obtained in a maps are constructed chiefly by using the best
(point 3) for yielding. tension test over a wide range of strain rates and mechanistic models of each fracture process.
An important condition is reached when de- temperatures. At temperatures below approxi- For more on high-temperature fracture mecha-
formation of the workpiece is prevented in one mately one-half the melting point of a given ma- nisms, see Ref (12).
of the principal directions (plane strain). This terial (below the hot working region), a typical Failure in deformation processing at below
one-half the melting temperature, 0.5 TM, occurs
by ductile fracture. The three stages of ductile
fracture are shown schematically in Fig. 14. The
first stage is void initiation, which usually oc-
curs at second-phase particles or inclusions.
Voids are initiated because particles do not de-
form, and this forces the ductile matrix around
the particle to deform more than normal. This in
turn produces more strain hardening, thus creat-
Fig. 11 (a) Directions of principal stresses and (b) yield criteria with some typical stress states Fig. 13 Fracture mechanism map for nickel.
28 / Introduction
Dynamic recovery results from the annihila-

tion of dislocations due to ease of cross slip,
climb, and dislocation unpinning at the hot
working temperature. These deformation mech-
anisms produce a microstructure consisting of
elongated grains, inside of which is a well-de-
veloped, fine subgrain structure, typically of the
order of 1 to 10 mm. The stress-strain curve for a
metal undergoing dynamic recovery shows an
increase in flow stress up to a steady-state value
that corresponds to the development of a steady-
Fig. 14 Stages in the dimpled rupture mode of ductile fracture. (a) Void initiation at hard particles. (b) Void growth.
state substructure (Fig. 16, curve a). The level of
(c) Void linking
the steady-state value increases with strain rate
ing a higher stress in the matrix near the parti- the probability of wedge cracking varies with and a decrease in deformation temperature.
cles. When the stress becomes sufficiently large, the applied strain rate. If the strain rate is so high In dynamic recrystallization dislocation, anni-
the interface may separate, or the particle may that the matrix deforms at a faster rate than the hilation only occurs when the dislocation den-
crack. As a result, ductility is strongly dependent boundaries can slide, then grain-boundary slid- sity reaches such high levels that strain-free re-
on the size and density of the second-phase par- ing effects are negligible. crystallized grains are nucleated. Therefore, the
ticles, as shown in Fig. 15. For high-temperature fracture initiated by rate of strain hardening is high until recrystal-
The second stage of ductile fracture is void grain-boundary sliding, the processes of void lization occurs (Fig. 16, curve b). However,
growth, which is a strain-controlled process. growth and coalescence, rather than void initia- when it begins, the flow stress drops rapidly as
Voids elongate as they grow, and the ligaments tion, are the primary factors that control ductil- recrystallization progresses.
of matrix material between the voids become ity. When voids initiated at the original grain Materials that experience rapid recovery and
thin. Therefore, the final stage of ductile fracture boundaries have difficulty in linking because thus do not undergo dynamic recrystallization
is hole coalescence through the separation of the boundary migration is high as a result of dy- are body-centered cubic iron; beta-titanium al-
ligaments that link the growing voids. namic recrystallization, hot ductility is high. In loys; hexagonal metals such as zirconium; and
Ductile fracture by void growth and coales- extreme cases, this can lead to highly ductile high stacking-fault energy, face-centered cubic
cence can occur by two modes. Fibrous tearing rupture, as shown in Fig. 12. metals such as aluminum. Face-centered cubic
(mode I) occurs by void growth in the crack Compressive stresses superimposed on tensile metals with lower stacking-fault energy, such as
plane that is essentially normal to the tensile or shear stresses by the deformation process can austenitic iron, copper, brass, and nickel, experi-
axis. In mode II void growth, voids grow in have a significant influence on closing small ence dynamic recrystallization in hot working.
sheets at an oblique angle to the crack plane cavities and cracks or limiting their growth and In these materials, dislocation climb is difficult.
under the influence of shear strains. This type of thus enhancing workability. Because of this im- This leads to higher dislocation densities than in
shear band tearing is found on the surface of the portant role of the stress state, it is not possible materials whose deformation is controlled by
cone in a ductile cup-and-cone tensile fracture. to express workability in absolute terms. dynamic recovery. Additional discussion and
It commonly occurs in deformation processing Workability depends not only on material char- references pertaining to the mechanisms of hot
in which friction and/or geometric conditions acteristics but also on process variables, such as working can be found in Chapter 3, Evolution
produce inhomogeneous deformation, leading to strain, strain rate, temperature, and stress state. of Microstructure during Hot Working, in this
local shear bands. Localization of deformation Hot Working. Many deformation processes Handbook.
in these shear bands leads to adiabatic tempera- for metals and alloys are performed at tempera- Flow Localization. Workability problems
ture increases that produce local softening. tures greater than 0.5 TM, the hot working range. can arise when metal deformation is localized to
Increasing the temperature of deformation Three principal benefits accrue from hot work- a narrow zone. This results in a region of differ-
leads to significant changes in deformation being. First, because the flow stress is lower at ent structures and properties that can be the site
havior and fracture mode. At temperatures above higher temperatures, it offers an economical of failure in service. Localization of deformation
one-half the melting point, particularly at low method for size reduction of large workpieces. can also be so severe that it leads to cracking in
strain rates, grain-boundary sliding becomes Second, metals at high temperatures are gener- the deformation process. In either mode, the
prominent. This leads to wedge-shaped cracks ally capable of achieving larger deformation presence of flow localization needs to be recog-
that propagate along grain boundaries and result strains without fracture than at lower tempera- nized and dealt with.
in low ductility. Such cracking is common at the tures. However, for many alloy systems, the tem- Flow localization is commonly caused by the
low strain rates found in creep but is not a fre- perature and strain rate must be appropriately formation of a dead-metal zone between the
quent occurrence at the faster strain rates in de- chosen. Lastly, high-temperature deformation workpiece and the tooling. This can arise from
formation processes unless there are brittle pre- assists in the homogenization of the ingot struc-
cipitates at the grain boundaries. This is because ture (chemical segregation) and in the closing
of internal voids. These benefits come at the cost
of surface oxidation, poorer dimensional accu-
racy, and the need to heat and cool the work
appropriately.
The mechanisms of hot working are rather
complex and vary considerably from alloy to
alloy. First, one needs to understand some termi-
nology. Those processes that occur during defor-
mation are called dynamic processes, while
those that occur between intervals of deforma-
tion or after deformation is completed are called
static processes. The two dynamic processes in- Fig. 16 (a) Stress-strain curve for a metal that under-
Fig. 15 Effect of volume fraction of second-phase
volved in hot working are dynamic recovery and
goes dynamic recovery in the hot working
particles on the tensile ductility of steel. region. (b) Stress-strain curve for a metal that undergoes
Source: Ref 13 dynamic recrystallization. dynamic recrystallization in the hot working region
presence of low-melting-point compounds that

frequently occur at grain boundaries. Deforma-
tion at temperatures too close to the melting
point of these compounds may lead to grain-
boundary cracking when the heat developed by
plastic deformation increases the workpiece
temperature and produces local melting. This
fracture mode is called hot shortness. It can be
prevented by using a sufficiently low deforma-
tion rate that allows the heat developed by de-
formation to be dissipated by the tooling, by
using lower working temperatures, or by sub-
jecting the workpiece to a homogenization heat
treatment prior to hot working. The relationship
between the workability of cast and wrought
structures and temperatures is shown in Fig. 19.
The intermediate temperature region of low
ductility shown in Fig. 19 is found in many met-
allurgical systems (Ref 17). This occurs at a
temperature that is sufficiently high for grain-
boundary sliding to initiate grain-boundary
Fig. 17 Consequences of friction illustrated in the upsetting of a cylinder. (a) Direction of shear stresses. (b) cracking but not so high that the cracks are
Consequent rise in interface pressure. (c) Inhomogeneity of deformation. ti, average frictional shear stress;
p, normal pressure; sf, flow stress; pa, average die pressure sealed off from propagation by a dynamic re-
crystallization process.
The relationship between workability and
poor lubrication at the workpiece-tool interface. conventionally cast ingot structures, workability temperature for various metallurgical systems is
Figure 17 illustrates the upsetting of a cylinder is lower, because cracks may initiate and propa- summarized in Fig. 20. Generally, pure metals
with poorly lubricated platens. When the work- gate easily along the grain boundaries. and single-phase alloys exhibit the best worka-
piece is constrained from sliding at the interface, Moreover, with cast structures, impurities are bility, except when grain growth occurs at high
it barrels, and the friction-hill pressure distribu- frequently segregated to the center and top or to temperatures. Alloys that contain low-melting-
tion is created over the interface. The inhomo- the surface of the ingot, creating regions of low point phases (such as gamma-prime-strength-
geneity of deformation throughout the cross sec- workability. Because chemical elements are not ened nickel-base superalloys) tend to be difficult
tion leads to a dead zone at the tool interface and distributed uniformly on either a micro- or a to deform and have a limited range of working
a region of intense shear deformation. A similar macroscopic scale, the temperature range over temperature. In general, as the solute content of
situation can arise when the processing tools are which an ingot structure can be worked is rather the alloy increases, the possibility of forming
cooler than the workpiece; in this case, heat is limited. low-melting-point phases increases, while the
extracted at the tools. Consequently, the flow Typically, cast structures must be hot worked. temperature for precipitation of second phases
stress of the metal near the interface is higher The melting point of an alloy in the as-cast con- increases. The net result is a decreased region
because of the lower temperature. dition is usually lower than that of the same for good forgeability (Fig. 21).
However, flow localization may occur during alloy in the fine-grain, recrystallized condition During the breakdown of cast ingots and the
hot working in the absence of frictional or chill- because of chemical inhomogeneities and the subsequent working by forging nonuniformities
ing effects. In this case, localization results from
flow softening (negative strain hardening). Flow
softening arises during hot working as a result of
deformation heating or microstructural instabili-
ties, such as the generation of a softer texture
during deformation or dynamic spheroidization.
Flow softening has been correlated with materi-
als properties determined in uniaxial compres-
sion (Ref 14,15) by the parameter:
g -1
ac =
m (Eq 42)
where g, the normalized flow-softening rate, is

given by g = (1/s) ds/de, and m is the strain-
rate sensitivity. Nonuniform flow in compres-
sion is likely when ac 5. For plane-strain com-
pression, as in side pressing, flow localization is
determined by ap = g/m 5. Figure 18 shows a
fracture that initiated at a shear band during
high-speed forging of a complex austenitic
stainless steel.
Metallurgical Considerations. Workabil-
Fig. 18 Austenitic stainless steel high-energy-rate forged extrusion. Forging temperature: 815 C (1500 F); 65% re-
ity problems depend greatly on grain size and duction in area; e = 1.4 103 s-1. (a) View of extrusion showing spiral cracks. (b) Microstructure at the tip
grain structure. When the grain size is large rel- of one of the cracks in area A of extrusion. Note that the crack initiated in a macroscopic shear band that formed at the
ative to the overall size of the workpiece, as in bottom of the field of view. Source: Ref 16
30 / Introduction
V = A0L0 = A1L1 (Eq 43)
If the plastic deformation is expressed as true

strain, then the constancy-of-volume condition
results in the following expression for the prin-
cipal strains:
e1 + e2 + e3 = 0 (Eq 44)
Deformation in metalworking is often ex-

pressed by the cross-sectional area reduction, R:
A0 - A1
R=
A0 (Eq 45)
From constancy of volume, Eq 43, and the

definition of true strain, one can write:
L1 A 1
e = ln = ln 0 = ln
L0 A1 1- R (Eq 46)
Strain rate has three chief effects in metal de-

formation processes: increases in strain rate
raise the flow stress, especially in strain-rate-
sensitive materials with a high m; the tempera-
ture of the workpiece is increased by adiabatic
heating, because there is little time for the heat
to dissipate; and there is improved lubrication at
the tool-metal interface, so long as the lubricant
film can be maintained.
If one considers a cylinder of height h being
upset in compression, then the strain rate is
given by:
de dh 1 v
e = = =
dt h dt h (Eq 47)
where dh/dt is the deformation velocity, v, and h

is the instantaneous height of the cylinder.
However, for most processes, the deformation
velocity is not a constant value. It is usual to de-
termine a time-averaged strain rate, e . For exam-
ple, for hot rolling:
e ln(h0 / h1 ) h 2 prn
e = = = ln 0
t L/v h1 r Dh (Eq 48)
Fig. 19 Relative workabilities of cast metals and wrought recrystallized metals at cold, warm, and hot working tem-
peratures. The melting point (or solidus temperature) is denoted as MPc (cast metals) or MPw (wrought and
recrystallized metals).
where L is the horizontal projection of the arc of
in alloy chemistry, second-phase particles, inclu- Process Variables Determining contact, r is the roll diameter, and n is the angu-
sions, and the crystalline grains themselves are Workability lar speed in revolutions per second.
aligned in the direction of greatest metal flow. Temperature. Metalworking processes are
This directional pattern of crystals and second- commonly classified as hot working or cold
phase particles is known as the grain flow pat- Strain. Stress and strain are defined in the working operations. Hot working refers to de-
tern. This pattern is responsible for the familiar section Stress, Strain, and Stress-Strain Curves formation under conditions of temperature and
fiber structure of wrought metal products (Fig. earlier in this chapter. This section relates the deformation velocity such that restorative
22). It also produces directional variation in such subject more closely to deformation processing processes (recovery and/or recrystallization)
properties as strength, ductility, fracture tough- and workability. The principal objective in plas- occur simultaneously with deformation (see the
ness, and resistance to fatigue. This anisotropy tic deformation processes is to change the shape section Hot Working in this chapter). Cold
in properties is greatest between the working of the deformed product. A secondary objective working refers to deformation carried out under
(longitudinal) direction and the transverse direc- is to improve or control the properties of the de- conditions for which restorative processes are
tion (Fig. 23). In a properly designed forging, formed product. not effective during the process. In hot working,
the largest stress should be in the direction of the In dense metals, unlike porous powder com- the strain hardening and the distorted grain
forging fiber, and the parting line of the dies pacts, the volume, V, of the workpiece remains structure produced by deformation are elimi-
should be located so as to minimize disruption to constant in plastic deformation as it increases in nated rapidly by the restorative processes during
the grain flow lines. cross-sectional area, A, and decreases in length, L: or immediately after deformation.
Fig. 22 Flow lines in a forged 4140 steel hook.

Specimen was etched using 50% HCl. 0.5
Fig. 20 Typical workability behavior exhibited by different alloy systems. TM: absolute melting temperature. Source:
Ref 18
ously with deformation. Therefore, the total de-

formation possible before fracture is less for cold
working than for hot working, unless the effects
of strain hardening are relieved by annealing.
Approximately 95% of the mechanical work
expended in deformation is converted into heat.
Some of this heat is conducted away by the tools
or lost to the environment. However, a portion Fig. 23 Anisotropy in wrought alloys
remains to increase the temperature of the work-
increases the propensity for fracture. If friction
piece. The faster the deformation process, the
is high, seizing and galling of the workpiece sur-
greater the percentage of heat energy that goes
faces occur, and surface damage results.
to increase the temperature of the workpiece.
The mechanics of friction at the tool-work-
Flow localization defects are enhanced by tem-
piece interface are very complex; therefore, sim-
perature buildup in the workpiece (see the sec-
plifying assumptions are usually used. One such
tion Flow Localization in this chapter).
assumption is that friction can be described by
Friction. An important concern in all practi-
Fig. 21 Influence of solute content on melting and so- Coulombs law of friction:
lution temperatures and therefore on forge- cal metalworking processes is the friction be-
ability tween the deforming workpiece and the tools
ti
and/or dies that apply the force and constrain the m=
Very large deformations are possible in hot shape change. Friction occurs because metal sur- p (Eq 49)
working, because the restorative processes keep faces, at least on a microscale, are never perfectly
pace with the deformation. Hot working occurs smooth and flat. Relative motion between such where m is the Coulomb coefficient of friction, ti
at essentially constant flow stress. Flow stress de- surfaces is impeded by contact under pressure. is the shear stress at the interface, and p is the
creases with the increasing temperature of defor- The existence of friction increases the value stress (pressure) normal to the interface.
mation. In cold working, strain hardening is not of the deformation force and makes deformation Another simplification of friction is to assume
relieved, and the flow stress increases continu- more inhomogeneous (Fig. 17c), which in turn that the shear stress at the interface is directly
32 / Introduction
proportional to the flow stress, s0, of the mate- trated and discussed in Chapter 12, Workability strain rate are critical parameters in ensuring that
rial: Theory and Application in Bulk Forming a deformation process achieves success, particu-
Processes, in this Handbook are a direct out- larly at elevated temperature where many differ-
s0 growth of the previously mentioned concept of ent mechanisms that lead to failure can occur. It
ti = m
3 (Eq 50) workability. would be advantageous if a processing map
The simplest and most widely used fracture could be developed that considered all of the
where m, the constant of proportionality, is the criterion (Ref 19) is not based on a microme- failure mechanisms that can operate in a mate-
interface friction factor. (The context of the situ- chanical model of fracture but simply recognizes rial over a range of strain rates and temperatures.
ation usually allows one to differentiate whether the joint roles of tensile stress and plastic strain This kind of processing map was developed for
m is friction factor or strain-rate sensitivity.) For in producing fracture: aluminum and is based on theoretical models of
given conditions of lubrication and temperature fracture mechanisms (Ref 20); however, it
ef
s * agrees well with experimental work. As Fig. 25
and for given die and workpiece materials, m is
usually considered to have a constant value in-
0
s
s
de = C
(Eq 52) shows, a safe region is indicated in which the
dependent of the pressure at the interface. Values material is free from cavity formation at hard
where s is the effective stress, s* is the maxi- particles, leading to ductile fracture, or wedge
of m vary from 0 (perfect sliding) to 1 (no slid-
ing). In the Coulomb model of friction, ti in- mum tensile stress, and C is a material constant cracking at grain-boundary triple points. The
creases with p up to a limit at which interface evaluated from the compression test. The effec- processing map predicts that, at constant tem-
shear stress equals the yield stress of the work- tive strain is defined by: perature, there should be a maximum in ductility
piece material. 1 with respect to strain rate. For example, at 500 K
2 2
Control of friction through lubrication is an e = (e12 + e 22 + e 32 )
important aspect of metalworking. High friction 3 (Eq 53)
leads to various defects that limit workability.
However, for most workability tests, conditions This fracture criterion indicates that fracture oc-
are selected under which friction is either absent curs when the tensile strain energy per unit vol-
or easily controlled. Most workability tests make ume reaches a critical value.
no provision for reproducing the frictional con- Use of this fracture criterion is shown in Fig.
ditions that exist in the production process; con- 24. The values of the reduction ratio at which
sequently, serious problems can result in the cor- centerburst fracture occurs in the cold extrusion
relation of test results with actual production of two aluminum alloys are illustrated. The en-
conditions. ergy conditions for different die angles are given
by the three curves that reach a maximum. The
fracture curves for the two materials slope down
Workability Fracture Criteria to the right. Centerburst occurs in the regions of
reduction, for which the process energies exceed
Workability is not a unique property of a the material fracture curve. No centerburst oc-
given material. It depends on such process vari- curs at small or large reduction ratios.
ables as strain, strain rate, temperature, friction
conditions, and the stress state imposed by the Process Maps Fig. 25 Composite processing map for aluminum
process. For example, metals can be deformed to showing the safe region for forming.
a greater extent by extrusion than by drawing be- Boundaries shift with microstructure. Instabilities due to
Mapping Based on Deformation Mecha- purely continuum effects, such as shear localization in
cause of the compressive nature of the stresses in nisms. It has been seen that temperature and sheet metal forming, are not considered. Source: Ref 20
the extrusion process that makes fracture more
difficult. It is useful to look at workability
through the relationship:
Workability = f1 (material) f2(process) (Eq 51)
where f1 is a function of the basic ductility of the

material, and f2 is a function of the stress and
strain imposed by the process. Because f1 de-
pends on the material condition and the fracture
mechanism, it is a function of temperature and
strain rate. Similarly, f2 depends on such process
conditions as lubrication (friction) and die
geometry. Therefore, to describe workability in
a fundamental sense, a fracture criterion must be
established that defines the limit of strain as a
function of strain rate and temperature. Also re-
quired is a description of stress, strain, strain
rate, and temperature history at potential frac-
ture sites. The application of computer-aided fi-
nite-element analysis of plastic deformation and
heat flow has facilitated this goal (see Chapters
10, Thermomechanical Testing, and 11,
Design for Deformation Processes, in this Fig. 24 Application of the Cockcroft-Latham fracture criterion (maximum tensile strain energy) to predict occur-
Handbook). The forming-limit diagrams illus- rence of centerburst fracture in aluminum alloy 2024 extrusions
(227 C, or 440 F), ductility should be at a max-

imum at a strain rate of 10-3 to 1 s-1. Below the
lower value, wedge cracking occurs; above this
level, ductile fracture reduces ductility.
The safe region shown in Fig. 25 is sensitive
to the microstructure of the metal. Decreasing
the size and volume fraction of hard particles
moves boundary 1 to the left. Increasing the size
or fraction of hard particles in the grain bound-
ary makes sliding more difficult and moves
boundary 2 to the right.
While a processing map is a very useful guide
for the selection of deformation processing con-
ditions, maps that are constructed from mecha-
nistic models of fracture are limited in practical
application as a design tool. The analyses can be
made only for pure metals and simple alloys, not
for complex engineering materials in which
strain-rate sensitivity is a function of tempera-
ture and strain rate. Moreover, the numerous ma-
terial parameters, such as diffusivity, that must Fig. 26 Dynamic material modeling processing map for the nickel-base superalloy Nimonic AP1. (a) Three-dimen-
sional plot of efficiency of power dissipation as a function of temperature and strain rate. (b) The corre-
be introduced into the models are difficult to ob- sponding contour map with numbers representing constant efficiency of power dissipation. Source: Ref 25
tain for complex engineering alloys. Of more
importance, the location of the boundaries in the or
processing map is very sensitive to microstruc-
ture and to prior thermomechanical history. It is P=G+J (Eq 55)
difficult to account for these factors in the mech-
anistic models. where G is the power dissipated by plastic work
Dynamic Material Modeling. Many work- (most of it converted into heat), and J is the dis-
ability problems arise when deformation is lo- sipator power co-content, which is related to the
calized into a narrow zone, resulting in a region metallurgical mechanisms that occur dynami-
of different structure and properties that can be a cally to dissipate power. A strong theoretical
site of failure. In severe cases, this leads to fail- basis for this position has been developed from
ure during processing. When less severe, it continuum mechanics and irreversible thermo-
serves as a potential site of failure in service. As dynamics (Ref 21, 22). Figure 27 illustrates the
has been seen, such flow localization can occur definitions of G and J. At a given deformation
in hot working in the absence of die chilling or temperature and strain:
frictional effects. Flow softening causes local- s s e m
ization of deformation due to deformation heat-
ing or microstructural instabilities, such as tex-
J=
0
e d s =
m +1 (Eq 56)
ture softening or dynamic spheroidization.
Because it is impossible to model from first where the constitutive equation for the material
principles so many structural phenomena, espe- is Eq 31, and m is the strain-rate sensitivity. The
cially in complex alloys, a different approach is value of J reaches its maximum when m = 1.
needed to create a processing map. The ap- Therefore:
proach that has evolved uses continuum princi-
ples with macroscopic determination of flow s e
J max =
stress as a function of temperature and strain rate 2 (Eq 57)
and applies appropriate criteria of instability to
identify the regions of T and e that should be This leads to the chief measure of the power
avoided in processing (Ref 21, 22). Then, micro- dissipation capacity of the material, the dimen-
scopic studies identify the nature of the instabil- sionless parameter called the efficiency of power
ity. Safe regions for processing are those that dissipation, h:
promote dynamic recrystallization, dynamic re-
covery, or spheroidization. Regions to avoid in J 2m Fig. 27 Schematic of constitutive relation of material
h= = system as energy converter (dissipator). (a)
processing are those that produce void formation J max m + 1 (Eq 58) Material system as nonlinear (general case) energy dissipa-
at hard particles, wedge cracking, adiabatic tor. (b) Material system as linear (special case) dissipator
shear band formation, and flow localization. Deformation processing should be focused on
The technique known as dynamic material the regions of maximum efficiency of power dis- The shaded region in Fig. 28 shows the pro-
modeling (DMM) maps the power efficiency of sipation, unless structural instabilities, for exam- cessing conditions to be avoided because of flow
the deformation of the material in a strain ple, flow localization, intrude (Fig. 26). The lo- localization. The instability criteria used in Eq
rate/temperature space (Fig. 26). At a hot work- cation of regions of microstructural instability 59 are based on the principle of maximum en-
ing temperature, the power per unit volume, P, are found by mapping the instability parameter, tropy production (Ref 23), that is, when the rate
absorbed by the workpiece during plastic flow is: x (e), where: of entropy production by a microstructural
change in the material is lower than the applied
ln[ m / ( m + 1)] rate of entropy, the material exhibits instabilities
e s x(e ) = +m<0
P = se =
0
sde +
0
e ds (Eq 54) ln e (Eq 59) during flow.
34 / Introduction
deform the material, the flow stress, and by its 11. V. Vujovic and A.H. Shabaik, A New
propensity for fracture. The flow characteristics Workability Criterion for Ductile Metals, J.
can be measured by methods described in Eng. Mater. Technol. (Trans. ASME), Vol
Chapter 4, Bulk Workability Testing, in this 108, 1986, p 245249
Handbook. Also discussed there are tests ex- 12. T.H. Courtney, Mechanical Behavior of
pressly designed to evaluate fracture. Materials, 2nd ed., McGraw-Hill, 2000, p
Most commercial metals are complex alloys 522550
that undergo a wide range of structural behavior 13. T. Gladman, B. Holmes, and L.D McIvor,
over the range of strain, strain rate, and temper- Effect of Second-Phase Particles on the
ature common to deformation processing. These Mechanical Properties of Steel, Iron and
mechanisms of fracture and instability must be Steel Institute, 1971, p 78
understood and identified. Ideally, enough infor- 14. S.L. Semiatin and G.D. Lahoti, The
mation is available to plot a process map that de- Occurrence of Shear Bands in Isothermal
lineates the safe and unsafe regions for process- Hot Forging, Metall. Trans. A, Vol 13, 1982,
ing. One way of doing this is with the p 275288
methodology called DMM. Another is to use the 15. J.J. Jonas, R.A. Holt, and C.E. Coleman,
tests described in Chapter 4, Bulk Workability Plastic Stability in Tension and
Testing, in this Handbook and in more detail in Compression, Acta Metall., Vol 24, 1976, p
other chapters of this Handbook to simulate the 911
process of interest and to identify the safe pro- 16. M.C. Mataya and G. Krauss, A Test to
Fig. 28 Finished processing map for the nickel-base cessing regions. Evaluate Flow Localization During
superalloy Nimonic AP1. Obtained by super-
position of instability regions determined with Eq 59 with
Forging, J. Appl. Metalwork., Vol 2, 1981, p
contours of percent efficiency of power dissipation. REFERENCES 2837
Shaded region corresponds to conditions of flow instabil- 17. F.N. Rhines and P.J. Wray, Investigation of
ity. Source: Ref 25 1. G.E. Dieter, Mechanical Metallurgy, 3rd the Intermediate Temperature Ductility
ed., McGraw-Hill, 1986, p 1836, 289290 Minimum in Metals, Trans. ASM, Vol 54,
The DMM methodology describes the dy- 2. P.W. Bridgman, Trans. ASM, Vol 32, 1944, 1961, p 117
namic path a material element takes in response p 553 18. A.M. Sabroff, F.W. Boulger, and H.J.
to an instantaneous change in e at a given T and 3. N.E. Dowling, Mechanical Behavior of Henning, Forging Materials and Practices,
e. As such, it is a map that graphically describes Materials, Prentice-Hall, 1993, p 165 Reinhold, 1968
power dissipation by the material in stable and 4. J.M. Holt, Uniaxial Tension Testing, 19. M.G. Cockcroft and K.J. Latham, Ductility
unstable ways. These boundaries correspond to Mechanical Testing and Evaluation, Vol 8, and Workability of Metals, J. Inst. Met., Vol
safe and unsafe regions on a processing map. ASM Handbook, ASM International, 2000, 96, 1968, p 3339
The use of the DMM methodology is in its in- p 131132 20. R. Raj, Development of a Processing Map
fancy, but it appears to be a powerful tool for 5. J. Datsko, Mechanical Properties and for Use in Warm-Forming and Hot-Forming
evaluating workability and controlling mi- Manufacturing Processes, John Wiley & Processes, Metall. Trans. A, Vol 13, 1982, p
crostructure by thermomechanical processing in Sons, 1966, p 1820 275288
complex alloy systems (e.g., see Ref 24 and 6. W.B. Morrison, Trans. ASM, Vol 59, 1966, p 21. H.L. Gegal, Synthesis of Atomistics and
Chapter 22, Multidisciplinary Process Design 824 Continuum Modeling to Describe
and Optimization: An Overview in this 7. J.W. House and P.G. Gillis, Testing Microstructure, Computer Simulation in
Handbook). A major compendium of processing Machines and Strain Sensors, Mechanical Materials Processing, ASM International,
maps has been published (Ref 25). Testing and Evaluation, Vol 8, ASM 1987
Handbook, ASM International, 2000, p 22. S.V.S. Narayana Murty, B. Nageswara Rao,
7892 and B.P. Kashyap, Instability Criteria in Hot
Summary 8. D. Zhao, Testing for Deformation Deformation of Materials, Int. Mater. Rev.,
Modeling, Mechanical Testing and Vol 45 (No.1), 2000, p 1526
The workability of metals is a complicated Evaluation, Vol 8, ASM Handbook, ASM 23. H. Ziegler, in Progress in Solid Mechanics,
subject. Workability is both a material attribute International, 2000, p 798810 I.N. Sneddon and R. Hill, Ed., Vol 4, North-
and a process characteristic. The two aspects in- 9. G.R. Johnson, J.M. Hoegfeldt, U.S. Holland, Amsterdam, 1963
teract in complex ways. The process establishes Lindholm, and A. Nagy, Response of 24. J.C. Malas and V. Seetharaman, Using
the state of stress in the material as well as its Various Metals to Large Torsional Strains Material Behavior Models to Develop
temperature, strain, and strain profile. In addi- Over a Large Range of Strain Rates, J. Eng. Process Control Strategies, JOM, Vol 44,
tion, the process imposes frictional conditions Mater. Technol. (Trans. ASME), Vol 105, (No. 6) 1992, p 813
that affect the stress state. 1983, p 4253 25. Y.V.R.K. Prasad and S. Sasidhara, Hot
The material response to the process environ- 10. H. Inoue and M. Nishihara, Ed., Hydrostatic Working Guide: A Compendium of
ment is expressed by the level of stress needed to Extrusion, Elsevier, 1985 Processing Maps, ASM International, 1997
Chapter 3
Evolution of Microstructure during

Hot Working
S.L. Semiatin, Air Force Research Laboratory, Materials and Manufacturing Directorate
IN PROCESS DESIGN, a compromise is namic recrystallization, metadynamic recrystal- often contains a collection of equiaxed subgrains
often necessary in the selection of conditions lization, static recovery, static recrystallization, (with low misorientations across their bound-
that give optimal workability versus those that and grain growth (Ref 1, 2). aries) contained within elongated primary grains
yield a desired microstructure. The term worka- Dynamic Recovery and Recrystallization. (Ref 2, 3). An example of such microstructural
bility is frequently applied in connection with As its name implies, dynamic recovery and re- features for an aluminum-lithium alloy is shown
the prevention of cracking or defects during crystallization occurs during hot working. As in Fig. 1. Furthermore, the dynamic-recovery
working and the determination of flow stresses metals are worked, defects are generated in the process leads to low stresses at high tempera-
and total working loads in metal-forming prac- crystal lattice. The most important defects are tures. Thus cavity nucleation and growth are re-
tice. These factors determine the necessary con- line defects known as dislocations. As deforma- tarded, and ductility is high. The evolution of
ditions of plastic flow (such as temperature, tion increases, the deformation resistance in- microstructure in high-SFE (and some low-SFE)
strain, and strain rate) and the required press ca- creases due to increasing dislocation content. materials worked at lower temperatures, such as
pacity, die materials and wear, and certain other However, the dislocation density does not those characteristic of cold-working, is similar.
die-design features. In addition, the develop- increase without limit because of the occurrence At these temperatures, subgrains may also form
ment of microstructure during bulk working, es- of dynamic recovery and dynamic recrystalliza- and serve as sinks for dislocations. However, the
pecially at hot-working temperatures, is an tion. subgrains are more stable. Thus, as more dislo-
equally important consideration when selecting In high stacking-fault-energy (SFE) metals cations are absorbed into their boundaries, in-
processing conditions. (e.g., aluminum and its alloys, iron in the ferrite creasing misorientations are developed, eventu-
The development of desirable microstructure phase field, titanium alloys in the beta phase ally giving rise to an equiaxed structure of
during hot working is a critical factor in process field), dynamic recovery (DRV) predominates. high-angle boundaries. Such a mechanism forms
design, and the description of microstructure During such processes, individual dislocations the basis for grain refinement in so-called severe
evolution (based on both phenomenological as or pairs of dislocations are annihilated because plastic deformation (SPD) processes such as
well as mechanism-based approaches) is reach- of the ease of climb (and the subsequent annihi- equal channel angular extrusion (ECAE) (Ref
ing a state of maturity. Thus, the design of ther- lation of dislocations of opposite sign) and the 46). This mechanism of grain refinement is
momechanical processes to control and optimize formation of cells and subgrains that act as sinks sometimes called continuous dynamic recrystal-
microstructure for specific service applications for moving (mobile) dislocations. Because sub- lization (CDRX) because of the gradual nature
is becoming increasingly widespread. In this grains are formed and destroyed continuously of the formation of high angle boundaries with
chapter, the general aspects of microstructure during hot working, the hot deformed material increasing strain.
evolution during thermomechanical processing
are reviewed briefly. The effect of thermome-
chanical processing on microstructure evolution
is briefly summarized to provide insight into this
aspect of process design. Attention is focused on
hot working and the key processes that control
microstructure evolutionthat is, dynamic and
static recovery and recrystallization and grain
growth. The discussion is divided into three
main subsections dealing with the mechanisms,
phenomenology, and modeling of microstruc-
ture evolution.
Mechanisms of
Microstructure Evolution
The key mechanisms that control microstruc-
ture evolution during hot working and subse- Fig. 1 Evolution of microstructure during hot-rolling of an aluminum lithium alloy undergoing dynamic recovery. (a)
Optical micrograph showing heavily deformed elongated initial grains and (b) TEM micrograph showing
quent heat treatment are dynamic recovery, dy- equiaxed subgrains. Source: K.V. Jata, Air Force Research Laboratory
36 / Introduction
In low-SFE materials (e.g., iron and steel in For low-SFE metals, a number of static soften- tallization (MDRX); this is followed by static re-
the austenite phase field, copper, and nickel), dy- ing processes may occur, depending on the level covery and recrystallization. Partially worked re-
namic recovery occurs at a lower rate under hot of deformation during hot working. These are gions may also undergo metadynamic recovery,
working conditions because mobile dislocations summarized in Fig. 3 for pure nickel (Ref 8). For thus reducing the stored energy needed for nucle-
are dissociated. Therefore, climb is difficult. prestrains much less than those required to initiation of static recrystallization and thus the abil-
This leads to somewhat higher densities of dislo- ate DDRX, sufficient stored energy to nucleate ity to obtain a fully softened condition.
cations than in materials whose deformation is static recrystallization is not available, and only Following static recovery and recrystalliza-
controlled by dynamic recovery. Furthermore, as static recovery occurs. Therefore, some dislocation, grain growth also occurs frequently (Ref
the temperature is increased, the mobility of tion substructure is retained, and full softening is 1). As in recovery and recrystallization, the driv-
grain boundaries increases rapidly. Differences not obtained. At somewhat higher prestrains just ing force for grain growth is a reduction in
in dislocation density across the grain bound- below those at which DDRX is initiated, static restored energy. For grain growth, the stored en-
aries, coupled with high mobility, lead to the nu- covery is followed by static recrystallization, ergy is in the form of grain-boundary energy.
cleation and growth of new, strain-free grains via which involves a nucleation-and-growth process. Grain growth may lead to very large grain sizes
a discontinuous dynamic recrystallization In this case, a fully annealed/softened mi- (and sharp crystallographic textures) particularly
process (DDRX) (Ref 2, 7). The evolution of a crostructure is obtained. For prestrains that ex- in single-phase metals heat treated at high tem-
dynamically recrystallized microstructure is il- ceed those at which DDRX was initiated during peratures. In alloys containing second-phase
lustrated in Fig. 2. At large strains, a fully re- hot working, residual DDRX nuclei undergo very particles, an equilibrium grain size may be
crystallized structure is obtained. However, even rapid growth without an incubation period in a reached due to pinning of the grain boundaries
at this stage, recrystallized grains are being fur- process known as post- or meta-dynamic recrys- by the second phase.
ther strained and thus undergo additional cycles
of dynamic recrystallization. Nevertheless, a
steady state is reached in which the rate of dislo-
cation input due to the imposed deformation is
balanced by dislocation annihilation due to the
nucleation and growth of new grains (as well as
some dynamic recovery). Hence, although a
nominally equiaxed grain structure is obtained at
large strains, the distribution of stored energy is
not uniform.
The presence of second-phase particles may
affect the evolution of microstructure during hot
working of both high- and low-SFE materials. In
high-SFE materials, particles may affect the ho-
mogeneity and magnitude of dislocation sub-
structure that evolves. In low-SFE materials,
particles may affect the evolution of substruc-
ture, serve as nucleation sites for dynamic re-
crystallization, and serve as obstacles to bound- Fig. 2 Discontinuous dynamic recrystallization (DDRX) in an initially coarse-grained nickel-base superalloy. (a)
Initial stage of DDRX and (b) nearly fully recrystallized microstructure
ary migration during the recrystallization
process.
Static Softening Processes and Micro-
structure Evolution. Residual dislocations
from hot working play an important role in the
evolution of microstructure during heat treat-
ment following hot working. In high-SFE mate-
rials, static recovery or static recovery and re-
crystallization may occur, depending on the
level of stored work, the rate at which the mate-
rial is reheated, and the annealing tempera-
ture/boundary mobility. Static recovery is simi-
lar to dynamic recovery in that climb of
dislocations and the absorption of dislocations
into subboundaries occur. Furthermore, subgrain
growth may occur, further reducing dislocation
density. However, a totally strain-free material
may not be obtained even after long annealing
times in the absence of static recrystallization. In
high-SFE metals such as aluminum that contain
second-phase particles, static recrystallization
may also occur as a result of particle-stimulated
nucleation (PSN) (Ref 1). In these cases, sub-
grains in regions of locally higher deformation
adjacent to the particles serve as nuclei for re-
crystallization. These subgrains grow rapidly,
consuming the substructure of the surrounding Fig. 3 Schematic illustration of the effect of hot-working prestrain on the subsequent static softening mechanisms and
relative magnitudes of softening due to each for pure nickel. The cross-hatched areas indicate conditions
material. under which incomplete softening occurs. Source: Ref 8
Chapter 3: Evolution of Microstructure during Hot Working / 37
Phenomenology of Plastic Flow and A somewhat more quantitative insight into the rate and temperature to a first order (Ref 11).
Microstructure Evolution shape of the flow curve for cases involving dy- Thus, the strain rate and temperature dependence
namic recovery may be obtained by an analysis of the rate of dislocation multiplication is deter-
of the overall rate of change of (mobile) disloca- mined principally by (e, T). An example
Some of the key phenomenological des- tion density, , with strain e, d/de, i.e.: of such a dependence for a microalloyed (Nb-B)
criptions of plastic flow and microstructure evo- steel is shown in Fig. 5(a) (Ref 11).
lution are summarized in this section. The quanti- To a first order, the flow stress under work-
tative description of plastic flow (i.e., stress-strain d/de d/destorage d/derecovery (Eq 2)
ing conditions is given by the following expres-
or flow curves) and microstructure evolution de- sion:
pends on the specific mechanism controlling de- The specific functional form of the dislocation
formation (and heat-treatment) response. storage and annihilation terms in Eq 2 can be ex-
pressed in the following form (Ref 10, 11): Gb (Eq 4)
Flow Curves in which denotes a constant whose magnitude

d/de U (Eq 3)
is between 0.5 and 1.0, G is the shear modulus,
The stress-strain (flow) curves that are meas-
In Eq 3, U denotes the rate of dislocation gener- and b is the length of the Burgers (slip) vector.
ured under hot working conditions are a function
ation due to strain hardening, and is a factor Inspection of Eq 3 and 4 reveals that the strain-
of the predominant dynamic softening mecha-
describing the rate of dynamic recovery. The rate rate and temperature dependence of the overall
nism.
of recovery is also directly proportional to the in- rate of hardening in the flow curve is largely de-
Flow Curves for Dynamic Recovery. As
stantaneous level of dislocation density, . At hot termined by . The strain-hardening rate is fre-
mentioned earlier, the hot-working response of
working temperatures, U is independent of strain quently quantified in terms of plots of d/de
high-SFE metals is controlled by dynamic re-
( ) as a function of . Typical plots for low-
covery. In such cases, dislocation generation is
carbon steel are shown in Fig. 5(b) (Ref 11).
offset by dislocation annihilation due to recov-
Equation 3 reveals that a steady-state disloca-
ery processes. The flow curve thus exhibits an
tion density, ss, is reached when U , or:
initial stage of strain hardening followed by a
steady-state (constant) flow stress. Typical flow
curves for iron in the ferrite phase field are ss U/ (Eq 5)
shown in Fig. 4 (Ref 9). The magnitude of the
steady-state flow stress ss decreases with in- The steady-state flow stress, ss, is thus given by
creasing temperature, T, and decreasing strain the following relation:
rate, e, typically according to a relation as fol-
lows (Ref 1, 2): s ss = aGb rss = aGb U/ W (Eq 6)
Z e exp(Qdrv/RT ) C1 sinh(C2ss)ndrv (Eq 1) There are a number of alternate approaches to

the modeling of dynamic recovery under the
in which Z denotes the Zener-Hollomon param- broad framework of Eq 2. For example, Kocks
eter, Qdrv is an apparent activation energy, R is (Ref 12) has shown that a linear dependence
the gas constant, and C1, C2, and ndrv are con- of strain-hardening rate (d/de) on stress ()
stants. is consistent with the following relation for
d/de:
dr / de = ( k1 r - k 2r)/ b (Eq 7)
in which k1 and k2 are constants. Similarly, for a

strain-hardening rate that varies linearly with
1/, Roberts (Ref 13) has shown that the follow-
ing relation applies:
dr /de = k3 - k 4 r (Eq 8)
in which k3 and k4 are constants.

Flow Curves for Discontinuous Dynamic
Recrystallization. Flow curves for materials
undergoing discontinuous dynamic recrystal-
lization (DDRX) have shapes that are distinc-
tively different from those that characterize ma-
terials that soften solely by dynamic recovery.
Those for DDRX exhibit an initial strain hard-
ening transient, a peak stress, flow softening,
and, finally, a period of steady-state flow.
Fig. 5 Characterization of flow behavior under hot Typical curves for two austenitic stainless steels
working conditions. (a) as a function of tem- are shown in Fig. 6 (Ref 14).
perature and strain rate for a Nb-B microalloyed steel and
(b) the overall hardening rate, , d/de as a function of
Dynamic recrystallization typically initiates
Fig. 4 Flow curves for Armco iron deformed under
at a strain of approximately five-sixths of the
hot-working conditions in the ferrite-phase stress, , at a strain rate of 2 s1 and various temperatures
field. Source: Ref 9 for a low carbon steel. Source: Ref 11 strain corresponding to the peak stress. Because
38 / Introduction
of this behavior, the peak stress for a material rise to grain-size coarsening or a reduction of Xdrx 1 exp{ln 2[(e ec)/e0.5]n6} (Eq 11)
that undergoes dynamic recrystallization is less less than one-half of the starting grain size ex-
than that which would be developed if the mate- hibit the oscillatory behavior (Ref 18). In Eq 10 and 11, Xdrx represents the fraction re-
rial softened solely by dynamic recovery (Fig. crystallized; C4, n5, and n6 are (positive) con-
7a). The strain at which DDRX initiates as well Phenomenological Models for stants; and ec and e0.5 denote the critical strain
as the steady-state flow stress that would be de- Microstructure Evolution for the onset of DDRX ( 5ep/6) and the strain
veloped in the absence of DDRX are readily defor 50% recrystallization, respectively. Equation
termined from a plot of d/de as a function of Dynamic Recovery. The key parameter that 11 is a classical Avrami-type relation (Ref 1, 20)
(Fig. 7b). characterizes the microstructure that evolves in for the nucleation-and-growth type processes
From a phenomenological standpoint, the materials that dynamically recover is the steady- that characterize DDRX. The Avrami exponent,
strain at the peak stress ep is usually found to de- state subgrain size dss. Two of the common cor- denoted as n6 above, typically has a value of
pend on the initial grain size, do, and the Zener- relations of dss to process variables are those de- approximately 2 for DDRX, but higher and
Hollomon parameter, Z, per an expression of the veloped by McQueen and Hockett (Ref 3) and lower values are not uncommon depending on
following form (Ref 1517): Derby (Ref 19). In the former work, dss was the type (e.g., site saturation versus continuous)
found to vary linearly with log Z (Fig. 8a). In the and sites (e.g., grain edges versus triple points)
ep C3 don3Z n4 (Eq 9) work of Derby, the normalized subgrain size of nucleation and the dimensionality (e.g., two-
(dss/b) of a number of materials was shown to dimensional or 2D versus three-dimensional or
in which C3, n3, and n4 are material-specific vary inversely with the normalized steady-state 3D) of growth (Ref 2022). The strain e0.5 for
constants. The activation energy specific to flow stress (ss/G) (Fig. 8b). 50% recrystallization is a weak function of Z,
DDRX is used in the determination of Z. Discontinuous Dynamic Recrystallization but depends strongly on the initial grain size do.
The regime of steady-state flow in stress- (DDRX). The key parameters that characterize Typically, e0.5 do0.3. Such a dependence is not
strain curves for materials that undergo dynamic DDRX are the dynamically-recrystallized grain surprising in view of the relatively smaller vol-
recrystallization may be smooth or exhibit an os- size ddrx and the kinetics of recrystallization as a ume fraction of nucleation sites (usually grain
cillatory behavior that dampens with increasing function of strain e. Devadas et al. (Ref 16) have boundaries) in coarser-grain materials. Hence,
strain. Temperature-strain rate conditions for summarized a number of such measurements. greater amounts of hot work may be required to
which the dynamically recrystallized grain size Typical relations are of the following form: fully recrystallize a cast ingot material via
is less than one-half of the initial grain size show DDRX as compared to a fine-grain wrought ma-
the former behavior. By contrast, those that give ddrx C4 Zn5 (Eq 10) terial unless nucleation occurs at sub-boundaries
that develop within the coarse grains during hot
working.
Metadynamic and Static Recrystallization.
Although the nucleation-and-growth mecha-
nisms are different for metadynamic (MDRX)
Fig. 7 Schematic illustration of work hardening behav-

ior for a material undergoing dynamic recrystal-
lization at hot-working temperatures. (a) Stress-strain
curve and (b) corresponding plot of d/de as a function of
Fig. 6 Flow curves for 316 and 317 stainless steels deformed under hot-working conditions. Source: Ref 14 stress,
Aluminum
Reciprocal
Fig. 8 Equilibrium subgrain sizes dss developed due to dynamic recovery during hot-working. (a) dss vs. Z for pure aluminum and (b) /G vs. dss/b for various materials. Source:
Ref 3, 19
and static (SRX) recrystallization, as discussed e* C9don14ep (Eq 14c)
earlier, the kinetics of both processes are also
usually well fit by an Avrami relation of the fol- and C7, C8, C9, n11, n12, n13, and n14 denote pos-
lowing form (Ref 15, 16): itive constants.
Grain Growth. Following recrystallization
Xmdrx,srx 1 exp[ln 2(t/t0.5)n7] (Eq 12) and the elimination of dislocation substructure,
the principal driving force for microstructural
in which Xmdrx,srx denotes the fraction trans- changes is the reduction of grain-boundary en-
formed by metadynamic or static recrystalliza- ergy. At elevated temperatures, classical isother-
tion, and t and t0.5 are time and time for 50% re- mal grain growth in single-phase alloys is
crystallization, respectively. The Avrami usually modeled with a phenomenological ex-
exponent n7 takes a value typically of the order pression of the following form:
of 1 (MDRX) or 2 (SRX). For static recrystal-
lization, the functional form for t0.5 depends on Dn Don Kt exp(Qg/RT) (Eq 15)
whether the hot-working prestrain is greater or
less than that required to initiate DDRX: in which D and Do denote the initial and final Fig. 9 Comparison of the measured dependence of the
grain sizes, n is the grain-growth exponent ratio of stable grain size (Ds) to particle size (dp)
t0.5 C5 en8 n9
do exp(Qs/RT) (e 5ep /6) (Eq 13a) (whose value typically ranges between 2 and 4), on volume fraction of particles ( fv) (data points) and
Zener-model predictions ( fv 0.01) or computer simula-
t0.5 C6 Zn10 exp(Qs/RT) (e 5ep /6) (Eq 13b) K is a constant, t is time, and Qg is the activation tion predictions of Hazzledine and Oldershaw ( fv 0.01)
energy for grain growth. Care must be exercised (solid lines). Source: Ref 24
In Eq 13a and 13b, C5, C6, n8, n9, and n10 are pos- in applying this expression particularly for ma-
itive constants, and Qs is the activation energy for terials that contain or develop noticeable crys-
static recrystallization. Similar expressions for t0.5 tallographic textures in which the grain-growth ical and computer models, respectively, for the
apply for the static recrystallization of high-SFE exponent n may vary during growth (Ref 1). In stable, pinned grain size for higher volume frac-
energy metals that undergo only dynamic recov- addition, alloys containing second phase parti- tions of particles (Fig. 9). For example, the for-
ery during hot working (Ref 15); in these cases, cles may exhibit parabolic growth kinetics ini- mer work leads to the following expression:
different fits are used depending on whether the tially, but eventually develop a stable grain size
prestrain is greater or less than that at which the due to particle pinning. For volume fractions fv Ds dp(6/fv)1/3 (Eq 16b)
maximum (steady-state) flow stress is achieved. of several percent or less of particles whose av-
The statically recrystallized grain size, dsrx, in erage diameter is dp, the stable grain diameter
metals that undergo DDRX is also a function of Ds due to such Zener pinning is given by (Ref
hot-working variables. Typical expressions are 1): Mechanistic Models for
as follows (Ref 15, 16): Microstructure Evolution
Ds 2pdp/3fv (Eq 16a)
dsrx C7don11en12 (e e*) (Eq 14a) A number of mechanism-based approaches
dsrx C8Z n13 (e e*) (Eq 14b) in which p is a constant between 0.25 and 1.0. have been developed to model microstructure
Hellman and Hillert (Ref 23) and Hazzledine evolution during hot working and annealing.
in which and Oldershaw (Ref 24) have developed analyt- These models incorporate deterministic and sta-
40 / Introduction
tistical aspects to varying degrees. In this sec- cleation conditions, the overall probability of size on the flow curve and necklace-recrystal-
tion, several broad classes of models that are nucleation is first calculated as PN Ndt/NCA, lization behavior (Fig. 11).
currently under development are discussed. in which dt is the time step, and NCA is the num- One of the major challenges of CA (and other
These include cellular automata, Monte-Carlo ber of CA cells. A cell is then selected at ran- kinds of discrete models) lies in the calibration
techniques, analytical models of grain (and sub- dom, and a random number between 0 and 1 is of the simulations relative to real time and spa-
grain) growth, and texture evolution models. generated. If the random number is greater than tial scales. To address some of these difficulties,
Each is briefly discussed subsequently. PN, the cell becomes a nucleus. The growth of a Davies (Ref 26) and Raabe (Ref 30) have used
nucleus to consume cells in its neighborhood two different methods to fix the interface migra-
may be automatic (i.e., every cell with a previ- tion rate in CA simulations of static recrystal-
Cellular Automata (CA) ously recrystallized cell in its neighbor under- lization. One is based on the theoretical relation
goes recrystallization) or specified based on between grain boundary velocity, grain bound-
Cellular automata (also sometimes referred to some sort of probability-based rule as well. ary mobility (as a function of diffusion parame-
as cellular automaton) is a numerical procedure Typical predictions of the evolution of the re- ters), and the dislocation density difference (i.e.,
used to model a number of metallurgical crystallized microstructure and fraction recrys- pressure difference) across the moving inter-
processes based on nucleation and growth tallized from a CA simulation assuming site sat- face. The other is semiempirical and is based on
or growth alone. These include static and dy- uration conditions are shown in Fig. 10. the Cahn-Hagel interface-averaged migration
namic recrystallization, precipitation, and grain Refinements of CA static recrystallization rate (Ref 31).
growth. Irrespective of the specific process models include the incorporation of the effect of
under investigation, rules for nucleation and inert second-phase particles on recrystallization
growth are specified and applied to a material as well as the extension to the modeling proce- Monte-Carlo Techniques
domain composed of discrete cells. Each cell is dure to DDRX conditions. For example, in the
surrounded by a neighborhood of adjacent cells former instance (Ref 28), a CA routine has been Monte-Carlo (MC) techniques for simulating
of a given geometry that determines the shape of developed in which the effect of particles is microstructure evolution were first developed in
the transformed phase. The rules may be applied purely geometric; that is, growth of a recrystal- the 1980s to investigate problems such as grain
at random sites, thus introducing a stochastic el- lized grain was assumed to cease on impinge- growth, static recrystallization, and dynamic re-
ement into the simulations as well. Because of ment with a particle. The simulations lead to no- crystallization. The foundation for MC methods
the discrete nature of the calculations, spatial as ticeable deviations from Avrami-like behavior lies in statistical mechanics (Ref 32) and the der-
well as average (temporal) predictions of struc- (e.g., Eq 12) with increasing particle-aspect ratio ivation of the Boltzmann factor. The Boltzmann
ture evolution are obtained. and, to a lesser extent, with increasing particle factor describes the ratio of the probabilities, P,
An example of the application of CA is static fraction and size. In the CA simulation of dy- that a system can be found in two distinct states
recrystallization (Ref 2527). In the simulation, namic recrystallization (Ref 29), rules for dislo- with energies of G1 or G2, namely:
cells may be either unrecrystallized or recrystal- cation input and recovery, nucleation, and
lized. A nucleation rule based on site saturation growth are specified. Specific sites at which re- P(G2)/P(G1) [exp(G2/kbT )] / [exp(G1/kbT)]
(all nuclei seeded at the beginning of the simu- covery and nucleation occur may be chosen ran- exp(G/kbT ) (Eq 17)
lation) or continuous nucleation is most com- domly or using deterministic methods based on
mon. The specific sites chosen as nuclei are se- local dislocation density. It has been shown that in which kb is the Boltzmann constant.
lected using a random number generator. For CA can replicate a number of the important fea- Perhaps the simplest problem that can be ad-
example, under continuous (constant rate N) nu- tures of DDRX such as the effect of initial grain dressed using the MC method is grain growth. In
Fig. 10 Cellular-automata predictions for static recrystallization under grain-boundary nucleation, site-saturation conditions. (a) Microstructure evolution and (b) fraction recrys-
tallized. Source: Ref 27
Fig. 11 Cellular-automata predictions for dynamic recrystallization. (a) Microstructure evolution and (b) stress-strain curves at various strain rates. Source: Ref 29
these simulations, the material domain, be it 2D The time increment in the MC technique cor- 18). An additional energy term (which may vary
or 3D, is discretized into material sites or points, responds to the Monte Carlo Step (MCS), which from grain to grain, depending on crystal orien-
each of which has one of a set of L possible crys- is equivalent to Ns trial flips, where Ns corre- tation) accounts for the difference in stored dis-
tallographic orientations. Grain boundaries di- sponds to the number of sites in the simulation. location energy across boundaries. Furthermore,
vide groups of sites with identical orientation. In initial MC investigations of grain growth nucleation sites are selected at random, usually
To simulate the kinetics of boundary motion, a (Ref 33), attention was focused on 2D simula- assuming site saturation or a constant nucleation
lattice site is selected at random, and a new trial tions. Furthermore, the variation of grain- rate. The probability of an (unrecrystallized to
orientation is also chosen at random from one of boundary energy and mobility with misorienta- recrystallized) transition at a given site is then
the orientations of the first or second nearest tion was not taken into account, and boundary determined as before.
neighborhood sites. The energy of the system migration was determined solely from the appli- The MC approach for dynamic recrystalliza-
before and after the flip is calculated from the cation of Eq 18 and 19. Nevertheless, normal tion (Ref 37) is similar to that for static recrys-
Hamiltonian H: growth was well simulated, yielding a growth tallization except that there is yet another term
exponent (Eq 15) of n 2.38. Subsequent work accounting for the input of dislocation energy at
H J(SiSj 1) (Eq 18) (Ref 34) considered boundary-energy variations a specified rate. Refined MC dynamic recrystal-
in calculating G as well as the effect of relative lization models (Ref 38) also incorporate the re-
in which Si is one of the L orientations on site i boundary mobility and relative boundary energy duction of dislocation density/stored energy due
(1 Si L) and hl is the Kronecker delta. on the flip probability. Such simulations pre- to dynamic recovery in addition to the specified
Thus, nearest neighbor pairs contribute J to the dicted grain-growth exponents between 2.38 and dislocation input. Hybrid MC-CA models that
system energy when they are of unlike orienta- 4. More recently, MC grain-growth approaches incorporate the influence of curvature driven and
tion and zero otherwise. The change in free en- have been extended to 3D problems and have in- stored-energy-driven transformation are also
ergy G before and after the flip is calculated as corporated crystallographic texture effects. For under development (Ref 39).
the difference between the two Hamiltonians. example, using 3D MC, Ivasishin and his col-
The transition probability, W, is then given by: leagues (Ref 35) have shown that texture influ- Models for Grain Growth
ences can give rise to values of n that vary dur-
ing the growth process (Fig. 12); such A number of analytical models have been de-
W exp(G/kbT ) G 0 (Eq 19a) phenomena were correlated with the interaction veloped to predict grain growth. Most of these
W1 G 0 (Eq 19b) of texture evolution and grain growth. In addi- are based on the assumption of curvature-driven
tion, in certain cases, n was shown to reach lev- boundary migration. In such cases, the driving
For G 0, the transition probability is com- els as high as 6. pressure is equal to s/R, where R is the radius of
pared to a number between 0 and 1 obtained The Monte-Carlo technique has also been ap- curvature of the grains. The velocity v ( dR/dt)
from a random-number generator. If W is greater plied to the analysis of both static and dynamic is then taken as proportional to the driving pres-
than that number (as well as if G 0), the flip recrystallization. The approach for static recrys- sure with the proportionality constant being the
is accepted. Successful transitions at a grain tallization (Ref 36) is similar to that for MC sim- mobility, M (Ref 1):
boundary to the orientation of a nearest neighbor ulation of grain growth in that material sites at a
grain correspond to boundary migration. grain boundary are assigned an energy (per Eq dR/dt M(s/R) (Eq 20)
42 / Introduction
If 2R represents the average grain size, Eq 20 in the presence of pinning particles. In such Texture Evolution Models
is easily integrated to yield a relation of the form cases, the curvature driving pressure is offset by
of Eq 15 with n 2. a restraining, or drag, pressure. For small vol- Texture evolution models fall into two main
Hillert (Ref 40) derived an expression similar ume fractions of particles, the drag pressure is categories, those principally for the prediction of
to Eq 20 that was capable of describing the that associated with Zener pinning and is pro- either deformation textures or recrystallization
growth of grains whose size is not uniform. The portional to s fv /dp. This assumption, in addi- textures. Deformation texture modeling is much
differential equation describing the growth of a tion to the generalization of Eq 20 to allow for more advanced compared to efforts for predict-
grain of radius R is: grain growth characterized by n p 2, yields ing recrystallization textures.
a differential equation of the following form for Deformation texture modeling treats the
dR/dt Ms[(1/Rcrit) (1/R)] (Eq 21) the average grain diameter, D, as a function of slip and twinning processes and the associated
time t (Ref 44, 45): crystal rotations to predict anisotropic plastic
Grains with R Rcrit were found to grow, and flow and texture evolution. Models of this sort
those with R Rcrit shrink. The critical grain dD/dt 2M(C*s)(p1)[(1/D) (Cfv /dp)](p1) include lower- and upper-bound approaches.
size Rcrit, which separated the two regimes, was (Eq 22) The lower-bound (or Sachs) method assumes
shown to be equal to the average grain size. that each grain of an aggregate is subjected to
Abbruzzese and Lucke (Ref 41), Eichelkraut et in which C*, C, and p are constants. The rela- the same stress state as the imposed macroscopic
al. (Ref 42), and Humphreys (Ref 43), among tion shows that the growth rate may be high stresses. The most highly stressed slip (or twin-
others, extended the Hillert model to take into when D is small but decreases as D approaches ning) system in each grain is used to establish
account the effect of texture/boundary misorien- the stable grain size ( dp/Cfv). The application yielding; a simple average of these stresses for
tation on mobility and boundary energy in the of Eq 22 to model the growth of alpha grains in all grains determines aggregate behavior. By
analysis of grain and subgrain growth. the presence of stable second-phase gamma par- contrast, the upper-bound method, introduced by
Relations similar to Eq 20 have also been de- ticles in a gamma titanium aluminide alloy is il- Taylor and Bishop and Hill (Ref 47, 48), as-
veloped to describe the kinetics of grain growth lustrated in Fig. 13 (Ref 46). sumes that each grain undergoes the same strain
as that imposed on the aggregate and that the
strain is uniform within the grains. The accom-
modation of an arbitrary imposed strain state
(with five arbitrary strain components) requires
the activation of at least five slip (or twinning)
systems. In the Taylor approach, the selection of
the specific set of systems is based on the prin-
ciple of minimum internal (virtual) work. On the
other hand, the Bishop-and-Hill approach makes
the selection of the required systems on the basis
of the maximum external work, a requirement
that ensures that the strain-increment vector is
normal to the yield locus so determined. It has
been shown that the two different approaches
give the same result. The Taylor/Bishop-Hill
methods have been applied to a very large num-
ber of metals with different crystal structures. A
number of software codes are available to con-
duct these analyses. One of the most popular
codes is the polycrystal plasticity program de-
veloped at Los Alamos National Laboratory and
known as LApp (Ref 49).
Although the upper-bound models give rea-
sonable estimates of deformation texture evolu-
Fig. 12 Monte Carlo (3D) model predictions of (a, b, c) grain structure (2D sections after 1000 MC steps) and (d)
grain-growth behavior for materials with various starting textures and assumed grain-boundary properties. (a) Fig. 13 Comparison of measured and predicted
Case A, isotropic starting texture and isotropic boundary properties (normal grain-growth case); (b) Case B, initial, sin- equiaxed alpha grain-growth kinetics for a
gle component texture, weakly anisotropic grain-boundary properties; and (c) Case C, initial, single-component texture, near-gamma titanium aluminide alloy annealed in the
strongly anisotropic grain-boundary properties. Source: Ref 35 alpha gamma phase field. Source: Ref 46
tion in many cases, a number of refinements have 2. J.J. Jonas and H.J. McQueen, Recovery of Steel Strip during Hot Rolling, Metall.
been made to such techniques (Ref 50). These in- and Recrystallization during High Temper- Trans. A, Vol 22, 1991, p 335
clude the development of a relaxed-constraints, ature Deformation, Treatise on Materials 17. G. Shen, S.L. Semiatin, and R. Shivpuri,
upper-bound approach in which the requirement Science and Technology, R.J. Arsenault, Modeling Microstructural Development
of five independent deformation systems is re- Ed., Academic Press, 1975, p 394 During the Forging of Waspaloy, Metall.
laxed for grain structures that are (or have be- 3. H.J. McQueen and J.E. Hockett, Micro- Mater. Trans. A, Vol 26, 1995, p 1795
come during deformation) elongated. A further structures of Aluminum Compressed at 18. T. Sakai and J.J. Jonas, Dynamic Recrys-
refinement relates to the fact that classical upper- Various Rates and Temperatures, Metall. tallization: Mechanical and Microstructural
bound models enforce strain compatibility, but Trans., Vol 1, 1970, p 2997 Considerations, Acta Metall., Vol 32, 1984,
typically give rise to stress states that are not to- 4. V.M. Segal, Equal Channel Angular p 189
tally compatible with the macroscopic stress Extrusion: From Macromechanics to 19. B. Derby, The Dependence of Grain Size on
state. To remedy this deficiency, self-consistent Structure Formation, Mater. Sci. Eng. A, Stress During Dynamic Recrystallization,
approaches, which allow the strain to vary from Vol 271, 1999, p 322 Acta Metall. Mater., Vol 39, 1991, p 955
grain to grain while enforcing the desired overall 5. R.Z. Valiev, R.K. Islamgaliev, and I.V. 20. M. Avrami, Kinetics of Phase Change: I.
imposed strain, have been developed. Alexandrov, Bulk Nanostructured General Theory, J. Chem. Phys., Vol 7,
Another refinement to deformation texture Materials from Severe Plastic Deforma- 1939, p 1103
modeling is that based on the finite element tion, Prog. Mater. Sci., Vol 45, 2000, p 103 21. M. Avrami, Kinetics of Phase Change:
method (FEM) and thus known as crystal-plas- 6. D.A. Hughes and N. Hansen, Charac- II. Transformation-Time Relations for
ticity FEM (Ref 51, 52). In this technique, each terization of Sub-Micrometer Structures in Random Distribution of Nuclei, J. Phys.
grain is divided into a number of elements. Each Heavily Deformed Metals Over the Entire Chem., Vol 8, 1940, p 212
of the elements has its own constitutive equation Misorientation Angle Range, Ultrafine 22. J.W. Cahn, The Kinetics of Grain
depending on its prior history (e.g., level of Grained Materials, R.S. Mishra, S.L. Boundary Nucleated Reactions, Acta
hardening or softening) and instantaneous crys- Semiatin, C. Suryanarayanan, N.N. Metall., Vol 4, 1956, p 449
tallographic orientation. With these local consti- Thadhani, and T.C. Lowe, Ed., TMS, 2000, 23. P. Hellman and M. Hillert, On the Effect of
tutive equations, the FEM analysis is then per- p 195 Second-Phase Particles on Grain Growth,
formed in a manner analogous to conventional, 7. R.D. Doherty, D.A. Hughes, F.J. Scand. J. Metall., Vol 4, 1975, p 211
continuum simulations. Although considerably Humphreys, J.J. Jonas, D. Juul Jensen, M.E. 24. P.M. Hazzledine and R.D.J. Oldershaw,
more computation intensive than other texture- Kassner, W.E. King, T.R. McNelley, H.J. Computer Simulation of Zener Pinning,
calculation techniques, the crystal-plasticity McQueen, and A.D. Rollett, Current Issues Philos. Mag., Vol 61A, 1990, p 579
FEM approach allows deformation nonunifor- in Recrystallization: A Review, Mater. Sci. 25. H.W. Hesselbarth and I.R. Gobel,
mities within grains, as well as from grain to Eng. A, Vol 238, 1997, p 219 Simulation of Recrystallization by Cellular
grain, to be established in addition to local tex- 8. T. Sakai, M. Ohashi, K. Chiba, and J.J. Automata, Acta Metall. Mater., Vol 39,
tures. The approach is also useful for the deter- Jonas, Recovery and Recrystallization of 1991, p 2135
mination of local conditions that may give rise to Polycrystalline Nickel after Hot Working, 26. C.H.J. Davies, Growth of Nuclei in a
cavitation, provided that the physics associated Acta Metall., Vol 36, 1988, p 1781 Cellular Automaton Simulation of Recrys-
with such a process can be quantified in terms of 9. J.-P.A. Immarigeon and J.J. Jonas, The tallization, Scr. Mater., Vol 36, 1997, p 35
the field variables used in the code. Deformation of Armco Iron and Silicon 27. R.L. Goetz and V. Seetharaman, Static
Recrystallization Texture Models. A rela- Steel in the Vicinity of the Curie Temper- Recrystallization Kinetics with Homoge-
tively recent development in texture modeling is ature, Acta Metall., Vol 22, 1974, p 1235 neous and Heterogeneous Nucleation Using
that associated with the recrystallization phenom- 10. A. Yoshie, H. Mirikawa, and Y. Onoe, Cellular Automata, Metall. Mater. Trans.
ena during (or following) hot deformation. During Formulation of Static Recrystallization of A, Vol 29, 1998, p 2307
hot working, the textures that evolve are a result of Austenite in Hot Rolling Process of Steel 28. C.F. Pezzee and D.C. Dunand, The
both dislocation glide and dynamic recrystalliza- Plate, Trans. ISIJ, Vol 27, 1987, p 425 Impingement Effect of an Inert, Immobile
tion. Texture evolution can be quantified by mech- 11. A. Laasraoui and J.J. Jonas, Prediction of Second Phase on the Recrystallization of a
anisms such as oriented nucleation and selective Steel Flow Stresses at High Temperatures Matrix, Acta Metall. Mater., Vol 42, 1994,
growth. In the former mechanism, recrystalliza- and Strain Rates, Metall. Trans. A, Vol 22, p 1509
tion nuclei are formed in those grains that have 1991, p 1545 29. R.L. Goetz and V. Seetharaman,
suffered the least shear strain (i.e., dislocation 12. U.F. Kocks, Laws for Work Hardening and Modeling Dynamic Recrystallization
glide). Such crystal orientations are usually estab- Low-Temperature Creep, J. Eng. Mater. Using Cellular Automata, Scr. Mater., Vol
lished from Taylor/Bishop-Hill models. Selective Techn., Trans. ASME, Vol 98, 1976, p 76 38, 1998, p 405
(i.e., faster) growth is then assumed to occur for 13. W. Roberts, Dynamic Changes That Occur 30. D. Raabe, Introduction of a Scalable
nuclei of particular misorientations with respect to During Hot Working and Their Significance Three-Dimensional Cellular Automaton
the matrix. The details of this approach have been Regarding Microstructural Development and with a Probabilistic Switching Rule for the
reviewed (Ref 53). Models incorporating oriented Hot Workability, Deformation, Processing, Discrete Mesoscale Simulation of Recrys-
nucleation and selective growth have also been ap- and Structure, G. Krauss, Ed., ASM, 1984, tallization Phenomena, Philos. Mag., Vol
plied to static recrystallization (Ref 54). Such p 109 79, 1999, p 2339
models are currently being integrated with cellular 14. H.D. Ryan and H.J. McQueen, Compar- 31. J.W. Cahn and W. Hagel, Theory of the
automata and crystal-plasticity FEM techniques to ative Hot Working Characteristics of 304, Pearlite Reaction, Decomposition of
predict the spatial and temporal evolution of static 316, and 317 Steels, Both Cast and Austenite by Diffusional Processes, V.F.
recrystallization. Worked, International Conf. on New Zackay and H.I. Aaronson, Ed., Interscience
Developments in Stainless Steel Technology, Publishers, 1962, p 131
R.A. Lula, Ed., ASM, 1985, p 293 32. C. Kittel, Thermal Physics, Wiley, 1969
REFERENCES 15. C.M. Sellars, Modeling Microstructure 33. M.P. Anderson, D.J. Srolovitz, G.S. Grest,
Evolution, Mater. Sci. Technol., Vol 6, and P.S. Sahni, Computer Simulation of
1. F.J. Humphreys and M. Hatherly, Recrys- 1990, p 1072 Grain GrowthI. Kinetics, Acta Metall.,
tallization and Related Phenomena, 16. C. Devadas, I.V. Samarasekara, and E.B. Vol 32, 1984, p 783
Elsevier, 1995 Hawbolt, Thermal and Metallurgical State 34. G.S. Grest, D.J. Srolovitz, and M.P.
44 / Introduction
Anderson, Computer Simulation of Grain 41. G. Abbruzzese and K. Lucke, A Theory of Cubic Crystal, Stephen Timoshenko 60th
GrowthIV. Anisotropic Grain Boundary Texture Controlled Grain GrowthDeriva- Anniversary Volume, Macmillan Company,
Energies, Acta Metall., Vol 33, 1985, p 509 tion and General Discussion of the Model, 1938, p 218
35. O.M. Ivasishin, S.V. Shevchenko, N.L. Acta Metall., Vol 34, 1986, p 905 48. J.F.W. Bishop and R. Hill, A Theory of
Vasiliev, and S.L. Semiatin, 3D Monte- 42. H. Eichelkraut, G. Abbruzzese, and K. Plastic Distortion of a Polycrystalline
Carlo Simulation of Texture Evolution and Lucke, A Theory of Texture Controlled Aggregate Under Combined Stresses,
Grain Growth during Annealing, Metal Grain Growth, Part II: Numerical and Philos. Mag., Vol 42, 1951, p 414
Physics and Advanced Technologies, 2001, Analytical Treatment of Grain Growth in 49. J.S. Kallend, U.F. Kocks, A.D. Rollett, and
Vol 23, p 1569 the Presence of Two Texture Components, H.-R. Wenk, Operational Texture Analysis,
36. D.J. Srolovitz, G.S. Grest, and M.P. Acta Metall., Vol 36, 1986, p 55 Mater. Sci. Eng. A, Vol 132, 1991, p 1
Anderson, Computer Simulation of 43. F.J. Humphreys, A Unified Theory of 50. U.F. Kocks, C.N. Tome, and H.-R. Wenk,
Recrystallization, Part I: Homogeneous Recovery, Recrystallization, and Grain Texture and Anisotropy, Cambridge Univer-
Nucleation and Growth, Acta Metall., Vol Growth Based on the Stability and Growth of sity Press, 1998
34, 1986, p 1833 Cellular Microstructures, Part I: The Basic 51. G.B. Sarma and P.R. Dawson, Effect
37. A.D. Rollett, M.J. Luton, and D.J. Model, Acta Mater., Vol 45, 1997, p 4231 of Interaction among Crystals on
Srolovitz, Microstructural Simulation of 44. I. Andersen and O. Grong, Analytical the Inhomogeneous Deformations of
Dynamic Recrystallization, Acta Metall., Modeling of Grain Growth in Metals and Polycrystals, Acta Mater., Vol 44, 1996, p
Vol 40, 1992, p 43 Alloys in the Presence of Growing and 1937
38. P. Peczak and M.J. Luton, A Monte Carlo Dissolving Precipitates, Part I: Normal 52. P. Bate, Modeling Deformation Micro-
Study of the Influence of Dynamic Grain Growth, Acta Metall. Mater., Vol 43, structure with the Crystal Plasticity Finite-
Recovery on Dynamic Recrystallization, 1995, p 2673 Element Method, Philos. Trans. R. Soc.
Acta Metall. Mater., Vol 41, 1993, p 59 45. G. Grewel and S. Ankem, Modeling (London) A, Vol 357, 1999, p 1589
39. A.D. Rollett and D. Raabe, A Hybrid Matrix Grain Growth in the Presence of 53. J.J. Jonas, Modeling the Length Changes
Model for Mesoscopic Simulation of Growing Second Phase Particles in Two That Take Place During The Torsion Test,
Recrystallization, Fourth International Phase Alloys, Acta Metall. Mater., Vol 38, Int. J. Mech. Sci., Vol 35, 1993, p 1065
Conf. on Recrystallization and Related 1990, p 1607 54. G. Gottstein and R. Sebald, Modeling of
Phenomena, T. Sakai and H.G. Suzuki, Ed., 46. V. Seetharaman and S.L. Semiatin, Analy- Recrystallization Textures, Thermo-
Japan Institute of Metals, 1999, p 623 sis of Grain Growth in a Two-Phase Gamma mechanical Processing of Steel, S. Yue and
40. M. Hillert, On the Theory of Normal and Titanium Aluminide Alloy, Metall. Mater. E. Essadiqi, Ed., Canadian Institute of
Abnormal Grain Growth, Acta Mater., Vol Trans. A, Vol 28, 1997, p 947 Mining, Metallurgy, and Petroleum,
13, 1965, p 227 47. G.I. Taylor, Analysis of Plastic Strain in a Montreal, 2000, p 21
G.E. Dieter, H.A. Kuhn, and S.L. Semiatin, editors, p 47-56 All rights reserved.
Chapter 4
Bulk Workability Testing

WORKABILITY is a complex property of a specimen. This introduces a triaxial tensile stress deform the specimen and the angle of twist, q, or
material, as indicated in Chapter 2, Bulk state and leads to fracture. the number of turns (q = 2p rad per turn). The
Workability of Metals. It is difficult to isolate For most metals, the uniform strain that pre- shear stress, t, on the outer surface of the speci-
the intrinsic workability, because this property is cedes necking rarely exceeds a true strain of 0.5. men is given by the following:
strongly influenced by stress state, which is in For hot working temperatures, this uniform
turn affected by friction and the geometry of the strain is frequently less than 0.1. Although ten- M (3 + m + n )
t=
tools and the workpiece. It has also been shown sion tests are easily performed, necking makes 2 pr 3 (Eq 1)
that the workability of a material is strongly in- control of strain rate difficult and leads to uncer-
fluenced by metallurgical structure, which can tainties about the value of strain at fracture be- where r is the specimen radius, m is the strain-
be a complex function of temperature and strain cause of the complex stresses that result from rate sensitivity found from plots of log M versus
rate. At the current state of development, the necking. Therefore, the utility of the tension test log q at fixed values of q, and n is the strain-
ability to model a forging process by calculating is limited in workability testing. This test is pri- hardening exponent obtained from the instanta-
stress, strain rate, and temperature throughout a marily used under special, high-strain-rate, hot neous slope of log M versus log q.
deforming workpiece with a computer-based fi- tension test conditions to establish the range of The engineering shear strain, G, and shear
nite-element technique exceeds the ability to hot working temperatures. A description of this strain rate, G, are given by:
predict the workability of the material. test method can be found in Chapter 7, Hot
A large number of tests are currently used to Tension Testing. rq
G=
evaluate the workability of a material. The pri- L (Eq 2)
mary teststension, torsion, compression, and Torsion Test
bendare discussed in this chapter. These are and
tests for which the state of stress is well defined In the torsion test, deformation is caused by r q
and controlled. Of these four tests, the compres- pure shear, and large strains can be achieved G =
sion test has been the most highly developed as L (Eq 3)
without the limitations imposed by necking (Ref
a workability test. 2, 3). Because the strain rate is proportional to
Specialized workability tests that have been where r is the radius of the specimen and L is the
rotational speed, high strain rates are readily ob- gage length. These values of shear stress and
developed from the four primary tests are also tained (Table 1). Moreover, friction has no effect
covered. Each of these tests provides informa- shear strain are typically converted to effective
on the test, as it does in compression testing. The , and effective strain, e, by means of the
stress, s
tion that is not readily available from the pri- stress state in torsion may represent the typical
mary tests. von Mises yielding criterion (see Chapter 2,
stress in metalworking processes, but deforma- Bulk Workability of Metals):
A number of workability tests that are espe- tion in the torsion test is not an accurate simula-
cially applicable to the forging process are dis- tion of metalworking processes, because of ex- s = 3t (Eq 4)
cussed in Chapter 13, Workability in Forging. cessive material reorientation at large strains.
Because of the previously mentioned advan- and
tages, the torsion test is frequently used to meas-
Primary Tests ure the flow stress and the stress-strain curve G
e=
(flow curve) under hot working conditions. 3 (Eq 5)
The primary tests for workability are those for Figure 1 shows typical flow curves as a function
which the stress state is well known and con- of temperature and strain rate. In the torsion test, versus
Figure 2 shows agreement in plots of s
trolled. Generally, these are small laboratory measurements are made of the moment, M, to e for stress-strain data determined in torsion,
simulation tests.
Table 1 Torsional rotation rates corresponding to various metalworking operations
Tension Test von Mises effective Corresponding surface shear
Operation strain rate ()(a), s-1 strain rate in torsion (), s-1 Rotation rate(b), rpm
The tension test is widely used to determine Isothermal forging 10-3 1.73 10-3 0.02
the mechanical properties of a material (Ref 1). Hydraulic press forging 1 1.73 16.5
Uniform elongation, total elongation, and reduc- Extrusion 20 34.6 330.4
Mechanical press forging 50 86.6 827.0
tion in area at fracture are frequently used as in- Sheet rolling 200 346.4 3307.9
dexes of ductility. The extent of deformation Wire drawing 500 866.0 8269.7
possible in a tension test, however, is limited by
(a) e = G3. (b) Assuming specimen geometry with r/L = 1.0
the formation of a necked region in the tension
48 / Workability Testing Techniques
conditions in hot working. The principal disad-

vantage of the compression test is that tests at a
constant, true strain rate require special equip-
ment.
Compression Test Conditions. Unless the
lubrication at the ends of the specimen is very
good, frictional restraint retards the outward mo-
tion of the end face, and part of the end face is
formed by a folding over of the sides of the orig-
inal cylinder onto the end face in contact with
the platens. The barreling that results introduces
a complex stress state, which is beneficial in
fracture testing but detrimental when the com-
pression test is used to measure flow stress. The
frictional restraint also causes internal inhomo-
geneity of plastic deformation. Slightly deform-
ing zones develop adjacent to the platens, while
severe deformation is concentrated in zones that
occupy roughly diagonal positions between op-
posing edges of the specimen (see Fig. 17 of
Chapter 2, Bulk Workability of Metals).
Figure 4 shows the hot upsetting of a cylinder
under conditions of poor lubrication in which
the platens are cooler than the specimen. The
cooling at the ends restricts the flow so that the
deformation is concentrated in a central zone,
with dead-metal zones forming adjacent to the
platen surfaces (Fig. 4a).
As deformation proceeds, severe inhomo-
geneity develops, and the growth of the end
faces is attributed entirely to the folding over of
the sides (Fig. 4b). When the diameter-to-height
ratio, D/h, exceeds approximately 3, expansion
of the end faces occurs (Fig. 4c).
The conditions described previously are ex-
treme and should not be allowed to occur in hot
compression testing unless the objective is to sim-
ulate cracking under forging conditions. Adequate
lubrication cannot improve the situation so that
homogeneous deformation occurs; however, with
glass lubricants and isothermal conditions, it is
possible to conduct hot compression testing with-
out appreciable barreling (Ref 6). Isothermal test
conditions can be achieved by using a heated sub-
assembly, such as that shown in Fig. 5, or heated
Flow curves for Waspaloy. (a) Effect of temperature at a fixed effective strain rate of 1 s-1. (b) Effect of strain dies that provide isothermal conditions (Ref 8).
Fig. 1
rate at a fixed test temperature of 1038 C (1900 F). Flow softening at the higher temperature The true strain rate, e, in a compression test is:
is a result of dynamic recrystallization. Source: Ref 4
de - dh / h 1dh v
tension, and compression. The agreement be- considered to be a standard bulk workability e = = =- =-
comes much better at hot working temperatures. dt dt hdt h (Eq 6)
test. The average stress state during testing is
Fracture data from torsion tests are usually re- similar to that in many bulk deformation
ported in terms of the numbers of twists to fail- processes, without introducing the problems of where v is the velocity of the platen, and h is the
ure or the surface fracture strain to failure. necking (in tension) or material reorientation (in height of the specimen at time, t. Because h de-
Figure 3 shows the relative hot workability of a torsion). Therefore, a large amount of deforma- creases continuously with time, the velocity must
number of steels and nickel-base superalloys, as tion can be achieved before fracture occurs. The decrease in proportion to (-h) if e is to be held
indicated by the torsion test. The test identifies stress state can be varied over wide limits by constant. In a normal test, if v is held constant, the
the optimal hot working temperature. For details controlling the barreling of the specimen engineering strain rate, e, remains constant:
on torsion testing methods, see Chapters 8 and 9, through variations in geometry and by reducing
Torsion Testing to Assess Bulk Workability friction between the specimen ends and the anvil de - dh / h0 1 dh - v
e = = =- =
and Hot Working Simulation by Hot Torsion with lubricants (see Chapter 5, Cold Upset dt dt h0 dt h0 (Eq 7)
Testing, respectively. Testing).
Compression testing has developed into a The true strain rate, however, is not constant.
Compression Test highly sophisticated test for workability in cold A machine called a cam plastometer can be used
upset forging, and it is a common quality-control to cause the bottom platen to compress the spec-
The compression test, in which a cylindrical test in hot forging operations. Compression forg- imen through cam action at a constant true strain
specimen is upset into a flat pancake, is usually ing is a useful method of assessing the frictional rate to a strain limit of e = 0.7. The use of cam
Chapter 4: Bulk Workability Testing / 49
Fig. 2 Comparison of effective stress-strain curves determined for type 304L stainless steel in compression, tension, and torsion. (a) Cold working and warm working temperatures.
(b) Hot working temperatures. Source: Ref 2
Fig. 5 Heated subassembly with specimen in position

used to achieve isothermal test conditions.
Fig. 3 Ductility determined in hot torsion tests. AISI, Thermocouple is removed prior to compression. Source:
American Iron and Steel Institute. Source: Ref 2 Ref 7
plastometers is limited; there probably are not servohydraulic testing machines. For an essen-
more than ten in existence. However, an essentially homogeneous upsetting test, a cylinder of
tially constant true strain rate can be achieved on diameter D0 and initial height h0 is compressed
a standard closed-loop servocontrolled testing to height h and spread out to diameter D1 ac-
machine. Strain rates up to 100 s-1 have been Fig. 4 Deformation patterns in nonlubricated, non-
isothermal hot forging. (a) Initial barreling. (b) cording to the law of constancy of volume:
achieved (Ref 6, 9). Barreling and folding over. (c) Beginning of end face ex-
When a constant true strain rate cannot be ob- pansion. Source: Ref 5
D 20 h0 = D2h (Eq 9)
tained, the mean strain rate may be adequate.
The mean true strain rate e for constant veloc- Flow Stress in Compression. Ideally, the If friction can be neglected, the uniaxial com-
ity, v0, when the specimen is reduced in height determination of flow stress in compression is pressive stress (flow stress), s, corresponding to
from h0 to h, is given by: carried out under isothermal conditions (no die a deformation force P is:
chilling) at a constant strain rate and with a min-
v ln(h0 /h) imum of friction in order to minimize barreling. P 4P 4 Ph
e = 0 s0 = = =
2 (h0 - h) (Eq 8) These conditions can be met with conventional A pD2 pD02 h0 (Eq 10)
If substantial friction is present, the average pressive flow stress. This assumes a von Mises area increase as the specimen is compressed, a
pressure, p, required to deform the cylinder is yield criterion
and an interface friction factor maximum results in the flow stress curve.
greater than the flow stress of the material, s0: m = 3ti /s0 where ti is the interfacial shear Analysis of the compression process indicates
(friction) component and where s0 is: that the plastic deformation is stable (no maxi-
2
p h 2 ma / h 2ma
mum in the flow curve) as long as (g + m) 1,
= e - -1 -1 where g is the dimensionless work-hardening
s 0 4ma h (Eq 11) 2 pR13 m
s 0 = P A + coefficient, and m is strain-rate sensitivity. Both
3 3h (Eq 13) of these material parameters are defined subse-
where a is the radius of the cylinder, and m is the quently (Ref 13,14). A material with a high
Coulomb coefficient of friction. The true com- A is the mean cross-sectional area obtained by di- strain-rate sensivity is more resistant to flow lo-
pressive strain, e, is given by: viding the volume of the specimen by the instan- calization in the tension test (necking), but in
taneous height of the compression specimen. R1 is compression testing, a higher rate sensitivity
h the radius of the specimen at the die-specimen in- leads to earlier flow localization.
e = ln 0 terface, and P is the compression load to produce
h (Eq 12) Flow softening or negative strain hardening
a specimen height h. Equation 13 assumes a rigid can also produce flow localization effects in
The effects of friction and die chilling can be plastic material that follows a power-law relation- compression independent of the effects of die
minimized through the use of a long, thin speci- ship for strain hardening. It gives excellent results chilling or high friction. The constant strain-rate,
men. Therefore, most of the specimen volume is in room-temperature comparisons but makes no isothermal hot compression test is useful for de-
unaffected by the dead-metal zones at the allowance for strain-rate-sensitive behavior such tecting and predicting flow localization.
platens. However, this approach is limited, be- as would be found in metals at elevated tempera- Nonuniform flow in compression is likely if a
cause buckling of the specimen occurs if h/D ex- ture. This subject is pursued further in Chapter 6, flow parameter, ac, exceeds a certain value:
ceeds approximately 2. Hot Compression Testing.
An extrapolation method involves testing Ductility Testing. The basic hot ductility
cylinders of equal diameters but varying heights, test consists of compressing a series of cylindri-
so that the D0/h0 ratio ranges from approxi- cal or square specimens to various thicknesses,
mately 0.5 to 3.0 (Ref 10). A specific load is ap- or to the same thickness with varying specimen
plied to the specimen, and the load is removed. length-to-diameter (length-to-width) ratios. The
The new height is determined in order to calcu- limit for compression without failure by radial
late a true strain. On relubrication, the specimen or peripheral cracking is considered to be a
is subjected to an increased load, unloaded, and measure of workability. This type of test has
measured. The cycle is then repeated. been widely used in the forging industry.
The same test procedure is followed with each Longitudinal notches are sometimes machined
specimen, so that the particular load levels are into the specimens before compression, because
duplicated. The results are illustrated in Fig. 6. the notches apparently cause more severe stress
For the same load, the actual strain (due to concentrations, thus providing a more reliable
index of the workability to be expected in a com- Specimens of Ti-10V-2Fe-3Al from isothermal
height reduction) is plotted against the D0/h0 Fig. 8
plex forging operation. hot compression tests. (a) to (c) Tested at 704 C
ratio for each test cylinder. A line drawn through (1300 F). (d) to (f) Tested at 816 C (1500 F). Strain rates
the points is extrapolated to a value of D0/h0 = 0. Plastic Instability in Compression. Sev-
were 10-3 s-1 (a, d), 10-1 s-1 (b, e), and 10 s-1 (c, f). Before
This would be the anticipated ratio for a speci- eral types of plastic instabilities can be devel- testing, the alloy had been b annealed to yield an
men of infinite initial height for which the end oped in the compression test. The first type is as- equiaxed b starting microstructure. Source: Ref 16
effects would be restricted to a small region of sociated with a maximum in the true stress-
the full test height. The true stress corresponding strain curve. The second type concerns inhomo-
to each of these true strains is given by Eq 10, geneous deformation and shear band formation.
with h values taken from Fig. 6 at D0/h0 = 0. Figure 7 shows the type of plastic instability that
Advanced Analysis. The analysis of the occurs in some materials in hot compression
compression of a cylinder so as to account for testing. At certain temperatures and strain rates,
the influence of bulging on the axial flow stress some of the typical strengthening mechanisms
has become a classical problem in mechanics. A become unstable (see Chapter 3, Evolution of
recent analysis provides useful information on Microstructure during Hot Working). Because
the problem (Ref 11). Starting with Hills varia- the rate of flow softening exceeds the rate of
tional principle (Ref 12), the authors developed
the following expression for the uniaxial com-
Fig. 6 Extrapolation method to correct for end effects Fig. 7 Example of compressive flow stress curve
in compressive loading. Source: Ref 10 showing strain softening Fig. 9 Three-point bend test
( g - 1) oped during bending are defined in Fig. 10. The pression test is primarily used to measure flow
ac = 5 critical parameter is width-to-thickness ratio, w/t. properties at room temperature, it can also be
m (Eq 14)
if w/t > 8, bending occurs under plane-strain con- used for elevated-temperature tests (see Chapter
where ditions (e2 = 0), and s2/s1 = 0.5. If w/t > 8, the 6, Hot Compression Testing).
bend ductility is independent of the exact w/t ratio. Because of its plane-strain geometry, this test
If w/t < 8, then stress state and bend ductility de- is more applicable to rolling operations than to
1 ds
g= pend strongly on the width-to-thickness ratio. forging.
s de (Eq 15)
For pure plastic bending, in which elastic de- The true stress and true strain determined
formation can be ignored, the maximum tensile from the test for plane-strain compression, epc,
and fiber strain is: can be expressed as:
ln s D log s Ro
m= T ,e T ,e e 0 = ln P
ln e D ln e (Eq 16) Ri (Eq 17) p=
wb (Eq 18)
Figure 8 illustrates the differences in deforma- where Ro is the radius of curvature on the outer
tion of titanium alloy samples. The specimens in (tensile) surface, and Ri is the radius of curvature t0
e pc = ln
Fig. 8(a) to (c) were deformed at a temperature, T, on the inner (compressive) surface (Ref 15). t (Eq 19)
at which ac was high. In Fig. 8(a), e = 10-3 s-1, When this strain is entered into the stress-strain
and ac = 2. In Fig. 8(b), e = 10-1 s-1, and ac = 5. equation or curve for the material, it gives the Because of the stress state associated with
flow stress for the material, s . Because of the
In Fig. 8(c), e = 10 s1, and ac = 5. However, the plane-strain deformation, the mean pressure on
specimens in Fig. 8(d) to (f) were deformed at a plane-strain
condition, the maximum fiber stress the platens is 15.5% higher in the plane-strain
/ 3
is 2 s
temperature at which ac was less than 0. . compression test than in uniaxial compression
testing. The true stress-strain curve in uniaxial
compression (s0 versus e) can be obtained from
Bend Test Specialized Tests the corresponding plane-strain compression
The bend test is useful for assessing the worka- curve (p versus epc) by:
bility of thick sheet and plate. Generally, this test In the plane-strain compression test, the
is most applicable to cold working operations. difficulties encountered with bulging and high 3 p
friction at the platens in the compression of s0 = p=
Figure 9 shows a plate deformed in three-point 2 1.155 (Eq 20)
bending. The principal stress and strains devel- cylinders can be minimized (Ref 10). As shown
in Fig. 11, the specimen is a thin plate or sheet
and
that is compressed across the width, w, of the
strip by narrow platens that are wider than the
strip. The elastic constraints of the undeformed 2
e= e pc = 1.155e pc
shoulders of material on each side of the platens 3 (Eq 21)
prevent extension of the strip in the width di-
mension; hence, the term plane strain. The plane-strain compression test is generally
Deformation occurs in the direction of platen accepted as one of the most reliable methods for
motion and in the direction normal to the length the generation of flow stress data (s versus e) for
of the platen. To ensure that lateral spread is neg- thermomechanical processing. For reproducible
ligible, the width of the strip should be at least measurements, corrections must be made for lat-
six to ten times the breadth of the platens. To en- eral spread of the specimen and deviation from
sure that deformation beneath the platens is es- ideal plane-strain behavior (Ref 17).
sentially homogeneous, the ratio of platen
breadth to strip thickness, b/t, should be between
2 and 4 at all times. It may be necessary to
change the platens during testing to maintain
Fig. 10 Schematic of the bend region defining direc- this condition. True strains of 2 can be achieved
tion of principal stresses and strains
by carrying out the test in increments in order to
provide good lubrication and to maintain the
proper b/t ratio. Although the plane-strain com-
Fig. 13 Secondary-tension test showing the geome-

Fig. 12 Partial-width indentation test. L @ h; b = h/2; tries of holes and slots. L @ h; wa 2 = h; b =
Fig. 11 Plane-strain compression test wa = 2L; I = 4L h/4; D = h/2
Fig. 15 Variation in shape of ring test specimens deformed the same amount under different frictional conditions.
Left to right: undeformed specimen; deformed 50%, low friction; deformed 50%, medium friction; de-
formed 50%, high friction
mately 440 C. Rolling tests on the material

showed that edge cracking was just beginning
when rolled at a temperature of 450 C. Good
correlation between loss of ductility in the STS
and incipient edge cracking was similarly ob-
tained for the 5182 alloy.
Ring Compression Test. When a flat, ring-
Fig. 14 Results of secondary-tension test on alu-
shaped specimen is upset in the axial direction,
minum alloy 7075. Source: Ref 20
the resulting change in shape depends only on
The partial-width indentation test is a test the amount of compression in the thickness di-
that is similar to the plane-strain compression rection and the frictional conditions at the die-
test for evaluating the workability of metals. ring interfaces. If the interfacial friction was
However, the partial-width indentation test does zero, the ring would deform in the same manner
not subject the test specimen to true plane-strain as a solid disk, with each element flowing out-
conditions (Ref 18). In this test, a simple slab- ward radially at a rate proportional to its dis-
shaped specimen is deformed over part of its tance from the center.
width by two opposing rectangular anvils hav- In the case of small, but finite, interfacial fric-
ing widths smaller than that of the specimen. tion, the outside diameter is smaller than in the
On penetrating the workpiece, the anvils longi- zero-friction case. If the friction exceeds a criti-
tudinally displace metal from the center, creat- cal value, frictional resistance to outward flow
ing overhangs (ribs) that are subjected to sec- becomes so high that some of the ring material
ondary, nearly uniaxial tensile straining. The flows inward to the center. Measurements of the
material ductility under these conditions is indi- inside diameters of compressed rings provide a
cated by the reduction in the rib height at frac- particularly sensitive means of studying interfa-
ture. The test geometry has been standardized cial friction, because the inside diameter in- Fig. 16 Theoretical calibration curve for standard
right with an outside-diameter-to-inside-
(Fig. 12). creases if the friction is low and decreases if the diameter-to-thickness ratio of 6 to 3 to 2
One advantage of this test is that it uses a friction is higher (Fig. 15).
specimen of simple shape. In addition, as-cast The ring test, then, is a compression test with
materials can be readily tested. One edge of the a built-in frictional measurement. Therefore, it is depending on the actual friction conditions.
specimen can contain original surface defects. possible to measure the ring dimensions and Under conditions of maximum friction, the
The test can be conducted hot or cold. The par- compute both the friction value and the basic largest usable specimen height is obtainable
tial-width indentation test, therefore, is suitable flow stress of the ring material at the strain under with rings of dimensions in the OD-to-ID-to-
not only for determining the intrinsic ductilities the given deformation conditions. thickness ratio of 6 to 3 to 1. Under conditions
of materials but also for evaluating the inhomo- Analysis of Ring Compression. The me- of low friction, thicker specimens can be used
geneous aspects of workability. This test has chanics of the compression of flat, ring-shaped while still satisfying the previously mentioned
been used to establish the fracture-limit loci for specimens between flat dies have been analyzed assumption. For normal lubricated conditions, a
ductile metals (Ref 19). using an upper bound plasticity technique (Ref geometry of 6 to 3 to 2 can be used to obtain re-
The secondary-tension test, a modification 21, 22). Values of p/s0 (where p is the average sults of sufficient accuracy for most applica-
of the partial-width indentation test, imposes forging pressure on the ring, and s0 is the flow tions.
more severe strain in the rib for testing highly stress of the ring material) can be calculated in For experimental conditions in which speci-
ductile materials. In this test, a hole or a slot is terms of ring geometry and the interfacial shear men thicknesses are greater than those permitted
machined in the slab-type specimen adjacent to factor, m. In these calculations, neither s0 nor by a geometry of 6 to 3 to 1 and/or the interface
where the anvils indent the specimen. Preferred the interfacial shear stress, t, appears in terms of friction is relatively high, the resulting side bar-
dimensions of the hole and slot are given in Fig. independent absolute values but only as the ratio reling or bulging must be considered. Analytical
13. With this design, the ribs are sufficiently m. treatment of this more complex situation is
stretched to ensure fracture in even the most The plasticity equations have been solved for available in Ref 24.
ductile materials. The fracture strain is based on several ring geometries over a complete range of The ring compression test can be used to
reduction in area where the rib is cut out, so that m values from 0 to unity (Ref 23), as shown in measure the flow stress under high-strain practi-
the fracture area can be photographed or traced Fig. 16. The friction factor can be determined by cal forming conditions. The only instrumenta-
on an optical comparator. measuring the change in internal diameter of the tion required is that for measuring the force
The secondary-tension test (STS) was used to ring. needed to produce the reduction in height. The
assess the workability in hot rolling of two alu- The ring thickness is usually expressed in re- change in diameter of the 6-to-3-to-1 ring is
minum alloys, alloy 5182 and 7075 (Ref 20). lation to the inside diameter (ID) and outside di- measured to obtain a value of the ratio p/s0 by
Figure 14 shows STS results as a function of ameter (OD). The maximum thickness that can solving the analytical expression for the defor-
temperature for the 7075 alloy. These data indi- be used while still satisfying the mathematical mation of the ring or by using computer solu-
cate that a loss of workability occurs at approxi- assumption of thin-specimen conditions varies, tions for the ring (Ref 25). Measurement of the
Fig. 17 The Gleeble test unit used for hot tension and compression testing. (a) Specimen in grips showing attached thermocouple wires and linear variable differential transformer
for measuring strain. (b) Closeup of a compression test specimen. Courtesy of Dynamics Systems, Inc.
area of the ring surface formerly in contact with high-temperature testing machine (Ref 26). A crosshead speed can be maintained as a constant
the die and knowledge of the deformation load solid buttonhead specimen that has a reduced di- throughout the test, but the true strain rate de-
facilitate calculation of p and, therefore, the ameter of 6.4 mm (0.250 in.) and an overall creases until necking occurs, according to the re-
value of the material flow stress, s0, for a given length of 89 mm (3.5 in.) is held horizontally by lationship:
amount of deformation. Repetition of this water-cooled copper jaws (grips) through which
process with other ring specimens over a range electric power is introduced to resistance heat de 1dL
of deformation allows the generation of a com- the test specimen (Fig. 17). Specimen tempera- e = =
dt L dt (Eq 22)
plete flow stress-strain curve for a given material ture is monitored by a thermocouple welded to
under particular temperature and strain-rate de- the specimen surface at its midlength. The ther-
formation conditions. This method is particu- mocouple, with a function generator, controls When the specimen necks, the strain rate in-
larly attractive for hot deformation processes the heat fed into the specimen according to a creases suddenly in the deforming region, be-
where die chilling of the workpiece makes it dif- programmed cycle. Therefore, a specimen can cause deformation is concentrated in a narrow
ficult to simulate the test conditions to measure be tested under time-temperature conditions that zone. Although this variable strain-rate history in-
the flow stress. simulate hot working sequences. troduces some uncertainty into the determination
Hot Tension Testing. Although necking is The specimen is loaded by a pneumatic-hy- of strength and ductility values, it does not negate
a fundamental limitation in tension testing, the draulic system. The load can be applied at any the utility of the hot tension test. Moreover, a pro-
tension test is nevertheless useful for establish- desired time in the thermal cycle. Temperature, cedure has been developed that corrects for the
ing the temperature limits for hot working. The load, and crosshead displacement are measured change in strain rate with strain so that stress-
principal advantage of this test for industrial ap- as a function of time. In the Gleeble test, the strain curves can be constructed (Ref 27).
plications is that it clearly establishes maximum
and minimum hot working temperatures (see
Chapter 7, Hot Tension Testing).
Most commercial hot tensile testing is done
with a Gleeble unit, which is a high-strain-rate,
Fig. 18 Reduction in area versus test temperature ob-

tained by hot tension testing on heating.
Specimens were heated to the test temperature, held 5 Fig. 19 (a) Localized strains on the bulging cylindrical surface of a compression test specimen. (b) Variation of
min, and pulled to fracture. strains with aspect ratio (h/D) of specimen and friction conditions. Source: Ref 28
Fig. 20 (a) Flanged and (b) tapered prebulged com-

pression test specimens. The lateral spread of
the interior material under compression expands the rim
circumferentially while little axial compression is applied.
The percent reduction in area is the primary re-

sult obtained from the hot tension test. This meas-
ure of ductility is used to assess the ability of the
material to withstand crack propagation.
Reduction in area adequately detects small ductil-
ity variations in materials caused by composition
or processing when the material is of low-to-
moderate ductility. It does not reveal small ductil-
ity variations in materials of very high ductility.
A general qualitative rating scale between re-
duction in area and workability is given in Table Fig. 21 Fracture limit lines for 2024 aluminum alloy in the T351 temper, measured by compression tests at room
temperature and at 250 C (480 F). Source: Ref 28
2. This correlation was originally based on su-
peralloys. In addition to ductility measurement,
the ultimate tensile strength can be determined
with the Gleeble test. This gives a measure of the
force required to deform the material. Typical
hot tension test curves are shown in Fig. 18.
Workability Analysis Using the

Fracture Limit Line
Workability is determined by two main fac-
tors: the ability to deform without fracture, and
the stress state and friction conditions present in
the bulk deformation process. These two factors
are brought together in an experimental worka-
bility analysis that has been further advanced
(Ref 28). The method is most applicable when
workability is limited by surface cracking, as in
edge cracks in rolling or contact surface cracks
in forging.
The first step is to determine the fracture
limit line for the material in question at the
temperature of interest. Details are given in
Chapter 5, Cold Upset Testing. While much
work has been done for room-temperature de-
formation processes, it is possible to measure Fig. 22 Example of workability analysis. (a) Upsetting of a bar with diameter d to head with diameter D (b) Material
fracture limit lines are superimposed on the strain paths by which the process achieves the final required
the curve at elevated temperature. Small com- strain. Strain path b (low friction) prevents fracture for both materials. Material B, with higher ductility, avoids fracture
pression specimens are used to establish the for either strain path. Source: Ref 28
line. Variation of the h/D ratio and the end lu-
brication are used to change the ratio of the cir- used to measure the flow stress. When friction e1. Changes in the geometry of the compression
cumferential tensile strain, e1, to the axial com- is present at the die contact surfaces, bulging specimen are used to extend the range of sur-
pressive strain, e2, in the specimen. Figure 19 occurs, and the strain path curves upward, as face strains toward the vertical (tensile) strain
shows that for frictionless (ideal) compression, shown in Fig. 19. axis. This is done by prebulging the specimens
the strain ratio is e1/e2 = -1/2, and s1 = 0. The The fracture limit line results from establish- by machining a flange or a taper on the cylin-
deformation in this case is referred to as homo- ing the values of e1 and e2 at which a surface ders (Fig. 20). Fracture limit curves take one of
geneous compression, because the only stress fracture can be just observed. Usually, this re- two forms: a straight line parallel to the line for
acting is an axial stress, and it is uniform quires compressing the specimen incrementally homogeneous compression, but displaced
throughout the specimen. This is the ideal con- with increasing strain or the testing of a series along the tensile strain axis, or a curve with two
dition to achieve when the compression test is of identical specimens to increasing levels of straight segments. At low compressive strains,
Fig. 23 Roll groove geometry for rolling square bars into round wire. Dimensions given in millimeters. Source: Ref 28
the line has a slope of -1, while at larger

strains, the slope is -1/2. Figure 21 shows frac-
ture limit lines for 2024 aluminum alloy that
are determined at room temperature and 250
C. Note how the tapered and flanged speci-
mens are useful in the strain region close to the
y-axis.
Workability Analysis. As a simplified ex-
ample of the use of the fracture locus curve in
workability analysis, consider a bolt-heading
process (Fig. 22). A bolt head, D, must be
formed from a rod with diameter, d. This re-
quires achieving a tensile circumferential true
strain of ln(D/d). However, the strain paths to
achieve this strain depend on process parame-
ters, such as h/d and friction, shown in Fig. 19.
The fracture limit line for two materials, A and
B, is shown. Now, if the strain path at the free
surface is denoted by curve b, working with ma-
terial A, the required strain is obtained without
crossing the fracture limit line. Therefore, there
is no fracture. However, if the lubrication in the
cold-heading process is inadequate, the strain
path may be curve a, and to reach ln(D/d) re-
quires crossing the fracture limit line. Fracture Fig. 24 Measured localized strains during the rolling of lead bars. Left side shows longitudinal strain versus verti-
ensues. There are two ways to solve this prob- cal compressive strain. Right side shows longitudinal strain versus cross-section reduction of area at room
temperature. Source: Ref 28
lem. Shifting to a material with a higher fracture
limit, B, solves the problem. If this is impossi- Table 2 Qualitative hot workability ratings for specialty steels and superalloys
ble, then improving the lubrication to move the
Hot tensile reduction Expected alloy behavior under normal hot
strain path back to curve b is called for. In other in area(a), % reductions in open-die forging or rolling Remarks regarding alloy hot working practice
more complex situations, the process variables
that can be changed include the die design, the <30 Poor hot workability, abundant cracks Preferably not rolled or open-die forged;
extrusion may be feasible; rolling or
workpiece (preform) design, and the distribution forging should be attempted only with
of lubricant. light reductions, low strain rates,
A more realistic example concerns the pro- and an insulating coating.
duction of aluminum 2024 round bar by hot 3040 Marginal hot workability, numerous cracks This ductility range usually signals the minimum
hot working temperature; rolled or press
rolling. Fracture occurs as a result of bulging of forged with light reductions and lower-than-
the free edges of the bars during rolling. Thus, it usual strain rates
should be possible to predict fracture in bar 4050 Acceptable hot workability, few cracks Rolled or press forged with moderate reductions
rolling from compression tests used to develop and strain rates
5060 Good hot workability, very few cracks Rolled or press forged with normal reductions
the fracture limit curve. Figure 23 shows the se- and strain rates
quence of passes required to roll a round bar. To 6070 Excellent hot workability, occasional cracks Rolled or press forged with heavier reductions
determine the strains developed on the surface of and higher strain rates than normal, if desired
the edge of the bar, the rolling process was mod- >70 Superior hot workability, rare cracks. Ductile Rolled or press forged with heavier reductions
ruptures can occur if strength is too low. and higher strain rates than normal, if alloy
eled with lead bars rolled at room temperature. strength is sufficiently high to prevent ductile
Surface strains were determined from measure- ruptures
ment of a grid scribed on the edges with a sharp
(a) Ratings apply for Gleeble tension testing of 6.4 mm (0.250 in.) diam specimens with 25 mm (1 in.) head separation. Source: Ref 26
tool. In other situations, the strains could be cal-
culated from analytical methods or by finite-ele-
ment analysis (Ref 29). Figure 24 shows the duced in going from a square to a diamond cross the correlation between tensile strain and area
measured localized strains for the various types section, and the least strain occurs on going from reduction is given. The fracture limit lines for
of roll passes. The greatest tensile strain is pro- a square to an oval one. To the right in Fig. 24, the material were given in Fig. 21. Superposition
14. S.L. Semiatin and J.J. Jonas, Formability

and Workability of Metals: Plastic Instabil-
ity and Flow Localization, American
Society for Metals, 1984
15. P. Dadras, Stress-Strain Behavior in Bend-
ing, Mechanical Testing and Evaluation,
Vol 8, ASM Handbook, ASM International,
2000, p 109114
16. S.L. Semiatin, Workability in Forging,
Workability and Process Design, ASM
International, 2003
17. N.J. Silk and M.R. van der Winden, Inter-
pretation of Hot Plane Strain Compression
Testing of Aluminum Specimens, Mater.
Sci. Technol., Vol 15, 1999, p 295300
18. S.M. Woodall and J.A. Schey, Development
of New Workability Test Techniques, J.
Mech. Work. Technol., Vol 2, 1979, p
367384
19. S.M. Woodall and J.A. Schey, Determina-
tion of Ductility for Bulk Deformation,
Formability TopicsMetallic Materials,
STP 647, ASTM, 1978, p 191205
20. D. Duly, J.G. Lenard, and J.A. Schey,
Applicability of Indentation Tests to Assess
Ductility in Hot Rolling of Aluminum
Alloys, J. Mater. Process. Technol., Vol 75,
Superposition of fracture limit line (dashed line) from Fig. 21 for 250 C on the measured strains in rolling
1998, p 143151
Fig. 25 21. B. Avitzur, Metal Forming: Processes and
lead at room temperature (model material). Source: Ref 28
Analysis, McGraw-Hill, 1968
of the curve for 250 C on the plot of strain path of Slender Cylinders, J. Eng. Ind. (Trans. 22. B. Avitzur and C.J. Van Tyne, Ring Forming:
for the square-to-diamond pass yields Fig. 25. ASME), Vol 104, 1982, p 79 An Upper Bound Approach, J. Eng. Ind.
This shows that the limiting reduction for hot 6. G. Fitzsimmons, H.A. Kuhn, and R. (Trans. ASME), Vol 104, 1982, p 231252
rolling of a square-to-diamond pass is approxi- Venkateshwar, Deformation and Fracture 23. A.T. Male and V. DePierre, The Validity of
mately 27%. Testing for Hot Working Processes, J. Met., Mathematical Solutions for Determining
Many other examples of the use of workabil- May 1981, p 1117 Friction from the Ring Compression Test, J.
ity analysis in forging and extrusion, and exam- 7. J.F. Alder and V.A. Phillips, The Effect of Lubr. Technol. (Trans. ASME), Vol 92, 1970,
ples of workability analysis applied to other Strain Rate and Temperature on the p 389397
forms of fracture, such as central burst and sur- Resistance of Aluminum, Copper and Steel 24. V. DePierre, F.J. Gurney, and A.T. Male,
face fir-tree defect, are given (Ref 28). to Compression, J. Inst. Met., Vol 83, Mathematical Calibration of the Ring Test
Workability analysis can be used as a trouble- 19541955, p 8086 with Bulge Formation, Technical Report
shooting methodology when problems arise in 8. F.J. Gurney and D.J. Abson, Heated Dies AFML-TR-37, U.S. Air Force Materials
production, or better yet, as a tool for the design for Forging and Friction Studies on a Laboratory, March 1972
of new processes. Modified Hydraulic Forge Press, Metall. 25. A.T. Male, V. dePierre, and G. Saul, ASLE
Mater., Vol 7, 1973, p 535 Trans., Vol 16, 1973, p 177184
9. J.G. Lenard, Development of an 26. E.F. Nippes, W.F. Savage, B.J. Bastian, H.F.
Experimental Facility for Single and Multi- Mason, and R.M. Curran, An Investigation
REFERENCES stage Constant Strain Rate Compression, J. of the Hot Ductility of High Temperature
Eng. Mater. Technol. (Trans. ASME), Vol Alloys, Weld. J., Vol 34, April 1955, p
1. P. Han, Ed., Tesile Testing, ASM Interna- 107, 1985, p 126131 183196s; see also http://www2.gleeble.
tional, 1992 10. A.B. Watts and H. Ford, On the Basic Yield com/gleeble/
2. S.L. Semiatin, G.D. Lahoti, and J.J. Jonas, Stress Curve for a Metal, Proc. Inst. Mech. 27. R.L. Plaut and C.M. Sellars, Analysis of
Application of the Torsion Test to Deter- Eng., Vol 169, 1955, p 11411149 Hot Tension Test Data to Obtain Stress-
mine Workability, Mechanical Testing, Vol 11. F.-K. Chen and C.-J. Chen, On the Strain Curves to High Strains, J. Test Eval.,
8, Metals Handbook, 9th ed., American Nonuniform Deformation of the Cylinder Vol 13, 1985, p 3945
Society for Metals, 1985, p 154184 Compression Test, J. Eng. Mater. Technol. 28. H.A. Kuhn, Workability Theory and
3. S.L. Semiatin and J. Jonas, Torsion Testing (Trans. ASME), Vol 122, 2000, p 192197 Application in Bulk Forming Processes,
to Assess Bulk Workability, Workability and 12. R. Hill, A General Method of Analysis for Forming and Forging, Vol 14, ASM Hand-
Process Design, ASM International, 2003 Metal-Working Processes, J. Mech. Phys. book, ASM International, 1988, p 388404
4. S. Fulop, K.C. Cadien, M.J. Luton, and H.J. Solids, Vol 11, 1963, p 305326 29. D.L. Dewhirst, Finite Element Analysis,
McQueen, J. Test. Eval., Vol 5, 1977, p 419 13. J.J. Jonas, R.A. Holt, and C.E. Coleman, Materials Selection and Design, Vol 20,
5. J.A. Schey, T.R. Venner, and S.L. Plastic Stability in Tension and Compres- ASM Handbook, ASM International, 1997,
Takomana, Shape Changes in the Upsetting sion, Acta Metall., Vol 24, 1976, p 911 p 176185
Chapter 5
Cold Upset Testing

Howard A. Kuhn, Consultant
DUCTILE FRACTURE is an extremely com-

plex phenomenon that is dependent on the mi-
crostructure of the material and the complex
stress and strain states developed during the de-
formation process (Ref 1, 2). The mechanism of
ductile fracture is not completely understood but
can be described by the process of void coales- Fig. 2 Cylindrical (left), tapered (center), and flanged
cence under shear deformation (Ref 3, 4). There (right) compression-test specimens
is also some evidence that local plastic instabil-
ity is associated with ductile fracture in metal- mens is 20. In no case should the end-face
working (Ref 5, 6). diameter be less than one-half of the overall
In evaluating the potential for fracture in a specimen height.
material during a particular process, it is gener- Friction at the die contact surface also has a
ally necessary to combine a workability test significant effect on the strains occurring at the
with a fracture criterion. The workability test free surface, particularly for cylindrical com-
establishes the ductility of the material under pression specimens. In the present work, the end
standard conditions. The function of the fracture faces of each specimen were polished to avoid
criterion is to extend this result to the stress and Fig. 1 Circumferential tension and axial compression
lubricant entrapment. Die conditions include
stresses at the equator of an upset cylinder
strain conditions existing in the deformation dies with knurled surfaces, dies with polished
process of interest. Attention must be given, surfaces (2 in. finish), and polished dies with
then, to the validity of the workability test in pression-test specimens can be measured from MoS2 lubricant. In the Male and Cockroft ring
simulating the type of fracture that occurs in the grid marks. Measurement of strains at fracture compression test (Ref 10), these friction condi-
actual process. for a wide range of test conditions leads to con- tions give friction factors, respectively, of 0.45,
Most metalworking processes involve compres- struction of a fracture-limit line for the material. 0.21, and 0.085.
sive deformation, and so the uniaxial compression Comparison of the strains in a material during an Each combination of friction condition and
test has been widely used for studying deforma- actual deformation process with the fracture- cylindrical aspect ratio gives a different combi-
tion behavior. However, the presence of friction limit line for the material indicates the possibil- nation of axial compressive strain and circum-
between workpiece and tool in the compression ity of fracture (Ref 9). ferential tensile strain at the free surface of a
test makes analysis complicated and restricts its cylindrical specimen. Friction has the greatest
application as a basic test of mechanical behavior. effect on these strains. For the flanged and ta-
A cylindrical specimen compressed with friction Upset Test Technique pered compression-test specimens, the height of
at the die surfaces does not remain cylindrical in the midplane cylindrical section affected the
shape but becomes bulged or barreled. Severity of Test-Specimen Geometry and Friction strains, but friction had very little effect. Each
the bulged cylindrical surface increases with in- Conditions. The aspect (height-to-diameter) test, with its own combination of surface
creasing friction and decreasing height-to-diame- ratio affects the strains occurring at the bulging strains, provides a point on the fracture-strain-
ter ratio (Ref 7). Tensile stresses associated with free surface. The upper limit on this ratio is 2.0 limit line.
the bulging surface make the upset test a candidate because of the possibility of buckling. The lower Strain Measurements. Use of the upset test
for workability testing (Fig. 1). limit is based on a convenient height for appli- for fracture studies, as first described by Kudo
The effects of the bulged surface in cylindri- cation of the grid marks. Normally, specimen as- and Aoi (Ref 11), involves measurement of the
cal compression tests can be enhanced by the pect ratios range between 0.75 and 1.75. axial and hoop strains at the equatorial surface.
use of shaped compression specimens. These Tapered and flanged test specimens are ma- Preliminary tests showed that the hoop strain
specimens contain flanged or tapered sections at chined cylinders with a section of the original was uniform around the circumference of the
midheight, which increase the circumferential surface remaining at midheight, as shown in barreled surface. The hoopstrains, therefore, can
tensile strain and decrease the axial compressive comparison with the cylindrical compression- be determined from measurements of the equa-
strain (Ref 8). test specimen in Fig. 2. The height of the cylin- torial diameter.
Collectively, the cylindrical, tapered, and drical surface at midheight ranges from 0.2 to The axial strain requires measurement of the
flanged compression-test specimens provide a 0.75 times the specimen height. The reduced separation of two gage marks placed symmetri-
wide range of circumferential tension/axial com- diameter of the flanged compression-test speci- cally above and below the midplane (Fig. 3).
pression strain states. mens is 0.8 times the original cylinder diameter, One of the following methods can be used for
Strains at the equatorial surfaces of the com- and the angle of the tapered compression speci- applying gage marks:
Two strips of cellophane tape are wrapped tainty affects the accuracy of the results. characteristics can be determined, unadulterated
around the cylindrical surface, such that Because large cracks grow rapidly, however, the by the effects of friction.
there is a thin band of bare metal around the strains measured at the first sight of a crack are Frictional constraint at the end faces of the
circumference at midheight. The specimens very close to those that existed when the crack specimen under conventional compression pre-
are then sprayed with blue steel dye, and the was actually initiated. vents uniform deformation and leads to a bulged
strips of tape are removed when the dye has Material Considerations. Because cracks surface at the midheight of the specimen, as
dried. The gage band is approximately 3 mm are initiated at the bulged surface, the ductility shown in the center specimen in Fig. 4. Such
(0.12 in.) wide and has sharply defined of the specimen during upsetting is significantly bulging leads to circumferential tensile stresses
edges that deform with the specimen surface influenced by surface-condition factors such as (depicted in Fig. 1) and the possibility of frac-
without cracking as the test proceeds. surface defects, decarburization, and residual ture.
The gage band is produced by electrochem- stress. It has been shown (Ref 12) that longitudi- The inhomogeneous deformation occurring
ical etching using the electromark system or nal surface defects on a spheroidized high- during compression with friction leads to three
by the photo-resist method. carbon steel sample drastically reduce its ductil- zones of deformation, as depicted in Fig. 5:
The gage band is marked by scribing fine ity during uniaxial compression testing due to
Metal in contact with the top and bottom
lines approximately 3 mm (0.12 in.) apart circumferential tensile-stress concentrations at
surfaces of the compression platens remains
with a carbide-tipped height gage. surface defects..
almost stationary. Such areas are also called
Four gage points are indented at midheight Material inhomogeneities of any type have an
dead-metal zones, or DMZs (region I in
of the cylindrical specimen with a Vickers or adverse effect on workability. Localized varia-
Fig. 5).
diamond brale hardness indenter. tions that can reduce the workability of a steel
Metal near the outer surface of the cylinder
include ferrite grain size, nonmetallic inclusion
Scribed lines or hardness indentations should deforms as a result of compression, resulting
content, banding, center segregation, and de-
be applied only to very ductile materials, be- in a bulged surface (region III in Fig. 5).
carburization. In all cases, therefore, the test-
cause these gage marks could be potential stress The most severe deformation is concen-
specimen free surfaces should contain the same
concentrations, leading to premature fracture trated in zones of shear just outside the
surface structural features as those of the actual
during upset testing. DMZs near each contact surface (region II
material to be used in the deformation process of
Test specimens complete with gage bands are in Fig. 5).
interest.
compressed incrementally between flat dies
Another condition to be taken into account is In the case of tapered or flanged compression-
under one of the friction conditions until cracks
the effect of mechanical anisotropy on ductile test specimens, the interior deformation of the
at the bulged surface are observed visually. For
fracture: the direction of the inclusion alignment cylinder expands the central region, accentuat-
each incremental deformation, displacement of
relative to the secondary tensile hoop stress gen- ing the circumferential tension. Because the free
the gage band is measured with a tool-makers
erated by friction during the uniaxial compres- surface at midheight is not directly in contact
microscope to an accuracy of 0.013 mm
sion test strongly influences the strain to frac- with the die surfaces along a straight line, com-
(0.0005 in.), and the diameter is measured
ture. It is necessary, therefore, that the test pression of this section is less than that in cylin-
with a micrometer to the same degree of accu-
specimen be prepared so that the direction of drical compression.
racy. Average values are calculated from three
tensile stresses, relative to the direction of inclu- Free-Surface Strains. At the free surfaces of
measurements in an effort to obtain more accu-
sion alignment, is the same during testing as it is the compressed cylinders, the strains consist of
rate data.
in the actual process. circumferential tension and axial compression.
The axial and circumferential true strains on
For frictionless (homogeneous) compression of
the barreled surface are calculated by using the
cylindrical specimens, the tensile strain is equal
relations given in Eq 1 and 2: Test Characteristics to one-half the compressive strain. With increas-
h
Axial e z = ln (Eq 1) Deformation. In a cylindrical compression
h0
test carried out under frictionless conditions, de-
W D formation of the test specimen is uniform, and
Circumferential e q = ln or 1n (Eq 2) no bulging is produced, as depicted in the sam-
W0 D0 ple on the right in Fig. 4. Frictionless conditions
can be achieved through the use of a synthetic
where h0 and h are initial and final gage heights, fluorine-containing resin film under laboratory
respectively; W0 and W are initial and final gage conditions. In this way, the material flow-stress
widths, respectively; and D0 and D are initial
and final diameters, respectively (Fig. 3).
Crack Detection. Most materials exhibit
some orange peel or surface roughening shortly Fig. 4 Cylindrical compression-test specimens: (left)
undeformed specimen; (center) specimen com-
after plastic flow begins. Fine networks of mi- pressed with friction (note crack); (right) specimen com-
crocracks may also appear if the surface is ob- pressed without friction
served at a magnification of approximately 30.
These cracks are stable and do not grow in size
as deformation progresses. They should be ig-
nored, because they do not represent a limitation
to useful deformation of the material. After fur-
ther deformation, one or more large cracks form
and grow rapidly; these cracks can easily be
seen with the unaided eye. The strains are meas-
ured, in accordance with Eq 1 and 2, when such
large cracks are initiated.
There is always some uncertainty regarding
the initiation of a large crack, and this uncer- Fig. 3 Grids for strain measurements on upset Fig. 5 Deformation zones in a longitudinal section of
cylinders an upset cylinder (see text)
Chapter 5: Cold Upset Testing / 59
Fig. 6 Strain paths in upset-test specimens
ing frictional constraint, bulge severity increases,

the tensile strain becomes larger, and the com- Fig. 7 Stresses at the equatorial surface of an upset-test specimen
pressive strain decreases. Effects of the bulged
profile, which develops naturally in cylindrical
compression specimens, are imposed artificially where der, de, and dez are incremental strains in tions permits evaluation of the effects of varia-
through tapered and flanged compression tests. the r, , and z-directions; r, , and z are tions in stress and strain states on the occurrence
Figure 6 summarizes the effects of friction, stresses in the r, , and z-directions; and d is a of fracture. The most convenient representation
aspect ratio, and specimen profile on the meas- proportionality constant that depends on mate- of fracture limits is a plot of circumferential and
ured free-surface strains at midheight. Measured rial and strain level and that is given by de/. For axial strains at fracture.
strain paths are shown in terms of circumferen- the surface area under consideration (Fig. 1), the Previous studies have shown that the results
tial versus axial strain. Starting with the strain stress in the r-direction is zero (r 0), and the of homogeneous cylindrical compression tests
ratio of one-half for the case of homogeneous stress state becomes plane stress. fit a straight line having a slope of one-half for
deformation, the strain-path slope increases with By manipulation of Eq 35, the stress compo- all materials (Ref 7, 9). The projected intercept
increasing friction. For a given value of friction, nents during deformation can be calculated ac- with the vertical axis depends on the material.
a decreasing aspect ratio slightly increases the cording to the following equations: Some examples are shown in Fig. 8. In addition,
strain-path slope. Tapered compression speci- fracture lines have been generated for a wide va-
mens further increase the strain-path slope, and riety of materials by various investigators. A
s 2a + 1 summary of these results is given in Ref 15, in-
flanged compression specimens result in strain sz = (Eq 8)
3 a2 + a + 1 dicating the value of the intercept for each mate-
paths that lie nearly along the circumferential
tensile-strain axis. rial. As expected, materials having greater duc-
s a+2 tility have higher values of the fracture line
Empirical relationships between test condi- sq = (Eq 9)
tions and the strain paths shown in Fig. 6 have 3 a2 + a + 1 intercept.
been derived in Ref 13. By approximating the Use of flanged and tapered upset-test speci-
strain paths as second order equations, the coef- mens expands the range of strains available for
where dez /de, which is determined graphi-
ficients of the linear and quadratic terms are cor- testing. Figures 9 and 10 show the combined
cally from the strain paths (Fig. 6). The effective
related statistically to the specimen aspect ratio results of cylindrical, tapered, and flanged
stress (
) is taken from the stress-strain curve of
(H/D) and friction. In addition, finite element upset tests on two alloys. The combination of
the material as obtained by a homogeneous com-
analyses of the upsetting process (e.g., Ref 14) results from all tests provides a continuous
pression test.
have been performed to determine the strain spectrum of strain states, with the tapered and
From Eq 8 and 9, the stress ratio, , is
paths as a function of the process parameters. flanged specimens adjoining, in sequence, the
defined as:
Stress States. Stresses at the free surfaces of cylindrical upset-test results. For the aluminum
compressed specimens can be calculated from sq a+2
alloy (Fig. 9), the data fit a straight line of con-
measured strains and plasticity equations. The b= = stant slope, but the medium-carbon steel (Fig.
s z 2a + 1
Levy-Mises equations relating plastic strain 10) shows bilinear behavior, with a larger slope
increments to stress for an isotropic material, in the small strain region (Ref 8).
For example, in homogeneous deformation,
expressed in cylindrical coordinates, are:
2, and becomes zero. Consequently, the
der d[r ( z)/2] (Eq 3) hoop stress becomes zero, and no fractures occur
in this case. At the other extreme, when 12,
de d[ (r z)/2] (Eq 4) becomes positive, and both and z are ten-
dez d[z ( r)/2] (Eq 5) sile. When 0, z 12 , which is the stress
state in plane strain. Figure 7 illustrates typical
The equivalent strain increment, d e, and changes in and z in a cylindrical compres-
equivalent stress,
, are given as: sion test. Note that, at large deformations, as the
bulge severity increases, z becomes tensile.
2
]
1/ 2
de = ( de r - d e z ) 2 + ( d e z - d e q ) + ( d e q - d e r ) 2
3 Fracture Limits
(Eq 6)
As indicated in the previous sections, a wide
1
[ ]
1/ 2
s= (s r - s z ) 2 + (s z - s q ) 2 + (s q - s r ) 2 range of stress and strain conditions can be gen-
2 erated at the free surfaces of cylindrical, tapered, Fig. 8 Fracture loci in cylindrical upset-test specimens
(Eq 7) or flanged test specimens. This range of condi- of two materials
Fig. 9 Fracture loci in cylindrical, tapered, and flanged

upset-test specimens of aluminum alloy 2024-
T351
Fig. 11 Comparison of strain paths and fracture locus lines
Conclusions 2. A.L. Hoffmanner, Metal Forming: Inter-

relation Between Theory and Practice,
A fracture criterion based on the linear rela- Plenum Press, 1971, p 349
tions between total surface strains at fracture 3. H.C. Rogers, in Ductility, American Society
has been presented. Plotting of the fracture- for Metals, 1968, p 31
strain line on a graph of tensile surface strain 4. K.E. Puttick, Philos. Mag., Vol 4 (Ser. 8),
(circumferential strain, e) versus compressive 1959, p 964
surface strain (axial strain, ez) yields a straight 5. P.F. Thomason, Int. J. Mech. Sci., Vol 11,
line with a slope of one-half, except in some 1969, p 189
exceptional cases where a bilinear relationship 6. H.A. Kuhn and P.W. Lee, Metall. Trans.,
exists at the low-strain region. In any case, the Vol 2, 1971, p 3197
fracture line represents the material limits to 7. P.W. Lee and H.A. Kuhn, Metall. Trans.,
Fig. 10 Fracture loci in cylindrical, tapered, and
Vol 4, 1974, p 969
flanged upset-test specimens of type 1045 deformation. The strain paths (relation between
cold-finished steel tensile and compressive strains during defor- 8. E. Erman and H.A. Kuhn, Novel Test
mation) are functions of the process parame- Specimens for Workability Measurement,
ters. Together, the fracture line and strain paths Proc. ASTM Conf. on Compression Testing,
give a graphical representation of the material ASTM, 35 March 1982
Workability Diagram 9. H.A. Kuhn, P.W. Lee, and T. Erturk, J. Eng.
and process factors in workability. Comparison
of the strain-path terminations required for Mater. Technol. (Trans. ASME), Vol 95H,
The results of Fig. 810 constitute workabil- complete formation of the part with the limit- 1973, p 213
ity diagrams for free-surface fracture in bulk ing fracture strains of the material indicates 10. A.T. Male and M.G. Cockroft, J. Inst. Met.,
forming operations. They embody, graphically, whether or not fracture is likely and suggests Vol 93, 1964, p 38
both the material influence in the fracture-strain- possible alterations of the material or process 11. H. Kudo and K. Aoi, J. Jpn. Soc. Techol.
limit line and the process influence through the to avoid fracture. Plast., Vol 8, 1967, p 17
variation of strain paths with die geometry, 12. P.F. Thomason, Int. J. Mech. Sci., Vol 11,
workpiece geometry, and friction. To illustrate 1969, p 187
this interrelationship, consider the bolt-heading ACKNOWLEDGMENT 13. J.J. Shah and H.A. Kuhn, Application of a
process in Fig. 11. If material A is used for the Forming Limit Concept to Upsetting and
product, strain path a crosses the fracture line on This chapter, originally published in Worka- Bolt Heading, Proc. ASTM Conf. on Com-
its way to its position in the final deformed bility and Testing Techniques, American Society pression Testing. ASTM, 35 March 1982
geometry, and cracking is likely. Two options for Metals, 1984, is revised from the 1984 text 14. H.P. Ganser, A.G. Atkins, O. Kolednik, F.D.
are open for avoiding defects: use material B, written by Peter W. Lee (deceased) and Howard Fischer, and O. Richard, Upsetting of
which has a higher forming-limit line; or alter Kuhn. Cylinders: A Comparison of Two Different
the strain path to follow b, which, in this case, is Damage Indicators, Trans. ASME, J. Eng.
accomplished through improved lubrication. REFERENCES Mater. Technol., Vol 123, 2001, p 9499
Applications of this workability concept in de- 15. A. Jenner and B. Dodd, Cold Upsetting and
formation process design are given in Chapter 1. G.E. Dieter, in Ductility, American Society Free Surface Ductility, J. Mech. Work.
12 for a wide variety of materials and processes. for Metals, 1968, p 1 Technol., Vol 5, 1981, p 3143
Chapter 6
Hot-Compression Testing
THE UTILITY OF THE COMPRESSION which sheet or thin plates are indented with be flat and parallel to within 0.0002 (ASTM
TEST in workability studies has been shown in long narrow platens E209).
previous chapters. A cylindrical specimen is rel- Lubrication at the interface between the
atively easy to prepare, and loading in compres- platen and the specimen is important in achiev-
sion simplifies the design of grips. Moreover, the Cylindrical Compression Test ing uniform deformation. However, at elevated
absence of necking, as occurs in a tension test, temperature, this can be difficult. Typical lubri-
means that a compression test can be conducted Testing Apparatus (Ref 2). Uniaxial com- cants are graphite sheet, water-base graphite,
to large strains that are comparable to those pression testing for workability analysis is usu- boron nitride solutions, molybdenum disulfide,
found in deformation processes such as forging ally carried out in a servohydraulic machine and low melting glasses. To retain the lubricant
or extrusion. However, as discussed in Chapter such as that shown schematically in Fig. 1. at the ends of the specimen, various types of
4, Bulk Workability Testing, barreling of the Because strain-rate effects are large in high-tem- grooved patterns are used. Application of the lu-
specimen due to friction at the platens compli- perature testing, it is important to use an appara- bricant to the specimen must be done carefully.
cates the determination of the true axial com- tus with high machine stiffness. The furnace that Usually this is done in two steps, allowing time
pression stress when the test is used to measure surrounds the specimen is usually of resistance, for drying between applications.
the stress-strain curve (flow properties) of the radiant-heating, or induction type. It is important Effect of Temperature. The effect of temper-
material. However, as Chapter 5, Cold Upset to protect the dynamic load cell and the hy- ature on the compressive flow curve for metals
Testing, describes, this feature of the compres- draulic actuator with cooling blocks. is shown in Fig. 2. Curves A and B are for de-
sion test can be used as a workability test by The top and bottom anvils are usually made of formation at cold working temperatures, T
measuring the propensity for surface cracking in stainless steel for tests below 1000 C (1832 F), 0.3Tm, where Tm is melting temperature. Curve
deformation processes. while nickel-base superalloys or Ti-Zr-Mo A, tested at a low strain rate, shows extensive
This chapter discusses the use of the compres- (TZM) alloy are used for higher temperatures. strain hardening, while curve B, tested at a high
sion test at elevated temperatures, a topic that is The diameter of the anvils should be at least strain rate, shows flow softening from a temper-
just touched on in previous chapters. Elevated three times that of the specimen. The compres- ature rise due to deformation heating. Curves C
temperature introduces the obvious experimental sion platens are made from tool steel, tungsten and D are illustrative of deformation at hot
issues of providing a uniform and controlled carbide, TZM, or ceramic composite, depending working temperatures, T > 0.6Tm. Curve C is
temperature, as well as dealing with more diffi- on the temperature. Compression platens should typical of metals that undergo dynamic recovery,
cult problems of lubrication of the platens.
Incremental loading, which can be used for pre-
cise flow-stress measurements at room tempera-
ture, becomes problematic in elevated-tempera-
ture testing. In addition, the behavior of metals at Crosshead
elevated temperature becomes more varied. Load data
Load cell
While most metals become softer as temperature acquisition
increases, and often more ductile, it is possible to Top anvil
encounter temperature regions of embrittlement.
Cooling
Also, because of thermomechanical processes Top platen Temperature
pipe (2)
such as dynamic recovery and recrystallization control and
Specimen
that take place at hot working temperatures (see data
Chapter 3, Evolution of Microstructure during Loading frame Thermocouple acquisition
Hot Working) plastic instabilities can develop in Furnace
the compression test. Finally, metals become Bottom platen
much more strain-rate sensitive at elevated tem- Displacement
perature (Ref 1). This makes strain-rate control Bottom anvil
control and
in testing very important. data
Actuator Cooling
Three forms of compression tests are dis- acquisition
cussed in this chapter: block (2)
The cylindrical compression test, or hot

upset test
The hot compression of thin ring specimens
The hot plane-strain compression test, in Fig. 1 Compression-testing system. Source: Ref 2
given by its strain-rate sensitivity, m. It is the in- testing since it expresses the position of the
crease in stress (s) needed to cause a certain in- platen in terms of strain rate. The distance of
A crease in plastic-strain rate (e) at a given level of crosshead travel is controlled by the voltage
plastic strain (e) and a given temperature (T ). input as a function of time. Constant strain rate
B up to about 100 s1 can be achieved (Ref 6, 7).
D log s Effect of Deformation Heating. Only about
m=
D log e e,T (Eq 1) 3 to 5% of the energy of plastic deformation
stays in the material as stored energy. Thus, most
of the energy that is required to cause plastic de-
Figure 4 shows how the dependence of flow
True stress
formation appears as deformation heating that is

stress on strain-rate sensitivity increases rapidly available to raise the temperature of the test
when the homologous temperature (T/Tm) ex- specimen. At slow rates of deformation, most of
ceeds a ratio of 0.6. Note that the symbol m is this heat escapes through the anvils and to the
also used below, in accordance with standard environment, but at high strain rates there is not
convention, to denote the interface friction fac- time for much heat flow and the specimen tem-
C tor. Usually the context of the situation will help perature rises. Since one aims to measure flow
identify which concept is meant by the symbol. curves at constant temperature and high strain
D
The true strain in a compression test is a func- rate, it is necessary to make a correction for de-
tion of the height of the specimen: formation heating.
An expression for the temperature increase
h Dh due to deformation heating, including an esti-
e = ln = ln1 -
True strain h0 h0 (Eq 2) mate of the heat generated by friction at the
platen/specimen interface, was developed by
Fig. 2 Typical flow curves for metals deformed at cold
Zhao (Ref 8):
working temperatures (A and B) and at hot In Eq 2, h and h0 are the instantaneous and initial
working temperatures (C and D). Source: Ref 3
specimen height, respectively. Dh h0 h is the
distance the crosshead of the testing machine has ha s a mva h t
while Curve D exhibits flow softening due to de- DT = ae + 1 - exp - T
formation heating and/or microstructural insta-
traveled since the compression test started. The hT ha crha (Eq 6)
negative sign obtained when evaluating Eq 2 de-
bilities such as the generation of a softer texture
notes that this is a compressive strain.
during deformation or dynamic spheroidization
The true strain rate in the compression test is where sa is the measured average stress, ha is the
of second phases. average height of the specimen, hT is the heat-
given by:
To accurately measure the temperature of the transfer coefficient between the platen and the
specimen, a thermocouple should be welded to de d (ln h / h0 ) 1 dh v specimen, m is the friction factor between the
the surface of the specimen or inserted in a small e = = = = platen and the specimen, t is the time, and va is
dt dt h dt h (Eq 3)
hole drilled into the specimen (Ref 3). In setting the average velocity at the platen/specimen in-
up the tests, it is important to establish the soak
where v is the crosshead velocity. Equation 3 terface. Also, a 0.95 is the fraction of defor-
time needed for the specimen to come to equi- mation energy converted to heat, c is the mate-
shows that for a constant crosshead velocity the
librium with the test temperature. The use of a
true strain rate will increase as the specimen rial specific heat, and r is its density.
heated guard ring box (Fig. 3) minimizes heat Knowing the temperature rise due to defor-
height decreases. Thus, to conduct a constant-
loss in inserting the compression specimen into mation heating, one can estimate the correction
strain-rate test the velocity of the moving platen
the testing machine. Other ideas for conducting
should decrease continuously, while maintaining that must be added to sa. First make a plot of sa
tests isothermally are given in Ref 4 and 5.
a constant (v/h) ratio. To see how this is done, Eq versus the corrected temperature (Ttest DT)
Effect of Strain Rate. The extent to which (Fig. 5). The slope of this plot gives ds/dT to be
3 is written as:
strain rate affects the flow stress of a material is used in Eq 7 to calculate the stress increase:
e et (Eq 4)
ds
Ds = DT e , e
Combining this with Eq 2 gives: dT (Eq 7)
Dh h0[1 exp (et)] (Eq 5) Finally, the isothermal stress-strain curve is ob-
tained by adding the stress increase from Eq 7 to
Equation 5 is the basic equation for machine the average true stress sa. Figure 6 shows the
control for constant strain rate in compression
Constant
Constant
True stress ()
d
dT
1
Corrected temperature, Ttest T

Fig. 4 Dependence of strain-rate effect on homologous
Fig. 3 Guard-ring box, with specimen in position. 1 in. temperature, T/Tm for specimens strained to Fig. 5 The determination of derivative of stress in terms
= 25.4 mm. Source: Ref 4 40% reduction. Source: Ref 4 of temperature. Source: Ref 2
Chapter 6: Hot-Compression Testing / 63
250
Corrected
Uncorrected
10.0 s1
200
1.0 s1
True stress, MPa
150
0.1 s1
100
0.01 s1
0.001 s1
50
0
0 0.2 0.4 0.6 0.8
True strain, mm/mm (in./in.)
Fig. 6 True-stress/true-strain curves in compression,

before and after temperature correction. Tests
are for HY-100 steel at 1000 C (1832 F) and various strain
rates. Note effect of strain rate on both level of flow curves
and the amount of temperature correction. Source: Ref 9
compression stress-strain curves for a high-

strength structural steel tested at 1000 C (1832 Fig. 7 Relationship between friction factor m and barreling coefficient B, for various values of specimen aspect ratio
r/h. Data are for T/Tm 0.8 and e 0.01 s1. Source: Ref 11
F). The corrected curves used the method de-
scribed previously. Note how the influence of
deformation heating on the flow curves is most factor is linearly related to the square root of B Plastic Instability in Compression. In tension
marked at higher strain rates. (Fig. 7). This figure shows that barreling is testing the onset of necking indicates unstable
Determination of the Flow Curve. The strongly dependent on the slenderness ratio, r/h, flow, characterized by a rapid decrease in diame-
basic data obtained from the compression test even for low values of interface friction. The ter localized to the neck region. It is this plastic
are the load and displacement (stroke). The true data in Fig. 7 can be used with Eq 9 to provide a instability that limits the use of the tension test for
stress is the load P divided by the instantaneous correction to the measured axial stress. determining the flow stress at strains much
cross-sectional area with diameter (D). If the Evans and Scharning (Ref 11) studied the sys- greater than about e 0.5. In the tension test,
compression deformation is homogeneous, then tematic errors in flow-stress determination in the necking occurs when the rate of strain hardening
this is given by: hot-compression test due to frictional forces at is no longer able to compensate for the decrease
the interfaces and deformation heating during in cross-sectional area of the specimen.
4P 4 Ph straining. They studied more than 3000 finite-el- In compression testing, a similar behavior oc-
sa = s0 = =
pD2 pD02 h0 (Eq 8) ement analyses using the six variables D/h, spec- curs when strain softening is prevalent. Under
imen volume, friction factor, homologous tem- normal conditions during compression, the
where sa is the average true axial stress, which perature, strain rate, and strain. As might be cross-sectional area of the specimen increases,
is equal to the flow stress s0 or effective stress s expected, the most important variables are fric- which increases the load-carrying capacity of
for homogeneous deformation. When the friction and specimen geometry. Strain and strain the specimen. However, when strain softening
tion at the specimen/platen interface is described rate are of intermediate importance, with speci- occurs its load-carrying capability is decreased.
by the interface friction factor m = 3 ti /s0, both men volume and temperature least important. A When the rate of decrease in the strength of the
the slab method of plasticity analysis and the general interpolation function was determined material due to strain softening exceeds the rate
upper-bound analysis (Ref 10) give the average so that the relative errors in stress can be calcu- of increase in the area of the specimen, an un-
axial flow stress as: lated for any of the values in the experimental stable mode of deformation occurs in which the
conditions. It is suggested that, with further specimen kinks as it is compressed (Fig. 8).
4P mD work for validation, these equations could be Instability in tension and compression can be
sa = = s 0 1 + used to correct the measured sa to the value of described with the aid of the Considre construc-
pD2 3 3h (Eq 9)
flow stress s0. tion (Fig. 9). Instability occurs when the slope of
Discussion so far about determining the the load-elongation (P-e) curve becomes zero:
Equation 9 shows how the measured flow stress stress-strain curve has focused on the determina-
is in excess of the effective or flow stress when tion of stress. The true strain in the hot-com-
friction, and barreling, occur in the test. dP d(sA) sdA Ads 0
pression test is found from Eq 2, making sure
Extensive finite-element modeling of the hot- that correction is made for the elastic deflection
compression test using realistic parameters for or
of the testing machine (Ref 12). Since barreling
hot-worked alloys showed that an observed bar- leads to nonuniform deformation, this raises the
reling coefficient can be used to determine the ds/s dA/A de de(1 e)
question of whether Eq 2 is suitable to express
friction factor for use in Eq 9 (Ref 11). The bar- the effective strain. Fortunately, it has been
reling coefficient is defined as: shown that the mean effective strain for a bar- from the relationship between true strain, e, and
reled specimen is the same, to a first approxima- engineering strain, e, e ln (1 e). Thus:
h(rmax )2
B= tion, as in the case of a compression specimen
h0 (r0 )2 (Eq 10) that did not barrel (Ref 13). This result provides ds/de s/(1 e) (Eq 11)
the theoretical justification for using the axial
where rmax is the maximum diameter of a speci- compressive strain as the effective strain in con- at the point of instability. Note that Fig. 9 has axes
men deformed to height h. The interface friction structing the stress-strain curve. of true stress versus engineering strain. Instability
conditions at the platen/ring interfaces. If the in-

terfacial friction were zero, the ring would de-
form in the same way as a solid disk, with each
element flowing radially outward at a rate pro-
portional to its distance from the center; that is,
the internal diameter would increase. In the case
of small but finite interfacial friction, the outside
diameter of the ring is smaller than for the zero
friction case. If friction exceeds a critical value,
it is energetically favorable for only part of the
ring to flow outward and for the remainder of the
ring to flow to the center, decreasing the inside
diameter. Thus, measurements of the inside di-
Fig. 8 Hot-compression test specimens of titanium
ameter of compressed rings provide a particu-
alloy Ti-10V-2Fe-3Al. Specimens (a), (b), and (c)
were tested at 704 C and (d), (e), and (f) at 816 C. Test larly sensitive means of studying interfacial fric-
strain rates were 103 s1 (a) and (d), 101 s1 (b) and (e), tion, because the inside diameter increases if the
and 10 s1 (c) and (f). All specimens had an equiaxed-b
microstructure before testing. When deformed at 704 C
friction is low and decreases if the friction is
(top row) the values of ac were 2.5, 5, 5 (left to right). At larger, as shown in Fig. 11.
816 C, the value of ac was less than zero at all strain The ring compression test is particularly at-
rates. Note the correlation with nonuniform flow and flow tractive for the measurement of friction in metal-
localization and ac. Source: Ref 14 Fig. 10 Formation of shear bands in compression
working in that no direct measurement of force
is required and no flow stress values for the de-
d ln s e ds forming material are needed.
m= = Analytical Basis. Male and Cockcroft (Ref
C d ln e s de (Eq 13)
17, 18) showed the potential of the ring com-
pression test, and Avitzur provided an analysis
Tension determined at fixed e and T. Nonuniform plastic
for the problem of the axial compression of flat,
deformation in compression occurs at strain,
ring-shape specimens between flat dies (Ref 19,
strain rate, and temperature conditions when the
20). The upper-bound solution developed by
e parameter ac 5 (Ref 15, 16).
Avitzur is quite complex and is usually pre-
sented as curves of percent change in inside di-
g -1 ameter versus percent reduction in the thickness
ac =
m (Eq 14) of the ring (Fig. 12). Different calibration curves
Compression are found for ring specimens of different ratios
C
The conditions that promote flow softening of outside diameter to inside diameter to thick-
include dynamic recovery and dynamic recrys- ness. The most-used configurations are when
1 tallization during hot working. Another source is these dimensions are in the ratio 6 to 3 to 2 or 6
the rapid spheroidization of pearlite and other to 3 to 1. An example of the use of the test is
lamellar microstructures and the coarsening of given in Ref 21.
Fig. 9 Considres construction showing point of insta-
precipitates during dynamic hot deformation. From the Avitzur analysis it is possible to cal-
bility in tension testing (due to decreasing strain-
hardening rate) and in compression testing (due to strain Much more detail on flow-softening mecha- culate values of pa/s0 at the instant when defor-
softening) nisms can be found in Chapter 3, Evolution of mation stops in terms of the ring geometry and
occurs when the slope of the curve equals the ratio Microstructure during Hot Working. the interfacial friction factor, m. In these equa-
of the true stress to (1 e). For the tension curve, tions neither the basic yield stress s0 nor the in-
necking occurs when the stress-strain curve terfacial shear stress, ti, appears as independent
Ring Compression Test value, but only as the ratio m:
reaches point C. This defines the ultimate strength
of the material in tension. For the compression
curve, Fig. 9 shows that for a strain-softening ma- Background. The ring compression test was ti t
developed to provide a measurement of the in- m = 3m = 3 3 i
terial unstable flow occurs at point C. The ulti- pa s0 (Eq 15)
mate strength of the material for this case of work terface friction between the specimen and the
softening is defined by the stress at point C. platens, but it also can be used to provide rea-
Another manifestation of strain softening is sonable values for the flow stress in compres- where m is the friction factor, m is the Coulomb
the formation of internal shear bands. Figure 10 sion. When a flat, ring-shaped specimen is com- coefficient of friction, ti is the interfacial shear
shows schematically how these form upon in- pressed in the axial direction, the change in stress, pa is the average axial pressure, and s0 is
creasing compressive deformation. See Chapter dimensions depends on the amount of compres- the axial flow stress. The assumption in this
13, Workability in Forging, Fig. 20 and 22 for sion in the thickness direction and the frictional analysis is that this ratio remains constant for the
actual microstructures. The two parameters that
predict the propensity for shear banding are the
normalized flow-softening rate, g, and the strain-
rate sensitivity index, m.
1 ds
g=
s de (Eq 12)
where s(e) is the flow curve determined under Fig. 11 Variation in shape of ring test specimens deformed the same amount under different frictional conditions. Left
to right: undeformed specimen; deformed 50%, low friction; deformed 50%, medium friction; deformed 50%,
constant e. high friction
Detailed comparisons of the results of many in- The true stress and true strain are determined
vestigators who have used the hot ring compres- from the plane-strain compression test by:
sion test are given in Ref 25.
P
s1 = pa =
Plane-Strain Compression Test wb
and
Testing Conditions. The fundamental diffi-
culties associated with high friction at the
platen/specimen interface and barreling of the h0
e1 = e pc = ln
free surface can be minimized to a large degree h
in the plane-strain compression test (Ref 26).
Moreover, as Fig. 13 shows, this test is ideally where the dimensions are indicated in Fig. 13
suited for testing sheet or thin plate. In the plane- and the one-direction is taken parallel to the
strain compression test, a metal sheet is com- platen motion. However, because the stress and
pressed across the width of the sheet by narrow strain state in the test is that of plane strain, the
platens that are wider than the strip. The elastic mean pressure on the platens, pa, is 15.5%
constraints of the undeformed shoulders of the higher than it would be in a uniaxial compres-
material on each side of the platens prevent ex- sion test. To use these values to construct a uni-
tension of the strip in its width dimension, hence axial compression flow curve (s- versus e- ), they
the name plane strain. There is deformation in must be converted to effective stress and strain
the direction of the platen movement and in the (see Chapter 2, Bulk Workability of Metals).
Fig. 12 A typical calibration curve for the ring com- direction normal to the length of the platen.
pression test. Change in internal ring diameter To ensure that plane-strain deformation is
versus change in specimen height, for a 6:3:2 ring. Source: 3 pa
achieved, the width of the strip should be at least s = s0 = pa =
Ref 21 2 1.155 (Eq 16)
6 to 10 times the breadth of the platens. To ensure
that deformation under the platens is essentially
and
given material and deformation conditions. If homogeneous, the ratio of platen width to strip
the analysis is carried out for a small increment thickness (w/h) at any instant in the test should be
of deformation, s0 and ti can be assumed to be between 2 and 4. When the plane-strain compres- e=
2
e pc = 1.155e pc
approximately constant for this increment, and sion test is done at room temperature, it is often 3 (Eq 17)
the solution is valid. On the assumption that the carried out incrementally so as to provide replen-
friction factor is constant for the entire deforma- ishment of lubrication and to change the platens so Note that Eq 16 assumes that friction between
tion, it is justifiable to continue the analysis in a as to maintain the proper w/h ratio. In this way it the platen and the strip is negligible. This can be
series of small deformation increments using the is possible to achieve true strains of around 2. achieved by incrementally applying lubrication
final ring geometry from one increment as the during the test or by using a very low friction
initial geometry for the subsequent increment. material such as polytetrafluoroethylene tape on
Moreover, using this procedure strain hardening the faces of the platens. Friction becomes a
can be accommodated, although a high strain- much more serious consideration in the hot
rate sensitivity may not be (Ref 22). plane-strain compression test, discussed next.
The analysis of Avitzur was modified to de- Hot Plane-Strain Compression Test. The
velop a reliable method for treating results from plane-strain compression (PSC) test is finding
the ring compression test with bulge formation growing use for making reliable and repro-
and varying friction factor (Ref 23). It was ducible measurements of flow curves at elevated
shown how the ring compression test can be temperature. It is interesting to note that steps
used for valid measurement of flow stress. A are underway to develop a good practice guide
ring specimen is deformed to a fixed h under the for the test through the offices of the National
required conditions of temperature and strain Physical Laboratory (United Kingdom). A nice
rate, and the deformation load is recorded. The feature of the test is that since a reasonably large
change in shape of the ring is measured at room specimen can be tested, it provides a good op-
temperature and the value of the ratio pa/s0 is portunity for studying microstructure develop-
obtained from the computer solutions to ment. Incremental tests are difficult to do in a
Avitzurs equations. Reference 23 gives the hot PSC test, and since lubrication also is more
equations, which are detailed and require a com- difficult at elevated temperature, it is not sur-
puter program for explicit solution of the ratio prising that load-displacement data for hot PSC
pa/s0. Measurement of the area of the ring sur- tests require corrections for the friction between
face formerly in contact with the platens, to- the platens and the specimen, and for lateral
gether with knowledge of the deformation load, spread (Ref 27). Figure 13(b) defines the dimen-
allows calculation of pa and s0 is obtained. True sions of interest. The dimension of concern is
strain is determined from Eq 2. Repeating this the breadth (width) b. Lateral spread is in the
process with other ring specimens over a range width direction and is evidenced by a bulge at
of deformations allows the generation of a com- the free surfaces. Thus, the stress state is not
plete stress-strain curve for the material under truly plane strain, and a correction needs to be
the particular conditions of temperature and made to the measured pa.
strain rate. Calibration curves similar to Fig.12 Fig. 13 Plane-strain compression (PSC) test. (a) Di-
The development of lateral spread means that
mensions before deformation. (b) Dimensions
have also been established by finite-element after deformation. Note bulge that is prominent in the hot the PSC test is not being carried out under truly
modeling of hot ring compression tests (Ref 24). PSC test. Source: Ref 27 plane-strain conditions. Therefore, corrections
for spread need to be applied to pa and epc to give pa w w2 2

= 1+ - (b - w ) + w
values close to the true axial values. Detailed fi- 2k 4h 12 hb (Eq 22) 3
nite-element analysis of the hot PSC test was f =
b (Eq 25)
carried out to find the effects of the test variables Equations 21 and 22 are only valid when slid-
on spread (Ref 27). A thermomechanically cou- ing or sticking conditions apply over the entire where b is the instantaneous breadth obtained
pled model that allowed for deformation heating face of the platens. However, it is very likely to from Eq 18. The width of the platen, w, does not
and modeled the interfacial friction using have an intermediate situation where sticking change appreciably with deformation.
Coulomb friction was used. The chief findings exists in the central region of the platens and
were: sliding friction conditions exist at the outer
2
For b0 /w 5 the spread increases almost edges. The location of this transition, measured e = f ln h0 / h
exponentially with decreasing instanta- from the centerline of the platen, is given by zo: 3 (Eq 26)
neous thickness. However, raising initial
width to large values incurs a strong penalty h 1 Thus, the flow stress determined with the hot
z0 = ln
on the load required for deforming the spec- 2m 2m (Eq 23) plane-strain compression test is obtained by
- -
imen. plotting s from Eq 24 versus e obtained from Eq
In addition, lateral spread decreases with in- The 12-term equation for this situation is not 26. When the corrections discussed previously
creasing friction coefficient and decreasing given here. It should be noted that Eq 21 and 22 are made to the data, tests between different lab-
initial specimen thickness. contain extra terms over the usual equations for oratories agree to 5% (Ref 30).
There is a strong dependence of side curva- yielding in plane-strain compression because they The plane-strain compression test is useful
ture on friction coefficient. have been corrected with the Orowan and Pascoe for measuring the workability of metals that are
The spread is not sensitive to the strain rate equations (Ref 29) that account for spread in sensitive to shear. The deformation between the
or to the material. rolling when b 6h. This correction cannot be platens includes crossed shear bands from each
ignored for the hot PSC test because it is common corner of the platens. This intense shear can
An empirical relationship was established to
to use specimens with b 5h (Ref 28). cause unstable deformation of hot deformed
evaluate instantaneous breadth b as a function of
The equations that relate frictional conditions material in much the same way as it occurs in
instantaneous thickness, h.
to the flow stress, although complex, can be han- the cylindrical compression tests referred to
h
s
dled rather simply with modern computers. The previously.
b = b0 1 + C C value of z0 is computed for each increment of
h0 (Eq 18)
deformation by using the instantaneous thick-
ness h. This determines which of the friction Conclusions
where s is the spread exponent, with a typical equations prevail, and this allows a calculation
value of 0.2, and C is defined as the spread co- of pa/2k. With pa known from Eq 20, the flow The compression test is the most general-pur-
efficient determined from the final breadth bf stress in plane strain 2k is readily determined. pose test for workability. Carrying out the test at
and thickness hf of a specimen, according to: The final step is to convert the plane-strain flow elevated temperature imposes issues with fric-
stress into the equivalent axial flow stress: tion and temperature and strain-rate control that
bf / b0 - 1 must be addressed with proper test equipment
C=
1 - (hf / h0 ) s (Eq 19) 3 1 and experimental planning. In addition, the ex-
s = s 0 = 2k
2 f (Eq 24) perimenter must be cognizant of the forms of
The flow stress in plane strain is by convention plastic instability that can arise in hot-compres-
denoted as 2k, where k is the flow stress in pure The term f is a correction to allow for the fact sion tests. The ring compression test provides
shear (torsion). The average pressure on the that because of bulging the stress state in the hot for the best routine measurement of friction in
platens, pa, will be greater than 2k, depending on PSC test is not truly plane strain. This is dis- hot deformation processing. However, it is an
friction conditions. The instantaneous value of cussed below. average value and cannot detect variations in
average pressure is given by: The displacement measurements Dh from friction over an area. Hot plane-strain compres-
which strain is calculated must be corrected for sion offers distinct advantages when the objec-
errors in the zero position of displacement. tive is to provide flow curves for modeling
pa P/wb (Eq 20)
These occur due to slight misalignment of the plane-strain deformation processes such as
tool and/or specimen faces not being perfectly rolling (Ref 31).
where P is the load on the platens and b is given
by Eq 18. There are three types of friction con- parallel (Ref 28). Also, zero displacement error
ditions that need to be considered, sliding fric- can occur when a thick glass lubricant film is
squeezed out from the platens during initial ACKNOWLEDGMENT
tion, sticking friction, and a combination of the
two. stages of deformation.
The material in this article is largely taken
When the Coulomb friction is low, the aver- Finally, the axial strain measurement should
from:
age pressure relative to the plane-strain flow be corrected with the f factor to determine the
stress (Ref 28) is given by: nominal equivalent strain. To convert the thick- H.A. Kuhn, Uniaxial Compression Testing,
ness strain by the factor 2/3, as in Eq 17, would Mechanical Testing and Evaluation, Vol 8,
be an overestimate because only part of the ASM Handbook, ASM International, 2000,
pa 1 2 h 2 (b - w )h mw 2 h
= + exp -1 - metal under the platen is in plane strain. This p 143151
2 k bw m 2 m h mb correction is based on the concept that at the A.T. Male and G.E. Dieter, Hot
(Eq 21) ends of the platens at the free surface the mate- Compression Testing, Workability Test
rial is almost free to move in both the length and Techniques, American Society for Metals,
where m is the friction coefficient. For sticking width directions. Therefore, the deformation in 1984, p 5072
friction, relative motion between the metal and these narrow regions is more like axisymmetric D. Zhao, Testing for Deformation
the platens occurs by shearing of the metal at the compression. Thus, a plane-strain condition ex- Modeling, Mechanical Testing and
interface, rather than sliding. Under these condi- ists only in the central (b-w) region of the speci- Evaluation, Vol 8, ASM Handbook, ASM
tions, pa/2k is given by: men (Ref 30). International, 2000, p 798810
REFERENCES Machines and Strain Sensors, Mechanical 23. V. DePierre and F.J. Gurney, A Method for
Testing and Evaluation, Vol 8, ASM Hand- Determination of Constant and Varying
1. S.L. Semiatin and J. Jonas, Formability and book, ASM International, 2000, p 7992 Factors During Ring Compression Tests, J.
Workability of Metals, American Society for 13. F.-K. Chen and C.-J. Chen, On the Lubr. Technol. (Trans. ASME), Vol 96, 1974,
Metals, 1984 Nonuniform Deformation of the Cylinder p 482488
2. D. Zhao, Testing for Deformation Model- Compression Test, J. Eng. Mater. Technol. 24. N.T. Rudkins, P. Hartley, I. Pillinger, and D.
ing, Mechanical Testing and Evaluation, (Trans. ASME), Vol 122, 2000, p 192 Petty, Friction Modelling and Experimental
Vol 8, ASM Handbook, ASM International, 197 Observations of Hot Ring Compression
2000, p 798810 14. S.L. Semiatin, Workability in Forging, Tests, J. Mater. Process. Technol., Vol 60,
3. D. Zhao and S. Lampman, Hot Tension Chapter 13, Workability and Process 1995, p 349353
and Compression Testing, Mechanical Design, ASM International, 2003 25. K.P. Rao and K. Sivaram, A Review of
Testing and Evaluation, Vol 8, ASM 15. J.J. Jonas, R.A. Holt, and C.E. Coleman, Ring-Compression Testing and Applic-
Handbook, ASM International, 2000, p Plastic Stability in Tension and Compres- ability of the Calibration Curves, J. Met.
152163 sion, Acta Metall., Vol 24, 1976, p 911 Process. Technol., Vol 37, 1993, p
4. J.F. Alder and V.A. Phillips, J. Inst. Met., Vol 16. S.L. Semiatin and G.D. Lahoti, The 295318
83, 19541955, p 80 Occurrence of Shear Bands in Isothermal 26. A.B. Watts and H. Ford, An Experimental
5. F.J. Gurney and D.J. Abson, Met. Mater., Vol Hot Forging, Metall. Trans. A, Vol 13A, Investigation of the Yielding of Strip be-
7, 1973, p 535 1982, p 275 tween Smooth Dies, Proc. Inst. Mech. Eng.,
6. G. Fitzsimmons, H.A. Kuhn, and R. 17. A.T. Male and M.G. Cockcroft, A Method Vol B1, 1952, p 448453
Venkateshwar, J. Met., May 1981, p 11 for the Determination of the Coefficient of 27. M.S. Mirza and C.M. Sellars, Modelling the
7. High Strain Rate Tension and Compression Friction of Metals under Conditions of Bulk Hot Plane Strain Compression Test-Effect
Tests, Mechanical Testing and Evaluation, Plastic Deformation, J. Inst Met., Vol 93, of Friction and Specimen Geometry on
Vol 8, ASM Handbook, ASM International, 19641965, p 3846 Spread, Mater. Sci. Technol., Vol 17, 2001, p
2000, p 429446 18. A.T. Male, Variations in Friction Coeffi- 11421148
8. D. Zhao, Temperature Correction in Com- cients of Metals During Compressive 28. N.J. Silk and M.R. van der Winden, Interpre-
pression Tests, J. Mater. Process. Technol., Deformation, J. Inst. Met., Vol 94, 1966, p tation of Hot Plane Strain Compression
Vol 36, 1993, p 467471 121125 Testing of Aluminum Specimens, Met. Sci.
9. M. Thirukkonda, D. Zhao, and A.T. Male, 19. B. Avitzur, Metal Forming Processes and Technol., Vol 15, 1999, p 295300
Materials Modeling Effort for HY-100 Analysis, McGraw-Hill, 1968, p 8193 29. E. Orowan and K.J. Pascoe, Special Report
Steel, Technical Report TR 96-027, 20. B. Avitzur and C.J. Van Tyne, Ring No. 34, Iron and Steel Institute, 1946
NCEMT, National Center for Excellence in Forming: An Upper Bound Approach, J. 30. H. Shi, A.J. McLaren, C.M. Sellars, R.
Metalworking Technologies, Johnstown, Eng. Ind. (Trans. ASME), Vol 104, 1982, p Shahani, and R. Bolingbroke, Hot Plane
PA, March 1996 231252 Strain Compression Testing of Aluminum
10. E.M. Mielnik, Metalworking Science and 21. F. Wang and J.G. Lenard, An Experimental Alloys, J. Test. Eval., Vol 25 (No. 1), 1997,
Engineering, McGraw-Hill, 1991, p 241243 Study of Interfacial FrictionHot Ring p 6173
11. R.W. Evans and P.J. Scharning, Axisym- Compression, J. Eng. Mater. Technol., Vol 31. S.B. Davenport, N.J. Silk, C.N. Sparks, and
metric Compression Test and Hot Working 114, 1992, p 1318 C.M. Sellars, Development of Constitutive
Properties of Alloys, Mater. Sci. Technol., 22. G. Garmong, N.E. Paton, J.C. Chestnutt, Equations for Modelling of Hot Rolling,
Vol 17, 2001, p 9951004 and L.F. Nevarez, Metall. Trans. A, Vol 8A, Mater. Sci. Technol., Vol 16, 2000, p
12. J.W. House and P.P. Gillis, Testing 1977, p 2026 539546
Chapter 7
Hot-Tension Testing
P.D. Nicolaou, R.E. Bailey, and S.L. Semiatin
THE DEVELOPMENT of successful manu- tensile loading at the strain rates to be imposed heating techniques are direct-resistance heating
facturing techniques for metallic materials by the specific hot-working operation. (in the case of Gleeble systems) and indirect-
requires reliable information regarding hot- The ideal hot-workability test is one in which resistance or induction heating with conventional
working characteristics. The proper hot-working the metal is deformed uniformly, without insta- load frames. In some cases, universal test ma-
temperature and deformation rate must be estab- bility, at constant true strain rate under well- chines may include a special chamber for testing
lished to produce high-quality wrought products controlled temperature conditions with con- in either vacuum or controlled atmosphere.
of complicated geometries. It is also important tinuous measurement of stress, strain, and Specimen (testpiece) temperatures typically are
that product yield losses (from either grinding to temperature during deformation followed by in- monitored and controlled by thermocouples,
remove surface cracks or excessive cropping to stantaneous quenching to room temperature. which may be attached on the specimen surface
remove end splits) be held to a minimum, while Two types of hot-tension tests are discussed in or located very close to the specimen. In some
avoiding the formation of internal cavities this chapter: the Gleeble test and the conven- cases, temperature is measured by optical or in-
(pores). Severe cracking is ordinarily the result tional isothermal hot-tension test. The major ad- frared pyrometers. Accurate measurement and
of high surface tensile stresses introduced when vantage of the hot-tension test is that its control is very critical for obtaining reliable data.
hot working is conducted either above or below stress/strain state simulates the conditions that To this end, the use of closed-loop temperature
the temperature range of satisfactory ductility. promote cracking in most industrial metalwork- controllers is indispensable.
Similarly, cavitation is associated with internal ing operations. However, even though the ten- The occurrence of deformation heating may
tensile stresses, which, for a given material, de- sion test is simple in nature, it may provide mis- also be an important consideration, especially at
pend on the temperature, the deformation rate, leading information if not properly designed. high strain rates, because it can significantly
and workpiece/die geometry. Specifically, parameters such as the specimen raise the specimen temperature.
The first and most important step in specify- geometry, tension machine characteristics, and
ing appropriate hot-working practice is to strain rate and temperature control all influence Gleeble Testing Equipment
determine suitable hot-working conditions. In the results of the tension test. Therefore, the ten-
particular, the tensile ductility (e.g., fracture sion test should be designed and conducted care- The Gleeble system (Ref 5) has been used
strain), the flow stress, and cavity formation fully, and testing procedures should be well doc- since the 1950s to investigate the hot-tension be-
conditions should be established as a function of umented when data are reported. havior of materials and thus to generate impor-
temperature and strain rate. A curve of ductility tant information for the selection of hot-working
versus temperature or strain rate shows what de- parameters. A Gleeble unit is a high-strain-rate,
gree of deformation the material can tolerate Equipment and Testing Procedures high-temperature testing machine where a solid,
without failure. On the other hand, a plot of the buttonhead specimen is held horizontally by
flow stress versus temperature, along with work- The apparatus used to conduct hot-tension water-cooled grips, through which electric
piece size and strain rate, indicates the force lev- tests comprises a mechanical loading system and power is introduced to resistance heat the test
els required of the hot-working equipment. Last, equipment for sample heating. A variety of specimen (Fig. 1). Specimen temperature is
a curve of cavity volume fraction versus strain, equipment types are used for applying forces monitored by a thermocouple welded to the
strain rate, and temperature shows what process- (loads) to test specimens. These types range specimen surface at the middle of its length. The
ing parameters should be selected in order to from very simple devices to complex systems thermocouple, with a function generator, con-
produce high-quality products. that are controlled by a digital computer. The trols the heat fed into the specimen according to
Although commercial metalworking opera- most common test configurations utilize univer- a programmed cycle. Therefore, a specimen can
tions cannot be analyzed in terms of a simple sal testing machines, which have the capability be tested under time and temperature conditions
stress state, workpiece failures are caused by lo- to test material in tension, compression, or bend- that simulate hot-working sequences.
calized tensile stresses in most instances (Ref ing. The word universal refers to the variety of Contemporary Gleeble systems (e.g., see
14). In rolling of plate, for example, edge stress states that can be applied by the machine, www.gleeble.com) are fully integrated servohy-
cracking is caused by tensile stresses that form in contrast to other conventional test machines draulic setups that are capable of applying as
at bulged (unrestrained) edges (Ref 1, 3, 4). The that may be limited to either tensile loading or much as 90 kN (10 tons) of force in tension at
geometry of these unrestrained surfaces affects compressive loading, but not both. Universal test displacement rates up to 2000 mm/s (80 in./s).
the magnitude of tensile stresses at these loca- machines or tension-only test frames may apply Different load cells allow static-load measure-
tions. Moreover, tensile stresses are also created loads by a gear (screw)-driven mechanism or hy- ment to be tailored to the specific application.
on the unrestrained surface of a round billet draulic mechanisms, as discussed in more detail Control modes that are available include
being deformed with open dies. Therefore, to in the section Frame-Furnace Tension-Testing displacement, force, true stress, true strain, engi-
obtain a practical understanding of how well a Equipment in this chapter. neering stress, and engineering strain.
material will hot work during primary process- The heating method used for hot-tension test- The direct-resistance heating system of the
ing, it is essential to know how it will respond to ing varies with the application. The most common Gleeble machine can heat specimens at rates of
Chapter 7: Hot-Tension Testing / 69
Fig. 1 Gleeble test unit used for hot-tension and hot-compression testing. (a) Specimen in grips showing attached thermocouple wires and linear variable differential transformer
(LVDT) for measuring strain. (b) Close-up of a test specimen. Courtesy of Duffers Scientific, Inc.
reduction of area is diminished. On the other Strain rate is another important variable in the
hand, a long sample length generally produces hot-tension test. However, strain rate varies
higher apparent ductility/elongation values. For during hot-tension testing under constant-
example, Smith, et al. (Ref 6) have shown that a crosshead-speed conditions and must be taken
grip separation of 36.8 mm (1.45 in.) produces a into account when interpreting test data. An
hot zone about 12.7 mm (0.5 in.) long. When analysis of the strain-rate variation during the
specimen diameter is increased, as is necessary hot-tension test and how it correlates to the
in testing of extremely coarse-grain materials, strain rates in actual metalworking operations is
the grip separation should also be increased pro- presented later in this chapter.
portionately to maintain a constant ratio of hot- The load may be applied at any desired time
zone length to specimen diameter. in the thermal cycle. Temperature, load, and
Fig. 2 Typical specimen used for Gleeble testing
Test Procedures. It is essential that hot- crosshead displacement are measured versus
tension tests be conducted at accurately con- time and captured by the data acquisition sys-
trolled temperatures because of the usually tem. From these measurements, standard me-
up to 10,000 C/s (18,000 F/s). Grips with high strong dependence of tensile ductility on this chanical properties such as yield and ultimate
thermal conductivity (e.g., copper) hold the process variable. To this end, temperature is tensile strength can be determined. The reduc-
specimen, thus making the system capable of monitored by a thermocouple percussion welded tion of area at failure is also readily established
high cooling rates as well. Thermocouples or py- to the specimen surface. Using a function gener- from tested samples.
rometers provide signals for accurate feedback ator, heat input to the specimen is controlled ac- If hot-working practices are to be determined
control of specimen temperature. Because of the cording to a predetermined programmed cycle for an alloy for which little or no hot-working in-
unique high-speed heating method, Gleeble sys- chosen by the investigator. However, the temper- formation is available, the preliminary test pro-
tems typically can run hot-tension tests several ature measured from this thermocouple junction cedure usually comprises the measurement of
times faster than conventional systems based on does not coincide exactly with the specimen data on heating. In such tests, samples are
indirect-resistance (furnace) heating methods. temperature because (a) heat is conducted away heated directly to the test temperature, held for 1
A digital-control system provides all the sig- from the junction by the thermocouple wires, to 10 min, and then pulled to fracture at a strain
nals necessary to control thermal and mechani- and (b) the junction resides above the specimen rate approximating the rate calculated for the
cal test variables simultaneously through the surface and radiates heat at a rate higher than metalworking operation of interest. The reduc-
digital closed-loop thermal and mechanical that of the specimen itself. Consequently, the tion of area for each specimen is plotted as a
servo systems. The Gleeble machine can be op- thermocouple junction is slightly colder than the function of test temperature; the resulting on-
erated totally by computer, by manual control, or test specimen. Furthermore, specimen tem- heating curve will indicate the most suitable
by any combination of computer and manual perature is highest midway between the grips temperature range to be evaluated to determine
control needed to provide maximum versatility and decreases toward the grips. In general, the the optimal preheat* temperature. This tempera-
in materials testing. specimen will fracture in the hottest plane per- ture, as indicated from the plot in Fig. 3, lies
Sample Design. A typical specimen configu- pendicular to the specimen axis. Therefore, it is between the peak-ductility (PDT) and zero-
ration used in Gleeble testing is shown in Fig. 2. important to place the thermocouple junction ductility (ZDT) temperatures.
This solid buttonhead specimen, with an overall midway between the grips in order that the To confirm the appropriate selection of pre-
length of 88.9 mm (3.5 in.), has an unreduced hottest zone of the specimen, which will be the heat temperature, specimen blanks should be
test-specimen diameter of 6.25 mm (0.25 in.). zone of fracture, is monitored. The longitudinal heat treated at the proposed preheat temperature
The length of the sample between the grips at thermal gradient does not present a serious prob-
the beginning of the test is also an important lem because the specimen deforms in the local-
consideration. Generally, this length is 25.4 mm ized region where the temperature is monitored. *In the context of this chapter, preheat temperature is the tem-
perature at which the test specimen or workpiece is held prior
(1 in.). Shorter lengths produce a narrow hot Consequently, the measured values of reduction to deformation at lower temperatures. In actual metalworking
zone and restrict hot deformation to a smaller, of area and ultimate tensile strength represent operations, preheat temperature usually refers to the actual
constrained region; consequently, the apparent the zone where the thermocouple is attached. furnace temperature.
geometry depend on the specific features of the

frame and the heating unit as well as the testing
conditions. Temperature is measured by thermo-
couples attached on or located very near to the
specimen. In some cases, a pyrometer can also
be used.
The most common methods of heating in-
clude induction heating and indirect-resistance
heating in chamber. Typical examples are shown
in Fig. 5. Induction heating (Fig. 5a) usually al-
lows faster heating rates than indirect heating
does, but accurate temperature control requires
extra care. Induction-heating systems can reach
testing temperatures within seconds. Induction
heating heats up the outer layer of the specimen
first. Furnaces with a lower frequency have bet-
ter penetration capability. Coupling the heating
coil and the specimen also plays an important
role in heating efficiency. The interior of the
specimen is heated through conduction. With the
rapid heating rate, the temperature is often over-
shot and nonuniform heating often occurs.
Fig. 3 Hypothetical on-heating Gleeble curve of specimen reduction of area as a function of test temperature Indirect-resistance heating may provide better
temperature control/monitoring than induction
heating can. Indirect-resistance heating can be
for a time period equal to that of a furnace soak driven mechanism or servohydraulic actuator. combined readily with specially designed cham-
commensurate with the intended workpiece size Screw-driven (or gear-driven) machines are typ- bers for testing either in vacuum or in a con-
and hot-working operation. These specimens ically electromechanical devices that use a large trolled atmosphere (e.g., argon, nitrogen, etc.).
should be water quenched to eliminate any actuator screw threaded through a moving Vacuum furnaces are expensive and have high
structural changes that could result from slow crosshead. The screws can turn in either direc- maintenance costs. The furnace has to be
cooling. Subsequently, the specimens should be tion, and their rotation moves a crosshead that mounted on the machine permanently, making it
tested by heating to the proposed furnace tem- applies a load to the specimen. A simple balance inconvenient if another type of heating device is
perature, holding at this preheat temperature for system is used to measure the magnitude of the to be used. The heating element is expensive and
a moderate period of time (1 to 10 min) to redis- force applied. oxidizes easily. The furnace can only be opened
solve any phases that may have precipitated, Loads may also be applied using the pressure at relatively lower temperatures to avoid oxida-
cooling to various temperatures at intervals of 25 of oil pumped into a hydraulic piston. In this tion. Quenching has to be performed with an
or 50 C (45 or 90 F) below the preheat tem- case, the oil pressure provides a simple means of inert gas, such as helium.
perature, holding for a few seconds at the measuring the force applied. Closed-loop servo- Environmental chambers (Fig. 5b), which are
desired test temperature, and finally pulling in hydraulic testing machines form the basis for the less expensive than vacuum chambers, have a
tension to fracture at the calculated strain rate. most advanced test systems in use today. circulation system to maintain uniform tempera-
These on-cooling data demonstrate how the Integrated electronic circuitry has increased the ture inside the furnace. Inert gas can flow
material will behave after being preheated at a sophistication of these systems. Also, digital through the chamber to keep the specimen from
higher temperature. Testing on cooling is nec- computer control and monitoring of such test oxidizing. Temperature inside the chamber can
essary because the relatively short hold times systems have steadily developed since their in- be kept to close tolerance (e.g., about 1 C, or
during testing on heating may not develop a troduction around 1965. Servohydraulic test ma- 2 F). However, the maximum temperature of
grain size representative of that hold tempera- chines offer a wider range of crosshead speeds an environmental chamber is usually 550 C
ture and may be insufficient to dissolve or pre- of force ranges with the ability to provide eco- (1000 F), while that of a vacuum furnace can be
cipitate a phase that may occur during an actual nomically forces of 4450 kN (106 lbf ) or more. as high as 2500 C (4500 F). The chamber can
furnace soak prior to hot working. Also, most in- Screw-driven machines are limited in their abil- either be mounted on the machine or rolled in
dustrial hot metalworking operations are con- ity to provide high forces due to problems asso- and out on a cart. Split-furnace designs (Fig. 5c)
ducted as workpiece temperature is decreasing. ciated with low machine stiffness and large and are also cost effective and easy to use. When not
The on-cooling data will indicate how closely expensive loading screws, which become in- in use, it can be swung to the side. The split fur-
the ZDT can be approached before hot ductility creasingly more difficult to produce as the force nace shown in Fig. 5(c) has only one heating
is seriously or permanently impaired. In addi- rating goes up. zone. More sophisticated split furnaces have
tion, if deformation heating (Ref 7) during For either a screw-driven or servohydraulic three heating zones for better temperature con-
onheating tests has resulted in a marked un- machine, the hot-tension test system is a load trol. Heating rate is also programmable. When
derestimation of the maximum preheat tempera- frame with a heating system attached. A typical furnace heating is used, it is a common practice
ture, this will be revealed and can be rectified by servohydraulic universal testing machine with a to use a low heating rate. In addition, the speci-
examination of on-cooling data. high-temperature chamber is shown in Fig. 4. men is typically soaked at the test temperature
The system is the same as that used at room tem- for about 10 to 30 min prior to the application of
perature, except for the high-temperature capa- the load.
Frame-Furnace Tension-Testing bilities, including the furnace, cooling system, The mechanical and thermal control systems
Equipment grips, and extensometer. In this system, the grips are similar to those described in the previous
are inside the chamber but partly protected by section on the Gleeble testing apparatus. The
Universal testing machines and tension-test refractory from heating elements. Heating ele- main advantage of hot-tension test machines is
frames can be used for hot-tension tests by at- ments are positioned around a tensile specimen. that the test specimen is heated uniformly along
taching a heating system to the machine frame. Thermocouple and extensometer edges touch its entire gage length, and hence other useful
The frame may impart loading by either a screw- the specimen. The grip design and the specimen materials properties such as total tensile elonga-
tion, plastic anisotropy parameter, cavity forma-

tion, and so forth can be determined in addition
to yield/ultimate tensile strength and reduction
of area. On the other hand, the overall time
needed to conduct a single test may be longer
than in the Gleeble test method.
Specimen Geometry. In tension testing,
elongation values are influenced by gage length
(see Chapter 2 Bulk Workability of Metals). It
is thus necessary to state the gage length over
which elongation values are measured. When
the ductilities of different materials (or of a sin-
gle material tested under different conditions)
are compared in terms of total elongation, the
specimen gage length also should be adjusted in
proportion to the cross-sectional area. This is of
great importance in the case of small elonga-
tions, because the neck strain contributes a sig-
nificant portion to the total strain. On the other
hand, the neck strain represents only a small por-
tion of the elongation in the case of superplastic
deformation. Unwins equation (Ref 8) also
shows the rationale for a fixed ratio of gage
length with cross-sectional area, expressed as a
fixed ratio of gage length to diameter (for round
bars) or gage length to the square root of the
cross-sectional area (sheet specimens). This re-
inforces the importance of stating the gage
length used in measuring elongation values.
Usually, the length-to-diameter ratio is between
4 and 6.
In addition to the gage length, the specimen
shoulder geometry and hence the gripping sys-
tem are also important design considerations. It
is desirable that specimen deformation takes
place only within its gage length; the shoulder
should remain undeformed. This is not always
the case, as can be seen from Fig. 6. The macro-
graphs of Fig. 6(a) (Ref 9) show the initial and
Fig. 4 Typical servohydraulic universal testing machine with a chamber and instrumentation for high-temperature deformed condition of a specimen in which
testing measurable deformation has occurred within the
Fig. 5 Typical examples of heating methods for load-frame tension testing. (a) Induction heating. (b) Environmental chamber. (c) Split-furnace setup
Figure 3 illustrates how hot-tension data are

used to select a hot-working temperature. The
safe, maximum hot-working temperature lies
between the PDT and the ZDT. In this hypothet-
ical curve of on-heating data, the PDT is 1095
C (2000 F) and the ZDT is 1200 C (2200 F).
On-cooling data should be determined using
preheat temperatures between the PDT and the
ZDT. For example, 1095, 1150, 1175, and 1200
C (2000, 2100, 2150, and 2200 F) would be
good preheat temperatures for on-cooling
studies. Typical on-cooling results are de-
picted in Fig. 7. A 1200 C (2200 F) preheat
temperature results in marginal or poor hot
workability over the possible working range,
Fig. 6 Initial specimen geometry and deformed specimen for cases in which (a) shoulder deformation occurred (Ref whereas an 1175 C (2150 F) preheat tempera-
9) or (b) the shoulder remained undeformed (Ref 10) ture results in acceptable hot workability over a
relatively narrow temperature range. Both 1150
and 1095 C (2100 and 2000 F) preheats result
grip area. On the other hand, the specimen with below approximately 30 to 40%. The maximum in good hot ductility over a relatively narrow
a different shoulder geometry (Fig. 6b) de- hot-working temperature is determined from temperature range. The 1150 C preheat temper-
formed essentially only along its gage length on-cooling data. The objective is to determine ature is preferred over the 1095 C preheat tem-
(Ref 10). The shoulder-deformation problem is which preheat temperature provides the highest perature because it provides good hot ductility
not insurmountable. In this regard, analyses and ductility over the broadest temperature range over a broader temperature range.
techniques, such as those developed by without risking permanent structural damage by Hot workability usually is enhanced by
Friedman and Ghosh (Ref 9), should be applied overheating. greater amounts of prior hot deformation. This
in order to eliminate the effect of shoulder de- An alloy with hot-tensile ductility rated as occurs because second phases and segregation-
formation from measured hot-tension data. marginal or poor may be hot worked, but smaller prone elements are distributed more uniformly
reductions and fewer passes per heating are re- and the grain structure is refined. Deformation at
quired, perhaps in combination with insulating high and intermediate temperatures during com-
Hot Ductility and Strength Data coatings and/or coverings. In extreme instances, mercial hot-working operations often refines the
from the Gleeble Test it may be necessary to minimize development of grain structure by dynamic (or static) recrystal-
tensile strains by employing special dies for lization, thereby augmenting subsequent hot
The reduction of area (RA) and strength are deforming under a strain state that more nearly ductility at lower temperatures. Because a spec-
the key parameters measured in hot-tension tests approaches hydrostatic compression (e.g., extru- imen tested on cooling to the low-temperature
conducted with a Gleeble machine (Ref 1, 11, sion).** end of the hot-working range has not been
12). Because RA is a very structure-sensitive It should be emphasized that the hot-tension deformed at a temperature where grains dynam-
property, it can be used to detect small ductility test reflects the inherent hot ductility of a mate- ically recrystallize, the grain structure is unre-
variations in materials of low to moderate rial, that is, its natural ability to deform under fined. Thus, ductility values will tend to be
ductility, such as specialty steels and superal- deformation conditions. If a workpiece pos- somewhat lower than those experienced in an
loys. However, it should be recognized that RA sesses defects or flaws, it may crack due to actual metalworking operation in which defor-
will not effectively reveal small variations in localized stress concentration in spite of good mation at higher temperatures has refined the
materials of extremely high ductility (Ref 2). inherent hot ductility. structure. The fact that the low-temperature end
Yield and tensile strength can be used to select of the on-cooling ductility range is lower than
required load capacity of production processing **E. Siebel, Steel, Vol 93, 1933 the values that would result in plant metalwork-
equipment.
Ductility Ratings Table 1 Qualitative hot-workability ratings for specialty steels and superalloys
Hot-tension reduction Expected alloy behavior under normal
Experience has indicated that the qualitative of area(a), % hot reductions in open die Remarks regarding alloy hot-working practice
ratings given in Table 1 for hot ductility as a 30 Poor hot workability. Abundant cracks Preferably not rolled or open-die forged. Extrusion may
function of Gleeble reduction-of-area data can be feasible. Rolling or forging should be attempted
be used to predict hot workability, select hot- only with light reductions, low strain rates, and an
working temperature ranges, and establish hot- insulating coating.
3040 Marginal hot workability. Numerous This ductility range usually signals the minimum
reduction parameters. Normal reductions* cracks hot-working temperature. Rolled or press forged
may be taken on superalloys when the reduction with light reductions and lower-than-usual strain
of area exceeds 50%, but lighter reductions are rates.
necessary when ductility falls below this level. 4050 Acceptable hot workability. Few Rolled or press forged with moderate reductions and
cracks strain rates
Thus, in this rating system, the minimum hot- 5060 Good hot workability. Very few cracks Rolled or press forged with normal reductions and
working temperature is designated by the tem- strain rates
perature at which the reduction of area falls 6070 Excellent hot workability. Occasional Rolled or press forged with heavier reductions and
cracks strain rates.
70 Superior hot workability. Rare cracks. Rolled or press forged with heavier reductions and
Ductile ruptures can occur if higher strain rates than normal provided that alloy
*Normal reductions as used in this chapter depend on both strength is too low. strength is sufficiently high to prevent ductile ruptures.
the alloy system being hot worked and the equipment being
used. For example, normal reductions for low-carbon steels (a) Ratings apply for Gleeble tension testing of 6.25 mm (0.250 in.) diam specimens with 25.4 mm (1 in.) head separation.
would be much greater than those for superalloy systems.
ing operations is not sufficient to alter the prac- Strength Data negligible. However, the difference between ul-
tical translation of the results. This feature timate tensile stress and flow stress increases as
serves as a safety factor for establishing the min- In the hot-working temperature range, temperature decreases because restoration
imum hot-working temperature. strength generally decreases with increasing processes cease. Furthermore, the Gleeble ten-
Although some alloys will recover hot ductil- temperature. However, the strength data plotted sile test does not accurately determine flow
ity when cooled from temperatures in the vicin- in Fig. 8 demonstrate that deformation resist- stress because the strain rate is not constant.
ity of the ZDT, it is nonetheless wise to avoid ance does not vary with preheat temperature to Nonetheless, the test still provides useful, com-
hot-working preheat temperatures approaching the same degree as does ductility. Furthermore, parative information concerning how the
the ZDT in plant practice because interior re- strength measured on heating is usually strength of an alloy varies as a function of tem-
gions of the workpiece may not cool sufficiently greater than that measured on cooling. perature within a given strain-rate range. For
to allow recovery of ductility, thereby causing To calculate the force required to deform a example, by analyzing strength values for com-
center bursting. Because industrial furnaces do metal in an industrial hot-working operation, mon alloys in relation to the load-bearing ca-
not control closer than approximately 14 C accurate measurement of flow stress is desir- pacity of a given mill, it may be possible to use
(25 F), the recommended furnace tempera- able. Ultimate tensile stress measured in the test data for a new or unfamiliar alloy in judg-
ture ordinarily should be at least 14 C (25 F) hot-tensile test is only slightly greater than ing whether the equipment is capable of form-
lower than the maximum temperature indicated flow stress at the high-temperature end of the ing the new or unfamiliar alloy.
by testing on cooling. hot-working range because work hardening is
Hot-Tension Data for

Commercial Alloys
For illustrative and comparative purposes,
Gleeble hot ductility and strength curves for
some commercial alloys are presented in Fig. 9.
The nominal compositions of these materials are
given in the table accompanying Fig. 9.
The hot-tensile strengths for the cobalt- and
nickel-base superalloys over the hot-working
temperature range are substantially higher than
those for the high-speed tool steel and the high-
strength alloy, which are iron-base materials.
Furthermore, the ductility data reveal that Ren
41 has the narrowest hot-working temperature
range (T) of 140 C (250 F) of the three su-
peralloys. Hot working of this alloy below 1010
C (1850 F) will lead to severe cracking. This
characteristic, coupled with its high deformation
resistance, makes this alloy relatively difficult to
hot work. On the other hand, HS 188 has high
deformation resistance, but it has high ductility
over a broad hot-working temperature range
from 1190 C (2175 F) to below 900 C (1650
F). Therefore, the permissible reduction per
Fig. 7 Typical on-cooling Gleeble curves of specimen reduction of area as a function of test and preheat tempera-
draft may be relatively small if the hot-working
tures with typical hot-workability ratings indicated
equipment is not capable of high loads, but HS
188 can be hot worked over a broader tempera-
ture range than Ren 41. However, if the equip-
ment has high load capacity, then heavier reduc-
tions can be taken on HS 188 than on Ren 41.
From the hot-working curves established for the
iron-base, high-strength alloy AF 1410, the low
deformation resistance coupled with high ductil-
ity over a broad temperature range indicate that
this material has extremely good hot workabil-
ity. Mill experience has verified this. The curves
shown for M42, the high-speed tool steel, reveal
that it is intermediate in hot workability between
the superalloys and AF 1410; this conclusion has
also been verified by mill experience.
Figures 10 and 11 illustrate the variation of
hot-tensile ductility values at various tempera-
tures. These results were correlated to the
fracture surfaces and structures of the test speci-
mens. For the high-strength iron-base alloy AF
1410, the on-heating curve in Fig. 10 shows
that the ZDT is never reached at practical upper-
Fig. 8 Typical on-heating deformation-resistance data obtained in Gleeble testing limit hot-working temperatures. At the highest
temperature tested (1230 C, or 2250 F) and at not occur, and the fracture surfaces indicate a - incipient melting was not evident in the
the PDT (1120 C, or 2050 F), where hot- less-ductile fracture mode. The correlation microstructure. Microstructural evidence of in-
tensile ductility is extremely high, the fracture among fracture appearance, microstructure, cipient melting at the ZDT is observed for some
appearance is ductile and dynamic recrystalliza- and hot-tensile ductility was even more evident alloys, but not for others (Ref 13).
tion occurs, leading to an equiaxed grain struc- for a developmental solid-solution-strengthened An example of the sensitivity of the hot-
ture. At the higher temperature, a coarser grain cobalt-base superalloy (Fig. 11). At the PDT tensile Gleeble test is shown in Fig. 12 for
structure results from grain growth, which ac- (1150 C, or 2100 F), dynamic recrystallization iron/nickel-base superalloy Alloy 901 (Ref 14).
counts for the drop in ductility. At the opposite occurs, fracture was primarily transgranular, and A small amount of lanthanum added to one heat
end of the hot-working temperature range (842 the fracture appearance was ductile. At the (top curve) was sufficient to reduce the analyzed
C, or 1548 F), the elongated grain structure ZDT (1200 C, or 2200 F), both static recrys- sulfur content to the 1 to 5 ppm range. This re-
reveals that dynamic recrystallization does tallization and grain growth were obvious, but sulted in a small improvement in the hot-tensile
ductility according to Gleeble hot-tensile data.
Isothermal Hot-Tension Test Data
From the isothermal hot-tension test, infor-

mation can be obtained about a number of mate-
rial parameters that are important with regard to
metalworking process design. These include
plastic-flow (stress-strain) behavior, plastic
anisotropy, tensile ductility, and their variation
with the test temperature and the strain rate.
StressStrain Curves
Engineering stress-strain curves from isother-
mal hot-tension tests are constructed from load-
elongation measurements. The engineering, or
nominal, stress is equal to the average axial stress
and is obtained by dividing the instantaneous load
by the original cross-sectional area of the speci-
men. Similarly, the engineering, or nominal,
strain represents the average axial strain and is
obtained by dividing the elongation of the gage
length of the specimen by its original length.
Hence, the form of the engineering stress-strain
curve is exactly the same as that of the load-
elongation curve. Examples of engineering stress-
strain curves obtained from hot-tension testing of
an orthorhombic titanium aluminide alloy (Ref
15) at 980 C (1800 F) and a range of nominal
(initial) strain rates are shown in Fig. 13. The
curves exhibit a stress maximum at strains less
than 10%, a regime of quasi-stable flow during
which a diffuse neck develops and the load drops
gradually, and, lastly, a period of rapid load drop
during which the flow is highly localized (usually
in the center of the specimen gage length) and
failure occurs. The engineering stress-strain curve
does not give a true indication of the deformation
characteristics of a metal because it is based en-
tirely on the original dimensions of the specimen.
These dimensions change continuously during
the test. Such changes are very significant when
testing is performed at elevated temperatures.
The true stress and true strain are based on ac-
tual (instantaneous) cross-sectional area and
length measurements at any instant of deforma-
Material Alloy type Nominal composition, wt% tion. The true-stress/true-strain curve is also
Ren 41 Nickel-base superalloy 0.09 C, 19 Cr, 10 Mo, 11 Co, 3 Ti, 1.5 Al, 1.35 Fe, bal Ni known as the flow curve since it represents the
Alloy 718 Nickel-base superalloy 0.05 C, 18 Cr, 3 Mo, 5 Nb, 1 Ti, 0.5 Al, 19 Fe, bal Ni basic plastic-flow characteristic of the material
HS 188 Cobalt-base superalloy 0.08 C, 22 Cr, 22 Ni, 14 W, 1.5 Fe, 0.05 La, bal Co under the particular (temperature-strain rate)
M42 High-speed tool steel 1.10 C, 3.8 Cr, 9.5 Mo, 8 Co, 1.5 W, 1.2 V, bal Fe testing conditions. Any point on the flow curve
AF 1410 High-strength steel 0.16 C, 2 Cr, 1 Mo, 14 Co, 10 Ni, bal Fe
can be considered the yield stress for a metal
Fig. 9 Typical on-cooling Gleeble curves of strength and ductility as functions of test temperature for several com- strained in tension by the amount shown on the
mercial alloys. curve. An example of the variation of the true
stress versus true strain for Al-8090 alloy de- The variation of true stress with true strain can 17). These curves reveal that deformation at low
formed under superplastic conditions (T 520 also give insight into microstructural changes that temperatures, at which nonsuperplastic condi-
C, ee 7.8 104 s1) is shown in Fig. 14 (Ref occur during hot deformation. For example, for tions prevail, is characterized by an increase of
16). Under these conditions, it is apparent that superplastic materials, an increase in the flow flow stress with strain due to the strain hardening.
the flow stress is almost independent of strain. stress with strain is normally indicative of strain- At higher temperatures, the effect of strain on the
For ideal superplastic materials, the flow stress enhanced grain growth. A decrease in flow stress, flow stress decreases until it becomes negligible
is independent of strain. A nearly constant, or particularly at high strains, can often imply the at the highest test temperature, thus indicating the
steady-state, flow stress is also frequently ob- development of cavitation damage (see the sec- occurrence of superplastic flow.
served at hot-working temperatures in materials tion Cavitation During Hot-Tension Testing in
that undergo dynamic recovery. In these cases, this chapter) or the occurrence of dynamic recrys- Material Coefficients from Isothermal
steady-state flow is achieved at strains of the tallization. As an example, true-stress/true-strain Hot-Tension Tests
order of 0.2, at which the rate of strain harden- curves for a -TiAl submicrocrystalline alloy de-
ing due to dislocation multiplication is exactly formed at temperatures between 600 and 900 C A number of material coefficients can be ob-
balanced by the rate of dislocation annihilation (1110 and 1650 F) and a nominal (initial) strain tained from isothermal hot-tension tests. These
by dynamic recovery. rate of 8.3 104 s1 are shown in Fig. 15 (Ref include measures of strain and strain-rate
hardening and plastic anisotropy. The strain-
hardening exponent (usually denoted by the
symbol n) describes the change of flow stress
(with an effective stress, s) with respect to the
effective strain, e, such that:
Temperature, C
871 927 982 1038 1093 1149 1204 ln s
90 n= (Eq 1a)
PDT ln e
(1120 C)
Gleeble reduction of area, %
For a uniaxial tension test, and prior to the de-

80 Elongated grains indicative of
no dynamic recrystallization velopment of a neck, the distinction of effective
Drop in ductility from stress and strain is not necessary because they
grain growth are equal to the axial stress s and strain e, so that
70
the expression is simply:
Less ductile
ln s
60 n= (Eq 1b)
fracture ln e
The strain-hardening exponent may have val-

50
1500 1600 1700 1800 1900 2000 2100 2200 2300 ues from n 0 for a perfectly plastic solid to n
1 for an elastic solid; negative values of n may
Temperature, F
also be found for materials that undergo flow
softening due to changes in microstructure or
Fig. 10 Typical Gleeble curve of reduction of area versus test temperature for an aircraft structural steel (AF 1410).
crystallographic texture during deformation.
At the PDT, dynamic recrystallization occurs leading to an equiaxed grain structure. Fracture appearance is
ductile. According to Eq 1, if the constitutive equation
for stress-strain behavior is of the form s Ken,
then a logarithmic plot of true stress versus true
strain results in a straight line with a slope equal
to n. However, this is not always found to be the
case and reflects the fact that this relationship is
Temperature, C only an empirical approximation. Thus, when
the plot of ln(s) versus ln(e) [or the plot of
871 927 982 1038 1093 1149 1204
70 log(s) versus log(e)] results in a nonlinear value
of n, then the strain-hardening exponent is often
PDT defined at a particular strain value. In general, n
60 increases with decreasing strength level and de-
Gleeble reduction of area, %
Equiaxed grains Grain

creasing ease of dislocation cross slip in a poly-
ductile fracture growth crystalline material.
50 The strain-rate sensitivity exponent (usually
denoted by the symbol m) describes the variation
of the flow stress with the strain rate. In terms of
40 effective stress (s) and effective strain rate (e), it
is determined from the following relationship:
ZTD
30 ln s
m= (Eq 2a)
ln e
Elongated grains
Less ductile
20 which is simplified for the condition of pure uni-
1500 1600 1700 1800 1900 2000 2100 2200 2300 axial tension as:
Temperature, F
ln s
m= (Eq 2b)
Fig. 11 Typical Gleeble curve of reduction area versus test temperature for a cobalt-base superalloy. ln e
120
Engineering stress, MPa

100
1.67 102
80
8.33 103
60
1.67 103
40
8.33 104
20
4
0 1.67 10
0 50 100 150 200 250
Engineering strain, %
Fig. 13 Engineering stress-strain curves for an or-

thorhombic titanium alloy (Ti-21Al-22Nb)
tested at 980 C and a range of initial strain rates (s1).
Source: Ref 15
The R value of a sheet material may be sensi-

Fig. 12 Gleeble ductility curves for lanthanum-bearing and standard Alloy 901 tested on cooling from 1120 C. Note tive to the testing conditions and in particular to
that the lanthanum-bearing heat displays slightly higher ductility. Specimens represent transverse orientation the strain rate and temperature. This is a result of
on a nominal 25 cm square billet. Specimen blanks were heat treated at 1095 C for 2 h and then water quenched prior
to machining. Specimens were heated to 1120 C, held for 5 min, cooled to test temperature and held for 10 s before
variation of the mechanism that controls defor-
being tested at a nominal strain rate of 20 s1 (crosshead speed 5 cm/s; jaw spacing, 2.5 cm). After Ref 14 mation (e.g., slip, grain-boundary sliding, etc.)
with test conditions. For the orthorhombic tita-
nium aluminide material discussed previously,
Deformation tends to be stabilized in a mate- axial tension. However, caution should be exer- the normal plastic anisotropy parameter shows a
rial with a high m value. In particular, the pres- cised when making such measurements to en- very weak dependence on strain, but a notice-
ence of a neck in a material subject to tensile sure that the stress state along the gage length is able variation with strain rate (Fig. 18). This
straining leads to a locally higher strain rate and uniaxial. Therefore, measurements in regions trend can be attributed to the presence of me-
thus to an increase in the flow stress in the near the sample shoulder and the failure site chanical and crystallographic texture and the ef-
necked region due to strain-rate hardening. (where a stress state of hydrostatic tension may fect of strain rate on the operative deformation
Such strain-rate hardening inhibits further devel- develop during necking) should be avoided. mechanism.
opment of the strain concentration in the neck. Figure 17 shows an example of such data from a
Thus, a high strain-rate sensitivity imparts a high Ti-21Al-22Nb sample pulled to failure in uniax-
ial tension at a nominal strain rate of 1.67 Effect of Test Conditions on
resistance to necking and leads to high tensile
elongation or superplasticity. Materials with val- 104 s1 and at a temperature of 980 C (1800 Flow Behavior
ues of m equal to or greater than approximately F). Within experimental scatter, the R value is
constant for the majority of deformation. When considering the effect of test conditions
0.3 exhibit superplasticity, assuming cavitation
Apparently, lower values of R at low strains on flow behavior, it must be understood that test-
and fracture do not intercede. An empirical rela-
(near the specimen shoulder) or very high strains ing for the modeling of deformation processes is
tion between tensile elongation and the m value
(at the fracture tip) are invalid due to constraint very different from testing for static mechanical
is revealed in the data collected by Woodford
(Ref 18) shown in Fig. 16. In addition, a number or flow-localization effects, respectively, and
of theoretical analyses have been conducted to hence conditions that are not uniaxial. 1100
600 C 650 C
relate m and tensile failure strain, ef (Ref 1921).
1000
For example, Ghosh (Ref 19) derived:
900
ef m ln(1 f 1/m) (Eq 3)
800
700 C
in which f denotes the size of the initial geometric
700
(area) defect at which flow localization occurs.
Strength, MPa
The plastic anisotropy parameter R character- 600

izes the resistance to thinning of a sheet material
during tension testing and is defined as the slope 500
of a plot of width strain, ew, versus thickness 750 C
strain, et, (Ref 22), that is: 400
de w 300 800 C
R= (Eq 4)
de t
200
850 C
A material that possesses a high R value has a 100 900 C
high resistance to thinning and hence good
formability, especially during deep-drawing op- 0
0 20 40 60 80 100 120
erations. Materials with values of R greater than
Elongation, %
unity have higher strength in the thickness direc-
tion than in the plane of the sheet.
The plastic anisotropy parameter can be read- Fig. 14 True-stress/true-strain data for an Al-8090 Fig. 15 True-stress/true-strain curves obtained from
alloy deformed in tension at 520 C and a true tension testing of submicrocrystalline TiAl
ily measured using specimens deformed in uni- strain rate of 7.8 04 s1. After Ref 16 samples. After Ref 17
properties at very low (quasi-static) loading superplastic region II is displaced to higher strain-rate variation involves continuously
rates. Testing conditions for deformation strain rates as temperature is increased and/or changing the crosshead speed during the tension
processes must cover a range of strain rates and grain size is decreased. Moreover, the maximum test to achieve nearly constant strain rate. This
may require high strain rates of 1000 s1 or observed values of m increase with similar approach assumes uniform deformation along
more. For tension testing, conventional test changes in these parameters. the gage length and no end effects and leads to
frames are applicable for strain of rates less than The stress-strain curve and the flow and frac- the following relation between crosshead speed
0.1 s1, while special servohydraulic frames ture properties derived from the hot-tension test
, desired strain rate e, the initial gage length lo,
have a range from 0.1 to 100 s1 (see are also strongly dependent on the temperature and time t:
Introduction to High Strain Rate Testing in at which the test is conducted. In both single-
Mechanical Testing and Evaluation, Volume 8 of crystal and polycrystalline materials, the
e lo exp(e t) (Eq 5)
the ASM Handbook, 2000, p 427). For strain strength decreases with temperature because the
rates from 100 to 1000 s1, the Hopkinson critical resolved shear stress decreases sharply
(Kolsky)-bar method is used. This chapter and with an increase in temperature. On the other
the following discussions only consider isother- hand, the tensile ductility increases with temper-
mal conditions and strain rates below 0.1 s1, ature because of the increasing ease of recovery
where inertial effects can be neglected. and recrystallization during deformation.
Effect of Strain Rate and Temperature on However, the increase in temperature may also
Flow Stress. At hot-working temperatures, most cause microstructural changes such as precipita-
metals exhibit a noticeable dependence of flow tion, strain aging, or grain growth that may af-
stress on strain rate and temperature. For in- fect this general behavior.
stance, the variation of flow stress with strain The flow stress dependence on temperature
rate for Ti-6Al-4V (with a fine equiaxed mi- and strain rate is generally given by a functional
crostructure) deformed at 927 C is shown in form that incorporates the Zener-Hollomon pa-
Fig. 19 (Ref 23). For the strain-rate range shown rameter, Z e exp (Q/RT) (Ref 24) in which Q
in Fig. 19, a sigmoidal variation of the flow is the apparent activation energy for plastic flow,
stress with strain rate is observed. From these R the universal gas constant, and T is the ab- Fig. 17 Width versus thickness strain (e w versus et) for
data, the strain-rate sensitivity (m value) can be solute temperature. an orthorhombic titanium aluminide speci-
readily calculated. The result of these calcula- Effect of Crosshead Speed Control on Hot men deformed at 980 C and a nominal strain rate of
1.67 104 s1. Source: Ref 10
tions (Fig. 20) shows that m is low at low strain Tension Data. The selection of constant-strain-
rates and then increases and passes through a rate versus constant-crosshead-speed control in
maximum after which it decreases again. This conducting isothermal, hot-tension tests is an
behavior is typical of many metals with fine- important consideration, especially for materials
grain microstructures and reveals that superplas- that are superplastic. When experiments are con-
ticity is not manifested in either the low-stress, ducted under constant-crosshead-speed condi-
low-strain-rate region I or the high-stress, high- tions, the specimen experiences a decreasing
strain-rate region III. Rather, superplasticity is strain rate during the test, thus making the inter-
found only in region II in which the stress in- pretation of results difficult, especially in the su-
creases rapidly with increasing strain rate. The perplastic regime. A method to correct for the
Fig. 18 Anisotropy parameter R versus the local axial

true strain for various nominal strain rates.
Data correspond to a Ti-21Al-22Nb alloy. Source: Ref 10
Fig. 19 Flow stress as a function of strain rate and

grain size for a Ti-6Al-4V alloy deformed at
Fig. 16 Tensile elongation as a function of the strain-rate sensitivity. Source: Ref 18 927 C. The strain level was about 0.24. After Ref 23
Fig. 22 Strain distribution for 12.7 mm (a) and 63.5 mm (b) gage length specimens for two different strain rates.
Length strains are plotted versus original axial position along the gage length. Source: Ref 9
Fig. 20 Strain-rate sensitivity (m) versus strain rate (e)
for the data corresponding to Fig. 19. After
Ref 23
tion Frame-Furnace Tension-Testing Equip- accompanied by the development of a triaxial
ment, two different specimen designs are dis- (hydrostatic*) state of stress in the neck.
cussed (Fig. 6). For one of these designs, defor- Because the flow stress of a material is strongly
mation was limited essentially to the gage dependent on the state of stress, a correction
section, while the other had experienced defor- must be introduced to convert the measured av-
mation in the shoulder section. For the specimen erage axial stress into the effective uniaxial flow
geometry in Fig. 6(a), tension tests indicated that stress; that is:
significant straining can occur in the grip regions
and that large strain gradients exist within the s la n = s FT (Eq 6)
gage section of the specimen. The strain gradient
(variation) along the gage length and the defor-
mation of the grip section depend on the gage in which s av l denotes the average axial stress
length and tensile strain rate. As can be seen in required to sustain further deformation, s is the
Fig. 22, the strain gradient of the smaller gage effective (flow) stress, and FT is the stress triax-
length (12.7 mm, or 0.5 in.) specimen geometry iality factor. The magnitude of FT (which essen-
is much steeper than that of the larger one (63.5 tially determines the magnitude of the average
Fig. 21 Comparison of stress versus strain for constant
mm, or 2.5 in.). With regard to the smaller gage hydrostatic stress within the neck) depends on
nominal strain rate (constant crosshead speed, the specimen shape (round bar or sheet) and the
CHS) and constant true strain rate (e) for 5083 Al at 550 length specimen, it is observed that the strain
C. Source: Ref 26 gradient becomes steeper as the strain rate in- geometry of the neck. Bridgman (Ref 27) con-
creases. Furthermore, a reduction of deforma- ducted a rigorous, theoretical analysis with re-
tion in the shoulder can be achieved by decreas- gard to the stress state at the neck for both
ing the width of the gage section because of the round-bar and for sheet specimen geometries.
The crosshead-speed schedule embodied in decrease in deformation load and hence stress For a plastically isotropic material, the following
Eq 5 has been used successfully for a test of Ti- level generated in the shoulder. However, there equations were derived for the stress triaxiality
6Al-4V (Ref 25). Verma et al. (Ref 26) have also are constraints in gage-width reductions arising factor of round-bar (FTr ) and sheet (FTs ) speci-
shown the efficacy of this approach by conduct- from the microstructural characteristics of a par- mens in the symmetry plane of the neck:
ing tension tests at constant crosshead speed as ticular material; in some cases, there may be an
well as constant strain rate on superplastic 5083- insufficient number of grains across the speci- R a
-1
Al specimens. Figure 21 compares stress-strain men section. FTr = 1 + 2 ln 1 + (Eq 7)

characteristics determined under constant- a 2 R
crosshead-speed conditions with those from
constant-strain-rate tests for two different initial Modeling of the Isothermal R
1/ 2
FTs = 1 + 2
strain rates. Constant-crosshead-speed tests Hot-Tension Test
a
showed consistently lower strain hardening -1
(lower flow stresses) and larger strain to failure The detailed interpretation of data from the a 2 a 1/ 2 1 a 1/ 2
ln 1 + + 1+ - 1
(higher tensile elongations) than the correspon- isothermal hot-tension test frequently requires R R 2 R
ding constant-strain-rate tests did. The above some form of mathematical analysis. This analy- (Eq 8)
finding highlights the importance of the test con- sis is based on a description of the local stress
trol mode; in addition, this mode should be state during tension testing and some form of
clearly stated when elongation and/or flow stress numerical calculation. The approach is de- *The term hydrostatic stress is defined as the mean value of the
data are reported. scribed briefly in this section. normal stresses. The term triaxial stress is often used to imply
Effect of Gage Length on Strain Distri- the presence of a hydrostatic stress. However, the term triaxial
bution. Under superplastic deformation condi- Stress State at the Neck stress is not equivalent to hydrostatic stress, because the pres-
ence of a triaxial stress state could be a combination of shear
tions, specimen geometry (especially shoulder stresses and/or normal stresses or only normal stresses. The
design) plays an important role in the determi- Prior to necking, the stress state in the tension term hydrostatic stress is thus preferred and more precise in de-
nation of hot-tension characteristics. In the sec- test is uniaxial. However, the onset of necking is scribing solely normal stresses in three orthogonal directions.
in which a represents the specimen half radius or tility. A comparison of simulation results (e.g., the material elements experience.* The load-
width, and R is the radius of curvature of the nominal stress-strain curves, axial-strain distri- equilibrium condition is thus described by:
neck.* bution, and total elongation) obtained from FEM
The variation of the stress triaxiality factor for analyses to those of the direct-equilibrium si Albi / FTi = s j A lbj / FTj (Eq 10)
round-bar (FTr ) and sheet (FTs ) specimens as a method has shown that the latter approach gives
function of the a/R ratio is shown in Fig. 23. For realistic predictions (Ref 29). To this end, a brief or, using Eq 9:
a positive a/R value (concave neck profiles), FT description of this simpler method is given in the
is less than unity, thus promoting flow stabiliza- following paragraphs.
tion. On the other hand, for negative a/R (convex Model Formulation. The formulation of the e smi e sni A lbi / FTi = e smj e snj A lbj / FTj (Eq 11)
neck profiles), FT 1; thus, flow tends to be direct-equilibrium method is based on dis-
destabilized. cretization of the sample geometry, description in which the subscripts and/or superscripts i and
In a rigorous sense, the closed-form equations of the material flow behavior, and development j denote the corresponding parameters for ele-
for FT (Eq 7 and 8) are applicable only for the of the appropriate load-equilibrium equation. ments i and j, respectively, FT represents the
plane of symmetry at the neck. At other loca- The specimen geometry (dimensions, cross- stress triaxiality factor, and Alb is the load-
tions, the solution for the exact form of FT is not section shape, geometrical defects, etc.) is first bearing area.
available. However, as has been shown from fi- specified. The specimen is divided along the For the case in which the material cavitates
nite-element method (FEM) analyses (Ref 28, axial direction into horizontal slices/elements during tension testing (see the section
29), Eq 7 and 8 provide a good estimate for the (Fig. 24). For the material flow behavior, the Cavitation During Hot-Tension Testing in this
stress triaxiality factor in regions away from the simple engineering power-law formulation has chapter), Eq 10 and 11 must be modified. In par-
symmetry plane provided that the local values of been used in most modeling efforts, that is: ticular, the presence of cavities affects the exter-
a and R are inserted into the relations. nal dimensions of the specimen (because they
s = K e sn e sm (Eq 9)
lead to a volume increase) and hence the load-
bearing area, the stress triaxiality factor, and the
Numerical Modeling of the strain rate at which the material deforms.
Hot-Tension Test in which s, e s, and es denote the effective stress,
effective strain, and effective strain rate, respec- As discussed by Nicolaou et al. (Ref 33),
tively, of the material. K, n, and m represent the spherical and uniformly distributed cavities in-
Two types of methods have been employed to crease each of the three dimensions (length,
model the tension behavior of materials: the strength coefficient, strain-hardening exponent,
and the strain-rate-sensitivity index, respec- width, and thickness**) of the tension specimen
FEM and the somewhat simpler finite-difference by the same amount. The relationship between
(direct-equilibrium) method originally pre- tively.
At any instant of deformation, the axial load P the macroscopic area (i.e., the external area of
sented by Gsell et al. (Ref 30), Ghosh (Ref 31), the specimen) Am, the load-bearing area Alb, and
and Semiatin et al. (Ref 32). Both approaches in- should be the same in each element in order to
maintain force equilibrium. The load borne by the initial (uncavitated) area Asp
o is then simply:
volve solutions that satisfy the axial force equi-
librium equation and the appropriate boundary each slice is equal to the product of its load-
conditions. These models enable the prediction bearing cross-sectional area and axial stress; the Am Alb /(1 Cv)2/3 (Eq 12)
of important parameters such as neck profile, axial stress is equal to the flow stress corrected
failure mode, axial-strain distribution, and duc- for stress triaxiality due to necking and evalu- where Cv is the cavity volume fraction and Alb is
ated at a strain rate corresponding to that which given by:
A lb = Aosp exp ( - e s ) (Eq 13)

*For a/R 0 (in particular, for 2 a/R 0), the stress
triaxility factor for sheet tensile specimesns with a convex
-1 In addition, the length l, width w, and thickness

[
curvature is given by FT 2 Q arctan 1 /

( )]
Q - 1 ,

t, for a sheet specimen, or radius r of a round-bar
specimen increase according to:
where Q = (1 + 2R/a).
d (Eq 14)
d =
(1 - Cn )1/ 3
in which d denotes any of the dimensions (l,

w, t, or r) for the case when cavities are pres-
ent in the material, and d represents the respec-
tive dimension changes with strain alone.
The matrix strain rate es, can also be related to
the macroscopic sample strain rate e. Using
power-dissipation arguments, the relation be-
tween the two strain rates is found to be (Ref
34):
es (1/ ) e (Eq 15)
in which is the relative density of the specimen

( 1 Cv) and is the stress-intensification
Fig. 24 Discretization of the sheet specimen for the
simulations of the isothermal hot-tension test *In case of a cavitating material, the strain rate is that of the
Fig. 23 Stress triaxiality factor for sheet and round-bar (Ref 33). The specimen geometry corresponds to the spec- matrix-material element, not the matrix-cavity continuum.
specimens imen shown in Fig. 6(b) (Ref 10, 15) **Diameter for a round-bar specimen geometry
factor, which for spherical and uniformly dis- shown in Fig. 25 (Ref 29). The effect of the ab- primarily occurs in the grain boundaries during
tributed cavities is (Ref 33): sence of a taper on increased elongation is quite high-temperature deformation, is referred to as
dramatic, especially as the strain-rate sensitivity cavitation. In some cases, cavitation may lead to
1/ 2/3 (Eq 16) increases from m 0.02 to m 0.15. For mate- premature failure at levels of deformation far
rials deformed at cold-working temperatures less than those at which flow-localization-
Algorithm. After having specified the speci- (m 0.02), tensile flow will still localize in the controlled failure would occur. For a given ma-
men geometry and the material-flow relation, the absence of a taper because the reduced section it- terial, the extent of cavitation depends on the
equations for model formulation can be inserted self acts as the defect relative to the greater cross- specific deformation conditions (i.e., strain rate
into an algorithm to simulate the tension test. At sectional area of the shoulder (Ref 31, 32). In and temperature). A wide range of materials ex-
any instant of deformation, the axial variation in contrast to the results for samples with and with- hibit cavitation; these materials include alu-
strain rate is calculated based on the load-equi- out a 2% taper, the predictions for samples with a minum alloys (Fig. 27a), conventional titanium
librium equation, which provides the ratios of the 1% versus a 2% taper show much less difference. alloys (Fig. 27b), titanium aluminides, copper
strain rates in the elements, and the boundary With appropriate modification, the direct- alloys, lead alloys, and iron alloys (Ref 3638).
condition (e.g., constant crosshead speed), which equilibrium modeling approach may also be used An important requirement for cavitation dur-
provides the specific magnitudes of the strain to analyze the uniaxial hot-tension testing of ing flow under either hot-working or superplas-
rates. The strain rates are then used to update the sheet materials that exhibit normal plastic tic conditions is the presence of a tensile stress.
macroscopic (and microscopic) strain and cavity anisotropy (Ref 35). Selected results are shown On the other hand, under conditions of homoge-
volume fraction (for a cavitating material) in in Fig. 26. The engineering stress-strain curves neous compression, cavitation is not observed;
each element. The simulation steps are: exhibit a load maximum, a regime of quasi-stable in fact, cavities that may be produced under ten-
flow during which the diffuse neck develops and sile flow can be removed during subsequent
1. An increment of deformation is imposed, and the stress decreases gradually, and finally a pe- compressive flow. In addition, it has also been
a/R and FT are calculated for each slice. riod of rapid load drop during which flow is demonstrated that the superposition of a hydro-
2. From the true strain and the cavity-growth highly localized in the center of the gage length. static pressure can reduce or eliminate cavitation
rate (see the section Cavitation During Hot When the m value is low, an increase in R in- (Ref 39). Hot isostatic pressing can also heal the
Tension Testing), the cavity volume fraction creases the amount of quasi-stable flow; that is, it deformation damage of nucleated cavities.
is determined. stabilizes the deformation in a manner similar to Cavitation is a very important phenomenon in
3. The true-strain-rate distribution is calculated the effect of strain-rate sensitivity. In addition, hot working of materials because not only may
for each element, using the equilibrium equa- the simulation results reveal that the flow- it lead to premature failure during forming, but it
tion and the boundary condition. stabilizing effect of R decreases as m increases also may yield inferior properties in the final
4. From the true-strain, cavity volume fraction, and in fact becomes negligible for conditions that part. Therefore, it has been studied extensively,
and strain-rate distributions, the engineering approach superplastic flow (i.e., m 0.3). primarily via the tension test.
stress and strain are calculated.
5. Steps 1 to 4 are repeated until a sharp neck is
formed (localization-controlled failure) or the Cavitation Mechanisms/
Cavitation During Phenomenology
cavity volume fraction at the central element
reaches a value of 0.3 (fracture/cavitation- Hot-Tension Testing
controlled failure). Cavitation occurs via three often-overlapping
A large number of metallic materials form mi- stages during tensile deformation: cavity nucle-
Example Applications. Several results illus- croscopic voids (or cavities) when subjected to ation, growth of individual cavities, and cavity
trate the types of behavior that can be quantified large strains under tensile modes of loading. coalescence. Each stage is briefly described in
using the direct-equilibrium modeling approach. This formation of microscopic cavities, which the following sections, while a more detailed
The first deals with the effect of specimen taper
on tensile elongation. Tension-test specimens
usually have a small (2%) reduction in the
cross-sectional area from the end to the center of
the reduced section in order to control the loca-
tion of failure. The predicted effect of reduced-
section taper on the engineering stress-strain
curves for non-strain-hardening materials is
Fig. 25 Comparison of the engineering stress-strain

curves for non-strain-hardening samples with-
out or with a 1 or 2% taper predicted using the direct- Fig. 26 Direct-equilibrium simulation predictions of engineering stress-strain curves at hot-working temperatures for
equilibrium approach. Source: Ref 29 various values of the strain-rate sensitivity and the normal plastic anisotropy parameter. Source: Ref 35
review of ductile fracture mechanisms is in the point, this effect may thus lead one to conclude It should be noted that this theoretical rela-
article Mechanisms and Appearances of that cavities nucleate continuously rather than tionship between m and the cavity-growth pa-
Ductile and Brittle Fracture in Metals Failure merely become microscopically observable con- rameter for an individual cavity follows the
Analysis and Prevention, Volume 11 of the ASM tinuously. Irrespective of the exact mechanism, same general trend as the experimentally deter-
Handbook (2002, p 587626). it is thought that the assumption of continuous mined correlation between the strain-rate sensi-
Cavity Nucleation. Several possible cavity- nucleation of cavities of a certain size (e.g., 1 tivity and the apparent cavity-growth rate APP.
nucleation mechanisms have been established m) still produces the same mechanical effect The parameter APP, which is readily derived
including (a) intragranular slip intersections on failure via cavitation or flow localization as from experimental data (Ref 46), is defined by:
with nondeformable second-phase particles and the postulated actual physical phenomenon.
grain boundaries, (b) sliding of grains along Cavity-growth mechanisms can be classified Cv Cvo exp (APP e) (Eq 20)
grain boundaries that is not fully accommodated into two broad categories: diffusional growth
by diffusional transport into those regions, and and plasticity-controlled growth (Ref 43). in which Cv is the cavity volume fraction at a
(c) vacancy condensation on grain boundaries Diffusional growth dominates when the cavity true strain e, APP is the apparent cavity-growth
(Ref 40). A frequently used cavity-nucleation size is very small. As cavity size increases, dif- rate, and Cvo is the so-called initial cavity vol-
criterion based on stress equilibrium at the cav- fusional growth decreases very quickly, and ume fraction.
ity interface is: plastic flow of the surrounding matrix becomes Cavity coalescence is the interlinkage of
the dominant cavity-growth mechanism. An il- neighboring cavities due to a microscopic flow-
rc 2( p i)/s (Eq 17) lustration of a cavity-growth-mechanism map localization process within the material ligament
(Ref 44) is shown in Fig. 28. From an engineer- between them. Coalescence occurs when the
in which rc is the critical cavity radius above ing viewpoint, plasticity controlled growth is of width of the material ligament reaches a critical
which a cavity is stable; , p, and i denote the greatest interest. In such cases, the growth of an value that depends on initial cavity spacing and
interfacial energies of the void, the particle, and isolated, noninteracting cavity is described for the strain-rate sensitivity. Coalescence can occur
the particle-matrix interface, respectively; and s the case of uniaxial tension deformation by: along both the longitudinal and transverse direc-
is the applied stress. This criterion requires flow tions with the latter being more important be-
hardening, which is minimal in superplastic ma- V Vo exp (e) cause it eventually leads to failure. According to
terials except in cases of significant grain Pilling (Ref 47), cavity coalescence may be re-
growth, in order to nucleate cavities at less fa- or garded as a process that in effect increases the
vorable sites, such as smaller particles. In addi- mean cavity-growth rate. In particular, the effect
tion, such surface-energy considerations require of pairwise coalescence on the average cavity-
h
stresses for initiation and early growth that are r = ro exp e (Eq 18) growth rate dr/de can be estimated from:
3
unrealistically high. Therefore, the development
of other (constrained-plasticity) approaches
based on nucleation and growth from inhomo- in which V and r are the cavity volume and ra-
(
dr 8Cv Fh 0.13r - 0.37( dr de )i de + ( dr de )i
=
)
geneities/regions of high local stress triaxiality dius, respectively, Vo and ro are the volume and de 1 - 4Cv Fhde
has been undertaken (Ref 41). radius of the cavity when it becomes stable, e (Eq 21)
The cavity-nucleation rate N is defined as the denotes axial true strain, and is the individual
number of cavities nucleated per unit area and cavity-growth rate. in which Cv is the instantaneous volume fraction
unit strain. N may either be constant or decrease Several analyses have been conducted to corre- of cavities, is defined from Eq 18,
e (de) is
or increase with strain. However, such strain de- late the cavity-growth rate with material
pendencies are usually not strong. Measure- parameters and the deformation conditions. For
ments have shown than N can be bracketed example, Cocks and Ashby (Ref 45) derived the Plasticity-controlled growth
between 104 and 106 cavities per mm3 per unit following relation between and m for a planar
increment of strain (Ref 16, 41, 42). array of spherical, noninteracting, grain-boundary
The constrained plasticity analysis suggests cavities under tensile straining conditions:
that a size distribution of second-phase con-
stituents/imperfections may lead to a variety of
cavity-growth rates at the nano/submicron m + 1 2 (2 - m )
h = 1.5 sinh (Eq 19)
Cavity growth rate (dr/d)
cavity-size level. From an operational stand- m 3 (2 + m) Superplastic

diffusion
growth
Diffusion
growth
rosp rc rcsp
Cavity radius (r)
Fig. 28 Variation of the cavity-growth rate for different

mechanisms. rc, critical cavity radius; rosp,
cavity radius for onset of superplastic deformation; rcsp,
Fig. 27 Examples of cavitation. (a) In aluminum (Al-7475) alloy. Courtesy of A.K. Ghosh. (b) In titanium (Ti-6Al-4V) critical cavity radius for superplastic deformation. Source:
alloy. Source: Ref 37 Ref 44
a small increment of strain, (dr/de)i is the rate of behavior. Engineering stress-strain (S-e) curves Surprisingly, the engineering stress at a given
growth per unit strain of an isolated cavity for a range of strain-rate sensitivities (m values) elongation for a cavitating material is higher than
(r/3 from Eq 18), and is given by: and cavity-growth rates APP were predicted the corresponding stress (at the same elongation)
using the direct-equilibrium modeling approach for a noncavitating material. This intuitively un-
(1
e/3 (
e)2/27) (Eq 22) (Fig. 30); for all of the cases, the strain- expected result was interpreted by the examina-
hardening exponent n was 0. The cavity volume tion of the effective (load-bearing) area at the
The phenomenon of cavity coalescence was fraction (CV) in the central element at failure is same elongation of a cavitating and a noncavitat-
further investigated by Nicolaou and Semiatin also indicated in the plots. ing specimen. In particular, the analysis of
(Ref 48, 49), who conducted a numerical analy- Examination of the engineering stress-strain Nicolaou et al. (Ref 33) revealed that for a given
sis of the uniaxial tension test considering: the curves reveals that cavitation causes a noticeable elongation the effective area of the cavitating
temporal and spatial location of the cavities in- reduction in total elongation; this reduction is specimen is larger than the area of a noncavitating
side the specimen and the temporal cavity ra- quantified and discussed in more detail in the next one. Therefore, the load and hence the engineer-
dius. Two cases were considered: a stationary section. Figure 30 also shows that the stress-strain ing stress required to sustain deformation is
cavity array (similar to the analysis of Pilling) curves for cavitating and noncavitating samples higher in the case of a cavitating material.
and continuous cavity nucleation. The analysis with the same value of m are very close to each As shown in Fig. 30, the difference between
of the stationary cavity array led to a much sim- other, except at elongations close to failure. the engineering stress-strain curves of cavitating
pler expression than Eq 21, that is: and noncavitating materials is not very large.
Therefore, it can be deduced that Consideres
criterion, if implemented in the usual fashion
dr
= hr ( 1 3 + Cv ) (Eq 23) using data from a tension test (i.e., a plot of load
de versus the elongation of the gage section), can
be used to test whether fracture of a tensile spec-
This simple equation gives predictions very sim- imen occurs due to instability, regardless of the
ilar to the more complex relation of Eq 21. presence of extensive internal cavities in the ma-
With regard to the continuous cavity nucle- terial and whether the volume of the material is
ation case, it was found that the average cavity conserved (Ref 50).
radius was described by:
dr Failure Modes During

= hr (0.2 + Cv ) (Eq 24)
de Hot-Tension Testing
A comparison with experimental cavity size Cavitating hot-tension specimens may fail by
measurements (e.g., Fig. 29) revealed that actual either localized necking (flow localization) or
results are bounded by cavity-growth-and- fracture/cavitation. The second mode of failure
coalescence models that assume either a con- occurs without flow localization in the neck and
stant, continuous nucleation rate (lower limit) or resembles a brittle type of fracture because the
a preexisting cavity array with no nucleation of fracture tip has a considerable area. Micrographs
new cavities (upper limit), that is, Eq 24 and 23, of these modes of failure are presented in Fig.
respectively. 31. The localization type of failure shown in Fig.
31(a) is for an orthorhombic titanium aluminide
(Ti-21Al-22Nb) deformed at 980 C (1795 F)
StressStrain Curves and a nominal strain rate of 1.6 103 s1. On
the other hand, Fig. 31(b) displays the fracture-
The work of Nicolaou et al. (Ref 33) also shed controlled failure of a titanium aluminide alloy
light on the effect of cavitation on stress-strain (Ref 51) deformed in tension at 1200 C (2190
F) and a nominal strain rate of 103 s1.
The particular mode of failure of a material
tested under tension conditions can be predicted
by the magnitude of the strain-rate sensitivity m
and the apparent cavity-growth rate APP. The
corresponding failure-mechanism map for non-
strain-hardening materials is plotted in Fig. 32.
For deformation under superplastic conditions
(m 0.3) and APP 2, the map shows that
failure is fracture/cavitation-controlled. On the
other hand, flow-localization-controlled failure
is seen to predominate only for small values of
the cavity-growth rate. In Fig. 32, experimental
observations of the failure mode of and or-
thorhombic titanium aluminides are also plotted.
The solid data points correspond to fracture-
controlled failures, while the open data points
Fig. 30 Predicted engineering stress-strain curves for correspond to localization-controlled failures.
Fig. 29 Comparison of measurements and predictions tension testing of sheet samples with a 2% Given the assumptions made in deriving such
of the evolution of average cavity radius with taper, assuming strain-hardening exponent n 0, initial
cavity volume fraction C vo 103, various cavity-growth
maps, it can be concluded that the prediction of
strain for an Al-7475 alloy assuming continuous nucle-
ation (Eq 24) or a preexisting cavity array (Eq 23). After rates , and a strain-rate sensitivity exponent m equal to failure mode from the magnitudes of m and APP
Ref 49 (a) 0.1, (b) 0.3, or (c) 0.5. After Ref 33 provides good agreement with actual behavior.
Fig. 32 Failure-mode map developed from simula-

tions of the sheet tension test. Experimental
data points are also shown on the map.
Fig. 31 Micrographs. Orthorhombic titanium aluminide that failed in tension by flow localization. Source: Ref 10.
(b) Near- titanium aluminide that failed in tension by fracture (cavitation). Source: Ref 51
Total Tensile Elongations

As shown in Fig. 33, cavitation may lead to
premature failure and thus to a significant reduc-
tion in the tensile elongation compared to that
measured by Woodford (Ref 18) for noncavitat-
ing metals. For a fixed value of m, the reduction
in elongation for fracture-controlled failure de-
pends on several factors, such as the cavity-
nucleation rate, cavity-growth rate, cavity shape
and distribution, and the cavity architecture.
Several analyses have been conducted to
quantify the effect of cavitation on the tensile
ductility. These include the two-slice approach*
by Lian and Suery (Ref 52), micromechanical
approaches by Zaki (Ref 53), and Nicolaou and
Semiatin (Ref 54), as well as approaches based
on the direct-equilibrium approach described in
the section Numerical Modeling of the Hot-
Tension Test in this chapter. Several results
from the direct-equilibrium model serve to illus-
trate the efficacy of such techniques.
The results shown in Fig. 33 correspond to a
non-strain-hardening material (n 0) with a 2%
taper and Cvo 103. The topmost curve in this
plot depicts the total elongation as a function Fig. 33 Elongation as function of the strain-rate sensitivity and (apparent) cavity-growth rate predicted from direct-
of m for a noncavitating sample, that is, equilibrium simulations. The individual data points represent experimental data. Source: Ref 33
Woodfords trend line. For such a material, the
elongation is controlled, of course, by the onset largest for large values of APP and m, for which not reported. Therefore, a general comparison
of localized necking. The remaining curves in the critical volume fraction of cavities for frac- based only on the value of m can be made. From
Fig. 33 for cavitating samples indicate the ture (assumed to be 0.30) is reached much be- the results of Fig. 33, it is seen that most of the
decrement in elongation due to the occurrence of fore the elongation at which necking occurs. In data points overlie the predicted curves.
fracture prior to localized necking. For low val- fact, for APP 5, the total elongation is almost Comparisons of reported tensile elongations
ues of APP (2) and m (0.3), the decrement independent of m for m 0.3 because fracture data (Table 2) to predictions of a microscopic
is equal to zero because failure is still necking in these cases intercedes during largely uniform, model (Ref 54) in which the cavity architecture
controlled. On the other hand, the decrement is quasi-stable flow. has been taken into account (through the value of
In most cases, rigorous comparisons of pre- G*) are shown in Fig. 34. With the exception of
*The two-slice approach assumes that the specimen comprises dicted tensile elongations and experimental data one data point (No. 9), the major deviations of the
two regions, one of them consisting of the central plane of the (Fig. 33) cannot be made because the cavity-
specimen that contains an initial geometric or strength defect.
The deformation of each region obeys the flow rule while at
growth rate and the cavity-size population and *The parameter G is a factor that describes the geometry of the
any instant of deformation the load is the same in both regions shape were not measured, while other important ligament between two cavities as a function of the cavity ar-
(slices). parameters such as the specimen geometry were chitecture within the specimen.
Table 2 Experimental data from the literature for the deformation and failure of
cavitating materials
Data point Material m Tensile elongation, %
1 -TiAl (as received) 0.38 2.2 219

2 -TiAl (as received) 0.51 2.3 350
3 -TiAl (as received) 0.62 1.8 446, 532
4 -TiAl (heat treated) 0.18 3.4 93, 104
5 -TiAl (heat treated) 0.15 8.0 51
6 5083 Al 0.50 5.2 172
7 Zn-22Al 0.45 1.5 400
8 / brass 0.60 2.3 425
9 Coronze 638 0.33 4.5 275
predictions tend to be on the high side. These de- 13. R.E. Bailey, Met. Eng. Quart., 1975, p
viations could be a result of the neglect in the mi- 4350
croscopic model of the macroscopic strain gradi- 14. R.E. Bailey, R.R. Shiring, and R.J.
ent in the diffuse neck of real tension samples. In Anderson, Superalloys Metallurgy and
addition, as mentioned previously, specimen Manufacture, Proc. of the Third Interna-
geometry and deformation in the shoulder region tional Conference, Claitors Publishing, Fig. 34 Comparison of experimentally determined
of actual tension specimens have an effect on Sept 1976, p 109118 total elongations with (microscopic) model
predictions that incorporate the cavity architecture.
measured ductilities, which is difficult to quan- 15. P.D. Nicolaou and S.L. Semiatin, An Source: Ref 54
tify. Nevertheless, agreement between the meas- Investigation of the Effect of Texture on
ured and predicted ductilities is reasonably good. the High-Temperature Flow Behavior of
an Orthorhombic Titanium Aluminide
REFERENCES Alloy, Metall. Mater. Trans. A., Vol 28A
(No. 3A), March 1997, p 885893 Mater. Trans. A, Vol 27A (No. 7), July
1. Met. Ind., Vol 11, 1963, p 247249 16. J. Pilling and N. Ridley, Superplasticity in 1996, p 18891898
2. M.G. Cockroft and D.J. Latham, Ductility Crystalline Solids, The Institute of Metals, 27. P.W. Bridgman, Studies in Large Plastic
and the Workability of Metals, J. Inst. Met., London, UK, 1989 Flow and Fracture, McGraw-Hill, 1952
Vol 96, 1968, p 3339 17. R.M. Imayev and V.M. Imayev, Mechanical 28. A.S. Argon, J. Im, and A. Needleman,
3. J. Inst. Met., Vol 92, 1964, p 254256 Behaviour of TiAl Submicrocrystalline Distribution of Plastic Strain and Negative
4. G. Cusminsky and F. Ellis, An Investigation Intermetallic Compound at Elevated Tem- Pressure in Necked Steel and Copper Bars,
into the Influence of Edge Shape on peratures, Scr. Met. Mater., Vol 25 (No. 9), Metall. Trans. A, Vol 6A, 1975, p 815824
Cracking During Rolling, J. Inst. Met., Vol Sept 1991, p 2041 2046 29. C.M. Lombard, R.L. Goetz, and S.L.
95 (No. 2), Feb 1967, p 3337 18. D.A. Woodford, Strain-Rate Sensitivity as a Semiatin, Numerical Analysis of the Hot
5. E.F. Nippes, W.F. Savage, B.J. Bastian, and Measure of Ductility, Trans. ASM, vol 62 Tension Test, Metall. Trans. A, Vol 24A
R.M. Curran, An Investigation of the Hot (No. 1), March 1969, p 291293 (No. 9), Sept 1993, p 20392047
Ductility of High-Temperature Alloys, 19. A.K. Ghosh and R.A. Ayres, On Reported 30. C. Gsell, N.A. Aly-Helal, and J.J. Jonas, J.
Weld. J., Vol 34, April 1955, p 183s196s Anomalies in Relating Strain-Rate Sensi- Mater. Sci., Vol 18, 1983, p 17311742
6. D.F. Smith, R.L. Bieber, and B.L. Lake, A tivity (m) to Ductility, Metall. Trans. A, Vol 31. A.K. Ghosh, A Numerical Analysis of the
New Technique for Determining the Hot 7A, 1976, p 15891591 Tensile Test for Sheet Metals, Metall. Trans.
Workability of Ni-Base Superalloys, pre- 20. J.W. Hutchinson and K.W. Neale, Influence A, Vol 8A, 1977, p 12211232
sented at IMD-TMS-AIME meeting, 22 Oct of Strain-Rate Sensitivity on Necking 32. S.L. Semiatin, A.K. Ghosh, and J.J. Jonas, A
1974 Under Uniaxial Tension, Acta Metall., Vol Hydrogen Partitioning Model for Hydro-
7. S.I. Oh, S.L. Semiatin, and J.J. Jonas, An 25 (No. 8), Aug 1977, p 839846 gen Assisted Crack Growth, Metall. Trans. A,
Analysis of the Isothermal Hot Compres- 21. F.A. Nichols, Plastic Instabilities and Vol 16A, 1985, p 203947
sion Test, Metall. Trans. A, Vol 23A (No. 3), Uniaxial Tensile Ductilities, Acta Mater., 33. P.D. Nicolaou, S.L. Semiatin, and C.M.
March 1992, p 963975 Vol 28 (No. 6), June 1980, p 663673 Lombard, Simulation of the Hot-Tension
8. W.C. Unwin, Proc. Inst. Civil Eng., Vol 155, 22. M.A. Meyers and K.K. Chawla, Mechanical Test under Cavitating Conditions, Metall.
1903, p 170185 Metallurgy Principles and Applications, Mater. Trans. A., Vol 27A (No. 10), Oct
9. P.A. Friedman and A.K. Ghosh, Micro- Prentice Hall, 1984. 1996, p 31123119
structural Evolution and Superplastic 23. N.E. Paton and C.H. Hamilton, Micro- 34. S.L. Semiatin, R.E. Dutton, and S.
Deformation Behavior of Fine Grain structural Influences on Superplasticity in Shamasundar, Materials Modeling for the
5083Al, Metall. Mater. Trans. A, Vol 27A Ti-6Al-4V, Metall. Trans. A, Vol 10A, 1979, Hot Consolidation of Metal Powders and
(No. 12), Dec 1996, p 38273839 p 241250 Metal-Matrix Composites, Processing and
10. P.D. Nicolaou and S.L. Semiatin, High- 24. G.E. Dieter, Mechanical Metallurgy, 3rd Fabrication of Advanced Materials IV, T.S.
Temperature Deformation and Failure of an ed., McGraw-Hill, 1986 Srivatsan and J.J. Moore, Ed., The Minerals,
Orthorhombic Titanium Aluminide Sheet 25. A.K. Ghosh and C.H. Hamilton, Me- Metals, and Materials Society, 1996, p 3952
Material, Metall. Mater. Trans. A., Vol 27A chanical Behavior and Hardening Char- 35. P.D. Nicolaou and S.L. Semiatin, Scr.
(No. 11), Nov 1996, p 36753681 acteristics of a Superplastic Ti-6Al-4V Mater.,Vol 36, 1997, p 8388
11. R. Pilkington, C.W. Willoughby, and J. Alloy, Metall. Trans. A, Vol 10A, 1979, 36. M.M.I. Ahmed and T.G. Langdon, Excep-
Barford, The High-Temperature Ductility of p 699706 tional Ductility in the Superplastic Pb-62
Some Low-Alloy Ferritic Steels, Metal Sci. 26. R. Verma, P.A. Friedman, A.K. Ghosh, S. Pct Sn Eutectic, Metall. Trans. A, Vol 8A,
J., Vol 5, Jan 1971, p 1 Kim, and C. Kim, Characterization of 1977, p 18321833
12. D.J. Latham, J. Iron Steel Inst., Vol 92, Superplastic Deformation Behavior of a 37. S.L. Semiatin, V. Seetharaman, A.K. Ghosh,
19631964, p 255 Fine Grain 5083 Al Alloy Sheet, Metall. E.B. Shell, M.P. Simon, and P.N. Fagin,
Cavitation During Hot Tension Testing of ity Growth Maps for Superplastic Mater- 49. P.D. Nicolaou and S.L. Semiatin, The
Ti-6Al-4V, Mater. Sci. Eng. A, Vol A256, ials, J. Mater. Sci., Vol 21, 1986, p 2073 Influence of Plastic Hardening on Surface
1998, p 92110 2082 Deformation Modes Around Vickers and
38. C.C. Bampton and J.W. Edington, The 44. B.P. Kashyap and A.K. Mukherjee, Spherical Indents, Acta Mater, Vol 48,
Effect of Superplastic Deformation on Cavitation Behavior During High Tem- 2000, p 34413450
Subsequent Service Properties of Fine- perature Deformation of Micrograined 50. L. Weber, M. Kouzeli, C. San Marchi, and
Grained 7475 Aluminum, J. Eng. Mater. Superplastic MaterialsA Review, Res A. Mortensen, On the Use of Consideres
Tech., Vol 105 (No. 1), Jan 1983, p 55 Mech., Vol 17, 1986, p 293355 Criterion in Tensile Testing of Materials
60 45. A.C.F. Cocks and M.F. Ashby, Creep Which Accumulate Internal Damage, Scr.
39. J. Pilling and N. Ridley, Cavitation in Fracture by Coupled Power-Law Creep and Mater. Vol 41, 1999, p 549551
Aluminium Alloys During Superplastic Diffusion Under Multiaxial Stress, Met. 51. C.M. Lombard, Superplasticity in Near-
Flow, Superplasticity in Aerospace, H.C. Sci., Vol 16, 1982, p 465478 Gamma Titanium Aluminides, Ph.D. Thesis,
Heikkenen and T.R. McNelley, The 46. P.D. Nicolaou, S.L. Semiatin, and A.K. University of Michigan, Ann Arbor, MI, 2001
Metallurgical Society, 1988, p 183197 Ghosh, An Analysis of the Effect of Cavity 52. J. Lian and M. Suery, Effect of Strain Rate
40. G.H. Edward and M.F. Ashby, Intergranular Nucleation Rate and Cavity Coalescence on Sensitivity and Cavity Growth Rate on
Fracture During Powder-Law Creep, Acta the Tensile Behavior of Superplastic Failure of Superplastic Material, Mater.
Metall., Vol 27, 1979, p 15051518 Materials, Metall. Mater. Trans. A, Vol 31A, Sci. Technol., Vol 2, 1986, p 10931098
41. A.K. Ghosh, D.-H. Bae, and S.L. Semiatin, 2000, p 14251434 53. M. Zaki, Micronecking and Fracture in
Initiation and Early Stages of Cavity 47. J. Pilling, Effect of Coalescence on Cavity Cavitated and Superplastic Materials, Metall.
Growth During Superplastic and Hot Growth During Superplastic Deformation, Mater. Trans. A, Vol 27A, 1996, p 10431046
Deformation, Mater. Sci. Forum, Vol Mater. Sci. Technol., Vol 1 (No. 6), June 54. P.D. Nicolaou and S.L. Semiatin, A
304306, 1999, p 609616 1985, p 461466 Theoretical Investigation of the Effect of
42. S. Sagat and D.M.R. Taplin, Fracture of a 48. P.D. Nicolaou and S.L. Semiatin, Modeling Material Properties and Cavity Architecture/
Superplastic Ternary Brass, Acta Met- of Cavity Coalescence During Tensile Shape on Ductile Failure During the Hot
all.,Vol 24 (No. 4), April 1976, p 307315 Deformation, Acta Mater, Vol 47, 1999, Tension Test, Metall. Mater. Trans. A., Vol
43. A.H. Chokshi, The Development of Cav- p 36793686 29A, 1998, p 26212630
Chapter 8
Torsion Testing to Assess Bulk

Workability
J.J. Jonas, McGill University
WORKABILITY is generally defined as the plex velocity control is required in axial loading controlled failures occur. These failures are
ability to impart a particular shape to a piece of tests. The interpretation of torsion-test data, how- manifested by shear bands or regions of local-
metal under the load capacity of the available ever, is more complex than that used for axial- ized shear deformation because torsion imposes
tooling and equipment, without the introduction testing procedures because the strain and strain a mode of simple shear (Fig. 2). During torsion,
of fracture or the development of undesirable rate vary along the specimen radius. shear bands may originate at a point of material
microstructures. A complete description of the When failure modes are important, torsion is
workability of a material is therefore specified also a valuable diagnostic test. These failures are
by its flow-stress dependence on processing usually divided into two broad categories: frac-
variables (e.g., strain, strain rate, preheat tem- ture-controlled failures, in which deformation is
perature, and die temperature), its failure be- relatively uniformly distributed prior to failure,
havior, and the phase transformations that char- and flow-localization-controlled failures, in
acterize the alloy system to which it belongs. which plastic deformation has been concentrated
Very few mechanical tests are capable of pro- in a particular area of the metal specimen or
viding information about all of these aspects, workpiece prior to the actual fracturing process.
which is primarily a result of the large deforma- Fracture of metals occurs by a variety of mech-
tions that are common in massive forming anisms that depend largely on deformation rate
processes such as forging, extrusion, and rolling. and temperature. At cold working temperatures,
For most common mechanical testssimple or temperatures corresponding approximately to
tension and compression teststhe maximum one-fourth or less of the absolute melting or
uniform strains achieved are rather low because solidus temperature (T/TM 0.25), ductile frac-
of necking and barreling, respectively. By con- ture, which is characterized by void initiation at
trast, strains in excess of 0.3 to 0.7, the levels second-phase particles and inclusions, void
typical of these uniaxial tests, are readily growth, and final coalescence, is most prevalent.
achieved in the torsion test. At warm (0.25 T/TM 0.6) and hot (T/TM Fig. 1 Typical deformation-processing map for
austenitic stainless steel, showings regions of
The torsion test has been used for some 50 0.6) working temperatures, processes such as ductile fracture, wedge cracking, dynamic recrystalliza-
years as a means of hot workability assessment in wedge cracking (the opening of voids and their tion, and safe forming. Note that the boundaries for safe
forming (i.e., the loci of processing conditions between
metals and alloys (Ref 15). The principal virtue propagation at triple junctions, or the intersection ductile fracture and wedge cracking) vary depending on
of this straining mode is the fact that the test- of three grain boundaries) and cavitation (the for- the required ductility (or fracture strain ef). Dash-dot line,
pieces do not undergo significant shape change as mation of voids or cavities around second-phase wedge cracking; solid line, ductile fracture; dashed line,
they are deformed as long as the gage section is particles, particularly at grain boundaries) are the dynamic recrystallization. Source: Ref 8
restrained to a fixed length. Because of the large most important fracture mechanisms. Frequently,
strains that can be achieved under relatively uni- the deformation rate determines which of these
form deformation conditions, the torsion test is fracture mechanisms predominates. Typically, the
the preferred test for obtaining flow-stress data former mechanism occurs at high strain rates, and
and unambiguous indications of failure and mi- the latter occurs at low strain rates.
crostructural response from deformation process- The different fracture regimes are best illus-
ing (Ref 6, 7). Provided failure does not inter- trated by deformation-processing maps (Fig. 1),
cede, it is not unusual to be able to obtain which are frequently determined by tension or
flow-stress results in torsion to deformation lev- torsion tests. Such maps may also depict regions
els equivalent to a true axial strain of 5 or more in in which large strains may be achieved without
tension, or a reduction of 90 to 95% in compres- fracture, or so-called safe processing regimes,
sion. Another attractive feature of the torsion test as well as those in which certain microstructural
is that a constant true strain rate can be imposed changes, such as dynamic recovery or recrystal-
on any given annular region of the specimen by lization, can be expected during processing. Fig. 2 Analogy between (a) torsion of a thin-wall tube
and (b) simple shear. Note the equivalence be-
simply twisting one end relative to the other at a The torsion test can also be used to determine tween the deformations of initially circular grid elements
constant angular velocity. By contrast, more com- conditions under which flow-localization- in the two modes.
Chapter 8: Torsion Testing to Assess Bulk Workability/ 87
inhomogeneity or temperature nonuniformity. torsion test to gage flow-localization-controlled between the torsion axis and the mechanical
Under the former conditions, localization is failure and the effect of processing on mi- fiber. The relationship between ductility and
driven by flow softening (a decrease of flow crostructure using torsion is detailed as well. mechanical texture is analogous to the relation-
stress with increasing strain) due to: ship between these two properties in tensile tests
Dynamic recovery or dynamic recrystalliza- performed on specimens cut from the rolling di-
Material Considerations rection (higher ductility) and transverse direc-
tion
Microstructural instabilities, such as the tion (lower ductility) of rolled plate.
The microstructure of the material to be tested The mechanical behavior in torsion testing is
breakup and spheroidization of lamellar mi-
in torsion greatly affects specimen design and somewhat more complex, however, because of
crostructures
the interpretation of workability data. The most the tendency of the fiber to rotate as twisting
Texture softening
significant material characteristics are grain proceeds. Consider, for example, the torsion of a
Deformation heating
size, crystallographic texture, and mechanical solid round bar in which inclusions are initially
When deformation heating occurs, very high texture. Because metals are composed of indi- in the form of stringers parallel to the torsion
strain rates prevent the dissipation of heat, which vidual grains with specific crystallographic ori- axis (Fig. 3). After twisting, the inclusions at the
in conjunction with the large negative tempera- entations and directional plastic deformation center of the bar have not rotated, but those at
ture dependence of flow stress found in many properties, the torsion-specimen gage-section the surface have rotated to form an angle, the
metals leads to particularly strong flow localiza- diameter should include at least 20 to 30 grain magnitude of which depends on the gage length
tions sometimes referred to as adiabatic shear diameters. Torsion measurement thus reflects an and diameter and the amount of twist. It is ap-
bands. At moderate strain rates, deformation average of the flow and failure response of the parent that the generation of such a mechanical
heat generation and heat conduction can intro- material. This requirement on gage-section di- texture can substantially reorient the planes of
duce substantial temperature gradients during ameter may therefore result in specimens of weakness into the form of a wolfs ear, as evi-
torsion testing, as in conventional metalworking varying sizes, depending on whether they are cut denced by tensile tests on bars prestrained in tor-
processes, and can cause noticeable flow local- from fine-grained wrought products or coarse- sion (Fig. 4). Rotation of the mechanical texture
ization in materials with flow stresses that are grained cast products. and the general level of inclusions may have a
strongly temperature dependent. The presence of crystallographic texture must significant effect on the torsional fracture strain.
Because failure mode and microstructural de- also be taken into consideration. In specimens Because of this, torsion testing is frequently
velopment depend on process variables, torsion- taken from wrought bar stock, for example, wire used to uncover lot-to-lot variations in workabil-
testing equipment and procedures to assess work- textures (a specific crystallographic direction ity for a given material.
ability must be carefully designed. The ability to parallel to the bar axis) are common. The strong
control and measure strain and strain rate textures in plate materials should also be noted
(through monitoring of twist and twist rate) is im- when torsion specimens are fabricated from Specimen Design
portant, as is test temperature. Furthermore, the these materials. Samples taken from forgings or
ability to impose arbitrary deformation or loading extrusions often exhibit preferred crystallo- Although the torsion test has been used for
histories should be considered when designing graphic orientations, which also may vary from many years, there are still no standards for spec-
torsion apparatus to be used to simulate processes one location to another. imen design. Typical specimen geometries are
such as multistage rolling or wire drawing. Mechanical texture, or the presence of inclu- shown in Fig. 5. A certain sense of standardiza-
This chapter discusses equipment design, pro- sions, grain boundaries, or other microstructural tion does exist inasmuch as many researchers
cedures, and interpretation of torsion tests to es- features with a preferred direction, should also have used specimens with a gage-length-to-
tablish workability. Experimental considerations be taken into account when designing a torsion radius ratio of approximately 8 to 1 (Ref 1031),
are discussed, along with the application of tor- specimen. Such texture, or fibering as it is some- although data have been reported for specimens
sion testing to obtain flow-stress data and to times called, can greatly affect the torsional duc- with ratios from 0.67 to 1 (Ref 32) to 17 to 1 (Ref
gage fracture-controlled workability. Use of the tility, depending on the orientation relationship 33). Many have chosen to use solid-bar speci-
Fig. 3 Torsion specimen with two typical inclusions at Fig. 4 Tensile fracture in torsionally prestrained copper specimens. The shear strain in the surface of each specimen
the center and the surface. (a) Prior to twisting. is indicated below the photograph. Specimen marked 3 shows the tensile fracture after twisting to a shear
(b) After twisting strain of 3 and then completely untwisting. Source: Ref 9
mens, some with a reduced gage section and length is short (e.g., 6.5 mm, or 0.25 in.), a sharp microstructure and ductility data. They demon-
transition fillets, whereas others have preferred fillet radius (e.g., 0.25 mm, or 0.010 in.), which strated that, for aluminum and silicon steel,
tubular specimens. Solid specimens are favored is blended to the shoulder area by a sharp coni- which only exhibit dynamic recovery during
for high-strain studies because thin-wall tubes cal taper of about 120 included angle, is often straining, the equivalent stress-strain curves are
have a tendency to buckle in this regime. Tubular used to prevent excessive deformation outside reproducible to within normal experimental
specimens, on the other hand, have a distinct ad- the gage section. For longer gage-length speci- scatter when the length-to-radius ratio is greater
vantage from the point of view of the analysis for mens, a substantially larger fillet radii (e.g., 4 than 2 to 1. It was also shown that the strain to
stress and strain and are particularly valuable for mm, or 0.15 in.) can be used. maximum stress in types 304 and 321 stainless
studies of work hardening at low strains. Thin-wall tubular specimens provide an ad- steels at temperatures at which they dynamically
The most important components of design are vantage over solid-bar specimens in that shear- recrystallize decreases as the length-to-radius
gage section, fillet radius, shoulder length, and strain and shear-strain-rate gradients are virtu- ratio increases up to 2 to 1 (Fig. 6). They at-
grip design. Gage-section geometry determines ally eliminated in the design. This is desirable in tributed this latter effect to an increase in the ho-
the deformation level and deformation rate for a workability testing because ambiguities associ- mogeneity of recrystallization in the longer
given amount of twist and a given twist rate, ated with definition of the failure strain and specimens. It was also noted that significant de-
based on: strain rate are eliminated. Moreover, the use of formation occurs in the fillets between the gage
thin-wall tubes eliminates the variation of tem- section and the shoulders, which induces an
rq perature across the section thickness that occurs error in strain and strain-rate measurement. If
g= (Eq 1) in solid bars twisted at high rates. These temper- the fillet radius is the same for specimens of dif-
L
ature variations are a consequence of the varia- ferent length-to-radius ratios, the error in strain
and tions in shear strain and thus deformation work increases with a decrease in the ratio.
converted into heat. Thin-wall specimens must The effect of bore diameter in tubular speci-
have a rather small length-to-diameter ratio to mens was examined by Barraclough et al. (Ref
rq avoid buckling. Even small amounts of buckling 35) and was found not to be of critical impor-
g = (Eq 2)
L can noticeably affect the magnitude of the plas- tance in determining the strain to maximum
tic properties deduced from the hot torsion test stress for type 304 stainless steel (Fig. 7). The
where is the engineering shear strain (equal (Ref 34). Typical dimensions for thin-wall spec- choice of bore diameter, however, is influential
to twice the tensorial shear strain or 3 times imens are 12.5 mm (0.50 in.) outer diameter, 11 in determining the strain to failure. This be-
the von Mises effective strain), is twist angle mm (0.44 in.) inner diameter, and a gage length havior is discussed in more detail later in this
(in radians), r is the radial position, L is the of 2.5 mm (0.10 in.). chapter.
gage length, is the engineering-shear-strain Barraclough et al. (Ref 35) have examined the In both solid bars and tubular specimens, the
rate, and is the twist rate (in radians per unit effects of variations in specimen geometry on gage-length-to-diameter ratio may also have a
time). the stress-strain curves obtained, as well as on marked effect on the actual specimen tempera-
For a given amount of twist, large values of r
and small values of L promote high values of .
Similarly, large r and small L values result in
large values of for a given . Values for and
vary with radius in a solid bar or thick-wall
tube; they are greatest at the surface and 0 at the
center. Strain and strain rate are typically re-
ported as the values that pertain to the surface of
the specimen. When torsion data are compared
to measurements obtained from other deforma-
tion modes, it is this strain and this strain rate
that are usually important.
For wrought materials, typical geometry con-
sists of a gage diameter of 6.5 mm (0.25 in.) and
a length ranging from 6.5 to 50 mm (0.25 to 2.0
in.) for solid-bar samples. The shorter lengths
are used to obtain higher strain rates. The gage
length selected also influences the fillet radius
that should be used in the torsion specimen be-
cause some deformation usually occurs in this
portion of the specimen as well. When the gage
Fig. 5 Typical torsion specimen geometries used for

workability testing. See text for discussion of di- Fig. 6 Effect of gage-length-to-radius ratio on strain (em) to maximum flow stress or to the onset of steady-state flow.
mensions. The solid lines join data points obtained at similar strain rates and temperatures. Source: Ref 35
ture during moderate-speed ( 102 to 10 can be used without introducing errors in the still be straight (i.e., it will form a helix at a fixed
s1) torsion tests due to the effects of heat con- stress measurement. angle to the torsion axis), indicating fracture-
duction. Because of this, flow curves derived The grip ends of torsion specimens for work- controlled failure, or it will exhibit a kink at a
from data obtained at these rates tend to show a ability studies are generally of two types: larger angle to the torsion axis than the remain-
dependence on the length-to-diameter ratio threaded and geometric cross sections. For der of the line, indicating flow-localization-
(L/d). Flow curves for large L/d specimens tend threaded grip ends, a surface against which the controlled failure. In the former case, the surface
to fall below those for small L/d ratios, in which torque can be reacted is provided by making the shear strain at fracture (sf) is given by sf tan
most of the deformation heat is dissipated into major diameter of the threads less than the f, where f is the angle between the scribe line
the shoulders (Fig. 8). Interpretation of fracture shoulder diameter, or by including flanges be- and torsion axis at failure. This value of sf
strain data from such tests should take into tween the threaded end and the shoulders in the should agree with that obtained from the equa-
account not only the nominal (initial) test tem- specimen design. Triangular, hexagonal, or other tion given above for shear strain as a function of
perature, but also the temperature history during simple geometric cross sections can also be used r, , and L. When flow localization has occurred
the test. for grip ends. In these instances, the torque is re- prior to fracture, the tangent of the angle be-
When workability is limited by flow localiza- acted against the flat faces of the ends in slotted tween the line away from the localization region
tion, gage length can noticeably affect the failure holders. Selection of threaded ends versus trian- and the torsion axis provides an estimate of the
process as well, particularly at moderate defor- gular or hexagonal cross-sectional ends is based workability of the material. However, the tan-
mation rates. At these rates, heat conduction largely on whether torsion testing is conducted gent of the angle between the line and the tor-
along the length of the gage section and into the under load control or stroke control. In load sion axis in the localization region also yields an
shoulder can set up a substantial temperature control, axial loads are applied to prevent buck- estimate of the fracture strain, provided the test
gradient, the magnitude of which depends on ling or to enable combined tension-torsion de- has been carried out to this point.
the gage length, shoulder size, and deformation formation; threaded ends are required for this
rate. In turn, the temperature gradient influences type of test.
the deformation resistance along the length of Triangular or hexagonal ends are most com- Torsion Equipment
the bar and hence the strain profile. mon when tests are to be run in stroke control to
The design of the shoulder and grip ends of prevent changes in gage-section geometry that The torsional mode of mechanical testing, and
torsion specimens is determined largely by the may result from the axial extension that charac- hot torsion in particular, is not as widely applied
method of heating and the type of torsion ma- terizes the torsion testing of many metals. as tension, so that there are few commercially
chine to be used. In all cases, the shoulder diam- Frequently, threaded ends can also be used in available machines. Instead, the apparatus is
eter should be at least one-and-one-half times such situations; however, triangular or hexago- commonly designed for the specific needs of the
the gage-section diameter and preferably two to nal ends have the added advantage of allowing individual researcher. The testing machines de-
three times as large to prevent plastic deforma- rapid specimen removal for purposes of quench- scribed by early authors of hot-workability stud-
tion as well as to minimize elastic distortion. ing, which is beneficial in subsequent metallur- ies were designed to provide a range of constant
Shoulder length can be very short for torsion gical analysis. The possible development of twist rate only (Ref 15, 1113, 16, 17, 33,
tests conducted at room temperature. If testing is axial stresses and their effect on flow and failure 3641). Later test machines were constructed
to be carried out at elevated temperature, the response in fixed-end tests should be carefully that can deliver multistage deformation sched-
shoulder should be longer to ensure temperature considered, because these stresses can be a sig- ules with intermediate unloading of the speci-
uniformity, typically about 25 mm (1.0 in.) for nificant fraction of the torsional shear stress men (Ref 29, 31, 4251). These latter machines
specimens heated by induction and 25 to 50 mm (Fig. 9). The subject of axial stresses is dis- can be effective in simulating multistage work-
(1.0 to 2.0 in.) for specimens heated in furnaces. cussed in greater detail later in this chapter. ing operations.
An interesting aspect of shoulder design for The final step in the preparation of torsion All torsion machines have certain features in
torsion testing is that there are no end effects specimens for determination of workability common: a test frame, a drive system, twist and
in contrast to the considerations that apply to is the application of a fine axial line on the gage-
tension testing. This is often referred to as St. section surface. This may be done using a
Venants principle for tension, which signifies felt-tip pen for room-temperature tests or a fine-
that stress concentrations arising from changes metal scribing instrument for elevated-tempera-
in cross section persist for a distance of about ture tests. After twisting to failure, the line may
one diameter along the axis. In torsion, the prin-
ciple does not apply to the shear stresses that are
developed. Hence, very short, thin-wall tubes
Fig. 8 Stress-strain curves for solid torsion specimens

of 3.3% Si steel showing effect of gage-length-
to-diameter ratio (L/d) on flow stress at high strain rates
Fig. 7 Effect of bore diameter on strain (em) to maxi- when adiabatic heating occurs. The flow curves are in
mum stress of type 304 stainless steel specimens terms of von Mises effective stress-strain (
e), defined
with a 7 mm gage length and a gage-length-to-radius ratio by
3 and e /3 , where - is the shear- Fig. 9 Dependence of shear stress and mean axial
of 2 to 1. The solid lines join data points obtained at sim- stress/shear-strain curve obtained in torsion testing. stress on effective strain in fixed-end torsion
ilar strain rates and temperatures. Source: Ref 35 Source: Ref 35 tests at high temperatures. Source: Ref 12
torque monitoring devices, and a furnace. This the stiffness of the torsion machine decreases, Perhaps the most publicized electric-drive
section briefly reviews the choices made by and the possibility of problems associated with machine is that designed by Rossard (Ref 39,
some designers as a guide to the construction of wave effects and ringing in the load cells dur- 50) at IRSID in France. This machine has es-
a torsion facility. ing high-strain-rate experiments increases. The sentially all the features of devices of similar
top of the load train is attached to the upper vintage (Ref 11, 13, 17, 40, 41). It consists of a
Design of the Test Frame crosshead by a set of bolts and spherical washers 4 kW, 1500 rpm electric motor, a 30-to-1
that provide the primary means of aligning the continuously variable hydraulic speed reducer, a
An essential requirement of a hot torsion test system. If the grips and other components are 30-to-1 gear box and an electropneumatic
is that the specimen be twisted about its axis accurately machined and the system is accu- clutch-and-brake assembly. This arrangement
with no induced bending. Therefore, the test rately aligned, the total indicator runout on the provides a strain-rate range of 0.5 to 20 s1 for
frame should be of high rigidity and should have specimen rotation can be kept to several thou- specimens measuring 24 by 6 mm (0.9 by 0.2
a facility for the accurate alignment of the rotat- sandths of an inch with a modified testing main.). Lower strain rates can be obtained by in-
ing components. The two major types of setups chine system. Such tolerances are comparable to stallation of other fixed-ratio gear boxes be-
comprise horizontal load trains mounted on the tolerances and deflections maintained in a tween the hydraulic speed reducer and the clutch
lathe beds and vertical load trains mounted in linear actuator and are usually acceptable for assembly. The maximum torque available at the
modified test machines. Lathe beds (Fig. 10) most torsion work. These tolerances can only be
provide adequate stiffness and precision-ma- improved by placing special reinforcing struc-
chined guideways that enable accurate align- tures or die sets between the crossheads of the
ment of the tooling components. Typically, for machine, thereby substantially increasing the
test specimens 2 to 6 cm (0.8 to 2.5 in.) in cost of the equipment.
length, a bed of 2 to 3 m (6.5 to 10 ft) will ac-
commodate all the necessary devices of the load Motors
train. However, this type of arrangement does
not allow for control of axial loads. Tests can be Three types of motors are generally used for
conducted with the tailstock (saddle) either fixed torsion testing to assess workability: electric,
or floating. hydraulic, and hydraulic rotary actuator.
Control of axial loads can be achieved by Electric Motors. Variable-speed electric mo-
mounting the components in a modified test ma- tors are the simplest and were the first to be used
chine, such as the electrohydraulic setup shown for torsion testing. These motors can have speeds
in Fig. 11. A hydraulic motor is in series with an ranging from fractions of a revolution per minute
independently controlled, linear, hydraulic actu- (rpm) to several thousand rpm. However, the
ator that may be run either in load or stroke con- available speed range for a given motor may be
trol. The torque is reacted against the posts of limited. In addition, they are not readily adapted
the test frame using a reaction plate to which a to provide a variation in strain rate during each
set of cam followers is attached, which allow the deformation cycle, as is necessary for accurate
ram-hydraulic motor assembly to translate up simulation of plate- or strip-mill operations.
and down. A series of adapters and the torque Other limitations are the difficulty of maintaining
and axial load cells are located at the top of the a constant rotation rate if the torque requirement
load train. The use of many adapters as well as changes during a test and the generally low
individual load cells for axial load and torque is torques that such motors generate. Electric mo-
undesirable, because they increase the overall tors may be fitted with gearboxes and energy-
length of the load train. As the length increases, storage devices to prevent these problems.
Fig. 11 Electrohydraulic testing machine modified for

combined torsional and axial loading: 1,
upper crosshead; 2, tension load cell; 3, torque cell
adapter No. 1; 4, torque cell; 5, torque cell adapter No. 2;
6, water-cooled grip; 7, specimen holder; 8, specimen; 9,
induction coil; 10, ram; 11, lower crosshead; 12, linear
(tension-compression) actuator; 13, torque reaction plate;
14, hydraulic motor; 15, incremental optical encoder; 16,
electrical leads; 17, water lines; 18, hydraulic lines.
Fig. 10 A servocontrolled, hot torsion machine mounted on a lathe bed. Source: Ref 48 Source: Ref 52
highest strain rate is limited by the power rating A critical component of all electric-drive sys- accelerated from 0 to 3000 rpm in 2 to 3 ms, cor-
of the motor, which for the configuration de- tems is the clutch/brake assembly. Many of the responding to approximately 0.3 radians or 0.05
scribed previously is about 10 N m (7.5 lbf early systems were not equipped with brakes, revolutions. The brake, which is of similar de-
ft), which may be insufficient for simulation of and so there was a tendency for the end of the sign, has comparable responsiveness.
rolling of steel at lower temperatures. Although specimen to continue to rotate after the clutch Hydraulic Motors and Hydraulic Rotary
this difficulty can be solved by increasing the had been disengaged. Furthermore, the mechan- Actuators. To overcome the drawbacks of elec-
power of the motor, this increases the inertia of ical dog systems used on some machines had tric motors, hydraulic motors and hydraulic ro-
the drive system, which places more severe de- the tendency to cause ringing of the specimen tary actuators are used in sophisticated torsion
mands on the braking system. This is particu- after engagement so that a true constant rate of systems. Hydraulic motors are typically of the
larly troublesome at lower strain rates, where se- deformation was not achieved (Ref 11). The use fixed-displacement type, in which the rotation
vere strain overshoot can occur. of axial electromagnetic clutches eliminated this and displacement of the shaft is proportional to
Weiss et al. (Ref 47) overcame some of these problem, but provided a less-positive linkage the volume of oil flowing through the motor.
difficulties by the incorporation of three inde- with a consequent loss of acceleration. Those The torque is produced by the oil as it pushes
pendent drive motors rated at 11, 1.5, and 0.2 machines that have been designed to carry out against a set of spring-loaded vanes that rotate
kW (Fig. 12). This system also includes a fly- working-schedule simulation use conical elec- around the motor housing or spherical pistons
wheel that ensures that only a 1.5% reduction of tropneumatic clutch-and-brake assemblies (Ref that ride up and down a contoured cam. In such
speed occurs when the clutch is engaged. The 41, 4345, 47). Weiss et al. (Ref 47) have de- motors, multiple turns are achieved by supply-
machine has a speed of 0.01 to 3000 rpm in 14 scribed their design in detail (Fig. 13). The ing a continuous flow of oil. Typical speeds
steps, which for the specimen dimension used clutch develops a normal load of 25 kN (2.8 range from 1 to about 1000 rpm. Slower speeds
corresponds to a range of shear-strain rates up to tonf) at the frictional interface, thereby enabling are obtained by using gear reducers in series
300 s1, covering seven orders of magnitude. At a torque of 1000 N m (740 lbf ft) to be trans- with the motor.
the highest speed the maximum torque is limited mitted. The system, however, has a very short re- Fulop et al. (Ref 48) describe a machine (Fig.
by the 11 kW motor to 25 N m (18 lbf ft). sponse time, which allows the driven parts to be 10) that uses a piston-type hydraulic drive
controlled by a servovalve. The hydraulic motor
can develop a maximum torque of 120 N m
(88 lbf ft) at a maximum speed of 140 radians/s
(1325 rpm) with a fluid-delivery rate of 70
l/min.
One advantage of the servohydraulic drive
system is that it eliminates the need for clutches,
brakes, and gear boxes. The latter may be in-
cluded, however, to extend the strain-rate range
downward. The use of a closed-loop control sys-
tem provides high effective stiffness, so that the
specimen can be stopped and the specimen un-
loaded in a similar period. The ability to actually
unload the specimen by reverse rotation is very
important in mill-simulation tests and provides a
high degree of precision to the imposed strains.
In hydraulic rotary actuators, the torque is
also supplied by oil pressure against a set of
vanes. However, the rotation is limited to only a
fraction of a revolution. The specimen gage
length therefore must be short or the diameter
must be large to achieve large strains. Such ac-
tuators are suitable when loading histories in-
Fig. 12 Testing machine designed by Weiss et al. Source: Ref 47
volving rotation direction reversals are required
or when the system is to be used for torsional fa-
tigue testing as well.
In addition, rotary actuators can produce a
greater range of rotation rates than hydraulic
motors, particularly at the low end of the scale
where fractions of an rpm are possible. The ro-
tation rates required to simulate various metal-
working operations are given in Table 1.
The flow of oil through either hydraulic mo-
tors or rotary actuators may be at a given rate
that is determined by the hydraulic pump sup-
plying the oil (an open-loop arrangement) or
may be accurately controlled by a servovalve in
a closed-loop circuit. In an open-loop system,
the rotation rate of the motor can vary as the
torque requirement changes. This difficulty is
overcome in a closed-loop, servocontrolled sys-
tem. The servovalve is an electrically actuated
valve that forms part of a feedback circuit. A ro-
Fig. 13 Conical electropneumatic clutch-and-brake system after Weiss et al. (Ref 47) tary transducer measures the rotation of the
Table 1 Torsional rotation rates corresponding to various metalworking operations 1. An added advantage of an optical encoder is
von Mises effective strain Corresponding surface shear
that the angular displacement of the specimen is
Operation rate ( )(a), s1 strain rate in torsion (), s1 Rotation rate(b), rpm determined absolutely because the rotational
sense is also coded. When linked to a fast digi-
Isothermal forging 103 1.73 103 0.02
Hydraulic press forging 1 1.73 16.5 tal-to-analog (D/A) converter, it can produce an
Extrusion 20 34.6 330.4 analog output that is proportional to displace-
Mechanical press forging 50 86.6 827.0 ment, which makes it ideal for position control
Sheet rolling 200 346.4 3307.9 in servocontrolled hydraulic drive systems.
Wire drawing 500 866.0 8269.7
Rheostat shafts may also be connected to one
(a) e /3. (b) Assuming specimen geometry with r/L 1.0 end of the hydraulic motor. This device consists
of a dual-gang potentiometer, which is con-
nected to voltage conditioning and switching
circuits and an eight-bit D/A converter. For each
half revolution, one of the two potentiometers is
active. At the completion of each half turn, the
signal input to the conditioner is switched from
one potentiometer to the other, and, simultane-
ously, the D/A converter is incremented and its
output summed with the scaled output of the ac-
tive potentiometer. In this way a continuous out-
put ramp is obtained that has a slope of 0.2
V/rev. The resolution of this device is 1 mV or
0.03 rad. The minimum rotation speed is de-
termined by the smallest step of a digitally gen-
erated ramp and the resolution of the control
electronics and is about 0.03 rad/s. When typical
specimens are used, shear-strain rates from 4
103 to 100 s1 can be obtained.
The rotary variable-differential transformer is
perhaps the most accurate of the rotation-
measuring devices and is used primarily in
Fig. 14 Typical control and data acquisition system for torsion testing. Source: Ref 48 sophisticated rotary actuator systems. Like a lin-
ear variable-differential transformer, a rotary
variable-differential transformer generates a
variable voltage, as an iron core moves through
hydraulic motor or rotary actuator, and a func- imen. As mentioned previously, a rotation trans- a current-carrying coil.
tion generator provides a predetermined twist ducer is a component of the feedback circuitry When the rotation transducer is located in the
rate (or twist history) in terms of an electrical required to control the hydraulic motor. These load train close to the torsion specimen, the volt-
voltage that varies with time. A sample circuit of transducers are of three types: optical devices, age produced by it is often used to measure the
this type is shown in Fig. 14. variable resistors (rheostats), and rotary vari- actual twist applied to the specimen. However, if
During operation, oil is fed through the servo- able-differential transformers. the load train is long and the transducer is not
valve into the motor or rotary actuator. The rota- With the exception of the device described by close to the specimen, elastic deflections of the
tion is sensed by a transducer that sends a volt- Fulop et al. (Ref 48), measurement of angular system can introduce systematic errors in the
age to the servocontroller. The voltage is then displacement has typically been carried out opti- specimen-twist measurement. To overcome this
compared to the predetermined voltage from the cally. A typical example of such devices is that difficulty, an auxiliary twist transducer is some-
function generator for that given point in time. described by Weiss et al. (Ref 47), which uses a times placed near the specimen. For example, a
Depending on the difference between the two metal disk with holes that interrupts the path be- rheostat system to which a drive pulley is at-
voltages, the servovalve is either opened or tween a lamp and a phototransistor. A Schmidt tached may be coupled to the specimen grip
closed to increase or decrease the flow of oil to trigger circuit, in turn, operates a digital fre- using a tight-fitting rubber O-ring or gear mech-
the hydraulic motor or rotary actuator. This se- quency counter and provides a pulsed signal to a anism. This arrangement is satisfactory for rota-
quence of operations is performed continuously recorder. Two disks are described: one with 100 tion rates of about several hundred rpm.
during the torsion test to ensure that the proper holes, which monitors specimen rotation, and
twist history is being applied. one with 1000 holes, which measures that of the Torque Transducers
Servocontrolled hydraulic systems can also be gearbox output. This device is capable of a 200
programmed for arbitrary strain-rate profiles. In m response at the specimen, which means that Torque sensors for torsion testing are usually
this way, it is possible to impose deformation a shear strain of 0.8% can be resolved at 40 s1. of the reaction-torque type and are usually
steps that accurately follow the strain-rate varia- Improved strain resolution can be obtained by placed on the side of the specimen opposite the
tions that occur, for instance, in the entry of a the use of optical encoders, which use a trans- hydraulic motor or rotary actuator. Load trans-
rolling mill. Indeed, it is possible to examine the parent disk on which a coded mask is etched. A ducers that are capable of monitoring both axial
effects of mill geometry and layout on the mill series of phototransistors transmit the binary or and torsional loads are also available. In either
loads as well as the working characteristics of BCD (binary coded decimal) position code di- case, the torque is measured through a structural
the material. rectly to the display or recording system. A ten- member in the load cell, to which foil-resistance
bit shaft encoder can resolve position to about strain gages have been applied. As torque is
Twist Sensors 20 minutes of arc or 6 milliradians, while a applied, the structural member deflects elasti-
twelve-bit device can achieve a resolution of 1.5 cally. The response is measured with the strain
Specially designed transducers are necessary milliradians, that is, a shear strain of 0.02% on a gages. In practice, the torque can be calibrated
to measure the twist of the motor shaft and spec- specimen with a length-to-diameter ratio of 4 to indirectly using a calibration resistance in a
Wheatstone-bridge circuit or directly by impos- is of primary importance, such as with relatively

ing fixed amounts of torque by hanging weights low-strain-rate hot strength or ductility tests
from a lever arm attached to the torque cell. To (Ref 10, 25, 29, 30, 46, 53). With such units,
ensure that the measured torque is free of any rapid quenching is difficult because the speci-
frictional components, it is desirable that the men must first be removed from the furnace.
torque cell be mounted between the specimen Glover and Sellars (Ref 46, 53) obtained rapid
and the fixed-end bearing. Because this places quenching by withdrawing the specimen axially
the cell close to the furnace and the specimen ex- with the aid of a pneumatic actuator that posi-
tension bars, it is commonly necessary to water tioned the specimen beneath a quench spray.
cool the load cell in order to stabilize the gage This device could only be used either with free-
factor. end specimens or after specimen failure. On the
The capacity of the torque cell should accom- other hand, Weiss et al. (Ref 47) used a complex
modate the specimen geometry, test material, apparatus that first removed a split radiant bar
and test temperature. For typical specimen furnace from around the specimen and covered
geometries, torque cells with ratings of 0 to 565 the elements before in situ quenching with water
N m (415 lbf ft) are sufficient. When testing spray. Both schemes can quench the specimen
is carried out on very soft metals or on most within 0.1 to 1 s.
metals at hot working temperatures (at which the In order to facilitate specimen quenching as
flow stress is very low), a torque cell of one- well as to provide arbitrary temperature/time
tenth this capacity (55 N m, or 40 lbf ft) is profiles, many researchers have opted for induc-
required to maintain adequate signal-to-noise ration or radiant heaters (Ref 12, 21, 36, 40, 45,
tios for data acquisition. With modern electron- 48, 54, 55). Both types have very low thermal
ics, most torque cells can be calibrated and used capacity and thereby promote rapid heating and
at various percentages (usually 10, 20, 50, or cooling. From the point of view of ease of
100%) of their maximum rating, which allows a quenching, induction heaters are preferred.
particular transducer to be used over a range of In induction systems, the coil for torsion spec- Fig. 15 Water-cooled rotating grip for high-tempera-
temperatures and strain rates. With commer- imen heating should overlap the shoulders of the ture torsion testing. (a) Outer housing. (b)
cially available torque cells, torque can be meas- test specimen to ensure temperature uniformity. Inner core. Note that the outer housing, to which water
lines are attached, is held stationary during testing by re-
ured with a precision of 0.1% of the range and In general, the most efficient application of in- action rods against which the torque is reacted. X, O-ring
with a linearity of better than 0.05%. duction heating is for ferrous alloys (carbon and grooves; Y, water inlet/outlet; Z, tapped hole for reaction
In view of the amount of data processing re- ferritic alloy steels, austenitic stainless steels, rod
quired to extract data on shear stress and strain etc.) and nickel- and titanium-base alloys.
from the torque/twist curves, it is advantageous Generator frequencies for heating torsion
to directly digitize and store the data with the aid specimens of these alloys are generally in the stationary grip; it is accomplished by brazing
of a desktop computer. Computer-automated radio-frequency range (100 to 400 kHz). copper tubing that contains a water supply onto
data acquisition greatly enhances the versatility Aluminum and copper alloys do not induction the grip.
of the testing system, particularly when complex heat readily and thus should be heated by this Water cooling is more difficult to accomplish
working schedules are simulated. method only with the use of susceptors. in grips that rotate. In this case, the grip, or an
To prevent contamination of torsion speci- adjacent fixture comprising a core that can rotate
Heating Systems and mens during elevated-temperature torsion stud- and an outer casing that is held stationary must
Water-Cooled Grips ies, specimens may be tested in specially de- be constructed. A typical example of this type of
signed chambers or within quartz tubes in which fixture is shown in Fig. 15. Water leakage is pre-
Torsion tests to establish workability are fre- a controlled atmosphere, such as argon, is main- vented by a set of rubber O-rings. Depending on
quently conducted at elevated temperatures. For tained. A simpler, less expensive technique is to the actual test temperature, the water-cooled fix-
this purpose, specially designed heating arrange- apply protective coatings to the specimens. tures should be constructed of tool steel or stain-
ments and specimen grips are necessary. These coatings are usually vitreous and are sup- less steel. Grips or parts of tooling that are
Selection of a suitable specimen-heating system plied as frits or as commercial metalworking subjected to temperatures higher than approxi-
depends on the particular type of test that is en- glass lubricants. Frits are applied at room tem- mately 540 C (1000 F) should be fabricated
visioned. In general, the furnace type and perature by grinding them into a fine powder, from high-temperature superalloys, such as
arrangement should be such as to provide rapid preparing a slurry with an alcohol carrier, and Waspaloy, Inconel 718, or IN-100.
changes in test temperature as well as a means of dipping the specimens into the coating prepara-
fast quenching. The two most common methods tion. Commercial lubricants used for specimen
of heating are furnace and induction. For typical protection can be applied by a variety of meth- Flow-Stress Data
heating times, furnaces or induction generators ods such as dipping, brushing, and spraying. For
should have power ratings of 2 to 10 kW. both types of glass, the coating melts and forms When failure is controlled by fracture without
Furnaces are usually electric-resistance or a viscous, protective coating as the specimen is prior flow localization, flow curves and fracture
quartz-tube radiant heating types. The hot zone heated. Before using such coatings, however, strains can be readily derived from measure-
should be long enough to ensure uniform heat- initial trials should be conducted to verify that ments of torque versus twist during testing.
ing of the gage section. Initial temperature undesirable reactions with the workpiece do not Frequently, these data indicate increasing or
nonuniformities may give an incorrect picture of occur at elevated temperatures. nondecreasing torque with increasing rotation.
failure if the test material is sensitive to flow lo- Another problem associated with elevated- However, at hot working temperatures (and
calization. This is particularly true of titanium- temperature testing is the necessity of prevent- sometimes even at cold working temperatures),
and nickel-base alloys, for which the flow stress ing heat from being conducted into parts of the the material may show softening due to dynamic
is very sensitive to temperature in the hot-work- load train located away from the specimen grips recovery, dynamic recrystallization, microstruc-
ing regime, and the thermal conductivity is low. (load cells, motors, etc.). This is accomplished tural softening, or deformation heating. In these
Conventional resistance-wound furnaces have by providing water cooling to the grips. This cases, the only reliable method of confirming
been used when long-term temperature stability provision is relatively straightforward for the that failure is fracture controlled is by examina-
tion of the surface scribe line or the specimen Now, as the shear strain is given by Eq 1 and plots of log M versus log at fixed values of .
microstructure. If the deformation has been frac- dr Ld/, on changing variables, Equations 4 Thus, tests at a variety of strain rates are re-
ture controlled, the inclination of the scribe line and 5 become: quired for proper evaluation. If the material ex-
to the torsion axis will remain uniform, and the hibits power-law strain-rate hardening ( m),
microstructure in the gage section will be de- 2p g2 m* is equal approximately to the strain-rate sen-
formed uniformly. The tangent of the angle be- M= tg 2 dg (Eq 6) sitivity exponent, m. At cold working tempera-
q3 0
tween the scribe line and the torsion axis should tures, m* is usually small and can often be neg-
be equal to the shear strain corresponding to that or: lected with respect to the value of 3 n*. At hot
value of twist at which the torque-twist record working temperatures, m* can take values com-
shows a sudden drop, indicating fracture. 2p g2 parable to the values of m at these temperatures,
M= tg 2 dg (Eq 7) which typically range from 0.1 to 0.3, or values
q3 g1
that are not negligible in comparison to 3 n*.
Reduction of Torque-Twist Data to where 2 and 1 are the shear strains correspon- If the specimen design is a thick-wall tube
Shear Stress versus Shear Strain ding to the radii a2 and a1, respectively. (i.e., the wall thickness is comparable to the ra-
Since the measured parameter in the torsion dius), a variation of the Fields and Backofen
Methods for reducing torque-twist (M-) data test is the torque, M, analytical schemes are re- formula may be used to derive the - relation.
to the shear stress/shear strain (-) form vary quired to extract the shear stress from the inte- is taken to be the shear strain at the outer
with specimen geometry. Reduction of torque- grals in Eq 4 to 7. Nadais approach (Ref 56) is diameter, rs /L. However, the shear stress
twist data is easiest with thin-wall tubes. In this to assume that the shear stress in the body de- in this case is:
case, the shear strain is found from r*/L, pends only on the local value of the shear strain, 3 + n* + m *
where r* is the mean tube radius and L is the namely: (3 + n * + m*) M ri
t= 1 - (13)
gage length. A relation between the torque and 2 prs3 rs
the (assumed) uniform shear stress is found from
() (Eq 8)
the torque-equilibrium relation M 2r*2t, where n* and m* are the slopes of the two loga-
which results in M/2r*2t. Here, t is the Then, since the value of the integral in Eq 6 de- rithmic plots, and ri denotes the internal speci-
tube-wall thickness, assumed to be small in pends on the upper limit, differentiation with re- men radius.
comparison to r*. spect to yields: The Fields and Backofen analysis has been
For round solid bars, the derivation of - widely used as a conversion technique, but in
from M- is more complex. Several methods are 1 dM general the procedure has been abbreviated in
available; however, all overlook the occurrence ta = 3M + q (Eq 9)
2 pa 3 dq that the torque-twist curves are simply scaled
of deformation heating effects at high strain uniformly with constant values of m* and n*.
rates. Nadai (Ref 56) was apparently the first to or For hot-torsion tests, the rate sensitivity m* is
formally extend the elastic analysis of a tor- usually that obtained from plots of log M (steady
sional bar from Ludwik (Ref 57) to include the M state) versus log , and n* is taken as 0. This
case of fully plastic flow. He argued that the dis- ta =
2 pa 3
( 3 + dlnM / d ln q ) (Eq 10)
form of the procedure is convenient and pro-
tortions produced in an isotropic medium by vides reasonable values of steady-state shear
small angles of relative twist can be considered This expression serves as the basis of the well- stress, but it has the effect of greatly distorting
as simple shear strains that are proportional to known graphical technique for obtaining shear the initial work-hardening portion of the flow
the distance, r, of an element from the bar axis, stresses from torque-twist records for solid bars curve and therefore renders invalid the measure-
that is, Eq 1. This assumes that all radii remain (Fig. 16). Clearly, dM/d(/L) BC/DC, and, ments of work-hardening rates and yield stress.
straight during deformation, which is borne out since DC /L, then BC dM/d, so that: Even when the analytical procedure of Eq 12 is
by observation and symmetry considerations fully applied, a significant source of error arises
even at high strains as well as finite-element 1 from the formalism itself. It may be shown that
modeling (Ref 58). The total torque developed ta = (3BA + BC) (Eq 11)
2 pa 3 the relative error in shear stress is given by:
in the testpiece can be obtained by noting that a
solid bar can be considered as an assembly of In spite of its simplicity, Nadais method suf- Dt DM DM Dq dM 1
concentric, thin-wall tubes. fers from two serious limitations. First, the iden- = + 2 q +
t M dM M M dq dq (3 + n * + m*)
When the specimen is twisted, the component tity in Eq 8 takes no account of the strain-rate
of torque, dM, due to one such tube of radius r sensitivity of the shear stress, so that Eq 9 cannot (Eq 14)
and thickness dr is given by: be applied at high temperatures at which the rate
sensitivity is significant. Second, the accuracy
dM 2r 2 dr (Eq 3) with which the surface shear stress is determined
depends critically on the precision with which
where is the current shear stress in the elemen- the slope of the torque-twist curve is known.
tal tube. The total torque induced in the testpiece The analysis for torsion of a solid bar of rate-
is then: sensitive material is somewhat different. The
simplest method was developed by Fields and
a2
Backofen (Ref 36). In this technique, the shear
M = 2p tr 2 dr (Eq 4) strain in the - relation is taken to be that for
the outer-diameter surface of the solid bar (
0
rs/L). The corresponding shear stress is ob-

where a2 is the outer radius of the bar. If the tained from:
specimen is in the form of a tube of inner radius
a1 and outer radius a2, Eq 4 becomes: t = (3 + n * + m *) M/ 2 prs3 (Eq 12)
a2 The quantity n* is the instantaneous slope of log

Fig. 16 Graphical determination of shear stress from
torque-twist records for rate-insensitive mate-
M = 2p tr 2 dr (Eq 5)
a1 M versus log . The value of m* is found from rials. Source: Ref 56
where indicates the absolute error of the meas- latter quantity is the rate sensitivity measured by to buckling because of axial compression at am-
urement, and the differences as used in slope the instantaneous difference in torque induced bient temperatures or to rupture because of axial
determinations. It is important to note that the by a sudden change in . Although m may differ tension at elevated temperatures. Consequently,
imprecision in increases linearly with . This from m* by a factor as large as 1 to 2, / is large maximum strains can be attained. In a
effect is demonstrated in Fig. 17, where Eq 12 is nevertheless small compared with other intrinsic sense, the R1 bar serves as a rotating mandrel
applied directly to torque-twist data for elec- errors arising from this method and, for all prac- and supports the attached (R2 R1) layer.
trolytic tough-pitch copper. If the torque curve tical purposes, can be neglected. The difference between the two measured
had been smoothed prior to the conversion, a Another important consideration in the reduc- torques M1 and M2 determined at the same twist
smooth stress-strain curve would have been ob- tion of torque-twist data is the occurrence of de- rate is used to deduce the properties of the in-
tained, but the error in stress would still have formation in the fillet radius. A simple method to cremental layer. The average shear stress in the
been 10% (Ref 59). correct for such deformation in a solid-bar sam- layer between R1 and R2 is then given by:
Another source of error that must be kept in ple utilizes the definition of an effective gage
mind when Eq 12 is used arises from the as- length equal to the length of the reduced section 3 M2 - M1
sumption that the shear stress is uniquely de- plus an increment in length whose magnitude is t= (Eq 16)
2 p R23 - R13
fined when the shear strain and strain rate are a function of the gage diameter, fillet radius, m*,
known. In their original paper (Ref 36), Fields and n*. Pohlandt (Ref 62) derived a semiempir-
ical expression for the effective gage length. This shear stress is considered to be developed
and Backofen pointed to the importance of
Subsequently, Khoddam and his coworkers (Ref at the mean radius of the layer (R1 R2)/2, so
strain-rate history in determining stress, but
63, 64) applied the finite-element method (FEM) that the associated mean shear strain is given by
failed to mention the inappropriateness of strain
to validate the Pohlandt equation for materials ave [(R1 R2)/2](/L), where /L is the twist
as a state parameter. Canova (Ref 60) have
whose constitutive behavior is of the power-law per unit length of the specimen. However, be-
considered the error to be attributable to the
form. By contrast, FEM simulations revealed cause of the nature of the torque integral and the
choice of strain rate and strain as the appropriate
that the effective gage length varies with defor- flow-hardening characteristics of most materi-
state variables, rather than the more realistic
mation when the constitutive behavior is of the als, generally corresponds to the stress acting
selection of strain rate and hardness state (Ref
hyperbolic sine form. at a radius greater than (R1 R2)/2 and there-
61). They state that the relative error in stress is
fore to a shear strain greater than ave. The mag-
given by:
nitude of this error, which also affects the results
Shear-Stress Derivations for of the thin-wall tube tests, depends on the details
Dt (m * - m ) of the work-hardening relation as well as on
= (Eq 15) Arbitrary Flow Laws
t (3 + n * + m*) (R2 R1) and varies during a given test.
The methods described previously for con- The differential method suffers from errors
where m* and n* have the same meaning as in that are inherent to the definitions of the stress
verting the measured torque to shear stress at the
Eq 12 and where m ( ln M/ ln )state. This and the strain introduced above. This method is
outer radius of solid-bar samples and thick-wall
tubes are based on the analysis of Ref 36 and therefore best suited to samples with a small ra-
therefore apply only to materials obeying simple dius difference (R2 R1). However, under these
constitutive relations such as the parabolic conditions, a large scatter in the values of the
(power) law. When accurate determinations of torque difference M2 M1 exists due to the
the large-strain effective stress versus effective usual sources of experimental error.
strain relations are required for materials that ex- Multiple Testpiece Method. The shortcom-
hibit more complex behaviors, FEM or alternate ings of the differential method can be overcome
experimental methods must be used. The former by using multiple testpieces of increasing radius
approach (Ref 6567) involves inverse numeri- (Ref 59), at the cost, however, of increasing the
cal calculations and the application of standard complexity of the test. The multiple testpiece
optimization techniques. The alternate experi- method is based on the following derivative of
mental methods comprise: the torque integral with respect to the outer ra-
dius R at a given twist (), twist rate ( ), and
Thin-wall tube testing outer radius (Ro):
Differential testing
Multiple testpiece method M
= 2 p t Ro Ro2 (Eq 17)
These techniques are more accurate than the R q,q
methods discussed previously in this chapter for
metals subject to flow softening, dynamic re- Thus, the precise value of the current shear
crystallization, dynamic recovery, and so forth. stress at Ro can be determined from a knowledge
Generally, they are appropriate when the flow of the slope of the torque/radius curve at this
stress at a given strain depends on the tempera- radius:
ture and strain-rate history of the test, because
these techniques are based on deducing the
properties of a thin incremental layer of the sam- M
ple without requiring any prior assumptions R q,q
t Ro = (Eq 18)
about the material behavior. The thin-wall tube 2 pRo2
method was described previously in this chapter
and is not discussed here. The magnitudes of the corresponding strain and
Differential testing is a modification of the strain rate are well defined at this position and
Fig. 17 Shear-stress/shear-strain curve calculated di- thin-wall technique that requires the use of two are given by Ro /L and Ro /L.
rectly from experimental torque-twist curve
for electrolytic tough pitch copper at room temperature
solid samples of slightly different radius, R1 and The results of a series of experiments ana-
and 0.01 turns/s using the method of Fields and Backofen. R2. In contrast to the tube method, samples lyzed in this way are given in Fig. 18. Ten sam-
Source: Ref 36, 59 tested according to this technique are not subject ples of increasing diameter (5 to 10 mm, or 0.2
Fig. 19 Torque/radius data taken from the results of

Fig. 18 at 35 radians. The influence of the
degree of the log M versus R polynomial on the smooth-
ing of the curve is apparent. Source: Ref 59
(d)/c. The parameter is the fraction of

the deformation work per unit volume (d)
converted into heat, is the material density, and
c is the specific heat. The value of usually
ranges from 0.90 to 1.00; 0.95 is the value most
frequently assumed. The plastic work per unit
volume (d) is simply the area under the -
curve. Once the temperature increase (T) has
been estimated, the shear stress associated with
Fig. 18 Experimental torque/twist curves determined in torsion on annealed electrolytic tough pitch copper. Twisted the test strain rate and particular values of and
at room temperature at 0.01 turns/s. Source: Ref 59 T (equal to nominal test temperature plus T)
can be determined.
to 0.39 in.) were tested. These curves were used moment M by one of the methods described By conducting torsion tests at several temper-
to construct a set of torque/radius curves at se- previously. atures at a given adiabatic strain rate, a high-
lected intervals of twist; the relation for 35 rate isothermal torsion flow curve can be deter-
radians ( 5.6 turns) is illustrated in Fig. 19. Each
Deformation Heating mined by estimating the T values and plotting
curve was fitted by means of a log M versus R versus the actual temperature at various levels
polynomial, where a polynomial of degree 5 was When the surface-shear-strain rate in torsion of shear strain. This procedure is shown in Fig.
found to produce satisfactory results. The deriv- is less than approximately 0.01 s1, the - 20. For each level of , a smooth curve is drawn
ative of this polynomial gives the slope curves established from the measurement of through the data points. Equivalent isothermal
(M/R)|,, corresponding to a selected Ro torque and twist may be assumed to be repre-
(e.g., 3.5 mm, or 0.14 in., in Fig. 19), from sentative of isothermal (constant temperature)
which the value of Ro, can be found from Eq 18. behavior. Above this strain rate, thermal conduc-
The complete versus curve is obtained from tion is not sufficiently rapid to dissipate the de-
a series of such determinations at increasing an- formation heat into the specimen shoulders (for
gles of twist . typical metals, alloys, and specimen geometries)
Torque and Angle of Elastic Unloading. (Ref 69). Because of this, the - results are rep-
When the torsion testing of solid bars is carried resentative of flow behavior at temperatures that
out at ambient temperatures (i.e., on rate- increase with the amount of deformation.
insensitive materials and in the absence of Furthermore, because the temperature increase
anelastic effects), a torque cell is not absolutely T is a function of , a radial, as well as an
indispensible for the measurement of the mo- axial, temperature gradient can be developed in
ment or couple. The latter can be deduced from solid-bar specimens. For this reason, tubular
the elastic unloading angle per unit length */L, specimens, in which radial effects are negligible,
as determined from (Ref 68): are preferable for high-rate torsion tests.
The temperature increase that occurs during a
p q* torsion test is a function of the amount of heat
M= GR 4 (Eq 19)
2 o L generated and the amount of heat conducted into
the shoulders. The amount of heat conducted
where G is the elastic shear modulus of the ma- into the shoulders depends on specimen geome-
terial and Ro is the outer radius. By progres- try and requires numerical analysis.
sively applying measured amounts of twist During an extremely rapid test ( typically
and determining the unloading angle * at each greater than or equal to about 20 s1), heating is
step as the experiment proceeds, the full torsion essentially adiabatic. Under these conditions, all Fig. 20 Stress-temperature plots used to estimate
stress-strain curve can be deduced. Here, the of the deformation work converted to heat is equivalent isothermal high-strain-rate flow
curves. The intersections of the dashed lines with the flow-
accumulated strain is given by Ro/L, where assumed to be retained in the gage section, giv- stress/temperature curves yield estimates of the isothermal
, and must be derived from the ing rise to a temperature increase of T flow behavior at temperatures T1, T2, and T3.
Fig. 21 Finite-element-method predictions of (a) temperature rise at the sample midlength and (b) effective stress-strain curves from simulations of the torsion of aluminum
alloy 5252 under various heat-transfer conditions. In (b), the simulation results are compared to experimental data. (a) To 450 C (840 F), P 4.8. (b) To 500 C
(930 F), Dg 10 mm (0.4 in.), Dh 20 mm (0.8 in.), Lg 10 mm (0.4 in.). Source: Ref 72
flow curves are then obtained at various temper- no length change or axial stress induced. By radial direction of the sample in the sense oppo-
atures by reading points off the various curves contrast, when it is inclined either away from or site to that of the shear, so that the (111) planes
at these fixed temperatures. This procedure as- toward the fixed end of the specimen, lengthen- do not lie exactly in the transverse plane, and the
sumes that shear strain is a state variable, that is, ing or shortening of the sample, respectively, oc- 110 directions are not aligned exactly with the
that the shear stress is uniquely determined by curs (free-end conditions). Alternatively, under shear direction. This inclination is responsible
the value of , regardless of prior shear-strain fixed-end conditions, compressive or tensile for the compressive stresses (fixed ends) and
history. Over small intervals of and T, such stresses are developed, respectively. lengthening (free ends) observed in this strain
an assumption is probably sound from an engi- The inclination of the mean Burgers vector range (Ref 74, 75).
neering viewpoint. depends in turn on the texture that is developed At deformations of 2 to 4, the above compo-
Finite-element analysis may also be used to by twisting and, in particular, on the rotation of nent is replaced by the {001}110 orientation,
deduce the temperature nonuniformity in torsion the ideal orientations away from strict coin- which is rotated in the same sense as the shear.
tests that are neither purely isothermal nor adia- cidence with the axial (shear plane normal) and Such an inclination of this component also pro-
batic. For example, Zhou and Clode (Ref 7072) tangential (shear direction) directions of the duces a compressive force (or lengthening).
have applied such a technique for the analysis of sample. For example, when copper is deformed Finally, at strains greater than about 5, a steady
the torsion of aluminum alloy 5252 to predict at 100 to 300 C (212 to 570 F), the state of flow is attained, in which the
the temperature uniformity as well as constitu- {111}110/{111}110 component predomi- {111}110/{111}110 set again predominates,
tive behavior (Fig. 21), among other quantities. nates at strains of 1 to 2. This set of ideal orien- inclined in this case in the same sense as the
Heat losses, such as those that may occur during tations indicates that two particular (111) glide shear. This inclination is in the opposite sense to
torsion tests on high-temperature alloys using an planes lie in the transverse plane of the specimen the one observed at low strains and is responsi-
induction method for sample heating, have also (in different grains) and that these are oriented ble for the tensile stress that develops at large
been treated using the finite-element method so that certain 110 glide directions are aligned strains (Fig. 23). In a similar manner, specimen
(Ref 73). along the shear direction. Actually, however, this shortening is induced when testing is carried out
set of orientations is rotated slightly about the under free-end conditions.
Texture Development and Axial
Stresses During Torsion Testing
The axial stresses developed during fixed-end
torsion testing are related directly to the defor-
mation textures produced by straining (Ref
7485) as are the length changes observed dur-
ing free-end testing. The sign and magnitude of
these effects depend on the orientation of the
mean value of all the Burgers (i.e., slip) vectors
involved in a given increment of twist (Fig. 22).
The applied shear stress causes slip (shear) to
occur on a variety of crystallographic slip
(shear) planes lying either in the transverse
plane of the specimen or on planes that are in-
clined with respect to the transverse plane. Such
shear displacements are produced as multiples
of the elemental unit of shear, the magnitude and
direction of which are given by the Burgers vec-
tor. In most metals, the latter corresponds to a
single atom diameter. Fig. 22 Effect of the inclination of the mean Burgers vector b on specimen length change during simple shear de-
formation. (a) When the mean Burgers vector b is inclined away from the fixed end, specimen lengthening
When the average Burgers vector lies in the occurs. res is resolved shear stress; app is applied shear stress; z is change in specimen length. (b) When the mean
transverse (shear) plane of the specimen, there is Burgers vector b is inclined toward the fixed end, specimen shortening is observed.
The principal textures developed in face-

centered cubic (fcc) metals are listed in Table 2,
with their associated axial forces or length
changes. Some texture components, for example,
the {112}110/{112}110 set, do not lead to any
axial force or length change, even when the pole
figure is rotated about the radial direction away
from full symmetry with respect to the shear
plane normal (axial) and shear (tangential) direc-
tions. By contrast, other texture components, for
example, the {111}112/ {111}112 set, pro-
duce axial effects even when they are in the max-
imum symmetry position. Some examples of the
textures produced in copper that is twisted at 20
to 500 C (68 to 930 F) are illustrated in Fig. 24.
The textures developed in twisted body-cen-
tered cubic (bcc) metals differ somewhat from
those of fcc metals and are listed in Table 3. A
given bcc texture component of the form (hkl)
Dependence of the axial force on temperature in copper. e 5 103 s1 from 20 to 200 C (70 to 390
uvw produces an axial effect that is qualita-
Fig. 23 tively similar to that of an fcc component of the
F); e 5 102 s1 from 300 to 500 C (570 to 930 F). Source: Ref 74
form (uvw) hkl. Rigorously, the stress deviator
Table 2 Common texture components and axial forces observed during the torsion tensors for the bcc and fcc cases are related
testing of fcc metals through a simple symmetry operation. The twist-
Axial force(c) ing of bcc iron, which induces a compressive
Rotated in the sense opposite No Rotated in the same sense as force (lengthening) at small strains, as in
Type(a) Miller indices(b) to the shear rotation the shear the case of the fcc metals, is due to the inclina-
A {111} 110 C 0 T tion of the {011}111/{011}111 set in the
A {111} 110 C 0 T sense opposed to the shear. In a similar manner,
A*1 {111} 112 T T T the tensile stresses (shortening) developed at
{111} 112
A*2
B {112} 110
C
0
C
0
C
0
large strains (e 5) are associated with the
B {112} 110 0 0 0 {112}111 component, which in this case is ro-
C {001} 110 T 0 C tated in the same sense as the shear.
Crystal plasticity models have been quite
(a) The components A/A, A*1/A*2, and B/B are observed in pairs, as required by the symmetry of the torsion test. The C component is self-symmetric.
(b) The plane {hkl} is the crystallographic plane parallel to the macroscopic shear (transverse) plane; the direction uvw is the crystallographic di-
successful in accounting for the development of
rection parallel to the macroscopic shear (circumferential) direction. (c) C, compression or lengthening; T, tension or shortening. Source: Ref 75 deformation textures and for the axial effects
Fig. 24 Dependence of copper textures on temperature. (a) At 20 C (68 F), e 4.7. (b) At 100 C (212 F), e 4.65. (c) At 125 C (225 F), e 5.89. (d) At 150 C (300 F), e 10.85.
(e) At 200 C (390 F), e 31. (f) At 300 C (570 F), e 31. (g) At 400 C (750 F), e 31. (h) At 500 C (930 F), e 31. Same strain rates as in Fig. 23. Source: Ref 74
observed at room temperature as well as at ele- under strain paths that characterize forging, ex- evolution can be quantified by mechanisms such
vated temperatures prior to the initiation of dy- trusion, and so forth. as oriented nucleation and selective growth. In
namic recrystallization (Ref 76 79, 82, 83). When fcc metals (except aluminum) are de- the former mechanism, recrystallization nuclei
This comment applies to both fcc and bcc met- formed at elevated temperatures, dynamic re- are formed in those grains that have suffered the
als in which lengthening (free-end testing) or crystallization is initiated after a critical strain. least shear strain (i.e., dislocation glide).
the development of compressive stresses (fixed- In these instances, the texture changes have been Selective growth is assumed to occur for nuclei
end testing) is generally observed. Never- modeled for the case of free-end torsion (Ref of particular misorientations with respect to the
theless, it should be kept in mind that the spe- 8082). The textures that evolve are a result of matrix. With this approach, Toth and Jonas (Ref
cific textures developed during torsion are both dislocation glide and dynamic recrystalliza- 8082) demonstrated that torsion textures can be
usually not relatable to those that may occur tion. The effect of recrystallization on texture predicted quantitatively for copper (Fig. 25).
Specifically, it was shown that the {100}011
texture component is gradually replaced by the
Table 3 Common texture components and axial forces observed during the torsion {111}011 and {111}211 components. It is
testing of bcc metals this substitution that is responsible for convert-
Axial force(c)
ing the behavior from lengthening to shortening
(free-end) or from compression to tension
Rotated in the sense opposite Rotated in the same sense
Type(a) Miller indices(b) to the shear No rotation as the shear
(fixed-end) (Ref 84). Simulations of a compara-
ble nature have also been carried out for bcc ma-
D1 {112} 111 C C C terials (Ref 85).
D2 {112} 111 T T T
E {011} 111 C 0 T
E {011} 111 C 0 T Reduction of - Data to

F* {110} 001 0 0 0
To make use of shear-stress/shear-strain (-)
(a) The E/E orientations occur as a twin symmetric set, as required by the symmetry of the torsion test. The D1, D2, and F* components are self-
symmetric. (b) The plane {hkl} is the crystallographic plane parallel to the macroscopic shear (transverse) plane; the direction uvw is the crystallographic data for the prediction of load and the analysis of
direction parallel to the macroscopic shear (circumferential) direction. (c) C, compression or lengthening; T, tension or shortening. Source: Ref 75 metal flow in actual metalworking operations,
these quantities can be converted to effective
stress (
) and effective strain (e). These stresses

and strains are the equivalent quantities that
would result in an identical amount of deforma-
tion work in a state of uniaxial compression or
tension. For a given deformation level, then:
d
de
de (Eq 20)
where
-e is the uniaxial flow curve. In uniaxial
tension of a material with a random crystallo-
graphic texture,

, and e e. For uniaxial
compression,

, and e e.
In torsion, the relation between - and
-e
depends on the plasticity theory formulation
that is used. The most common concept, first
proposed by von Mises (Ref 86) and extended
by Shrivastava et al. (Ref 87), leads to the rela-
tions e /3 and
3. Note that the
concepts of effective stress and effective strain
are also approximations, because the
-e rela-
tions obtained from different mechanical tests
are often different, depending on the particular
material and test conditions. Nevertheless, for
load estimation, these constructs are sufficiently
accurate for engineering purposes. Unless oth-
erwise specified, the effective stress, effective
strain, and effective strain-rate definitions used
in this chapter are based on the von Mises for-
mulation.
Effective Stress for Free-End versus Fixed-
End Testing. The simple expressions for effec-
tive stress and strain given in the previous sec-
tion apply solely to the case in which sample
geometry does not change and no axial stresses
are developed during torsion testing. In practice,
a choice must be made between the fixed- and
free-end conditions of straining. Generally,
fixed-end conditions are preferred. If the free-
Fig. 25 Textures in terms of (111) pole figures and orientation distribution functions (ODFs) for copper deformed in
end method is chosen, the length can increase (at
torsion at 300 C to a shear strain of 11. (a) Measurements and (b) predictions based on the initial texture
and simulations of crystallographic slip and dynamic recrystallization. Isovalues on all plots are 0.8, 1.0, 1.3, 1.6, 2.0, ambient temperatures) or decrease (at elevated
2.5, 3.2, 4.0, 5.0, and 6.4. Source: Ref 80, 82 temperatures) by about 10 to 40% (Ref 12). The
derived shear stress , or

z, is proportional to warm working temperatures. Only at 20 C (68 and a relatively small strain-rate sensitivity (as
M/R3 (see above), where M is the measured F) and at 200 C (390 F) and a von Mises sur- shown by the initial portions of the flow curves,
torque and R is the current sample radius. In face effective strain rate of es s/3 10 in which thermal effects are unimportant).
most tests, the radius change is not monitored, s1 do the flow curves exhibit softening. Under Isothermal flow curves for 10 s1 can be de-
and the initial value is used instead. Thus, a these conditions, scribe-line measurements indi- duced by estimating the associated T values
length change of 10%, which involves a radius cate that flow localization had occurred prior to and by constructing
-T plots. This leads to
change of about 5%, leads to an error in the de- fracture. All other scribe-line observations indi- isothermal high-strain-rate flow curves that are
rived shear stress
z of about 15%. This error cate fracture-controlled failure. consistently above the lower-strain-rate flow
increases with the 3/2 power of the length In contrast to these flow curves are those ob- curves.
change and can become quite significant. tained at hot working temperatures, all of which In contrast to the trends at cold and warm
By contrast, if the length is held constant, reveal maxima followed by decreasing flow working temperatures, the type 304L flow re-
axial stresses,
zz, develop, which range from 2 stresses and eventually steady-state flow. This sponse in the hot working regime reveals a
to 20% of the developed shear stress
z. The trend is indicative of dynamic microstructural noticeable strain-rate effect. Under these condi-
axial stress
zz(R) at the surface of a solid tor- changes characterized by recrystallization phe- tions, the high-strain-rate curves are consider-
sion bar sample can be estimated from the mean nomena. However, scribe-line measurements ably above their low-strain-rate counterparts at a
stress
m F/R2 by using the formula derived confirmed that, despite the flow-stress decrease, given test temperature. Such a response is the
in Ref 74: failure was fracture controlled in all cases, due result of the high strain-rate sensitivity of most
to the stabilizing influence of the increased metals at hot working temperatures. The strain-
F 1 ln F 1 ln F strain-rate sensitivity. rate sensitivity effect offsets any possible
s zz ( R) = 1+ + (Eq 21) The flow curves for type 304L stainless steel crossover due to deformation heating at the
pR 2 2 ln N 2 ln N
shown in Fig. 26 also indicate the influence of higher strain rates. Flow stresses in this temper-
strain rate on flow behavior. At cold and warm ature regime are much lower than those at cold
where F is the axial force, and N and N are the
working temperatures, strain rate has only a and warm working temperatures. Because of
number of revolutions and twist rate, respec-
slight effect on flow response. In fact, the 10 s1 this, T values associated with the higher strain
tively. This relation only applies rigorously to
curve at a given temperature eventually drops rate, which vary with the magnitude of and ,
materials displaying simple hardening laws, as
discussed previously in the section on the Fields below the curve measured at 0.01 s1. The lower tend to be smaller at the higher temperatures.
rate can be considered isothermal and the higher Flow-stress data from the torsion test for other
and Backofen relation for deriving the shear
rate adiabatic. Thus, the crossover of flow curves alloys (such as Waspaloy, Fig. 27) exhibit simi-
stress
z (referred to as ), from the moment M.
at the two rates is a result of deformation heating lar variations with temperature and strain rate.
When
zz and the circumferential stress

are nonzero (the radial stress
rr can be taken to
be zero at the surface of a torsion specimen), the
following relation applies for effective stress
(Ref 74):
1/ 2
[
s = 3s q2z + (s qq
2 2
+ s zz ]
- s qq s zz ) (Eq 22)
Detailed calculations (Ref 74, 75) show that the

correction associated with the inner bracketed
term on the right side of Eq 22 is only about 1%
for fcc metals and attains a maximum of about
8% for bcc metals, which tend to develop higher
ratios of axial to transverse shear stress. Because
the magnitude of the correction depends on the
square of the
zz/
z ratio, where the latter is al-
ways 1, the error introduced by neglecting
zz in the calculation of the effective stress is

generally small.
Typical Flow Curves

Flow-stress results for type 304L austenitic
stainless steel at two strain rates and a variety of
cold, warm, and hot working temperatures are
shown in Fig. 26. Note that the shear-strain lev-
els in these curves are about 5, corresponding to
von Mises effective strains in excess of 2.5 or to
values much greater than those obtained in ten-
sion or compression tests. All tests shown in Fig.
26 were taken to failure. However, the flow
stresses for the hot working tests are only shown
for 5.
The flow-stress behavior observed in torsion
is best interpreted in terms of test temperature
and strain rate. For the type 304L alloy (Fig. 26),
the flow stresses generally increase with strain Fig. 26 Flow curves from type 304L stainless steel torsion tests. (a) Cold and warm working temperatures. (b) Hot
(or remain constant with strain) at cold and working temperatures. Source: Ref 88
Fig. 28 Correlation of torsional flow stress data for a

0.25% C steel using a temperature-compen-
sated strain-rate parameter (the Zener-Hollomon parame-
ter, Z ). Source: Ref 89
stress data using the von Mises relations. In this

form, they are useful for load-prediction and
metalworking analysis. However, even when
converted to effective terms, flow-stress data ob-
tained from torsion tests do not always show
perfect agreement with data from other mechan-
ical tests.
The differences are usually greatest at cold-
working temperatures. They are a consequence
of one or a combination of several factors, such
as the effect of deformation path (tension, com-
pression, etc.) on the development of crystallo-
graphic texture (Ref 9092) and on the nature of
the microscopic slip or twinning processes that
control the observed macroscopic strain-harden-
ing rate. The variability of such factors is less
likely at hot working temperatures because of the
dynamic restorative processes that may prevent
Fig. 27 Flow curves for Waspaloy. (a) Effect of temperature at a fixed effective strain rate of 1 s1. (b) Effect of strain sharp textures from being formed. Also, a signif-
rate at a fixed test temperature of 1038 C (1900 F). Flow softening at the higher temperature is a result of
dynamic recrystallization. Source: Ref 48 icant amount of deformation occurs at grain
boundaries as well as in the matrix of metals. The
At cold and warm working temperatures, strain energy for the flow process involved, R is the gas stress obtained in torsion often lies below that ob-
hardening often persists to large strains, except constant, and e /3
. For many single-phase tained by axisymmetric methods. Such differ-
for high-rate tests, which are characterized by materials, the relationship between Z and the ences have been noted at low (Ref 9294) as well
deformation heating and a general decrease in steady-state values of is of the form: as hot working temperatures (Ref 32, 59, 96, 97).
flow stress with temperature. At hot working Nadai and Davis (Ref 95) used an expression for
temperatures, flow curves frequently show flow- ( A sinh 3a t) m = Z (Eq 23) equivalent strain similar to that of Eichinger (Ref
stress maxima followed by flow softening and a 98) that reduced the discrepancy.
steady-state flow stress. This behavior is associ- Sakai and Jonas (Ref 99) have noted that the
in which A, , and m are material constants. An
ated with a variety of microstructural changes. strain to the onset of dynamic recrystallization is
example of the fit for a 0.25% C steel is shown
Furthermore, the strain-rate sensitivity tends to delayed in torsion when compared with com-
in Fig. 28.
be small (m 0.02), except under hot working pression test results at the same equivalent strain
conditions, at which the value of m is typically rate. The maximum stress, however, does not
between 0.1 and 0.3 for many metals and alloys. Comparison of Torsion Flow-Stress differ markedly between the two testing meth-
The steady-state flow stress under hot work- Data to Data from Other ods. Inasmuch as a critical level of work harden-
ing conditions measured in torsion tests (and a Workability Tests ing is thought to be required to initiate dynamic
variety of other mechanical tests) frequently is a recrystallization at a given strain rate and tem-
function of the Zener-Hollomon parameter, Z Shear-stress/shear-strain data from torsion perature, an increase in the critical strain (by a
e exp (Q/RT), where Q is the apparent activation tests are often converted to effective flow- factor of 2 to 3 times) can be understood if the
Fig. 29 Comparison of effective stress-strain curves determined for type 304L stainless steel in compression, tension, and torsion. (a) Cold and warm working temperatures. (b)
Hot working temperatures. Source: Ref 100
work-hardening rates in torsion differ from strain are employed, the differences between the Figure 32 illustrates results for a material with
those in tension and compression. Canova et al. curves cannot be eliminated. However, an esti- a crystal structure different from the fcc materi-
(Ref 59) have shown that the material flow in the mate of the working loads can still be derived als in Fig. 30 and 31. These effective stress-
surface of a torsion bar follows the pattern of from torsion data plotted in von Mises terms. In strain curves are for bcc carbon steels that were
simple shear. These observations are consistent contrast to the 20, 400, and 800 C (68, 750, and tested in torsion and tension. To achieve the high
with the conclusion that the state of strain in tor- 1470 F) behaviors, comparison of type 304L deformation levels in tension, samples were pre-
sion is one of simple shear (Ref 87). It can then torsion data to tension and compression data is strained by wire drawing. The von Mises flow
be argued (Ref 59) that at large strains disloca- quite good at the hot working temperature of curves for torsion lie below those for tension for
tion motion would tend to be restricted to trans- 1000 C (1830 F) (Fig. 29b). This is most likely
verse planes in the torsion bar and that the active a result of the absence of marked textural and
Burgers vectors would be close to the circumfer- strain-hardening effects.
ential direction. Accordingly, strain is essentially Other metals show the same divergence be-
accumulated by macroscopic slip on a single set tween torsion flow curves and those obtained by
of parallel planes, and the work-hardening rate other test techniques. Data for copper and alu-
should be lower than in tension or compression, minum (tested at room temperature) are given in
where two or more macroscopic slip planes are Fig. 30 and 31. Comparisons with tension and
active. This view adequately accounts for the plane-strain compression results are given in
discrepancies between the flow behavior of ma- Fig. 30 and 31, respectively. The effective stress-
terials in torsion and in tension and compres- strain curves from torsion show consistently
sion, and it preserves the validity of the expres- lower levels of strain hardening as well as lower
sions for equivalent stress and strain. It raises flow stresses.
serious questions, however, as to the true equiv-
alency of alternative testing procedures when the
microstructural state is of prime importance, as
it is in hot working. It also has impact on the
question of hot-ductility determination.
In Fig. 29, torsion flow-stress data for type
304L are compared to compression and tension
data in terms of von Mises effective stress and
strain. At cold and warm working temperatures,
as well as at low hot working temperatures (800
C, or 1470 F), the flow curves from the vari-
ous tests do not coincide. Generally, there is a
lower level of strain hardening in torsion. Thus,
although the overall stress levels are similar, the
actual shapes of the curves are quite different. Fig. 30 Flow curves determined at room temperature Fig. 31 Comparison of room-temperature flow curves
in tension and torsion on oxygen-free high- from torsion and plane-strain compression
Even if other definitions of effective stress and conductivity copper. Source: Ref 87 tests on copper and aluminum. Source: Ref 32
both the low-carbon steel and the pearlitic, near- Interpretation of Torsion men radius could be rationalized in terms of the
eutectoid high-carbon steel. Fracture Data associated change in surface-to-volume ratio
Also shown in Fig. 32 are effective stress- (Ref 21).
strain curves from torsion tests that were calcu- In tubular specimens, it has been found that
lated on the basis of the Tresca criterion, in The fracture strains measured in torsion are the bore diameter and wall thickness do signifi-
which the effective stress is equal to 2 and the highly reproducible and provide a good quanti- cantly influence the measured ductility (Fig. 34).
effective strain to /2. A divergence between tor- tative measure of the ductility of materials un- This observation points to an effect that is possi-
sion and tension is still present. dergoing high-strain deformation. Frequently, bly the source of the enhanced ductility some-
The differences between torsion flow stresses torsion tests are conducted to determine the ef- times observed in torsion: the constraint of the
and results from other tests described previously fects of process conditions (e.g., strain rate and solid core. In thin-wall tubes, cracks generated
have much less of an effect on load predictions temperature) and material composition on work- at the outer surface, at the highest stress, only
than on the estimation of formability such as in ability. A large number of hot ductility tests have have to grow a distance equal to the wall thick-
sheet stretching. In these operations, workability been performed successfully on a variety of ma- ness to produce complete failure. This occurs
is controlled by the onset of instability and flow terials over a wide range of temperatures and rapidly because the stresses are generally high
localization, which are influenced strongly by strain rates (Ref 3, 1114, 24, 27, 2932, 35, 37, throughout the wall. By contrast, surface-initi-
the rate of strain hardening, as expressed by the 40, 45, 55, 91, 102108). It is not the purpose of ated cracks in solid-bar specimens must grow
strain-hardening exponent, n. Because the n val- this section to review the test results obtained in along the entire radius, under the influence of
ues obtained from torsion tests are typically these studies, but rather to discuss the test pro- stresses that decrease markedly with distance.
below values obtained from the other tests, they cedures, some typical results, and their applica- The central core of a solid specimen, therefore,
should be used with caution in formability tion. For extensive reviews of these and other acts to inhibit complete failure and thereby en-
modeling. data, the reader is referred to a number of excel- hances ductility.
lent reviews on the topic (Ref 22, 89, 109, 110). Variations with specimen geometry may also
occur when experiments are run at moderate
Effect of Specimen Size strain rates or at rates at which specimen geom-
etry affects the temperature profile due to heat
For round-bar samples, the surface strain at conduction. In these cases, the fracture strain,
fracture is usually relatively independent of initial test temperature, and estimated tempera-
specimen design (e.g., length-to-radius ratio), as ture at fracture should be reported together.
shown by the results from torsion tests at hot
working temperatures for type 304 stainless
steel and aluminum given in Fig. 33. Effect of Temperature and Alloying on
In another study of the hot ductility of ferrous Torsional Ductility
alloys, Reynolds and Tegart (Ref 14) reported
that the number of twists to failure increased In torsion, the ductility of most metals is
with radius when specimens of constant length- moderate at cold working temperatures (f 1
to-radius ratio were used. It has been shown in a to 5), least at warm working temperatures, and
number of studies (Ref 27, 45, 55) that ductility greatest at hot working temperatures, at which
in torsion is determined by the growth rate of at fracture often exceeds 10 or more. At cold
surface cracks that propagate inward as straining working temperatures, fracture occurs by ductile
proceeds. Because surface oxidation promotes fracture initiated at second-phase particles and
initiation and propagation of cracks at elevated inclusions. The decrease in ductility at warm
temperatures, the change in ductility with speci- working temperatures occurs because of the
Fig. 32 Flow curves determined via torsion testing

and tension testing (following wire drawing).
(a) 0.06% C steel. (b) 0.85% C steel (in pearlitic condi-
tion). Note that the torsion data are expressed in terms of
both the von Mises and the Tresca effective stress-strain
definitions. Source: Ref 101
Fig. 33 Effect of the gage length-to-radius ratio on the

effective strain to failure (ef) in torsion tests. Fig. 34 Effect of bore diameter on strain to failure (ef) of type 304 stainless steel specimens with a 7 mm (0.3 in.) gage
Lines join results at similar strain rate and temperature. diameter and a gage-length-to-radius ratio of 2:1 (lines join data points obtained at similar strain rates and
Source: Ref 35 temperatures). Source: Ref 35
thermal activation of grain-boundary sliding, for forging or extrusion. Above this temperature, ified, it is often used to document the effect of
which culminates in brittle intergranular fail- ductility may decrease because of effects such as temperature and to establish optimal working
ures. At hot working temperatures, dynamic grain growth, deformation heating, or incipient conditions. Examples of this type of torsion data
restoration processes such as recovery and remelting (hot shortness). Increases in grain size obtained from hot torsion experiments are shown
crystallization act to heal incipient voids (due to promoted by very high temperatures increase the in Fig. 36 for several steels and nickel alloys.
cavitation) and microcracks, thereby increasing tendency for intergranular fracture, while re- As shown in Fig. 36, carbon and alloy steels
ductility substantially. tained deformation heat at moderate to high such as 1040 and 4340 are very workable. These
These processes are most rapid in single- strain rates may be sufficient to increase the steels fall into group VII of Fig. 35. As the tem-
phase pure metals, which tend to have the largest instantaneous local temperature above the incip- perature is decreased, a two-phase ferrite plus
ductilities at hot working temperatures. Alloys ient melting or solidus temperature, thus result- austenite structure ( ) is formed from the
with solute elements tend to have lower ductili- ing in poor ductility. single-phase austenite (), causing a sharp de-
ties because of the increased difficulty of dy- Figure 35 summarizes the qualitative depend- crease in ductility. This is also shown for a high-
namic restoration. Moreover, in such materials, ence of workability on temperature for a wide oxygen iron alloy in Fig. 37.
the onset of true hot working conditions occurs range of alloys. These trends are classified by Types 304 and 410 stainless steels are gener-
at higher temperatures than in pure metals. This the number of phases, the types of compounds ally single-phase alloys at hot working tempera-
is also true for alloys with two or more phases in that are present at various temperatures, and so tures and exhibit increasing ductility with tem-
which deformation tends to be inhomogeneous on and are a valuable reference tool when new perature. However, at very high temperatures,
because of the variation of properties between alloys for which quantitative data are unavail- ferrite is formed in type 410, causing a drop in
the two phases, which results in poor ductility. able are being formed. workability. Therefore, type 304 exhibits a
In the hot working regime, ductility does not Quantitative fracture data from torsion tests behavior like a group I material, and type 410
increase monotonically with temperature up to are usually reported in terms of twists to failure behaves like a group V alloy in Fig. 35.
the melting or solidus temperature. Frequently, or surface fracture strain. Although using twists
ductility passes through a maximum, which usu- to failure as a measure of ductility is somewhat
ally correlates well with the optimal temperature qualitative unless the specimen geometry is spec-
Fig. 36 Ductility determined in hot torsion tests.

Source: Ref 112, 113
Fig. 37 Variation of ductility with temperature for a

Fig. 35 Typical workability behaviors exhibited by different alloy systems. TM melting temperature. Source: high-oxygen Swedish iron tested in torsion at
Ref 111 an effective strain rate of 0.5 s1. Source: Ref 14
The nickel-base superalloys in Fig. 36 show higher. This trend is the result of a decrease in mal working temperature may be shifted. An ex-
inferior ductility compared to types 304 and 410 the volume fraction of manganese sulfide and ample of this effect is shown in Fig. 41 for the
steels due to the generally high alloy content of oxide inclusions. nickel-base superalloy Udimet 700. For this ma-
the former and the formation of hard, second- Quantitative analysis of the effect of sulfides terial, the ductility maximum is shifted to higher
phase particles at low to moderate hot working on the workability of steels is shown in Fig. 39, temperatures with increases in strain rate. This
temperatures in these materials. These alloys ex- in which the ductility of several rimmed steels is behavior occurs because the alloy is single phase
hibit group VIII behavior. Because the second presented as a function of sulfide content. As in- at the temperatures under consideration, and the
phase is formed at relatively high temperatures, clusion content increases, the fracture strain de- flow stress of single-phase alloys is a function of
the superalloys must be worked close to the creases and thus workability during an actual the Zener-Hollomon parameter (Z e exp
melting point. Hot torsion data in this range forming operation is affected deleteriously. (Q/RT )). Thus, as the strain rate is increased,
often show a decrease in ductility with tempera- Figure 40 gives the torsional ductility of pure similar flow and fracture processes should occur
ture because of incipient melting problems. aluminum and a series of aluminum-magnesium at correspondingly higher temperatures. As
Several other alloying effects in steels and alloys as a function of temperature. At the test- shown in Fig. 41, however, the peak ductility is
aluminum alloys are illustrated in Fig. 38 to 40. ing temperatures used, magnesium is totally in not as large at high strain rates. This can be as-
Figure 38 shows the effects of carbon on the tor- solution. The observed trend is caused by in- cribed to the increased amount of deformation
sional ductility of plain-carbon steels and elec- creased amounts of grain-boundary sliding in heating and the temperature increase at these
trolytic iron-carbon alloys at 0.5 TM. For both the magnesium alloys as well as by a decrease in rates, which raise the alloy to its incipient melt-
types of alloys, workability is controlled by duc- the kinetics of the dynamic softening processes. ing point.
tile fracture due to cavitation at precipitates and The latter leads to an increase in flow stress, The influence of strain rate on torsional duc-
inclusions. thereby reducing the ability of the alloy to ac- tility is also illustrated in Fig. 42(a) for type
For the plain-carbon steels, the ductility in- commodate stress concentrations at inclusions 304L stainless steel, in which the ductilities
creases initially with carbon content because of and grain boundaries. from tests conducted at von Mises surface effec-
a decrease in the number of oxide inclusions, tive strain rates of 0.01 and 10.0 s1 are plotted.
which serve as void nucleation sites. Similarly, Effect of Strain Rate on The values of sf from the lower rate test are
with carbon contents above 0.1%, ductility de- plotted versus the (constant) test temperature at
creases with carbon content as a result of the in-
Torsional Ductility which they were conducted. The high-strain-rate
creasing volume percent of carbides at which The effect of strain rate on workability should ductilities are plotted versus the estimated tem-
voids may also nucleate. The variation of ductil- be considered when attempting to apply labora- perature at fracture. These estimated tempera-
ity with carbon content is similar for the elec- tory torsion measurements to metalworking tures are equal to the nominal test temperature
trolytic alloys, although the level is generally processes carried out at much higher speeds. In plus the adiabatic T from the plots shown in
many cases, the general dependence of ductility Fig. 42(b). Ductility data for high-strain-rate
on temperature will be similar at different strain tests performed at 20 C (68 F) and 200 C (390
rates. However, the ductility peak and thus opti- F), in which flow localization occurred, are also
shown. In these two instances, sf was obtained
from scribe-line measurements, and the T ver-
sus plot was extrapolated to obtain the fracture
temperature.
The low-strain-rate data for type 304L shown
in Fig. 42(a) illustrate a classical dependence on
temperature, that is, a modest ductility at cold
working temperatures and a ductility minimum
at warm working temperatures. The major effect
of the higher strain rate is a translation of the
lower-strain-rate data to higher temperatures.
For example, the ductility minimum appears to
Fig. 38 Influence of carbon on the ductility of bcc iron be shifted.
tested in torsion at 0.03 s1 and 650 C (1200 A more detailed interpretation of low- and
F), or one-half the absolute melting point. Source: Ref 21 high strain-rate-data can be made through
the Zener-Hollomon parameter, Z e exp
(Q/RT) { exp (Q/RT)}/3 . For this alloy,
Fig. 40 Effect of the amount of magnesium in solid so-

lution in aluminum on torsional ductility.
Fig. 39 Relation between manganese sulfide content Tested at an effective strain rate of 2.3 s1. With 5% Mg,
and ductility in hot torsion tests on a variety of ductility is severely reduced over the entire temperature Fig. 41 Effect of test temperature on the torsional duc-
rimmed steels. Source: Ref 31 range. Source: Ref 29 tility of Udimet 700. Source: Ref 114
the steady-state flow stress at equal values of Z to be the same as the activation energy from Correlation of Torsional Ductility Data
is nearly the same. Therefore, equal failure flow-stress data (98.2 kcal/mol, or 411 kJ/mol). to Other Workability Data
strains can be expected at fixed values of Z. This The results of the calculation are shown in Fig.
hypothesis may be checked by translating the 42(a) and appear to be in fair agreement with the Because metalworking processes are not car-
low-strain-rate fracture locus fa (Tfa) to fb(Tfb) measured high-strain-rate fracture locus. ried out under a state of pure torsional or shear
defined by fb fa and b exp (Q/RT) a Nicholson et al. (Ref 115) corrected their loading, it is often necessary to convert the
exp/(Q/RTfa), where b/ a 10/0.01 103. For ductility-temperature curves for deformation- workability parameter measured in torsion to an
these type 304L stainless steel results, Q is taken heating effects in a similar fashion (Fig. 43). index that is compatible with other deformation
modes. Previously, attempts were made to corre-
late torsion, tension, and other types of data
using the effective strain concept. Although a
definite relationship exists between torsion and
tension effective fracture strains (Fig. 44), such
a method is incapable of explaining how free-
surface fracture is avoided in homogeneous
compression by preventing barreling. For frac-
ture to occur, deformation must also involve ten-
sile stresses to promote ductile fracture, wedge
cracking, or some other failure mechanism.
One of the most successful hypotheses incor-
porating the effects of deformation and tensile
stress is that proposed by Cockcroft and Latham
(Ref 116). They postulated that fracture occurs
after the maximum tensile stress (
T) carries out
a fixed amount of work through the applied ef-
fective strain, or:
ef
0
s T de = a constant (C ) (Eq 24)
The constant in the relation is a function of ma-

terial, purity, and test temperature. In a torsion
test, the maximum tensile stress occurs at 45 to
the torsion axis and is equal to . As mentioned
previously, e for torsion is equal to /3 . If the
material under consideration exhibits power-law
strain hardening, Kn, then C is equal
to (K n1sf
)/(3 (n 1), where sf is the shear
fracture strain. If the material does not harden in
a power-law manner, the integration to evaluate
C at fracture can be performed graphically.
In uniaxial tension,
T is the axial tensile
stress, and e is the axial strain. Upon necking,
T
is higher than the effective flow stress by a cor-
rection factor that can be estimated from the
work of Bridgman (Ref 117). This factor is
on the order of 0 to 33% for true axial strains
of 0 to 2.5 in the necked region. For a power-
law-hardening material,
Ken. For tension,
C (1 CF) {Kenf+1/(n 1)}, where CF
is the average correction factor and ef is the axial
fracture strain determined by reduction-in-area
measurements and the constant volume assump-
tion of plastic flow.
To apply the Cockcroft and Latham criterion
to other deformation modes, the critical value
of C must be known, and the maximum tensile
stress and effective strain must be estimated.
For arbitrary deformation paths, estimating
these quantities can be very difficult and often
requires sophisticated mathematical tech-
niques, such as finite-element methods. Once

T and e are known, however, numerical tech-
Fig. 42 (a) Fracture-strain data from type 304L austenitic stainless steel torsion tests and (b) estimated temperature niques can be used to estimate the value of the
changes during high-rate tests. Low-strain-rate (e 0.01 s1) data are plotted versus the actual test temper-
ature. High-strain-rate (10.0 s1) data are plotted versus temperatures estimated from (b) a T- plot and the nominal maximum work integral, which can then be
test temperature, which is shown beside each data point. Source: Ref 88 compared with the critical value of C at frac-
ture determined by a simple workability test

such as the torsion test.
The Cockcroft and Latham criterion has
found its greatest success in correlating tensile
and torsion fracture strains. The degree of suc-
cess in these instances is judged by comparing
the critical values of C established from flow
and fracture results for the different tests. An ex-
ample of such a comparison for type 304L stain-
less steel is given in Table 4. At 20 and 400 C
(68 and 750 F), the material strain hardens in a
power-law manner, enabling closed-form inte-
gration to obtain C values. At the two higher
temperatures of 800 and 1000 C (1470 and
1830 F), a graphical integration procedure
using flow curves (Fig. 29) was used. The inte-
gration procedure was straightforward for
torsion, because deformation was uniform to
fracture.
For tension, in which necking precludes ob-
taining flow stress data to fracture, the initial
Fig. 43 Effect of strain rate on temperature dependence of fracture strain in torsion for type 321 stainless steel. The
portion of the tensile curves was extrapolated to
influence of adiabatic heating is shown by the corrected curves. After Nicholson et al. (Ref 115)
the required large strains using the compression
and torsion curves as models. Results of the C
parameter calculations for a given test tem-
perature agreed within 10% except at 800 C
(1470 F), where the two values differ by ap-
proximately 15%. The magnitudes of these
differences are considered typical and accept-
able, particularly because a gross continuum
model was used for a mechanism that greatly de-
pends on the presence of microstructural fea-
tures such as inclusions. The same continuum
model was successful over a wide range of tem-
peratures, within which the fracture mechanism
may vary significantly.
Attempts to include microstructural features
and fracture mechanisms have met with less suc-
cess than the Cockcroft and Latham formula-
tion. One such attempt is that of Hoffmanner
(Ref 118), who proposed a model that incorpo-
rates a factor describing the dependence of frac-
ture strain on the magnitude of the normal stress
perpendicular to the mechanical texture. Such a
concept may have application in the analysis of
fracture in torsion, during which the mechanical
texture rotates with respect to the stress axes.
Because of the stress-state dependence of
ductile failure, the axial stresses that sometimes
develop during torsion may affect ductility. The
ratio of the axial to the shear component of
stress seldom exceeds 0.4, so the axial compo-
nent has a limited influence on the effective
stress. The axial stress, however, plays a signifi-
cant role in determining the ductility measured
by the shear or effective strain to failure. This
has been clearly demonstrated by external appli-
cation of longitudinal tension or compression
during torsion (Ref 40, 45, 91, 105); data for
Inconel 600 are shown in Fig. 45. When the ratio

A/ is negative (compressive axial stress), the
shear strain at fracture increases almost linearly
with the magnitude of the ratio. In the regime of
positive axial stresses, the curves show a marked
upward curvature, which may, in part, reflect the
Fig. 44 Relation between effective fracture strain from tension and from torsion tests for several alloys. Source: development of geometric instabilities as the
Ref 114 shear-to-normal stress ratio typical of tensile
Table 4 Cockcroft and Latham criterion for annealed type 304L fracture data
Strength
Strain at fracture Strain- Average
coefficient,
(torsional shear hardening correction
Test temperature G or K C parameter
Deformation or tensile), exponent, factor,
C F mode sf or f MPa ksi n CF MPa ksi
20 68 Torsion 4.01 517 75 0.25 ... 1360 197

Tension 1.26 1172 170 0.352 0.075 1275 185
400 750 Torsion 5.17 310 45 0.137 ... 1020 148
Tension 1.51 772 112 0.307 0.105 1120 162
800 1470 Torsion 5.89 ... ... ... ... 445 64.5
Tension 1.94 ... ... ... 0.12 525 76.1
1000 1830 Torsion 8.33 ... ... ... ... 215 31.5
Tension 2.75 ... ... ... 0.14 235 34.1
Fig. 45 Ratio of maximum axial stress to maximum
Note: strain rate 0.01 s1. Source: Ref 100
shear stress versus values of shear strain at fail-
ure for Inconel 600. Source: Ref 45
testing is approached. Although it was demon-

strated that the
A/ ratio influences initiation of
cracks, the dominant influence is through the
propagation phase. Dragon and his coworkers,
(Ref 91, 105) have considered these and other
observations and have argued that the influence
of axial stress on ductility can be represented by:
eT e0(1

)
A (Eq 25)
where eT is the total effective strain at failure on

the application of an axial compressive stress

A, e0 is the strain at fracture in the absence of
an axial stress, and is a material constant that
ranges from 2.5 to 3 for carbon steel, and from
10 to 20 for Inconel 600 at 900 C (1650 F),
These authors suggest that e0 is a measure of the
intrinsic ductility of the material. Perhaps
ductility in simple shear would be a more ap-
propriate term, but the effects of the core con-
Fig. 46 Comparison of optimal ductility temperature determined in hot torsion tests at 0.5 s1 with operating tem-
straints mentioned previously are not considered peratures for piercing, rolling, and forging of carbon steels. Dashed curve is for torsion; closed circles,
in this definition. Mannesmann tube mill; open circles, Stiefel tube mill; closed triangles, slabbing mill; open triangles, general rolling mill;
closed diamonds, forging. Source: Ref 120
Correlation of Torsional Ductility

Data with Working Practice
points in Fig. 46). The hot torsion test is shown to process variables that contribute to the forma-
The hot torsion test has been used extensively be particularly effective in predicting the optimal tion of shear bands during metalworking. Either
to grade materials and assess optimal tempera- temperature for rotary piercing, which has been solid or tubular specimens can be used for test-
tures for particular working operations. Studies confirmed by other authors (Ref 3, 14). By con- ing. When solid specimens are used, however,
of the mechanisms that control hot ductility at trast, this test tends to overestimate the ideal tem- data analysis and interpretation are usually more
high strains have met with reasonable success at perature for forging and extrusion. This discrep- complex than when tubular specimens are used.
relatively low temperatures for materials that un- ancy has been attributed in part to a propensity This difficulty arises because of the variation of
dergo dynamic recrystallization (Ref 27, 89, for deformation heating in torsion and piercing strain, strain rate, and (at moderate to high strain
119). They have been less successful for materi- where high shear rates occur in a relatively small rates) temperature across the specimen diameter.
als that undergo only dynamic recovery during volume of material (Ref 109). The low-tempera-
straining (Ref 89, 119). As yet, there remains no ture limit for a particular working operation cor- Observation of Flow Localization in
systematic basis for translating these basic con- relates with a specific total effective strain in tor- Torsion
cepts into a framework for workability predic- sion. Reynolds and Tegart (Ref 14), for example,
tions. On the other hand, measurement of hot showed that an effective strain of between 3 and The development of flow localization during
ductility in torsion can successfully provide esti- 4 at e 1 s1 is necessary in torsion testing to torsion is usually detected by scribe-line meas-
mates of optimal working temperatures for sev- ensure successful extrusion, whereas e 18 is urements, variations in microstructure along the
eral working operations. Zidek (Ref 120) has car- required for successful rotary piercing. gage length, or torque-twist behavior. The first
ried out a relatively complete survey of the two of these techniques are unequivocal.
relation between torsional ductility and hot work- Examples of flow localization detected by these
ing performance for carbon steel. His data are Measuring Flow-Localization- two methods for the shear bands developed in
summarized in Fig. 46. These data confirm a rea- Controlled Workability tubular type 304L stainless steel torsion speci-
sonable correlation between the temperature of mens twisted at es 10 s1 at room temperature
maximum ductility in torsion (dashed line in Fig. Because torsion consists essentially of defor- are shown in Fig. 47 and 48.
46) and the optimal working temperature for a mation in simple shear, torsion testing is fre- The use of the torque-twist behavior, on the
variety of practical working processes (data quently used to determine the material and other hand, to measure flow localization may be
basic instability of the acicular microstructure

was deduced to be the main reason for the local-
ization behavior. For either case, however, a lo-
calization analysis that deals with the problem as
a process rather than as an event is required to
fully understand such observations.
Flow Localization Analyses
Fig. 49 Comparison of experimental and theoretical The main variables that control flow localiza-
torque-twist curves for (equiaxed alpha) tion during torsion are material properties (me-
microstructure Ti-6242Si hot torsion specimens. Tested at
e 0.9 s1; T 913 C (1675 F). Source: Ref 52 chanical and thermal), twist rate, and the presence
of material or specimen imperfections. Of the ma-
terial properties, strain- and strain-rate-hardening
rates, temperature sensitivity of the flow stress,
and thermal conductivity are the most important.
Twist rate affects heat conduction and heat trans-
fer. Material or torsion specimen imperfections
offer sites for the initiation of flow localization.
Such sites are particularly important when torsion
is conducted at very low or very high rates, at
which the axial temperature field is relatively uni-
Fig. 47 Failed type 304L stainless steel torsion speci- form. Under these circumstances, temperature
mens from e 10 s1 tests showing evidence gradients and heat-transfer phenomena, which
of flow localization. (a) 20 C (68 F), average s 1.3. (b) serve as a prime source of flow localization, are
200 C (390 F), average s 2.9. Magnification: 2.
Source: Ref 88 absent. The interpretation of torsion data under
Fig. 50 Comparison of experimental and theoretical flow-localization conditions is discussed subse-
torque-twist curves for (Widmansttten quently for two cases. In one, heat-transfer effects
alpha) microstructure Ti-6242Si hot torsion specimens. are not considered, while in the other their influ-
Tested at e 0.9 s1; T 816 C (1500 F). Source: Ref

52
ence is taken into account during the analysis.
particularly when the test alloy possesses a sig- Analysis of Flow Localization in the
nificant strain-rate sensitivity index. Absence of Heat Transfer
The difficulties associated with the interpre-
tation of torque-twist behavior are shown in Flow Localization Parameter. When heat-
Fig. 49 and 50, in which experimental data are transfer effects are minimal, the study of flow
presented for hot torsion tests on thick-wall tu- localization in torsion aids understanding of the
bular specimens of Ti-6Al-2Sn-4Zr-2Mo-0.1Si occurrence of shear bands in low-strain-rate
(Ti-6242Si) twisted at rates sufficiently high to isothermal metalworking operations, as well as
minimize the effects of axial heat transfer. In those at very high strain rates. Examples of these
Fig. 49, the alloy had a starting microstructure processes include isothermal forging (e 103
of equiaxed alpha ( microstructure); in s1) and high-energy-rate forming (e 103
Fig. 50, it had one of acicular, Widmansttten s1). In these cases, a defect is assumed to be the
alpha ( microstructure). The two torque-twist source of localization. For a torsion specimen,
curves were similar in that they both exhibited the defect may be a deficiency in radius (or wall
torque maxima early in the deformation and thickness), strength coefficient, or another mate-
subsequently decreasing torque with increasing rial property. The simplest imperfection to visu-
twist. For the microstructure, however, alize is the geometric (or radius/wall thickness)
the softening rate was substantially greater. defect, although the localization behavior is usu-
Scribe-line measurements revealed that the ally similar for similar sizes of geometric or ma-
specimen deformation was uniform, terial-property inhomogeneities.
whereas the specimen had undergone flow When a torsion specimen with a geometric
localization. defect is twisted, variations in twist, twist rate,
These results can be explained on the basis of and temperature exist between the defect region
a parameter that relates the degree of flow local- and the nominally uniform region from the onset
Fig. 48 Micrographs from room-temperature (e 10
s1) torsion tests on type 304L stainless steel. ization to the ratio of the torque-softening rate to of deformation. Generally, all of these quantities
(a) e 1.16 (outside shear band). (b) e 1.16 (inside shear strain-rate-sensitivity index. For the mi- will be higher in the defect area. These differ-
band). Magnification: 310. Source: Ref 88 crostructure test, the detrimental effects of torque ences are generated to maintain torque equilib-
softening were counterbalanced by a rather high rium, so that the defect and uniform regions
misleading. Although a monotonically increas- strain-rate sensitivity, and localization was pre- transmit an identical torque. They change as the
ing torque-twist curve usually indicates homo- vented in much the same way that it is during the deformation proceeds, depending on the mate-
geneous or nearly homogeneous flow, a curve tensile testing of superplastic materials. In con- rial properties. Thus, localization, or the lack of
that exhibits a torque maximum (often referred trast, for the microstructure Ti-6242Si speci- it, must be viewed as a process. In contrast to the
to as torque instability) and torque softening men, which also has a high strain-rate sensitivity, localization process and analysis, shear banding
does not necessarily signify flow localization, the larger amount of flow softening caused by the is often interpreted in terms of an instability con-
dition. Such treatments are concerned only with A somewhat modified form of Eq 28 is useful Substituting this into Eq 30 and assuming m*
material properties and with determining the in obtaining a quantitative estimate of the level m yields:
twist, under a nominally homogeneous deforma- of flow localization that can be expected during
tion field, at which the torque maximum (dM torsion testing. This form is obtained by
d ln t
0) occurs. The results for Ti-6242Si discussed substituting:

previously demonstrated that a torque maximum d ln e dg g
a= =- 3 (Eq 33)
is not a sufficient condition for localization. de m
Rapid localization cannot occur until after the ( )
3 L e
( 3 )( Le) (Eq 29)

instability condition is satisfied. However, if the q = and q =

rs rs or
material shows only a small amount of flow soft-
ening or has a large positive rate sensitivity, lo- where rs and L are the outer radius and gage
calization and the occurrence of shear banding d ln s
length of the specimen, respectively, into this de
in torsion will be minimal. Thus, the quantitative (Eq 34)
equation to yield: a = - e

degree of localization cannot be predicted by the m
instability (torque maximum) analysis alone.
As its name implies, flow-localization analy-

d ln e

=-
( 3 )( L) G
sis concerns variations of twist, twist rate, and de rs m* (Eq 30)

temperature between a defect region and a nom-
inally uniform region. During the entire strain- The left side of Eq 30 is known as the alpha in terms of von Mises effective stress and strain.
concentration process, the torque must be () parameter. Much experimental and theoret- Equation 34 indicates that the parameter for
uniform along the axis, or M 0, where de- ical work has demonstrated that when is a thin-wall torsion-test specimen can be calcu-
notes a variation. In the present context, it is equal to 5 or more, noticeable flow localization lated directly from the normalized flow softening
used to define variations of the various field should be expected, either in torsion, compres- rate (at constant strain rate) and the strain-rate
quantities between the two regions of the torsion sion, or some other deformation mode. When G sensitivity index. Specimen geometry has no ef-
specimen. If M M(, , T, R), where R is the 0, which corresponds to the torque instabil- fect on the results. Furthermore, Eq 34 suggests
specimen radius, the equilibrium condition M ity condition, dM 0, is equal to 0, and min- that other mechanical tests may be useful in esti-
0 may be used to obtain: imal localization is expected at this point. mating values of (d ln
/de)|e. In particular, tests
Beyond the torque instability, sufficiently large in which flow localization can be avoided are
ln M d ln R
preferable. The mechanical test in which local-
- G + amounts of torque softening and/or low rate
d ln q ln R q,q T dq sensitivity are required to obtain values of ization occurs the most slowly is the uniaxial
=
(Eq 26) on the order of 5. Only when m* ( m) is very compression test due to the stabilizing effect of
dq ln M
small (as at cold working temperatures) will increases in cross-sectional area. To obtain val-

ln q q,T , R the onset of noticeable flow localization follow ues of the parameter for thin-wall torsion tests,
soon after the occurrence of the torque compression data from tests conducted at con-
where G denotes the normalized torque harden- maximum. stant strain rates equivalent to those in torsion
ing (or softening) rate at a fixed : Application of the Flow-Localization can be conducted to estimate the torque-soften-
Parameter. An examination of Eq 30 indicates a ing rate and strain-rate sensitivity parameters.
dependence of the flow-localization parameter When a thick-wall tube or solid-bar torsion
d ln M test is to be analyzed, compression data are also
G on specimen geometry, G, and m*. As stated in
dq q
Eq 27, G is the normalized torque softening rate useful in estimating the magnitude of the pa-
under constant conditions. This rate varies rameter. Assuming that localization has been
ln M ln M avoided in the compression tests, flow stresses

q q ,T , R
dq +
T q,q , R
dT with , leading to the conclusion that , and thus
= (Eq 27) the rate of localization, can vary during torsion as a function of strain, strain rate, and tempera-
dq testing. ture can be converted to the equivalent shear
Difficulty arises, however, when attempting to stress as a function of , , and T. For a given
The derivation of Eq 26 and a detailed discus- use the measured M- curve from a torsion test torsional , the shear stress can then be used via
sion of flow-localization analyses are presented in which localization has occurred, because the numerical integration to obtain the torque (M
in Ref 121. Once localization becomes notice- strain rate will have varied along the gage length 2r2dr) required for the specific torsion speci-
able, lnR, and the second term in the even though the overall twist rate and average men geometry. Note that this is the torque that
braces in Eq 26 becomes negligible. Defining strain rate may have been held constant. Thus, it would be required under uniform deformation
( ln M/ ln ),T,R as m*, Eq 26 reduces to an is often necessary to estimate G from the meas- conditions.
expression for the torsional flow-localization pa- ured torque-twist behavior or to use some other Calculations of M versus in such a case,
rameter A: mechanical test in which localization does not which were based on the compression data, were
occur. performed for the thick-wall torsion specimens
The values of G and m* are readily estimated of the Ti-6242Si alloy discussed previously. The
d ln q - G results (Fig. 51) demonstrate that the absolute
A = (Eq 28) for thin-wall tube specimens. In this instance,
dq m*
m* is identical to the strain-rate sensitivity magnitude of G (d ln M/d) was certainly
index, m. G is obtained from the torque expres- much greater for the microstructure than
The parameter A describes the variations of twist sion for this geometry: microstructure. This behavior reflects the
and twist rate that can be sustained under equi- larger amount of flow softening that occurs in
librium conditions, as a function of the material uniaxial compression at a variety of deformation
M 2r s2t (Eq 31)
properties through their influence on G and m*. rates.
A material whose constant-strain-rate flow An estimated constant-strain-rate M- curve
Resulting in:
curve shows a large amount of flow softening is also plotted with each of the experimental re-
will exhibit large negative G values. Similarly, a sults in Fig. 49 and 50. For the mi-
high strain-rate sensitivity, m ln / ln , G=
d ln M d ln t rs d ln t
= = (Eq 32)
crostructure, the predicted curve agrees quite
will result in large values of m*. dq dq L dg well with the measured curve, confirming that
the assumption of uniform deformation is valid. ization. In addition, the parameter in this case and is readily determined from constant-strain-
The maximum value of over the twist range is found to be as high as 14.3much greater rate compression tests.
investigated is only 3.0. In contrast, the pre- than the critical of 5. As in torsion, criteria based on a c 5 or
dicted M- curve for the microstructure in Fig. Application of the Flow-Localization p 5 are useful for predicting the occurrence
50 lay substantially above the experimental re- Parameter to Other Deformation Modes. of flow localization. Examples are given in Fig.
sults. This divergence, or the fact that deforma- The rate of flow localization in torsion is related 52 and 53 for the isothermal hot compression
tion proceeded at torques below those required to the parameter, as defined by Eq 30. and isothermal plane-strain sidepressing, respec-
for uniform flow, is an indication of flow local- Basically, this expression establishes that tively, of Ti-6242Si. In each case, the mi-
marked flow localization occurs only after a crit- crostructure specimens, which developed low
ical value of the ratio of the torque softening rate values of the parameter (c or p), deformed
to the rate sensitivity of the torque is achieved. uniformly. However, the microstructure speci-
Related metalforming research has demon- mens in both cases had sufficiently high degrees
strated that similar parameters can be used to of flow softening to promote 5 and thus de-
gage the rate of localization in isothermal uniax- veloped regions of nonuniform flow.
ial compression and plane-strain forging opera- Effect of Temperature and Strain Rate on
tions. In compression, is defined as: the Parameter. The parameter used to gage
flow localization generally shows a sharp de-
g - 1 pendence on temperature and strain rate. These
a c = - (Eq 35) variations can be predicted, at least qualitatively,
m
from the magnitudes of the terms that comprise
and is used to predict when unstable, localized it. For simplicity, only p, the flow-localization
bulging may occur. In isothermal plane-strain parameter for the occurrence of shear bands in
forging processes, an defined by: plane-strain forging operations, is discussed
here.
g As defined by Eq 36, p depends on and m.
ap = (Eq 36) In turn, is specified by Eq 37. At cold work-
m
ing temperatures, is usually positive due to
can be used to predict the formation of shear strain hardening. At high strain rates, however,
Fig. 51 Torque-twist curves for Ti-6242Si predicted
bands. In both cases, m is the strain-rate sensi- the effect of thermal softening in Eq 37 may out-
from a numerical-deformation/heat-transfer
simulation and measured compression flow-stress data. tivity and is the normalized strain-hardening weigh the strain-hardening effect (the first par-
Results are for testing at 913 C (1675 F) and various av- (or softening) rate at fixed strain rate: enthetical term on the right side of the equation),
erage effective strain rates. Average effective strain rate resulting in an overall negative , or flow soft-
0.6 surface effective strain rate for the tubular specimen ening. Furthermore, at cold working tempera-
design used. Source: Ref 52 ln s ln s
de + dT tures, m is usually small (usually between 0 and
1 ds e e T
e,e
g = = 0.02). Therefore, small amounts of flow soften-
s de e ,T
de ing coupled with low m may be sufficient to gen-
(Eq 37) erate 5 and hence cause noticeable flow
Fig. 52 Specimens of Ti-6242Si from isothermal hot

compression tests. Tested at 913 C (1675 F);
e 2 s1. Starting microstructures were (a) Fig. 53 Transverse metallographic sections of Ti-6242Si bars. (a) (equiaxed alpha). (b) (Widmansttten
(equiaxed alpha). (b) (Widmansttten alpha). alpha) microstructures. Isothermally sidepressed at 913 C (1675 F); e 2 s1. Magnification: 5. Source:
Magnification: 3. Source: Ref 122 Ref 52
localization. Shear bands generated at cold specimen. In this case, the analysis involves the dynamic recovery and dynamic recrystalliza-
working temperatures at high strain rates are development of a torque-equilibrium equation as tion, which are the main mechanisms controlling
often called adiabatic shear bands. well as a relation to describe the heat-transfer as- microstructural development for single-phase
At hot working temperatures, flow softening pects (Ref 121, 123, 124). In its most basic form, metals and alloys, and dynamic spheroidization
(negative ) is prevalent not only as a result of the heat-transfer relation considers axial heat and strain-induced precipitation in two-phase
thermal softening effects at high strain rates, but conduction and deformation heat generation. In materials.
also because of microstructural softening at most applications, radiation, convection, and ra- When microstructural observations are made,
all deformation speeds. These microstructural dial-heat-transfer effects can be neglected. The a particular type of cross section, which is nei-
effects, quantified by the first parenthetical equilibrium and heat-transfer equations are ther axial nor transverse, should be selected.
term on the right side of Eq 37, are due to solved incrementally, subject to the imposed This section should be one in which the outer-
softening processes such as dynamic recovery, boundary conditions (typically, a constant over- diameter surface, or near-surface region of
dynamic recrystallization, and the breakup all twist rate). the round gage section, is examined. Axial-
of Widmansttten microstructures to form Calculations have been carried out to deter- tangential sections reveal the true simple shear
equiaxed microstructures. During plane-strain mine the effect of heat transfer on flow localiza- nature of the deformation.
forging, it is also possible for the strain path in tion during the torsion of type 304 stainless steel Because the deformation varies along the ra-
crystallographically hard grains to change torsion specimens at room temperature (Ref dius of the gage section in torsion, the correla-
from homogeneous pure shear to localized sim- 123). The material coefficients required for the tion of observed microstructural features to the
ple shear due to texture-related flow softening at analysis were determined from low-speed tests measured shear stress, strain, and strain rate
a grain scale. On the other hand, at hot working at which localization does not occur. Using a must be done with care. Barraclough et al. (Ref
temperatures, m values tend to be substantially specimen with a premachined 8% defect in ra- 35) have examined the problem in detail and
larger (usually between 0.1 and 0.3) than at dius at the center of the gage section, localiza- have devised the concept of an effective ra-
cold working temperatures. Thus, substantial tion occurred during tests at 0.05 s1. The dius, am, for handling the difficulty of mi-
amounts of flow softening, much more than at localization rate (measured using scribe lines) crostructure variation across the section. This ra-
lower temperatures, are required to produce showed good agreement with the localization dius is that whose local shear stress * would
marked flow localizations. simulation based on material parameters, when generate the measured torque if that same shear
the additional effect of geometry changes oc- stress were developed across the entire section
curring during testing (because the specimen and was derived as:
Analysis of Flow Localization in the ends were not clamped) was taken into account
Presence of Heat Transfer (Fig. 54).
am =
3

[
a2( 3+ m + n ) - a1( 3+ m + n ) ]
3 + m + n a23 - a13 (Eq 38)
At intermediate strain rates, heat-generation
and heat-transfer effects must be taken into ac- Microstructure Development
count to describe flow localization during tor- During Deformation Processing in which a2 and a1 denote the outer and inner
sion testing. Axial temperature (and hence flow radii, respectively, of the torsion specimen, and
stress) gradients are established as a result of m and n are material coefficients used to fit a
In previous sections of this chapter, the appli-
uneven deformation heating caused by the pres- constitutive equation:
cation of torsion testing to determine workabil-
ence of defects, as well as by the conduction of ity limited by excessive loads, fracture-con-
heat into the colder shoulders of the torsion trolled failure, and flow-localization-controlled K n m (Eq 39)
failure has been discussed. Often, however, the
ability to form a piece of metal into a particular
shape with the available equipment comprises Dynamic Recovery and
only the basic considerations of workability. It is Recrystallization in Single-Phase
usually important to control the microstructures Materials
that are developed, which ultimately determines
the properties of the finished product and its In single-phase materials, the primary mi-
suitability for further deformation processing, crostructural features are dislocations and grain
heat treatment, final machining, or service. The boundaries. The torsion test is used to determine
ability to accurately control test variables during how these characteristics are affected by defor-
torsion makes the technique attractive to estab- mation temperature and deformation rate.
lish the processing parameters that are required Dynamic recovery at hot working temperatures
to produce the desired microstructures. leads to a reduction in the number of disloca-
At cold and warm working temperatures, tions at a given strain, without noticeably affect-
changes in microstructure are largely a distor- ing the gross deformation of the grains. By con-
tion of the metal grains (observable at optical trast, dynamic recrystallization is characterized
magnifications) and a process of dislocation by the motion of grain boundaries and annihila-
multiplication and dynamic recovery, which is tion of large numbers of dislocations in a single
only detectable with the aid of transmission event, thereby producing new strain-free grains.
electron microscopy. Dynamic recovery consists Metals with high-stacking fault energies (e.g.,
Fig. 54 Comparison of experimentally observed local- of the regrouping of individual dislocations to pure aluminum and -iron), in which climb of
ization kinetics (data points) with simulation form cells and subgrains. However, recovery, as dislocations is easy because they are not dissoci-
results (solid lines) in type 304 stainless steel specimens.
Specimens had premachined radius defects at the center well as many other microstructural transfor- ated, tend to soften primarily by dynamic-recov-
of the gage section and were tested in torsion at room tem- mations, occurs much more readily under hot ery processes. On the other hand, low stacking
perature. The simulations were run with two rate sensitiv- working conditions. The hot torsion test is thus fault energy materials (e.g., -iron, copper, and
ities: m* 0.01 and m* 0.005, whose values bounded frequently used to detect the effects of deforma- nickel), in which dislocations are dissociated
those measured in torsion tests on specimens without geo-
metric defects. Average surface shear strain rate was ap- tion and deformation rate on these changes. and thus able to climb and cross-slip only with
proximately 0.05 s1 in both experiments and simula- Significant microstructural phenomena that can difficulty, tend to store large reservoirs of strain
tions. Source: Ref 123 be studied with the hot torsion test include energy at hot working temperatures. Dynamic
recrystallization is initiated from these large tion exceeds the rate of recovery. This is fol- The torsional flow curves for materials that
stores of dislocations. lowed by a plateau in flow stress and steady-state recrystallize dynamically at hot working temper-
The flow curve from a torsion test on a metal flow. At this point, well-defined, equiaxed sub- atures are often quite different from those typi-
in which dynamic recovery predominates at hot grains form whose walls (subboundaries) consist cal of metals in which dynamic recovery pre-
working temperatures has an appearance typified of fairly regular arrays of dislocations. These dominates. These curves may be used to
by iron tested in the regime at 700 C (1290 subgrains remain equiaxed during the remainder determine the onset of dynamic recrystallization
F) (Fig. 55). Initially, strain hardening occurs, of the deformation process, even though the ac- as a function of strain rate and temperature.
during which the rate of dislocation multiplica- tual grain boundaries do not, as shown by the op- Stress-strain curves for nickel and carbon
tical micrographs of Fig. 56. Figures 56(a) and steel (in the single-phase austenite range) are
(b) are macrographs of two different hot torsion shown in Fig. 57 and 58. The gross behavior
specimens twisted a total of ten revolutions. The consists of an initial strain-hardening stage, dur-
sample in Fig. 56(a) was given ten twists in the ing which the dislocation density increases rap-
forward direction, while that in Fig. 56(b) was idly. At some point, a flow-stress maximum is
twisted forward for five turns and then given five achieved, after which marked flow softening oc-
turns of reverse twist. The grains in both speci- curs. The strain at the peak flow stress decreases
mens underwent a large degree of deformation, with increasing temperature and decreasing
which increased from the center to the surface strain rate. Such flow softening can be identified
because of the strain variation inherent in the tor- with the onset of dynamic recrystallization. At
sion of solid round bars. The large degree of de- relatively low temperatures and/or high strain
formation can be readily seen from the micro- rates, the flow curve eventually achieves a
graphs of the first test specimen, Fig. 56(c), in steady state, indicative of ongoing (discontinu-
which the highly distorted nature of the twisted
grains is evident at the bottom of the micrograph.
The micrograph of the second test sample of
identical geometry that was twisted five revolu-
tions in one direction and then five in the reverse
direction is shown in Fig. 56(d). It is apparent
that the original equiaxed grain structure has
Fig. 55 Stress-strain curves for Armco iron. Strain-rate
been restored by this deformation schedule, even
dependence of the flow stress at 700 C (1290
F), or 0.54 the absolute melting point, is evident. Data though these grains have received the same
are from compression tests; torsion results exhibit similar equivalent strain as those of Fig. 56(c).
behavior. Source: Ref 125
Fig. 57 Stress-strain curves derived from hot torsion

data for nickel at an effective strain rate of
0.016 s1. The dependence of flow behavior on test tem-
perature in a material that undergoes dynamic recrystal-
lization is shown. Source: Ref 26
Fig. 58 Flow curves for 0.25% low-carbon steel in the

austenitic state tested in torsion at 1100 C
Fig. 56 Effect of twist reversal on specimen appearance and structure of superpurity aluminum after deformation at (2010 F). The strong influence of strain rate and a behav-
400 C (750 F) and a strain rate of 2 s1. (a) 5 5 revolutions (2). (b) 5 5 revolutions (2). (c) 5 5 ior indicative of dynamic recrystallization are shown.
revolutions (25). (d) 5 5 revolutions (25). Source: Ref 54 Source: Ref 39
ous) dynamic recrystallization. At low strain single-peak and multiple-peak (periodic) flow occurs, the shape of the flow curve may be used
rates and/or high temperatures (as under condi- curves. This curve is labeled Zc-Do. In addition, to determine the temperatures and strain rates
tions approaching those typical of creep defor- a curve representing Z versus 2Ds is shown. at which refinement of the grain size, a charac-
mation), the flow curve following the peak flow The similarity between the Zc-Do and Z-2Ds teristic important with respect to service prop-
stress often exhibits cyclic hardening and soft- curves establishes that the transition in flow be- erties, occurs. Moreover, it can be used to
ening. havior occurs when the equilibrium or stable, establish equivalent combinations of tempera-
The dynamically recrystallized grain size dynamically recrystallized size is equal to ap- ture and strain rate at which a given grain size is
formed under equilibrium conditions is a func- proximately one-half of the initial grain size. produced.
tion of the deformation temperature and strain Thus, grain refinement leading to at least a halv-
rate. For example, the stable grain size in type ing of the starting grain size produces a single- Development of Microstructure in
304L stainless steel twisted at several tempera- peak flow curve, whereas grain coarsening (or
tures and strain rates is shown in Fig. 59, which refinement of less than one-half) results in cyclic
Alloys with More Than One Phase
illustrates that grain size increases with increas- flow curves (Ref 99). The torsion test has also been used to deter-
ing temperature and decreasing strain rate. As These observations are useful in rationalizing mine the effects of deformation parameters on
with the flow stress and fracture behavior at hot the dependence of the shape of the torsional the microstructures developed in two-phase and
working temperatures, the dependence on these flow curve on temperature and strain rate. As multiphase alloys. Torsion testing has the ability
two variables is best expressed in terms of the mentioned previously, cyclic curves are most to impose large strains at rates up to and includ-
product e exp (Q/RT ), or the Zener-Hollomon frequently observed at low strain rates and high
parameter (Z ). The strong correlation between temperatures, or the regime in which the stable,
the reciprocal of the average recrystallized grain recrystallized grain size is large. Because all but
size and Z is demonstrated for copper and nickel coarse-grained materials undergo grain coarsen-
in Fig. 60. ing during torsion under these conditions, the
A similar dependence between Z and the sta- flow curves are cyclic. These coarsening cycles
ble, dynamically recrystallized grain size, Ds, continue to be observed until the recrystallized
for a 0.16% C steel tested in the austenite regime grain size attains the equilibrium value.
is shown in Fig. 61. Also illustrated are the crit- Thus, the torsion test can be very useful in
ical Z, or Zc, values as a function of starting establishing the occurrence of recovery or re-
grain size, Do; the values of Zc represent the con- crystallization during the hot working of single-
ditions associated with the transition between phase materials. When dynamic recrystallization
Fig. 60 Dynamically recrystallized grain sizes of cop-

per and nickel as a function of the Zener-
Hollomon parameter (Z). Source: Ref 126
Fig. 61 Dependence of the critical parameter Zc on

initial austenite grain size Do in a 0.16% C
steel (open data points). The solid line fitted to the filled
points is the Z-Ds relationship (Ds stable grain size
Fig. 59 Micrographs of type 304L stainless hot torsion specimens tested under various strain-rate/temperature con- achieved during dynamic recrystallization). Note that the
ditions. (a) 0.01 s1, 800 C (1470 F) (e 1.99). (b) 10 s1, 800 C (1470 F) ( e 3.81). (c) 0.01 s1, 1000 Z-2Ds (broken line) and Zc-Do relations are nearly coinci-
C (1830 F) (e 3.73). (d) 0.01 s1, 1200 C (2190 F) (e 4.64). Magnification: 350. Source: Ref 88 dent. Source: Ref 99
ing those used in commercial metalworking op- diffusion processes, resulted in a large interpar- almost identical conditions of strain, strain rate,
erations. For example, torsion testing in two- ticle spacing, indicative of a coarse dispersion of and temperature. In addition, the microstructures
phase alloys can be used to study flow softening cementite. developed in torsion and extrusion at the higher
and the breakup of unstable microstructures. In a similar two-phase system, torsion has temperature of 1150 C (2100 F) (Fig. 66)
These effects are especially strong in materials been used to study the modification of the closely resembled one another. These results and
with lamellar or Widmansttten phases, such as Widmansttten microstructure of - titanium similar ones on other alloys establish the torsion
carbon steels and - titanium alloys. alloys subjected to deformation below the -tran-
In carbon steels, torsion may be used to deter- sus temperature. As with the eutectoid steel, tests
mine the flow response and microstructural of this type have shown that coarsening prevails
changes that occur in pearlite subject to defor- at temperatures near the transus, whereas spher-
mation below the lower critical temperature. For oidization results from deformation at high strain
steel of a eutectoid composition, large amounts rates and lower subtransus temperatures.
of flow softening were measured in torsion, ex- The torsion test can also be used to study mi-
cept when tested above the critical temperature crostructural modification in multiphase alloys
of 742 C (1368 F) (Fig. 62). This softening and to establish processing conditions based on
was associated with the breakup and spher- this information. It is particularly useful for
oidization of the cementite (Fe3C) lamellae. As nickel-base superalloys. These materials are ex-
the temperature was lowered (or the strain rate pensive and have a very limited working tem-
was increased), the torsion results suggested that perature range due to the presence of second
the rate of spheroidization increased. At the phases (, intermetallic carbides), which only
highest subcritical temperature (715 C, or 1320 go into solution at relatively high temperatures
F), the amount of softening was rather low, sug- (if at all), and because of the problems of grain
gesting that microstructural changes were not as growth or incipient melting when deformation is
drastic as at lower temperatures, consisting pri- performed near the melting point.
marily of the development of coarse cementite. Figures 64 through 66 illustrate a typical ap-
This conclusion was supported by microstruc- plication of the torsion test to study the develop-
tural examination of the hot torsion specimens, ment of microstructure in the nickel-base alloy
which indicated a definite relationship among Udimet 700. In the as-received condition (Fig.
strain rate, temperature, and spacing of the 64), the material had fine , carbides, and
spheroidal particles after a fixed amount of borides, dispersed through the matrix, whose
deformation (Fig. 63). Again, the strain rate grain structure was not resolvable. After torsion
and temperature dependence was expressed at approximately 1060 C (1940 F), the struc-
through the Zener-Hollomon parameter, e exp ture consisted of a well-defined fine grain
(Q/RT). When high strain rates or low tempera- structure with carbides and borides situated at
tures were used, a fine, closely spaced disper- the grain boundaries Fig. 65(a).
sion was produced. On the other hand, low strain This microstructure is very similar to that ob-
rates and high temperatures, which enhance served in Udimet 700 material extruded under
Fig. 63 The mean free path () between spheroidite
particles in hot-worked eutectoid steels as a
function of the Zener-Hollomon parameter. The right ordi-
nate is scaled to show the strain rate at 500 C (930 F),
which would produce the indicated spacing after large
plastic deformation. Data are from torsion, compression,
and rolling experiments. Source: Ref 127
Fig. 62 Effect of test temperature on the torsional flow curve of a high-purity 0.8% C pearlitic iron-carbon alloy. Fig. 64 Starting microstructure of Udimet 700 billet
Numbers in parentheses refer to the number of twists to fracture. The flow softening at the three lower tem- material used in torsion and extrusion studies.
peratures can be attributed to pearlite spheroidization. Source: Ref 21 Magnification: 465. Source: Ref 114
test as a valuable means of determining the mi- normal torsion tests in which the test specimen loop test systems with computers or function
crostructures that can be developed during ac- is heated to temperature, soaked for a period of generators that provide the proper control sig-
tual metalforming processes. time, and then twisted. Furthermore, during de- nals representing the rotation rate-time depend-
formation processing, the strain rate is rarely ence needed.
constant, unlike that in the conventional torsion Figures 67 and 68 illustrate the effects of
Processing History Effects experiment. temperature and strain rate history during tor-
With proper controls, the effects of tempera- sion on the flow stress of -iron and copper,
Torsion testing describes the broad patterns of ture and strain-rate history on workability (flow- respectively. The -iron was tested at a constant
deformation, failure resistance, and microstruc- stress levels, fracture behavior, and microstruc- strain rate, but the specimen temperature was in-
ture development for a given alloy. In actual ture development) can be assessed using the creased or decreased at a continuous rate of 50
metalforming processes, however, the thermo- torsion test, provided means exist to replicate the C/min (90 F/min). The flow stresses from such
mechanical history of the workpiece is rarely so thermal and/or deformation-rate history. The tests were compared to isothermal stresses. If
simple. Frequently, it will be preheated in a fur- thermal effect of greatest importance is cooling the specimen was heated during testing, the flow
nace and transferred to the processing equip- during processing. Cooling histories are best stress was higher than the isothermal stress. The
ment (forging press, rolling mill, etc.). During controlled during testing through forced-air or reverse was true if the specimen was cooled.
this dwell period, it will have cooled a certain argon convection around the specimen at rates During either heating or cooling, the dislocation
amount. Also, the workpiece will chill when it is that must be determined experimentally to ob- substructure does not change instantaneously.
in contact with the tooling during conventional tain the desired results. Strain-rate histories are During the heating experiments, a lower-
metalworking processes. This is in contrast to more readily controlled by interfacing closed- temperature, less highly recovered substructure
is retained, giving rise to higher flow stresses
than the isothermal tests. Similarly, during cool-
ing, a softer, more highly recovered substructure
leads to lower flow stresses than observed in
isothermal experiments.
The effect of strain-rate history on the flow
stress of copper is shown in Fig. 68. Under con-
stant-strain-rate conditions, harder substructures
were produced at higher strain rates. However, if
the strain rate was increased or decreased during
torsion testing, the inertia of the acquired sub-
structure prevented changes in flow stress as
Fig. 65 Comparison of (a) torsion and (b) extrusion microstructures in Udimet 700. Deformed under nearly identi-
cal conditions of e 7 s1, T 1060 C (1940 F), and e 1.85. Magnification 465. Source: Ref 114
Fig. 67 Effect of continuous heating or cooling on the

steady-state flow stress of vacuum-melted iron.
Deformed in torsion at an effective strain rate of 1.5
3 1
10 s . Source: Ref 128
Fig. 68 Effect of increasing or decreasing strain rate on

Fig. 66 Comparison of (a) torsion and (b) extrusion microstructures in Udimet 700. Deformed under nearly identi- the flow stress of copper deformed in torsion
cal conditions of e 4 s1, T 1145 C (2090 F), and e 2.15. Magnification: 465. Source: Ref 114 at 750 C (1380 F). Source: Ref 129
large as those observed in a series of constant-

strain-rate tests. Thus, strain-rate sensitivities
measured in rate-change tests are often lower
than those based on constant-strain-rate or
so-called continuous flow curves in materials
such as copper, aluminum, and austenitic stain-
less steels. In materials that exhibit dynamic
strain aging (e.g., carbon steels at cold working
temperatures), the relationship between the two
rate-sensitivity parameters may be reversed, de-
pending on the strain-rate regime and the kinet-
ics of strain aging (Fig. 69).
The torsion test is especially useful to study
the effects of history during deformation
processes such as bar, plate, and sheet rolling
(Ref 54, 131145). For example, Fig. 70 shows
the type of flow-stress behavior that might be
expected for high-purity aluminum during
rolling in which the temperature decreases con-
tinuously. As the temperature decreased, the
flow stress increased, but not as much as would
be expected based on isothermal measurements.
This phenomenon was particularly evident once
the temperature dropped below 450 C (840 F)
and was a result of the retention of a soft high-
Fig. 69 Typical stress-strain curves for tests involving instantaneous changes in strain rate for an austenitic stainless temperature substructure.
steel and a ferritic low-alloy steel. Note that the rate-change strain-rate sensitivity is either lower (stainless Simulation of the rolling of high-strength
steel) or higher (low-alloy steel) than that based on continuous (constant-strain-rate) torsion tests. Source: Ref 130
low-alloy steels through the torsion test (Fig.
7174) has been used to provide insight into
how equipment requirements and final mi-
crostructures can be determined by this tech-
nique. In this instance, torsion tests were con-
ducted in a servohydraulic test machine at a
fixed strain rate. However, temperature was con-
trolled to decrease continuously at a rate almost
equivalent to that measured during actual pro-
duction (Fig. 71). Under these conditions, the
torsion flow stresses increased rapidly (Fig. 72).
Using these data, the roll-separating force and
rolling torque were estimated for the rolling
schedule under study using standard formulas.
These were compared to actual measurements
made at the individual stands of the rolling facil-
ity. Agreement was fairly good (Fig. 73), indi-
cating the usefulness of the torsion test in as-
sessing the influence of processing parameters
(rolling speed, temperatures, dwell time between
stands, etc.) on equipment requirements. In ad-
dition, the final microstructure from the torsion
simulation was almost identical to that from the
production run (Fig. 74).
Other experiments using similar computer-
controlled torsion setups have provided valuable
information on the required loads and mi-
crostructures developed during rolling of a wide
range of alloy steels.
The torsion test can also be used to simulate
the effects of die chilling on workability in mul-
tiphase alloys. In particular, the effects of solu-
tioning of second phases (during preheating) and
subsequent reprecipitation during working be-
cause of chilling are readily simulated by the
test. This is accomplished by preheating the tor-
sion specimen to a high temperature and testing
it on cooling. The torsional ductility obtained in
Fig. 70 Deformation-temperature-time schedule and resulting flow behavior of superpurity aluminum deformed in such tests frequently may be used to determine a
torsion at an effective strain rate of 2.3 s1. Source: Ref 54 potential workability problem.
Fig. 71 Deformation-temperature-time sequence imposed during torsion testing of microalloyed steels. The temper-
ature-time profile followed in a production plate mill (dashed curve) is compared with that experienced by
the sample in the torsion machine (solid curve). Source: Ref 131
Fig. 73 Comparison of the predicted and meas-

ured (a) roll-separating forces and (b) roll
torques associated with rolling of a niobium-vana-
dium microalloyed steel. The pass number is shown
beside each data point. Source: Ref 131
Fig. 72 Flow behavior for a niobium-vanadium microalloyed steel deformed in 17 passes in a torsion machine. The
specimen temperatures are represented by the upper bold line. Source: Ref 131
Soc. Test. Mater., Vol 57, 1957, p

12591272
37. T. Morushima, Tetsu-to-Hagan (J. Iron
Steel Inst. Jpn.), Vol 44, 1958, p 552559,
660668
38. C. Rossard and P. Blain, Rev. Met., Vol 55,
1958, p 573594, 595598
39. C. Rossard, Mtaux-Corros.-Ind., Vol 35,
1960, p 102115, 140153, 190205
40. G.E. Dieter, J.V. Mullins, and E. Shapiro,
Deformation under Hot Working Condi-
tions, Iron and Steel Institute, London,
1968, p 713
41. J.A. Bailey and S.L. Haas, J. Mater., Vol 7,
1972, p 813
Fig. 74 Ferrite structure obtained in a niobium-vanadium microalloyed steel. (a) After 17 passes in the torsion ma-
chine. (b) After 17 passes in a production plate mill. Source: Ref 131
1962, p 223236
1964, p 949961
ACKNOWLEDGMENT 16. P.J. Regenet and H.P. Stuwe, Z. 44. T.B. Vaughan and W.C. Aston,
Metallkde., Vol 54, 1963, p 275 Metallurgia, Vol 471, 1969, p 3944
This article drew upon: 17. H.P. Stuwe and H. Turck, Z. Metallkde., 45. E. Shapiro and G.E. Dieter, Metall. Trans.,
S.L. Semiatin, G.D. Lahoti, and J.J. Jonas, Vol 55, 1964, p 699703 Vol 1, 1970, p 17111719
Application of the Torsion Test to Deter- 18. H.P. Stuwe, Z. Metallkde., Vol 56, 1965, 46. G. Glover and C.M. Sellars, Metall.
mine Workability, Mechanical Testing, Vol p 633642 Trans., Vol 4, 1973, p 756775
8, ASM Handbook, American Society for 19. C.M. Sellars and W.J.McG. Tegart, Mem. 47. H. Weiss, D.H. Skinner, and J.R. Everett,
Metals, 1985, p 154184 Sci. Rev. Met., Vol 63, 1966, p 731746 J. Phys. E, Sci. Instrum., Vol 6, 1973,
M.J. Luton, Hot Torsion Testing, Work- 20. C. Rossard and A. LeBon, Hart.-Tech. p 709714
ability Test Techniques, American Society Mitt., Vol 24, 1967, p 221225 48. S. Fulop, K.C. Cadien, M.J. Luton, and
for Metals, 1984, p 95133 21. J.L. Robbins, O.C. Shepard, and O.D. H.J. McQueen, J. Test. Eval., Vol 5, 1977,
Sherby, Trans. ASM, Vol 60, 1967, p 419426
REFERENCES p 205216 49. J.J. Mills, K. Nielsen, and W. Merriam,
22. J.L. Robbins, H. Wagenaar, O.C. Shepard, Novel Techniques in Metal Deformation
1. C.L. Clark and J.J. Russ, Trans. AIME, Vol and O.D. Sherby, J. Mater., Vol 2, 1967, p Testing, TMS, Warrandale, PA, 1983,
167, 1946, p 736748 271299 p 343357
2. H.K. Ihrig, Trans. AIME, Vol 167, 1946, 23. C. Rossard, Rev. Met., Vol 65, 1968, 50. P. Choquet, A. LeBon, C. Rossard, C.
p 749790 p 181195 Perdrix, and G. Joannes, Proc. Int. Conf.
3. D.E.R. Hughes, J. Iron Steel Inst., Vol 170, 24. F.E. White, Rev. Met., Vol 63, 1966, Physical Metallurgy of Thermomechanical
1952, p 214220 p 991998 Processing of Steels and Other Metals
4. J. Danvergne, M. Pelabon, and J. Ivernal, 25. H.P. Stuwe, Deformation under Hot (Thermec88), Iron and Steel Institute of
Rev. Met., Vol 51, 1954, p 254264 Working Conditions, Iron and Steel Japan, Tokyo, 1988, p 729736
5. K. Tajima and K. Kugai, Tetsu-to-Hagan Institute, London, 1968, p 16 51. B.M. Perrett, Mechanical Testing of
(J. Iron Steel Inst. Jpn.), Vol 42, 1956, 26. M.J. Luton and C.M. Sellars, Acta Metall., Materials, Institute of Metals and
p 980985 Vol 17, 1969, p 10331043 Materials Austalasia Ltd., Melbourne,
6. W. Precht and J.R. Pickens, Metall. Trans. 27. M.J. Luton and W.J.McG. Tegart, Met. Sci. Australia, 1994, p 4144
A, Vol 18A, 1987, p 16031611 J., Vol 3, 1969, p 142146 52. S.L. Semiatin and G.D. Lahoti, Metall.
7. M.P. Clode, Alum. Ind., Vol 11, 1992, 28. J.P. Sah, G.J. Richardson, and C.M. Trans. A, Vol 12, 1981, p 17191728
p 3439 Sellars, J. Aust. Inst. Met., Vol 14, 1969, p 53. G. Glover and C.M. Sellars, Metall.
8. R. Raj, unpublished research, Cornell 292297 Trans., Vol 3, 1972, p 22712280
University, Ithaca, NY, 1981 29. J. Cotner and W.J.McG. Tegart, J. Inst. 54. M.M. Farag, C.M. Sellars, and W.J.McG.
9. W.A. Backofen, A.J. Shaler, and B.B. Met., Vol 97, 1969, p 73 Tegart, Deformation under Hot Working
Hundy, Trans. ASM, Vol 46, 1954, 30. G.A. Redfern and C.M. Sellars, J. Iron Conditions, Iron and Steel Institute,
p 655680 Steel Inst., Vol 208, 1970, p 576587 London, 1968, p 6067
10. H. Ormerod and W.J.McG. Tegart, J. Inst. 31. G. Mima and F. Inoko, Trans. Iron Steel 55. E. Shapiro and G.E. Dieter, Metall. Trans.,
Met., Vol 89, 196061, p 9496 Inst. Jpn., Vol 10, 1970, p 216221 Vol 2, 1971, p 13851391
11. D. Hardwick and W.J.McG. Tegart, J. Inst. 32. F.A. Hodierne, J. Inst. Met., Vol 91, 56. A. Nadai, Theory of Flow and Fracture,
Met., Vol 90, 196162, p 1720 196263, p 267273 Vol 1, 2nd ed., McGraw-Hill, 1950,
12. D. Hardwick and W.J.McG. Tegart, Mem. 33. C. Rossard and P. Blain, Rev. Met., Vol 56, p 349
Sci. Rev. Met., Vol 58, 1961, p 869880 1959, p 175180 57. P. Ludwik, Elemente der Technologischen
13. J.L. Robbins, O.C. Shepard, and O.D. 34. A. Johansen, B. Ronning, and N. Ryum, Mechanic, Springer Verlag, Berlin, 1909,
Sherby, J. Iron Steel Inst., Vol 199, 1961, Proc. Sixth Int. Conf. Aluminum Alloys, p 132
p 175180 Japan Institute of Light Metals, Tokyo, 58. H.S. Kim, Mater. Sci. Eng. A, Vol A299,
14. R.A. Reynolds and W.J.McG. Tegart, J. 1998, p 559564 2001, p 305308
Iron Steel Inst., Vol 200, 1962, p 35. D.R. Barraclough, H.J. Whittaker, K.D. 59. G.R. Canova, S. Shrivastava, J.J. Jonas,
10441059 Nair, and C.M. Sellars, J. Test. Eval., Vol and C.GSell, Formability of Metallic
15. F.E. White and C. Rossard, Rev. Met., Vol 1, 1973, p 220230 Materials2000 A.D, STP 753, ASTM,
59, 1962, p 237241 36. D.S. Fields and W.A. Backofen, Proc. Am. 1982, p 189210.
60. G.R. Canova, Masters thesis, McGill 88. S.L. Semiatin and J.H. Holbrook, Metall. FormingInterrelation between Theory
University, Montreal, 1979 Trans. A, Vol 14, 1983, p 20912099 and Practice, A.L. Hoffmanner, Ed.,
61. E.W. Hart, C.Y. Li, H. Yamada, and G.L. 89. W.J.McG. Tegart, in Ductility, American Plenum Press, 1971
Wire, Constitutive Equations in Plasticity, Society for Metals, 1968, p 133 115. A. Nicholson, D. Smith, and P. Shaw,
A.S. Argon, Ed., MIT Press, 1975, p 149 90. T.L.F. Muller, G.J. Richardson, and H.P. Deformation under Hot Working Condi-
62. A. Graber and K. Pohlandt, Steel Res., Vol Stuwe, Deformation under Hot Working tions, Iron and Steel Institute, London,
61, 1990, p 212218 Conditions, Iron and Steel Institute, 1968, p 161
63. S. Khoddam, Y.C. Lam, and P.F. Thomson, London, 1968, p 131 116. M.G. Cockroft and D.J. Latham, J. Inst.
Steel Res., Vol 66, 1995, p 4549 91. I. Dragon, M.Sci. thesis, University of Met., Vol 96, 1968, p 3339
64. S. Khoddam, Y.C. Lam, and P.F. Thomson, Aston in Birmingham, 1967 117. P.W. Bridgman, Studies in Large Plastic
J. Test. Eval., Vol 26, 1998, p 157167 92. R. Hill, The Mathematical Theory of Flow and Fracture, McGraw-Hill, 1952,
65. S. Khoddam, Y.C. Lam, and P.F. Thomson, Plasticity, Oxford University Press, 1950, p9
Steel Res., Vol 67, 1996, p 3943 p 346 118. A.L. Hoffmanner, Metal Forming
66. S. Khoddam, Y.C. Lam, and P.F. Thomson, 93. P. Ludwik and R. Schen, Stahl Eisen, Vol Interrelation Between Theory and
Proc. Int. Conf. Thermomechanical Pro- 45, 1925, p 373 Practice, A.L. Hoffmanner, Ed., Plenum
cessing of Steels and Other Metals 94. C. Zener and J.H. Hollomon, Trans. ASM, Press, 1971, p 349391
(Thermec97), T. Chandra and T. Sakai, Ed., Vol 33, 1944, p 163 119. A. Gittins and C.M. Sellars, Met. Sci. J.,
TMS, Warrendale, PA, 1997, p 911917 95. A Nadai and E.A. Davis, J. Appl. Phys., Vol 6, 1972, p 118
67. A. Gavrus, E. Massoni, and, J.-L. Chenot, Vol 8, 1937, p 213 120. M. Zidek, Sbornik VSB Ostrave
Steel Res., Vol 70, 1999, p 259268 96. D.J. Lloyd, Met. Sci., Vol 14, 1980, (Hutnicka), Vol 11, 1965, p 687
68. J.J. Jonas, F. Montheillet, and S. p 193198 121. S.L. Semiatin and J.J. Jonas, Formability
Shrivastava, Scr. Metall., Vol 19, 1985, 97. C.M. Sellars, Philos. Trans. R. Soc. and Workability of Metals: Plastic
p 235240 (London) A, Vol A288, 1978, p 147 Instability and Flow Localization, Ameri-
69. T. Spittel, M. Spittel, and J. Suchanek, 98. A. Eichinger, Festigkeits Theoretische can Society for Metals, 1984
Neue Htte, Vol 31, 1986, p 233238 Untersuchungenn in Hanbuch der Werk- 122. S.L. Semiatin, G.D. Lahoti, and T. Altan,
70. M. Zhou and M.P. Clode, Mater. Des., Vol stoffprufung, 2nd ed., Springer Verlag, Process ModellingFundamentals and
17, 1996, p 275281 Berlin, 1955 Applications to Metals, T. Altan et al., Ed.,
71. M. Zhou and M.P. Clode, Mater. Sci. 99. T. Sakai and J.J. Jonas, Acta Metall., Vol American Society for Metals, 1980, p
Technol., Vol 13, 1997, p 818824 32, 1984, p 189209 387408
72. M. Zhou and M.P. Clode, Finite Elements 100. S.L. Semiatin, J.H. Holbrook, and M.C. 123. E. Rauch, G.R. Canova, J.J. Jonas, and
in Analysis and Design, Vol 31, 1998, p Mataya, Metall. Trans. A, Vol 16A, 1985, S.L. Semiatin, Acta Metall., Vol 33, 1985,
114 p 145148 p 465476
73. S. Khoddam, Y.C. Lam, and P.F. Thomson, 101. E. Aernoudt and J.G. Sevillano, J. Iron 124. E. Rauch, M. Eng. thesis, McGill Uni-
Mech. Mater., Vol 22, 1996, p 19 Steel Inst., Vol 211, 1973, p 718 versity, Montreal, Quebec, Canada, 1983
74. F. Montheillet, M. Cohen, and J.J. Jonas, 102. F.E. White and C. Rossard, Deformation 125. J.-P.A. Immarigeon and J.J. Jonas, Acta
Acta Metall., Vol 32, 1984, p 20772089 under Hot Working Conditions, Iron and Metall., Vol 22, 1974, p 12351247
75. F. Montheillet, P. Gilormini, and J.J. Steel Institute, London, 1968, p 1420 126. H.J. McQueen and J.J. Jonas, Treatise on
Jonas, Acta Metall., Vol 33, 1985, 103. P. Choquet, J. Nucl. Mater., Vol 52, 1975, Material Science and Technology, Vol 6,
p 705717 p 3444 Plastic Deformation of Materials, R.J.
76. L.S. Toth, J.J. Jonas, P. Gilormini, and B. 104. J. Kortmann, Neue Htte, Vol 2, 1977, Arsenault, Ed., Academic Press, 1975,
Bacroix, Int. J. Plast., Vol 6, 1990, p 371375 p 393493
p 83108 105. I. Dragon, C.G. Radn, and C. Vaida, 127. O.D. Sherby, M.J. Harrigan, L.
77. K.W. Neale, L.S. Toth, and J.J. Jonas, Int. Metallurgia (Bucharest), Vol 30, 1978, Chamagne, and C. Sauve, Trans. ASM, Vol
J. Plast., Vol 6, 1990, p 4561 p 439443 62, 1969, p 575
78. L.S. Toth and J.J. Jonas, Textures 106. T. Nishumura, H. Miguro, and S. Kikuchi, 128. G. Glover, Ph.D. thesis, University of
Microstruct., Vol 10, 1989, p 195209 Denki Seiko (Electric Furn. Steel), Vol 49, Sheffield, 1969
79. L.S. Toth, P. Szaszvari, I. Kovacs, and J.J. 1978, p 187195 129. R. Bromley, Ph.D. thesis, University of
Jonas, Mater. Sci. Technol., Vol 7, 1991, 107. H.J. McQueen, J. Sankar, and S. Fulop, Sheffield, 1969
p 458463 Mechanical Behavior of Materials, 130. D.R. Barraclough and C.M. Sellars,
80. L.S. Toth and J.J. Jonas, Scr. Metall. Cambridge, England, Vol 2, 1979, Mechanical Properties at High Rates of
Mater., Vol 27, 1992, p 359363 p 675684 Strain, Institute of Physics Conference
81. J.J. Jonas and L.S. Toth, Scr. Metall. 108. F.H. Hammad, G.A. Hassan, and M.M. Series No. 21, J. Harding, Ed., Institute of
Mater., Vol 27, 1992, p 15751580 Dougol, Aluminum, Vol 55, 1979, Physics, London, 1974, p 111
82. J.J. Jonas, Inter. J. Mech. Sci., Vol 35, p 457461 131. I. Weiss, J.J. Jonas, and G.E. Ruddle,
1993, p 10651077 109. C.M. Sellars and W.J.McG. Tegart, Inter. Process Modelling Tools, J.F. Thomas, Jr.,
83. J. Baczynski and J.J. Jonas, Acta Mater., Met. Rev., Vol 17, 1972, p 124 Ed., American Society for Metals, 1981,
Vol 44, 1996, p 42734288 110. J. Coupry, Rev. Alum., Vol 451, 1976, p 97125
84. L.S. Toth, Thermomechanical Processing p 259270 132. N.D. Ryan, The Science and Technology
of SteelJ.J. Jonas Symposium, S. Yue 111. A.M. Sabroff, F.W. Boulger, and H.J. of Flat Rolling, B. Fazan et al., Ed.,
and E. Essadiqi, Ed., CIM, Montreal, Henning, Forging Materials and Prac- IRSID, Mazieres-Le-Metz, France, 1987,
Canada, 2000, p 109120 tices, Rheinhold, 1968 p 17.117.9
85. J. Baczynski and J.J. Jonas, Metall. Mater. 112. Evaluating the Forgeability of Steel, The 133. N.D. Ryan and H.J. McQueen, Mater. Sci.
Trans. A, Vol 29A, 1998, p 447462 Timken Company, Canton, OH, 1974 Technol., Vol 7, 1991, p 818826
86. R. von Mises, Gttringer Nachrichten, 113. H.J. Henning and F.W. Boulger, Mechan- 134. N.D. Ryan and H.J. McQueen, J. Mater.
Math.-Phys. K1, 1913, p 582 ical Working of Steel I, P.H. Smith, Ed., Proc. Technol., Vol 36, 1993, p 103123
87. S. Shrivastava, J.J. Jonas, and G. Canova, J. Gordon and Breach, 1964, p 107 135. S. Yue, F. Boratto, and J.J. Jonas, Proc.
Mech. Phys. Solids, Vol 30, 1982, p 7590 114. C.M. Young and O.D. Sherby, Metal Conf. on Hot- and Cold-Rolled Sheet
Steels, R. Pradhan and G. Ludkovsky, 138. L.N. Pussegoda, S. Yue, and J.J. Jonas, 143. T.M. Maccagno, S. Yue, J.J. Jonas, and K.
Ed., TMS, Warrendale, PA, 1988, Metall. Trans. A, Vol 21A, 1990, Dyck, Metall. Trans. A, Vol 24A, 1993, p
p 349359 p 153164 15891596
136. F.H. Samuel, R. Barbosa, F. Boratto, S. 139. L.N. Pussegoda and J.J. Jonas, ISIJ Int., 144. T.M. Maccagno, J.J. Jonas, S. Yue, B.J.
Yue, and J.J. Jonas, Proc. Int. Conf. Vol 31, 1991, p 278288 McCrady, R. Slobodian, and D. Deeks,
Physical Metallurgy of Thermomechan- 140. A. Najafi-Zadeh, J.J. Jonas, and S. Yue, ISIJ Int., Vol 34, 1994, p 917922
ical Processing of Steels and Other Metals Metall. Trans. A, Vol 23A, 1992, 145. L.P. Karjalainen, P. Kantanen, T.M.
(Thermec88), Iron and Steel Institute of p 26072617 Maccagno, and J.J. Jonas, Proc. Int. Conf.
Japan, Tokyo, 1988, p 721728 141. J.J. Jonas and C.M. Sellars, Iron Thermomechanical Processing of Steels
137. F.H. Samuel, S. Yue, J.J. Jonas, and B.A. Steelmaker, Vol 19, 1992, p 6771 and Other Metals (Thermec97), T.
Zbinden, ISIJ Int., Vol 29, 1989, 142. P.R. Cetlin, S. Yue, and J.J. Jonas, ISIJ Chandra and T. Sakai, Ed., TMS,
p 878886 Int., Vol 33, 1993, p 488497 Warrendale, PA, 1997, p 819825
Chapter 9
Hot Working Simulation by Hot

Torsion Testing
Joseph R. Pickens, Concurrent Technologies Corporation
PROCESS PARAMETERS for commercial processes through analytical relationships. By temperature tensile testing is necking, which
hot working processes such as rolling, extrusion, using hot working physical simulation, one in- prematurely limits or obscures the strain to
and forging are normally based on prior experi- vestigator can quickly generate the data neces- which the specimen may be deformed. Elevated-
ence with similar alloys. If the parameters are sary to improve or optimize a production hot temperature tensile testing is often used to assess
successful in producing material that meets working schedule. In addition, the data gener- superplasticity by determining the uniform de-
quality control and property targets at economic ated can be used as input for process modeling, formation up to necking.
production rates, little additional effort is gener- increasing the accuracy of computer simulations Uniaxial compression of cylindrical speci-
ally extended to optimize the process. Produc- of the process. mens is another test by which the flow stress can
tion equipment and personnel are often not Primary input variables in hot working physi- be measured accurately and reproducibly. How-
available for the experimentation required to cal simulations are the deformation temperature, ever, uniaxial compression specimens undergo
perform an optimization study, and the cost of Tdef, and strain rate, e. The strain, e, to which the barreling due to friction at the contact surfaces,
processing (then possibly scrapping) many large simulation experiment is performed depends on which limits uniform deformation to a true strain
ingots or billets is excessive. the process being simulated and the amount of of about 0.8. A more important limitation is that
In addition, alloy developers are often faced deformation occurring in the process. The most the test does not provide a practical measure of
with the dilemma of selecting hot working pa- useful output variables are flow stress, so, and working limits.
rameters to produce experimental material that strain to failure, e f. Basically, so is a measure of The cam plastometer is a hot working sim-
may be expensive and in short supply. They are the resistance of a material to deformation, and ulator in which two flat, opposing platens impact
primarily concerned with producing material in e f is a measure of the deformation limits of a a flat, rectangular specimen. The platens are
the desired product form without losing material material. moved together rapidly by energy stored in a ro-
due to cracking or loading the process equip- tating flywheel acting through a cam, which
ment excessively. Conservative hot working pa- produces strain rates as high as 500 s1. Cam
rameter selection may result in microstructures Types of Hot Working plastometers are expensive and also have the
and properties that do not reach the ultimate po- Simulation Tests disadvantage that the working limits of the ma-
tential of the experimental material. terial are not measured.
One solution to these problems is to perform The capabilities of various hot working simu- Hot torsion testing involves subjecting a
a scaled-down physical simulation of the hot lation tests are summarized in Table 1 (Ref 1). dog-bone-shaped specimen to torsion on a
working process by deforming relatively small These test methods are described in more detail machine that can be loosely described as an in-
amounts of material under carefully controlled in preceding chapters, and so the following is strumented lathe. One end of the specimen is
conditions. The investigator can thereby use just a brief summary of hot tension, compres- prevented from rotating, while the other is sub-
specimens weighing hundreds of grams as op- sion, and torsion testing used in the assessment jected to torque provided by a motor. A tube fur-
posed to production ingots weighing hundreds of workability. nace is used to heat the specimen in place.
or thousands of kilograms. Testing is carried out Hot tension testing can be used to simu- Angular displacement is monitored from the
at temperatures and strain rates in the range of late hot working operations at low strain rates. drive-motor grip system and converted to
commercial hot working practice. The deforma- The peak flow stress, or stress for steady-state torsion strain. The fixed end of the specimen
tion parameters used and output data generated deformation, gives a measure of the material de- is connected to a transducer that measures
during the test can be related to commercial formation resistance. A limitation of elevated- torque, which is converted to torsion stress.
Table 1 Comparison of hot working simulators

Working limit Temperature Flow stress Ease of Ease of Limiting
Simulators e, max e, max measurement control measurement microexam quench instabilities
Tension 101 0.3 No Good Good Poor Fair Necking

Uniaxial compression 101 0.8 No Fair Good Good Fair Barreling
Cam plastometer 102 2 No Fair Good Good Fair Friction
Torsion 103 20 Yes Good Fair Fair Good None
Note: Modified from Ref 1. max, maximum

Chapter 9: Hot Working Simulation by Hot Torsion Testing / 123
Sophisticated hot torsion machines use dual axis creasing linearly to a maximum value at the sur- ture. For this hypothetical example, the output
transducers on the fixed end to measure both face. The maximum shear strain is: variables, equivalent tensile flow stress, or so,
torque and axial load. Changes in axial load and equivalent tensile strain to failure, or ef, are
occur during torsion if the material undergoes R/L (Eq 1) plotted versus Tdef for the various strain rates, as
recrystallization or other internal material trans- shown in Fig. 3.
formations. where R is the radius of the test specimen, L is Interpreting the results, clearly the flow stress
the length of the gage section, and is the angle decreases significantly and monotonically with
of rotation of the movable end relative to the Tdef; however, at the highest strain rate, e 3, the
Hot Torsion Testing Practice fixed end. flow stress approaches zero at the highest Tdef.
Because the shear strain at all points is be- Metallographic evaluation was carried out in this
This section illustrates the use of hot tor- yond the elastic limit, the shear stress at all sample, which showed that the vanishing flow
sion testing for optimization of hot working points across the cross section is in the plastic stress resulted from hot shortness, that is, local-
processes. Billets of an experimental aluminum range, and, neglecting strain hardening or strain- ized melting. Also, ef shows a peak at all strain
alloy have been homogenized by a conservative rate hardening, the shear stress is uniform, ex- rates at the intermediate deformation tempera-
temperature-time schedule and are ready for trial cept at the central axis. From equilibrium, the ture, T2.
processing. What temperature and strain rate shear stress is related to the torque and specimen Thus, the extrusion preheat temperature
should be used to maximize the probability that geometry by: selected is slightly below T2, where ef is near its
useful product can be extruded? maximum. The highest strain rate, e3, is avoided,
A transverse slice from one experimental bil- 3M/2(R3o R3i ) (Eq 2)
because it may lead to hot shortness at T1. To
let is cut, and torsion specimens of the geometry generate the desired strain rate, the appropriate
shown in Fig. 1 (Ref 2) are machined in a where is the shear stress, M is the torque, Ro is billet and extrusion process parameters can be
chordal orientation. This ensures that the shear the outside radius, and Ri is the inside radius. obtained from (Ref 4):
strain during torsion testing is in the same orien- (For a solid test specimen, Ri 0.) For materi-
tation as the dominant shear deformation during
als that strain harden or strain-rate harden, the e 6VD b2 lnR tan/(Db3 D3e) (Eq 6)
extrusion. Fifteen specimens are deformed, one radial variation in strain and strain rate leads
at each combination of five working tempera-
to a modification of Eq 2 for solid specimens where e is the mean equivalent strain rate in ex-
tures and three torsional strain rates. These pa- (Ref 3): trusion; Db is the extrusion chamber diameter;
rameters are based on previous plant experience De is the extrusion product diameter; is the
for similar alloys. During each test, the torque,
M, is measured as a function of angular dis- (3 m n) M/2 Ro3 (Eq 3)
placement.
As described in more detail in the preceding where m is the strain-rate hardening exponent
of the material, and n is the strain-hardening ex- 3
chapter, the shear strain in torsion () increases
Yield strength or flow stress (o)

linearly, starting from zero at the axis and in- ponent.
For the 15 tests performed, calculations using 2
Eq 1 and 2 or 3 are plotted in Fig. 2 to determine 1
o, the shear yield strength, and f, the shear
strain to failure.
For isotropic materials, the shear yield
strength and strain-to-failure values can be con- T1
verted to equivalent tensile stress and strain
through the relationships (Ref 4):
so = t 3 (Eq 4) Deformation temperature (Tdef)

ef = g f 3 (Eq 5) (a)
where so is tensile yield strength (or flow T2

stress), and ef is tensile strain at failure or frac-
Tensile strain at fracture (f)
o
1
Shear stress ()
Deformation temperature (Tdef)
f (b)
Shear strain ()
Fig. 3 Hypothetical dependence of (a) flow stress and
Fig. 2 Typical torsion stress-strain curve at elevated (b) strain to failure on deformation temperature
Fig. 1 Typical torsion specimen temperatures at three strain rates where e1 e2 e3
semidie angle of extrusion die; V is the extrusion the more rapid deformation on the surface of the mercial alloys are generally wide and allow the
ram speed; and R is the extrusion ratio billet extrusion billet. Such high strain rates may lead producer to have narrower composition sweet
area/extruded cross-section ratio Db2 /D2e. to hot shortness. spots for specific end-use applications. For ex-
This example illustrates the details in applica- The results show that slight differences in ample, an alloy may have alloying element com-
tion of hot torsion testing. Following are real- flow stress exist (Fig. 4), but for the limited position in the upper part of the allowable range
world examples of the use of this approach. number of specimens tested, the differences be- to enable service in strength-critical applica-
tween treatments are not statistically significant. tions, such as extrusions under heavy compres-
However, the e f data (Fig. 5) show a clear, sta- sive loading. Alternatively, for corrosion-critical
Hot Torsion Application Examples tistically significant superiority for treatment 3. applications, the alloying element content may
While variations in homogenization treatments be toward the lower end of the allowable range
Example 1: Extrusion of Al-Mg Alloy. An do not significantly alter the resistance of the but just high enough to reliably meet minimum
extrusion fabricator considered extruding a pro- material to deformation, they do have a large ef- strength requirements. Such changes in compo-
prietary aluminum-magnesium (5xxx) alloy that fect on the limits to which the material can be sition, which are within specification but often
is very similar in composition to an alloy that is hot worked. The most advantageous homoge- not highlighted to the user, can have dramatic ef-
usually hot rolled. From rolling practice, it is nization schedule (schedule 3) is identified with- fects on hot working behavior.
known that this alloy is sensitive to homoge- out taking the extrusion press out of production. For example, a vertically integrated extrusion
nization schedule, and four candidate schedules Note also that a few tests were run at 425 C fabricator wanted to make a highly alloyed ver-
are to be considered. The alloy has high magne- (800 F), beyond the hot working range ex- sion of an Al-Zn-Mg-Cu alloy for weight-critical
sium content, which is well known to produce pected by plant metallurgists, and hot ductility applications where high strength could allow
high flow stress. The production supervisor was (e f) was still increasing. Based on the hot torsion down-gaging. The standard version of this alloy
concerned that the high flow stress might stall results, the production supervisor attempted to (7xxxA, hereafter) has relatively low alloying
the press, and removing a stalled billet from the extrude this alloy with exit temperatures higher content in the allowable composition range to
container wastes time. The press was to be taken than those previously thought permissible. Plant minimize stress-corrosion-cracking susceptibil-
out of production temporarily for experimental trials confirmed that this was indeed possible. ity. The fabricator wanted to know what the ef-
extrusion of several billets of the alloy, and the Example 2: Determining the Effect of fect of the increased alloying content would be
supervisor did not want the press to be unavail- Alloy Variation Within Specification. Many on extrusion breakout pressure at various ram
able for production any longer than necessary. metal producers have internal versions of stan- speeds. One log was cast for 7xxxA and an-
To solve this problem, torsion specimens were dard alloys, and the versions are used for other for a more highly alloyed variant (7xxxB,
machined from billets that were subjected previ- specific applications, depending on user require- hereafter). Extrusion billets were machined from
ously to the various homogenization treatments. ments. The composition specifications for com- each log. Torsion specimens were machined
The specimens were tested at temperatures
spanning the expected extrusion range of 340 to
400 C (650 to 750 F) and at a strain rate typi-
cal of commercial extrusion. Some specimens
were tested at very high strain rates to simulate Temperature, C
350 375 400 425

2.2
#3
Temperature, C 2.0 Homogenization

group
325 350 375 400 425 Low strain rate
24 1.8
Homogenization group #4 = 0.04 = 2.0
#4 = 0.04 = 2.0 1.6 #1 #1
22 #1 #1 #2 #2
#2 #2 #3 #3
#3 1.4 #2
#3 #4 #4
High strain rate #4 #4 #1
Strain to failure (f)
20 1.2
Flow stress (o)
#1 1.0
18
#3
#2 1.4
16
#3
1.2 High strain rate
#4
14 1.0
#2
Low strain rate
#3 0.8
12 #1 #1
#2
#4 0.6
10
600 650 700 750 800
600 650 700 750 800 850
Temperature, F
Temperature, F
Fig. 4 Flow stress as a function of deformation temper-

ature for an aluminum-magnesium alloy ho- Fig. 5 Strain to failure as a function of deformation temperature for an aluminum-magnesium alloy homogenized by
mogenized by different schedules (#1, #2, #3, #4) different schedules
from each log, with the long axis of the speci- whether to invest in EMC equipment, uncertain- Flow stress for both 5182 DC and 5182 EMC
mens parallel to the casting direction. Hot tor- ties were raised, such as: Could EMC reduce were essentially unaffected by homogenization
sion testing was performed at strain rates corre- homogenization costs? Would the finer mi- time for the times investigated (Fig. 7). In addi-
sponding to the mean-equivalent strain rate for crostructure of the EMC material lead to re- tion, strain to failure for 5182 DC was largely
the extruded shape in question, based on ram duced edge cracking and result in higher mill unaffected by homogenization time at 523 C
speeds that have been successful for 7xxxA. recovery? The company did not want to invest in (975 F) (Fig. 7). However, 5182 EMC dis-
These strain rates were calculated from Eq 6. purchasing several 10,000 kg (22,000 lb) in- played a dramatic increase in strain to failure, re-
The standard billet preheat temperature of 315 gots to answer these questions by a plant trial. sulting in a 30% increase in hot ductility relative
C (600 F) for 7xxxA was used. However, investing in a small hot working sim- to 5182 DC. Similar results were observed at
Flow stress increased monotonically with in- ulation study was deemed worthwhile (Ref 5). other homogenization temperatures. Based on
creasing strain rate for each alloy, as expected Torsion specimens were machined from the this inexpensive study, the company believed
(Fig. 6). However, the flow stress for 7xxxB was 5182 EMC ingot and a slice of a conventional that an optimization study could lead to reduced
significantly higher than that for 7xxxA, which 5182 DC ingoteach in the as-cast condition. homogenization costs for 5182 EMC, and, fur-
was directly related to the increased alloying Unfortunately, the DC ingot supplied was signif- thermore, edge cracking would likely be lower
content. No difference in strain to failure was icantly lower in magnesium content than the for the EMC material. As an added benefit to
discerned. Subsequent extrusion of 7xxxB at 315 EMC ingot (Table 2), and magnesium is well this simulation, plant metallurgists evaluated the
C (600 F) did indeed stall the press. Successful known to be a major determiner of flow stress in data and then questioned whether the existing
extrusion was conducted at higher temperatures. aluminum alloys. Specimens were homogenized homogenization practice for 5182 DC was
Example 3: Torsion Test in a Cost-Benefit for different times (from 3 to 9 h) at 523 C (975 accomplishing anything, so they initiated a hot
Analysis of EMC Ingot. An aluminum company F), and others were homogenized for 6 h at torsion investigation to improve the homoge-
considered investing in electromagnetic casting temperatures ranging from 482 to 548 C (900 to nization practice.
(EMC) equipment, with the hope of increasing 1020 F). A representative strain rate of 1.2 s1 Example 4: Determining Advantageous
casting recovery rates and reducing production was calculated using commercial rolling param- Rolling Temperatures. An aluminum company
costs. A slice of an EMC ingot of aluminum- eters and the following equation (Ref 6): was uncertain as to the preferred rolling temper-
magnesium alloy 5182 was purchased from a ature for high-volume aluminum-magnesium
company that was casting the alloy by EMC.
Visual inspection of the ingot clearly showed
e = (2 p / 60) N [ ]
( R / H 2 ) (1 - r ) / r ln[1 / (1 - r )] alloy 5052. The conventional practice was to
use the standard homogenization practice and
(Eq 7)
superior surface finish relative to conven- begin rolling at the homogenization temperature
tional direct-chill-cast (DC) ingots. In addition, of 549 C (1020 F) as the ingot was quickly
metallographic evaluation revealed that the where r is the reduction (H1 H2)/H1; H1 is transferred to the reversing mill from the ho-
EMC microstructure was finer than that of the the entry thickness; H2 is the exit thickness; N is mogenization furnace. To identify a more advan-
DC ingot. As the aluminum company was per- the angular speed of rolls in revolutions/min; tageous rolling temperature, a hot torsion inves-
forming the cost-benefit analysis to determine and R is the deformed roll radius. tigation was initiated (Ref 7). Torsion specimens
were machined from various regions of a cross
section of a commercial-scale ingot and were
homogenized by the standard practice (which
was previously determined by hot torsion test-
ing). The strain rate calculated from Eq 7, based
on typical commercial rolling parameters, was
1.2 s1 for the first pass on the reversing mill.
104 Specimens were deformed at temperatures from
Flow stress (o), MPa
425 to 560 C (800 to 1040 F).

Results showed that the flow stress decreases
monotonically with increasing deformation
87
7 xxx B temperature (Fig. 8). However, a surprising
peak in strain to failure was observed at
7 xxx A approximately 500 C (930 F). The plant met-
T def = 315 C (600 F) allurgists had assumed that hot ductility in-
73 creased monotonically with increasing defor-
0 0.010 0.020 0.030 0.040 0.050 0.060 mation temperature until the onset of hot
shortness. This surprising result suggested a re-
Strain rate (), (s1)
vision in the plant rolling practice. The temper-
ature of the homogenization furnace could be
Fig. 6 Flow stress for two versions of a 7xxx aluminum alloy as a function of strain rate at Tdef 315 C (600 F) lowered toward the end of the treatment so that
the ingots could begin rolling at a temperature
slightly higher than the peak in Fig. 8. Thus, as
the ingot cooled during hot rolling, it would be
deformed at temperatures around the strain-to-
Table 2 Composition of 5182 DC and 5182 EMC ingots (balance aluminum) failure peak; that is, the amount of cooling dur-
Composition, wt% ing a typical 20-pass reversing mill operation
Ingot Si Fe Cu Mn Cr Zn Ti Na Mg was measured, and that temperature change
was placed symmetrically around the 500 C
EMC 0.133 0.214 0.032 0.383 0.010 0.014 0.009 0.0002 4.39
DC 0.100 0.223 0.017 0.330 0.035 0.012 0.010 0.0001 3.87
(930 F) rolling temperature, where the peak in
strain to failure is observed. These predictions
Note: EMC, electromagnetic cast; DC, direct chill cast
were confirmed in a plant trial with 20 produc-
tion ingots.
13 Deformation temperature ( = 1.2 s1), C

450 475 500 525 550
12 80
85 11
Flow stress, MPa

70
Flow stress, MPa

10
Flow stress, ksi
12
Flow stress, ksi

5182 EMC 9
80 60
8
7 50
11 75 6 All material homogenized 40
5182 DC 5 at 549 C (1020 F), 4 h
4
800 850 900 950 1000 1050
10 Deformation temperature ( = 1.2 s1), F

0 3 6 9 12 (a)
Homogenization time at 523 C (975 F), h
Deformation temperature ( = 1.2 s1), C
(a)
450 475 500 525 550
10.0
4.5 8.0
Strain to failure
6.0
4.0 5182 EMC 4.0

Strain to failure
2.0
0
800 850 900 950 1000 1050
Deformation temperature ( = 1.2 s1), F
3.5
(b)
Fig. 8 (a) Flow stress and (b) strain to failure as a func-

tion of deformation
5182 DC
3.0
0 3 6 9 12
stress for the two impurity levels (Fig. 9).
Homogenization time at 523 C (975 F), h However, strain to failure peaked at approxi-
(b) mately 315 C (600 F), and, importantly, the
enhanced-purity version displayed an 40%
Fig. 7 (a) Flow stress and (b) strain to failure as a function of homogenization time at 523 C (975 F). Deformation improvement in strain to failure (Fig. 9).
temperature is 493 C (920 F). The decision was made to develop a two-stage
homogenization schedule, and the first step, 16 h
at 427 C (800 F), was selected based on dif-
ferential thermal analysis to reduce coarse com-
Example 5: Determining Commercial determine whether an enhanced-purity version position gradients in this overalloyed material.
Rolling Parameters. For army applications, a of the alloy would have superior edge-cracking This first step increased strain to failure for both
stronger aluminum-base armor alloy has been resistance relative to a standard-purity version, purity levels (Fig. 10). A fixed, 8 h second ho-
sought for many years as a replacement for Al- to develop a viable homogenization schedule, to mogenization step was performed at tempera-
4.4wt%Mg alloy 5083. One obvious way to in- identify an optimal deformation temperature, tures from 470 to 505 C (880 to 940 F), and
crease strength of 5083-type alloys is to increase and to elucidate the effect of strain rate on edge- the enhanced-purity material displayed further
the magnesium content, but this raises the issue cracking resistance. The strain to failure in hot improvements in strain to failure. However, the
of stress-corrosion-cracking susceptibility torsion was used as an assessment of resistance standard-purity version actually displayed a sur-
(when sensitized by exposure to warm tempera- to edge cracking. prising decrease in strain to failure with homog-
tures for long periods) and decreased workabil- Lab-scale ingots were cast at the two impu- enization temperature (Fig. 10). This decrease
ity. In fact, the propensity for edge cracking of rity levels, and torsion specimens were ma- was accompanied by a change in fracture mode
aluminum-magnesium alloys during hot rolling chined from locations equidistant from the sur- to quasi-cleavage, which may have been caused
generally increases with magnesium content and faces of the ingots to minimize effects of by solid metal induced embrittlement by the im-
impurity levels. segregation in the ingots. Specimens were de- purities (Ref 8). This surprising decrease in hot
The army was evaluating a candidate alu- formed at a typical commercial strain rate for ductility and edge-cracking resistance was found
minum-magnesium alloy with 8.5 wt% Mg, and alloy 5083 over a wide range of deformation efficiently and inexpensively by hot torsion sim-
funds were not available to produce the several temperatures. In the as-cast condition, that is, ulation of hot rolling.
10,000 kg (22,000 lb) ingots needed to develop not homogenized, the flow stress decreased The homogenization study was repeated with
commercial rolling parameters for the alloy. A monotonically with deformation temperature, specimens deformed at a strain rate an order of
hot torsion investigation was initiated (Ref 8) to and there was essentially no difference in flow magnitude lower than typical for 5083 (0.1 s1
Deformation temperature, F instead of 1.2 s1). Strain to failure improved dra- recommended that the U.S. Army evaluate the
500 600 700 800 matically for both purity levels, with the en- effects of this slow strain rate on overall cost.
280
hanced-purity material displaying a relatively
Equivalent tensile flow stress, MPa
Equivalent tensile flow stress, ksi

260 high strain to failure of 2.7 at the lower strain rate.
36
240 34
The lab-scale ingots were homogenized and REFERENCES
32 rolled, with visual edge cracks correlating with
220 1. H.J. McQueen and J.J. Jonas, Hot Work-
30 the strain-to-failure values. A homogenization
200 28 and a rolling schedule for enhanced-purity ability Testing Techniques, Metal Forming:
26 material were recommended to the U.S. Army, Interrelation Between Theory and Practice,
180
Al-8.5wt%Mg 24 emphasizing that the lower strain rate would 1971, p 93428
160 As-cast, = 1.2 s1 2. C. Rossard, IRSID, Iron and Steel Institute,
22 produce acceptably low edge cracking. It was
Enhanced purity France
140 20
Standard-purity limits
120 18 3. D.S. Fields and W.A. Backofen, Proc.
200 250 300 350 400 450 ASTM, Vol 57, 1957, p 1259
Homogenization temperature, F
Deformation temperature, C 4. G.E. Dieter, Mechanical Metallurgy, 2nd
800 880 900 920 940
(a) 3.0 ed., McGraw-Hill, Inc., 1976, p 385, 652
= 0.1 s1
Deformation temperature, F
Equivalent tensile strain to failure, f
5. J.R. Pickens, W. Precht, and J.J. Mills, Hot
Al-8.5wt%Mg Rolling Simulation of Electromagnetically
2.5
500 600 700 800 Enhanced purity
Cast and Direct-Chill Cast 5182 Aluminum
Equivalent tensile strain to failure
Commercial purity
2.0 Tdef = 427 C (800 F) Alloy by Hot Torsion Testing, Proceedings
0.8
= 0.1 s1
of Mater., 30th Sagamore Army Research
1.5
Conference: Innovations in Materials Pro-
0.6 cessing, Aug 1983, G. Bruggeman and V.
Weiss, Ed., Plenum Press, 1985, p 101116
1.0
0.4 = 1.2 s1 6. A. Gittins, J.R. Everett, and W.J.M. Tegart,
Metalwork. Technol., Vol 4, 1977, p 377383
0.2 0.5
= 1.2 s1 7. W. Precht and J.R. Pickens, A Study of the
Hot Working Behavior of Al-Mg Alloy
0 0
250 300 350 400 450 500 420 427 470 480 490 500 510 5052 by Hot Torsion Testing, Metall. Trans.
16 h A, Vol 18, Sept 1987, p 16031611
Deformation temperature, C Homogenization temperature, C 8. J.R. Pickens and F.H. Heubaum, A Study
(b)
of the Hot Workability of Al 8.5 wt. Pct. Mg
Fig. 10 Alloy purity and strain-rate effects on equiva- Alloys for Armor Plate Applications, final
Fig. 9 (a) Equivalent tensile flow stress and (b) strain to lent tensile strain to failure for the second ho-
report on contract DAAG46-85-C-0034,
failure of as-cast specimens deformed by hot mogenization step of 8 h at different temperatures follow-
torsion at various temperatures ing a first homogenization of 427 C (800 F) for 16 h Report MTL TR 89-3, Jan 1989
Chapter 10
Thermomechanical Testing
Stphane Guillard and Koushik Ray, Concurrent Technologies Corporation
THERMOMECHANICAL PROPERTIES chanical processes. A series of examples, each Typical Types of Problems
TESTS, as covered in this chapter, are used to using the same general approach, is also in-
gain insight into the causes of problems that cluded. They illustrate how the process, the prob- Essentially three types of problems can occur
arise during a given thermomechanical process. lem, and previous knowledge influence the way during a thermomechanical process, and each
Material data can be divided into two categories: the test was selected, conducted, and analyzed. type requires the use of a specific test or analy-
mechanical properties and thermophysical prop- To get information about a specific thermo- sis tool. First, the part might not get formed, or
erties. Tests needed to obtain the mechanical mechanical process, the testing must resemble the final geometry might differ from expecta-
properties for most process modeling endeavors the process. Table 1 highlights how different tions. An example would be a lack of complete
are described in previous chapters. Because this processes correspond to various types of me- die filling during a forging operation. This type
chapter deals with thermomechanical properties chanical tests. The state of stress experienced of problem is usually solved through redesign of
testing, it is important to mention at least some during the actual metalworking operation should the die, die series, or preform, or by adjusting
of the tests needed to obtain properties, such as be present during the test. This correlation is the process variables (e.g., temperature, strain rate).
specific heat, coefficient of thermal expansion, most direct way of ensuring that phenomena ob- Such design issues are outside the scope of this
thermal conductivity/diffusivity, and density, all served during the tests can be expected during chapter; however, physical modeling, using wax
of which can significantly affect the behavior the actual process. or plasticine, can be one effective way of ex-
of a material during processing. Descriptions One key element is to determine what param- ploring such problems.
of these tests are given in the section Thermo- eters will be considered as possible influences Recently, this type of issue has been tackled
physical Properties Tests in this chapter. on the phenomenon being observed. For exam- very effectively through the use of process
Successful outcomes from thermomechanical ple, when grain size during a metalworking modeling by finite-element-based software
testing require proper test-selection, appropriate process is the chief concern of the process de- codes, such as DEFORM (Scientific Forming
experiment design to maximize efficiency in ob- signer, the parameters likely to influence grain Technologies Corp.) or MSC.MARC (MacNeal-
taining test results, and using those results in an size should be considered. Some parameters are Schwendler Corp.), or by recently developed
effective manner. Table 1 provides information easily contemplated, such as initial grain size, finite-volume-based software, such as MSC.
about the most common processes, but the poten- temperature, strain, and deformation rate. Other SUPERFORGE (MacNeal-Schwendler Corp.).
tial number of process and problem combinations parameters are less obvious, such as prior ther- Independent of the type of analysis used, the ac-
is extremely high. Therefore, the intention here is momechanical history or frictional effects. The curacy of the results provided by these codes not
not to cover every possible situation, but rather to numbered examples provided in this chapter only depends on how well the process model
present through examples how to approach solv- demonstrate how significant parameters were matches the actual process but also on how ac-
ing typical problems associated with thermome- selected for specific tests. curate and reliable the material data are.
A second type of problem commonly experi-
enced arises when the desired geometry is ob-
tained, but cracking occurs or, more generally,
Table 1 Guide to selecting a thermomechanical process test based on problem some type of defect is observed. Examples in-
encountered and process used clude edge cracking during rolling operations,
Process Problem Test central bursting during extrusion, cavitation dur-
Rolling Surface cracks, edge cracking, barreling, center Upset, plane-strain indentation, partial width
ing superplastic forming of aluminum alloys, or
bursts indentation, hot tensile (f, Af) even cracking of the tooling itself. In these in-
Forging Hot shortness, center bursts, triple-point cracks/ Wedge forging, side pressing, isothermal stances, combining the use of damage criterion
fracture, grain-boundary cavitation/fracture, tensile or compressive, temperature-cycling (Ref 15) with process modeling generally al-
shear bands/fracture tensile
Closed-die forging Laps or buckles, die fill Wedge forging, side pressing, isothermal
lows enough insight to generate a solution that
tensile or compressive, temperature-cycling usually consists of redesigning dies/pass sched-
tensile ules or adjusting the processing parameters.
Extrusion, wire drawing Center bursts, fracture Hot tensile (f, Af) Another approach, which involves extensive
Wire/rod drawing Surface defects: fir-tree cracks, splitting, crows Boundary layer lubrication studies
feet, orange peel; center bursts (due to
testing, consists of determining a map showing
inhomogeneous deformation) the occurrence of damage as a function of vari-
Sheet metal forming Tearing (localized necking), wrinkling and Limiting dome height, tensile (R, n values), ous processing parameters. The well-known
earring, Lders bands, orange peel, fracture punch (forming-limit diagram) forming-limit diagram (Ref 6, 7) is used to avoid
failure through proper design of strain paths
Note: Besides thermomechanical testing, a number of tools can be useful in solving the type of problems listed above. Among them are reviewing tool
and die design rules, using finite-element analysis, physical modeling, and metallography, including optical and scanning electron microscopes.
(Fig. 1). Another example of this approach is the
strain-rate sensitivity (m) map (Fig. 2).
Chapter 10: Thermomechanical Testing / 129
Although Fig. 2 does not specifically identify ically through necking but through cavitation. resistance. Large grains, on the other hand, in-
regions where damage will occur, it does high- The cavitation, which is the nucleation and crease high-temperature creep resistance. As a
light regions of high strain-rate sensitivity. growth of tiny cavities, probably occurs at the result, inadequate properties can generally be
These regions occur where damage is less likely intersection of surfaces of grain-boundary slid- traced back to inadequate microstructure.
because necks tend to push deformation to ing or cooperative grain-boundary sliding within Figure 3 relates the overall approach for solv-
other, nonnecking areas, thus diffusing the neck- the material (Ref 8). ing thermomechanical processing problems to
ing mechanism. A strain-rate sensitivity about Consequently, although a given processing microstructural evolution. It highlights test
0.2 to 0.3 is considered quite satisfactory for map might seem suitable to design how a mate- choice as a function of both the thermomechan-
most typical forming operations where necking rial should be processed under specific conical process used and the problem experienced.
is a concern. ditions, it might not correspond to the controlling This chapter addresses which tests should be
Of course, avoiding failure at strains much mechanism throughout the planned processing conducted, how they should be conducted, and
larger than those experienced during forming range of the material. In this particular situation, how they should be interpreted to gain sufficient
operations can be achieved with materials hav- a cavitation map would be used in conjunction insight into the microstructural evolution taking
ing very high strain-rate sensitivity. For exam- with the strain-rate sensitivity map to ensure safe place in a material during thermomechanical
ple, m reaches values above 0.5 and as high as processing throughout the intended strain range. processing. Once combined with other tools at
0.9 in so-called superplastic materials, such as Therefore, limitations in using processing maps the process designers discretion, this informa-
Ti-6Al-4V or aluminum 7475. Interestingly, the show how carefully they must be utilized. Indeed, tion can be used so that the process yields the
failure of 7475 under these conditions is not typ- the various phenomena likely to occur must be desired properties in the final product.
known to determine what processing map or
combination thereof should be considered.
Physical modeling can also be used to solve Thermophysical Properties Tests
defect problems in thermomechanical process-
ing. However, it is only helpful when an analog In addition to thermomechanical behavior,
exists, such as plasticine for hot steel, and when thermophysical properties are also of particular
the damage is not caused by a microstructural- interest when designing or optimizing a metal-
level phenomenon but rather by a stress- or working process. Thermophysical properties
strain-state effect. For example, central bursting include specific heat, coefficient of thermal ex-
during extrusion can be physically modeled be- pansion, thermal conductivity/diffusivity, and
cause it is caused by tensile stresses along the density. In addition, interfacial heat-transfer co-
center of the product. Cavitation in aluminum efficients between workpiece and tooling is im-
cannot be physically modeled because it is portant.
caused by accommodation of a microstructural Specific Heat. Enthalpy is the overall heat
phenomenongrain-boundary sliding. (or energy) at constant pressure. The average
A third type of problem relates to poor prop- specific heat, Cp, over a temperature range at
erties of the product itself, which arise from constant pressure is defined as the change in en-
inadequate microstructure. The link between thalpy divided by the temperature change:
properties and microstructure is generally
well understood. For example, small-grained Cp d(HT H273)p/dT (Eq 1)
Fig. 1 Forming limit diagram for 301 stainless steel in
microstructures are known to favor room-
the 14 hard condition. Samples were machined
in the rolling direction. Source: Ref 7 temperature tensile strength and fatigue crack
Fig. 2 Map depicting the evolution of strain-rate sensitivity as a function of temperature and strain rate for CF-8M Fig. 3 Overall approach for solving problems encoun-
stainless steel. Source: Ref 7 tered during thermomechanical processing
where HT is the enthalpy at a given temperature, materials. Temperatures can range up to 2000 C tially force heat to flow through both the speci-
H273 is the enthalpy at 0 C, and T is the tem- for some systems. men and a known conductivity reference mate-
perature. The specific heat influences the rate of The three most common methods of measur- rial. Because lateral heat flow is prevented
cooling and heating of the workpiece and how ing CTE are through the use of a quartz through the use of various guards, the amount of
much the temperature changes with local heat- dilatometer, a vitreous silica dilatometer, or an heat flowing through the sample and the refer-
ing by deformation and friction. interferometer. The quartz dilatometer (Fig. 6) ence material is the same. Therefore, measuring
The two most common methods of measuring uses a quartz tube as a reference for the expan- the temperature gradient through the specimen
specific heat are differential thermal analysis sion of the specimen. Quartz is a crystalline provides an indication of how easily heat flows
(DTA) and differential scanning calorimetry form of silica that has an extremely low and through it. For example, a small temperature dif-
(DSC). Figure 4 shows a DTA cell in which a well-defined thermal expansion coefficient. The ference means high conductivity.
specimen and a reference are heated by a single specimen is contained in the quartz tube and The laser flash technique, a newer method to
heating element while each temperature is mon- pressed against a quartz push rod whose motion measure thermal diffusivity, utilizes a transient
itored. These temperature curves can be used to is monitored by a linear variable differential or sudden change method as opposed to a
derive the specific heat of the sample, but not transformer (LVDT). The specimen temperature steady-state method. The sample surface is
with great accuracy. is monitored by a thermocouple maintained in blackened by carbon to give it an emissivity re-
The DSC cell (Fig. 5) has separate heaters for an inert gas environment. The ratio of specimen sembling a black body. It is then suspended in a
the sample and the reference material, with the strain to temperature change gives the CTE. furnace maintained at a known temperature. A
temperatures of each monitored and controlled. A vitreous silica dilatometer essentially uses sudden flash from a laser heats one side of the
The amount of energy required to force both to glassy silica as the reference material. Current sample. The emission of infrared radiation from
follow the same heating, or cooling, curve is systems use optical encoders or optical interfer- the other side is monitored, and the emission
measured, which provides a direct, accurate in- ometers to measure the displacement. However, curve is analyzed to determine the heat diffusion
dication of the specific heat. the overall principle does not change. The spec- rate. Essentially, the diffusivity is the measure-
Thermal Expansion and Density. The coef- imen is positioned in a highly uniform and accu- ment of the emitted infrared heat curve by using
ficient of thermal expansion (CTE) is a meas- rately controlled furnace, and the extension is the time, t 1/2, to reach half of the maximum tem-
urement of strain resulting from a temperature measured and compensated for the small expan- perature. A heat loss parameter, w, is a geomet-
change. Instruments that measure CTE are based sion of the silica. ric constant calculated for the specimen geome-
on a precisely controlled and highly uniform Density is obtained quite easily as a function try. The specimen thickness, L, is also used:
temperature furnace combined with a system for of temperature once the CTE as a function of
measuring the change in length of the specimen temperature is known, based on conservation of D wL/t1/2 (Eq 3)
or a change relative to a reference material. mass.
Available temperature ranges are usually only Thermal conductivity and thermal diffu- Figure 7 shows a schematic of a laser flash ex-
limited by the availability of stable reference sivity define the ability of the workpiece or tool- perimental setup.
ing to conduct or spread heat, respectively. The
difference between the two comes from the heat
capacity and density of the material:
K D Cp (Eq 2)
where K is the thermal conductivity, D is the

thermal diffusivity, and is the density.
Therefore, materials with the highest diffusivity
have high conductivity and low heat capacity
and density.
During metalworking, the thermal conductiv-
ity of both the workpiece and the tooling can be
important. Two common methods for measuring
these properties are the heat flow meter and the
guarded plate method. These techniques essen- Fig. 7 Laser flash system
Fig. 4 Differential thermal analysis cell. S, specimen;

R, reference; T, temperature
Fig. 5 Differential scanning calorimetry cell. S, speci-

men; R, reference Fig. 6 Quartz dilatometer. LVDT, linear variable differential transformer
Testing for the Heat-Transfer Coefficient. Table 2 Example of lower and upper
Deformation processing often involves heat bounds selected for three factors
transfer, and one of the difficulties in modeling is considered to optimize a process
determining the input value of the heat-transfer Factor Lower bound () Upper bound ()
coefficient. The most commonly used method of
determining the heat-transfer coefficient is the 1. Strain, m/m 0.2 1.5
2. Strain rate, m/ms 0.1 10
inverse method. In this method, the temperatures 3. Temperature, C 900 1150
of the workpiece and tooling are measured at dif-
ferent locations as a function of time. Subse-
quently, the process is modeled with trial values
of the heat-transfer coefficient to match the tem-
perature measured in the experiments. The trial- Table 3 23 factorial design of experiments
and-error method for determining the heat- Test Factor 1 Factor 2 Factor 3 Fig. 8 Wedge test specimen. (a) As-machined speci-
transfer coefficient can be programmed to loop men. (b) Specimen after deformation. Source:
the simulations until an optimal value of the co- 1 1 1 1 Ref 12
2 1 1 1
efficient is determined (see the section Testing 3 1 1 1
for the Heat-Transfer Coefficient in the article 4 1 1 1
enough to be representative of the behavior of
Testing for Deformation Modeling in Volume 5 1 1 1 the cast material are machined from the cast
8 of ASM Handbook for a discussion of this for 6 1 1 1 structure whose microstructural evolution prop-
7 1 1 1 erties are to be determined. Typically, 12 to 15
rolling operations). 8 1 1 1
grains across the diameter of the specimen is
considered the minimum size to obtain a repre-
Designing Thermomechanical Tests Note: For simplification, the lower bound of each variable is represented
by 1 and the upper bound by 1. Refer to Table 2 for the appropriate sentative behavior. The wedge specimens are
values. heated to different preheat temperatures. These
The testing should be designed to maximize specimens are then deformed to one specified
the chances of deriving useful information. The height reduction and annealed at different tem-
key element is to identify the mathematical form peratures for various time lengths to obtain a re-
used to express the influence of the processing lower bounds of the factors are chosen (Table 2). crystallized microstructure. Finally, the speci-
parameters and other parameters on the variable These upper and lower bounds are the practical mens are quenched. Through microstructural
of interest. For example, when looking at the re- and realistic limits within which the value of the examination of each quenched specimen, the
crystallization behavior of a material, the fol- parameter is expected to lie. Tensile test meas- process engineer can evaluate how strain affects
lowing can be assumed to be correct: urements are made on samples processed under the recrystallization behavior. Therefore, for one
the combination of factors listed in Table 3. The combination of annealing temperature and hold
X 1 exp[0.693(t/t0.5)n] (Eq 4) two values of each factor combine with the two time, the engineer needs only one test to deter-
values of every other factor, thus leading to 2n mine the ideal strain to input into the material. It
where X is the percentage of recrystallization, t tests (n number of factors). Following this is possible to obtain the entire range of strain
is the dwell time, t0.5 is the time needed to obtain method, the factors, and more importantly the values from only one sample. From the collec-
50% recrystallization at that temperature and combination of factors, having the greatest in- tion of samples, it is also possible to determine
strain rate, and n is a material constant. Equation fluence on the quality of the product (i.e., tensile how the strain effect varies with the other pro-
4 requires tests for determining the effect of tem- strength) are known. cessing parameters and to specify how much
perature, strain, and strain rate on t0.5. Variations on this method exist, such as the strain should be input into the material depend-
Factorial Design Testing. Knowing the fractional factorial method. A number of excel- ing on the various processing parameters.
mathematical form that the results should take is lent reference books should be consulted prior to
not a prerequisite. For example, techniques such using any of these techniques (Ref 911). More
as response surface determination can be used to recently, various software packages have been Obtaining, Analyzing, and Using
obtain polynomial expressions that do not con- developed and are available commercially to Thermomechanical Testing Results
tain physical meaning, but are often sufficient speed up both the design of experiments and the
for material and process modeling applications analysis of results. These software packages in- When attention has been paid to choosing the
(e.g., example 3 in this chapter). A simple, yet clude Design-Expert (STAT-EASE Inc.) and right test and carefully designing the test matrix,
effective, technique is called factorial design of ECHIP Inc. (Hockessin, DE). analyzing the results becomes a simple task.
experiments. Improving Testing Efficiency. It has been as- Indeed, the form the results should take is usu-
Often, the process parameters that can affect sumed that one data point, such as a grain size ally determined beforehand, and this step then
the quality of the product are known, but the value, is obtained from each sample. However, essentially consists of characterizing the sam-
relative importance is not clear. A sensitivity samples themselves can be designed so that ples. The principal tool is undoubtedly the opti-
analysis is then needed, and factorial design of many data points are obtained from each sample. cal microscope. In most situations, the mi-
experiments aids in understanding the effects of The wedge-test specimen, because of its shape, crostructural features of interest to the process
the process parameters (now called factors) in allows the metallurgist to evaluate the effect of engineer are the grain size, percent recrystalliza-
the most efficient manner. The method of facto- uniformly varying amounts of strain on the mi- tion, amount and distribution of second phase,
rial design is best demonstrated by example. crostructural evolution of a material in a single and texture. These features directly influence the
Obtaining a superior microstructure in a ma- test (Fig. 8). The relevant limits of strain can be properties of the product. Image analysis can be
terial through optimization of the factors (tem- fixed by designing the geometry of the wedge very helpful at this stage, especially when a
perature, strain rate, strain) is desired. The meas- appropriately (Ref 12). large number of samples is involved.
ure of superior microstructure will be in the An important practical application of this test Use of the Results. At this stage, the process
form of tensile strength measurements made helps the process engineer determine the optimal engineer should be able to express the relation-
from samples that have undergone different pro- reduction-anneal sequence for obtaining a re- ship between parameters that can be controlled
cessing parameter combinations. A schedule of crystallized structure after breakdown of a cast and variables that influence, in a known fashion,
experiments is set up in which the upper and ingot. In this instance, wedge specimens large the properties of the product. The most effective
approach is then to combine this relationship size was measured and remained almost constant tural evolution. The process engineer must opti-
with a numerical model of the process, such as a at approximately 11.5 m for deformation up to mize all these variables to produce the most de-
finite-element model. This combination allows a strains between 1 and 1.4. This result is in con- sired microstructure. In addition, because of the
what if scenario to be considered, with the mi- trast to other findings that show increasing grain high costs involved in trying something new dur-
crostructure of the product as one of the aspects size with increasing strain. The explanation is ing an actual cogging operation, most operators
to optimize. two-fold. First, the other studies were performed only apply small changes to a process known to
using uniaxial tensile tests, which do not accu- work at least reasonably well.
rately reflect the biaxial state of stress present Therefore, obtaining relevant information
Examples of Thermomechanical during superplastic forming. Second, the other from laboratory-scale testing is the key to finding
studies considered a starting material that had out what works. Although laboratory-scale work
Testing: Design, Experiment, equiaxed grains. The process engineer had no- is performed to simplify real-life situations, over-
and Analysis ticed that the starting material had slightly elon- simplification is detrimental. When interested in
gated grains, which could influence the results. cogging or radial forging to break down cast in-
Four examples illustrate how the various con- As a consequence of conducting tests more gots, performing single-hit laboratory-scale tests,
siderations in testing are successfully used to directly applicable to the process, the process for example, is unlikely to yield appropriate re-
solve practical thermomechanical processing engineer was able to ascertain that grain growth sults. For this reason, double-hit and even four-
problems. would not be a problem in manufacturing the hit tests have been considered by process design-
Example 1: Superplastic Forming of component. Had the specific tests not been run, ers (Ref 13, 14). Even a factor as simple as load
Aluminum Alloy 7475. This first example il- the component might have been determined a is unlikely to be represented accurately if a sin-
lustrates how a biaxial test (closely resembling very risky candidate for superplastic forming. gle-hit test is used to represent the full range of
the intended process) allows the process engi- The process engineer might have either in- strain expected during the operation, when the
neer to verify the viability of superplastic form- creased the level of process control or selected total strain is, in fact, obtained from four succes-
ing for a sheet of aluminum alloy 7475. In su- another manufacturing method, both of which sive hits (Fig. 10). In this situation, using the re-
perplastic forming, a key factor is to keep the likely would have resulted in increased manu- sults from a single-hit test would be accurate for
grain size under control to within a small level facturing costs. the first hit, would lead to overprediction of the
(less than 15 m), knowing the behavior of Example 2: Cogging of Nickel-Base load during most of the second hit, and would
grain-size growth during the process. Before Superalloy 718. Another example of how a test lead to underestimation of the maximum loads
committing funds to making tooling and actually should be designed to closely fit the thermome- for the third and fourth hits. The process designer
forming a component, the process designer de- chanical process can be seen in how compres- could not even conclude whether the test repre-
cided to verify that the grain size would not be a sion tests were designed by various researchers sents an upper or a lower bound.
problem. To do so, the material was tested by su- to depict phenomena occurring during the ingot Adding to the confusion, the previous obser-
perplastically forming a double-well pan, which breakdown by cogging of superalloy ingots. vation is itself a function of the processing tem-
is a fairly standard test item for superplastic Cogging consists of a succession of open-die perature. Indeed, Fig. 11, which represents the
forming (Fig. 9). forging-type hits along the length of an ingot to same comparison between single-hit and four-
This test was chosen as opposed to a standard break down the casting structure and promote
uniaxial tensile test because the stress state recrystallization and grain refinement. As in
needed to be biaxial to accurately reflect what other processes, the resulting microstructure is
happens in the aluminum sheet during actual strongly influenced by the way the ingot is
forming. Following deformation, samples were processed. In particular, the amount of strain re-
cut from the pan at various locations and pre- ceived during each hit, the number of hits, and
pared for metallographic examination. The grain the dwell time between hits affect microstruc-
Fig. 10 Stress-strain curves comparing single-hit (open

circles) and multiple-hit (continuous lines)
tests for wrought alloy 718 at 950 C. Source: Ref 13
Fig. 11 Stress-strain curves comparing single-hit (open

circles) and multiple-hit (continuous lines)
Fig. 9 Double-well pan superplastically formed using aluminum alloy 7475 tests for wrought alloy 718 at 1050 C. Source: Ref 13
hits tests but at 100 C higher than in Fig. 10, re- Table 4 illustrates the levels of the three fac- that log e was used instead of e itself, for con-
veals that at 1050 C using the single-hit test re- tors for a central composite rotatable design. As venience only):
sults to predict the load would yield acceptable the 1 values in Table 3 show the normalized
results. However, because of the poor results ob- levels for the extrema values of the parameters X 3268.63 7.017 log e 4.7605T
tained at 950 C, the process engineer would for the trials, the 1 values in Table 4 represent
56.84e 0.001776T 2 12.52e2 (Eq 6)
probably not trust the results at 1050 C. normalized values of corresponding parameters.
The uncertainty is not limited to loads. However, in a central composite rotatable de-
where T is temperature in Kelvin. Equation 6 is
Microstructurally, single-hit tests yield increased sign, the factorial design trials are only a subset
a response surface equation, and contour plots of
levels of recrystallization, which is explained by of all the trials. So-called star trials, with values
that surface can be drawn. For example, Fig. 13
the fact that they favor adiabatic heating and, exceeding the extrema values, are also included
depicts the influence of strain and strain rate on
therefore, higher temperatures. Also, extensive (represented in Table 4 by normalized values of
percent spheroidization at 1330 K. The process
deformation is applied without much chance for 1.68, 1.68, and the square values). Ultimate
engineer can use this plot, or others at various
recovery (Ref 13). In a multiple-hit test, dwell explanation for these values is based on group
temperatures, to determine the optimal process-
times are available for recovery to occur, thereby theory. The additional star trials make the com-
ing regime for this specific material if it is
reducing the driving force for recrystallization. posite rotatable design of experiments a more
known that maximizing the percent spheroidiza-
Whatever phenomena might occur during the robust method than factorial design.
tion is beneficial to the properties of the product.
process, and especially when those phenomena Compression tests were conducted for all
A full derivation of the central composite rotat-
are unknown, testing must closely match the ac- conditions. Due to friction at the sample/platen
able design for this specific example, which also
tual process in terms of temperature, stress state, interfaces, the samples exhibited some amount
includes consideration of the spheroidized grain
strain rates, strain paths, lubrication, or prior of barreling. As a result, the strain across the
size, is provided in Ref 15.
thermomechanical history of the material. midplane of the sample, perpendicular to the
Example 3: Processing of Near-Gamma compression axis, was not uniform. Because
Titanium Aluminide. This example illustrates strain was being studied for its influence on per-
how statistical design of experiments can be cent spheroidization, it was critical to know
used to maximize the amount of information ob- what strain specific portions of the samples had
tained from very few tests. The challenge con- experienced. In fact, because the central rotat-
sisted of quantitatively determining the influ- able design called for specific strains to be con-
ence of processing on percent spheroidization sidered, finite-element modeling was used
(Ref 15). The processing parameters selected to predict what areas of the samples experi-
were temperature, strain, and strain rate during enced those specific strains. Figure 12 shows
the deformation of the material. The type of ex- where a micrograph frame had to be taken to be
perimental design selected was composite rotat- sure that the material had experienced a strain
able design (Ref 11), which is just a more ad- of 0.70 0.05.
vanced version of factorial design. However, the Using an image-analysis system, the percent
composite rotatable design allows the determi- spheroidization was measured for all samples.
nation of a response surface equation. Therefore, The results were input into a treatment ma-
if percent recrystallization could be measured on trix where the coefficients corresponding to
samples having been deformed under conditions temperature, strain, strain rate, and the combi-
of temperature, strain (e) and strain rate (e) rec- nations were determined. Although it is not the
ommended by the design, then a polynomial intent of this chapter to provide details about
equation could be obtained: any of these methods, the central composite
rotatable design does allow determination of
the significance of the various effects so that
X ae be cT de2 ee2 f T 2 those with little importance can be deleted
geT heeT iee (Eq 5) from the final mathematical expression (note
Table 4 Central composite rotatable design of experiments for 3 factors

Test T 2 2 2 T T
1 1 1 1 1 1 1 1 1 1
2 1 1 1 1 1 1 1 1 1
3 1 1 1 1 1 1 1 1 1
4 1 1 1 1 1 1 1 1 1
5 1 1 1 1 1 1 1 1 1
6 1 1 1 1 1 1 1 1 1
7 1 1 1 1 1 1 1 1 1
8 1 1 1 1 1 1 1 1 1
9 1.68 0 0 2.83 0 0 0 0 0
10 1.68 0 0 2.83 0 0 0 0 0
11 0 1.68 0 0 2.83 0 0 0 0
12 0 1.68 0 0 2.83 0 0 0 0
13 0 0 1.68 0 0 2.83 0 0 0
14 0 0 1.68 0 0 2.83 0 0 0
1520 0 0 0 0 0 0 0 0 0 Fig. 12 Strain contours obtained after upsetting a
compression sample in a median plane (a)
parallel to and (b) perpendicular to the compression axis.
Note: The values 1.68 and 2.83 are chosen to yield special properties that facilitate the analysis of the results. Only the first three columns are used to
define the values of the processing factors during the experiments. The other columns provide information used to determine the values of the param-
Note the micrograph frame positioned to ensure observa-
eters corresponding to the respective effects and interactions at the top of each column. tion of microstructure resulting from imposition of 0.7
strain.
Example 4: Using Damage Modeling to mum tensile stress, is the equivalent stress, mizing the processing parameters, such as die
Eliminate Defects During Ingot Side and ep is the equivalent strain. design, temperature, and ram speed; otherwise,
Pressing. The objective in calculating damage in There are two limitations when using the cracking might occur. Because of the high cost
a metalforming operation is to determine regions damage parameter. First, the numbers are purely involved with full-scale trials to validate optimal
where failure is most likely to occur. Because a relative measure of the probability of damage parameters, physical modeling with plasticine
failure usually is caused by accumulation of ten- and cannot be used to determine with certainty can be used. In one instance, the plasticine ingot
sile strain, the damage number should reflect the whether damage will occur. Secondly, there cracked during the side-pressing operation. The
accumulation of tensile strain at a point along might be other modes of damage not incorpo- cracking is shown in Fig. 14 as voids adjacent to
the path of the evolution during metalforming. rated in the damage parameter, such as what the inserted hard beads. The challenge was to
Damage evaluation considers more than strain might cause the material to fail at otherwise- optimize the parameters in such a way to prevent
accumulation. It also considers the effective act- predicted low damage parameter values. The the cracking. The process engineer constructed a
ing stress that might have assuaged or aggra- knowledge, then, to be gleaned from damage finite-element analysis model and plotted the
vated the accumulation of the strain, which predictions is where the most critical areas dur- damage values associated with it during the
makes the number a more realistic reflection of ing thermomechanical processing are and how forming operation (Fig. 15). Cracking corre-
how damaging the strain has been to the ma- could an adjustment of the processing parame- sponded to the region with the highest damage
terial. A damage value can be determined: ters reduce the probability of failure. value in the model, which is expected. Next, the
ef
The concept of using damage values as poten- processing parameters were modified in various
s * tial indicators of areas of concern can be shown ways to lower the damage contour values. By ju-
s de
0
p =C (Eq 7)
by damage modeling a side-pressing operation. dicious changes in the platen design, the process
K-Monel has limited ductility at warm working engineer was able to lower the contours (Fig.
where ef is the fracture strain, * is the maxi- temperatures. Hence, there is a need for opti- 16). Incorporating the suggested changes in the
Fig. 14 Plasticine ingot representing side pressing of a

K-Monel ingot. Hard beads have been embed-
ded to illustrate the presence of tensile stresses at the cen-
ter of the ingot. Damage seen is directly related to the ten-
sile stresses.
Fig. 13 Effect of strain and strain rate on percent spheroidization of Ti-49Al-2V at 1330 K. Source: Ref 15
Fig. 15 Model showing damage values corresponding to the process used to manu- Fig. 16 Model showing damage values obtained after modification to the platen
facture the part shown in Fig. 14 used to manufacture the part shown in Fig. 14
REFERENCES 8. A.H. Chokshi, A.K. Mukherjee, and T.G.

Langdon, Mater. Sci. Eng. Res., Vol 10,
1. M.G. Cockcroft and D.J. Latham, Ductility 1993, p 237274
and the Workability of Metals, J. Inst. Met., 9. C.R. Hicks, Fundamental Concepts in the
Vol 96, 1968, p 3339 Design of Experiments, Holt, Rinehart, and
2. M. Oyane, Criteria of Ductile Fracture Winston, 1973
Strain, Bull. Jpn. Soc. Mech. Eng., Vol 15, 10. W.G. Hunter and J.S. Hunter, Statistics of
1972, p 15071513 Experimenters, John Wiley & Sons, 1978
3. A.M. Freudenthal, The Inelastic Behaviour 11. W.J. Diamond, Practical Experiment
of Solids, John Wiley & Sons, 1950 Design for Engineers and Scientists, Van
4. D. Zhao, J.P. Bandstra, and H.A. Kuhn, A Nostrand Reinhold, 1989
New Fracture Criterion for Fracture 12. S.L. Semiatin, Workability Testing Tech-
Prediction in Metalworking Processes, Con- niques, George Dieter, Ed., American
current Engineering Approach to Materials Society of Metals, 1984, p 202
Processing, S.N. Dwivedi, A.J. Paul, and 13. M.C. Mataya, Simulating Microstructural
F.R. Dax, Ed., TMS, 1992, p 107119 Evolution during the Hot Working of Alloy
Fig. 17 Successfully side-pressed plasticine ingot after
5. J.P. Bandstra, 3D Extension of Kuhn 718, JOM, Jan 1999, p 1826
platen modification suggested by model
shown in Fig. 16. Note that the plasticine around the hard Surface Fracture Criterion, CTC/JPB- 14. D. Zhao, S. Guillard, and A.T. Male, High
beads does not flow in this instance, suggesting that dam- M0469-95, CTC Memorandum, Concurrent Temperature Deformation Behavior of Cast
age has been minimized.
Technologies Corporation, 1995 Alloy 718, Superalloys 718, 625, 706 and
6. S.S. Hecker, Sheet Met. Ind., Vol 52, 1975, Various Derivatives, E.A. Loria, Ed., TMS,
plasticine model obtained defect-free parts (Fig. p 671675 1997, p 193204
17). The damage modeling exercise, in this case 7. Atlas of Formability Bulletins, National 15. S. Guillard, High Temperature Micro-
coupled with physical modeling for further cost Center for Excellence in Metalworking Morphological Stability of the (2 )
minimization, helped avoid the cost of running Technology, Concurrent Technologies Cor- Lamellar Structure in Titanium Aluminides,
in-plant experiments and prototypes. poration, p 15 Ph.D. dissertation, Clemson University, 1994
Chapter 11
Design for Deformation Processes
PRODUCT FORMS may be shaped by a wide gorization of decisions is somewhat arbitrary, as Why Use Deformation Processes?
variety of metalworking methods, and the suc- processing and part decisions can be interre-
cess or failure of a bulk working process de- lated. Preform (initial workpiece) geometry, the Cost, dimensions and tolerances, surface fin-
pends on the interaction of many variables, such deformation temperature, amount of force and ish, throughput, available equipment, and part-
as process conditions (strain rate, temperature, strain rate, the friction conditions, and the met- performance requirements dictate the material
and shape) and metallurgical conditions (e.g., allurgical condition of the workpiece are all in- and process selection for production. The sim-
flow stress, workability, and microstructure). terrelated factors in the design. Nonetheless, plest practice that still achieves the desired prod-
There are four major design considerations in Table 1 identifies some major factors in metal- uct form should be used. Deformation offers
selecting a deformation process: working design. These decisions are needed to many advantages, especially in terms of mi-
accentuate the advantages of deformation pro-
The workpiece material and its flow-stress crostructural benefits, but these processes also
cessing (Table 2), while still preventing prob- have disadvantages, the main ones being the cost
behavior
lems of inferior parts or poor equipment utiliza-
The fracture behavior of the material and of equipment and tooling.
tion (Table 3). It is also interesting to note that Advantages and Disadvantages of Defor-
the effects of temperature, stress state, and
many of the advantages of deformation can be- mation Processes. Discussions of process ad-
strain rate on ductility and fracture (i.e.,
come disadvantages if the process is poorly de- vantages must be approached carefully because
workability)
signed and/or poorly executed.
The desired microstructure needed to pro- most final parts are generally subjected to more
This chapter briefly introduces the basic de- than one type of manufacturing process. For ex-
duce an acceptable product and a determi-
sign concepts of bulk deformation processes that ample, a screw-machined part has been first cast,
nation of which process should be used to
are described in more detail in subsequent chap- hot rolled to bar stock, and possibly cold rolled
produce this microstructure
ters. This chapter also introduces reasons behind
The added constraints of available equip- or drawn prior to screw machining. A forging
the selection of a deformation process as a pro- may have been cast, hot rolled to bar form,
ment and economics in addition to flow
duction method for producing a part or product cropped into a billet, forged through multiple
stress, forming, and part-performance con-
form. Product form may be produced by a wide stations, and then finish machined. There are
siderations
variety of methods, and the basic goal of a de- steps needed to produce the starting material
The fourth consideration usually dominates the formation process is the same as other manufac- (ingot or cast shape), intermediate steps needed
other considerations, sometimes to the detriment turing processes; that is, a desired form of a par- to shape the material into a manageable interim
of the material being worked. It is also necessary ticular material is sought with certain form (bar, plate, tube, sheet, wire), and then
to consider the effect of friction at the tool/work- mechanical and/or physical characteristics at a steps needed to make the final part. Both
piece interface, as well as the ability of the tool to minimum cost. What are the characteristics of processes to make interim product forms or
withstand the loads and temperatures required to deformation processing that are desirable for de- stock and processes to make parts are included
deform the workpiece into the desired geometry. signers? What are the advantages of bulk form- here.
Proper design of deformation processes in- ing as opposed to achieving the shape by solidi- The objective of hot forging or hot rolling of
volves many decisions. The design includes both fication, by molding of powder, or by cast materials is to refine the structure that re-
part-related and process-related decisions, such machining? These general questions are dis- sults from solidification. To alter the inhomoge-
as those listed in Table 1. This division or cate- cussed briefly in this chapter. For a more com- neous structure due to solidification and to pro-
plete discussion of the equipment and methods
used for deformation processes, see Ref 1 to 4
Table 1 Design decisions associated with and subsequent chapters in this book. Table 3 Potential disadvantages of
deformation processes deformation processing
Part-related decisions Fracture-related problems
Part or product material selection Table 2 Advantages of deformation Internal bursts or chevron cracks
Geometry and dimensions to be produced processing Cracks on free surfaces
Required properties (mechanical, physical, and Cracks on die contacted surfaces
metallurgical) Improved internal quality due to compressive deformation Metal-flow-related problems
Process-related decisions Uniform grain structure End grain and poor surface performance
Equipment selection (type, rate, and load requirements) Elimination of casting porosity Inhomogeneous grain size
Starting material geometry (plate, bar, sheet, etc.) Breakup of macrosegregation patterns Shear bands and locally weakened structures
Workpiece temperature and tooling temperature Beneficial grain-flow pattern for improved part performance Cold shuts, folds, and laps
Orientation of part during deformation step(s) Improved toughness due to grain flow and fibering Flow-through defect
Location of flash or scrap loss Improved fatigue resistance due to grain-flow pattern Control, material selection, and utilization problems
Number of deformation steps Controlled surface quality Underfill, part distortion, and poor dimensional control
Lubrication and method of application Burnished surface can have improved fatigue resistance Tool overload and breakage
Starting microstructure and control of microstructure due to quality of as-forged surface Excessive tool wear
during forging sequence (preheat practice and High throughput due to potentially high rates of forming High initial investment due to equipment cost
intermediate heating steps, if any) Ability to produce a net-shape or near-net-shape part Poor material utilization and high scrap loss
140 / Process Design and Workability
duce a more workable microstructure, cast in-

gots and continuously cast slabs and blooms are
typically hot worked into interim product forms,
that is, plate, bars, tubes, or sheet. Large defor-
mation in combination with heat is very effec-
tive for refining the microstructure of a metal,
breaking up macrosegregation patterns, collaps-
ing and sealing porosity, and refining the grain
size.
Many design decisions are required in order
to take advantage of the benefits of deformation
processing while avoiding potential problems of
flow-related defects: fracture or poor mi-
crostructure. Some of the advantages of a part
produced by deformation are listed in Table 2. In
addition, the process can be tailored to achieve
tight control of dimensions for mass production,
and typically some net surfaces can be achieved.
While the goal is to achieve a net shape, it is rare
that a totally net shape is produced by bulk de-
formation processes, and some machining is
typically needed to produce a usable part. Sheet-
forming processes, however, often result in net
functional surfaces. The production rate for
many deformation processes can be high, so that
high-volume production requirements can be
met with efficient machinery utilization.
Disadvantages of deformation processes are
listed in Table 3. It is interesting to note that
many of the advantages of deformation also
show up on the list of disadvantages. If the de-
formation process is poorly designed and/or Fig. 1 The nine classes of manufacturing processes. The first row contains the primary forming (shaping) processes.
poorly executed, the sought-after advantages The processes in the vertical column below are the secondary forming and finishing processes. Source: Ref 6
will not be realized, and instead, an inferior part
will be produced. The categorization in Table 3
of a design. Further details on the selection of readily processed material and (2) it is the
is somewhat arbitrary because metal flow, frac-
manufacturing processes are described in Ref 7. final part cost that counts, not the initial ma-
ture, die wear, and tool stresses are so inter-
Process-selection charts introduced by Ashby terial cost.
linked. The decisions that the designer must
(Ref 6) are also useful in comparing manufactur- Fit the design to the manufacturing process.
make concerning the preform or initial work-
ing processes (e.g., casting, metalworking, poly- This argues for a process-first, materials-
piece geometry, the deformation temperature,
mer processing, power fabrication, and machin- second approach to materials selection. For
amount of force and forging speed, the friction
ing) during the conceptual stage of design. the many applications where performance is
conditions, and the metallurgical condition of
Much attention has also been directed in re- not paramount or difficult to achieve, this is
the workpiece are all interrelated. Decisions
cent years to developing design rules and com- the approach to use.
about the deformation process (e.g., Table 1)
puter methods for enhancing manufacturability Design each part to be easy to make. The in-
must be made to accentuate the advantages
through proper design. This approach, some- herent capabilities and constraints of each
listed in Table 2 and to overcome or avoid dis-
times referred to as design for manufacturing manufacturing process must be known and
advantages listed in Table 3.
(DFM), integrates product and process concepts considered by the design team. With this
to ensure ease of manufacture by matching prod- knowledge, the designer should specify tol-
uct and process requirements. The DFM ap- erances easily met by the process and try to
Characteristics of Manufacturing proach examines both product and process de- minimize the need for fixturing and second-
Processes (from Ref 5) sign in an integrated way. While DFM methods ary operations.
have not given much attention to the materials
Additional information on DFM can be found in
Figure 1 (Ref 6) gives a breakdown of manu- selection, the ideas of DFM are important to
Ref 9 to 11.
facturing processes into nine broad classes, and consider when selecting a material in relation-
the goal is to select processes that maximize ship to possible manufacturing processes. The
quality and minimize the cost of the part. product (including material selection) and the Materials Aspects
Selection depends on the basic characteristics of process are designed together to ensure effective
manufacturing processes in terms of: production. Typical objectives of DFM, as out-
The selection of a material must be closely
lined by Bralla (Ref 8) include:
Material factors coupled with the selection of a manufacturing
Size and shape factors Minimize total number of parts. process. This is not an easy task, as there are
Process factors Standardize design of products, subassem- many processes that can produce the same part.
blies, components, modules, and individual In a very general sense, the selection of the ma-
These factors are described briefly in rather ele- parts. terial determines a range of processes that can be
mentary form, although the intent is to provide Use readily processed materials with two used to process parts from the material. Table 4
sufficient detail for identification of likely important caveats: (1) never compromise shows the manufacturing methods used most
processes at the conceptual or embodiment stages the quality of a part by selecting a more frequently with different metals and plastics
Chapter 11: Design for Deformation Processes / 141
Table 4 Compatibility between materials and manufacturing processes

Aluminum Copper Zinc Magnesium Titanium Nickel
Cast Carbon Alloy Stainless and alumi- and copper and zinc and magnes- and titanium and nickel Refractory Thermo- Thermoset
Process iron steel steel steel num alloys alloys alloys ium alloys alloys alloys metals plastics plastics
Casting/molding
Sand casting X X
Investment casting X X
Die casting X X X X X X X X X
Injection molding X X X X X X X X X X X
Structural foam molding X X X X X X X X X X X X
Blow molding (extrusion) X X X X X X X X X X X X
Blow molding (injection) X X X X X X X X X X X X
Rotational molding X X X X X X X X X X X X
Forging/bulk forming
Impact extrusion X X X X X X
Cold heading X X X X X
Closed-die forging X X X X
Pressing and sintering (P/M) X X X X
Hot extrusion X X X X
Rotary swaging X X X X
Machining
Machining from stock
Electrochemical machining X X
Electrical discharge machining (EDM) X X X
Wire EDM X X
Forming
Sheet metal forming X X X X
Thermoforming X X X X X X X X X X X X
Metal spinning X X X
, Normal practice; , less-common practice; X, not applicable. Source: Adapted from Ref 9
(Ref 9). The melting point of a material and level design objectives and required properties is se- shape. The overall guide should be to select a pri-
of deformation resistance (hardness) and ductil- lected. However, as previously noted, the impor- mary process that makes the part as near to final
ity determine these relationships. Low-melting- tant objective is to minimize final part cost. A shape as possible (near-net-shape forming) with-
point metals can be used with any of a large more expensive material may result in a less ex- out requiring additional secondary processes such
number of casting processes, but as the melting pensive product because it can employ a more as machining or grinding. Sometimes the form of
point of the material rises, the number of avail- economical manufacturing process. the starting material is important. For example, a
able processes becomes limited. Similarly, yield The act of selecting a material is a form of en- hollow shaft can be made best by starting with a
strength, or hardness, determines deformation gineering decision making. It is important to con- tube rather than a solid bar.
limits, and forging and rolling loads are related sider a spectrum of materials during conceptual Maximum size often is controlled by equip-
to yield strength. Some materials are too brittle stage of design and then narrow the list during ment considerations. Overall size of the part is
to be plastically deformed; others are too reac- embodiment or configuration design. The materi- expressed by volume, projected area, or weight.
tive to be cast or have poor weldability. als-selection decision is ultimately a trade-off be- For example, Fig. 2 plots the typical regions of
The ultimate criterion for selection of materi- tween performance and cost. In many materials- part size and complexity for deformation pro-
als is the cost to produce a quality part. For se- selection problems, one property stands out as the cessing, casting, and fabrication operations. In
lection of the best material for producing a part most dominant service requirement. In this case,
the following factors must be considered: a reasonable selection criterion may be a method
based on cost per unit property. Cost per unit
Material composition: grade of alloy or
property is a type of performance index for mate-
plastic
rial cost. Processing factors also influence cost,
Form of material: bar, tube, wire, strip,
and trade-offs may be needed in deciding whether
plate, powder, and so forth
to use material A with process B versus material
Size: dimensions and tolerances
C with manufacturing process D. For geometri-
Heat treated condition
cally simple parts, such as straight shafts, the
Directionality of mechanical properties
most economical raw material form and method
(anisotropy)
of manufacture are readily apparent. As the shape
Surface finish
of the part becomes more complex, the applica-
Quality level: control of impurities, inclu-
bility of two or more forms and methods of man-
sions, and microstructure
ufacture add complexity to the selection process.
Quantity: volume of production (batch size)
The trade-offs depend on the materials-selection
Ease of manufacture: workability, weldabil-
factors and processing factors.
ity, castability, and so forth
Ease of recyclability
Cost of material Part Size, Shape, and Tolerances
Cost is the basis to which the other factors are Each process has associated with it a range of
reduced to reach a decision on materials selec- part shapes and sizes that can be produced. Thus,
tion. Cost of material may be a dominant factor, the first decision in selecting a process is its capa- Fig. 2 Ashby process-selection chart for the relation-
ship between part size and complexity of the
as the least expensive material consistent with bility of producing parts of the required size and shape. Source: Ref 6
this case, part complexity is expressed in terms of research in manufacturing, but there is no uni- uses a classification and coding system to iden-
of the information content of the part, that is, the versally accepted system. In general, shape is tify and understand part similarities and to es-
number of independent dimensions that must be often characterized by aspect ratio, the surface- tablish parameters for action. Because similar
specified to describe the shape. Machining adds to-volume ratio, or the web thickness-to-depth shapes tend to be produced by similar process-
precision and allows production of parts with a ratio. Overall complexity of a shape is influ- ing methods, the aim of GT is to classify simi-
greater range of complexity (Fig. 2). Simple enced by a variety of factors such as: larity of parts for more effective design and
shapes require only a few bits of information; for manufacturing (Ref 13). Benefits include:
Requirements to achieve a minimum sec-
example, the casting for an engine block might
tion thickness Ability to quickly generate designs for sim-
have 103 bits of information, but after machin-
Lack of symmetry ilar-shaped parts
ing, the complexity increases both by adding
Surface details (e.g., the smallest radius of Control of part proliferation and redundant
new dimensional features and improving the
curvature at a corner is a surface detail that designs
precision (or dimensional tolerances).
influences complexity) Standardization of design features such as
Part Shape and Complexity. Shape is an
The presence of undercuts and reentrant an- chamfers, corner radii, and tolerances
essential feature of all manufactured parts; com-
gles Grouping of machine tools into more pro-
plexity of this shape often determines what
The presence of internal cavities ductive units
processes can be considered for making it. In the
Development of jigs and fixtures that can
most general sense, increasing complexity nar- Figure 3 (Ref 12) is one example of some basic
accommodate different members of a part
rows the range of processes and increases cost. shapes with increasing spatial complexity.
family
A cardinal rule of design is, therefore, to keep Another scheme for shape classification is shown
Development of computer-based process-
the shape as simple as possible. This rule may, in Fig. 4, where the complexity or difficulty of
selection methods
however, be broken if a more complex shape al- making the shape increases down and to the right.
lows consolidation of several parts and/or elimi- Group technology (GT) is a term that refers to Manufacturing engineers use GT to decide on
nation of one or more manufacturing steps. the classification of shapes suitable for process- more efficient ways to increase system flexibil-
The classification of shapes is an active field ing by the same techniques. Group technology ity by streamlining information flow, reducing
Fig. 3 The choice of possible manufacturing methods is aided by classifying shapes according to their geometric features. Source: Ref 12
In general, configuration design should seek to

reduce the number of geometric and dimen-
sional tolerances and the stringency or tightness
of the tolerance requirements. By way of exam-
ple, the following are some general tolerance
guidelines for aluminum extrusion:
Avoid tight straightness tolerances on long,
thin-wall sections. Add small ribs to help
achieve straightness.
Angles of unsupported wall sections are dif-
ficult to control precisely. Hence, tolerances
on size dimensions involving long legs are
difficult to control. If necessary for toler-
ance control, support long legs with webs
connecting to other parts of the extrusion.
(This may create a more costly hollow
shape, but if tolerances are critical, it may
be necessary.)
Using uniform wall thickness throughout
the cross section makes all dimensions eas-
ier to control.
Rounding inside corners makes angles eas-
ier to control than when corners are sharp.
The dimensions of symmetrical shapes are
generally easier to control than nonsymmet-
rical shapes.
In forgings, shrinkage and warpage contribute
largely to the difficulties of controlling toler-
ances. Forged parts with simple shapes and com-
mon proportions can help minimize the effects
of shrinkage and warpage. It is also important to
provide for well-rounded corners, fillets, and
Fig. 4 Classification of shapes by complexity and process manufacturing difficulty where the complexity or difficulty
edges. Thin sections, such as webs, should be
of making the shape increases down and to the right. 3-D, three-dimensional. Source: Ref 6
kept to a minimum in number and should be
made as thick as possible.
setup time and floor space requirements, and process has the capability of producing a part
standardizing procedures for batch-type produc- with a certain range of tolerance and surface fin-
tion. Design engineers use GT to reduce design ish (Fig. 5). Every manufacturing process has Process Factors
time and effort as well as part and tooling prolif- standard or commercial tolerances that are gener-
Key manufacturing process factors include:
eration. With increasing emphasis on flexible ally well known to manufacturing process engi-
and integrated manufacturing, GT is also an ef- neers, although standard tolerance values are not Cycle time: Once the process has been set
fective first step in structuring and building an always explicitly defined in print for designers. up and is operating properly, cycle time is
integrated database. Standardized process plan- Manufacturing costs increase with tighter di- the time required to produce one unit. The
ning, accurate cost estimation, efficient purchas- mensional tolerances. The manufacturing toler- inverse of cycle time is production rate.
ing, and assessment of the impact of material ances are those that can be produced with the Quality: A widely inclusive term describing
costs are benefits that are often realized. A more normal attention paid to process control and in- fitness for use. It may include dimensional
complete discussion of GT, especially as it ap- spection. Part designs based on standard manu- tolerance, freedom from defects, and per-
plies to manufacturing, is given in Ref 14. facturing tolerances will be the least expensive formance properties.
The method of GT for part coding and classi- to produce. Moreover, the cost of manufacturing Flexibility: The ease of adapting a process
fication began with metal-removal processes, but also depends on the number of tolerances that to produce different products or variations
coding and classification schemes have been de- must be critically controlled. Controlling one or of the same product. Product customization
veloped for metalforming operations (e.g., Ref two critical dimensions, unless they are of an es- is becoming more important, and so this
15). Coding details are based on part geometry, pecially difficult type or extremely tight, is often characteristic has gained importance.
tolerance level, surface finish, material, heat relatively easy to do if the other dimensions of Flexibility is influenced greatly by the time
treatment, and defect level. Hybrid component the part do not need special control. to change and set up tooling.
classification systems have also been developed At the configuration stage of design for parts, Materials utilization: The amount of
where part families are defined by codes of both even before exact dimensions are assigned, de- processed material used in the final product.
geometrical and technological factors of a part signers can and should think ahead about both Machining operations can generate 60 to
(Ref 16). This allows for more effective plan- dimensional tolerances and geometric toler- 80% scrap. Net-shape forging and injection
ning, estimating, and flexible manufacturing. ances (e.g., flatness) that will be needed to molding are at the other extreme. As mate-
Dimensional Tolerances. Part complexity achieve the desired functionality. Geometric tol- rials costs become the greater part of the
also depends on the dimensional tolerances and erances refer to such issues as flatness, straight- product cost, materials utilization becomes
the roughness of the surface finish. Tolerance is ness, parallelism, perpendicularity, and so forth. of greater importance.
the degree of deviation from ideal that is permit- These issues are discussed in other references Operating costs and equipment utilization:
ted in the dimensions of a part, and it is closely re- such as ASME standard Y 14.5M and the Operating costs involve both the capital
lated to the surface finish. Each manufacturing Handbook of Geometric Tolerancing (Ref 17). cost (plant infrastructure, machinery, and
quired to produce one part. The most commonly

used manufacturing processes are evaluated with
respect to these characteristics in Table 7 (Ref
20). These relative values are meant to serve
only as general guidelines.
Categories of Deformation
Processes
Dieter (Ref 21) categorizes deformation
processes into five broad classes:
Direct-compression processes: Force is ap-
plied directly to the surface of the work-
piece and material flow is normal to the ap-
plication of the compressive force;
examples are open-die forging and rolling.
Indirect-compression processes: Deforma-
tion is imposed by compressive loads gener-
ated as the workpiece is pushed or pulled
through a converging die. The direction of
the external load applied to the workpiece is
in the direction of workpiece motion. Ex-
amples include extrusion, wire drawing, and
deep drawing.
Tension-based processes: Tensile loading is
developed in the workpiece to cause thin-
ning, with stretch forming being a primary
example.
Fig. 5 Approximate values of surface roughness and tolerance on dimensions typically obtained with different man-
Bending processes: A bending moment is
ufacturing processes. ECM, electrochemical machining; EDM, electrical discharge machining. Polymers are
different from metals and ceramics in that they can be processed to a very high surface smoothness, but tight tolerances applied to cause a geometry change, the de-
are seldom possible because of internal stresses left by molding and creep at service temperatures. Source: Ref 12 formation being limited to the local region
of the bend. Sheet bending, rod bending and
coiling, and plate bending are example
tooling) and the labor costs of setting up riod). In general, a larger production quantity
processes.
and running the process. Process selection justifies greater investment in dies, equipment,
Shearing processes: Metal deformation is
often is constrained by the available equip- and automation, whereas small quantities are
highly localized in a workpiece as offset
ment, particularly if an alternate process is often made with more labor input.
blades moving in opposite directions gener-
costly to install. In any manufacturing sit- The total number of parts is frequently pro-
ate a plane of intense shear to intentionally
uation, one available decision is to out- duced in lots. Lot size used to be determined to
cause a shear failure. Hole punching, plate
source the production to a qualified sub- a large extent by total quantity, with lot size cho-
shearing, blanking, and slitting are exam-
contractor. sen to provide a supply for several days or
ples of shearing processes.
weeks. Some standard components such as fas-
A rating system for evaluating these five process teners are still produced in large lots by hard au- Other terminology is also recognized in the in-
characteristics is given in Table 5. This is ap- tomation. However, the spread of just-in-time dustry. Bulk forming processes are processes that
plied to rating the most common manufacturing delivery schedules and the increasing use of have large volumes of material participating in
processes in Table 6. quick die-changing techniques and flexible au- the deformation and may be termed three-dimen-
Production Volume. An important practi- tomation have contributed to shrinking lot sizes sional processes. The starting material in bulk de-
cal consideration is the quantity of parts re- even in mass production; the limit is reached formation processes is a slab, ingot, billet, and so
quired. For each process, there is a minimum when a single part constitutes a lot. forth, produced by casting into stationary molds
batch size below which it is not economical to It is not feasible to state hard rules of what the or by continuous-casting techniques. Primary de-
go because of costs of tooling, fixtures, and economical production quantity and lot size is. formation processes such as hot rolling, tube
equipment. Thus, process choice is heavily in- Much depends on the equipment, degree of au- piercing, extrusion, and open-die forging are then
fluenced by the total number of parts to be made tomation, and process control available in a used for converting the cast structure. The product
and by the required rate of production (i.e., the given plant. Also related to part cost is the pro- may be suitable for immediate application, but in
number of parts produced in a given time pe- duction rate or the cycle time, that is, the time re- many cases it serves as the starting material for
Table 5 Scale for rating manufacturing processes

Rating Cycle time Quality Flexibility Materials utilization Operating costs
1 >15 min Poor quality, average reliability Changeover extremely difficult Waste > 100% of finished component Substantial machine and tooling costs
2 5 to 15 min Average quality Slow changeover Waste 50 to 100% Tooling and machines costly
3 1 to 5 min Average to good quality Average changeover and setup time Waste 10 to 50% Tooling and machines relatively inexpensive
4 20 s to 1 min Good to excellent quality Fast changeover Waste < 10% finished part Tooling costs low/little equipment
5 <20 s Excellent quality No setup time No appreciable waste No setup costs
Rating scheme: 1, poorest; 5, best. Source: Ref 18
Table 6 Rating of characteristics for common manufacturing processes another deformation process, the so-called sec-
ondary deformation processes, such as drawing
Cycle Material Equipment
Process Shape time Flexibility utilization Quality tooling costs and hot and cold forging.
Much primary working is conducted at elevated
Casting
temperature, in the hot-working temperature range
Sand casting 3D 2 5 2 2 1 (typically above one-half the melting point on the
Evaporative foam 3D 1 5 2 2 4
Investment casting 3D 2 4 4 4 3
absolute temperature scale) where dynamic recov-
Permanent mold casting 3D 4 2 2 3 2 ery and recrystallization ensure low flow stress
Pressure die casting 3D solid 5 1 4 2 1 and high workability. Cold working produces bet-
Squeeze casting 3D 3 1 5 4 1 ter surface finish, can impart higher strength, and
Centrifugal casting 3D hollow 2 3 5 3 3
Injection molding 3D 4 1 4 3 1
allows thinner products, but does so at the expense
Reaction injection molding (RIM) 3D 3 2 4 2 2 of higher forces. Warm working in an intermediate
Compression molding 3D 3 4 4 2 3 temperature range can combine some of the bene-
Rotational molding 3D hollow 2 4 5 2 4 fits of hot and cold working. Many processes are
Monomer casting, contact molding 3D 1 4 4 2 4
capable of producing net-shape or near-net-shape
Forming products, and Table 8 refers to these. Limits of size
Forging, open die 3D solid 2 4 3 2 2 and web thickness are often set by the pressure in-
Forging, hot closed die 3D solid 4 1 3 3 2 crease due to friction at high width-to-thickness
Sheet metal forming 3D 3 1 3 4 1
Rolling 2D 5 3 4 3 2
ratios (Fig. 6) (Ref 12).
Extrusion 2D 5 3 4 3 2 Rolling is an important primary working
Superplastic forming 3D 1 1 5 4 1 process, practiced in highly specialized, high-pro-
Thermoforming 3D 3 2 3 2 3 ductivity, capital-intensive plants. The products of
Blow molding 3D hollow 4 2 4 4 2
Pressing and sintering 3D solid 2 2 5 2 2
flat hot rolling are relatively thick plate, sheet, and
Isostatic pressing 3D 1 3 5 2 1 strip, typically down to 1.6 mm (0.060 in.). Cold
Slip casting 3D 1 5 5 2 4 rolling is capable of producing the thinnest of all
Machining metal products (strip and foil, to a few microns or
Single-point cutting 3D 2 5 1 5 5
a fraction of a thousandth of an inch) to very tight
Multiple-point cut 3D 3 5 1 5 4 tolerances and controlled surface finish. Most of
Grinding 3D 2 5 1 5 4 these products are further processed in sheet met-
Electrical discharge machining 3D 1 4 1 5 1 alworking operations. Hot-rolled bar and wire
Joining serve as starting materials for other bulk deforma-
Fusion welding All 2 5 5 2 4 tion processes or machining. Shapes are mostly
Brazing/soldering All 2 5 5 3 4 hot rolled or hot extruded for structural applica-
Adhesive bonding All 2 5 5 3 5 tions, although cold-rolled precision shapes fill an
Fasteners 3D 4 5 4 4 5
important niche. All these processes yield prod-
Surface treatment ucts of two-dimensional configuration, often in
Shot peening All 2 5 5 4 5 very long lengths, and are capable of producing
Surface hardening All 2 4 5 4 4 high-quality (close-tolerance) products at very
CVD/PVD All 1 5 5 4 3
high rates and low cost. Closer to final application
Rating scheme: 1, poorest; 5, best. Ratings from Ref. 19 are three-dimensional products of hot and cold
ring rolling (rotating rings of jet engines, ball-
bearing races), forge rolling for complex shapes
Table 7 Manufacturing processes and their attributes

Surface Dimensional Production Production Relative Size
Process roughness accuracy Complexity rate run cost (projected area)
Pressure die casting L H H H/M H H M/L

Centrifugal casting M M M L M/L H/M H/M/L
Compression molding L H M H/M H/M H/M H/M/L
Injection molding L H H H/M H/M H/M/L M/L
Sand casting H M M L H/M/L H/M/L H/M/L
Shell mold casting L H H H/M H/M H/M M/L
Investment casting L H H L H/M/L H/M M/L
Single point cutting L H M H/M/L H/M/L H/M/L H/M/L
Milling L H H M/L H/M/L H/M/L H/M/L
Grinding L H M L M/L H/M M/L
Electrical discharge machining L H H L L H M/L
Blow molding M M M H/M H/M H/M/L M/L
Sheet metal working L H H H/M H/M H/M/L L
Forging M M M H/M H/M H/M H/M/L
Rolling L M H H H H/M H/M
Extrusion L H H H/M H/M H/M M/L
Powder metallurgy L H H H/M H H/M L
Key:
H >250 <0.005 High >100 >5000 High >0.5
M >63 and <250 >0.005 and <0.05 Medium >10 and <100 >100 and <5000 Medium >0.02 and <0.5
L <63 >0.05 Low <10 <100 Low <0.02
Units min. in Parts/h Parts m2
Source: Ref 20
Table 8 General characteristics of bulk deformation processes

Deformation process
Cold
Hot forging Hot forging, Shape Shape Transverse
Characteristic Open die Impression extrusion extrusion drawing rolling rolling
Part
Material (wrought) All All All All All All All
Shape(a) RO3; B; T1, 2; F0; Sp6 R; B; S; T1, 2, 4; (T6, 7); Sp R; B; S ;SS; T1, 4; Sp Same as hot R0; B0; S0; T0 R0; B0; S0 R12; T12; Sp
Size, kg 0.1200,000 0.01100 1500 0.00150 101000 101000 0.00110
Minimum section, mm 5 3 1 (0.005) 1 0.1 0.5 1
Minimum hole diameter, mm (10) 20 10 20 (1) 5 0.1
Surface detail(b) E C BC AB A AB AC
Cost
Equipment(b) AD AB AB AC BD AC AC
Die(b) F BC CD AB CD AC AC
Labor(b) A BD BC CE CE CE CE
Finishing(b) A BC CD DE E E DE
Production
Operator skill(b) A BC CE CE DE B BC
Lead time Hours Weeks Daysweeks Weeks Days Weeks Weeksmonths
Rates, pieces/machine 150 per h 10300 per h 10100 per h 10010,000 per h 102000 m/min 20500 m/min 1001000 per h
Minimum quantity or length (m) 1 1001000 110 1000100,000 1000 m 50,000 m 100010,000
(a) From Fig. 3. (b) Comparative ratings with A indicating the highest value, E the lowest. Values in parenthesis may be obtained by special techniques. Source: Ref 12
the die-to-billet interface. Two-dimensional shapes

of great complexity and thin walls (e.g., architec-
tural extrusions) can be made at a relatively low
die cost. In the absence of a lubricant, divided ma-
terial strands can be reunited and welded in a
bridge-type die, allowing extrusions with one or
more closed cavities, in a very wide size range
(multihole tubes are extruded with walls as thin as
0.25 mm, or 0.010 in.). Hot extrusion of higher-
melting alloys requires a high-temperature lubri-
cant, often a glass; shapes are more limited and
wall thickness is greater. The purpose may be that
of producing a two-dimensional semimanufac-
tured product or, if part of the billet is retained at
the starting diameter, that of making near-net-
shape three-dimensional products such as large
valves and fasteners. The skill of the die designer
is called upon to ensure a sound product. Extrusion
is the preferred way for making bar stock from dif-
ficult-to-work high-temperature alloys.
Cold extrusion is always conducted with an ef-
fective lubricant (often with the lubricant super-
imposed on or reacted with a conversion coating
on the metal) and is normally used to create near-
net-shape products such as fasteners, automotive
Fig. 6 For a given process, the minimum web thickness increases with the distance over which material must move. components, and so forth. Impact extrusion is a
Source: Ref 12 term employed for the cold extrusion of thin-
walled products such as toothpaste tubes.
(cutlery), gear rolling, and cross (transverse) reduction is achieved by drawing over a plug or Forging. The aim in most forging opera-
rolling for shapes of axial symmetry (preforms for a bar (also called mandrel, from which the tube tions is that of producing a part as close to the
forging double-ended wrenches). Seamless tubes is released by cross rolling). Die costs are rela- finished shape as possible. Open-die forging
are made by hot piercing or hot extrusion and are tively low and production rates can be high, uses simple tools in a programmed sequence of
further reduced by special rolling processes for making for low-cost, high-precision products of basic operations (upsetting, drawing out),
immediate use (for example, in oil fields) and for wide size and wall-thickness range. mostly in the hot-working temperature range,
further working, including cold reduction and Extrusion. A billet held in a container is and the products (ranging from the one-off prod-
drawing. pushed against a die (direct or forward extru- ucts of the blacksmith to huge turbine rotors)
Drawing. A rolled preform is drawn sion), or the die penetrates the billet (indirect or usually require finishing by machining. Rotary
through a converging gap, usually cold. Round reverse extrusion), either hot or cold. Friction forging and swaging on special-purpose ma-
bars and wires as well as two-dimensional sec- over the container and die surfaces has a decisive chines produce parts of axial symmetry to much
tions of often complex cross-sectional shape are influence on material flow and die pressures. tighter tolerances (axles, gun barrels).
drawn through stationary dies or, less frequently, Only lower-melting alloys, including aluminum Hot impression-die forging (sometimes termed
rollers. Tubes may be simply drawn through a alloys, can be extruded hot without a lubricant, and closed-die forging) shapes the part between two
die (tube sinking), but more often wall-thickness then deformation occurs with shearing the alloy at die halves; thus productivity is increased, albeit at
the expense of higher die costs. Excess metal is Cold Working with a high K require large machines for defor-
allowed to escape in the flash; thus pressure is mation. Work hardening is a measure of how the
kept within safe limits while die filling is ensured. Cold working results in a deformed, unrecrys- resistance to plastic flow increases as the metal
More complex shapes, thinner walls, and thinner tallized grain structure with the grains being is deformed. Typically, n has values of 0.1 to 0.5
webs may necessitate forging in a sequence of die elongated in the direction of metal flow. for cold working, with 0 being a perfectly plas-
cavities, as for connecting rods and crankshafts. Deformation begins in the grain interiors when tic metal (no work hardening). A metal with a
Die design calls for a thorough knowledge of ma- the critical resolved shear stress of the material high work-hardening exponent but a low
terial flow and is greatly aided by computer mod- is exceeded. At cold-working temperatures, the strength coefficient will achieve a high strength
els and expert systems. At high width-to-thick- grain boundaries are more resistant to deforma- level after a large amount of deformation.
ness ratios, friction sets a limit to minimum web tion, so workpieces with fine grains and a large Copper, brasses, and low-carbon steels are typi-
thickness (Fig. 6) that decreases with effective lu- amount of grain-boundary area are stronger than cal examples of metals that are cold worked to
brication. In true closed-die forging the material coarse-grained material of the same alloy. This produce improved hardness and strength in the
is trapped in the die cavity. dependence of yield strength on grain size and formed part. Table 9 contains some values of K
With dies heated to or close to forging tem- the amount of grain-boundary area in a material and n for these metals (Ref 21). For steels, K in-
perature (isothermal or hot-die forging), cooling is captured by the Hall-Petch equation: creases with carbon content, while n generally
is prevented and thin walls and webs can be pro- decreases. Both copper and brass have a much
duced, provided the die material is stronger than s y = s i + k/ D (Eq 1)
higher work-hardening exponent than steel.
the workpiece material at the temperatures and Both K and n are affected not only by chemistry,
strain rates (e) prevailing in the process. This is but also by prior history and the microstructure.
relatively easy for aluminum alloys (airframe where sy is the yield stress of polycrystalline This is shown in Fig. 7 (Ref 22) for the work-
parts); low press speeds (low e) help to keep metal, si is the stress related to resistance of dis- hardening exponent for a variety of steels and
stresses and forces low. Titanium alloys and su- location motion within a grain, k is the parame- microstructures.
peralloys can be forged in the superplastic state ter relating grain-boundary-hardening effect, Over the range of strain rates at which cold-
(jet-engine fan blades and turbine disks). and D is the average grain diameter. Equation 1 deformation processes are conducted
The sequence of operations can be carried out indicates that the yield strength of an alloy in- (0.1100/s), the sensitivity to strain rate for most
by moving the heated end of a bar through the creases as the grain size becomes finer, where metals is low. Strain level rather than strain rate
die cavities in an upsetter, achieving high pro- yield strength is the initial flow stress. controls the flow stress, in addition to the initial
duction rates. Mechanized transfer between cav- Once deformation is initiated, the moving dis- strength coefficient K. Grain growth is not a fac-
ities in conventional presses is also possible. In locations interact with each other and with the tor in cold working. However, grain flow and a
all impression-die forging, die design calls for grain boundaries to make continued yielding change in grain-aspect ratio is very much a fac-
considerable knowledge and die cost can be high more difficult. This is work hardening, and a fur- tor. As the grains distort, a well-defined grain-
(Table 8), but the product often has superior ther feature of cold forming is that work-harden- flow pattern is developed due to grain-boundary
properties because material flow can be directed ing effects continue to build with continued de- alignment. Nonmetallic inclusions may also par-
to give the best orientation of the structure rela- formation. An empirical relationship for cold ticipate and further define a definite directional-
tive to loading direction in the service of the working between flow stress and plastic strain is: ity in the microstructure and mechanical proper-
part. ties due to this mechanical fibering (see Chapter
Cold forging is related to cold extrusion and, s0 = K en (Eq 2)
2, Bulk Workability of Metals, Fig. 22).
when a complex shape is to be formed in a sin- Extremely deformed microstructures, as are
gle step, requires special lubricants, often with a present in cold-rolled sheet products, may also
conversion coating, as in making spark-plug where s0 is the flow stress, K is the strength co- show alignment of crystallographic planes or
bodies. Alternatively, the shape is developed by efficient (stress when e = 1.0), e is the plastic texture, as well as grain-boundary alignment.
moving the bar or slug through a sequence of strain, and n is the work-hardening exponent. The result is anisotropic behavior of the de-
cavities, using a liquid lubricant. Cold forging is From Eq 2, a high strength coefficient indicates formed material, either in service or in subse-
often combined with cold extrusion. It is the pre- a high initial resistance to plastic flow. Metals quent deformation steps. The designer must be
ferred process for mass producing near-net-
shape parts such as bolts, nuts, rivets, and many
automotive and appliance components.
Table 9 Values for the work-hardening

exponent and strength coefficient for
selected metals
Work- Strength
hardening coefficient (K)
Metal Condition exponent (n) MPa ksi
0.05% C steel Annealed 0.26 531 77

4340 steel Annealed 0.15 641 93
0.6% C steel Quenched and 0.10 1572 228
tempered
540 C
(1000 F)
0.6% C steel Quenched and 0.19 1227 178
tempered
700 C
(1300 F)
Copper Annealed 0.54 317 46
70/30 brass Annealed 0.49 896 130
Source: Ref 21 Fig. 7 Relationship between yield strength and the strain-hardening exponent (n) for a variety of steel microstruc-
tures. Source: Ref 22
aware of the effects of microstructural features

such as fibering or preferred orientation on me-
chanical properties and the relationship of mi-
crostructural alignment and performance
stresses. This is especially critical in cases where
fatigue and fracture toughness are design issues
(for more details, see Chapter 3, Evolution of
Microstructure during Hot Working).
Hot Working
Hot working takes place roughly above a ho-

mologous temperature of 0.5 TM, with typical
hot-working temperatures being 70 to 80% of
the absolute melting temperature. At these tem-
peratures, there is a high amount of internal en-
ergy available, and a number of deformation
mechanisms, in addition to slip, are also avail-
able. These additional mechanisms include
power-law creep mechanisms such as disloca-
tion glide and climb and diffusional flow such as
diffusion of vacancies and boundary motions. Fig. 9 Flow-stress curves representative of (a) dynamic
recovery during hot working and (b) dynamic
Hot working also allows static or dynamic re- recovery and dynamic recrystallization. j, shear-strain
covery and recrystallization of the microstruc- rate; T, temperature; C, constant. Source: Ref 3
ture. Dynamic recovery and dynamic recrystal-
lization occur during deformation, while static crostructures, and in extreme cases, shear crack-
recovery and recrystallization occur after defor- ing. This is mainly true in cases where this phe-
mation while the workpiece is still hot. Dynamic nomenon occurs over a narrow range of temper-
recovery and recrystallization ensure low flow atures or strain rates.
stress and high workability. Recovery and recrystallization are thermally
At high temperature, work hardening is low Fig. 8 Hot-working effects on microstructure. (a)
activated, and therefore these mechanisms are
Rolling with a thickness strain of 50%. (b)
and the flow-stress curve becomes very different Extrusion with a strain of 99%. Source: after Ref 3 very dependent on temperature. They are also
from that of cold deformation. Because the re- very dependent on the level of deformation be-
covery processes take time, flow stress (s0) dur- (Fig. 8b). Dynamic recovery occurs when there cause strain is indicative of a growing volume
ing hot working is a function of strain rate: is sufficient atomic mobility to balance or nearly fraction of defects in the crystal structure and
balance work hardening. That is, dislocations greater internal energy or stress within the
s0 = C e m (Eq 3) are sufficiently active to move in response to grains. The benefit of recovery is that work-
local stresses associated with dislocation tangles hardening effects can be minimized by allowing
where C is the strength coefficient (decreasing and forests, the presence of second phases, and atomic rearrangement to reduce the internal
with increasing temperature) and m is the strain- other local stress concentrations. A metal that is stress within the grain. This is why hot-working
rate-sensitivity exponent. A high m value means undergoing dynamic recovery during hot work- processes can accomplish large deformation lev-
that an incipient neck becomes stronger and ing will exhibit negligible work hardening, with els while maintaining relatively low working
spreads to neighboring material, allowing more most low-carbon and low-alloy steels being pri- loads. For processes such as hot rolling, extru-
deformation in tension. In some very fine-grained mary examples. Figure 9(a) shows a flow stress- sion, and forging, the time in the deformation
metals the value of m may reach 0.4 or 0.5, but strain curve for a metal that dynamically recov- zone is short. Grain refinement is accomplished
only at very low strain rates and in a limited tem- ers during hot working. by static recrystallization after hot working. A
perature range, and then superplastic deformation However, a high deformation level is required high level of hot deformation followed by a hold
is possible to large strains and with low stresses. to produce recrystallization in metals with high time at an elevated temperature causes static re-
The stacking-fault energy relates to the dislo- stacking-fault energy, while metals with low covery and recrystallization to result in a fine
cation structure of the crystal. Low stacking- stacking-fault energy can recrystallize at a lower grain size. This may occur in hot rolling, where
fault energy results in wide stacking faults that level of deformation. Recrystallization occurs if there is time between roll passes, or after hot
have a relatively high resistance to thermally ac- a critical level of strain energy is achieved so forging where the workpiece slowly cools in a
tivated mechanisms, and these metals strain that a new set of grains forms. If recrystalliza- bin. Metallurgical specifications more fre-
harden rapidly. Metals with high stacking-fault tion occurs during hot deformation, the result is quently include grain size limits, and therefore it
energy have narrower stacking faults, the dislo- flow softening as shown in Fig. 9(b) (Ref 3). is becoming more critical to control workpiece
cations are more mobile, and as a result the rate Examples of metals that may flow soften during temperature, deformation rate, the amount of de-
of work hardening is low. Metals with low stack- hot working include nickel-aluminum bronze, formation per working step, and the time be-
ing-fault energies include brass and austenitic commercial-purity titanium, super a2 titanium tween steps in order to control the microstruc-
stainless steels, and metals having high stack- aluminide, Ti-6Al-4V, and 7075 aluminum. ture of the deformed workpiece.
ing-fault energy include aluminum and nickel Whether or not recrystallization and flow soft- The other microstructural phenomenon that
alloys. The effect of hot deformation on changes ening occur is a function of the working temper- can occur during hot-working processes is grain
in grain structure depends on stacking-fault en- ature and strain rate and the local deformation growth. The natural drive for a polycrystalline
ergies and the level of deformation (Fig. 8). level. In practice, most conventional hot-work- material is to minimize internal energy, and be-
Dynamic recovery occurs in all cases with ing processes are too fast for dynamic recrystal- cause grain boundaries are regions of higher in-
high or low stacking-fault energies for both lization to occur. Flow softening can lead to ternal energy, grain growth is a way for nature to
moderate deformation (Fig. 8a) and high strain strain localization and highly nonuniform mi- minimize energy by minimizing the grain-
boundary content. Grain growth is also ther- Grain growth is retarded by the presence of Chapter 12, Workability Theory and
mally driven, and because hot-working stable second-phase particles, as these tend Application in Bulk Forming Processes.
processes may hold a workpiece at a high tem- to hinder grain-boundary movement. Different metals have different workability
perature for a long time, grain growth can occur. Aluminum-killed steels have finer grain size limits, and the limits change with chemistry,
In fact, in an extended hot-working process such than nonkilled steels because of the presence grain size, temperature, second-phase content,
as ingot breakdown rolling, a cyclic history of of AlN precipitates that lock up grain bound- and possibly with strain rate. The remainder of
grain deformation, recrystallization, and growth aries. Other common second phases that can this chapter briefly reviews the forgeability of
is established for each deformation step. The be used to control grain size in iron and other common alloys. These brief summaries of alloy
ability to put work into the grains at a level suf- metals include thermally and chemically sta- forgeability are only meant for the purpose of
ficient to cause recrystallization is the reason ble carbides, nitrides, and oxides. broad comparison. Typical forging temperatures
that fine grains can be developed from a coarse- and lubrication are also discussed. Lubrication
grained structure by hot working. Hot-working More details on the microstructural changes dur- variations cause local velocity gradients and sec-
processes must balance recovery and recrystal- ing hot working are described in Chapter 3, ondary tensile stresses in the plane of the flow-
lization against grain growth in order to be ef- Evolution of Microstructure during Hot ing surface. Even though there is compression
fective in refining large-grained microstructures Working. normal to the workpiece surface, the local in-
or in homogenizing microstructures of mixed plane tension may be high enough to cause
grain sizes. For the designer, this is important cracking. Once a small crack forms, the tension
because grain size has such a pronounced effect Forgeability of Alloys is relieved so the crack does not propagate.
on mechanical performance of the part. These cracks usually do not propagate deeply
At temperatures above the equicohesive tem- Workability is the ability to shape metals by into the workpiece, but can result in unaccept-
perature, the grain interiors are more resistant to bulk deformation, and various techniques are able surface quality. Cracking on surfaces in
deformation than the grain boundaries, and the used to evaluate workability, as described in pre- contact with a die is a common problem, and it
grain boundaries can sustain deformation. If a vious chapters of this book. For example, ductil- can lead to unacceptable parts in net-shape forg-
very fine grain size can be achieved and main- ity as measured by the tensile test is a good way ing or unacceptable machining depths if that sur-
tained during deformation at a low strain rate, to compare the deformation resistance of differ- face is to be finish machined.
grain-boundary sliding can occur. This mecha- ent materials. However, ductility alone is insuf-
nism, in combination with other thermally acti- ficient in the evaluation of workability, because
vated deformation mechanisms, is used success- workability may involve several factors besides Carbon and Alloy Steels
fully to deform very-fine-grained metals to large fracture susceptibility of a material. For exam-
deformation levels, and this is termed superplas- ple, workability may be defined in more general Carbon and alloy steels are by far the most
tic behavior. Superplastic behavior is not a pri- terms by a variety of factors, such as the gener- commonly forged materials and are readily
mary feature in most hot-working processes be- ation of a rough surface finish or the inability to forged into a wide variety of shapes using hot-,
cause of the required low deformation rates and achieve a required tolerance on a critical dimen- warm-, or cold-forging processes and standard
extremely fine grain size. Creep forming, hot- sion. From this practical point of view, a worka- equipment. Despite the large number of avail-
die forging, isothermal forging and sizing, and bility problem occurs whenever the produced able compositions, all of the materials in this
isothermal rolling are processes that rely in part part is unacceptable and must be scrapped or re- category exhibit essentially similar forging char-
on grain-boundary sliding and other thermally worked. acteristics. Exceptions to this are steels contain-
activated deformation mechanisms. Workability is a complex characteristic that ing free-machining additives such as sulfides;
Some general observations concerning recov- depends on the specific stress-state conditions of these materials are more difficult to forge than
ery, recrystallization, and grain growth include: a particular metalworking process. For example, are non-free-machining grades.
the favorable compressive stresses generated One common means of measuring the forge-
If the metal has been previously deformed, during extrusion allow a high deformation ca- ability of steels is the hot-twist test. As the name
grain growth is preceded by recovery and by pacity in comparison to other deformation implies, this test involves twisting of heated bar
recrystallization. processes (see Fig. 2 in Chapter 20, Extru- specimens to fracture at a number of different
The driving force for recrystallization ex- sion). Therefore, it is possible to extrude some temperatures selected to cover the possible hot-
ceeds that of grain growth; thus recrystal- metals that can only be slightly deformed by working temperature range of the test material.
lization can occur at lower temperatures other methods. Another comparison of worka- The number of twists to fracture, as well as the
than grain growth. bility for various processes is shown in Fig. 10 torque required to maintain twisting at a constant
If the workpiece is stress free, grain growth (Ref 23) for free-surface cracking that can occur rate, are reported. The temperature at which the
begins without recovery or recrystallization as a workpiece expands due to a deformation number of twists is the greatest, if such a maxi-
occurring, but the temperature must be rela- process. The strain path that is experienced at a mum exists, is assumed to be the optimal hot-
tively high. free surface is a function of friction and defor- working temperature of the test material.
The rate of grain growth is a function of the mation-zone geometry, where a steeper strain Selection of forging temperatures for carbon
starting grain size, grain shape, and, most path is observed for either higher friction or and alloy steels is based on carbon content, alloy
importantly, temperature. The driving force lower workpiece aspect ratio (which for upset- composition, the temperature range for optimal
for grain growth is the minimization of in- ting is the height/diameter ratio of the work- plasticity, and the amount of reduction required
ternal energy, with the amount of grain- piece). For forging processes, this would include to forge the workpiece. Of these factors, carbon
boundary area representing internal energy. cracking on the exposed surfaces during upset- content has the most influence on upper-limit
Therefore, a fine-grained workpiece will ex- ting and cracking on surfaces at the leading edge forging temperatures. Table 10 lists the typical
perience a higher rate of grain growth than of localized extrusion during forging. For rolling hot-forging temperatures for a variety of carbon
a coarse-grained workpiece will at the same processes, this may be edge cracking of rolled and alloy steels; it can be seen that, in general,
temperature. slabs, plates, or rings. In these processes, the forging temperatures decrease with increasing
As the workpiece is held at temperature, a uniaxial compression test can be used to con- carbon and alloy content.
saturation grain size will be reached, and struct workability limits to evaluate materials for The hot forging of carbon and alloy steels into
holding the workpiece for a longer time will free-surface cracking (e.g., see Chapter 5 Cold intricate shapes is rarely limited by forgeability
not result in further appreciable grain Upset Testing). Localized stress and strain con- aspects with the exception of the free-machining
growth. ditions on workability also are described in grades mentioned previously. Section thickness,
lubricants, including water/graphite mixtures

and water-base synthetic lubricants. Each of the
commonly used lubricants has advantages and
limitations (Table 11) that must be balanced
against process requirements.
Lubricant selection for forging is based on
several factors, including forging temperature,
die temperature, forging equipment, method of
lubricant application, complexity of the part
being forged, and environmental and safety con-
siderations. At normal hot-forging temperatures
for carbon and alloy steels, water-base graphite
lubricants are used almost exclusively, although
some hammer shops may still employ oil-based
graphite.
The most common warm-forming tempera-
ture range for carbon and alloy steels is 540 to
820 C (1000 to 1500 F). Because of the sever-
ity of forging conditions at these temperatures,
billet coatings are often used in conjunction with
die lubricants. The billet coatings used include
graphite in a fluid carrier or water-based coat-
Fig. 10 Workability limit for upsetting, bending, and rolling with varying aspect ratios and friction conditions. ings used in conjunction with phosphate conver-
Source: Ref 23 sion coating of the workpiece.
For still lower forging temperatures (less than
Table 10 Typical forging temperatures for various carbon and alloy steels about 400 C, or 750 F), molybdenum disulfide
Typical forging temperature has a greater load-carrying capacity than does
Steel Major alloying elements C F graphite. Molybdenum disulfide can either be
Carbon steels applied in solid form or dispersed in a fluid car-
1010 ... 1315 2400
rier.
1015 ... 1315 2400 Microalloyed Steel Forgings. Microalloy-
1020 ... 1290 2350 ingthe use of small amounts of elements such
1030 ... 1290 2350 as vanadium and niobium to strengthen steels
1040 ... 1260 2300
1050 ... 1260 2300
has been in practice since the 1960s to control
1060 ... 1180 2160 the microstructure and properties of low-carbon
1070 ... 1150 2100 steels. Most of the early developments were re-
1080 ... 1205 2200 lated to plate and sheet products in which mi-
1095 ... 1175 2150
croalloy precipitation, controlled rolling, and
Alloy steels modern steelmaking technology combined to in-
4130 Chromium, molybdenum 1205 2200 crease strength significantly relative to that of
4140 Chromium, molybdenum 1230 2250 low-carbon steels.
4320 Nickel, chromium, molybdenum 1230 2250
The application of microalloying technology
4615 Nickel, molybdenum 1205 2200 to forging steels has lagged behind that of flat-
5160 Chromium 1205 2200 rolled products because of the different property
6150 Chromium, vanadium 1215 2220 requirements and thermomechanical processing
of forging steels. Forging steels are commonly
used in applications in which high strength, fa-
Source: Ref 24 tigue resistance, and wear resistance are re-
shape complexity, and forging size are limited Table 11 Advantages and limitations of the principal lubricants used in the hot forging of
primarily by the cooling that occurs when the steels
heated workpiece comes into contact with the Type of lubricant Advantages Limitations
cold dies. For this reason, equipment that has
relatively short die contact times, such as ham- Water-base micro-graphite Eliminates smoke and fire; provides Must be applied by spraying for
die cooling; is easily extended best results
mers, is often preferred for forging intricate with water
shapes in steel. Generally, the hot forgeability of Water-base synthetic Eliminates smoke and fire; is Must be sprayed; lacks the lubricity
carbon and alloy steels improves as deformation cleaner than oils or water-base of graphite for severe forging
rate increases. The improvement in workability graphite; aids die cooling; is easily operations
diluted, and needs no agitation after
has been primarily attributed to the increased initial mixing; reduces clogging of
heat of deformation generated at high deforma- spray equipment; does not transfer dark
tion rates. pigment to part
Oil-base graphite Fluid film lends itself to either spray Generates smoke, fire, and noxious
Forging Lubricants (Ref 25). For many or swab application; has good performance odors; explosive nature may shorten
years, oil-graphite mixtures were the most com- over a wide temperature range die life; has potentially serious
monly used lubricants for forging carbon and (up to 540 C, or 1000F) health and safety implications for workers
alloy steels. Recent advances in lubricant tech- Source: Ref 25
nology, however, have resulted in new types of
quired. These requirements are most often filled 0.040S-0.10V-0.02Ti). The carbon content of among the methods used to forge stainless steel.
by medium-carbon steels. Thus, the develop- these steels was reduced to between 0.10 and As in the forging of other metals, two of these
ment of microalloyed forging steels has centered 0.30%. They are produced with either a ferrite- methods are sometimes used in sequence to pro-
around grades containing 0.30 to 0.50% C. pearlite microstructure or an acicular-ferrite duce a desired shape. Roll forging can be used to
The metallurgical fundamentals described structure. The latter results from the suppression forge specific products, such as tapered shafts. It
previously were first applied to forgings in the of pearlite transformation products by an addi- is also used as a stock-gathering operation prior
early 1970s. A West German composition, tion of about 0.10% Mo. to forging in closed dies. Ring rolling is used to
49MnVS3 (nominal composition: 0.47C-0.20Si- Titanium additions have also been made to produce some ringlike parts from stainless steel
0.75Mn-0.060S-0.10V), was used successfully these steels to improve impact toughness even at lower cost than by closed-die forging. The
for automotive connecting rods. The steel was further. Titanium-treated microalloy steels are techniques used are essentially the same as those
typical of the first generation of microalloy currently in production in the United States, for the ring rolling of carbon or alloy steel.
steels, with a medium carbon content Germany, and Japan. The resistance to grain Stainless steels, based on forging pressure and
(0.350.50% C) and additional strengthening coarsening imparted by titanium nitride precipi- load requirements, are considerably more diffi-
through vanadium carbonitride precipitation. The tation increases the toughness of the forgings. cult to forge than carbon or low-alloy steels, pri-
parts were subjected to accelerated air cooling di- One of the primary concerns of any steel user marily because of the greater strength of stain-
rectly from the forging temperature. The AISI is the consistency of finished-part properties. less steels at elevated temperatures and the
grade 1541 microalloy steel with either niobium Heat treatment has successfully addressed this limitations on the maximum temperatures at
or vanadium has been used in the United States concern, and a method must be found to ensure which stainless steels can be forged without in-
for similar automotive parts for many years. the consistency of finished-part properties for mi- curring microstructural damage. Forging-load
The driving force behind the development of croalloy steels. One disadvantage of ferrite- requirements and forgeability vary widely
microalloyed forging steels has been the need to pearlite microalloy steels is that the finished among stainless steels of different types and
reduce manufacturing costs. The objective of mi- strength and hardness are functions of the cooling compositions; the most difficult alloys to forge
croalloy forging steels is to obtain the enhanced rate. Cooling rate can vary because of either are those with the greatest strength at elevated
mechanical properties of hot-formed steel parts process changes or part geometry. The ultimate temperatures.
while simultaneously eliminating the need for strength of first- and second-generation microal- Open-die forging (hand forging) is often used
heat treating of the steel. Elimination of the heat loy steels is adequate for many engineering appli- for smaller quantities for which the cost of
treating operation reduces energy consumption cations, but these steels do not achieve the tough- closed dies cannot be justified and in cases in
and processing time as well as the material in- ness of conventional quenched-and-tempered which delivery requirements dictate shortened
ventories resulting from intermediate processing alloys under normal hot-forging conditions. lead times. Generally, products include round
steps. This is accomplished by means of a sim- Third-generation microalloy forging steels bars, blanks, hubs, disks, thick-wall rings, and
plified thermomechanical treatment (that is, a went into commercial production in the United square or rectangular blocks or slabs in virtually
controlled cooling following hot forging) that States in 1989. This generation of microalloy all stainless grades. Forged stainless steel round
achieves the desired properties without the sepa- forging steels has five to six times the toughness bar can also be produced to close tolerances on
rate quenching and tempering treatments re- at -30 C (-20 F) and twice the yield strength radial forge machines. Although massive forg-
quired by conventional carbon and alloy steels. of second-generation materials. No special forgings are normally associated with open-die forg-
Recent advances in titanium-treated and di- ing practices are required except for the use of a ing, most stainless steel open-die forgings are
rect-quenched microalloy steels provide new op- water-cooling system. produced in the range of 10 to 900 kg (25 to
portunities for the hot forger to produce tough, These steels differ from their predecessors in 2000 lb).
high-strength parts without special forging prac- that they are direct quenched from the forging Closed-die forging is extensively applied to
tices. Product evaluations of these microalloy temperature to produce microstructures of lath stainless steel in order to produce blocker-type,
steels indicate that they are comparable to con- martensite with uniformly distributed temper car- conventional, and close-tolerance forgings. Selec-
ventional quenched-and-tempered steels. Warm bides. Without subsequent heat treatment, these tion from these closed-die types invariably de-
forging continues to make steady progress as a materials achieve properties, including toughness, pends on quantity and the cost of the finished part.
cost-effective, precision-manufacturing tech- similar to those of standard quenched-and-tem- The relative forgeability characteristics of
nique because it significantly reduces machining pered steels. The metallurgical principles behind stainless steels can be most easily depicted
costs. Microalloy steels austenitized at 1040 C this development are based on: through examples of closed-die forgings.
(1900 F), cooled to a warm forging temperature Niobium additions sufficient to exceed the Stainless steels of the 300 and 400 series can be
of 925 C (1700 F), forged, and cooled by air or solubility limit at the forging temperature, forged into any of the hypothetical parts illus-
water (depending on composition), will produce so that undissolved Nb(CN) retards the re- trated in Fig. 11. However, the forging of stain-
a range of physical properties. The resulting cost crystallization and grain growth of austenite less steel into shapes equivalent to part 3 in
savings has the potential to improve the compet- during forging, trimming, and entry into the severity may be prohibited by shortened die life
itive edge that forging has over other manufac- quenchant (20 to 35% of that obtained in forging such a
turing techniques. Composition control to ensure that the shape from carbon or low-alloy steel) and by the
First-generation microalloy forging steels martensite finish temperature is above 200 resulting high cost. For a given shape, die life is
generally have ferrite-pearlite microstructures, C (400 F) shorter in forging stainless steel than in forging
tensile strengths above 760 MPa (110 ksi), and A fast cold-water quench performed on a carbon or low-alloy steel.
yield strengths in excess of 540 MPa (78 ksi). moving conveyor through a spray chamber Forgings of mild severity, such as part 1 in
The room-temperature Charpy V-notch tough- or by other appropriate equipment Fig. 11, can be produced economically from any
ness of first-generation forgings is typically 7 to stainless steel with a single heating and about
14 J (5 to 10 ft lbf), ambient. It became appar- The relatively high martensite finish tempera- five blows. Forgings approximating the severity
ent that toughness would have to be significantly ture, combined with the mass effect of a forging, of part 2 can be produced from any stainless
improved to realize the full potential of microal- results in an autotempered microstructure with steel with a single heating and about ten blows.
loy steel forgings. excellent toughness. For any type of stainless steel, die life in the
Second-generation microalloy forging steels Stainless Steels forging of part 1 will be about twice that in the
were introduced in the mid-1980s. These are forging of part 2.
typified by the West German grade 26MnSiVS7 Forging Methods. Open-die, closed-die, Part 3 represents the maximum severity for
(nominal composition: 0.26C-0.70Si-1.50Mn- upset and roll forging, and ring rolling are forging all stainless steels and especially those
with high strength at elevated temperature, over a wide range of temperatures above 930 C water sprays, such as might be employed to cool
namely, types 309, 310, 314, 316, 317, 321, and (1700 F), and because they do not undergo dies, should be avoided, because they would
347. Straight-chromium types 403, 405, 410, major phase transformation at elevated tempera- cause cracking of the forging. Forgings of the
416, 420, 430, 431, and 440 are the easiest to ture, they can be forged at higher temperatures martensitic steels are often tempered in order to
forge into a severe shape such as part 3 (al- than the martensitic types (Table 12). Exceptions soften them for machining. They are later
though type 440, because of its high carbon con- to the above statements occur when the composi- quench hardened and tempered.
tent, would be the least practical). Types 201, tion of the austenitic stainless steel promotes the Maximum forging temperatures for these
301, 302, 303, and 304 are intermediate between formation of d-ferrite, as in the case of the 309S, steels are low enough to avoid the formation of
the two previous groups. 310S, or 314 grades. At temperatures above 1100 d-ferrite. If d-ferrite stringers are present at forg-
Upset forging is sometimes the only suitable C (2000 F), these steels, depending on their ing temperatures, cracking is likely to occur.
forging process when a large amount of stock is composition, may form appreciable amounts of Delta-ferrite usually forms at temperatures from
needed in a specific location of the workpiece. d-ferrite. Delta-ferrite formation adversely af- 1095 to 1260 C (2000 to 2300 F). Care must
For many applications, hot upset forging is used fects forgeability, and compensation for the be exercised so as not to exceed this temperature
as a preforming operation to reduce the number amount of ferrite present can be accomplished during forging and to avoid rapid metal move-
of operations, to save metal, or both when the with forging-temperature restrictions. ment that might result in local overheating.
forgings are to be completed in closed dies. Equally important restrictions in forging the Surface decarburization, which promotes ferrite
The rules that apply to the hot upset forging of austenitic stainless steels apply to the finishing formation, must be minimized.
carbon and alloy steels are also applicable to temperatures. All but the stabilized types (321, The d-ferrite formation temperature decreases
stainless steel; that is, the unsupported length 347, 348) and the extra-low-carbon types should with increasing chromium content, and small
should never be more than two times the diameter be finished at temperatures above the sensitizing amounts of d-ferrite reduce forgeability signifi-
(or, for a square, the distance across flats) for sin- range (~815 to 480 C, or 1500 to 900 F) and cantly. As the d-ferrite increases above about
gle-blow upsetting. Beyond this length, the un- cooled rapidly from 870 C (1600 F) to a black 15%, forgeability improves gradually until the
supported stock may buckle or bend, forcing heat. The highly alloyed grades, such as 309, structure becomes entirely ferritic. Finishing
metal to one side and preventing the formation of 310, and 314, are also limited with regard to fin- temperatures are limited by the allotropic trans-
a concentric forging. Exceeding this limitation ishing temperature, because of their susceptibil- formation, which begins near 815 C (1500 F).
also causes grain flow to be erratic and nonuni- ity at lower temperatures to hot tearing and s However, forging of these steels is usually
form around the axis of the forging and encour- formation. A final annealing by cooling rapidly stopped at about 925 C (1700 F), because the
ages splitting of the upset on its outside edges. from about 1065 C (1950 F) is generally ad- metal is difficult to deform at lower tempera-
The size of an upset produced in one blow also vised for nonstabilized austenitic stainless steel tures.
should not exceed two diameters (or, for a square, forgings in order to retain the chromium car- Sulfur or selenium can be added to type 410
two times the distance across flats). This varies to bides in solid solution. to improve machinability. These elements can
some extent, depending on the thickness of the Finishing temperatures for austenitic stainless cause forging problems, particularly when they
upset. For extremely thin upsets, the maximum steels become more critical where section sizes form surface stringers that open and form
size may be only two diameters, or even less. increase and ultrasonic testing requirements are cracks. This can sometimes be overcome by ad-
Forgings of the severity represented by hypo- specified. During ultrasonic examination, justing the forging temperature or the procedure.
thetical parts 4, 5, and 6 in Fig. 12 can be hot coarse-grain austenitic stainless steels fre- With sulfur additions, it may be impossible to
upset in one blow from any stainless steel. quently display sweep noise that can be exces- eliminate all cracking of this type. Therefore, se-
However, the conditions are similar to those en- sive due to a coarse-grain microstructure. The lenium additions are preferred.
countered in hot-die forging. First, with a stain- degree of sound attenuation normally increases Ferritic Stainless Steels. The ferritic
less steel, die wear in the upsetting of part 6 will with section size and may become too great to straight-chromium stainless steels exhibit virtu-
be several times as great as in the upsetting of permit detection of discontinuities. Careful con- ally no increase in hardness upon quenching.
part 4. Second, die wear for the forming of any trol of forging conditions, including final forge They will work harden during forging; the de-
shape will increase as the elevated-temperature reductions of at least 5%, can assist in the im- gree of work hardening depends on the temper-
strength of the alloy increases. Therefore, type provement of ultrasonic penetrability. ature and the amount of metal flow. Cooling
410, with about the lowest strength at high tem- Martensitic stainless steels have high hard- from the forging temperature is not critical.
perature, would be the most economical stain- enability to the extent that they are generally air The ferritic stainless steels have a broad range
less steel for forming any of the parts, particu- hardened. Therefore, precautions must be taken of forgeability, which is restricted somewhat at
larly part 6. Conversely, type 310 would be the in cooling forgings of martensitic steels, espe- higher temperature because of grain growth and
least economical. cially those with high carbon content, in order to structural weakness but is closely restricted in
The austenitic stainless steels are more dif- prevent cracking. The martensitic alloys are gen- finishing temperature only for type 405. Type
ficult to forge than the straight-chromium types, erally cooled slowly to about 590 C (1100 F), 405 requires special consideration because of
but are less susceptible to surface defects. Most either by burying in an insulating medium or by the grain-boundary weakness resulting from the
of the austenitic stainless steels can be forged temperature equalizing in a furnace. Direct development of a small amount of austenite. The
other ferritic stainless steels are commonly fin-
ished at any temperature down to 705 C (1300
Fig. 11 Three degrees of forging severity. Dimensions are given in inches. Fig. 12 Three degrees of upsetting severity
Table 12 Typical compositions and forging temperature ranges of high-temperature of blows are required to achieve metal flow equiv-
alloys alent to that of the other types.
Typical composition, % Temperature
During trimming, the forgings must be kept
Alloy C Cr Ni Mo Co Other C F hot enough to prevent the formation of flash-line
cracks. To prevent these cracks, it is often neces-
More difficult to hot work
sary to reheat the forgings slightly between the
Carpenter 41 0.09 19.0 Bal 10.0 11.0 3.1 Ti, 1.5 Al, 0.005 B 10401175 19002145 finish-forging and trimming operations. Cooling,
Pyromet 718 0.10 18.0 55.0 3.0 ... 1.3 Ti, 0.6 Al, 5.0 Nb 9251120 17002050
M252 0.15 18.0 38.0 3.2 20.0 2.8 Ti, 0.2 Al 9801175 18002145
especially the cooling of the martensitic grades,
Waspaloy 0.07 19.8 Bal 4.5 13.5 3.0 T1, 1.4 Al, 0.005 B 10101175 18502145 must be controlled to avoid cracking.
Pyromet 860 0.1 14.0 45.0 6.0 4.0 3.0 Ti, 1.3 Al, 0.01 B 10101120 18502050 Die Lubrication. Dies should be lubricated
Carpenter 901 0.05 12.5 42.5 6.0 ... 2.7 Ti, 0.2 Al, 0.015 B 10101120 18502050 before each blow. For forging in shallow im-
N155 0.12 21.0 20.0 3.0 19.5 2.4 W, 1.2 Nb, 0.13 N 10401150 19002100
V57 0.05 15.0 27.0 1.3 ... 3.0 Ti, 0.2 Al, 0.01 B, 0.3 V 9551095 17502000
pressions, a spray of colloidal graphite in
A-286 0.05 15.0 25.0 1.3 ... 2.1 Ti, 0.2 Al, 0.004 B, 0.3 V 9251120 17002050 kerosene or in low-viscosity mineral oil is usu-
Carpenter 20Cb-3 0.05 20.0 34.0 2.5 ... 3.5 Cu 9801230 18002245 ally adequate. Ordinarily, dies are sprayed man-
Pyromet 355 0.12 15.5 4.5 3.0 ... 0.10 N 9251150 17002100 ually, but in press forging, automatic sprays
Type 440F 1.0 17.0 ... 0.5 ... 0.15 Se 9251150 17002100
Type 440C 1.0 17.0 ... 0.5 ... ... 9251150 17002100
timed with the press stroke are sometimes used.
19-9DL/19DX 0.32 18.5 9.0 1.5 ... 1.4 W plus Nb or Ti 8701150 16002100 For deeper cavities, however, it is often neces-
Types 347 and 348 0.05 18.0 11.0 ... ... 0.07 Nb 9251230 17002245 sary to use a supplemental spray (usually man-
Type 321 0.05 18.0 10.0 ... ... 0.40 Ti 9251260 17002300 ual) to reach the deep areas of the cavity or to
AMS 5700 0.45 14.0 14.0 ... ... 2.5 W 8701120 16002050
Type 440B 0.85 17.0 ... 0.5 ... ... 9251175 17002145
swab the cavity with a conventional forging oil.
Type 440A 0.70 17.0 ... 0.5 ... ... 9251200 17002200 Forging oils are usually mixtures of oil and
Type 310 0.15 25.0 20.0 ... ... ... 9801175 18002145 graphite; the oil should be free of lead and sul-
Type 310S 0.05 25.0 20.0 ... ... ... 9801175 18002145 fur. Forging oils are often purchased as greases
17-4 PH 0.07 17.0 4.0 ... ... 3.03.5 Cu, 0.3 Nb + Ta 10951175 20002145
15-5 PH 0.07 15.0 5.0 ... ... 3.5 Cu, 0.3 Nb + Ta 10951175 20002145
and are then diluted with mineral oil to the de-
13-8 Mo 0.05 13.0 8 2.25 ... 0.901.35 Al 10951175 20002145 sired viscosity. Any volatile lubricant should be
Type 317 0.05 19.0 13.0 3.5 ... ... 9251260 17002300 used sparingly. With even a slight excess, vapor
Type 316L 0.02 17.0 12.0 2.5 ... ... 9251260 17002300 explosions are likely, and greater amounts can
Type 316 0.05 17.0 12.0 2.5 ... ... 9251260 17002300
Type 309S 0.05 23.0 14.0 ... ... ... 9801175 18002145
cause explosions that will eject the workpiece,
Type 309 0.10 23.0 14.0 ... ... ... 9801175 18002145 possibly injuring personnel.
Type 303 0.08 18.0 9.0 ... ... 0.30 S 9251260 17002300 Glass is sometimes used as a lubricant or bil-
Type 303Se 0.08 18.0 9.0 ... ... 0.30 Se 9251260 17002300 let coating in press forging. The glass is applied
Type 305 0.05 18.0 12.0 ... ... ... 9251260 17002300
by dipping the heated forging in molten glass or
Easier to hot work by sprinkling the forging with glass frit. Glass is
Types 302 and 304 0.05 18.0 9.0 ... ... ... 9251260 17002300 an excellent lubricant, but its viscosity must be
UNS S21800 0.06 17 8.5 ... ... 8.0 Mn, 0.12 N 10951175 20002145 compatible with the forging temperature used.
No. 10 0.05 16.0 18.0 ... ... ... 9251230 17002245
Lapelloy 0.30 11.5 0.30 2.8 ... 0.3 V 10401150 19002100
For optimal results, the viscosity of the glass
Lapelloy C 0.20 11.5 0.40 2.8 ... 2.0 Cu, 0.08 N 10401150 19002100 should be maintained at 10 Pa s (100 cP).
636 0.23 12.0 0.8 1.0 ... 0.3 V, 1.0 W 10401175 19002145 Therefore, when different forging temperatures
H46 0.17 12.0 0.5 0.8 ... 0.4 Nb, 0.07 N, 0.3 V 10101175 18502145 are used, a variety of glass compositions must be
AMS 5616 (Greek 0.17 13.0 2.0 0.2 ... 3.0 W 9551175 17502145
Ascoloy)
stocked. Another disadvantage of glass is that it
Type 431 0.16 16.0 2.0 ... ... ... 9001200 16502200 will accumulate in deep cavities, solidify, and
Type 414 0.12 12.5 1.8 ... ... ... 9001200 16502200 impair metal flow. Therefore, the use of glass is
Type 420F 0.35 13.0 ... ... ... 0.2 S 9001200 16502200 generally confined to shallow forgings that re-
Type 420 0.35 13.0 ... ... ... ... 9001200 16502200
Pyromet 600 0.08 16.0 74.0 ... ... 8.0 Fe 8701150 16002100
quire maximum lateral flow.
Type 416 0.1 13.0 ... ... ... 0.3 S 9251230 17002245
Type 410 0.1 12.5 ... ... ... ... 9001200 16502200
Superalloys
Type 404 0.04 11.5 1.8 ... ... ... 9001150 16502100 Because superalloys are designed to resist de-
Type 501 0.2 5.0 ... 0.5 ... ... 9801200 18002200
Type 502 0.05 5.0 ... 0.5 ... ... 9801200 18002200
formation at high temperatures, it is not surpris-
HiMark 300 0.02 ... 18.0 4.8 9.0 0.7 Ti, 0.1 Al 8151260 15002300 ing that they are very difficult to hot work; duc-
HiMark 250 0.02 ... 18.0 4.8 7.5 0.4 Ti, 0.1 Al 8151260 15002300 tility is limited, and the flow stress is high.
Carpenter 7-Mo 0.08 28.0 5.8 1.6 ... ... 9251095 17002000 Further, any alloying addition that improves the
(Type 329)
Type 446 0.1 25.0 ... ... ... ... 9001120 16502050
service qualities usually decreases workability.
Type 443 0.1 21.0 ... ... ... 1.0 Cu 9001120 16502050 Moreover, machining of heat-resistant alloys is
Type 430F 0.08 17.0 ... ... ... 0.3 S 8151150 15002100 difficult and expensive and can sometimes
Type 430 0.06 17.0 ... ... ... ... 8151120 15002050 amount to 40% of the cost of production.
Source: Ref 26 Therefore, the use of computer-aided design and
modeling are particularly significant in the forg-
ing of heat-resistant alloys, because of the pre-
F). For type 446, the final 10% reduction site transformation and precipitation. They are the mium placed on near-net-shape manufacturing.
should be made below 870 C (1600 F) to most difficult to forge and will crack if tempera- In addition, the thermomechanical conditions of
achieve grain refinement and room-temperature ture schedules are not accurately maintained. The forging are often designed to impart desired
ductility. Annealing after forging is recom- forging range is narrow, and the steel must be re- properties and controlled microstructures. The
mended for ferritic steels. heated if the temperature falls below 980 C (1800 complexity of these demands makes computers
Precipitation-Hardening Stainless Steels. F). They have the least plasticity (greatest stiff- very relevant in analyzing and simulating the
The semiaustenitic and martensitic precipitation- ness) at forging temperature of any of the classes forging of heat-resistant alloys.
hardening stainless steels can be heat treated to and are subject to grain growth and d-ferrite for- Methods (adapted from Ref 27). Forged
high hardness through a combination of marten- mation. Heavier equipment and a greater number superalloy components are produced by:
Die forging timize properties (as defined by the component F); the lubricant used is an important limitation
Upsetting specification) in all sections of the component. on maximum die temperature. Die temperature
Extrusion forging This may require generating a grain size that is is controlled by the use of temperature-sensitive
Roll forging within a defined size range. The soundness and crayons or surface pyrometers.
Swaging (or versions using proprietary ro- uniformity of the forging also must be ensured. Lubrication. Dies should be lubricated be-
tary forging machines) Most forgings are inspected by ultrasonic test- fore each forging. For shallow impressions, a
Ring rolling ing, macroetch, and mechanical test of integral spray of colloidal graphite in water or in mineral
Two or more of these methods used in se- coupons. In order to impart optimal work during oil is usually adequate. Dies are usually sprayed
quence each stage, it may even be necessary to include manually, although some installations include
The die-forging categories can be further subdi- redundant work if work penetration in the sub- automatic sprays that are timed with the press
vided as: sequent processing sequence is not likely to be stroke. Deeper cavities, however, may require
uniform. The forging process today does not op- the use of a supplemental spray (usually manu-
Open-die forging erate as a stand-alone function as it did at the ally controlled) to ensure coverage of all sur-
Open-die forgings formed with the aid of start of the superalloy age. faces, or they can be swabbed with a conven-
plugs and rings to impart certain shapes Controlling the deformation process. Re- tional forging oil. These oils are readily
Closed-die blocker-type forgings crystallization must be achieved in each opera- available as proprietary compounds.
Closed-die finish forgings tion to obtain the desired grain size and flow Cooling. Specific cooling procedures are
Which type of method is used depends on com- characteristics in a forged superalloy. Recrys- rarely, if ever, needed after the forging of heat-
plexity of shape and tolerances required. For ex- tallization also helps to eliminate the grain- and resistant alloys. If forging temperatures are cor-
ample, closed-die finish forgings have much twin-boundary carbides that tend to develop rectly maintained, the forgings can be cooled in
thinner ribs and webs, tighter radii, and closer during static heating or cooling. Nonuniform still air, after which they will be in suitable con-
tolerances than blocker forgings (see Table 13 distribution of inhomogeneities will likely lead dition for heat treating.
for some conventional superalloys). Closed-die to problems. Up to 80% of metal reduction is Forgeability Ratings. Table 14 lists the
forging is widely used for forging heat-resistant accompanied by recrystallization, usually com- most commonly forged heat-resistant alloys, and
alloys. The procedures are generally different pleted over falling temperatures; the remaining their forging temperatures and forgeability rat-
from those used for similar shapes from carbon 20% can be as warm work at lower tempera- ings. Generally, these alloys can be grouped into
or low-alloy steels in that preforms (produced by tures for additional strengthening. However, two categories: solid-solution-strengthened al-
open-die forging, upsetting, rolling, or extru- the range of applications for superalloy forg- loys and precipitation strengthened alloys. The
sion) are used to a greater extent for the closed- ings is diverse, and in some circumstances, the latter group is much more difficult to forge than
die forging of heat-resistant alloys than for steel. aim of the forging process may be to produce a the former. These alloys are usually worked with
Regardless of the method used, the forging of duplex, not a single, grain size in the finished the precipitates dissolved, but the higher con-
superalloys generally should be done as part of component. centration of dissolved alloying elements
total thermomechanical processing. In other During the latter quarter of the 20th Century, (4050% total) gives rise to higher flow stress,
words, shaping should not be the only factor in a trend developed to lower the strain rate and to higher recrystallization temperature, and lower
forging. Work energy can be introduced and heat the dies. Faster strain rates lead to frictional solidus temperature, thus narrowing the useful
managed via temperature and deformation con- heat buildup, nonuniform recrystallization, and temperature range for hot forming.
trols to impart the most useful or desired design metallurgical instabilities, and they are also Iron-base superalloys evolved from austenitic
qualities in a component. This discussion as- likely to cause radial-type ruptures, especially in stainless steels and are based on the principle of
sumes that forging intends to create both shape high g alloys. Superalloys can be forged by a combining a closed-packed face-centered cubic
and properties. variety of methods, and two or more of these (fcc) matrix with (in most cases) both solid-solu-
In some cases, forgings are deliberately methods are often used in sequence. A particular tion hardening and precipitate-forming elements.
processed for better tensile properties, stress- outcome of lower strain rate was the introduc- The austenitic matrix is based on nickel and iron,
rupture behavior, creep strength, or low-cycle fa- tion of isothermal superplastic forging/forming with at least 25% Ni needed to stabilize the fcc
tigue life. Therefore, the objectives for the forg- or, at least, isothermal forging (see the discus- phase. Other alloying elements, such as
ing cycle may be: sion of Isothermal Forging in the section on chromium, partition primarily to the austenite for
Nickel-Base Alloys in this article). solid-solution hardening. The strengthening pre-
Uniform grain refinement Die heating. Dies are always heated for the cipitates are primarily ordered intermetallics, such
Control of second-phase morphology forging of heat-resistant alloys. The heating is as Ni3Al (or g ) Ni3Ti (or h), and Ni3Nb (or g ),
Controlled grain flow usually done with various types of burners, al- although carbides and carbonitrides may also be
Structurally sound components though embedded elements are sometimes used. present. Elements that partition to grain bound-
Fine grain size is not necessarily a desired Optimal die temperature for conventional hot aries, such as boron and zirconium, perform a
outcome for all alloys. The objectives are to op- forging varies from 150 to 260 C (300 to 500 function similar to that which occurs in nickel-
Table 13 Design guides for some conventional superalloy forgings

Type of Web thickness, min Rib width, min. Thickness tolerance Corner radii, min. Fillet radii, min.
Alloy forging mm in. mm in. mm in. mm in. mm in.
A-286, Inco 901, Blocker 19.131.8 0.751.25 19.125.4 0.751.00 4.66.4 0.180.25 15.8 0.62 19.131.8 0.751.25
Hastelloy X, Finish 12.725.4 0.501.00 15.819.8 0.620.78 3.04.6 0.120.18 12.7 0.50 15.825.4 0.621.00
Waspaloy, Udimet
630, TD-nickel(a)
Inco 718, Ren 41, Blocker 25.438.1 1.001.50 25.431.8 1.001.25 5.16.4 0.200.25 19.1 0.75 25.450.8 1.002.00
X-1900(a) Finish 19.131.8 0.751.25 19.825.4 0.781.00 3.85.1 0.150.20 15.8 0.62 19.138.1 0.751.50
Astrology, B-1900(a) Blocker 38.163.5 1.502.50 31.838.1 1.251.50 6.47.6 0.250.30 25.4 1.00 31.863.5 1.252.50
Finish 25.438.1 1.001.50 25.431.8 1.001.25 4.66.4 0.180.25 19.1 0.75 25.450.8 1.002.00
Note: For forgings over 258,064 mm2 (400 in.2) in plan area. For forgings of 64,516258,064 mm2 (100400 in.2) plan area, design allowables can be reduced 25%. For forgings under 64,516 mm2 (100 in.2), design allowables
can be reduced 50%. Recommended draft angles are 57. Machining allowance for finish forgings is 3.816.35 mm (0.150.25 in.). Some shapes can require higher minimum allowables than shown above. (a) Based on lim-
ited data. Source: Ref 27
base alloys; that is, grain-boundary fracture is sup- wt% Ni, and 20 wt% Cr along with additions of Nickel-base Superalloys (see also Nickel-
pressed under creep-rupture conditions, resulting molybdenum, tungsten, and niobium for im- Base Alloys). As shown in Table 14, all but one
in significant increases in rupture life. proved solid-solution strengthening. It is single of the nickel-base superalloys are less forgeable
The inclusion content of the alloys has a sig- phase (austenitic) and strengthened primarily by than the iron-base alloys. Astroloy (UNS N13017)
nificant effect on their forgeability. Alloys con- work hardening. Maximum starting temperature and alloy U-700 are the two most difficult to forge
taining titanium and aluminum can develop ni- is 1230 C (2250 F), and minimum finishing nickel-base alloys. For a given percentage of upset
tride and carbonitride segregation, which later temperature is 955 C (1750 F). Hot-cold work- reduction at a forging temperature of 1095 C
appears as stringers in wrought bars and affects ing may be done down to 760 C (1400 F), but (2000 F), these alloys require about twice the spe-
forgeability. This type of segregation has been reduction should exceed 10% below 980 C cific energy needed for the iron-base A-286. In the
almost completely eliminated through the use of (1800 F) to prevent formation of coarse grains forgeability ratings listed in Table 14, Astroloy and
vacuum melting. Precipitation-hardening iron- on solution treating (Ref 30). U-700 alloys have about one-fifth the forgeability
base alloys are electric-furnace or vacuum-in- Alloy 800 (UNS N08800) is workable, both of Alloy 600 (UNS N06600). However, these rat-
duction melted and then vacuum-arc or elec- hot and cold. The major part of the forging ings reflect only a relative ability to withstand de-
troslag remelted. Double-vacuum-induction should be done between 1010 and 1230 C formation without failure; they do not indicate the
melting may be employed when critical applica- (1850 and 2250 F) metal temperature. Light energy or pressure needed for forging, nor can the
tions are involved. working without tensile or bending stresses ratings be related to low-alloy steels and other al-
Alloy A-286 (UNS S66286) is among the most could be continued as low as 870 C (1600 F), loys that are considerably more forgeable.
forgeable of the heat-resistant alloys (Table 14), but no work should be attempted between 650 The forging of nickel-base alloys requires close
and its forgeability approaches that of AISI type and 870 C (1200 and 1600 F) because of the control over metallurgical and operational condi-
304 stainless steel. For example, Fig. 13(a) (Ref susceptibility to cracking within that range. tions. Particular attention must be given to control
28) plots the effect of temperature on the forging Exposure at high temperatures to sulfurous at- of the work metal temperature. Recording usually
pressure required for a 10% upset reduction of mospheres or to other sources of sulfur must be is required for data on transfer time, soaking time,
various steels, where the forging pressures for A- avoided. The furnace atmosphere for heating the finishing temperature, and percentage of reduc-
286 and AISI type 304 converge at 1100 C (2000 material should be slightly reducing with ap- tion. Critical parts are usually numbered, and pre-
F). However, alloy A-286 requires more power proximately 2% CO. The rate of cooling is not cise records are kept. These records are useful in
and more frequent reheating. Forging tempera- critical with respect to thermal cracking, but the determining the cause of defective forgings, and
tures range from 980 to 1175 C (1800 to 2150 alloy is susceptible to carbide precipitation be- they permit metallurgical analysis so that defects
F). Reductions of at least 15% must be used tween 535 and 760 C (1000 and 1400 F). can be avoided in future products.
under 980 C (1800 F) to prevent formation of The material is cold worked by much the Nickel-base alloys are sensitive to minor varia-
coarse grains on solution treating (Ref 29). same practice that is used for stainless steel. tions in composition, which can cause large varia-
Alloy 556 (UNS R30556, or N-155) is an iron- However, it work hardens to a slightly less de- tions in forgeability, grain size, and final proper-
rich superalloy containing about 20 wt% Co, 20 gree than does stainless steel. ties. In one case, wide heat-to-heat variations in
Table 14 Forging temperatures and forgeability ratings for heat-resistant alloys

Forging temperature(a)
Upset and
breakdown Finish forging Forgeability
Alloy UNS designation C F C F rating(b)
Iron-base alloys
A-286 S66286 1095 2000 1035 1900 1
Alloy 556 R30556 1175 2150 1175 2150 3
Alloy 800 N08800 1150 2100 1035 1900 1
U-57 1095 2000 1035 1900 1
16-25-6 1095 2000 1095 2000 1
Nickel-base alloys
R-235 ... 1205 2200 1205 2200 3
Astroloy N13017 1120 2050 1120 2050 5
Hastelloy W W80004 1205 2200 1035 1900 4
Hastelloy X N06002 1175 2150 1175 2150 3
Alloy 214 ... 1160 2125 1040 1900 3
Alloy 230 ... 1205 2200 1205 2200 3
Inconel 600 N06600 1150 2100 1035 1900 1
Inconel 700 ... 1120 2050 1105 2025 4
Inconel 718 N07718 1095 2000 1035 1900 2
Inconel X-750 N07750 1175 2150 1120 2050 2
Inconel 751 N07751 1150 2100 1150 2100 3
Incoloy 901 N09901 1150 2100 1095 2000 2
M-252 N07252 1150 2100 1095 2000 3
Rene 41 N07041 1150 2100 1120 2050 34
U-500 N07500 1175 2150 1175 2150 34
U-700 ... 1120 2050 1120 2050 5
Waspaloy N07001 1160 2125 1040 1900 3
Cobalt-base alloys
J-1570 ... 1175 2150 1175 2150 2
J-1650 ... 1150 2100 1150 2100 2
S-816 R30816 1150 2100 1150 2100 4
HS-25 (L-605) ... 1230 2250 1230 2250 3
Haynes 188 R30188 1205 2200 1175 2150 3
(a) Lower temperatures are often used for specific forgings to conform to appropriate specifications or to achieve structural uniformity. Fig. 13 Forging pressure required for upsetting versus
(b) Based on the considerations stated in text. 1, most forgeable; 5, least forgeable. (a) forging temperature and (b) percentage of
upset reduction. Source: Ref 28
grain size occurred in parts forged from alloy 901 pressures required in forging them are several quired to reach temperature uniformity in re-
(UNS N09901) in the same sets of dies. For some times greater than those for the iron-base alloys. heating. Consequently, because reheating time is
parts, optimal forging temperatures had to be de- Even when forged at its maximum forging shorter, the reheating temperature may some-
termined for each incoming heat of material by temperature, Alloy-25 work hardens; therefore, times be increased 30 to 85 C (50 to 150 F)
making sample forgings and examining them after forging pressure must be increased with greater above the initial forging temperature without
heat treatment for variations in grain size and other reductions. Accordingly, this alloy generally re- harmful effects. However, if the part receives
properties. Improved ingot metallurgy is making quires frequent reheating during forging to pro- only small reductions in subsequent forging
the forging operation more consistent and easier to mote recrystallization and to lower the forging steps, forging should be continued at the lower
monitor, and wide variations in product structure pressure for subsequent steps. temperatures. These small reductions, in turn,
are less frequent than in the past. Forging conditions (temperature and reduc- must be in excess of about 5 to 15% to avoid ab-
In the forging of nickel-base alloys, the forg- tion) have a significant effect on the grain size of normal grain growth during subsequent anneal-
ing techniques developed for one shape usually cobalt-base alloys. Because low ductility, notch ing. The forging temperatures given in Table 14
must be modified when another shape is forged brittleness, and low fatigue strength are associ- are usually satisfactory.
from the same alloy; therefore, development ated with coarse grains, close control of forging
time is often necessary for establishing suitable and of final heat treatment is important. Cobalt- Refractory Metals (Ref 31)
forging and heat treating cycles. This is espe- base alloys are susceptible to grain growth when
cially true for stronger, more advanced alloys heated above about 1175 C (2150 F). They Refractory metals are forged from as-cast ingots
such as Waspaloy and Astroloy. heat slowly and require a long soaking time for or from billets that have been previously broken
Cobalt-Base Alloys. Many of the cobalt- temperature uniformity. Forging temperatures down by forging or extrusion. Forgeability de-
base alloys cannot be successfully forged be- and reductions, therefore, depend on the forging pends to some extent on the method used to work
cause they ordinarily contain more carbon than operation and the part design. the ingot into a billet. The forging characteristics of
the iron-base alloys and therefore greater quan- The alloys are usually forged with small re- refractory metals and alloys are listed in Table 15.
tities of hard carbides, which impair forgeability. ductions in initial breakdown operations. The re- Niobium and Niobium Alloys. Niobium
The cobalt-base alloys listed in Table 14 are ductions are selected to impart sufficient strain and several of its alloys, notably, Nb-1Zr and Nb-
forgeable. The strength of these alloys at ele- to the metal so that recrystallization (and usually 33Ta-1Zr, can be forged directly from as-cast
vated temperatures, including the temperatures grain refinement) will occur during subsequent ingot. Most impression-die forging experience,
at which they are forged, is considerably higher reheating. Because the cross section of a partly however, has been with unalloyed niobium.
than that for iron-base alloys; consequently, the forged section has been reduced, less time is re- Unalloyed niobium, Nb-1Zr, and Nb-33Ta-1Zr
Table 15 Forging characteristics of refractory metals and alloys

Approximate Recrystallization Hot-working
solidus temperature, temperature, Forging
temperature minimum minimum(a) temperature
Metal or alloy C F C F C F C F Forgeability
Niobium and niobium alloys

99.2% Nb 2470 4475 1040 1900 815 1500 201095 702000 Excellent
Nb-1Zr 2400 4350 1040 1900 1150 2100 201260 702300 Excellent
Nb-33Ta-1Zr 2520 4570 1205 2200 1315 2400 10401480 19002700 Good
Nb-28Ta-10W-1Zr 2590 4695 1230 2250 1315 2400 12601370(b) 23002500 Good(b)
Nb-10Ti-10Mo-0.1C 2260 4100 1205 2200 1370 2500 10401480 19002700 Moderate
Nb-10W-1Zr-0.1C 2595 4700 1150 2100 1205 2200 10951205(b) 20002200 Moderate(b)
Nb-10W-2.5Zr ... ... 1150 2100 1260 2300 12051425(b) 22002600 Good(b)
Nb-15W-5Mo-1Zr 2480 4500 1425 2600 1650 3000 13151650 24003000 Fair
Nb-10Ta-10W 2600 4710 1150 2100 1315 2400 9251205 17002200 Good
Nb-5V-5Mo-1Zr 2370 4300 1150 2100 1315 2400 12051650 22003000 Moderate(b)
Nb-10W-10Hf-0.1Y ... ... 1095 2000 1205 2200 10951650(b) 20003000 Good(b)
Nb-30Ti-20W >2760 >5000 1260 2300 1150 2100 11501260 21002300 Good
Tantalum and tantalum alloys
99.8% Ta 2995 5425 1095 2000 1315 2400 201095(b) 702000 Excellent(b)
Ta-10W 3035 5495 1315 2400 1650 3000 9801260(b) 18002300 Good(b)
Ta-12.5W 3050 5520 1510 2750 >1650 >3000 >1095(b) >2000 Good(b)
Ta-30Nb-7.5V 2425 4400 1150 2200 1540 2800 11501315(b) 22002400 Good(b)
Ta-8W-2Hf 2980 5400 1540 2800 >1650 >3000 >1095(b) >2000 Good(b)
Ta-10Hf-5W 2990 5420 1315 2400 1650 3000 >1095(b) >2000 Good(b)
Ta-2.5W >2760 >5000 1260 2300 1150 2100 201150 702100 Excellent
Molybdenum and molybdenum alloys
Unalloyed Mo 2610 4730 1150 2100 1315 2400 10401315 19002400 Good
Mo-0.5Ti 2595 4700 1315 2400 1480 2700 11501425 21002600 Good-fair
Mo-0.5Ti-0.08Zr 2595 4700 1425 2600 1650 3000 12051480 22002700 Good
Mo-25W-0.1Zr 2650 4800 1425 2600 1650 3000 10401315 19002400 Fair
Mo-30W 2650 4800 1260 2300 1370 2500 11501315 21002400 Fair
Tungsten and tungsten alloys
Unalloyed W 3410 6170 13701595 25002900 ... ... 12051650 22003000 ...
W-1ThO2 3410 6170 15951650 29003000 ... ... 13151925 24003500 ...
W-2ThO2 3410 6170 16501760 30003200 ... ... 13151370 24002500 ...
W-2Mo 3385 6125 15401650 28003000 ... ... 11501370 22002500 ...
W-15Mo 3300 5970 14801595 27002900 ... ... 10951370 20002500 ...
W-26Re 3120 5650 >1870 >3400 ... ... >1480 >2700 ...
W-0.5Nb 3405 6160 17051870 31003400 ... ... 12051650 22003000 ...
(a) Minimum hot-working temperature is the lowest forging temperature at which alloys begin to recrystallize during forging. (b) Based on breakdown forging and rolling experience
can be cold worked. Other alloys, such as Nb- cause internal cracking. Forgeability of the tan- Tungsten requires considerably higher forg-
15W-5Mo-1Zr, generally require initial hot work- talum alloys decreases sharply as tungsten con- ing pressures than molybdenum; therefore, in-
ing by extrusion to break down the coarse grain tent exceeds about 12.5%. Interstitial elements process annealing is often necessary to reduce
structure of as-cast ingots before finish forging. such as carbon, oxygen, and nitrogen also have a the load requirements for subsequent forging op-
The billets are usually heated in a gas furnace deleterious effect on forgeability. erations. Because the need for lateral support
using a slightly oxidizing atmosphere. Niobium Two types of coatingsglasses and alu- during forging is greater for tungsten than for
alloys tend to flow laterally during forging. This minideshave been successfully used to protect molybdenum, the design of preliminary forging
results in excessive flash that must be trimmed tantalum from oxidation during forging. A 0.076 tools is more critical. This is especially true for
from forgings. Niobium and its alloys can be pro- mm (3 mil) thick coating of aluminum has pro- pressed-and-sintered billets, which have some
tected from oxidation during hot working by dip- vided protection for the Ta-10W alloy when it porosity and are less than theoretical density.
ping the billets in an Al-10Cr-2Si coating at 815 was heated in air at 1370 C (2500 F) for 30 Tungsten oxide, which becomes molten and
C (1500 F), then diffusing the coating in an inert min. Glass coatings are generally preferred for volatilizes at forging temperatures, serves as an
atmosphere at 1040 C (1900 F). The resulting their lubricating properties. Various borosilicate effective lubricant in the forging of tungsten.
coating is about 0.05 to 0.1 mm (2 to 4 mils) thick glasses are available that can be used for forging Mixtures of graphite and molybdenum disulfide
and provides protection from atmospheric con- operations carried out in the range of 1095 to are also used. Sprayed on the dies, these films
tamination at temperatures to 1425 C (2600 F). 1315 C (2000 to 2400 F). provide lubricity and facilitate removal of the
Glass frit coatings can also be applied to the work- Tungsten and Tungsten Alloys. Tungsten- part from the dies. Glass coatings are also used,
piece before heating in a gas-fired furnace. base materials, like the other refractory alloy sys- but they can accumulate in the dies and interfere
Molybdenum and Molybdenum Alloys. tems, can be classified into two broad groups: un- with complete die filling.
The forging behavior of molybdenum and molyb- alloyed tungsten and solid-solution or
denum alloys depends on the preparation of the dispersion-strengthened alloys. These classifica- Aluminum Alloys (Ref 32)
billet. Billets prepared by pressing and sintering tions are convenient because they group the alloys Compared to the nickel/cobalt-base alloys and
can be forged directly. Large billets are open-die in terms of metallurgical behavior and applicable titanium alloys, aluminum alloys are consider-
forged or extruded before closed-die forging. Arc- consolidation methods. Solid-solution alloys and ably more forgeable, particularly in conven-
cast billets are usually brittle in tension; they can- unalloyed tungsten can be produced by powder- tional forging-process technology, in which dies
not be forged before extruding, except at ex- metallurgy or conventional melting techniques; are heated to 540 C (1000 F) or less. Figure 14
tremely high temperatures. A minimum extrusion dispersion-strengthened alloys can be produced illustrates the relative forgeability of ten alu-
ratio for adequate forgeability is 4 to 1. only by powder-metallurgy methods. minum alloys that constitute the bulk of alu-
Workpieces subjected to large reductions usu- The forgeability of tungsten alloys, like that minum alloy forging production. This forgeabil-
ally exhibit anisotropy and will recrystallize at of molybdenum alloys, is dependent on the con- ity rating is principally based on the deformation
lower temperatures than parts given less reduction. solidation technique used. Billet density, grain per unit of energy absorbed in the range of forg-
Forging temperature and reduction must be care- size, and interstitial content all affect forgeabil- ing temperatures typically employed for the al-
fully controlled to avoid premature recrystalliza- ity. Metallurgical principles in the forging of loys in question. Also considered in this index is
tion in service and the resulting loss in strength. tungsten are much the same as those for molyb- the difficulty of achieving specific degrees of
Gas- or oil-fired furnaces can be used to heat denum. Tungsten is usually forged in the severity in deformation as well as the cracking
molybdenum and its alloys to approximately hot/cold-working temperature range, in which tendency of the alloy under forging-process con-
1370 C (2500 F). Induction heating is required hardness and strength increase with increasing ditions. As shown in Fig. 14, there is consider-
for higher forging temperatures. Above 760 C reductions. Both systems exhibit increasing able variation in the effect of temperature on
(1400 F), molybdenum forms a liquid oxide that forgeability with decreasing grain size. forgeability of aluminum alloys. The high-sili-
volatilizes rapidly enough that surface contami- con alloy 4032 shows the greatest effect, while
nation is rarely a problem. If metal losses are ex- the high-strength Al-Zn-Mg-Cu 7xxx alloys dis-
cessive, protective atmospheres such as argon, play the least effect.
carbon monoxide, or hydrogen can be used dur- The 15 aluminum alloys that are most com-
ing heating. The liquid oxide formed during heat- monly forged, as well as recommended temper-
ing also serves as a lubricant. Glass coatings are ature ranges, are listed in Table 16. All of these
also used; in addition to providing lubrication, alloys are generally forged to the same reduc-
glass coatings reduce heat losses during forging. tion, although some alloys may require more
Molybdenum disulfide and colloidal graphite are forging power and/or more forging operations
suitable lubricants for small forgings. than others. The forging temperature range for
Tantalum and Tantalum Alloys. Unalloyed
tantalum and most of the single-phase alloys listed Table 16 Recommended forging
in Table 15 can be forged directly from cast ingots. temperature ranges for aluminum alloys
However, breakdown operations are usually re- Forging temperature range
quired in order to avoid laps, wrinkles, internal Aluminum alloy C F
cracks, and other forging defects. The breakdown
1100 315405 600760
temperature is 1095 to 1315 C (2000 to 2400 F). 2014 420460 785860
After about 50% reduction, the forging tempera- 2025 420450 785840
ture may be permitted to drop slightly below 1095 2219 425470 800880
C (2000 F). Billets produced by powder-metal- 2618 410455 770850
3003 315405 600760
lurgy techniques do not lend themselves to direct 4032 415460 780860
forging and must be subjected to breakdown. 5083 405460 760860
Most of the forging experience to date has 6061 430480 810900
been with the Ta-10W alloy. Billets are heated to 7010 370440 700820
Fig. 14 Forgeability and forging temperatures of vari-
7039 380440 720820
1150 to 1205 C (2100 to 2200 F) in gas-fired ous aluminum alloys. There are wrought alu-
minum alloys, such as 1100 and 3003, whose forgeability 7049 360440 680820
furnaces using an oxidizing atmosphere. would be rated significantly above those presented; how- 7050 360440 680820
Breakdown forging below 980 C (1800 F) or ever, these alloys have limited application in forging be- 7075 380440 720820
continued working below 815 C (1500 F) can cause they cannot be strengthened by heat treatment. 7079 405455 760850
most alloys is relatively narrow (generally <55 C, Aluminum alloy forgings are produced on a are blocker-type (finish forging only), conven-
or 100 F), and for no alloy does the range extend wide variety of forging equipment. Aluminum tional (block-and-finish forging or finish forging
beyond 85 C (155 F). Obtaining and maintain- alloy forgings are produced on the full spectrum only), high-definition (near net shape), and preci-
ing proper metal temperatures in the forging of of forging equipment, ranging from hammers sion (no draft, net shape). These closed-die forg-
aluminum alloys is critical to the success of the and presses to specialized forging machines. ing types are illustrated in Fig. 15.
forging process. Die temperature and deformation Selection of forging equipment for a given forg- Blocker-type forgings (Fig. 15a) are produced
rates play key roles in the actual forging tempera- ing shape and type is based on the capabilities of in relatively inexpensive, single sets of dies. In
ture achieved. Unlike some forging processes for the equipment, forging design sophistication, dimensions and forged details, they are less re-
carbon and alloy steels, the dies used in virtually desired forging process, and cost. Deformation fined and require more machining than conven-
all hot-forging processes for aluminum alloys are or strain rate is also a critical element in the suc- tional or high-definition closed-die forgings. A
heated in order to facilitate the forging process. cessful forging of given alloy. The deformation blocker-type forging costs less than a compara-
Therefore, die temperature is another critical ele- or strain rate imparted by equipment varies con- ble conventional or high-definition forging, but
ment in the forgeability and forging-process opti- siderably, ranging from very fast (for example, it requires more machining.
mization of this alloy class. >10 s-1 on equipment such as hammers, me- Conventional closed-die forgings (Fig. 15b)
Table 17 summarizes the die-temperature chanical presses, and high-energy-rate ma- are the most common type of aluminum forging.
ranges typically used for several aluminum forg- chines) to relatively slow (for example, <0.1 s-1 They are produced with either a single set of fin-
ing processes. The criticality of die temperature in on equipment such as hydraulic presses). ish dies or with block-and-finish dies, depending
the optimization of the process depends on the Although aluminum alloys are generally not on the design criteria. These forgings have less
forging equipment being employed, the alloy considered to be as sensitive to strain rate as machine stock and tighter tolerances than
being forged, and the severity of the deformation other materials (such as titanium and blocker-type forgings, but require additional
or the sophistication of the forging design. For nickel/cobalt-base superalloys), selection of the cost (both for the dies and for fabrication) to
slower deformation processes, such as hydraulic strain rate in a given forging process or differ- produce.
press forging, die temperature frequently controls ences in deformation rates inherent in various High-definition, near-net-shape closed-die
the actual metal temperature during deformation, types of equipment affect the forging-pressure forgings (Fig. 15c) are a result of improved forg-
and, in fact, aluminum alloys forged in hydraulic requirements, the severity of deformation possi- ing equipment and process control. They offer
presses are isothermally forged; that is, metal and ble, and therefore the sophistication of the forg- forging design and tolerance enhancement over
dies are at the same temperature during deforma- ing part that can be produced. conventional or blocker-type forgings to affect
tion. Therefore, the die temperatures employed Open-die forging is used frequently to produce further reduction in machining costs. High-defi-
for hydraulic press forging exceed those typical of small quantities of aluminum alloy forgings nition forgings are produced with multiple die
more rapid deformation processes, such as ham- when the construction of expensive closed dies is sets, consisting of one or more blocker dies and
mers and mechanical presses. Both remote and not justified or when such quantities are needed finish dies, and are frequently used with some
on-press die-heating systems are used in the forg- during the prototype fabrication stages of a forg- as-forged surfaces remaining unmachined by the
ing of aluminum alloys. ing application. The quantity that warrants the purchaser.
Forging Methods. Aluminum alloys are use of closed dies varies considerably, depending Precision forgings (Fig. 15d) represent net-
produced by all of the current forging methods on the size and shape of the forging and on the shape products that require no subsequent ma-
available, including open-die (or hand) forging, application for the part. However, open-die forg- chining. Net-shape aluminum forgings are pro-
closed-die forging, upsetting, roll forging, or- ing is by no means confined to small or prototype duced in two-piece, three-piece through-die,
bital (rotary) forging, spin forging, mandrel quantities, and in some cases it may be the most and/or multiple-segment wrap-die systems to
forging, ring rolling, and extrusion. Selection of cost-effective method of aluminum forging man- very restricted design and tolerances necessary
the optimal forging method for a given forging ufacture. For example, as many as 2000 pieces of for assembly. Precision aluminum forgings are
shape is based on the desired forged shape, the biscuit forgings have been produced in open dies produced with very thin ribs and webs; sharp
sophistication of the forged-shape design, and when it was desired to obtain the properties of a corner and fillet radii; undercuts, backdraft,
cost. In many cases, two or more forging meth- forging, but closed dies did not provide sufficient and/or contours; and, frequently, multiple part-
ods are combined in order to achieve the desired economic benefits. ing planes that may optimize grain-flow charac-
forging shape and microstructure. For example, Open-die forgings in aluminum alloys can be teristics. Design and tolerance criteria for preci-
open-die forging frequently precedes closed-die produced to a wide variety of shapes, ranging sion aluminum forgings have been established to
forging to prework the alloy (especially when from simple rounds, squares, or rectangles to provide a finished product suitable for assembly
cast ingot forging stock is used) and to preshape very complex contoured forgings. In the past, or further fabrication. Precision aluminum forg-
(or preform) the metal to conform to the subse- the complexity and tolerances of the open-die ings do not necessarily conform to the tolerances
quent closed dies and to conserve input metal. forging of aluminum and other materials de- provided by machining of other product forms;
pended on the skill of the press operator; how- however, as outlined in Table 18, design and tol-
ever, with the advent of programmable com- erance criteria are highly refined in comparison
Table 17 Die temperature ranges for the puter-controlled open-die forging presses, it is to other aluminum alloy forging types and are
forging of aluminum alloys possible to produce such shapes to overall thick- suitable for the intended application of the prod-
Die temperature ness/width tolerance bands of 1.27 mm (0.050 uct without subsequent machining by the pur-
Forging process/equipment C F in.). Because the open-die forging of aluminum chaser.
Open-die forging alloys is also frequently implemented to pro- Precision aluminum forgings are typically
Ring rolling 95205 200400 duce preforms for closed-die forgings, these produced on hydraulic presses, although in some
Mandrel forging 95205 200400 state-of-the-art forging machines also provide cases mechanical and/or screw presses have
Closed-die forging very precise preform shapes, improving the di- been effectively employed. Although many pre-
Hammers 95150 200300 mensional consistency and tolerances of the re- cision aluminum forgings have been produced
Upsetters 150260 300500 sulting closed-die forging and reducing closed- on small-to-intermediate hydraulic presses with
Mechanical presses 150260 300500 die forging cost through further input material capacities in the range of 9 to 900 kN (1 to 100
Screw presses 150260 300500
Orbital (rotary) forging 150260 300500 conservation. tonf), the size of precision parts demanded by
Spin forging 150315 200600 Closed-Die Forging. Most aluminum alloy users has increased, and so heavy hydraulic
Roll forging 95205 200400 forgings are produced in closed dies. The four presses in the range of 135 to 310,000 kN (15 to
Hydraulic presses 315430 600800 types of aluminum forgings shaped in closed dies 35,000 tonf) have been added or upgraded to
Table 18 Design and tolerance criteria for

aluminum precision forgings
Characteristics Tolerance
Draft outside 0 +30, -0

Draft inside 1 +30, -0
Corner radii 1.5 0.75 mm (0.060 0.030 in.)
Fillet radii 3.3 0.75 mm (0.130 0.030 in.)
Contour 0.38 mm (0.015 in.)
Straightness 0.4 mm in 254 mm (0.016 in. in
10 in.)
Minimum web 2.3 mm (0.090 in.)
thickness(a)
Minimum rib 2.3 mm (0.090 in.)
thickness
Length/width +0.5 mm, -0.25 mm (+0.020 in.,
tolerance -0.010 in.)
Die closure +0.75, -0.25 mm (+0.030,
tolerance -0.010 in.)
Mismatch 0.38 mm (0.015 in.)
tolerance
Flash extension 0.75 mm (0.030 in.)
(a) Web thicknesses as small as 1.5 mm (0.060 in.) have been produced
in certain forging designs
tion formers and is frequently used to produce

forging shapes that are characterized by surfaces
of revolution, such as bolts, valves, gears, bear-
ings, and pistons. Upset forging may be the sole
process used for the shape, such as pistons, or it
can be used as a preliminary operation to reduce
the number of impressions, to reduce die wear,
or to save metal when the products are finished
in closed dies. Wheel and gear forgings are typ-
ical products for which upsetting is advanta-
geously used in conjunction with closed-die
forging. As a rule, in the upset forging of alu-
minum alloys, the unsupported length of forg-
ings must not exceed three diameters for a round
shape or three times the diagonal of the cross
section for a rectangular shape.
Roll forging can be used as a preliminary pre-
form operation to reduce metal input or to re-
duce the number of closed-die operations. In roll
forging, the metal is formed between moving
rolls, either or both containing a die cavity, and
is most often used for parts, such as connecting
rods, where volume is high and relatively re-
stricted cross-sectional variations typify the part.
Orbital (rotary) forging is a variant of closed-
die mechanical or hydraulic press forging in
which one or both of the dies is caused to rotate,
usually at an angle, leading to the incremental
deformation of the workpiece. Orbital forging is
used to produce parts with surfaces of revolution
with both hot and cold aluminum alloy forging
processes, and it provides highly refined close-
tolerance final shapes.
Fig. 15 Types of aluminum closed-die forgings and tolerances for each. (a) Blocker-type. (b) Conventional. (c) High- Spin forging, a relatively new aluminum alloy
definition. (d) Precision forging technique, combines closed-die forging
and computer numerically controlled (CNC)
produce this product. Forging-process criteria ment in precision aluminum forging. The die lu- spin forgers to achieve close-tolerance axisym-
for precision aluminum forgings are similar to bricants employed, although of the same generic metric hollow shapes. Because spin forging is
those for other aluminum alloy forging types, al- graphite/mineral-oil formulations used for other accomplished over a mandrel, inside-diameter
though the metal and die temperatures used are aluminum forging processes, frequently use contours are typically produced to net shape, re-
usually controlled to near the upper limits of the other organic and inorganic compounds tailored quiring no subsequent machining. Outside-di-
temperature ranges outlined in Tables 16 and 17 to the process demands. ameter contours can be produced net or with
to enhance producibility and to minimize forg- Upset forging can be accomplished in special- very little subsequent machining and to much
ing pressures. As with other aluminum forging ized forging equipment called upsetters (a form tighter out-of-round and concentricity tolerances
processes, die lubrication also is a critical ele- of mechanical press) or high-speed multiple-sta- than competing forging techniques, such as for-
ward or reverse extrusion (see below), resulting of aluminum alloys, it is generally desirable to ther by the lubricant manufacturers or by the forg-
in material savings. Parts with both ends open, have billets near the high side of the forging ers themselves. Lubricant composition varies
one end closed, or both ends closed can also be temperature range when forging begins and to with the demands of the forging process used and
produced. finish the forging as quickly as possible before the forging type. The major active element in alu-
Ring rolling is also used for aluminum alloys the temperature drops excessively. Open-die minum alloy forging lubricants is graphite; how-
to produce annular shapes. The procedure used forging and multiple-die closed-die forging of ever, other organic and inorganic compounds are
to ring roll aluminum alloys is essentially the aluminum alloys are frequently conducted with- added to colloidal suspensions in order to achieve
same as that used for steel. Both rectangular and out reheating as long as critical metal tempera- the desired results. Carriers for aluminum alloy
contoured cross section rolled rings, with or tures can be maintained. forging lubricants vary from mineral spirits to
without subsequent machining by the forger, are Aluminum alloys have a relatively narrow mineral oils to water.
produced in many aluminum alloys. The tem- temperature range for forging. Therefore, care- Lubricant application is typically achieved by
peratures employed for the ring rolling of alu- ful control of the temperature in preheating is spraying the lubricant onto the dies while the lat-
minum alloys are quite similar to those for other important. The heating equipment should have ter are assembled in the press; however, in some
forging processes, although care must be taken pyrometric controls that can maintain 5 C cases, lubricants are applied to forging stock
to maintain metal temperature. The deformation (10 F). Continuous furnaces used to preheat prior to reheating or just prior to forging. Several
achieved in the ring rolling of aluminum typi- aluminum typically have three zones: preheat, pressurized-air or airless spraying systems are
cally results in the predominant grain flow in the high heat, and discharge. Most furnaces are employed, and with high-volume highly auto-
tangential or circumferential orientation. If pre- equipped with recording/controlling instruments mated aluminum forging processes, lubricant
dominant grain flow is desired in other direc- and are frequently surveyed for temperature uni- application is also automated by single- or mul-
tions, such as axial or radial, other ringmaking formity in a manner similar to that used for so- tiple-axis robots. Lubricant can be applied with
processes, such as hollow-biscuit open-die forg- lution treatment and aging furnaces. Heated alu- or without heating. State-of-the-art lubricant-ap-
ings, mandrel forging, or reverse/forward extru- minum alloy billets are usually temperature plication systems have the capability of applying
sion, can be employed. The economy of ring checked by using either contact or noncontact very precise patterns or amounts of lubricant
rolling in aluminum alloys depends on the vol- pyrometry based on dual-wavelength infrared under fully automated conditions such that the
ume, size, and contour of the forging. For some systems. This latter technology, although sensi- forging processes are optimized and repeatable.
ring parts, it may be more economical to pro- tive to emissivity, has been successfully incorpo-
duce the shape by mandrel forging or to cut rings rated into the fully automated temperature-veri-
from hollow extruded cylinders. Both tech- fication systems used in automated high-volume Copper and Copper Alloys
niques are discussed below. aluminum forging processes to provide signifi-
Mandrel forging is used in aluminum alloys to cantly enhanced temperature control and process Copper and copper alloy forgings offer a
produce axisymmetric, relatively simple, hollow repeatability. number of advantages over parts produced by
ring or cylindrical shapes, in which the metal is As previously noted, the methods of die heat- other processes, including high strength as a re-
incrementally forged, usually on a hammer or ing include remote die-heating systems and on- sult of working, closer tolerances than compet-
hydraulic press, over a mandrel. In the incre- press die-heating systems. Remote die-heating ing processes such as sand casting, and modest
mental forging process, the wall thickness of the systems are usually gas-fired die heaters capable overall cost. The most forgeable copper alloy,
preform is reduced, and this deformation en- of slowly heating the die blocks. These systems forging brass (alloy C37700), can be forged into
larges the diameter of the piece. The mandrel are used to preheat dies to the desired tempera- a given shape with substantially less force than
forging of aluminum has been found to be eco- ture prior to assembly into the forging equip- that required to forge the same shape from low-
nomical for relatively low-volume part fabrica- ment. On-press die-heating systems range from carbon steel. A less forgeable copper alloy, such
tion and/or in the fabrication of very large ring relatively rudimentary systems to highly engi- as an aluminum bronze, can be forged with ap-
shapes (up to 3.3 m, or 130 in., in diameter). neered systems designed to maintain very tight proximately the same force as that required for
With control of the working history of the input die-temperature tolerances. On-press die-heat- low-carbon steel. Copper and copper alloy forg-
material and the mandrel-forging process, man- ing systems include gas-fired equipment, induc- ings, particularly brass forgings, are used in
drel-forged rings can be produced with either tion-heating equipment, and/or resistance-heat- valves, fittings, refrigeration components, and
circumferential or axial-predominant grain ori- ing equipment. In addition, presses used for the other high-pressure liquid and gas-handling ap-
entations. precision forging of aluminum alloys frequently plications. High-strength bronze forgings find
Reverse or forward extrusion, a variant of have bolsters that can be heated or cooled as application as mechanical parts such as gears,
closed-die forging for aluminum, can be used to necessary. bearings, and hydraulic pumps.
produce hollow, axisymmetric shapes in alu- State-of-the-art on-press aluminum die-heating Copper C10200 and the copper alloys most
minum alloys with both ends open or with one equipment can hold die-temperature tolerances commonly forged are listed in Table 19. They
end closed. Extrusion also frequently plays an within 15 C (25 F) or better. Specific on-press comprise at least 90% of all commercially pro-
important role in the closed-die forging of alu- die-heating systems vary with the forging equip- duced copper alloy forgings. Forging brass, the
minum alloy parts other than hollow shapes ment used, the size of the dies, the forging process, least difficult alloy to forge, has been assigned
(such as wheels). More information on extrusion and the type of forging produced. an arbitrary forgeability rating of 100. Some
of aluminum is available in Chapter 20, Die lubrication is another critical element in copper alloys cannot be forged to any signifi-
Extrusion. the aluminum forging process and is the subject cant degree because they will crack. Leaded
Heating and Temperature Control. Metal of major engineering and developmental empha- copper-zinc alloys, such as architectural
temperature is a critical element in the alu- sis, both in terms of the lubricants themselves and bronze, which may contain more than 2.5% Pb,
minum forging process. Aluminum alloys form a the lubricant-application systems. The lubricants are seldom recommended for hot forging.
very tenacious oxide coating upon heating. The used in aluminum alloy forging are subject to se- Although lead content improves metal flow, it
formation of this coating is self-limiting; there- vere service demands. They must be capable of promotes cracking in those areas of a forging,
fore, aluminum alloys do not scale to the same modifying the surface of the die to achieve the de- particularly deep-extruded areas, that are not
extent that steel does. However, most aluminum sired reduction in friction, withstand the high die completely supported by, or enclosed in, the
alloys are susceptible to hydrogen pickup during and metal temperatures and pressures employed, dies. This does not mean that the lead-contain-
reheating operations such that reheating equip- and yet leave the forging surfaces and forging ing alloys cannot be forged, but rather that the
ment and practices are also critical elements of geometry unaffected. Lubricant formulations are design of the forging may have to be modified
forging-process control. In the open-die forging typically highly proprietary and are developed ei- to avoid cracking.
Table 19 Relative forgeability ratings of Table 20 Recommended forging The die cavity is usually polished to a better
commonly forged copper alloys temperature ranges for copper alloys surface finish for forging copper and copper
Ratings are in terms of the most forgeable alloy, Temperature range
alloys.
forging brass (C37700). Alloy C F
Die materials and hardnesses selected for forg-
Relative C12200 730845 13501550 ing copper alloys depend on part configuration
forgeability(a),
Alloy Nominal composition %
C18200 650760 12001400 (forging severity) and number of parts to be pro-
C37700 650760 12001400 duced. Figure 16 illustrates the forging severities
C10200 99.95 min Cu 65 C46400 595705 11001300
C62400 705815 13001500 of parts listed in Table 21. Whether the dies are
C10400 Cu-0.027 Ag 65
C11000 99.9 min Cu 65 C64200 730900 13501650 made entirely from a hot-work steel such as H11
C11300 Cu-0.027 Ag + O 65 C67000 595705 11001300 or H12 or whether or not inserts are used depends
C14500 Cu-0.65Te-0.008P 65 C67300 595730 11001350 largely on the size of the die. Common practice
C18200 Cu-0.10 Fe-0.90Cr-0.10 Si-0.05Pb 80 C67400 595730 11001350
C67500 595705 11001300 is to make the inserts from a hot-work steel and
C37700 Cu-38Zn-2 Pb 100
C46400 Cu-39.2Zn-0.8Sn 90 to press them into rings or holders made from a
C48200 Cu-38Zn-0.8Sn-0.7Pb 90
temperature is detrimental, because it causes low-alloy die block steel (Table 21) or L6 tool
C48500 Cu-37.5Zn-1.8 Pb-0.7Sn 90
grain growth and increases the amount of scale. steel. Hardness of the ring or holder is seldom
C62300 Cu-10Al-3Fe 75 critical; a range of 341 to 375 HB is typical.
C63000 Cu-10Al-5Ni-3Fe 75 Reheating Practice. When forging in ham-
C63200 Cu-9Al-5Ni-4Fe 70 mers, all of the impressions are usually made in Die Lubrication. Dies should be lubricated
C64200 Cu-7Al-1.8Si 80
one pair of dies, and reheating is rarely required. before each forging operation. A spray of col-
C65500 Cu-3Si 40
In press forging, particularly in high-production loidal graphite and water is usually adequate.
C67500 Cu-39Zn-1.4Fe-1Si-0Mn 80 Many installations include a spray that operates
applications, blocking is often done separately,
(a) Takes into consideration such factors as pressure, die wear, and hot
followed by trimming before the forging is com- automatically, timed with the press stroke.
plasticity
pleted. The operations are likely to be performed However, the spray is often inadequate for deep
in different presses; therefore, the partially com- cavities and is supplemented by swabbing with a
The solubility of lead in b-brass at forging pleted forging is reheated to the temperature conventional forging oil.
temperatures is about 2% maximum, but lead is originally used. Forging Methods. Most copper alloy forg-
insoluble in b-brass at all temperatures. Heating of Dies. Dies are always heated for ings are produced in closed dies. The sequence of
Consequently, although a lead content of up to forging copper and copper alloys, although be- operations is the same as that used for forging a
2.5% is permissible in Cu-40Zn a-b brasses, cause of the good forgeability of copper alloys, similar shape from steel, that is, fullering, block-
lead in excess of 0.10% in a Cu-30Zn a-brass die temperature is generally less critical than for ing, and finishing, as required. However, it is es-
will contribute to catastrophic cracking. Other forging aluminum. Dies are seldom preheated in timated that 90% of the forgings produced from
copper alloys, such as the copper-nickels, can be ovens. Heating is usually accomplished by ring forging brass are forged completely in one or two
forged only with greater difficulty and at higher burners. Optimal die temperatures vary from blows in a finishing die. The starting slugs or
cost. The copper-nickels, primarily because of 150 to 315 C (300 to 600 F), depending on the blanks are usually cut from extruded bars or
their higher forging temperatures, are sometimes forging temperature of the specific alloy. For al- tubes to eliminate the blocking operation.
heated in a controlled atmosphere, thus compli- loys having low forging temperatures, a die tem- Excessive flash is produced, but it is easily
cating the process. The silicon bronzes, because perature of 150 C (300 F) is sufficient. Die trimmed and remelted. In the forging of parts of
of their high forging temperatures and their temperature is increased to as much as 315 C mild to medium severity, in plants where remelt-
compositions, cause more rapid die deterioration (600 F) for the alloys having the highest forging facilities are available, cutting slugs from
than the common forging alloys. ing temperatures shown in Table 20. bars or tubes is usually the least expensive ap-
Heating Practices. Optimal forging tem- Dies designed for forging copper or copper proach. However, in plants that do not remelt
perature ranges for ten alloys are given in Table alloys usually differ from those designed for their scrap, the flash must be sold as scrap, and it
20. Atmosphere protection during billet heating forging the same shapes from steel, as follows: is sometimes more economical to use blocking.
is not required for most alloys, especially when Cylindrical slugs are sometimes partially flat-
forging temperatures are below 705 C (1300 The draft angle can be decreased for forging tened before forging to promote better flow and
F). For temperatures toward the top of the range copper (3 max and often less than 3). consequently better filling of an impression.
in Table 20, a protective atmosphere is desirable The die cavity is usually machined to di- This can usually be done at room temperature
and is sometimes required. An exothermic at- mensions that are 0.005 in./in. less than between flat dies in a hammer or a press. A rec-
mosphere is usually the least costly, and it is sat- those for forging steels. tangular slug is occasionally obtained by extrud-
isfactory for heating copper alloys at tempera-
tures above 705 C (1300 F).
Gas-fired furnaces are almost always used,
and furnace design is seldom critical. Open-fired
conveyor chain or belt types are those most com-
monly used. Any type of temperature control
that can maintain temperature within 5 C (10
F) is suitable. As billets are discharged, a peri-
odic check with a prod-type pyrometer should
be made. This permits a quick comparison of
billet temperature with furnace temperature.
Heating Time. The time at temperature is
critical for all copper alloys, although to varying
degrees among the different alloys. For forging
brass (alloy C37700), the time is least critical,
but for aluminum bronze, naval brass, and cop-
per, it is most critical. Time in excess of that re-
quired to bring the billet uniformly to forging Fig. 16 Forged copper alloy parts of varying severity. See Table 21 for recommended die materials.
Table 21 Recommended die materials for the forging of copper alloys ception is the high-zinc alloy ZK-60, which
Part configurations of varying severity are shown in Fig. 16.
sometimes contains small amounts of the low-
melting eutectic that forms during ingot solidifi-
Total quantity to be forged cation. Forging of this alloy above about 315 C
10010,000 10,000
Maximum severity Die material Hardness, HB Die material Hardness, HB
(600 F)the melting point of the eutecticcan
cause severe rupturing. This problem can be
Hammer forging minimized by holding the cast ingot for ex-
Part 1 H11 405433 H12 405448
6G, 6F2 341375
tended periods at an elevated temperature to re-
Part 2 6G, 6F2 341375 6G, 6F2 341375 dissolve the eutectic and to restore a higher
H12(a) 405448 solidus temperature.
Part 3 6G, 6F2 269293 6G, 6F2 302331 Hydraulic presses or slow-action mechanical
Part 4 H11 405433 H11 405433
Part 5 6G, 6F2 302331 6G, 6F2(b) 302331
presses are the most commonly used machines
for the open-die and closed-die forging of mag-
Press forging nesium alloys. In these machines, magnesium
Part 1 H12 477514 H12 477514
6G, 6F2 341375
alloys can be forged with small corners and fil-
Part 2 6G, 6F2 341375 H12 477514 lets and with thin web or panel sections. Corner
Part 3 Part normally is not press forged from copper alloys radii of 1.6 mm (1/16 in.), fillet radii of 4.8 mm
Part 4 H11 405433 6G, 6F2(c) 341375 (3/16 in.), and panels or webs 3.2 mm (1/8 in.)
Part 5 6G, 6F2 341375 H12 477514
thick are not uncommon. The draft angles re-
(a) Recommended for long runsfor example, 50,000 pieces. (b) With either steel, use H12 insert at 405448 HB. (c) With either steel, use H12 in- quired for extraction of the forgings from the
sert at 429448 HB
dies can be held to 3 or less.
Magnesium alloys are seldom hammer forged
or forged in a rapid-action press because they
ing rectangular-section bar stock and sawing Magnesium Alloys will crack unless exacting procedures are used.
slugs from it. Alloys ZK60A, AZ31B, and HM21A are more
Upset forging is used less frequently for cop- Magnesium and its alloys, like other ductile easily forged by these methods than AZ80A,
per alloys than for steels, primarily because cop- metals, can be formed by any of the bulk work- which is extremely difficult to forge. Cracking
per alloys are so easily extruded. A part having a ing processes, including forging, extrusion, and can occur also in moderately severe, unsupported
long shaftlike section and a larger-diameter head rolling. Because of its hexagonal crystal struc- bending. Magnesium alloys generally flow later-
can often be made at less cost by extruding the ture, magnesium has relatively low cold worka- ally rather than longitudinally. This characteristic
smaller cross section from a larger one than by bility. Magnesium alloys also are good conduc- must be considered in the design of tools.
starting with a small cross section and upsetting tors of heat; therefore, dies must be heated to An important factor in the forging of magne-
to obtain the head. temperatures not much lower than those used to sium alloys is to refine the grain size. Alloys that
In the upsetting of copper alloys, the same heat the stock (Table 22). are subject to rapid grain growth at forging tem-
rule applies for maximum unsupported length as The forgeability of magnesium alloys depends peratures (AZ31B, AZ61A, and AZ80A) are
is used for steels, that is, not more than three on three factors: the solidus temperature of the generally forged at successively lower tempera-
times stock diameter. For the forging of brass, alloy, the deformation rate, and the grain size. tures for each operation. Common practice is to
single-blow upsetting as severe as 3 to 1 (upset Only forging-grade billet or bar stock should be reduce the temperature about 15 to 20 C (25 to
three times starting diameter) is considered rea- used in order to ensure good workability. This 35 F) after each step. For parts containing re-
type of product has been conditioned and in- gions that receive only small reductions, all
sonable. In practice, however, upsets of this
spected to eliminate surface defects that could forging is often done at the lowest practical tem-
severity are rare. The degree of allowable upset
open during forging, and it has been homoge- perature to permit strain hardening. Grain
for other copper alloys is somewhat less than
nized by the supplier to ensure good forgeability. growth in ZK60A and HM21A is slow at forging
that for forging brass, generally in proportion to
Table 22 lists commonly forged magnesium al- temperatures, and there is little risk of extensive
forgeability (Table 19). In most designs, the
loys along with their forging temperatures. grain growth.
amount of upset can be reduced by using slugs Magnesium alloy forgings are water
Magnesium alloys are often forged within 55
cut from specially shaped extrusions or by using C (100 F) of their solidus temperature. An ex- quenched directly from the forging operation to
one or more blocking impressions in the forging
sequence. Table 22 Recommended forging temperature ranges for magnesium alloys
Ring rolling is sometimes used as a means of
saving material when producing ring gears or sim- Recommended forging temperature(a)
Workplace Forging dies
ilar ringlike parts. Temperatures are the same as Alloy C F C F
those for forging the same alloy in closed dies.
Cost usually governs the minimum practical size Commercial alloys
ZK21A 300370 575700 260315 500600
for ring rolling. Most rings up to 300 mm (12 in.) AZ61A 315370 600700 290345 550650
in outside diameter are more economically pro- AZ31B 290345 550650 260315 500600
duced in closed dies. However, if the face width is High-strength alloys
less than about 25 mm (1 in.) it is often less ex- ZK60A 290385 550725 205290 400550
pensive to produce rings no larger than 200 mm (8 AZ80A 290400 550750 205290 400550
in.) in outside diameter by the rolling technique. Elevated-temperature alloys
The alloy being forged is also a factor in selecting HM21A 400525 750975 370425 700800
EK31A 370480 700900 345400 650750
ring rolling or closed-die forging. For example, al-
loys such as beryllium copper that are difficult to Special alloys
ZEA42A 290370 550700 300345 575650
forge are better adapted to ring rolling. For these ZE62 300345 575675 300345 575675
alloys, ring rolling is sometimes used for sizes QE22A 345385 650725 315370 600700
smaller than the minimum practical for the more (a) The strain-hardening alloys must be processed on a declining temperature scale within the given range to preclude recrystallization
easily forged alloys.
prevent further recrystallization and grain cross section and therefore the rate of ingot con- In open-die forging, a series of moderate re-
growth. With some of the age-hardening alloys, version. duction passes along the entire length of the
the quench retains the hardening constituents in General Guidelines for the Finish Forging forging is preferred. In working a square section
solution so that they are available for precipita- of Nickel-Base Alloys. Figure 17 shows the into a round, the piece should be worked down
tion during subsequent aging treatments. temperature ranges for the safe forging of 12 in the square form until it approaches the final
nickel-base alloys. Use of the lower part of the size. It should then be converted to an oversize
Nickel-Base Alloys (Ref 33) temperature range may be required for the de- octagon before finishing into the round. Billet
velopment of specific mechanical properties. corners that will be in contact with dies should
Most nickel-base alloys are stronger and Closed-die forging of nickel-base alloys is be chamfered rather than left square. The work
stiffer than steel, but alloy 200 (UNS N02200) generally done below the g solvus temperature should be lifted away from the dies occasionally
and alloy 400 (UNS N04400) are softer than in order to avoid excessive grain growth. to permit relief of local cold areas.
many steels. For example, Table 23 lists the Approximately 80% of the reduction is sched- Other Considerations. The precipitation-
pressures developed in the roll gap at 20% re- uled in the recrystallization temperature range, hardenable nickel alloys are subject to thermal
duction in hot rolling for five nickel-base alloys with the remaining 20% done at lower tempera- cracking. Therefore, localized heating is not rec-
and two steels at four hot-working temperatures. tures to introduce a certain amount of warm ommended. The entire part should be heated to
Higher pressures indicate greater resistance to work for improved mechanical properties. the forging temperature.
hot deformation. Sufficiently powerful equip- Preheating of all tools and dies to about 260 C If any ruptures appear on the surface of the
ment is of particular importance when forging (500 F) is recommended to avoid chilling the metal during hot working, they must be removed
alloys 800 (UNS N08800), 600 (UNS N06600), metal during working. at once, either by hot grinding or by cooling the
625 (UNS N06625), and the precipitation-hard- Forging Rate. A very rapid rate of forging work and cold overhauling. If the ruptures are
enable alloys such as 718 (UNS N07718) and X- often causes heat buildup (due to friction and de- not removed, they may extend into the body of
750 (UNS N07750). These alloys were specifi- formation heating), a nonuniform recrystallized the part.
cally developed to resist deformation at elevated grain size, and mechanical-property variations. For sections equal to or larger than 400 mm
temperatures. Susceptibility to free-surface ruptures also in- (16 in.) square, precautions should be taken in
General Guidelines for the Breakdown of creases with forging rate (and forging tempera- heating precipitation-hardenable alloys. They
Nickel-Base Alloys (Ref 34). Because of their ture). Therefore, slow strain rates are typically should be charged into a furnace at 870 C (1600
high alloy content and generally narrow working used during the initial closed-die reductions of F) or colder and brought up to forging temper-
temperature range, nickel-base alloys must be such alloys as Astroloy (UNS N13017) and Ren ature at a controlled rate of 40 C/h (100 F/h).
converted from cast ingots with care. Initial 95 (Ni-14Cr-8Co-3.5Mo-3.5W-3.5Nb-3.5Al- Cooling after Forging. The rate of cooling
breakdown operations are generally conducted 2.5Ti). With proper selection of starting stock after forging is not critical for alloys 200, 400,
well above the g solvus temperature, with subse- and forging temperature, however, the forging and 625. Alloys K-500 (UNS N05500) and 301
quent deformation completed below it but still rate is less critical. For example, some Astroloy (UNS N03301) should be water quenched from
high enough to avoid excessive warm working turbine components are currently hammer forging temperatures to avoid the excessive
and an unrecrystallized microstructure. The orig- forged. hardening and cracking that could occur if they
inal cast structure must be completely refined Forging Reduction. A sufficient amount of were cooled slowly through the age-hardening
during breakdown, that is, before final forging, recrystallization is necessary in each of a series range and to maintain good response to subse-
particularly when substantial levels of reduction of closed-die forging operations to achieve the quent aging. Alloy 825 (UNS N08825) should
not imposed during closed-die forging. desired grain size and to reduce the effects of the be cooled at a rate equal to or faster than air
Good heat-retention practice during ingot continuous grain-boundary or twin-boundary cooling.
breakdown is an important factor in obtaining a carbide networks that develop during heating Alloys 800 and 600 are subject to carbide pre-
desirable billet microstructure. Rapid transfer of and cooling. This condition contributes more to cipitation during heating in or slow cooling
the ingot from the furnace to the forging press, mechanical-property and other problems than through the temperature range of 540 to 760 C
as well as the use of such techniques as reheat- any other single factor. Poor weldability, low- (1000 to 1400 F). If sensitization is likely to
ing during breakdown, is necessary to promote cycle fatigue, and stress-rupture properties are prove disadvantageous in the end use, parts
sufficient recrystallization during each forging associated with continuous grain-boundary car- made of these alloys should be water quenched
pass. In addition, it has been found that diffusion bide networks. Heat treatment can do very little or cooled rapidly in air.
of precipitation-hardening elements is associ- to correct this problem without creating equally The precipitation-hardenable alloys should, in
ated with recrystallization during ingot conver- undesirable mechanical-property problems general, be cooled in air after forging. Water
sion. Mechanical factors such as cycling speed when higher solution-treatment temperatures are quenching is not recommended, because of the
(which affects heat losses), reduction, length of used. All portions of a part must receive some possibility of thermal cracking, which can occur
pass, die design, and press capacity all influence hot work after the final heating operation in during subsequent heating for further forging or
the degree of work penetration through the billet order to achieve uniform mechanical properties. heat treating.
Forging Practice for Specific Alloys. The
following practices are used in the forging of
Table 23 Hot-forming pressures for several nickel-base alloys
nickel-base alloys. However, as noted in the sec-
Pressures developed in the hot forming of 1020 steel and AISI type 302 stainless steel are shown for comparison. tion on Superalloys, state-of-the-art forging
Pressure developed at working temperature(a)
practices for nickel-base alloys depend on ther-
870 C (1800 F) 1040 C (1900 F) 1095 C (2000 F) 1150 C (2100 F) momechanical process design for the develop-
Alloy UNS No. MPa ksi MPa ksi MPa ksi MPa ksi ment of desirable properties and microstruc-
400 N04400 124 18 106 15.3 83 12 68 9.8 tures. Variations from these procedures may be
600 N06600 281 40.8 239 34.6 195 28.3 154 22.3 necessary for some specialized applications (see
625 N06625 463 67.2 379 55 297 43 214 31 the sections Thermomechanical Processing
718 N07718 437 63.3 385 55.8 333 48.3 283 41 and Isother-mal Forging in this article).
X-750 N07750 335 48.6 299 43.3 265 38.4 230 33.3
1020 steel G10200 154 22.4 126 18.3 99 14.3 71 10.3 Alloy 200 should be charged to a hot furnace,
Type 302 stainless steel S30200 192 27.8 168 24.3 148 21.4 124 18 withdrawn as soon as the desired temperature
(a) Pressure developed in the roll gap at 20% reduction in hot rolling
has been reached, and worked rapidly. The rec-
ommended range of forging temperatures is 650
perature of 790 C (1450 F) or higher. If the

piece is allowed to cool slowly, it will age
harden to some extent, and stress will be set up
that may lead to thermal splitting or tearing dur-
ing subsequent reheating. In addition, quenched
material has better response to age hardening
because more of the age-hardening constituent is
retained in solution.
Alloy 600 has a normal forging range of 870 to
1230 C (1600 to 2250 F). Heavy hot work
should be done in the range from 1040 to 1230
C (1900 to 2250 F). Light working can be con-
tinued down to 870 C (1600 F). Generally,
forging should not be done between 650 and 870
C (1200 and 1600 F) because of the low duc-
tility of the alloy in this temperature range.
Fig. 17 Forging temperature ranges for 12 nickel-base alloys Judicious working at a temperature below 650 C
(1200 F) will develop higher tensile properties.
to 1230 C (1200 to 2250 F). Because the metal Reheat. The rate of cooling after forging is not critical
stiffens rapidly when cooled to about 870 C Forge to a section having about 5% larger with respect to thermal cracking. However, alloy
(1650 F), all heavy work and hot bending area than the final shape (take at least 25% 600 is subject to carbide precipitation in the
should be done above that temperature. High reduction). range between 540 and 760 C (1000 and 1400
mechanical properties can be produced by work- Cool to 705 C (1300 F). F), and if subsequent use dictates freedom from
ing lightly below 650 C (1200 F). The best Finish to size (5% reduction). sensitization, the part should be rapidly cooled
range for hot bending is 870 to 1230 C (1600 to through this temperature range.
2250 F). High-tensile forgings, as described in certain Alloy 625 should be heated in a furnace held at
Alloy 301 has an optimal forging range of military specifications, also require a minimum 1175 C (2150 F) but no higher. The work
1065 to 1230 C (1900 to 2250 F). Light fin- of 30 to 35% reduction after the last reheat. This should be brought as close to this temperature as
ishing work can be done down to 870 C (1600 is taken in the following manner: conditions permit. Forging is done from this tem-
F). Finer grain size is produced in forgings by perature down to 1010 C (1850 F); below 1010
Reheat.
using 1175 C (2150 F) for the final reheat tem- C (1850 F) the metal is stiff and hard to move,
Forge to a section having an area about 25%
perature and by taking at least 30% reduction of and attempts to forge it may cause hammer splits
larger than the final shape (take about 5%
area in the last forging operation. at the colder areas. The work should be returned
reduction).
After hot working, the alloy should be to the furnace and reheated to 1175 C (2150 F)
Cool to 705 C (1300 F).
quenched from a temperature of 790 C (1450 whenever its temperature drops below 1010 C
Finish to size (25% reduction).
F) or above. Quenching retains the strain hard- (1850 F). To guard against duplex grain struc-
Grain refinement is achieved by using a
ening imparted by the forging operation and pro- ture, the work should be given uniform reduc-
temperature of 1095 C (2000 F) for the
duces better response to subsequent age harden- tions. For open-die work, final reductions of a
final reheat and by increasing the amount of
ing. Quenching in water containing about 2 minimum of 20% are recommended.
reduction taken after the last reheat.
vol% alcohol results in less surface oxidation. Alloy 718 is strong and offers considerable re-
Material that must be cooled prior to subse- Alloy K-500 has a maximum recommended sistance to deformation during forging. The
quent hot working should also be quenched. heating temperature for forging of 1150 C forces required for hot deformation are some-
Slow cooling may cause age hardening, which (2100 F). Metal should be charged into a hot what higher than those employed for alloy X-
sets up stresses in the workpiece that can cause furnace and withdrawn when uniformly heated. 750. Alloy 718 is forged in the range from 900
cracking during subsequent reheating. Prolonged soaking at this temperature is harm- to 1120 C (1650 to 2050 F). In the last opera-
Alloy 400 has a maximum heating tempera- ful. If a delay occurs such that the material tion, the metal should be worked uniformly with
ture for forging of 1175 C (2150 F). Prolonged would be subject to prolonged soaking, the tem- a gradually decreasing temperature, finishing
soaking at the working temperature is detrimen- perature should be reduced to or held at 1040 C with some light reduction below 955 C (1750
tal. If a delay occurs during processing, the fur- (1900 F) until shortly before working is to F). In heating for forging, the material should
nace temperature should be reduced to 1040 C begin, then brought to 1150 C (2100 F). When be brought up to temperature, allowed to soak a
(1900 F) and not brought to 1175 C (2150 F) the piece is uniformly heated, it should be with- short time to ensure uniformity, and withdrawn.
until operations are resumed. drawn. In the event of a long delay, the work Alloy 718 should be given uniform reductions
The recommended metal temperature for should be removed from the furnace and water in order to avoid duplex grain structure. Final re-
heavy reductions is 925 to 1175 C (1700 to quenched. ductions of 20% minimum should be used for
2150 F). Light reductions may be taken at tem- The forging range is 870 to 1150 C (1600 to open-die work, and 10% minimum for closed-
peratures down to 650 C (1200 F). Working at 2100 F). Heavy work is best done between die work. Parts should generally be air cooled
the lower temperatures produces higher mechan- 1040 and 1150 C (1900 and 2100 F), and from the forging temperature, rather than water
ical properties and smaller grain size. working below 870 C (1600 F) is not recom- quenched.
A controlled forging procedure is necessary to mended. To produce finer grain in forgings, Alloy 706 (UNS N09706) is similar to alloy
meet the requirements of some specifications for 1095 C (2000 F) should be used for the final 718, except that alloy 706 is more readily fabri-
forged hot-finished parts. Both the amount of re- reheat temperature, and at least 30% reduction cated, particularly by machining. Forging should
duction and the finishing temperature must be of area should be taken in the last forging oper- be done using the same procedures and temper-
controlled in order to develop the desired prop- ation. atures as those for alloy 718.
erties. When forging has been completed or when it Alloy X-750 has a forging range of 980 to
One procedure for producing forgings to such is necessary to allow alloy K-500 to cool before 1205 C (1800 to 2200 F). Below 980 C (1800
specifications consists of taking a 30 to 35% re- further hot working, it should not be allowed to F), the metal is stiff and hard to move, and at-
duction after the final reheat. This is done by: cool in air, but should be quenched from a tem- tempts to work it may cause splitting. All heavy
forging should be done at about 1040 C (1900 reduction, these alloys should be heated between temperatures slightly above the recrystallization
F), and the metal should be reheated whenever 995 and 1050 C (1825 and 1925 F). The final temperature led to the development of finer
it cools to below that temperature. Forgings can heating for alloys 907 and 909 should be 980 to grain sizes (ASTM 5 to 6). Coupling this with
be finished with some light reduction between 1025 C (1800 to 1875 F) for a minimum re- modified heat treating practices resulted in ex-
980 and 1040 C (1800 and 1900 F). duction of 20% over a falling temperature range cellent combinations of tensile, fatigue, and
As a general rule, alloy X-750 should be air (finishing at 925 C, or 1700 F). The final creep properties.
cooled rather than liquid quenched from the heating for alloy 903 should be 870 C (1600 F) Forging temperature is carefully controlled
forging temperature. Liquid quenching can with a final forging reduction of 40% minimum. during the thermomechanical processing of
cause high residual stresses that may result in Thermomechanical processing (TMP) nickel- and nickel-iron-base alloys to make use
cracking during subsequent heating for further refers to the control of temperature and defor- of the structure control effects of second phases
hot work or for heat treatment. Parts with large mation during processing to enhance specific such as g. Above the optimal forging tempera-
cross sections and pieces with variable cross properties. Special TMP sequences have been ture range (Table 25), the structure-control phase
sections are especially susceptible to thermal developed for a number of nickel-base alloys, goes into solution and loses its effect. Below this
cracking during cooling. In very large cross sec- and the design of TMP sequences relies on a range, extensive fine precipitates are formed,
tions, furnace cooling may be necessary to pre- knowledge of the melting and precipitation tem- and the alloy becomes too stiff to process.
vent thermal cracking. peratures for the precipitates in the alloy of in- Some examples of specific TMP sequences
Alloy 800. Hot working of alloy 800 starts at terest. Precipitates in nickel-base (as well as are given below. State-of-the-art TMP practices
1205 C (2200 F), and heavy forging is done at iron- and cobalt-base) alloys include various for nickel-base alloys rely on the following mi-
temperatures down to 1010 C (1850 F). Light types of metal carbides such as MC carbides (M crostructural effects (Ref 2):
working can be accomplished down to 870 C = titanium, niobium, etc.), M6C carbides (M =
(1600 F). No working should be done between molybdenum and/or tungsten), or M23C6 car- Dynamic recrystallization is the most im-
870 and 650 C (1600 and 1200 F). As with bides (M = chromium). However, the primary portant softening mechanism during hot
alloy 600, thermal cracking is not a problem, precipitate of concern in the processing of pre- working.
and workpieces should be cooled rapidly cipitation-strengthened nickel-base alloys is the Grain boundaries are preferred nucleation
through the range between 540 and 760 C g strengthening precipitate. g is an ordered sites for recrystallization.
(1000 and 1400 F) to ensure freedom from sen- face-centered cubic (fcc) compound in which The rate of recrystallization decreases with
sitization. aluminum and titanium combine with nickel to the temperature and/or the extent of defor-
Alloy 825 has a forging range of 870 to 1175 form Ni3(Al,Ti). In nickel-iron alloys such as mation.
C (1600 to 2150 F). It is imperative that some alloy 718, titanium, niobium, and, to a lesser ex- Precipitation that may occur during the re-
reduction be accomplished in the range between tent, aluminum combine with nickel to form or- crystallization can inhibit the softening
870 and 980 C (1600 and 1800 F) during final dered fcc g or ordered body-centered tetragonal process.
forging to ensure maximum corrosion resist- g. Nickel-iron-base alloys are also prone to the Recrystallization cannot be completed until
ance. formation of other phases, such as hexagonal the precipitate coarsens to a relatively inef-
Cooling after forging should be done at a rate close-packed Ni3Ti (h), as in titanium-rich alloy fective morphology.
equal to or faster than air cooling. Heavy sec- 901, or orthorhombic Ni3Nb (d) in niobium-rich
tions may become sensitized during cooling alloy 718. Waspaloy. A typical TMP treatment of
from the forging temperature and therefore be Table 24 lists first melting and precipitation nickel-base alloys is that used for Waspaloy
subject to intergranular corrosion in certain temperatures for several nickel-base alloys. (UNS N07001) to obtain good tensile and creep
media. A stabilizing anneal of 1 h at 940 C Early forging practice of nickel- and nickel-iron- properties. This consists of initial forging at
(1725 F) restores resistance to corrosion. If the base alloys consisted of forging from and solu- 1120 C (2050 F) and finish forging below ap-
forged piece is to be welded and used in an en- tion heat treating at temperatures well in excess proximately 1010 C (1850 F) to produce a
vironment that could cause intergranular corro- of the g solvus temperature. High-temperature fine, equiaxed grain size of ASTM 5 to 6.
sion, the piece should be given a stabilizing an- solution treatment dissolved all of the g, an- Solution treatment is then done at 1010 C (1850
neal to prevent sensitization from the heat of nealed the matrix, and promoted grain growth F), and aging is conducted at 845 C (1550 F)
welding, regardless of the cooling rate after (typical grain size ASTM 3 or coarser). This for 4 h, followed by air cooling plus 760 C
forging. was followed by one or more aging treatments (1400 F) for 16 h and then air cooling.
Alloy 925. The hot-working characteristics that promoted controlled precipitation of g and Ren 95. Initial forging of Ren 95 is done
of alloy 925 (UNS N09925) are similar to those carbide phases. Optimal creep and stress-rupture between 1095 and 1140 C (2000 and 2080 F).
of alloy 825 at temperatures to 1095 C (2000 properties above 760 C (1400 F) were thus Following an in-process recrystallization anneal
F). At higher temperatures, alloy 925 has lower achieved. Later in the development of forging at 1175 C (2150 F), finish forging (reduc-
ductility and higher strength. The forging range practice, it was found that using preheat furnace tion 40 to 50%) is then imposed below the g
is 870 to 1175 C (1600 to 2150 F). For maxi-
mum corrosion resistance and highest mechani-
Table 24 Critical melting and precipitation temperatures for several nickel-base alloys
cal properties after direct aging, final hot work-
ing should be done in the range of 870 to 980 C First melting temperature Precipitation temperature
Alloy UNS No. C F C F
(1600 to 1800 F).
Alloys 722 and 751 (UNS N07722 and Alloy X N06002 1260 2300 760 1400
N07751, respectively) are forged using the same Alloy 718 N07718 1260 2300 845 1550
procedures and temperatures as those for alloy Waspaloy N07001 1230 2250 980 1800
Alloy 901 N09901 1200 2200 980 1800
X-750. Alloy X-750 N07750 1290 2350 955 1750
Alloys 903, 907, and 909 (UNS N19903, M-252 N07252 1200 2200 1010 1850
N19907, and N19909, respectively) are best Alloy R-235 ... 1260 2300 1040 1900
forged in three stages in order to obtain the de- Ren 41 N07041 1230 2250 1065 1950
U500 N07500 1230 2250 1095 2000
sired properties after aging. The initial break- U700 ... 1230 2250 1120 2050
down of 40% minimum reduction should be per- Astroloy N13017 1230 2250 1120 2050
formed at 1060 to 1120 C (1940 to 2050 F). Source: Ref 35
For intermediate forging at a minimum of 25%
Table 25 Structure control phases and working temperature ranges for various heat- any other commonly forged material. Therefore,
resistant alloys the forging process in titanium alloys is used not
Working temperature range
only to create cost-effective forging shapes, but
Alloy UNS No. Phases for structure control C F also, in combination with thermal treatments, to
create unique and/or tailored microstructures to
Nickel-base alloys
Waspaloy N07001 g (Ni3(Al,Ti) 9551025 17501875
achieve the desired final mechanical properties.
Astroloy N13017 g (Ni3(Al,Ti) 10101120 18502050 By the design of the working process history
IN-100 ... g (Ni3(Al,Ti) 10401175 19002150 from ingot to billet to forging, and particularly
Ren 95 ... g (Ni3(Al,Ti) 10251135 18752075 the selection of metal temperatures and defor-
Nickel-iron-base alloys mation conditions during the forging process,
901 N09901 h (Ni3Ti) 940995 17251825 significant changes in the morphology of the al-
718 N07718 d (Ni3Nb) 915995 16751825 lotropic (a, b) phases of titanium alloys are
Pyromet CTX-1 ... h (Ni3Ti), d (Ni3Nb), or both 855915 15751675
achieved that in turn dictate the final mechanical
Source: Ref 36 properties and characteristics of the alloy. The
classes of titanium alloys (a, a + b, b) exert a
strong influence on forging part design and forg-
solvus, typically between 1080 and 1105 C and 901 have even lower volume fractions of ing-process selection.
(1975 and 2025 F). The large grains formed precipitate and are therefore even less frequently Classes of Titanium Alloys. Titanium and
during high-temperature recrystallization are used in isothermal forging. its alloys exist in two allotropic forms: the
elongated and surrounded by small recrystal- As the term implies, isothermal forging con- hexagonal close-packed (hcp) a phase and the
lized grains that form during finish forging. sists of forging with the workpiece and the dies body-centered cubic (bcc) b phase. The more
Alloy 901. The thermomechanical processing at the same temperature. Because this tempera- difficult to deform a phase is usually present at
of alloy 901 is often done to produce a fine-grain ture is often of the order of 980 to 1095 C (1800 low temperatures, while the more easily de-
structure that enhances fatigue strength. This is ac- to 2000 F), the dies are usually made of molyb- formed b phase is present at high temperatures.
complished by using the h (Ni3Ti) phase, which is denum for elevated-temperature strength. The However, the addition of various alloying ele-
introduced in a Widmansttten form at the begin- isothermal forging system must be operated in a ments (including other metals and such gases as
ning of processing by a heat treatment at 900 C vacuum or inert atmosphere to protect such die oxygen, nitrogen, and hydrogen) stabilizes ei-
(1650 F) for 8 h. Forging is then conducted at 955 materials from oxidation. ther the a or b phase. The temperature at which
C (1750 F), which is below the h solvus; the Compared to conventional forging, isothermal a given titanium alloy transforms completely
forging deformation is completed below the re- forging deformation rates are slow; hydraulic from a to b is termed the beta transus (bt). The
crystallization temperature. A fine-grain structure press speeds of approximately 2.5 mm/min (0.1 bt is a critical temperature in titanium alloy forg-
is generated by a subsequent recrystallization in./min) are typical. However, the slower pro- ing-process criteria.
treatment below the h solvus. The needlelike h duction rate is largely offset by the ability to Based on the predominant allotropic form(s)
phase will become spherical during forging and forge complex shapes to closer tolerances, present at room temperature, titanium alloys are
will restrict grain growth. Aging is then conducted which leads to less machining and substantial divided into three major classes:
according to standard procedures. material savings. In addition, a large amount of
a/near-a alloys
Isothermal Forging. Nickel-base alloys deformation is accomplished in one operation,
a + b alloys
that are hard to work or are typically used in the pressures are low, and uniform microstructures
b/metastable b alloys
cast condition can be readily forged when in a are achieved. For example, the as-forged weight
powder-consolidated form. The most common of a finish-machined 68 kg (150 lb) Astroloy Table 26 lists most of the commonly forged ti-
forging technique using powder preforms is disk is about 110 kg (245 lb) for a conventional tanium alloys by alloy class, along with the
isothermal forging. In this process, powder is forging versus 72 kg (160 lb) for the correspon- major alloying elements constituting each alloy.
produced by inert gas atomization and is com- ding isothermal forging. Each of these types of titanium alloys has unique
pacted into billet form by extrusion. The billets forging-process criteria and deformation behav-
are fabricated below the g solvus temperature Titanium Alloys (Ref 37) ior. Further, the forging-process parameters,
for alloys such as IN-100 in order to maintain a often in combination with subsequent thermal
fine grain size and a fine distribution of precipi- As a class of materials, titanium alloys are treatments, are manipulated for each alloy type
tates. In this condition, the material exhibits su- among the most difficult metal alloys to forge, to achieve the desired final forging microstruc-
perplastic properties that are characterized by ranking behind only refractory metals and ture and mechanical properties (heat treatment
large tensile elongations (during sheet forming) nickel-cobalt-base superalloys. Therefore, tita- serves a different purpose in titanium alloys
and good die-filling capacity (during forging). nium alloy forgings, particularly closed-die forg- from that in aluminum alloys or alloy steels, as
Multiples of the extruded bar are then isother- ings, are typically produced to less highly refined discussed in the paragraphs that follow).
mally forged into a variety of complex turbine final forging configurations than are typical of Alpha/near- titanium alloys contain ele-
engine and other high-temperature parts. aluminum alloys. However, precision titanium ments that stabilize the hcp a phase at higher
The key to successful isothermal forging of alloy forgings can be produced to the same de- temperatures. These alloys (with the exception
nickel-base alloys is the ability to develop a fine sign and tolerance criteria as aluminum alloy of commercially pure titanium, which is also an
grain size before forging and to maintain it dur- precision forgings. Titanium alloy forgings also a alloy) are among the most difficult titanium al-
ing forging. With regard to the latter, a high vol- may be equivalent to, or more refined than, car- loys to forge. Typically, a/near-a titanium alloys
ume percentage of second phase is useful in pre- bon or low-alloy steel forgings, because of re- have modest strength but excellent elevated-tem-
venting grain growth. Therefore, alloys such as duced oxidation or scaling tendencies in heating. perature properties. Forging and TMP processes
IN-100, Ren 95, and Astroloy, which contain Thermomechanical processing is another im- for a alloys are typically designed to develop
large amounts of g, are readily capable of de- portant aspect in forging of titanium alloys. optimal elevated-temperature properties, such as
veloping the superplastic properties necessary in Working history and forging parameters used in strength and creep resistance. The bt of a/near-a
isothermal forging. In contrast, Waspaloy, which titanium alloy forging have a significant impact alloys typically ranges from 900 to 1065 C
contains less than 25 vol% g at isothermal forg- on the final microstructure (and therefore the re- (1650 to 1950 F).
ing temperatures, is only marginally superplas- sultant mechanical properties) of the forged tita- Alpha-beta titanium alloys represent the most
tic. Nickel-iron-base alloys such as alloys 718 nium alloysperhaps to a greater extent than in widely used class of titanium alloys (Ti-6Al-4V
Table 26 Recommended forging temperature ranges for commonly forged titanium alloys to whether the forging temperature is below or
t Forging temperature(b)
above the bt. The two basic categories are (a)
Alloy C F Process(a) C F conventional (a-b) forging, where the alloy tem-
perature during forging is predominantly below
/near- alloys
Ti-C.P.(c) 915 1675 C 815900 15001650
the bt and (b) b forging, where the alloy temper-
Ti-5Al-2.5Sn(c) 1050 1925 C 9001010 16501850 ature during forging is predominantly above the
Ti-5Al-6Sn-2Zr-1Mo-0.1Si 1010 1850 C 900995 16501925 bt. These are the two principal metallurgical ap-
Ti-6Al-2Nb-1Ta-0.8Mo 1015 1860 C 9401050 17251825 proaches to the forging of titanium alloys.
B 10401120 19002050
Ti-6Al-2Sn-4Zr-2Mo(+0.2Si)(d) 990 1815 C 900975 16501790
However, within these fundamental approaches,
B 10101065 18501950 there are several possible variations that blend
Ti-8Al-1Mo-1V 1040 1900 C 9001020 16501870 these two techniques into processes that are used
IMI 685 (Ti-6Al-5Zr-0.5Mo-0.25Si)(e) 1030 1885 C/B 9801050 17951925 commercially to achieve controlled microstruc-
IMI 829 (Ti-5.5Al-3.5Sn-3Zr-1Nb-0.25Mo-0.3Si)(e) 1015 1860 C/B 9801050 17951925
IMI 834 (Ti-5.5Al-4.5Sn-4Zr-0.7Nb-0.5Mo-0.4Si-0.06C)(e) 1010 1850 C/B 9801050 17951925
tures that tailor the final properties of the forging
to specification requirements and/or intended
- alloys service applications. In particular, manipulation
Ti-6Al-4V(c) 995 1825 C 900980 16501800
B 10101065 18501950
of the a phase is important during the conven-
Ti-6Al-4V ELI 975 1790 C 870950 16001740 tional (a-b) forging of the a and a-b alloys. By
B 9901045 18151915 the selection of metal temperatures and defor-
Ti-6Al-6V-2Sn 945 1735 C 845915 15501675 mation conditions, the forging process can have
Ti-6Al-2Sn-4Zr-6Mo 940 1720 C 845915 15501675
B 9551010 17501850
a significant influence on the a-phase morphol-
Ti-6Al-2Sn-2Zr-2Mo-2Cr 980 1795 C 870955 16001750 ogy, which in turn dictates the mechanical prop-
Ti-17 (Ti-5Al-2Sn-2Zr-4Cr-4Mo(f) 885 1625 C 805865 14801590 erties and characteristics of the alloy. In fully b-
B 900970 16501775 stabilized alloys, manipulation of the a phase
Corona 5 (Ti-4.5Al-5Mo-1.5Cr) 925 1700 C 845915 15501675
B 9551010 17501850
through forging-process techniques is less
IMI 550 (Ti-4Al-4Mo-2Sn) 990 1810 C 900970 16501775 prevalent; therefore, fully b-stabilized alloys are
IMI 679 (Ti-2Al-11Sn-4Zr-1Mo-0.25Si) 945 1730 C 870925 16001700 typically forged above the bt of the alloy.
IMI 700 (Ti-6Al-5Zr-4Mo-1Cu-0.2Si) 1015 1860 C 800900 14701650 Table 26 lists recommended metal tempera-
/near-/metastable alloys tures for 27 commonly forged a, a-b, and b ti-
Ti-8Al-8V-2Fe-3Al 775 1425 C/B 705980 13001800 tanium alloys. With some exceptions, these al-
Ti-10V-2Fe-3Al 805 1480 C 705785 13001450 loys can be forged to the same degree of
B 815870 15001600
Ti-13V-11Cr-3Al 675 1250 C/B 650955 12001750
severity; however, the power and/or pressure re-
Ti-15V-3Cr-3Al-3Sn 770 1415 C/B 705925 13001700 quirements needed to achieve a given forging
Beta C (Ti-3Al-8V-6Cr-4Mo-4Zr) 795 1460 C/B 705980 13001800 shape may vary with each individual alloy and
Beta III (Ti-4.5Sn-6Zr-11.5Mo) 745 1375 C/B 705955 13001750 particularly with alloy class. As a general guide,
Transage 129 (Ti-2Al-11.5V-2Sn-11Zr) 720 1325 C/B 650870 12001600
Transage 175 (Ti-2.7Al-13V-7Sn-2Zr) 760 1410 C/B 705925 13001700
metal temperatures of bt - 28 C (50 F) for a-b
forging and bt + 42 C (75 F) for beta forging,
(a) C, conventional forging processes in which most or all of the forging work is accomplished below the bt of the alloy for the purposes of desired
mechanical property development. This forging method is also referred to as a-b forging. B, b forging processes in which some or all of the forging
are recommended.
is conducted above the bt of the alloy to improve hot workability or to obtain desired mechanical property combinations. C/B, either forging method- The upper limits of the temperature ranges in
ology (conventional or b) is employed in the fabrication of forgings or for alloys, such as b alloys, that are predominately forged above their bt but Table 26 are based on (a) prudent proximity
may be finish forged at subtransus temperatures. (b) These are recommended metal temperature ranges for conventional a-b, or b forging processes
for alloys for which the latter techniques are reported to have been employed. The lower limit of the forging temperature range is established for open- (from furnace temperature variations and minor
die forging operations in which reheating is recommended. (c) Alloys for which there are several compositional variations (primarily oxygen or other composition variations) to the nominal bt of the
interstitial element contents) that may affect both bt and forging temperature ranges. (d) This alloy is forged and used both with and without the sili-
con addition; however, the bt and recommended forging temperatures are essentially the same. (e) Alloys designed to be predominately b forged. (f) alloy in the case of conventional (a-b) forging
Ti-17 has been classified as an a-b and as a near-b titanium alloy. For purposes of this article, it is classified as an a-b alloy and (b) forging without undue metallurgical
risks (such as excessive grain growth) in the case
of b forging. The lower limit of the specified
is the most widely used of all titanium alloys) with good fracture toughness and excellent fa- ranges is the temperature at which forging
and contain sufficient b stabilizers to stabilize tigue characteristics; therefore, forging and TMP should be discontinued in the case of open-die
some of the b phase at room temperature. Alpha- processes are designed to optimize these prop- forging to avoid excessive cracking and/or other
beta titanium alloys are generally more readily erty combinations. The bt of b titanium alloys surface-quality problems.
forged than a alloys and are more difficult to ranges from 650 to 870 C (1200 to 1600 F). Conventional (-) forging of titanium alloys,
forge than some b alloys. Typically, a-b alloys Forging Processes. Titanium alloy forgings in addition to implying the use of die tempera-
have intermediate-to-high strength with excel- are produced by all of the typical mechanical tures of 540 C (1000 F) or less, is the term
lent fracture toughness and other fracture-related methods of forging, including open-die (or used to describe a forging process in which most
properties. Forging and TMP processes for a-b hand) forging, closed-die forging, upsetting, roll or all of the forging deformation is conducted at
alloys are designed to develop optimal combina- forging, orbital forging, spin forging, mandrel temperatures below the bt of the alloy. For a, a-
tions of strength, fracture toughness, and fatigue forging, ring rolling, and forward and backward b, and metastable b alloys, this forging tech-
characteristics. The bt of a-b alloys typically extrusion. Selection of the optimal forging nique involves working the material at tempera-
ranges from 870 to 1010 C (1600 to 1850 F). method for a given forging shape is based on the tures where both the a and b phases are present,
Beta/Metastable Alloys are those alloys desired forging shape, the sophistication of the with the relative amounts of each phase being
with sufficient b stabilizers that the bcc b-phase design of the forged shape, the cost, and the de- dictated by the composition of the alloy and the
is the predominant allotropic form present at sired mechanical properties and microstructure. actual temperature used. With this forging tech-
room temperature. Beta titanium alloys are usu- In many cases, two or more forging methods are nique, the resultant as-forged microstructure is
ally easier to fabricate than other classes of tita- combined to achieve the desired forging shape, characterized by deformed or equiaxed primary
nium alloys, although b alloys may be equiva- to obtain the desired final part microstructure, a in a transformed-b matrix. The volume frac-
lent to, or more difficult to forge than, a-b alloys and/or to minimize cost. tion of primary a is dictated by the alloy com-
under certain forging conditions. Beta titanium From a metallurgical standpoint, the methods position and the actual working history and tem-
alloys are characterized by very high strength of forging titanium also are classified according perature. Alpha-beta forging is typically used to
develop optimal strength/ductility combinations Ti-10V-2Fe-3Al b alloy displays even less metal 6Al-4V and 815 C in Fig. 18c for Ti-10V-2Fe-
and optimal high/low-cycle fatigue properties. temperature sensitivity, also at higher levels of 3Al). The differences in strain-softening behav-
With conventional (a-b) forging, the effects of total strain. At 815 C (1500 F), Ti-10V-2Fe- ior are a function of the differences in mi-
working on microstructure, particularly a-phase 3Al is being deformed above its bt with an at- crostructure present during the deformation
morphology changes, are cumulative; therefore, tendant reduction in flow stresses in comparison above or below the transus of the alloy. The
each successive a-b working operation adds to to subtransus deformation at 760 C (1400 F). equiaxed a in a b-matrix structure, typical of
the structural changes achieved in earlier opera- However, at this high strain rate, the flow-stress subtransus forging, has been found to redistrib-
tions. reduction achieved by deforming the b alloys ute strain and to promote dislocation movement
Beta forging, as the term implies, is a forging above its bt is less than the flow-stress reduction more effectively than acicular a in a transformed
technique for a, a-b, and metastable b alloys in achieved by deforming the a-b alloys above its b structure, leading to increased strain softening
which most or all of the forging work is done at bt. in the former.
temperatures above the bt of the alloy. In com- As with other forged materials, many titanium Flow stresses describe the lower limit of the
mercial practice, b forging techniques typically alloys display a strain-softening behavior at the deformation resistance of titanium alloys as rep-
involve supertransus forging in the early and/or strain rates typically used in conventional forg- resented by ideal deformation conditions and are
intermediate stages with controlled amounts of ing techniques. As shown in Fig. 18(a) to (c), therefore rarely present during actual forging
final deformation below the bt of the alloy. strain softening is typically observed when such
Actual final subtransus working criteria are de- alloys are forged below their bt. It is observed to
pendent on the alloy, the forging design, and the a much lesser extent when these alloys are b
mechanical property combinations sought. forged (for example, 1000 C in Fig. 18b for Ti-
In b forging, the working influences on mi-
crostructure are not fully cumulative; with each
working-cooling-reheating sequence above bt,
the effects of the prior working operations are at
least partially lost because of recrystallization
from the transformation upon heating above the
bt of the alloy. Beta forging techniques are used
to develop microstructures characterized by
Widmansttten or acicular primary a morphol-
ogy in a transformed b matrix. This forging
process is typically used to enhance fracture-re-
lated properties, such as fracture toughness and
fatigue crack propagation resistance, and to en-
hance the creep resistance of a and a-b alloys.
In fact, several a alloys are designed to be b
forged to develop the desired final mechanical
properties. There is often a loss in strength and
ductility with b forging as compared to a-b
forging. Beta forging, particularly of a and a-b
alloys, has the advantages of significant reduc-
tion in forging unit pressures and reduced
cracking tendency, but it must be done under
carefully controlled forging-process conditions
to avoid nonuniform working, excessive grain
growth, and/or poorly worked structures, all of
which can result in final forgings with unac-
ceptable or widely variant mechanical proper-
ties within a given forging or from lot to lot of
the same forging.
The effect of temperature on the flow
stresses of titanium alloys varies with alloy
class. This effect is illustrated in Fig. 18 for three
alloys, each representative of one class of tita-
nium alloy. For example, it is evident that the
more difficult-to-forge a alloy (Ti-8Al-1Mo-1V
in Fig. 18a) displays the greatest sensitivity to
metal temperature. The flow stress at 10/s and
900 C (1650 F) is two to three times that of the
alloy at 1010 C (1850 F) (the latter tempera-
ture is below the bt of the alloy). In comparison,
the a-b alloy (Ti-6Al-4V in Fig. 18b) also dis-
plays sensitivity to metal temperature, but to a
lesser extent than the Ti-8Al-1Mo-1V alpha
alloy, especially at higher levels of total strain. Fig. 19 Effect of three strain rates (0.001, 0.1, and
In Fig. 18(b), at 1000 C (1830 F), Ti-6Al-4V 10/s) on flow stress of three titanium alloys
Fig. 18 Effect of forging temperature on flow stress of forged at different temperatures. (a) a alloy Ti-8Al-1Mo-1V
is being deformed at or above the nominal bt of titanium alloys at 10/s strain rate. (a) a alloy at 955 C (1750 F). (b) a-b alloy Ti-6Al-4V at 900 C
the alloy, in which the structure is entirely bcc Ti-8Al-1Mo-1V. (b) a-b alloy Ti-6Al-4V. (c) Metastable b (1650 F). (c) Metastable b alloy Ti-10V-2Fe-3Al at 815 C
and considerably easier to deform. Finally, the alloy Ti-10V-2Fe-3Al (1500 F)
processes. However, flow-stress information, as Temperature Control and Heating. As sary and introduce the probability of excessive
a function of such forging-process variables as noted, titanium alloys have a relatively narrow scale or a case. Generally, soaking times should
temperature and strain rate, is useful in design- temperature range for conventional forging. be restricted to 1 to 2 h, and if unavoidable de-
ing titanium alloy forging processes. Because of Titanium alloy heating equipment should be lays are encountered, where soaking time may
other forging-process variables, such as die tem- equipped with pyrometric controls that can exceed 2 to 4 h, removal of the metal from the
perature, lubrication, prior working history, and maintain 14 C (25 F) or better. Titanium furnace is recommended.
total strain, actual forging pressures or unit pres- alloy stock heating equipment is often tempera- Heating of Dies. Dies are always preheated
sure requirements may significantly exceed the ture uniformity surveyed in much the same man- in the closed-die conventional forging of tita-
pure flow stress of any given alloy under similar ner as with heat treating furnaces. Continuous nium alloys, as noted in Table 27, with die tem-
deformation conditions. rotary furnaces used for titanium alloys typically perature varying with the type of forging equip-
Effect of Deformation Rate. Deformation have three zones: preheat, high heat, and dis- ment used. Dies for titanium alloy forging are
resistance of titanium alloys is very sensitive to charge. Most furnaces are equipped with record- usually preheated in remote die-heating systems,
strain rateconsiderably more so than alu- ing/controlling instruments, and in some batch although on-press equipment is sometimes used.
minum alloys or alloy steels. The strain-rate furnace operations separate load thermocouples Remote die-heating systems are usually gas-
sensitivity for representative alloys from each of are used to monitor furnace temperature during fired die heaters, which can slowly heat the die-
the three classes is shown in Fig. 19 for an a preheating operations. blocks to the desired temperature range before
alloy (Ti-8Al-1Mo-1V), an a-b alloy (Ti-6Al- In addition to highly controlled heating equip- assembly into the forging equipment.
4V), and a b alloy (Ti-10V-2Fe-3Al). For each ment and heating practices, the temperature of With some conventional forging processes,
of these alloys, as the deformation rate is reheated titanium alloy billets can be verified with particularly the hydraulic press forging of tita-
duced from 10/s to 0.001/s, the flow stress can contact pyrometry or noncontact optical pyrome- nium alloys, the temperature of the dies may in-
be reduced by up to ten times. For example, the ters. The latter equipment must be used with care crease during forging. Die damage may occur
flow stress for Ti-6Al-4V at 900 C (1650 F), because it is emissivity sensitive and may provide without appropriate cooling. Therefore, titanium
50% strain, and 10/s is 205 MPa (30 ksi); at different temperature indications when the metal alloy dies are often cooled during forging using
0.001/s, the flow stress is 50 MPa (7 ksi), a four- is observed inside the hot furnace versus when the wet steam, air, or occasionally water.
fold reduction. metal has been removed from the furnace. In most For those conventional forging processes in
From the known strain-rate sensitivity of tita- closed-die and open-die forging operations, it is which die temperatures tend to decrease, on-
nium alloys, it appears to be advantageous to de- desirable to have titanium alloy metal tempera- press heating systems ranging from rudi-
form these alloys at relatively slow strain rates in tures near the upper limit of the recommended mentary to highly sophisticated are used. The
order to reduce the resistance to deformation in temperature ranges. In open-die forging, the techniques used include gas-fired equipment,
forging; however, under the nonisothermal con- lower limit of the recommended ranges is usually induction-heating equipment, resistance-
ditions present in the conventional forging of the point at which forging must be discontinued heating equipment, or combinations of these
titanium alloys, the temperature losses encoun- to prevent excessive surface cracking. methods.
tered by such techniques far outweigh the bene- Heating Time. It is good practice to limit the Lubrication is also a critical element in the
fits of forging at slow strain rates. Therefore, in exposure of titanium alloys in preheating to conventional forging of titanium alloys and is
the conventional forging of titanium alloys with times just adequate to ensure that the center of the subject of engineering and process-develop-
relatively cool dies, intermediate strain rates are the forging stock has reached the desired tem- ment emphasis in terms of the lubricants used
typically employed as a compromise between perature in order to prevent excessive formation and the methods of application. With titanium
strain-rate sensitivity and metal temperature of scale and a case. Actual heating times will alloy conventional forging, a lubrication system
losses to obtain the optimal deformation possi- vary with the section thickness of the metal is used that includes ceramic precoats of forging
ble with a given alloy. In isothermal forging, being heated and with furnace capabilities. stock and forgings, die lubrication, and, for cer-
major reduction in resistance to deformation of Because of the relatively low thermal conductiv- tain forging processes, insulation.
titanium alloys can be achieved by slow-strain- ity of titanium alloys, necessary heating times Ceramic Glass Precoats. Most titanium
rate forging techniques under conditions where are extended in comparison to aluminum and alloy forging stock and forgings are precoated
metal temperature losses are minimized through alloy steels of equivalent thickness. Generally, with ceramic precoats prior to heating for forg-
dies heated to temperatures at or close to the 1.2 min/mm (30 min/in.) of ruling section is suf- ing. These ceramic precoats, which are formu-
metal temperature. ficient to ensure that titanium alloys have lated from metallic and transition-element ox-
With rapid-deformation-rate forging tech- reached the desired temperature. ides and other additives, provide several
niques, such as the use of hammers and/or me- Heating time at a specific temperature is crit- functions, such as:
chanical presses, deformation heating during the ical in titanium alloys for the reasons outlined
Protection of the reactive titanium metal
forging process becomes important. Because ti- previously. Long soaking times are not neces-
from excessive contact with gaseous ele-
tanium alloys have relatively poor coefficients of
ments present during heating
thermal conductivity, temperature nonunifor-
Table 27 Die temperature ranges for the Insulation or retardation of heat losses dur-
mity may result, giving rise to nonuniform de-
conventional forging of titanium alloys ing transfer from heating to forging equip-
formation behavior and/or excursions to temper-
ment
atures that are undesirable for the alloy and/or Die temperature
Lubrication during the forging process
final forging mechanical properties. As a result, Forging process/equipment C F
in the rapid-strain-rate forging of titanium al- Open-die forging The formulation of the ceramic precoat is varied
loys, metal temperatures are often adjusted to Ring rolling 150260 300500 with the demands of the forging process being
account for in-process heat-up, or the forging 95260 200500 used, the alloy, and the forging type. Modifi-
process (sequence of blows, and so on) is con- Closed-die forging cation of the ceramic precoat formulation usu-
trolled to minimize undesirable temperature in- Hammers 95260 200500 ally affects the melting or softening temperature,
creases, or both. Therefore, within the forging Upsetters 150260 300500 which ranges from 595 to 980 C (1100 to 1800
temperature ranges outlined in Table 26, metal Mechanical presses 150315 300600 F) for most commercially available precoats for
Screw presses 150315 300600
temperatures for optimal titanium alloy forging Orbital forging 150315 300600 titanium alloys. Experience has shown that ce-
conditions are based on the type of forging Spin forging 95315 200600 ramic precoats with a viscosity of 20 to 100
equipment to be used, the strain rate to be em- Roll forging 95260 200500 Pa s (200 to 1000 P) at operating temperature
ployed, and the design of the forging part. Hydraulic presses 315480 600900 provide optimal lubricity and desired continuous
film characteristics for protecting the metal dur- modification of die-cavity dimensions to ensure 6. M.F. Ashby, Materials Selection in Mechan-
ing heating and for preventing galling and metal the dimensional integrity of the finished forging. ical Design, 2nd ed., Butterworth-Heine-
pickup during forging. The actual formulations Insulative blankets are usually applied to the man, 1999
of ceramic precoats are often proprietary to the dies immediately before insertion of the hot 7. J. Schey, Manufacturing Processes and Their
forger or the precoat manufacturer. Ceramic pre- metal for forging. Selection, Materials Selection and Design,
coats are usually colloidal suspensions of the ce- Vol 20, ASM Handbook, 1997, p 687704
ramics in mineral spirits or water, with the latter 8. J.G. Bralla, Design for Excellence, McGraw-
being the most common. Finally, most conven- Summary Hill, 1996
tional titanium forging die design techniques in- 9. G. Boothroyd, P. Dewhurst, and W. Knight,
clude allowances for ceramic precoat thickness Engineering materials represent the func- Product Design for Manufacture and As-
in sinking the die cavity to ensure the dimen- tional connection between design and manufac- sembly, 2nd ed., Marcel Dekker, 2002
sional integrity of the final forging. turing, and the selection of a given manufactur- 10. H.E. Trucks, Designing for Economical
Ceramic precoats are applied using painting, ing process is very much dependent on the Production, 2nd ed., Society of Manufac-
dipping, or spraying techniques; state-of-the-art specific product design and the selected mate- turing Engineers, 1987
dipping and/or spraying processes are fully auto- rial. Selection and application of a manufactur- 11. R. Bakerjian, Ed., Design for Manufactura-
mated. Necessary ceramic precoat thicknesses ing method and material depend on many factors bility, Vol 6, 4th ed., Tool and Manufacturing
vary with the precoat and the specific forging during the various stages of conceptual, config- Engineers Handbook, Society of Manufac-
process, but generally fall in the range of 0.01 to uration, and detailed design. This is discussed in turing Engineers, 1992
0.1 mm (0.0005 to 0.005 in.). Most ceramic pre- ASM Handbook, Volume 20, Material Selection 12. J.A. Schey, Introduction to Manufacturing
coats require a curing process following applica- and Design. This chapter, which is adapted from Processes, 3rd ed., McGraw-Hill, 2000
tion to provide sufficient green strength for han- content of that book, briefly summarizes some 13. C.C. Gallagher and W.A. Knight, Group
dling. Curing procedures range from drying at of the general design factors that influence the Technology Production Methods in
room temperature to automated furnace curing at selection and application of metal bulk forming Manufacture, John Wiley & Sons, 1986
temperatures to approximately 150 C (300 F). operations, as discussed in more detail in subse- 14. I. Ham, Group Technology, Chapter 7.8,
Die lubricants are also used in the conven- quent chapters of this book. This chapter also Handbook of Industrial Engineering, G.
tional closed-die forging of titanium alloys. summarizes the forgeability characteristics of Salvendy, Ed., John Wiley & Sons, 1982
Such die lubricants are subject to severe de- major alloy systems and some key factors that 15. J.-Y. Jung and R.S. Ahluwalia , FORCOD:
mands and are formulated to modify the surface may influence forging practices, based on con- A Coding and Classification System for
of the dies to achieve the desired reduction in tent from ASM Handbook, Volume 14, Forming Formed Parts, J. Manuf. Syst., Vol 10 (No.
friction under conditions of very high metal tem- and Forging. 3), 1991, p 223232
peratures and die pressures and yet leave the 16. M. Tisza, Expert Systems for Metal
forging surfaces and forging geometry unaf- Forming, J. Mater. Process. Technol., Vol 53
ACKNOWLEDGMENT
fected. Die-lubricant formulations for titanium (No. 12), Aug 1995, p 423432
alloys are usually proprietary, developed either 17. G. Henzold, Handbook of Geometric
by the forger or the lubricant manufacturer. Die- Portions of this article were adapted from Toleranc-ing, John Wiley, 1995
lubricant composition is varied with the de- content in Forming and Forging, Vol 14, ASM 18. L. Edwards and M. Endean, Ed., Manufac-
mands of the specific forging process; however, Handbook (formerly 9th ed., Metals Handbook, turing with Materials, Butterworths, 1990
the major active element in titanium alloy die-lu- 1988) and the following articles in Materials 19. Data Card Index, Manufacturing with
bricants is graphite. In addition, other organic Selection and Design, Vol 20, ASM Handbook, Materials, The Open University, Milton
and inorganic compounds are added to achieve 1997: Keynes, 1990
the desired results because of the very high tem- G.E. Dieter, Relationship between 20. E.B. Magrab, Integrated Product and Proc-
peratures present. Carriers for titanium alloy die Materials Selection and Processing ess Design and Development, CRC Press,
lubricants vary from mineral spirits to mineral B.L. Ferguson, Design for Deformation 1997
oils to water. Processes 21. G.E. Dieter, Mechanical Metallurgy, 3rd
Titanium alloy die lubricants are typically ap- J.A. Schey, Manufacturing Processes and ed., McGraw-Hill, 1986
plied by spraying the lubricant onto the dies. Their Selection 22. M. Gensamer, Strength and Ductility, Trans.
Several pressurized-air or airless systems are em- H.W. Stoll, Introduction to Manufacturing ASM, Vol 36, 1945, p 3060
ployed, and with high-volume, highly automated and Design 23. P.W. Lee and H.A. Kuhn, Fracture in Cold
titanium alloy forging processes, die-lubricant ap- Upset ForgingA Criterion and Model,
plication is also automated by single or multiaxis Metall. Trans., Vol 4, 1973, p 969974
robots. Some state-of-the-art application systems REFERENCES 24. J.T. Winship, Fundamentals of Forging, Am.
can apply very precise patterns or amounts of lu- Mach., July 1978, p 99122
bricant under fully automated conditions. 1. Forming and Forging, Vol 14, ASM 25. D.W. Hutchinson, The Function and
Insulation. In the conventional forging of ti- Handbook (formerly 9th ed. Metals Proper Selection of Forging Lubricants,
tanium alloys in relatively slow-strain-rate Handbook), ASM International, 1988 Acheson Colloids Co., 1984
processes such as hydraulic-press forging, insu- 2. T.G. Byrer, Ed., Forging Handbook, 26. Open Die Forging Manual, 3rd ed., Forging
lative materials in the form of blankets are often Forging Industry Association and American Industry Association, 1982, p 106107
used to reduce metal-temperature losses to the Society for Metals, 1985 27. M.J. Donachie and S.J. Donachie,
much cooler dies during the initial deformation 3. K. Lange, Handbook of Metal Forming, Superalloys: A Technical Guide, ASM
stages. The insulative blankets are usually fabri- McGraw-Hill, 1985 International, 2002, p 9295
cated from fiberglass that is formulated to pro- 4. S. Kalpakjian, Manufacturing Processes for 28. H.J. Henning, A.M. Sabroff, and F.W.
vide the necessary insulative properties. Blanket Engineering Materials, 3rd ed., Addison- Boulger, A Study of Forging Variables,
thickness varies with specific materials of fabri- Wesley, 1996 Report ML-TDR-64-95, U.S. Air Force,
cation and desired insulative properties, but gen- 5. G.E. Dieter, Overview of Materials 1964
erally ranges from 0.25 to 1.3 mm (0.010 to Selection Process, Materials Selection and 29. W.D. Klopp, A-286 Datasheet (March 1987),
0.050 in.). If insulative blankets are used, al- Design, Vol 20, ASM Handbook, 1997, p Aerospace Structural Metals Handbook,
lowance is made in die-sinking tolerances for 246248 CINDAS/USAF Handbooks Operation and
Purdue Research Foundation, Code 1601, Handbook), ASM International, 1988, p Ackgerman, and H.J. Henning, Forging
1995, p 6 241254 Equipment Materials, and Practices, MCIC-
30. W.D. Klopp, N-155 Datasheet (June 1989), 33. H.H. Ruble and S.L. Semiatin, Forging of HB-03, Metals and Ceramics Information
Aerospace Structural Metals Handbook, Nickel-Base Alloys, Forming and Forging, Center, 1973
CINDAS/USAF Handbooks Operation and Vol 14, ASM Handbook (formerly 9th ed. 36. D.R. Muzyka, MiCon 78: Optimization of
Purdue Research Foundation, Code 1602, Metals Handbook), ASM International, Processing, Properties, and Service
1995, p 2 1988, p 261266 Perfor-mance Through Microstructural
31. Forging of Refractory Metals, Forming and 34. A.J. DeRidder and R. Koch, MiCon 78: Control, H. Abrams et al., Ed., American
Forging, Vol 14, ASM Handbook (formerly Optimization of Processing, Properties, and Society for Testing and Materials, 1979, p
9th ed. Metals Handbook), ASM Interna- Service Performance Through Microstruc- 526
tional, 1988, p 237238 tural Control, H. Abrams et al., Ed., Ameri- 37. G.W. Kuhlman, Forging of Titanium Alloys,
32. G.W. Kuhlman, Forging of Aluminum can Society for Testing and Materials, 1979, Forming and Forging, Vol 14, ASM Handbook
Alloys, Forming and Forging, Vol 14, ASM p 547 (formerly 9th ed. Metals Handbook), ASM
Handbook (formerly 9th ed. Metals 35. T. Altan, F.W. Boulger, J.R. Becker, N. International, 1988, p 267287
Chapter 12
Workability Theory and Application in

Bulk Forming Processes*
Howard A. Kuhn, Consultant
WORKABILITY, as described in previous specific process measurements for an experi- 1 1

chapters, is not merely a property of a material mental approach to workability evaluation. e 1 = l s 1 - s 2 - s 3
2 2
but a characteristic of the material/process sys- Theoretical models and fracture criteria are then
tem. The role of the material is measured by a described and compared with experimental re- 1 1
e 2 = l s 2 - s 3 - s1
simple test or tests and should be expressed in a sults. Finally, the application of workability con- 2 2
quantitative form that is applicable universally. cepts to forging, rolling, and extrusion processes 1 1
e 3 = l s 3 - s1 - s 2
This measure is taken to be a basic property of is discussed. 2 2 (Eq 1)
the material composition and structure, and it re-
flects the macroscopic outcome of the micro- where e denotes strain, denotes stress, and the
mechanisms of plastic flow perturbed by such Stress and Strain States subscripts 1, 2, and 3 designate the three direc-
inhomogeneities as voids, inclusions, and grain tions in an orthogonal coordinate system. Here,
boundaries. Such phenomena are dependent not Forging, extrusion, and rolling processes gen- l is a proportionality factor dependent on the
only on the material structure but also on the erally are considered to involve the application deformation history and flow stress curve of the
process parameters (strain rate and temperature), of compressive force to material to impart a material. The resulting strain in a given direction
which define the role of the process in determin- change in shape and dimensions. On close ex- is affected by the stress in all three coordinate
ing workability. amination, however, it is clear that deformation directions. In addition, these relationships sat-
The micromechanisms of ductile fracture in resulting from the applied load causes secondary isfy the volume constancy condition, e1 e2
bulk forming processes are strongly influenced stress and strain states that vary from point to e3 0.
by the stress and strain environment imposed by point throughout the deforming workpiece. To illustrate these relationships, it is useful to
the process. As shown in Fig. 10 of Chapter 2, These stress and strain states may include ten- consider a two-dimensional case of plane stress,
overall measure of strain to fracture can be re- sion, so fracture can occur at certain locations in s3 0:
lated to an overall value of hydrostatic stress in a material even though the primary (applied)
the material during processing. In the spectrum load is compressive.
1
of processes (extrusion, rolling, forging, and To explore this possibility further, it is useful e 1 = l s 1 - s 2
2
wire drawing), the average hydrostatic stress be- to review briefly the von Mises yield criterion,
comes increasingly tensile, and the strain to which is the foundation of plasticity theory for 1
e 2 = l s 2 - s 1
fracture progressively decreases. Within each of isotropic materials. It gives the relationships be- 2 (Eq 2)
these processes, however, the stress and strain tween normal and shear stress components at
states can be considered on a localized basis in yielding. From this, relationships between stress Referring to Fig. 1 and Eq 2, a reduction in
the specific regions in which fractures initiate. and strain components are derived: thickness (compressive strain in the 2-direction)
These localized conditions are controlled by the
geometry of the workpiece, die design, and fric-
tion at the die/workpiece interface. These three
factors, in addition to the strain rate and temper-
ature parameters already mentioned, embody the
role of the process in determining workability.
This article focuses on the effects of mechan-
ical plasticity on workability; that is, process
control of localized stress and strain conditions
to enhance workability. First, the nature of local
stress and strain states in bulk forming processes
is described, leading to a classification scheme
that facilitates the application of workability
concepts. This defines testing procedures and Fig. 1 Similarity of deformation under (a) horizontal tension and (b) vertical compression
*This article is revised from Howard A. Kuhn, Workability Theory and Application in Bulk Forming Processes, Forming and Forging, Vol 14, ASM Handbook, ASM International, 1988, p 388404.
Chapter 12: Workability Theory and Application in Bulk Forming Processes / 173
can be accomplished either by a compressive creases, and the localized axial compressive strains e1 and e 2 in the plane of the surface, as in
stress, P2 (or s2), or by a tensile stress, s1. strain e 2 (0) decreases. From Eq 3 as e1/e 2 be- Fig. 2 to 4. In Fig. 5, however, this surface is also
Although the nature of the deformation is the comes more negative (12), s1 becomes more acted upon by pressure P3 normal to the plane.
same in both cases, accomplishing that deforma- positive. Therefore, increasing bulging due to Similarly, at internal locations of the workpieces
tion in the latter case could risk fracture because friction during the compression of a cylinder in- in such processes as forging or wire drawing
it involves a tensile stress. creases the secondary tensile stress s1 and en- (Fig. 6), a material element of the central longi-
For more general cases, Eq 2 can be re- hances the likelihood of fracture. tudinal plane is subjected to strains e1 and e 2 and
arranged to determine the following stresses: Similarly, at the edges of bars during rolling stress normal to the plane.
(Fig. 3), the elongation strain e1 is determined In Fig. 5 and 6, the material element can be
4 1 4 1 e2 by the overall reduction in area. The localized thought of as the plane-stress elements in Fig. 2
s1 = e1 + e 2 = e 1 1 + compressive vertical strain e 2, however, depends to 4 with s3 acting normal to the plane. If e1 and
3l 2 3l 2 e 1
on the shape of the edge. Greater convexity and e 2 are taken to be the strains in this plane, then
4 1 4 1 e2 sharpness of the edge decrease the compressive Eq 3 becomes:
s2 = e 2 + e1 = e1 +
3l 2 3l 2 e 1 vertical strain for a given reduction, which, from
the first part of Eq 3, increases the secondary 4 1
s3 = 0 (Eq 3) tensile stress s1 at the edge. Therefore, edge s1 = e1 + e 2 + s 3
3l 2
cracking during rolling is also due to secondary
Therefore, the stresses depend on the localized stress states. 4 1
s2 = e 2 + e1 + s 3
strains e1 and e2 that result from metal flow. More complex cases of the same type of sec- 3l 2
Equation 3 is a more convenient representation ondary tensile stress states occur in forging (Fig. s3 = s3
of plasticity relationships for workability study. 4). During the forging of a hub shape, for exam-
For example, the cylindrical surface of a com- ple, the top surface of the hub is subjected to bi-
pression test undergoes various combinations of axial tension because of friction during flow In other words, for the same deformation (that is,
axial and circumferential strains, depending on around the die radius (Fig. 4a). This is identical the same values of e1 and e 2 as in Fig. 2 to 4), the
the aspect ratio and the friction at the die contact to the conditions present at the nose of a billet stresses s1 and s2 in Eq 3 are biased by s3. The
surfaces (Fig. 2). When no friction exists, the that is being extruded or rolled. Similarly, during stress normal to the surface, then, increases the
ratio of circumferential strain to axial strain is forging, the top surface of a rib undergoes hydrostatic stress component of the stress state
e1/e 2 12. According to the first part of Eq 3, tensile strain in the direction of curvature, and by s3. This reflects the basic concept that hydro-
s1 0 for this case. The deformation in this essentially no strain occurs along the length di- static stress does not affect yielding or plastic de-
case is referred to as homogeneous compression, rection of the rib (Fig. 4b). In both cases, local- formation. As shown in Fig. 10 of Chapter 2,
because the only stress acting is s2 and it is uni- ized secondary tensile stresses are generated that Bulk Working Behavior of Metals, however,
form throughout the specimen. Therefore, the may cause fracture. These stresses can be calcu- the hydrostatic stress has a significant effect on
homogeneous compression test is suitable for lated from measured strain values using Eq 3. fracture. In Fig. 5, the compressive stress (die
measuring flow stress. Figures 2 to 4 show examples of plane stress, pressure) P3 would increase the strains at frac-
When friction exists at the die contact sur- with the stress normal to the free surface being ture, but in Fig. 6 the internal stress s3 may be
faces, material at these surfaces is constrained zero. Other regions of workpieces in bulk defor- tensile and decrease the strains at fracture.
from moving outward, while material at the mid- mation processes, however, are subjected to The preceding discussion of stress and strain
plane is not constrained. As a result, bulging three-dimensional stress states. For example, states in various regions of workpieces suggests
occurs, as shown in Fig. 2(a). Under these con- material at the die contact surfaces in forging, a convenient method of classifying cracking de-
ditions, the circumferential strain e1 (0) in- extrusion, and rolling (Fig. 5) is subjected to fects. Figures 2 to 4 in this article illustrate the
locations at which free surface cracks occur in
forging and rolling, as well as at the nose ends of
billets in extrusion or rolling. Figure 5 in this ar-
ticle shows examples of circumstances in which
contact surface cracking occursfor example,
the fir tree defect in extrusion or the longitudinal
surface cracks in rolled plates.
Empirical Criterion of Fracture

The stress and strain environments described
in the previous section in this article suggest
that a workability test should be capable of sub-
jecting the material to a variety of surface strain
combinations. A capability for testing under
superimposed normal stress would also be
desirable.
When considering workability tests, it is im-
portant to recognize that fractures initiate in lo-
calized regions where interaction between the
stress and strain states and the material structure
reaches a critical level. Orientation, shape, and
volume fraction of inclusions and other inhomo-
geneities have a dominant effect on the fracture
Fig. 2 Localized strains on (a) the bulging cylindrical surface of an upset test and (b) their variation with aspect ratio process. Therefore, it is critically important that
and friction conditions. Source: Ref 1 workability test specimens contain material hav-
Fig. 3 Localized strains at the edges of bars during (a) rolling and (b) their variation with edge profile. Source: Ref 2
Fig. 4 Strains at the free surfaces of forgings. (a) Axisymmetric hub. (b) Rib-web forging
Fig. 6 An element of material at the center of a forging undergoing (a) double extrusion and (b) at the center of
wire being drawn. The element undergoes strains e1, and e 2, as in Fig. 2 to 4, with stress s3 normal to the
12 plane.
Fig. 5 An element of material at the die contact sur-

faces during (a) forging, (b) drawing or extru-
sion, and (c) rolling
ing the same microstructural features as the ma-

terial in the localized, potential fracture regions
of the actual process.
Specifically, when evaluating a workpiece for
surface fractures, specimen surfaces must con-
tain the as-received surface of the workpiece
Fig. 7 Compression tests on 2024-T35 aluminum
under consideration because it may contain laps, alloy. Left to right: undeformed specimen, com-
seams, a decarburized layer, and so on, which af- pression with friction (cracked), compression without
fect fracture initiation. By the same argument, friction (no cracks)
Fig. 8 (a) Flanged and (b) tapered prebulged compres-
evaluation of material for internal fractures such sion test specimens. Lateral spread of interior
as central burst must involve test specimens material under compression expands the rim circumferen-
tially while little axial compression is applied (see Fig. 9).
taken from the middle of the workpiece, where, friction and geometry conditions. The influence
for example, segregation of second phases may of friction and consequent bulging on circumfer-
have occurred. Because of possible anisotropy ential tensile stress development is clearly
effects, orientation of the critical stresses with shown in Fig. 7. Compression with friction pro- strain axis, e1 (Ref 3). Test specimens are artifi-
respect to any inclusion alignment must be the duces circumferential tension that leads to frac- cially prebulged by machining a taper or a
same in the test specimens as it is in the actual ture, while frictionless compression prevents flange on the cylinders (Fig. 8). Compression
process and material of interest. barreling, tension, and cracking as described in then causes lateral spread of the interior mate-
The compression test has become a standard Fig. 2 and Eq 3. rial, which expands the rim circumferentially
for workability evaluation. As shown in Fig. 2, a Alterations of the compression test geometry while applying little axial compression to the
range of strain combinations can be developed at have been devised to extend the range of surface rim. Therefore, the tapered and flanged upset
the cylindrical free surface simply by altering strains available toward the vertical, tensile test specimens provide strain states consisting of
small compressive strain components. Each ture, and 2024-T4 alloy at a hot-working tem- The strain paths at potential fracture sites in
combination of height, h, and thickness, t, gives perature. In some cases, the fracture strains fit a material undergoing deformation processing
a different ratio of tensile to compressive strain. straight line of slope 12; in others, the data fit a (determined by measurement or mathematical
The strain states developed at the surfaces of dual-slope line with slope 12 over most of the analysis) can then be compared to the fracture
straight, tapered, and flanged compression test range and slope 1 near the tensile strain axis. strain loci. Such strains can be altered by process
specimens are summarized in Fig. 9. Similar data have been obtained for a wide vari- parameter adjustment, and they represent the
The variety of strain combinations available ety of materials. In each case, the straight line process input to workability. If the process
in compression tests offers the possibility for behavior (single or dual slope) appears to be strains exceed the fracture limit lines of the ma-
material testing over most of the strain com- characteristic of all materials, but the height of terial of interest, fracture is likely. Other ap-
binations that occur in actual metalworking the line varies with the material, its microstruc- proaches to establishing fracture criteria, as well
processes. A number of samples of the same mature, test temperature, and strain rate. as applications of the criteria, are given in the
terial and condition are tested, each one under The nature of the fracture loci shown in Fig. following two sections of this article, Theo-
different friction and geometry parameters. Tests 10 to 12 suggests an empirical fracture criterion retical Fracture Models and Criteria and
are carried out until fracture is observed, and the representing the material aspect of workability. Applications.
local axial and circumferential strains are meas-
ured at fracture. Figures 10 to 12 give some ex-
amples of results for AISI 1045 carbon steel,
2024-T351 aluminum alloy at room tempera-
Fig. 9 Range of free surface strain combinations for

compression tests having cylindrical (Fig. 2),
tapered, and flanged (Fig. 8) edge profiles. The ranges
shown are approximate and they may overlap a small
amount. Fig. 11 Fracture locus for aluminum alloy 2024-T351 at room temperature
Fig. 12 Fracture locus for aluminum alloy 2024-T4 at

room temperature and at 300 C (570 F).
Fig. 10 Fracture locus for AISI 1045 cold-drawn steel e 0.1 s1
Theoretical Fracture Models Void Growth Model. Microscopic observa- which localized instability cannot occur, has
and Criteria tions of void growth and coalescence along prompted consideration of the effects of inho-
planes of maximum shear leading to fracture mogeneities in the material. For example, a
have led to the development of a model of hole model of localized thinning due to a small inho-
Fracture criteria for metalworking processes growth (Ref 4). Plasticity mechanics is applied mogeneity has been devised (Ref 5). Beginning
have been developed from a number of view- to the analysis of deformation of holes within a with the model depicted in Fig. 15, plasticity
points, as described in Chapters l and 2. The shear band. When the elongated holes come into mechanics is applied to determine the rate of
most obvious approach involves modeling of the contact, fracture is considered to have occurred thinning of the constricted region tB in relation
void coalescence phenomenon normally associ- (Fig. 13). When the McClintock model is evalu- to that of the thicker surrounding material tA.
ated with ductile fracture. Another approach in- ated for a range of applied stress combinations, When the rate of thinning reaches a critical
volves a model of localized thinning of sheet a fracture strain line can be constructed. value, the limiting strains are considered to have
metal that has been adapted to bulk forming Figure 14 shows the calculated results from been reached, and a forming limit diagram can
processes. In addition to models of fracture, cri- the McClintock model in comparison with the be constructed. The analysis includes the effects
teria have been developed from macroscopic experimental fracture line. The predicted frac- of crystallographic anisotropy, work-hardening
concepts of fracture. The Cockcroft criterion is ture strain line has a slope of 12 over most of its rate, and inhomogeneity size.
based on the observation that both tensile stress length, matching that of the experimental frac- This model was applied to free surface frac-
and plastic deformation are necessary ingredi- ture line. Near the tensile strain axis, the slope of ture in bulk forming processes because of evi-
ents, which lead to a tensile deformation energy the predicted line is 1, matching that of actual dence that localized instability and thinning also
condition for fracture. The upper bound method material results shown in Fig. 10 and 12. precede this type of ductile fracture (Ref 6). Two
has been used to predict fracture in extrusion Localized Thinning Model. In sheet form- model geometries were considered, one having a
and drawing. Other approaches are based on the ing, the observation that a neck forms before groove in the axial direction (Z-model) and the
calculation of tensile stress by slip-line fields. fracture, even under biaxial stress conditions in other having a groove in the radial direction (R-
model) as shown in Fig. 15. Applying plasticity
mechanics to each model, fracture is considered
to have occurred when the thin region, tB, re-
duces to zero thickness. When these fracture
strains are plotted for different applied stress ra-
tios, a fracture strain line can be constructed. As
shown in Fig. 16, the predicted fracture line
matches the essential features of the experimen-
tal fracture lines. The slope is 12 over most of
the strain range and approximately 1 near the
tensile strain axis. Again, this model is in gen-
eral agreement with the dual-slope fracture loci
shown in Fig. 10 and 12.
Cockcroft Model. The Cockcroft criterion of
fracture is not based on a micromechanical
model of fracture but simply recognizes the
macroscopic roles of tensile stress and plastic
deformation (Ref 7). It is suggested that fracture
occurs when the tensile strain energy reaches a
critical value:
Fig. 13 McClintock model of void coalescence by shear from (a) initial circular voids, through (b) growth, and (c) ef
void contact s*de = C
0
Fig. 14 Fracture strain locus predicted by the Fig. 15 Models for the analysis of localized thinning Fig. 16 Fracture strain locus predicted by the model of
McClintock model of void growth. The and fracture at a free surface. (a) R-model. (b) localized thinning. The shaded area repre-
shaded area represents typical experimental fracture loci Z-model. Source: Ref 6 sents typical experimental fracture loci.
such as Fig. 10 to 12.
where s* is the maximum tensile stress; e is the lower drawing force. In the range of die angles rate hardening characteristics of the material
equivalent strain; and C is a constant determined between 1 and 2, central burst is energetically have been included in the analysis (Ref 10), but
experimentally for a given material, tempera- more favorable (Ref 9). the microstructural characteristics have been
ture, and strain rate. The criterion has been suc- Repeated calculations using the upper bound omitted. Therefore, when operating in the cen-
cessfully applied to cold-working processes. It method provide the combinations of die angle tral burst range illustrated in Fig. 19, fracture
has also been reformulated to provide a pre- and reduction that cause central burst (Fig. 19). can occur; whether or not it will depends on the
dicted fracture line for comparison with the ex- Friction at the die contact surface affects the re- material structure (voids, inclusions, segrega-
perimental fracture strain line. Figure 17 shows sults. If operating conditions are in the central tion, and so on). When operating in the safe area
that the fracture strain line predicted by the burst region, defects can be avoided by decreas- shown in Fig. 19, central burst will not occur, re-
Cockcroft criterion is also in reasonable agree- ing the die angle and/or increasing the reduction gardless of the material structure.
ment with experimental results. This criterion so that the operating conditions are in the safe Tensile Stress Criterion. The role of tensile
does not show the dual-slope behavior of the zone. This example is a clear illustration of the stress in fracture is implicit yet overwhelmingly
previous models and some actual materials. The role of process parameters (in this case, geomet- clear throughout the discussion of fracture and
question of why some materials exhibit the dual- ric conditions) in workability. An application of fracture criteria. The calculation of tensile
slope fracture locus and others only a single this method is given in the section Appli- stresses in localized regions, however, requires
slope remains the subject of speculation, discus- cations in this article. the use of advanced plasticity analysis methods
sion, and further study. It should be pointed out that the upper bound such as slip-line fields or finite-element analysis.
The upper bound method of plasticity analy- method for defect prediction gives only a neces- One result of slip-line field analysis that has
sis requires the input of a flow field in mathe- sary condition. The strain-hardening and strain wide application in workability studies is dis-
matical function form. The external work cussed here.
required to produce this flow field is determined Double indentation by flat punches is a classi-
through extensive calculation. This value for ex- cal problem in slip-line field analysis (Fig. 20).
ternal work is an upper bound on the actual work
required. Through optimization procedures, the
flow field can be found that minimizes the cal-
culated external work done, and this flow field is
closest to the actual metal flow in the process
under analysis. The upper bound method has
been applied to a number of metalworking
processes (Ref 8).
In consideration of metal flow fields, pertur-
bations can be incorporated that simulate defect
formation. In some situations, the external work
required to create the flow field with defects is
less than the work required to create the flow
field without defects (sound flow). According to
the upper bound concept, defects would occur in
these situations.
The upper bound method has been applied
to the prediction of conditions for central burst
formation in extrusion and wire drawing. As
shown in Fig. 18, for die angles less than 1,
sound flow requires less drawing force than flow Fig. 18 Variation in mode of flow with die angle in Fig. 19 Upper bound prediction of central burst in
wire drawing. The mode requiring the small- wire drawing. Increasing friction, expressed
with central burst formation. Above 2, extru- est force at any die angle is the active mode. This is a by the friction factor, m, increases the defect region of the
sion with a dead metal zone near the die requires schematic for one value of reduction. Source: Ref 9 map. Source: Ref 10
Fig. 17 Fracture strain locus predicted by the

Cockcroft criterion Fig. 20 Slip-line fields for double indentation at different h/b ratios. Source: Ref 11
The boundaries of the deformation zone change processesfor example, those in Fig. 21. In a For example, in extrusion, h/b is approxi-
as the aspect ratio h/b (workpiece thickness-to- very detailed study of double indentation, re- mated by:
punch width) increases. For h/b 1, the slip- markable agreement was found between experi-
line field meets the centerline at a point, and for mental results and slip-line field results (Ref 12).
h/b 1, the field is spread over an area nearly as For workability studies, however, it is neces-
h/b=
[
a 1 + (1 R) ]
1/ 2 2
large as the punch width. sary to locate and to calculate the critical tensile R
This tooling arrangement and deformation stresses. It can be shown that the hydrostatic
geometry approximates several other metal- stress is always greatest algebraically at the cen- where is the die half-angle and R is area re-
working processes, as shown in Fig. 21. For sim- terline of the material and that this stress is ten- duction. Taking h/b 1.8, the relationship be-
ilar h/b ratios in these processes then, the stresses sile for h/b 1.8. The calculated results for tween and R that produces tensile hydrostatic
throughout the deformation zone of the process punch pressure and centerline hydrostatic stress stress at the centerline can be calculated. The re-
can be approximated by those calculated from are given in Fig. 22. Therefore, it is necessary to sult is given in Fig. 23, which is shown to be
slip-line field analysis for double indentation. specify die and workpiece geometric parameters similar to the relationship predicted by upper
Most of the work on double indentation has such that h/b 1.8 in order to avoid tensile bound analysis (Fig. 19). The correlation is re-
focused on the calculation of punch pressure and stress and potential fracture at the centerline of markable in view of the dissimilarity in die
the extrapolation of these results to other the processes shown in Fig. 21. shape between extrusion and double indentation.
Furthermore, the similarity emphasizes that the
flow mode for defect formation in the upper
bound method is physically equivalent to the de-
velopment of tensile hydrostatic stress.
As in the case of the upper bound method, the
existence of a tensile hydrostatic stress does not
ensure fracture, but it is a necessary ingredient.
The material structure must be considered in
conjunction with the tensile stress. In other
words, the upper bound and tensile stress crite-
ria are useful for defining approximate deforma-
tion limits and successful process parameters,
but more detailed criteria or models are required
to provide more exact values. An experimental-
analytical approach in which an experimental
value reflecting the inherent material ductility is
determined first would be most useful (this value
is used to define a point on the fracture strain
locus), followed by development of the rest of
the fracture-limit line, as in Fig. 10 to 12, 14, 16,
Fig. 21 Slip-line fields for (a) rolling, (b) drawing, and (c) side pressing. These fields are similar to those for dou- and 17.
ble indentation shown in Fig. 20.
Applications
The fracture criteria discussed previously in

this article can be used as tools for troubleshoot-
ing fracture problems in existing processes or
for designing/modifying processes for new
products. In either case, graphical representation
of the criteria permits independent consideration
of the process and material parameters in quan-
titative or qualitative form.
An example is the bolt-heading process
shown in Fig. 24(a). If it is required to form a
bolt head diameter D from the rod of diameter d,
the required circumferential strain is ln (D/d ),
indicated by the horizontal dashed line in Fig.
24(b). The strain paths that reach this level,
however, depend on process parameters, as
shown previously in Fig. 2(b), and the fracture
strain loci vary with material, as shown in Fig.
10 to 12. Referring to Fig. 24(b), if the strain
path labeled a describes the strain state at the ex-
panding free surface for one set of processing
conditions and the material used has a forming
limit line labeled A, then, in order to reach the
Fig. 22 Variation of the normalized indentation pres- Fig. 23 Prediction of central burst in wire drawing by required circumferential strain, the strain path
sure (P/Y where Y is the yield strength) and the the tensile stress criterion and slip-line field
normalized centerline hydrostatic stress (h/Y ) with h /b analysis of double indentation. The range of predictions must cross the fracture line, and fracture is likely
ratio as calculated from slip-line field analysis by upper bound analysis (Fig. 19) is shown by dashed to occur. As shown, one option for avoiding de-
lines.
fects is to use material B, which has a higher cause it presents a graphical and quantitative de- Such a study is illustrated by the analysis of
fracture limit. Another option is to alter the scription of the relationship between material cracking during the rolling of 2024-T351 alu-
process so that strain path b is followed by the and process parameters. Some examples of the minum alloy bars. The intent was to roll square
material. The latter option represents a process application of the workability analysis proce- bars into round wire without resolutioning.
change, which in this case involves improved lu- dures already described follow. Rolling was done on a two-high reversible bar
brication, as shown in Fig. 2. This procedure has Bar Rolling. As shown in Fig. 3, the strains at mill with 230 mm (9 in.) diam rolls at 30 rpm
been quantified and implemented in a computer- the edges of bars during rolling are similar to (approximate strain rate: 4s1). The roll groove
ized tool for upsetting process design (Ref 13). those at the bulging free surface of a cylinder dur- geometry is shown in Fig. 25. Defects occurred
In more complex cases, other means are ing compression. It should be possible, then, to primarily in the square-to-diamond passes (12
available for altering the strain path, such as predict fracture in bar rolling from compression and 34), but the two diamond-to-square passes
modification of die design, workpiece (preform) tests on the alloy of interest. This is pertinent in (23 and 45), the square-to-oval pass (56),
design, and redistribution of lubricant. Examples current attempts to roll ingots of high alloy con- and the oval-to-round pass (67) also were ex-
of application to powder forging preform design tent into bar form. The complete workability amined for completeness.
and other metalworking processes are given in study of bar rolling includes physical modeling of Lead was used as the simulation material for
Ref 14 and Ref 15. bar rolling to obtain the strain states at the edges the physical modeling of bar rolling. Pure
The workability concept presented earlier in of the bar, compression tests to obtain the mate- (99.99%) lead was cast and extruded into 25 mm
this article provides a useful supplement to the rial fracture limits, and comparison of the two sets (1 in.) round bars and then squared in the box
experience and intuition of the die designer, be- of results to establish roll pass reduction limits. pass (step 1, Fig. 25). Grids were placed on the
lateral edges of the bars by an impression tool,
and the grid spacing was measured before and
after each pass for calculation of the longitudi-
nal, e1, and vertical, e 2, strains. Different reduc-
tions in area were achieved by feeding various
bar sizes and by changing roll separation dis-
tances. A transverse slice was cut from the bars
after each pass for measurement of the cross-
sectional area and calculation of the reduction.
Results of the strain measurements are sum-
marized in Fig. 26, in which tensile strain is
plotted simultaneously with the compressive
strain and reduction. As expected, the square-to-
diamond passes involve the least compressive
vertical strain, and the square-to-oval pass has
the greatest compressive strain. The tensile
strain versus reduction plot is the same for all
cases, reflecting volume constancy.
Compression tests were performed on the
2024 aluminum alloy at room temperature and at
250 C (480 F) at a strain rate of 4s1 to deter-
mine fracture limit lines. Straight, tapered, and
flanged specimen profiles were used. Results are
given in Fig. 27. Superposition of Fig. 26 onto
Fig. 27 gives the rolling deformation limits.
To test the workability predictions, aluminum
alloy bars were rolled at room temperature and
at 250 C (480 F). Grid and area reduction
measurements were made for the square-to-
Fig. 24 Upsetting (a) of bar diameter d to head diameter D. (b) Material fracture strain limits are superimposed on
diamond passes. Figure 28 shows the measured
strain paths reaching the final required strain. Strain path b (low friction) prevents fracture for both materi-
als. Material B avoids fracture for either strain path. strains at room temperature, which agree with
Fig. 25 Roll groove geometry for rolling square bars into round wire. Dimensions given in millimeters
Similarly, in the hot rolling of this aluminum

alloy, the reduction limit for diamond-to-square
passes would be approximately 27%; for oval-
to-round passes, approximately 30%; and for
square-to-oval passes, approximately 40%. The
latter two are beyond the reduction normally
used because of fin formation, so cracking oc-
curs rarely in such passes.
Example 1: Preform Design for a Ball
Bearing Race. A low-load high-torque ball
bearing outer race was cold forged from a low-
alloy steel powder preform (Fig. 30). The pre-
forms were compacted from 4600 grade powder
with carbon added to give 0.20% C in the sin-
tered material. The sintered preforms were 80%
of theoretical density.
Initial efforts led to cracking through the pre-
form along a diagonal beginning at the point of
contact between the punch and preform (Fig.
31). The large shear stress developed by the con-
tact was beyond the fracture limit of the porous
preform, and two solutions were considered that
would avoid such stresses (Fig. 32). The first so-
lution was the use of a flat preform that involves
Fig. 26 Measured localized strains during the rolling of lead bars. Left side shows longitudinal tensile strain versus back extrusion flow into the outer rim, and the
vertical compressive strain. Right side shows longitudinal strain versus cross-sectional area reduction at second was the use of a tall, thin-wall preform
room temperature.
that involves radial inward flow into the inner
flange. The first option was rejected because it
would generate circumferential tension that
would most likely cause fracture. The second
option is desirable because compression is ap-
plied at the top face; this option was pursued
through physical modeling.
The primary concern with the second option
(Fig. 32b) was the large amount of radially in-
ward deformation required to form the inner
flange. As a result, this option was examined by
physical modeling. Model preforms were pro-
duced from sintered 601AB aluminum alloy
powder and gridded on the inside surface (Fig.
33a). Grid displacements were measured after
each of several increments of deformation, and
the calculated strains were plotted along with the
fracture line of the material (Fig. 33b). It is clear
that both the axial and circumferential strains are
compressive throughout the process and do not
exceed the fracture line. Some wrinkling of the
inside surface occurred, but this was smoothed
out when the surface contacted the mandrel
under pressure.
Actual production of straight-wall preforms
as in Fig. 33(a) was not feasible, because the
height-to-thickness ratio is too large for com-
paction. A compromise was developed in which
the preform angle was 17 instead of 30, as
Fig. 27 Fracture strain lines for 2024 aluminum alloy in the T351 temper, measured by compression tests at room
used in the original preform, or 0, as used in the
temperature and at 250 C (480 F)
physical model (Fig. 33b). This ensured initial
punch contact at the top face of the preform and
those measured in lead bars for the same pass C (480 F), there was conformance between generated compressive strains on the inside sur-
(Fig. 26). Open circles indicate fracture, and fracture in bar rolling (Fig. 29) and the fracture face, as in Fig. 33(b). Cold-forging trials on
closed circles indicate no fracture. The fracture line of the alloy (Fig. 28). In this case, the limit- these preforms produced no cracks, developed
line for the aluminum alloy at room temperature ing reduction is approximately 25%. the desired full density in the ball path region,
is superimposed as the dashed line. It is clear Extrapolating the preceding results for cold and showed the added benefit of a smooth ball
that edge cracking in bar rolling conformed with rolling, the limiting reduction for diamond-to- path surface that did not require grinding.
the material fracture line, and the limiting reduc- square passes would be approximately 15%; for Example 2: Back Extrusion of Copper
tion is approximately 13% for this combination the oval-to-round pass, approximately 20%; and Alloy. A low-ductility dispersion-strengthened
of material and pass geometry. Similarly, at 250 for the square-to-oval pass, approximately 25%. copper alloy was back extruded into a cup shape,
as shown in Fig. 34. The deformation was car- or 0.75 in.), as shown in Fig. 35(a). Deformation strain is ln (0.75/0.625) 0.18. Workability
ried out at room temperature on a mechanical of such preforms involves circumferential ex- analysis would then require only measurement
press. Crack formation on the rim caused high pansion strain (equal to ln (D/d ), where D is the of the material fracture line and comparison with
rejection rates. The original slugs for this die bore diameter and d is the slug diameter) the required circumferential strain 0.18.
process were smaller in diameter (16 mm, or along with very little compressive strain at the Fracture strains were measured in flange com-
0.625 in.) than the die inside diameter (19 mm, rim (Fig. 35b). For this case, the circumferential pression tests as shown in Fig. 36, giving a min-
imum circumferential strain of 0.2, which is suf-
ficiently above the required strain for avoidance
of fracture. A hydraulic press was used (giving a
strain rate of approximately 0.5 s1), on the as-
sumption that there is no strain rate effect at
room temperature. Because the workability
analysis showed that fracture should not be a
Fig. 28 Superposition of fracture line (dashed) on measured strains during rolling of 2024-T351 aluminum alloy bars Fig. 30 Ball bearing outer race that was cold forged
at room temperature. Solid line represents the strain path measured during rolling of the lead model mate- from sintered powder preform of 4620 low-
rial shown in Fig. 26. alloy steel
Fig. 31 Cracks initiated at the point of contact be-

Fig. 29 Superposition of fracture line (dashed) on measured strains during the rolling of 2024-T351 aluminum tween the punch and preform in the original
alloy bars at 250 C (480 F). Solid line represents the strain path measured during rolling of the lead preform design. The preform had a taper of 30 on the
model material shown in Fig. 26. inside diameter.
problem, the effect of strain rate was explored mechanical press. Figure 37 shows the surpris- on the production press. The problem was
further. ing result that the fracture limit line decreases corrected by using slugs of larger diameter to
Tests were performed on the same alloy using with increasing strain rate. In particular, the min- decrease the circumferential tension and by pre-
controlled strain rate servohydraulic test equip- imum circumferential strain falls below the re- forming a taper on the top face (Fig. 38), which
ment at strain rates of 5, 10, and 15 s1; the third quired value of 0.18 for successful forming of produced some axial compression in the mate-
strain rate given is close to that in the production the rim; this explains the occurrence of fracture rial at the rim. The strain path then avoided
crossing the fracture line (Fig. 39), and the re-
jection rate during production on the mechanical
press was nil.
Contact Surface Fracture and Internal
Fracture. All of the previous applications and
examples involved free surface fractures and
could be treated directly by the fracture line.
Consideration of contact surface fractures (Fig.
5) and internal fracture (Fig. 6), however, re-
quires modification of this approach or use of a
new approach. In the following, an example is
given of the application of the upper bound and
tensile stress criteria to central burst in extru-
sion. The empirical workability concept de-
scribed previously is then modified for applica-
tion to contact surface fracture as well as central
burst.
Example 3: Central Burst During Extru-
sion. Central burst can occur in extrusion when
light reductions and large die angles are used
(Fig. 19), and it is encountered in the production
of shafts for transmissions and suspension sys-
tems. A test of the central burst criterion was
carried out by processing shafts from hot-rolled
1024 steel bars 22 mm (78 in.) in diameter (Ref
Fig. 32 Preform alternatives for forging the ball bearing outer race shown in Fig. 30. (a) Back extrusion. (b) 16). The processing sequence consisted of initial
Compression and radial inward flow
drawing followed by three extrusion steps in a
boltmaker:
Process Reduction, % Die half-angle, degrees
Drawing 8 9
Extrusion 22 22.5
Extrusion 23 22.5
Extrusion 16 22.5 or 5
All passes are in the central burst area of Fig. 40,

except for the last pass with a 5 die angle.
A total of 1000 shafts were processed with the
22.5 die, and 500 shafts were processed with
the 5 die. All shafts were tested ultrasonically
for internal defects. Central bursting was de-
Fig. 33 (a) Physical model for the second option (Fig. 32b) and (b) the measured strains during forging of the pre-
form. The heavy line is the material fracture line. It is clear that the strain path never crosses the fracture
line and that defects are prevented. Fig. 34 Part that was back extruded from copper
Fig. 36 Fracture strain line for copper alloy, deter-

mined by flanged and tapered compression
tests. Specimen geometries used for each test are shown
also.
monitored is subjected to stress normal to that

surface. It was shown in Eq 3 that stress states
leading to a given set of surface strains differ
only by a hydrostatic stress component, and this
Fig. 35 Back extrusion of the cup shape shown in Fig. 34. (a) The preform slug was 16 mm (0.625 in.) in diameter, component is equal to the applied stress normal
the die was 19 mm (0.75 in.) in diameter. (b) Strains at the cup rim where fracture occurred consist of cir-
cumferential tension to a value of 0.18 and very little compressive strain. The heavy line is the material fracture strain
to the surface on which the strains are moni-
line. tored. Experience shows that this hydrostatic
stress affects fracture, and it should also affect
tected in 4.5% of the shafts extruded with the fracture that are controlled by stress and defor- the fracture strain locus. It should be possible,
22.5 die, and no defects were detected in the mation. The experimental fracture locus is also then, to use the theoretical fracture criteria to
shafts extruded with the 5 die. These results reproduced by several theoretical fracture crite- predict the effects of hydrostatic stress on the
show that the upper bound central burst criterion ria (Fig. 14, 16, and 17). fracture strain locus.
is a necessary condition. It was further shown in For contact surface and internal fractures, The simplest criterion described previously in
Ref 16 that central burst was avoided in other however, the surface on which the strains can be this article is that due to Cockcroft; therefore, it
heats with slightly different compositions be- was modified to predict the effects of stress nor-
cause their strain-hardening coefficients were mal to the plane (Fig. 5 and 6) on the fracture
larger than the original heat. This confirmed the strains e1 and e 2. The result (Fig. 41) shows that
predicted results in Ref 10. superimposed pressure (P 0) increases the
Modified Empirical Criterion. It was shown
previously in this article that measured free sur-
face strains at fracture fit a linear or bilinear line
that constitutes a fracture locus for the material
tested (Fig. 10 to 12). This is a convenient
representation of the complexities of ductile
Fig. 37 Decrease in circumferential tensile strain at Fig. 39 Comparison of measured strains at the cup
fracture with increasing strain rate for the cop- Fig. 38 Modified preform slug for the back extrusion rim during back extrusion of the modified
per alloy tested in Fig. 36. Results are for thin-flanged of the cup shape shown in Fig. 34. The slug di- preform slug shown in Fig. 38. The strains do not exceed
compression specimens, which have the lowest fracture ameter is 18.8 mm (0.74 in.) and has a 5 taper on the top the material fracture strain line for low or high strain rate
strain. surface. forming.
height of the fracture strain line and also in- is compressed between dies, material flows radi- (die orifice diameter), the hydrostatic stress be-
creases its slope slightly. Superimposed tension ally inward and then vertically into the opposed comes tensile, so the fracture line decreases in
(P 0) decreases the height of the fracture line, hubs. This develops a hydrostatic tensile stress height. The strains at the center continue to rise,
decreases its average slope, and gives it a slight state at the center (Fig. 6a), and fracture is a however, and cross the fracture line, leading to
downward curvature. It is clear that the increase strong possibility. the central burst. The calculated hydrostatic ten-
in strains to fracture due to additional pressure is Through visioplasticity analysis on split, grid- sion at fracture was 0.3Y. This approach could
unlimited as pressure increases, but the strains to ded specimens (Fig. 42), the strain and stresses be used for predicting central burst in drawing
fracture due to additional tension are limited by at the center of the workpiece were calculated and extrusion to provide a material-dependent
zero as tension increases. This result is discussed for several increments of deformation. The criterion, as opposed to the more simplistic
with regard to internal fracture and die contact hydrostatic stress state at the center of the spec- upper bound and tensile stress criteria described
surface fracture in the following paragraphs. imen is not always tensile; initially, it is com- previously.
Central Burst in Forgings. Internal fractures pressive, and then it reverses, becoming tensile Die Contact Surface Fracture. Frequently,
along the centerline of extruded or drawn bars as the flange thickness is reduced and flow into cracks occur during forging on surfaces that are
were discussed earlier (Fig. 18, 19, 23, and 40). the hub occurs. Meanwhile, the strains at the in contact with the dies (Fig. 5). One common
Similar fractures are observed in forged shapes center are increasing monotonically as deforma- location of such defects is the vicinity of a die or
such as that shown in Fig. 42 for heat-treated tion progresses. This is illustrated in Fig. 43 by punch corner. From the observation of a variety
6061 aluminum alloy. Here, as the outer region the steps 0-1-2-3. As deformation proceeds, the of such defects, it appears that a common char-
strains at the center increase, but the hydrostatic acteristic is an abrupt change in frictional shear
pressure is also increasing, so the fracture line traction distribution in the region of the crack.
moves upward. Then, as the flange thickness High friction to retard metal flow in advance of
approaches one-half of the hub base diameter the crack location is one method for preventing
such defects.
A technique for studying die contact surface
cracks was developed by means of a disk com-
pression test and dies having a rough surface in
the central region and a smooth surface in the
outer region. Figure 44 shows the top view of a
6061 aluminum alloy disk compressed between
such dies. In the transition region between the
rough central die surface and the smooth outer
region, radial cracks initiate and propagate out-
ward. Such cracks occurred at approximately
30% reduction when the smooth outer region
was lubricated with Teflon. The cracks occurred
at approximately 45% reduction when grease lu-
brication was used in the outer smooth region.
No cracks occurred even for very large reduc-
Fig. 40 Location of process conditions on a theoreti- tions when the smooth outer region was not lu-
cal central burst map. For an angle of 22.5, bricated.
central burst occurred in 4.5% of the extruded shafts. For
a die angle of 5, no central burst occurred.
Fig. 41 Movement of the fracture strain line due to su-

perimposed hydrostatic stress. Applied stress Fig. 43 Progression of surface strains and fracture line
is represented in terms of multiples of the yield strength, Fig. 42 Internal fracture during the double-extrusion at the central internal location of the double-
Y. Negative values of P indicate hydrostatic tension. forging of aluminum alloy 6061. Grid defor- extrusion forging shown in Fig. 42. The fracture line rises
Calculations are based on a modification of the Cockcroft mations on the middle longitudinal plane are shown. from 0 to 1 to 2 as internal pressure increases and then
criterion. Stress and strain states are defined by Fig. 6(a). falls to point 3 as the internal stress becomes tensile.
Grid marks placed on the die contact surface normal pressure during compression of a disk Such defects occurred only when the thickness
of the disks were used to measure the distribu- with grease lubricant in the outer region (indi- of the rib was greater than approximately one-
tion of surface strains in the radial direction. cated by the increments of reduction to 45%). half the web length; that is, extrusion reduction
Figure 45 gives an example of such measure- The fracture line increases at a slower rate than was less than one-half. Each crack formed by
ments. In the rough central region the strains are the strains increase with increasing pressure, and shear at the corner as material flowed from the
zero, while in the smooth outer region the strains at 45% reduction the strain path exceeds the web into the vertical rib.
are equal and constant. In the transition, how- fracture line and cracks are observed. For Teflon The stress and strain state on a surface mate-
ever, the circumferential strain, e , jumps lubricant, the crossover occurs at about 30% re- rial element is shown in Fig. 48. As material in
abruptly from zero to its constant value in the duction, while in the case of no lubricant the the web is compressed, the surface element ex-
outer region, and the radial strain, er, overshoots fracture line moves progressively away from the periences tensile strain in the direction of flow,
to a very high value before returning to its constrain path. and the strain increases as the element ap-
stant value in the smooth outer region. Example 4: Fir Tree Defect. The criterion proaches the die corner. At the same time, there
The strains shown in Fig. 45 were the same for contact surface fracture can be applied qual- is a compressive stress from the die onto the sur-
regardless of the friction condition in the smooth itatively for interpretation of the fir tree defect in face element. This pressure diminishes, how-
outer region. Therefore, fractures in the transi- extrusion. Such defects occur on the surfaces of ever, as the element nears the die corner and
tion region occur because of the combination of extruded bars as well as in localized areas of almost disappears as the element moves around
large tensile surface strains and low hydrostatic forgings containing ribs. In a section of a rib- the corner.
stress state. This explains the occurrence of web forging from a preform of sintered alu-
cracks at low reduction when Teflon is used, and minum alloy powder, small cracks formed with
no occurrence of fracture when no lubricant is a regular spacing on the rib surfaces (Fig. 47).
used. The Teflon, having a near-zero friction co-
efficient, results in very low radial back pressure
on the transition region, while grease and no
lubricant provide progressively larger back
pressures.
By means of visioplasticity analysis, the
stresses were determined at the contact surface
in the vicinity of the transition region. The re-
sulting normal die pressure plus the surface ra-
dial and circumferential strains define the stress
and strain states in the transition region and can
be illustrated on a forming limit diagram. Figure
46 shows the change in surface strains and the
increase in the fracture line due to increasing
Fig. 46 Progression of surface strains and fracture line

at the transition region between rough and
smooth zones of the compressed disk shown in Fig. 45.
Fig. 45 Radial variation of contact surface strains after Points 1, 2, 3, and 4 represent 10, 20, 30, and 45% re-
30% compression of the disk shown in Fig. 44 duction, respectively.
Fig. 44 (a) Top view of aluminum alloy 6061 disk

compressed between dies. (b) Cracks form at Fig. 47 Die contact surface cracking during forging extrusion of aluminum alloy powder compact. (a) Cross sec-
the transition region between rough and smooth areas of tion. (b) Normal to vertical rib surface. Note also the cracks at the top free surface. Stress and strain states
the die. are defined in Fig. 4(b).
Fig. 49 Progression of strain e1, and decline of frac-

ture line due to decrease in pressure P for an
element moving from the web to rib section during the
forging extrusion shown in Fig. 48. At stage 3, the strain
exceeds the fracture line.
8. B. Avitzur, Metal FormingProcesses and

Analysis, McGraw-Hill, 1968
9. W.M. Evans and B. Avitzur, Die Design for
Drawing Extrusion, Paper MF67582,
Fig. 48 Increase in strain e1, and decrease in die contact pressure P on a surface element as it moves from the Society of Manufacturing Engineers,
web into the rib section of the extrusion forging shown in Fig. 47. 1967
10. B. Avitzur, Strain Hardening and Strain
Rate Effects in Plastic Flow through Con-
ical Converging Dies, Paper 66-Prod-17,
This can be illustrated schematically on the vide a method of predicting the draft angle in American Society of Mechanical Engineers,
fracture strain diagram shown in Fig. 49. order to prevent fracture. 1966
Because the deformation is in a state of plane 11. R. Hill, On the Inhomogeneous Deform-
strain, the strain path is represented as a vector REFERENCES ation of a Plastic Lamina in a Compression
of increasing length along the vertical axis. Test, Philos. Mag., Vol 41, 1950, p 733
Meanwhile, the fracture line decreases in 1. H.A. Kuhn, P.W. Lee, and T. Erturk, A 12. J.F. Nye, Experiments on the Plastic
height because the pressure acting normal to Fracture Criterion for Cold Forging, J. Eng. Compression of a Block between Rough
the element is progressively decreasing. When Mater. Technol. (Trans. ASME), Vol 95, Plates, J. Appl. Mech. (Trans. ASME), Vol
the strains cross the fracture line, fracture 1973, p 213218 19, 1952, p 337
occurs. 2. H. Ortiz and H.A. Kuhn, Physical Modeling 13. J.J. Shah and H.A. Kuhn, An Empirical
This phenomenon does not take place when of Bar Rolling for Workability Study, Physi- Formula for Workability Limits in Cold
extruding ribs of small thickness, because the cal Modeling of Bulk Metalworking Proc- Upsetting and Bolt Heading, Proceedings of
extrusion reduction, and therefore the pressure, esses, Conf. Proc., ASM International, 1987 13th NAMRC, Society of Mechanical
is larger, which maintains the fracture line at a 3. E. Erman and H.A. Kuhn, Novel Test Engineers, 1985
high level. For thick ribs, two solutions were Specimens for Workability Measurements, 14. C.L. Downey and H.A. Kuhn, Application
considered. One approach is to increase friction Compression Testing of Homogeneous of a Forming Limit Criterion to Design of
along the rib walls by roughening the die surface Materials and Composites, STP 808, Preforms for Powder Forging, J. Eng.
or avoiding lubrication of the die rib. This pro- ASTM International, 1983, p 279290 Mater. Technol. (Trans. ASME), Vol 97H,
duces greater back pressure at the die corner, el- 4. F.A. McClintock, J. Appl. Mech. (Trans. 1975, p 121
evating the fracture line and preventing crack- ASME), Vol 90, 1968, p 363 15. H.A. Kuhn, Deformation Processing
ing. Such an approach is difficult to implement 5. Z. Marciniak and K. Kuczynski, A Model of of Sintered Powder Materials, Powder
and can be used only with segmented dies be- Localized Thinning in Sheet Metalforming, Metallurgy Processing, H.A. Kuhn and
cause the formed rib cannot be removed from Int. J. Mech. Sci., Vol 9, 1967, p 609 A. Lawley, Ed., Academic Press, 1978,
the die. The second approach is to use a draft 6. P.W. Lee and H.A. Kuhn, Fracture in Cold p 99
angle on the rib, which has the same effect as in- Upset ForgingA Criterion and Model, 16. Z. Zimerman, H. Darlington, and E.H.
creased friction. An angle of 10 prevented frac- Metall. Trans. A, Vol 4, 1973, p 969 Kottcamp, Jr., Selection of Operating
ture in the current case, but other alloys may re- 974 Parameters to Prevent Central Bursting dur-
quire a smaller or larger angle. A quantitative 7. M.G. Cockcroft and D.J. Latham, Ductility ing Cold Extrusion, Mechanical Working
analysis combining the pressure effect on the and the Workability of Metals, J. Inst. Met., and Steel Processing, The Metallurgical
fracture line and plasticity analysis would pro- Vol 96, 1968, p 3339 Society, 1970, p 405
Chapter 13
Workability in Forging
WORKABILITY IN FORGING depends on a metal flow in open-die forging (amount of re- may take a wide variety of forms, several of
variety of material, process-variable, and die- duction, frictional boundary conditions, and heat which are summarized in Table 1. For this rea-
design features. In general, forging is a term that transfer between the dies and the workpiece). son, it is not surprising that no single workabil-
can be used to describe a wide variety of bulk Simple closed-die forgings may also be largely ity test may be relied on for determination
metalworking processes, and forging engineers axisymmetric or plane strain in nature. Gears for of forgeability. However, a number of test
must consider a variety of workability tests to di- automotive and industrial uses and rotor blades techniques have been developed for gaging
agnose forging problems for a wide range of for steam and gas turbine engines are among the forgeability depending on alloy type and mi-
possible forging operations. In basic form, most common parts in these categories that are crostructure, die geometry, and process vari-
however, all forging processes consist of com- made using closed-die techniques. In addition, ables. This chapter summarizes some of the
pressive deformation of a piece of metal, known closed-die forging is often employed for manu- more common tests and illustrates their applica-
as the workpiece, between a pair of dies. facture of more complex parts in which the tion in practical forging situations.
Depending on the geometry of the dies, varying metal flow is three-dimensional in nature. These The term workability is also used in reference
amounts of lateral constraint may be imposed on include structural shapes consisting of thin webs to the determination of flow stresses and total
the workpiece, a fact that enables forging opera- and long, thin protrusions known as ribs, a class working loads in metal-forming practice (which
tions to be classified very broadly into two cate- of parts with wide application in the aircraft thus determines equipment utilization factors
gories: open-die forging and closed-die forging industry. such as required press capacity, die materials
(Ref 13). The complete description of any forging and die wear, and certain other die-design fea-
In open-die forging, the lateral constraint is process, be it basically an open-die or a closed- tures). Such topics are summarized in other
minimal, and the amount and distribution of lat- die process, requires specification of various chapters of this Handbook and in other sources
eral metal flow are controlled by factors such as process variables besides the die geometry and (Ref 1, 4) and thus are not discussed here. In ad-
total reduction in the workpiece dimension par- workpiece material. These include working dition, workability can also refer to the design of
allel to the forging direction, frictional boundary speed, die temperature, workpiece preheat tem- thermomechanical processes to control and
conditions, and heat-transfer effects (when the perature, and lubrication. Working speed is de- optimize microstructure for specific service ap-
dies and workpiece are at different initial tem- termined by the kind of equipment used and plications. The development of desirable mi-
peratures). Simple two-dimensional examples of varies depending on whether hydraulic presses crostructure during forging plays a critical role
such operations include axial compression of (relatively low forging speeds), mechanical and in the design of forging processes. The general
right-circular cylinders between flat, parallel screw presses (intermediate speeds), or hammers aspects of microstructure evolution during ther-
dies (a generic upsetting operation) and pressing (high speeds) are selected (Ref 1). momechanical processing are briefly reviewed
of long, round bars along their lateral surfaces, Because forging is a relatively complex in Chapter 3, Evolution of Microstructure dur-
also between flat, parallel diesa process aptly process, fracture or generation of undesirable ing Hot Working, in terms of both phenomeno-
known as sidepressing. The former is perhaps defects in the workpiece during deformation logical as well as mechanism-based approaches.
the simplest type of axisymmetric deformation,
and the latter the simplest type of plane-strain
deformation commonly used in forging practice.
Nevertheless, such basic operations find wide Table 1 Common metallurgical defects in forging
commercial application in processing sequences Metallurgical defects in:
for complex parts. For instance, upsetting is Temperature Cast grain Wrought (recrystallized)
often used to pancake round billets prior to regime structure grain structure
secondary processing via a closed-die forging
Cold working (a) Free-surface fracture
operation. Furthermore, open-die forging is fre- Dead-metal zones (shear bands, shear cracks)
quently used in conjunction with annealing Centerbursts
treatments during the breakdown of cast ingots Galling
in order to obtain recrystallized, wrought mi- Warm working (b) Triple-point cracks/fractures
Grain-boundary cavitation/fracture
crostructures. Hot Working Hot shortness Shear bands/fractures
As its name implies, closed-die forging is Centerbursts Triple-point cracks/fractures
done in closed or impression dies that impart a Triple-point cracks/fractures Grain-boundary cavitation/fracture
well-defined shape to the workpiece. As might Grain-boundary cavitation/fracture Hot shortness
Shear bands/fractures
be expected, the degree of lateral constraint
varies with the shape of the dies and the design (a) Cold working of cast structures is typically performed only for very ductile metals (e.g., dental alloys) and usually involves many stages of work-
of the peripheral areas where flash is formed, as ing with intermediate recrystallization anneals. (b) Warm working of cast structures is rare.
well as with the same factors that influence
Chapter 13: Workability in Forging / 189
Workability Tests for Open-Die ture range over which ingot structures can be working of cast (and wrought) structures can be
Forging of Cast Structures forged is very limited. Typically, they must be further restricted by the effects of die chilling.
hot worked, that is, worked at temperatures Because hot working is conventionally carried
greater than three-fifths of the absolute melting out with a hot workpiece and usually much
In open-die forging, metal-flow patterns and point of the alloy. The upper temperature limit in cooler tool steel dies, heat transfer brought about
stress states are not highly complex, and forge- this range is usually slightly below the melting by die chilling may lead to workpiece tempera-
ability is determined primarily by material struc- point of the metal. The melting point of an alloy tures during deformation that are much lower
ture and properties and process conditions and in the as-cast condition is usually lower than that than the preheat temperature. In terms of die de-
only secondarily by die geometry. Material- of the same alloy in the finer-grain, recrystal- sign, secondary tensile stresses are often set up
structure variables include grain structure and lized state because of chemical inhomogeneities in open-die forging (Ref 7).These stresses may
texture, crystal structure, and the presence of sec- and the presence of low-melting-point com- assist the grain-boundary fracture process, giv-
ond phases or solid-solution elements. Material pounds, which are often found at the grain ing rise to gross centerbursts, and should be
properties include temperatures at which melt- boundaries. Forging at temperatures too close to evaluated in workability tests as well.
ing, recrystallization, and phase changes occur; the melting points of these compounds may lead From the previous discussion, it is clear that
the variation of flow stress with strain, strain rate, to grain-boundary cracking when heat devel- workability tests for open-die forging of cast
and temperature; and physical properties such oped during deformation raises the workpiece structures should be designed in order to deter-
as density, specific heat, and thermal conductiv- temperature and causes melting of the com- mine realistic temperature regimes in which a
ity. Of the process variables mentioned previ- pounds. This fracture mechanism is called hot recrystallized structure may be obtained and
ously, workpiece preheat temperature is the most shortness and is sometimes avoided by forging fracture avoided. Two such tests are the wedge-
important. at deformation rates sufficiently low to allow the forging test and the sidepressing test.
The variation of workability with workpiece heat developed by deformation to be dissipated The wedge-forging test (Ref 2, 8) is ideal for
preheat temperature depends a great deal on into the dies, by use of lower working tempera- determining breakdown temperature ranges for
grain size and grain structure, factors that influ- tures, or by use of homogenization heat treat- ingot structures. In this test, a wedge-shaped
ence the choice of workability tests (Fig. 1). ments prior to working. The last of these piece of metal is machined from the ingot (Fig.
When the grain size is large, as in conventionally methods is especially useful for the breakdown 2) and forged between flat, parallel dies. The
cast ingot structures, cracks may initiate easily of highly alloyed metals such as nickel-base overall dimensions of the wedge are selected on
(and then propagate along the grain boundaries) superalloys. the basis of initial grain size. Large-grain mate-
due to the large stress concentrations associated The lower temperature limit for hot forging of rials require large specimens, and fine-grain ma-
with slip within the grains or unrelaxed sliding cast structures is the temperature at which there terials require small specimens. In either case, a
at triple points (where three grain boundaries in- is some, but not a great deal of, dislocation mul- specimen large enough to be representative of
tersect). Also, as is often the case in cast struc- tiplication. Dislocations are needed to build a the bulk of the ingot should be chosen. In many
tures, impurities are segregated to the center and reservoir of stored energy that, in combination cases, a number of specimens are required be-
top of the ingot (producing areas of low worka- with thermal energy (supplied during hot defor- cause of the variety of grain shapes (columnar
bility there), and chemical elements are not dis- mation or during postdeformation heat treat- versus equiaxed), grain sizes, and grain orienta-
tributed uniformly on either a microscopic or a ment), is useful in breaking down the coarse tions relative to the working direction that per-
macroscopic scale (Ref 5). Thus, the tempera- grains and generating a much finer recrystallized tain to ingot breakdown. Specimen design and
grain structure. However, the benefit of disloca- subsequent interpretation of workability data
tion generation must be balanced against the may require that the crystallographic textures
reduction with temperature of grain-boundary that are developed during ingot solidification
mobility, a phenomenon that leads to dynamic also be taken into account. For example, colum-
recrystallization and offsets the tendency for in- nar grains typically grow via solidification along
tergranular fracture due to grain-boundary slid- a preferred (rapid-growth) direction and thus
ing (Ref 6). Low forging temperatures may also possess noticeable fiberlike textures.
lead to precipitation of alloying elements that re- The wedge test is different from other worka-
main in solution at higher temperatures, an oc- bility tests (e.g., tension test, uniaxial compres-
currence that tends to pin matrix dislocations sion test) designed for small specimens of
and also to cause grain-boundary failure. material with relatively fine wrought grain sizes
Precipitation at grain boundaries may also be that are subjected to nominally uniform defor-
harmful when it provides sites for grain- mation throughout the test section. In contrast,
boundary cavitation. From a process-variable wedge specimens forged between flat, parallel
standpoint, the lower temperature limit for dies suffer strains that vary with position. The
variation of strain can be estimated to a first
order from measurements of local thickness re-
duction and width increase. If higher accuracy is
desired, finite-element-method (FEM) simula-
tions can be used to predict local strains and
strain rates and thus to correlate process vari-
ables to failure or microstructure evolution (Ref
9, 10). Typical results for the effective strain de-
veloped in wedge forging are shown in Fig. 3.
Machined wedge specimens are forged (or,
equivalently, flat rolled) in the equipment in
which the actual forging is to take place. This
Fig. 1 Schematic illustration of relative workability of permits evaluation of the effects of working
cast metals and wrought and recrystallized met- speed, and hence of die chilling due to heat
als at cold, warm, and hot working temperatures. The
melting point (or solidus temperature) is denoted as MPc Fig. 2 Specimens for the wedge test. (Top) As-machined transfer, on workability. Wedge preheat temper-
(cast metals) or MPw (wrought and recrystallized metals). specimen. (Bottom) Specimen after forging atures are selected on the basis of previous ex-
kinds of results that can be obtained from a

wedge test in which deformation is imposed by
rolling rather than by forging. The material was
an experimental high-nitrogen stainless steel
(nominal composition, in weight percent: 24
Cr, 16 Ni, 5 Mn, 0.1 C, 0.65 N) produced by
plasma arc remelting. The von Mises effective
strain, e, (Ref 7, 12) at various locations was
estimated solely from measurements of the
final thickness of the wedge specimen in order
to obtain the compressive thickness strain pro-
file. Because the wedge was rolled, rather than
forged between flat, parallel dies, the width
strain was small; therefore, the longitudinal
tensile strain was taken to be equal in magni-
tude to the thickness strain because of the con-
stant-volume assumption.
Estimates of the percentage of recrystallized
structure at the wedge specimen midwidth for the
high-nitrogen stainless steel for specimens oil
quenched immediately after deformation (Fig. 4)
illustrate the important influence of preheat tem-
perature and strain on dynamic recrystallization
(Ref 11, 13). The data were taken from test spec-
imens deformed at temperatures between 1090
and 1230 C (1995 and 2245 F), with each spec-
imen providing a data point for each constant-e
curve. Also illustrated in Fig. 4 is the fact that
cracking of wedge specimens occurred for T
Fig. 3 Finite-element-model predictions for the wedge-forging test, assuming a friction shear factor of 0.4. (a) Grid 1090 C (1995 F) and T 1200 C (2190 F).
distortions. (b) Strain contours along the symmetry plane. Source: Ref 9
In both regimes, failure was intergranular. It was
found from metallography and isothermal-trans-
perience with the alloy or, for new alloys, previ- Following deformation, wedge specimens formation results that the alloy formed chromium
ous experience with similar alloys. If no such in- are quenched or furnace annealed to determine nitride precipitates, the kinetics of formation of
formation is available, a temperature of approx- dynamically and statically recrystallized grain which were greatest at approximately 1050 C
imately 0.8 to 0.9 times the measured solidus sizes as functions of working temperature, (1920 F) (Ref 14). Because the rolls were oper-
temperature is a good starting point. This choice amount of deformation, and annealing temper- ated at a temperature much lower than the wedge
can be modified based on examination of data ature and time. Figures 4 and 5 illustrate the preheat temperatures, a large amount of chilling,
for possible binary and multicomponent phases
that may be formed from the alloy constituents
and that are known to possess low melting points
or that precipitate from solution at high temper-
atures. Even if no such phases are expected to be
present, very high temperatures often are not ad-
visable, because recrystallization may be ac-
companied by grain growth and a less workable
microstructure.
Fig. 4 Dependence of recrystallization on rolling tem-

perature and deformation for high-nitrogen
stainless steel wedge-test specimens cut from cast ingot. Fig. 5 Percentage of recrystallized structure in wedge-test specimens after various thermomechanical treatments per-
Source: Ref 11 formed on a high-nitrogen stainless steel. OQ, oil quench. Source: Ref 11
or cooling of the workpiece surface in contact lish the ability of residual hot work to promote in Ref 17 and 18, the maximum tensile stresses
with the rolls during deformation, must have oc- recrystallization of a cast structure. in this instance occur at the center of the bar at
curred. Because of this effect, specimens pre- A workability method that is very similar to the beginning of the test (Fig. 7). As deforma-
heated to 1090 C (1995 F) actually experienced the wedge test involves hot forging of so-called tion proceeds, the bar assumes more of a rec-
much lower temperatures, at which precipitation double-cone samples whose size is established tangular cross section. As this occurs, the mag-
of any dissolved chromium and nitrogen was based on the grain size of the cast microstructure nitudes of the tensile stresses decrease, and the
likely. This precipitation would most likely occur (Fig. 6a). Finite-element modeling is often used location of the maximum stress is shifted away
on dislocations, impeding their motion and lead- to establish local deformations for such tests in from the center of the bar. Similar changes in
ing to intergranular failure. Precipitation in this order to quantify fracture and microstructure magnitude and location of the maximum tensile
alloy was also found to occur along the grain observations, much as in the wedge test. Typical stress can be made to occur at early stages of
boundaries, preventing recrystallization and results from such simulations are shown in deformation by changing from flat dies to dies
healing of intergranular microcracks. Fig. 6(b). that encompass more of the workpiece (Fig. 7);
Data in Fig. 5 for rolled and subsequently an- Sidepressing Test. Because specimen size that is, secondary tensile stresses are minimized
nealed wedge specimens of the high-nitrogen can be varied, the sidepressing test is another as less of an open-die geometry is employed
stainless steel demonstrate the usefulness of the method well suited for establishing the worka- and thus as more constraint is imposed.
wedge test not only as a workability technique bility of large-grain cast structures. It is particu- Moreover, friction and chilling may act in con-
but also as a means of devising detailed thermo- larly useful for estimating the interaction of cert to promote bulging and thus to increase (or
mechanical processing sequences for obtaining incipient grain-boundary cracks, on the one maintain) high levels of tensile stress during
recrystallized microstructures in cast ingots. hand, and secondary tensile and hydrostatic sidepressing above that level resulting from
These static recrystallization trends are similar stresses, often present in open-die forging, on geometry alone.
to those for annealing of cold-worked and hot- the other. In this test, round bars are laterally The application of sidepressing as a workabil-
worked fine-grain metals (Ref 5, 15) and estab- pressed between flat, parallel dies. As discussed ity test for cast materials can also be illustrated
using breakdown of ingots of high-nitrogen
stainless steel as an example (Fig. 8). When
round ingots were forged round-to-round via
sidepressing at 1150 C (2100 F), at which
temperature dynamic recrystallization was
found to occur, centerbursting was observed
(Fig. 8a). This must be attributed to the inability
to recrystallize weak interfaces early enough in
the deformation so as to avoid defect initiation
and growth. Such an occurrence could be
avoided (Fig. 8b) by forging at 1125 C (2060
F), at which dynamic recrystallization was also
favorable, and by forging/sidepressing in a se-
quence of round-to-square operations. (Round-
to-square forging at 1150 C (2100 F) also
eliminated centerbursts, but minor surface
cracks were still evident, presumably because of
chilling.) In the latter instance, the decrease in
tensile stresses can be assumed to have been
beneficial in eliminating centerbursting and is in
line with the typical press-forging practice for
ingot breakdown.
Workability Tests for Hot/Warm

Open-Die Forging of
Recrystallized Structures
Breakdown of coarse-grain cast structures to
produce fine-grain wrought-and-recrystallized
structures generally leads to improved workabil-
ity in forging, as shown schematically in Fig. 1.
Even though most wrought metals can be
worked over wide ranges of temperature and de-
formation rate, care must be exercised in selec-
tion of forging temperature. Hot and warm forg-
ing require lower working loads than cold
forging (performed at temperatures less than
one-fourth of the melting or solidus tempera-
ture), but several types of defects can still arise
in hot and warm open-die (and closed-die) forg-
Fig. 6 The double-cone test. (a) Typical sample geometry. (b) FEM-predicted strain and strain-rate contours, assuming. These defects include many of those previ-
ing a friction shear factor of 0.5. Source: Ref 16 ously discussed for cast structures (Table 1).
Examination of the common defects in hot and best ductility/forgeability except when grain
warm working of wrought-and-recrystallized growth occurs at high temperatures (Ref 2123)
structures (Table 1) reveals a convenient grouping (Fig. 10). On the other hand, alloys that contain
that is helpful in devising different types of work- low-melting-point phases (e.g., iron alloys with
ability tests for open-die forging. This consists of sulfur) or intermetallic second phases (e.g.,
two classes of workability problems: those related gamma-prime-strengthened nickel-base superal-
to fracture-controlled failure and those related to loys) tend to be difficult to forge and have lim-
flow-localization-controlled failure. The first ited forging temperature ranges. Such behavior
group comprises hot shortness, triple-point crack- for alloys that contain strengthening phases is
ing, and grain-boundary cavitation (Fig. 9). The further illustrated in Fig. 12. This figure shows
second group consists of shear banding (which the relative forgeability of such an alloy system
can lead to fracture) that may be initiated by heat- as a function of temperature and of percentage
transfer effects (due to chilling) as well as by flow of alloy content that forms second phases. The
softening (in the absence of chilling). trends shown are very representative of nickel-
base superalloys, among which a lean alloy
Workability Tests for Fracture- might be 718, and an alloy that is highly
Controlled Defect Formation strengthened by precipitates might be alloy 700.
As the amount of alloying increases, the proba-
Workability tests for fracture-related prob- bility of a low-melting-point phase being formed
lems in forging must be able to determine the increases in much the same way as the liquidus
interactions among material aspects, process temperature in a binary eutectic system de-
variables, and secondary tensile stresses. Ma- creases as the eutectic composition is ap- Fig. 8 Sidepressed bars of a high-nitrogen stainless
steel cut from a cast ingot. (a) Bar forged round-
terial aspects include those mentioned before proached. Further, the temperature for precipita- to-round at 1150 C (2100 F), which developed center-
such as crystal structure; phase changes; solidus, tion increases as the amount of second-phase burst. (b) Bar successfully forged round-to-square at 1125
solutioning, and recrystallization temperatures; elements, and hence the possibility for super- C (2060 F). Source: Ref 11
and mechanical and physical properties. Often, saturation, increases.
simple tests such as the hot compression, ten-
sion, and torsion tests are useful in the initial se- Workability Tests for Establishing
lection of forging temperature and strain rate Effects of Process Variables
and in gross estimation of forgeability. Usually,
the estimate of forgeability is based on a param- Tests for establishing the effects of process
eter such as upset-test reduction at fracture, re- variables on workability in forging include the
duction of area in tension test, or shear strain to high-speed isothermal tension test (or compres-
failure in torsion. A useful summary of such sion test) and the on-cooling tension test. The
measurements has been made for different types former test gives some estimate of the effects of
of alloy systems (Fig. 10, 11) (Ref 2). Generally, strain rate and deformation heating on ductility.
pure metals and single-phase alloys exhibit the Done under nominally isothermal conditions,
Fig. 9 Nucleation of grain-boundary voids (cavitation)

and triple-point cracks at warm and hot working
temperatures. (a) Schematic illustration showing how
grain-boundary voids are formed under the action of ma-
trix deformation and how grain-boundary sliding in the
absence of grain-boundary migration and recrystallization
may cause cracks to open at triple points. (b) Examples of
grain-boundary voids and triple-point cracking at the prior
beta grain boundaries in hot-forged Ti-6Al-2Sn-4Zr-2Mo-
Fig. 7 Schematic illustration of the effects of billet shape and degree of enclosure on stress state in forging with good 0.1Si with a colony-alpha starting microstructure. Source:
lubrication and no chilling. Source: Ref 17 Ref 19, 20
the high-speed tension test (see Chapter 7, Hot in establishing the effects of die chilling in forg- improve forgeability. For example, a pretreat-
Tension Testing) may be used to gage when ing on workability. In such a technique, a series ment to precipitate and overage (i.e., coarsen)
deformation-induced heating may lead to tem- of nominally isothermal tension tests is first per- second phases in various iron- and nickel-base
perature increases sufficient to cause incipient formed over a range of temperatures by simply alloys has been found to be very beneficial in
melting (Ref 22). The temperature rise is easily heating to the test temperature and running the subsequent working (Ref 24), an effect that can
estimated if the specimen gage length is great tests. Then, a second set of tests is run by heat- be evaluated most readily by means of simple
enough and the test fast enough to give rise to ing to a variety of temperatures, holding at those workability tests.
adiabatic conditions at the center of the gage temperatures for fixed or variable times, and Hot forging is often a multistep process, with
section. Under these circumstances, the temper- then running the test after cooling to a lower intermediate dwells or reheats between each in-
ature rise is derivable from the measured flow temperature. It is obvious that such a test is use- crement of deformation and changes in loading
curve and the material density and specific heat ful for determining the effects of chilling as a direction. Because of this, microstructure evolu-
by assuming that a certain fraction (usually ap- function of working speed and thermal proper- tion and the propensity for fracture may be
proximately 0.95) of the deformation work is ties on the forgeability of alloys that form sec- dependent on the precise thermomechanical his-
transformed into heat. When the gage section of ond phases on cooling or that undergo grain tory to which the workpiece is subjected. Hence,
the tension specimen is short or the deformation growth during high-temperature preheating. a number of studies have been undertaken to es-
slow, such that a substantial amount of heat Further information on such testing is given in tablish the effect of single- versus multi-blow
transfer occurs between the gage section and the Chapter 7, Hot Tension Testing, of this deformation and strain-path changes on mi-
undeformed shoulders, temperature changes and Handbook. crostructure development (Ref 2529). It is
workability observations must be interpreted The standard tension test and on-cooling ten- likely that similar studies will be undertaken in
with the aid of a numerical thermal simulation. sion test may also be useful for evaluation of the future to establish the path-dependence of
The on-cooling tension test (Ref 22) is useful heat treatments prior to deformation which may fracture as well.
Tests for Establishing Effects of

Secondary Tensile Stresses
on Forgeability
Sometimes the uniaxial tension and compres-
sion tests are insufficient for determining forge-
ability in actual open-die and closed-die forging
operations. This may arise because of secondary
and hydrostatic tensile stresses that can exacer-
bate a marginal workability problem. Several
techniques have been developed specifically to
establish forging guidelines in these instances.
These include the notched-bar or U-notch upset
test, the double-cone test, and pancake forging
itself.
The notched-bar upset test (Ref 2, 8, 30),
developed by the Ladish Company, is similar to
the conventional upset test, except that axial
notches are machined into the test specimens
prior to compression. The notch magnifies the
effects of secondary tensile stresses that may
arise during a conventional upset test due to
bulging caused by chilling and/or poor lubrica-
tion. The higher levels of tensile stress in the test
are supposedly more typical of those in actual
forging operations. In preparation of specimens
for the test, a forging billet is quartered longitu-
dinally, exposing center material along one cor-
ner of each test specimen. After sectioning,
notches with either 1.0 or 0.25 mm (0.04 or 0.01
in.) radii are machined into the faces (Fig. 13); a
weld button is often placed on one corner to
identify center and surface material in metals
that exhibit forgeability problems because of
segregation. Assuming a uniform bulge, the
faces with the deeper (1.0 mm, or 0.04 in.)
notches are subjected to the highest tensile
stresses and thus are more likely to indicate a
workability problem.
The notched-bar specimens are heated to one
or a number of possible forging temperatures
and upset forged to a reduction in height of
approximately 75%. Because of the stress-
Fig. 10 Typical forgeability behaviors exhibited by different alloy systems. Source: Ref 2 concentration effect, fractures in the form of
Fig. 12 Influence of solute content that forms second

phases on melting and solution temperatures
and thus on forgeability
Fig. 11 Chart illustrating interaction of workability, flow strength, and die-filling capacity in forging. Large shaded
arrow indicates increasing ease of die filling. Source: Ref 2
ruptures are most likely to occur in the notched ing numbers indicate increasing incidence and
areas and may be assumed to have been initiated depth of rupture.
by hot shortness, triple-point cracking, grain- Examples of the application of the notched-
boundary cavitation, and so on. These ruptures bar test are those reported in Ref 2 and 8. In the
may be classified according to the rating system former example, rolled rings of type 403 stain-
shown in Fig. 14. A rating of 0 is applied if no less steel with a notched-bar rating of 0 were
ruptures are observed. If they are small, discon- sound, whereas those with a rating of 4 ruptured
tinuous, and scattered, the rating is 1. Higher rat- extensively. In a similar vein, the rupture sensi-
Fig. 14 Suggested rating system for notched-bar upset

test specimens that exhibit progressively
poorer forgeability. A rating of 0 would indicate freedom
Fig. 13 Method of preparing specimens for notched-bar upset forgeability testing. Source: Ref 2, 30 from ruptures in the notched area. Source: Ref 2, 30
tivity of Inconel 718 was demonstrated with the The comparison of FEM solutions for the its thickness. Because of the large reduction,
notched-bar upset test (Fig. 15). The alloy was stress field in the double-cone test with actual friction, and chilling (during conventional hot
shown to be much more notch-sensitive as the observations has confirmed the usefulness of forging), substantial barreling of the free surface
test temperature was increased. The combined this technique as a workability test. For example, occurs, giving rise to substantial tensile stresses.
effects of interface frictional and deformation cavitation observations in double-cone samples These stresses may cause free-surface fracture
heating in conjunction with the stress concentra- of Ti-6Al-2Sn-4Zr-2Mo-0.lSi with a colony- as well as subsurface cavitation without gross
tion were able to produce very large ruptures alpha microstructure forged isothermally (dies fracture. The analysis of free-surface fracture is
when the alloy was forged at and above 1150 C and workpiece at the same initial temperature) at described in Chapter 4, Bulk Workability
(2100 F). Such an effect is not surprising in 940 C (1725 F) and a strain rate of approxi- Testing. Because cavities can detract from sub-
view of the fact that Inconel 718 starts to melt at mately 2.5 s1 were found to correlate well with sequent load-carrying capability and fatigue
approximately 1200 C (2190 F) (Ref 31). the (peripheral) regions in which circumferential resistance during service, it is important to quan-
From these results, one may conclude that the tensile stresses were predicted to be developed tify the occurrence of such defects, particularly
notched-bar upset test is more sensitive than the (Ref 32, 33). Furthermore, the analysis sug- when their severity is just below the level that
simple upset test. In fact, it has been reported gested that the defects originated at low strains, causes surface fracture.
that unnotched billets from heats having a at which the tensile stresses were relatively A method to quantify and interpret cavitation
notched-bar rating of 3 are perfectly sound after large, and that the tensile stresses diminished as based on pancake forging is described in Ref 35.
similar reductions in simple upsetting. Thus, the the amount of compression increased, giving In this work, Ti-6Al-4V samples with a colony-
simple upset test may indicate a deceptively rise to less taper on the forging. In effect, the alpha microstructure were hot forged to various
higher degree of workability than can be real- simulation suggested that the defects move to- reductions. Following deformation, the depth
ized in an actual forging operation, and the ward the outside of the disk as the reduction in- below the surface at which cavities were found
notched-bar upset test may be particularly useful creased, leaving behind a sound structure, a (at 500) was determined in each sample. The
for identifying materials having marginal trend verified by comparison with experimental strain and stress state that corresponded to the
forgeability. observations. Therefore, the test may be used to regions of cavitation were determined using a
The double-cone test is another method em- determine when voids may be formed and when FEM technique (Fig. 16). In a similar manner,
ployed to study the effects of secondary tensile they may be healed by further deformation. cavitation was quantified in hot tension samples
stresses on defect formation. Similar to the The double-cone test is also finding increased deformed at comparable strain rates and temper-
notched-bar approach, this test involves a simple use as a method to establish microstructure evo- atures. A comparison of the results from the ten-
upsetting operation between flat, parallel dies. lution during hot working of wrought alloys. For sion and forging experiments demonstrated that
However, the tensile stresses are developed by a example, it has been used to determine the effect cavity initiation (as well as gross surface frac-
much different specimen geometry (Fig. 6). As of process parameters on recrystallization and ture) in the two different deformation modes
the sample is upset, secondary tensile stresses abnormal grain growth during the thermome- could be correlated based on local values of the
are generated in the outside portion of the spec- chanical processing of nickel-base superalloys Cockroft and Latham (Ref 36) maximum tensile
imen. The generation of this stress state is anal- and microstructure evolution during hot working work, C (st /s)de, in which st and s denote
ogous to that in sidepressing of round bars. of beta-titanium alloys (Ref 16, 34). the maximum tensile stress and the effective
Because of secondary tensile stresses, voids ini- Pancake forging comprises the upsetting of a stress, respectively, and e is the effective strain.
tiated by grain-boundary cavities and/or triple- round cylinder to large strains, thus producing a (In uniaxial tension involving minimal necking,
point cracks and so on may be opened. pancake whose diameter is much greater than st s, the critical values of C for cavity initia-
tion (C *i ) and fracture (Cf*) are simply the corre-
sponding axial strains.) An example of this cor-
relation is shown in Fig. 17. Hence, a critical
damage parameter can be determined from pan-
cake forging trials and accompanying FEM
analysis. This critical damage parameter can
then be used to predict the possible occurrence
of cavitation in other forgings for which FEM
simulations are also conducted to determine
local values of the maximum tensile work.
Mechanistic Modeling of Cavitation.
Unlike the well-developed phenomenological
approaches for ductile fracture, such as that
based on the Cockcroft and Latham criterion
(Ref 36), the development of physics-based
(mechanistic) models for the prediction of cavi-
tation during hot forging is still in a relatively
early stage of development. To a certain extent,
this is a result of the complexity of the nucle-
ation-and-growth processes themselves as well
as the influence of complex stress states that
characterize forging. For example, the classical
surface-energy criterion for cavity nucleation at
grain-boundary particles (Eq 17 in Chapter 7,
Hot Tension Testing) has a number of draw-
backs. To replicate experimental observations of
continuous nucleation, the criterion requires
flow hardening in order to nucleate cavities at
Fig. 15 Nickel alloy 718 notched-bar upset-test specimens showing rupture sensitivity associated with forging tem- less favorable sites compared to those at which
perature. Source: Ref 8. Courtesy of Ladish Company nucleation occurs initially. In addition, the
Fig. 16 Finite-element-modeling calculations of (a) Cockcroft and Latham tensile work ((st / s)de), (b) st / s, and (c) effective strain, e, as a function of stroke for a Ti-6Al-4V sam-
ple with a colony-alpha microstructure forged at 815 C (1500 F), 0.1 s1 to a final height reduction of 50%. The results are plotted for material tracking points located
at various distances from the free surface on the equatorial plane. The stroke at which cavitation was predicted to initiate (based on uniaxial tension measurements) is indicated by
the intersections with the C Ci* curves in the various figures. Source: Ref 35
stresses required for initiation and early growth in which e and e denote the far-field strain and
are unrealistically high. Therefore, the formula- strain rate, respectively. The dependence of D on
tion of other (constrained-plasticity) approaches stress was shown to be as follows:
based on cavity nucleation and early growth
from preexisting inhomogeneities/nanovoids (at D 0.558 sinh(3sm/2s)
which locally higher stress triaxiality is devel-
oped) has been undertaken (Ref 37). In such ap- 0.008 n cosh(3sm/2s) (Eq 4)
proaches, the degree of constraint decays as the
cavity grows from the nanometer to the microm- in which n denotes a function of the strain rates
eter scale, at which classical general plasticity- defined as n 3e2/(e e3), with e1 e2 e3
controlled analyses, such as are discussed in representing the principal strain rates. The de-
Chapter 7, Hot Tension Testing, apply. pendence of D on the ratio of the mean to the ef-
For cavities whose size is of the order of sev- fective stress, sm/s, is shown in Fig. 18(a). For
eral micrometers or larger, a simple method was example, for uniaxial stress conditions, sm/s
developed (Ref 38) for assessing the effect of 1
3, n 1.0, and D DU 0.30. Equation 3
stress state on the kinetics of cavity growth. The thus incorporates the effect of the local stress
approach was based on the plasticity-controlled state (via D) on the growth of an individual
growth of an isolated spherical cavity. Under cavity.
purely uniaxial tension conditions (i.e., sm/s It was postulated (Ref 38) that the increase of Fig. 17 Comparison of FEM calculations of the
Cockcroft and Latham tensile work ((st / s)de)
1
3, in which sm is the hydrostatic stress, and s the cavity growth rate parameter, o, (Eq 1) due at the end of the stroke to the critical values established
denotes the effective stress), such growth is de- to stress triaxiality is proportional to D/DU, that from tension tests for cavity initiation (Ci*) and surface
scribed by the following relation: is: fracture (Cf*) in pancake forgings of Ti-6Al-4V with a
colony-alpha microstructure conducted to various height
reductions at 815 C (1500 F), 0.1 s1. The predicted
ho h = ho
D
h = ho F(s )
depths of cavitation and the critical reduction for surface
R = Ro exp e (Eq 1) or (Eq 5) fracture were well predicted by such comparisons.
3 DU
Source: Ref 35
In Eq 1, R and Ro are the instantaneous and in which denotes the cavity growth rate pa-
initial cavity radii, respectively, e is the axial rameter under the complex stress state, and Testing, expressions are readily determined to
strain, and o is the volumetric cavity growth rate F(s)D/DU. F(s), and hence the average cavity relate to the overall evolution of cavitation
parameter under uniaxial stress conditions (o growth rate, increases as sm/s increases; for ex- when continuous nucleation and coalescence
dlnV/de). The effect of stress triaxiality on cavity ample, at the same strain level, the cavity size occur in addition to the growth of single cavities.
growth as described in the preceding expression and volume fraction would be higher in the case
was incorporated by using the Rice and Tracey of biaxial tension compared to uniaxial tension. Workability Tests for
analysis (Ref 39). This prior analysis led to the The suitability of this approach was validated to Flow-Localization-Controlled Failure
determination of a measure of the cavity growth a first order by a comparison to the cavitation
rate known as the void growth factor, D: observed in Ti-6Al-4V tension samples with Nonisothermal Upset Test. Workability
sharp necks and thus high levels of hydrostatic problems in hot and warm open-die forging that
R d ln R stress (Ref 40) (Fig. 18b). In contrast, by reduc- are related to flow localization are very com-
D= = (Eq 2) ing the level of hydrostatic stress, cavity growth mon, and various terms, such as chill zone,
e R de
can be suppressed. An example of such an effect dead-metal zone, and locked metal, have become
is the elimination of cavitation in a gamma- synonymous with defects that result from this
or titanium-aluminide alloy by using superimposed phenomenon. In open-die forging, where lateral
pressure during hot forging (Ref 41). constraint to metal flow is minimal, the occur-
R Ro exp(D e) (Eq 3) As described in Chapter 7, Hot Tension rence of flow localization is a result of heat
Fig. 18 Effect of stress state on cavitation. (a) Dependence of the void growth factor, D, on the ratio of the mean to effective stress (sm/s). Source: Ref 38, 39. (b) Comparison of
measured and predicted values of the ratio of the cavity growth parameter at the center and edge (c/e) of Ti-6Al-4V hot tension specimens as a function of the local
value of the ratio of sample radius to profile radius (a/R). Source: Ref 40
transfer between the hot workpiece and cold dies (and thus strain) gradients: the material flow- pressing test may be attributed to two main
or of excessive friction. As such, it is relatively stress dependence on temperature, the material factors. First, the amount of contact area in side-
easy to devise workability tests for investigating strain-rate sensitivity, and the magnitude of the pressing of round bars starts at zero (i.e., line
and resolving these problems when they occur. temperature gradient. The first two are proper- contact) and builds up rather slowly at the be-
Probably the simplest workability test that can ties that are measured in isothermal hot com- ginning of deformation. Second, because the op-
be employed to study the effects of heat transfer pression tests, whereas the third can be predicted eration is basically a plane-strain operation, sur-
on flow localization is the nonisothermal upset from heat-transfer analysis. For Ti-6242Si, the faces of zero extension, along which block
test. As in the uniform, isothermal hot compres- dependence of flow stress on temperature at typ- shearing can initiate and propagate, are present
sion test, a cylindrical workpiece is upset ical forging temperatures is very strong; also, the in sidepressing. These surfaces are the natural
between flat, parallel dies. Workpiece and die rate sensitivity is modest (Ref 4244). The heat- ones along which shear strains can concentrate
temperatures, as well as working speed, lubrica- transfer analysis incorporates the densities and into shear bands. Formation of well-defined chill
tion, and dwell time on the dies prior to forging, thermal properties (specific heat, thermal con- zones prior to shear-band formation would prob-
can be varied. Following deformation, speci- ductivity) of the dies and the workpiece, the
mens are sectioned and metallographically pre- initial die and workpiece temperatures, the de-
pared in order to determine the extent of chilling formation rate, and the heat-transfer coefficient
and formation of shear bands (regions of intense characterizing the interface. Simplifying as-
localized deformation) between the chill zones sumptions can be made regarding heat genera-
and the remainder of the deforming bulk. As tion and heat transport that make a closed-form
shown in Fig. 19 for several Ti-6Al-2Sn-4Zr- solution possible. With these data and assump-
2Mo-0.1Si (Ti-6242Si) specimens that had a tions, the extent of chill zones in simple tests
starting equiaxed-alpha microstructure, the chill such as those done on Ti-6242Si (Fig. 19) can be
zones are usually revealed as areas that etch dif- predicted with fairly good accuracy (Ref 42).
ferently than the rest of the specimen. The nonisothermal sidepressing test is an-
By varying the process variables in a series of other workability technique that can be used to
nonisothermal upset tests, the extent of chill gage the interactions of material properties and
zones and the severity of shear banding can be process variables during flow localization in hot
estimated. These results may be interpreted forging. As in the nonisothermal upset test, sev-
using a variety of mathematical techniques, such eral test specimens are sidepressed between flat,
as the upper-bound technique (Ref 4) and a sim- parallel dies at a variety of workpiece tempera-
plified analytical treatment based on a one- tures, die temperatures, working speeds, and so
dimensional heat-transfer analysis and the on, and the formation of defects is determined Fig. 19 Axial cross sections of specimens of Ti-6Al-
stress-equilibrium equation (Ref 42). Although by metallography. Unlike the upset test, with its 2Sn-4Zr-2Mo-0.1Si with an equiaxed-alpha
very approximate in nature, the latter technique axisymmetric chill zones, however, flow local- starting microstructure that were nonisothermally upset at
954 C (1750 F ) to 50% reduction in a mechanical press
is capable of quantifying the three most impor- ization is manifested primarily by shear-band (e- 30 s1) between dies at 191 C (375 F ). Dwell times
tant material/processing components that deter- formation and propagation. The absence of well- on the dies prior to deformation were (a) 0 s and (b) 5 s.
mine the magnitude and extent of strain-rate defined chill zones in the nonisothermal side- Source: Ref 42
ably involve more deformation work and hinder The effect of lower working speed, and there- effect may occur during hot working of metals
the propagation of shear bands. This is not to fore greater heat transfer, on shear-band forma- because of dynamic recrystallization, an unsta-
say, however, that temperature gradients are not tion in sidepressing was also determined in ble microstructure, changes in texture, redistrib-
important in supplying a driving force for flow sidepressing tests on Ti-6242Si and serves to il- ution of solid-solution elements, changes in the
localization. lustrate the extremes in behavior that can be in- morphology of second phases, and so on (Ref
Sidepressing results for the Ti-6242Si alloy il- vestigated by means of this test. These experi- 45).
lustrate the kinds of localization behavior that ments were conducted in a hydraulic press The constant-strain-rate, isothermal hot com-
can be observed (Ref 42). In Fig. 20, cross sec- (strain rate 1 s1); for a given reduction, depression test (see Chapter 6, Hot Compression
tions of bars sidepressed in a mechanical press formation thus lasted approximately 30 times as Testing) is the most useful workability test that
(strain rate 30 s1), using dies preheated to long as in the mechanical-press trials. At low re- can be used to detect or predict flow localization
191 C (375 F), are shown. When the bars were ductions, shear bands similar in appearance to during forging in which the dies are preheated to
preheated to a temperature much below the tran- those observed previously were formed in the the same temperature as the workpiece, that is,
sus temperature (913 C, or 1675 F), at which hydraulic-press specimens (Fig. 22). On the during isothermal forging. Flow-stress data and
the flow stress is very sensitive to temperature, other hand, at high reductions, shear cracks were observations are useful in establishing guide-
shear bands whose intensity increases with re- found to develop along these shear bands. The lines for flow localization for various alloy sys-
duction were developed (Fig. 20ac). It has been same effect was found regardless of whether the tems and various deformation states. To this end,
found that, in sidepressing, the propagation workpiece preheat temperature was much below the flow-stress data are analyzed to determine
stages of shear banding appear to follow an the transus temperature or rather close to it (Ref the normalized flow-softening rate,
(1/s)
identical pattern, regardless of die temperature, 42), a trend that, as has been mentioned, was ds/de, where s(e) is the flow curve determined
working speed, and so on, which points out the significant for the mechanical-press tests. From under constant-strain-rate conditions, and the
importance of geometry in determining the these results, it may be concluded that long de- strain-rate-sensitivity exponent is m logs/
shapes of the shear bands. These so-called prop- formation times may drop the average work- loge at fixed strain and temperature, where e de-
agation stages are summarized somewhat piece temperature from one at which the flow notes strain rate. This determination is made as a
schematically in Fig. 21, which is discussed stress is not too temperature-sensitive to one at function of strain for flow curves measured at a
more thoroughly in Ref 42. In contrast to the which it is and at which thermal gradients may variety of strain rates and temperatures. It has
previously mentioned results, no shear banding lead to shear bands, and long deformation times been found that for materials that exhibit c
was observed (Fig. 20df ) when the workpiece many drop the average workpiece temperature (
1)/m 5 at a given strain, strain rate, and
temperature is approximately 982 C (1800 F), into a regime of low ductility. The latter conclu- temperature, nonuniform flow in uniaxial com-
which is close to the beta transus temperature for sion explains why cracking was observed in pression is likely (Ref 47, 48).
this alloy. At this temperature, the dependence of hydraulic-press trials, inasmuch as process- The previously mentioned conclusion is sup-
flow stress on temperature is minimal. Hence, simulation results revealed that the average ported by data for a number of alloys, such as
the high strain rates employed in this instance, in workpiece temperature had indeed fallen to a Ti-6242Si, Ti-l0V-2Fe-3Al, and nickel alloy
conjunction with the low-temperature sensitivity level at which ductility is low. 700. Observations for Ti-6242Si deformed at a
of the flow stress, may be surmised to have min- Isothermal Compression and Sidepressing temperature of 913 C (1675 F) and a strain rate
imized stress gradients, and hence strain gradi- Tests. It has also been found that flow localiza- of 2 s1 are shown in Fig. 23 (Ref 47). When the
ents, during sidepressing. tion may occur during hot forging in the absence
of chilling or frictional effects (Ref 43, 45, 46).
In these cases, localization is a result of flow
softening, or negative strain hardening. Such an
Fig. 20 Transverse metallographic sections of speci-

mens of Ti-6Al-2Sn-4Zr-2Mo-0.1Si with an
equiaxed-alpha starting microstructure that were non- Fig. 22 (a, b) Transverse metallographic sections and
isothermally sidepressed with zero dwell time in a me- (c) micrograph of region with shear band and
chanical press (e- 30 s1) between dies heated to 191 C crack from section shown in (b) of Ti-6Al-2Sn-4Zr-2Mo-
(375 F ). Specimen preheat temperatures, Ts, and percent 0.1Si specimens with an equiaxed-alpha starting mi-
reductions, R (relative to the initial specimen diameter), crostructure that were nonisothermally sidepressed in a
were (a) Ts 913 C (1675 F ); R 14, (b) Ts 913 C hydraulic press (e 30 s1). Specimen preheat tempera-
(1675 F ); R 54, (c) Ts 913 C (1675 F ); R 77, (d) Fig. 21 Schematic representation of the mechanism of ture, die temperature, and dwell time were 913 C (1675
Ts 982 C (1800 F ); R 21, (e) Ts 982 C (1800 F ); shear-band formation in sidepressing. Source: F ), 191 C (375 F ), and 14 s, respectively. Reductions
R 57, (f ) Ts 982 C (1800 F ); R 79. Source: Ref 42 Ref 42 were (a) 25% and (b) 53%. Source: Ref 42
alloy had an equiaxed-alpha microstructure, c at 704 C (1300 F), the maximum values of c initiate and propagate in a manner similar to that
values less than zero were estimated from the over the strain interval from 0 to 0.7 were ap- previously discussed. In addition, in analogy to
flow-stress data, and upset testing gave rise to proximately 2.5 (e 103 s1), 5 (e 101 the c 5 criterion for nonuniform flow in
uniform deformation at this temperature and s1), and 5 (e 10 s1). At 816 C (1500 F), isothermal compression, it has been found that
strain rate (Fig. 23a). In contrast, when the alloy c was less than zero at strain rates of 103, shear bands in isothermal plane-strain operations
had a nonequilibrium colony-alpha microstruc- 101, and 10 s1. Observation of nonuniform such as sidepressing typically occur when pro-
ture, the flow curves exhibited large amounts of flow and flow localization at 704 C (1300 F) cessing conditions are such that p 5 (Ref 46).
flow softening. At 913 C (1675 F) and e 2 and e 101 and 10 s1 and the absence of such Figures 25 and 26 illustrate this behavior for al-
s1, c values of approximately 5 were esti- localization at 704 C (1300 F) and e 103 loys Ti-6242Si and Ti-l0V-2Fe-3Al, respectively.
mated using the flow curves; the very nonuni- s1 and at 816 C (1500 F) and e 103, 101, Figure 25 also shows the results of a FEM simu-
form flow observed in compression (Fig. 23b) and 10 s1 (Fig. 24) are in line with the previous lation of sidepressing of Ti-6242Si that confirms
was ascribed to the high levels of c. It is also in- results for Ti-6242Si. the usefulness of the p-parameter correlation.
teresting to note that the nonuniform flow ob- A related parameter, p
/m, where
and Information on the effects of temperature and
served in isothermal compression is analogous m have the same definitions as before, may be strain rate on flow localization, such as that ob-
to the formation of chill zones in nonisothermal used to estimate the tendency for flow localiza- tained in isothermal compression and sidepress-
upsetting, inasmuch as it is evidenced on a bulk tion in plane-strain forging operations such as ing tests, is often summarized in workability
scale rather than as an internal defect similar to sidepressing. As in nonisothermal sidepressing, diagrams such as that shown in Fig. 27 for the oc-
a shear band. localization under isothermal conditions due to currence of shear bands in plane-strain sidepress-
The occurrence of nonuniform deformation flow softening is in the form of shear bands that ing of Ti-6242Si (Ref 46). Closed and open cir-
during hot compression testing of beta-annealed cles, denoting process conditions for which shear
Ti-l0V-2Fe-3Al further demonstrates the appli- bands were and were not observed, are noted on
cability of the c parameter in predicting the diagrams, one for an equiaxed-alpha mi-
nonuniform flow. When this alloy was deformed crostructure and the other for a colony-alpha mi-
crostructure obtained via beta annealing. It can be
seen that the two regions (called Safe and
Fail in analogy to sheet metal forming-limit di-
agrams) are fairly well delineated by lines that
correspond to temperature and strain-rate condi-
tions for which p 5 pertains over a large range
of strain. Workability diagrams for other hot
working defects, such as cavitation and triple-
point cracking, may also be obtained from exper-
imental workability tests as well as predicted
from mechanistic models (Ref 49).
Workability Tests for Cold Open-

Die Forging of Recrystallized
Structures
As the forging temperature is lowered into the
cold working regime (i.e., to temperatures less
than or equal to one-fourth of the melting or
solidus temperature), workability problems
Fig. 23 Specimens of Ti-6Al-2Sn-4Zr-2Mo-0.1Si from
change from failure related largely to intergran-
isothermal hot compression tests at 913 C

(1675 F ), e 2 s1. Starting microstructures were (a) Fig. 25 (a,b) Transverse metallographic sections of ular cracking and cavitation to failure by a mode
bars of Ti-6Al-2Sn-4Zr-2Mo-0.1Si
equiaxed alpha and (b) colony alpha. Source: Ref 47 isothermally sidepressed at 913 C (1675 F ), e 2 s1. (c,
involving transgranular ductile fracture or shear
d) FEM-simulation predictions of contours of constant banding. Workability tests used to diagnose
strain rate. Specimen in (a) and simulation in (c) are for the these problems are discussed in turn.
alloy with an equiaxed-alpha starting microstructure (no Workability Tests for Ductile Fracture
shear bands), and specimen in (b) and simulation in (d) are
for the alloy with a colony-alpha starting microstructure
During Cold Forging. Ductile fracture consists
(that exhibited shear bands). Source: Ref 46 of the nucleation of cavities at second-phase
particles and inclusions and the growth and even-
tual coalescence of these cavities under the ac-
tion of tensile stresses to cause fracture. In cold
forging, ductile fracture occurs most readily at
free surfaces where barreling has set up second-
ary tensile stresses. A simple workability test for
analyzing these kinds of failures is the upset test
in which various combinations of tensile and
compressive stress fields are developed by use of
Fig. 24 Specimens of Ti-10V-2Fe-3Al from isothermal various lubrication conditions and initial speci-
hot compression tests at (a, b, c) 704 C men height-to-diameter ratios. From tests of this
(1300 F ) and (d, e, f) 816 C (1500 F). Strain rates were sort, it has been found that surface fracture often
(a, d) 103 s1, (b, e) 101 s1, and (c, f ) 10 s1. Prior to Fig. 26 Transverse metallographic section of specimen
testing, the alloy had been beta annealed to yield an of Ti-10V-2Fe-3Al isothermally sidepressed at follows a critical strain criterion, e1 ae2 con-
equiaxed-beta starting microstructure. 704 C (1300 F), e 10 s1, which exhibited shear bands stant, where e1 and e2 are the tensile and com-
Fig. 27 Workability maps for occurence of shear bands in isothermal sidepressing of Ti-6Al-2Sn-4Zr-2Mo-0.1Si with (a) equiaxed-alpha starting microstructure and (b) colony-
alpha starting microstructure. Workability predictions based on p 5 (). Forging conditions in which shear bands were () and were not (o) observed are noted. Source:
Ref 46
pressive principal surface strains at fracture, re- material. In this regard, it is similar to the previ- do
spectively, and a denotes the slope, usually equal ously discussed notched-upset test used for hot
to 12 (Ref 50, 51). This test technique and the workability. It has been suggested (Ref 54) that
resulting criterion are discussed in detail in this test be used for establishing an index of sur-
Chapter 5, Cold Upset Testing. Determination face quality for steel wire used in cold heading.
of values of the constant in this fracture criterion The truncated-cone test (Ref 58) was devel-
is very useful in ranking the resistance of various oped in an attempt to minimize the effects of
materials and different lots of a given material to surface flaws and the random errors they pro-
ductile fracture. Because the determination of duce on workability data. Unlike the simple
such fracture loci involves surface strains, a mod- upset and grooved-upset tests, the truncated-
ified upset technique, known as the collar test, cone test involves indentation by a conical tool zo
has been developed (Ref 52, 53). In this method, of a cylindrical specimen whose initial diameter
circular cylinders with a flange (collar) at mid- is greater than that of the truncation (Fig. 29).
height (Fig. 28) are upset to the point of surface By this means, cracking is made to occur
fracture. The hoop (tensile) and axial (compres- beneath the surface of the workpiece at the tool-
sive) strains are determined simply by measuring material interface, thereby eliminating the influ-
the final diameter and thickness of the flange. ences of free-surface flaws and surface finish on
Several tests other than those employing sim- workability limits. The reduction (measured at
ple upsetting of cylindrical specimens have been the specimen axis) at which cracking occurs
devised to obtain a measure of workability may be used to compare the workability of dif-
in cold forging. These include the grooved- ferent materials. Alternatively, the reduction at Fig. 28 Typical specimen geometry for the collar test.
compression test and an indentation-type method which a fixed crack width is obtained or the Source: Ref 52
using a truncated cone. Both are useful for deter- width of the crack at a given reduction has been
mining the deformation responses of metals, suggested as an index of workability. play a large role in such failures. In addition, the
such as low-carbon steels, that do not readily As for hot working, FEM modeling provides voids generated by these stresses may form
fracture in ordinary upsetting. In the grooved- a useful tool for the evaluation of the workabil- gross centerbursts, as in hot forging. Methods of
compression test (Ref 5457), cylindrical speci- ity of complex cold forgings. Detailed stress, establishing the sources of centerbursts are dis-
mens with various initial aspect ratios and with strain, and strain-rate histories are predicted and cussed in Chapter 21, Workability and Process
axial grooves are upset under various conditions can be coupled with an appropriate fracture cri- Design in Extrusion and Wire Drawing, inas-
of lubrication in much the same way that stan- terion, such as the ductile-fracture relation of much as these are the principal operations in
dard upset testing for workability analysis is con- Cockcroft and Latham (Ref 36) or that of Lee which these defects occur.
ducted. Fracture is initiated in the groove at the and Kuhn (Ref 51) (which is derivable from the Workability Tests for Shear Banding dur-
specimen midheight (where tensile stresses are Cockcroft and Latham criterion), to provide ing Cold Forging. Shear bands are defects that
greatest), and the average circumferential frac- overall deformations at which failure should may be found in cold forging as well as in hot
ture strain at that point is taken as a measure of occur. The usefulness of such a method has been forging. In open-die forging, shear-band forma-
workability. Available data also show that, for a described (Ref 59), and researchers successfully tion may be attributed almost wholly to the for-
given material, the value of this circumferential predicted critical reductions for surface fracture mation of dead-metal zones as a result of high
fracture strain is independent of initial groove during upsetting of aluminum alloy 7075-T6 levels of friction at the die-workpiece interface.
depth; however, as groove depth is increased, the using a FEM simulation and a measured fracture Therefore, workability techniques are relied on
axial reduction at which this strain is reached is criterion of the Lee-Kuhn type. to evaluate the effects of lubricants on friction
diminished. For this reason, the grooved-com- Ductile fracture may also occur in the work- and metal flow.
pression test appears to be useful for establishing piece interior during cold forging (Ref 60). As in The most common test for evaluation of fric-
the effects of surface flaws on forgeability and in free-surface fracture and centerbursting at hot tion is the ring compression test (Ref 61)
ranking the quality of different lots of the same working temperatures, secondary tensile stresses (Chapter 6, Hot Compression Testing). In this
test, rings of the workpiece material are upset to

various reductions using different lubricants.
With good lubrication, the inside diameter of the
ring increases, whereas the reverse occurs under
conditions of poor lubrication. Through meas-
urements of reduction and percentage change in
inside diameter, a quantitative measure of lu-
bricity may be obtained. This measure may be in
terms of a coefficient of friction, , or a friction
factor, m, depending on the analysis used to re-
duce the test data. In addition, the flow curve,
with corrections for friction, can also be deter-
mined through analysis of ring-test data.
Other techniques for evaluation of friction in-
clude those in which deformation patterns are
visualized through the use of grid wires inserted,
or small-diameter holes drilled, into the work-
piece (Fig. 30). Either method may be used to
study the formation of dead-metal zones. As
shown in Fig. 30, the dead-metal zones in sim-
ple upsetting form early in deformation and tend
not to deform at all until the reduction is high
enough to bring them into contact. Prior to this,
shear bands may develop between the dead-
metal zones and the deforming bulk, moving ra-
dially outward. Once the dead-metal zones come
into contact, however, deformation becomes
more uniform, and high levels of hydrostatic
compression are set up. This stress state tends to
close holes in test specimens with such inten-
tional initial defects (Fig. 30).
Fig. 29 Relationship between crack width and stroke in truncated-cone indentation test for workability of various
steels at cold forging temperatures. Source: Ref 58
Workability in Closed-Die Forging
The defects mentioned previously may also

occur in closed-die or impression-die forging of
metals. Thus, the workability tests discussed
previously may be used to evaluate the forma-
tion of defects under conditions in which more
constraint is applied to metal flow. In addition,
other defects occur in closed-die forging that
may or may not involve formation of fractures or
regions of localized flow. These defects often re-
sult from factors such as improper choice of the
starting (or preform) shape of the workpiece that
is used in the forging operation, poor die design,
and poor choice of lubricant or process vari-
ables. All of these may contribute to formation
of flaws such as laps, flow-through defects, ex-
trusion defects, cold shuts, and so on.
Laps are defects that form when metal folds
over itself during forging (Ref 2, 64). This may
occur in rib-web forgings at a variety of locations.
One such location is the web in a forging in which
the preform web is too thin. During finish forging,
such a web may buckle, causing a lap to form
(Fig. 31). Another location is a rib in which metal
is made to flow nonuniformly (Fig. 32, 33). Most
often, a lap at this location is a result of an exces-
sively sharp fillet radius in the forging die.
Flow-through defects are flaws that form
when metal is forced to flow past a recess after
it has filled or has ceased to deform because of
chilling (Ref 2) (Fig. 34, 35). Similar to laps in
appearance, flow-through defects may be shal-
Fig. 30 Deformation patterns in cylindrical billets upset between flat, parallel dies. Source: Ref 2, 62, 63 low but are indicative of an undesirable grain-
flow pattern or shear band that extends much Flow-through defects may also occur when As summarized in Ref 64 and 69, three pri-
deeper into the forging. trapped lubricant forces metal to flow past an mary techniques are available for investigation
A special compression test using cylindrical impression. and correction of metal-flow defects such as
samples with a reduced section (Fig. 36a) has Extrusion-type defects are formed when cen- those described previously. These techniques in-
been developed to quantify the effect of material trally located ribs formed by extrusion-type volve the application of empirical guidelines in
properties on the severity of flow localization flow draw too much metal from the main body designing preforms and finish-forging dies,
that may give rise to flow-through and other or web of the forging. A defect similar to a pipe physical modeling studies, and computer-aided
kinds of defects (Ref 67, 68). Following forging, cavity is thus formed (Fig. 37) (Ref 2). Means design techniques. Computer-aided design tech-
a parameter know as the distributed gage volume of minimizing the occurrence of these defects niques are useful in the calculation of required
(DGV) is calculated from measurements of the include increasing the thickness of the web or volume distributions in forging preforms and the
original gage volume, Vo, and final apparent designing the forging with a small rib opposite determination of neutral planes and general
gage volume, Vf, between the specimen ends, the larger rib. metal-flow patterns. With modern FEM tech-
that is, DGV [(Vo Vf)/Vo] 100%. The Most of the defects just summarized are found niques, preform geometries and metal-flow pat-
DGV provides a measure of the ability to dis- in hot forging, which is most common for im- terns that avoid defect formation in complex
tribute deformation. Greater ability to distribute pression-die forging. Because of this, defect for- forgings are readily established.
deformation results in greater penetration of the mation may also involve entrapment of oxides Empirical Guidelines for Die and
reduced section into the specimen ends and thus (and lubricant). When this occurs, the metal is Preform Design. Several simple guidelines for
higher values of DGV. Processing conditions incapable of rewelding itself back together design of forging dies and preforms to avoid
that produce high values of DGV are attractive, under the high forging pressures. The term cold
in that they are least likely to give shear bands, shut is often applied in conjunction with laps,
flow-through defects, and so on. Typical results flow-through defects, and so on in describing the
from the test are shown in Fig. 36(b). flaws that are generated.
Fig. 31 Typical deformation sequence in closed-die forging of a rib-web part, showing how laps can be generated if Fig. 32 Lap formation in the rib of a rib-web part due
preform geometry is selected improperly. Source: Ref 2 to improper preform geometry. Source: Ref 65
defect formation have been developed through Draft angles are needed to allow removal of angles (7 or greater) because of higher forging
years of experience. By and large, these guide- parts after forging; they are generally in the pressures and a greater tendency for sticking in
lines are for conventional hot forging in which range of 1 to 7, with lower values being used the dies (Ref 2).
the dies are much cooler than the workpiece. when part-knockout mechanisms are employed Radii are also important in die design, be-
From the viewpoint of dies, the most important or when surface area must be minimized because an insufficient radius may lead to lap for-
design parameters are draft angles, corner and cause of frictional or chilling effects. In addi- mation or incomplete die fill. As with draft an-
fillet radii, and rib and web dimensions (Ref tion, easily forged materials, such as aluminum gles, required radii increase with increasing
13). Some of these guidelines are summarized alloys, tend to require small draft angles (1), forging difficulty of the workpiece alloy. Typical
subsequently. More information on this topic carbon and alloy steels require moderate draft values for making 2.5 cm (1 in.) high ribs are
can be found in Chapter 14, Process Design in angles (3 to 5), and hard-to-work alloys, such given in Table 2. Guidelines for minimum di-
Impression Die Forging. as nickel-base superalloys, need large draft mensions of ribs and webs in conventional hot
forgings may also be suggested, based on expe-
rience (Table 3) (Ref 2). The important or decid-
ing factors here are the cooling conditions of the
forging, which depend on the workpiece alloy
and deformation speed, among other things, and
the minimum web dimension, which determines
the amount of lateral constraint.
Because the formation of metal-flow defects
also depends a great deal on the shape of pre-
forms, it is not surprising that guidelines for
preform design have been devised. Primarily in-
tended for conventional hot forging, these guide-
lines provide a basic idea of how much metal
flow can be imposed under typical forging con-
ditions. A number of useful relations between
preform and finish forging dimensions have
been suggested for steel, aluminum-alloy, and ti-
tanium-alloy rib-web structural forgings (Ref 1).
Those for aluminum and titanium alloys are
summarized in Table 4.
Physical Modeling. Although FEM com-
puter modeling has become the most common
technique for the design of preform and finish
forging dies, an alternative approach, physical
modeling, is still finding use as a heuristic tool
as well as a method to validate the detailed flow
patterns predicted by FEM analyses. Based on
the application of the principles of similarity,
physical modeling makes use of small-scale dies
and model materials.
For the model system to have applicability, it
is best that the laws of similarity be adhered to.
The laws of similarity consist of rules of geo-
metric, physical, and boundary-condition simili-
tude. From a geometric standpoint, model dies
Fig. 33 Lap defect in Ti-6Al-4V bulkhead forging. Source: Ref 66 must be designed so as to approach, as nearly as
possible, exact scale models of the envisioned
forging diesthat is, all linear dimensions,
radii, angles, and so forth of the model dies must
be in fixed proportions to those of the real tool-
ing. A similar relation between model and actual
workpiece dimensions should be followed.
Physical similarity pertains largely to selection
of a model material whose deformation and ther-
mal response are similar to those of the intended
workpiece material. Boundary-condition simi-
larity is related to interface friction and heat-
transfer considerations.
In physical modeling of forging, model spec-
imens/workpieces are usually sectioned and
placed back together after a grid pattern has
been inscribed on one of the flat faces. By this
means, the flow pattern may be visualized, and
regions where defects are likely to form in an
actual forging may be determined. Such a tech-
Fig. 34 Metal-flow patterns in forgings without and with flow-through defects. Source: Ref 2 nique is similar to the grid technique used
on for diagnosing problems in sheet metal With these materials, model dies need not be of hand, it may be possible to adjust the frictional
forming. high strength and can be made from inexpensive boundary conditions to compensate for this
Physical modeling has been applied often to steels, aluminum alloys, or even hard plastics. deficiency.
simulation of both conventional and isothermal Because the tests are usually conducted at room The work described in Ref 74 provides a good
hot forging (Ref 7073). These studies are usu- temperature, boundary-condition similarity with example of the usefulness of physical modeling.
ally performed at room temperature with model regard to heat transfer is satisfied for isothermal In this work, densification behavior during the
materials that are rate-sensitive and non-strain- but not for conventional hot forging. Therefore, forging of ferrous-powder compacts was ration-
hardening, as are typical forging alloys at hot information regarding the possible effects of alized based on the deformation of wedge sam-
working temperatures. Lead, plasticine, and wax chilling on metal flow may not be obtainable ples made from plasticine (Fig. 38). Variations in
are the most common choices for materials. from simple modeling studies. On the other both normal and shear strains, readily apparent
on the surface and cross sections of the plas-
ticine samples, were used to explain density gra-
dients in actual forgings.
Figure 39 gives an example of how a lap de-
fect is visualized in a physical model of a
closed-die-forging process (Ref 75). Here, the
grid lines give valuable information not only on
the development of the lap in the forging but also
on the occurrence of a dead-metal zone in con-
tact with the bottom die. Similar work has been
done in demonstrating the occurrence of laps in
an engine impeller part (Ref 72). It was shown
by the use of the model forgings that the laps
could be avoided by changes in preform
design.
Fig. 35 Flow-through defect in Ti-6Al-4V rib-web structural part. Source: Ref 66
Fig. 37 Extrusion-type defect in centrally located rib

(top), and die-design modification used to
avoid defect (bottom). w, width; h, height; t, thickness.
Source: Ref 2
Table 2 Fillet and corner radii suggested

for forgings of several alloys with 2.5 cm
(1 in.) high ribs
Fillet radius, Corner radius,
Alloy mm mm
2014 (aluminum) 6 3
AISI 4340 (alloy steel) 1013 3
H11 (hot work steel) 1013 5
17-7PH (stainless steel) 613 5
A286 (superalloy) 1319 6
Ti-6Al-4V 1316 6
Unalloyed molybdenum 13 13
Fig. 36 The reduced-section compression test. (a) Typical specimen geometry. (b) Specimens of JBK-75 hot forged at Note: From the viewpoint of the forging or workpiece. Source: Ref 2
650 C (1200 F ) (top) or 760 C (1400 F ) (bottom). Source: Ref 67, 68
Summary available. These methods include physical mod-

eling, in which easy-to-deform model materials,
Workability in forging depends on a variety of such as lead, plasticine, and wax, are deformed
material, process-variable, and die-design fea- in inexpensive model dies. In such a way, fea-
tures. Because of the almost unlimited range of tures of preform or die design that may cause
possible forging operations, the forging engineer laps or other undesirable defects may be estab-
must use a variety of workability tests to diag- lished economically. The use of computer simu-
nose forging problems. For example, open-die lation techniques to diagnose workability prob-
forging operations do not impose the same lems in both open- and closed-die forging has
amount of lateral constraint on the workpiece as become widespread with the advent of inexpen-
closed-die forging, and thus, different tech- sive computers and powerful FEM software.
niques are used in the evaluation of workability With such tools, the forging engineer can readily
for open-die forging and closed-die forging. determine stresses and metal-flow patterns and
Workability may also be associated with equip- thus predict the occurrence of metal-flow de-
ment use (e.g., press capacity, die materials, die, fects, gross shear bands, voids, and so on.
etc.) and evolution of microstructure, as dis-
cussed in other chapters of this Handbook.
For open-die forging, the wedge-forging (or
rolling), sidepressing, and double-cone tests are
useful for determining optimal processing pa-
rameters for breaking down cast structures and
for avoiding fracture. For forging of wrought
and recrystallized structures, a number of tests
are available for establishing the effects of
process variables, such as temperature and de-
formation rate, and of secondary tensile stresses
on working limits controlled by fracture or flow
localization preceding fracture. These include
standard tension, compression, and sidepressing
tests as well as variations of these tests intro-
duced to simulate chilling (e.g., nonisothermal
compression test) and the development of com-
plex stress states (notched-upset test, double-
cone test).
Techniques for predicting the occurrence of
forging defects in closed-die forging, which may
not involve fracture or flow localization, are also
Fig. 38 Physical modeling of the wedge-forging test.
(a) Preform geometry. (b) Plan view, longitudi-
nal section, and transverse section of forged plasticine
wedge sample. Source: Ref 74
Table 3 Suggested minimum section thicknesses for rib-web forgings of several alloys
Minimum rib thickness(a), Minimum web thickness(a),
mm, for forgings of mm, for forgings of
given plan area, cm2, of: given plan area, cm2, of:
Alloy Up to 25 25 to 250 250 to 750 Up to 25 25 to 250 250 to 750
2014 (aluminum) 3 5 ... 3 6 8

AISI 4340 (alloy steel) 3 5 ... 5 8 13
H11 (hot work steel) 3 5 ... ... 10 ...
17-7PH (stainless steel) ... 6 ... ... 10 ...
A286 (superalloy) ... 6 ... ... 10 ...
Ti-6Al-4V ... 6 ... ... 8 ...
Unalloyed molybdenum ... 10 ... ... 10 ...
(a) Values listed are for rib-web forgings with 7 draft and standard fillet and corner radii. Source: Ref 2
Table 4 Relations between preform and finished forging dimensions for rib-web
structural parts
Dimensions in preform(a)
Dimensions in finished
forgings(a) Aluminum alloys Titanium alloys
Web thickness, tF tp (11.5) tF tp (1.52.2) tF

Fillet radii, RFF RPF (1.22) RFF RPF (23) RFF
Corner radii, RFC RPC (1.22) RFC RPC (2) RFC
Draft angle, F p F (25) p F (35) Fig. 39 Forging sequence of gridded lead billet used in
Width of rib, WF W p WF 0.8 mm W p WF (1.63.2 mm) physical modeling studies to establish the ef-
fect of die and preform design on the occurrence of metal-
(a) The first subscript letter for each dimension indicates finish forging (F) or preform (P). Source: Ref 1 flow defects such as laps. Source: Ref 75. Courtesy of S.
Kobayashi
REFERENCES 15. J.J. Jonas, Recovery, Recrystallization, 30. R.P. Daykin, Ladish Co., unpublished re-
and Precipitation Under Hot-Working search, 22 June 1951
1. T. Altan, F.W. Boulger, J.R. Becker, N. Conditions, Proc. Fourth Int. Conf. on the 31. Anonymous, High-Temperature, High-
Akgerman, and H.J. Henning, Forging Strength of Metals and Alloys, 30 Aug3 Strength Nickel-Base Alloys, The Inter-
Equipment, Materials, and Practices, Sept 1976 (Nancy, France), p 976 national Nickel Company, Inc., 1977
Report MCIC-HB-03, Metals and Ceramics 16. M. Jackson, R.J. Dashwood, L. 32. J.C. Malas, H.L. Gegel, S.I. Oh, and G.D.
Information Center, Battelles Columbus Christodoulou, and H.M. Flower, Applica- Lahoti, Metallurgical Validation of a Finite-
Laboratories, Columbus, OH, Oct 1973 tion of Novel Technique to Examine Ther- Element Program for the Modeling of
2. A.M. Sabroff, F.W. Boulger, and H.J. momechanical Processing of Near Alloy Isothermal Forging Processes, Experi-
Henning, Forging Materials and Practices, Ti-10V-2Fe-3Al, Mater. Sci. Technol., Vol mental Verification of Process Models,
Rheinhold Book Corp., 1968 16, 2000, p 1437 American Society for Metals, 1983, p 358
3. J.A. Schey, Principles of Forging Design, 17. A.L. Hoffmanner, Plasticity Theory as 33. S.I. Oh, Finite Element Analysis of Metal
American Iron and Steel Institute, 1965 Applied to Forging of Titanium Alloys, Forming Problems with Arbitrarily Shaped
4. B. Avitzur, Metal Forming: Processes and paper presented at the Symposium on the Dies, Int. J. Mech. Sci., Vol 24, 1982, p 479
Analysis, McGraw-Hill, 1968 Thermal-Mechanical Treatment of Metals, 34. B. Antolovich and M. Evans, Predicting
5. R.E. Reed-Hill, Physical Metallurgy 1 May 1970 (London), Institute of Metals Grain Size Evolution of Udimet Alloy 718
Principles, Van Nostrand, 1973 18. E.P. Unksov, An Engineering Theory of During the Cogging Process Through the
6. F.N. Rhines and P.J. Wray, Investigation of Plasticity, Butterworth and Co., 1961 Use of Numerical Analysis, Superalloys
the Intermediate Temperature Ductility 19. S.L. Semiatin, Battelles Columbus Labora- 2000, T.M. Pollock, R.D. Kissinger, R.R.
Minimum in Metals, Trans. ASM, Vol 54, tories, unpublished research, March 1979 Bowman, K.A. Green, M. McLean, S. Olson,
1961, p 117 20. H. Gegel, S. Nadiv, and R. Raj, Dynamic and J.J. Schirra, Ed., TMS, 2000, p 39
7. W.A. Backofen, Deformation Processing, Effects on Flow and Fracture During 35. S.L. Semiatin, R.L. Goetz, E.B. Shell, V.
Addison-Wesley, 1972 Isothermal Forging of a Titanium Alloy, Scr. Seetharaman, and A.K. Ghosh, Cavitation
8. A.J. DeRidder and R. Koch, Forging and Metall., Vol 14, 1980, p 241 and Failure During Hot Forging of Ti-6Al-
Processing of High-Temperature Alloys, 21. K.A. Bywater and T. Gladman, Influence of 4V, Metall. Mater. Trans. A, Vol 30, 1999,
MiCon 78: Optimization of Processing, Composition and Microstructure on Hot p 1411
Properties, and Service Performance Workability of Austenitic Stainless Steels, 36. M.G. Cockcroft and D.J. Latham, Ductility
Through Microstructural Control, ASTM Met. Technol., Vol 3, 1976, p 358 and Workability of Metals, J. Inst. Met., Vol
STP 672, H. Abrams, G.N. Maniar, D.A. 22. R.E. Bailey, Predicting the Hot-Working 96, 1968, p 33
Nail, and H.D. Solomon, Ed., American Characteristics of Superalloys by Hot 37. A.K. Ghosh, D.H. Bae, and S.L. Semiatin,
Society for Testing and Materials, 1979, Tensile Testing, Met. Eng. Q., Vol 15, May Initiation and Early Stages of Cavity
p 547 1975, p 43 Growth During Superplastic and Hot Defor-
9. J.J. Park and S. Kobayashi, Three-Dimen- 23. N. Cederblad and N.J. Grant, Hot mation, Mater. Sci. Forum, Vol 304/306,
sional Finite-Element Analysis of Block Workability of Nimonic 115 as a Function 1999, p 609
Compression, Int. J. Mech. Sci., Vol 26, of Strain Rate, Metall. Trans. A, Vol 6, 1975, 38. P.D. Nicolaou and S.L. Semiatin, The Effect
1984, p 165 p 1547 of Stress Triaxiality on Tensile Behavior of
10. V.K. Jain, L.E. Matson, H.L. Gegel, and R. 24. R.J. Quigg, Effect of Phase Changes on the Cavitating Specimens, J. Mater. Sci., Vol 36,
Srinivasan, Physical Modeling of Metal- Workability and Mechanical Properties of 2001, p 5155
working Processes II: Comparison of Udimet 700, High Temperature Materials 39. J.R. Rice and D.M. Tracey, On Ductile
Viscoplastic Modeling and Computer II, G.M. Ault, W.F. Barclay, and H.P. Enlargement of Voids in Triaxial Stress
Simulation, Physical Modeling of Metal- Munger, Ed., Interscience Publishers, 1963, Fields, J. Mech. Phys. Solids, Vol 17, 1969,
working Processes, E. Erman and S.L. p 245 p 201
Semiatin, Ed., TMS, 1987, p 127 25. A. Laasraoui and J.J. Jonas, Recrystal- 40. P.D. Nicolaou and S.L. Semiatin, An
11. D.D. Bhatt, G.E. Meyer, and A.L. lization of Austenite After Deformation at Experimental and Theoretical Investigation
Hoffmanner, Optimum Parameters for High Temperatures and Strain Rates of the Influence of Stress State on
Deformation Processing of a Plasma-Arc Analysis and Modeling, Metall. Trans. A, Cavitation During Hot Working, Acta
Melted High-Nitrogen Stainless Steel, Vol 22, 1991, p 151 Mater., Vol 51, 2003, p 613
paper presented at the Soviet-American 26. M.C. Mataya, Simulating Microstructural 41. D. Watkins, H.R. Piehler, V. Seetharaman,
Symposium on Problems of Electro- Evolution During the Hot Working of Alloy C.M. Lombard, and S.L. Semiatin, Effect of
metallurgy and Welding, Sept 1977 (Kiev, 718, JOM, Vol 51 (No. 1), Jan 1999, p 18 Hydrostatic Pressure on the Hot Working
Ukraine) 27. A.A. Korshunov, F.U. Enikeev, M.I. Behavior of a Gamma Titanium Aluminide,
12. G.E. Dieter, Mechanical Metallurgy, Mazurskii, G.A. Salishchev, A.V. Muravlev, Metall. Trans. A, Vol 23, 1992, p 2669
McGraw-Hill, 1976 P.V. Chistyakov, and O.V. Dmitriev, Effect 42. S.L. Semiatin and G.D. Lahoti, The
13. G.F. Torkhov, Y.V. Latash, R.R. Fessler, of Method of High-Temperature Loading on Occurrence of Shear Bands in Noniso-
A.H. Clauer, E.E. Fletcher, and A.L. Transformation of Lamellar Structure in hermal, Hot Forging of Ti-6Al-2Sn-4Zr-
Hoffmanner, Development of Melting and VT9 Titanium Alloy, Russ. Metall., Vol 3, 2Mo-0.lSi, Metall. Trans. A, Vol 14, 1983,
Thermomechanical-Processing Parameters 1994, p 103 p 105
for a High- Nitrogen Stainless Steel Pre- 28. C.M. Sellars and Q. Zhu, The Effect of 43. S.L. Semiatin, G.D. Lahoti, and S.I. Oh,
pared by Plasma-Arc Remelting, J. Met., Strain Path on Mechanical Behaviour and The Occurrence of Shear Bands in Me-
Vol 30 (No. 12), Dec 1978, p 20 Microstructure, 20th Ris Int. Symp. on talworking, Material Behavior under High
14. E.E. Fletcher and R.R. Fessler, Character- Materials Science, J.B. Bilde-Srenson et Stress and Ultrahigh Loading Rates,
ization of an Ingot of High-Nitrogen al., Ed., Ris, 1999, p 167 Plenum Press, 1983, p 119
Stainless Steel Produced by Plasma-Arc 29. S.B.Davenport, R.L. Higginson, and C.M. 44. S.L. Semiatin and G.D. Lahoti, Deforma-
Remelting, paper presented at the Soviet- Sellars, The Effect of Strain Path on tion and Unstable Flow in Hot Forging of
American Symposium on Problems of Material Behaviour during Hot Rolling Ti-6Al-2Sn-4Zr-2Mo-0.lSi, Metall. Trans.
Electrometallurgy and Welding, 57 Sept of FCC Metals, Philos. Trans. R. Soc. A, Vol 12, 1981, p 1705
1977 (Kiev, Ukraine) (London) A, Vol 357, 1999, p 1645 45. J.J. Jonas and M.J. Luton, Flow Softening at
Elevated Temperatures, Advances in Defor- Mater. Process. Technol., Vol 60, 1996, p 269 telles Columbus Laboratories, Columbus,
mation Processing, J.J. Burke and V. Weiss, 56. H. Sueyoshi, K. Suenaga, and R. Tanaka, OH, July 1977
Ed., Plenum Press, 1978, p 215 Cold Forgeability and Machinability after 65. A. Chamouard, Closed-Die Forging, Part I,
46. S.L. Semiatin and G.D. Lahoti, The Oc- Cold Forging of Hypo-Eutectoid Graphitic Dunod, Paris, 1964 (in French)
currence of Shear Bands in Isothermal, Hot Steels, J. Jpn. Inst. Met., Vol 53, 1989, p 206 66. F.N. Lake and D.J. Moracz, Comparison of
Forging, Metall. Trans. A, Vol 13, 1982, 57. K. Olsson, S. Karlsson, and A. Melander, Major Forging Systems, Report AFML-
p 275 The Influence of Notches, Testing Geo- TR-71-112, TRW, Inc., Cleveland, May
47. S.L. Semiatin, G.D. Lahoti, and T. Altan, metry, Friction Conditions, and Micro- 1971
Determination and Analysis of Flow Stress structure on the Cold Forgeability of Low 67. M.C. Mataya and G. Krauss, A Test to
Data for Ti-6242 at Hot-Working Tem- Carbon Steels, Scand. J. Metall., Vol 15, Evaluate Flow Localization during Forging,
peratures, Process ModelingFundamen- 1986, p 238 J. Appl. Metalwork., Vol 2, 1981, p 28
tals and Applications to Metals, T. Altan, H. 58. T. Okamoto, T. Fukuda, and H. Hagita, 68. M.C. Mataya, M.J. Carr, and G. Krauss,
Burte, H. Gegel, and A. Male, Ed., Material Fracture in Cold Forging Flow Localization and Shear Band Forma-
American Society for Metals, 1980, p 387 Systematic Classification of Working Meth- tion in a Precipitation-Strengthened
48. J.J. Jonas, R.A. Holt, and C.E. Coleman, ods and Types of Cracking in Cold Forging, Austenitic Stainless Steel, Metall. Trans. A,
Plastic Stability in Tension and Com- Sumitomo Search, No. 9, May 1973, p 46 Vol 13, 1982, p 1263
pression, Acta Metall., Vol 24, 1976, p 911 59. S.I. Oh and S. Kobayashi, Workability of 69. N. Akgerman, J.R. Becker, and T. Altan,
49. R.C. Koeller and R. Raj, Diffusional Aluminum Alloy 7075-T6 in Upsetting and Preform Design in Closed-Die Forging,
Relaxation of Stress Concentration at Rolling, J. Eng. Ind. (Trans. ASME), Vol Metall. Met. Form., Vol 40 (No. 5), May
Second Phase Particles, Acta Metall., Vol 98B, 1976, p 800 1973, p 135
26, 1978, p 1551 60. S.K. Suh and H.A. Kuhn, Three Fracture 70. T. Altan, H.J. Henning, and A.M. Sabroff,
50. P.W. Lee and H.A. Kuhn, Fracture in Cold Modes and Their Prevention in P/M The Use of Model Materials in Predicting
Upset ForgingA Criterion and Model, Forging, Modern Developments in Powder Forming Loads in Metalworking, J. Eng.
Metall. Trans., Vol 4, 1973, p 969 Metallurgy, Vol 9, Met. Powder Ind. Tech., Ind. (Trans. ASME), Vol 92B, 1970,
51. H.A. Kuhn, P.W. Lee, and T. Erturk, A 1977, p 407 p 444
Fracture Criterion for Cold Forming, J. Eng. 61. V. DePierre, G. Saul, and A.T. Male, The 71. K. Yagishita, H. Tsukamoto, T. Egawa, S.
Mater. Technol. (Trans. ASME), Vol 95H, Relative Validity of Coefficient of Friction Oomori, and J. Ibushi, A Study of Sim-
1973, p 213 and Interface Friction Shear Factor Under ulative Model Test for Metal Forming Using
52. R. Sowerby, N. Chandrasekaran, N.L. Conditions of Bulk Plastic Deformation, Plasticine, Mitsubishi Technical Bulletin
Dung, and O. Mahrenholtz, An Analysis of Report AFML-TR-70243, Air Force Mate- 91, Mitsubishi Heavy Industries, Ltd., July
Some Upsetting Tests for Assessing the rials Laboratory, Wright-Patterson Air 1974
Cold Workability of Steels, Fracture 84, Vol Force Base, OH, Oct 1970 72. C.W. Corti, G.H. Gessinger, and A.H.
5, Pergamon Press, 1984, p 3239 62. H. Wagner and F.W. Boulger, A Study of Shabaik, Superplastic Isothermal Forging:
53. M. Kaiso and M. Katsumata, Cold Forge- Possible Methods for Improving Forging and A Model Metal Flow Study, J. Mech. Work.
ability in Medium Carbon Steel with In- Extruding Processes for Ferrous and Non- Technol., Vol 1, 1977, p 35
sufficiently Spheroidized Microstructure, Ferrous Materials, final report on Air Force 73. E. Erman and S.L. Semiatin, Physical
Tetsu-to-Hagane (J. Iron Steel Inst. Jpn.), Contract 33(600)-2627Z, Battelle Memorial Modeling of Metalworking Processes,
Vol 84, 1998, p 721 Institute, Columbus, OH, June 1957 TMS, 1987
54. P.F. Thomason, The Free Surface Ductility 63. A. Tomlinson and J.D. Stringer, The 74. T.L. Hartman, D.F. Bickford, and H.R.
of Uniaxial Compression Specimens with Closing of Internal Cavities in Forgings by Piehler, Densification and the Nature of
Longitudinal Surface Defects, Int. J. Mech. Upsetting, J. Iron Steel Inst., Vol 188, 1958, Residual Porosity in Hot-Deformed Ferrous
Sci., Vol 11, 1969, p 65 p 209 Preforms, Physical Modeling of Metal-
55. L. Janicek and B. Maros, The Determination 64. T.L. Subramanian, N. Akgerman, and T. working Processes, E. Erman and S.L.
of the Cold Forgeability for Specimens with Altan, Application of CAD/CAM to Pre- Semiatin, Ed., TMS, 1987, p 9
Axial Notches of Heat-Resisting and cision Isothermal Forging of Titanium 75. S. Kobayashi, University of California, un-
Corrosion-Resisting Chromium Steels, J. Alloys, Report AFML-TR-77-108, Bat- published research, 1981
Chapter 15
Modeling Techniques in
Forming Processes
W.T. Wu, J.T. Jinn, and C.E. Fischer, Scientific Forming Technologies Corp.
THE OBJECTIVE OF MANUFACTURING This chapter reviews the overall development method. For detailed discussion, refer to text-
is the production of a consistent quality product of modeling techniques for forming processes, books, e.g., Ref 5 to 9. The FEM is discussed in
at a minimal cost. Generally effective goals in- including: detail in a separate section.
clude shortening the lead time in the design Slab Method. As implied in its name, the slab
Slab method
cycle, reducing tooling cost and machine down- method is based on the assumptions such that the
Slip-line method
time at the production stage, and developing a deformation of a workpiece can be approximated
Upper bound method
stable process with a minimal reject rate. The with the deformation of a series of slabs, and the
Finite element method
ability to predict the performance of a particular shape of slab (either flat or cylindrical shape) does
manufacturing process and to compare it with not change in the course of deformation. Each slab
alternative manufacturing processes at an early having an infinitesimal thickness is sliced so that it
stage in the process design cycle furthers these Modeling Techniques contacts with the tool. These methods assume that
goals by reducing costly trial-and-error design the shear stress due to friction between the work-
iterations using production equipment. The most fundamental calculations used in piece and tool does not affect the stress distribu-
Numerous analytical techniques have been metal forming analysis involve a forming load tion within the slab, but it is considered in the
developed to improve the process designers estimate, which is useful in selecting the size of force equilibrium. The stress distribution within
ability to evaluate a process and to predict vari- equipment required to form the product. The each slab is simplified further such that the stress
ous aspects of the metal forming process. Early simplest formula takes the form: component normal to the thickness direction can
methods relied on simple analytical techniques, be varied as a function of the thickness coordinate.
such as the slab method, the slip-line method, A
PK (Eq 1) The other nontrivial stress components are then re-
the upper bound method, and heuristics to pre- lated to the stress component normal to the thick-
dict forming load, critical ratios, workability where P is the forming load, is the mean flow ness direction according to the yield criterion and
limits, and die design features. Complex analyt- stress of a workpiece material under an idealized the flow rule for a rigid-perfectly-plastic material.
ical equations were converted to charts, which state of deformation, A is the planar area of the Then, the differential equilibrium equation ex-
could be applied by the designer. workpiece, and K is an empirically determined pressed in terms of stresses and the thickness co-
As computer technology became prevalent in correction factor for a particular forming process. ordinate is solved with the associated boundary
the engineering and manufacturing industry, The correction factor reflects on the effects of conditions. The slab method has been applied
analytical techniques, such as the upper bound nonuniformity of deformation and friction be- widely to various plane strain or axisymmetric
method, were used to develop specialized com- tween the workpiece and tool. forming problems such as upsetting, extrusion,
puter programs, which could be used to analyze a This estimation has been improved upon by drawing, and rolling.
particular process, such as tube sinking, strip more elaborated approximate solution techniques Figure 1 illustrates a simple example of the
rolling, or extrusion. The finite element method such as the slab method, the slip-line method, the slab method applied to a ring compression prob-
(FEM) for metal forming applications was first upper bound and lower bound methods, Hills gen- lem. In fact, the ring compression test has been
introduced in early 1970 (Ref 1). The continuous eral method, and, finally, the finite element meth- used widely as a convenient tool to evaluate the
improvements in computer technology and FEM ods. These methods have been employed to allow friction factor between the tool and workpiece.
finally made an important impact in the metal for the estimation of not only the forming load, but In bulk metal forming industries, the constant
forming industry in the mid-1980s. Due to its also the material flow pattern and the stress distri- shear friction model (based on shear strength of
unique capability in describing complex shapes, bution. The first two methods, i.e., the slab method the workpiece) has been preferred to the well-
boundary conditions, and realistic material ther- and the slip-line method, solve the field equations known Coulomb friction model (based on con-
momechanical response, the development of a in a differential form directly with certain degrees tact pressure). In Fig. 1(a), the inner radius Ri,
general-purpose metal forming analysis software of simplification. On the other hand, the other the outer radius Ro, and the height H character-
has been realized. The method has been used as methods solve the field equations in an integral ize the geometry of the ring. Now consider an
an essential tool for product and process design form derived from the associated energy theorems. infinitesimal element depicted in Fig. 1(b). The
engineers to reduce development time and cost. Except for the FEMs, application is generally lim- equilibrium equation in the r direction can be
Due to the demand from the industry to produce ited to metal forming processes with simple geom- written as follows:
more accurate simulation models, the FEM has etry and idealized materials laws.
continuously evolved from two-dimensional The following sections briefly review some s r Hrdq - (s r + ds r ) H (r + dr ) dq
analysis into the true three-dimensional models classical solution techniques for the slab method, dq
+ 2s q sin H dr m 2 mkr dr dq = 0
since the late 1980s and early 1990s (Ref 24). the slip-line method, and the upper bound 2 (Eq 2)
Chapter 15: Modeling Techniques in Forming Processes / 221
state of stresses at some points. The slip-line

method has also been applied successfully to var-
ious plane-strain forming problems such as in-
dentation, extrusion, drawing, and rolling.
Upper-Bound Method. Unlike the two pre-
viously discussed methods, the upper-bound
method (UBM) is based on the energy principle,
known as the upper-bound theorem. The upper-
bound theorem states that the rate of total en-
ergy associated with any kinematically admissi-
ble velocity field defines an upper bound to the
actual rate of total energy required for the de-
formation. Hence, for a given class of kinemat-
ically admissible velocity fields, the velocity
field that minimizes the rate of total energy is
the lowest upper bound, and therefore is nearest
the actual solution. Here, the kinematically ad-
missible velocity field is used to denote a veloc-
ity field that satisfies the incompressibility re-
quirement for a rigid-plastic material and the
prescribed velocity boundary conditions. How-
ever, the velocity field may be discontinuous on
a finite number of imaginary internal surfaces.
Fig. 1 A ring compression problem. (a) Geometry of ring. (b) Slab element The rate of total energy generally consists of
three terms such that:
where m and k are the shear friction factor and the axial component of stress can be obtained as
the equivalent shear stress, respectively, and the follows:
sign is used to express that the direction of
E T = s e dV + k Dv
V SD
t dS + m k Dv
SF
s dS
(Eq 11)
frictional stress changes from outward to inward z Y r (Eq 8)
at the neutral point as r increases. The neutral
where Y is the yield stress of a material. It is where and e are the equivalent stress and the
point is denoted by the radius rn. Assuming that
r and neglecting higher order terms, Eq 2 noted that Y 3 k for a material complying equivalent strain rate, respectively; vt is the
can be simplified as: with von Mises yield criterion. The forming load magnitude of velocity discontinuity tangent to
can then be evaluated as: the velocity discontinuity surfaces SD; and vs is
dr 2k dr (Eq 3) the magnitude of sliding velocity on the contact
rn Ro surface SF.
where a =
m
is a constant representing the
Ri
P = s z 2 p r dr + s z 2 p r dr
rn Each term in the right-hand side of Eq 11 rep-
H = KSlab s A (Eq 9) resents the rate of plastic deformation energy, the
coining effect, that is, a portion of the total form- rate of energy dissipation associated with internal
ing load required to overcome friction increases velocity discontinuity, and the rate of energy dis-
as the height of ring decreases. It is noted that where KSlab is the correction factor or normal- sipation due to friction between the tool and
the assumption r holds when the radial ized forming load defined by: workpiece, respectively. The second term, also
component of velocity u is a function of homo- known as the jump condition, can be omitted
geneous degree one in r so that 1 m when a class of continuous velocity fields is con-
KSlab = 1 + ( Ro - Ri )
2 3 H (Eq 10) sidered. Among various classical solution meth-
du u ods, the upper-bound method has been applied
=
dr r The slip-line method is another simple and most extensively to various two-dimensional or
or powerful classical solution method, although its three-dimensional forming problems because it
application is limited to plane-strain problems delivers a fast and accurate solution as long as
e r = e q (Eq 4) for a rigid-perfectly plastic material (Ref 5, 6). In the trial velocity field can be provided closer to
this method, the equilibrium equations for a the actual velocity field. However, it is not easy
By applying the boundary condition r 0 at plane-strain state are first transformed into the to choose a good trial velocity field using a
r Ri and r Ro, Eq 4 can be integrated in the hyperbolic differential equations expressed in combination of analytic functions for geometri-
form: terms of the mean stress, the maximum/mini- cally complicated problems. In order to relax in
mum shear stress, and the direction of maximum/ such a difficulty, the upper-bound elemental
2a k (r - Ri ) when r rn minimum shear stress. The characteristics of the technique (UBET), based on the concept of a
sr =
2a k ( Ro - r ) when r rn (Eq 5) hyperbolic differential equations are known as unit rectangular deforming region (Ref 10)
the slip-lines. The slip-line field then can be con- has been developed and applied to rather com-
Since the stress must be continuous at rn, the lo- structed by networking two kinds of slip-lines plex forming problems and preform design
cation of neutral point is determined as: representing the maximum and minimum con- applications.
stant shear lines that are orthogonal to each other. In order to compare the characteristics of so-
1 Several useful techniques have been proposed to lutions with different solution methods, i.e., the
rn = ( Ro + Ri )
2 (Eq 6) construct the slip-line field graphically depend- upper-bound and slab methods, the ring com-
ing on the configuration of a problem and the as- pression problem illustrated in Fig. 1 is em-
By applying von Mises yield criterion: sociated boundary conditions. The forming load ployed as an example here. Also, a simple form
can then be obtained by determining integral of the trial velocity field is chosen so that an ex-
Y
(Eq 7) constants for particular slip lines from the known plicit form of solution can be obtained. The trial
z r
velocity field used in the present example has and the radial velocity u defined in Eq 12 into Eq some understanding of the three-dimensional
the form (Ref 11): 16, the rate of total energy E T can be integrated forming process. Although a considerable
explicitly as: amount of research has been done in develop-
D r2 ing the FEM for metal forming simulation
u( r ) = r- n
2 r
E T 1
since the pioneering work (Ref 1) was pre-
= rn4 + 3 Ro4 - rn4 + 3 Ri4 sented in 1973, rigorous three-dimensional
v=0 2 p k VD 2
simulation of metal forming problems still re-
w( z ) = - D z (Eq 12)
2
rn + rn4 + 3 Ro4
2 mains a challenging task from the standpoint of
rn2 R
- log 2 - log o computational efficiency, solution accuracy,
VD 2 r + r 4 + 3 R 4 Ri graphics visualization, mesh generation and au-
where D = , VD is the die velocity, H is the n n i
H tomatic remeshing, and so on. As computer
m 4 3 1
height of ring, and rn is the radius at a neutral
H 3 3
(
+ rn - ( Ri + Ro )rn2 + Ri3 + Ro3 )
technology and FEM advance, wider and more
point. It is noted that this velocity field contains complicated metal forming processes are being
only one unknown, i.e., rn, and it is adequate for (Eq 18) investigated. It is believed that the further de-
investigating the effect of the location of neutral velopment of FEM will be continuously chal-
point on the forming load. The neutral radius rn The unknown coefficient, i.e., the neutral radius lenged by the need from the industry to make
is defined by: rn, is then determined so that the rate of total en- the modeling more accurate, more practical,
ergy attains to its minimum value such that: and more affordable.
u0 at r rn Since the two-dimensional FEM implementa-
dE T tion has been discussed elsewhere in detail (Ref
u0 for r rn =0 12), this section focuses on the three-dimensional
drn (Eq 19)
u0 for r rn (Eq 13) implementation.
After some manipulation, the neutral radius can
Because the radial component of velocity, u, is a be written as:
Preliminary Assumptions
function of r only, the present trial velocity field
cannot reproduce the so-called barreling or In order to narrow down the discussion to the
b2 + 3 + b4 most practical applications among a variety of
bulging phenomenon due to friction. 1 1 H o o
rn = ( Ri + Ro ) + log metal forming simulations, the following pre-
All nonzero strain-rate components are then 2 4 m b2 + 3 + b4 (Eq 20)
i i liminary assumptions are first introduced.
written as:
Quasi-Static Analysis. In most metal form-
D r2 ing processes, dynamic effects can be neglected
e r = 1 + n2 rn r
where b i = and b o = n . Equation 20 can then except for high strain-rate processes in which a
2 r Ri Ro realistic deformation mode cannot be obtained
D r2 be solved for rn numerically with specific values without considering the effect of stress wave
e q = 1 - n2 of the ring geometry, i.e., Ri, Ro, and H, and the
2 r propagation and in which the magnitude of ki-
shear friction factor m. netic energy is comparable to that of deforma-
e z = - D (Eq 14) The forming load can be evaluated directly tion energy. In cases where dynamic effects can
by substituting the value of neutral radius into be neglected, the progress of deformation is an-
It is noted that the velocity field satisfies the in- Eq 18 and with use of the relation: alyzed in a manner such that every instantaneous
compressibility condition such that: state of a body in the course of deformation is
E T satisfied with the equilibrium conditions. Such a
P=
er e ez 0 (Eq 15) VD (Eq 21) method of analysis is generally called the quasi-
static analysis.
Since the present velocity field satisfies the It is noteworthy that even for forming
The correction factor or normalized forming
continuity requirement, the incompressibility processes between medium and low strain-rate
load KUBM can be defined by:
condition, and the prescribed velocity boundary ranges, the explicit method of solution originally
condition, it is proved to belong to a class of designed for dynamic analyses is sometimes
kinematically admissible velocity fields. It is P used for the sake of computational efficiency.
K UBM =
also noted that the second term in Eq 11 can be sA (Eq 22) For the explicit method, however, the size of
omitted because there is no internal velocity time step must be small enough, typically the
discontinuity. order of microseconds, to satisfy the stability
The rate of total energy in Eq 11 can be Finite Element Method criterion of the explicit time marching scheme. It
rewritten as: means that the computational efficiency may not
Metal forming simulation is often classified be achieved because the number of solution
as a class of highly nonlinear continuum me- steps increases so greatly compared with that of
E T = s e dV + mk Dv
V SF
s dS chanics problems because it is accompanied by the implicit method. Thus, the so-called mass
large deformation (geometric nonlinearity), scaling technique (Ref 13), which may be inter-
Ro Ro
s e (2 prH ) dr + 2 nonlinear materials behavior (material nonlin- preted as the introduction of artificial inertia
= Ri
Ri
mk u (2 pr ) dr
earity in both deformation and temperature), term, is usually employed in order to increase
(Eq 16) and frictional contact (nonlinear boundary con- the size of time step. In these cases, the mass
dition). Starting from the mid-1980s, the FEM scaling factor must be selected carefully to get
After substituting the following expression of has shown great success in the axisymmetric the solution with reasonable accuracy and simu-
the equivalent strain-rate: applications such as disk forging, cold forging lation time. In sheet forming applications, the
of cylindrical fasteners, and so forth. The ap- explicit method has been very successful.
proximation of a two-dimensional cross section Rigid-plastic analysis is more advantageous
4
1 rn in a three-dimensional part, using the plane- for computational efficiency and robustness
e = D 1 +
3r (Eq 17) strain assumption, is an alternative to achieving than elasto-plastic analysis is. This method has
been used predominantly for the majority of particle. The motion of a mesh can be chosen that the stress measure in Eq 23 must be the
bulk forming processes where the elastic defor- arbitrarily, but it must preserve the boundary of first kind of Piola-Kirchhoff stress tensor ac-
mation is negligible compared with the plastic a continuum in the course of deformation. When cording to the framework of the referential de-
deformation and the distribution of residual the motion of mesh is set the same as the mo- scription of continuum mechanics. However, it
stresses is not of major concern. By neglecting tion of material, it becomes the same as the can be assumed that the reference configuration
the elastic portion of deformation, the rigid- Lagrangian methods. Several different schemes is updated as frequently as required, so it is
plastic formulation (Ref 14) turns out to be very regarding the selection of mesh motion have therefore not distinguished from the Cauchy
similar to that of fluid flow problems except for been proposed according to the specific area of stress tensor.
the presence of yielding and so it is sometimes applications. The ALE method has been consid- Equation 24 represents the incompressibility
called the flow formulation. The velocity field ered a prominent method, especially for the condition where ui is the velocity vector.
satisfying the equilibrium equations, constitu- steady state metal forming applications. How- Equation 25 represents the constitutive equation
tive equations, and boundary conditions instan- ever, it has disadvantages, such as the increase of based on the so-called J2 flow rule. Here, eij is
taneously is obtained at each state in the course problem size due to additional variables for the strain-rate tensor or the symmetric part of the
of deformation. Therefore, it is necessary to defining the mesh motion and the requirement of velocity gradient tensor as defined in Equation
adopt an appropriate scheme for updating de- additional computation due to the evaluation of 26. and e are the effective stress and the ef-
formed configurations from the velocity field convection terms in updating the state variables. fective strain-rate, respectively, defined by:
obtained at each state.
1 1
Unlike the elasto-plastic formulation, the rigid- Description of the Problem s= [ 3
2 s ijs ij ] 2
and e = [ e ]
2
3 ij
2
(Eq 33)
plastic formulation does not have any ambiguity
related to the choice of the objective stress rates The governing partial differential equations
and the decomposition of deformation gradient and the associated boundary conditions for the Equation 27 represents an implicit form of the
into the elastic and plastic parts (Ref 15, 16). rigid-plastic and rigid-viscoplastic problems can yield criterion as a function of the effective
Updated Lagrangian (UL) Formulation. be written as (Ref 12): strain e , the effective strain rate e , and the tem-
There are several different formulations (Ref 17, perature T. Provided the increment of time be-
18) for continuum mechanics problems with large ij,j ij,j m,i 0 in (Eq 23)
tween any two adjacent referential configura-
deformation and/or large rotation, e.g., total tions is small enough, the effective strain can be
Lagrangian (TL), updated Lagrangian (UL), evaluated approximately such that:
ui,i 0 in (Eq 24)
Eulerian, and arbitrary Lagrangian Eulerian
e
(ALE) methods. In the first two methods, the new e old e t (Eq 34)
motion of a continuum is described in terms of 2s
s ij = e ij in W (Eq 25)
the coordinates of a material particle in an arbi- 3 e where t denotes the increment of time. Equa-
trarily chosen reference configuration. The TL tion 27 represents the most popular form of the
method uses the initial undeformed configura- e ij = 12 (ui,j + u j,i ) in W (Eq 26) rigid-plastic constitutive equations, but more
tion as a fixed reference configuration, but the complicated forms of constitutive equations can,
UL method uses the most updated configuration (e, e , T )
in (Eq 27) of course, be used within this structure of for-
during the progress of deformation as the refer- mulation such as the constitutive equations hav-
ence configuration. Both methods are amenable ui i on u (Eq 28)
ing internal state variables with the associated
to solid mechanics problems in which the con- evolution equations.
figuration of a boundary is not fixed in space and Equations 28 and 29 are the prescribed veloc-
changes in the course of deformation. Compared ijnj t i on t (Eq 29) ity and traction boundary conditions, respectively.
with the TL method, the UL method can use the The hat symbol on ui and ti is used to denote pre-
simplified kinematics of a continuum assuming Friction & contact conditions on c (Eq 30) scribed values, and the vector ni denotes the out-
that the reference configuration is updated con- ward normal vector to the body at a point on the
tinuously within a small amount of increment of Here, the open domain and its associated boundary. The friction and contact conditions are
deformation according to the desired degree of boundary represent the current configuration discussed in detail in the subsequent section.
solution accuracy. Thus, the UL method has of a body according to the UL formalism. The
been used widely in most metal forming simula- subscripts u, t, and c in are used to denote
tion. It is worthwhile to mention that this method three different types of the boundary associated Friction and Contact Conditions
generally requires frequent remeshing (or rezon- with the boundary conditions: the prescribed ve-
locity, the prescribed traction, and the frictional The friction and contact conditions in this sec-
ing) when mesh is severely distorted in the
contact conditions, respectively. Indices i, j, and tion are described between any two boundaries
process of large deformation. Furthermore, it may
k are used to denote the components of a tensor whether they belong to the same body (i.e., self-
not be an effective method for a certain class of
and a comma denotes the spatial derivative with contact), rigid and deformable bodies, or two or
forming problems, such as extrusion, rolling,
respect to the current configuration. more deformable bodies. Although several differ-
and machining, in which the steady state solu-
In Eq 23, i.e., the equilibrium equation, ij , is ent methods (Ref 1921) have been presented in
tion is of major concern.
the stress tensor, and ij and m denote its devi- describing the macroscopic contact and friction
In the Eulerian method, the motion of a con-
atoric and volumetric components, respectively, phenomena within the framework of continuum
tinuum is described in terms of the spatial coor-
such that: mechanics, the method of pointwise description
dinates of a material particle in the current con-
of friction and contact conditions has been used
figuration. Therefore, this method is particularly
ij ij mij (Eq 31) in most practical applications. The contact condi-
suitable for well-constrained fluid flow prob-
tion and friction can be summarized as:
lems in which the domain of interest is fixed in
the space. However, there are inherent difficul- and Contact condition (non-penetration condi-
ties in applying this method to the moving tion):
boundary or free surface problems. The ALE s m = 13 s kk (Eq 32) a. Any material particle of a given body
method has been developed to remove these cannot penetrate into another
difficulties by introducing the concept of mesh where ij is the Kronecker delta, and the re- b. The normal component of contact traction
motion independent of the motion of material peated index denotes the summation. It is noted must be compressive for each body
c. A pair of contact points can separate only classical theory of plasticity. The slip function is so-called weak formulation or the principle of
when the contact traction vanishes or be- defined by: virtual work, although certain mathematical fea-
comes tensile tures of variational problems with the quadratic
(pi) (pt)i pn (Eq 40) functional such as the existence and uniqueness
Friction condition:
of solutions and the stability and accuracy of fi-
a. The magnitude of the tangential compo-
and nite element solutions cannot be stated. Here,
nent of contact traction must be less
the weak formulation is used to accommodate a
than or equal to that of the normal com-
(pi) (pt)i mk (Eq 41) broad class of plastic constitutive models to the
ponent multiplied by a coefficient of
same framework of formulation.
friction.
corresponding to the Coulombs law of friction The constraint conditions such as the incom-
b. The instantaneous relative motion in the
tangential direction for a pair of contact and the shear friction law, respectively, where pressibility condition and the contact condition
is the coefficient of friction, k is the shear yield can generally be incorporated into the variational
points can take place when the equality
stress of the weaker material, and m is a constant formulation by using one of two techniques: the
in (a) above holds.
friction factor. penalty method or the Lagrange multiplier
c. The tangential relative motion must be
It is noted that the frictional dissipation func- method. The penalty method has the advantage
along the same line as the tangential
tional associated with the friction law in Eq 38 of simple implementation, but it has a drawback
component of contact traction but in the
or 39 has a non-differentiable form with respect such that it can result in an overconstrained prob-
opposite direction.
to its primary variables, i.e., the velocities of lem or an underconstrained problem depending
In most metal forming applications, two dif- each body. In the area of metal forming simula- on the choice of the penalty parameter. The over-
ferent types of friction laws have been widely tion, the following form of regularized friction constraint means the volumetric locking or the
used: the Coulombs law of friction and the laws has been widely used. For any (z t)i: locking of contact surfaces, and the undercon-
shear friction law. Statement (a) under Friction straint means the inaccuracy of solution in the
condition represents the Coulombs law of fric- 2 sense of incompressibility or nonpenetration. On
( zt ) i ( zt ) i
tion, but it also represents the shear friction law (p t )i = m p n tan -1 ( z ) the other hand, the Lagrange multiplier method
by replacing the normal component of contact p zo t i (Eq 42) can avoid the drawback of the penalty method,
traction by the shear yield stress of the weaker but it has a disadvantage concerning the increase
material. Also, the coefficient of friction is gen- and of problem size because the Lagrange multipliers
erally referred to as the constant factor in the are treated as additional solution variables such
shear friction law. as the velocity of material particles. The
This statement of contact and friction condi- 2 ( zt )i ( zt ) i Lagrange multiplier can be interpreted as the hy-
(p t )i = - m k tan -1
tions can be expressed in a mathematical form as: p zo ( zt )i (Eq 43) drostatic stress for the incompressibility con-

straint and the normal contact traction for the
(p ) nonpenetration condition.
zi = l n I (p n )ni + l t I ( f ) t i where zo is a positive constant. These regular-
(p t ) i (Eq 35) In areas of metal forming simulation, it is pop-
ized friction models approach the original fric- ular to use the Lagrange multiplier method for
tion laws asymptotically as zo approaches zero. the incompressibility condition and the penalty
where z is the relative velocity between two However, a very small value of zo can make a method for the contact condition. However, the
points in contact at an instant as: difficulty in convergence, and a larger value may penalty method has also been used successfully
give a solution deviated from the original fric- with a certain class of finite elements with the
z i uai ubi (Eq 36) tion laws. selective reduced integration scheme.
where the superscripts a and b denote the corre- Mixed Variational Formulation Finite Element Formulation
sponding contact surfaces chosen arbitrarily, ni
is the outward normal vector with respect to the Boundary value problems (BVP) in contin- The finite element method can be distin-
contact surface a, p is the contact traction vec- uum mechanics can be expressed in two different guished from other approximate methods by the
tor on the contact surface a as: ways: the partial differential equations (PDE) way it constructs the trial solution (e.g., kine-
with the associated boundary conditions and the matically admissible velocity field) with a finite
pi pai pbi (Eq 37) variational equations with the appropriate func- number of piecewise continuous trial functions
tion space. The solutions of these two different (polynomial functions in most applications)
where the subscripts n and t denote the normal forms of BVP are referred to as the strong solu- (Ref 2224).
and tangential components, respectively, as: tion and the weak solution, respectively. The For a particular class of metal forming
term weak is used in the sense that the require- processes, it is always important to select an ap-
z i z nni (z t)i and pi pnni (pt)i (Eq 38) ment of continuity (differentiability) of solution propriate type of element (i.e., the order of poly-
is weakened in the variational form of BVP. If a nomials, the geometric shape, the rule of numer-
Continuing with the description of terms in Eq strong solution requires the existence of a second ical integration, and so on). Some issues related
35, the symbol denotes the absolute magni- derivative, the corresponding weak solution re- to the selection of element are discussed subse-
tude of a vector, and ln and lt are non-positive quires only the existence of a first derivative in quently. From the standpoint of polynomial
constants. I(g) is an indicator defined for any the sense of distribution; that is, the first deriva- order, linear elements are generally preferred to
scalar-valued function g as: tive needs to be continuous within each of a fi- quadratic or higher-order elements for most metal
nite number of subdomains but not necessarily forming applications in which the friction and
if g = 0 across the interboundary between subdomains. contact conditions are always present. With use
1,
I ( g) = The variational form of BVP can be obtained of the friction and contact conditions described
0, if g 0 (Eq 39)
only when the quadratic form of functional ex- previously, the so-called node-to-segment contact
ists so that the set of Euler equations, obtained situation cannot be avoided because these con-
In order to describe the friction laws effectively, by the vanishing of the first variation of the func- straints need to be imposed point-wise. The node-
a slip function f is introduced for each pair of tional, is identical to the original PDE. However, to-segment contact can be treated in a simpler
contact points similar to the yield function in the the same form can be obtained by using the manner with linear elements rather than higher-
order elements. Also, linear elements are gener- Some disadvantages of CST elements are MINI element, the shape function for the veloc-
ally easier to use without a-priori knowledge listed as follows. The CST elements are apt to ity is enriched with bubble terms associated with
of solution than higher-order elements with the have directionality in the mesh topology and an additional bubble node, although the shape
same degrees of freedom for a given problem. thus in the solution because the mesh topology function for the mean stress remains as a linear
Next, from the standpoint of geometric shape, directly reflects on the form of trial solutions. function interpolated with the values at four ver-
there are two different kinds of linear elements: Here, the directionality implies that triangular tex nodes. The bubble node is located at the cen-
triangular or quadrilateral for two-dimensional and tetrahedral elements are globally biased in troid of tetrahedron and has only the velocity de-
elements and tetrahedral or hexahedral for three- a certain direction. Also, the CST elements grees of freedom. Here, the word bubble means
dimensional elements. Shape functions for trian- may show more tendency of volumetric locking that its value always vanishes along the bound-
gular and tetrahedral elements contain polyno- due to the incompressibility constraint condition ary of the element. This element does not have
mial terms such as (1, , ) and (1, , , ), than quadrilateral and hexahedral elements. The the volumetric locking problem. Moreover, the
respectively. On the other hand, those for quadri- term volumetric locking means that each ele- total number of equations can be maintained
lateral and hexahedral elements contain polyno- ment endowed with a linear velocity field cannot as the same as that with the tetrahedral elements
mial terms such as (1, , , ) and (1, , , , , deform properly when the same degree of satis- because the velocity components at the bubble
, , ), respectively. Triangular and tetrahe- faction is required elementwise for both the de- nodes can be eliminated at the element level by
dral elements are known as constant stress/strain formation and the constraint condition. In order means of the static condensation.
elements (CST) because all derivatives with re- to avoid such a locking phenomenon, the degree Although the bubble node is introduced mainly
spect to any component of coordinates vanish. In of satisfaction for the constraint condition needs for a systematic stabilization of the ill-posed fi-
the rigid-plastic formulation, vanishing deriva- to be relaxed in a certain manner such as with nite element system, a considerable amount of
tives imply constant strain rates and, therefore, the selective reduced integration scheme or with computational effort is additionally required for
constant stresses, within an element. On the other the mixed formulation as described previously. the static condensation and recovery of the ve-
hand, the quadrilateral and hexahedral elements Another important aspect to be considered in de- locity of the bubble node. However, such addi-
have polynomial terms in the product form of termining the element shape, especially for large tional efforts can be minimized with a few mod-
coordinates, and thus, the velocity gradient in deformation problems, may be whether an effi- ifications of the standard MINI element. In fact,
one component of coordinates is linear with re- cient and robust method for automatic remesh- it turns out that the derivation of this modified
spect to the other components of coordinates. ing is available with the shape of element. In MINI element can be conceived as a systematic
Strain rates and stresses are linear within an ele- areas of metal forming simulation, quadrilateral stabilization of the standard mixed tetrahedral
ment accordingly. Both kinds of linear elements elements are preferred to triangular elements for element because the effect of the bubble node
have been used widely by weighing the pros and the two-dimensional analysis, but tetrahedral appears on only the block diagonal terms associ-
cons for a particular application. elements have been used more than hexahedral ated with the mean stress.
Generally speaking, triangular and tetrahedral elements for the three-dimensional analysis be-
elements have more flexibility in filling meshes cause of the versatility in automatic remeshing. Heat Transfer
into any complicated shape than quadrilateral As mentioned previously, linear tetrahedral
and hexahedral elements. It is noted that triangu- elements cannot be used for metal forming sim-
lar and tetrahedral elements in the group of CST ulation without an appropriate resolution of the In the metal forming industry, heating is fre-
elements must be distinguished from the so- volumetric locking problem. The most common quently used to increase material workability and
called degenerated elements whose strain rates approach to enforce incompressibility includes control forming loads. During the forming
and stresses are not constant within an element. the penalty method and the Lagrange multiplier process, heat is generated from plastic and fric-
Degenerated elements can be obtained from as discussed previously. These methods, how- tion work. Heat is also lost through contact with
coalescing adjacent nodes into the same node, for ever, are limited to the quadrilateral element in colder dies and through convection and radiation
instance, mapping from the four-node quadrilat- two-dimensional and hexahedral element in three- with the environment. Since materials properties
eral parent element into a three-node triangular dimensional. Furthermore, the resulting matrix vary considerably with temperature, accurate
shape or mapping from the eight-node hexahe- is ill conditioned for a conjugate-gradient solver temperature prediction is required. For conven-
dral parent element into a four-node tetrahedral due to large penalty values for incompressibility. ience, the deformation analysis and thermal
shape. The performance of degenerated elements The mixed formulation uses the Lagrange multi- analysis can be loosely coupled in such a way that
is not as good as that of the CST elements or the plier method to achieve incompressibility condi- plastic work and friction are considered as heat
original quadrilateral and hexahedral elements. tion. Generally speaking, the order of polynomi- source in the thermal analysis while the updated
Also, a numerical integration scheme for the als for the velocity shape function needs to be temperature field is used to determine the flow
CST elements in evaluating the stiffness matrix is higher than that for the mean stress shape func- stress behavior during the deformation analysis.
unnecessary because they have constant values tion because the mean stress is originally a dual- The governing equation for heat transfer can
of strain rate and stresses within an element. In ity variable paired with the volumetric strain rate be expressed as:
other words, the integration is always exact since that is the trace of the velocity gradient. For in-
the integrand is constant with respect to coordi- stance, when the linear shape function is used k1 Tii q CT 0 (Eq 44)
nates. On the other hand, for quadrilateral and for both variables, the volumetric strain rate be-
hexahedral elements, a numerical integration rule comes constant within an element, but the mean where is the density, C is the specific heat ca-
based mostly on the Gaussian quadrature for- stress is linear. It is obvious that a linear function pacity, T is the temperature, t is time, k is the
mula is required in evaluating the stiffness ma- cannot be matched with a constant value in a thermal conductivity, and q is a heat generation
trix. The numerical integration is exact when a general case and a mesh system may tend to lock term. Heat generation in metal forming is due to
real element keeps the same shape as the parent as a result. When the quadratic shape function is work of plastic deformation and friction. Heat
element, i.e., a rectangular shape or a brick shape, used for the velocity in order to make the well- generation due to plastic deformation is given by
although it is not possible except for problems posed problem, a considerable increase of prob-
with a very simple geometry. In such cases, the lem size cannot be avoided.
determinant of the Jacobian matrix mapping be- The MINI element is effectively designed for
q pw = k s e dV (Eq 45)
tween the domains of the parent element and a the three-dimensional simulation of incom-
real element is constant. Errors originated from pressible problems such as fluid flow, rubber where k is a deformation efficiency term, repre-
the numerical integration increase as the shape of elasticity, and rigid-plasticity problems under senting the fraction of the work of deformation
element is apart from that of the parent element. the framework of the mixed formulation. In the converted to heat.
The boundary condition for the tool-work- compression, and three methods (slab, UBM, is based on a rectangular workpiece shape. The
piece contact surface includes friction heating and FEM) are used to study the process. The radial velocity is a function of the radius and the
and heat exchange via temperature difference of second example is the cold forging of an elec- neutral radius, the axial velocity is a function of
two objects. trode, the third example is the hot forging of a the height position, and the neutral radius is a
crankshaft, and the last example considers mate- function of friction factor and shear stress.
rial cutting. Only FEM is used in the last three This simplification/assumption introduces
q1 = hfS1
s1 us1 dS1 + HDTdS
S1
1 (Eq 46) examples due to the complexity of the processes. error when the ID and OD surfaces of workpiece
start to bulge during the compression. As the
where us is the sliding velocity, fs is the friction Ring Compression workpiece height increases, it is known that the
stress, is the percentage of friction energy ab- unstable flow (or buckling mode) will gradually
sorbed by the object, H is the lubricant heat The ring compression test is one of the meth- dominate the deformation. In such situations, the
transfer coefficient, and T is the temperature ods used to determine the friction factor. The assumed UBM velocity field is moving away
difference between two objects. process is generally assumed to be two-dimen- from the true velocity field. To demonstrate the
The boundary condition of the free surface in- sional or axisymmetric. Due to the simplicity of buckling mode, another ring compression exam-
cludes convection heat and radiation heat from/to the workpiece and die geometry, ring compres- ple with OD, ID, and height ratio of 6:3:6 was
the environment: sion is used as the first example to demonstrate carried out. In this case, the flow stress is assumed
and compare the capability of slab, UBM, and to be in the form of 10e0.1. A friction factor
FEM. of 0.2 was used. The predicted FEM mesh and
q2 = hc (T - T ) dS2 + se(T 4 - T4 )dS2 (Eq 47) Two-ring test FEM simulations with outer di- various deformation stages are shown in Fig. 4.
S2 S2
ameter (OD), inner diameter (ID) and height ra- Figure 5 shows the effect of friction on the lo-
tios of 6:3:2 and 6:3:0.5 were carried out with a cation of a neutral point with three different as-
where hc is the convection heat transfer coeffi- pect ratios of a ring: i.e., 6:3:0.5, 6:3:1, and 6:3:2,
cient, T is the environment temperature, is constant friction factor of 0.4. A constant flow
stress of 70 MPa (10 ksi) was used in the simu- in terms of the OD, the ID, and the H, in se-
the Stefan-Boltzmann radiation constant, and e quence. From this figure, it is noted that the
is the emissivity of the surface. It is also noted lation. The predicted shapes using the FEM and
the associated neutral line within the workpiece upper-bound solutions approach to the mean ra-
that the inclusion of the view factor for radiation dius of ring asymptotically as the magnitude of
heat calculation (which is not addressed here) at different stages of deformation in both cases
are shown in Fig. 2 and 3, respectively. In the shear friction factor increases. It is because, ex-
can improve the accuracy of the thermal model, cept for point sticking, neither local nor global
especially for the hot forging condition. case with the ratio 6:3:2, the neutral line within
the workpiece is a function of height. In the case sticking phenomenon can be represented with
By substituting Eq 45 into Eq 44 and intro- the present velocity field. It is also shown that the
ducing a small, arbitrary temperature variation with 6:3:0.5, the neutral line remains to be a ver-
tical line. Furthermore, the ID and OD surface upper-bound solutions correlate better with the
T, and applying the divergence theorem, Eq 44 FEM solutions as the height of ring reduces.
can be written in the form: bulged due to the friction force on the two con-
tacting surfaces.
Ti dTj dV + rCT dT dV To simplify the mathematical derivation, the
k
V
1
V
assumed velocity field (Eq 12, 13) in the UBM
+ ks ij e ij dT dV - q n dT dS = 0
Sq
V (Eq 48)
where qn is the heat flux across the boundary. It

includes the convection heat and radiation heat
to the environment for the free surface and fric-
tion heat and heat gain or loss to the contacting (a) (b) (c)
surface. The temperature distribution function
Fig. 2 The effect of friction on the location of the neu-
can be expressed through nodal temperatures tral point for a thick ring compression test. (a) (a) (b) (c) (d)
and shape functions (Ref 12). 2%. (b) 10%. (c) 20%. The cross section of the ring is
After discretization, Eq 48 can be further ex- shown with the curved line denoting the neutral point Fig. 4 A two-dimensional buckling simulation. (a) 1%.
over the height. (b) 20%. (c) 35%. (d) 50%
pressed in matrix form (Ref 24) as:
CT KcT Q (Eq 49)
where C is the heat capacity matrix, Kc is the (a)

heat conduction matrix, T is the node point tem-
perature vector, and T is a vector containing the
time rate of change of temperature of node
points. The heat flux vector, Q, for metal form-
ing simulations considers plastic work of defor-
mation, heat generation due to sliding contact (b)
friction, and heat flux due to lubricant conduc-
tion, convection, and radiation.
(c)
Example Simulations
Fig. 3 The effect of friction on the location of the neu-
tral point in a thin ring compression test. (a) 4%.
To demonstrate the capability of the modeling
techniques, four examples are included in the
(b) 20%. (c) 40%. The cross section of the ring is shown Fig. 5 The effect of friction on the location of the neu-
with the curved line denoting the neutral point over the tral point. UBM, upper-bound method; FEM,
following section. The first example is the ring height. finite-element method
Fig. 7 Progression of electrode forging
mensions, and the final forming was modeled in ternal fold is successfully predicted in the
three dimensions. The shearing and squaring op- zoomed area of the figure. Due to the axisym-
erations were not modeled in this study. metric assumption, the internal fold is not seen
Fig. 6 The variation of forming load with the magnitude Two laps were developed during the back- in the simulation results. Further study, consid-
of friction factor. UBM, upper-bound method ward extrusion as shown in Fig. 8(a) and (b). ering tooling eccentricity, will be carried out in
One lap is seen on the outside wall (Fig. 8b) and the near future. Due to the symmetry condition,
The normalized forming loads obtained using extends the entire circumference. The other lap 1/12 of the workpiece was modeled in the simu-
the UBM are compared with those obtained occurred only partially on the inside wall (Fig. lation to reduce the simulation time.
using the slab method presented in Eq 10 as well 8a). It is believed that eccentricity of the forming The predicted geometry at early stage of the
as the corresponding FEM solutions. It is note- process caused the internal lap. From visual in- tooth forming is shown in Fig. 10. It is clearly
worthy that the forming load estimated using spection, the lap occurred in every tooth in the seen that the material is pushed both upward and
both the UBM and the FEM based on the veloc- final part as shown in Fig. 8(c). Figure 8(c) also downward through the contact of the punch.
ity formulation can be used as a measure of the shows a change in the texture at the inner wall of Folds occurred on the wall above the teeth and
upper limit of the actual load. As shown in Fig. 6, the electrode, above the teeth. Figure 8(d) shows below the teeth. It is also noted that the partially
the forming load estimated using the FEM is a lap that occurred at the bottom of the cavity, formed tooth is seen as a concave shape. As the
smaller than that using the UBM, and thus it can which is seen as a round pattern pointed to by punch moved farther down, the fold began to
be said that the present UBM solution becomes the arrow. Due to the size of the part, it is very wrap around the top of the teeth (Fig. 11). Figure
less accurate than the FEM solution as the mag- difficult to visualize the fold, other than the one 12 shows that this material peeled down into the
nitude of friction factor increases. On the other on the tooth tip. bottom of the cavity as shown in Fig. 8(d). Also,
hand, the load estimated using the slab method The strain distribution and predicted geome- the fold above the teeth is shown to smear,
can be used as a measure of lower limit when the try at the end of each operation through extru- which correlates well to Fig. 8(c) as the cause of
stress field used in the slab method satisfies all sion are shown in Fig. 9. It is seen that the ex- the uneven texture. The final modeled part is
requirements of the kinetically admissible stress seen in Fig. 13. As shown in the real part, the
field: the equilibrium condition, the yield crite- tooth tip is the last area to fill. A clear fold is pre-
(a) (b)
rion, and the prescribed traction boundary con- dicted in the same area.
dition. As shown in Fig. 6, however, the present
slab method does not provide a lower-bound so-
lution because the stress field does not take into
account the shear stress due to friction, and
hence, it violates the Cauchy stress law on the
traction boundary. (c)
(d)
Electrode Cold Forging Process
Folding, also known as lapping, is a severe
forging defect that process designers would Fig. 10 Early stage of tooth forming. The material flow
like to predict and avoid in the manufacturing can be seen as driven up and down in forming
process. There are many challenges to model- Fig. 8 Arrows highlight the defects seen in the forming the teeth.
of the electrode.
ing a fold properly, especially in the three-
dimensional UL method. The methodologies,
to mesh a workpiece with very thin fold and to
account for self-contact condition in the FEM
formulation, are important for a robust and ac-
curate FEM model.
This section demonstrates the modeling of the
forming of a copper electrode (Ref 25). The ac-
tual progression is shown in Fig. 7. The work-
piece undergoes a total of five operations: shear-
ing, squaring, pancaking, backward extrusion,
and final forming. The pancaking and backward Fig. 9 Forming results after squaring and backward Fig. 11 As forming of the teeth continues, note the
extrusion operations were modeled in two di- extrusion. smearing of the material above the teeth.
Fig. 14 Outline of forging process of crankshaft forging
to the temperature distribution inside the work-

piece. Being able to describe the flow stress be-
havior within the process window is extremely
Fig. 12 The material peeling down as the teeth are important for an accurate simulation. Further-
formed
more, the flow stress is generally sensitive to
the strain rate at a hot forging condition; ram
speed will also play an important role, not only
to predict accurately in flow pattern, but also in Fig. 15 Crankshaft forging after buster stage
the prediction of the load. Interface properties
such as friction and heat transfer coefficient are
also important variables that will influence the
heat loss rate of the workpiece, material flow,
and forging load. Therefore, an accurate model
should take into account the coupling between
the process variables, for example, ram speed,
friction factor, heat transfer coefficient, and the (a)
Fig. 13 The final part shape material data (e.g., flow stress representation
and thermal data).
Toward this goal, a parametric study was first
Due to the small part size, it is very difficult to carried out to evaluate these critical process vari-
visualize the folds as well as understand the ables to assure the accuracy of the model. The
overall folding development. The simulation variables under study included the flow stress,
has, however, clearly revealed the complete his- friction factor, heat transfer coefficient, and bil-
tory of the fold development. (b)
let temperature.
The materials properties for AISI-1045 and Fig. 16 Comparison of buster results for (a) high heat
transfer and (b) low heat transfer
Crankshaft Hot Forging Process AISI-1055 were selected for the simulation.
There are three stages to forge the crankshaft:
This forging process and the corresponding busting, blocking, and finishing. profile reduces as the heat transfer coefficient
simulation (Ref 26) were carried out at the The workpiece is first heated to 1200C increases. This is due to the fact that heat loss
Hyundai Motor Company. The objective of the (2192 F). All dies are heated to 200C (392 from the die-workpiece interface is faster using
numerical simulation was to evaluate the forma- F). As shown in Fig. 14, after 10 s of air trans- a higher lubricant heat transfer coefficient. The
bility and to predict the forging load for both a fer, the workpiece is placed in the buster dies difference in temperature distribution influ-
crankshaft and a connecting rod. In this section, for the first forging operation. After 1.5 s air ences the material flow in the die cavity and
only the crankshaft simulation is presented. transfer, the workpiece is removed from buster flash.
Specific interests included: and placed in the blocker for the second forg- At the end of the blocker operation, the pre-
ing process. After another 1.5 s air transfer, the dicted shape and the actual shape are in excel-
The flow pattern and potential defects such
workpiece is removed from blocker and placed lent agreement as shown in Fig. 17. It is also
as under fill or folding
in the finisher for the final forging process. A noted that the folding defects (also known as
The forging load for tool stress analysis and
6500 ton mechanical press is used in the forg- laps) are successfully predicted by the simula-
press selection
ing and simulation. To accurately model the tion as shown in Fig. 18 and 19. The prediction
Stress and strain distribution for possible
temperature evolution during the entire form- of the lap shape and the propagation pattern are
evaluation of microstructure
ing processes, the simulation was carried out well correlated with the corresponding experi-
Improvement and optimization of the exist-
in a nonisothermal manner, including heat mental results.
ing forging processes
transfer analysis to account for the air transfer To validate the model further, the flash thick-
An accurate hot forging model involves many time. ness and the corresponding forging load are
material variables and process variables. It re- During the parametric study, it was found that measured and compared with the numerical pre-
quires proper characterization of the material an accurate lubricant heat transfer coefficient diction with excellent agreement as shown in
such as flow stress behavior and thermal data as was very critical to obtain a similar flash profile Fig. 20.
well as interface properties such as the friction as the actual part. The actual workpiece at the
factor and the lubricant heat transfer coefficient. end of busting is shown in Fig. 15. The pre- Cutting Modeling
In the case of a hot forging process, the thermal dicted shapes based on two lubricant heat trans-
data will affect the accuracy of the temperature fer coefficients are given in Fig. 16. From these Machinability is of primary interest in the
prediction. The deformation pattern is sensitive figures, it is noted that the waviness of the flash materials cutting process. It is affected by many
(a)
Fig. 20 Results of actual and simulated forming loads
(b)
Fig. 17 Comparison of (a) actual to (b) simulated
blocker result
Fig. 21 Initial and final stages of Lagrangian analysis of chip formation
Fig. 18 Actual lap on crankshaft forging lation for transient analysis and Eulerian formu- updated so that it satisfies the free surface con-
lation for steady state has been developed (Ref dition and contact condition with the given ve-
27). In this hybrid procedure, the transient UL locity field. The first part is based on the stan-
method is first used to predict the initial chip for- dard flow formulation of plastic materials. For
mation as shown in Fig. 21. The solution at the the sake of computational effectiveness, the sec-
end of the transient run is then used as an initial ond part is divided into two levelsthat is, the
guess for the steady state analysis. determination of the position of surface nodes
In most solid mechanics problems, the do- and internal nodes.
main, i.e., the steady state configuration of a Employing surface elements that have both
deformable body, is unknown a priori and must membrane and edge bending stiffness then
be determined as a part of the solution. From solves the position of surface nodes. The free
the standpoint of the FEM, each node has de- surface condition is also embedded into the ele-
grees of freedom for both the velocity and the mental equations so that each free surface ele-
position. Thus, the velocity field needs to be ment is positioned to be parallel to the velocity
satisfied for the equilibrium equations, the con- at the center of element. Since the resulting set
stitutive equations and the boundary conditions, of equations is nonlinear due to the geometric
and the configuration needs to be satisfied for nonlinearity, the Newton-Raphson iterative
Fig. 19 Simulated lap on crankshaft forging
the free surface condition and the contact con- method is used. Once the position of surface
dition. Here, the free surface condition implies nodes is obtained, employing incompressible
factors, among which are materials behaviors, that the velocity on the free surface must be elastic solid elements solves for the position of
insert shape, and cutting condition. A good un- tangent to the surface and there is no traction internal nodes.
derstanding of the interactions among the chip on the surface. State variables such as stresses and strains are
flow, heat generation, residual stress, tool stress, A great number of different solution methods updated in two different ways: simple interpola-
and tool wear is crucial in order to optimize the have been presented, including ALE methods, tion scheme of existing solutions and streamline
design of the process and tooling (Ref 27). pseudo-solid domain mapping methods, adaptive tracking scheme, depending on whether the up-
To understand the thermal-mechanical re- h-p FE methods, streamline tracking methods, dated Lagrangian solutions are available. A re-
sponse on an insert (tool wear) during a long and methods of splines. In this current approach, sult of steady state thermomechanical machining
period of cutting, the conventional updated two sets of coupled governing equations are simulation is presented in Fig. 22. The steady
Lagrangian transient approach, especially for solved for both the velocity and the position it- state insert temperature is shown in Fig. 23. The
three dimensional analysis, is not efficient since eratively. First, the Eulerian velocity solution is work to validate the cutting force, temperature,
it requires enormous CPU time to reach a steady obtained based on a given configuration. Next, and wear using the model is being carried out
state. A hybrid procedure using both UL formu- the new configuration, i.e., the nodal position, is (Ref 27 and 28).
plastic deformation using rigid-plastic and

rigid-viscoplastic formulation. For processes
in which spring back or residual stress is of
interest, an elasto-plastic formulation should
be used. At high temperatures, materials
can be highly rate sensitive, in which case
an elasto-viscoplastic formulation should be
considered.
Microstructure modeling. Currently, most
modeling considers isotropic behavior.
Recrystallization, texture, damage, phase
transformation, grain size effect, precipitate
size, and distribution could potentially af-
fect the deformation behavior. Understand-
(a) (b)
ing the microstructure properties evolution
Fig. 22 Chip shape before and after Eulerian calculation during the thermal mechanical process and
its associated effect with respect to the me-
The forging load can be predicted for equip-
chanical properties, such as flow stress,
ment selection.
Youngs modulus, and creep behavior, is
The solution accuracy depends largely on the va- very crucial to the model accuracy since
lidity of the assumptions that are made for the these properties are generally the input data
simulation. Minimizing the time and cost of pro- to the FEM analysis.
duction trials and design iterations continues to
be the goal of FEM development. Process Conditions, Including the Friction and
Toward this goal, ongoing research is focused the Lubricant Heat Transfer Coefficient. During
on the following areas. the forming process, the surface of the workpiece
Computational Efficiency. Three-dimen- is severely deformed/stretched, wherein the lu-
sional FEM transient metal forming analysis is bricant film thickness and film temperature are
very computationally intensive. Although the continuously changing in time. Consequently,
computing power has been improving dramati- this will affect the material flow behavior. Proper
cally every year, the size of the FEM model for characterization of the lubricant behavior is im-
simulation has also been increasing quickly to portant for the accuracy of the model.
improve the model accuracy. Methodologies to Integrated FEM for Complete Manufac-
improve the computing speed, such as parallel turing Processes. Most discrete mechanical
Fig. 23 Insert temperature after cutting
components go through a manufacturing cycle,
computing and efficient contact searching algo-
rithms, are under continuous exploration and de- which typically includes ingot breakdown, forg-
velopment. Steady state or quasi-steady state ing, trimming, heat treating, welding, machin-
Current and Future Works methods are being investigated for certain classes ing, and assembling and installation. A complete
of forming processes, such as extrusion, rolling, material flow and thermal mechanical history is
and ring rolling. important to model the behavior properly under
Modeling techniques for metal forming the service condition. It is also necessary to have
Robustness and Ease of Use. The required
have continuously evolved throughout the a better understanding of the material evolution
geometry data is generated by a computer-aided
years. Because of the unique capability of FEMs and modeling throughout the entire manufactur-
design (CAD) system. Generally speaking, the
to describe the complex geometry and bounding process.
data may not meet the requirement for com-
ary condition of the forming process, the Optimization. The FEM provides a predic-
puter-aided engineering (CAE) analysis. An
method has proven itself by numerous success tion of the result of a proposed manufacturing
error-tolerant system is necessary for a robust
stories experienced by the industry or reported process but still relies on an experienced de-
CAE tool. In addition, realistic graphics presen-
by research organizations, such as reducing signer to interpret the results of the analysis and
tation is also important for post-processing the
the trials and errors on the shop floor, and modify the process based on prior knowledge
simulation result. To facilitate the use of the
shortening the lead time to develop a new and experience. Some recent research efforts have
FEM model, special purpose preprocessors
product, and so on. sought to use computational resources to en-
aimed for specific manufacturing process are
Specifically, the FEM has been used by the hance and optimize process designs based on a
being developed.
forming industry primarily for the following starting design, and improvement of the design
Accuracy. The FEM approach is an error
purposes: is based on the sensitivity analysis of the design
optimization process. Generally speaking, the
The forming flow can be analyzed to evalu- larger the model is, the better the solution accu- variables. Because the amount of simulations are
ate the die/model design and to avoid flow- racy is expected. Certain assumptions, due to the generally large for the sensitivity analysis during
related defects such as folds, suck-in de- lack of understanding of many real phenomena, optimization, these techniques are still some-
fects, flow-through defects, under fill, and may introduce inevitable error. Specifically, they what limited to two-dimensional analysis but are
ductile fracture. are as follows: emerging and evolving quickly.
The tool stress can be analyzed to improve Material Characterization: With advances in the FEM, more manufactur-
the tool life. ing processes/problems with greater complexity
The process window control (e.g. press Flow stress data should cover the process are being studied and investigated. In addition to
speed, forging temperature, lubricant, and window of the actual forming process, in- the development of an integrated manufacturing
heat treatment procedure) can be modeled cluding temperature, strain, and strain rate. software, utilizing optimization techniques to
to ensure that the resulting microstructure The extrapolation of the data range usually systematically achieve design objectives is also
properties (such as grain size) meet the results in unexpected error. the focus of current and future work. Due to the
requirement. Elastic response has been neglected for large required computing resource in optimization and
to minimize the users time, the speed and the Kobayashi, Mechanics of Plastic Deforma- Int. J. Solids Struct., Vol 6, 1970, p. 1069
accuracy, the robustness of the FEM procedure, tion in Metal Processing, Macmillan, New 1086
and computer graphics in three dimensions will York, 1965 18. J. Wang and M.S. Gadala, Formulation and
continuously be challenged. 6. W. Johnson and P.B. Mellor, Engineering Survey of ALE Method in Nonlinear Solid
Plasticity, Van Nostrand and Reinhold, Mechanics, Finite Elem. Anal. Des., Vol 24,
London, 1973 1997, p 253269
ACKNOWLEDGMENTS 7. B. Avitzur, Metal Forming: Processes and 19. J.T. Oden and J.A.C. Martins, Models and
Analysis, McGraw-Hill, New York, 1968 Computational Methods for Dynamic Fric-
The authors would like to thank Professor
8. R.A.C. Slater, Engineering Plasticity: tion Phenomena, Comp. Meth. Appl. Mech.
Soo-Ik Oh, Seoul National Unversity, for pro-
Theory and Application to Metal Forming Eng., Vol 52, 1985, p 527634
viding the upper bound method of ring com-
Processes, Macmillan, London and Basing- 20. A. Curnier, A Theory of Friction, Int. J.
pression, and Mr. D.H. Seo, Hyundai Motor
stoke, 1977 Solids Struct., Vol 20, 1984, p. 637647
Company, for their crankshaft forging example.
9. T. Altan, S.I. Oh, and H.L. Gegel, Metal 21. A. Klarbring, General Contact Boundary
The authors would also like to thank Mr. Tony
Forming: Fundamentals and Applications, Conditions and The Analysis of Frictional
Perez and Dr. Daqing Jin, both of the Timken
American Society for Metals, Metals Park, Systems, Int. J. Solids Struct., Vol 22, 1986,
Company, for their collaborate effort on material
Ohio, 1983 p 13771398
cutting modeling under the DOE funded CTMP
10. H. Kudo, Some Analytical and Experimental 22. G. Strang and G.J. Fix, An Analysis of The
project.
Studies of Axi-Symmetric Cold Forging and Finite Element Method, Prentice-Hall,
Extrusion, Parts I & II, Int. J. Mech. Sci., Vol Englewood Cliffs, New Jersey, 1973
REFERENCES 2, 1960, p 102127; Vol 3, 1961, p 91117 23. J.T. Oden and J.N. Reddy, An Introduction
11. S.I. Oh, Ring Compression Modeling to the Mathematical Theory of Finite Ele-
1. C.H. Lee and S. Kobayashi, New Solutions Using Upper Bound Method, Private Com- ments, John Wiley & Sons, New York, 1976
to Rigid-Plastic Deformation Problems munication, 2002 24. O.C. Zienkiewicz and R.L. Taylor, Chapter
Using a Matrix Method, J. Eng. Ind. (Trans. 12. S. Kobayashi, S.I. Oh, and T. Altan, Metal 10, The Finite Element Method, 4th ed. Vol
ASME), Vol 95, 1973, p 865 Forming and The Finite Element Method, 2, McGraw-Hill Book Company
2. J. J. Park and S.I. Oh, Application of Three Oxford University Press, New York, 1989 25. M. Foster, Scientific Forming Technologies
Dimensional Finite Analysis to Metal 13. W.J. Chung, J.W. Cho, and T. Belytschko, Corporation Simulation Database Archive
Forming Process, Proc. NAMRC XV, On the Dynamic Effects of Explicit FEM 26. D.H. Seo, Numerical Simulation of Crank
(Bethlehem, PA), 1987, p 296 in Sheet Metal Forming Analysis, Eng. Shaft Forging Process, DEFORM Korean
3. T. Coupez, N. Soyris, and J.L. Chenot, 3-D Comput., Vol 15, 1998, p 750776 Users Group Meeting, 29 Oct 2002,
Finite Element Modeling of the Forging 14. R. Hill, The Mathematical Theory of Plastic- Changwon, Korea
Process with Automatic Remeshing, J. ity, Oxford University Press, London, 1971 27. Y.C. Yen, A. Jain, C. Avanachand, W.T. Wu,
Mater. Process. Technol., Vol 27, 1991, 15. L.E. Malvern, Introduction to The Mechanics and T. Altan, Computer Simulation of
p 119133 of a Continuous Medium, Prentice-Hall, Orthogonal Cutting Using a Tool with
4. G. Li, W.T. Wu, and J.P. Tang, DEFORM- Englewood Cliffs, New Jersey, 1969 Multiple Coatings, submitted to 2003
3D: A General Purpose 3-D Finite Element 16. A.E. Green and P.M. Naghdi, A General College International pour lEtude Scienti-
Code for the Analysis of Metal Forming Theory of Elastic-Plastic Continuum, fique des Techniques de Production Meca-
Processes, Metal Forming Process Simula- Arch. Rational Mech. Anal., Vol 18, 1965, nique (CIRP)
tion in Industry, 2830, Sept 1994, (Baden- p 251281 28. C. Pavanachand, J.T. Jinn, A. Perez, D.Q.
Baden, Germany), Internationale Kongress 17. H.D. Hibbit, P.V. Marcal, and J.R. Rice, Jin, et al., Machining Modeling, SFTC,
und Tagungsorganisation Finite Element Formulation for Problems Timken, DOE Project Final Report, under
5. E.G. Thomsen, C.T. Yang, and S. of Large Strain and Large Displacements, preparation, 2003
Chapter 16
Rolling
ROLLING OF METALS is perhaps the most (1.0 in.) may take place on several in-line Uniform reduction in thickness with no
important metalworking process, because a mills with the slab running out between the change in width: This is the case with strip,
greater volume of material is worked by rolling roll pairs (stands). sheet, or foil rolling where the deformation
than by any other deformation process. A signif- Thick strip (band) is hot rolled to a thick- is in plane strain, that is, in the directions of
icant portion (on the order of ~90%) of steel, ness of approximately 1.5 to 5 mm (0.06 to rolling and sheet thickness. This type of
aluminum, and copper products may go through 0.2 in.); if quantities are large, a continuous metal flow exists when the width of the de-
the rolling process at least one time during the hot strip mill is employed in which the formation zone is at least 20 times the
production process. The principal advantage of stands are placed close enough together for length of that zone.
rolling lies in its ability to manufacture products the strip to run through all stands simultane- Uniform reduction in thickness with an in-
from relatively large pieces of metals at very ously, with controlled tensions between crease in width: This type of deformation
high speeds in a somewhat continuous manner. stands. The hot strip may be the finished occurs in the rolling of blooms, slabs, and
Other methods of metalworking, such as forg- product in the form of a coil, cut-to-length thick plates. The material is elongated in the
ing, are slower. Most ingots are processed by hot plate, or sheet. rolling (longitudinal) direction, is spread in
rolling into blooms, slabs, and billets (Fig. 1), Cold rolling on reversing or tandem mills is the width (transverse) direction, and is com-
which are subsequently rolled into other prod- done if a thinner gage, a smoother surface pressed uniformly in the thickness direc-
ucts such as plate, sheet, bar, structural shapes, finish, or a strain-hardened end product is tion.
rod (for drawing into wire), and rounds for mak- required. The strip is cold rolled with accu- Moderately nonuniform reduction in cross
ing seamless tubing. rate control of coiler and/or interstand ten- section: In this case, the reduction in the
The largest quantities of rolled forms are flat sions. If the material strain hardens exces- thickness direction is not uniform. The
products (plate, sheet, strip, foil) from the sively, intermediate process anneals are metal is elongated in the rolling direction, is
processes of hot rolling followed by cold rolling. carried out. The resulting wide strip or sheet spread in the width direction, and is reduced
Smaller quantities are rolled into shapes (sec- is typically 0.2 to 2 mm (0.008 to 0.08 in.) nonuniformly in the thickness direction.
tions), usually hot. All flat products are made in thick. For small quantities, individual sheets Along the width, metal flow occurs only to-
a fairly standard sequence of operations: may be rolled by hand. ward the edges of the section. The rolling of
The thinnest strip (less than, say, 0.05 mm, an oval section in rod rolling or of an airfoil
The starting material is an individually cast or 0.002 in.) is called foil and is produced in section would be considered to be in this
ingot or continuously cast slab that is hot re- specialized cold rolling mills. The work category.
duced (broken down) on two- or four-high rolls are invariably backed up; the number Highly nonuniform reduction in cross sec-
reversing mills. Unless the end product is of backup rolls ranges from two (four-high tion: In this type of deformation, the reduc-
wide plate, rolling is continued to form a mill) to eighteen (Sendzimir mill). tion in the thickness direction is highly
long, coiled, wide band. If production vol- nonuniform. A portion of the rolled section
umes warrant it, rolling from approximately Narrow strip with rounded corners is obtained is reduced in thickness, while other portions
100 mm (4 in.) to approximately 25 mm by flattening of wire. may be extruded or increased in thickness.
The primary objectives of the rolling process As a result, in the width (lateral) direction,
are to reduce the cross section of the incoming metal flow may be toward the center. Of
material, to improve its properties, and to obtain course, in addition, the metal flows in the
the desired section at the exit from the rolls. In thickness direction as well as in the rolling
rolling, a squeezing type of deformation is ac- (longitudinal) direction.
complished by using two work rolls rotating in
opposite directions. The process can be carried These points illustrate different deformation
out hot, warm, or cold; thus, rolling processes conditions of rolling. Except for strip rolling,
are often classified as hot or cold. From a funda- metal flow in rolling is in three dimensions (in
mental point of view, however, it is more appro- the thickness, width, and rolling directions). In
priate to classify rolling processes on the bases addition, material flow during the rolling of
of the complexity of metal flow during the shapes from cast billets of rectangular or round
process and the geometry of the rolled product. cross section is much more complicated than flat
Thus, this chapter, based on adapted text from rolling. Determination of metal flow and rolling
Ref 1 and 2, briefly introduces the rolling stresses in shape rolling is very important in the
process for the basic product forms of strip, design and operation of rolling mills, and, so,
plate, and shapes. numerical techniques are used to simulate metal
Fig. 1 Rolling sequence for fabrication of bars, shapes, In general, the rolling of solid sections can be flow in such complex rolling operations. How-
and flat products from blooms, billets, and slabs divided into the subsequent categories: ever, this chapter only introduces general con-
Chapter 16: Rolling / 233
cepts on rolling of strip, plate, and shapes. From the geometry of the pass, the maximum hardly be overstated. In a typical rolling situa-
Numerical techniques are not discussed in de- possible draft is: tion, forward slip seldom exceeds 15%; the rest
tail, but some references to numerical techniques of the speed differential is taken up in the back-
are cited. Dhmax = m2 R (Eq 2) ward slip zone. Therefore, the strip enters the roll
gap at a high interface sliding velocity, which di-
where R is the radius of the roll, and m is the co- minishes very rapidly, within a space of typically
efficient of friction. By definition, F/P = m, 20 mm (0.8 in.) or less, to zero in the neutral
Flat Rolling which can be expressed also as tan f (where f is zone, only to pick up again as the strip leaves the
the friction angle). Therefore, it is usual to state roll gap. This relative sliding and its reversal,
Flat rolling may appear to be a very simple that the section will enter the rolls unaided only while helpful in many respects, also account for
process. The workpiece volume remains con- if the friction angle f exceeds the contact angle most of the problems encountered with lubrica-
stant, and the reduction in thickness is first de- a. This represents one of the instances where tion in rolling, as discussed in more detail in the
scribed under the idealized conditions of plane friction is desirable. This also would appear to section Relative Slip, in this chapter.
strain and homogeneous deformation. That is, offer a very simple way of determining the coef-
the idealized (simplified) case of flat rolling as- ficient of friction: the roll gap is increased until The Friction Hill. The amount of slip and the
sumes a rectangular body that is homogeneously the slab enters the rolls without being pushed. location of the neutral plane are influenced by
deformed into another rectangular body of However, the value of this test is limited, be- the friction that opposes the relative slip.
greater length. The reduction in thickness results cause the frictional force is unidirectional at the Frictional stresses oppose the slip, and the direc-
mainly in a change of length of the material point of acceptance, and lubricant films may be tion of the frictional stresses point (act) toward
under the condition of plane strain. The defor- scraped off; neither of these conditions is repre- the neutral plane. These frictional stresses are
mation also is assumed to be homogeneous sentative of steady-state rolling. Nevertheless, described in terms of a so-called friction hill,
that is, all vertical sections of the piece before the method is convenient under conditions of dry which is related to the distribution of the pres-
rolling remain vertical during the rolling lubrication and for a very approximate ranking sure (p) between the rolls and the section in the
process. These two simplifying assumptions are of lubricants. contact arc (Fig. 3). The peak of the friction hill
useful as first approximation, although they are occurs at the neutral point, and the shape of it
only valid under limited conditions. Plane-strain depends on the pass geometry and the magni-
conditions are approximated when a flat is wide, Speed and Stress Distribution tude of the friction. For example, as the amount
and the assumption of homogeneous deforma- of reduction is increased, the position of the neu-
tion restricts the validity of solutions to large L/h Once the slab is accepted and drawn through tral point approaches the exit. When the maxi-
ratios, where L is the projected length for the arc the roll gap, the situation changes considerably. mum possible reduction is attempted, the neutral
of contact in the deformation zone, and h is the If idealized conditions of homogeneous defor- point reaches the exit. This situation is very un-
mean thickness of the section (as shown in Fig. mation and plane strain are assumed, there is no stable, because any slight increase in reduction
2). Friction and the pass geometry both influ- spread, and so the reduction in thickness (Dh) or drop in friction will cause the strip to stop
ence the assumption of homogeneous deforma- causes an increase in length. Under these condi- moving and the rolls to start skidding over the
tion, and inhomogeneous deformation has sev- tions, continuity can only be maintained if the strip.
eral consequences (as noted in the section products of thickness and velocity are constant Figure 4 illustrates some different situations and
Strain Distribution in this chapter). at all points along the zone (arc) of contact. That the shape of the friction hill. With very low friction
To initiate rolling, the workpiece first must be is: (Fig. 4a), the inclination of the roll surfaces causes
drawn into the roll gap by friction. This condi- most of the material to flow backward, thus mov-
tion is reached when the horizontal component v0h0 = vnhn = v1h1 ing the neutral plane close to the exit. Deformation
of the frictional force F is just equal to the hori- is almost homogeneous. The friction hill is low,
zontal component of the radial force Pr (Fig. 2). where v0 and h0 are the initial velocity and
Thus: height before entering the deformation, and
where vn, hn, and v1h1 are equivalent products
F cos a Pr sin a (Eq 1a) anywhere in the deformation zone and at the
exit, respectively (Fig. 2). Accordingly, the exit
or speed of the slab, vl, must increase in proportion
to the elongation (which in turn is proportional
F to reduction). The rolls also move at some speed
> tan a (v, Fig. 2) intermediate between the entry and
Pr (Eq 1b)
exit speeds. Therefore, there is only one point in
the arc of contact where the slab and roll move
at the same speed (v = vn). There is no relative
slip at the point where v = vn, and so, the point is
h = h0 + h1
Pr 2
described as the no-slip point or neutral point. In
F a three-dimensional presentation, it is referred to
v as the neutral plane.
With these variations in relative motion, slip
v0 vN v1
h0 h1 between the slab and the roll occurs everywhere
in the contact zone except at the neutral plane (or
point). Between the entry point and neutral point,
the work material moves more slowly than the
L
rolls; this is referred to as backward slip.
Between the neutral point and exit point, the strip
moves more quickly than the roll; this is referred
to as forward slip. The importance of relative slip Fig. 3 Positive and negative friction directions during
Fig. 2 Simplified schematic of rolling conditions between the workpiece and roll surfaces can strip rolling
these examples, inhomogeneous deformation re-

sults only from high friction. At lower L/h ratios,
however, and especially when L/h < 1, inhomo-
geneity results from the process geometry itself.
The friction hill may show a double hump (Fig.
4d). This situation is typical of early passes dur-
ing the hot rolling of thick slabs, and separation
of the effects of friction and process geometry
becomes extremely difficult.
The prediction of the shape of the friction hill
is one of the principal aims of rolling theory, but
few theories predict the shape of the friction hill
and the distribution of shear stresses in full
agreement with actual measurements. To some
extent, there is doubt also about the accuracy of
the measurements themselves. In practice, the
average pressure and roll force (discussed in the
section Modeling of Strip Rolling) are more
practical and reliable indicators of the magni-
tude of friction; with higher friction, roll force is
also higher.
Relative Slip
The Neutral Angle. As shown in Fig. 4, the
neutral plane moves toward the middle of the arc
of contact with increasing friction. Therefore, the
angle of the neutral plane an (the angle between
the neutral and exit planes in Fig. 3) is a sensitive
indicator of friction balance and thus of the mag-
nitude of friction. The neutral angle may be read-
ily calculated if some simplifying assumptions are
accepted. One of the most widely employed for-
mulae (Ref 4) assumes that deformation is homo-
geneous, that slipping friction exists everywhere
except in the neutral plane, and that radial roll
Fig. 4 Material flow, pressure, and shear stress distributions for various friction conditions and L/h ratios. (a) Very low pressure p and interface friction m are constant
friction, which moves the neutral plane close to the exit. (b) Intermediate friction, with a steeper friction hill along the arc of contact. These two latter assump-
and a neutral point further from the exit. If friction is high enough, sticking occurs, and the neutral plane broadens into tions are certainly unjustified, but it appears that
a neutral zone. (c) Sticking within the entire contact zone. (d) Friction hill with double hump at low L/h ratios. See text
for discussion.
many of the errors offset each other. Then, from
the horizontal components of acting forces, the
position of the neutral plane is:
and most of the pressure is attributable to the criterion, the condition for sticking is sometimes
plane-strain flow stress (s0), equal to 1.15 s0 (or considered to occur when mmax = 0.577, but this 2
a 1 a
2k) if the material yields according to the von is really true only when full surface conformity an = -
Mises criterion (where k is the shear flow stress ac- is reached at p = s0 (Ref 3). With sticking, slid- 2 m 2 (Eq 3)
cording to von Mises). In cold working, the mate- ing is arrested, and the neutral plane broadens
rial strain hardens during its passage through the into a neutral zone. Shear stresses reach a limit- where the a angle sustained at entry can be ex-
rolls, and the friction hill appears as in Fig. 4(a). ing value, and correspondingly, the friction hill pressed from:
Frictional stresses are low and frequently de- is rounded (Fig. 4b). Deformation is inhomoge-
scribed by a constant m, implying the ti distribu- neous, because a dead-metal zone forms adja- 1/ 2
tion shown in Fig. 4(a) by broken lines. Direct cent to that part of the contact arc in which stick- L h0 - h1
sin a = =
measurements generally show a decay of the inter- ing friction prevails. Modeling of the shear R R (Eq 4)
face shear (friction) stress (ti) toward the neutral stress distribution becomes more difficult. On
plane; even if there is some error in the measure- rough rolls under unlubricated conditions, stick- Substituting into Eq 3 obtains:
ments themselves, a variable m (or friction factor ing may exist within the entire contact zone.
m) within the arc of contact is more realistic. Deformation becomes highly inhomogeneous,
h - h1 1 h0 - h1
With intermediate friction (Fig. 4b), the fric- resembling backward and forward extrusion. an = 0 -
tion hill is steeper, the neutral point is farther Instead of a neutral zone, it is more realistic to 4R m 4R (Eq 5)
from the exit, and, if friction is high enough, the speak of a flow-dividing zone. Because defor-
following condition for sticking is satisfied near mation is concentrated away from the roll sur- and thus, an is completely determined by pass
the neutral plane: face, shear stresses on the roll decrease, and the geometry and the coefficient of friction and,
friction hill is heavily rounded (Fig. 4c). most conveniently, is independent of the flow
ti = m p > k The previous examples are based on the situ- stress of the material. This offers one of the most
ation of relatively thin stripthat is, L/h > 2 reliable methods of friction determination.
where k is the shear flow stress according to von (where L is the projected length of the arc of Rolling a strip into a coil permits the applica-
Mises. Because k = 0.577 s0 with the von Mises contact, and h is the mean strip thickness). In tion of tensions. Back tension retards the move-
ment of the strip and thus shifts the neutral plane Forward slip values are comparable only for Strain Distribution
toward the exit; front tension has the opposite ef- preset tensions; otherwise, l1 must be extracted
fect. For rolling with tensions, the position of the from Eq 6 or some similar formula. The tech- In the idealized case of flat rolling, a rectan-
neutral plane can be calculated only if the mean niques proposed in Ref 6 and 7 derive m from si- gular body is homogeneously deformed into an-
plane-strain flow stress (2k) is known. multaneously measured roll force, torque, and other rectangular body of greater length. In real-
According to Ref 5: forward slip values. Their great advantage is that ity, inhomogeneous deformation develops in the
2k need not be known, and the coefficient of thickness, width, and length directions due to the
a 2 k (h0 - h1 ) + h0 t 0 - h1t1 friction is obtained even in the presence of ten- effects of friction and the effects of pass geome-
f= -
2 4kR m (Eq 6) sions (Ref 6): try. The ideal of homogeneous deformation is
approached only in rolling with a large L/h ratio
where t0 is the back and t1 the front tension, both T and with low enough friction to maintain sliding
m= along the whole arc of contact (Fig. 4a).
expressed in units of stress; R is the deformed 2 Sf h1
roll radius (see Eq 28). PR 1 - As soon as a zone of sticking develops, the
h0 - h1 (Eq 10) formation of a dead-metal zone (Fig. 4b, c) leads
When friction is high enough for a sticking
zone to develop, the surface of the strip exhibits to the extrusion effect already mentioned, caus-
forward slip only from the front end of the stick- where P is measured roll force, T is torque, Sf ing a backward convexity of originally vertical
ing zone. At a first glance, if sticking prevails represents forward slip values, and R is the lo- planes in the entry zone and a forward curvature
over the whole arc of contact, forward slip cally increased radius of the rolls due to slight in the exit zone. Deformation may be likened to
should reduce to zero, at least on the strip sur- flattening under load. compression between inclined plates, and there-
face. However, the center of the strip is now sub- For a constant m, forward slip reaches a max- fore, the angle of entry a, or, more realistically,
jected to severe deformation, resembling for- imum when the angle of contact equals the fric- the (h0 h1)/R ratio, affects the position of the
ward and back extrusion, as noted in conjunction tion angle. This would allow direct determina- flow-dividing plane. Friction determines, to-
with Fig. 4(d). At the point of exit, the strip sur- tion of m by simply rolling with gradually gether with the L/h ratio, the extent of the stick-
face is accelerated to keep up with the center, increasing reductions until the maximum for- ing zone; therefore, it also affects inhomogene-
and the position of the flow-dividing plane is ward slip value is found. Unfortunately, this pro- ity. Furthermore, frictional drag may cause a
still a sensitive indicator of frictional conditions. cedure is valid only if m remains unchanged with severe retardation of surface layers.
Friction from Forward Slip. The roll and increasing pass reduction, a condition seldom Through-Thickness Inhomogeneity. Inho-
strip move at the same speed at the neutral plane, fulfilled. mogeneity of through-thickness deformation
and any reduction in thickness between the neu- Friction from Skidding. With increasing re- may cause variations in properties such as hard-
tral and exit planes must result in a higher exit ductions, the neutral angle moves toward the ness. One example of this is where an inhomo-
velocity v1. Forward slip (Sf ) is defined as: exit, and, at some critical reduction when the geneity factor is defined as the difference in
neutral plane reaches the exit, skidding sets in. A hardness between the core and the surface, ex-
value of m or m can then be calculated (Ref 8). pressed as a percentage of core hardness (Ref
v1 - v In the absence of tensions, and when a is small 11). With this factor, inhomogeneity is shown to
Sf =
v (Eq 7) so that in Eq 4 sin a = a, the angle of entry at decrease with reductions in friction (Fig. 5), but,
skidding is: at very light reductions, inhomogeneity was evi-
It increases with f and thus with increasing dent even with the best lubricant, because the
friction. From the geometry of the pass, forward 1/ 2 L/h ratio was small. At high L/h values, defor-
slip is approximately equal to: h - h1
a skid = 0 = 2m mation became homogeneous irrespective of the
R (Eq 11) lubricant used. Deformation can be inhomoge-
1 2 R neous even in the rolling of very thin strip, if the
Sf = f 2 - 1
2 h1 (Eq 8) The onset of skidding also can be deter- pass reduction (or, rather, the L/h ratio) is small.
mined, in principle, from a single experiment if This condition is intentionally induced in temper
Forward slip can be simply measured by a wedge-shaped specimen is rolled (Ref 9). (skin pass) rolling of steel.
scratching one or more lines on the roll surface Once the critical reduction is reached, skidding The homogeneity of deformation also affects
parallel with the roll axis; the imprints of the sets in, and the entry angle is twice the friction the location at which new surfaces are gener-
lines will be clearly visible on the rolled strip. At angle. ated. The absolute increase in surface area is a
a given roll speed v, the rolls take a certain time Other Methods of Friction Determination. function only of pass reduction; thus, if a 50%
to travel the carefully measured distance (l0) be- A method related to forward slip measurement reduction is taken, half of the surface is old (cov-
tween successive scratch lines; during the same has been proposed (Ref 10). Small irregularities, ered with adsorbed films, oxides, contaminants,
time period, the strip travels at a higher speed v1, such as pickup on the roll surface, cause etc.) and half of it is virgin metal. In well-lubri-
and the distance between the imprints of the two scratches to appear on the surface of a polished cated rolling where sliding friction predomi-
marks is greater (l1). Therefore: strip. If rolling is carried out slowly enough to
stop the rolls with the strip in the gap, the length
and direction of the scratches may be observed
l1 - l0
Sf = under a microscope. From the relative lengths of
l0 (Eq 9) forward and reverse scratches, the position of
the neutral plane can be determined, and, from
From Eq 8, the position of the neutral plane, Eq 5, m can be derived. The scratches become
and from Eq 5, the average external coefficient hooks in the spread zone, allowing fully rolled
of friction m can be calculated, with the advan- strips to be examined, although some of the tail-
tage that 2k need not be known and that no spe- ing scratches formed in the backward slip zone
cial instrumentation is needed. Some precau- tend to be obscured by later contact. The posi-
tions are necessary, though, when roll flattening tion of the neutral plane can be determined also
is significant or when the shrinkage of the l1 dis- from backward slip, provided that some accurate
tance on cooling must be taken into account for means of measuring entry velocity v0 and roll Fig. 5 Effect of lubrication on the inhomogeneity of
rolling at elevated temperatures. velocity v is available. deformation. Source: Ref 11
nates, the new surfaces are generated along the roll; for the same roll diameter, spread increases heavy reductions taken on a relatively thin mate-
arc of contact, and, presumably, the lubricant ef- with the L/w ratio, because a longer arc of con- rialthat is, with L/h > 2. Relatively light re-
fectively separates the two surfaces. With poor tact presents greater frictional resistance in the ductions on a thick workpiece (L/h < 1) lead to
lubrication or in the absence of a lubricant, the rolling direction. These observations have been double barreling (Fig. 8d), which is only weakly
surfaces break up just before entering the roll repeatedly confirmed. A partial review is given dependent on friction.
gap; virgin surfaces enter the contact zone, and in Ref 13. Confusion arises only when dry and Through its influence on the homogeneity of
metal-to-metal contact could occur. The conse- lubricated experiments are conducted at a con- deformation, friction also affects the occurrence
quences of this depend on the prevailing lubrica- stant roll gap setting; lubricated rolling gives of rolling defects. Edge cracking is a result of
tion mechanism. heavier reductions and therefore greater spread, longitudinal secondary tensile stresses generated
Occasionally, surface cracking is encountered creating the false impression that spread de- at the edges of the rolled workpiece, where the
in the hot rolling of difficult-to-work materials. creases with friction. roll pressure drops (Fig. 7). Secondary tensile
The cause is a combination of low ductility (at The friction hill shapes shown in Fig. 4 apply stresses become especially large when the work-
less than optimal hot working temperatures) and only to the central portion of a slab where plane piece barrels severely; therefore, the danger of
secondary tensile stresses generated by cooling strain prevails. Pressures gradually decay toward edge cracking increases with increasing friction.
of the workpiece surface in contact with the the spread zones, as already shown (Ref 12) Friction, especially sliding friction, has also
rolls. Localized deformation associated with (Fig. 7). At the edge, the normal pressure drops been suspected of contributing to opening up of
sticking friction contributes to these stresses. below 2k, and yielding takes place in a stress the billet in the horizontal center plane (croco-
Lubricants alleviate the problem partly by in- state composed of the normal stress and a longi- diling or alligatoring), although no direct evi-
creasing the homogeneity of deformation and tudinal tensile stress induced by the elongation dence or detailed analysis is available to support
partly through their heat-insulating properties. of material adjacent to the edge (secondary ten- this assumption.
When friction is higher on one roll than on the sile stress). Using a large number of pressure Strip Shape. Control of strip shape requires
other, the larger elongation on the well-lubri- transducers embedded in the roll surface along a an understanding of the interactions of many
cated surface causes the rolled slab to curl line parallel to the roll axis, it has been shown variables. A flat product is rolled only if reduc-
around the roll with higher friction. This also (Ref 14) that without a lubricant, the pressure tions and, consequently, elongations are uniform
happens when one of the rolls is smaller. drop-off is steep and limited to the edges, across the whole width of the sheet. If the strip
Friction and inhomogeneity of deformation are whereas in lubricated rolling, the pressure be- was originally of uniform thickness, the roll gap
also important factors in achieving bonding in gins to drop off, very gradually, from the center must be parallel; if the strip is rolled with a
roll cladding and in rolling of bimetallic strip. of the strip. slightly thicker center to ensure good tracking,
Localized deformation due to friction and pass Friction also affects the development of the the roll gap must be proportionally thicker in the
geometry also can affect the rate of texture de- side surfaces of the rolled material. In ideal ho- middle. The roll gap referred to here is that de-
velopment in cold-rolled sheet. mogeneous deformation, the edge profiles fined by the rolls while the rolling load is ap-
Lateral Inhomogeneity. The assumption of would remain straight (Fig. 8). However, friction plied; because the rolls deflect under the applied
plane-strain deformation implies that material restricts material flow at the interface in the loads, cylindrical rolls would present a substan-
flow is directed only in the longitudinal direc- width direction, too, and therefore, higher fric- tially larger gap in the middle of the strip. Thus,
tion. In practice, it is inevitable that some lateral tion results not only in greater spread but also in greater elongations would be imposed on the
flow (spread) should also occur. One early ex- greater barreling of the side faces (Fig. 8b). strip edges, making them longer and therefore
perimental demonstration of this is outlined in Under sticking conditions, some of the spread wavy (Fig. 9).
Ref 12. Lateral flow is clearly evident, as is the actually develops by folding over of the side sur- These effects may be counteracted in several
resultant increase in width w (Fig. 6). Most of faces (Fig. 8c). Single barreling is typical of ways:
the spread develops in the zone of backward
The rolls may be ground to a barrel shape,
slip; as such, it is sensitive to the position of the
with the crown (camber) calculated to give
neutral plane. Spread increases with increasing
a parallel gap under load.
friction, but other factors enter, too. For a given
Heat input into the rolls causes them to ac-
pass reduction, spread is greater with a larger
quire a thermal camber, because much of
the heat is extracted through the roll necks.
Frictional heating contributes to the roll
camber. Luckily, the larger thermal camber
is at least partially compensated for by the
larger roll deflection due to higher roll
forces.
The temperature profile and thus the thermal
camber of the rolls are affected by the quan-
tity, distribution, and heat capacity of the lu-
bricant.
The work rolls of a four-high or multiroll
mill may be mechanically deflected to con-
trol the shape of the roll gap very rapidly.
Fig. 7 Interface pressure distribution in dry rolling of In addition to ground and thermal cambers
aluminum strip on rough rolls. Source: Ref 12 and roll bending, strip shape is also affected
Fig. 6 Directions of material flow in dry rolling of alu-

minum strip (plan view of contact zone over
half-width). Source: Ref 12 Fig. 8 Edge profiles of rolled slabs
it must be understood that, because of the preva-

lence of the mixed-film lubrication mechanism,
these values change greatly with process condi-
tions. Lubricants may be of an oil or a water
base. Because of the predominance of the
mixed-film mechanism, boundary and/or ex-
treme pressure additives are essential. Oil-based
lubricants are used, in general, only when water
staining would develop, as in cold rolling of alu-
minum, or when an aqueous lubricant would
prove inadequate in terms of reduction of fric-
tion or prevention of adhesion.
The lubricating power of emulsions can be in-
creased by reducing their stability. It is also pos-
sible to combine the high lubricating power of a
neat fat or oil with the high cooling power of
water by applying the lubricant to the strip and
the water to the roll, or by mechanically mixing
the two at the point of application. The same
principles apply to the hot rolling of sections
(shapes) and tubes, except that the spatial con-
figuration of the deformation zone becomes
complex, and large speed differentials develop
around the circumference of the section. Filling
Fig. 9 Elastic deflection of rolls and its effect on strip shape. (a) Long, wavy shape at the strip edges due to inade- of the shape is generally hindered by friction,
quate crown (thermal or ground) of the rolls. (b) Long, wavy shape in the middle region due to excessive
crown. (c) Parallel gap between rolls, and no waviness on the strip
but friction is actually helpful in filling a leg
formed by indirect draft.
In cold rolling, one of the main purposes of
by tensions, because, for a given reduction, hot rolling depends on the adhesion between lubrication is that of reducing friction. Because
roll forces are lower and roll deflections are workpiece and roll materials. If adhesion is high, of the need for some minimum friction, hydro-
smaller if higher tensions are applied. as in rolling of aluminum, lubrication is essen- dynamic lubrication is impractical, but hydrody-
tial for the control of roll coating buildup. If ad- namic theory is of value in predicting the thick-
The delicate balance may be upset by any hesion is low, as in rolling of steel, the cooling ness of the film entrained for mixed-film
change from steady-state conditions. For exam- and wear-reducing functions of the lubricant lubrication. The gradually converging entry
ple, during mill start-up or when cooling by the prevail. At high L/h ratios, lubricants aid in re- zone, periodic contact with the roll surface, and
lubricant is excessive, sufficient heating does not ducing roll force and power. limited sliding in the contact zone make rolling
develop the thermal camber of the rolls, and the The most frequently used lubricants are tabu- responsive to lubrication with fluids. As ex-
strip is rolled with a long (wavy) edge (Fig. 9a). lated in Table 1. For purposes of preliminary, ap- pected from basic principles, friction in the
If, on the other hand, insufficient coolant is approximate calculations, typical average coeffi- mixed-film lubrication regime decreases with
plied or the lubricant is too warm, excessive cient of friction values are also shown, although decreasing roll entry angle (smaller reduction
thermal camber leads to rolling with a long
(wavy) middle (Fig. 9b). When the waviness
shifts off center, it is usual to speak of a quarter- Table 1 Commonly used rolling lubricants and typical values
wave. Highly localized poor shape may develop Hot rolling Cold rolling
as a result of a blocked coolant nozzle, which al-
Material Lubricant(a) (b) Lubricant(a) (c)
lows local heating.
Steels None + cooling water ST Emulsion, 36% conc., of MF
Emulsion of fat (+ E.P.) 0.4 synthetic palm oil
Lubrication Fat (ester) (+ E.P.) + cooling 0.3 Synthetic palm oil + water MF
water
Friction is needed to draw the workpiece into Stainless steels As for steel M.O. (1020) with MF
the roll gap and, once the workpiece is ac- and Ni alloys Cl additive
Al and Mg Emulsion, 215% conc., of M.O. 0.4 M.O. (420) with MF
cepted, to ensure its passage through the defor- alloys (20100) with <20% fatty 15% fatty acid, alcohol,
mation zone. The minimum value of friction re- acid, alcohol, ester ester
quired for acceptance is twice that needed for As above, but synthetic mineral MF
continuous rolling. Lubricants are applied pri- oil
Foil: as above, but (1.56) MF
marily to reduce friction and wear and to en- Cu and Cu Emulsion, 28% conc., of M.O. 0.3 Emulsion, 210% conc., of M.O. MF
sure temperature control. The relative impor- alloys (80400) with fat (80400) with fat
tance of these functions depends on the process M.O. (850) with fat MF
and the workpiece material. In cold rolling, the (fatty acid)
Ti alloys None ST Oxidized surface, with:
dominant mechanism is abrasive and adhesive Fat (+ water) ST Esters or soaps 0.2
wear accompanied by spalling, a form of fa- Castor oil (fatty oil) 0.2
tigue wear. Compounded M.O. (410) 0.2
In hot rolling, abrasive and thermal fatigue Refractory Canning + lubricant for can 0.4 M.O. with boundary and MF
metals material extreme pressure agents
wear dominate. Lubrication then serves to re- Bare, dry 0.3
duce abrasive and adhesive wear and to mini-
(a) M.O., mineral oil; viscosity in centiStokes at 40 C (105 F) in parentheses. (b) ST, sticking friction. (c) MF, mixed-film lubrication; = 0.10 at
mize the thermal excursions that culminate in low speeds, dropping to 0.03 at high speeds and viscosities
thermal fatigue. The purpose of lubrication in
and larger roll diameter), increasing viscosity, d (sh) rolling processes, are presented to demon-
and increasing speed. Decreasing friction means = pv (sin f m m cos f) strate the merits of the proposed method.
dx (Eq 12)
lower forces and, for a given roll gap setting, A Plane-Strain Elastoplastic Finite-
thinner issuing gage. Thus, on acceleration of Element Model for Cold Rolling of Thin
where the minus sign refers to the zone of back-
the mill, the strip thickness decreases; this speed Strip (Ref 31): Elastic roll deformation in
ward slip and the plus sign to the zone of for-
effect is reinforced by changes in oil film thick- cold rolling of thin strip is an important con-
ward slip. The horizontal compressive stress s is
ness in mills equipped with full-fluid-film bear- cern, and a coupled FEM is formulated,
assumed to be uniformly distributed over the
ings. meshing a global strip-roll system with in-
height h of the section.
With increasing pass reduction, lubrication ternal interface with sliding and friction.
A more rigorous analysis (Ref 16) eliminated
shifts from predominantly hydrodynamic to pre- The model is two-dimensional and only an-
some of the unrealistic assumptions, notably, ab-
dominantly boundary in character. Forward slip alyzes roll flattening, but application to sev-
sence of strain hardening in the workpiece mate-
increases and roll forces rise; the rise in roll eral kinds of rolling passes is examined
rial, inhomogeneous deformation, and the pres-
force becomes very steep once conditions of (temper rolling, thin foil rolling), with spe-
ence of a sticking zone. The analysis was
limiting reduction are approached. Very heavy cial emphasis on roll-deformed shape and
applied in the mid-1900s for calculation of
reductions and/or high speeds thin the film to the behavior of metal in the roll gap (sliding/
rolling forces (Ref 17, 18). The roll-separating
point where breakdown occurs, resulting in sticking zones, elastic/plastic zones).
force and the roll torque can be estimated with
heavy smearing and roll pickup with materials A Finite Element Computer System for
various levels of approximations by such mathe-
prone to adhesion (such as aluminum) and the Analysis of Mechanical and Thermal
matical techniques as the slab method, the
development of a localized heat streak defect Behaviors in Hot Strip Rolling (Ref 32): A
upper-bound method (Ref 19), or the slip-line
with other materials (such as steel). Increasing two-dimensional, Eulerian finite-element-
method of analysis (Ref 20, 21). In more recent
roll surface roughness generally shifts the mech- based model is presented for the prediction
developments, computerized numerical tech-
anism toward the boundary regime but can be of heat transfer and metal flow occurring in
niques have been used to estimate metal flow,
beneficial in delaying the onset of the heat streak the strip, and heat transfer and thermoelas-
stresses, roll-separating force, elastic deflection
defect. tic phenomena occurring in the work roll
of the rolls, thermal conditions, and strain distri-
during hot strip rolling. Basic FEMs are de-
butions by viscoplastic analysis (Ref 19, 2227).
scribed, with emphasis on combining each
Modeling of Strip Rolling Viscoplastic analysis allows the construction of
model to deal rigorously with the coupled
slip-line fields and gives information on pressure
aspect of the thermomechanical behaviors.
distribution and interface conditions. The basic
This section, adapted from Ref 2, provides a A Three-Dimensional Finite Element
formulation for a general viscoelastic material
brief review of strip rolling theory and analysis. Method for a Nonisothermal Aluminum Flat
and constitutive equations are described in Ref
In strip rolling, the width of the strip is much Strip Rolling Process (Ref 33): A three-di-
28, and the formulation of flow in rolling is in-
greater than its thickness, and the deformation mensional coupled thermoelastic-plastic
troduced in Ref 29.
process is regarded as a problem of plane strain FEM of nonisothermal rolling is described
Of course, numerical solutions continue to
with sliding friction along the arc of contact (ex- for the analysis of strip curvature caused by
become more refined and robust with advances
cept at the neutral point). Models might also as- the difference in the heat-transfer boundary
in computer technology. Finite-element meth-
sume homogeneous deformationthat is, origi- conditions of the upper and lower rollers. In
ods (FEM) are particularly important, because
nally vertical sections remain vertical during addition, the changes in shape, temperature
they are very flexible and can be applied for
rolling. This assumption restricts the validity of field, and strain field of the strip during the
various situations ranging from constant fric-
solutions to large L/h ratios and an ideal rigid- various stages were analyzed to obtain the
tion, mixed sliding, and non-steady-state condi-
plastic material of constant flow strength. lateral plastic flow of the strip.
tions. In FEM, the complex shape under the roll
The researcher von Karman (Ref 15) was the Three-Dimensional Rigid-Plastic Finite
gap is divided into cells or elements with simple
first to write the differential equations for the Element Analysis of Strip Crown and
three-dimensional shapes. Through the analysis
equilibrium of horizontal forces acting on a ver- Flatness During Rolling With Constraint
of these elements one at a time, the deformation
tical section of thickness dx taken at a distance x Boundary Conditions Proposed by Shohet
pattern in a complex shape can be determined.
from the exit plane (Fig. 10). Assuming that fric- (Ref 34): Three-dimensional rigid-plastic
Such calculations can take hours, depending on
tion is low and x is small relative to the roll ra- FEMs and three-dimensional elastic FEMs
the complexity of the calculation and the speed
dius R, the radial roll pressure p may be consid- were used to investigate the behavior of the
of the computer, but new generations of com-
ered to be approximately equal to the vertical contact boundary between the surface of a
puter hardware with greater speed continue to
component pv, and equilibrium can be expressed roll and a strip in strip rolling. The con-
enhance the use of FEM. A more detailed re-
as: straint boundary condition proposed by
view of FEM is beyond the scope of this chap-
Shohet is incorporated into the whole sys-
ter, but the following abstracts are a small sam-
tem to evaluate both strip crown and flat-
ple of FEM applications in strip rolling
ness.
analysis:
Simplified Method for Estimating Roll-
An ALE Hydrodynamic Lubrication Finite
Separating Force. The strip rolling process and
Element Method With Application to Strip
symbols used in all subsequent equations are il-
Rolling (Ref 30): A method that incorpo-
lustrated in Fig. 11. Because of volume con-
rates the hydrodynamic lubrication analysis
stancy, the following relations hold:
into the arbitrary Lagrangian Eulerian
(ALE) finite-element analysis is developed
for steady-state strip rolling simulation. By w h0 v0 = w h v = w h1 v1 (Eq 13)
employing the ALE formulation, only part
of the workpiece, which is subjected to where w is the width of the strip; h0, h, and h1
large plastic deformation within the roll-bite are the thicknesses at the entrance, in the defor-
region, is modelled, so that the computa- mation zone, and at the exit, respectively; and v0,
Fig. 10 Stresses acting on a vertical section of a rolled tional cost is substantially reduced. Two nu- v, and v1 are the velocities at the entrance, in the
workpiece merical examples, aluminium and steel strip deformation zone, and at the exit, respectively.
In order to satisfy Eq 13, the exit velocity v1 plane (roll angle a) in the deformation zone, and by following the theory presented in Ref 16.
must be larger than the entrance velocity v0. vR is the roll surface velocity. At the entrance However, these calculations are quite complex
Therefore, the velocity of the deforming mate- plane: and require numerical techniques in order to
rial in the x or rolling direction must steadily in- avoid an excessive number of simplifying as-
crease from entrance to exit. At only one point vz = 2vR sin aD; h = h0 sumptions. A computerized solution, with all
along the roll-strip interface is the surface veloc- necessary details and the listing of the Fortran
ity of the roll vR equal to the velocity of the strip. At the exit plane: computer program, is also given in Ref 22.
This point is called the neutral point or neutral For a numerical/computerized calculation of
plane, indicated by N in Fig. 11. vz = 0; h = h1
rolling stresses, the deformation zone can be di-
The interface frictional stresses are directed vided into an arbitrary number of elements with
from the entrance and exit planes toward the By taking a simple average of these two lim- flat, inclined surfaces (Fig. 13). The element, il-
neutral plane, because the relative velocity be- iting values, an approximate value of strain rate lustrated in this figure, is located between the neu-
tween the roll surface and the strip changes its is obtained: tral and exit planes, because the frictional stress t
direction at the neutral plane. This is considered is acting against the direction of metal flow. When
later in estimating rolling stresses. this element is located between the entrance and
e = [2v R sin a D /h0 + 0] / 2
An approximate value for the roll-separating (Eq 16) neutral planes, t acts in the direction of metal flow.
force can be obtained by approximating the de- The stress distribution within this element can be
formation zone, shown in Fig. 11, with the ho- A more accurate value can be obtained by cal- obtained by use of the slab method, as applied to
mogeneous plane-strain upsetting process. With culating an integrated average of ea (Eq 15) plane-strain upsetting (Ref 36):
this assumption, Eq 14 is valid; that is, the load throughout the deformation zone. Then, an aver-
per unit width (P/w) of the strip is given by: age approximate value is (Ref 35): K2 h1
sz = ln + s z1
1/ 2 K1 h0 + K1 X
P 2s v 2(h - h1 ) (Eq 18)
= 1 +
ml
l e = R 0
w 3 4h h0 R (Eq 17)
(Eq 14) where
However, in this case, the following approxi- The simplified (linear) stress roll pressure
distribution in strip rolling is illustrated in Fig. K1 = 2 tan a (Eq 19)
mations must be made:
12. The maximum stress is at the neutral plane
Average strip height h = 0.5(h0 + h1). N. These stresses increase with increasing fric- 2 s K1
Average length of the deforming strip l = tion and length of the deformation zone XD. K2 = - + 2 t(1 + tan 2 a )
3 (Eq 20)
RaD, with cos aD = 1 (h0 h1)2R. In the Tensile stresses applied to the strip at entrance or
literature, it is often recommended that the exit have the effect of reducing the maximum
value of the projection of strip length xD stress (by an amount approximately equal to t = m s/ 3 (Eq 21)
(Fig. 11) be used for l; however, considering Dsz, in Fig. 12b) and shifting the position of the
the effect of friction on the roll-strip inter- neutral plane. The analogy to plane-strain upset- Following Fig. 13, for x = Dx, h0 + K1x = h1,
face length, RaD, it is more appropriate to ting is illustrated in Fig. 12(a). and therefore, Eq 18 gives sz = sz1, the bound-
use l = RaD. The stress distribution can be calculated by ary condition at x = Dx, which is known. For x =
To estimate average flow stress s (e , e, T ) at using the equations derived in most textbooks or 0:
a given rolling temperature T, the average strain
e is obtained from the thickness reduction; that
is, e = ln(h0 /h1). The strain rate e is given by:
e a = v z / h = 2v R sin a / h
= [2v R sin a ] / [h1 + 2 R(1 - cos a )] (Eq 15)
where vz is the velocity at a given plane in the z-

direction (Fig. 11), h is the thickness at a given
(a)
(b)
Fig. 12 Stress distribution in rolling. (a) With no ten-

Fig. 11 Representation of strip rolling. The strip width sile stresses at entry or exit. (b) With tensile Fig. 13 Stresses in a deformation element used in
w is constant in the y (width) direction. stress sze at exit computerized calculation of rolling stresses
K 2 h1
Roll-Separating Force and Torque. The in- contact length. There are several numerical
s z = s z0 = ln + s z1 tegration of the stress distribution over the methods for calculating the elastic deformation
K1 h0
length of the deformation zone gives the total of the rolls (Ref 23). A method for approximate
roll-separating force per unit width in strip correction of the force and torque calculations
If the element shown in Fig. 13 is located be- rolling. In addition, the total torque is given by: for roll flattening entails replacement of the
tween the entrance and neutral planes, then the original roll radius R with a larger value R . A
sign for the frictional shear stress t must be re- XD value of R is suggested in Ref 40 and is re-
versed. Thus, Eq 18 and 19 are still valid but: T=
0
RdF
(Eq 24)
ferred to extensively in the literature. This is
given as:
2s
K2 = - K1 - 2 t(1 + tan 2 a ) where XD is the length of the deformation zone 16(1 - n 2 ) p
3 (Eq 22) (Fig. 14), R is roll radius, and F is the tangential R = R1 +
pE(h0 - h1 ) (Eq 28)
force acting on the roll. Assuming that all energy
In this case, the value of the boundary condi- is transmitted from the roll to the workpiece by
tion at x = 0, that is, sz0, is known, and sz1 can frictional force: where v is Poissons ratio of the roll material, p
be determined from Eq 18: is the average roll pressure, and E is the elastic
dF = tds (Eq 25) modulus of the roll material.
It is obvious that R and p influence each
K h1
s z1 = s z 0 - 2 ln other. Therefore, a computerized iteration proce-
K1 h0 + K1Dx (Eq 23)
ds = dx / cos a = 1 = tan 2 adx (Eq 26)
dure is necessary for consideration of roll flat-
tening in calculating rolling force or pressure.
The stress boundary conditions at exit and en- In the deformation zone, the frictional force is Thus, the value of p is calculated for the nomi-
trance are known. Thus, to calculate the com- in the rolling direction between entry and neutral nal roll radius R. Then R is calculated from Eq
plete stress (roll pressure) distribution and to de- planes. It changes direction between the neutral 28. If R /R 1, the calculation of p is repeated
termine the location of the neutral plane, the and exit planes. Thus, the total roll torque per with the new R value, and so on, until R /R has
length of the deformation zone XD (Fig. 11, 12) unit width is: approximately the value of 1.
is divided into n deformation elements (Fig. 14). Hydrodynamic Lubrication (Adapted
Each element is approximated by flat top and XN XD
from Ref 41). Lubricant is applied at the en-
bottom surfaces. Starting from both ends of the
T = Rt

0
(1 + tan 2 a )dx
XN
(1 + tan 2 a )dx
trance side of each pair (stand) of rolls. At the
deformation zone, that is, entrance and exit (Eq 27) entrance side, the layers of lubricant that are in
planes, the stresses are calculated for each ele- contact with the rolls or with the strip adhere to

ment successively from one element to the next. where t equals ms /3; R is roll radius; a is roll their respective metal surfaces and move in-
The calculations are carried out simultaneously angle (Fig. 11); XN is the x distance of the neu- ward toward the exit. An inlet entry zone,
for both sides of the neutral plane. The location tral plane from the entrance (Fig. 14); and XD is which is shaped as a wedge, also forms. The
of the neutral plane is the location at which the the length of the deformation zone (Fig. 14). outer layers of this wedge move inward, and a
stresses, calculated progressively from both exit Elastic Deflection of Rolls. During rolling of return flow of lubricant (in the form of an eddy
and entrance sides, are equal. This procedure has strip, especially at room temperature, a consid- current) occurs between the surface layers (Fig.
been computerized and extensively used in cold erable amount of roll deflection and flattening 15a). At low rolling speeds, the entry zone
and hot rolling of sheet, plane-strain forging of may take place. In the width direction, the rolls wedge is negligibly small. With increasing
turbine blades (Ref 37), and in rolling of plates are bent between the roll bearings, and a certain rolling speeds (or increasing liquid viscosity or
and airfoil shapes (Ref 38, 39). amount of crowning, or thickening of the strip, increasing values of the Sommerfeld number),
occurs at the center. This can be corrected by ei- the wedge increases both in thickness and
ther grinding the rolls to a larger diameter at the depth. The Sommerfeld number (S) is:
center or by using backup rolls.
In the thickness direction, roll flattening Viscosity Strip exit velocity
S=
causes the roll radius to enlarge, increasing the s 0 hf (Eq 29)
Fig. 14 Calculation of stress distribution by dividing

the deformation zone into a number of ta-
pered elements. In this case, tensile stresses in the strip are
zero at both entrance and exit. Fig. 15 (a) The lubrication entry zone and (b) surface irregularities associated with strip rolling
where hf is the final (exit) strip thickness, and s0 rolls to decrease. With the increasing thickness The entry zone expands, and the corner of
is the flow stress. With increasing values of S, of the lubricant film and reduced mill flexing, the liquid wedge moves further away from
the point where metal-to-metal contact between the thickness of the emerging strip reduces. The the entrance and closer to the exit. The
the strip and the rolls is established moves fur- actual gap between the rolls is larger than the meaningful slope of the friction hill on the
ther toward the exit. thickness of the strip by twice the film thickness. entrance side starts at the area around the tip
The surfaces of both rolls and strip are not Because friction drag decreases with increasing of the entry zone.
perfectly smooth surfaces in that they contain ir- rolling speed, the neutral point approaches the The roll separation force gets lower.
regularities in the form of peaks (or crests) and exit, and forward slip is reduced. Roll bending/flattening and mill stretching
valleys (cavities) (Fig. 15b). Some lubricant Two critical points may now be reached. The are reduced.
passes from the entrance to the exit side through first is skidding due to insufficient friction drag, Strip thickness is reduced.
the labyrinth of channels created by these irreg- while the second is the establishment of hydro- For relatively thick strip (or low dry-friction
ularities. As the strip is deformed, the crests are dynamic lubrication. If the strip is thin enough, value m0), a critical roll speed of the first
flattened and the entrapped lubricant is pressur- hydrodynamic lubrication is reached first, and kind may be reached. When this occurs,
ized in the diminishing volume of the cavities. skidding will not develop. When speed continues friction becomes so low (below the critical
At slow rolling speeds, the excess lubricant in to increase after hydrodynamic film lubrication is value required) that the neutral point is at
the diminishing cavity space is squeezed to flow established, friction drag and forward slip, which the exit, and skidding will commence (lines
back into the entry zone. At higher rolling speed, have already reached their respective minimum 5 and 6). At the moment that skidding be-
the escape of excess fluid from the diminishing points, begin to increase (lines 1 to 4 in Fig. 17). gins, the strip stops.
gap of the labyrinth between the rolls and the For increasing values of strip thickness (lines 2, For relatively thin strip, another critical
strip is relatively slower. The entrapped lubricant 3, and 4 in Fig. 17), the forward slip decreases, speed (the critical speed of the second kind)
is then pressurized and causes partial separation and the critical value of roll speed at which the can be reached, even before skidding com-
between the rolls and the strip. At low speed, neutral angle and forward slip reach a minimum mences. When this occurs, the point of the
pressure is transmitted from the rolls to the strip is increased. The values of the neutral angle and entry zone will reach the exit, and hydrody-
through metal-to-metal contact. At higher forward slip at which hydrodynamic lubrication namic lubrication will commence. A further
speeds, more of the pressure is transmitted begins increase with increasing thickness values. increase in speed will cause a thicker lubri-
through the entry zone and through the en- At a critically high strip-thickness value, hy- cating film to separate the rolls from the
trapped lubricant. As the pressure transmitted drodynamic lubrication commences when for- strip and make the rolling more stable (lines
through metal-to-metal contact is reduced with ward slip and the value of the neutral angle are 1, 2, and 3 in Fig. 17).
higher speeds, the friction is decreased. zero (line 4 in Fig. 17). When the thickness of For relatively thick strip, even when a criti-
Eventually, at high speeds, no metal-to-metal the strip is above the stated critical value, for cal speed of the first kind commences first,
contact exists. lines 5 to 7, friction drops to below the minimum a further increase in speed may ultimately
When conditions for complete hydrodynamic required for rolling before hydrodynamic lubri- produce a critical speed of the second kind
lubrication are reached (Fig. 16), friction is at its cation can be established. Skidding then com- and reestablish hydrodynamic lubrication
minimum value. Therefore, with increased mences at the critical roll speed. It can be ob- (lines 4 to 7 in Fig. 17).
speeds, high shear rates are created in the liquid served that at higher values of speed, referred to
Changes in the entry zone, the friction hill, the
where the shear stress is proportional to the here as the minimum required speed, rolling
neutral point, and strip thickness as a function of
shear strain rate to the second power, and fric- with hydrodynamic lubrication may be reestab-
speed for thick and thin strip are shown
tion rises at a mild rate. Friction values are much lished. It should also be noted that the minimum
schematically in Fig. 18.
lower when hydrodynamic lubrication prevails required speed can be reduced by the use of lu-
than when metal-to-metal contact is prevalent. bricants of higher viscosity.
With increasing speed, the friction hill effect The following variations in the friction hill are Mechanics of Plate Rolling
is also reduced, and roll-separation force and expected with increasing speed:
roll flattening become less pronounced. With re-
In rolling of thick plates, metal flow occurs in
duced roll-separation force, the elastic stretching The peak of the friction hill gets lower and,
three dimensions. The rolled material is elon-
of the mill as well as roll bending and roll flat- together with the neutral point, shifts closer
gated in the rolling direction as well as spread in
tening are reduced, causing the gap between the to the exit.
the lateral or width direction. Spread in rolling is
usually defined as the increase in width of a plate
Fig. 17 Forward slip and position of the neutral point

versus rolling velocity. See text for explana-
Fig. 16 Hydrodynamic lubrication during strip rolling. (a) Overall schematic. (b) Shear in the lubricant film tion of numbered curves.
or slab expressed as a percentage of its original volves a rather complex analysis and requires (c), respectively. As expected from the slab
width. The spread increases with increasing re- the use of computerized techniques (Ref 38, 45, analysis, the stress distributions are very similar
duction and interface friction, decreasing plate 46). A modular upper-bound method has been to those illustrated for strip rolling in Fig. 12 to
width-to-thickness ratio, and increasing roll-di- used to predict metal flow, spread, elongation, 14. By use of a numerical approach similar to
ameter-to-plate-thickness ratio. In addition, the and roll torque (Ref 38). The principles of this that discussed for strip rolling, detailed predic-
free edges tend to bulge with increasing reduc- method are described subsequently. Figure 19 tions of stresses, in both the longitudinal and lat-
tion and interface friction. The three-dimensional (Ref 38) illustrates the coordinate system, the di- eral directions, can be made. The stresses are
metal flow that occurs in plate rolling is difficult vision of the deformation zone into elements, calculated by assuming the frictional shear stress
to analyze. Therefore, most studies of this and the notations used. The spread profile is de- t to be constant, as in the case of upper-bound
process have been experimental in nature. fined in terms of a third-order polynomial w(x), analysis. Thus, the stress distribution at various
Several empirical formulas have been established with two unknown coefficients, a1 and a2. The planes along the width, or y, direction (Fig. 20)
for estimating spread (Ref 4244), with some location of the neutral plane xN is another un- is linear on both sides of the plane of symmetry.
theoretical attempts to predict elongation or known quantity. The following kinematically The stress distribution in the rolling, or x, direc-
spread (Ref 4547). Once the spread has been es- admissible velocity field (Ref 48) is used: tion is calculated along the streamlines of metal
timated, the elongation can be determined from flow (Fig. 19). At each node of the mesh, the
the volume constancy, or vice versa. vx = 1/[w(x)h(y)] (Eq 31) lower of the sz values is accepted as the actual
An Empirical Method for Estimating stress. Thus, a tentlike stress distribution is ob-
Spread. Among the various formulas available tained (Fig. 21). Integration of the stresses act-
for predicting spread, one formula (Ref 43) is 1 d 1 ing on the plane of contact gives the roll-sepa-
vy =
used most extensively and is given as: h( x ) dx w( x ) rating force.
(Eq 32)
w1/w0 = abcd(h0/h1)P (Eq 30)
1 d 1 Shape Rolling
vz =
w( x ) dx h( x ) (Eq 33)
where w1 and w0 are the final and initial widths
of the plate, respectively; h1 and h0 are the final Rolling of shapes, also called caliber rolling,
and initial thicknesses of the plate, respectively; Using Eq 31 to 33, the upper-bound method is one of the most complex deformation
P equals 10(1.269) (w0/h0)(h0/D)0.556; D is the ef- can be applied to predict spread. A computer processes. A round or round-cornered square
fective roll diameter; and a, b, c, and d are con- program, SHPROL, can be used for some steps bar, billet, or slab is rolled in several passes into
stants that allow for variations in steel composi- in the analysis. More information on SHPROL is relatively simple sections, such as rounds,
tion, rolling temperature, rolling speed, and roll available in the section Shape Rolling in this squares, or rectangles, or into complex sections,
material, respectively. These constants vary chapter. such as L, U, T, H, or other irregular
slightly from unity, and their values can be ob- Prediction of Stresses and Roll-Separating shapes (Ref 49). For this purpose, certain inter-
tained from the literature (Ref 38, 43, 47). Force. Once the spread (the boundaries of the mediate shapes or passes are used, as shown in
An empirical formula for predicting spread, deformation zone) has been calculated, this in- Fig. 22 for the rolling of angle sections (Ref 50).
such as Eq 30, gives reasonable results within formation can be used to predict the stresses and The design of these intermediate shapes, that is,
the range of conditions for the experiments from the roll-separating force. The computerized pro- roll pass design, is based on experience and dif-
which the formula was developed. There is no cedure used here is, in principle, the same as the fers from one company to another, even for the
formula that will make accurate predictions for method discussed earlier for predicting the same final rolled section geometry.
all the conditions that exist in rolling. Thus, it is stresses in strip rolling (Ref 38). The art of roll pass design, known since the
often necessary to attempt to estimate spread or The deformation zone under the rolls is di- late 17th century, has turned into a modern tech-
elongation by theoretical means. vided into trapezoidal slabs by planes normal to nology based on scientific foundations (Ref 51).
The theoretical prediction of spread in- the rolling direction and along the stream tubes, The purpose of roll pass design is to ensure the
as illustrated in Fig. 13 and 20 (Ref 38). The production of the correct size and shape of a
stresses acting on strips in the rolling and trans- product with defect-free surface and intended
verse directions are illustrated in Fig. 20(b) and mechanical properties, and at the same time en-
sure maximum output at lowest cost, ease the
working conditions of the rolling crew, and min-
imize the roll wear. In the past, relatively few
quantitative data on roll pass design have been
available in the literature. Today, computer-aided
roll design and FEMs are established for design
of roll passes with computer numerical control
machines for roll turning and computer-inte-
grated roll manufacturing. In the future, roll pass
design is expected to become more advanced for
thermomechanical processing (Ref 51).
Basically, there are two methods for rolling
shapes or sections. The first method is universal
rolling (Fig. 23). The second method is caliber
rolling (Fig. 22, 24). In universal rolling, the mill
and stand constructions are more complex.
However, in the rolling of I-beams or other sim-
ilar sections, this method allows more flexibility
than does caliber rolling and requires fewer
passes. This is achieved because this method
Fig. 18 Variations in the friction hill versus speed. Fig. 19 Configuration of deformation and the grid
provides appropriate amounts of reductions,
The parameters m, R0, sxb, sxf, and percent system used in the analysis of the rolling of
reduction are constant. thick plates. Source: Ref 38 separately in webs and flanges.
two plate sections (A and B in Fig. 25) with in-

dividual cross-sectional areas Aa and Ab, then the
following weighted-average formula can be
used:
A0 A + Ab 0 A l + Ab1 l b
lm = = a0 = a1 a
A1 AA1 + Ab1 Aa1 + Ab1 (Eq 34)
where l is the elongation coefficient (that is, the

cross-sectional area at the entrance divided by
the cross-sectional area at the exit); A is the
cross-sectional area; m is a subscript denoting
average; a and b are subscripts denoting section
portions A and B; and 0 and 1 are subscripts de-
noting entrance and exit values, respectively.
Computer-Aided Roll Pass Design. Estima-
tion of the number of passes and of the roll geom-
etry for each pass is the most difficult aspect of
shape rolling. To accomplish this, certain factors,
discussed subsequently, must be considered.
Fig. 20 Stress analysis of the rolling of plates. (a) Top view of the rolled plate. (b) Stresses in the rolling direction. (c)
Stresses in the transverse direction. Source: Ref 38
For successful rolling of shapes, it is neces- and B, were completely independent of each
sary to estimate the following for each stand: the other (Fig. 25b), zone B would be much more
roll-separating force and torque, the spread and elongated than zone A. However, the two zones
elongation, and the appropriate geometry of the are connected and, as part of the rolled shape,
roll cavity or caliber. The force and torque can must have equal elongation at the exit from the
be estimated either by using empirical formulas rolls. Therefore, during rolling, metal must flow
or by approximating the deformation in shape from zone B into zone A so that a uniform elon-
rolling with that occurring in an equivalent plate gation of the overall cross section is obtained
rolling operation. In this case, the equivalent (Fig. 25c). This lateral flow is influenced by the
plate has initial and final thicknesses that corre- temperature differences that exist in the cross
spond to the average initial and final thicknesses section because of variations in material thick-
of the rolled section. The load and torque calcu- ness and heat flow.
lations can be performed for the equivalent To estimate the overall elongation, it is neces-
plate, as discussed earlier in this chapter for sary to divide the initial section into a number of
plate rolling. The results are approximately valid equivalent plates (A, B, C, and so forth), as
for the rolled shape being considered. shown in Fig. 25. The elongation for an individ-
Estimation of Elongation. During the ual section, without the combined influence of
rolling of a given shape or section, the cross other portions of the section, can be estimated by
section is not deformed uniformly, as can easily using both the plate rolling analogy and the
be seen in Fig. 24. This is illustrated further in techniques discussed in this chapter. The com-
Fig. 25 for a relatively simple shape. The reduc- bined effect can be calculated by taking a
tions in height for zones A and B are not equal weighted average of the individual elongations.
(Fig. 25a). Consequently, if these two zones, A For example, if the original section is to be di-
vided into an equivalent system consisting of
Fig. 23 Universal rolling of flanged beams. (a) Univer-

Fig. 21 The calculated stress (sz) distribution in plate Fig. 22 Five possible roll pass designs for the rolling sal roll stand. (b) Edging stand. (c) Finishing
rolling shown three-dimensionally. Source: Ref 38 of a steel angle section. Source: Ref 50 stand
tain an acceptable degree of fill of the roll cal-

iber used for that pass.
Computer-Aided Roll Pass Design of
Airfoil Sections. Two examples of computer
programs developed for the analysis of rolling
airfoil sections are the SHPROL and ROLPAS
programs (Ref 39). The first of these programs,
SHPROL, uses upper-bound analysis in a nu-
merical form to predict spread and roll torque.
SHPROL is based on the following simplifying
assumptions:
The initial contact between the rolls and the
entrance section can be approximated as a
straight line. (This is only correct if the
upper and lower surfaces of the initial sec-
tion already have the shape of the rolls.)
An airfoil shape can be considered as an ag-
gregate of slabs, as shown in Fig. 26 (Ref
39).
Plane sections perpendicular to the rolling
Fig. 24 Analysis of a roll stand used in rolling of rails. Sketches 1 through 5 illustrate the stock in broken lines and
direction remain plane during rolling. Thus,
the roll in solid lines at various positions in the deformation zone.
the axial velocity (velocity in the rolling, or
x, direction) at any section perpendicular to
The Characteristics of the Available Installa- are predicted by using an empirical formula. If
the rolling direction is uniform over the en-
tion. These include diameters and lengths of the the elongation per pass is known, it is then pos-
tire cross section.
rolls, bar dimensions, distance between roll sible, by use of computer graphics, to calculate
The velocity components in the transverse,
stands, distance from the last stand to the shear, the cross-sectional area of a section for a given
or y, direction and in the thickness, or z, di-
and tolerances that are required and that can be pass, that is, the reduction and the roll geometry.
rection are functions of x and linear in the y
maintained. The roll geometry can be expressed parametri-
and z coordinates, respectively.
The reduction per pass must be adjusted so that cally (in terms of angles, radii, and so forth).
the installation is used at a maximum capacity, the These geometric parameters can then be varied In Fig. 26, each element is considered to be a
roll stands are not overloaded, and roll wear is to optimize the area reduction per pass and ob- plate for which it is possible to derive a kine-
minimized. The maximum value of the reduction matically admissible velocity field. The total en-
per pass is limited by the excessive lateral metal ergy dissipation rate of the process ET is:
flow, which results in edge cracking; the power
and load capacity of the roll stand; the require- ET = EP + ED + EF (Eq 35)
ment for the rolls to bite in the incoming bar; roll
wear; and tolerance requirements. where EP is the energy rate of plastic deforma-
At the present stage of technology, the previ- tion and is calculated for each element by inte-
ously mentioned factors are considered in roll grating the product of flow stress and the strain
pass design by using a combination of empirical rate over the element volume, ED represents the
knowledge, some calculations, and some edu- energy rates associated with velocity discontinu-
cated guesses. A methodical way of designing
roll passes requires not only an estimate of the
average elongation, as discussed earlier, but also
the variation of this elongation within the defor-
mation zone. The deformation zone is limited by
the entrance, where a prerolled shape enters the
rolls, and by the exit, where the rolled shape
leaves the rolls. This is illustrated in Fig. 24. The
deformation zone is cross-sectioned with several
planes (for example, planes 1 to 5 in Fig. 24; 1
is at the entrance, 5 is at the exit). The roll posi-
tion and the deformation of the incoming billet
are investigated at each of these planes. Thus, a
more detailed analysis of metal flow and an im-
proved method for designing the configuration
of the rolls are possible. It is evident that this
process can be drastically improved and made
extremely efficient by the use of computer-aided
techniques.
In recent years, most companies that produce
shapes have computerized their roll pass design
procedures for rolling rounds or structural
shapes. In most of these applications, the elon- Fig. 25 Nonuniform deformation in the rolling of a
gation per pass and the distribution of the elon-
shape. (a) Initial and final sections. (b) Two Fig. 26 Configuration of deformation zone in the ap-
zones of the section considered as separate plates. (c) plication of numerical upper-bound analysis
gation within the deformation zone for each pass Direction of lateral metal flow to the rolling of airfoil shapes. Source: Ref 39
ities and is due to internal shear between the el- ability. One example, based on a simplified fi- of computer graphics for roll caliber and roll
ements, and EF is the energy rate due to friction nite-element-slab method, is discussed in the pass design. In this case, the cross section of a
between the rolls and the deforming material. section Modeling of Strip Rolling in this chap- rolled shape is described in general form as a
The total energy dissipation rate ET is a func- ter. Methods of experimental design are also polygon. Each corner or fillet point of the poly-
tion of unknown spread profiles w1 and w2 (Fig. used with the FEM of process simulation to argon is identified with the x- and y-coordinates
26) and the location of the neutral plane xN. As rive at robust roll pass designs (Ref 56). and with the value of the corresponding radius
in the analysis discussed earlier for plate rolling, One example of a computer program called (Fig. 28) (Ref 58). Thus, any rolled section can
the unknown coefficients of w1, w2, and xN are for establishing roll cross sections and pass be represented by a sequence of lines and cir-
determined by minimizing the total energy rate. schedules is RPDROD (Ref 57). RPDROD uses cles. This method of describing a rolled section
The computer program SHPROL uses the follow- an empirical formula for estimating the variation is very general and can define a large number of
ing as input data: roll and incoming shape geom- of the spread in the roll bite and parametrically sections with a single computer program. Lines
etry, friction, flow stress, and roll speed. SH- described alternative roll caliber designs. When or circles that are irrelevant in a specific case can
PROL can predict the energy dissipation rates, the using this program, the designer obtains an opti- be set equal to zero. Thus, a simpler section,
roll torque, and, most important, the amounts of mal roll pass schedule by evaluating a number of with a smaller number of corner and fillet points,
elongation and spread within one deformation alternatives in which individual pass designs are can be obtained. For example, in the rolling of
zone in the rolling of any airfoil shape. selected from a variety of caliber shapes com- the symmetric angle section shown in Fig. 22,
The second program, called ROLPAS, uses monly used in rod rolling. several intermediate section passes are required.
interactive graphics and is capable of simulating The computer program RPDROD consists of Such an intermediate section is shown in para-
the metal flow in the rolling of relatively simple four modules, called Stock, Schedule, Groove, metric representation in Fig. 29. In this figure,
shapes, such as rounds, plates, ovals, and airfoils and Metal Flow. The Stock design module al- all the geometric variables can be modified to
(Fig. 27). ROLPAS uses the following as input: lows the user to design/specify the entry cross change the cross-sectional area and/or the
the geometry of the initial section, the geometry section for the first pass in the schedule. A amount of reduction per pass. These variables,
of the final section, the flow stress of the rolled square, rectangular, or round stock cross section which fully describe this section, are:
material and the friction factor, and the varia- can be defined. The Schedule design module al-
SELA, length (of one leg) at centerline
tions in elongation and spread in the rolling di- lows the user to design the roll pass schedule by
BETAG, angle at top corner
rection, as calculated by the SHPROL program. providing specific functions:
RK, radius at top corner
To simulate the rolling process, ROLPAS di-
Add a new pass to the roll pass schedule, by AL, length of straight portion at top
vides the deformation zone into a number of
estimating alternative roll cross section di- RD, radius of leg at top
cross sections parallel to the roll axis (Fig. 19,
mensions from design data provided by PRST, projection of draft angle
27). The simulation is initiated by considering
spread/elongation calculations RRU, radius at lower tip of leg
the cross-sectional area, stresses present, and the
Delete pass design data from the schedule in RH, radius at bottom corner
roll-separating force and torque for the first sec-
order to investigate alternative pass designs
tion. These same analyses can then be performed In establishing the final section geometry, the
Review and/or provide hard copy of exist-
on any succeeding section. designer assigns desired values to the variables
ing pass design data
Computer-Aided Roll Pass Design for listed previously and, in addition, inputs the de-
Round Sections. Computer-aided methods for The Schedule design module allows the user sired cross-sectional area and the degree of cal-
designing caliber rolls for rod rolling have been to design an optimal roll pass schedule by inves- iber fill, for example, the desired ratio of rolled
used since the1960s and 1970s (Ref 5255), and tigating various alternative pass design and/or section area versus section area on the caliber
advances continue to be made in improving shape combinations. In principle, any roll cross- rolls. Thus, there is only one geometric variable
product quality and reducing dimensional vari- sectional shape considered by the program could
be used for a given pass in the schedule.
However, RPDROD has facilities for checking
input data and thus for preventing the selection
of an illogical pass design or the inappropriate
selection of roll cross-sectional shape combina-
tions.
The Groove design module can be used to
change the initially suggested roll cross section
dimensions, as the user deems appropriate. As in
the Schedule module, input-checking facilities
ensure that specified roll cross-sectional dimen- Fig. 28 Geometric representation of a rolled section
as a polygon. Source: Ref 58
sions are consistent with the chosen roll cross-
sectional shape and bar entry cross section.
The Metal Flow design module provides the
user with details of metal-flow simulation, in-
cluding the calculated cross sections of the de-
forming bar in the roll bite, stresses in the de-
forming material, roll-separating load, and roll
torque. For this purpose, this module uses the
ROLPAS program discussed earlier for the
rolling of airfoil shapes.
Computer-Aided Roll Pass Design of
Structural and Irregular Sections. Many com-
panies use computer-aided graphics for the de-
sign and manufacture of the caliber shapes for
Fig. 27 Selected simulation steps as displayed by
the rolling of structural sections. Reference 58 Fig. 29 Parametric representation of an intermediate-
ROLPAS for a test airfoil shape cold rolled shaped rolling pass for a symmetric angle sec-
from rectangular steel stock gives an excellent summary of the practical use tion. See also Fig. 30. Source: Ref 58
that is calculated by the computer program, and templates and the graphite electrical discharge flow and the bulge profile in flat rolling of slabs,
that is the thickness of the leg of the angle sec- machining electrodes used in manufacturing the the shape rolling of a simple H section, and
tion. In the example shown in Fig. 30(a) (Ref necessary cutting tools for roll machining (Ref the rolling of a practical H-beam section. The ef-
58), the leg thickness is calculated to be 18.2 58). fect of roll pass sequence design on the roll
mm (0.72 in.). The designer compares this sec- Finite-element modeling (FEM) is used in forces and the quality (distribution of strain) of
tion geometry (Fig. 30a) with the caliber geom- the analysis of three-dimensional (3-D) metal the rolled product can be studied. With reduced
etry of the next pass that has been generated in a flow and of stress and strain distributions in simulation time per pass, the code has potential
similar way. Assume that the section shown in shape rolling; the laborious task of designing use in the redesign of roll pass schedules for im-
Fig. 30(a) appears to be too long; that is, SELA rolls for complex shapes can be simplified by proved product properties (Ref 60).
is 67.5 mm (2.66 in.) and should be reduced to FEM. Continued improvements in the speed of
65 mm (2.56 in.) without modifying the other computer calculations enhance the use of FEM,
variables. The interactive program is rerun with and computer programs have been developed Summary
the new value for SELA. The modified section, from the simplification of 3-D numerical meth-
shown in Fig. 30(b), is slightly thicker than the ods to simulate the shape-rolling process. For This chapter briefly introduces the mechanics
original section in order to maintain the same example, shape rolling can be modeled by a fi- of rolling with descriptions of plane-strain
cross-sectional area. nite- and slab-element method (FSEM), which rolling, strain distributions, and lubrication.
This interactive graphics program does not in- uses a rigid-plastic two-dimensional FEM for Some friction is needed to draw the pieces into
volve any analysis of metal flow or stresses. the generalized plane-strain condition combined the roll gap, and frictional stresses increase in
Nevertheless, it is extremely useful to the de- with the slab method (Ref 59). This reduces the the deformation zone until interface pressure
signer for modifying section geometries quickly computational effort without losing much accu- reaches an apex roughly at the neutral point
and accurately, calculating cross-sectional areas, racy obtained in the 3-D computer simulation of where the surface speed of the workpiece and
and cataloging all this geometrical information the shape-rolling process. The FSEM has been roll are equal. The mechanics of this process are
systematically. The program also automatically used to develop a computer program, called briefly described along with the factors that in-
prepares engineering drawings of the sections TASKS, for 3-D analysis of shape rolling as a fluence the homogeneity of deformation in strip,
and the templates for quality control as well as kinematically steady-state process. The TASKS plate, and shape rolling. Friction, together with
tapes for numerically controlled milling of the program has been used to simulate the metal pass geometry, determines the homogeneity of
deformation in several ways, as discussed in this
chapter. The idealized case of plane-strain
rolling is introduced first, and strip-rolling the-
ory is described in more detail. Plate rolling me-
chanics and shape rolling are also discussed.
ACKNOWLEDGMENTS
This article was adapted from Tribology in

Metalworking, by John Schey, American Society
for Metals, 1983, p 249333 and from Flat,
Bar, and Shape Rolling, by G.D. Lahoti and
S.L. Semiatin, Forming and Forging, Volume 14
of the ASM Handbook, ASM International,
1988, p 343360.
REFERENCES
1. J. Schey, Rolling, Tribology in Metalwork-

ing, American Society for Metals, 1983, p
249
2. G.D. Lahoti and S.L. Semiatin, Flat, Bar, and
Shape Rolling, Forming and Forging, Vol 14,
ASM Handbook, 1988, p 343360
3. J. Schey, Friction in Metalworking, Tribol-
ogy in Metalworking, American Society for
Metals, 1983, p 15
4. S. Ekelund, Steel, Vol 93 (Aug 21), 1933, p
2729
5. H. Ford, F. Ellis, and D.R. Bland, J. Iron
Steel Inst., Vol 168, 1951, p 5772
6. W.L. Roberts, Cold Rolling of Steel,
Dekker, 1978
7. H. Inhaber, J. Eng. Ind. (Trans. ASME) B,
Vol 88, 1966, p 421429
8. W. Evans and B. Avitzur, J. Lubr. Technol.
(Trans ASME) F, Vol 90, 1968, p 7280
9. W. Dobrucki and E. Odrzywolek, J. Mech.
Fig. 30 (a) Alternative design for an intermediate-angle pass obtained by assigning values to the parameters given in Fig.
Work. Technol., Vol 4, 1980, p 263284 (in
29. (b) Modified design of the intermediate-angle pass shown in (a), with new dimensions (left) and added
and subtracted areas (right). Source: Ref 58 French)
10. J.M. Capus and M.G. Cockroft, J. Inst. plastic Finite-Element Model for Cold 46. R. Kummerling and H. Lipmann, On
Met., Vol 92, 19631964, p 3132 Rolling of Thin Strip, Int. J. Mech. Sci., Vol Spread on Rolling, Mech. Res. Commun.,
11. B.B. Hundy and A.R.E. Singer, J. Inst. Met., 34 (No. 3), March 1992, p 195210 Vol 2, 1975, p 113
Vol 83, 19541955, p 401407 32. C.G. Sun and S.M. Hwang, A Finite 47. H. Neumann, Design of Rolls in Shape
12. E. Siebel and W. Lueg, Arch. Eisenhtten- Element Computer System for Analysis of Rolling, VEB Deutscher Verlag, 1969 (in
wes., Vol 15, 1933, p 114 (in German) Mechanical and Thermal Behaviors in Hot German)
13. G.D. Lahoti, N. Akgerman, S.I. Oh, and T. Strip Rolling, Second International Con- 48. R. Hill, A General Method of Analysis for
Altan, J. Mech. Work. Technol., Vol 4, 1980, ference on Modelling of Metal Rolling Metalworking Processes, Inst. J. Mech. Sci.,
p 105119 Processes, 911 Dec 1996 (London), Vol 16, 1974, p 521
14. M. Vater, G. Nebe, and J. Petersen, Stahl Institute of Materials, 1996, p 8491 49. R.E. Beynon, Roll Design and Mill Layout,
Eisen, Vol 86, 1966, p 710720 (in German) 33. Z.-C. Lin and C.-C. Shen, A Three- Association of Iron and Steel Engineers,
15. T. von Karman and Z. Angew, Mat. Mech., Dimensional Finite Element Method for a 1956
Vol 5, 1925, p 139141 Nonisothermal Aluminum Flat Strip 50. A. Schutza, Comparison of Roll Pass
16. E. Orowan, The Calculation of Roll Rolling Process, J. Mater. Eng. Perform., Designs Used for Rolling Angle Sections,
Pressure in Hot and Cold Flat Rolling, Proc. Vol 5 (No. 4), Aug 1996, p 452461 Stahl Eisen, Vol 90, 1970, p 796 (in
Inst. Mech. Eng., Vol 150, 1943, p 140 34. K. Narita and K. Yasuda, Three- German)
17. J.T. Hockett, Calculation of Rolling Forces Dimensional Rigid-Plastic Finite Element 51. S.-E. Lundberg, Roll Pass Design: The Key
Using the Orowan Theory, Trans. ASM, Vol Analysis of Strip Crown and Flatness Function in Control of Shape, Dimension
52, 1960, p 675 During Rolling with Constraint Boundary and Mechanical Properties of Hot Rolled
18. R.B. Sims, The Calculation of Roll Force Conditions Proposed by Shohet, METEC Products, Scand. J. Metall., Vol 26 (No. 3),
and Torque in Hot Rolling Mills, Proc. Inst. Congress 94, Second European Continuous June 1997, p 102114
Mech. Eng., Vol 168, 1954, p 191 Casting Conference and Sixth International 52. P.S. Raghupathi and T. Altan, Roll Pass
19. B. Avitzur, An Upper-Bound Approach to Rolling Conference, Vol 2, 2022 June Design in Shape Rolling, unpublished re-
Cold Strip Rolling, J. Eng. Ind. (Trans. 1994, Verein Deutscher Eisenhuttenleute view of German literature, Battelle
ASME), Feb 1964, p 31 (Dusseldorf), 1994, p 430435 Columbus Laboratories, 1980
20. E.G. Thomsen, C.T. Yang, and S. 35. G.E. Dieter, in Mechanical Metallurgy, 53. H. Neumann and R. Schulze, Programmed
Kobayashi, in Mechanics of Plastic Defor- McGraw Hill, 1961, p 488 Roll Pass Design for Blocks, Neue Htte,
mation in Metal Processing, Macmillan, 36. T. Altan and R.J. Fiorentino, Prediction of Vol 19, 1974, p 460 (in German)
1965, p 373 Loads and Stresses in Closed Die Forging, 54. H. Gedin, Programmed Roll Pass Design
21. G.W. Rowe, in Principles of Industrial J. Eng. Ind. (Trans. ASME), May 1971, p for Quality Steels, Kalibreur, Vol 11, 1969,
Metalworking Processes, Edward Arnold, 477 p 41 (in German)
1968, p 208 37. N. Akgerman and T. Altan, Application of 55. U. Suppo, A. Izzo, and P. Diana, Electronic
22. J.M. Alexander, On the Theory of Rolling, CAD/CAM in Forging Turbine and Computer Used in Roll Design Work for
Proc. R. Soc. (London) A, Vol 326, 1972, p Compressor Blades, J. Eng. Power (Trans. Rounds, Kalibreur, Vol 19, Sept 1973, p 3
535 ASME) A, Vol 98 (No. 2), April 1976, p 290 56. K. Yoshimura, S. Kini, and R. Shivpuri, A
23. G.D. Lahoti, S.N. Shah, and T. Altan, 38. G.D. Lahoti et al., Computer Aided Technique for Robust Design of Roll Passes
Computer Aided Analysis of the Analysis of Metal Flow and Stresses in for Consistent Rod Quality, 38th Mechan-
Deformations and Temperatures in Strip Plate Rolling, J. Mech. Work. Technol., Vol ical Working and Steel Processing Confer-
Rolling, J. Eng. Ind. (Trans. ASME), Vol 4, 1980, p 105 ence Proceedings, Vol 34, USA, 1316 Oct
100, May 1978, p 159 39. N. Akgerman, G.D. Lahoti, and T. Altan, 1996 (Cleveland, OH), Iron and Steel
24. H. Ford and J.M. Alexander, Simplified Computer Aided Roll Pass Design in Society/American Institute of Mining,
Hot-Rolling Calculations, J. Inst. Met., Vol Rolling of Airfoil Shapes, J. Appl. Metallurgical and Petroleum Engineers,
92, 19631964, p 397 Metalwork., Vol 1, 1980, p 30 1997, p 241250
25. D.J. McPherson, Contributions to the 40. J.H. Hitchcock, Roll Neck Bearings, re- 57. K.F. Kennedy, G.D. Lahoti, and T. Altan,
Theory and Practice of Cold Rolling, search committee report, American Society Computer Aided Analysis of Metal Flow
Metall. Trans., Vol 5, Dec 1974, p 2479 of Mechanical Engineers, 1935; cited in Stresses and Roll Pass Design in Rod
26. G.J. Li and S. Kobayashi, J. Eng. Ind. L.R. Underwood, The Rolling of Metals, Rolling, Iron Steel Eng., Vol 60 (No. 6),
(Trans. ASME) B, Vol 104, 1982, p 5564 Vol I, John Wiley and Sons, 1950, p 1516 June 1983, p 5054
27. P.F. Thomson and J.H. Brown, Int. J. Mech. 41. B. Avitzur, Friction during Metal Forming, 58. F. Schmeling, Computer Aided Roll Pass
Sci., Vol 24, 1982, p 559576 Friction, Lubrication, and Wear Technology, Design and Roll Manufacturing, Stahl
28. R.H. Wagoner and J.-L. Chenot, in Funda- Vol 18, ASM Handbook, 1992, p 6365 Eisen, Vol 102, 1982, p 771 (in German)
mentals of Metal Forming, John Wiley & 42. S. Ekelund, in Roll Pass Design, VEB 59. N. Kim, S. Kobayashi, and T. Altan, Three-
Sons, 1997 Deutscher Verlag, 1963, p 48 (in German) Dimensional Analysis and Computer
29. R.H. Wagoner and J.-L. Chenot, Metal 43. Z. Wusatowski, Hot Rolling: A Study of Simulation of Shape Rolling by the Finite
Forming Analysis, Cambridge University Draught, Spread and Elongation, Iron Steel, and Slab Element Method, Int. J. Mach.
Press, 2001, p 210212 Vol 28, 1955, p 69 Tools Manuf., Vol 31 (No. 4), 1991, p
30. V.-K. Hu and W.K. Liu, An ALE Hydrody- 44. L.G.M. Sparling, Formula for Spread in Hot 553563
namic Lubrication Finite Element Method Rolling, Proc. Inst. Mech. Eng., Vol 175, 60. W. Shin, S.M. Lee, R. Shivpuri, and T.
with Application to Strip Rolling, Int. J. 1961, p 604 Altan, Finite-Slab Element Investigation of
Numer. Methods Eng., Vol 36 (No. 5), 45. S.I. Oh and S. Kobayashi, An Approximate Square-to-Round Multi-Pass Shape Rolling,
March 1993, p 855880 Method for Three-Dimensional Analysis of J. Mater. Process. Technol., Vol 33 (No.
31. P. Gratacos, C. Fromholz, J.L. Chenot, and Rolling, Int. J. Mech. Sci., Vol 17, 1975, p 12), Aug 1992, p 141154
P. Montmitonnet, A Plane-Strain Elasto- 293
Chapter 17
Thermomechanical Processing by
Controlled Rolling
THERMOMECHANICAL PROCESSING may eliminate the need for subsequent heat In contrast to CHR methods, controlled
(TMP) is a term referring to a variety of metal treatment. rolling involves special methods to control the
forming processes that involve careful control microstructural condition of austenite mi-
of thermal and deformation conditions to crostructure during hot rolling. For example, if
achieve advantageous microstructures and im- Fundamentals of Controlled Rolling the hot rolling takes place below the recrystal-
provements in properties (particularly the re- (Ref 1) lization stop temperature (TRXN), the austenite
finement of grain size to improve strength and grains become highly elongated and, at a suffi-
toughness). Perhaps the most common type of In general terms, controlled rolling refers to ciently large strain, become filled with intra-
TMP application is the controlled rolling of mi- rolling processes designed for strict temperature granular defects such as deformation bands and
croalloyed steels. Controlled rolling involves and deformation control to obtain specific ob- twins. This technique, which perhaps represents
the careful conditioning of austenite during hot jectives for austenite conditioning. Obviously, the most well-known and common variant of
deformation so that the austenite transforms to a all hot-rolling practices occur under some sort of thermomechanical processing, is referred to as
fine-grain ferrite in the final as-rolled product. temperature and deformation control, but not all conventional controlled rolling (CCR). In the
Similar concepts also apply to steel bar and rolling practices are designed to manipulate the CCR method, finish rolling is typically done
forgings, although TMP applications for these condition of austenite prior to transformation. below the recrystallization stop temperature
types of microalloying steel products have The general difference between conventional hot (method B in Fig. 1) so that deformation results
lagged behind that of flat-rolled steel products. rolling (CHR) and various forms of thermome- in very elongated (pancake-like) austenite grains
The basic objective of TMP, regardless of form, chanical processing is perhaps most simply de- with intragranular crystalline defects, which
is to ensure and/or improve properties through scribed using Fig. 1. The method denoted as A in then transform into very fine ferrite grain sizes
the control of microstructural changes during Fig. 1 is conventional hot rolling, which shows during cooling.
hot deformation. The concept of TMP also ap- that the reheating, rough rolling, and finish In another method of controlled rolling, if the
plies to nonferrous systems, such as the forging rolling all occur at the highest possible tempera- rolling temperatures are high to allow recrystal-
of titanium alloys and nickel-base superalloys tures. The primary goal of conventional hot lization, then deformed equiaxed grains of
(e.g., see Chapter 11, Design for Deformation rolling is to optimize productivity. austenite recrystallize into a different set of
Processes).
This chapter focuses on the controlled rolling
of steel, which is probably the most significant
form of TMP application. The application of
TMP methods for microalloyed steel bar and
forgings, which involve different methods and
justifying objectives than flat-rolled product, is
briefly addressed in Chapter 11, Design for
Deformation Processes. In addition, Chapter 3,
Evolution of Microstructure during Hot
Working, describes the general effects of ther-
momechanical processing on microstructure
evolution with attention on the key processes
that control microstructure evolutionthat is,
dynamic and static recovery and recrystalliza-
tion and grain growth. These topics are ad-
dressed in more detail in this chapter, as they re-
late to the methods of controlled rolling. By
careful process control (and the use of microal-
loying), controlled rolling of steels results in im-
proved properties and advantages in subsequent
operations. For example, improvement in
strength from controlled rolling may allow re-
duction in carbon contents, which thus enhances
the weldability of the product. Improved me- Fig. 1 Comparison of selected thermomechanical treatments based on critical austenite temperatures, transformation
temperatures, and rough and finish rolling operations. A, conventional hot rolling; B, conventional controlled
chanical properties of as-rolled products also rolling; C, intensified controlled rolling; D, recrystallization controlled rolling
Chapter 17: Thermomechanical Processing by Controlled Rolling / 249
equiaxed grains that differ from the original tate, or second-phase, compounds. The extent pect ratio of the deformed grains are required to
chiefly in size. This type of hot-rolling process, and conditions of this partitioning of solutes be- determine the grain boundary surface per unit
which leads directly to fine equiaxed grains, is tween solid solution and precipitates in austenite volume. In general, because the ferrite grain size
referred to as recrystallization controlled rolling have a strong influence on the migration and is related to the number of sites for ferrite nucle-
(RCR). This form of thermomechanical process- movement of austenite grains during hot defor- ation, and because this number is related to the
ing involves repeated recrystallization of austen- mation. This is explained in more detail in the austenite grain boundary area per unit volume,
ite by both rough rolling and finish rolling above section Restoration Processes. The ability of the ferrite grain size will decrease as this bound-
the recrystallization stop temperature (method D an element to remain a solute or precipitate is di- ary area increases. Furthermore, finer austenite
in Fig. 1). The success of this technique depends rectly controlled by its solubility product, which grains reduce the chances of forming upper bai-
not only on achieving a fine austenite grain size is a function of temperature and alloying con- nite during transformation because the austenite
by repeated recrystallization but also on main- centrations. Elements such as niobium, titanium, grain volume can be completely consumed by
taining a fine grain by inhibiting grain-coarsen- and vanadium have been shown to have solubil- the ferrite and pearlite reactions. Although both
ing mechanisms. These different TMP tech- ity products as to make them particularly useful of the issues were qualitatively recognized in the
niques of controlled rolling are discussed in for TMP conditioning of austenite. However, 1960s (for example, ferrite grain size was re-
more detail after a brief review of the basic met- further discussion of solubility products of these duced by lowering the finish rolling temperature
allurgy of austenite conditioning. The principal microalloying elements is a subject of physical of austenite), it was not until the mid-1970s that
goals are to obtain large amounts of fine ferrite metallurgy that is beyond the scope of this chap- effects such as these could be quantitatively as-
grains and small amounts of lower temperature ter. The intent is to describe only briefly the im- sessed.
transformation products that contain cementite portance of microalloying in the TMP condi- The understanding of the hot deformation of
(such as pearlite or upper bainite), which de- tioning of austenite. austenite in general and its influence on final
grade toughness. Comparison of the various precipitation sys- ferrite grain size was improved dramatically
Microalloying. Thermomechanical process- tems is done through the use of the appropriate through research (Ref 26) in the mid-1970s. A
ing of ferritic steel product involves control of solubility relations as a function of temperature. summary of this work is presented in Ref 6. The
austenite microstructure, but it is important to Titanium precipitates are very effective for high- key to understanding the process by which
note that improvements in the strength and temperature control (for example, during reheat- austenite links alloying and hot rolling with the
toughness depend on the synergistic effect of ing), while niobium precipitates are very effec- final ferrite grain size was based on quantita-
microalloy additions and thermomechanical tive for intermediate temperatures (for example, tively assessing the number of pre-existing or
conditions. This synergism, recognized from control of recrystallization stop temperature). strain-induced heterogeneities introduced into
early research on controlled rolling, is evidenced Vanadium, on the other hand, displays supersat- the austenite during rolling that could act as sites
by the following observations: uration at the very lowest range of hot-rolling for ferrite nucleation (Ref 2). Earlier research
temperatures and can be retained easily in solid (Ref 7) using quantitative metallography led to
It is very difficult to improve the mechani- solution in the austenite for the eventual precip- the definition of a parameter referred to as the
cal properties by thermomechanical pro- itation (VC and VN) hardening of the transfor- austenite interfacial area (Sv), which attempts to
cessing without microalloying elements. mation product. account for all the near planar crystalline defects
Steels with microalloying elements, but Austenite Conditioning. The metallurgical that may exist in a polycrystalline aggregate.
without thermomechanical processing, dis- condition of the austenite at the transformation The parameter Sv has units of mm1 or equiva-
play a lower toughness than their plain-car- temperature depends on a complex interplay be- lently area (mm2) per unit volume (mm3).
bon steel counterparts do. tween the stored energy of deformation and the Researchers (Ref 2) applied the parameter Sv,
restoration processes of recovery and recrystal- which they defined as the total effective interfa-
The importance of microalloying elements on lization. These restoration processes drive the cial area per unit volume, to the problem of
austenite during thermomechanical processing metallurgical state back to a condition that is en- quantitatively assessing the number of sites in
can be understood by the three critical tempera- ergetically comparable to or lower than the con- thermomechanically processed austenite for po-
tures of austenite metallurgy: dition existing before deformation. However, tential nucleation of ferrite and developed the
various thermal and metallurgical factors may correlation shown in Fig. 2.
Grain-coarsening temperature, TGC also retard or limit the restoration processes that Attempts to quantify the influence of Sv on
Recrystallization stop temperature, TRXN, are driven, in part, by the stored energy of de- final ferrite grain size have shown that the ferrite
below which recrystallization of austenite formation. Therefore, the austenite conditioning grain size decreases continuously with increases
ceases depends on a complex interplay of thermome- in Sv (Ref 2, 8, 9). There is still some disagree-
Transformation temperatures, such as the chanical and metallurgical factors that drive or ment over the exact relationship, but the general
critical temperature when austenite begins retard the restortion processes. This interaction form has been well defined (Fig. 3). Hence, from
to transform to ferrite upon cooling (Ar3), or is described in a little more detail in the section the standpoint of ferrite grain refinement, a well-
the bainite-start temperature (Bs) above Restoration Processes. conditioned austenite exhibits a large Sv. For ex-
which austenite will not transform to bainite In general, as-rolled austenite can show three ample, if the rolling takes place above the re-
upon cooling types of microstructures: crystallization temperature of the austenite, then
the deformed austenite grains are replaced with
Fully recrystallized
Critical transformation temperatures are an a different set of equiaxed, recrystallized grains
Partially recrystallized
important part of microalloying, but they are not that differ from the original chiefly in size. For
Fully unrecrystallized
discussed in this chapter because they are related equiaxed grains:
more to hardenability concepts and heat treat- While the grain size of fully recrystallized
ment. For more details, see Ref 1. austenite is a simple matter to comprehend and 3
Solubility, which is the extent that an element determine, the determination of a fully unrecrys- Sv =
Dg (Eq 1)
can be maintained in solid solution, is a key as- tallized microstructure is not. In order to specify
pect of how microalloying elements behave in the grain size of fully recrystallized austenite,
austenite during TMP. Microalloying elements only an average grain diameter is required; this where Dg is the diameter of the austenite grains.
during the deformation of austenite may remain fully describes the grain boundary area per unit A reduction in grain size by hot rolling thus
in solid solution, or they may chemically bond volume. In the case of fully unrecrystallized leads to an increase in grain boundary area and,
with another alloying element to form a precipi- austenite, both the starting grain size and the as- hence, an increase in Sv.
Grain Coarsening
Grain coarsening is driven by the minimization
of grain boundary energy (or area) per unit vol-
ume. Like surface energy, grain-boundary en-
ergy tends to minimize itself when possible by
decreasing the total grain-boundary area of a
system. Thus, one grain grows at the expense of
another to reduce the overall grain-boundary
area per unit volume. Grain boundaries that have
sharp corners or bends tend to smooth and
straighten, and small grains (with high surface-
to-volume ratios) become absorbed by larger
grains.
Grain growth can be complicated (or slowed)
when the movement of grain boundaries is
restricted by metallurgical conditions such
as:
Solute atoms that cause grains to grow more
Fig. 4 Effect of rolling reduction on Sv for cube-shaped
slowly
austenite grains. Source: Ref 10
Second-phase particles or precipitates that
pin or restrict the movement of grain
Austenite interfacial area per unit volume (Sv), mm1
Austenite interfacial area per unit volume (Sv), in.1

Fig. 2 Variation of effective austenite interfacial area boundaries
(Sv) with initial grain size and amount of reduc-
tion below the recrystallization stop temperature for a These two mechanisms, which are key fac-
0.03 wt% Nb steel. Source: Ref 2
tors in the restriction of grain growth, are de-
scribed briefly here. They are not, however, the
Austenite interfacial area per unit volume (Sv), in.1
exclusive mechanisms. For example, another
reason for slow grain growth may be the pres-
ence of a strongly preferred crystal orientation
or texture. With respect to other restoration
processes, grain growth tends to occur much
more slowly than recrystallization, but this de-
pends on the temperature and the amount of
stored energy that may drive recrystallization
processes. In particular, inhibition of grain
growth is important in RCR methods, where
finish rolling is done at relatively higher tem-
Fig. 5 Increase in Sv because of deformation bands. peratures (i.e., above TRXN).
Austenite grain diameter, Dg = 100 mm (0.004 Solute Drag. Solutes can cause grains to
in.); SvGB grain boundary planar-linear model; SvDB de-
formation band. SvTotal = SvGB + SvDB. Source: Ref 10
grow more slowly through the stabilization of
grain boundaries. Solute atoms lower the grain-
Austenite interfacial area per unit volume (Sv), mm1
This work has been summarized (Ref 10) and is boundary energy and therefore prevent the
Fig. 3 Ferrite grain sizes produced from recrystallized illustrated in Fig. 4 and 5. The variation of Sv grain boundary from moving as rapidly as it
and unrecrystallized austenite at various Sv val- was calculated for a hypothetical array of cube- would in the absence of a solute atom. This is
ues. Source: Ref 10 shaped grains of variable initial size that were referred to as solute drag. The effect of solutes
subjected to various plane strain deformations, in retarding grain growth depends on the solute
If the hot rolling takes place below the recrys- all to simulate the behavior of grains during con- involved; elements that distort the lattice the
tallization temperature, the austenite grains be- trolled rolling (Ref 10). most impede grain growth the most. The effec-
come highly elongated and, at a sufficiently tiveness of solute drag has been shown to be
large strain, become filled with intragranular de- related to differences in size and valence be-
fects such as deformation bands and twins. In Restoration Processes (Ref 1) tween the solute and solvent atoms (Ref 13, 14,
this case, the final Sv comprises two terms, one 15) and has been discussed extensively in the
describing the grain boundary surface area per The metallurgical condition of the austenite at literature (Ref 16, 17, 18, 19, 20). However,
unit volume and the other describing the surface the transformation temperature is dependent on unlike the effects of particle pinning, quantita-
area of the deformation bands and twins con- a complex interplay between the stored energy tive treatment of solute-drag effects are not
tained within the elongated grains such that: of deformation, restoration processes, and the established.
forces (or factors) that retard restoration. The Grain-Boundary Pinning by Particles. The
Sv = Sv(GB) + Sv(IPD) (Eq 2) following discussions briefly describe the forces effects of particle pinning on grain growth have
(or factors) that drive and retard the restoration been expressed in more quantitative terms. The
where Sv(GB) is the total effective area per unit processes of grain growth, recovery, and recrys- first attempt to treat the retarding effects of sec-
volume from the grain boundary contribution, tallization. In general, restoration processes in- ond-phase particles on a migrating grain bound-
and Sv(IPD) is the contribution from intragranu- volve the movement of grain boundaries so that ary was made by Zener (Ref 21). He indicated
lar planar defects. The important parameters in- stored energy tends to become more dispersed. that when particles were present in the vicinity
clude the extent of deformation below the re- However, the mechanisms of grain-boundary of a grain boundary, the effective energy of the
crystallization temperature and the grain size of movement also can be retarded by solutes and grain boundary would be lowered. The reduction
the austenite prior to pancaking (Ref 10, 11, 12). precipitates that act on the grain boundaries. in grain-boundary energy occurs because the
surface area of second-phase particles would re- undissolved precipitates can no longer suppress (Ref 28). Reheating controls not only the grain
place part of the grain boundary. Therefore, the grain growth. The influence of various microal- size, but also the initial composition of the
movement of a grain boundary away from parti- loying elements on grain growth during reheat- austenite (Ref 25, 28). Niobium is not fully
cles would require work as the effective grain- ing is shown in Fig. 6 (Ref 10). While it does not taken into solution until a steel is heated to ap-
boundary area is increased (Ref 22). Zeners appear that the final average austenite grain size proximately 100 C (180 F) above its grain-
original theory was later quantified with the as- is very dependent on the reheated grain size coarsening temperature (Ref 25).
sumption of the rigid motion of grain boundaries prior to multipass deformation (Ref 25), it does Similar results have been found in a separate
through a regular array of spherical particles appear likely that the distribution of grain sizes work (Ref 29). The factors that control the com-
(Ref 23). It was shown that the pinning force for about the average is much smaller when the re- position of the austenite at various stages of pro-
each particle (F) and the particle radius (r) were heating temperature is kept below the grain- cessing have been reviewed (Ref 30).
related by: coarsening temperature (Ref 26).
During reheating, a slab is slowly heated from
F = 4rsI (Eq 3) room temperature to the reheating temperature, Recovery and Recrystallization
where it is soaked for an appropriate period of
Recovery and recrystallization are driven by
where sI is the interfacial energy per unit area of time and after which it is available for hot rolling.
During this heating and soaking period, events the stored energy of deformation. The concen-
boundary. An expression was derived subse-
that ultimately control the grain size and compo- trations of carbon, nitrogen, microalloying ele-
quently for the critical particle size (rc) below
sition of austenite prior to hot rolling occur. The ments (niobium, titanium, vanadium), the degree
which grain boundaries are pinned (Ref 23):
ultimate response of a given steel to subsequent of strain, the time between passes, the strain rate,
-1 processing will, to a large extent, be strongly in- and the temperature of deformation all influence
6 R0 f v 3 2 recrystallization during hot working.
rc = - fluenced by the reheating stage. The major events
p 2 z (Eq 4) The driving force for recovery and recrystal-
that may occur during reheating include:
lization has been shown to be equal to the dif-
The relationship shown by Eq 4 was obtained Nucleation, growth, and possible coarsen- ference in dislocation density that is present be-
by equating the rates of the grain-boundary ening of austenite grains tween strained and strain-free local volumes.
ergy increase during grain growth and the parti- Elimination of inhomogeneities in solute This driving force has been estimated and, based
cle pinning force. In this expression, R0 is the distribution upon the area under the flow curve (Ref 31), is
mean radius of the matrix grain, and fv is the Dissolution of microalloy precipitates that expressed by:
volume fraction of second-phase particles. The were inherited from the original
term z represents the ratio of the radii of grow- microstructure Ds = 0.2mb Dr (Eq 6)
ing and matrix grains. It was deduced (Ref 24)
that z can range in value between 2 to 2 Austenite first forms at ferrite-carbide (ce- where Ds is the increment in flow stress due to
throughout the grain-growth process. The im- mentite) interfaces (Ref 27), and the increase in work hardening, m is the shear modulus, b is the
portant point to be realized from Eq 4 is that a the amount of austenite is accompanied by a de- Burgers vector, and Dr is the excess dislocation
high volume fraction of small particles inhibits crease in the amount of cementite, at least in the density.
boundary migration. early stages of austenite formation (Ref 27). Recrystallization Stop Temperature
Equation 3 can be expanded to account for the Also, although the kinetics of austenite forma- (TRXN). Perhaps the most important effect of mi-
total pinning force (FPIN) that a number of partition are very rapid at normal reheating tempera- croalloying elements is in controlling the recrys-
cles per unit area (Ns) exert on a migrating tures, the elimination of heterogeneities in solute tallization stop temperature of austenite (Ref
boundary. This total pinning force is expressed distribution is very sluggish (Ref 27). 31). Austenite recrystallization ceases at temper-
as: Once the austenite grains have formed from atures below TRXN, and the influence of several
the ferrite-pearlite starting structure, they will microalloying elements on the recrystallization
FPIN = 4r sI Ns (Eq 5) coarsen with time and temperature. This coars- stop temperature is shown in Fig. 7. As shown in
ening behavior has been reviewed in some detail Fig. 7, niobium has the most profound effect on
Grain Coarsening Temperature (TGC). The
grain coarsening temperature (TGC) is defined to
be that temperature above which grain coarsen-
ing by secondary recrystallization commences
and refers to the temperature above which the
Fig. 6 Austenite grain-coarsening characteristics in

steels containing various microalloying addi-
tions. Source: Ref 10 Fig. 7 Effect of microalloying additions on the recrystallization stop temperature of austenite. Source: Ref 31
TRXN, and this accounts for its selection as the is indicative of the weak retarding effect of 3 fv
primary microalloying element in modern high- solute drag as a retarding force. However, when N sR =
2 pr 2 (Eq 9)
strength, low-alloy (HSLA) steels. Some early precipitation accompanies the early stages of de-
steels used vanadium in conjunction with nitro- formation as it does in the vanadium plus nitro- The rigid boundary model, however, was rather
gen additions to produce a moderate effect in regen steel at low temperatures and the two nio- simple and unrealistic since it did not account
tarding austenite recrystallization, but such prac- bium steels at higher temperatures, there is a for any flexibility of the austenite grain bound-
tices are not compatible with current ideas for drastic retardation of dynamic recrystallization, aries (Ref 22).
limiting free nitrogen concentrations, nor were and the curves display a plateau. Notice that A model at the opposite extreme of the rigid
they particularly effective. Consequently, the when precipitation occurs, the curves are virtu- boundary model was proposed (Ref 31). This
predominant use of vanadium is for precipitating ally identical. They are only displaced on the model was termed the flexible boundary model.
vanadium carbide in the low-temperature trans- temperature scale. It assumed that an infinitely flexible boundary is
formation products. Similarly, titanium is not Models of Recrystallization Suppression. capable of interacting with every particle in the
considered useful for retarding austenite recrys- Three models have been postulated to explain three-dimensional array until it is fully pinned
tallization but is highly effective in preventing how microalloy precipitates suppress austenite (Ref 31). This led to the relationship:
grain growth (Fig. 6). The grain-refining bene- recrystallization (Ref 16, 22, 34). All of these
fits of titanium have another primary use in models are based on the general form of Eq 5.
welding that is unrelated to the benefits during Their differences arise from the method by rN v
N sF =
austenite processing, and this has led to wide- which Ns is calculated. The first of these models ( f v )1/ 3 (Eq 10)
spread use of approximately 0.015 wt% Ti in was an extension of earlier research (Ref 23).
plate and pipeline steels. This approach is not This model is often referred to as the rigid which, when combined with Eq 8, yields the
readily accessible when nitrogen concentrations boundary model because it assumes the motion number of particles per unit area assuming a
are above 0.008 wt% N or when the ingot cast- of a rigid grain boundary, which is capable of in- flexible boundary model, NsF:
ing route is involved. teracting with those particles lying within r of
The data shown in Fig. 6 and 7 reveal that the the boundary plane. Under this set of conditions,
various microalloying elements all show the NsR is defined by: 3( f v ) 2 / 3
N sF =
same general behavior. The differences between 4 pr 2 (Eq 11)
them are predominantly the temperature range NsR = 2r Nv (Eq 7)
over which they exert their influence and the The most realistic of these models was the
level of intensity of that influence. This differ- one proposed in Ref 34 and 35. This model con-
ence in behavior is due chiefly to differences in where Nv is the number of particles per unit vol- sidered the effect of a precipitate distribution
the solubility of the various precipitates in ume. Assuming spherical particles having radius that could exist on austenite subgrain boundaries
austenite. The difference in the temperature r and volume fraction fv, Nv is defined by the fol- prior to the start of recrystallization. Assuming
range of precipitation in austenite is shown lowing expression: that the average subgrain intercept is l, the sur-
clearly in Fig. 8 for vanadium and niobium car- face area per unit volume for such subgrain
bonitrides (Ref 32, 33). In Fig. 8, the deforma- 3 fv boundaries would be 2/l, and the number of par-
tion temperature is plotted against the deforma- Nv = ticles per unit subgrain boundary area would be
4 pr 3 (Eq 8)
tion time required to initiate dynamic given by (Ref 34):
recrystallization (i.e., time-to-peak stress, tp),
which is one form of austenite restoration. When Combining Eq 7 and 8 yields the number of par-
lN v
no strain-induced precipitation forms, the curves ticles per unit area assuming a rigid boundary N sS =
are only slightly offset from the base steel. This model, NsR: 2 (Eq 12)
As before, combining Eq 8 and 12 gives the

number of particles per unit area assuming a
subgrain boundary model, NsS:
3lf v
N sS =
8pr 3 (Eq 13)
Substitution of the expressions for NsR, NsF, and

NsS (Eq 9, 11, 13) into Eq 5 gives the respective
pinning force for each model.
Conventional Controlled Rolling
Conventional controlled rolling was the first

type of controlled rolling to come into regular
commercial use. About 8 to 10% of the total
steel tonnage rolled annually in the 1990s was
produced in this way. This process was origi-
nally developed for the production of plate
grades for the manufacture of oil and gas
Dependence of the time-to-peak stress (tp) on deformation temperature at a low strain rate (e is 0.006 s1) for
pipelines with required minimum yield
Fig. 8 strengths ranging from 350 to 490 MPa (50 to
selected microalloyed steels containing two levels of nitrogen (0.006 and 0.025% N). The 0.025% N steel is
a reference steel. Source: Ref 32 70 ksi). Because of the need for good weld-
ability, low concentrations of carbon and carbon Dynamic recrystallization that might occur tion stop temperature. Titanium and zirconium
equivalents were specified. These were obtained during deformation in the roll gap nitride formed during solidification and upon
readily by reducing carbon concentrations to Metadynamic recrystallization that, if it oc- cooling of the slab do not readily dissolve upon
0.06 or 0.07%; small amounts of niobium (about curred, would begin in the roll gap and go to reheating to hot-rolling temperatures. Although
0.04%), in combination with vanadium (up to completion in the time period between de- these nitrides may prevent grain coarsening
0.1%) and molybdenum (up to 0.30%), were formations upon reheating, they are not effective in prevent-
added for higher-strength grades. ing recrystallization because insufficient tita-
As noted previously, the goal of conventional nium or zirconium remains in solution at the
Static recrystallization is the more dominant
controlled rolling is to produce very fine ferrite rolling temperature to precipitate on deformed
form of recrystallization during industrial pro-
grain sizes by conditioning the austenite through austenite boundaries during hot rolling and thus
cessing because the time required for substantial
extensive rolling in the non-recrystallization re- suppress austenite recrystallization. Vanadium,
amounts of dynamic recrystallization is in orders
gion of austenite. During roughing operations, on the other hand, is so soluble that precipitation
of magnitude longer than the times actually
the coarse reheated austenite grains in a slab are does not occur readily in the austenite at normal
available in the roll gap (Ref 3841). Suppres-
first refined by repeated recrystallization, bring- hot-rolling temperatures.
sion of static recrystallization thus is a critical
ing the grain sizes down to about 20 mm (0.8 Effect of Niobium Precipitation on Ar3.
factor, and its arrest is caused by the copious
mil) or less. The transfer bar can then cool below When microalloyed austenite undergoes multi-
precipitation of Nb(C,N) during delays between
the TRXN during transfer from roughing to the pass hot rolling at temperatures where large su-
mill passes. Because the precipitation of
finishing facilities. When rolling is restarted or persaturations exist, various amounts of strain-
Nb(CN) in the austenite during hot rolling re-
continued below the TRXN, recrystallization is induced precipitation may occur depending on
tards recrystallization and raises the recrystal-
no longer possible, and the austenite structure is factors such as the finishing temperature and
lization-stop temperature, a broader temperature
progressively flattened in an operation known as the interpass holding time. The effect on the
range is possible for hot working the steel to
pancaking. For pancaking to be successful, the amount of precipitation caused by variations in
produce highly deformed austenite.
accumulated reductions applied in this tempera- deformation temperature, extent of recrystal-
Niobium is the most effective alloying ele-
ture range must add up to at least 80%. Finally, lization, and interpass holding time is shown in
ment for grain refinement by conventional con-
when the flattened austenite grains go through Fig. 9. Figure 9 shows that the amount of
trolled rolling. The optimum amount of niobium
their transformation to ferrite, the ferrite pro- strain-induced precipitation, which attends hot
to suppress recrystallization between passes can
duced has a very fine grain structure because of rolling, is a variable, and hence, the amount of
be as little as 0.02%. During the rolling reduc-
the large number of nucleation sites available on niobium in solution can also vary. Because it is
tions at temperatures below 1040 C (1900 F),
the expanded surfaces of the pancaked austenite known that the Ar3 temperature varies with the
the niobium in solution suppresses recrystalliza-
grains. This leads to ferrite grain sizes in the amount of niobium in solution, it should be ex-
tion by solute drag or by strain-induced Nb(C,N)
range of 5 to 10 mm (ASTM grain size numbers pected that the Ar3 temperature may also vary
precipitation on the deformed austenite slip
10 to 12). The fine-grain ferrite is responsible with the details of the rolling practice. This is
planes. The strain-induced precipitates are too
for the attractive combination of good toughness shown in Fig. 10, which plots the dependence
large to affect precipitation strengthening but are
and strength. of Ar3 versus the amount of deformation that
beneficial for two reasons: they allow additional
Suppression of Recrystallization. Austenite takes place below the recrystallization stop
suppression of recrystallization by preventing
pancaking is only possible in the absence of re- temperature (Ref 43).
migration of austenite grain subboundaries, and
crystallization. The three different forms of re- Intensified Controlled Rolling. A variation
they provide a large number of nuclei in the de-
crystallization that might take place during mul- in conventional controlled rolling practice,
formed austenite for the formation of fine ferrite
tipass hot rolling are (Ref 36, 37): known as intensified controlled rolling, is shown
particles during cooling.
in Fig. 1. Note that intensified controlled rolling
Static recrystallization that might occur in Titanium, zirconium, and vanadium are not as
differs from conventional controlled rolling in
the time period between deformations effective as niobium in raising the recrystalliza-
that finish rolling not only occurs below the
TRXN but also extends to temperatures below the
Ar3. The goal of intensified controlled rolling is
to increase the strength and toughness of the as-
rolled microalloyed steel over what could be
achieved through conventional controlled
rolling. The combination of the lower reheat and
Fig. 9 Isothermal recrystallization and precipitation in a niobium steel after 50% deformation. The steel composition Fig. 10 Increase in Ar3 temperature as a result of con-
was 0.10C-0.99Mn-0.04Nb-0.008N. Source: Ref 42 trolled rolling. Source: Ref 43
lower rolling temperatures leads to finer as- local volume fractions of fine particles are effec- 6 mm (0.12 to 0.24 mil), as opposed to 5 to 8
rolled austenite. Furthermore, because a portion tive. In this regard, particles such as titanium ni- mm (0.2 to 0.3 mil) for the latter process (Ref
of the rolling occurs in the intercritical or two- tride have been shown to be particularly well 45). However, such low-temperature finishing
phase (a+g) region, a certain amount of proeu- suited for the suppression of grain coarsening. increases the rolling load, and it can also make
tectoid ferrite would be deformed during this These particles must have sufficient stability to mill control more difficult because of the load
rolling. This deformed ferrite is partially respon- resist Ostwald ripening (solution-crystallizer drop associated with the initiation of dynamic
sible for the higher strengths observed with in- phenomenon in which small crystals, more solu- recrystallization. It is important to note that
tensified controlled rolling. ble than large ones, dissolve into larger parti- under industrial rolling conditions, CCR, RCR,
cles). and DRCR can all occur to different degrees
An alternate grain-coarsening inhibition during a given operation. This can happen when
Recrystallization Controlled Rolling mechanism involves intense solute drag as the the processing parameters have not been opti-
critical element. It has been found that high lev- mized so as to favor only strain-induced precip-
Recrystallization controlled rolling is an at- els of soluble molybdenum and niobium can be itation and austenite pancaking in the case of
tractive alternative to conventional controlled very effective in retarding boundary motion at CCR, conventional recrystallization in the case
rolling when low-temperature working is im- high reheating temperatures (Fig. 11). Hence, of RCR, and dynamic recrystallization in the
practical for certain products (hot forgings or these solute effects act to retard grain coarsening case of DRCR.
thick plate) or with the equipment limitation of both during reheating and after post-deformation
an underpowered mill. Conventional controlled static recrystallization.
rolling is generally based on the use of low fin- Dynamic Recrystallization Controlled Thermomechanical Processing
ishing temperatures (that is, in the vicinity of Rolling (DRCR) (Ref 44). When the interpass of Microalloyed Bar (Ref 47)
800 to 900 C, or 1470 to 1650 F). However, time is short, as in the case of rod, hot strip, and
such finishing at such low temperatures is inap- certain other rolling processes, insufficient time Plate-rolling operations lend themselves to
propriate for certain products, such as heavy is available for conventional recrystallization considerable control of the thermomechanical
plates and thick-walled seamless tubes, due to during the interpass delay. The amount of car- treatment. The slab reheat temperature can be re-
excessive rolling loads. For such applications, it bonitride precipitation that can take place is also duced if desired. In fact, some rolling strategies
is desirable to condition austenite at a higher severely limited. As a result, an alternative form involve only reheating to 960 C (1760 F) prior
rolling temperature in the crystallization regime. of recrystallization, known as dynamic recrys- to rolling. Delays can be built into the rolling op-
By carefully controlling the recrystallization of tallization, is initiated. Dynamic recrystalliza- eration (although with some penalty in produc-
austenite and arranging for it to occur at succes- tion involves the nucleation and growth of new tivity), and a considerable range of finishing
sively lower temperatures during finish rolling, grains during (as opposed to after) deformation temperatures can be achieved. The operation can
recrystallization controlled rolling achieves a (Ref 45, 46). This also requires the accumulation accommodate the most severe controlled-rolling
fine, equiaxed austenite. of appreciable reductions, of the order of 100%, schedules, including the deformation of austen-
The success of recrystallization controlled to enable the recrystallization process to spread ite-ferrite mixtures.
rolling depends on two key factors. First, the re- completely through the microstructure inherited In contrast to the plate-rolling process, the
crystallization process should not be sluggish or from the roughing process. Austenite grain sizes possible thermomechanical treatment on a mod-
suppressed so that the times required for recrys- as small as 10 mm (0.4 mil) can be achieved ern bar mill (Fig. 12) is somewhat limited in
tallization are not too long. The second factor is with DRCR (Ref 23). scope. For example, the temperature-time pro-
the inhibition of grain growth. Rapid grain Low-temperature finishing by DRCR has the file for the rolling of 44 mm (1 in.) diam bar
coarsening can occur during the time between advantage of producing finer ferrite grain sizes shown in Fig. 13 can be compared with the
rolling passes, and so steels processed at the after transformation than CCR does; that is, 3 to plate-rolling profile shown in Fig. 14. There are
higher temperatures of the RCR method require
microalloying to inhibit grain coarsening. For
this purpose, sufficient titanium is added to have
about 0.01% available for the formation of fine
particles of TiN during cooling after continuous
casting. When this dispersion has an appropriate
size and frequency distribution, it can almost
completely prevent grain growth of the austenite
after each cycle of recrystallization.
In general, methods of grain-growth inhibi-
tion can take one of two forms. The most com-
mon method relies on the pinning force (Eq 5)
from stable particles. As noted previously, the
driving force for normal grain coarsening is typ-
ically smaller than for recrystallization, depend-
ing on the temperature and deformation condi-
tions. The driving force for grain growth can be
three orders of magnitude smaller than that for
recrystallization (Ref 22). Because grain growth
is slower, the pinning force needed for inhibition
of grain growth is less than that needed for sup-
pression of recrystallization. The particle pin-
ning force also varies with fv /r, and so particle
distributions with rather small fv and large r can
still be effective barriers to grain coarsening.
This, of course, would not be true in the sup- Fig. 11 Effect of molybdenum addition on austenite grain size after reheating at 1150 C (2100 F) and holding at
pression of recrystallization, where only large, this temperature for various times. Courtesy of C.I. Garcia and A.J. DeArdo, University of Pittsburgh, 1991
Even though the lower reheat temperatures

typical of bar products place some limita-
tions on the use of different microalloying
elements, these lower temperatures do pro-
vide for a finer as-reheated austenitic grain
size than is typical of slabs reheated for con-
version to plate. With a small titanium addi-
tion and continuous casting, as-reheated
austenitic grain sizes of 50 to 60 mm (2 to
2.4 mils) can be achieved in billets destined
Fig. 12 Controlled-rolling process for microalloyed steel bar for bar.
Finishing temperatures in bar rolling are rel-
atively high, even with the use of interstand
cooling.
Consequently, recrystallization controlled

rolling becomes quite important in bar rolling,
and the rolling strategy must be designed to pro-
duce the finest possible recrystallized austenitic
grain size. Subsequent control of the austenite-
to-ferrite transformation range is still important
to maximize ferritic grain refinement. Neverthe-
less, the ferritic grain size that can be produced
on transformation from a recrystallized austenite
is limited compared with the grain size that can
be produced on transformation from austenitic
grains that have been flattened by rolling below
the recrystallization temperature. Thus, while
moderate grain refinement can be achieved in an
as-rolled microalloyed bar, this grain size will be
Fig. 13 Temperature-time profile for the controlled rolling of 44 mm (13/4 in.) diam microalloyed steel bar. Compare
somewhat coarser than the grain size of con-
with Fig. 14.
trolled-rolled HSLA plates.
Because the degree of ferritic grain refine-
ment possible in as-rolled microalloyed bar
steels is somewhat limited, and because sub-
structural strengthening is not possible, alterna-
tive strengthening mechanisms are employed to
reach yield strength levels comparable to those
of plate grades. For example, in the alloy design
of microalloyed bar steels, precipitation and
pearlite strengthening must be relied on to a
greater extent than in the design of plates. In
view of the limited solubility of niobium or tita-
nium at the reheat temperatures used in bar pro-
cessing, vanadium is usually used to obtain the
required level of precipitation strengthening in
HSLA bar grades. Precipitation of V(C,N) dur-
ing or after transformation can provide signifi-
cant strengthening increments. In this regard, ni-
trogen level is also of importance. Judicious
selection of both the vanadium and nitrogen lev-
els is required to produce the desired level of
precipitation strengthening. Similarly, an in-
crease in the carbon content and thus the pearlite
volume fraction of a bar steel can also be used to
increase strength. Of course, these two strength-
Fig. 14 Temperature-time profile for controlled rolling of 19 mm (3/4 in.) thick microalloyed steel plate. TR, recrys-
ening mechanisms have very deleterious effects
tallization temperature
on toughness. Toughness levels currently avail-
able in commercial microalloyed bar steels may
clear differences between these rolling amount of niobium that can be dissolved upon be adequate for many applications, but there is
processes: reheating. For example, in a 0.20% C steel, considerable interest in improving the toughness
only about 0.01% Nb is soluble at 1100 C of microalloyed bar grades.
Lower reheat temperatures, typically in the (2010 F). In contrast, vanadium is still read- Dynamic Recrystallization in Bar Rolling.
range of 1100 to 1200 C (2010 to 2190 F), ily soluble at bar reheat temperatures. Con- Similar to the controlled rolling of plate, dy-
are used in bar rolling. This lower tempera- sequently, in HSLA bar grades, vanadium is namic recrystallization can be a factor in the
ture, in combination with the generally higher the microalloying element commonly used to rolling of bar, rod, or tube. At the finishing end
carbon levels in bar products, limits the obtain the highest possible strength levels. of hot-rolling processes such as rod and bar
rolling, the stock moves quickly, and the inter- al., Ed., Union Carbide Corporation, 1977, 27. C.I. Garcia and A.J. DeArdo, Metall. Trans.
pass intervals are of the order of 0.010 to 0.1 s. p 120 A, Vol 12, 1981, p 521
Under these conditions, there may not be 3. T. Tanaka, N. Tabata, T. Hatomura, and C. 28. A.J. DeArdo, J.M. Gray, and L. Meyer,
enough time for significant amounts of conven- Shiga, Microalloying 75, M. Korchynsky et Niobium, H. Stuart, Ed., The Metallurgical
tional static recrystallization to take place, paral., Ed., Union Carbide Corporation, 1977, Society of AIME, 1984, p 685
ticularly when low finishing temperatures were p 107 29. E.J. Palmiere, Ph.D. thesis, University of
employed. The strain, therefore, accumulates 4. H. Sekine and T. Maruyama, Trans. Iron Pittsburgh, 1991
from pass to pass until the retained strain Steel Inst. Jpn., Vol 16, 1976, p 427 30. J.M. Gray and A.J. DeArdo, HSLA Steels:
reaches the critical level required for the initia- 5. W.B. Morrison and J.A. Chapman, Philos. Metallurgy and Applications, J.M. Gray et
tion of dynamic recrystallization (DRX), fol- Trans. R. Soc. (London) A, Vol A282, 1976, al., Ed., American Society for Metals, 1986,
lowed immediately upon unloading by post- p 289 p 83
dynamic static recrystallization (PDSR). This 6. T. Tanaka, International Metallurgical 31. L.J. Cuddy, Thermomechanical Processing
has important consequences for the rolling loads Reviews, 1981, p 185 of Microalloyed Austenite, A.J. DeArdo,
and grain size evolution. The influences of the 7. E.E. Underwood, Quantitative Metallogra- G.A. Ratz, and P.J. Wray, Ed., The
deformation parameters and chemical composi- phy, McGraw-Hill, 1968, p 77 Metallurgical Society of AIME, 1982, p 129
tion on PDSR are described in Ref 48. 8. W. Roberts, H. Lindefelt, and A. Sandberg, 32. I. Weiss et al., Thermomechanical Process-
In another analysis (Ref 49), data from pass- Hot Working and Forming Processes, The ing of Microalloyed Austenite, A.J. DeArdo,
by-pass evolution of austenite grain size during Metals Society, London, 1980, p 38 G.A. Ratz, and P.J. Wray, Ed., The
the rod rolling of plain carbon steel was organ- 9. L.J. Cuddy, Metall. Trans. A, Vol 15, 1984, Metallurgical Society of AIME, 1982, p 33
ized into a commercial computer spreadsheet p 87 33. K. Tiitto, G. Fitzsimons, and A.J. DeArdo,
program with previously developed equations 10. G.R. Speich, L.J. Cuddy, C.R. Gordon, and Acta Metall., Vol 31, 1983, p 1159
relating grain size and hot-working parameters. A.J. DeArdo, Phase Transformations in 34. S.S. Hansen, J.B. Vander Sande, and M.
By considering the substanial redundant Ferrous Alloys, A. Marder and J. Goldstein, Cohen, Metall. Trans. A, Vol 11, 1980, p 387
strains developed during rod rolling, the analysis Ed., The Metallurgical Society of AIME, 35. J.G. Speer and S.S. Hansen, Metall. Trans.
revealed that metadynamic recrystallization 1984, p 341 A, Vol 20, 1989, p 25
(MRX) is the dominant microstructural process. 11. Y. Zheng, G. Fitzsimons, and A.J. DeArdo, 36. J.J. Jonas, C.M. Sellars, and W.J. McG.
The spreadsheet was also employed to examine HSLA Steels: Technology and Applications, Tegart, Met. Rev., Vol 14, 1969, p 1
the effects of modifications to rod rolling prac- American Society for Metals, 1984, p 85 37. H.J. McQueen and J.J. Jonas, Plastic
tice aimed at refining the austenite grain size. 12. M.R. Blicharski, C.I. Garcia, S. Pytel, and Deformation of Materials, R.J. Arsenault,
Little benefit is obtained by increasing the strain A.J. DeArdo, Processing, Microstructure Ed., Academic Press, 1975, p 393
rate or by reducing the distance from the final and Properties of HSLA Steels, A.J. 38. J.D. LEcuyer and G. LEsperance, Acta
rolling pass to the laying head. On the other DeArdo, Ed., The Metallurgical Society of Metall., Vol 37, 1989, p 1023
hand, increasing the cooling rate on the forced- AIME, 1988, p 317 39. J.J. Jonas, International Conference on
air cooling deck should lead to measurable grain 13. R.A. Oriani, Acta Metall., Vol 7, 1959, p Physical Metallurgy of Thermomechanical
refinement. Lowering the temperature during 62 Processing of Steel and Other Metals, I.
rolling can lead to even more refinement, and 14. J.L. Walter and C.G. Dunn, Trans. AIME, Tamura, Ed., The Iron and Steel Institute of
various strategies to achieve this refinement are Vol 215, 1959, p 465 Japan, 1988, p 59
discussed in Ref 49. 15. E.L. Holmes and W.C. Winegard, Acta 40. O. Kwon and A.J. DeArdo, Interactions
Metall., Vol 7, 1959, p 411 Between Recrystallization and Precipitation
16. R.A.P. Djaic and J.J. Jonas, Metall. Trans. in Hot-Deformed Microalloyed Steels, Acta
Summary and Acknowledgment A, Vol 4, 1973, p 621 Metall. Mater., Vol 39 (No. 4), April 1991,
17. R.A. Petkovic, M.J. Luton, and J.J. Jonas, p 529538
Thermomechanical processing of steel by Can. Metall. Q., Vol 14, 1975, p 137 41. A.J. DeArdo, Mathematical Modelling of
controlled rolling is an ongoing activity of re- 18. M.J. Luton, R. Dorvel, and R.A. Petkovic, Hot Rolling of Steel, S. Yue, Ed., CIM,
search and development because property im- Metall. Trans. A, Vol 11, 1980, p 411 1990, p 220
provements in as-rolled steels products can have 19. G.L. Wang and M.G. Akben, Processing, 42. T.M. Hoogendoorn and M.J. Spanraft,
important engineering and economic implica- Microstructure and Properties of HSLA Microalloying 75, M. Korchynsky et al.,
tions. Bibliographic searches of current litera- Steels, A.J. DeArdo, Ed., The Metallurgical Ed., Union Carbide Corporation, 1977, p 75
ture on this topic provide many examples, and Society of AIME, 1988, p 79 43. C. Ouchi, T. Sampei, and L. Kozasu, Trans.
thorough coverage is beyond the scope of this 20. L.J. Cuddy, J.J. Bauwin, and J.C. Raley, Iron Steel Inst. Jpn., Vol 22, 1982, p 214
chapter. By necessity, this chapter only briefly Metall. Trans. A, Vol 11, 1980, p 381 44. R.I.L. Guthrie and J.J. Jonas, Steel
describes some of the basic concepts and meth- 21. C. Zener, Trans. AIME, Vol 175, 1949, p 15 Processing Technology, Properties and
ods of controlled rolling. 22. M.F. Ashby, Recrystallization and Grain Selection: Irons, Steels, and High-
Substantial portions of this chapter were Growth of Multi-Phase and Particle Performance Alloys, Vol 1, ASM Handbook,
adapted from Ref 1 as noted in the section head- Containing Materials: 1st RISO ASM International, 1990, p 117
ings. The editors extend their thanks for the International Symposium on Metallurgy 45. F.H. Samuel, S. Yue, J.J. Jonas, and K.R.
reuse of this material. and Material Science, N. Hansen, A.R. Barnes, Effect of Dynamic Recrystallization
Jones, and T. Leffers, Ed., RISO National of Microstructural Evolution During Strip
Laboratory, Roskilde, Denmark, 1980, p Rolling, ISIJ International, Vol 30 (No. 3),
REFERENCES 325 March 1990, p 216225
23. T. Gladman, Proc. R. Soc. (London) A, Vol 46. L.N. Pussegoda, S. Yue, and J.J. Jonas,
1. A.J. DeArdo, C.I. Garcia, and E.J. Palmiere, 294, 1966, p 298 Laboratory Simulation of Seamless Tube
Thermomechanical Processing of Steels, 24. M. Hillert, Acta Metall., Vol 13, 1965, p 227 Piercing and Rolling Using Dynamic
Heat Treating, Vol 4, ASM Handbook, ASM 25. L.J. Cuddy, Metall. Trans. A, Vol 12, 1981, Recrystallization Schedules, Metall. Trans.
International, 1991, p 237255 p 1313 A, Vol 21 (No. 1), Jan 1990, p 153164
2. I. Kozasu, C. Ouchi, T. Sampei, and T. 26. A.J. DeArdo, University of Pittsburgh, un- 47. Bulk Formability of Steels, Properties and
Okita, Microalloying 75, M. Korchynsky et published research, 1983 Selection: Irons, Steels, and High-
Performance Alloys, Vol 1, ASM Handbook, Transformations During the Thermal/ 49. T.M. Maccagno, J.J. Jonas, and P.D.
ASM International, 1990, p 587 Mechanical Processing of Steel, Vancouver, Hodgson, Spreadsheet Modelling of Grain
48. J.J. Jonas; T.M. Maccagno, and S. Yue, The 2024 Aug 1995, Canadian Institute of Size Evolution during Rod Rolling. ISIJ
Role of Dynamic Recrystallization in Mining, Metallurgy and Petroleum, 1995, p International, Vol 36 (No. 6), 1996, p
Industrial Hot Working, Conf. Proc.: Phase 179193 720728
Chapter 18
Workability and
Process Design in Rolling
John G. Lenard, University of Waterloo
THE MANUFACTURE of successful prod- and thermal energies takes place. In the flat or of the books devoted exclusively to rolling in-
ucts requires knowledge about the three basic shape rolling process, the machine refers to the clude Ref 2 to 11. Texts concerned with the
elements of any metal forming operation: the rolling mill, including the driving motor, the theory of plasticity and metal forming include
machine, the workpiece, and the interface where work rolls, the back-up rolls, the screw-down chapters on the rolling process. In addition to the
the transfer of mechanical and thermal energies system, the mill frame, the bearings, the drive traditional texts (Ref 1218), the books cited in
takes place. These three elements determine spindles, the scale breakers, the coilers, loopers, Ref 19 to 21 should be mentioned. Publications
both the workability conditions and the process- and the control system of the mill. The parame- in various technical journals, concerned with
design factors for successful manufacture of a ters of relevance that may affect workability in- mechanical and metallurgical aspects of flat or
part. In the overall context of this book, work- clude the available power and speed of the driv- shape rolling, are numerous and the list is much
ability can be defined in general terms as any ing motor, the mill stretch and stiffness, the too long to be given here.
conditional limit on the successful creation of a resonant frequencies, the lubricant, emulsion and The Forces, Torques, and Speeds in Flat
part. Likewise, this is closely related to process the coolant delivery systems, the manufacture of Rolling (Ref 22). While the flat rolling process
design, where the three elements are properly the roll, material, strength, hardness, diameter is well known, the sequence of events experi-
configured in technological terms to manufac- and crown, the furnaces, and the responsiveness enced by the rolling mill and the metal indicate
ture products with acceptable quality in terms of of the control systems. its complexity, and the fundamental ideas de-
the mechanical, metallurgical, and geometrical The interface indicates the contact between scribed next (adapted from Ref 22) briefly con-
requirements of the customer. the work roll and the rolled metal, and the pa- sider the forces acting in the roll-metal system.
In this chapter, the subject of workability and rameters describing its mechanical and tribolog- The strip enters the deformation zone because
process design is reviewed for metal rolling ical behavior include the roughness and hard- of the friction forces exerted by the work rolls on
based on some current studies, as well as im- ness of the roll and the strip, the presence of a it, and as a result, it first experiences elastic de-
portant studies conducted earlier. Substantial lubricant, an emulsion, a coolant, and the pres- formation. The limit of elasticity is reached soon
portions of this chapter are due to Scheys orig- ence of oxides and/or scales. The workpiece is after entry. The permanent deformation regime
inal publication on this topic (Ref 1), and many the rolled metal, which may be flatthat is, of is thus in existence through most of the roll gap
of the comments and explanations of the issues rectangular cross sectionor may have a partic- region, followed by the elastic unloading
are as valid now as they were in 1984. Since the ular, nonrectangular cross section. The attributes regime. These regimes are illustrated in Fig. 1(a)
publication of the original review of workability and parameters of significance refer to the re- to (c), which show a schematic of a two-high
in rolling, numerous examinations of the topic sistance of the workpiece to deformation as a mill and a strip in various stages of the process.
have been presented in the technical literature. function of mechanical and metallurgical vari- In Fig. 1(a) the strip is ready to be rolled and
While many of these considered physical simu- ables. It is argued that while the list is certainly entry is imminent. In Fig. 1(b) it is rolled part-
lations and the development of failure criteria, a incomplete, it includes the most significant pa- way through to roll gap and in Fig. 1(c) the
significant portion of the studies involved the rameters. Further, it is well understood that no rolling process is continuous and the forces and
application of mathematical models to the pre- realistic analysis of the metal rolling system can torques acting on the work roll are also shown.
diction of the occurrence of defects. In what take all of these factors into consideration, and The conditions shown describe either a labora-
follows, the phenomenological issues of rolling mathematical models for acceptance and use by tory situation where no front and back tensions
are discussed first. Mathematical models of rolling mill engineers invariably contain simpli- exist, or a single stand, reversing roughing mill.
various complexities are presented next. These fying assumptions and approximations. A fairly In Fig. 1(a), where the strip is about to make
will then be applied to workability in general recent addition to the family of mathematical contact, if the coefficient of friction is larger
terms and then in reference to rolling. A presen- models is the use of artificial intelligence tech- than the tangent of the bite angle, a relationship
tation of process design in rolling follows. The niques, which, if well designed, may in fact take that is often used to determine the minimum
chapter closes with some general and specific all parameters and possibilities into considera- friction necessary to start the rolling process, the
conclusions. tion and may be continuously upgraded using strip enters the deformation zone. In a laboratory
the most current data. mill, the usual practice is to push the strip,
This section briefly reviews some phenome- placed on the delivery table, toward the work
The Rolling Process nological aspects of rolling, which is also intro- rolls and allow the friction forces to cause entry;
duced in Chapter 16, Rolling. The flat and under certain circumstances it is necessary to
As noted, the three basic elements of a form- shape rolling processes, given their highly devel- taper the leading edge of the strip to facilitate the
ing process are the machine, the workpiece, and oped economic significance, are also reviewed bite. In a strip mill, edge rolls force the strip into
the interface where the transfer of mechanical in detail in a large number of publications. Some the roll gap in the first stand, and the momentum
Chapter 18: Workability and Process Design in Rolling / 259
duced; however, the writer is not familiar with and back tensions are present, as would be the
publications that may have resulted from the case under industrial conditions, their effect on
research. the longitudinal stresses at the entry and exit
Figure 1(b) shows the strip about halfway should be included in the definitions of the
through the deformation zone. As previously boundary conditions.
mentioned, the entering metal first experiences The surface velocities of the roll and the strip
elastic deformation, and when and where the should also be considered. It may be assumed
yield criterion is first satisfied, plastic flow is ob- that the rolls rotate at a constant angular velocity,
served. These two regimes are separated by an even though there may be some slow down under
elastic-plastic boundary, the location of which high loads. The strip usually enters the roll gap at
should be determined by an analysis of the a surface velocity less than that of the roll. The
rolling process. In the elastic region, the theory friction force always points in the direction of the
of elasticity describes the deformation. In the relative motion, and on the strip it acts to aid its
permanent deformation region, the criterion of movement. As the compression of the strip pro-
yielding, the associated flow rule and the condi- ceeds, its velocity increases and it approaches
(a) tion of incompressibility describe the situation. that of the surface of the roll. When the two ve-
The rolls are further flattened. The magnitude locities are equal, the no-slip region is reached,
of the roll stresses should not exceed the yield often referred to as the neutral point, or more cor-
strength of the roll material. The theory of elas- rectly, a neutral region, where the strip and the
ticity is to be used to determine the roll distor- roll move at the same speed (see Chapter 16,
tion and the corresponding changes of the length Rolling). If the neutral point is between entry
of contact. and exit, beyond it the strip experiences further
In Fig. 1(c), the leading edge of the rolled compression, and its surface velocity exceeds
metal has exited and the process is continuing. that of the roll. Between the neutral point and the
The figure shows the pressures, forces, and exit, the friction force on the strip has changed
torques acting on the roll and strip. These in- direction and is now retarding its motion.
clude the roll pressure distribution and the inter- Figure 2 illustrates the side-view of the flat
(b) facial shear stress, the integrals of which over rolling process, showing L, the projected contact
the contact length lead to the roll separating length, and j, the independent variable in the
force and the roll torque. These are the variables rolling direction. It is generally accepted that in
the models are designed to determine. If front strip rolling the field variablesstress, strain,
(c)
Fig. 1 (a) Schematic of the entry of the strip into the

roll gap. (b) The strip is partially in the deforma-
tion zone. (c) Free body diagram of the work roll and the
rolled strip
of the strip exiting from there carries it into the

deformation zone of the next stand. In either
case, entry creates some longitudinal compres-
sion at the leading edge of the strip, where there
will be some initial thickening as well. This is
accompanied by local, elastic deformation of the
work rolls, indicating that the usual simplifica-
tion about the entry pointlocated where the
edge of the strip encounters the undeformed,
perfectly cylindrical rolldoes not represent re-
ality very well. Relatively little has been done to
analyze the exact entry conditions of the strip
into the roll gap, indicated by the circle in Fig.
1(a). In one of the attempts, high-speed photog-
raphy was used (Ref 23) to allow visualization
of the entry conditions and the length of contact
in the hot-rolling process. A doctoral disserta- Fig. 2 Schematic of the flat rolling process, showing the distorted work roll, the roll pressure, and the interfacial shear
tion (Ref 24) on the topic has also been pro- stress distributions
rate of strain, and temperaturedont vary in ductions of thin, hard strips using large diameter the average thickness of the rolled metal to the
the direction perpendicular to the plane shown. rolls are analyzed. projected contact length, h/L. The description of
This is a consequence of the very low straining The shear stress distribution in Fig. 2 shows the role of this ratio, given in Ref 1 is most rele-
the rolled metal experiences in the width direc- the location of the no-slip point, which does not vant and in what follows, it is repeated essen-
tion and makes the assumption of plane-strain necessarily coincide with the location of the tially unchanged. When h/L 1, deformation
deformation realistic. A similar schematic of the maximum roll pressure, as demonstrated expe- becomes somewhat similar to that observed in
rolling process appeared in the first edition (Ref rimentally (Ref 2830). Also, when strips are the indentation of a slab with two opposing nar-
1), but in more idealized form with undeformed rolled at higher speeds to higher reductions and row tools. At high values of h/L, deformation is
work rolls and the traditional friction hill, in lubricants of higher viscosities are used, no neu- concentrated in zones adjacent to the rolls (Fig.
which the location of the saddle point is taken as tral point is located, the strip exits at a velocity 4) and does not penetrate the full thickness of
the location of the no-slip region (Fig. 3). The less than the surface velocity of the work roll, the workpiece. Deformation against the restraint
sharp saddle point, or cusp, in Fig. 3 results from and the forward slip becomes negative. These presented by the rigid (or only elastically de-
one-dimensional analysis of the abrupt change differences are of importance when cold rolling formed) central portions of the workpiece neces-
in the direction of frictional resistance at the of thin strips and foils are considered. They are sitates a rise in interface pressures (Ref 31) for
neutral point. If a more realistic friction model is of less importance in the study of the hot rolling forging and then recognized as relevant for
used to relate relative slip between the roll and of strips and plates when the resistance of the rolling (Ref 32). Because rolling may be re-
strip, then the sharp cusp in the traditional fric- rolled metals is substantially lower. garded as a progression of indentation (Ref 33),
tion hill is replaced by a smooth-hill top. Homogeneous Versus Non-Homogeneous the interface pressure can be calculated by anal-
Figure 2 indicates the deformation of the Deformation. The usual assumption of homo- ogy to the plane-strain indentation of a slab,
work roll, albeit in a highly exaggerated fashion geneous deformation, that is, planes remaining either from the curve given by Hill from slip-line
and gives its shape as y f(x), to be determined planes during a rolling pass, made in most one- field analysis (Ref 12), or from the experimental
as part of any mathematical analysis. The roll dimensional mathematical models of the flat relationship determined by Smirnov and re-
pressure distribution is also different, depicting a rolling process, represents the actual conditions ported in Ref 9:
more realistic, smooth shape, not the sharp very well, provided the ratio of the roll diameter
0.4
saddle-point obtained in simplified one-dimen- (D) to the strip thickness (h) is significantly p h
= 2.57
sional analyses. It is recalled that in the tradi- larger than unity. This condition is well satisfied 2k L
tional, one-dimensional analyses of the flat when cold rolling of strips and foils is consid-
rolling process (Ref 2527), the roll pressure ered, in which case the D/h ratio may approach where p is the average interface pressure and 2k
distribution is obtained by integrating the differ- 103. When hot rolling strips in the finishing train is the plane-strain flow stress of the workpiece
ential equations of equilibrium from the entry of strip mills, this ratio is smaller but the defor- material. It is immaterial whether the interface is
and from the exit, and the location of the neutral mation is not too far from being homogeneous. lubricated; the pressure increase is due to geo-
point is taken as the place where the two curves The deformation may well deviate, in a signifi- metric restraint and not to friction.
meet. However, it is also well known that the cant manner, from the homogeneous condition The transition from almost homogeneous to
one-dimensional analysis of the friction hill ap- during rough rolling of strips, plate rolling, or highly inhomogeneous deformation is, of course,
proach can fail when interface friction assump- bar rolling, however. The important parameter is not abrupt. At intermediate values of h/L, the de-
tions fail. One such case occurs when large re- the shape factor (Ref 1), defined as the ratio of formation zone may be extremely complex (Ref
(b)
Fig. 4 Highly inhomogeneous plane-strain deformation

at large h/L ratios. (a) Deformation. (b) Pressure
Fig. 3 Idealized view of homogeneous plane-strain rolling without deformation of work rolls. Source: Ref 1, p 270 distribution in arc of contact. Source: Ref 1, p 272
trend is indeed found when roll pressures cal-

culated from numerous production roll-force
measurements are analyzed (Ref 34). Much more
difficult is a theoretical treatment of the inho-
mogeneous deformations themselves, but the
deformation-zone geometry and the relative ve-
locities for several geometries typical of hot
rolling were identified. It was discovered that a
single dead-metal zone (Fig. 7a) extends from
entry to exit at combinations of relatively low
pass reduction and roll-radius to strip-thickness
ratio. Their relationship can be reduced to the
statement that a single, fully extended dead-
metal zone exists at h/L 0.3 (Ref 36), or at h/L
0.5. At lower h/L values, the rigid zone is
bounded by two intense deformation zones in
which relative motion occurs at the strip/roll in-
Fig. 5 Deformation zones at intermediate h/L ratios. (I) Dead-metal zone; (II) Intense deformation. (III and IV) Indirect
terface (Fig. 7b). This solution puts the transi-
deformation zones. Source: Ref 1, p 273
tion from full to partial sticking at a lower h/L
value than that found in the experiments re-
ported in Ref 11. In these latter experiments, un-
usually smooth rolls were used for unlubricated
hot rolling of aluminum and copper, but an ap-
proach to full sticking was found only at h/L 2.
The distributions of stresses, strains, rates of
strains and temperatures, supporting the above
conclusions, can be determined by the use of
mathematical models.
Mathematical Models of the Flat Rolling
Process. The effects of process, material, and
geometrical parameters on the rolling mill and
the rolled material may be predicted using math-
ematical models of various sophistication. The
output of the models depends on their mathe-
matical rigor and the assumptions made during
Fig. 6 Pressure-multiplying factor as a function of h/L ratio. Source: Ref 1, p 274
the derivations. They may include the roll sepa-
rating force and the roll torque; the radius of the
deformed roll, the roll pressure and the interfa-
cial shear stress; or, if two- or three-dimensional
finite element or finite difference approaches are
used, the distributions of the displacements, ve-
locities, strains, strain rates, stresses, tempera-
tures and grain sizes. Further sophistication also
allows the calculation of the post-rolling attrib-
utes: the yield strength, the hardness and the size
(a)
of the austenite/ferrite/pearlite/bainite grains.
The models, discussed in brief subsequently,
may be used to shed light into various mecha-
nisms that operate during the rolling pass and
from that point of view, are of immense use
when workability limits are analyzed. Further,
the models may be used for the predictions of
the mechanical parameters and material attrib-
utes, and in this context their contributions are of
(b) importance in the design of the hardware or of
the rolling process. The models may be on-
Fig. 7 Slip-line fields for hot rolling. (a) Single dead-metal zone extending from entry to exit. (b) Relative motion line or off-line. While in the former, the time
between strip and roll surface at entry and exit. Source: Ref 1, p 274
taken for computation must be short, in the lat-
ter the time required for number crunching may
11). Rigid (dead-metal) zones extend partway action of the deforming zone with the rigid parts be long due to the iterative calculations.
into the body of the workpiece from the sticking of the outlying portions of the workpiece. The models of the rolling process, for potential
zone on the roll surface (zones I in Fig. 5). The combined effects of inhomogeneous de- on-line use, include both empirical and semi-
Intense deformation occurs between these two formation and friction on roll forces can be read- empirical types (Ref 27, 37, 38). Simple relations
zones, whereas minor plastic deformation takes ily calculated, at least approximately, by com- for the roll separating force and the roll torque,
place outside the entry and exit where the work- bining the effects of the h/L ratio and the friction for use during hot rolling of carbon steels, are
piece actually thickens slightly while contracting hill; the resulting curve shows (Fig. 6) a mini- given in Ref 37. Simplified formulae for the
longitudinally; this is a consequence of the inter- mum of the ratio p/2k at around h/L 1. Such a same parameters are presented in Ref 27 and 38.
The curves presented in Ref 39 may also be used,

but some method of transferring the values, pre-
dicted by the curves, to a computer needs to be
devised. A more elaborate, iterative one-
dimensional model (Ref 25) includes the effect
of roll flattening, the contributions of the elastic
entry and exit regions, and strain hardening as the
strip or plate is traveling through the roll gap.
The models presented in Ref 26 and 27 are both
obtained from the Orowan model (Ref 25)
through various simplifications. A statistical
analysis of the predictive capabilities of Orowans
and Ford and Alexanders (Ref 37) models indi- (a)
cated that the standard deviation of the difference
between measurements and predictions de-
creased with the increasing sophistication of the
model (Ref 40).
The one-dimensional models, most of which
assume that planes remain planes, that is, the
rolled strip experiences homogeneous compres-
sion, are useful for the prediction of the roll sep-
arating forces, as long as two conditions are sat-
isfied. The first is a good understanding of the
tribological conditions at the contact surfaces,
allowing one to use an appropriate value for the
coefficient of friction. The second is a mathe-
matical limitation; the roll diameter to strip
thickness ratio should not be excessive since
(b)
otherwise the models will not converge. Further,
since the calculations of the roll torque are per-
formed by integrating over the contact surface
and the true extent of that is not known accu-
rately, torque predictions are seldom accurate
and consistent.
If more accuracy and consistency of predic-
tions are required, a two-dimensional rigid-
plastic finite element model (FEM) may be
necessary. Many of these models (Abaqus, LS-
Dyna, MARC; Deform; Forge, for example) are
commercially available. Many of them are based
on the minimization of the required energy, and
they determine the roll torque from the power
thus resulting in improved prediction of both the
roll separating force and the torque. For proper
functioning, they need an appropriate descrip-
tion of the boundary conditions, and they also (c)
need appropriate values of the coefficients of
friction and heat transfer, both of which may be
difficult to provide.
The rigid-plastic formulation is considered
adequate for the simulation of hot rolling. For
suitable analysis of cold rolling, an elastic-
plastic model is necessary. Rigorous analysis of
shape rolling needs three-dimensional models.
Models, capable of predicting the mechanical
and metallurgical attributes after hot rolling, are
also available (Ref 22).
Stress, Strain, Strain Rate and Tempera-
ture Distribution in the Flat Rolling Process.
As just stated, the states of stress, strain, rate of
strain and temperature in the rolled sample will
determine the limits of workability of the rolled
metal. These distributions are shown in Fig. 8, 9,
and 10, obtained by a two-dimensional rigid- (d)
plastic finite element approach, using the com- (a) The effective strain distribution; hentry 300 mm (12 in.); h/L 0.98. (b) The effective strain rate distribu-
mercially available program Elroll (Ref 22). Hot
Fig. 8
tion; hentry 300 mm (12 in.); h/L 0.98. (c) The effective stress distribution; hentry 300 mm (12 in.); h/L
rolling of a C-Mn steel plate, reduced by 40% in 0.98. (d) The temperature distribution; hentry 300 mm (12 in.); h/L 0.98
a shape factor of h/L 0.98. In Fig. 9, the same

distributions are given, but the entry thickness is
reduced to 50 mm (2 in.), leading to an interme-
diate shape factor of 0.4. Finally, in Fig. 10, the
distributions are repeated, and a thin strip of
2 mm (0.08 in.) entry thickness is considered. In
this case, the mean thickness to projected con-
tact length ratio is 0.08.
(a) The expected nonhomogeneity of the defor-
mation is clearly observable in Fig. 8(a), (b), (c),
and (d). The largest effective strains are located
near the exit plane, at the contact surface in
between the work roll and the rolled strip. The
strains reduce toward the centerline, and the
strain gradient in the through-thickness direction
is high, pointing to the localization of the strains
near the exit. The nonhomogeneity of the defor-
mation is also observable at the entry. The rates
of strain are also distributed in a nonhomoge-
neous manner, as noted in Fig. 8(b). The maxi-
mum strain rates, however, are predicted to be
(b) near the entry point, and they diminish toward
the exit. The distribution of the stresses (Fig.
8c), indicates the highest concentration near the
entry, as was the case with the rates of strain.
This is expected, since at the temperatures con-
sidered, the resistance of the steel to deforma-
tion is mostly dependent on strain rates. The
stress maxima are observed in the contact zone
where intense temperature gradients are also
found, as shown in Fig. 8(d). The distributions of
the temperatures are also highly nonhomoge-
neous and the surface of the rolled strip is pre-
dicted to cool to a low of 700 C (1300 F), cre-
ating a material with sharply reduced ductility.
(c) The nonhomogeneity is reduced, but still no-
ticeable, when the shape factor is reduced to 0.4,
as indicated in Fig. 9(a) to (d), showing the re-
sults when the entry thickness of the hot strip
was reduced to 50 mm (2 in). The distributions
of the effective strains and strain rates are not
significantly different from that shown in Fig.
8(a) and (b); the largest strains are near the exit
and the zone of contact and the largest rates of
strain are near the point of entry. Both indicate a
certain measure of nonhomogeneity. The largest
stresses are now found near the possible location
of the maximum roll pressure and nearer the
centerline of the rolled strip, demonstrating the re-
(d) duced nonhomogeneity. The magnitudes of the
strain rates and, hence, the stresses are larger
Fig. 9 (a) The effective strain distribution; hentry 50 mm (2 in.); h/L 0.40. (b) The effective strain rate distribution; than those shown in Fig. 8(b) and (c) for the
hentry 50 mm (2 in.); h/L 0.40. (c) The effective stress distribution; hentry 50 mm (2 in.); h/L 0.40. thicker strip, necessitated by the creation of a
(d) The temperature distribution; hentry 50 mm (2 in.); h/L 0.40
larger portion of stressed metal. Not much dif-
ference in the temperature distributions is ob-
one pass is presented, similar to a pass in the of the carbon content, temperature, strain rate served when Fig. 8(d) and 9(d) are compared;
roughing stand of a hot strip mill. The calcula- and strain, are used as the constitutive model, the coolest part is again near the surface of con-
tions were performed and the figures prepared with the carbon equivalent taken as 0.3. The FE tact. The loss of the temperature is reduced;
by Prof. M. Pietrzyk, (Akademia Grniczo- model also requires the coefficient of heat trans- however, the surface is warmer and correspond-
Hutnicza, Krakow, Poland) using Elroll, a 2D fi- fer, and in the figures that follow, 50,000 W/m2K ingly more ductile. Almost complete homoge-
nite element program. was used. In all calculations, the roll was as- neous compression is observed in Fig. 10(a) to
The slab is heated to a uniform temperature sumed to remain rigid, not accurate, but not un- (d) where the shape factor is 0.08. The major dif-
of 950 C (1740 F) before rolling. The process reasonable, in the hot-rolling process. ferences concern the very much larger rates of
parameters are as follows. The roll radius is Figure 8 demonstrates the distributions of the strain and the very high stresses as a result.
500 mm (20 in.), the roll velocity is 50 rpm, and strains (a), the rates of strains (b), the stresses These stress and strain distributions may be
the friction coefficient is 0.4. The equations in (c) and the temperatures (d), considering a thick used to indicate the potential locations of duc-
Ref 41 describing the flow strength as a function plate of 300 mm (12 in.) entry thickness, having tile fracture, based in the flow localization
designing the rolling process, discussed in

some detail in the section Process Design.
The limits contributed by the workpiece are
twofold. The first set refers to mechanical phe-
nomena and is concerned with the ability of the
rolled metal to deform as the rolling process
demands (Ref 52). These involve the nonhomo-
geneity of the variables strains, rates of
(a) strain, stresses, and the temperatures dis-
cussed previously. The second limitation is
metallurgical in nature and is caused by the
often-observed ductility trough. This is most
significant when hot rolling of plates and strips
is considered. The limits indicate the reaction
of the material to the thermal-mechanical treat-
ment it is receiving, including the effect of the
treatment on ductility.
The limits associated with the interface, the
contact zone between the rolled metal and the
roll, affect the quality of the surfaces. These in-
clude friction, heat transfer, and the develop-
(b) ment of the roughness of the rolled metal, all of
which depend on tribological phenomena and
are influenced by the transfer of mechanical and
thermal energies. These factors are not directly
related to fracture, and they are not often con-
sidered in the usual treatments of workability.
However, the interface has much to do with frac-
ture. The heat transfer at the interface and the
nature of surface roughness (at the interface) can
affect surface fracture. Indirectly, friction also
affects internal stresses and even the formation
of the edge profile, which can have a large influ-
ence on edge cracking.
(c)
Workability Limits due to

Nonhomogeneity
In the idealized situation of homogeneous
deformation under conditions of plane-strain
rolling, straight vertical sections in the work-
piece remain almost straight (Fig. 3). No sec-
ondary tensile stresses are developed during
rolling, and no residual stresses remain in the
product. This idealized result from elementary
(d) plasticity theory assumes that an ideal rigid-
plastic workpiece is rolled on relatively large-
diameter rigid rolls at heavy reductions. Of
Fig. 10 (a) The effective strain distribution; hentry 2 mm (0.08 in.); h/L 0.08. (b) The effective strain rate distribu-
tion; hentry 2 mm (0.08 in.); h/L 0.08. (c) The effective stress distribution; hentry 2 mm (0.08 in.); h/L course, this idealization is not completely realis-
0.08. (d) The temperature distribution; hentry 2 mm (0.08 in.); h/L 0.08 tic in terms of deformations. Even under condi-
tions of plane-strain (approximated under condi-
tions of thin products with large widths and
lengths), deformation can be frequently inhomo-
technique. It is realistic to conclude that frac- face of fracture may be predicted when a fail- geneous in the through-thickness direction. At
ture occurs (depending on the chosen failure ure criterion is defined in conjunction with large h/L ratios, for example, secondary tensile
criterion) where the stresses, strains, and rates the shear stresses (e.g., see Chapter 12, Work- stresses may develop in the centerplane and
of strains are the highest and where their gradi- ability Theory and Application in Bulk Form- could lead to center bursts. Conversely, with de-
ents are high. These, combined with the loss of ing Processes). creasing h/L ratios, the danger of inhomoge-
ductility resulting from roll chilling, are usu- neous deformation now shifts to the surface.
ally responsible for the types of defects that Wherever the actual point of transition from ho-
cause the limit of workability to be breached. Workability in Rolling mogeneous to inhomogeneous deformation may
Another approach, not used often to predict be, it is evident that a variety of situations can
failures, is the application of the three- The limits introduced by the rolling mill refer develop that may be harmful in rolling of mate-
dimensional Mohrs circle, which allows the to its load, torque, and power carrying capabil- rials, as described subsequently and in Chapter
visualization of potential surfaces of fracture. ity, including its deformation and dynamic re- 12, Workability Theory and Application in
If the state of stress can be determined, the sur- sponse, and these are best considered when Bulk Forming Processes.
A detailed, systematic study of the limits of and, as shown in Fig. 8(b), 9(b), and 10(b), the countered as a result. As the scale layer cracks,
workability during hot rolling of aluminum slabs localization of the strain rates is highest at the the islands become separated as the strip is
was conducted, using samples of rectangular entry. The rates drop as the exit is approached. elongated. Hot metal then extrudes between the
and tapered cross sections. The state of stress in Contributing to this is the possibly incomplete islands and sticks to the rolls while the sliding
the hot rolled slabs was examined and the possi- coverage of the rolls surface by the layer of the islands move farther apart and promote tensions
ble planes of fracture were identified using the mineral seal oil, some of which may have burnt applied to the sticking portion, thereby reduc-
three-dimensional Mohr circle analysis (Ref 43). off as the pass was nearing completion. The po- ing the load. At the same time, simultaneous
Two alloys were selected for this investigation: a tential loss of lubrication is overwhelmed by the smoothing and roughening of the rolled surface
5182 alloy, containing 0.06% Cu, 0.37% Mn and strain rate localization, and the trailing portion occurs, which is visible after pickling and may
4.65% Mg, which has a susceptibility to alliga- of the strip, where strain rates are more uniform, well cause difficulties in subsequent cold roll-
toring and edge cracking during the early stages does not demonstrate surface cracking. Once ing passes (Ref 47). The surface of a carbon
of hot rolling, following reheating, that is known sufficient lubrication was available, no cracks steel strip, rolled at 877 C (1610 F) shows no
to be very high, and a 7075 alloy containing were produced. The 7075 alloy, rolled at 460 C surface cracking, as in Fig. 14. When the tem-
1.45% Cu and 2.5% Mg, which is normally (860 F), also suffered some minor surface perature was 842 C, severe cracking was ob-
extruded rather than rolled, but possesses hot cracking, as in Fig. 12. While the cracks appear served as shown in Fig. 15, again indicating the
ductility that is very different from that of the to cover all of the surface of the strip, from entry chilling and the loss of surface ductility (Ref
5182 alloy. During the process, the work rolls to exit, they dont extend over the complete 48) in addition to the very strong dependence of
were covered with a thin layer of mineral seal oil width. It is possible that surface lubrication was workability on the temperature.
to eliminate sticking. The speed of rolling was more effective in the test shown in Fig. 12. The Reference 1 describes the tensile stresses that
50 rpm, corresponding to a surface speed of ap- strain rates in the two tests were similar, so the may be imposed on the surface and may cause
proximately 0.4 m/s (1.5 ft/s). The average superior ability of the 7075 alloy to avoid sur- surface cracks; similar conclusions are presented
shape factor in the tests was 0.6, near the shape face cracking is demonstrated. in Ref 35. Surface cracking may also be due to
factor of 0.4, analyzed previously and for which Cracking of the surfaces during the hot rolling stress corrosion (Ref 16) or by sudden changes
the distributions are given in Fig. 9(a) to (d). of steel is influenced by the development of the of tribological conditions as the rolling pass pro-
Several types of workability limits were dis- layer of tertiary scales. In the finishing stands of ceeds (Ref 44). This was also experienced by the
covered, including surface cracking, alligator- the hot strip mill, the thickness of this layer is in author of this article in an experiment while at-
ing, edge cracking, and a combination of these. the order of 10 to 20 mm and is likely quite hard tempting to overcome the minimum friction nec-
These limits are discussed below, first consid- with limited ductility (Ref 45). essary to cause bite. A tapered leading edge was
ering the rectangular cross-section samples, A schematic of the cross section of the hot prepared and kept dry. Neat oil was applied only
followed by those with tapered edges. steel is reproduced here as Fig. 13, indicating some distance away from the edge. Successful
Surface Cracking. The cracking of the sur- the potential surface problems that may be en- entry was achieved, but the steel strip tore vio-
faces is caused by the combination of roll chill- lently as the lubricated portion was reached. The
ing, as indicated in Fig. 8(d), 9(d), and 10(d), effect of changing surface conditions on surface
and the corresponding loss of metal ductility. cracking was also observed by others (Ref 44,
The temperature gradients shown in Fig. 8(d), 49, 50). Surface cracking during direct rolling of
9(d), and 10(d) indicate the severity of the cool- continuously cast low-alloy steel slabs was stud-
ing of the surfaces. This phenomenon, combined ied by means of hot tensile tests (Ref 51). The
with the possible loss of lubricants, either due to loss of ductility was explained in terms of dy-
lubricant breakdown as the temperatures rise or namic precipitation of carbides and nitrides.
due to improper lubricant delivery, results in Edge cracking. As stated in Ref 1, the most
the dramatic increase of the interfacial shear Fig. 12 Surface and edge cracking of a 7075 Al alloy, severe problem in the loss of workability is the
stresses, approaching or surpassing the metals rolled at 460 C (860 F). Source: Ref 43 cracking of the edges. This cracking is attributed
resistance to deformation (Ref 44). The cracks
usually occur in a direction parallel to the rolls
axis of rotation.
Surface cracking is demonstrated in Fig. 11,
showing a sample of the 5182 alloy, rolled at 560
C (1040 F) to a strain of 0.58. It is interesting to
observe that the surface cracks are deepest near
the edges and are located near the leading edge
of the sample. They appear to vanish as the
rolling pass nears completion. The resistance of
the metal to deformation at the temperature of
the pass is highly dependent on the rate of strain,
Fig. 11 Surface cracking of a 5182 Al alloy, rolled at Fig. 13 Schematic of the cross section of a hot steel strip, indicating the cracked layer of scale. Trs, depth of roller
560 C (1040 F) to a strain of 0.58. Source: scale; Toxr, thickness of oxide on roller; Toxs, thickness of oxide on stock; Tss, depth of scale on stock. Source:
Ref 43 Ref 46
Fig. 16 No ductility problems of a 5182 Al alloy,

rolled at 520 C (968 F). Source: Ref 43
Fig. 17 Edge cracking of the 7075 Al alloy, rolled at

460 C (860 F). Source: Ref 43
Fig. 14 No surface cracking on a strip of low carbon

steel, rolled at 877 C (1611 F). Source: Ref 48
Fig. 15 Cracked surface of a hot steel strip, rolled at

842 C (1548 F). Source: Ref 48
to the secondary tensile stresses, developed as a

result of nonhomogeneity of deformation in slabs
of finite widths. The effect of the tensile stresses
is enhanced by the temperature losses and the
added loss of ductility near the edges, which may
well be up to 100 C (212 F) cooler than the
centers of the rolled metal. Schey (Ref 1) also
describes the effects of double barreling at high
h/L ratios, found when the shape factor is greater
than 1.8 (Ref 6163). The workability of Al alloy
7075-T6 in upsetting and rolling at room temper-
ature was investigated (Ref 54) with reference
to the occurrence of free-surface cracks. A work-
ability chart was constructed by combining the
fracture criterion, the complete solution for var-
ious friction conditions, and several specimen
dimensions. With this theory it was possible to
predict workability in multipass rolling with ref-
erence to edge cracking. The limiting reductions
in thickness were given as functions of work-
piece dimensions for various friction values.
Experiments demonstrated that the predictions
were very good. The hot workability of the Al Fig. 18 The relationship between thickness reduction and tensile fracture strain. Source: Ref 56
alloy was established by tensile and torsional
testing (Ref 55). Three-dimensional diagrams,
showing the effects of preliminary deformation A slab of the 5182 alloy, rolled at 520 C strip, made of Al 2024 T351 alloy. The rela-
and temperature on workability and the effects of (970 F) to a true strain of 0.53, showed no tionship in between the reduction of thickness
the stress state and deformation rate on workabil- ductility problems (Fig. 16). The 7075 alloy ex- and the tensile fracture strain is shown in Fig.
ity show that the optimum rolling temperature hibited some edge cracking when rolled at 460 18. The figure indicates the dependence of the
range for this alloy is between 440 and 500 C C to a strain of 0.6 (Fig. 17). Reference 56 dis- possible thickness reduction on the manner of
(820 and 930 F). The workability is minimal at cussed the effect of edge cracking, which pro- edge preparation. The state of stress was found
30 to 50% hot reduction. gresses to the center of the square-edged rolled to be most severe when chamfered edge strips
were rolled and least severe when square edges broadly. Reference 1 quotes h/L values of 1.3 strips in terms of the shape factor, was devel-
were used. (Ref 53). While rolling aluminum (Al-8 Mg) bil- oped. It was concluded that as long as the shape
The criterion from Ref 57 was applied suc- lets, alligatoring appeared at shape factors of 0.5 factor, defined as the ratio h/L, is less than 0.75,
cessfully (Ref 58) to correlate crack formation to 0.7 (Ref 63). In the study quoted in Ref 43, neither defect will occur, somewhat contradic-
and stresses, computed by an FEM of the rolling the 7075 alloy showed alligatoring when rolled tory to the conclusions reached in Ref 43. The
process. A three-dimensional model with thermal- at 510 C (950 F) (Fig. 19). An upper bound ap- results were presented in plots of the relative re-
mechanical coupling, using flow formulation and proach (Ref 64), which successfully predicted duction versus the relative thickness and the safe
the Norton-Hoff viscoplastic material descrip- the occurrence of central burst and alligatoring and nonsafe zones identified. The results are
tion, was developed. Experiments in flat and during cold rolling of copper and aluminum shown in Fig. 20, obtained for a perfectly plastic
shape rolling were conducted, using plasticine material. The figure gives the safe zone, the cen-
as the model material. The crack opening crite- tral burst zone, and the zone where both defects
rion function was defined as: are expected in terms of the relative reduction
and the relative thickness. These defects tend to
C pass max[(snc)nc,0]d e be promoted by small roll radii and large initial
workpiece thickness.
where s is the Cauchy stress tensor, nc is the unit Limits of Workability While Rolling
normal, and e is the equivalent strain. Tapered-Edge Samples. The study of workabil-
Edge cracking of low carbon steel sheets, pro- ity limits of aluminum samples also considered
duced in a mini-mill, was examined (Ref 59, the effect of tapered edges (Ref 43), which cre-
60). It is recognized that while the rolling of ated complex stress states, and as a result, a
thinner gages in mini-mills enhances the in- combination of edge and surface cracking in
dustrys competitive advantage, cracking of the addition to alligatoring was observed. Surface
edges was found to cause difficulties. The reason cracking and alligatoring were observed when
for the problem was identified as embrittlement the 7075 alloy was rolled at 460 C (860 F);
at high temperatures. The authors concluded that the edges were tapered at 15 from the vertical
the increase in manganese content and smaller (Fig. 21). Recall that the same alloy, having rec-
amounts of FeS helped in the suppression of tangular edges and rolled at 460 C (860 F),
edge cracking. Isothermal preheat treatment also suffered minor cracking of the edges (Fig. 17).
helped in reducing the cracking of the edges. Fig. 19 Alligatoring of the 7075 Al alloy, rolled at
Similar behavior is observed when the 5182
510 C (950 F). Source: Ref 43
Reference 61 considers defects due to porosity
or shrinkage, which are present in all steels, fol-
lowing solidification. Process conditions under
which the internal stresses are able to eliminate
porosity are defined. It was concluded that when
the reduction was in excess of 30% all pores
closed.
Centerburst defects are more commonly
found in the extrusion process but can occur in
slab rolling when the slab product thickness is
large, such as for some ship plates or pressure
vessel plates. There are few public reports on the
subject because it has not been treated by aca-
demic institutions, and so it is not surprising that
the search of the literature yielded no recent pub-
lications concerning the occurrence of center-
burst during rolling of metals. As noted in Ref 1,
internal defects due to secondary tensile stresses
generated in the center of the slab may occur at
early stages of hot rolling when h/L is very large,
but the temperature is still high enough to ensure
reasonable ductility in all but the most brittle ma-
terials. Defects that form also have a chance of
being healed by forge welding. Most practical
hot rolling is performed in the finishing stands,
and as shown in Fig. 10, this causes little inho-
mogeneity. It is conceivable, however, that this
danger exists more generally during the rolling
of superalloys, refractory-metal alloys and tool
steels, particularly if restricted mill power limits
the attainable reductions and thus keeps h/L high.
Centerburst of a steel beam during rolling was
demonstrated (Ref 62).
Alligatoring. This defect is also related to
nonhomogeneity in the rolling process, in the
through-thickness direction. It affects metals
of lower ductility but not exclusively. The shape
factor at which alligatoring is observed varies Fig. 20 The safe, central burst, and alligatoring zones during bar rolling. Ref 64
shearing (location 3), except at the roll/sample

interface (location 1). The stresses in the direc-
tion of rolling and in the transverse direction are
practically identical near the longitudinal center-
line, whereas near the edge, the magnitude of the
transverse stress decreases. The shear stresses
are approximately half of the normal stresses
differences.
According to Trescas criterion for yielding,
deformation will occur when the maximum
shear stress reaches the yield strength of the
metal in pure shear. Further, the most probable
surface of fracture may be located by searching
Fig. 21 Alligatoring of the 7075 Al alloy, rolled at for the plane on which shear stress of the great-
460 C (860 F). Source: Ref 43
est magnitude is acting. Thus, the description of
Fig. 22 Alligatoring of the 5182 Al alloy, rolled at maximum shear stresses, as determined by 3D
alloy, with edges making 45 with the vertical, 558 C (1036 F). Source: Ref 43 Mohrs circle, gives the planes on which fracture
was rolled at 558 C (1036 F) (Fig. 22). The is likely to propagate (but not necessarily the
tapered edges produced a significant reduction initiation of fracture). In alligatoring, for exam-
traditional grinding technique and compared
of the workability, reinforcing the conclusions ple, the most likely cause is excessive shearing
the results with those obtained when the surface
reached in Ref 56. Chamfered, sheared, or fully stresses at the roll/metal interface, creating a
of the roll was prepared by electrical discharge
rounded edges also affect workability in cold state of stress such as the one shown in Fig.
machining (EDM), which produced a random
rolling. Data obtained in a rolling mill (Ref 65) 24(a), near the exiting portion of the rolled strip
roughness direction. Using the ground roll, the
with rolls of 250 mm (10 in.) diameter and (location 2). Separation of the rolled aluminum
roll force increased exponentially with increas-
rolling strips of 2.5 by 63 mm (0.10 by 2 in.), re- begins on the plane of absolute maximum shear
ing reductions, and the strip tore at a reduction
ported in Ref 1, are repeated in the table below: stress, defined in this figure. As to the edge, the
of about 45%. Rolling with the EDM surface al-
stress in the transverse direction reduces from a
lowed reductions of 78% with no tearing. The
fairly large magnitude to zero. There are two
Reduction at fracture (%) beneficial and efficient distribution of the lubri-
planes within the element at location 5 in Fig. 23
cant by the randomly placed open and closed
Edge 1.3% C steel 1.5% W steel Al-5Mg Al-7Mg on which the shearing stresses are at their maxi-
pores was credited with the increased rollability.
Square, ground 5864 53 90 60 mum level, as in Fig. 24(b). It is the resistance of
The state of stress can be derived from sim-
Fully rounded 1530 10 80 48 the material to deformation at that location that
30 chamfer 1417 ... ... ... ple mechanical considerations at different loca-
will determine on which of these planes fracture
Sheared 3152 ... ... ... tions of the sample passing through the roll gap
may occur.
As-received 2631 ... ... ... (Fig. 23). The stress in the longitudinal direction
This simple approach neglects many parame-
is mostly tensile. The stress in the transverse
ters, such as material anisotropy, material resist-
direction is generally compressive, but in the
ance to deformation, friction effects, tempera-
Roll Surface Roughness. The effect of the tapered edge or in a bulged side (location 5), it
ture distribution, or thermal history, all of which
direction of the surface roughness of the work may vanish. The shear stresses are most severe
have potentially important effects. For instance,
roll on the attainable reduction was examined near the surface at the edge (location 2), where
even though the temperature distribution within
(Ref 66). The researchers cold rolled Al-Mg-Cu they act in two directions. The elements located
the furnace is uniform, the heating rate is not,
strips, using rolls with surfaces prepared by the at the vertical centerline do not experience
with the surfaces reaching the furnace tempera-
ture sooner than the central portion. The portions
of the metal near the surface are kept therefore at
the furnace temperature for longer time periods.
If the materials attributes are sensitive to the rate
of heating, as should be the case for partly ho-
mogenized materials for instance, the balance
between the edge cracking and alligatoring will
be affected. Edges, especially those tapered to
15 and 30, are also at considerably lower tem-
peratures than those at the center of the work-
pieces.
Adequate modeling of the failure mechanisms
observed during rolling tests thus requires a
quantitative estimate of the thermomechanical
history of the different zones of the samples, a
precise estimate of the deformation path, taking
into account materials properties and friction,
and the use of fracture criteria (for crack initia-
tion) and stress analysis to identify the probable
planes of crack propagation. Stress analysis may
be very simple, such as Trescas criterion, or
more complex. In addition to Trescas criterion,
many existing criteria of fracture also show that
the hydrostatic stress state has a dominant effect
Fig. 23 The state of stress in a rolled sample. Source: Ref 43 on the probability of fracture along planes of
Fig. 25 The split ends of a hot rolled bar. Source: Ref 64
(a) pass and by a circular pass were considered. A

stress state index was defined as:
sm
ns =
si
where sm is the mean stress and si is the flow

stress. The index is related to the limit strain:
eiL K1 exp(K2ns)
where K1 and K2 are to be determined and two

testsa tension and a torsion testare needed.
The actual and predicted values of the limit
strain intensity were found to be close. In a dis-
cussion of the effect of shape factor on internal
defects (Ref 64), the split ends of a steel bar,
which initiates a crack, are shown (Fig. 25).
Workability Limited by Metallurgical

Phenomena
Charles writes in a keynote address about
(b) processability (Ref 68) that the behavior of
the material during hot processing by mechani-
Fig. 24 Potential surfaces of fracture. (a) Near the surface and strip edge shown as location 2 of Fig. 23. (b) Interior cal work is governed by its intrinsic structural
shown as location 5 of Fig. 23. Source: Ref 43
strength and ductility as a function of the mi-
crostructure and by the presence of gross imper-
fections. . . . He classifies the problems encoun-
maximum shear. This problem can be addressed whether they are on the surface or in the center.
tered as surface defects, porosity, segregation,
properly only with the assistance of two-dimen- The purpose of roll-pass design is the equaliza-
and inclusions. Reference 69 discusses the met-
sional or, most preferably, three-dimensional, tion of these strains.
allurgical causes of ductile fracture. It describes
FEM. Additional information on fracture criteria Uneven cooling of thinner portions can be
cavity-nucleated fractures, shear bands, cavita-
is given elsewhere in this book (e.g, see Fig. especially troublesome in hot rolling, and edge
tion, and grain anisotropy. Reference 70 presents
2529 in Chapter 12, Workability Theory and cracking due to cold corners is avoided by de-
five phenomena and the events that control
Application in Bulk Forming Processes, for signing roll passes with more rounded corners.
them, all of which affect formability:
edge cracking in bar rolling). In general, it is not meaningful to speak of a par-
Workability in Shape Rolling. By necessity, ticular h/L ratio in shape rolling because both the Ductility, controlled by the state of stress,
rolling of sections and shapes generates large contact surface and the deformed zone are very temperature, inclusions, segregation of re-
secondary tensile stresses because a more com- complex in shape. siduals
plex, nonrectangular shape is gradually devel- Workability depends on the chemical compo- Anisotropy, controlled by texture, thermal-
oped from a simple starting cross section. Various sition, structural state, and temperature of the mechanical history, elongated inclusions,
portions of the cross section are subjected to dif- material, in addition to the state of stress in the second phase particles, stacking fault energy
ferent reductions, and the differential elongation deformation zone (Ref 67). The rolling of a rec- Strength level, controlled by chemical com-
leads to cracking in the less-deformed zones, tangular cross-sectioned bar by a two-arc oval position, grain size, precipitates, solutes
Work hardening, controlled by tempera- that of austenitic grades 304 and 316 are while annealing at 1473 K yielded a slightly
ture, solutes, strain gradients, degree of de- strongly temperature dependent. Boron and tita- lower value than that of the as-cast material.
formation nium were found to increase the ductility of the A typical industrial hot-rolling process is
Surface properties, controlled by oxide 304 and 316 grades, respectively, while carbon, simulated by means of an interrupted torsion
films, lubricants, surface plastic deforma- nitrogen, chromium, nickel, molybdenum, sul- test, using 5083 aluminum alloy as the test ma-
tion, hard particles fur, manganese, and silicon lowered it. terial (Ref 76). In addition, the effect of soak-
The workability of cast pure chromium pro- ing time on the hot workability of the material
The hot workability of a large number of ma- duced by an induction-slag-melting process was is explored, where it is shown that the resist-
terials are compared and metallurgical mecha- investigated through the study of rolling and ance to deformation of the alloy when soaked
nisms that affect the limits are identified in Ref isothermal upsetting (Ref 75). The effect of for 24 hours is as much as 20% less than that
71. As the researchers write, plastic deformation rolling, which was performed to change the when soaked for 4 hours.
occurs by dislocation glide. Reduction in the re- coarse cast structure to a finer one, on the duc- Improvement of hot workability is shown to
sistance of the metals to deformation is observed tile-to-brittle transition temperature (DBTT) was depend upon the reduction of the denuded zone
when slip and climb mechanisms cause disloca- examined. The results are summarized as fol- around the dendritic cast structure and on the re-
tion annihilation. The restoration processes that lows. The DBTT of the as-cast specimen tested fining of the precipitated second phases that
resultrecovery and recrystallization, dynamic, in tension at a strain rate of 2.4 102 s1 was occur with increasing deformation. The flow and
static or metadynamicaffect ductility in a sig- approximately 500 K. A minimum ductility and fracture of MP35N (35Co, 35Ni, 20Cr, 10Mo)
nificant manner. High levels of dynamic recov- a maximum flow stress appeared at approxi- have been studied by uniaxial compression and
ery prevent triple-point cracking. At lower levels mately 973 K, owing to dynamic strain aging. plane strain bending in the temperature range
grain-boundary fissuring may be noted. Poor The forming limit in upsetting, the percent re- 1000 to 1200 C (1800 to 2200 F) and strain rate
ductility may also result in the prevention of dy- duction of height at the beginning of cracking on range 0.01 to 10 s1, which cover the normal
namic recrystallization or its retardation by low the specimen wall, increased to 80% at 773 K, bar rolling production conditions (Ref. 77). The
temperatures. When particles pin grain boundary and no crack appeared at temperatures over strain to fracture in plane strain bending was
movement, the metal may experience intergran- 873 K. At all temperatures, there occurred cracks found to increase with increasing strain rate,
ular fracture. The ductility troughs demonstrated in the as-cast specimen deformed by conven- roughly coinciding with the increase of the strain
by C-Mn, C-Mn-Al, C-Mn-Nb-Al, C-Mn-V-Al, tional rolling. Cracks occurred slightly in the to the peak stress in the flow curves. Within most
and C-Mn-Ti-Al steels, all of which fall in the specimen even in sandwich rolling by which a of the temperature and strain rate ranges investi-
range of temperatures used in the finishing sintered chromium was successfully deformed. gated, and under plane strain bending deforma-
stands of hot strip mills, are shown in Fig. 26. Cracking in the as-cast specimen was mainly due tion conditions, microvoid nucleation was found
The effect of minor elements on hot workability to the coarse structure with large crystal grains. to be concurrent with or greatly enhanced by the
are discussed in Ref 73. It was concluded that in Once the structure was changed to a finer one by onset of dynamic recrystallization. A small ad-
Nb-added two-phase stainless steels reduction upsetting, the specimen was easily rolled to a dition of boron improves hot workability of
of the phosphorus content and the additions of thin sheet without crack occurrence. The DBTT austenitic stainless steels (Ref 78), and it is ex-
yttrium and/or cerium improve ductility. Accord- of the specimen rolled to 80% and, followed by pected that boron should have the same effect on
ing to Ref 74, while the ductility of ferritic stain- annealing at 1273 K, was higher by approxi- ultra-high-strength metastable austenitic stainless
less steels is little dependent on the temperature, mately 200 K than that of the as-cast specimen, steels. Examination of hot workability and duc-
tility at room temperature using hot rolling, hot
tensile, notched tensile, and Erichsen tests
showed that the mechanical properties are related
to boron content. Hot workability is improved
with increasing boron addition. However, exces-
sive boron content has a deleterious effect, caus-
ing reduced room temperature ductility in steels
aged at various temperatures after cold rolling.
This significant loss of ductility is attributed
mainly to borides that precipitate after annealing.
Mathematical Models for

Workability Prediction
The physical phenomena of the loss of work-
ability may be translated into mathematical
terms. This section reviews some of these math-
ematical models. While some of these models
are applied directly to rolling, some address the
general problem of bulk forming. A general re-
view of prediction of damage in materials pro-
cessing is given in Ref 79.
The model of Montmitonnet has been referred
to (Ref 58) in the context of correlating crack
formation and the stress field. As mentioned, a
finite element program that uses a viscoplastic
material model and a steady state approach with
free surface updating by minimizing the material
flux through the surface was developed. Hot
Fig. 26 The ductility trough for several alloy steels. Source: Ref 72 workability is presented from the systems point
of view (Ref 80). Based on a polar reciprocity tal data available in the relevant literature are elastic-plastic finite strain model, based on irre-
relationship between the stress and strain rate used to verify these predictions. It can be con- versible thermodynamics. Their model accounts
spaces, a measure of hot workability, known as cluded that the new damage model can success- for the degradation of elastic properties, plastic
hardening power index, is defined. Deformation fully reveal the ductile damage evolution during dilatation caused by the growth of voids and
processing maps of this parameter are plotted for the plastic deformation process. Thus, it opens microcracks and strain softening. They success-
a titanium alloy IMI685 (Ti-6Al-5Zr-0.2Mo- the way to predict product quality and materials fully predicted the occurrence of defects in the
0.1Si) using flow data from high-temperature workability for metal-forming processes. In Ref process of extrusion. They emphasize the impor-
axial compression tests in the temperature range 84, plasticity theory is modified for compressi- tance of stress and strain history in addition to
800 to 1000 C (1470 to 1830 F) and strain rate ble materials and the upper bound theorem or materials parameters. Reference 88 discusses
range 103 to 102 s1. Maps plotted at different the second extremum principle, which incorpo- and compares damage models and underlines
strain values reveal the optimum processing con- rates a normal velocity discontinuity, considered two aspects of the fracture mechanisms in metal
ditions and indicate the probable occurrence of to be a measure of fracture. The theorem is ap- forming processes. When progressive ductile
instabilities. plied to predict the occurrence of fracture and damage development is the mechanism, material
Reference 81 describes a new method of central bursting in extrusion or drawing. The oc- models that include an account of voids or
preform design in multistage metal forming currence of fracture appears to depend on the cracks are needed. When fracture is occasioned
processes considering workability, limited by value of the parameters in the theorem; these by plastic instabilities, increasing void volume
ductile fracture. The FE simulation combined parameters may be related to the ductility or fraction is not an appropriate measure. The au-
with the ductile fracture criterion has been per- the workability of the material in question. The thor uses a perturbation technique for the detec-
formed in order to predict ductile fracture. The upper bound technique is then applied to multi- tion of instabilities. The authors of Ref 89 pres-
artificial neural network using the Taguchi stage extrusion of a carbon steel for which the ent an optimization technique for process design
method has been implemented for minimizing values of the parameters are given with some as- with damage minimization as the main objec-
objective functions relevant to the forming sumptions. It is thus shown that the stage at tive. Others (Ref 90) used the Oyane criterion
process. The combinations of design parameters which central bursting occurs agrees with exper- (Ref 91) and a three-dimensional FEM to predict
used in FE simulation are selected by an ortho- imental results and that it apparently depends on the occurrence of surface defects during hot
gonal array in the statistical design of experi- the extrusion condition. Reference 85 presents rolling of heavy ingots. The Oyane criterion,
ments. The orthogonal array and the result of a ductile-damage model based on the precise based on the critical void volume, is defined in
simulation are used as training data for artificial measurement of relative density changes. The terms of the Oyane parameter, B, given by:
neural networks. The cold heading process is rationality of the proposed model is verified e
sm
taken as an example of designing preforms,
which do not form any fracture in the finished
through observation of the change in the volume
fraction of voids during the tensile test, by the 1 + A s
0
de=B
eq
product. The results of analysis to validate the use of scanning electron microscopy. An exam-
proposed design method are presented. The use ple to demonstrate the application of the model where A and B are materials characteristics, de-
of ductile fracture criteria in conjunction with is given. Combining the proposed model with a pending on thermal conditions; and sm and seq
the FEM for predicting failures in cold bulk proper method of the analysis of metal-forming are the mean stress and the equivalent stress, re-
metal forming is described in Ref 82. processes, it is possible to predict the evolution spectively. The authors state that while the
Four previously published ductile fracture of ductile damage in the interior of the work- Cockroft and Latham criterion could not fit their
criteria are selected, and their relative accuracy piece and thus the workability of the material. data, the Oyane criterion did.
for predicting and quantifying fracture initia- Reference 86 examines the behavior of a lon-
tion sites is investigated. Experiments with ring, gitudinal V-shaped crack on the surface of a con- Process Design
cylindrical, tapered, and flanged upset samples tinuously cast steel slab during hot rolling. The
are performed to investigate the validity of the analysis is carried out by means of the commer-
In Ref 92, a simplified description is given of
workability criteria under conditions of stress cial FE-code LS-DYNA3D. Process parameters
the Taguchi methods (Ref 93) as applied to
and strain similar to those usually found in bulk obtained from industry are used as a reference.
quality engineering, a topic of which process de-
metal forming processes. The implementation of The slab of initial width of 1000 mm (39 in.) and
sign is an essential part. Off-line quality control
ductile fracture criteria into a rigid-plastic FE 220 mm (9 in.) thickness is rolled down to
consists of two stages: product design and process
computer program is also presented. Local stress 30 mm (1.2 in.). It is assumed that the material
design (Ref 93). Process design is, in fact, the
and strain distributions throughout the defor- can be treated as rigid-perfectly plastic and that
essence of manufacturing engineering (Ref 92).
mation are computed and compared with ex- the cracks do not propagate. The latter assump-
The process specifies the equipment and sets work
perimental measurements. In general, a good tion is in agreement with industrial observations
standards as well as workable specifications.
agreement is found. However, only two of these for a steel grade similar to that analyzed here.
When the rolling process, or in industry parl-
workability criteria have successfully predicted The aim of the study is to investigate the possi-
ance, the draft schedule, is designed, several
the location at which fracture initiates for all the bility of controlling the plastic deformation so
questions need to be answered. These questions
upset tests performed in this work. The paper that the cracks disappear or so that their deterio-
concern the three components of the rolling sys-
concludes with a discussion of the importance of rating effects are minimized. The analysis is fo-
tem: the mill, the workpiece, and their interface.
the critical damage at fracture which is to remain cused upon the influence of friction, roll radius,
As far as the rolling mill is concerned, these
independent of the technological processes. and rolling schedule on the change in the shape
questions may include:
In the model described in Ref 83, irreversible of a crack of initial depth 20 mm (0.8 in.) and a
thermodynamics and the orthogonal flow rule crack angle of 6. The reliability of the simula- Are the mill stands of appropriate stiffness,
are used in formulating a microdamage evolu- tions is checked by pilot-plant experiments providing satisfactory mill stretch?
tion equation, taking advantage of the specific using aluminum as the model material for steel. Are the drive spindles capable of transmit-
free energy and plastic potential with internal The results indicate that it is not possible to pre- ting the necessary torque?
variables. Furthermore, by analogy with the vent the bottom side surfaces of the crack from Are the driving motors of sufficient power?
microdamage evolution equation, a macrodam- coming into contact, especially not for small re- Is the roll crown satisfactory?
age evolution equation is derived according to duction-pass and small roll radii. The influence Is the speed of the mill sufficient?
the principle of minimum strength. Afterward, of friction was found to be marginal. Are the roll cooling facilities appropriate?
predictions are made for proportional loading In another study (Ref 87), researchers incor- Is the surface roughness of the work roll
and nonproportional loading, and the experimen- porated ductile damage mechanisms in their adequate?
Is lubricant/emulsion delivery suitable? (2190 and 2280 F), and the fact that not all ele- Cooling Tables. After exiting the last stand of
Are the furnaces adequate? ments are dissolved is simply accepted. the finishing train, the strip is cooled at a prede-
Time in the reheat furnace is another vari- termined rate to a final temperature before it is
To include the attributes of the metal to be
able of the process, which has been established coiled for shipment. The cooling process has a
rolled, the designer may need to know:
by long experience. The usual time slabs spend very significant effect on the final product prop-
What is the ductility of the metal, especially in the reheat furnace is in the order of 2 to 3 erties and there are a large number of designs in
the ductility vs. temperature curve? hours, which is believed to be sufficient for them use in the steel mills.
What is the constitutive relation of the metal? to reach a uniform temperature distribution. Coilers. Here the strips are coiled, allowed to
What are the metallurgical attributesi.e., Descaling and Rough Rolling Practice. In cool, and are prepared for shipment.
the hardening and recovery processesof the roughing mill the slabs thickness is reduced
the metal? from approximately 200 to 300 mm (8 to 12 in.)
When the interface is considered, the limits
to about 25 to 30 mm (0.98 to 1.2 in.) in several Thermomechanical Processing
steps. Descaling before and after the rough
concern three events:
rolling passes, which is necessary to avoid poor Of the items in the preceding section, one im-
The ability of the two metal surfacesthat surfaces, is accomplished by high-pressure water portant topic that has received the most attention
of the roll and of the rolled metalto suc- jets. These practices also have been used in steel of researchers is the development of the draft
cessfully entrain the lubricant or the emul- companies for some time, and they are not eas- schedule on the finishing stands of a hot strip
sion ily changed. The flying shears remove the dis- mill. Most of the advances in process design
After entrainment, how well the lubricant is torted edges from the rolled metal. originate in the realization that the attributes of
distributed in the contact zone Time on the Transfer Table or in the Coil the hot-rolled product are influenced by both the
The boundary conditions as a result of the Box. In older mills the bar waits to enter the fin- mechanical and the thermal treatment it receives
contact of the rolls and the metal; note that ishing train on the transfer table. In more rein the finishing train of the hot strip mill. This is
these usually refer to the coefficients of cently built mills, the coil box, placed between referred to as thermomechanical processing, or
friction and heat transfer at the contact the rougher and the finishing train and in which in the case of rolling, controlled rolling (see
zone the steel is coiled up on exit from the rougher and Chapter 17, Thermomechanical Processing by
uncoiled in order to enter the finishing stands, al- Controlled Rolling).
When the process of the rolling of shapes is de-
lows the development of some temperature uni- The design of thermal-mechanical treatment
signed, all of these considerations are of some
formity and leads to improved homogeneity of on the finishing train of hot strip mills has been
importance. Of prime significance, however, is
the finished product. The temperature of the considered in detail (Ref 94). A schematic of
the process of roll pass design, which, in spite
entry of the transfer bar to the first stand of the several possibilities is shown in Fig. 28. The fig-
of a large number of technical publications, is
finishing train is one of the most important pa- ure shows four different designs for the draft
most often an experience-driven, trial-and-error
rameters because it controls the metallurgical at- schedules. The first one indicates normal rolling
procedure.
tributes of the finished product. There are some of carbon steels, showing that both the roughing
compromises to be made in deciding the entry and the finishing processes are performed at
Example temperature. Values that are too high reduce the high temperatures, well above that required for
loads on the mill but may create a strip that is too recrystallization. In the second example, still for
As the preceding discussion indicates, these soft. Temperatures that are too low would pro- carbon steels, finish rolling is performed at sig-
phenomena are interdependent, making process duce higher strength and hardness but may over- nificantly lower temperatures, but they are still
design a difficult and crucial undertaking. A load the drive train and increase roll wear. within the dynamic recrystallization range. The
well-designed draft schedule for a hot strip mill, The Draft Schedule on the Finishing Train. third and the fourth schedules are for microal-
a schematic of which is shown in Fig. 27, in- This part of the hot strip mill contributes most to loyed steels, containing various amounts of nio-
cludes considerations of the following. the attributes of the product. The final geometry bium. In both cases, finish rolling is completed
The Temperature of the Reheat Furnace. of the strip, its surface quality, and its metallurgy at temperatures below that required for dynamic
While the objective of reheating the slab prior to and the attendant mechanical properties are de- recrystallization, taking advantage of the result-
rough rolling is to allow all alloying elements to termined by the reductions and their sequence ing elongated g grains.
enter into solid solution of the matrix, the usual on the five, six or seven stands of the finishing Thermal-mechanical working, in which the
practice is to use a preset temperature, arrived at train. While the speed of the passes has a major effects of hot working on the kinetics of recrys-
most often by a trial-and-error procedure. Some impact on the development of the properties of tallization are used to produce a fine, uniform
of the alloying elements dissolve at relatively the product, engineers may freely decide only microstructure, is industrys favored technique
low temperatures, while others (for example, ti- the speed of the first stand. The speeds of the toward creating the required attributes in the
tanium and vanadium) require heating in excess subsequent stands must follow the laws of mass rolled metal. The alloys that respond best are
of 1300 C (2370 F). The usual temperature of conservation, and if these are disobeyed, tearing among the family of microalloyed steels, most
the reheat furnace is between 1200 and 1250 C or cobble may occur. often containing strong carbide formers such
as niobium, vanadium, molybdenum, titanium,
boron or aluminum, singly or in combinations.
Of these, niobium is used most, and its principal
advantage is its contribution to strengthening
through precipitation (Ref 95). The other major
benefits are grain refinement and the retardation
of recovery and recrystallization of austenite.
Vanadium is often added as a second alloying el-
ement with the objective of providing increased
strength. In addition, the extent of retardation of
the recovery processes, with multiple additions
of alloys, is expected to be greater than the sum
of the individual components. For automotive
Fig. 27 Schematic of a hot strip mill applications, yield strengths of up to 700 MPa
New steels are being developed constantly

and information about their response to thermal-
mechanical treatment is always required by the
producers. A very informative approach is to
subject samples of the steel to compression
under closely controlled conditions in which
the rates of strain are kept constant during the
process. Study of the resulting true stress-strain
curves leads to conclusions that may, in turn,
lead to the efficient scheduling of strip mill
operations.
The Web site of the American Iron and Steel
Industry AISI includes up-to-date information
concerning the description and the processing of
recently developed steel alloys. As given in the
site, advanced high-strength steels (AHSS) in
use in the automotive industry include the dual
Fig. 28 Thermal-mechanical treatment. Source: Ref 94 phase steels (DP), the microstructure of which
includes ferrite and up to 20 and 70% volume
fraction of martensite. While the use of bainite
(120 Ksi) may be required, establishing the need rolling schedules, steels containing a large vari- helps to enhance the capability to resist stretch-
for bainitic microstructures, obtained through ety of microalloys, including niobium, of up to ing on a blanked edge, the ferrite phase leads
the control of transformation kinetics (Ref 95). 0.118% by weight were considered (Ref 98). It to high ductility and creates high work harden-
The technical literature contains a large number was found that in multistage plane strain coming rates that give the DP steels higher tensile
of publications that detail the effects of these el- pression, carried out at temperature ranges that strength than conventional steels. Further, trans-
ements on the hot, as well as the room tempera- varied from one that would produce complete formation induced plasticity (TRIP) steels are
ture, strength of the steels. A detailed examina- recrystallization to one that would cause pan- also used, the microstructure of which consist of
tion of the chemical compositions of the metals caking only, most of the reduction of the grains a ferrite matrix containing a dispersion of hard
tested, however, reveals that most of the work would occur in the first stage. In a later study second phasesmartensite and/or bainite in ad-
concerned steels with niobium contents under (Ref 99) steels with the niobium content up to dition to retained austenite in volume fractions
0.06%. 0.112% by weight were considered. Niobium greater than 5%. During deformation, the hard
Other researchers (Ref 96) worked with contents up to 0.16% were examined (Ref 100). second phases create a high work hardening
several niobium steels, containing up to The alloys were hot rolled in several passes at rate, while the retained austenite transforms to
1.03%, by mass, of the alloy. They listed the various temperatures, and it was observed that martensite, increasing the work hardening rate at
important features of thermal-mechanical pro- the roll separating force remained independent higher strain levels. The complex phase (CP)
cessing, which are the control of the initial of the niobium content in the first three passes. steels consist of a very fine microstructure of
austenite grain size and the volume of carbides Beyond a total reduction of 70% however, the ferrite and a higher volume fraction of hard
and nitrides in the solution; the mechanical loads increased with the amount of niobium, in phases, which are strengthened further by fine
treatment; the composition; and the interpass an apparent contradiction to the conclusions precipitates. In the martensitic (MART) steels,
time. Increasing the amount of niobium re- reached in Ref 96. Another researcher (Ref 101) the austenite that exists during hot rolling or an-
fined the austenite grains at all temperatures also indicated that there are beneficial effects to nealing is transformed almost entirely to marten-
up to 1200 C (2190 F). Beyond the stoichio- be gained from increased niobium content be- site during quenching on the run-out table or in
metric Nb:C ratio, slight coarsening of the yond 0.08% by weight. Reference 102 discusses the cooling section of the annealing line. All
grains was noted. Solute drag appeared to be the role of microalloying elements in the modi- AHSS are produced by controlling the cooling
the mechanism of retardation of the restora- fication of the softening behavior of the steels, rate from the austenite or austenite plus ferrite
tion processes at higher temperatures, while at mentioning the three possibilities: solute ef- phase, either on the runout table of the hot mill
lower temperatures, strain-induced precipita- fects, strain-induced precipitation, and a combi- (for hot-rolled products) or in the cooling sec-
tion of carbonitrides caused the delay. At the nation of both. Studying single alloy additions, tion of the continuous annealing furnace (con-
lower temperatures, the strengthening was it was concluded that the most likely effect is tinuously annealed or hot dip coated products).
caused by the precipitation of fine, planar the combination of the two mechanisms. One of Advanced high-strength cooling patterns and
Nb(CN), coherent with the ferrite matrix, the the steels tested contained 0.035% Nb. When resultant microstructures are schematically illus-
carbides having been in solution in the austen- deformed at 900 C (1652 F), strain-induced trated on the continuous cooling-transformation
ite grains during the reheating process. Ther- precipitation began in approximately 10 s. The diagram, available for examination on the AISI
momechanical treatment was by hot rolling at effect of high-temperature deformation on the Web site (www.steel.org). The cooling patterns
temperatures of 900 to 950 C (1650 to 1740 size distribution of Nb(CN) was examined (Ref are designed on the bases of mathematical mod-
F) and 1200 to 1250 C (2190 to 2280 F), 20 103). In addition, the researchers studied the els, which attempt to predict the structures and
to 50% reduction per pass, followed by quench- retardation of dynamic recrystallization due to properties resulting from the processing tech-
ing in ice brine. Rolling the 0.16% Nb steel at precipitation. The interaction of manganese nique. Recent research concentrated on the de-
950 C (1740 F) to 50% reduction caused with niobium and vanadium has been studied velopment of these models.
only partial recrystallization. (Ref 104, 105) with attention focused on dy- According to Ref 106, most important mod-
Using a 0.073% Nb steel, researchers (Ref namic precipitation. Increasing the manganese els, developed for all aspects of the hot rolling of
97) noted sharply increased retardation of dy- content from 1.25% by weight to 1.9% decreased steel, are those that attempt to predict the evolu-
namic recrystallization at lower rates of strain, the rate of precipitation of niobium carboni- tion of the microstructureincluding the prop-
caused by precipitation. At a strain rate of trides. While dynamic precipitation started ear- erties after hot rolling and cooling in addition to
13 s1, the addition of 0.14% V caused further liest in the single-alloy niobium steel, the addi- rolling load prediction. The models are now
retardation of the beginning of dynamic re- tion of other microalloying elements retarded used, offline, to design the process of hot rolling
crystallization. In an attempt to optimize plate the kinetics of precipitation. of steels. Several of these models were reviewed
and discussed (Ref 22). The models include tory. They also describe processes that create REFERENCES
those found in Ref 107 to 118. In these reports, ultra-fine ferrite grains either through strain-
various closed form equations are presented, de- induced transformation or by creating new plant 1. J.A. Schey, Workability in Rolling, Work-
scribing the processes of recrystallization and configurations. Various processes and their de- ability Testing Techniques, G.E. Dieter, Ed.,
grain growth. Their predictive capabilities under tails are listed in a comprehensive table, includ- ASM International, 1984, p 269314
laboratory and industrial conditions have also ing conventional and recrystallization controlled 2. W.L. Roberts, Cold Rolling of Steel,
been discussed and, in general, found reason- rolling, accelerated cooling, direct quenching, Marcel Dekker Inc., 1978
able; see for example, the experiments and cal- quenching and self-tempering, rod cooling, warm 3. W.L. Roberts, Hot Rolling of Steel, Marcel
culations of the size of the ferrite grains after forming, and intercritical rolling. They conclude Dekker Inc., 1983
hot rolling experiments described in Ref 118. by writing that thermomechanical processing 4. W.L. Roberts, Flat Rolling of Steel, Marcel
Niobium-carrying steels were hot rolled, cooled has not changed in a marked fashion over the Dekker Inc., 1978
in air, and the resulting ferrite grain sizes were past few decades and that most of the recent re- 5. V.B. Ginzburg, High-Quality Steel Rolling:
determined. As observed in the following table search was devoted to the development of math- Theory and Practice, Marcel Dekker, 1993
(reproduced from Ref 22), the comparison of the ematical models of the process. 6. Z. Wusatowski, Fundamentals of Rolling,
measurements and the predictions is good. Pergamon, Oxford, 1969
7. E.C. Larke, The Rolling of Strip, Sheet and
Conclusions Plate, 2nd ed., Chapman and Hall, 1965
Measured and calculated ferrite grain size
8. C.W. Starling, The Theory and Practice of
Center Surface While the limits of workability in the rolling Flat Rolling, University Press, London,
Test, C (F)
(Start temperature) Measured Calculated Measured Calculated process depend on a large number of materials 1962
and process parameters, the most important con- 9. A. Tselikov, Stress and Strain in Metal
1187 (2169) 7.3 8 5.5 6.8
1168 (2134) 8.4 8.2 6.4 6.7
tributor appears to be the geometrical parame- Rolling, MIR, Moscow, 1967
1163 (2125) 8.4 7.4 7.0 6.5 terthe shape factordefined as the ratio of 10. L.R. Underwood, The Rolling of Metals,
1209 (2208) 8.4 7.4 7.0 6.5 the average thickness of the rolled metal to the John Wiley & Sons, 1950
1169 (2136) 7.4 6.1 6.3 5.0 projected contact length of the pass. When the
1187 (2169) 4.8 6.1 3.3 4.6
11. I. Ya. Tarnovskii, A.A. Pozdeyev, and V.B.
1169 (2136) 6.3 7.7 3.6 6.0
shape factor is larger than unity, the distributions Lyashkov, Deformation of Metals during
of the field variables, i.e., the stresses, strains, Rolling, Pergamon, Oxford, 1965
strain rates, and temperatures, are highly nonho- 12. R. Hill, The Mathematical Theory of
The Symposium on Emerging Steel Pro- mogeneous and will result in flow localization. Plasticity, Oxford University Press, 1950
cessing Technologies, held in conjunction with The probability of reaching the limits of worka- 13. O. Hoffman and G. Sachs, Introduction to
the 43rd Mechanical Working and Steel Pro- bility is very high. This phenomenon, combined the Theory of Plasticity for Engineers,
cessing Conference (October, 2001), considered with a possible loss of ductility, is the likeliest McGraw-Hill, 1953
the application of these approaches to various cause of ductile fracture, whether in the form of 14. B. Avitzur, Metal Forming: Processes and
steel grades. Bake hardening and aging of TRIP edge cracking, alligatoring, or surface cracking. Analysis, McGraw-Hill, 1968
steels are considered in Ref 119. The develop- When the shape factor is much less than unity, 15. G.W. Rowe, Principles of Industrial
ment of texture in dual phase steels and of TRIP such as in the cold-rolling process or in the fin- Metalworking Processes, Edward Arnold,
steels were the topics of Ref 120 and 121, re- ishing stands of the hot strip mill, the most likely London, 1977
spectively. In each of these studies the effect cause of failure is probably the tensile cracking 16. W.A. Backofen, Deformation Processing,
of the thermal treatment on the resulting proper- of the edges of the rolled strip. In the rolling Addison-Wesley, 1972
ties was examined. Higher intercritical holding of shapes of non-rectangular cross sections, the 17. W.F. Hosford and R.M. Caddell, Metal
temperatures resulted in higher strength, higher definition of the shape factor is of little use. Still, Forming: Mechanics and Metallurgy,
elongation, and higher formability index. The failure is most likely caused by flow localization 2nd ed., Prentice Hall, Englewood Cliffs,
annealing cycles used in industrial conditions and the loss of ductility. 1993
were presented in Ref 122, and the best mechan- Mathematical models, attempting to predict 18. W. Johnson and P.B. Mellor, Plasticity
ical properties were obtained at an overaging the limits of workability, are also available in the for Mechanical Engineers, Van Nostrand,
temperature of 400 C (752 F). The effect of technical literature. It appears, however, that London, 1962
different prior austenite conditions on transfor- the Cockroft and Latham criterion (Ref 57) is 19. J. Lubliner, Plasticity Theory, Macmillan,
mation kinetics and on the resulting properties the model of choice because it is the one whose New York, 1990
of TRIP steels was examined in Ref 123. The predictions have been most often substantiated. 20. E.M. Mielnik, Metalworking Science and
thermal schedule was given and it was con- Research publications concerning process de- Engineering, McGraw-Hill, 1991
cluded that the volume fraction of the retained sign in rolling are few, probably because of the 21. R.H. Wagoner and J.-L. Chenot, Funda-
austenite is not strongly dependent on the re- proprietary nature of the processes. When the mentals of Metal Forming, John Wiley &
finement of ferrite. Further, the effect of the draft schedule for a particular steel has been de- Sons, 1996
retained austenite on the mechanical properties veloped and has been shown to be successful, 22. J.G. Lenard, M. Pietrzyk, and L. Cser,
was more significant than the ferrite grain re- the details are understandably guarded closely Mathematical and Physical Simulation of
finement. Strain-induced transformation was by the companies. Experimental simulation of the Properties of Hot Rolled Products,
used to create ultra-fine ferrite grains of 1 to the hot-rolling process of microalloyed steels Elsevier, 1999
3 mm size (Ref 124). The success in producing shows the success of thermal-mechanical treat- 23. D. Kobasa and R.A. Schultz, Experimental
these very small grains depended on the fine tun- ment in producing high-quality, high-strength Determination of the Length of the Arc of
ing of the thermal-mechanical process. The rate steels. Contact in Cold Rolling, Iron and Steel
of strain was found to be one of the most impor- Eng., 1968, p 16
tant parameters. ACKNOWLEDGMENT 24. J.R. Yow, Behavior of the Elastic En-
The state-of-the-art of thermomechanical trance Region in Cold Strip Rolling, Ph.D.
processing is discussed in Ref 125. The authors Substantial portions of this chapter are due dissertation, North Carolina State Univer-
divided the techniques into those that involve to Scheys original publication (Ref 1) on this sity, Raleigh, NC, 1970
control of the deformation process and those topic, and several portions of his text are repro- 25. E. Orowan, The Calculation of Roll Pres-
that control the post-deformation cooling his- duced, essentially unchanged. sure in Hot and Cold Flat Rolling, Proc.
Inst. Mech. Eng., Vol 150, 1943, p 140 Rolling, J. Inst. Met., Vol 92, 1964, p 254 ing and Steel Processing Conf. Proc., Vol
167 256 38, Iron and Steel Society, 2000, p
26. D.R. Bland and H. Ford, The Calculation 45. D.T. Blazevic, Tertiary Rolled in Scale 311320
of Roll Force and Torque in Cold Strip the Hot Strip Mill Problem of the 1990s, 61. A. Wang, P.F. Thomson, and P.D.
Rolling with Tensions, Proc. Inst. Mech. 37th MWSP Conf. Proc., ISS-AIME, Hodgson, A Study of Pore Closure and
Eng., Vol 159, 1948, p 144153 XXXIII, 1996, p 3338 Welding in Hot Rolling Process, J. Mater.
27. R.B. Sims, The Calculation of Roll Force 46. Y.H. Li and C.M. Sellars, Modelling Process. Technol., Vol 60, 1996, p 95102
and Torque in Hot Rolling Mills, Proc. Inst. Surface Temperatures during Hot Rolling 62. S. Turczyn and Z. Malinowski, Split Ends
Mech. Eng., Vol 168, 1954, p 191200 of Steel, Modelling of Metal Rolling and Central Burst Defects in Rolling,
28. L.S. Lim and J.G. Lenard, A Study of Processes 3, J.H. Beynon, M.T. Clark, Materials Processing Defects, S. K. Ghosh
Friction in Cold Strip Rolling, J. Eng. P. Ingham, P. Kern, K. Waterson, Ed: and M. Predeleanu, Ed., Elsevier, p 401416
Mater. Technol. (Trans. ASME), Vol 106, IOM Communications, London, 1999, 63. J.A. Schey, J. Inst. Met., Vol 94, 1966, p
1984, p 139146 178186 193200
29. A.N. Karagiozis and J.G. Lenard, Tem- 47. M. Krzyzanowski and J.H. Beynon, The 64. S. Turczyn, The Effect of the Roll-Gap
perature Distribution in a Slab During Hot Tensile Failure of Mild Steel Oxides under Shape Factor on Internal Defects in
Rolling, J. Eng. Mater. Technol. (Trans. Hot Rolling Conditions, Steel Res., Vol 70, Rolling, J. Mater. Process. Technol., Vol
ASME), Vol 110, 1988, p 1721 1999, p 2227 60, 1996, p 275282
30. B. Hum, H.W. Colquhoun, and J.G. 48. J.B. Tiley, J.G. Lenard, and Y. Yu, Roll Bite 65. W.C.F. Hessenberg and L. Bourne, BISRA
Lenard, Measurements of Friction During Deformation of the Thin Scale Layer on a Report MW/A/51/49, 1949
Hot Rolling of Aluminum Alloys, J. Mater. Plain Carbon Steel during Hot Rolling, 66. S. Sheu, L.G. Hector and O. Richmond,
Process. Technol., Vol 60, 1996, p 331338 42nd Mechanical Working and Steel Pro- Tool Surface Topographies for Controlling
31. L. Prandtl, Anwendungsbeispiele zu einem cessing Conf. (Toronto), Vol 38, Iron and Friction and Wear in Metal Forming
Henckyschen Satz ber das plastische Steel Society, 2000, p 215222 Processes, J. Tribology (Trans. ASME), Vol
Gleich-gewicht, ZAMM, Vol 3, 1926, p 49. A. Jones and B. Walker, Met. Technol., Vol 120, 1998, p 517527
401406 1, 1974, p 310315 67. P. Macura and J. Petruska, Numerical and
32. E. Orowan and K.J. Pascoe, A Simple 50. E.A. Leach, P. Gregory and R. Eborall, Experimental Simulation of Pass Rolling,
Method of Calculating Roll Pressure and J. Inst. Met., Vol 83, 195455, p 347353 J. Mater. Process. Technol., Vol 60, 1996,
Power Consumption in Hot Flat Rolling, 51. Y. Maehara, K. Nakai, K. Yasumoto, and p 5560
Special Report 34, Iron and Steel Institute, T. Mishima, Hot Cracking of Low Alloy 68. J.A. Charles, Hot Processability and the
1946, p 124 Steels in Simulated Continuous Casting Optimization of Hot-Worked Properties,
33. F. Krber and E. Siebel, ber die Direct Rolling Process, Trans. Iron Steel Hot Working and Forming Processes, C.M.
Beanspruchungsverhltnisse beim Schmie- Inst. Jpn., Vol 28, 1988, p 10211027 Sellars and G.J. Davies, Ed., The Metals
den und Walzen, Mitt. Kaiser-Wilh.-Inst. 52. W. Dahl, E. Wildschtz, and W. Schiffgen, Society, 1979, p 8798
Eisenforsch., Vol 10, 1928, p 1522 Arch. Eisenhttenwes, Vol 39, 1968, p 501 69. D.V. Wilson, Influences of Metallurgical
34. O. Pawelski, Stahl und Eisen, Vol 83, 509 Structure on Formability, Formability and
1963, p 14401451 53. F. Kasz and P.C. Varley, J. Inst. Met., Vol Metallurgical Structure, A.K. Sachdev and
35. F.A.A. Crane and J.M. Alexander, J. Inst. 76, 194950, p 423428 J.D.Embury, Ed., The Metallurgical
Met., Vol 96, 1968, p 289300 54. S. Kobayashi and S.I. Oh, Workability of Society, Warrendale, PA, 1987, p 332
36. P. Dewhurst, I.F. Collins, and W. Johnson, Aluminum Alloy 7075-T6 in Upsetting 70. J.D. Embury, Microstructural Aspects of
J. Mech. Eng. Sci., Vol 15, 1973, p 439 and Rolling, J. Eng. Ind. (Trans. ASME B), Formability, Formability and Metallurgi-
447 Vol 98, 1976, p 800806 cal Structure, A.K. Sachdev and J.D.
37. H. Ford and J.M. Alexander, Simplified 55. V.K. Vorontsov, A.I. Baganov, B.V. Embury, Ed., The Metallurgical Society,
Hot-Rolling Calculations, J. Inst. Met., Vol Kucheryaev, V.V. Brinza, and V.P. Gusev, Warrendale, PA, 1987, p 101116
92, 196364, p 397403 Study of Workability of AK4-1ch Alloy, 71. H.J. McQueen, and D.L. Bourell, Hot
38. J.A. Schey, Introduction to Manufacturing Tekhnol. Legk. Splavov, Vol 7, 1981, p 1519 Workability of Metals and Alloys, J. Met.
Processes, 3rd ed., McGraw-Hill, 2000 56. A L. Hoffmanner, Metal Forming: Inter- 1987, p 2835
39. P.M. Cook and A.W. McCrum, The Cal- relation Between Theory and Practice, 72. D.N. Crowther, Z. Mohamed, and B.
culation of Load and Torque in Hot, Flat Plenum, 1971 Mintz, Influence of Micro-Alloying Addi-
Rolling, British Iron and Steel Research 57. M.H. Cockroft, and D.J. Latham, Ductility tions on the Hot Ductility of Steels Heated
Association, 1958 and Workability of Metals, J. Inst. Met., Directly to the Test Temperature, Trans.
40. A. Murthy and J.G. Lenard, Statistical Vol 96, 1968, p 3339 Iron Steel Inst. Jpn., Vol 27, 1987, p 366
Evaluation of Some Hot Rolling Theories, 58. P. Montmitonnet, J.L. Chenot, C. Bertrand- 375
J. Eng. Mater. Technol., Vol 104, 1982, Corsini, C. David, T. Iung, and P. Buessler, 73. H. Kobayashi, S. Yamaguchi, and M.
p 4752 A Coupled Thermomechanical Approach Endo, Effect of Minor Elements on Hot
41. S. Shida, Effect of Carbon Content, for Hot Rolling by a 3D Finite Element Workability of Two-Phase Stainless Steel,
Temperature and Strain Rate on Com- Method, J. Eng. Ind. (Trans. ASME), Vol Hot Working and Forming Processes, C.M.
pression Flow Stress of Carbon Steels, 114, 1992, p 336344 Sellars and G.J. Davies, Ed., The Metals
Hitachi Research Report, 1974, p 19 59. J.-H. Kwak, J.-H. Chung, and K.M. Cho, Society, 1979, 133139
42. J.A. Schey, Fracture in Rolling Processes, Optimum Mn/S Ratio and Heat Treatment 74. S.R. Keown, Hot Ductility of Wrought
J. Appl. Metalwork., Vol 1, 1980, p 4859 of Low Carbon Steels in Mini-Mill Process Austenitic Stainless Steels, Hot Working
43. D. Duly, J.G. Lenard, and J.A. Schey, to Prevent Edge Cracking, J. Korean Inst. and Forming Processes, C.M. Sellars and
Applicability of Indentation Tests to Assess Met. Mater. (South Korea), Vol 38, 2000, G.J. Davies, Ed., The Metals Society,
Ductility in the Hot Rolling of Aluminium p 213218 1979, 140147
Alloys, J. Mater. Process. Technol., Vol 75, 60. J.-H. Kwak, J.-H. Chung, and K.M. Cho, 75. Y. Harada, M. Ohmori, and S.-I. Ohnishi,
1998, p 143151 A Study on the Edge Cracking of Low Workability in Rolling and Upsetting of
44. Discussion: Edge Cracking, Lamination Carbon Steel Sheets Manufactured by Cast Chromium, J. Jpn. Inst. Met., Vol 54,
and Surface Cracking in Hot and Cold Mini-Mill Process, 42nd Mechanical Work- 1990, p 473479
76. M. Ueki, S. Horie, and T. Nakamura, Guerlet, Surface Defects in Hot Rolling of cessing, Proc. THERMEC 97, T. Chandra,
Simulation of the Hot Working of 5083 Copper Based Alloys Heavy Ingots: Three and T. Sakai, Ed., Wollongong, 1997, p
Aluminum Alloy By Means of the Torsion Dimensional Simulation Including Failure 311
Tests, J. Mech. Work. Technol., Vol 11, Criterion, Computational Methods for 109. A. Laasraoui and J.J. Jonas, Prediction of
1985, p 365376 Predicting Material Processing Defects, Temperature Distribution, Flow Stress and
77. G. Fitzsimons and H.A. Kuhn, Flow and M. Predeleanu, Ed., Elsevier, 1987, p 29 Microstructure during the Multipass Hot
Fracture of a Multiphase Alloy MP35N for 38 Rolling of Steel Plate and Strip, ISIJ Int.,
Study of Workability, Metall. Trans., Vol 91. M. Oyane, T. Sato, K. Okimoto, and S. Vol 31, 1991, p 95105
15A, 1984, p 18371848 Shima, Criteria for Ductile Fracture and 110. A. Laasraoui and J.J. Jonas, Recrystalliza-
78. K. Tomimura and K. Miyakusu, Effect of Their Applications, J. Mech. Work. Technol., tion of Austenite after Deformation at
Boron on Hot Workability and Ductility in Vol 4, 1980, p 6581 High Temperature and Strain Rates
Ultra-High-Strength Metastable Austenitic 92. M.P. Groover, Fundamentals of Modern Analysis and Modelling, Metall. Trans. A,
Stainless Steels, Nisshin Steel Tech. Rep. Manufacturing, 2nd ed., John Wiley and Vol 22A, 1991, 151160
(Japan), Vol 80, 2000, p 1016 Sons, 2002 111. P. Choquet, P. Fabregue, J. Giusti, B.
79. M. Predeleanu, Damage Mechanics 93. G. Taguchi, E.A. Elsayed, and T.C. Hsiang, Chamont, J.N. Pezant, and F. Blanchet,
Approach for Predicting Material Process- Quality Engineering in Production Systems, Modelling of Forces, Structure and Final
ing Defects, Surface Treatment V: Compu- McGraw Hill, 1989 Properties during the Hot Rolling Process
ter Methods and Experimental Measure- 94. I. Tamura, H. Sekine, T. Tanaka, and C. on the Hot Strip Mill, Proc. Mathematical
ments (USA), 2001, p 129139 Ouchi, Thermomechanical Processing of Modelling of Hot Rolling of Steel, S. Yue,
80. T. Rajagopalachary and V. V. Kutumbarao, High-Strength Low-alloy Steels, Butter- Ed., Hamilton, 1990, p 3443
Model for the Evaluation of Hot Work- worth and Co., 1988 112. P.D. Hodgson, and R.K. Gibbs, A Mathe-
ability of Metallic Materials. J. Mater. 95. G. Tither, Proc. Int. Conf. on HSLA Steels: matical Model to Predict the Mechanical
Process. Technol., Vol 24, 1990, p 333 Processing, Properties and Applications, Properties of Hot Rolled C-Mn and Micro-
342 G. Tither and S. Zhang, Ed., The Minerals, alloyed Steels, ISIJ Int., Vol 32, 1992,
81. D.-C. Ko, D.-H. Kim, and B.-M. Kim, Metals, and Materials Society, Warrendale, p 13291338.
Application of Artificial Neural Network PA, 1992, p 6168 113. H. Yada, Prediction of Microstructural
and Taguchi Method to Preform Design in 96. R.K. Amin and F.B. Pickering, Proc. Int. Changes and Mechanical Properties in Hot
Metal Forming Considering Workability, Conf. on Thermomechanical Processing Strip Rolling, Proc. Symp. Accelerated
Int. J. Mach. Tools Manuf., Vol 39, 1999, of Microalloyed Austenite, A.J. DeArdo, Cooling of Rolled Steel, G.E. Ruddle and
p 771785 G.A. Ratz, and P.J. Wray, Ed., AIME, A.F. Crawley, Ed., Pergamon Press,
82. B.P.P.A. Gouveia, J.M.C. Rodrigues, and Pittsburgh, PA, 1981, p 133 Winnipeg, 1992, p 105119
P.A.F. Martins, Fracture Predicting in Bulk 97. I. Weiss, G.L. Fitzsimons, K. Mielityinen- 114. J.H. Beynon, and C.M. Sellars, Modelling
Metal Forming, Int. J. Mech. Sci., Vol 38, Tiitto, and A.J. DeArdo, Proc. Int. Conf. Microstructure and Its Effects during
1996, p 361372 on Thermomechanical Processing of Micro- Multipass Hot Rolling, ISIJ Int., Vol 32,
83. M. Zheng, Z.J. Luo, and X. Zheng, alloyed Austenite, A.J. DeArdo, G.A. Ratz, 1992, p 359367
New Damage Model for Ductile Materi- and P.J. Wray, Ed., AIME, Pittsburgh, PA, 115. T. Sakai, Dynamic Recrystallization Micro-
als, Eng. Fract. Mech., Vol 41, 1992, 1981, p 3359 structures under Hot Working Conditions,
p 103110 98. L.J. Cuddy, Metall. Trans. A, Vol 12A, J. Mater. Process. Technol., Vol 53, 1995,
84. S. Shima and Y. Nose, Development of 1981, p 13131320 p 349361.
Upper Bound Technique for Analysis of 99. L.J. Cuddy, Proc. Int. Conf. on Thermo- 116. R. Kuziak, Y.-W. Cheng, M. Glowacki, and
Fracture in Metal Forming, Ingenieur- mechanical Processing of Microalloyed M. Pietrzyk, Modelling of the Micro-
Arch., Vol 60, 1990, p 311322 Austenite, A.J. DeArdo, G.A. Ratz, and P.J. structure and Mechanical Properties of
85. Z.J. Luo, W.H. Ji, N.C. Guo, X.Y. Xu, Q.S. Wray, Ed., AIME, Pittsburgh, PA, 1981, Steels during Thermomechanical Process-
Xu, and Y.Y. Zhang, Ductile-Damage p 129140 ing, NIST Technical Note 1393, Boulder,
Model and Its Application to Metal- 100. Y.C. Hirsch and B.A. Parker, Proc. Int. 1997
Forming Processes, J. Mater. Process. Conf. on Advances in the Physical Met- 117. C. Devadas, I.V. Samarasekera, and E.B.
Technol., Vol 30, 1992, p 3143 allurgy and Application of Steels, The Hawbolt, The Thermal and Metallurgical
86. E. Ervasti and U. Stahlberg, Behavior of Metals Society, Liverpool, 1981, p 26 State of Steel Strip during Hot Rolling,
Longitudinal Surface Cracks in the Hot 101. P.E. Repas, Proc. Int. Conf. on HSLA Part III: Microstructural Evolution, Metall.
Rolling of Steel Slabs, J. Mater. Process. SteelsTechnology and Applications, Trans., Vol 22A, 1991, p 335349
Technol., Vol 94, 1999, p 141150 (Philadelphia, PA), ASM International, 118. M. Pietrzyk, C. Roucoules, and P.D.
87. J.C. Gelin and M. Predeleanu, Finite Strain 1984, p 203208 Hodgson, Modelling the Thermomechanical
Elasto-Plasticity Including Damage 102. H.L. Andrade, M.G. Akben, and J.J. Jonas, and Microstructural Evolution during Roll-
Application to Metal Forming Problems, Metall. Trans. A, Vol 14A, 1983, p 1967 ing of a Nb HSLA Steel, ISIJ Int., Vol 35,
NUMIFORM 89, E.G. Thompson, R.D. 103. I. Weiss and J.J. Jonas Metall. Trans. A, 1995, p 531541
Wood, O. Zienkiewicz, and A. Samuelsson, Vol 11A, 1980, p 403 119. M. De Meyer, A.K. De, L. Tosal-
Ed., A.A. Balkema, 1989, p 151157 104. M.G. Akben, I. Weiss, and J.J. Jonas Acta Martinez, and B.C. De Cooman, The Bake
88. J.C. Gelin, Modelling of Damage in Metal Metall., Vol 29, 1981, p 111 Hardening and Ageing Behaviour of Cold
Forming Processes, J. Mater. Process. 105. M.G. Akben, B. Bacroix, and J.J. Jonas Rolled TRIP Steels, 43rd Mechanical
Technol., Vol 8081, 1998, p 2432 Acta Metall., Vol. 31, 1983, p 161 Working and Steel Processing Conf. Proc.,
89. P. Picart, O. Ghouati, and J.C. Gelin, 106. S. Yue, private communication, 2001 Vol 39, Iron and Steel Society, 2001,
Optimization of Metal Forming Process 107. C.M. Sellars, Modelling Microstructural p 349358
Parameters with Damage Minimization, Development During Hot Rolling, Mater. 120. T. Waterschoot, L. Kestens, S. Vandeputte,
J. Mater. Process. Technol., Vol 8081, Sci. Technol., Vol 6, 1990, p 10721081 and B.C. De Cooman, Development of the
1998, p 597601 108. C.M. Sellars, An Internal State Variable Texture in MnCr Hot Rolled Dual Phase
90. O. Branswyck, C. David, C. Levaillant, Approach to Modelling Microstructural Steel: Influence on the r-Value, 43rd
J.L. Chenot, J.P. Billard, D. Weber, and J.P. Evolution during Thermomechanical Pro- Mechanical Working and Steel Processing
Conf. Proc., Vol 39, Iron and Steel Society, ical Working and Steel Processing Conf. Grain Size Through the Spontaneous
2001, p 373383 Proc., Vol 39, Iron and Steel Society, 2001, Reverse Transformation Due to Adiabatic
121. M. De Meyer, L. Kestens, L. Tosal- p 411433. Deformation Heating, 43rd Mechanical
Martinez, and B.C. De Cooman, Texture 123. S. Jiao, F. Hassani, R.L. Donaberger, E. Working and Steel Processing Conf. Proc.,
Development in Intercritically Annealed Essadiqi, and S. Yue, The Effect of the Prior Vol 39, Iron and Steel Society, 2001, p
TRIP Steels, 43rd Mechanical Working and Heat Treatment on a Cold Rolled and 479488
Steel Processing Conf. Proc., Vol 39, Iron Annealed Mo-Nb Microalloyed TRIP Steel, 125. P.D. Hodgson and M.R. Barnett, The
and Seel Society, 2001, p. 333347 43rd Mechanical Working and Steel Thermomechanical Processing of Steels
122. A. Pichler, S. Traint, H. Pauli, H. Mildner, Processing Conf. Proc., Vol 39, Iron and Current Status and Future Trends, Metal
J. Szynyur, M. Blaimschein, P. Stiaszny, Steel Society, 2001, p 385395 Forming 2000, M. Pietrzyk, J. Kusiak, J.
and E. Werner, Processing and Properties 124. T. Yokota, K. Sato, M. Miyake, and M. Majta, P. Hartley, and I. Pillinger, Ed.,
of Cold Rolled TRIP Steels, 43rd Mechan- Niikura, Ultra Refinement in Austenite Balkema, 2000, p 2128
Chapter 19
Drawing of Wire, Rod, and Tube*
IN THE DRAWING PROCESS, the cross- filaments. Annealing might occasionally be nec- may break down, and the properties of the drawn
sectional area and/or the shape of a rod, bar, essary after several drawing passes before the product may suffer.
tube, or wire is reduced by pulling through a die. drawing operation is continued.
One of the oldest metalforming operations, Wire and bar drawing are basically the same.
drawing allows excellent surface finishes and In general, drawing is a steady-state process from Drawing of Bar and Wire
closely controlled dimensions to be obtained in both the mechanical and tribological viewpoints.
long products that have constant cross sections. The workpiece is long and is continuously drawn Significant quantities of bar and wire materi-
The deformation is accomplished by a combina- into the die. Even though non-steady-state condi- als are produced by drawing. In wire or rod
tion of tensile and compressive stresses that are tions exist during acceleration, they are signifi- drawing, the section is usually round but could
created by the pulling force at the exit from the cant in only a few instances. Unless the lubricant also be a shape. In the cold drawing of shapes,
die and by the die configuration (Fig. 1). breaks down, forces and temperatures attain the basic contour of the incoming shape is es-
In drawing, a previously rolled, extruded, or steady levels. Proper lubrication is essential in tablished by cold-rolling passes that are usually
fabricated product with a solid or hollow cross rod, tube, and wire drawing. In contrast to preceded by annealing. After rolling, the section
section is pulled through a die, sometimes at exit rolling, little or no friction is needed for wire shape is refined and reduced to close tolerances
speeds as high as several thousand feet per drawing, tube sinking, and tube drawing on a by cold drawing. Again, several steps may be
minute. The die geometry determines the final fixed plug. However, some minimum friction is necessary to eliminate the effects of strain hard-
dimensions, the cross-sectional area of the essential for drawing with a floating plug, and ening, that is, to reduce the flow stress and to in-
drawn product, and the reduction in area. The friction is helpful on the tube/bar interface in crease the ductility.
purpose of drawing might be simply to tighten drawing on a bar. Therefore, if at all possible, the A typical carbide drawing die is shown in Fig.
tolerances, improve surface finish, or increase lubricant is chosen to give lowest friction and 2. During drawing, the workpiece is pulled
the strength of the product; in this case, a single minimum wear. It is essential, though, that the through a stationary die with a converging chan-
draw (pass) at room temperature might suffice. heat generated be extracted, especially in high- nel, where reduction of the cross-sectional-area
More frequently, the end product is of a smaller speed drawing; if this is not done, the lubricant from Ao to A1 takes place. The wire or rod makes
cross section than can be produced by hot work-
ing. In this case, a sequence of passes may be
employed, usually at cold working temperatures,
with interpass process anneals if required.
Occasionally, material of limited ductility is
drawn warm or hot. Drawing is usually con-
ducted at room temperature using a sequence of
passes or reductions through consecutively lo-
cated dies. An important exception is the warm
drawing of tungsten to make incandescent lamp
Fig. 2 Cross section of a typical carbide die for wire drawing, 5.5 mm (0.218 in.) diam rod to 4.6 mm (0.180 in.) diam
Fig. 1 The drawing process. Source: Ref 1, p 345 wire (17% reduction per pass)
*Portions adapted from Tribology in Metalworking: Friction, Lubrication and Wear by John Schey, American Society for Metals book, 1983, pages 343393 (Ref 1) and from Drawing of Rod
and Wire, by Roger N. Wright, Encyclopedia of Materials Science and Engineering, Volume 2, M.B. Bever, Ed., Pergamon Press and The MIT Press, 1986, pages 12271231 (Ref 2).
Chapter 19: Drawing of Wire, Rod, and Tube / 279
contact in the drawing cone along the approach increasing friction and height-to-length ratio
angle and is reduced to the dimensions of the (h/L), of approximately 55% under the best con-
drawing cone exit. To prevent sudden decom- ditions. However, no wire can be perfect, and
pression of lubricant film and to avoid a rapid in- drawing with critical reductions would lead to
crease in diameter on die wear, a parallel land frequent wire breaks. Therefore, practical reduc-
(bearing) is always provided. The bearing region tions are limited to 20 or 30% per pass and
involves no further reduction and allows the die rarely greater than 30 to 35%. A particularly
to be refinished without a change in the exit di- common reduction is that of an American Wire
mensions of the drawing cone. The back relief Gauge of 1, or about 20.7%. Multiple reductions
ensures a gradual drop in stresses in the die and or drawing passes are needed to achieve a large
reduces the amount of abrasion that takes place overall reduction, or larger reductions can be
if the drawing stops or if the die is out of align- achieved in a single operation with extrusion.
ment. Lubricant is introduced at the bell portion Alternatively, drawing can be used to generate
of the die and is pulled into the die/wire inter- larger quantities of small-diameter product (for
face by the moving wire. example, 0.01 mm, or 0.0004 in.) with excellent
Reduction is defined as the engineering com- dimensional control (assuming proper die main-
pressive strain tenance).
The force required for drawing can be calcu-
A0 - A1 lated by a variety of methods. By analogy, all
r= solutions can be brought to the general form:
A0 (Eq 1) Fig. 3 Components of draw force and optimal die half
angle (for 20% reduction in area). Source: Ref 1,
p 346
or t h
p = s 0 f e; i ; = s 0 Q
a L (Eq 4) sile stresses in the less-deformed parts of the
A - A1 workpiece. If these secondary tensile stresses
r= 0 100(%)
reach high values, a workpiece material of low
A0 where s 0 is the mean flow stress, and p is the ap-
ductility may suffer internal fracture (called an
plied drawing pressure. The bracketed term
For purposes of calculation, the true strain e is (pressure-multiplying factor Q) is a function
arrowhead defect). Even if there is no failure dur-
preferred: ing deformation, residual stresses may remain in
with three factors that define the three types of
the workpiece, which can lead to subsequent dis-
work performed by the drawing force. The first
tortion or to delayed failure due to stress corro-
A0 term is the portion of the drawing force that does
e = ln sion. Any inhomogeneously deformed workpiece
A1 (Eq 2) work by pure (homogeneous) deformation of the
also undergoes differential strain hardening and,
worked material. Pure deformation work occurs
if subsequently annealed, will show large grain-
Because the volume of the wire must remain from the homogeneous strain (e), which is di-
size and property variations.
constant, the wire elongates. If drawn at a speed rectly proportional to the flow stress s0 (or, in
In many processes, friction effects and inho-
v1, it enters the die at a lower v0 speed, which cold working, the mean flow stress, s 0) of the
mogeneity induced by process geometry occur
can be calculated from constancy of volume: wire material.
simultaneously, sometimes reinforcing and
The other two terms in the bracketed function
sometimes opposing each other. For a given re-
of Eq 4 are the force components that overcome
v0 A0 v1 A1 (Eq 3) duction in wire drawing, the friction contribu-
friction and expend work in the form of inhomo-
tion decreases and the contribution of redundant
geneous deformation (known as redundant
However, in contrast to rolling, there is no work increases with die angle. Therefore, an op-
work). Friction and redundant work, which are
neutral plane here, and the wire always slides timal half angle can be found, as discussed fur-
always additional to the work required for pure
over the whole contact zone. To a close enough ther in the next section on approach angle. The
deformation, depend on the die geometry and
approximation, it can be assumed that the inter- important point is that the effects of friction can-
work interface between the die and drawn mate-
face sliding velocity increases from entry to exit not be artificially separated from the effects of
rial. The contribution of friction increases with
linearly in plane-strain drawing and paraboli- inhomogeneous deformation or redundant work.
increasing interface shear stress (ti) and, for a
cally in axially symmetrical drawing in a conical Friction effects and redundant work are de-
given reduction, decreases with increasing die
die. Sliding generates a frictional stress oppos- scribed in more detail below, but a simple and
half angle a (Fig. 3) because of a decrease in
ing the movement of the wire (Fig. 1), and the useful formula for the drawing pressure is:
sliding length. The interface is usually described
contribution of this stress to the draw force is of by a constant coefficient of friction (m) even
immediate concern. Deformation occurs under though evidence indicates that friction varies p s 0 Q s0 (1 m cot a)fe (Eq 5)
the combination of the longitudinal (draw) stress along the contact surface with the die. Friction on
and the indirect compressive stress generated in where f allows for inhomogeneous deformation.
the die land also contributes to the draw force.
the die. Friction is equivalent to a back tension Its value can be taken, on the basis of experi-
The contribution of redundant work is the ad-
and thus lowers the interface pressure, but at the ments, for round wire as:
ditional work expended due to inhomogeneous
expense of higher draw forces. deformation. As in other processes, the amount
h
of redundant work increases with an increasing f = 0.88 + 0.2
Stresses and Forces (Ref 2) h/L ratio. If the contact length (L) is smaller than L (Eq 6a)
the mean thickness or diameter, deformation will
The pulling force, or drawing stress, cannot be inhomogeneous, as the center would deform and for plane-strain drawing as:
exceed the strength of the wire or rod being less than would be expected. Because deforma-
drawn (otherwise, fracture or unstable deformation proceeds against the restraining effect of h
f = 0.8 + 0.2
tion would occur). In fact, practical considera- the less-deformed center, deformation stresses, L (Eq 6b)
tions often limit the drawing stress to about 60% forces, and power requirements increase.
of the as-drawn flow stress. This sets a limit on Usually, a more important consequence of inho- There is no general agreement on an appropriate
the attainable reduction, which decreases with mogeneous deformation is the generation of ten- formula; the above formulas can give underesti-
mates, and other experiments give better agree- m

1/ 2 studies suggest that the redundant work factor F
ment with different formulas. D min @ 4.9 can be estimated to be:
For a given workpiece material and process ln(1 / 1 - r ) (Eq 8)
geometry, a larger draw force generally indicates F D/6 1 (Eq 10)
higher friction. As in rolling, a back pull (back where m is the coefficient of friction between the
tension) can be applied to the wire. The resultant wire and the drawing cone. The drawing stress The effects of redundant work are considered in
drop in interface pressure can be beneficial in sd can be usefully approximated as: great detail by Blazynski (Ref 3) and Avitzur
terms of the lubrication mechanism and wear, (Ref 4). Inhomogeneous deformation also has
but at the expense of greater draw force. several other consequences besides the effect on
3.2
The approach angle is perhaps the most im- s d = s 0 (a + m )
D + 0.9 (Eq 9) the drawing pressure. These include differential
portant feature of the die for most drawing ap- strain hardening, periodic centerbursts, and bulge
plications. As previously noted, an optimal half formation.
angle also can be found, because the frictional where s 0 is the yield or flow stress of the wire
component of Eq 4 decreases with an increase during the drawing pass.
of the die angle, while the amount of redundant Determination of Friction. In principle, the Heat Generation
work increases with an increase of the die magnitude of friction should be obtainable by
The management of heat is of great concern in
angle. The optimal half angle is usually be- various techniques. In practice, however, there
drawing. Much heat is generated directly by the
tween 5 and 7, where draw force is minimum. are several problems. Because the contributions
plastic deformation, and this heat is only par-
Lower angles are permissible only with very of friction and inhomogeneous deformation to
tially removed by interpass cooling; practical
low friction. With an optimal die angle, the con- the total force are of similar magnitude when an
cold-drawing operations can involve wire-tem-
tributions of friction and inhomogeneous defor- optimal approach angle is used, back-calculation
perature increases of a few hundred degrees
mation are of the same order of magnitude. of the magnitude of friction from the total draw
Kelvin. Most of the total work expended in wire
With increasing reduction, die pressures drop, force is difficult. Simultaneous measurement of
drawing is converted into heat. The die is of
deformation becomes more homogeneous, and draw force and normal force requires more so-
small mass and is in continuous contact with the
the efficiency of the drawing process (the ratio phisticated equipment and instrumentation, but
wire, and thus heating presents an even greater
of total work to pure deformation work) in- gives a direct method of measuring frictional
danger than in rolling. The dies extract little heat
creases. The optimal angle shifts to higher val- forces. (Ref 1, p 347).
under commercial conditions and become very
ues. Because of this, the trumpet-shape dies The interface shear strength cannot be deter-
hot.
used in the past actually gave a better approxi- mined from the total frictional force in drawing
At high drawing speeds, most of the heat is re-
mation of the optimal die profile. over a plug or a bar because there is no assur-
tained in the wire; still, the die also heats up, and
The approach angle also cannot easily be con- ance that md and mp (or md and mb) are equal.
in multihole drawing the entry temperature of
sidered independent of section reduction and Measurement of die pressure presents the same
the wire also rises. The surface temperature of
metal flow. In drawing theory, a D parameter is difficulties as in bar drawing, although strain
the wire is higher because of concentration of re-
used: gages have been successfully applied to the die
dundant work and frictional heating. From the
circumference. Much work has, therefore, been
tribochemical view point, peak temperatures are
D (a/r)[1 (1 r)1/2]2 (Eq 7) conducted with plane-strain drawing simula-
important. On emergence of the wire from the
tion (see Ref 1, p 204). Plane-strain drawing
die, temperatures rapidly equalize, and the main
where a is the approach semiangle (one-half the simulation tests are designed to evaluate spe-
concern is the overall temperature level.
included angle) in radians and r is the fractional cific tribological aspects of deformation
Therefore, cooling becomes an important func-
drawing reduction, given by Eq 1 as r 1 processes such as rolling, wire drawing, and
tion of the lubricant.
A1/A0,where A0 and A1 are the starting and fin- sheet drawing. The incentive for developing
Under adiabatic conditions, the temperature
ishing cross-sectional areas, respectively. these test methods is the desire to have better
increase (DTd) associated with plastic deforma-
Commercial die design often involves approach control over process variables, and tests are
tion in a single pass is approximately:
semiangles in the range of 6 to 10 and drawing usually designed to facilitate quantitative meas-
reductions of about 20%. The corresponding D urements of friction and wear. Their common
Fs 0 ln[1 / (1 - r )]
values typically range from 2 to 3, with higher feature is that interface shear stress (ti) (or m) DTd =
values corresponding to lower reductions and can be obtained without knowing flow stress Cr (Eq 11)
higher die angles, and lower values correspon- (sf) and without recourse to theory. A common
ding to higher reductions and lower die angles. feature of these tests is that new surfaces are where C and r are the heat capacity and density
For simple compression, D h/l, but at an generated by means of deforming the specimen of the wire, respectively. Additional heat gener-
oblique angle, then D reduces to Eq 7. through its entire thickness. ation is associated with frictional work. This
Effect of Friction. Basically, low D values Redundant Work of Shear Deformation. heat is concentrated at the die/wire interface and
may involve excessive frictional work between Redundant work refers to the additional energy can lead to diminished lubrication, further heat-
the wire and the drawing cone, and high D values expended due to inhomogeneous (shear) defor- ing, and catastrophic lubricant breakdown.
involve redundant work or plastic strain beyond mation, in addition to the work expended for Accompanying problems include poor wire-
that calculable from the reduction in area of the pure deformation and friction. As previously surface quality and metallurgical changes near
pass. Some degree of redundant work exists for D noted, redundant work increases with an in- the wire surface. If the coefficient of friction is
> 1, with redundant work increasing as D increasing h/L ratio. The ratio can be calculated not influenced by D, frictional heating is aggra-
creases, much as frictional work can increase as from various formulas, but, as shown in Fig. 1, it vated by low D processing. Fortunately, there is
D decreases. The net effect is that some interme- can be simply taken as the mean diameter (or, in a tendency for low approach angles (and thus
diate value of D involves the minimum work and plane-strain drawing, the mean thickness) of the low D) to foster hydrodynamic lubrication and a
therefore the minimum drawing force, because workpiece divided by the contact length L. For a reduced coefficient of friction.
the drawing force multiplied by the drawing ve- given reduction, the ratio increases with increas-
locity is the work consumed per unit time. ing a (Fig. 3). Redundant work is also expressed Flow through Conical Dies
Similarly, the drawing stress equals the work per in terms of the redundant work factor or the ratio
unit volume of wire drawn. The D for minimum of total plastic deformation work to the work im- Relative Drawing Stress and Power
drawing stress can be approximated by: plied by dimensional change. Experimental Consumption (Ref 5). Wire drawing can be
modeled in terms of relative drawing stress and distribution also involves some other factors, as
2 a
the relative amount of power consumed by inter- Ws = 2 - cot a described in this section.
nal deformation, redundant work, and friction. 3 sin a (Eq 14)
Residual Stresses. The first consequence of
The symbols used in this formulation are shown inhomogeneous strain is that residual stresses
in Fig. 4 for wire drawing through a converging The relative portion of power consumed by fric- are set up in the wire. At low h/L ratios the sur-
die where section size is reduced from R0 to Rf. tion (Wf ) along the conical surface of the die is: face residual stress is tensile. For a given die
As it passes through the die, the wire rubs angle a, stresses decrease with increasing pass
against the conical and cylindrical surfaces of 2 R reduction and improved lubrication, both of
the die and encounters friction resistance. The W f = m (cot a )ln 0 which serve to make deformation more homoge-
3 Rf (Eq 15)
characteristics of the die and the flow patterns in neous. Consequently, the service properties of
Fig. 4 are common to wire drawing, open-die ex- the wire also change. In a bar of large enough di-
trusion, and hydrostatic extrusion. where m is an interface friction factor (see Eq 50 ameter, inhomogeneity is also reflected in hard-
Under steady-state conditions, a typical solu- in Chapter 2, Bulk Workability of Metals). ness variations. Hundy and Singer (Ref 6) found
tion (Ref 4) expresses the relative drawing stress Relative drawing stress is plotted in Fig. 5 ver- this variation to increase with friction for a 20
by the ratio of the drawing stress (or front ten- sus the semicone angle (a) of the die for various die angle, but to be independent of friction for a
sion, sxf) and the flow stress (s0) such that: values of the friction factor (m). Each curve in 30 angle, presumably because inhomogeneity
Fig. 5 demonstrates a minimum at some optimal had an overwhelming effect. The surface defor-
s xf s xb angle. For angles smaller than the optimal angle, mation associated with very high h/L ratios can
= + W i + Ws + W f
s0 s0 (Eq 12) friction losses (W f ) are excessive. Friction losses be beneficial because residual compressive
drop with increasing die angles, but redundant stresses impart greater fatigue resistance.
where sxb is the back tension (Fig. 4), W i is the power losses increase with increasing die angle. Die Geometry. The simple conical die used
relative (dimensionless) portion of power con- Beyond the optimal angle, excessive distortion in practice does not ensure the most homoge-
sumed by internal deformation, Ws is the portion occurs with increasing die angles, and the draw- neous deformation, and ideal die profiles can be
of power consumed by shear or redundant work, ing stress increases. The optimal die angle that theoretically calculated for more uniform strain
and Wf is the power portion consumed by fric- minimizes the drawing stress increases with the or strain-rate distribution while also minimizing
tion. The values of these dimensionless compo- increase in reduction and friction. With higher friction. However, the improvement is seldom
nents for relative portion of power consumption friction it is advisable to use larger die angles. sufficient to justify the cost of making dies with
are calculated from the geometry as follows. More expressions have been derived for the convex, sigmoidal, or other complex shapes.
The relative portion of power consumed by determination of process limitations due to tear- Centerbursts. Surface compressive stresses
internal deformation (W i) is expressed as: ing, dead zone formation, and shaving (Ref 5). are balanced by central tensile stresses. At high
When the phenomenon of central burst is ana- h/L ratios, triaxial tensile stresses are set up in
R lyzed, one finds that increasing friction deters the center. The density of wire decreases be-
W i = 2 f (a )ln 0 central burst during extrusion, but promotes it cause of void formation, and discontinuity in
Rf (Eq 13)
during drawing (Fig. 6). material flow can result in periodic centerburst
Strain Distribution. Inhomogeneous strain defects* in wire of moderate ductility, especially
where the function of the semicone angle (a) is: distribution is governed by the h/L ratio, but it if it is subjected to a succession of high h/L
also increases with friction. Draw forces are af-
fected by inhomogeneity of deformationas ex-
*For an example of periodic centerburst defects in extended
1 11 2 pressed by the h/L ratio. Inhomogeneous strain product, see Fig. 5 in Chapter 21.
f (a ) = 1 - cos a 1 - sin a
sin 2 a 12

1 1 + 11 / 12
+ ln
(11 12) 11 / 12 (cos a ) +
(1 - [11 / 12 ] sin 2
a )

The relative portion of power consumed by the

shear (or redundant power) component (W s ) is:
Fig. 4 Wire drawing or extrusion processes. Source:

Ref 5, p 59 Fig. 5 Relative drawing stress as a function of semicone angle and friction factor (m). Source: Ref 5, p 61
Fig. 7 Condition for bulge formation in drawing with light reductions. Source: Ref 9
Fig. 8 Pressure-lubricating shaving/drawing die.

Fig. 9 Configurations of roller dies. Source: Ref 1, p 352
Source: Ref 10
when friction is high (Fig. 7) (Ref 8, 9). At yet the bar is fully enclosed only in the plane of the
higher angles the surface of the wire is shaved axes, and that gaps open farther back. This lim-
off by a process related to cutting with a nega- its the reduction attainable without flash forma-
tive-rake-angle tool. Such shaving is an impor- tion, and successive die sets must be set at a dif-
tant step in making wire free of surface defects ferent angular position so that any incipient flash
inherited from the hot processing stages. In one is rolled back into the body of the bar. Some
version of pressure die for lubrication, the inlet spread is unavoidable, and the pass must be
sealing is replaced with a shaving die (Fig. 8). slightly opened to accommodate it. The repeated
Drawing of Profiles. An increasingly impor- deformation places greater demands on work-
tant activity is the drawing of bars with more-or- piece ductility. However, sliding friction is re-
less complex cross sections, often to close toler- placed by the much lower rolling friction, and
ances, and in difficult-to-draw materials. the tribological demands are much less strin-
Drawing stresses are somewhat higher, but the gent: the danger of die pickup and die wear are
major concern is uniformity of material flow, much reduced. Lubrication practices are similar
which can be promoted with convex dies. to those of cold rolling.
Drawing of profiles presents some special lubri- From a mechanical view point, the process is
cation difficulties; as material flow is less uni- equivalent to rolling with a high enough front
form, local stress concentrations arise at corners tension to obviate the need for the rolls to be
and grooves, and surface expansion can be much driven. This simplifies construction, but limits
greater than in drawing of a round to a round. the attainable reduction. This limit is removed
Fig. 6 Criteria for central burst in (a) drawing and (b) Lubricants may be evaluated by drawing profiles when the rolls are driven. Three-roll arrange-
extrusion. The safe region is above or to the left representing varying degrees of difficulty, and
of each line. Source: Ref 5, p 61
ments can become quite complex, but are feasi-
the order of merit of lubricants may change rel- ble even for small-diameter wire (Ref 11). It is
ative to wire drawing. not actually necessary to have profiled rolls. As
passes. Because friction increases the back ten- Roller Dies and Rolling. Many of the lubri- pointed out by Sayer and Moller (Ref 12), duc-
sion, it aggravates the situation, and the theoret- cation problems in wire drawing can be by- tile wires can be reduced in a succession of flat
ical safe zone for drawing is reduced (Fig. 6a). passed by using roller dies or by switching to rolls and then shaped in the last rolls or finished
Of course, this represents only a necessary but cold rolling of sections. Rectangular sections by drawing.
not a sufficient criterion for the onset of center- with varying width-to-thickness ratios can be
burst: it is also necessary that the material have drawn through turks heads in which four rollers
low ductility. A model based on the volume frac- are mounted adjustably, with their axes in the Tube Drawing
tion of voids has been proposed (Ref 7). same plane.
External Deformation and Shaving. High Roller dies are usually constructed with two With four exceptions, the methods and equip-
h/L ratios affect the external deformation, too. or three rollers (Fig. 9) mounted in a frame with ment used for cold drawing tubes in straight
First, a bulge forms at the die entry; for a given their axes in a single plane. A consideration of lengths are basically identical to those used for
die angle, bulging sets in at a lower reduction the three-dimensional geometry will show that bar drawing. The four exceptions are:
Some tubes require more than one drawing

pass.
Tubes are usually longer than bars. Draw-
benches for tubes are usually correspond-
ingly longer, some permitting drawn lengths
of more than 30 m (100 ft).
Tube diameters are generally larger than bar
diameters, ranging to about 305 mm (12 in.).
The bigger tube drawbenches have larger
components than do bar drawbenches.
Tubes require internal mandrels or bars for
simultaneous working or support of the in-
terior surface during drawing. Tube draw-
benches are usually equipped with one of
several available devices, usually powered,
for ready assembly of the cleaned, coated,
and pointed workpiece onto internal bars or
rod-supported mandrels. If rod-supported
mandrels are used, they are usually air-op-
erated so that the mandrel can be placed and
maintained in the plane of the draw die after
pulling starts. Butt or electric-welded tubes Fig. 10 Tube drawing processes. (a) Tube sinking. (b) Drawing with a fixed cylindrical plug. (c) Drawing with fixed
are sometimes drawn to smooth the weld conical plug. (d) Floating plug. (e) Drawing on a bar (or mandrel). Source: Ref 1, p 353
seams and tube walls.
Tubes, particularly those having small diam- draw. The additional friction, mp, on the plug in-
eters and requiring working only of their outer creases the draw force. This effect can be allowed
surfaces, are produced from cold-drawn coils for in Eq 5 by replacing m cot a with the term:
on machines that straighten the stock and cut it
(md mp)/(tan a tan b) (Eq 16)
to required lengths. As with bars, however,
most tubes are produced from straight lengths The optimal die half angle is usually 12 to 15 in
rather than coiled stock. The surface finish of cooperation with b at 10 to 11 for conical plugs
a tube blank can be improved, and its wall (Fig. 10c).
thickness and/or diameter reduced, by cold Tube Drawing with Floating Plug. In draw-
rolling on a pilger mill or, more frequently, by Fig. 11 Change in wall thickness in tube sinking with
ing long and small-diameter tubes, the plug bar
cold drawing. 15% reduction in diameter, in dies with 9
half angle. Source: Ref 14 might stretch and even break. In such cases, it is
In tube drawing both the diameter and wall advantageous to use a floating plug (Fig. 10d).
thickness change. In calculating strain from Eq This process can be used to draw any length of
1 and 2, Ao and Al now are cross-sectional areas ness increase and tube elongation depends on tubing by coiling the drawn tube at speeds as
of the tube before and after drawing, respec- the flow stress of the drawn part, die geometry, high as 10 m/s (2000 ft/min). If the plug is de-
tively. In all forms of tube drawing, inhomo- and interface friction. Its diameter is reduced signed so that the frictional force keeps it in the
geneity of deformation arises when h/L is high; while the wall slightly thickens and the inside deformation zone, the holding bar can be omit-
here, h stands for mean wall thickness. As in surface roughens. Thickening of the wall is a ted. Such a floating plug allows drawing of long
bar drawing, most of the total work is trans- function of friction and of the tube wall-thick- lengths on a drum (bull block), and heavier re-
formed into heat, and the temperature of the ness-to-diameter ratio (Fig. 11) (Ref 14). The ef- ductions can be taken. Plugs can be designed
tube rises. fects of friction are the same as in bar drawing, from basic principles if friction can be charac-
Seamless tubes are mostly produced by one of but interface pressures are lower and optimal die terized. Here, too, there is an optimal angle.
the hot tube-rolling or tube-extrusion tech- angles are larger. Dvnyi (Ref 15) observed op- Tube Drawing with Moving Mandrel. In
niques. Only the Ehrhardt process utilizes a timal die angles of about 12 at 10% reduction drawing on a moving bar (also called a man-
technique related to drawing. Hot, heavy-wall, and about 20 at 40% reduction in drawing alu- drel), the tube material is forced to slide over a
closed-ended tube blanks are pushed through a minum tubes. bar (Fig. 10e). The mandrel travels at the identi-
series of roller dies in mechanical push benches Drawing with a fixed plug (Fig. 10b, c) is cal speed at which the section exits the die. This
to reduce the tube wall. For example, the used for drawing large-to-medium diameter process, also called ironing, is widely used for
Ehrhardt Push-Bench Method is effective in pro- straight tubes. The plug, when pushed into the thinning the walls of drawn cups or shells in, for
ducing seamless steel pipes, as described Ref 13 deformation zone, is pulled forward by the fric- example, the production of beverage cans or ar-
for production of seamless pipe with outside ditional force created by the sliding movement of tillery shells. A frictional stress is set up that
ameters of 165 to 950 mm (6.5 to 37 in.) for the deforming tube. Therefore, the plug must be transfers the drawing stresses from the tube wall
power plants, nuclear power plants, chemical held in the correct position by a bar. The internal to the bar (Fig. 7). Because friction on the bar
plants, and construction purposes. diameter and surface finish are controlled when opposes friction on the die, the stress in the tube
Sinking. Figure 10 illustrates the basic types drawing is done on a plug. In addition to the di- wall can be approximated by substituting in Eq
of tube drawing. Sinking (Fig. 10a) is closest to ameter (sink) being reduced, the wall thickness 5 for m cot a the term:
wire drawing in that the tube is drawn through a (draw) is also reduced to a dimension defined by
die without a mandrel (or plug). In tube sinking, the diameter of a cylindrical plug (Fig. 10b) or (md mb)/tan a (Eq 17)
the tube is initially pointed to facilitate feeding the position of a conical plug (Fig. 10c). The
through the die; it is then reduced in outside di- sink-to-draw ratio affects interface pressures, If md < mb (a quite likely circumstance), the
ameter while the wall thickness and the tube sliding velocities, and surface extension. The stress imposed on the tube wall drops below that
length are increased. The magnitudes of thick- severity of operation increases with increasing prevailing in frictionless drawing. Of course, the
force necessary to overcome friction on the bar either with a softer metal or with a conversion drawing slip-type machines. On these machines,
has to be provided. After drawing, the tube has coating. Copper or tin can be chemically de- the surface speed of the capstans, except for the
to be reeled (cross rolled) off the bar, as in the posited on the surface of the metal. This thin final (pullout) capstans, exceeds the speed of the
Ehrhardt process. layer of softer metal acts as a solid lubricant dur- wire being drawn, thus creating slip of the wire
ing drawing. Conversion coatings may consist of on the capstans. Brighter surface finishes are
Lubrication sulfate or oxalate coatings on the rod; these are generally produced with these machines, but the
then typically coated with soap, as a lubricant. machines are limited to smaller reductions per
Proper lubrication is essential in rod, tube, Polymers are also used as solid lubricants, such pass than with dry-drawing nonslip continuous
and wire drawing. In contrast to rolling, no fric- as in the drawing of titanium. machines.
tion is needed at all for wire drawing, tube sink- In the case of steels, the rod to be drawn is With wet-drawing slip-type machines, the
ing, and tube drawing on a fixed plug. However, first surface treated by pickling. This removes drawing operation is generally confined to an
some minimum friction is essential for drawing the surface scale that could lead to surface de- enclosed chamber, with the lubricant bathing the
with a floating plug, and friction is helpful on fects and therefore increases die life. For the dry dies and wire as it is being drawn. These ma-
the tube/bar interface in drawing on a bar. Thus, drawing of ferrous metals, four types of nonslip chines are less complicated electrically than
if at all possible, the lubricant is chosen to give continuous machines are in general use: nonslip machines, and only one drive system is
lowest friction and minimum wear. It is essen- employed. They are designed with either tandem
Accumulating-type machines
tial, though, that the heat generated should be or cone-type configurations, usually with hori-
Double-block accumulating-type machines
extracted, especially in high-speed drawing, be- zontal spindles, but sometimes with a vertical
Controlled-speed machines
cause otherwise the lubricant can fail and the spindle for the finishing capstan. Cone capstans
Straight-through machines
properties of the wire suffer. have drawing surfaces (usually hardfaced) that
An accumulating-type multiblock continuous are stepped outward to provide increasing pe-
Dry and Wet Drawing wire-drawing machine is equipped with electro- ripheral speeds. This compensates for the elon-
magnetic block clutches. Photocells sense high gation and increasing speed of the wire as it is
These lubrication and cooling requirements and low wire accumulation on each block and reduced in diameter during drawing.
can be satisfied by one of two techniques: disengage or engage appropriate block clutches.
Double-block accumulating machines have indi- Plastohydrodynamic (Full-Fluid-Film)
Dry drawing, where the lubricant is chosen
vidually driven blocks. Wire is transferred from
for its tribological attributes and the wire is Lubrication
the first drawing block by means of an interme-
cooled while it resides on the internally
diate flyer sheave that reverses the direction of As in other processes, mixed-film lubrication
cooled draw drums (capstans)
the wire (without twisting it) onto a coiling
Wet drawing, where the lubricant is chosen is often encountered, but the high speeds of
block mounted immediately above the first drawingcoupled with a favorable process
both for its tribological attributes and for its
drawing block. The wire is then held temporar- geometrymake plastohydrodynamic (PHD)
cooling power
ily in storage until demanded by the second lubrication possible and even desirable, at least
A transition between the two techniques is drawing block. for early draws in which a rougher surface can
sometimes employed, particularly in low-speed On controlled-speed machines, the wire fol- be tolerated. In principle, a thick film can de-
drawing of bar and tube. A high-viscosity liquid lows an essentially flat path from block to block velop with any lubricant if other conditions are
or semisolid is applied to the workpiece and/or with a constant, unvarying amount of wire stor- conducive to it. Under some very specific condi-
die. Drawing speeds range from 0.3 to 2 m/s (59 age without twisting and slipping. A tension arm tions, it is possible to maintain a continuous (or
to 390 ft/min) for bar and heavy tube, through 7 between the blocks, activated by a loop of the full-fluid) film between the die and workpiece in
m/s (1380 ft/min) for heavy (say, 5 mm, or 0.2 wire being drawn, regulates the speed of the ad- plastic deformation processes.
in.) wire, to 40 m/s (7870 ft/min) for fine steel justable-speed direct-current motor on the pre- Early development of hydrodynamic lubrica-
wire and 70 m/s (13,780 ft/min) for nonferrous ceding block. Straight-through machines, with- tion to wire drawing is described in Ref 17 and
metals. The limit is often set by die life and coil out tension arms, are also available. The spindles 18. More detailed information on the phenome-
length, which would force a die change or are often canted from the vertical axis to accom-
rethreading at unreasonably short intervals. modate wire buildup on the blocks and to pro-
Dove (Ref 16) provides additional details on the vide unimpeded, straight entry into the succeed-
lubrication of ferrous wire. ing die; this is usually done when large-size
In dry drawing, the lubricant is usually a dry workpieces are required. Skillful operators are
soap powder, placed in a die box and picked up necessary because torque adjustments may need
by the wire surface upon its passage through the to be altered at each block when stringing up the
box. This technique is used for steel wire larger machines in order to make the electrical system
than 0.5 to 1 mm (0.02 to 0.04 in.) in diameter, function properly.
for which the relatively rough surface produced In wet drawing, the lubricant is chosen both
is acceptable. For the most severe draws and for for its tribological attributes and for its cooling
tubes, the soap is often preapplied from a solu- power and can be either oil-based or aqueous.
tion, if necessary, over a conversion coating; the The lubricant/coolant can be applied to the die
soap must be allowed to dry. Additionally, exter- inlet, to the wire, and often also to the capstan,
nal air cooling of the wire coil and water cooling or the entire machine can be submerged in a
of the die holder are possible. If water is applied bath. When the machine operates with slip, the
to the wire at all, it must be totally removed be- lubricant must reduce wear of the capstan while
fore the wire enters the next die. maintaining some minimum friction. This wet-
These techniques are customarily described as drawing practice is typical of all nonferrous met-
dry drawing (a process that has nothing to do als and of steel wires less than 0.5 to 1 mm (0.02
with dry working in the sense of an unlubricated to 0.04 in.) in diameter.
interface). With high-strength materials such as The continuous drawing of nonferrous rod
steels, stainless steels, and high-temperature al- and wire, as well as some intermediate and fine Fig. 12 Inlet tube for hydrodynamic lubrication. See
loys, the surface of the rod or wire can be coated sizes of ferrous wire, is generally done on wet- Ref 5, p 62
non of hydrodynamic lubrication during flow The angle q or the entire entry geometry
through converging dies can be found in Ref 1, changes by deformation (rounding) of the
4, 5, and 19. The analysis of hydrodynamic lu- workpiece just before entry into the die
brication by Hillier (Ref 20) offers a simple and (Fig. 13a).
applicable treatment. In early studies of the The fluid pressure at the die entry is in-
equipment design for analysis of hydrodynamic creased (as in wire drawing, hydrostatic ex-
lubrication, a long tube with a narrow gap be- trusion, or deep drawing) (Fig. 13b).
tween the tube and the wire was firmly attached The angle q is reduced by elastic deflection
at the entrance side of the die (Fig. 12). The lu- of the die.
brication adhered to the wire and was dragged The workpiece surface roughness helps
into the clearance between the tube and wire. At carry lubricant into the deformation zone
about 3 m/s (590 ft/min), the pressure that built (Fig. 13c). Fig. 15 Setup for hydrodynamic lubrication with dry
up at the approach to the die reached 70 to 275 soap. Source: Ref 5, p 62
Viscosity is also a function of pressure and tem-
MPa (10 to 40 ksi), and the liquid formed a film
perature. The magnitude of the viscosity-pres-
between the wire and the die (Ref 21). film with a solidlike substance. Hydrodynamic
sure coefficient (j) is an influential factor in de-
Treatment of PHD lubrication has many sim- lubrication through a pressure box is shown in
termining film thickness, as follows:
ilarities to elastohydrodynamic (EHD) lubrica- Fig. 15. The pressure chamber is inserted into
tion, and much of the work came from research the soap bin of a conventional wire-drawing bull
on EHD lubrication. In its simplest form (with 6 h0 jv block. The bin preceding the entrance die to the
h0 =
the die and workpiece assumed to be rigid prior tan q[1 - exp( -js f )] (Eq 19) chamber is filled with powdered soap. The wire
to entry in the deformation zone), the entrained running through the powder supply drags pow-
film thickness (h0) is: where h0 is the viscosity at atmospheric pres- der into the chamber through the narrow gap be-
sure. tween the incoming wire and the approach die.
6 hv In high-speed wet drawing, there is little doubt When the wire is pulled through the soap,
h0 = that film thickness increases with velocity, al- some soap is dragged into the chamber. In the
s 0 tan q (Eq 18)
though this is not necessarily proof of a full-fluid- small clearance gap, the soap powder shears,
film mechanism. The hydrodynamic effect does heats because of this shear, and then melts.
where h is the viscosity, v is the mean surface
not depend entirely on the speed, viscosity, or Liquid lubricating soap then enters the pressure
speed, s0 is the flow strength of the workpiece
chamber/tube length. As noted, hydrodynamic lu- chamber. The faster the wire is drawn, the higher
material, and q is the angle between the con-
brication can be accomplished at lower velocities are the temperatures and pressures introduced in
verging surfaces (Fig. 13a). The entrained film
and with lower fluid viscosities when higher pres- the chamber causing hydrodynamic lubrication.
thickness increases with increasing viscosity (h)
sure is applied. Pressurized chambers (Fig. 14) A chemical bond between the wire and the
and surface speed (v) or with diminishing wedge
have been used, whereby the lubricating liquid is molten soap is also produced. As soon as the
angle q (a for a conical die) and with dimin-
supplied to the chamber at high pressure by ex- wire exits the die, its temperature drops and the
ishing flow stress sf (or, more relevantly, with
ternal means (Ref 22). There is experimental evi- soap freezes and forms a layer on the wire.
die pressure pm, which is governed by reduction, There is now a layer of lubricant performing the
dence to suggest that a full film can be generated
die angle, friction, and back tension). The film duties of hydrodynamic lubrication, that is,
with a high-viscosity lubricant in drawing of soft
thickness increases in steady-state processes if:
metals. Felder (Ref 23) observed thick (15 mm) keeping full separation between the deforming
films (as judged from weight differential) and no wire and the next five or six dies. The pressures
electrical contact in drawing soft aluminum and produced in these chambers, without any exter-
copper wires with a chlorinated oil of 200 Pa s nal or auxiliary agent, are in excess of 275 MPa
viscosity, at speeds from 0.5 to 5 m/s (98 to 980 (40 ksi).
ft/min). The presence of a bamboolike defect (a One difficulty associated with the use of a
periodic variation of diameter) also suggested powdered-soap bin is in maintaining steady drag
full-fluid-film lubrication on soft copper. of powder into the chamber. Inconsistencies in
However, only short pieces of wire were drawn, the particle size and dryness of the powder vary
and it is not clear whether fluid-film conditions the effectiveness of its adhesion to the wire and
could have been maintained on long lengths of the quantity of powder dragged. At the extreme,
wire after attaining a thermal equilibrium. a hollow channel in the powder surrounds the in-
Hydrodynamic Lubrication in Dry coming wire and no soap is dragged in. Greater
Drawing. In high-speed dry drawing, there is uniformity of the powder and constant agitation
more opportunity for forming a fully separating (directly or through the box) may improve the
performance. Proper surface preparation also
may be called for. A recent development utilizes
the conventional powder spray system with elec-
trostatically charged particles.
With phosphating, the most popular method, a
predetermined layer of a spongy phosphate coat-
ing is applied to the surface of the wire elec-
trolytically. This sponge can absorb and retain a
large volume of lubrication liquid. Even without
the lubricant, the phosphate sponge provides an
effective separation between the die and the
workpiece, thereby minimizing friction and wear.
Fig. 13 Geometric factors that influence thickness of
Soaps in Hydrodynamic Lubrication. The
lubrication film and the occurrence of plasto-
hydrodynamic lubrication. (a) Rounding off. (b) Fluid pres- Fig. 14 Wire drawing through a pressurized chamber. rheological behavior of soaps is highly effective
sure. (c) Surface roughness. Source: Ref 1, p 69 Source: Ref 5, p 62 in allowing hydrodynamic lubrication of metal-
Melting point is important in controlling feed- the half angle must be kept at about 6 instead of
ing. Excess soap picked up by the wire is re- at the previously used 8 to 10; obviously, the
jected by the die; if it flows back in a molten lower angle is needed to maintain a sufficiently
form, the clumps thus formed deny further ac- thick film at the higher temperatures generated
cess of soap to the die. Sodium soaps are prone by the higher speeds. For this reason, cooling
to caking because of their hygroscopic nature. must be improved as well. Land length is also
Tube Drawing. The possibility of increasing important. An excessively long land strips the
the film thickness in tube drawing has also at- wire of its soap coating, whereas a short land
tracted attention. In tube sinking, a pressure tube wears rapidly. Schmidt (Ref 28) suggests that,
or multiple dies can offer benefits (Ref 25). In for steel wire 2.0 to 6.5 mm in diameter, land
drawing on a plug, the critical problem is to in- length (in mm) = 5/v, where v is drawing speed
crease lubricant supply to the plug/inner-tube in- in m/s. A similar relation may well hold for other
terface. Pumping of the lubricant through a hol- materials and lubricants.
low plug is feasible, but in itself does not help to Under mixed-film conditions, friction and
build up a lubricant film. For this, a double-plug wear may be difficult to characterize. There is no
arrangement, with supply of pressurized lubri- single value of coefficient of friction or other
Fig. 16 Apparent viscosity of various soaps versus tem-
cant to the enclosed space, is needed. Some im- descriptor can possibly characterize the condi-
perature. Source: Ref 24
provement in plug drawing is achieved simply tions existing for a wide range of speeds, geome-
by using a suitable plug profile, as reported by tries, and reductions. As Schey notes in Ref 1 (p
drawing processes. The viscosity of soap is re- Rees (Ref 26). The oscillating floating plug (Fig. 372): It would be futile to compile lists of m
duced at high temperature (Fig. 16) and a higher 17a) is designed to trap lubricant; the pressure in values; in addition to the effects of process vari-
drawing speed is then required for full-film lu- the lubricant is assumed to force the shoulder ables, m values are influenced also by the
brication. The effect of speed is twofold; first it back against the spring, creating an oscillating- method of their determination (plane-strain
increases film thickness, but then heat genera- pumping action. The adjustable floating plug drawing, split die, strain-gaged die) and by the
tion causes a drop in viscosity. Thus, the indus- (Fig. 17) creates a lubricant cavity and combines method of calculation from measured draw
trial practice of using leaner (higher lime-con- the advantages of straight and tapered plugs. The force. Furthermore, many laboratory measure-
tent) soaps for heavier gages and harder alloy is fluid pressure generated in the cavity prevented ments are conducted at very low speeds and
justified. galling in drawing of a limited number of stain- sometimes on lengths as short as 25 mm, and
In addition to lime (CaO), other thickeners are less steel tubes (Ref 24). thus the results are quite unrepresentative of
also incorporated. Borax and soda ash (Na2O) production practices. For predictive calcula-
are used instead of or in addition to lime. Iron Mixed-Film Lubrication tions, the values listed in Table 1 are suggested
sulfate replaces the hydrated iron oxides found by Schey.
on sull-coated wire, and imparts a brown color Mixed-film lubrication occurs when there is
to the drawn wire. Graphite is added for a black an interruption in the conditions for the continu- Solid-Film Lubrication
finish, and TiO2 (which also acts as a polishing ous (full-film) lubrication. Average film thick-
agent) for white color. Mica and talc, even ness has little meaning under these conditions. This section describes lubricants that do not
though of lamellar structure, act only as parting Nevertheless, it is a useful concept for visualiz- need a hydrodynamic wedge action to develop a
agents. Often, MoS2 is added to a sodium soap, ing process conditions. As indicated by Eq 18, film.
sometimes in company with graphite and lime, film thickness increases with decreasing entry Layer-Lattice Compounds. Graphite and
mostly for difficult-to-draw, strong alloys. angle. Indeed, it has been repeatedly observed MoS2 are seldom used on their own, except for
Sulfur added to steel-drawing soaps acts as an that lubricant flow increases greatly with very high-temperature work. Their friction is found to
extreme-pressure agent. Evidently, experience low (about 2) half angles. Even in low-speed be independent of speed (Ref 29). The effect of
has demonstrated that film thickness is fre- experiments, a transition from thick-film to pressure is controversial. With a truly continu-
quently insufficient to ensure complete mixed-film lubrication may be observed on in- ous, well-developed film, m should be constant.
die/workpiece separation. It is not clear whether creasing to some critical value. In sheet drawing, m has been observed to drop or
or not such additives affect lubricant rheology to Die geometry is important in industrial prac- rise with pressure in different situations. It is
a significant degree. They can, however, inter- tice. Schmidt (Ref 28) notes that in drawing of conceivable that in some instances the film was
fere with film formation. Occasionally, an addi- steel wire 5 or 6 mm (0.2 or 0.24 in.) in diame- burnished and improved, while it may have lost
tive, such as graphite, can be used as a coating ter at speeds greater than 8 m/s (1580 ft/min), its continuity in other tests. Generally, m rises
applied prior to drawing. On nonferrous wire,
sodium and potassium soaps are used as coat-
ings without carriers, but their rheology is most
likely affected by the changes due to pressure,
temperature, and reactions with the substrate.
In considering the rheology of soap, one must
remember that the soap undergoes changes in
service. Reactions with the wire, and especially
reactions with oxides remaining after descaling
and with borax, are also possible. Of course,
soap viscosity and composition matter little if
the wire coating and/or wire roughness are inad-
equate to carry soap into the die. The soap must
be of controlled particle size; beaded (granu-
lated) soaps help avoid tunneling and reduce
dusting. Vibration of the die box or agitation of
the soap is necessary when the wire runs free of Fig. 17 Profiles of floating plugs with full-film lubrication. (a) Oscillating floating plug. (b) Adjustable floating plug.
vibration or whipping that would aid feeding. Source: Ref 27
Table 1 Commonly used drawing lubricants and typical values (lubricants listed working applications, only thermoplastic poly-
according to increasing severity of conditions) mers with appropriate glass-transition tempera-
Material Lubricant(a) (b)
ture (Tg) are of interest. The film thickness of a
solid coating deposited from a solvent can be
Wiredrawing controlled quite independently of process condi-
Steels Over 1 mm: dry (Ca-Na) soap on
lime or borax MF
tions; this is a great advantage in low-speed bar
Over 1 mm: phosphate + soap MF and tube drawing. The friction of coatings is in-
Under 1 mm: EM (MO + fat + EP) 0.07 dependent of speed as long as no major temper-
Under 1 mm: phosphate + EM 0.1 ature rise is encountered. Polytetrafluoroethyl-
Metal (Cu, Zn, brass) + EM MF
Stainless steels and Ni alloys EM (MO + Cl) (on lime) 0.1
ene (PTFE) has often been investigated, but
MO + Cl additive (on lime) 0.07 other polymers, closely related to the coatings
Chlorinated paraffin, wax 0.05 used on sheet have found use. Numerous poly-
Oxalate + soap 0.05 mers are useful in bearing applications, but for
Metal (Cu) + EM (or oil) MF
Al and Mg alloys MO + fatty derivatives MF
metalworking purposes only thermoplastic poly-
Synthetic MO + fatty derivatives MF mers of the appropriate glass-transition temper-
Cu and Cu alloys EM (MO + fat) (+ EP) MF ature are of interest.
Metal (Sn) + EM or MO MF Polyethylene (Tg 120 C, or 185 F;
Ti alloys Oxidized + Cl oil (wax) 0.15
Fluoride-phosphate + soap 0.1
melting temperature, or Tm 137 C, or 280 F)
Metal (Cu or Zn) + soap or MO 0.07 is readily available in sheet form and possesses
Refractory metals Hot: graphite coating 0.15 substantial elongation (100 to 300%). In combi-
warm (cold): graphite or MoS2 0.1 nation with mineral oil, sheets 2.5 to 7.5 mm
Oxidized + wax 0.15
Metal (Cu) + MO 0.1
thick provide excellent lubrication.
Bar and tube drawing Polypropylene (Tg 18 C, or 0.5 F; Tm
Steels Heavy oil, soap-fat paste, grease MF 176 C, or 350 F) has similar properties.
Heavy oil, soap-fat, paste, grease (+ EP) (+ MoS2, etc.) MF Polytetrafluoroethylene (Tg 50 C, or
Polymer coating + EP oil 0.07
Phosphate + soap 0.05
60 F; Tm 327 C, or 620 F) sheets have
Metal + MO (or EM) MF limited ductility, but well-adhering films of con-
Stainless steels and Ni alloys MO + Cl additive 0.15 trolled thickness and good extendability can be
Chlorinated wax 0.07 deposited from a tricholorotrifluorethylene dis-
Polymer (chlorinated) (+ MO) 0.07
Oxalate + soap 0.05
persion of telomers (which have aliphatic end
Metal (Cu) + MO MF chains grafted onto the PTFE). Sputtering is also
Al and Mg alloys MO + fatty derivatives MF possible.
Soap coating 0.07 Polyvinylchloride (Tg 87 C, or 190 F;
Wax (lanolin) coating 0.05
Polymer coating 0.05
Tm 212 C, or 415 F) is available both in
Cu and Cu alloys EM (fat) MF sheet form and as a deposited coating.
MO (+ fat) (+ EP) MF Polymethylmethacrylate (PMMA, Tg 105
Soap film 0.05 C or 220 F; Tm 240 C, or 465 F) and
Ti alloys Polymer coating 0.07
Oxidized + Cl oil (wax) 0.15
acrylics in general have the advantage that they
Fluoride-phosphate + soap 0.1 are readily deposited from either solvent solu-
Metal + soap 0.07 tions or emulsions, and they can be modified so
Refractory metals Hot: graphite coating 0.15 as to endow them with boundary-lubrication
Warm (cold): graphite or MoS2 0.1
Oxidized + wax 0.15
properties.
Metal (Cu) + MO 0.1 Acetal resins contain the linkage CH2O
(a) EM, emulsion of ingredients shown in parentheses. MO, mineral oil; of higher viscosity for more severe duties, limited by staining. EP, EP com-
CH2O and have found some limited applica-
pounds (S, Cl, and/or P). (b) MF, mixed-film lubrication; = 0.15 at low speeds, dropping to 0.3 at high speeds. Source: Ref 1, p 373 tion.
Polyimides formed with aromatic radicals
have outstanding temperature stability.
with reduction, most likely because surface ex- phating. Phosphating was originally developed Metal coatings are among the most powerful
tension exposes more asperities. for steel, but modifications for other metals are aids to lubrication. Even though thin metal coat-
Conversion Coatings. By definition, conver- also available: ings can provide lubrication by reducing adhe-
sion coating involves the reaction of the surface sion or friction, they are almost always used in
Zinc and its alloys can be phosphated in so- conjunction with a liquid or semisolid lubricant.
atomic layers of the workpiece material with the
lutions enriched with iron. Thus, they serve more as a lubricant carrier, with
anions of a selected medium. It is similar to ex-
Aluminum is phosphated mostly in propri- the lubricant chosen for the coating rather than
treme-pressure (EP) lubrication in that a con-
etary solution containing chromates and flu- the substrate metal. The metal film may improve
trolled corrosion process produces a firmly
orides in addition to phosphates. performance of another (usually liquid) lubri-
bonded surface layer. The difference is that a
Titanium can be treated in a solution con- cant either by providing better entrapment
conversion coating does not lubricate, but acts
taining sodium orthophosphate, potassium through favorable surface configuration or by al-
only as a parting agent and lubricant carrier. A
fluoride, and hydrochloric acid. lowing chemical reactions. The film must be
superimposed and sometimes reacted lubricant
layer is necessary to complete the lubrication Steels containing more than 5% Cr cannot be thick enough to ensure continuous coating, and
system. phosphated. Therefore, stainless steels, the metal film can only be effective if it is well
Conversion coatings are indispensable for chromium-molybdenum steels, nickel-base su- bonded to the substrate.
tube drawing, low-speed bar drawing, and wire peralloys, and other high-chromium alloys are
drawing with heavy reductions on materials that oxalated. Surface Preparation
are otherwise difficult to lubricate. Coating Polymer Coatings. Deposition or bonding of
types include sull coating, oxide coatings, sulfu- polymer films provides lubrication in a wide Any surface oxide or dirt carried into the die
ration, chromating, and most significantly phos- variety of bearing or wear applications. In metal- greatly accelerates die wear and lubricant deteri-
oration. Therefore, the first step for successful Severe wear at the point where the wire en- of the wear situation. For the different types of
lubrication is preparation of the surface. Both ters the die results in so-called ringing. The wear situations, there are four general ways (or
mechanical and chemical techniques are used. In change in entry geometry changes the film types of mechanisms) by which material wear
drawing steel wire, blast cleaning with cast iron thickness and thus the surface finish of the occurs:
or cut-wire shot is used. Steel wires also can be product.
descaled with the scale itself. Much but not all of Wear of the draw cone changes the geome- Adhesive processes
the scale can be removed by reverse bending, try of operation and affects film thickness. Abrasive or deformation processes
provided that the sum of the wrap angles ex- Wear of the die land is directly responsible Fatigue or fatiguelike processes
ceeds 180 or that a total elongation of 5 to 10% for loss of tolerances and also affects the Oxidative or corrosive processes
is achieved. Elongation of the wire reduces the surface finish of the product. If wear of a With adhesive processes, wear occurs as a re-
scale residue to some 0.6%. Normally, the wire die is nonuniform, the wire acquires a sult of the bonding that takes place between two
residue must be removed by pickling, but Bernot noncircular profile. surfaces in contact. With subsequent separation
(Ref 27) found wet blasting (with scale in water) of the two surfaces, material from either surface
to be an ecologically and economically attrac- Wear Mechanisms may be pulled out, resulting in wear. Adhesive
tive alternative. Abrasive belts have been used wear processes normally occur only with slid-
on heat treated rod. Continuing efforts are made Deformation would proceed without wear, but ing. However, adhesive wear processes can
to increase the efficiency of pickling, not only by only if the die and workpiece were separated by occur under nominal impact and rolling condi-
control of chemical composition but also by ap- a thick film of nonreactive lubricant without any tions, because of the slip that is often present in
plication of ultrasonics. Recycling of spent foreign particles. However, PHD lubrication is those situations (Ref 34).
liquors is increasingly practiced. Even though seldom achieved; lubrication is mostly of the Abrasive wear processes are those fracture,
these examples refer to steel, similar problems mixed-film or boundary type. Thus, reactions cutting, and plastic deformation processes that
exist, to varying degrees, with other metals. with surface typically occur and result in pro- can occur when a harder surface engages a softer
Prevention of oxidation eliminates the need gressive loss (wear) of the interface materials. surface. These mechanisms tend to produce ma-
for descaling. Steel wire rod cooled by water or Mild wear of the workpiece material may be chining-chip-like debris. Fatigue or fatiguelike
nitrogen has a thin scale (4.5 mm max) consist- considered normal and acceptable. Wear of the wear processes are those associated with crack
ing of the less-abrasive FeO and could be drawn, die, or severe wear of the workpieceperhaps initiation and propagation or progressive defor-
at least experimentally, without descaling, in a coupled with scoring of the workpiece surface mation as a result of repeated contact, such as a
lubricant consisting of a stearate soap with 20% is obviously undesirable and must be prevented ratchet mechanism. Corrosive wear processes
inorganic and 12% EP additives (Ref 30). with proper lubrication. are those associated with the loss of wear of in
Copper annealed in a protective atmosphere can Wear generally follows Eq 20 and increases lin- situ formed reaction product (e.g., oxide layers).
be drawn without descaling. early with the length (or weight) of wire drawn. The relative wear rates of these mechanisms
Under steady-state conditions, the wear volume can differ by several orders of magnitude, depend-
(V) inversely proportional to the materials harding on the wear situation. Table 2 shows the ob-
Wear ness (H) and is proportional to the distance (l) or served range of a dimensionless wear coefficient
time of sliding with a normal load (P) such that: for each of these mechanisms under sliding condi-
One can compensate for wear in rolling by re- tions. This coefficient is obtained by dividing the
setting the roll gap. No such opportunity exists amount of wear by the distance of sliding and the
in drawing through a die. Thus, wear results in V P
=K load and multiplying by the hardness of the wear-
an increase in the dimension of the product, and l H (Eq 20)
ing surface. The value for each of the mechanisms
the tolerances set an absolute limit on the allow- is based on data obtained from tribosystems in
able wear. The geometry of the converging hole where K is some constant of the wear situation. which that mechanism is considered to be the pre-
also changes, and thus the lubrication mecha- Compilations of the K coefficient are given by dominant one. The higher the value, the more se-
nism is affected, lubricant breakdown can occur, Rowe (Ref 32) and Rabinowicz (Ref 33), but vere the wear situation. Generally, engineering ap-
and the product finish suffers. Therefore, mini- minor variation in the interface conditions can plications require that this coefficient be in the
mization of wear is a principal preoccupation of have a much more marked effect on the wear rate range of 105 or less. In some cases, very small
tribology in drawing. than on friction and the K (for ostensibly identi- values are required. Table 2 shows that the most
In a typical die, wear occurs at three major cal conditions). The chief difficulty of wear the- severe forms of wear are associated with adhesive
points (Fig. 18): ories is the interpretation and prediction of K for and abrasive wear mechanisms.
various wear situations and mechanisms. All wear mechanisms can be active in a given
In general terms, three general types of wear sit- situation, and the foregoing mechanisms are not
uations are common, as noted in Ref 34. The first mutually exclusive. They can coexist and interact
occurs when two solid bodies are in contact and to form more complex wear processes, and when
move relative to one another. This first situation is worn surfaces are examined, features indicative of
generally subdivided by the predominant nature of more than one mechanism are usually found.
the relative motion, such as sliding, rolling, or im- However, in most cases, one type of mechanism
pact. The second situation occurs when the wear is
caused by a liquid moving relative to a solid sur-
face. Wear in that situation is often called erosion Table 2 Sliding-wear coefficients
or erosive wear. The third situation, which is gen- Sliding wear
erally referred to as abrasive wear, occurs when Wear mechanism coefficient(a)
the wear is caused by hard particles. Erosion and Adhesive wear 107 to 101
abrasive wear situations can also be subdivided Abrasive wear 106 to 101
into more specific categories. Examples of these Corrosive war 107 to 102
are cavitation erosion, solid-particle erosion, Fatigue wear 106
gouging abrasion, and slurry erosion. (a) (V H)/(P S), where V is the volume of wear, H is the hardness
(load per unit area), P is the applied load, and S is the sliding distance.
Fig. 18 Section of typical worn wire-drawing die. The actual mechanisms of wear are complex Source: Ref 34
Source: Ref 31 and may vary in extent depending on the nature
diameter, a given amount of absolute wear rep- steels are still used for sections and large-diam-
resents a greater percentage increase for a thin- eter tubes and bars, as they are for roller dies and
ner wire. This means that more wear-resistant rolls. They could, no doubt, be improved with
die materials must be found for finer wires if tol- coatings, but some of the incentive for doing so
erances are to be held. Wear rates are expressed has disappeared with the ready availability of
as the length of wire that can be drawn before large carbide dies. Hard-chromium plating is
the increase in wire diameter reaches 1 mm. commonly used for plugs. Boronizing has been
Typical values are: 1 to 2 km (0.6 miles) for low- reported to give better results. Ceramic coatings,
carbon steel with die steel dies, 10 km (6 miles) particularly ZrO2, look promising for drawing of
for patented wire, and up to 500 km (310 miles) stainless steel. Pure zirconia plugs have also
for low-carbon steel wire with tungsten carbide been used.
Fig. 19 Effect of wear rate on die life for a given toler-
dies. Lives are 10 to 200 times longer for dia-
ance range. Source: Ref 36
mond dies (e.g., 1.6 105 km, or 100,000 miles,
of copper wire may be drawn before repolish- Conclusions
tends to predominate and ultimately be the con- ing). All die materials have their places; discus-
trolling one. In wire drawing, wear mechanisms sions of various types are given in the steel-wire No friction between workpiece and die is
have been studied extensively by Wistreich (Ref handbooks. needed in wire drawing and tube drawing on a
31), and the applications of wear theory have been Die life is also increased by minimizing the fixed plug; some moderate friction is required on
reviewed by Shatynski and Wright (Ref 35). There presence of abrasive oxides and removing debris the plug in drawing tubes with a floating plug,
is a phase of early rapid wear (Fig. 19), which from the lubricant. The surface finish obtained and friction on the bar is beneficial in drawing
Papsdorf (Ref 36) attributed to removal of the as- on finish polishing is equally important and on a bar.
perities in a new die, and an acceleration in the last must be measured. Early polishing is important The frictional force is additional to the force
stage, which is associated with heavy ringing. (particularly for the prevention of incipient ring- required to perform the deformation (work of
In general, the following wear mechanisms ing), and checking the die profile is needed to homogeneous deformation) and the force arising
are most active: ensure lubricant film formation. Cast silicone from the inhomogeneity of deformation (redun-
Abrasive wear is unavoidable. Most metals rubber replicas allow inspection without special dant work). The frictional force increases,
carry a harder oxide; even if the heavy oxide equipment. A drastic improvement is obtained whereas the force for inhomogeneous deforma-
inherited from hot working is removed, fresh by the use of roller diesan improvement that is tion decreases, with increasing die angle, and
oxide always forms. Some lubricants and equivalent to a tenfold to 100-fold increase in therefore an optimal die angle is found. Because
coatings are mildly abrasive; others contain die life. Wear often begins as an angular ring on the contributions of friction and inhomogeneous
abrasive constituents (e.g., SiO2 or Al2O3 in the approach angle of the die. Die life may be in- deformation to the total force are of similar mag-
lime). Wear debris in wet drawing, unless creased by as much as 200% if the die is re- nitude, back-calculation of friction from the
continuously removed, creates a three-body moved and repolished at the first appearance of total draw force is difficult. However, for a given
wear condition. In a simulation test with a this ring (Ref 38). workpiece material and process geometry, a
copper disk rotating against a diamond flat, Table 3 (Ref 38) gives recommended materi- larger draw force is always indicative of higher
Harper et al. (Ref 37) found wear to double als for wire-drawing dies. For round wire, dies friction.
when a new lubricant was recirculated and to made of diamond or cemented carbides are al- Yielding takes place under the combined ef-
increase sevenfold when Cu2O particles were ways recommended without regard to composi- fects of axial tension and radial compression,
added. A used lubricant, in which fines were tion or quantity of the metal being drawn. For and interface pressures are below the flow stress
entrapped in the soap-fat emulsion, gave sim- short runs or special shapes, hardened tool steel of the material except at low die angles, when
ilarly high rates of wear. is less costly, although cemented carbides give they can rise to higher values. Even though pres-
Adhesive wear is likely whenever metal-to- superior performance in most applications. Die sures are, normally, not excessive, the process
die contact occurs, although the severity of
wear is highly dependent on the specific
workpiece-die combination.
Surface fatigue wear has been suspected as Table 3 Recommended materials for wiredrawing dies
the main culprit in ringing. The high stress Wear size Recommended die materials for:
gradient, even if loading is continuous, is Metal to be drawn mm in. Round wire Special shapes
damaging in itself and is aggravated by vi-
bration of the wire. It has also been found Carbon and alloy steels <1.57 <0.062 Diamond, natural or M2 or cemented
synthetic tungsten carbide
that entering of the wire at an angle greatly >1.57 >0.062 Cemented tungsten
increases wear rates; therefore, accurate carbide
alignment and elimination of vibration by Stainless steels; titanium, <1.57 <0.062 Diamond, natural or M2 or cemented
guide pulleys reduce wear. tungsten, molybdenum synthetic tungsten carbide
and nickel alloys >1.57 >0.062 Cemented tungsten
Thermal fatigue wear can lead to crazing of carbide
steel dies but is, generally, not a major prob- Copper <2.06 <0.081 Diamond, natural or D2 or cemented
lem. synthetic tungsten carbide
>2.06 >0.081 Cemented tungsten
Catastrophic failure (fracture) of dies is usually carbide
due to overloading. The danger is greater with Copper alloys and <2.5 <0.100 Diamond, natural or D2 or cemented
lower die angles, and brittle, hard dies must be aluminum alloys synthetic tungsten carbide
>2.5 >0.100 Cemented tungsten
encased in steel rings. carbide
Magnesium alloys <2.06 <0.081 Diamond, natural or
synthetic
Die Wear >2.06 >0.081 Cemented tungsten
carbide
Because die wear is proportional only to the Source: Ref 39
length of wire drawn and is not related to wire
presents severe tribological conditions because drawing speed, especially for lubricants operat- 18. Recent Progress in Metalworking, Ameri-
sliding prevails over the whole contact zone, and ing in the mixed-film regime. can Elsevier, 1964
heating due to deformation and friction raises 19. B. Avitzur, Handbook of Metalforming
the temperature. In multihole drawing the high Processes, John Wiley, 1983
temperature of the entering wire aggravates the ACKNOWLEDGMENT 20. M.J. Hillier, A Hydrodynamic Lubrication
situation. in Wire Drawing, Int. J. Prod. Res., Vol 5,
Lubricants are applied to reduce friction, This article was adapted from substantial con- 1967, p 171
wear, and temperature. A good lubricant reduces tent contained in Ref 1 and 2. Additional content 21. G.H. Tattersall, Theory of Hydrodynamic
temperature by reducing friction. Nevertheless, was adapted from Wire, Rod, and Tube Lubrication in Wire Drawing, Report
in high-speed multihole drawing the lubricant Drawing, Forming and Forging, Volume 14, MW/D/46/59, British Iron and Steel
must also have adequate cooling capacity to re- ASM Handbook, 1988, p 330342. Research Association
move heat from the product. The method of 22. A. Bobrowsky, Pressure Box, U.S. patent
coolant application becomes most important. 3,417,589, 24 Dec 1968
Wear is brought within tolerable limits, in part, REFERENCES 23. E. Felder, Elastohydrodynamics and
by the selection of highly wear-resistant die ma- Related Topics, Proc. Fifth Leeds-Lyon
terials, primarily tungsten carbide and diamond. 1. J.A. Schey, Tribology in Metalworking, Symposium on Tribology, D. Dowson et al.,
Drawing without a lubricant would result in im- Friction, Lubrication, and Wear, ASM Ed., Mechanical Engineering Publications,
mediate pickup. So-called dry drawing is con- International, 1983 London, 1979, p 365369
ducted with a nonliquid lubricant, usually a soap, 2. Drawing of Rod and Wire, R.N. Wright, 24. O. Pawelski, W. Rasp, and T. Hirouchi, in
whereas wet drawing is performed with viscous Encyclopedia of Materials Science and Tribologie, Springer, Berlin, 1981, p
oils or aqueous emulsions. The distinction is prac- Engineering, Vol 2, M.B. Bever, Ed., 479506 (in German)
tical rather than fundamental: under the conditions Pergamon Press and the MIT Press, 1986, p 25. V.L. Kolmogorov, S.I. Orlov, and K.P.
prevailing at the interface, soap can be modeled as 12271231 Selischev, Drawing under Conditions of
a viscous fluid of temperature- and strain-rate-sen- 3. T.Z. Blazynski, Metal Forming, Tool Hydrodynamic Lubrication, National
sitive viscosity, or as a Bingham substance. Profiles, and Flow, Halstead Press, 1976 Lending Library for Science and
Drawing is a steady-state process with a fa- 4. B. Avitzur, Metal Forming: Process and Technology, Boston Spa, Yorkshire,
vorable entry geometry for the entrainment of Analysis, McGraw-Hill, 1968, Krieger, re- England, 1968
lubricants. Full-fluid-film lubrication can be en- vised 1979 26. T.W. Rees, J. Appl. Metalwork., Vol 1 (No.
couraged by special die constructions that aid in 5. B. Avitzur, Friction During Metal Forming, 4), 1981, p 5357
developing hydrodynamic films or that allow the Friction, Lubrication, and Wear Technol- 27. J. Bernot, Wire J., Vol 12 (No. 12), 1979, p
introduction of externally pressurized fluids. ogy, Vol 18, ASM Handbook, 1992, p 5969 881
The most practicable of these is the pressure die 6. B.B. Hundy and A.R.E. Singer, J. Inst. Met., 28. W. Schmidt, Stahl Eisen, Vol 91, 1971, p
for soaps; it builds up sufficient pressure by Vol 91, 19621963, p 401407 13741381 (in German)
purely hydrodynamic action to cause yielding of 7. M. Mohamdein and T. Vinh, CIRP, Vol 25, 29. H. Kudo, S. Tanaka, K. Imamura, and K.
the wire prior to entry into the die. The surface 1976, p 169172 Suzuki, CIRP, Vol 25, 1976, p 179184
finish of the workpiece produced under full- 8. B. Avitzur, Wire J., Vol 7 (No. 11), 1974, p 30. O. Miki, S. Fuji, and M. Vemura, Wire J.,
fluid-film conditions is rough. 7786 Vol 12 (No. 5), 1979, p 5660
Most practical drawing is conducted under 9. R.W. Johnson and G.W. Rowe, Proc. Inst. 31. J.G. Wistreich, Met. Rev., Vol 3, 1958, p
mixed-film conditions. Initial draws may be Mech. Eng., Vol 182, 19671968, p 521 97142
made with a dry lubricant, but brighter, 526 32. C.N. Rowe, Wear Control Handbook, M.B.
smoother surfaces can be produced only by de- 10. R.W. Gottschlich and N.N. Breyer, J. Met., Peterson and W.O. Winer, Ed., American
liberately thinning the film to the point where Vol 15, 1963, p 364367 Society of Mechanical Engineers, 1980, p
boundary lubrication predominates. This is ac- 11. G. Properzi, Wire J., Vol 12 (No. 12), 1979, 143160
complished by increasing the die angle (al- p 5862 33. E. Rabinowicz, Wear Control Handbook,
though this increases also the inhomogeneity of 12. R. Sayer and R. Moller, Wire J., Vol 11 (No. M.B. Peterson and W.O. Winer, Ed.,
deformation and could lead to internal arrow- 2), 1978, p 6872 American Society of Mechanical Engineers,
head fracture) and by drawing with a low-vis- 13. N. Koichiro, M. Tawara, Y. Hayase, T. 1980, p 475506
cosity lubricant. The potential for die pickup ne- Katsube, M. Kyoda, and H. Nakagawa, 34. R.G. Bayer, Design for Wear Resistance,
cessitates the use of boundary or EP additives. Effective Production in the Erhardt Push Materials Selection and Design, Vol 20,
Solid film lubrication is of greatest importance Bench Pipe Manufacturing, Tetsu-to- ASM Handbook, 1997, p 603614
in warm and hot drawing and in cold drawing of Hagane (in Japanese); J. Iron Steel Inst. 35. S.R. Shatynski and R.N. Wright, Wire
adhesion-prone alloys at low speeds, in which an Jpn., Vol 78 (No. 11), Nov 1992, p Technol., Vol 7 (No. 4), 1979, p 5962, 6669
effective fluid film cannot be maintained. T209T212 36. W. Papsdorf, Stahl Eisen, Vol 72, 1952, p
Tube drawing on a fixed or floating plug pres- 14. O. Pawelski and V. Rdiger, Arch. 393399 (in German)
ents special difficulties of lubricant application Eisenhttenwes., Vol 47, 1976, p 483487 37. S. Harper, A.R. Goreham, and A.A. Marks,
to ensure effective lubrication of the inner tube (in German) Met. Mater., Vol 4, 1970, p 335339
surface. 15. G. Dvnyi, Draht, Vol 13, 1962, p 223231 38. The Selection of Materials for Tools for
Many problems of lubrication and wear can (in German) Drawing, Wire, Bar, and Tubing, Properties
be bypassed by replacing the draw die with a ro- 16. A.B. Dove, Steel Wire Handbook, The Wire and Selection of Metals, Vol 1, 8th ed.,
tary die composed of two to four rollers, partic- Association International, 1980 Metals Handbook, 1961, p 752
ularly in drawing of more ductile materials. 17. A.F. Gerds and F.W. Boulger, Rod, Wire and 39. The Selection of Materials for Tools for
Lubricant recommendations are summarized Tube Drawing, DMIC Report 226, Metal Drawing, Wire, Bar, and Tubing, Properties
in Table 1. For preliminary calculations, coeffi- Deformation Processing, Vol II, Defense and Selection: Stainless Steels, Tool
cient of friction values are given with the under- Metals Information Center, Battelle Materials, and Special-Purpose Metals, Vol
standing that their magnitude is a function of Memorial Institute, 7 July 1966, p 78124 3, 9th ed., Metals Handbook, 1980, p 522
Chapter 20
Extrusion*
EXTRUSION is a deformation process used of mean hydrostatic pressure (sm) and flow stress scribed in more detail in the section
to produce long, straight, semifinished metal (s0). This is also borne out from practice, because Hydrostatic Extrusion in this chapter).
products such as a bar, solid and hollow sec- alloys such as tin bronze (CuSn8) and free-cutting Hydrostatic extrusion also allows greater reduc-
tions, tubes, wires, and strips. Practically all aluminum (AlCuMgPb) are almost impossible to tions in area (higher extrusion ratios) than con-
metals can be extruded, although the extrudabil- hot roll without cracking but can be extruded with ventional extrusion. Hydrostatic extrusion is
ity of many alloys (including many extrudable a perfect finish (Ref 1). Nonetheless, successful done hot or cold, but hot hydrostatic extrusion is
alloys of aluminum and magnesium) falls within production by extrusion frequently requires a nar- not very commonif done at all.
a narrow range of processing conditions. row range of processing conditions that balance Conventional extrusion can be done hot, warm,
Extrudability depends on a variety of applica- extrusion temperature with extrusion speed, as or cold, depending on the work material and the
tion-specific factors, such as: briefly described in this chapter. size and shape of the desired product. For exam-
This chapter also briefly introduces the meth- ple, aluminum alloys are very suitable for eco-
Mechanical workability (W), or resistance to
ods and mechanics of extrusion, with an emphasis nomical production of parts by cold (i.e., room-
cracking with plastic deformation, defined in
on the unique metalworking aspects of extruding. temperature) extrusion, even when parts are large
broad terms as: W jf /s0 (where jf is de-
For example, one unique aspect of extrusion is the or complex in shape. Cold extrusion of steel is typ-
formation to fracture, and s0 is flow stress)
complex role of friction, which may be beneficial ically confined to relatively smaller parts (with
Extrusion temperature
or detrimental, depending on the conditions of starting plugs seldom more than 10 kg, or 25 lb),
Permitted temperature range
metal flow (see the section Friction and depending on the availability of equipment, tool
Load or specific pressure required (which
Lubrication in this chapter). Extrusion is, with materials, and the plasticity of the steel. Cold ex-
depends on the work material and geometry
few exceptions, a batch process that operates trusion is always conducted with an effective lu-
of the extruded section)
under discontinuous conditions. In extrusion, a bricant. In contrast, hot extrusion may be done by
Exit speed at a constant load
second billet is not loaded until the first has been either unlubricated or lubricated methods (Fig. 4),
Maximum extrusion speed (to the onset of
extruded. Non-steady-state conditions occur at the depending on the extrudability of the work mate-
hot shortness)
beginning and end of the extrusion process, and rial (see Hot Extrusion in this chapter). Hot ex-
Maximum extrusion ratio (RE), which is the
flow is nonsteady due to temperature variations trusion is used to produce long, straight metal
ratio of the initial cross-sectional area (A0)
during the extrusion cycle and variations of tem- products of constant cross section, such as bars,
to the final cross-sectional area (A1) after
peratures and friction over the length of the billet. solid and hollow sections, tubes, wires, and strips,
extrusion (RE = A0/A1)
Thus, nonuniform flow and property variations from materials that cannot be economically
Special techniques to extrude different mate- occur over the length of the part, as briefly deformed by cold extrusion.
rials are, to a large extent, dependent on the ex- scribed in the section Non-Steady-State Effects The mechanical process of extrusion is further
trusion temperature. Cold (room-temperature) and Defects in this chapter. classified as either direct or forward extrusion
extrusion is economically feasible for low-melt- (Fig. 5a) or indirect or backward extrusion (Fig.
ing alloys and some aluminum alloys, while hot 5b). In some cases, the extrusion process may be
extrusion is required for harder alloys. Figure 1 Extrusion Methods designed as a combined (direct and indirect) ac-
(Ref 1) is a general classification of alloys in tion. In forward (direct) extrusion, the die and
terms of extrusion temperature. The process of extrusion can be done by sev- ram are at opposite ends of the extrusion billet,
Metal flow from extrusion through a conical eral different methods, which can be generally and the product and ram travel in the same di-
die is analogous to drawing (described in Chapter classified as either conventional extrusion or hy- rection. Also, there is relative motion between
19, Drawing of Wire, Rod, and Tube, in this drostatic extrusion. In conventional extrusion, a the extrusion billet and the die. In backward (in-
Handbook), although there are some important billet (typically in a container to support the direct) extrusion, the die is on a moving ram, and
differences in these processes. One basic differ- workpiece under high extrusion pressures) is the product travels in the direction opposite that
ence is that the workpiece during drawing is forced through the die by a ram acting directly of the ram, either around the ram (as in the im-
pulled through a die, while during extrusion, the on the workpiece or billet (Fig. 3a). In contrast, pact extrusion of cylinders) or up through the
workpiece is either pushed through a die (direct hydrostatic extrusion is performed by liquid center of a hollow ram.
extrusion) or a die is pushed over a stationary pressure (Fig. 3b) rather than by direct applica- The shape of the part is usually the primary
workpiece (indirect extrusion). The favorable tion of the load with a ram. In hydrostatic extru- factor that determines the use of direct or indi-
compressive state of stress during extrusion al- sion, the work material is completely sur- rect extrusion. For example, many cuplike parts
lows a high capacity for deformation, and so it is rounded by a fluid, which is sealed off and is are produced by backward extrusion, while
possible to extrude metals that can be only pressurized sufficiently to extrude the billet shaftlike parts and long, hollow shapes can usu-
slightly deformed by other methods. This is through the die. Hydrostatic extrusion can be ally be produced more easily by forward extru-
shown schematically in Fig. 2, where the defor- used to extrude brittle materials that cannot be sion. For many shapes, both forward and back-
mation to fracture (jf) is plotted against the ratio processed by conventional extrusion (as de- ward extrusion are used. Other factors that
*Portions adapted from publications listed in the Acknowledgments at the end of this chapter.
Fig. 1 Classification of extruded alloys on the basis of temperature. Source: Ref 1
influence procedure are the composition and cal) dies (Fig. 6). Spider or bridge dies also are been developed to reduce the costs of designing
condition of the work material, the required di- used for producing hollow extrusions such as and manufacturing extrusion dies, and a few ex-
mensional accuracy, quantity, and cost. tubing. Flat-face dies (also termed square dies) amples are noted in Ref 2 to 4. Computer simu-
have one or more openings (apertures) that are lation programs also have been developed to
Dies similar in cross section to that of the desired ex- model the metalworking process to more effec-
truded product. Dies for lubricated extrusion tively select process variables, such as extruding
The tooling for extrusion consists of such (also called shaped, converging, or streamlined speed and billet temperatures. However, finite-
components as containers, container liners, dies) often have a conical entry opening with a element modeling of deformation by extrusion
stems (rams), dummy blocks, mandrels, and circular cross section that changes progressively can be complicated, because small variations in
dies. There is considerable variation in the tool- to the final extruded shape required. Flat-face die shape, friction, or temperature distribution
ing practice and design, depending on the pro- dies are easier to design and manufacture than may result in a large change in the profile of the
cessing temperature, lubrication, loading re- shaped dies and are commonly used for the un- section. Moreover, deformation is complicated
quirements, and whether the process is lubricated hot extrusion of aluminum alloys. by the presence of quasi-shear surfaces close to
backward or forward extrusion. Thus, tooling is Shaped dies are more difficult and costly to de- the die exit (Ref 5). For these reasons, very few
described in subsequent sections for the various sign and manufacture, and they are generally computer codes are available for three-dimen-
types (direct/indirect) and processing conditions used when lubrication is required; this includes sional simulation; most extrusion models are for
(hot/cold) of extrusion. However, one of the cold extrusion and hot extrusion of harder alloys axisymmetric problems.
most critical aspects of tooling is die design. Die such as steels, titanium alloys, and nickel alloys. Luckily, some internal equalization takes
design is a crucial part of successful extrusion The design of extrusion dies, whether of the place in the product, and the process is fairly for-
that depends very much on the judgment, skill, flat or conical type, is still an art rather than a giving; in an example of computer-aided die de-
resourcefulness, intuition, and experience of the science. Nonetheless, rational design techniques sign for extrusion of a T-section, material flow
individual die designer and maker. and empirical guidelines are employed, often was much more uneven than anticipated, and
In general, the two most common types of ex- with the assistance of computer programs. yet, the extruded product was sound (Ref 6). The
trusion dies are flat-face dies and shaped (coni- Computer-aided design and manufacture has increased frictional resistance along the longer
Chapter 20: Extrusion / 293
Fig. 3 Comparison of (a) conventional extrusion and

(b) hydrostatic extrusion
Fig. 2 Deformation to fracture, jf, versus the ratio of mean hydrostatic stress, sm = 1/3 (s1 + s2 + s3), and flow stress,
s0, in relation to the point of fracture for (a) various deformation processes and (b) in tension, compression,
and torsion testing. kf = s1 - s3 (see Fig. 8b). Source: Ref 1
die perimeter and the more complex flow in- and die deflection (both dies and deep
Fig. 4 Schematic of conventional hot extrusion (a)
crease the punch pressures relative to extrusion tongues) without lubrication and (b) with lubrication
of a round bar of equal A1 cross-sectional area. Optimization of bearing lengths to increase
General Die Design Considerations. The productivity dressed here. The key is to have a close working
basic considerations of die design for direct ex- relationship between the designer, die maker,
trusion are (Ref 7): However, optimal design depends on a large press operator, and die corrector.
number of factors, including the size of the shape Most design-for-manufacturing considerations
The number of die openings based on the to be produced, the maximum and minimum wall for extrusions involve the difficulty of forcing the
shape and size of the profile and the nature thickness, the press capacity, the length of the metal to flow uniformly from a large, round billet
of exiting tooling runout table, the stretcher capacity, the tool- through small, complex die openings. This leads
Location of die openings relative to the bil- stacking limitations, an understanding of the to some general rules that can help guide the de-
let axis properties and characteristics of the metal to be sign process. For example, even for the more eas-
Orientation of the openings around their extruded, and the press operating procedures and ily extrudable aluminum parts, sections with both
centroids to match the handling system maintenance. Complete treatment of these fac- thick and thin sections are to be avoided. Metal
Determination of the final openings based tors is beyond the scope of this Handbook, so flows faster through a larger opening than a
on thermal shrinkage, stretching allowance, only some very basic factors in die design are ad- smaller one, and so flow is faster where thicker
bearing length needs to be smaller to balance

flow in a thinner section. The treatment of bear-
ing surface at the front and back of the die aper-
ture is termed the choke and relief, respectively
(Fig. 7) (Ref 7). A choke can be provided on cer-
tain portions of the bearing surface if the die de-
signer anticipates difficulty in filling sharp cor-
ners or completing thin sections of the extruded
product. This slows the rate of metal flow and
Fig. 5 Basic methods of conventional extrusion. (a) Forward (direct). (b) Backward (indirect). 1, billet; 2, container;
consequently fills the die aperture. For direct ex-
3, die; 4, stem; 5, dummy block; 6, die backer
trusion of hard aluminum alloys, the front of the
bearing is generally choked at an angle up to 3
(Ref 7). Increasing the amount of back relief at
the exit side of the bearing surface increases the
rate of metal flow by decreasing the original
bearing length.
The Extrusion Process (Ref 8)
In common with wiredrawing, the process of

extrusion can be described by metal flow
through a conical channel (Fig. 8). The starting
material is usually a cylindrical, cast, or previ-
ously extruded or rolled billet that is placed into
a container and pressurized, either by means of
a punch attached to the ram of a hydraulic or me-
chanical press (direct extrusion) or by a die at-
tached to a moving ram (indirect extrusion).
Fig. 6 Two basic types of extrusion dies. (a) Shaped (or conical-type) dies for lubricated extrusion. (b) Flat (or square- Deformation in extrusion tends to be nonuni-
type) dies typical of unlubricated hot extrusion
form (described more later), but it is still cus-
The next step in the design process is the de- tomary to calculate an average strain as either re-
sections occur, which thus results in distortions in
termination of die-bearing lengths. The die bear- duction, r, in area:
the extruded shape. This must be compensated for
in the design of dies. For example, when a section ing at any position is dependent on the section
to be extruded has a thick wall and a thin wall, var- thickness at that position and on its distance A0 - A1
r=
ious means are employed to retard metal flow from the die center. In the direct extrusion A0 (Eq 1)
through the thick section and to increase the flow process, the frictional resistance at the billet-
rate through the thin section of the die. container interface slows down metal flow near
or extrusion ratio:
Long, thin wall sections should also be the billet surface, and so the center of the billet
avoided, because such shapes are difficult to moves faster than at the periphery. To balance
keep straight and flat. If such sections are ab- flow, bearing length must be inversely propor- A0
RE =
solutely necessary, then the addition of ribs to tional to its distance from the surface. Similarly, A1 (Eq 2)
the walls help distribute the flow evenly. Hollow
sections are quite feasible, and the designer can
select various types of dies for hollow sections.
These are discussed in the section Tube
Extrusion in this chapter. It is best if hollow
sections can have a longitudinal plane of sym-
metry. Semihollow features may also be consid-
ered but should be avoided, because a semihol-
low feature requires the die to contain a very
thinand hence, relatively weakneck.
After die layout, tool strength analysis is also
needed to determine correct openings with die
cave and die deflection from the bending and
shear stresses in the die, backer, and bolster due
to the extrusion pressure. The need for conform-
ing tools to support the die is also determined by
the tool strength analysis. Die deflection analy-
sis includes estimation of the bending of the die
at various locations. This information is helpful
in predicting the dimensional changes in the die
orifices under load during extrusion. The dimen-
sions of the die orifices can be modified to cor-
rect for these dimensional changes to obtain the Choke and relief in die bearing. (a) Choke at front of bearing. (b) Increased relief angle at the back or exit side
Fig. 7
desired tolerances in the extruded shape. of the bearing. Source: Ref 7
Fig. 8 Forward extrusion (a) with conical die entry (half arrows indicate interface shear stresses acting on the material. (b) Schematic variation of axial pressure, s3, and radial
pressure, s1 = s2, over the length of the container for lubricated and unlubricated extrusion. kf = s1 - s3
which conveniently leads to the true strain, e: teract. Deformation becomes more homoge- tending the range over which sound extrusions
neous with a decreasing h/l ratio (Fig. 8), which can be obtained (Fig. 10) (Ref 10).
A0 implies, for a constant reduction, a smaller die On the basis of pressure considerations, a
e = ln RE = ln
A1 (Eq 3) half-angle, a. This results, however, in a longer conical die entry would often be optimal.
surface over which friction must be overcome. However, it has the disadvantage that the press
The extrusion ratio can reach very high values, Conversely, a larger die angle reduces the fric- stroke must be arrested before the punch can
in excess of 400 to 1. tional components but leads to increased inho- touch the die, thus increasing material lost in the
Extrusion Pressure. Most extrusion theo- mogeneity; therefore, as in wiredrawing, an op- remnant (butt). Furthermore, the extrusion must
ries lead to solutions in which the extrusion timal die angle is found, the value of which be pushed back out of the die if the remnant is to
pressure pe at the end of the stroke is propor- increases with increasing extrusion ratio and be cut off. Alternatively, the remnant may be
tional to strain: friction (Fig. 9) (Ref 9). pushed through with a deformable follower
Flow becomes truly streamlined only when block (e.g., a graphite block), or another billet
friction is very low. It can be shown that die must follow the first billet without removal of
pe = s 0 (a + be) (Eq 4)
shapes other than simple cones ensure even the remnants. In the latter case, the interfaces be-
greater homogeneity of flow, especially with high tween the billets must be clean to facilitate weld-
where s0 is the mean flow stress of the material, friction. With low friction, the improvement does ing and prevent the development of internal de-
and the parenthetical term allows for the pres- not, in general, justify an extra expense in fects, and thus the use of a lubricant is
sure required for uniform deformation, the inho- diemaking, even though with numerically con- precluded. For practical reasons, extrusion is
mogeneity of deformation (redundant work), trolled machining this expense may be slight. often conducted with a flat-face die (a = 90),
and die-face friction. The constants a and b may Stresses in the deformation zone are fully especially at heavier reductions where there is
be calculated or determined from experiments. compressive when h/l > 1, and, for this reason, no danger of a centerburst defect.
For very approximate calculations, it may be as- extrusion is particularly favorable for working of Extrusion Force. The pressure calculated
sumed that a = 0.8 and b = 1.2. materials of limited ductility. If, however, the ex- from Eq 4 applies to the container cross-sec-
As in other processes, the effects of inhomo- trusion ratio is low and the die angle is large tional area A0, and thus the extrusion force is:
geneity of deformation and interface friction in- (and thus h/l is large, too), a hydrostatic tensile
stress can develop in the center of the extrusion,
Pe = pe A0 (Eq 5)
and a centerburst defect similar to that found in
wiredrawing occurs. Because increasing die
friction increases the pressure required for ex- For measurement of this net extrusion force, a
trusion and thus generates a hydrostatic pres- billet may be pushed through the extrusion die
sure, higher friction is actually desirable in ex- without the support of a container (Fig. 11a). To
avoid upsetting of the unsupported billet, the
pressure pe must be less than the flow stress (s0)
of the workpiece material. Thus, the effects of
friction and die angle can be explored but only
for a rather limited range of reductions. This
technique is, nevertheless, suitable for lubricant
ranking. In order for the interface shear strength
ti (or, if preferred, m or friction factor, m) to be
derived, the value of the average or mean flow
stress (s0) must be accurately known. Alterna-
tively, the axial and normal (radial) forces can be
simultaneously measured. Thus, the split die
principle was employed for axial symmetry (Ref
11) and for plane strain (Ref 12).
If heavier reductions are to be explored, a
conventional extrusion configuration is needed.
It is possible, however, to measure separately the
forces acting on the extrusion die by leaving a
small, well-lubricated clearance between die and
Fig. 9 Effect of lubrication on optimal die angle in for- Fig. 10 Effect of friction factor (m) on danger of cen- container, so that no fin is extruded, but free
ward extrusion of aluminum. Source: Ref 9 terburst in extrusion. Source: Ref 10 movement of the die is ensured even under load
Flow pattern C (Fig. 13d) occurs in the hot

extrusion of materials having inhomogeneous
properties when the friction is high (as in flow
pattern B) and when the flow stress of the mate-
rial in the cooler peripheral regions of the billet
is considerably higher than that in the center.
The billet surface forms a relatively stiff shell.
Therefore, the conical dead-metal zone is much
larger and extends from the front of the billet to
the back. At the start of extrusion, only the ma-
terial inside the funnel is plastic, and it is se-
Fig. 11 Measurement of die friction by (a) containerless extrusion and (b) separation of die force verely deformed, especially in the shear zone, as
it flows toward the die. The stiff shell and the
(Fig. 11b). Such measurements have been found Flow pattern A (Fig. 13b) occurs in homoge- dead-metal zone are in axial compression as the
to be reproducible (Ref 13, 14). Load cells for neous materials when there is virtually no fric- billet length decreases; consequently, the dis-
punch and die loads and strain gages on the con- tion between the container and the billet but sig- placed material of the outer regions follows the
tainer were used for measurement of internal nificant friction at the surface of the die and its path of least resistance to the back of the billet,
pressure in cold forward and backward extrusion holder. This retards the radial flow of the pe- where it turns toward the center and flows into
(Ref 15). ripheral zones and increases the amount of the funnel.
The extrusion force is limited by the potential shearing in this region. The result is a slightly This type of flow is found in the (a + b)
failure of the ram or container. A short ram (or larger dead-metal zone than that in flow type S, brasses, in which the cooling of the peripheral
punch) can fail by upsetting, a long one by buck- along with a correspondingly wider deformation regions of the billet leads to an increase in flow
ling. The container is an internally pressurized zone. However, deformation in the center re- stress, because the flow stress of the a-phase is
vessel that can be reinforced by a shrink-ring mains relatively uniform. Flow patterns of this much higher than that of the b-phase during hot
construction. Nevertheless, internal pressure is type are seldom observed in nonlubricated ex- working. As in the (a + b) brasses, flow type C
typically limited to 1.2 GPa (175 ksi) in hot extrusion; instead, they occur during the lubricated occurs when there is a hard billet shell and, at
trusion and to 2.5 GPa (360 ksi) in cold extru- extrusion of soft metals and alloys, such as lead, the same time, the friction at the container wall
sion. This means that for any given material, the tin, a-brasses, and tin bronzes, and during the is high. It can also occur without any phase
extrusion ratio is limited according to Eq 4 (Fig. extrusion of copper billets covered with oxide change that leads to a higher flow stress if there
12) (Ref 16). (which acts as a lubricant). is a large temperature difference between the bil-
Typical Flow in Extrusion (Ref 1). Metal Flow pattern B (Fig. 13c) occurs in homoge- let and the container. This can take place in the
flow varies considerably during extrusion, de- neous materials if friction exists at both the con- extrusion of tin as well as of aluminum and its
pending on the material, the material-tool inter- tainer wall and at the surfaces of the die and die alloys.
face friction, and the shape of the section. Fig. holder. The peripheral zones are retarded at the Heat Generation (Ref 8). The work of
13 shows the four types of flow patterns that billet-container interface, while the lower resist- plastic deformation and friction is converted into
have been observed. ance causes the material in the center to be ac- heat. Because strains can be very large and fric-
Flow pattern S (Fig. 13a) is characterized by celerated toward the die. The shear zone between tional energy is expended in a thin surface layer,
the maximum possible uniformity of flow in the the retarded regions at the surface and the accel- heat generation in extrusion is substantial. With
container. Plastic flow takes place primarily in a erated material in the center extends back into the rapid extrusion, most of the heat is carried away
deformation zone directly in front of the die. The billet to an extent that depends on the extrusion by the extruded product, but heat is conducted
major part of the nonextruded billet, pushed as a parameters and the alloy. Therefore, the dead- back into the undeformed part of the billet when
rigid body through the die, remains undeformed; metal zone is large. At the start of extrusion, the extrusion is relatively slow. In both cases, the
therefore, the front of the billet moves evenly shear deformation is concentrated in the periph- butt end of the extrusion is heated more than the
into the deformation zone. eral regions, but as deformation continues, it ex- nose end. Preheating temperature in hot extru-
tends toward the center. This increases the danger sion is usually high, partly to reduce mean flow
of material flowing from the billet surfacewith stress (s0) and thus allow a large extrusion ratio,
impurities or lubricantalong the shear zone and partly to ensure adequate ductility. It is pos-
and finishing up under the surface of the extru- sible that the temperature rise due to deforma-
sion. In addition, the dead-metal zone is not com- tion and friction could heat the emerging prod-
pletely rigid and can influence, even if to a lim- uct above the melting point (or, in alloys, above
ited degree, the flow of the metal. Flow type B is the solidus temperature). The parallel die land,
found in single-phase (homogeneous) copper al- while contributing only a few percent to the ex-
loys that do not form a lubricating oxide skin and trusion force, is a major source of heating. Thus,
in most aluminum alloys. the temperature is highest on the surface, and
Fig. 12 Process limitations for hot extrusion arising

from container friction and workpiece heat- Fig. 13 Four types of flow patterns observed in the extrusion of metals. (a) Flow pattern S. (b) Flow pattern A. (c)
ing. Source: Ref 16 Flow pattern B. (d) Flow pattern C. See text for details. Source: Adapted from Ref 1
grain-boundary melting causes a disintegration

of the product in the form of the characteristic
fir-tree cracking (speed cracking). Grain
growth can also occur on the surface due to the
heat generation. Extrusion must be slowed down
to minimize the rate of heat generation (Fig. 12).
Alternatively, the billet can be unevenly pre-
heated, with a temperature gradient dropping
from front to back to compenstate for normal
temperature rise at the butt end of the extrusion.
Non-Steady-State Effects and Defects (Ref
8). Non-steady-state conditions always exist at
the beginning and end of the extrusion process.
Non-steady-state conditions are more predomi-
nate in extrusion of short billets for the purpose
of making parts, where end defects cannot be al-
lowed to develop, because the unextruded por-
tion of the workpiece forms part of the finished
component. In contrast, steady-state conditions
prevail during extrusion of long, semifinished
products such as bars, sections, and tubes. In this Fig. 14 Typical defects observed in cold extrusion of aluminum alloys. (a) Deep scoring. (b) Surface roughening
case, end defects may be allowed to develop to a from the presence of thick lubricant film. (c) Bambooing. (d) Shear cracks. (e) Machining marks stabilized
by lubricant. Source: Ref 18
limited degree, because the residue of the re-
maining billet in the container is discarded.
Initiation of Extrusion. A pressure peak defect of similar appearance can also be attrib- Another important aspect of extrusion is the
(breakthrough pressure) is observed just at the uted to the simultaneous occurrence of slip on complex role of friction. In wire drawing, the
point where extrusion is initiated. In lubricated the die face and sticking on the die land (Ref elimination of friction is desirable, because the
extrusion, it indicates that the lubrication mech- 17). introduction of friction promotes the formation
anism typical of steady-state conditions has not Other external defects include: of centerbursts during drawing (see Fig. 6a in
yet been established. Thus, it is most marked Chapter 19, Drawing of Wire, Rod, and Tube,
when a square-ended billet is extruded through a Subsurface defects caused by intermittent in this Handbook). In extrusion, however, fric-
tapered die; the sharp edge of the billet may ac- stick-slip on the container wall may emerge tion may deter centerbursts during extrusion.
tually scrape off whatever lubricant film is pres- at the surface in the form of laminations or Nonetheless, there also are a number of friction-
ent on the die, and if this occurs, a new, low-fric- flakes. related surface defects, from either hot or cold
tion layer can develop only after the die throat Pickup on the die or on the die land leads to extrusion, that can make a product unsuitable for
has been filled. deep scoring of the extruded surface (Fig. use. Friction during extrusion can lead to the de-
End of Extrusion. When extrusion proceeds to 14a) (Ref 18). Pickup tends to be cumula- velopment of tensile stresses at the surface of
the point where the end face touches the defor- tive and may force a die change and heterogeneous alloys. Thus, lubrication is often
mation zone, flow becomes nonuniform even in cleanup, often after extrusion of each billet required, although in some cases, hot extrusion
lubricated flow. The flow rate from the sides to in hot extrusion. is done without lubrication (especially for alu-
the center is insufficient, and a crater (pipe) In lubricated extrusion, the surface rough- minum alloys).
forms. The depth of the pipe increases with in- ens in the presence of a thick lubricant film Of all bulk deformation processes, extrusion
creasing container and die friction but is reduced (Fig. 14b), a phenomenon often described is perhaps the most sensitive to lubrication,
by friction on the back face of the billet. In ex- as orange peel. An excessively thick, un- partly because the non-steady-state conditions at
trusion of semifabricated products, the pipe in- stable film leads to bambooing (Fig. 14c) the beginning and end of extrusion are con-
creases the discard, especially if it fills up with and, in some less-ductile materials, even to ducive to instabilities, and partly because any
oxides and lubricant residues; it can form a dif- shear cracks (Fig. 14d). If the billet has been unwanted change in lubrication leads to defects
ficult-to-detect defect in the rear end of the prod- turned in preparation for extrusion, the ma- that affect not only the appearance but also the
uct. chining marks are stabilized by the lubricant properties and integrity of the product. In many
External Extrusion Defects. In addition to in- (Fig. 14e). respects, the problems of lubrication are similar
ternal defects (such as centerburst) or metallur- to those of lubrication in forging.
gical deficiencies (from variations in grain size From the standpoint of friction, it is of great-
and metallurgical structure, resulting from re- Friction and Lubrication (Ref 8) est importance whether the billet moves relative
crystallization after inhomogeneous deforma- to the container (direct extrusion) or is at rest in-
tion), there are also external, friction-related de- Extrusion is one of the processes in which a side the container (indirect extrusion). In either
fects in both hot and cold extrusion that, when thick lubricant film can be developed and main- case, the billet is first upset to fill out the con-
severe, make a product unsuitable for use. tained. The similarity to wiredrawing is evident; tainer of cross section A0. The pressure then
Surface cracking, whether circumferential (fir- the difference is that lubricant pressure at the die rises until plastic flow begins, and a product of
tree cracking), or at roughly 45 to circumferen- inlet can be higher (even in nonhydrostatic extru- cross section A1, which has a short, parallel land,
tial (crows feet cracking), or even longitudi- sion) and thus, thick-film formation is promoted. emerges from the die. In principle, it is immate-
nal (splitting), is a common problem in both Film thickness is governed by a number of fac- rial whether the extrusion is a round bar, a
extrusion and drawing (see Chapter 21, tors. As would be expected from the plastohy- shaped section, or a tube. Products vary greatly
Workability and Process Design in Extrusion drodynamic lubrication theory, film thickness in- in cross-sectional dimensions and in length, but
and Wire Drawing, in this Handbook for illus- creases with increasing viscosity and decreasing they typically range from, say, 5 mm (0.2 in.) di-
trations). The fir-tree effect (speed effect) occurs die half-angle. Film thicknesses may range from ameter wire to 500 mm (20 in.) diameter tubes
in hot-short materials with friction on the die excessively thick full-fluid films to mixed films or sections. Presses range from 2 to 150 MN
land as a contributory cause. A cold extrusion with predominantly boundary contact. (200 to 17,000 tonf) in capacity.
Table 1 Summary of the extrusion methods used for different materials furnaces, with or without a protective at-
Method of extrusion
mosphere. If scaling of steel is allowed, the
scale may be removed with high-pressure
Without lubrication, Without lubrication, With lubrication,
without shell with shell without shell water, but great care must be taken to avoid
scale pockets that would break through any
Type of material
lubricant film.
Materials with a lubricating Materials that flow according Materials with container
oxide skin (e.g., Cu) and to type C (e.g., brass) and flow characteristics that
materials that flow materials that readily adhere can be made similar to Solid-Film Lubrication. Because of the
according to type B and to the container wall flow type A by using a high interface pressures, solids with pressure-in-
in which lubrication (e.g., complex Al bronzes) lubricant and usually sensitive shear strengths can be very useful as
would lead to a poor-quality Materials that rapidly oxi- conical dies lubricants. With all forms of solid-film lubrica-
surface finish (e.g., Al alloys) dize during billet heating, Materials that are worked
i.e., those in which any by cold extrusion, including tion, pressures in direct extrusion of long billets
flowing in of the oxide hydrostatic extrusion, as may reach high enough values to satisfy the con-
skin in type B flow must well as alloys that are dition of sticking friction (Fig. 15). Sticking fric-
be prevented at all costs difficult to extrude, such tion occurs when the frictional interface shear
(low-alloy Cu materials as Ni alloys, steels, and high-
such as CuCr, CuZr, etc.) melting-point metals that stress (ti) is higher than the condition for inter-
cannot be extrudedor nal shearing of the material instead of sliding
only uneconomically against the die surface. The condition of sliding
without lubrication friction (Fig. 15a) is not necessarily realistic for
Source: Ref 1 metalworking, and the coefficient of friction (m)
becomes meaningless with a high interface pres-
sure (Fig. 15b). When sticking friction occurs
The importance of lubrication during direct die, mandrel, or container on prolonged ex- with a long billet, then the billet length has to be
extrusion of a billet in a container is discussed in posure and thus when, in the absence of such reduced or the process changed to reverse extru-
more detail in the section Forward (Direct) insulation, the die would close in or open up sion. Solid lubricants may be applied to the bil-
Extrusion in this chapter. It is important to note (die wash), the mandrel would thin out or let when the pressure medium lacks sufficient
that lubrication can have a detrimental effect on tear off, and the container would open up. lubricating properties. Continued development
the surface finish if the metal flow is inhomoge- Protection against oxidation is vital when of a model for solid lubricant entrainment in
neous (flow type B or C in Fig. 13). If type B or the workpiece material forms an abrasive hydrostatic extrusion has allowed for pressure-
C flow cannot be excluded, then container lubri- oxide, or when the lubricant does not wet as sensitive shear strength (Ref 19).
cation should be avoided. This is generally the a result of oxidation. Because oxidation can Oxides. Among the oxides grown on metals,
case for aluminum alloys and for those copper be controlled not only by a preapplied lubri- only some (e.g., on copper) can serve as lubri-
alloys that are extruded without a shell and do cant film but also by the preheating atmos- cants. Others (e.g., on titanium) serve merely as
not have a lubricating oxide skin (Ref 1, p 112). phere, the method of heating must be cho- parting agents. Externally introduced oxides
However, many difficult-to-extrude materials re- sen with the entire tribological system in never repaid the effort put into investigating
quire lubrication to reduce extrusion load or to mind. Induction heating, resistance heating, them, although researchers found that solids
protect the die (Table 1). and glass or salt baths can all be considered such as lime or slate gave adequate lubrication in
Lubrication is advisable during cold extrusion in addition to gas-fired or electric muffle the form of a friable pad trapped between die
or extrusion at low temperatures (less than 300
C, or 570 F). Lubrication may also be advis-
able in hot extrusion of some difficult alloys
through conical dies. Lubricants can be classi-
fied into two groups by temperature (Ref 1):
Greaselike lubrication below 1000 C (1830
F) with grease, graphite, MoS2, mica, talc,
soap, bentonite, asphalt, and plastic (e.g.,
high-temperature polymides)
Glasslike lubrication above 1000 C (1830
F) with glass, basalt, and crystalline pow-
ders
Attributes of Extrusion Lubricants. An
extrusion lubricant must have the usual attrib-
utes for metalworking, but some requirements
assume special significance:
An uninterrupted separating film between
billet, die, and container and its low shear
strength are important not only because they
determine the allowable extrusion ratio and
billet length in direct extrusion, but also be-
cause they control material flow and thus
the properties of the product and the length Fig. 15 Shear stress and coefficient of friction in (a) sliding friction and (b) sticking friction. (a) In sliding friction, the
of the discard. coefficient of friction, m, is roughly constant, and the frictional (or interface) shear stress, ti, increases in pro-
portion to the interface pressure, p. This condition is not necessarily realistic in metalworking when ti exceeds condi-
Heat insulation is imperative in nonisother- tions for internal shear of the workpiece. (b) Sticking friction occurs when ti = mp > k, where k is the value of mmax where
mal extrusion when the extrusion tempera- relative sliding does not occur at the interface. Calculated m actually drops with increasing hydrostatic stress (or inter-
ture is high enough to cause softening of the face pressure, p). sf = flow stress
and billet (Ref 20). Lubrication must be attrib- largest quantities of steel are extruded by phase- lubricants. In hot extrusion, they may be gradu-
uted to a controlled form of wear (and perhaps change lubrication of glass. ally melted to provide a form of phase-change
melting) of the pad. Glasses for lubrication are often proprietary lubrication.
Canning and Lubricants. Simple canning developed for minimum cost compatible with
techniques can be used with lubrication in the performance requirements for uniform applica-
form of canning powders or glass lubricants for tion of heat-insulating, low-friction glass. The Forward (Direct) Extrusion
difficult-to-extrude materials in a steel can. desirable glass property is usually defined as a
Lubricants are applied as a coating or, if a viscosity of 101 to 102 Pa s at the billet tem- A typical sequence of operations for the for-
thicker coating is required or a powder is to be perature, although a viscosity-temperature curve ward extrusion of a solid section is:
extruded, a simple canning technique can be of low slope would appear to be needed for op-
1. The heated billet and the dummy block are
used with glass as the lubricant. To ensure timum melt-away performance, and thermal
loaded into the container.
smooth, defect-free material flow, extrusion is properties such as heat conduction and specific
2. The billet is extruded by the force of the
invariably done with lubrication, and then it is heat should be just as important. Window glass
punch being pushed against it. This upsets the
preferable to have a well-fitting billet with a or its equivalent, in powder or fiber form, mixed
billet, then forces the metal to flow through
conical nose. The can is often evacuated through with a binder, is satisfactory for many steels and
the die. During extrusion, a thin shell of ma-
a tube to prevent oxidation of the powder and even for stainless steels. Some producers replace
terial (butt) may be left in the container.
puffing up of the can. Because the can material borosilicate glass with the cheaper basalt for Ni-
3. The container is separated from the die, the
is removed after extrusion, preferably by me- Cr alloys and with basalt mixed with borosili-
extruded section with the butt, and the
chanical means, a parting agent is often placed cate glass for Ni-Cu alloys extruded at lower
dummy block.
on the billet-can interface. In some instances, a temperatures. Several improvements have been
4. The discard (butt) is sheared off.
thin sheet of diffusion barrier is also used. By suggested, usually in the form of patents. For ex-
5. The shear die, the container, and the ram are
the proper choice of can material, extrusion ample, it has been suggested that a desirable vis-
returned to their initial (loading) positions.
pressures can be reduced, as in extrusion of a cosity gradient could be established by the use
high-strength Al-Cu-Mg alloy canned in pure of glass layers of differing viscosities, either be- In forward or direct extrusion, the billet is
aluminum (Ref 21). However, the can material tween the billet and container or between the pushed bodily along the container wall, and the
must not be too soft or too thick, lest it be ex- front and back of the billet bore. The viscosity of extrusion emerges in the same direction.
truded ahead of the billet or behave similar to a glass used for making commercial foams is too However, the crucial aspect is that the billet
fluid film with a typical unsteady bambooing high, and the graphite used as a foaming agent moves relative to the container; therefore, in ad-
flow. A can material of one-half to one-third the interferes with wetting of the metal; however, dition to friction on the die, friction on the con-
strength of the billet material gives satisfactory foam is successful as a carrier for powder. tainer wall has to be overcome, too. This has a
results. An example of a complex canning Various methods of glass application have profound effect on all aspects of extrusion.
arrangement for extrusion of a tungsten tube is been used. For example, a pad of plate glass
shown in Fig. 16. The can metal, because of its placed in front of the billet cracks in contact Process Characteristics
lower strength, allows roughening of the extru- with the billet. Pads made of about 100-mesh
sion surface. powder have become standard. They are self- Container Friction. As previously noted,
Glass Lubrication. The geometry of the ex- sintered or are made with binders such as the large pressures of most practical extrusion
trusion process makes it particularly favorable sodium silicate or bentonite clay. Other forms of processes require a billet supported by a con-
for lubrication with glasses that melt gradually glass such as wool, cloth, and foam were also tainer. At a first approximation, it can be as-
to form a lubricating film. As a lubricant, glass tried but with less success or economy. As ex- sumed that pressure within the container propa-
exhibits unique characteristics, such as its ability trusion begins, a thin film of glass melts off the gates as in a hydraulic medium and that the
to soften selectively during contact with the hot pad. If melting characteristics are properly cho- extrusion pressure prevails at the billet-container
billet and, simultaneously, to insulate the hot bil- sen and the press speed is adequate (typically, interface (in reality, pressures are lower).
let material from the tooling. For example, glass 0.15 to 0.5 m/s), the pad supplies glass to the end Because even the minimum pressure (Eq 4) can
lubricants are used in the Sjournet process, of the stroke. Another method is rolling a pre- reach high values at large extrusion ratios, fric-
which is a common method of hot extrusion for heated billet down an inclined table covered tion at the container-to-billet interface assumes
steels and titanium alloys (see the section Hot with glass powder or fiber so that it becomes great importance. From the practical point of
Extrusion in this chapter). Since about 1950, coated with a layer of molten lubricant. For tube view, there are only two possibilities: either the
the technique of glass lubrication of the Ugine- extrusion, a woven glass sock is pulled over the interface is arrested (sticking friction), or sliding
Sjournet process and its further developments mandrel, or glass powder is sprayed into the bil- of the interface is ensured (lubricated flow). No
has spread to become dominant for extrusion of let bore. intermediate condition can be allowed, because
tubes, sections, and bars from steels and other Other Solid Lubricants. Layer-lattice com- partial sticking and partial sliding invariably
materials when quantities are insufficient or pounds are widely employed. When they are the lead to unsteady material flow and a defective
shapes too complex to allow rolling. By far, the only means of lubrication, uniform coverage of extruded product.
the billet is critical, because localized film fail- Material flow is affected by friction of the
ure would lead to die pickup and workpiece billet with the container wall, as previously
damage. The shear stress of the interface in- noted (Fig. 13). These flow patterns extend to
creases with temperature; lower friction is en- the product, as illustrated schematically in Fig.
sured with grease-based lubricants filled with 17. In the total absence of friction, grid lines in-
copper and lead powder (Ref 22). To minimize scribed on the center plane of a split billet re-
breakthrough pressure, the billet should be main undistorted until they reach the deforma-
nosed (Ref 23). Because of their limited spread- tion zone just ahead of the die (Fig. 17a). Even
ing ability, these lubricants are more suitable for at the prevailing high interface pressures, this
short billets. condition can be maintained only as long as
Graphitic lubricants are often used for selec- ti << k (where k is the shear flow stress accord-
tive lubrication of die and mandrel, either alone ing to von Mises criterion).
Fig. 16 Complex canning arrangement, including
or in conjunction with another lubricant. With increasing friction, friction on the con-
composite billet for extruding tungsten tubes
at 1210 C (2210 F) Polymers are used in cold extrusion as solid-film tainer wall causes the grid lines to curve. In
continues until a butt of 5 to 20% of the weight

of the billet is left. The extruded product is then
examined for soundness.
Effect of Friction on Forces. In direct ex-
trusion of a billet, the pressure required to move
the billet against the frictional stress, ti, has to be
added to the extrusion pressure, pe. The calcula-
tion is simple when ti is a constant, and also for
sticking friction when ti = k s0 /3. The area to
be sheared is of circumference d0p and length L,
where L is the length of the unextruded billet
measured to the die entry (Fig. 8) or to the be-
ginning of the dead-metal zone (Fig. 18). Thus,
the pressure at any point in the extrusion stroke
is:
pL = pe + 4ti (L / d0) (Eq 6)
The punch force drops gradually (Fig. 20) to the

minimum pe measured at the point where the die
Fig. 17 Material flow in forward extrusion. (a) Without container friction. (b) With container friction. (c) With cool- or dead-metal zone is touched. Sometimes a
ing. (d) Extrusion defect higher initial breakthrough pressure is regis-
seeking minimum energy flow, movement at the subjected to subsequent heat treatment, to blis- tered, which can be attributed to the initiation of
transition to the flat die is arrested, and a dead- tering. flow in the workpiece material and/or to tran-
metal zone is formed (Fig. 17b). Deformation When a hot workpiece is extruded in a colder sients associated with the development of the lu-
now occurs in the body of the billet along the container, chilling contributes to the inhomo- bricating film.
boundary of the dead-metal zone. The internal geneity of deformation, and a very complex flow In extrusion with sliding friction, the magni-
die angle thus formed increases with decreasing pattern develops (Fig. 17c). In later stages of ex- tude of ti must be known. The shear strengths of
friction and increasing reduction. Dead-metal trusion, the dead-metal zone spreads throughout some lubricating substances (for example, soft
zone formation in lubricated extrusion may be the entire length of the billet. Material near the metals) are insensitive to interface pressure. In
avoided if the die is made to the appropriate punch face moves toward the center and carries this case, then Eq 6 applies. If, on the other
angle. Sliding then takes place over the entire with it surface oxide films. This leads to the de- hand, the lubricant is pressure-sensitive and ti
container-die interface but only if the lubricant velopment of the so-called extrusion defect (also can be described by a constant m (as is the case
thins out uniformly without suffering local called piping): a ring of oxide inclusions causes for many boundary lubricants and polymers),
breakdowns. Otherwise, an incipient dead-metal complete material separation in the form of a the punch pressure at a distance L can be ob-
zone forms, and localized subsurface defects are cone, the diameter of which increases as extru- tained by integration in the following form (Ref
generated. Extension of the lubricant is substan- sion proceeds (Fig. 17d). Chilling in the con- 8, p.411):
tial; even at a modest extrusion ratio of 16 (ex- tainer can further complicate material flow when
truding a round bar to one-quarter of the billet the billet is allowed to rest directly on the con- pL = pe exp(4mL/d0) (Eq 7)
diameter), the surface increases fourfold. tainer. Uneven temperature distribution also re-
If the previous conditions cannot be fulfilled, sults in nonuniform metal flow, which in turn and the extrusion pressure rises exponentially.
it is much better to extrude without any lubricant leads to internal stresses and curving of the ex- If, however, the condition ti = mpL is satisfied
whatsoever. Then, sticking friction prevails over truded product. anywhere along the container wall, sliding at
the entire container-die interface, and extrusion Unlubricated (hot) extrusion with sticking that point is arrested, even in the presence of a
proceeds by shearing along the container wall friction has its advantages, too. The extruded lubricant (Fig. 20). The increased pressure (and
(Fig. 18). A dead-metal zone of maximum angle product is smooth and shiny, because its surface particularly, breakthrough pressure) limits the
forms, and this angle increases with reduction is formed by shearing inside the body of the bil-
ratio. It is essential, though, that no trace of a lu- let. Undesirable side effects can be neutralized
bricant or contaminant film be present on the by extruding with a punch (or follower block) of
billet surface, because otherwise, the sticking slightly smaller diameter than the container, thus
condition (A-B in Fig. 19a) is locally disturbed, leaving a skull that traps all oxides and other for-
and the lubricant is washed in below the product eign material. The formation of the extrusion de-
surface (at F-E in Fig. 19b) (Ref 24). This can fect can be prevented by leaving a large butt, and
lead to surface lamination and, if the product is yet, for reasons of economy, extrusion usually
Fig. 20 Variation in extrusion pressure during direct

Fig. 19 Material flow in unlubricated extrusion (a) extrusion under various container-friction
under normal conditions (left side) and (b) and thermal conditions. m, friction factor; tf, shear flow
Fig. 18 Forward extrusion with sticking friction with intermittent lubrication (right side) stress
length of billet that can be directly extruded Special Processes one method uses a fixed or floating mandrel, the
(Fig. 12). other uses a bridge-type die (Fig. 23).
Evaluation of Friction. In principle, measure- Billet-on-billet extrusion is a special tech- Tube Extrusion with Mandrel. A hole pierced
ment of the punch force at two points in the nique for alloys that are easily welded together at or machined in a billet can be maintained in the
stroke offers a means of determining the inter- the extrusion temperature and pressure. Perfect course of extrusion by a fixed or floating man-
face shear strength from Eq 6 or the coefficient welding of the billet in the container with the drel long enough to reach into the die land (Fig.
of friction from Eq 7. Of course, any such eval- next billet must take place as the joint passes 23a). A solid billet can be used if a separately ac-
uation can be valid only if all other variables re- through the deformation zone, and the following tuated piercing ram is available. Lubricated ex-
main constant. This condition is seldom satis- requirements must be fulfilled (Ref 7): trusion is possible only with a hollow billet; in
fied, because the mean or average flow stress working with a piercing ram, sticking friction
Good weldability at the deformation tem-
(s 0) is bound to vary even in hot extrusion and must be encouraged in order to avoid subsurface
perature
is known to vary greatly in cold extrusion. If the defects on the internal tube surface. The effect of
Accurate temperature control
billet heats up in the course of extrusion (as in friction on material flow is similar to that found
Cleaned billet surface
cold extrusion and isothermal hot extrusion), the in extrusion of bars. Marker pins radially in-
Sawn, cleaned billet end free from grease
extrusion force drops steeply (Fig. 20), and a serted into the billet show the development of a
Bleeding of air from the container at the
misleadingly high friction value is calculated. In dead-metal zone in unlubricated extrusion (Fig.
start of extrusion, to avoid blisters and other
contrast, chilling in nonisothermal hot extrusion 24a), whereas pins are successively extruded in
defects (Fig. 21)
raises s 0, and quite unrealistically low friction lubricated extrusion (Fig. 24b) (Ref 25). The
may be calculated. Sometimes, when the extru- Two methods of billet-on-billet extrusion residual skull clearly shows that the original skin
sion force actually rises during the stroke (Fig. have been developed. In one method, the discard remains in the container in the absence of a
20), a negative friction value is obtained. is removed, and the following billet is welded to lubricant but is drawn in during lubricated
Therefore, coefficients of friction quoted with- the one remaining in the welding or feeder plate extrusion.
out further specification should be looked on (Fig. 22). In the other method, there is no dis- Tube Extrusion with Bridge-Type Die.
with suspicion. card. The subsequent billet is directly pressed Alternatively, a solid billet is extruded through a
Nevertheless, the extrusion force and, even onto the billet still in the container. bridge, spider, or porthole die (Fig. 23b) that di-
more so, its variation in the course of the stroke Extrusion of Shapes. One of the main ad- vides and then reunites the flowing material to
provide important clues regarding the efficiency vantages of extrusion relative to rolling of form a hollow product with several internal
of lubrication. They can be used for comparative shapes is that the dimensions or shape of the ex- pressure welds. All traces of lubricant must be
lubricant evaluation, provided that other condi- truded product can be changed relatively easily carefully kept out; otherwise, the weld quality is
tions are not allowed to change. More reliable and inexpensively by fabricating a new die or die impaired. This technique is suitable only for ma-
quantitative data could be obtained by direct insert. If the extruded product is to be straight terials that can be extruded at low enough tem-
measurement of the frictional shear stress. and free of internal stresses, the rate of extrusion peratures to retain the strength of the die (in
However, because the high pressures and tem- must be the same in all parts of the extruded sec- practice, lead and the more readily weldable alu-
peratures often encountered present serious ex- tion. The rate of flow reflects the resistance that minum alloys). Die design calls for great skill.
perimental difficulties, few attempts have been a given part of the section encounters in exiting The additional frictional surface created on
recorded. from the die. Friction on the die face and over the mandrel or around the bridge adds consider-
the die land hinders free flow and can be used to ably to the extrusion pressure. Toward the back
equalize flow rates. A thicker section emerges of the billet, sticking may prevail between man-
more rapidly than a thinner one and therefore drel and billet, but toward the deformation zone,
curves the resulting extrusion. Two ways to the inner surface of the tube slides on the man-
counteract the tendency of the extrusion to curve drel, subjecting it to large frictional forces.
is to increase the land length or reduce the die Because the punch is exposed to high tempera-
half-angle to increase frictional retardation. tures in hot extrusion, it may not be able to with-
Products of small cross-sectional area usually stand these forces and may be torn off or thinned
are extruded through multiple holes. Flow rates out by plastic flow. Therefore, lubrication of the
can then be equalized by similar techniques. For mandrel surface is usually mandatory, at least
example, another way to counteract this ten- above aluminum extrusion temperatures.
dency for uneven flow is to incorporate what is Friction-Assisted Extrusion. There are
called a bleeder hole in the die near the thin por- several ingenious schemes (Ref 26) in which
tion of an extrusion. The difference between the friction is actually exploited to aid extrusion. In
rates of flow through two holes of different di- friction-assisted forward extrusion, the container
ameters can be used for evaluating frictional is moved at a somewhat higher speed than, but in
conditions. the same direction as, the punch (Fig. 25), and
Fig. 21 Bleeding out air during upsetting Tube Extrusion. Hollow products with a thus, friction on the container wall actually aids
variety of cross-sectional shapes may be ex- the movement of the billet, ensures more homo-
truded, in principle, by two direct techniques: geneous material flow, and reduces the remnant.
Fig. 22 Continuous-type extrusion using a welding

plate in front of the die Fig. 23 Forward extrusion of a tube with (a) a fixed mandrel and (b) a bridge-type die
Backward (Indirect) Extrusion 3. The billet is extruded, leaving a butt. container. As a result, machined billets are used
4. The die and the butt are separated from the in many cases. In addition, the cross-sectional
Indirect extrusion is also called inverse, re- section. area of the extrusion is limited by the size of the
verse, or back(ward) extrusion, because the billet hollow stem.
Backward extrusion offers a number of ad-
is stationary in the container cavity, and a punch The components of a typical tool assembly
vantages:
moving against the billet forces the extrusion to used for the backward extrusion of steel parts
emerge in a direction opposite to the punch A 25 to 30% reduction in maximum load are identified in Fig. 28. There is considerable
movement (Fig. 26a). The crucial point is that the relative to direct extrusion variation in the tooling practice and design de-
billet remains at rest in the container, and thus, Extrusion pressure is not a function of bil- tails of tool assembly components. A major
container friction does not come into play. let length, because there is no relative dis- problem in indirect extrusion is the construction
Therefore, the extrusion load and the temperature placement between the billet and the con- of the hollow ram and the die; advances in the
generated by deformation and friction are retainer. Therefore, billet length is not limited design of these components have contributed to
duced, effectively increasing the load capacity of by the load required for this displacement the extended application of the process (Ref 1,
the press. Indirect extrusion is often used for ex- but only by the length and stability of the 27). An entirely different form of back extrusion
trusion of harder alloys when shape is critical. hollow stem needed for a given container is obtained when a solid punch penetrates the
The sequence of operations for the backward length. billet.
extrusion of a solid section is: Extrusion pressure is more uniform as a Material Flow (Ref 8). In indirect extru-
function of ram travel compared to direct sion, as in direct extrusion, material flow in the
1. The die is inserted into the press. extrusion (Fig. 27). vicinity of the die depends on die geometry, re-
2. The billet is loaded into the container. No heat is produced by friction between the duction, and the presence of a lubricant. With a
billet and the container; consequently, no good lubricant and a tapered die entry, a homo-
temperature increase occurs at the billet sur- geneity of flow comparable to that of hydro-
face toward the end of extrusion, as is typi- static extrusion can be obtained. In lubricated
cal in the direct extrusion of aluminum al- flow, the dead-metal zone is limited to a thin
loys. Therefore, in backward extrusion, layer on the die face. The nonextruded part of
there is a lesser tendency toward cracking of the billet shows no effect of deformation (Fig.
the surfaces and edges, and extrusion speeds 29b). Because only the extrusion pressure, pe,
can be significantly higher. needs to be developed, heat generation is also
The service life of the tooling is increased, reduced. As can be deduced from Fig. 12, for a
especially that of the inner liner, because of given punch and container pressure, either a
reduced friction and temperatures. larger extrusion ratio can be taken or, more im-
portantly for difficult-to-extrude materials,
The disadvantage of backward extrusion is
workpiece temperature can be lowered and ex-
that impurities or defects on the billet surface af-
trusion speeds increased without running the
fect the surface of the extrusion and are not au-
danger of incipient melting. In direct extrusion,
tomatically retained as a shell or discard in the
the temperature in the unextruded portion of the
billet keeps rising, whereas in indirect extru-
sion, it remains essentially constant. This is of
particular importance in extrusion with sticking
friction.
Fig. 24 Material flow in tube extrusion shown by

marker pins. (a) Unlubricated. (b) Lubricated
Fig. 25 Friction-assisted forward extrusion

Fig. 26 (a) Backward extrusion and (b) resultant ma-
terial flow
Fig. 27 Variation of load or pressure with ram travel

for both direct and indirect extrusion. Region
I: The billet is upset, and pressure rises rapidly to a peak
value. Region II: So-called steady-state, where pressure
drops with direct extrusion. Region III: Pressure reaches a Fig. 29 Material flow in hot extrusion of a composite
minimum and then rises sharply as the discard is com- Fig. 28 Tools constituting a typical setup for the cold billet made up of disks of two aluminum al-
pacted. backward extrusion of steel parts loys. (a) Forward extrusion. (b) Backward extrusion
Combination of Direct (Friction-Assisted) tion and product quality. Nevertheless, they machining operations sometimes precede extru-
and Indirect Extrusion (Ref 8). A number of provide examples of using rather than combat- sion in order to produce a shape that can be more
attempts have been made to take advantage of ing friction. easily extruded.
container friction. Because friction is strongly The use of hot upset forgings as the starting
influenced by the movement of the billet rela- material is also common practice. Hot upsetting
tive to the container, it is recognized that many Cold Extrusion followed by cold extrusion is often more eco-
of these processes are combinations of direct nomical than alternative procedures for produc-
and indirect extrusion. In one technique (Ref Cold extrusion is normally used to create ing a specific shape. Axle shafts for cars and
28), the die is fixed in a hollow stem, which, at near-net shape products, such as fasteners, auto- trucks are regularly produced by this practice;
the beginning of extrusion, penetrates into the motive components, and so forth. Impact extru- the advantages include improved grain flow as
billet. Thus, extrusion begins without a high sion is a term employed for the cold extrusion of well as low cost. For example, the rear-axle
breakthrough pressure; the punch then ad- thin-walled products, such as toothpaste tubes. drive shaft illustrated in Fig. 30 was fabricated
vances, and the process continues as direct ex- Cold extrusion is used when the process is eco- by a three-step cold extrusion operation. This
trusion. nomically attractive because of: process improved surface finish (and thus fa-
Several methods of making the process essen- tigue resistance), maintained more uniform di-
tially continuous have been proposed. One Savings in material ameters and closer dimensional tolerances, in-
method (Ref 29) suggests that the material be Reduction or elimination of machining and creased strength and hardness, and simplified
wedged into the groove of a roller with the aid of grinding operations because of the good production.
a stationary shoe. Friction on the groove walls surface finish and dimensional accuracy of The problems most commonly encountered in
moves the billet, while friction on the shoe hin- cold-extruded parts cold extrusion are:
ders its movement, and the net extrusion force Elimination of heat treating operations be- Tool breakage
available is governed by the balance of the two. cause of the increase in the mechanical Galling or scoring of tools
This technique has been used for extrusion of properties of cold-extruded parts Workpieces sticking to dies
both aluminum and copper. In another technique Workpieces splitting on outside diameter or
(Ref 30), a closed pass is formed between a Cold extrusion is sometimes used to produce
cupping in inside diameter
grooved roll and a roll provided with a flange, only a few parts of a certain type, but it is more
Excessive buildup of lubricant in dies
and the frictional forces developed on the roll commonly used for mass production because of
surface and on the groove sides are used for ex- the high cost of tools and equipment. Cold ex- Table 2 lists the most likely causes of these prob-
trusion through a die. trusion is often combined with cold forging lems.
A longer contact length, and thus greater fric- (e.g., cold heading) for mass production of
tional force, can be obtained in a linear arrange- near-net or net-shape parts, such as bolts, nuts,
ment. In one process (Ref 31), articulated grips rivets, and many automotive and appliance Cold Extrusion Tooling
move a square bar from two sides, while lubri- components.
cated stationary dies prevent its spread in the Cold extrusion and cold heading are often Metals can be cold extruded by different tool-
lateral direction. In another version (Ref 32), combined to produce hardware items and ma- ing setups, depending mainly on the size and
the frictional force between a tightly fitting, chinery parts that require two or more diameters. shape of the workpiece, the composition of the
elastically deforming clamp and a precision bil- An intermediate-sized starting piece may be rework metal, and the quantity requirements. The
let is used to push the material through the ex- duced to the smallest diameter by cold extrusion, principal types of tooling employed are:
trusion die. It has also been proposed (Ref 33) followed by forming of the large-section portion
to remove the need for close-fitting clamps by by heading. Cold extrusion may also follow
using segmented grips moving around all four other operations that shape the starting piece.
sides of the billet. None of the previously men- The use of symmetrical slugs as the starting ma-
tioned processes has achieved production sta- terial for extrusion is common practice, but other
tus, partly because of difficulties of construc- shapes are often used. For example, one or more
Table 2 Problems in cold extrusion and some potential causes

Problem Potential cause
Tool breakage Slug not properly located in die

Slug material not completely annealed
Slug not symmetrical or not properly shaped
Improper selection or improper heat treatment of tool material
Misalignment and/or excessive deflection of tools and equipment
Incorrect preloading of dies
Damage caused by double slugging or overweight slugs
Galling or scoring of tools Improper lubrication of slugs
Improper surface finish of tools
Improper selection or improper heat treatment of tool material
Improper edge or bend radii on punch or extrusion die
Workpieces sticking to dies No back relief on punch or die
Incorrect nose angle on punch and incorrect extrusion angle of die Fig. 30 Rear-axle drive shaft of 1039 steel produced
by three-step cold extrusion of an upset forg-
Galled or scored tools ing (billet weight: 36 kg, or 79.5 lb). The drive shafts were
Workpieces splitting on outside diameter or hot upset forged to form the flange and to preform the
forming chevrons on inside diameter Slug material not completely annealed shaft and then were cold extruded to lengthen the shaft.
Reduction of area either too great or too small The flange could have been upset as a final operation after
Excessive surface seams or internal defects in work material the shaft had been cold extruded to length, but this would
Incorrect die angles have required more passes in the extrusion press than
Excessive buildup of lubricant on dies Inadequate vent holes in die space allowed. Hot upsetting and cold extrusion replaced
Excessive amount of lubricant used a hammer forging and machining sequence, after which
Lack of a means of removal of lubricant, or failure to prevent lubricant the flange, a separate piece, had been attached.
buildup by spraying the die with an air-oil mist Dimensions given in inches
Single-station tooling forms the part in one tios. The pressure pad, which transmits the load Table 3 Typical tool steels used in
stroke of the press. Additional operations from the back of the punch to the die set, should extruding aluminum
may be required for finishing. Closed-end be designed for economy, ease of replacement, Tool AISI steel Hardness, HRC
containers, such as toothpaste tubes, are and efficiency in reducing the number of punch
formed in this manner. failures. Die, solid W1 6567
Die sleeve(a) D2 6062
Multiple-station tooling involves a series of Die Design. In forward extrusion, the die is L6 5662
separate dies arranged so that the rough under maximum pressure, and this pressure is H13 4852
blank is made into a preform, which then not distributed uniformly. Therefore, the tool de- Die button(b) H11 4850
proceeds through successive operations signer must calculate the hoop (tensile) stresses H13 4850
L6 5052
until the required form is produced. on the inner die wall and provide adequate rein- H21 4750
Transfer presses are similar in concept to forcement. Ordinarily, pressures of less than ap- T1 5860
multiple-station tooling, because they can proximately half the yield strength of the die do Ejector D2 5557
perform several operations in succession by not require reinforcement, while those in excess S1 5254
Punch S1 5456
mechanical transfer of the workpiece from of this value do require reinforcement. D2 5860
one operation to the next. Extrusion dies are usually inserted in one or H13 5052
Upsetters or headers are used for continu- more shrink rings to provide reinforcement. Stripper L6 5658
ous operation, frequently incorporating both These rings prestress the die in compression by Mandrel, forward S1 5254
H13 5052
backward and forward extrusion and cold providing interference fits between rings and Holder H11 4248
heading. die. This results in lower working stress and H13 4248
Rotating dial or indexing can be applied for therefore longer fatigue life of extrusion tools. A 4130 3644
manual or automatic production, where similar technique is used to shrink radially seg- 4140 3644
multiple dies are held in an index dial on the mented die inserts together to prevent the seg- AISI, American Iron and Steel Institute. (a) Cemented carbide is some-
press table. ments from separating under load. Permanent times used for die sleeves. (b) Maraging steel is sometimes used for die
buttons.
shrink-fit assemblies are sometimes made by
Punch Design. A major problem in punch heating the outer ring to facilitate assembly.
design consists of assessing the nature and Interchangeable die inserts are usually force fit-
magnitude of the stresses to which the punch is ted mechanically, using a tapered press fit and
subjected in service. Because the stresses are molybdenum disulfide as a lubricant. Of the two ferred for extruding hollow parts, the angle vary-
dynamic, fatigue effects arise, and these fa- methods, shrinking-on by heating is generally ing inversely with wall thickness. Ejection pres-
tigue effects, in conjunction with the inher- preferred, because a cylindrical hole and shaft sure on the work increases with decreasing die
ently brittle nature of hardened tool steels, ne- are easier to fabricate than a tapered hole and angle, because greater friction must be over-
cessitate care in avoiding design features likely shaft. However, a taper fit has several advan- come. This pressure also increases with an in-
to produce stress concentrations. The stability tages, such as: crease in the length of the part. Extrusion pres-
problems that may arise when slender punches sure causes elastic expansion of the die, which
are used are affected by the accuracy of align- The hardness and yield strength of the vari- shrinks when the pressure is discontinued.
ment provided by the tool set or the press it- ous die components are not lowered (as they Accordingly, very high wall pressures are devel-
self, or by factors in the extrusion operation, would be by heating) and can be measured oped, and these require correspondingly high
such as punch wander, initial centering, and with dependable accuracy. ejection pressures.
use of distorted slugs. The ratio of punch The prestress value is ensured by strict con- Die Materials. Compressive strength of
length to punch diameter also affects stability; trol of the input measurements. the punch and tensile strength of the die are
a ratio of approximately 3 to 1 is probably the Release and exchange of the inner die bush- among the most important factors influencing
maximum for cold extrusion of steel using tool ings is quick, easy, and inexpensive. the selection of material for cold extrusion
steel punches. Die parts can be standardized. tools. Because the die is invariably prestressed
The design of the punch nose has a significant Hot working die steels are not required. in compression by the pressure of inner and
effect on extrusion pressures and tool life. In outer shrink rings, the principal requirement for
backward extrusion, acceptable results are ob- The most commonly used taper angle is 1/2 to a satisfactory die is a combination of tensile
tained with a nose profile consisting of a trun- 1. The conditions for obtaining the specified yield strength and prestressing that prevents
cated cone having an included angle of 170 to advantages of the taper force fit are careful failure. Punches require sufficient compressive
180, with an edge radius of 0.51 to 2.54 mm preparation of the taper shell surfaces and exact strength to resist upsetting without being haz-
(0.020 to 0.100 in.), and a land length of 1.27 to agreement between taper angles of correspon- ardously brittle. Thus, almost without excep-
1.9 mm (0.050 to 0.075 in.), with the shank re- ding contact faces. If the shell surfaces do not tion, and particularly for extruding steel, the pri-
lieved 0.1 to 0.2 mm (0.004 to 0.008 in.) on the provide uniform support over the entire die mary tools in contact with the work must be
diameter. Although they reduce initial punch length, the prestresses will be unequal, and the made from steels that through harden in the sec-
stresses, small cone angles or large radii are un- reinforcement will not be fully effective. tion sizes involved.
desirable because of rapid lubricant depletion In some setups, the first reinforcement is ap- Recommended materials for extrusion
and the risk of metal-to-metal contact. Design of plied by taper force fit and the second (outer) punches typically include M2 and M4 high-
the punch nose to distribute the lubricant prop- reinforcement by shrinking-on. It is advisable speed tool steels and tungsten carbide. Tool steel
erly during extrusion is essential for minimizing to standardize on the size of reinforcing ele- punches should be heat treated to a hardness of
the pressures developed. ments. In general, no further advantage is 62 to 66 HRC. Die inserts are usually fabricated
The area ratio between punch shank and head gained by making the outside diameter of a re- from such alloy tool steels as D2, M2, and M4
is also an important design factor. A large ratio inforcement more than four to five times the die and are heat treated to 58 to 64 HRC, depending
has the effect of spreading the punch load over diameter. on the steel. Typical tool steels and their work-
a large area of pressure pad. On the other hand, In forward extrusion, die angles are deter- ing hardnesses for the cold extrusion of alu-
it requires a wider block of metal for its fabri- mined by the shape of the workpiece and by the minum are given in Table 3.
cation, with a resultant cost increase. Because operating sequence. In general, an angle of 2a = Tungsten carbide is extensively used, because
pressure pads are less expensive than punches, 24 to 70 is selected for the forward extrusion of it provides good die life, high production rates,
it is generally advisable to favor the smaller ra- solids, and an angle of 2a = 60 to 126 is pre- and good dimensional control. Tungsten carbide
often finds application as a punch material in poor shearability and machinability; therefore, Boron-modified steels for heading and extrusion
backward extrusion. Retainer rings or housings some extrudability is occasionally sacrificed. are also available. The advisability of paying the
used for tungsten carbide dies should have suffi- Annealing techniques that produce a partly additional cost for cold extrusion quality B or
cient strength and toughness to prevent splitting pearlitic structure are ideal for many extrusion cold heading quality steel depends on the sever-
and failure of the working tools. Shrink rings applications in which shearability or machin- ity of extrusion, the quality requirements of the
should be fabricated from hot work die steels ability is important. extruded part, and the cost of rejected parts in
such as H11 or H13 heat treated to 46 to 48 Free-machining additives, such as sulfur or comparison with the extra cost for these steels.
HRC. Outer housings are often made from H13 lead, are likely to impair extrudability. Free-ma-
die steel or from 4340 alloy steel. chining steels, containing such additives as lead Cold Extrusion of
Effects of Lubricant Viscosity. Even and sulfur, are not preferred for cold extrusion. Aluminum and Its Alloys
though most extrusion of semifabricated prod- Extrusions from these steels are more suscepti-
ucts takes place at elevated temperatures, the ble to defects than extrusions from their non- Aluminum alloys are well adapted to cold (im-
softer aluminum alloys are extruded at room free-machining counterparts. In addition, be- pact) extrusion. The lower-strength, more ductile
temperature or at slightly elevated tempera- cause parts produced by cold extrusion generally alloys, such as 1100 and 3003, are the easiest to
tures with the aid of lubrication. Aluminum require only minimal machining (this is often extrude. When higher mechanical properties are
was extruded at various temperatures around the primary reason for using cold extrusion), required in the final product, heat treatable
150 C (300 F), using abietic acid so that vis- there is much less need for free-machining addi- grades are used. Although nearly all aluminum
cosity could be varied over a wide range (Ref tives than when parts are produced entirely by alloys can be cold extruded, the five alloys listed
34). At higher viscosities (lower temperatures) machining. in Table 4 are most commonly used. The alloys
and a lower half-angle, pressure built up fairly The successful extrusion of free-machining in Table 4 are listed in the order of decreasing ex-
gradually until extrusion began. The first part steels depends on the amount of upset, the flow trudability based on pressure requirements. The
of the section to emerge was fairly bright, typ- of metal during extrusion, and the quality re- easiest alloy to extrude (1100) has been assigned
ical of boundary contact, but then the surface quirements of the extruded part. Free-machining an arbitrary value of 1.0 in this comparison.
became duller, indicating the presence of a steels can generally withstand only the mildest The cold extrusion process of aluminum parts
thicker film. upset without developing defects. If it is under is often considered, because high production
When conditions are not favorable enough to compression at all times during flow, a free-ma- rates can be achieved. Even when parts are large
create a thick film, deformation begins in the die chining steel will probably extrude without de- or of complex shape, lower production rates
throat, and the surface is much brighter, with fects. However, rupture is likely if compressive may still be economical. The impact-extruded
only occasional hydrostatic pockets, which is in- force is suddenly changed to tensile force. part itself has a desirable structure. It is fully
dicative of a predominantly boundary-type, Nonmetallic inclusions, particularly the wrought, achieving maximum strength and
mixed-film lubrication. With excessively high silicate type, are detrimental to extrudability. toughness. It is a near-net shape. There is no
viscosity, the lubricant film becomes so thick The fewer the inclusions, the more desirable the parting line, and all that may be required is a
that it cannot maintain stability, and periodic steel is for cold extrusion. Silicate inclusions trim to tubular sections. Surface finish is good.
collapse leads to development of the bamboo de- have been found to be the most harmful. Impacts have zero draft angles, and tolerances
fect. It has been noted that the wavelength of the Therefore, some steels have been deoxidized are tight. Once impacted, sections can be treated
bamboo defect was equal to the length of the die with aluminum rather than silicon in an attempt in the same manner as any other piece of
land, with the defect attributed to intermittent lu- to keep the number of silicate inclusions at a wrought aluminum.
bricant failure (Ref 35). Most information on minimum. The aluminum-killed steels have bet- From a design standpoint, aluminum impacts
this defect comes from research related to hy- ter extrudability in severe applications. should be considered in the following situations:
drostatic extrusion. Extrusion Quality. Carbon steel bars are
available at additional cost in two classes of ex- For hollow parts with one end partially or
trusion quality: cold extrusion quality A and totally closed
Cold Extrusion of Steel cold extrusion quality B. The mill preparation When multiple-part assemblies can be re-
for cold extrusion quality A is the same as that placed with a one-piece design
The extrudability of steel decreases with in- used for special-quality bars; cold extrusion When a pressure-tight container is required
creasing carbon or alloy content. The cold extru- quality B is a still higher quality. Higher quality When bottoms must be thicker than the
sion of steels containing up to 0.45% C is com- refers primarily to fewer external and internal walls, or the bottom design includes bosses,
mon practice, and steels with even higher carbon defects. Hot scarfing and more rigorous inspec- tubular extensions, projections, or recesses
contents have been successfully extruded. tion of the billets are additional operations that When a bottom flange is required
However, it is advisable to use steels of the low- are performed at the mill to prepare cold extru-
est carbon content that will meet service re- sion quality B material.
quirements. Most carbon and alloy steels that are Alloy steel without a quality extra is used in
extruded contain 0.10 to 0.25% C. However, in applications similar to those of cold extrusion
some applications, steels with more than 0.45% quality A for carbon steel. Alloy steels are also
(especially alloy steels) are cold extruded. available as cold heading quality, which paral-
For a given carbon content, most alloy steels lels cold extrusion quality B for carbon steel.
are harder than plain carbon steels and are there-
fore more difficult to extrude. Most alloy steels Table 4 Relative pressure requirements
also work harden more rapidly than their carbon for the cold extrusion of annealed slugs of
steel counterparts; therefore, they sometimes re- five aluminum alloys (alloy 1100 = 1.0)
quire intermediate annealing. Steels that have Relative extrusion
been spheroidized by annealing are in their soft- Alloy pressure
est condition and are therefore preferred for ex-
1100 1.0
trusion. However, operations that precede or fol- 3003 1.2
low extrusion may make it impractical to have 6061 1.6
the steel in its softest condition. Extremely soft 2014 1.8
7075 2.3 Fig. 31 Three types of dies used in the cold extrusion
steels of low-to-medium carbon content have of aluminum alloy parts
When bottoms, sidewalls, or heads have tice is to extrude the slug in the O temper, solu- be given a phosphate treatment, followed by ap-
changes in section thickness tion treat the preform to the T4 temper, and then plication of a soap that reacts with the surface to
size or finish extrude. This procedure has two form a lubricating layer similar to that formed
Equipment is readily available that can pro- advantages. First, after solution treatment, the when extruding steel.
duce backward and forward extrusions up to 400 metal is reasonably soft and permits sizing or
mm (16 in.) in diameter. Backward extrusions additional working, and, second, the distortion
can be up to 1.5 m (60 in.) long. The length of caused by solution treatment can be corrected in Other Nonferrous Alloys
forward extrusions is limited only by the cross final sizing. After sizing, the part can be aged to
section of the part and the capacity of the press. the T6 temper, if required. Cold Extrusion of Copper and Copper
Hydraulic extrusion and forging presses, suit- Slugs for extrusions are obtained by blanking Alloy Parts. Oxygen-free copper (C10200) is
ably modified, are used for making very large from plate; by sawing, shearing, or machining the most extrudable of the coppers and copper-
extrusions. Parts up to 840 mm (33 in.) in diam- from bars; or by casting. Rolled aluminum alloy base alloys. Other grades of copper and most of
eter have been produced by backward extrusion plate is widely used as a source of cold extrusion the copper-base alloys can be cold extruded,
from high-strength aluminum alloys in a 125 stock. The high speed at which slugs can be pre- although there are wide differences in extrud-
MN (14,000 tonf) extrusion press. Similar extru- pared is the major advantage of blanking from ability among the different compositions. For
sions up to 1 m (40 in.) in diameter have been rolled plate. When slug thickness is greater than example, the harder copper alloys, such as
produced in large forging presses. Because alu- approximately 50 mm (2 in.) or when the thick- aluminum-silicon bronze and nickel silver, are
minum extrudes more easily, tools designed es- ness-to-diameter ratio is greater than approxi- far more difficult to extrude than the softer, more
pecially for extruding aluminum may be differ- mately 1 to 1, blanking from plate is uneconom- ductile alloys, such as cartridge brass (alloy
ent from those used for steel. For example, a ical, if not impossible. Blanking is also C26000), which can satisfactorily withstand
punch used for the backward extrusion of steel excessively wasteful of metal, which negates a cold reduction of up to 90% between anneals.
should not have a length-to-diameter ratio principal advantage of the cold extrusion Alloys containing as much as 1.25% Pb can be
greater than approximately 3 to 1; however, this process. successfully extruded if the amount of upset is
ratio, under favorable conditions, can be as high Sawing from bars is widely used as a method mild and the workpiece is in compression at all
as 17 to 1 for aluminum (although a 10 to 1 ratio of obtaining slugs. More accurate slugs are pro- times during metal flow. Copper alloys contain-
is usually the practical maximum). duced by sawing than by blanking; however, as ing more than 1.25% Pb are likely to fracture
Tooling. Three basic types of dies for ex- in blanking, a considerable amount of metal is when cold extruded.
truding aluminum are shown in Fig. 31. Solid lost. When doughnut slugs are required, they The pressure required for extruding a given
dies are usually the most economical to make. can be sawed from tubing, or they can be area for one of the more extrudable coppers or
Generally, a cavity is provided in each end so punched, drilled, or extruded. Machined slugs copper alloys (such as C10200 or C26000) is
that the die can be reversed when one end be- (such as those produced in an automatic bar ma- less than that required for extruding low-carbon
comes cracked or worn. Compared to the die chine) are generally more accurate but cost more steel. However, the pressure required for extrud-
cavities used in the backward extrusion of steel, than those produced by other methods. Cast ing copper alloys is generally two to three times
the die cavities for aluminum are notably shal- slugs can also be used; the selection of a cast that required for extruding aluminum alloys (de-
low, reflecting a major difference in the extru- slug is made on the basis of adequate quality at pending on the copper or aluminum alloy being
sion characteristics of the two metals. Steel is lower fabricating cost. Compositions that are not compared).
more difficult to extrude, requiring higher pres- readily available in plate or bar stock can some- The length of a backward-extruded section is
sures and continuous die support of the work- times be successfully cast and extruded. There is limited by the length-to-diameter ratio of the
piece throughout the extrusion cycle. In contrast, often a savings in metal when a preform can be punch and varies with unit pressure. This ratio
aluminum extrudes readily, and when the punch cast to shape. should be a maximum of 5 to 1 for copper. A
strikes the slug in backward extrusion, the metal Surface Preparation and Lubricants. ratio of 10 to 1 is common for the extrusion of
squirts up the sides of the punch, following the Slugs of the more extrudable aluminum alloys, aluminum, and ratios as high as 17 to 1 have
punch contours without the external restraint or such as 1100 and 3003, are often given no sur- been used. The total reduction of area for copper
support afforded by a surrounding die cavity. face preparation before a lubricant is applied or copper alloys, under the best conditions,
Holder-and-sleeve dies are used when extru- prior to extrusion. For slugs of the less extrud- should not exceed 93%. In applications involv-
sion pressures are extremely high. This type of able aluminum alloys or for maximum extrusion ing minimum-to-moderate severity, copper slugs
die consists of a shrink ring or rings (the holder), severity or both, surface preparation may be nec- are often extruded with no special surface prepa-
a sleeve, and an insert (button). The die sleeve is essary for retention of lubricant. One method is ration before the lubricant is applied. However,
prestressed in compression in the shrink ring to to etch the slugs in a heated caustic solution, fol- for the extrusion of harder alloys (aluminum
match the tension stress expected during extru- lowed by water rinsing, nitric acid desmutting, bronze, for example) or for maximum severity
sion. Horizontal split dies are composed of as and a final rinse in water. For the most severe ex- or both, best practice includes the following sur-
many as four parts: a shrink ring, a sleeve (intrusion, slug surfaces are given a phosphate face preparation before the lubricant is applied:
sert), and a one-piece or two-piece base. Figure coating before the lubricant is applied.
31 identifies the one-piece base as a die bottom, Aluminum and aluminum alloys can be suc- Cleaning in an alkaline cleaner to remove
and the components of the two-piece base as a cessfully extruded with such lubricants as high- oil, grease, and soil
holder and a backer. viscosity oil, grease, wax, tallow, and sodium-tal- Rinsing in water
Stock for Slugs. In general, the methods for low soap. Zinc stearate, applied by dry tumbling, Pickling in 10 vol% sulfuric acid at 20 to 65
preparing aluminum slugs are similar to those is an excellent lubricant for extruding aluminum. C (70 to 150 F) to remove metal oxides
for preparing slugs from other metals and are In applications in which it is desirable to remove Rinsing in cold water
therefore subject to the same advantages and the lubricant, water-soluble lubricants are used to Rinsing in a well-buffered solution, such as
limitations. Many extrusions are produced di- reduce the wash cycle. The lubricant should be carbonate or borate, to neutralize residual
rectly from slugs purchased in the O (annealed, applied to metal surfaces that are free from for- acid or acid salts
recrystallized) temper. In other applications, es- eign oil, grease, and dirt. Preliminary etching of Lubrication. Zinc stearate is an excellent lu-
pecially when slugs are machined from bars, the the surfaces (see previous information) increases bricant for extruding copper alloys.
slugs are annealed after machining and before the effectiveness of the lubricant. For the most Common practice is to etch the slugs as de-
surface preparation. When extruding alloys that difficult aluminum extrusions (less extrudable al- scribed previously and then to coat them by
will be heat treated, such as 6061, common prac- loys or greater severity or both), the slugs should dry tumbling in zinc stearate.
Impact Extrusion of Magnesium Alloys. the length-to-diameter ratio, and the press align- truded with a variety of grease/graphite or
Impact extrusion is used to produce symmetrical ment. Magnesium has a relatively high coefficient glass-based lubricants.
tubular magnesium alloy workpieces, especially of thermal expansion compared to steel. In nonlubricated hot extrusion, the material
those with thin walls or irregular profiles for Therefore, in order to ensure that the magnesium flows by internal shear, and a dead-metal zone is
which other methods are not practical. As ap- extrusion, when cooled to room temperature, will formed in front of the extrusion die.
plied to magnesium alloys, the extrusion process be within dimensional tolerance, it is necessary to Deformation occurs with shearing the alloy at
cannot be referred to as cold, because both multiply the room-temperature dimensions of the die-to-billet interface. Only lower-melting
blanks and tooling must be preheated to not less steel tools by a compensatory factor for the tem- alloys, including aluminum alloys, can be ex-
than 175 C (350 F); workpiece temperatures of perature at which the magnesium alloy is to be extruded hot without a lubricant. Two-dimensional
260 C (500 F) are common. Pressures for the truded. shapes of great complexity and thin walls (e.g.,
impact extrusion of magnesium alloys are ap- architectural extrusions) can be made at a rela-
proximately half those required for aluminum tively low die cost. In the absence of a lubricant,
and depend mainly on alloy composition, Hot Extrusion divided material streams can be reunited and
amount of reduction, and operating temperature. welded in a bridge-type die, allowing extrusions
Length-to-diameter ratios for magnesium extru- The techniques for extrusion of different ma- with one or more closed cavities, in a very wide
sions may be as high as 15 to 1. There is no lower terials are dependent, to a large extent, on the ex- size range (multihole tubes are extruded with
limit, but parts with ratios of less than approxi- trusion temperature. The essence of hot working walls as thin as 0.25 mm, or 0.010 in.).
mately 2 to 1 can usually be press drawn at lower is to reduce the yield stress and thus increase ex- Nonlubricated hot extrusion is a relatively
cost. A typical ratio is 8 to 1, and parts with higher trusion speed for a given press load. However, if straightforward process once the conditions
length-to-diameter ratios are more amenable to the exit temperature is too high, surface liqua- have been defined, as briefly described in the
forward extrusion than to backward extrusion. At tion can result in hot shortness (that is, surface section Unlubricated Hot Extrusion of
all ratios, the mechanical properties of magnesium and surface cracks, such as in Fig. 32). The op- Aluminum Alloys in this chapter.
extrusions normally exceed those of the blanks timal extrusion temperature, Toptimal, is found at Hot extrusion of higher-melting alloys re-
from which they are made, because of the benefi- the intersection of the load limiting curve and quires a high-temperature lubricant (usually
cial effects of mechanical working. the allowable extrusion speed for a given surface glass or grease) between the extruded billet and
Dies for the impact extrusion of magnesium liquation temperature (Fig. 33). In terms of the the die. Shapes are more limited, and wall thick-
alloys differ from those used for other metals, be- right-side curve for hot shortness of an alloy, a ness is greater. Lubrication in extrusion reduces
cause magnesium alloys are extruded at moder- lower overall exit temperature allows a higher load and energy requirements, reduces tool
ately elevated temperature (usually 260 C, or extrusion speed, because more surface heating wear, improves surface finish, and provides a
500 F). Common practice is to heat the die with (from friction and less time for heat dissipation) product with nearly uniform properties. This
tubular electric heaters. The die is insulated from can occur without surface melting. technique is commonly used in the extrusion of
the press, and an insulating shroud is built around When lubrication is used with extrusion, the shapes from steels, titanium alloys, and nickel
the die. The top of the die is also covered, except right-side curve in Fig. 33 is also shifted to the alloys. Proper die design is critical in lubricated
for punch entry and the feeding and ejection de- right, because friction is reduced. Conventional extrusion, especially when noncircular shapes
vices. The punch is not heated, but it becomes hot extrusion may be either nonlubricated or lu- are extruded. An effective die design must en-
hot during continuous operation; therefore, the bricated (Fig. 4). As previously noted in the sure smooth metal flow with consistent lubrica-
punch should be insulated from the ram. section Friction and Lubrication in this chap- tion. It is desirable to use shaped dies, which
Punches and dies are usually made of a hot ter, container lubrication should be avoided if provide a smooth transition for the billet from
work tool steel, such as H12 or H13, heat treated non-homogenous flow (type B or C flow) can- the round or rectangular container to the shaped-
to 48 to 52 HRC. In one application, tools made not be excluded. Generally, hot extrusion of die exit.
of heat treated H13 produced 200,000 extru- aluminum alloys is done without lubrication, Critical parameters for successful and eco-
sions. Carbide dies can be used and can extrude while copper alloys, titanium alloys, alloy nomical hot extrusion include the method of bil-
up to 10 million pieces. The sidewalls of the die steels, stainless steels, and tool steels are ex- let preparation and heating, the amount of pres-
cavity should have a draft of approximately
0.002 mm/mm of depth, which prevents the ex-
trusion from sticking in the cavity. In normal op-
eration, the part stays on the punch and is
stripped from it on the upward stroke.
The tolerances for magnesium alloy extrusions
are influenced by the size and shape of the part,
Fig. 33 Limit diagram of extrusion speed, V, versus

temperature for a given extrusion load and
the alloy limit for surface cracking (hot shortness). Note:
Fig. 32 Surface cracks (hot shortness) of different de- This optimal temperature only refers to extrusion speed Fig. 34 Examples of extruded sections produced from
grees of severity on an extruded CuSn8 tube and not metallurgical development of properties. easily extrudable aluminum alloys
sure and rate of speed used for extruding, and for cable. Molten lead is used instead of billets difficult or very difficult alloys are very low (Fig.
the type of lubricant employed. A billet temper- for many applications. Vertical extrusion presses 35). With a typical extrusion ratio of 40 to 1, exit
ature that is too high can cause blisters or other are sometimes used to produce protective speeds of the more difficult alloys are in the order
surface defects, including cracking. A tempera- sheathings of lead on electrical conductors. of 0.6 to 1.2 m/min (2 to 4 ft/min).
ture that is too low increases the pressure re- Aluminum and aluminum alloys are probably Extrudability of the moderately difficult or
quirements for the extrusion and shortens tool the ideal materials for extrusion. In the alu- very difficult alloys cannot be significantly in-
life. All metals also shrink on cooling after hot minum industry, rod, bar, wire tubular products, creased by hot extrusion technology because of
extrusion; therefore, a shrinkage allowance must and shapes are termed mill products, as they are the narrow temperature interval between the ex-
be provided in the design of the dies. Deforma- in the steel industry, even though they often are trusion-load limiting temperature and the tem-
tion of the die under high pressures and expan- produced by extrusion rather than by rolling. perature of hot shortness. In extrusion of high-
sion resulting from the high temperatures must Most commercially available aluminum alloys magnesium alloys, a short cylinder of soft
also be considered in die design. can be extruded, and any shape can be produced aluminum helps to initiate more homogeneous
from easily extrudable alloys. Examples of vari- flow (Ref 36). Billet temperatures generally
ous sections produced from easily extrudable range from approximately 300 to 595 C (575 to
Materials for Hot Extrusion aluminum alloys are shown in Fig. 34 (Ref 1). 1100 F), depending on the alloy. Typical billet
Most design-for-manufacturing considerations temperatures for some of the harder aluminum
Figure 1 classifies various materials on the for aluminum extrusions depend on the diffi- alloys are listed in Table 5. This does not include
basis of typical extrusion temperatures. culty of extruding through small, complex die variations in exit temperatures.
Commonly extruded metals include aluminum al- openings. Hollow sections are quite feasible, al- Magnesium and Magnesium Alloys. Extruded
loys, lead, tin, magnesium, zinc, and steel. Other though they cost approximately 10% more per magnesium and magnesium alloy products are
metals that are hot extruded include titanium and pound produced. The added cost is often com- used in the aircraft, aerospace, and nuclear power
titanium alloys, nickel and its alloys, superalloys, pensated for by the added torsional stiffness that industries. With similar billet temperatures, the
zirconium, beryllium, uranium, and molybde- the hollow shape provides. It is best if hollow extrudability of these materials is approximately
num. Some titanium alloys are more difficult to sections can have a longitudinal plane of sym- the same as that of aluminum, but longer heating
extrude than steels. Nickel alloys also can be very metry. Semihollow features should be avoided, periods are usually necessary to ensure uniform
difficult to extrude, and billet temperatures above because semihollow features require the die to temperatures throughout the billets.
995 C (1825 F) are used. All of these metals are contain a very thinand hence relatively Zinc and Zinc Alloys. The extrusion of zinc
extruded into tubes, rods, and bars. weakneck. Sections with both thick and thin and zinc alloys requires pressures that are higher
Lead and tin exhibit high ductility and are sections are to be avoided. Metal tends to flow than those necessary for lead, aluminum, and
easy to extrude. The addition of alloying ele- faster where thicker sections occur, giving rise to magnesium. Billet temperatures generally range
ments increases the force required, but extruding distortions in the extruded shape. from approximately 205 to 345 C (400 to 650
does not present a problem and is carried out The relative extrudability of aluminum alloys, F). Applications include rods, bars, tubes, hard-
with billets heated to a maximum temperature of as measured by extrusion rate, is given subse- ware components, fittings, and handrails.
approximately 300 C (575 F). Principal appli- quently for several of the more important com- Copper and copper alloy extrusions are
cations include pipes, wire, tubes, and sheathing mercial extrusion alloys: widely used for wire, rods, bars, pipes, tubes,
electrical conductors and connectors, and weld-
Alloy Extrudability, % of rate for 6063
ing electrodes. Architectural shapes are extruded
from brass but usually in limited quantities.
1350 160 Billet temperatures vary from approximately
1060 135
1100 135 595 to 995 C (1100 to 1825 F). Depending on
3003 120 the alloy, extrudability ranges from easy to diffi-
6063 100 cult. High pressures (690 MPa, or 100 ksi, or
6061 60 more) are necessary for the extrusion of many
2011 35
5086 25 copper alloys.
2014 20 Steels. For the hot extrusion of steel, it is nec-
5083 20 essary to use glass or some other high-tempera-
2024 15 ture lubricant to prevent the excessive tooling
7075 9
7178 8 wear that can result from the high billet temper-
atures required (995 to 1300 C, or 1825 to 2375
F). In addition, high ram speeds are required in
In general, higher alloy content and strength in- order to minimize contact time between the bil-
crease the difficulty of extrusion and lower its ex- lets and the tooling. The products produced in-
trusion rate. The easily extruded alloys can be eco- clude structural sections (generally required in
nomically extruded at speeds up to 100 m/min small quantities) and tubes with small bores. For
(330 ft/min) or faster, while speeds for moderately economic reasons, steel structural shapes, espe-
Table 5 Typical values of billet temperature and extrusion speed of some harder Al alloys
Billet temperature Exit speed
Alloy Type C F m/min ft/min
20142024 Heat treatable 420450 788842 1.53.5 511

5083, 5086, 5456 Non-heat-treatable 440450 824842 26 720
7001 Heat treatable 370415 700780 0.51.5 25
7075, 7079 Heat treatable 300460 572860 0.82 37
7049, 7150, 7178 Heat treatable 300440 572824 0.81.8 2.56
Note: Temperatures and extrusion speeds are dependent on the final shape and the extrusion ratio, and it may be necessary to start with lower billet
Fig. 35 Extrusion rate versus flow stress for various temperatures than mentioned in the table. Source: Ref 1, 7
aluminum alloys. Source: Adapted from Ref 1
cially those needed in large quantities, are better Taper heating (Fig. 21) is perhaps the most
suited to the rolling process. Alloy and stainless practical method of isothermal production in a
steels are usually extruded in the form of either production environment. Additional heaters can
solid shapes or tubes. be used to transfer more heat to the front of the
Metal powders are extruded into long shapes billet, or the back end can be cooled as the billet
by cold and hot processes, depending on the is transferred to the container. Several methods
characteristics of the powders. Aluminum, cop- for isothermal extrusion of aluminum alloys are
per, nickel, stainless steels, beryllium, and ura- also described in Ref 7.
nium are some of the powders that are extruded. Billet Preparation. The more common
The powders are often compressed into billets metals that are to be extruded are generally cast
that are heated before being placed in the extru- in the form of cylindrical logs measuring 3.7 to
sion press. For many applications, the powders 6 m (12 to 20 ft) or more in length. These logs
are encapsulated in protective metallic cans, are sawed or sheared into billets of varying
heated, and extruded with the cans. Extrusion of length, depending on the cross-sectional area
metal powders is discussed in detail in the arti- and the length of the product to be extruded.
cle Extrusion of Metal Powders in Powder Additional billet preparation is sometimes nec-
Metal Technologies and Applications, Volume 7 essary, depending on the material to be extruded.
of the ASM Handbook. For example, it is necessary to machine the outer
surfaces of some steel billets before they are
Process Factors heated; the outer surfaces must then be descaled
after being heated to the extrusion temperature.
Exit Temperature. The temperature of the
Best results are attained in backward extrusion
extruded product as it emerges from the die is
by scalping the billets before extrusion to re-
one of the essential factors that influence prod-
move oxides and other impurities from the billet
uct quality. The temperatures developed during
skin. If this is not done, these impurities would
extrusion significantly influence the speed at
find their way onto the surfaces of the extrusion
which the process can be carried out. This is es-
because of the inherent nature of the metal flow
pecially true for alloys that are difficult to ex-
Fig. 36 Increase in emergent temperature versus ram in backward extrusion.
travel in the extrusion of superpure alu- trude, where the flow stress of the extruded ma-
minum. Ram speeds are indicated on the curves. Billet di-
For the hot extrusion of such materials as
terial must be kept relatively low by increasing
ameter: 38 mm (1.5 in.); billet length: 51 mm (2 in.); ex- brass, bronze, and other soft metals, the dummy
the billet preheating temperature.
trusion ratio: 16:1; starting temperature: 20 C (70 F). block is made smaller in diameter than the billet.
Source: A.R.E. Singer and J.W. Coakham, Temperature In general, a higher rate of production can be
In extruding, no lubrication is provided between
Changes Occurring During the Extrusion of Aluminum, obtained by increasing extrusion speed and/or
Tin and Lead, J. Inst. Met., Vol 89, 19611962, p 177
the bore of the container liner and the outer sur-
the extrusion ratio (for a larger reduction in
face of the billet. Consequently, friction prevents
area). However, the flow stress of the extruded
the outer surface of the billet from sliding, and
material must be kept relatively low to maintain
the undesirable skin of the billet is left in the
extrusion pressure at an acceptable level. One
container as the dummy block shears the metal
way is to increase the billet preheating tempera-
during its forward stroke. An additional press
ture, but this combination of high billet temper-
stroke is required to remove this retained metal
ature with a large reduction in area and/or high
before the next billet can be charged into the
extrusion speed causes a considerable rise in
container.
temperature in the extruded material, especially
Pressure Requirements. The unit pres-
near the surface of the section.
sures needed for hot extrusion are significant
Due to the mechanics and thermodynamics of
considerations in press selection. The determi-
the extrusion process, exit temperatures also
nation of pressure requirements is difficult for
vary with a constant ram speed. Data from this
the extrusion of complicated shapes and sec-
effect are plotted for aluminum and aluminum
tionsespecially those with thin walls. Careful
alloys in Fig. 36 and 37, respectively. Exit tem-
judgments based on past experience must be
peratures are influenced by a complex set of re-
made for estimates. Formulas have been devel-
lationships among the material and process vari-
oped for estimating pressure requirements, using
ables, such as billet material and temperature,
shape, friction, and other parameters. However,
friction, tool material and temperature, extrusion
for less complicated shapes, such as round bars
speed, shape of the extruded section, and reduc-
and tubes, a fair approximation of pressure re-
tion in area. This complex thermal situation ex-
quirements can be calculated by:
ists as soon as a heated billet is loaded into the
preheated container and extrusion begins.
Various methods have been investigated and
A
used to maintain a more constant exit tempera- p = k ln
a (Eq 8)
ture during extrusion. These methods, referred
to as isothermal extrusion, include:
Fig. 37 Surface temperature of the extruded product
versus ram displacement for two aluminum
alloys. Ram velocities are indicated on the curves.
Reducing exit speed during extrusion with where p is the extrusion pressure required; k is a
Reduction ratio: 5:1; billet diameter: 71 mm (2.8 in.); bil- controls according to measurement of exit numerical value representing the resistance to
let length: 142 mm (5.6 in.); initial billet and tooling tem- temperature deformation, usually based on past experience in
perature: 440 C (825 F). Source: G.D. Lahoti and T. Reducing extrusion speed according to pre- extruding a specific metal at a specific tempera-
Altan, Prediction of Metal Flow and Temperatures in
Asymmetric Deformation Processes, Proceedings of the
selected speed programs ture; A is the cross-sectional area of the con-
21st Sagamore Army Materials Research Conference, Aug Nonuniform taper heating to give a lower tainer liner or, in the case of tubes or other hol-
1974 temperature at the back of the billet low shapes, the cross-sectional area of the liner
minus the cross-sectional area of the mandrel (in of iron- and nickel-base alloys and have also plastohydrodynamic (PHD) mechanism is infe-
square inches or square millimeters); and a is the found application as solid die inserts. The low rior to the accustomed surface. Graphite cannot
total cross-sectional area of the extruded product thermal conductivity of many high-temperature maintain sliding friction over the entire course
(the shape area times the number of openings in materials is a disadvantage in that high tool tem- of extrusion of a long billet, although both
the die) (in square inches or square millimeters). peratures can build up. The molybdenum-base graphite and MoS2 have been used for mandrel
The extrusion pressure requirements deter- alloy TZM has been shown to give a better com- lubrication in tube extrusion. Polytetrafluoroeth-
mined with Eq 8 are useful, but the values ob- promise in properties. ylene has been found to be useful in cable extru-
tained are only approximations. The factor k Follower blocks suffer mostly from thermal sion. However, unlubricated hot extrusion of
varies with such factors as billet temperature, die fatigue and, if overloaded, from plastic deforma- aluminum alloys remains dominant.
design, type of metal extruded, amount of re- tion. The materials employed are similar to those Unlubricated Hot Extrusion of Aluminum
duction (extrusion ratio), stem speed, and con- used for containers. Mandrels are subject to high Alloys. By far, the largest quantity of alu-
figuration of the extruded product. Billet length, heat and frequently fail by plastic deformation. minum alloys is hot extruded without any lubri-
nonhomogeneous metal flow, and friction also Therefore, they are often cooled internally or, cant whatsoever. As long as the billet is of good,
influence pressure requirements. between strokes, externally. uniform quality, the die is free of major pickup,
Unit pressures generally range from 450 to Lubricated Hot Extrusion. The Ugine- and no lubricant or contaminant finds it way into
760 MPa (65 to 110 ksi), with a maximum of ap- Sejournet process is the most commonly used the container, the quality of the extrusion will be
proximately 1035 MPa (150 ksi). When practi- for the extrusion of steels and titanium alloys. In excellent (Ref 36). For best surface quality, a
cal, it is generally desirable to use a press with a this process, the heated billet is rolled over a bed follower block of smaller diameter is used to
capacity exceeding that actually required. This of ground glass or is sprinkled with glass pow- leave a skull, which is removed by specially de-
allows lower billet temperatures and faster stem der to provide a layer of low-melting glass on signed follower blocks or in a separate stroke.
speeds to be used and provides improved prop- the billet surface. Before the billet is inserted The skull entraps imperfections of the billet sur-
erties in the extruded products. into the hot extrusion container, a suitable lubri- face. When extrusion is done without a skull, the
Ram Speeds. Optimal stem speeds are es- cating system is positioned immediately ahead billet is scalped for highest quality. Shearing
sential for hot extrusion. Excessive speed can of the die. This lubricating system can be a com- along the container wall contributes to heat gen-
cause overheating of the billet as well as tears pacted glass pad, glass wool, or both. The prelu- eration; in direct extrusion it may reach 50% of
and other surface defects. A speed that is too bricated billet is quickly inserted into the con- the total force (Ref 27) and thus greatly adds to
slow reduces productivity and increases the retainer, along with the appropriate followers or a the temperature rise. Reverse extrusion imposes
quired extrusion pressure because of billet cool- dummy block. The extrusion cycle is then much smaller strains. The work of shearing
ing. Slow speeds can also decrease tool life be- started. along the container wall does not contribute to
cause of prolonged contact time between the As a lubricant, glass exhibits unique charac- an increase in the temperature of the extruded
tools and the hot billet. Typical stem speeds for teristics, such as its ability to soften selectively product, and a threefold increase in speed is pos-
various metals are: during contact with the hot billet and, simulta- sible (Ref 36).
neously, to insulate the hot billet material from In nonlubricated hot extrusion of aluminum
Stem speed the tooling. The tooling is usually maintained at alloys, a flat-face (shear-face) die is often used.
Material mm/s in./s a temperature that is considerably lower than The main problem in extrusion with sticking
Steel 152203 68 that of the billet. In the extrusion of titanium and friction is the die. The die is designed and made
Copper 5176 23 steel, the billet temperature is usually 1000 to so that optimal material flow is ensured and an
Aluminum 12.725.4 -1 1250 C (1830 to 2280 F), but the maximum extrusion free of defects is produced. The die
Brass 2551 12 temperature the tooling can withstand is 500 to usually has a flat face, which is protected by the
550 C (930 to 1020 F). Therefore, compatibil- dead-metal zone. However, the high adhesion of
The use of variable-delivery pumps and ad- ity can be attained only by using the appropriate aluminum to steel soon results in the buildup of
justable valves facilitates control of stem speed. lubricants, insulative die coating, and ceramic a coating on the die and die land. Local lubrica-
Automatic control is available for maintaining die inserts and by designing dies to minimize tion of the die land or of the billet face with a
constant speed throughout the extruding cycle. tool wear. Glass lubricants have performed satis- graphitic lubricant is sometimes practiced, even
Lubrication is another important operating pa- factorily on a production basis in extruding long though the lubricant is lost soon after the extru-
rameter. The types of lubricants used and the ef- lengths. sion emerges (Ref 37) and extrusion proceeds
fects of lubrication are discussed in the section The choice between grease and glass lubri- over the bare die. No lubricant is allowed in
Lubricated Hot Extrusion in this chapter. cants is based mainly on the extrusion tempera- some press shops, because it may encourage at-
Dies and Tooling. The extrusion of long ture. At low temperatures, lubrication is used tempts at more extensive lubrication, with con-
bars and tubes imposes some of the most severe only to reduce friction. At moderate tempera- sequent material defects, as shown in Fig. 19.
demands on the various tooling elements. High tures, there is also some insulation between the Die coatings can be helpful. Pickup is typical of
pressures and sliding velocities combine with hot billet and the tooling from the use of par- bare tool steel.
adhesive and abrasive action and, in hot extru- tially molten lubricants and vapor formation in
sion, with sudden temperature fluctuations and addition to the lubrication effect. At tempera-
prolonged exposure to high temperatures. It is, tures above 1000 C (1830 F), the thermal in- Hydrostatic Extrusion
therefore, not surprising that some of the im- sulation of the tooling from overheating is of
provements in tool materials were first applied equal importance to the lubricating effect, par- In hydrostatic extrusion, the billet is pushed
to extrusion dies. ticularly with difficult-to-extrude alloys. The lu- through the die by the action of a liquid pressure
Hot working die steels are generally adequate brication film can also impede oxidation. medium rather than by direct application of the
except in isothermal extrusion of titanium, for Lubricated Hot Extrusion of Aluminum load with a ram. The billet is completely sur-
which superalloy dies are more satisfactory. Alloys. Most aluminum alloys are hot extruded rounded by a fluid that is sealed off and pressur-
Much work has gone into increasing wear resist- without any lubricant whatsoever. Lubricated ized to extrude the billet through the die (Fig. 3b).
ance by selection of appropriate steels, by use of hot extrusion of aluminum alloys has found only For all practical purposes, hydrostatic extrusion is
surface treatments such as nitriding and W2C limited application. It is not easy to identify an a form of lubricated extrusion, where pressure ap-
and W3C coating, and by use of tungsten carbide obviously best lubricant. There are few glasses plied to the inlet zone aids in developing a hydro-
inserts. Ceramic coatings, particularly ZrO2 and of the right viscosity for phase-change lubrica- dynamic film. Hydrostatic extrusion has attracted
Al2O3, have come into general use in extrusion tion, and the rough product surface typical of the a great amount of interest, because envelopment of
long billets are to be extruded. If the pressure

medium has lubricating qualities, friction is
much reduced over the die face. This permits the
use of lower die angles, which then ensures
greater homogeneity of deformation.
The hydrostatic pressure acting on the work-
piece can be increased by taking advantage of
low die friction, by extruding with a low die
half-angle (low h/l ratio), or by extruding into a
pressurized medium (Fig. 38b). With the latter
technique, even brittle materials can be de-
formed, although at the expense of high con-
tainer pressures. These parameters are described
in more detail in Ref 41, which describes an an-
alytical model for optimization in terms of vari-
ous process parameters, such as die cone angle,
the reduction ratio, the friction coefficient, the
extrusion and back-pressure ratio, and the die
shape.
Nonetheless, a basic problem of all of the pre-
Fig. 38 Basic process of hydrostatic extrusion. (a) Simple hydrostatic extrusion. (b) Extrusion into a pressurized
viously mentioned techniques is low production
medium. (c) Thick film of semisolid lubricant to the surface of the billet. (d) Dynamic seal with a controlled
leakage rate. (e) Punch with a high-pressure seal that bears on the billet. (f) Drawing augmented hydrostatic extrusion rate; the container has to be pressurized for each
billet, and large quantities of fluid have to be
supplied and drained. A thick film of semisolid
the billet in a fluid medium reduces or practically much reduced mandrel friction makes it advan- lubricant applied to the surface of the billet al-
eliminates major sources of friction such as con- tageous for extrusion of tubes. Applications of lows the billet to be loaded as in usual cold ex-
tainer friction and, in extrusion of tubes, friction hydrostatic extrusion include the production of trusion but into a preheated container in which
on the mandrel surface. These are major contribu- copper tubing, copper-clad aluminum wire, fine the lubricant melts to become the pressurized
tors to extrusion pressure and thus limit both billet wires of precious metals, aluminum alloy tubes, medium (Fig. 38c) (Ref 42). If the clearance
length and extrusion ratio for given maximum and niobium-titanium superconducting wires around the punch is small enough to act as a dy-
punch and container pressures (Fig. 12). (Ref 40). namic seal with a controlled leakage rate (Fig.
The concept of hydrostatic extrusion can be 38d), sufficient pressure for extrusion might be
traced back to 1893 (Ref 38), when an augmented Methods of Hydrostatic Extrusion developed. Less practically, lubricant may also
hydrostatic extrusion process, known as Hydraw, be trapped with a billet that has a flange at the
was patented. In the Hydraw process, a liquid in a Because hydrostatic extrusion requires less punch end. Extrusion at high velocities is also
pressure chamber causes the wire to extrude power than conventional extrusion, it continues helpful in minimizing leakage.
through a die, while speed control is maintained to attract attention. As noted, the primary advan- A common problem is that an excessively
by a moderate pull on the wire at the exit. This tages are: thick lubricant film not only roughens the prod-
process is referred to as augmented hydrostatic uct but may also generate periodic instability,
There is no friction between the billet and
extrusion process, because a slight drawing force leading to bambooing that manifests itself in pe-
the container. Therefore, the pressure at the
augments the extrusion pressure. Although this riodic variations in the surface finish and diame-
beginning of extrusion is much lower, and
particular method has been recognized as a po- ter of the extruded product. To suppress this ef-
billets of any length can theoretically be ex-
tential for economical production of fine wire fect, a billet-augmented process has been
truded into semifinished products. Billet
with the goal of replacing multiple-pass wire considered. In this technique, the punch has a
length is limited only by container construc-
drawing with large reductions in a single pass or high-pressure seal that bears on the back face of
tion, and, if desired, a spool of wire may be
to produce wire of materials that cannot be drawn the billet (Fig. 38e). Alternatively, the extruded
placed inside the container and long lengths
by conventional methods, the concept of aug- product may be subjected to a drawing force
of fine wire extruded.
mented hydrostatic extrusion for the production (Fig. 38f). The combined extrusion-drawing
Friction at the die can be significantly re-
of fine wire still has difficulties (Ref 39). process creates the advantage of lower die pres-
duced by a film of pressurized lubricant be-
This section focuses on the method of simple sures and the possibility of heavier reductions.
tween the deforming metal and the die sur-
hydrostatic extrusion (Fig. 38a) and some vari- In extrusion of tubes, drawing also reduces the
face.
ants of it (Fig. 38bf). The first commercial ap- frictional stresses on the mandrel. In extrusion
The lower extrusion pressures and the re-
plications of simple hydrostatic extrusion have through two dies in tandem, the first reduction is
duced die friction of hydrostatic extrusion
been in cases when conventional processes have equivalent to drawing under hydrostatic pressure
allow the use of either higher extrusion ra-
shortcomings. Hydrostatic extrusion is prima- with fluid-film lubrication, and the second re-
tios or lower extrusion temperatures.
rily used when conventional lubrication is inad- duction, where most of the work is done, corre-
The uniform hydrostatic pressure in the con-
equatefor example, in the extrusion of special sponds to hydrostatic extrusion.
tainer means that billets do not have to be
alloys, composites, clad materials, or brittle ma- The general limitations of the hydrostatic ex-
straight; coiled wire can also be extruded.
terials that cannot be processed by conventional trusion process include:
extrusion. Hydrostatic extrusion allows greater However, there are a number of practical dif-
reductions in area (higher extrusion ratios) than ficulties that have led to different techniques. In Containment of the fluid under high pres-
either cold or conventional hot extrusion. The the basic process of simple hydrostatic extru- sure (up to 2 GPa, or 290 ksi) requires reli-
large reduction capability of hydrostatic extrusion, the pressure is built up by the penetration able seals between the container bore sur-
sion has prompted much work into wire proof a punch that traps the fluid medium with a face and both the ram and die. The
duction. The greater homogeneity of deforma- high-pressure seal (Fig. 38a). Contact between technology required to achieve dependable
tion also makes hydrostatic extrusion container and billet surfaces is eliminated, seals at these points is widely available,
particularly favorable for cladding, and the which can be particularly advantageous when however. Also, sealing between the billet
nose and the die can easily be achieved by extruded, the mandrel. The container is sub- efficient of friction dropped with increasing vis-
chamfering or tapering the billet nose to jected to very high pressures (up to 2 GPa, or cosity (Ref 25). Bambooing occurred only with
match the entry angle of the die. 290 ksi), and a large hoop stress is imposed at the stiffer, less compressible castor oil and not
In addition to being tapered to match the die the bore. The ram is under a uniaxial compres- with higher-viscosity silicone oils. It has been
opening angle, the billet is also usually ma- sive load and generally has a much longer life shown that the extrusion actually comes to a mo-
chined all over to remove surface defects than the container. The die and mandrel have mentary rest but then accelerates sufficiently to
that would otherwise reappear on the ex- considerably shorter service lives than the con- establish, even though for only a brief time, a
truded product. This is especially true when tainer, because they are usually subjected to fluid film (Ref 46). Other research noted that the
cast billets are being used. larger stresses, higher temperatures, and more bamboo defect is suppressed with increasing ex-
wear. The stability of extrusion increases also trusion speed (Ref 47). Stick-slip combined with
Other limitations of the process arise when a
when an elastomer is used for the punch and a poor die lubrication can even result in firtreelike
relatively large volume of fluid is used com-
separate lubricant is applied to the billet. cracking, and therefore, lubrication reduces the
pared to the billet volume to be extruded. These
Tool materials used for hydrostatic extrusion defect.
include:
include alloy steels, higher-alloy tool steels, The origin of the bamboo defect lies in the lu-
Increased handling for injecting and remov- 18Ni maraging steels, nickel- and cobalt-base brication mechanism, but its development, mag-
ing the fluid for each extrusion cycle superalloys, and, in some applications, ce- nitude, and damping out depend on the entire
Reduced control of billet speed and stop- mented carbides (usually 5 to 15% Co binder system. Thus, billet augmentation suppresses
ping due to potential stick-slip and exces- phase). Table 6 lists some tooling applications of fluctuations not only by physically restraining
sive stored energy in the compressed fluid these materials, and more detailed information the billet but also by reducing the film thickness.
Reduced process efficiency in terms of bil- on the suitability of materials for hydrostatic ex- This, incidentally, increases die friction.
let-to-container volume ratio trusion tooling is available in Ref 43. Because the billet is now pushed (Fig. 38e), it
Increased complications when extruding at first upsets if the billet-to-container clearance is
elevated temperatures too large for the prevailing lubricant viscosity; in
Hydraulic Medium and Lubrication contrast, lubricant starvation sets in when the
The problems of billet speed and stopping (Adapted from Ref 8) clearance is too small. Bambooing may still
control can be reduced by using viscous
occur if hv/p is excessive, a is small, or the die
dampers and by improving lubrication at the bil-
The hydraulic medium must be chosen to de- land is slightly tapered inward. If the augment-
let-die interface. Another approach to minimiz-
velop the required pressure without solidification. ing pressure is too high, the billet is upset into
ing this and the other problems cited previously
Thus, a medium with good lubricating properties, the die throat, lubricant excess is reduced, and
is to keep the amount of pressurizing fluid to an
such as castor oil, may be used if pressures are not the entry geometry is impaired, resulting in a
absolute minimum, as in the Hydrafilm process
too high, as in model experiments with wax or in rise in friction and a change to boundary lubri-
(see the section The Hydrafilm Process in this
extrusion of soft metals. Otherwise, favorite sub- cation.
chapter).
stances are low-viscosity mineral oils, castor oil The same effects were noted in extrusion with
Equipment designs for hydrostatic extrusion
with methylated spirits (useful up to 1.4 GPa, or a controlled follower block clearance (Fig. 38d)
presses are becoming more similar to those of
200 ksi), glycerine with 25% ethylene glycol (2.0 (Ref 48). The whole billet is extruded if loss of
conventional extrusion presses, as many of the
to 2.8 GPa, or 290 to 400 ksi), and isopentane and fluid around the follower block is limited. The
problems have been resolved (Ref 43). The high
gasoline (to 3.0 GPa, or 435 ksi). leakage rate, Q, can be calculated from:
pressures and the low-viscosity media necessi-
Because of the limitation imposed by solidifi-
tate the use of special containers and seals to
prevent leakage of the pressure medium. Various
cation, lubrication is frequently of the mixed- pac 3 dp
film type, as evidenced by the need for the pres- Q=
methods are used to prevent the stick-slip phe- 6 h dl (Eq 9)
sure medium to be compounded with boundary
nomenon, which was once considered a serious
and extreme-pressure additives (and even with
problem in hydrostatic extrusion. Various ingen-
layer-lattice compounds) and by the response of where a is the radius of the container, c is the ra-
ious schemes have been proposed to extrude dial clearance, and dp/dl is the rate of pressure
surface finish to die angle, reduction, and the ef-
bars continuously and semicontinuously, as re- drop.
fect of viscosity, h, and the velocity-pressure
viewed in Ref 44 and 45. Mixed-film lubrication prevails also in hydro-
ratio, v/p, on film thickness (where film thick-
The essential components of tooling for hy- static extrusion/drawing (Fig. 38f), as shown by
ness is proportional to hv/p). Often, it is found to
drostatic extrusion are the high-pressure con- the roughening of the product surface with in-
be more practical and effective to apply a sepa-
tainer, the ram, the die, and, if tubing is being creasing speed (Ref 49). Deformation of the bar
rate lubricant film, such as a grease or, in the
case of steel, a conventional phosphate/soap prior to entry into the die and a rougher emerg-
coating to the billet, and to use the hydrostatic ing surface indicate a shift toward a more hydro-
Table 6 Tool materials for hydrostatic dynamic contribution at higher extrusion/draw
extrusion and some typical applications feature only to reduce friction and improve the
homogeneity of deformation. Oxidation of the stress ratios, at least for a strain-hardening mate-
Material Typical applications
surface prior to coating with soap has also been rial such as copper.
High-strength alloy steel Outer container rings found to be effective. Of course, under such con-
Hot work tool steels ditions, the benefits of hydrostatic operation be- Hot Hydrostatic Extrusion
(e.g., AISI H11, H12) Inner container liners; come rather dubious, because pressures are
rams for relatively
low working hardly reduced relative to those encountered in Hydrostatic extrusion was originally used
pressure; dies well-lubricated conventional extrusion. only for cold working; that is, the billet was not
18Ni maraging steels Dies; container liners Because pressures in hydrostatic extrusion are preheated. Recently, however, more emphasis
and rams for high sufficient to initiate yielding of the billet, the lu- has been placed on the development of hot hy-
working pressures
Nickel- and cobalt-base bricant film may easily thicken to the point drostatic extrusion processes. Hot hydrostatic
superalloys Dies for high-temperature where it becomes unstable, and bambooing is extrusion is particularly applicable to difficult-
applications often observed. In experiments on wax as a to-work materials, such as high-strength alu-
Cemented carbides Die inserts; mandrel tips model material, it was noted that the defect was minum alloys, titanium alloys, refractory metals
AISI, American Iron and Steel Institute. Source: Ref 43
associated with fluctuations between PHD and and alloys, bimetallic products, and multifila-
boundary conditions, and that the calculated co- ment superconductors. The hot process has also
been used for the production of copper tubing at also be used for some applications. Other poten- reasons, the fluid-to-fluid process may not be
extrusion ratios on the order of 500 to 1. tial advantages of the Hydrafilm process include suitable for many industrial applications.
One of the main problems associated with hot (Ref 51): The problem of extruding low-ductility met-
hydrostatic extrusion has been that the pressure als was approached in a different way by re-
The process is simplified and is very similar
media used in cold or warm processes (usually searchers at Battelle Columbus Division (Ref
in operation to conventional extrusion.
castor or other vegetable oils) ignite and burn at 52), who established that the cracks or fracture
Conversion of conventional presses to hy-
high temperatures. Therefore, research in the first developed in the rear section of the die
drostatic extrusion is facilitated.
area of hot hydrostatic extrusion has focused on land, immediately before the exit plane, and that
The Hydrafilm concept is applicable to
the development of processes using pressure the surface cracking resulted from residual ten-
high-speed mechanical presses; production
media that can withstand elevated temperatures. sile stresses as the product left the die. The
rates of 30 to 40 billets/min may be possi-
There is also the potential for hot extrusion with cracks observed were either longitudinal or
ble.
grease, polymer, or glass media, which melt at transverse across the extruded product, depend-
The process is adaptable to hot extrusion at
the extrusion temperature and thus transmit the ing on whether the predominating residual
billet temperatures up to and beyond 1200
extrusion pressure. stresses were longitudinal or circumferential.
C (2190 F).
One team of investigators proposed the use of This phenomenon was noted much earlier in rod
a viscoplastic pressure medium for hot hydro- The Hydrafilm process has been used for ex- and tube drawing (Ref 53). It was discovered
static extrusion (Ref 50). Because such materials truding tubes of various materials, including tub- that it is possible to reverse the residual stresses
are soft solids at room temperature, a piece of ing of Ti-6Al-4V, American Iron and Steel at the surface to compressive stresses by a sub-
the pressure medium can be introduced into the Institute (AISI) 4140 steel, AISI type 316 stain- sequent draw with a low reduction in area
container without the need for a charging pump. less steel, and Zircaloy-2; aluminum, copper, (<2%).
This simplifies machine design; the construction and steel rod; and copper and steel shapes. More The work of these investigators led to the de-
of a press used in this manner can be as simple information on these applications is available in velopment of the double-reduction die (Ref
as that of a conventional extrusion press. Ref 51. 53). Figure 40 compares a standard die with a
Viscoplastic pressure media used for hot hy- double-reduction die. The double-reduction die
drostatic extrusion include: Simple Hydrostatic Extrusion of Brittle used for the experiments was designed to give
Materials a 2% reduction in the second step. This method
Waxes or fats, such as beeswax, carnauba
has been successfully applied to the extrusion
wax, montan wax, lanolin, and complex
Most brittle materials are subject to circum- of brittle materials, including beryllium and
waxes
ferential (transverse) and longitudinal surface TZM molybdenum alloy, without any cracking.
Soap-type greases composed of petroleum
cracking during hydrostatic extrusion. This The lubricant used was polytetrafluoroethyl-
oil and such soaps as fatty acids or soaps of
cracking can be avoided through the use of ei- ene, and the pressurizing fluid was castor oil.
sodium, calcium, or lithium; such mixtures
ther fluid-to-fluid extrusion or double-reduction The results may be applicable to conventional
may also contain graphite.
dies. In fluid-to-fluid extrusion, the billet is hy- cold extrusion through a lubricated conical die
Mixtures of nonsoap greases and silica or
drostatically extruded into a fluid at a lower (Ref 52).
other metal oxides; mixtures of petroleum
pressure. This method has several disadvan- It is believed that the small second reduction
oil and bentonite are heat resistant up to
tages, including high tooling and operating prevents cracking by imposing an annular coun-
1200 C (2190 F).
costs, extrusion lengths that are limited to the terpressure on the extrusion as it exits the first
High-molecular-weight polymers, such as
length of the secondary chamber, and increased portion of the die. This counters the axial tensile
polyethylene; the properties of these materi-
fluid pressure required for extrusion. For these stresses arising from residual stresses, elastic
als depend on their molecular weight and
the additives used.
Some metal oxides, salts, and glass can also be
used as pressure media for hot hydrostatic extru-
sion, but these materials may adhere to the ex-
truded product and can be difficult to remove.
The Hydrafilm Process

The method known as the Hydrafilm, or
thick-film, process was developed at Battelle
Columbus Division to overcome many of the
disadvantages of pure hydrostatic extrusion, in
which a large volume of hydrostatic fluid is usu-
ally employed. The Hydrafilm process is charac-
terized by the relatively low volume of pressur-
izing fluid used and by the relatively thick film
of lubricant that is maintained around the billet
(Fig. 39).
The small amount of pressurizing fluid used
in the Hydrafilm process minimizes fluid-han-
dling time, thus increasing billet cycle rates. It
also minimizes billet chilling, which is impor-
tant in hot hydrostatic extrusion. For cold or
warm hydrostatic extrusion, the billet can be Fig. 39 Schematics of the Hydrafilm (thick-film) Fig. 40 (a) Standard die and (b) double-reduction die
process. (a) Before extrusion. (b) During ex- for the hydrostatic extrusion of brittle materi-
precoated with the pressurizing medium by dip- trusion. 1, extrusion ram; 2, billet; 3, film of pressurizing als. H, distance between the start of each bearing; q, in-
ping or spraying; lubricants that are different liquid; 4, separate film of lubricant; 5, sealing ring. cluded angle at second reduction. Dimensions given in
and separate from the pressurizing medium can Source: Ref 51 inches. Source: Ref 51
bending, and friction. Prevention of circumfer- truding hollow products with bridge, spider, shown, although it must be understood that
ential cracks on exit from the second portion of or porthole dies, because use of a lubricant under hydrodynamic or mixed-film condi-
the die is believed to be associated with the fa- would prevent rewelding of the separated tions, the coefficient of friction is greatly af-
vorable permanent change in residual stresses in material streams. Unlubricated extrusion is fected by process conditions. At the high
the workpiece caused by the small second re- desirable for hot extrusion of aluminum al- pressures developed with high extrusion ra-
duction (Ref 51). loys with flat dies, because the surface of tios, sticking may be attained even with rel-
the extruded product is freshly formed by atively good lubricants.
Summary (Ref 8) internal shear in the billet and is feasible
also for nonisothermal hot extrusion of cop-
ACKNOWLEDGMENTS
In extrusion of long, semifabricated prod- per and its alloys.
ucts, friction is generally unnecessary and Lubricated extrusion is essential for hot ex-
Significant portions of this chapter were
undesirable. Friction on the die increases trusion of steel, nickel- and titanium-base al-
adapted from:
extrusion pressures and, for a given die loys, refractory metal alloys, and high-
geometry, impairs the homogeneity of de- strength copper alloys, as well as for all cold Conventional Hot Extrusion, Forming and
formation. If, under the prevailing pres- extrusion. The geometry of the process makes Forging, Volume 14, ASM Handbook, 1988
sures, the shear strength of the interface exit particularly favorable for lubrication with Hydrostatic Extrusion, Forming and
ceeds the shear flow stress of the workpiece glasses that melt gradually to form a lubricat- Forging, Volume 14, ASM Handbook, 1988
material, a dead-metal zone forms. ing film. The billet surface becomes the sur- K. Laue and H. Stenger, Extrusion:
In direct (forward) extrusion, friction on the face of the extrusion, modified by surface ex- Processes, Machinery, and Tooling,
container wall increases the extrusion pres- tension and by the presence of the lubricant. American Society for Metals, 1981
sure. In the limit, sticking friction develops, Friction is reduced, and the homogeneity of J.A. Schey, Chapter 8: Extrusion of Semi-
and extrusion proceeds by shearing through flow is improved, by lubrication. Lubricant Fabricated Products, Tribology in
the billet, leaving behind a skull. For a given film thickness increases with viscosity, ex- Metalworking: Friction, Lubrication, and
press capacity, the length of billet that can trusion speed, and decreasing die angle. Wear, American Society for Metals, 1983, p
be extruded is limited by container friction. Tapering of the billet eliminates break- 403441
Friction on the container wall is immaterial through pressures. Film thickness increases
in indirect (reverse) extrusion, because there also with the pressure applied to the lubri-
is no relative movement. cant, and thus, hydrostatic extrusion not REFERENCES
Friction contributes to heat generation and only eliminates container friction but can
limits attainable reductions and speeds in also reduce die friction. It is particularly ef- 1. K. Laue and H. Stenger, Extrusion: Proc-
hot extrusion. Friction on the die land pro- fective in reducing friction on the mandrel esses, Machinery, and Tooling, American
motes surface cracking (speed cracking, fir- in tube extrusion. Excessive film thickness Society for Metals, 1981
tree effect). leads to rough surfaces and, possibly, to pe- 2. Y.-J. Huang, Integrated Computer Aided
Friction is beneficial only when there is riodic collapse of the film (bamboo defect). Extrusion Process Simulation, Die Design/
danger of internal (arrowhead) cracking or Severe sliding and prolonged exposure to Manufacturing, and Management Planning
when the container is moved so that the fric- high temperatures and high pressures com- System, Light Met. Age, Vol 56 (No. 56),
tional force contributes to the extrusion bine to make hot extrusion extremely de- June 1998, p 7681
force. Friction on the die and die land may manding on die materials. Adhesive and 3. C. Devadas, O.C. Celliers, D. Greliche, and
be used to equalize material flow in extru- abrasive wear, thermal fatigue, and plastic X. Zhang, Mathematical Modelling of
sion of complex shapes. deformation may all occur. Surface treat- Extrusion Process to Enhance Die Design,
Extrusion is a process that can only be con- ments such as nitriding are beneficial even Aluminum 2000 Conference, Second
ducted either totally unlubricated or fully lu- in unlubricated extrusion, and ceramic die International Congress on Aluminum, Vol II,
bricated. Intermediate conditions are undesir- coatings are often essential in high-temper- 29 March to 4 April 1993 (Florence, Italy),
able, because changeover from unlubricated ature glass-lubricated extrusion. Interall Publications, 1993, p 353365
to lubricated flow during extrusion of a billet The most commonly used lubricants are 4. C.O. Stockdale, The Application of CAD/
results in formation of subsurface defects. given in Table 7. For purposes of prelimi- CAM to Extrusion Die Design, Fifth
Unlubricated extrusion is essential for ex- nary calculation, typical m values are also International Aluminum Extrusion Technol-
ogy Seminar, Vol I, 1922 May 1992
Table 7 Commonly used extrusion lubricants and typical friction values (Chicago, IL), The Aluminum Association,
1992, p 391394
Hot extrusion Cold extrusion 5. R.H. Wagoner and J.-L. Chenot, Metal
Material Lubricant(a)(b) m(c) Lubricant m Forming Analysis, Cambridge University
Steels GR (for short pieces) 0.2 NA ... Press, 2001, p 224
Glass (10100 Pa s) pad and coating 0.02 6. V. Nagpal, C.F. Billhardt, and T. Altan, J.
Stainless steels and Ni alloys Glass (20200 Pa s) pad and coating 0.03 NA ... Eng. Ind. (Trans. ASME), Vol 101, 1979 p
Melting solids 0.05
Al and Mg alloys None ST Lanolin 0.07
319325
None (GR on die face) ST 7. P.K. Saha, Aluminum Extrusion Technology,
Cu and Cu alloys GR (for short pieces) 0.2 Castor oil (for hydrostatic 0.03 ASM International, 2000
extrusion) 8. J.A. Schey, Chapter 8: Extrusion of Semi-
None (with or without skull) ST
Glass (10100 Pa s) pad and coating 0.05
Fabricated Products, Tribology in Metal-
Ti alloys GR (for short pieces) 0.2 NA ... working: Friction, Lubrication, and Wear,
Glass (10100 Pa s) coating + GR on die 0.05 American Society for Metals, 1983, p 420,
Glass (10100 Pa s) pad and coating 0.03 435
Refractory metals Glass (10200 Pa s) coating + GR on die 0.05 NA ...
Glass (10200 Pa s) pad and coating 0.03
9. H. Kudo and H. Takahashi, CIRP Ann., Vol
13, 1965, p 7378
(a) GR = graphite, often with a polymeric binder. (b) Viscosities of glasses are given for typical extrusion temperatures. (c) When mp = k is satisfied, 10. B. Avitzur, Wire J., Vol 7 (No. 11), 1974, p
sticking friction (ST) sets in.
7786
11. G.H. Townend and D.G. Broscomb, Proc. 28. K.F. Ziehm, Light Met. Age, Vol 28 (No. 42. R.J. Fiorentino, G.E. Meyer, and T.G. Byrer,
Inst. Mech. Eng., Vol 169, 1955, p 671676 11/12), 1970, p 610 Metall. Met. Form., Vol 39, 1972, p 200203
12. N.H. Polakowski and L.B. Schmitt, Trans. 29. D. Green, J. Inst. Met., Vol 100, 1972, p 43. M. Yamaguchi, Plant Equipment for
AIME, Vol 218, 1960, p 409416 295300 Hydrostatic Extrusion, Hydrostatic Extru-
13. I. Perlmutter, V. DePierre, and C.M. Pierce, 30. B. Avitzur, Paper MF 75-140, SME, sion: Theory and Applications, N. Inoue
in Friction and Lubrication in Metal (Dearborn), 1975 and M. Nishihara, Ed., Elsevier Applied
Processing, ASME 1966, p 147161 31. J.T. Black and W.G. Voorhes, J. Eng. Ind. Science Publishers, 1985, p 164193
14. V. DePierre, J. Lubr. Technol. (Trans. (Trans. ASME), Vol 100, 1978, p 3742 44. J.M. Alexander, Mater. Sci. Eng., Vol 10
ASME), Vol 92, 1970, p 398405 32. F.J. Fuchs and G.L. Schmehl, Wire J., Vol 6 (No. 2), 1972, p 7074
15. B. Lengyel and D. Mutlu, in Proc. 19th Int. (No. 11), 1973, p 5357 45. E. Tuschy, Metall., Vol 36, 1982, p 269279
MTDR Conf., Macmillan, 1979, p 265269 33. B. Lengyel and S. Kamyab-Tehrani, in (in German)
16. T. Sheppard and A.F. Castle, in Proc. 16th Int. Proc. 17th Int. MTDR Conf., Macmillan, 46. D. Egerton and W.B. Rice, J. Eng. Ind.
MTDR Conf., Macmillan, 1976, p 535545 1977, p 423428 (Trans. ASME), Vol 98, 1976, p 795799
17. R.J. Wilcox and P.W. Whitton, J. Inst. Met., 34. P.W. Wallace, K.M. Kularni, and J.A. 47. P. Krhanek and J. Jakes, in Engineering
Vol 88, 19591960, p 145149 Schey, J. Inst. Met., Vol 100, 1972, p 7885 Solids Under Pressure, H.L.I.D. Pugh, Ed.,
18. R. Akeret and P.M. Stratman, Light Met. 35. A.W. Duffill and P.B. Mellor, Proc. Inst. Inst. Mech. Eng., 1971, p 117121
Age, Vol 31 (No. 3/4), 1973, p 6, 810; (No. Mech. Eng., Vol 180 (Pt. 3I), 19651966, p 48. K.M. Kulkarni and J.A. Schey, J. Lubr.
5/6), p 1518 260269 Technol. (Trans. ASME), Vol 97, 1975, p
19. J.R. Johnson and W.R.D. Wilson, ASLE 36. R. Chadwick, Met. Mater., Vol 4, 1970, p 2531
Trans., Vol 24, 1981, p 307316 201207 49. P. Dunn and B. Lengyel, J. Inst. Met., Vol
20. J.A. Rogers and G.W. Rowe, J. Inst. Met., 37. G. Merk and S.E. Naess, Z. Metallkd., Vol 100, 1972, p 317321
Vol 95, 1967, p 257263 68, 1977, p 683687 50. M. Nishihara, M. Noguchi, T. Matsushita,
21. H. Unckel, Aluminium, Vol 25, 1943, p 38. H.L.D. Pugh, Hydrostatic Extrusion, Cu 86 and Y. Yamaguchi, Hot Hydrostatic
342346 Copper Tomorrow; TechnologyProducts Extrusion of Nonferrous Metals, Proceed-
22. C.E.N. Sturgess and T.A. Dean, J. Mech. Research, Conference Proceedings, (Flor- ings of the 18th International MTDR Con-
Work. Technol., Vol 3, 1979, p 119135 ence, Italy), Europa Metalli-LMI, 1989, p ference, 1977, p 9196
23. P. Loewenstein, J.G. Hung, and R.G. 6783 51. R.J. Fiorentino, Selected Hydrostatic
Jenkins, Refractory Metals and Alloys, Vol 39. B. Avitzur, Limitations on Wire Size During Extrusion Methods and Extruded Materials,
111, The Metallurgical Society AIME, Drawing and Extrusion, Wire J. Int., Vol 25 Hydrostatic Extrusion: Theory and
Gordon and Breach, 1965, p 151167 (No. 3), Mar 1992, p 5156 Applications, N. Inoue and M. Nishihara,
24. R. Chadwick, Met. Rev., Vol 4, 1959, p 40. N. Inoue, Introduction, Hydrostatic Ed., Elsevier Applied Science Publishers,
189255 Extrusion: Theory and Applications, N. 1985, p 284322
25. A. Wuerscher and W.B. Rice, J. Eng. Ind. Inoue and M. Nishihara, Ed., Elsevier 52. R.J. Fiorentino, B.D. Richardson, and A.M.
(Trans. ASME), Vol 94, 1972, p 795799 Applied Science Publishers, 1985, p 16 Sabroff, Hydrostatic Extrusion of Brittle
26. B. Avitzur, Wire J., Vol 11 (No. 3), 1978, p 41. H. Elkholy, Parametric Optimization of Materials: Role of Design and Residual Stress
7884 Power in Hydrostatic Extrusion, J. Mater. Formation, Met. Form., 1969, p 107110
27. R. Chadwick, Int. Met. Rev., Vol 25, 1980, p Process. Technol., Vol 70 (No. 13), Oct 53. H. Buhler, Austrian Patent 139,790, 1934;
94136 1997, p 111115 British Patent 423,868, 1935
Chapter 21
Workability and Process Design

in Extrusion and Wire Drawing
Roger N. Wright, Rensselaer Polytechnic Institute
EXTRUSION AND WIRE DRAWING are ability so that the most efficient processing the allowable level of extrusion pressure (analo-
very analogous processes with regard to the schedule can be set forth. gous to the drawing-stress limitation in wire
manner of metal flow through the die, and they drawing) or by the onset of some degree of duc-
have somewhat comparable workability prob- tile fracture. Of course, the extrusion-pressure
lems. The major differences in workability be-
Multipass Workability versus limitation is not a workability limitation, per se,
havior involve the difference in hydrostatic Single-Pass Workability because workpiece defect development or frac-
stress state (the flowing metal is generally sub- ture is not implicit.
jected to much greater compression in extrusion It is not quite enough to define workability in
than in wire drawing), differences in working extrusion and wire drawing as the ability of the
temperature (extrusion is usually hot while metal to be worked without fracture or fracture- General Observations on Fracture
drawing is usually cold), and the obvious differ- type defects. It must be clear whether the ability and Flaw Development
ences in hardware and process layout. to be worked concerns a single pass or a series
Workability in these cases is the ability of the of passes. The question mostly applies to wire The limits of workability in extrusion and
workpiece metal to undergo extrusion or draw- drawing, because most extrusion is done as a drawing may be associated with the develop-
ing without fracture or defect development. As single-pass operation. In a single wire-drawing ment of localized surface cracks, centerbursts, or
in any of the deformation processes, the prob- pass, the material cannot be worked beyond a re- gross fracture. While the surface cracks and cen-
lems of metallurgical workability are difficult to duction that produces a drawing stress, sd, that terbursts are subcritical flaws not necessarily
separate from problems of mechanical process is equal to the flow (yield) strength (s0) of the producing gross fracture, they are perhaps even
optimization. In short, if process parameters are emerging wire. Otherwise it would simply break more important than fracture itself. In particular,
improperly chosen, even the best material may at the die exit (or at the point where the drawn they may go unnoticed, leading to inferior phys-
fracture. On the other hand, metal with out- wire contacts the capstan). The limiting single- ical properties in as-shipped material or to sud-
standing intrinsic workability will in many cases pass reduction, or the reduction at which sd = s0 den failures under apparently safe subsequent
withstand the abuse of improper processing is a function of die design, friction, and the process or service conditions.
without fracture or defect development. In any work-hardening character of the material. Thus, Again, it is worth noting that such fracture
case, it rarely makes sense for a manufacturer to for a reasonably ductile metal one could argue and defect development can be produced in su-
cope with inferior procedures by requiring met- that work-hardening behavior is an important perior materials if processing is improperly un-
als of unusually good workability. The costs factor in workability. Actually, the effect of work dertaken. For the discussions in this chapter, one
and/or supply problems for such superior mate- hardening on single-pass workability is rather should assume for the most part that proper or
rials provide considerable incentive for process small, and most commercial drawing passes in- orthodox procedures are being followed and that
optimization. Thus, mature, competitive manu- volve drawing stresses well below the strength defects or fractures truly reflect practical metal
facturing generally involves the use of processes of the wire at the die exit (typically, sd 0.5s0). workability limitations.
flexible enough to accommodate metal of stan- The more practical wire-drawing considera- Surface cracking, whether circumferential
dard or average workability, which is readily tion, then, is multipass workability, or the ability (fir-tree cracking), or at roughly 45 to cir-
available from competent, major suppliers. of the metal to withstand fracture or defect de- cumferential (crows-feet cracking), or even
This is not to say, however, that the measure- velopment in the face of progressive straining in longitudinal (splitting), is a common problem
ment or estimation of workability is a minor multiple drawing passes without the benefit of in both extrusion and drawing. A schematic rep-
consideration. Quite the contrary, it is essential intermediate annealing. Generally, multipass resentation of surface cracking is given in Fig. 1,
to monitor day-to-day fluctuations in standard workability is tied to the resistance to ductile and illustrative micrographs are shown in Fig. 2
products. Not only will material of completely fracture and not directly to factors such as draw- to 5. There are several explanations commonly
unacceptable quality be occasionally received, stress limitation and work-hardening character. given for these flaws. In wire drawing, they are
but fluctuations within acceptable limits can Of course, in some materials multipass work- observed under conditions of boundary-lubrica-
often be dealt with to advantage, with the supe- ability may be so limited that fracture occurs at tion breakdown and sometimes, under thick-
rior material being directed to the more demand- strains or reductions below those commonly un- film conditions, are developed with dry soaps
ing process sequences. Beyond the case of the dertaken in a single drawing pass (up to about (Ref 3). In the former case it is likely that local
standard material, new alloy development and 40%). lubrication breakdown in the die causes simulta-
new process development require that the work- In the case of extrusion, the single-pass re- neous sticking and slipping of adjacent areas of
piece metal be carefully characterized for work- duction of common interest may be limited by the wire surface. This sticking and slipping of al-
Chapter 21: Workability and Process Design in Extrusion and Wire Drawing / 317
(Ref 8). A particularly common cause of surface

cracking in hot extrusion is surface heating. The
extrusion temperature will often be specified so
as to be below temperature ranges associated
with poor ductility or hot shortness.
Unfortunately, surface frictional heating, inten-
sified by excessive ram speed, may raise the sur-
face temperature into the hot-shortness range
anyhow, leading to surface fractures. Such frac-
tures may be associated with low-melting-range
Fig. 1 Three types of surface cracks that occur in ex- second phases, or strain concentrating second
truded and drawn products. Source: Ref 1, p 252 phases, or may be intergranular in nature.
Yet another surface flaw observed in extru-
sion is the extrusion defect, an intrusion of
surface oxide fostered by high container wall
friction. Although this results in a defect that is
somewhat comparable to surface cracking, no
fracture is involved, and the extrusion defect is
Fig. 3 Gross roughness and apparent cracking, result- not a true workability problem. On the other
ing from thick-film lubrication, on the surface of hand, surface fractures can provide paths that
EC aluminum rod drawn in calcium stearate. Scanning
electron micrograph. Magnification: 320. Source: Ref 1,
allow lubricant and oxide to intrude into the in-
p 254 terior of the extruded product, producing some
of the same problems presented by the extru-
sion defect.
Centerbursts in their grossest form are
chevronlike shear fractures that form in the cen-
ter of the extruded or drawn product, as shown in
Fig. 5. In more subtle form, they may be little
more than aggregates of pores along the center-
line of the product. They may develop almost
Fig. 2 Crows-feet crack configurations on the sur-
spontaneously or may slowly build up with pro-
face of a drawn EC aluminum rod. Scanning
electron micrograph. Magnification: 248. Source: Ref 1, gressive working (as in multiple drawing
p 253 passes). They may lead to outright fracture dur-
ing the working process or may remain subcriti-
ternate areas presumably leads to intense shear- cal, reducing finished product ductility or posing
ing and tearing of local regions, which present the threat of catastrophic failure during service.
the crows feet appearance shown in Fig. 2 The basic cause of centerbursting is nonuniform
(Ref 4). In the case of thick-film dry-soap lu- metal flow through the die and the associated
brication, it is suggested that the cracking comes development of tensile stresses in the center of
from unconstrained, nonuniform surface flow the workpiece. Such nonuniform flow and cen-
permitted by the thick lubricant film and pro- A split in tungsten wire, induced by compres-
terline tensile-stress development may be pro-
duced by irregularities in the soap-particle dis- Fig. 4 jected from the shape of the deformation zone. A
sion. Source: Ref 2
tribution and by differences in plastic flow from practical description of shape can be gotten from
one grain to another (orange peel). In any case, the D parameter, as described by Backofen (Ref
the cross-hatched cracking shown in Fig. 3 is de- 9). Basically, D is the ratio of the height of the
veloped. In drawing of refractory metals, such as deformation zone (perpendicular to the axis of
tungsten, deep longitudinal surface cracking or extension or die axis) to the length of the de-
splitting is often observed. A split induced formation zone (parallel to the axis of extension
by indentation is shown in Fig. 4 (Ref 2). or die axis). For round wire drawing and extru-
Apparently, very weak longitudinal interfaces sion:
are developed in the drawing of tungsten, per-
haps as a result of mechanical fibering (Ref 5)
a
[ ]
2
and/or grain-boundary embrittlement. Restricted D= 1+ 1- r
r (Eq 1)
slip may also lead to such delamination (Ref 6).
Moreover, it has been shown that this weakness
is greatly increased by certain anneals (Ref 2). where a is the die semicone angle in radians and
During drawing, these weak interfaces are con- r is the reduction per pass in decimal form.
ceivably opened through the action of tensile Values of D less than one are associated with
surface residual stresses (Ref 7). uniform metal flow. Centerbursting is associated
Circumferential cracking and 45 shear crack- with the nonuniform flow of higher D passes,
ing are seen in extrusion. The local sticking/slip- usually where D is about 2.5 or more. The burst-
ping mechanism described previously for wire ing may be slow to develop and may require
drawing is felt to cause intense shear zones and considerable overall plastic strain before becom-
resulting surface cracks. On a somewhat grosser ing obvious. Thus, light extrusion passes (r <
scale, transverse cracking is believed to be 0.20) often will not result in centerbursting in
caused by momentary sticking in the die land Fig. 5 Centerbursts in sectional, cold extruded steel spite of high D values (low r produces high D).
followed by pressure buildup and breaking away bars. Courtesy of Bethlehem Steel Corporation Similarly, it may take many passes of a high-D
geometry before a failure-resistant material such

as electrolytic tough pitch (ETP) copper displays
centerbursts. This is not to say, however, that the
material is undamaged by the early passes. Quite
the contrary, the early stages of ductile fracture
can be detected by an appropriate experimental
technique (Ref 10, 11).
Beyond the semiempirical D-based criterion,
several elaborate analytical approaches have
been undertaken in order to provide a more de-
tailed theoretical description of the centerburst
process (Ref 1215). These efforts involve three
steps:
1. Establishment of a quantitative, generalized
fracture criterion
2. Preparation of a plastic-flow analysis suitable
for determining the stress-strain experience
at each point in the workpiece
3. Use of the plastic-flow analysis to determine
which processing conditions (reductions, die
angles, friction conditions, etc.) produce
stresses and strains that satisfy the fracture cri-
terion and presumably lead to centerbursting Fig. 6 Workability criteria for centerbursting in aluminum alloy 2024 based on a maximum tensile stress-strain en-
ergy criterion.Source: Ref 13
A commonly used fracture criterion is that of
Cockcroft and Latham (Ref 16), wherein frac- and M42) and to carbon steel in the warm work- 23). The effect of friction in enhancing center-
ture is projected to occur when: ing range has been set forth by Wright et al. (Ref line hydrostatic tension and centerbursting is
18). well worth noting. This has been shown quite
ef
s *
0
s
s
de = C
(Eq 2)
A novel approach to predicting centerbursts
has been employed by Avitzur (Ref 15). Using
clearly by Coffin and Rogers (Ref 10). More-
over, Avitzur projects an increase in the occur-
an upper-bound analysis, Avitzur has projected rence of centerbursting with an increase in fric-
where e is effective strain, ef is effective strain at die angles and reductions for which the metal tion (Ref 15). Sticking friction can produce a
is effective stress, s* is the highest
fracture, s being worked will tend to flow apart at the cen- centerline hydrostatic stress state that is more
tensile stress, and C is a material constant. terline. As has been noted by Backofen (Ref 19), tensile than the frictionless case by as much as
Determining a proper value of C has been a this approach seeks to provide a necessary, but three-tenths of the flow stress.
major problem, in that upset testing and tensile not sufficient criterion for centerbursts. That is For short-term problem solving, the following
testing can provide widely different values. This to say, the upper-bound analysis projects condi- perspective seems best. First, it should be recog-
problem with variable results in using the tions whereby metal will tend to flow apart at the nized that there is probably a potential for cen-
Cockcroft-Latham criterion can be resolved centerline, assuming that no energy is required terbursting anytime the D value for extrusion or
using finite element analysis, but differences ob- to fracture the metal. A fully sufficient criterion drawing exceeds about 2.5. Moreover, metals of
tained by tension testing and compression test- would require incorporation of the fracture re- limited workability will develop centerbursts
ing should be recognized. In work on aluminum sistance of the metal. Moreover, Lee and after small strains and certainly in single passes.
alloy 2024-T351, for example, C has been esti- McMeeking (Ref 20) have pointed out that the On the other hand, metals of great workability
mated at 0.3 times the average effective stress in lower-bound theorem indicates that the work- may show little or no evidence of centerbursting
tension and 0.08 times the average effective piece will always tend to flow apart, regardless until very high strains have accumulated in sin-
stress in compression (Ref 13, 14). Latham and of die angle and reduction (the only thing that gle or multipass reductions with D 2.5 (again,
Cockcroft (Ref 12) and Hoffmanner (Ref 13) apparently keeps the metal from doing so is its the r for D is the reduction in each pass). Such
have employed slip-line fields for plastic-flow fracture resistance, which is, of course, neg- progressive centerbursts lead to failures in wire
analysis, whereas Chen, Oh, and Kobayashi lected in the upper-bound approach). They argue drawing called cuppy-core failures, with the
have employed the matrix method (Ref 14, 17). that the critical conditions for centerbursts in cup presumably forming over many passes. Of
The projections of Hoffmanner are typical of the upper-bound analysis arise from the assumption course, many commercial extrusion and drawing
results of these efforts and are shown in Fig. 6. of an unrealistic velocity field for this phenome- operations involve D values above 2.5. Thus, to
The ordinate is simply the value of the lefthand non. some extent, a potential for centerbursting is the
side of Eq 2 divided by the average effective In an approach of relative simplicity, Coffin rule and not the exception. In this regard it can-
stress. When the value of this integral reaches or and Rogers (Ref 10, 11) related centerline void not be overemphasized that even where no gross
exceeds the product of C times the average ef- development and centerbursting to the level of centerbursting has occurred, damage in the form
fective stress, centerbursting is projected. Note hydrostatic tension, based on a slip-line field of incipient ductile fracture may be present.
that for reductions large enough for D to be less analysis. It seems quite clear that such damage Such damage (centerline porosity, etc.) can man-
than about 2.5, no centerbursting is projected. occurs with process geometries that produce hy- ifest itself in reduced tensile properties in the as-
No centerbursting is projected for small strains drostatic tension (as opposed to compression) at extruded or drawn product. Other, more insidi-
either. the centerline. Normally, with low friction, such ous problems may arise. For example, caps cut
With appropriate adjustment of the value of conditions are achieved with D values in excess from bar stock will occasionally be sources of
C, these analyses are in reasonable agreement of 2.5. In addition to the Hill and Tupper slip- vacuum leaks, due to centerline porosity.
with experimental findings. However, much re- line field analysis (Ref 21) used by Coffin and Fractures. Occasionally a surface crack or
finement of these approaches remains to be un- Rogers, the interested reader is directed to the centerburst will form to such an extent and with
dertaken. An application of such bulk workabil- hydrostatic tension projections of Chen, Oh, and such spontaneity as to cause outright fracture.
ity testing to limited workability steels (T440C Kobayashi (Ref 14) and of Lee et al. (Ref 22, Normally, however, the compressive stress state in
the die in extrusion and drawing limits such crack is important to consider the relevant mechanisms sensitive to strain rate. Generally, higher strain
growth, and truly spontaneous fracture is avoided. of fracture from a viewpoint basic enough to rates are consistent with increased stress re-
The important exception is fracture of the allow design of sound workability test proce- quirements and a greater likelihood of fracture.
wire in drawing due to excessive drawing load. dures or selection of pertinent technical data (An important exception is carbon steel in the
The drawing stress is composed of the work re- from published sources. 200400 C, or 390750 F, range, wherein dy-
quired for uniform deformation, redundant de- First of all, the driving forces for fracture, namic aging effects reverse the stress-strain rate
formation, and friction. Using Wistreichs re- stress, and strain have nothing directly to do relationship.) In extrusion and drawing, on the
dundant-work data (Ref 24), Wright has with the inherent resistance of the metal to frac- other hand, these effects are complicated by fric-
formulated the following relationship for practi- ture (or its workability). There will merely be tional heating. Moreover, the rather high strain
cal drawing-stress calculations (Ref 25): some limit of stress and/or strain beyond which rates involved in commercial drawing and extru-
cracking will occur. The more workable the sion are not easy to develop in simple, well-de-
sd 3.2 metal, the higher the limit. For example, as men- fined test formats (short of setting up a pilot
+ 0.9 (a + m )
s 0 D (Eq 3) tioned previously, Cockcroft and Latham (Ref model of the process itself). At a minimum,
16) have proposed that fracture occurs, at a however, the investigator should be aware that
where s 0 is average flow stress and m is coeffi- given temperature and strain rate, when the inte- flow stress and fracture behavior are fundamen-
cient of friction. Rarely is a commercial drawing gral in Eq 2 reaches a critical value of C. More tally strain-rate sensitive.
pass designed so that the extent of uniform and workable metals simply have higher values of C. Lastly, and of most practical interest, second-
redundant deformation involves excessive draw- Thus, in the discussion of general cracking phase content, inclusions and internal cavities,
ing stress. Rather, the occurrence of drawing observations, various causative factors involving and external damage are of great importance to
breaks is usually attributable to excessive fric- stress and strain have been alleged (frictional ef- workability. In the extreme case, they reduce
tion, to inertial loadings, and, most often, to fects, centerline tension, excessive drawing workpiece cross section or cause extraordinary
flaws in the wire, which reduce load-bearing ca- stress, etc.). Therefore, it is helpful to have some strain concentrations in their vicinities, to the ex-
pacity. These flaws may be gross inclusions, cut idea of the levels and natures of these stresses tent that fracture follows almost immediately.
or abraded surfaces, rolled-in dirt, or enlarged and strains in the process and to have compara- More commonly, they serve as sites (fracture
surface cracks or centerbursts developed in the ble states of stress and strain in any test that is to centers) for the development of pores or cavi-
manner discussed previously. Again, the well- be used as a basis for forecasting workability. ties in the early stages of ductile fracture. Thus,
known case of cuppy-core fracture in wire In most instances, fractures in extrusion and the larger or more numerous these defects are,
drawing is little more than a drawing break that drawing are ductile in nature. The mechanisms the sooner or more rapidly ductile fracture will
occurs at a point weakened through the growth of ductile fracture typically involve the growth develop. Even very fine scale precipitation can
of a centerburst. of pores with strain under a general tensile stress produce a marked acceleration in the develop-
In the case of hot drawing, fracture of the wire state (hydrostatic tension). Eventually the pores ment of ductile fracture. An interesting case of
often occurs at a point beyond the die where the become gross enough so that tearing occurs in this can be seen in the work of Coffin and
wire has cooled into a brittle range. This takes the intervening material, and a fracture results. Rogers, where centerline porosity (the initial
place most often on contact with the capstan. The fracture path may follow bands of intense stages of centerbursting) is shown to be more
The capstan, unless heated, chills the wire and shear strain or may be largely perpendicular to pronounced in drawn ETP copper than in drawn
simultaneously imposes tension as the wire as- the direction of the largest tensile stress. oxygen-free high-conductivity (OFHC) copper
sumes the capstan radius. The fundamental question is which factors in- (Ref 10, 11). Interesting observations on the
herent to the metal affect its ability to withstand evolution of centerburst fractures in extrusion
given stresses and strains without cracking. (A have been set forth by Pepe (Ref 26).
Shapes and Tubes comprehensive discussion of this question is be- Thus, for a given stress state, temperature, and
yond the scope of this chapter.) Such factors in- strain rate, the workability variations demon-
Most of the remarks in the previous section clude aspects of basic crystal structure; crystal- strated by an array of materials will strongly re-
address flaw development as it occurs in round defect structure; grain size, shape, and orientation; flect any variations in inclusions, second phases,
stock. In the case of tubular products, the re- inclusion and second-phase content; defects in- internal cavities, and external damage. This fact
marks regarding surface cracking and fracture herited from primary processing (porosity, ex- has led to extensive use of approaches such as
are still largely applicable. Although there may ogenous inclusions); and handling damage (cuts, metallographic characterization, nondestructive
be nothing quite like centerbursting in tubular scrapes, rolled-in particles, etc.). testing, and dye checking to predict workability.
products, cracks may in some cases develop on Crystal structure, substructure, and mi- These techniques can provide much useful in-
internal surfaces for some of the same reasons crostructure are factors related to the ability of a formation. However, they are no substitute for
that govern external surface behavior. The con- metal to deform readily without the need for mechanical testing.
sideration of shapes is quite complicated. For large stresses. Hot working, of course, allows
rather regular shapes (hexagonal cross sections, deformation to be undertaken at relatively low
ovals, etc.), the remarks concerning round stock stress levels. On the other hand, in some hot Projecting Workability from
are directly applicable. For shapes with thin sec- working ranges the development of fracture is Mechanical Tests
tions (I-beams, L-shapes, etc.), the surface-flaw nonductile, due to the melting (hot shortness) or
problems persist, but centerbursts are less likely considerable softening of certain second phases, The ultimate workability test is the process it-
to occur. In effect, D values are low in the thin or to excessive grain-boundary deformation. In self. Indeed, in some instances manufacturers may
sections. On the other hand, problems involving any case, it is important to recognize that the wish to set up a pilot line for the purpose of qual-
surface friction are apt to be more severe in ability of a metal to deform extensively without ifying and evaluating new materials. Beyond this,
tubes and thin section shapes. fracturing is very sensitive to temperature. Thus, careful observation of manufacturing operations
processing should reflect careful attention to can yield much information on workability, even
temperature optimization and control, and work- when the metal being processed is of acceptable
Some Comments on Fracture ability evaluations must be undertaken in the quality. Nonetheless, it is usually desirable to have
Mechanisms temperature ranges to be encountered during abstract mechanical test procedures at hand for the
processing. purpose of projecting workability.
Now that the general observations on cracking Somewhat similarly, the stresses and fracture Unfortunately, wire drawing and extrusion
in extrusion and drawing have been set forth, it tendencies developed by a deforming metal are have not had the benefit of the extensive devel-
opmental work that has gone into the projection ef

s * ef fracture in the compression of cylindrical speci-
of workability limits in, say, sheet metal defor- C= 0
s
s
de =
0
sde = s 0 e f
(Eq 4)
mens:
mation. Thus, it is not often possible to project
absolute levels of workability in wire drawing 1
and extrusion from simple mechanical test for- where s0 is again the average flow stress and ef e1 + e2 = a
is the true strain at fracture or ln (Ai/Af), where 2 (Eq 6)
mats (an important exception is that of the max-
imum possible single-pass wire drawing reduc- Ai and Af are the initial and final cross-sectional
areas, respectively. The use of the Bridgman cor- where e1 is the circumferential strain, e2 is the
tion, where rather good quantitative estimates axial strain at the equatorial free surface, and a
are readily made). Rather, this discussion fo- rection for stress in the necked region (Ref 28) is
recommended for accurate determination of the is a material constant. Thus the value of a may
cuses on testing that can reveal relative degrees
of workability from one sampling of feedstock average flow s0 stress. Even if the Cockcroft and be used to index workability at least as far as
Latham analysis cannot be applied to the predic- free-surface fracture is concerned. For example,
to the next. Of course, pertinent results from a values of 0.32 and 0.18 have been measured,
these tests can always be correlated with ob- tion of absolute workability, the value of C, or
even the area reduction itself, is an excellent respectively, for type 1020 steel and type 303
served processing limitations to allow empirical stainless steel (Ref 31). Unfortunately, the sur-
projections of absolute workability. index of relative workability in wire drawing.
In the case of very ductile metals, the area at faces undergoing fracture in extrusion and draw-
The tensile test is probably the most impor- ing generally are not free surfaces. Even so, ex-
tant of all mechanical tests, and it can be useful fracture may be very small and difficult to meas-
ure. In this regard, the scanning electron micro- ploration of the use of Eq 6 in assessing surface
in projecting drawing and extrusion behavior. workability in extrusion and drawing may be
The concept of the tensile test, as used here, scope (SEM) can be very useful. (As a word of
caution, the magnifications set forth on the SEM worthwhile.
should be extended to include aspects of vari- In summary, however, the compression test is
ous stretch tests as commonly employed in may not be as precise as desired for such meas-
urement. This problem can be surmounted to less suited to assessing workability in extrusion
the wire industry and short-time hot tensile and drawing than might be imagined.
tests such as can be run on a hot tensile some extent by placing objects of known dimen-
sions in the field of view.) The torsion, or twist, test is widely used to
(Gleeble) testing apparatus (Ref 27) (see chap- assess workability in cylindrical stock (see
ter 7). The first consideration is that the test Most of the other tensile test data are of little
use in workability assessment. In particular, the Chapter 8, Torsion Testing to Assess Bulk
temperature be relevant. Such a requirement of Workability). Even where the load is not meas-
course demands knowledge of the process it- elongation (uniform or total) has limited mean-
ing because it is largely dictated by the onset of ured, the number of turns to failure is felt to be a
self, and, particularly at the workpiece surface, useful measure of surface-region fracture resist-
processing temperature can be difficult to as- necking. Necking is a plastic instability that
does not occur in normal wire drawing and ex- ance. It should always be emphasized, however,
sess. Beyond this, the cumulative heating ef- that the strain is highest at the surface in this test.
fects of thermomechanical working must be trusion, and so the elongation has little bearing
on relative workability. It is unfortunate that That is:
considered, particularly in wire drawing, where
high-speed cold processing can generate wire such wide use is made of elongation or stretch-
ing data in an effort to forecast workability. RDf
surfaces that are actually at several hundred de- g=
grees Celsius. In any case, the test temperature Compression testing (see Chapters 5, L (Eq 7)
should be carefully considered. The selection of Cold Upset Testing and 6, Hot Compression
a relevant strain rate in a tensile test (or any Testing) can also be useful in projecting work- where g is the torsional shear strain, R is the ra-
simple mechanical test) can be very difficult in ability in drawing and extrusion. There are some dial position, L is the gage length, and Df is the
view of the wide range of strain rates, and the major limitations, however. First, the tensile hy- angle, in radians, through which one end of the
very high strain rates, encountered in the drostatic stress state germane to centerbursts is gage length has been twisted relative to the other
processes themselves. In general, it is recom- not developed. Second, the flow stress may be (one revolution = 2p radians). Moreover, the sur-
mended that the tests be run at the highest prac- obscured by uncertain friction-hill effects at face is a free surface. Thus, the fracture resist-
tical strain rate. the sample ends. Third, the fractures are gener- ance manifested in this test is only that of the
The tensile-test results that are germane to ally at the surface, and the workability of mate- surface material, and of the unconstrained sur-
workability in extrusion and drawing are the rial in the interior of the workpiece may not be face, at that. Thus, as in the compression test, the
flow-stress level and the area reduction at frac- reflected. On the other hand, knowledge of the torsion test has little relevance to centerburst
ture. Flow-stress data are of course directly use- resistance to surface cracking is important. problems and may not be an accurate indicator
ful in estimating limitations due to extrusion- With proper analysis of the stress and strain at of surface fracture behavior under constraint of
pressure capacity. The relevance to wire the specimen surface, the compression test may the extrusion or drawing die.
drawing is less clear-cut since the draw stress be used to determine a value of C for the Even so, the torsion test can be recommended
and capacity for draw stress are both based on Cockcroft and Latham analysis. Work along this for extrusion and drawing workability testing on
flow stress. Beyond this, the flow-stress level line has been undertaken by Hoffmanner (Ref a qualified basis. It has the advantage of allow-
has many uses in assessing frictional effects, die 13), who observed strain to fracture on barreled ing very large strains to be generated. Thus,
and tooling pressures, thermomechanical ef- cylindrical surfaces during upsetting. The strains growth of surface flaws can be rather carefully
fects, and other factors somewhat outside the could be monitored through the deformation of a evaluated in very ductile materials. In general,
scope of workability. However, the really im- grid printed on the specimen surface. Stress state fracture (or failure via observable surface
portant piece of data is the area reduction at could be related to strain through the relation: cracks, etc.) will occur when the maximum tor-
fracture. This is an expression of the intrinsic sional shear strain, gmax, reaches a critical level.
de 2 (2s 2 - s 3 ) The value of gmax is a measure of workability
ductility at fracture, in the presence of hydro- =
static tension, and should be very much related de 3 (2s 3 - s 2 ) (Eq 5) and may be calculated from:
to the capacity to resist centerbursts and ductile
process fractures in general. where e2 and s2 are the circumferential strain Rmax Nmax
g max =
From a somewhat more sophisticated basis, and stress, respectively, and where e3 and s3 are L (Eq 8)
the flow stress and the area reduction at fracture the strain and stress, respectively, in the axial di-
can be used to estimate the value of C in the rection. where Rmax is the actual radius of the cylindrical
analysis of Cockcroft and Latham (Ref 16). For The work of several authors (Ref 2931) has stock, and where Nmax equals 2pn, with n being
the case of a simple tensile test: resulted in the following criterion for surface the number of turns to fracture or failure. Of
course, for any fracture strain or level of worka- then be used as an index of centerburst re-
bility, a maximum number of turns exists and n sistance; the greater the total reduction, the
becomes an index of workability. higher the resistance. A series of 15% re-
Bend testing is also widely used to assess ductions at an a value of 12 (D = 5.2) is
workability of rod and wire. Basically, the rod or recommended.
wire is bent or wrapped around a mandrel. In As an alternative to waiting for gross cen-
bending, the strain is maximum along a surface terbursting to occur, one can assess ductile
locus, with fracture (or failure via observable damage or incipient ductile fracture at the
surface cracks, etc.) occurring when the maxi- centerline. This can, of course, be done met-
mum tensile strain (at the outer bend radius) allographically or by careful density meas-
reaches a critical level. This maximum tensile urements (Ref 10, 11). Pores on the order of
strain (emax) can be calculated from: 1 mm in diameter are fairly easy to resolve
with conventional metallographic tech-
1 niques, and Coffin and Rogers detected duc-
e max = tile damage-related density changes of a
1 + D
few parts in ten thousand.
d (Eq 9) Some indication of resistance to surface
cracking can be obtained by drawing the
where d is the rod diameter and D is the mandrel stock under conditions of little or no lubri- Fig. 7 Generic extrusion limit diagram. Source: Ref 32
diameter. For any given fracture strain (ef) a cation or under conditions of thick-film,
minimum value of D/d (or D) exists, such that: dry-soap lubrication. Scanning electron mi- is reduced. Thus, Vmax would be increased.
1
croscopy will reveal the extent of surface However, many extrusion operations are under-
( D / d ) min = - 1 cracking, and comparative evaluations can taken without lubrication, and the surface trac-
ef (Eq 10) be made at a given cumulative reduction tion is fixed at a large fraction of the extrudate
(Ref 3). Three or four passes of 20% each shear strength (at the prevailing temperature
If the rod or wire is itself used as a mandrel, the may be necessary to establish a consistent and strain rate). Except in the cases of certain
test is often called a wrap test. For a wrap test pattern of surface damage. With the incon- light sizing or shaping passes, or certain com-
D/d is 1.0, and emax is 0.5. Thus, if a rod or wire sistent boundary lubrication conditions that plex shapes, extrusion reductions are generally
passes a wrap test, ef exceeds 0.5. will develop with little lubricant, stick-slip so large that centerline stresses remain gener-
As in the case of torsion, bend testing is most behavior should promote surface cracking. ally compressive, and centerbursting is not a
relevant to workability limitations associated On the other hand, the rough, somewhat un- problem.
with surface cracking. It has no relevance to constrained surface that develops with dry Drawing-Process Design Implications.
workability limitations related to structure pecu- soaps will also provide local strain concen- From a single-pass perspective, there is a limit-
liar to the centerline. trations predisposed to cracking. The results ing reduction that may be taken, associated ap-
More Elaborate Test Formats. A major of such testing must be viewed with caution, proximately with the condition where sd /s0 is
problem with the previously described tests is however, owing to problems in maintaining unity. (The precise limit would be for the case
the absence of a die and the associated metal consistent lubricant performance. where the drawing stress equaled the flow stress
flow constraints. It is desirable to measure work- at the die exit, which may differ from the aver-
ability with a test that comes closer to process age flow stress.) This limiting reduction can be
mechanisms. The most practical approach to this Process Design maximized if the drawing pass is designed to
is to use a laboratory-scale drawing system. minimize the draw stress (see Chapter 19,
Even a tensile-testing machine, outfitted with a Extrusion-Process Design Implications. Drawing of Wire, Rod, and Tube). However,
die holder, can be used to draw short lengths of In many instances, extrusion-process design only rarely is this single-pass reduction limit ap-
stock. Of course, deformation at hot working maximizes productivity and thereby maximizes proached or desired in practice. Moreover, it is
temperatures and at high strain rates will require exit speed. Limitations to extrusion speed in- not a function of the intrinsic workability of the
more elaborate equipment. volve press capacity and surface temperature, as drawing stock.
In any event, some practical approaches for shown in the workability or extrusion-limit dia- Many multipass-drawing-process design con-
assessing workability on this basis are: gram of Fig. 7. This is a widely used format for cepts involve reductions consistent with stan-
extrusion-limit diagrams (Ref 33 and 34). The dard gages, and the remaining design issues can
Draw a series of specimens once each maximum extrusion speed, Vmax, is to be found be explained in terms of the D parameter. There
through respective dies with different die at the intersection of two curves. The scored area are a wide variety of drawing technologies that
angles, using the same reductions for each under the two curves is the practical process benefit from low D process designs. With lower
specimen. The value of D increases with in- regime, with maximum productivity associated, values of D (~1.0 to 2.0), one reduces redundant
creasing die angle, of course, and the die in principle, with the highest speed, at the top of work and general strain accumulation, one re-
angle at which centerbursting is first noticed the scored area. duces centerline tension and the propensity for
can be used as an index of centerburst re- The left-side curve in Fig. 7 expresses extru- centerbursts, and one lowers die pressure. The
sistance for the material; the higher the sion speed, for a given available press load, minimization of centerline tension and general
value of a, the better the resistance. A 10% with the speed increasing as flow stress dimin- strain accumulation should allow maximum
reduction with a values of 4 to 16 will pro- ishes with increasing temperature. The right- multipass reduction to be achieved without gen-
vide a range of D values from 2.7 to 10.6. side curve expresses allowable extrusion speed, eral ductile fracture or centerline fracture.
For materials with too much workability for for a given surface liquation temperature, with Intermediate D (~2.0 to 3.5) drawing-process de-
failure in a single pass, set up a sequence of speed increasing with lower overall exit tem- signs have been common, historically, and may
dies allowing successive reductions to be perature, since there can be more surface heat- offer the advantages of reduced fine develop-
taken at relatively high values of D (D 5). ing (from friction and lack of time for dissipa- ment (a fine is a small particle of the wire sur-
The reductions and die angles can be the tion) without surface melting. When lubrication face that becomes liberated during drawing)
same in each pass. The total reduction at is used with extrusion, the right-side curve in (Ref 35). Higher D drawing-process designs
which centerbursting is first observed can Fig. 7 should be shifted to the right, as friction may be employed when high friction is unavoid-
able or when a high die pressure is desired to 5. W.A. Backofen, Deformation Processing, 21. R. Hill and S.J. Tupper, J. Iron Steel Inst.,
iron the wire surface and brighten the wire. Addison-Wesley, 1972, p 303 Vol 159, Part 4, 1948, p 353
When drawing-temperature control is a prior- 6. D. Lee, Metall. Trans., Vol 6A, 1975, p 22. E.H. Lee, R.L. Mallett, and R.M.
ity, it may be useful to use tapered drawing 2083 McMeeking, Numerical Modeling for
schedules that reduce the drawing reduction as 7. R.N. Wright, Wire Technol., Vol 6 (No. 3), Manufacturing Processes, Jones, Armen,
the wire work hardens, thus achieving a rela- 1978, p 131 and Fong, Ed., ASME, 1977, p 19
tively constant draw stress, and relatively con- 8. R.J. Wilcox and P.W. Whitton, J. Inst. Met., 23. E.H. Lee, R.L. Mallett, and W.H. Yang,
stant degree of thermomechanical heating, from Vol 88, 19591960, p 145 Comput. Meth. Appl. Mech. Eng., Vol 10,
pass to pass. This concept is largely independent 9. W.A. Backofen, Deformation Processing, 1977, p 339
of the workability of the wire. Addison-Wesley, 1972, p 88 24. J.G. Wistreich, Proc. Inst. Mech. Eng., Vol
The achievement of maximum drawing pro- 10. L.F. Coffin, Jr. and H.C. Rogers, Trans. 169, 1955, p 654
ductivity through maximum drawing speed may ASM, Vol 60, 1967, p 672 25. R.N. Wright, Wire Technol., Vol 4 (No. 5),
be highly problematical where lubrication is 11. H.C. Rogers and L.F. Coffin, Jr., Int. J. 1976, p 57
marginal, owing to very high surface tempera- Mech. Sci., Vol 13, 1971, p 141 26. J.J. Pepe, Met. Eng. Q., Vol 16 (No. 1),
tures and related surface failure. The increased 12. D.J. Latham and M.G. Cockcroft, The 1976, p 46
speed not only limits the dissipation of frictional Effect of Stress System on the Workability 27. W.F. Savage, High Speed Testing, Vol III,
heating, but it limits the time available for cool- of Metals, Report No.216, National Interscience Publishers, 1962, p 55
ing between drawing passes. However, where Engineering Laboratory, Feb 1966 28. P.W. Bridgman, Trans. ASM, Vol 32, 1944,
good lubrication is readily achieved, increased 13. A.L. Hoffmanner, Technical Report AFML- p 553
speed may abet lubrication. TR-69-174, Air Force Materials Laboratory, 29. H. Kudo and K. Aoi, J. Jpn. Soc. Technol.
June 1969 Plast., Vol 8, 1967, p 17
14. S.-I. Oh, C.C. Chen, and S. Kobayashi, 30. S. Kobayashi, Trans. ASME, J. Eng. Ind.,
Trans. ASME, J. Eng. Ind., Vol 101, Series Vol 92, Series B, 1970, p 391
B, 1979, p 36 31. P.W. Lee and H.A. Kuhn, Metall. Trans.,
REFERENCES 15. B. Avitzur, Trans. ASME, J. Eng. Ind., Vol Vol 4, 1973, p 969
90, Series B, 1968, p 79 32. K. Laue and H. Stenger, Extrusion:
1. G.E. Dieter, Ed., Workability Testing 16. M.G. Cockcroft and D.J. Latham, J. Inst. Processes, Machinery, Tooling, American
Techniques, American Society for Metals, Met., Vol 96, 1968, p 33 Society for Metals, 1981
1984 17. C.H. Lee and S. Kobayashi, Trans. ASME, 33. W.Z. Misiolek, Engineering Concepts in
2. A.W. Funkenbusch, F. Bacon, and D. Lee, J. Eng. Ind., Vol 95, Series B, 1973, p 865 Extrusion, Aluminum Processing Program
Technical Information Series Report No. 18. R. N. Wright, T. A. Kircher, and J. R. Workshop, Rensselaer Polytechnic Institute,
78CRDO17, General Electric Corporate Vervlied, J. Met., Vol 39 (No. 10), 1987, p 18 April 1994
Research and Development, Feb 1978 26 34. H. Stenger, Drahtwelt, Vol. 59, 1973, p. 235
3. R.N. Wright and A.T. Male, Trans. ASME, 19. W.A. Backofen, Deformation Processing, and 371
Lubr. Eng., Vol 97, Series F (No. 4), 1975, p Addison-Wesley, 1972, p 150 35. G. Baker, Workpiece Wear Mechanisms in
134 20. E.H. Lee and R.M. McMeeking, Trans. the Drawing of Copper Wire, Ph.D. disser-
4. R.N. Wright, Wire J. Int., Vol 35 (No. 8), ASME, J. Eng. Ind., Vol 100, Series B, 1978, tation, Rensselaer Polytechnic Institute,
2002, p 86 p 386 1994
Chapter 22
Multidisciplinary Process Design and

Optimization: An Overview
H.L. Gegel, Air Force Research Laboratory, Wright-Patterson Air Force Base (Retired)
R. Grandhi and J.S. Gunasekera, Wright State University
Charles Gure, Consultant
ENGINEERING DESIGN is the reverse of Evolutionary computation, expert systems, A multistage design process is described to
engineering analysis. It is the process of product and numerical optimization to achieve more show the benefits of the MPDO concept in up-
realization that involves the synthesis of knowl- effective solutions than solutions created by stream process design of a forging. However,
edge and techniques from engineering, science, human engineers. (Evolutionary computa- MPDO can be utilized to create optimal designs
aesthetics, economics, and psychology in estab- tion is a term for biologically inspired artifi- for all metalworking processes as forging, extru-
lishing specifications for products and their as- cial-intelligence technology such as genetic sion, rolling, shape rolling, and cog milling. The
sociated production processes. Multidisciplinary algorithms and neural networks. The idea of design techniques are mostly the same for vari-
process design and optimization (MPDO) is an a genetic algorithm is based on the evolu- ous bulk forming processes, only the design
engineering environment that supports process tionary concept of mutation that breeds rules, material databases, and the equipment
design and optimization. Its purpose is to auto- solutions, which may be better than those characteristics will vary.
mate, optimize, and accelerate the design devised by human engineers. Neural- Additional examples of process design and op-
process. The MPDO design environment is the network methods refer to simulation models timization are also provided in the chapters that
process by which engineering descriptions and based on massively parallel processing sim- follow. Optimal Design of Thermomechanical
specifications are integrated with material ilar in concept to the structure of neurons in Processes, Chapter 23, demonstrates the appli-
process analysis techniques to ensure that a a brain. These programs learn by looking at cation of dynamical modeling and process design
product possesses the desired behavior, perform- examples as images or collections of pro- techniques using simplified analysis models and
ance, quality, and cost, not to mention satisfying cessing data. They presumably may spot re- trajectory optimization for solving a class of
the requirement of delivering the product when lationships that may escape human notice.) problems where shape is not a controlling factor.
the customer needs it. Computer-Aided Optimization for Improved
The MPDO concept represents an integrated These developments are part of a collection of Process Engineering Productivity of Complex
infrastructure of software, standards, and data- advanced engineering and business design tech- Forgings, Chapter 25, emphasizes the use of a
bases that allows companies to solve their tough- niques that provide design excellence and high- multiobjective system for the integrated design
est design problems by making more effective quality products for a given product line or com- of forging process conditions, preform shapes,
use of their design software and human re- pany. Additionally, these tools are used when a and stages. Nonlinear viscoplastic finite element
sources. The MPDO environment is a result of need exists to shorten lead times, get products to methods and mathematical optimization tools are
a series of technical developments that have market faster, enhance value-stream analysis, utilized for simulating metal forming and shape
occurred over a period of four decades of mate- and eliminate scrap or activities that do not add optimization for preform design. Process
rials and manufacturing research. Some of the value for the customer. They are capable of re- Design of Gas Turbine Engine Components
technical and business developments include the ducing design time and giving engineers the Using iSIGHT for Process Optimization,
following: means to focus on design decisions that maintain Chapter 24, demonstrates the benefits of using a
upstream and downstream quality and value for commercially available computer-aided opti-
Analytic and numerical mesh-based and the customer. mization (CAO) application (iSIGHT) to obtain
material models for understanding and eval- This chapter briefly introduces MPDO as a an optimal forging product and process design.
uating materials processing type of engineering productivity software. The
Computer-aided design/computer-aided commercial needs (Ref 1) for MPDO are cate-
manufacturing (CAD/CAM) and computer- gorized as: Why Use MPDO?
aided engineering (CAE) tools for creating
part and die geometries and automatically Modeling to reduce iteration of trials (speed Competitive forging suppliers are streamlin-
meshing the geometries for analysis and cost) ing their manufacturing processes to compete
Workflow process modeling, which in- Modeling to optimize an existing process to more effectively in the global market. By shift-
volves defining and implementing a collec- reduce cost ing the focus from manufacturing hours to man-
tion of activities (structured or ad hoc) that Modeling to improve a customer-specified ufacturing parts, a new performance indicator
must be performed with different types of quality (reduced process and product vari- for how well a manufacturing job has been done
human activities and processes performed ability) has evolved (Ref 2). Competitive companies
by tools Modeling to reduce product delivery time. now measure delivery performance based on the
326 / Multidisciplinary Process Design and Optimization
date that the customer wants the order, not on the ful impact of waste reduction on productivity. forged products, when compared with the tradi-
date someone at the company feels comfortable Cutting the amount of work-in-process (WIP) tional manufacturing hour business strategy, is
about pledging. for a given level of output is much greater than characterized by less human effort, less rework
Improving the practice of engineering design can be explained by reductions in working capi- and scrap, less time, less space, and smaller eco-
and manufacturing is essential for achieving in- tal. A reduction in WIP by one-tenth produced a nomic lot size. Process-activity diagrams de-
dustrial excellence (Ref 3 to 5). In the 1990s, 9% rise in TFP. Empirical evidence from study- scribe the workflow in terms of specific activi-
best practices in the industry, which focused the ing both Japanese and U. S. companies supports ties in engineering offices, purchasing offices,
vision on what could be done to convert losses the observed WIP relationship and TFP. Cutting sales departments, and activities in the forge
into gains for the industry, were identified. WIP generally leads to faster, more reliable de- shop. Causes for bottlenecks in a targeted work
Studies were done on different types of compa- livery times and lower rejection rates. Reducing area were identified to reduce the time parts are
nies to shed some light on the variables that in- WIP clearly cuts overhead costs, and WIP is processed out of sequence and to maximize
fluence productivity growth at the local level invariably a symptom of poor product and mate- value flow within the work area.
(Ref 6). The types of companies were: rial process design, lack of process understand-
ing, erratic process yields, unreliable equipment,
Process companies, where manufacturing is unfocused business processes incapable of Process Design Methods
characterized as being highly connected and producing outputs that satisfy customer require-
automated ments, poor responsiveness to changing cus- The traditional build-and-test methods of de-
Batch processing, where manufacturing is tomer expectations, and insufficient critical veloping a complete manufacturing process
characterized as a disconnected line-flow measures of performancecost, quality, speed, were heavily experienced based until the end of
organization of work and service. the 1980s (Ref 13). Major barriers existed to in-
Several different batch processes, where Dramatic increases in efficiency are thus pos- creasing the use of models in industry with how
manufacturing is characterized by a very sible by solving the problems that create them in to implement existing modeling technology.
rapid change in both product and processes the first place. Creating order and promoting un- One of the most important aspects was deciding
derstanding of business processes make for a whether to use analytical modeling or mesh-
In none of these companies did the traditional powerful combinationand a powerful lever for based models. Since then, many software appli-
profit and loss statements provide adequate up- competitiveness. cations have been developed for analyzing the
to-date information about performance. Lean Manufacturing. In the mid-1990s, process. By the end of the 1980s, mesh-based
A new measuring system called total factor high-technology suppliers began moving toward modeling became the dominant method for most
productivity (TFP) was devised to measure the lean manufacturing by exercising some com- industrial manufacturing processes, and power-
overall efficiency of a manufacturing facility mon-sense principles such as (Ref 7): ful mesh-generating and remeshing software be-
(Ref 6). It identified the following practices, came commercially available. The finite element
which, when managed correctly, really make a Precisely specifying value for each job ac-
method (FEM) was selected for the following
difference: tivity and product
reasons: (a) engineering applications tend to
Identifying the value stream for each
Investing in new equipment involve complex shapes; (b) the process physics
product
of the bulk metalworking processes is highly
Reducing waste Making value flow without interruptions
nonlinear, including geometry deformation and
Cutting work-in-process Letting the customer pull value from them
material and boundary condition behavior; and
Always pursuing perfection
The study showed unequivocally that capital in- (c) the stability of the process is governed by
vestment in new equipment and design tools are Lean manufacturing always starts with value as coupled phase transformations, plastic flow, and
essential for sustaining growth in TFP over a defined by the customer (Ref 810). A lean- heat flow. Additionally, by relaxing some condi-
long time; that is, a decade or more. A negative manufacturing business strategy includes close tions on the FEM, a one-step simplified finite el-
correlation between waste rates; that is, scrap contact with the client. To please the client, an ement method (SFEM) that has computational
and rework, was expected, but the magnitude of all out effort is made to find out all the necessary speeds corresponding to analytic models can be
the effect of waste on TFP was not anticipated. information needed to satisfy the clients ex- used to speed up the design process. Most in-
A one-tenth reduction in waste increased the pectations and to reach an optimal solution. dustrial design for manufacturing is done by
TFP by 3%. The strength of this relationship Competitiveness is achieved by applying both teams of specialists, not by a single individual.
was amplified when a decision to boost the pro- workflow and process modeling to the part man- Manufacturing processes inherently involve
duction throughput rate was made. This deci- ufacturing process. the processing of materials, and understanding
sion was made to theoretically raise TFP be- To compete effectively, some forging compa- these processes at some level of detail is neces-
cause of large-fixed components in labor and nies also have adopted a step-by-step strategy to sary if predictive models are to be used to opti-
capital costs. Unfortunately, the waste ratio in- reach a highly efficient mode of lean manufac- mize the processes. The MPDO techniques
creases, and this increase indicates the power- turing (Ref 11, 12). The lean manufacturing of require a menu of analysis tools, some complex
and others simple, which are capable of predict-
ing the time evolution of internal variables and
input-output relationships for the system. The
Table 1 Methods of deformation analysis and typical design capability ratings design focus places emphasis on solving pro-
Die Die pressure Shape Sculptured Strain Error, CPU cessing problems and using appropriate model-
Method(a) load distribution complexity surface rate Temperature % time Rating ing techniques to optimize the material, process,
SLAB G G F 25 15 s M and equipment variables within delivery time
UB(b) M G M (c) 15 1 min M and quality constraints set by the customer. The
UBET M M F (c) 10 35 min F type of model is related to design objectives
SLF M P P M 10 57 min M
SFEM M M G VG VG G 10 15 s G such as shape complexity, temperature, and al-
FEM G G G VG VG G 5 14 h G lowable error. A summary of various design ob-
jectives is given in Table 1 for several analytical
See text for discussion. (a) See also Chapter 1 for references and an introduction to modeling methods. (b) More suitable for metalworking operations and numerical methods used in the modeling of
that are steady state such as extrusion, drawing, and rolling. (c) A temperature module must be added separately. VG, very good; G, good; F, fair; M, deformation processes. These analytical and
moderate; P, poor.
numerical methods are described in Chapter
Chapter 22: Multidisciplinary Process Design and Optimization: An Overview / 327
1, Workability and Process DesignAn using an approximate method. order effects may be required. Only the FEM has
Introduction, while the following provides a Analysis by UBET provides a value that is the flexibility to handle this class of industrial
brief description of design capabilities of these equal to the actual force or higher, and the pre- metalworking problem.
methods. dicted total power consumption inside the die is In practical metalworking processes, a num-
Table 1 summarizes the results from the de- higher than the power actually consumed. In ber of preforming operations are required to
velopment of different analysis methods that are general, the approach to using this analysis transform a simple initial geometry into a com-
important in the context of industrial process method begins by assuming a pattern of defor- plex shape while achieving desired tolerances,
modeling to simulate metalworking processes. mation, which is usually based on the modelers residual stresses, microstructures, and proper-
The Sachs (Slab) method (SM), upper bound conception of the actual flow. The velocity com- ties. In a multistage forging process, the various
method (UBM), and the slip-line field (SLF) ponents satisfy the volume continuity of the ma- workpiece shapes are achieved by using dies of
theory, which are relatively simple methods terial at entry and exit of the die and boundary various shapes. Therefore, the advantages of the
compared to the FEM, use a rigid plastic ap- conditions. The strain-rate components that sat- FEM are needed to fully treat the nonlinearity of
proximation for the constitutive equation of the isfy the incompressibility condition are calcu- the boundary conditions of the dies and the con-
workpiece. The approach for the practical use of lated, making the velocity field kinematically stitutive behavior of the workpiece material in
the Slab and UB methods is to assume a veloc- admissible; (that is, the analytical method pos- the process analysis. The major disadvantage of
ity field and for this velocity field to estimate the tulates a velocity field that obeys continuity re- the conventional FEM (as Table 1 shows) is the
average strains, strain rates, and temperatures quirements and the prescribed boundary velocity central processing unit (CPU) time.
within each distinct zone of the velocity field. requirements). On obtaining the velocity field, In order to make the FEM useful in the early
Knowing the effective flow stress and friction, the total power consumption inside the die is design stage to correspond to the simplified
one applies the necessary equations for predict- computed numerically for different conditions method time scale for analysis, a simplified
ing the stress distribution and forming load in of friction, strain-rate, and flow stress. The aver- FEM can be used to speed up the design process.
the Slab method and the average forming pres- age pressure can be calculated since the total As the following discussions briefly note, the
sure in the UBM. power consumption is obtained. In its present simplified finite element method (SFEM) has
These simple models require simple geome- form, the UBET method appears to be close to the following characteristics:
tries and uniform boundary conditions even the FEM with the advantage that the velocity
Only one-step computation is required.
though they can cope with some complexity in field is truly compressible.
History variables are used as initial condi-
the constitutive behavior of a material (Ref 14).
tions only.
The slip-line field (SLF) is an older method that
Finite Element Method (FEM) Geometry-based FEM models can be used.
can be used for plane strain and plane stress
Parametric geometry input can be used,
problems, as it is not applicable for three-
specifically for die and preform shape
dimensional or axisymmetric problems. The ma- The FEM discretizes the continuum body,
design.
terial deforms plastically along trajectories of such as the billet, into many small units called
maximum shear stress called slip lines. Plastic elements just like the UBET method. The differ- The one-step FEM approach dramatically re-
equilibrium equations are made up of slip lines, ence is that the FEM has nodes, which, on the el- duces the computation time, because forming
and the characteristics for stress and velocity are ement boundary, connect neighboring elements. simulations usually require hundreds of steps.
parallel, resulting in the strain-rate for the plane The unknown parameters are located at the This use of the FEM is in correspondence with
strain and plane stress geometries being charac- nodes as nodal velocity, temperature, and so the simplified methods listed in Table 1 where
terized by a single slip line field (Ref 15). These forth. The velocity field or temperature field is these simplified methods are applied to analyze
simple models still have interest for industrial assumed within the element using shape func- a typical step of the forming process during the
process modeling for specific cases where they tions and the nodal unknowns. The material design stage.
are known to be reliable, because they provide a properties, which can be a function of tempera- In many steady-state forming processes such
basis for comparison. These specific cases in- ture, strain-rate, and strain, are located at the nu- as extrusion, profile rolling, ring rolling, and
clude those that can be modeled by steady-state merical integration points inside the elements. drawing, a one-step calculation is adequate in
approximations as extrusion, plane-strain forg- The local unknowns are then assembled into a the early design stages to obtain the forming
ing, rolling, and drawing. global linear equation system. After solving this load and material flow. History variables such as
Mesh-based models are needed, if the process normally large set of linear equations, the nodal temperature and strain, which are important for
design objective is to improve quality by reduc- unknowns are obtained such as velocity, temper- the transient problem, are not considered in the
ing variability in properties, microstructure, and ature, or displacements. Iterations are required SFEM. However, they can be used as initial con-
scrap. The upper-bound elemental technique and because of the nonlinearity of the material and ditions to obtain the correct material properties
the finite element method (FEM) are useful for the geometry deformation. The UBET method for the analysis. Since SFEM is not intended for
this class of industrial problem. does not appear to have as much flexibility as detail design, the effect of history variables can
the FEM when the components being modeled be ignored.
have complicated flow patterns, and three- Geometry-based FEM models allow the user
Upper Bound dimensional analysis is presently difficult to easily set up the FEM model. This type of
Elemental Technique Analysis (Ref 16). FEM model is set up by inputting the geometry
The FEM is widely used by industrial process and assigning boundary and initial conditions,
The upper bound elemental technique modeling teams, because the major advantage which are normally known to the designer re-
(UBET) is useful for parts that are mainly two- for this method is its ability to generalize (Ref gardless of what computation method he or she
dimensional or axisymmetrical. The part is auto- 1517). The FEM has applicability to a wide chooses to use. This is a good approach to use in
matically divided into rectangular or triangular class of boundary value problems with few re- industrial process modeling, because no FEM
zones where the velocity field has a prescribed strictions on geometry. To illustrate the advan- knowledge is required at this stage. The FEM
form. At the interelement boundaries, the normal tage of the FEM, if the objective of an industrial model is created internally based on very few
velocity is continuous, and the elements are design team is to improve the quality of a part control parameters such as the element size or
designed to ensure appropriate boundary condi- with a complex shape by reducing variability in time step. The geometry-based system is now
tions on the tools. As Table 1 shows, the com- the microstructure, mechanical properties, and possible because of the new technology of gen-
puting time is low for computing the forging waste, complex models of the workpiece, forge erating finite element meshes. The Delaunay and
force, material flow, and the local normal stress press, lubricant, and dies that consider second- Advancing Front triangulation methods allow
the user to generate unstructured finite element ous design alternatives can be investigated with simulation. The SFEM has the capability for
meshes automatically on any arbitrary two- less human effort. rapidly analyzing any type of 2D and 3D form-
dimensional and three-dimensional geometry. Comparison of One-Step FEM with the ing process while having many advantages not
The only input required from the user apart from Full FEM. A series of simplified finite element offered by other simplified analysis tools sum-
the part geometry is the element size or mesh method (SFEM) simulations can show how the marized in Table 1. One important advantage of
density. FEM can be made to be in correspondence with this method is that the process designer does not
In industrial process modeling, it is important the upper bound (UB) methods given in Table 1. have to be an expert in the FEM to create the
in the design to allow the design team to change The Figure 3 geometry represents the initial bil- simulation model, as the SFEM model is created
or modify their previous design easily and let geometry, which is a simple cylinder. Figures automatically by the analysis system. The forg-
quickly without having to set up the FEM model 4 and 5 represent intermediate shapes between ing designer does not have to be concerned
again. The parametric geometry attributes make the initial billet geometry and the final geometry about assuming an admissible velocity field or
rapid design changes possible. When one attrib- shown in Fig. 6. The final geometry in Fig. 6 deriving any analytical equations for any partic-
ute changes the whole geometry, the FEM model represents an engine disk typically made from a ular forming process. This is not offered by the
associated with it will change accordingly, with- high-temperature titanium alloy. These one-step other simplified analysis tools summarized in
out user intervention. This flexibility and au- simulations correspond to what a forging de- Table 1.
tomation of FEM model creation make the signer often describes as a short-shot when Table 1 thus summarizes some of the impor-
SFEM a useful analysis tool in conjunction with studying die filling by making partial forgings in tant analytical and mesh-based analysis tools
computer-aided optimization to automate mate- a trial-and-error design process. The reason for that are available for industrial process model-
rial process design and optimization. doing these one-step simulations is to provide a ing. The advantages of the FEM are mainly in its
Several SFEM simulations are subsequently means for the process designer to evaluate any
illustrated to demonstrate the usefulness of the preform shape and die geometry at any forming
SFEM for process design and optimization. stage of the forging process. The SFEM method
Compute Preform Size. Computing the pre- corresponds to the one-step UB analytical meth-
form size is the first thing that a designer will do. ods in terms of simulation time. Figure 7 shows
The SFEM allows the designer to compute the that the CPU time for the SFEM method is about
preform size based on the final part geometry at 70 times less than the time needed for the full
the click of a mouse. The examples in Fig. 1 and FEM simulation.
2 calculate the sizes of the different parts. An ad- The SFEM simulations and CPU times shown
vantage that this method has over other simpli- in Fig. 3 to 6 are for a full elastoviscoplastic ma-
fied analysis methods is that an inexperienced terial with thermal coupling and unloading,
analyst does not have to be concerned about as- which is required for analyzing residual stress
suming an admissible velocity field or deriving patterns in the forged part. The one-step SFEM
any analytical equations for any particular form- allows the forging designer to compute in a sin-
ing process. The SFEM approach to simplified gle step the forming loads, material flow, and die
modeling, like the full FEM, is less restrictive, stresses, and, in general, the material flow simu- Fig. 5 Partial die fill. One-step Analysis at 1.0 s. CPU
time 5.53 s
and, with parametric geometric attributes, previ- lations are comparable to the results of the full
Fig. 1 Engine disk. Central processing unit (CPU) time

less than five s using a Pentium personal com-
puter. Model was an elasto-plastic material with thermal Fig. 6 Die fill with flash. One-Step Analysis at 1.2 s.
analysis. Fig. 3 Initial billet geometry CPU time 7.20 s
Fig. 2 Three-dimensional (3D) wheel. CPU time was

less than five minutes for a one-step simulation Fig. 4 Partial die fill. One-step FEM analysis at 0.25 S. Fig. 7 Die fill with flash. Simulation time 2 S. CPU
of an automobile tire rim. CPU time 7.21 s time 488 s
flexibility for the treatment of complex, 3D force (AF) Manufacturing Science project in the An MPDO program automates existing de-
shapes and complex constitutive laws. The older 1980s. Followed by an air force Man-Tech pro- sign environments and allows engineers to ana-
analytical methods still have value because they gram to modernize the forging industry and to lyze a complex problem by predicting the results
provide a basis for the comparison of specific learn how to design an integral blade and rotor of a particular decision. The MPDO process in-
cases for which they give reliable results. The (IBR) forging. In the 1990s, DARPA funded a tegrates and couples process design simulation
barriers to a widespread use of meshed models similar program with the U.S. engine industry to codes and runs them automatically to produce an
no longer exist as commercial software for auto- reduce the cost of manufacturing precision cast optimal process design. An MPDO environment
matically generating an unstructured FEM mesh turbine engine components. The activity-based enables process designers to explore many more
is available. The FEM models with parametric model in Fig. 8 is still appropriate today, al- alternatives than an engineer can study manually
geometry attributes make it possible for the user though many more new software products are and reach optimal designs faster, resulting in re-
to modify the previous design easily and quickly commercially available compared to the analysis duced product cycle time, reduced scrap and re-
without having to set up the FEM model again, tools (like ALPID, NIKE and AFD) available work, lower manufacturing cost, and increased
as the FEM model associated with an attribute when Fig. 8 was developed. process productivity.
modification will change accordingly without The global approach to automation is straight- The MPDO process permits engineers to
user intervention. Complex shapes are nearly al- forward. Process design automation tools are focus on parts-manufacturing decision making,
ways found in secondary processing such as needed especially when the design team is chal- and it eliminates an iterative, costly trial-and-
forging, whereas primary processing, such as in lenged by customer-imposed quality standards, error manipulation of simulation codes by the
billet conditioning, nearly always produces sim- delivery time, and manufacturing cost and the engineer to obtain an optimal process design.
ple shapes. Thus, mesh-based methods are the team encounters a general problem of a new ma- Figure 9 shows different types of commercially
dominant methods for modeling secondary terial and new product shape, which they have available CAO software that can be packaged
processes and nonmesh, lumped parameters or a never previously manufactured. Because the po- with appropriate process modeling software,
state variable model are dominant methods in tentially viable processing routes are numerous, business models, and standards to form an
the primary processes. many process simulations are necessary to iden- MPDO infrastructure.
tify optimal methods, unless some means of re-
stricting the solution space are carried out. The
Computer Aided Optimization problem of large design solution spaces is not Manual Design Process
restricted to the design of the deformation
The competitive metalworking companies of process. It also applies to the selection of correct The manual design process is costly and com-
today are striving to meet the objectives of lean combinations of processing temperatures and plex. The complexity of the forging process is il-
manufacturing and value-stream analysis to strain-rates and to the identification of optimal lustrated in Fig. 10 by the physics of a non-
give customers exactly what they want on time. heat treatment techniques, material removal steady-state forging process. The complex
Conventional design techniques rely heavily on techniques, customized nondestructive evalua- physics can be broken down into seven design
handbooks and the experience of the design en- tion (NDE) and product testing. tasks (Ref 13). These tasks must be performed in
gineer. These design techniques provide ade- Multidisciplinary process design and opti- proper sequence, and the information generated
quate designs, which are validated through ex- mization (MPDO) provides the framework to in each task must be correctly transferred be-
tensive physical modeling, a process that conform metalworking processes to changing tween tasks, making certain that any redundancy
entails considerable expense and long lead customer and market requirements and to do it in work is eliminated. The manual process can
times. The conventional design techniques may fast. In this framework, a system is viewed as an be prone to error, if a uniform user interface is
not satisfy the speed requirements of a lean interconnection of various interacting compo- not maintained. An overview of the tasks that
manufacturing business strategy. Human in- nents, each being described by a set of relations must be performed in developing a forging man-
volvement is great with the conventional design describing the time evolution of internal vari- ufacturing plan is presented in Fig. 11. This
techniques. Too much time is spent trying to ables and input-output relationships (Ref 19). representation shows inputs, redesign when
squeeze more out of local performance, leaving The MPDO process consists of formulating ap- analysis invalidates a design, and database infor-
insufficient time for engineers to search for the propriate material, process, and equipment mod- mation used by a process designer or planner.
correct combinations of downstream process els together with carefully chosen design objec- The tasks that lead to the creation of a manufac-
parameters needed for arriving at a global-opti- tives and constraints. The selection of models, turing plan are:
mum solution. objectives, and constraints must be considered
Task 1: Integration of geometry repre-
New approaches to process design are now concurrently to be consistent.
sentation
being started in some supplier companies to im- The MPDO process offers a means for au-
Task 2: Initial-guess finished forging
prove the efficiency of upstream process design tomating and satisfying the process design activ-
Task 3: Forging sequence design
by carrying out computer-aided optimization ities. It allows a process design team to combine
Task 4: Processing parameter selection
(CAO) to achieve improvement and ever-higher analytic, numeric, evolutionary computing, and
Task 5: Forging sequence analysis
standards. Today, industrial process design knowledge-directed heuristic search techniques
Task 6: Die design and stress analysis
teams have a choice of commercially available to narrow the design space across engineering
Task 7: Generate key results
process modeling tools for simulating metal- disciplines. From a lean manufacturing perspec-
working processes. Cost-effective design opti- tive, using MPDO software to do iterative, This section provides a conceptual description
mization has been demonstrated using generic costly design tasks address the following basic of these tasks.
CAO software, which can run multivendor sim- issues: Design Task 1: Integration of Geometry
ulation applications, acting as a software robot Representation. Forging process design is con-
to drive these programs without human interven- ducted backwards. It starts with the finished-part
tion to an optimal solution (Ref 18). Activity- Integrating uniformly diverse engineering drawing and ends with a forging design of the
based models (Ref 13) of a modern forging op- and business disciplines part, including the preform shape, flash, and die
eration can also be connected to design activities Using engineering tools correctly design for the finisher die. The finished part rep-
(Fig. 8) by using computer-aided optimization Freeing the engineer from doing costly de- resentation is usually in the form of a geometry
(CAO) techniques. Figure 8 is an activity-based sign analysis iterations database plus the structure and property specifi-
model that was one of the first attempts to mod- Freeing the engineer to do other cross- cations. The latter is typically not archived in the
ernize the forging design process as part of an air functional manufacturing or business work database at this point in the design process.
Fig. 8 Activity-based model for planning and specification of forging process. Analysis tools (like ALPID, NIKE and AFD) in this diagram were those available through government
sources in the 1980s. ANVIL was a commercial CAD/CAM graphics package commonly used by forging designers. Today, many more and better software products are com-
mercially available for use.
This phase includes the collection of informa- that embodies the necessary constraints and re- nomic criteria are beginning to play a part. It
tion about the customers specifications. With a quirements. Fixed solution ideas often have an may be that several preform variants look
lean manufacturing strategy, close contact with adverse effect on the outcome, so at this stage of equally promising, but the final decision must be
the client is sought to discover if any client ex- the design process the following information based on a more concrete level of knowledge.
pectations for the product exist that are not de- must be obtained from the client: An overall objective is to find a preform shape
ducible from the purchase order description of that will fill the die cavity with the least amount
The objectives the client expects to have
the deliverable product, including the required of energy and pressure. Microstructure-property
satisfied
delivery time. The effort is to find out all of the relationships, equipment characteristics, and any
The properties the finished forging must
necessary information needed to satisfy client downstream requirements for heat treatment and
have
requirements and to reach an optimal solution material removal must be factored into the de-
Any general requirements not completely
sign. Using MPDO software as a software robot
specified by the customer
to run process simulation analyses can eliminate
Process design experts Design Task 2: Initial-Guess Finished most of the engineering oversight needed to cre-
Knowledge-directed Forging. The initial-guess finished forging ate an optimized solution by designing the pre-
search geometry is generally experientially based. A form shape and die geometry concurrently.
Knowledge-based design expert will take into account material and One way to softly automate this task is by
expert system (KBES) equipment variables such as the workpiece ma- using artificial intelligence techniques such as
terial, recommended process, and available genetic algorithms and neural network analysis
Numerical CAO equipment characteristics. The interdependence that automatically provide a list of possible
optimization Evolutionary
of forging process parameters is shown in Fig. process design alternatives and are dependent on
computing 12. A design approach that fully utilizes expert the functional requirements (FRs) defined dur-
knowledge and supplements it with numerical ing design clarification. This approach allows
Additional local optimization is warranted. Concept variants the design team to deal with real-world prob-
Novel global
performance must now be evaluated by analysis to eliminate lems, where the best is only a theoretical ideal
optimization
any variants that do not satisfy the demands of that is often unattainable or not cost effective.
Fig. 9 Multidiscipline process design and optimiza- the customers specifications. During this task, Through the use of soft optimization techniques,
tion (MPDO) the chief criteria are technical in nature, and eco- a modest goal of being just good enough can be
Fig. 10 Complex physics of a nonisothermal and non-steady-state forging process

Fig. 11 Precision forging design process. RFQ, request for quote
achieved even for problems in manufacturing

currently considered to be beyond reach by cal-
culus-based methods. This approach to concep-
tual design is expected to reduce the time for ar-
riving at a set of design parameters (DPs). It is
highly unlikely that efficient algorithms for the
solution of these problems of arbitrary size will
be found. Figure 8 shows an integrated com-
puter-aided value-flow diagram. The overall ob-
jective is to find a solution for the finished forg-
ing that will have an acceptable level of residual
stress and a uniform distribution of the residual
stress after quenching hot from the finisher die.
The objectives included getting the correct serv-
ice properties while minimizing any distortion
during subsequent machining (Fig. 13).
Design Task 3: Forging Sequence Design.
Forging sequence design requires tracing back-
wards to acceptable starting billet geometry.
This design task is required when several inter-
mediate shapes may be required to reach the de-
sired finished shape. An important aspect of se-
quence design is distributing uniformly the
Fig. 12 Interdependence of forging process parameters energy dissipated between the different se-
quence stages. Normally, the process designer (e.g., see Chapter 1, Workability and Process and an optimal set of forging microstructures
knows the principal design parameters that af- DesignAn Introduction, for a general intro- and properties.)
fect the design objective or violated constraint duction). One method is the so-called dynamic Besides stable material flow, the database in-
but does not know the direction or amount to material model (DMM) (Ref 20), which is a fun- formation correlating processing-microstructure
change these parameters. Frequently, evaluation damental methodology for selecting temperature relationships can be used to identify the process
of individual variations may lead to the selection and strain rate for stable deformation and mi- window within which both stable material flow
of one that is particularly promising but may be crostructure evolution. The DMM is a criterion and appropriate microstructure evolution is
further improved by considering other appropri- for mechanical and microstructure stability, achieved. The optimal combination of tempera-
ate combinations of process parameters and the which is based on continuous field variables that ture and strain-rate values will produce the least
elimination of weak links. A knowledge-based have an immediate and direct influence on the variation in service properties such as grain size,
expert system can support this scenario by in- process. Microstructure-based models, in con- tensile strength, fracture toughness, and fatigue
corporating other appropriate ideas and solu- trast, may be more difficult to use, as they are strength (Ref 19, 21, 22). When the workpiece
tions, allowing the designer to specify the pa- generally based on the secondary effect of the material is processed within the range of tem-
rameters to change and then call a numerical field variables on some microstructural condi- perature and strain rate given by the material sta-
optimization with just these design parameters tion such as grain growth above a certain size. bility analysis, the process is not sensitive to per-
to determine their best set of values. If no design This can require iterative coupling with the rest turbations of either process parameter, and the
knowledge is available, a hybrid optimization of the process model. Microstructural models material stability is path independent. The DMM
approach may be combined to get a better global also may be tough to use, because of the analysis also helps the design team to define
optimum. complex physical metallurgy of hot working process variables such as die temperature, the
Design Task 4: Processing Parameter processes. workpiece temperature before entering the die,
Selection. A strong relationship exists between The designer is usually concerned with defect and the workpiece temperature range during de-
the process variables and the machine variables, avoidance, for example, preform buckling; formation.
and this design task is concerned with establish- underfill; flow related defects such as shear A strong relationship exists between the
ing optimum process and machine variables. bands, lap or fold, suck-in, shearing, or flow process variables and the machine variables.
Process parameters include the workpiece defor- through defects; free surface cracking; grain Figure 12 shows the interdependence of process
mation rate, which is expressed as the slide boundary cavities; and triple junction cracks. variables. Forging companies have access to a
velocity and the effective strain rate, and the Typically, temperature and strain-rate ranges are variety of types of forging equipment (Ref 23).
temperatures of the workpiece before forging entered as nonholonomic constraints to avoid However; the choice of equipment to produce a
and the temperature during forging. These overconstraining the system. (A nonholonomic particular part is often limited to the equipment
process variables are evaluated during this stage constraint is an inequality that avoids overcon- that is available in the forge shop. Accordingly,
of design, with the objective of determining op- straining a set of equations, making it possible to it is not always possible to select the optimum
timum values in terms of overall workability. get a solution for the unknown parameters. This type of equipment for a particular job. It is for
This task can encompass the meaning of worka- approach is used as a means for incorporating this reason that it is important to understand the
bility in its widest sense; that is, workability cri- material behavior information into the FEM principal differences between the various classes
teria defined by defect prevention, effective model to define the stability ranges of tempera- of equipment and how these differences affect
equipment utilization, and proper development ture, strain rate, and strain for the workpiece ma- the metal forming process and die design deci-
of microstructure. terial that is akin to the concept of material sions. This particular knowledge is important for
Evolution of microstructure during hot work- workability. It is a consistent approach for the manufacturing of high quality precision
ing is a major factor in process design, and many applying process optimization methods to forgings. The important characteristics that must
methods have been investigated and developed achieve a stable process, preform and die shapes, be understood when selecting the forging equip-
ment are listed as follows:
Workpiece deformation rate: This parame-

ter is expressed as the slide velocity equip-
ment parameter and the effective strain-rate
of the workpiece material, a process vari-
able. The workpiece deformation rate
should be strictly referred to as the relative
velocity between the dies when referring to
the equipment. The choice of relative die
velocity is determined by the intrinsic work-
ability of the workpiece material, which is
determined by the prior thermomechanical
history of the forging billet.
Dimensional control: This process variable
reflects the capability of the equipment to
consistently achieve part tolerance. Dimen-
sional control must be achieved along with
achieving a controlled set of microstruc-
tures and properties in the finished product
on a repeatable basis. Consistency is
achieved only when all components in the
forging cell are individually in control.
Production rate: Production rate equates to
the number of parts per unit time. The pro-
Fig. 13 General flow chart of forging process design to obtain parts with acceptable properties, residual stress dis- duction rate is dependent on the type of
tribution, and dimensions press and the ability of the workpiece mate-
rial to respond to the demands of the niques may be required to provide the best solu- erally, making it feasible to fully utilize expert
process. The forging press and the work- tion. Optimization studies show that no single knowledge and to supplement it with other opti-
piece material must be properly matched to technique works for all design problems. Be- mization techniques. At this stage of process de-
achieve an optimal production rate. cause an experienced designer uses experiential sign, the designer typically compares the elastic
Materials such as titanium and nickel-base rules to decide sequencing, it is likely that a deflections of alternative die materials. A com-
alloys that have a high strain-rate sensitivity knowledge-directed approach supplemented by plete design interaction approach, where there
must be forged at lower die velocities as in another optimization method, for example, nu- are varying sizes, strain levels, and forces to
hydraulic presses to avoid a rapid rise in merical optimization, may be needed to get the specify the die diameter and thickness compati-
flow stress and die stress values and to avoid best solution. ble with the die support system, is generally
the formation of defects and plastic instabil- Design task 6: die design and stress analy- done.
ities. The equipment parameters that control sis provides the detailed loading distribution on The designer does a separate analysis of the
the production rate are the stroking rates, the die surface. Equilibrium elasticity analysis strain concentration factors. This analysis is
ram velocity, machine energy, and the de- of the dies is done to make certain enough die done within the die impression and on the die
gree of automation and computer control. material exists to distribute the load satisfacto- outer shapes such as ejector holes, and threaded
rily and to determine elastic die deflections. An surfaces. The local shape change produces ef-
Design Task 5: Forging Sequence Analysis. experienced designer does die block sizing gen- fects that are confined to the specified zone, and
The detail forging sequence design is the phase
of the design process in which the arrangement,
form, dimensions, surface properties of all the
individual components, for example, the equip- 1. Billet material specification 5. Post forge inspection
ment, workpiece material, lubricants, separating Alloy: Dimensional check:
agents, dies and tooling, are finally laid down. Specifications: Defects:
All of the materials used in the process are spec- Melting specs:
ified, and the technical and economic feasibility Material source:
Heat no.
are rechecked. All the drawings and other pro-
Conversion:
duction documents are produced. It completes Surface condition:
the substantiation of the forging process with 6. Heat treatment
final instructions about the layout, shapes, di-
Furnace preheat:
mensions, microstructures and properties, and a 2. Billet material quality Furnace setpoint:
final scrutiny of the manufacturing methods and assurance Time at temperature:
costs. (Mill data or acceptance tests) Cooling from furnace:
Unlike the finisher die and workpiece preform
design that goes on in the backward direction Composition: Repeat these
from the final requirements, the forging se- Ultrasonic: blocks as
Property data: appropriate
quence analysis task progresses in the forward
Macro etch:
direction. This reversal of methodology is done
Microstructure:
because the initial conditions are not generally
known for intermediate steps in a nonisothermal 7. Finishing operations
forging sequence without reheats. This design 3. Billet preparation
stage is quite crucial, as very often corrections Machine to sonic shape:
Conversion:
must be made during this work and preceding Billet shape: Cut test samples:
steps repeated. The crucial activities are opti- Cut length:
mization of the principle and optimization of the Conditioning:
layout and intermediate workpiece shapes. Identification:
These activities influence each other and overlap Coating:
considerably. In general, optimization of the in-
termediate workpiece shapes, geometries of the 8. Final quality assurance
4. Forging practice
dies, and, hence, optimizations of the manufac-
(This will be repeated for each
turing process begin to assume growing impor- step in forging sequence)
Test type:
tance as substantiation proceeds. Test locations:
Identification:
Sequence design involves studying a combi- Preform conditioning:
Ultrasonic:
nation of circumstances that, when considered Lubrication practice:
Mag particles:
together, will result in lowering the overall cost Fritting:
Die penetrant:
Furnace:
of the forging and satisfying customer specifica- Heating practice:
Microstructure:
tions. Attention must be given to optimize such Macro slice:
Temperature:
factors as throughput, material and energy con- Fracture toughness:
Maximum time:
Tensile:
sumption, and tool and die life extension, the Atmosphere control:
Fatigue:
number of tool changes, and tooling cost. Many Press:
Creep rupture:
intermediate results are obtained for the forging Tooling assembly ID:
Press controls:
sequence and process parameter designs. These Velocity:
results, in the manual approach, often require an Stroke:
experienced engineer to do trial-and-error simu- Load:
lations over several weeks, which involves Identification (serial no.):
costly, iterative manipulation of simulation pro- Trimming:
Hot inspection:
gram parameters. The tools of MPDO poten-
tially can help the engineer reduce the design
space, and a combination of optimization tech- Fig. 14 A completed schematic process and inspection traveler
these local zones are the ones where failure is forging for critical parts, which, if failure oc- integrated design approach that is based on the
likely to occur. If the die loading during the forg- curred in service, would threaten the lives of use of commercially available computer-aided
ing process is cyclical, the designer should do a human beings. A gas turbine aircraft engine disk optimization (CAO) tools for integrating, au-
stress factor calculation to include the effect of is an example of a forged critical component that tomating, optimizing, and accelerating the entire
fatigue on die life. A wear analysis can be done requires close process and inspection controls. A design process and using a knowledge-based
to show areas that will be prone to wear by abra- schematic of a process and inspection traveler is system for process design. The implementation
sion, and candidate surface regions can be iden- shown in Fig. 14. of such a system requires a detailed specification
tified for surface hardening by ion implantation A part traveler should be automatically gener- of the methodology that is involved. A represen-
or chemical vapor deposition. Die drawings are ated by an integrated software system that is ca- tation of an activity-based method description
prepared as required based on the detailed re- pable of fusing all of the important process and was shown in Fig. 8.
sults of simulation-based design. This activity material information into a unified document.
includes any open draw dies, preform dies and All key results should be made available, includ-
finish dies or inserts, including impressions, ing key intermediate results from process se- REFERENCES
bearing and cleared surfaces, locking and attach- quence design and process parameter design.
ing methods, ejection designs, and thermocou- Some key design results for these two critical 1. P. Sargent and H. Shercliff, Modelling
ple holes. design tasks are shown in Fig. 15. Materials Processing (Version 2).
Design Task 7: Generate Key Results. The Traveler documents are an authenticated, ex- [CD-ROM]. CUED/C-MATS/TR.206,
end product of process design and analysis is a tended, and complex substantiation of the total Cambridge, England: Department of
manufacturing plan. A complete process specifi- forging process, which is made up of compli- Engineering, Cambridge University,
cation called a process and inspection traveler is cated and related parts. The previous discussion Cambridge, CB2 1PZ England, November
created. The traveler generally accompanies the of the interconnected design tasks provides an 1993
2. J.R. Wright, Forging a Culture of
Competitiveness, Forg., Fall 1995, p 23
3. P.F. Drucker, The New Productivity
Challenge, Harvard Bus. Rev., Vol 69 (No.
6), 1991, p 6979
4. J.S. Brown, Research that Reinvents the
Organization, Harvard Bus. Rev., Vol 69
(No. 1), 1991, p 102111
5. M. Hammer, Reengineering Work: Dont
Automate, Obliterate, Harvard Bus. Rev.,
Vol 68 (No. 4), 1990, p 104112
6. R.H. Hayes and K.B. Clark, Why Some
Factories Are More Productive Than
Others, Harvard Bus. Rev., Vol 69 (No. 1),
1991, p 6673
7. M. Imai, Kaizen, McGraw-Hill Publishing
Company, 1986
8. R. Charan, How Networks Reshape
Organizations, Harvard Bus. Rev., Vol 69
(No. 5), 1991, p 104115
9. E.A. Haas, Breakthrough Manufacturing,
Harvard Bus. Rev., Vol 65 (No. 2), 1987, p
7581
10. Knowledge-Based Integrated Design
System, Computer-Aided Materials Selec-
tion During Structural Design, Committee
on Application of Expert Systems to
Materials Selection during Structural
Design, NMAB-467, National Academy
Press, Washington, D.C., 1995, p 5762
11. J.P. Womack and D.T. Jones, Lean
Thinking: Banish Waste and Create Wealth
In Your Corporation, Simon and Schuster,
1995, p 990
12. D.K. Sobek, II, J.K. Liker, and A.C. Ward,
Another Look at How Toyota Integrates
Product Development, Harvard Bus. Rev.,
JulyAugust 1998, p 36
13. H.L. Gegel, J.C. Malas, S.M. Doraivelu,
and V.A. Shende, Forging Process Design,
Forming and Forging, Vol 14, ASM
Handbook (formerly Metals Handbook, 9th
ed.) ASM International, 1988, p 409
14. T. Altan, S. Oh, and H. Gegel, Metal
Fig. 15 Examples of Key Intermediate Results from the critical steps of process-sequence design and process- FormingFundamentals and Applications,
parameter design ASM Series in Metal Processing, H.L.
Gegel, Ed., American Society for Metals, 19. J.C. Malas and W.G. Frazier, Optimal cesses, Ph.D. dissertation, College of
1983, p 131142 Design of Thermomechanical Processes Engineering and Technology, Ohio Uni-
15. R.H. Wagoner and J.-L. Chenot, Funda- Using Ideal Forming Concepts, The Inte- versity, November 1991
mentals of Metal Forming, Wiley, 1997 gration of Material, Process and Product 22. V. Srinivasan, E.A. Medina, J.C. Malas, III,
16. R.H. Wagoner and J.-L. Chenot, Metal Design, Zabaras et al., Ed., A.A. Balkema, S. Medeiros. W.G. Frazier, W.M. Mullins,
Forming Analysis, Cambridge University Rotterdam, 1999, p 229236 and R. Srinivasan, Optimization of Micro-
Press, 2001 20. H.L. Gegel, J.C. Malas, S.M. Doraivelu, structure During Deformation Processing
17. T. Altan, S. Oh, and H. Gegel, Metal J.M. Alexander, and J.S. Gunasekera, Using Control Theory Principles, Scr. Mater.,
FormingFundamentals and Applications, Material Modeling and Intrinsic Work- Vol 36 (No. 3), 1997, p 347353
ASM Series in Metal Processing, H.L. ability for Simulation of Bulk Deformation, 23. T. Altan, S. Oh, and H. Gegel, Metal
Gegel, Ed., American Society for Metals, Adv. Technol. Plast., Vol 1, Springer-Verlag, FormingFundamentals and Applications,
1983, p 329339 1987, p 1243 ASM Series in Metal Processing, H.L.
18. iSight Technical Overview, Engineous 21. J.C. Malas III, Methodology for Design Gegel, Ed., American Society for Metals,
Software, Inc., Morrisville, NC 27560, 1997 and Control of Thermomechanical Pro- 1983, p 536
Chapter 23
Optimal Design of
Thermomechanical Processes
J.C. Malas, W.G. Frazier and H.L. Gegel (Retired), Air Force Research Laboratory, Wright-Patterson Air Force Base
V. Srinivasan, Indira Gandhi Centre, Kalpakam, India
OPTIMAL DESIGN of a multiple-step pro- Improve product design jectory optimization are discussed with exam-
cessing system consists of formulating appropri- Reduce cost and time-to-market ples to illustrate potential benefits. The objective
ate models of materials, process, and equipment Use simplified models and concepts cost ef- is to show a system design approach that holds
behavior together with carefully chosen design fectively to identify optimal workpiece ma- promise for substantially reducing manufactur-
objectives and constraints. In order to be con- terial trajectories ing cost, risk, and delivery time to a customer,
sistent, the selection of materials and process while resulting in an improved product that sat-
models, objectives, and constraints must be con- Substantial improvements in effectiveness isfies customer requirements and expectations.
sidered as an integrated whole, and not inde- and efficiency can be realized through inte- Critical issues such as efficient material flow
pendently. The goal is to bring all of the process grated approaches that optimize the whole sys- and the sensitivity of the variation of finished
elements together to address the commercial tem performance and not just individual sub- parts properties to small changes in key process
need to reduce cost and to save time by reducing systems such as workpiece materials behavior, parameters are addressed using lumped-parame-
iteration of trials and improving quality by re- material flow in dies, and equipment responses. ter models of microstructure evolution and opti-
ducing product and process variability, while Vendors who find creative ways to reach out to mization techniques for controlling microstruc-
avoiding over-optimization of any particular ele- their customers to use products and services ture during thermomechanical processing. The
ment of the system. Parts suppliers are continu- more effectively become more indispensable to use of the state-space (lumped parameter) method
ally pushed to reduce prices, perhaps at the ex- their customers. provides considerable insight into the controlla-
pense of quality or other considerations. To deal Design engineers work under considerable bility of metalworking processes. It is a con-
with quality issues, companies need to become time pressure, and they need to have easy-to-use strained optimization method, where the overall
more effective in using their existing design tools to speed the design of manufacturing design is specified as constraints and objectives
software and adopt upstream design processes processes, such as hot metal forging. Metal- to accommodate multiple physical and econom-
that enable them to deliver customized services working industries seeking to be responsive to ical requirements. This approach identifies opti-
and products at relatively low cost. the competitive needs of their customers must mal workpiece material trajectories that are gen-
As noted in the preceding chapter, Multi- become more effective in using their existing de- erally needed for achieving customer product
disciplinary Process Design and Optimization: sign software and human resources. It is increas- specifications.
An Overview, multidisciplinary process design ingly important to adopt upstream technical and Optimization is just one method for perform-
and optimization (MPDO) is a software infra- business processes that enable them to deliver ing a process design. Whether the resulting de-
structure that links tools from different environ- customized services and products at relatively sign is the best design depends entirely on the
ments, while providing a software framework low costs and deliver them with the expected criteria specified by the product designer. As
for automating, optimizing, and integrating the quality when the customer expects the parts to illustrated in the section Material Trajectory
engineering design processes (Ref 15). In this be delivered. Basic design tasks, such as select- Optimization in this chapter, optimization tech-
framework, a connection exists between various ing the number of forming steps and specifying niques require the specification of two types of
interacting components of the metalworking the processing conditions for thermomechanical criteria: objectives (wants) and constraints
process, which includes models for the work- operations, may also result in feasible solutions (needs). To achieve the desired goals, the de-
piece, lubrication, tooling, metalworking equip- that are not necessarily optimal. Therefore, new signer must specify all relevant criteria and care-
ment, inspection tools, analysis, and so forth. process design and control methods are still fully determine whether the criteria are either
The benefits of MPDO are measured in terms of sought for optimization of cost, quality, and pro- objectives or constraints. For example, it may be
the ways it helps process design teams do their ductivity of thermomechanical processes. desirable to minimize the production costs (ob-
jobs. Multidisciplinary process design and opti- This chapter describes a strategy for system- jective) while maintaining specified product
mization helps the design team: atically calculating near optimal control param- quality standards (constraint). On the other hand,
eters for hot deformation processes. The focus is the opposite scenario may be desired, i.e., max-
Create innovative design alternatives on a design concept of ideal forming that is imize the product quality (objective) while not
Manage the dynamical and material stabil- based on state-space control-theory concepts, exceeding a specified cost (constraint). Effective
ity effects in the design of thermomechani- materials stability analysis, and geometry map- optimization strategies consider the entire pro-
cal materials processes ping between the initial and final states to pre- cessing design problem, not just some parts,
Handle explicitly design objectives and con- dict important process parameters. The applica- thereby avoiding over-optimization of parts of
straints on workpiece materials, processing tion of dynamical modeling and process design the process at the expense of the whole manu-
equipment, and tooling techniques using ideal forming concepts and tra- facturing enterprise.
Concepts of Dynamic Modeling in A material trajectory is defined here as the as factors in simplifying the process model. Time
Optimal Design time evolution of materials state with respect to constants for mechanisms that are appreciably
strain, strain rate, temperature, and temperature longer than another mechanism can be consid-
rate. The development of materials processing ered to be operating in a continuum and amenable
A metalworking manufacturing enterprise for models invariably involves quantitative descrip- to an analytic model.
the purpose of this discussion consists of a se- tions of the underlying process physics. Materials The use of the state-space (lumped parameter)
quence of thermomechanical processing steps to process models always involve descriptions in method provides considerable insight into the
produce a component. Each unit process, such as terms of sets of partial differential equations that controllability of metalworking processes. It is a
forging, heat treatment, material removal, and are dependent on time, spatial dimensions, field constrained optimization method in which the
nondestructive evaluation (NDE), can be decom- variables, and internal states. Because several dif- overall design is specified as constraints and ob-
posed into subsubsystems, i.e., the workpiece ferent processing histories (trajectories) can lead jectives to accommodate multiple physical and
material, tooling, equipment, control system, and to nearly the same end result, it is important to economical requirements. This approach identi-
so forth, which can be decomposed further. A be able to find the best trajectory to realize the fies optimal workpiece material trajectories that
typical block diagram for a hot forging equip- desired objectives. Consequently, models and de- are generally needed for achieving customer
ment system is shown in Fig. 1. Each component sign techniques for controlling microstructure product specifications. Critical issues such as
of the manufacturing system is represented by a during thermomechanical processing have been efficient material flow and the sensitivity of the
set of relations, known as state equations, that developed to address critical issues such as sta- variation of finished parts properties to small
describe the time evolution of internal variables bility, transient and steady state response, and changes in key process parameters can be ad-
and input-output relationships. When these state robustness of material trajectories. dressed by using lumped-parameter models of
equations and input-output relationships are If the required shape does not involve complex microstructure evolution and optimization tech-
viewed as a whole, they provide a complete sys- geometries (as in primary processing operations), niques for controlling microstructure during ther-
tem description (Ref 1). constraints are simplified, and the process model momechanical processing. Dynamic models of
The systems approach to process design, when can be simplified to solve a lumped-parameter equipment systems are also useful and essential
linked to computer-aided optimization models type problem. Lumped-parameter problems can in determining the desired adjustable parameter
for design and analysis, allows simplified form- be represented by a set of ordinary differential settings for coincident tracking of the equipment
ing models to be used. Manufacturing processes equations, and they can be solved using simpler response with optimized commands. In general,
can be mathematically modeled as nonlinear dy- means than the complex apparatus of numerical furnaces and forming machinery possess a range
namical systems using a state-variable formula- analysis that is required for solving partial differ- of time-varying performance capabilities that can
tion, i.e., a system of coupled, first-order non- ential equations. Whether the physical shape is be tuned to satisfy the needs of a given process.
linear differential equations. Symbolically, this is sufficiently complex to justify developing meshed In one example, a 1000 ton hydraulic forge press
often written in the form: models is a difficult decision to make. One does is evaluated using a dynamical model (Ref 6).
x (x, u) (Eq 1) not usually know a priori if some geometrical Materials Modeling and Stability. Dynamic
feature is significant. The idea of shape deter- models of materials behavior are especially im-
where x and u are vectors of the systems state mining a materials process model type may not portant because of the time-varying behavior of
and control variables, respectively. The function be a physical shape, but it could be the shape of quantities such as microstructure, flow stress,
f defines the relationship among the current state a heat field. In this case, the material determines and defect formation. Industrial process design
of the system, the current control variables, and shape simplification. Materials such as titanium-, and optimization depends on the type of materi-
the rate of change of the state. The important nickel-, and iron-base alloys have smaller heat als model created and the existence of a cost-
point concerning dynamical models is that they capacities and thermal conductivities than alu- effective and efficient method for obtaining ma-
are valid for a broad range of control signals, un- minum, beryllium, and copper alloys have. The terials parameters that must be measured (Ref 1,
like algebraic models, which are only valid for a materials with a sufficiently low heat capacity 4, 712). The type of materials model created
particular class of process controls, such as con- and thermal conductivity frequently can be mod- for industrial process simulation depends on
stant temperature and strain rate. The use of dy- eled as semi-infinite and highly symmetric, mak- two issues:
namical models provides greater predictive ca- ing an analytic model often entirely adequate.
pability over algebraic models and provides more Time constants for different mechanisms such as The existence of an industrially sensible
degrees of freedom in the time domain for opti- heat diffusion and bulk or grain-boundary diffu- method for obtaining the needed materials
mal process design. sion of mobile species should also be considered parameters
Fig. 1 Block diagram of metal forging system

Chapter 23: Optimal Design of Thermomechanical Processes / 339
The danger that measured parameters may deforming continuum. Steady-state phenomena Workpiece temperature after transferring it
have little value or relevance for another often represent a limiting condition that can be to the die system
material with a slightly different prior ther- achieved only under unique combinations of Workpiece temperature at the start of defor-
momechanical history temperature T and effective strain rate, e . There- mation
fore, it is highly desirable to be able to define Workpiece temperature at finish forging to
The latter item may be particularly troublesome stable regions in terms of the kinematic variables control shrinkage
because there are several reasons why materials T, effective strain (e), and e . Optimal die temperature and die material
properties may be variable or erratic. For exam- A solution to this problem requires identifica- Proper die velocity and material strain-rate
ple, even engineering alloys of similar composition of the limiting conditions for the loci of bi- range
tion can have lot-to-lot variations in properties, furcation points where two atomic mechanisms Press size needed for required force and
making it difficult to describe precisely the ini- are operating simultaneously to produce a max- energy
tial conditions needed for a materials model. imum in the energy dispersal rate at a unique Forge press type compatible with workpiece
Engineering alloys may also have multiple mech- combination of temperature T and effective strain-rate sensitivity
anisms that provide different degrees of freedom strain rate e . Under these unique conditions, the Prefinish shapes for proper metal flow
for dissipating energy while undergoing forced deformation process would be steady state. One Service properties of forged components
dissipative flow during deformation processing. branch of the deformation process would be sta- Lubricant type
Some of the dissipation mechanisms include ble, and the other would be unstable. Processes Maximum die life
lamellar kinking, adiabatic shearing, grain-bound- such as grain-boundary cavitation, wedge crack-
ary cracking, dynamic grain growth, grain-bounding, and cleavage are considered to cause the Industrial simulation of deformation processes
ary diffusion, dynamic recrystallization, and dy- deforming continuum to become unstable be- requires mathematical relationships that predict
namic recovery. These mechanisms may operate cause the free surfaces formed increase, rather how the workpiece material will respond to the
either in series or in parallel and at unpredictable than decrease, the free energy of the system. In demands of the process. Overall, this response is
times. Thus, the models of alloy deformation this sense, workability can be viewed as an in- expressed as a set of constitutive equations that
and workability can produce erratic results, espe- trinsic characteristic of the material; the ability relate the effective stress (effort variable), s, and
cially when materials stability is path dependent. of the material to dissipate power at any state of the effective strain rate (flow variable), e , in the
Processing predominantly in a region of ma- stress by favorable metallurgical mechanisms current state of the material system. The scalar
terials stability can reduce the sensitivity of er- generates entropy at a faster rate than it is ap- product of these two variables is the instanta-
ratic results due to materials variability. This is a plied and maintains the total energy of the sys- neous power. The state of the system is deter-
key point in the concept of process design opti- tem at the lowest level possible. mined by the prior thermomechanical history of
mization. A stable processing space is a region Identification of the limiting conditions for sta- the workpiece material and its current tempera-
defined by temperature, effective strain rate, and ble plastic deformation is an important concern, ture (Ref 9, 17).
time, where the desired microstructures and as- as discussed briefly in the next section, Materials It is not practical to develop a general consti-
sociated properties evolve in a stable fashion re- Stability during Forced Dissipative Flow. Mate- tutive equation for all materials. When large
gardless of the state of stress. In contrast, insta- rials stability criteria also can be easily incorpo- plastic deformations are involved, it is reason-
bility is defined as any set of conditions during rated in a simulation model as a so-called non- able to neglect the elastic strains. However, when
processing that cause plastic instabilities such as holonomic constraint (that is, a constraint based the elastic strains are of the same size as the
flow localization, fracture, grain-boundary cavi- on an inequality condition, which does not exces- plastic strains, such as when predicting residual
tations, and microstructural instabilities such as sively constrain a set of simulation equations) stresses, constitutive equations should include
dynamic grain growth to occur. The occurrences (Ref 13, 14). These stability limits are the con- elastic and plastic behavior. Two general ap-
of these defects are strongly material path- or straints needed for designing preform and other proaches are used to develop constitutive equa-
trajectory- (Ref 1) dependent when processing intermediate product shapes, which are required tions for material flow analysis. One approach is
occurs outside the limits for stable material flow. for optimizing metalworking processes used to micromechanical and the other is empirical. A
Materials Stability Criteria and Intrinsic produce components with complex shapes. lack of basic quantitative understanding of the
Workability. For lumped-parameter modeling, Materials Stability during Forced Dissipa- fundamental material energy dispersal mecha-
considerable attention is paid to materials stabil- tive Flow. Successful industrial process design nisms, their high variability from material to
ity criteria, constitutive laws, and physical in- and optimization demand serious consideration material, and different modeling needs for the
sight about the materials processes. The concept of materials data needs. In addition to funda- various processes all work against the usefulness
of materials stability is akin to the idea of work- mental knowledge about materials properties of micromechanical and general models, making
ability (Ref 14); it can be expressed in terms of such as the moduli of elasticity, heat capacity, empirical materials modeling the most useful for
key process parameters, temperature, and effec- thermal expansion, Poissons ratio, density, com- industrial process modeling.
tive strain-rate ranges as a function of time, where pressibility, and the transport properties, it is es- Empirical descriptions of materials behavior
microstructures and mechanical properties evolve sential that process designers understand how that are based on a series of well-lubricated iso-
without forming deleterious structures regard- the workpiece material responds to the demands thermal, constant strain-rate compression tests is
less of the state of stress. of the process. Whether material flow dissipates effective and affordable because it is a simple
Materials being formed (forged) undergo large, in nonhomogeneous flow and fracture or leads test that is easy to understand and analyze. When
nonlinear, irreversible deformation that is made to large-scale order depends on the nature of in- conducted over the process parameter range of
possible by linking together several atomic stabilities in the system (Ref 15). interest, all materials information needed for
processes. This string of processes, in addition Decisions about forging process parameters calculating the force and energy of forging and
to providing the degree of freedom needed for depend on important materials properties, which deciding the important process parameters
producing large deformation, is also responsible are based on fundamental thermodynamic stabil- needed for achieving mechanical and microstruc-
for the evolution of structure. Each of these ity parameters such as mechanical, thermal, and tural stability is generated (e.g., Ref 18, 19).
processes can be thought of as a channel for dis- diffusion stability. Following are some impor- This approach is based on careful measurements
persing energy supplied by the forging process, tant process parameters (Ref 16) that are con- of the material flow stress over a range of tem-
and each can give rise to entropy production. trolled by the workpiece material: perature and effective strain rates and strain. The
Forging, as a rule, is a shape-changing process in experimental results can be described analyti-
which the deformation is generally inhomoge- Workpiece temperature at the time it is re- cally to produce constitutive equations, or re-
neous and transient over a large volume of the moved from the heating equipment sults can be tabulated in spreadsheet form as a
function of temperature, effective strain, and stress (y at constant temperature and effective that the system is approaching steady state and
effective strain rate. The empirical approach is strain rate) approaches zero asymptotically as the thermodynamic stability criterion for me-
useful for industrial process modeling also, be- steady state is moved toward. When steady state chanical stability is satisfied. Similarly, another
cause it requires the least amount of information is reached, the processing map no longer changes system parameter s that can be used in the
needed for determining the intrinsic workability with time (strain) and the microstructure and Liapunov sense to evaluate microstructural sta-
of the workpiece material. The workability of properties are no longer evolving with deforma- bility has been defined. The Liapunov function
the workpiece material is derived from the con- tion. This materials behavior at large-scale, sta- for structural stability is Ls s(ln e ) 0, and
stitutive equation data. ble deformations helps to explain why simple the condition for structural stability in the
A constitutive equation for describing the compression test data can be used to simulate s
Liapunov sense is given by L s = < 0 . This
flow stress behavior based on compression test- different classes of bulk hot-working processes, ln e
ing data that is corrected for adiabatic heating is although the effective strain levels can be as large parameter ensures that the thermodynamic diffu-
expressed as: as 2.0% using constitutive equation data that sion stability criterion is satisfied and that all of
were generated as a function of effective strain the phases given by the material phase diagram
s ( e , T, t)
s (Eq 2) levels that were no larger than 0.4 and 0.6. can form. After the stability criteria are calcu-
Observations such as the latter imply that lated, the negative values are grouped to decide
This equation can be represented in differential steady state is a long-term attractor, and it is a the stable regions of the material in processing
form such that: fundamental materials behavior characteristic, space. If the workpiece material is processed
which can be approximated with a Liapunov under stable conditions, the necessary and suffi-
d ln s md ln e jdT y dt (Eq 3) function. This has led to the development of a cient conditions for stable material flow and mi-
method referred to as dynamic material model- crostructure evolution are satisfied regardless of
where m is the well-known strain-rate sensitivity ing (DMM), which is briefly introduced in Ref the state of stress. Conversely, if the workpiece
parameter, j is the temperature dependence of 14 and Chapters 1, Workability and Process material is deformed in an unstable region, it has
the flow stress, and y is the rate of change of the DesignAn Introduction, and 2, Bulk Worka- the necessary but not sufficient conditions to
flow stress. The coefficients are given by their bility of Metals, of this book. More detailed form a structural defect. In this case, the state of
stress, sm/s, may prevent a fracture-related de-
respective partial derivatives: discussions of the DMM methodology are de-
scribed in Ref 8 to 10, and Ref 8 is a com- fect from forming if this ratio is less than zero.
pendium of compression data for 160 alloys with Thus, the DMM methodology is a simple
ln s e s technique to use in an industrial environment. It
m= = processing maps in support of this modeling
ln e e ,T s e e ,T (Eq 4a) method. is the type of materials model for which the field
The DMM method identifies process design variables such as temperature, strain rate, and
ln s windows for materials stability based on constrain have an immediate, direct influence, and
f= cepts from continuum mechanics and irre- they appear in the stability criteria. It describes
T e , e (Eq 4b) the dynamic path a material element takes in re-
versible thermodynamics. The stability of the
workpiece material and the evolution of micro- sponse to an instantaneous change in e at a given
ln s structure and properties is described using a temperature T and e . As such, it is a map that
y= graphically describes power dissipation by the
t T , e (Eq 4c) classical Liapunov stability criterion that depicts
how fast the deforming workpiece material is workpiece in stable and unstable ways. In con-
approaching steady state conditions. The theo- trast, models involving real variables that can be
In plasticity analysis, these derivatives repre- retically derived Liapunov stability function measured, such as grain size, grain growth rate,
sent materials properties. They are as important cannot be solved directly because the forcing and fraction transformed, are very difficult to
as the flow stress for characterizing materials function cannot be specified a priori. The reason devise and calibrate because microstructure pa-
behavior during processing. These derivatives for this dilemma is that the initial state for the rameters tend to be affected by nearly every
have been used to define two parameters for workpiece material cannot be precisely speci- physical effect in the process. Thus, it is much
axiomatically defining the mechanical and struc- fied, and the plastic deformation process in- easier to optimize a process to avoid a defect than
tural stability constraints as a function of T, e , volves instabilities that are stochastic. Therefore, to predict its occurrence or the numeric level of
and e for use during materials process design some microstructurally related feature. Physi-
the Liapunov function must be solved axiomati-
and optimization. The coefficients m and j are cally based material models tend to make diffi-
cally by finding a sufficient number of dimen-
used to define arguments for two Liapunov ma- cult demands in terms of model parameters and
sionless system variables that describe its evolu-
terials stability criteria. The argument for me- the experimental data necessary for initializing
tion by trial and error. This is the concept of the
chanical stability is the strain-rate sensitivity pa- the model, and the results are often no better
DMM technique, and it is possible to determine
rameter m and the temperature coefficient of the than an empirical model.
whether the deforming system, which is con-
flow stress j leads to another coefficient s that Multidisciplinary Process Design and Opti-
trolled by selected process mechanisms, is ap-
is an entropy production rate parameter, which mization. An understanding of the interaction
proaching steady state. The system parameters
serves as another argument for deciding struc- between materials properties and processing
tural stability. It is important to realize that all of that control mechanical stability and microstruc- variables is important in controlling the micro-
the necessary information for designing a robust tural stability during forced dissipative flow are structures of forged parts. Multidisciplinary Pro-
forging process is contained in the constitutive the strain-rate sensitivity parameter m and an en- cess Design and Optimization is a tool for opti-
equation for the workpiece material. tropy production rate parameter s. Both of these mally controlling the deformation temperature
A stability map can be automatically created parameters are derived from the constitutive and deformation rate to produce a desirable set
for any effective strain level using an appropriate equations for the workpiece material that are used of microstructures and properties in a finished
software application to aid the process design for process simulation. forging. The workpiece stability maps of the
team in deciding the forging process parameters The mechanical and structural stability cri- DMM method can be used to define the critical
previously listed. Because the materials stability teria used for creating the intrinsic workability temperature and strain-rate range for evolving
map is independent of process equipment and maps are summarized subsequently. The the desired set of microstructures, which subse-
tooling parameters, it is representative of the so- Liapunov Function for mechanical stability is quently respond to heat treatment, to achieve the
called intrinsic workability of the workpiece ma- Lm m(ln e ) 0, and the mechanical stability objective service properties. Multidisciplinary
m
terial. The rate of change of the effective flow criterion is Lm = < 0 . This condition ensures process design and optimization can also estab-
ln e
lish the other key process temperatures parame- The use of optimization concepts to obtain Optimization of microstructures is one key
ters discussed previously. Strain in the critical valid and realistic design solutions applies to a application of state-space control theory. The fol-
regions of the part is optimized through die and variety of process analysis methods that are lowing section describes this concept in more
preform design, as discussed in more detail in available to process design teams for almost any detail, and two examples are given. One example
subsequent chapters. materials process design problem. When shape describes a state-space model for dynamic recrys-
The goal of process design optimization is to or volume is a relevant factor in optimization, tallization of steel, and the second example de-
create a robust process that meets product remesh-based numerical techniques can be used scribes trajectory optimization techniques of the
quirements and satisfies customer expectations. to analyze geometrical relationships between extrusion process (Ref 20). The extrusion process
Meeting these objectives is enhanced through the initial and final shapes. The simplified finite is an example of a steady state process that can be
optimization of important process interactions element method (SFEM), for example, corre- effectively modeled using this method of analy-
(e.g., see Chapter 14, Process Design in Im- sponds to analysis methods such as the Sachs sis. For extrusion, material flow occurs predomi-
pression Die Forging, Fig. 3, for interactions of (Slab) and upper bound (UB) methods with re- nantly along an axial direction, but high interface
the forging process). Important process variables spect to computation time and the analysis of a friction can affect the stresses and can cause non-
include the type of forge press, the flow stress, typical step of the forming process. Optimization homogeneous deformation in the radial direction.
the interface friction, and the part geometry. The of forging shape and properties may also involve In this class of problem, microstructure evolution
size and geometry of the part determine the factors associated with subsequent operations is influenced by the flow pattern, the rate of en-
forming load and energy, and the flow stress in- such as machining and heat treatment. The geom- ergy input, and the ability of the workpiece mate-
creases with increasing deformation rate and de- etry of a near-net-shape forging geometry is in- rial to dissipate energy under stable conditions.
creasing temperature. The magnitudes of these fluenced by several practical considerations.
variations depend on the specific material under For example, one optimization factor may be
consideration, and the frictional conditions dete- the uniformity of residual stress distributions. Optimization of Microstructure
riorate with die chilling. Another type of optimization factor may be die Development Using Control Theory
Temperature variations during deformation life, which is controlled by the stresses within (Ref 21)
processing are largely influenced by a number of the die impression and on the die outer shapes.
variables for a given workpiece initial tempera- For reasons such as these, designers do not usu- As shown in Figure 2, the design methodol-
ture. The key process variables include: ally know a priori if some geometric feature is ogy is separated into a microstructure optimiza-
sufficiently significant to rule out the use of tion stage (or problem) and a process optimiza-
Surface area of contact between the dies and
simplified process models, such as state-space tion stage (or problem). Goals of the first stage
the workpiece
(lumped-parameter) models. are to achieve enhanced workability and to obtain
Workpiece thickness and volume
Nonetheless, modeling of dynamical systems prescribed microstructural parameters. In this
Die temperature
in state-space form provides a natural frame- investigation, an attempt has been made to con-
Amount of heat generated by deformation
work for describing the time evolution of the trol microstructural parameters such as final grain
and friction
states and outputs of a system under the influ- size (average) and volume fraction recrystallized
Contact time under pressure
ence of multiple, time-varying inputs. This pro- during deformation. In the second stage, a pri-
The strain rate and temperatures, which are both vides an effective framework for diverse op- mary goal is to achieve the thermomechanical
extremely critical when processing high-temper- timization strategies over the entire materials conditions required by the first stage for prede-
ature alloys, are influenced by the characteristics processing design problem, not just some parts termined regions of the deforming workpiece. In
of the forging press used for processing. of the process at the expense of the whole man- Figure 2, desired microstructural features, such
ufacturing enterprise. This approach to opti- as volume fraction recrystallized and grain size,
mized process design is a stark contrast to the serve as inputs to the microstructure optimiza-
Material Trajectory Optimization traditional philosophy that a more predictive cation problem. Trajectories of strain, strain rate,
pability for materials process models leads to and temperature are the outputs, which depend
Optimization concepts are based on the notion better design solutions. Some designers run de- on the details of the microstructure development
that ideal deformation conditions can be defined tailed finite element simulations looking for model and the criterion used for optimization.
for a material and/or a process. In the context of trends in the behaviors of certain key process Next, these trajectories become the inputs to
stable dissipation under forced conditions (de- field parameters such as the material flow pattern the process optimization problem. The outputs
scribed by DMM charts), dynamic state-space and how they are affected by different values of of the process optimization are the die shape,
models of microstructure evolution can be de- die velocity, die temperature, and workpiece ram velocity, and billet temperature for the case
veloped from control-theory concepts to identify temperature. This ad hoc approach to materials of extrusion.
precise material trajectories for achieving the process design is frequently but mistakenly The design approach requires three basic com-
design objectives of finished shapes. Analytic called process optimization. It is very labor in- ponents for defining and setting up the opti-
models of microstructure evolution are applied tense and costly, and the ad hoc approach may mization problem: a dynamical system model,
in the stable regions, and the variable nature of not be effective in finding an optimal solution. physical constraints, and an optimality criterion.
materials behavior is managed by designing the
materials process to operate primarily in the re-
gions of materials stability for a process-param-
eter space. The deleterious effects from erratic
behavior are dampened in the regions of materi-
als stability, where the apparent activation ener-
gies that correspond to beneficial energy dissi-
pation mechanisms are relatively low compared
with materials failure mechanisms. In this ap-
proach, the analytic microstructure evolution
models are only reliable when applied in the sta-
ble regions, where microstructure and mechani-
cal property parameter variance is not sensitive
to material path trajectory. Fig. 2 Schematic representation of the two-stage approach
In metal forming, the system models of interest In Eq 6 and 7, t is time; x(t) is a vector of state
are material behavior and deformation process variables; u is the system control input; tf is the
models; constraints include the hot workability duration of the process; h(x(tf)) is the penalty as-
of the work piece and the limitations of the form- sociated with violating the desired final state;
ing equipment. Optimality criteria can be chosen g(x(t),u(t)) is the integrand of the penalty asso-
to achieve a particular set of final microstruc- ciated with deviating from desired state and con-
tural features, to regulate temperature, and to trol input trajectories; f[x(t), u(t)] is a vector
maximize deformation speeds. function that describes the process dynamics;
Materials Behavior and Process Modeling. and xo is the initial state condition. A graphical
Materials behavior models that describe the ki- description of the optimal control problem rep-
netics of primary metallurgical mechanisms such resented by Eq 6 and 7 is given in Fig. 3 for a
as dynamic recovery, dynamic recrystallization, case of a system with one input and one state.
and grain growth during hot working are required Suppose that an optimality criterion of the type
for analysis and optimization of materials sys- given in Eq 6 has been defined and that several
tem dynamics. As an illustration, consider a case possible input trajectories have been evaluated
of dynamic recrystallization for which a possible according to that optimality criterion. Figure
state-space model is: 3(a) shows several of the infinite number of tra-
jectories that the system input can follow. The
D f1 (T , e , D) corresponding trajectories of the state variable
(called trial trajectories) are given in Fig. 3(b).
c = f2 (T , e, D, c)

e
Figure 3(c) gives the value of the optimality cri-
u
terion that corresponds to each of these trial

T hse / r C p
(Eq 5) trajectories as a function of trial index, i. The
objective is to find the input trajectory that, to-
The current state at any point in a deforming gether with the corresponding trajectory of the
body is described by the state variables grain state variable, will give a minimum value of J,
size D; volume fraction recrystallized c; accu- labeled as Optimal, J* in Fig. 3(c). It is im-
mulated strain e; and current temperature T. The portant to note that minimization of an optimal-
time rates of change of these variables, i.e., D, c, ity criterion implies that the system has been
e, and T, are functions of the current state vari- optimized with respect to that particular crite-
ables and input variable, which is strain rate. In rion, and that whether the original design objec-
Eq 5, f1 and f2 are known functions; u is the sys- tives underlying the formulation of the optimal-
tem input, which is the strain rate in the present ity criterion have been met is a different issue.
case; h is a coefficient that determines how Optimality Criteria for Microstructure
much of the mechanical work is converted into Development. Careful selection of optimality
heat and contributes to the increase in tempera- criteria is critical for finding the most appropriate
ture of the billet; s is flow stress; and the prod- design solutions. In the control of microstructure
uct rCp is the heat capacity (r is density and Cp development during hot metal deformation, com- Fig. 3 A one-input, one-state example of the optimal
mon design criteria include producing specified control problem. (a) Several possible input tra-
is the specific heat) of the material. jectories. (b) Corresponding state trajectories. (c) Corre-
In addition to dynamic system models, the microstructural features and gradients of micro- sponding values of the optimality criterion
formulation of an optimal control problem re- structure within a specified variance on a repeat-
quires a statement of physical constraints and able basis. These optimality criteria and others
specification of an optimality criterion for pro- can usually be formulated as functions to be min- Table 1 shows some examples of typical opti-
ducing the desired hot-worked microstructural imized, and are often lumped together into a sin- mality criteria for microstructure development
characteristics. The limiting process conditions gle scalar optimality criterion J in the form: during hot metal deformation. Both final value
for acceptable hot workability are important ma- and trajectory specifications are shown. The gen-
terials behavior constraints in the first stage of J J 1F J F2 . . . J FNF J T1 J T2 eral formulation of this approach allows new
the control strategy. The ranges of temperature . . . J TNT (Eq 8) terms to be defined according to the specific
and strain rate over which the material exhibits a needs of each design problem. The quantities
table processing window have to be determined. where the superscripts F and T refer to require- fx(x,a) and fx(x,a,b) in Table 1 are penalty func-
This defines a region for any particular thermo- ments on desired final states and trajectories, tions that can be used to constrain optimized de-
mechanical trajectory that may be determined respectively. In the case where it is desired that sign solutions within acceptable process param-
using the prescribed optimality criterion. microstructure feature x achieve a value xdes at eter ranges imposed by material workability or
Formulation of the Optimal Control Prob- the termination of the deformation process, the equipment limitations. These functions evaluate
lem. The design problem is formulated into an corresponding term in J often has the form: to virtually zero for values of x in the acceptable
open-loop optimal control problem (Ref 22) that range and attain very high values when x is out-
can be formally stated as follows. Find input J1F b1(x(tf) xdes)2 (Eq 9) side that range. Scalars a and b define the ac-
variables u to minimize the optimality criterion: ceptable ranges for process parameters, such as
where b1 is a weight factor. This type of func- temperature or strain rate. An example of a
tf tion can also be used to include certain fixed penalty function is shown in Fig. 4.
J = h[ x (tf )] + g[ x(t ), u(t )]dt
t =0
(Eq 6) process parameters and other final values for
nonmicrostructural quantities such as strain and
The weight factors b, serve three purposes.
First, they are used to scale the terms in J so that
temperature in optimization calculations. The they have comparable influence in the overall
with the constraint that the system state obeys terms J Tj in the optimality criterion define re- optimality criterion. Second, they are increased
the state-space model: quirements on the desired state and control input for certain terms according to their relative im-
trajectories during the forming process and have portance to achieve overall design requirements.
x(t) [x(t),u(t)], x(0) xo (Eq 7) integral forms. Third, they may be adjusted in order to avoid
possible conflicts in design requirements and D and c and process parameters e , T, and e. The strate a systematic approach applicable to any
obtain a satisfactory compromise solution. volume fraction recrystallized c is zero until a problem, not to find a solution for plain carbon
The optimality criterion J, which is to be min- critical amount of strain ec has been imposed. steel only. It should be clear that, for many ma-
imized in order to determine, e, e , and T, can Beyond this critical strain, the kinetics of recrys- terials, the dependence of the size of the forming
incorporate a number of physically realistic re- tallization at any temperature T is characterized grains on the temperature, strain rate, and strain
quirements. For the specific problem of hot metal by e0.5, the strain required to reach 50% recrys- may be considerably more complex. In fact, the
deformation, one possible optimality criterion is: tallization. The time derivative of the fraction re- coupling among these parameters combined
crystallized can be approximated by applying with multiple constraints on temperature, strain
J = b1 ( D(tf ) - Ddes )2 + b 2 ( c(tf ) - c des )2 the chain rule of differentiation to the equation rate, and strain can easily eliminate straightfor-
tf for c in Table 2 with some simplifying assump- ward solution approaches, while the state-space

+ { f2w [e (t ), e min , e max ]
0
tions. The result is shown in Table 3.
If the desired objective is to achieve a speci-
trajectory optimization approach can reliably
solve the problem.
+ f1w [T (t ), Tmin , Tmax ]}dt (Eq 10) fied grain size, it is possible to obtain the desired The equation for the rate of change of tem-
e from the fourth equation in Table 2, assuming perature due to deformation given in Table 3

where D is the average recrystallized grain size; T is fixed. Because of the dependence of T on e states that a fraction h of the mechanical work is
Ddes is the desired final grain size; cdes is the de- (Table 3), this assumption is not valid, although converted to heat and increases the temperature.
sired final volume fraction recrystallized; e min it may be adequate in some applications. The The microstructural state of the material is given
and e max are the minimum and maximum limits purpose of this example, however, is to demon- by the state vector x [c, T, e]T, which evolves
of strain rate, respectively; Tmin and Tmax are the
minimum and maximum limits of temperature Table 1 Examples of typical terms for the optimality criterion for microstructure
for acceptable workability, respectively; e (t) is development during hot metal deformation
the nominal strain rate; and T(t) is the nominal
temperature. The functions f1 and f2 ensure that Design objective Term in the optimality criterion
the nominal strain rate and temperature during Achieve final average grain size Ddes JFi = bi (D (tf) Ddes)2
deformation will be kept within the prescribed
limits of the safe processing window.
Finding an analytical solution to the problem Achieve final strain of edes JFi = bi (e (tf) edes)2
given in Eq 10 is highly unlikely due to the com-
plexity of the resulting functional form. However, tf
it is possible to formulate a numerical algorithm Maintain strain rate between emin and emax because of workability considerations JT =
j b j (t ) f (e, e min , e max ) dt
that can yield a practical solution (Ref 22). 0
tf
Examples Limit deformation heating; initial temperature is T0

j
J T = b j (t ) (T - T0 )2 dt
0
The formulation of state-space equations is tf
described in the previous section for microstruc-
b (t ) f (e, e
T
Keep strain rate under emax because of equipment limitations Jj = j max ) dt
tural evolution during hot working. The follow- 0
ing two examples use trajectory optimization tf
with mathematical relationships for dynamic re-
crystallization of plain carbon steel. These equa-
Maintain temperature between Tmin and Tmax because of workability considerations JiT = b (t ) f (T , T
0
j min , Tmax ) dt
tions are valid within the stable processing win-
tf
dow determined by the DMM method.
Example 1: State-Space Models for Dy- Limit energy consumption; s (t) e (t) is a measure of power, and s is flow stress J jT = b (t ) s(t ) e (t ) dt
j
namic Crystallization of Steel. An empirical 0
model developed by Yada (Ref 23) and critically

reviewed and assessed by Devadas et al., (Ref Table 2 Yada equations for the dynamic recrystallization of steel
24) shown in Table 2, was used to describe the Parameter Equation
relationship between microstructural parameters
Volume fraction recrystallized c 1 exp (ln(2) ((e ec)/e0.5)2)
Critical strain ec 4.76 104 e8000/T
Plastic strain for 50% volume fraction recrystallization e0.5 1.144 103 D0.28
0 e 0.05 e6420/T
Average recrystallized grain size D 22,600 e 0.27 e0.27(Q/RT)
Activation enery and gas constant Q 267 kJ/mol, R 8.314 103 kJ/mol K
Table 3 Equations used in the state-space model of microstructural evolution

Parameter Equation
c de 2 ln 2
Time derivative of volume fraction recrystallized c = ( e - e c )(1 - c ) e
e dt ( e ) 2
0.5
h
T = s ( e, e , T ) e
r Cp
Time derivative of temperature
Flow stress (kPa) s sinh1 [(e/A)1/n eQ/nRT] / 0.0115 103
ln A (e) 13.92 9.023 / e0.502
n (e) 0.97 3.787 / e0.368
Fig. 4 A typical penalty function, f(x,a,b), for con-
Activation energy and gas constant Q (e) 125 133.3 / e0.393, R 8.314 103
straining process parameters between a and b
according to the equations given in Table 3 for

time derivatives of c and T and the obvious rela-
tionship between the input strain rate u and the
evolution of strain. Since the grain size does not
influence the other state variables c, e, and T, it
is treated as an output of the dynamical system
and not included as one of the state variables.
Example 2: Application of a Two-Stage
Approach for Optimization of Grain Size in
Extruded Steel. Trajectory optimization has
been applied to the design of extrusion processes
(Ref 20). Optimized material trajectories in
strain, strain-rate, and temperature space were
determined for extruding a plain carbon steel
into a 26 mm diameter bar shape with 26 mm
grain size. It was a two-stage approach. In the
first stage, the optimized strain, strain-rate, and
temperature solutions were obtained for the grain
size objective. In the second stage, the optimized
material trajectories were used to directly calcu-
late the corresponding extrusion die throat
geometry for achieving the optimized strain and
strain-rate trajectories.
Stage 1: Optimizing the Microstructural Tra-
jectories. For the case studied here, the optimal-
ity criterion was chosen so as to attain a given
final strain of 2, while ensuring that the recrys-
tallized grain size was kept at a desired value of
26 mm. The average grain size of the raw stock
prior to extrusion was 120 mm. The optimality
criterion chosen was:
tf

J = 10(e(tf ) - 2.0)2 + ( D(t ) - 26)2 dt
0
(Eq 11)
where a desired final strain of 2, with a weighing

factor of 10, and a desired grain size of 26 mm
have been specified. The results of the first stage
of the microstructural optimization problem are
shown in Fig. 5(a). Starting at an initial tempera-
ture of 1273 K, the temperature of the material
increases approximately to 1295 K for deforma-
tion to a strain of 2.0. The strain rate is initially
slightly below 1.0 s1 and increases gradually to a
little above 1.0 s1. The recrystallized grain size,
which is initially 120 mm, decreases to 26 mm
beyond the critical strain of approximately 0.25.
Subsequently, since the recrystallized grain size
depends both on T and e , the simultaneous in-
crease in both of these variables maintains the
grain size constant. The results of two additional
optimization runs to achieve grain sizes of 30 and
15 mm are shown in Fig. 5(b) and (c), respec-
tively. The initial billet temperature was 1273 K Fig. 5 Trajectories of strain, strain rate, temperature, and grain size for achieving the desired final grain size of (a) 26 mm,
(b) 30 mm, and (c) 15 mm
for the second case and 1223 K for the third case.
Stage 2: Optimizing the Process Parameters.
It is not physically possible to ensure that all the rameters that will attempt to achieve the desired sired strain and strain rate profiles at the center-
points in the deforming piece will undergo the trajectories at predetermined points in the mate- line of the piece. Given that r0 is the die entrance
strain, strain rate, and temperature trajectories rial piece. In such an approach, each evaluation radius (equal to the billet radius), L is the die
obtained in stage 1. However, process parame- of the objective function for the optimization length, and e(t) is the required strain trajectory,
ters such as die geometry, ram velocity, and bil- process usually requires a detailed analysis of the ram velocity is:
let temperature can be designed to ensure that the deformation process by the finite element
selected regions of the material will experience method or some other technique. L
trajectories that approximate those designed. It In the case of round-to-round extrusion, it is Vram = tf
e
e( t )
is feasible to formulate a second optimization possible to analytically calculate the die profile dt
problem that determines values for process pa- and ram velocity necessary for achieving the de- t =0 (Eq 12)
2. H. Cheng, R.V. Grandhi, and J.C. Malas, 14. J.C. Malas and V. Seetharama, Use of
Design of Optimal Process Parameters for Material Behavior Models in the Develop-
Non-Isothermal Forging, Int. J. Numer. ment of Process and Control, J. Met., Vol
Methods Eng., Vol 37, 1994, p 155177 44, 1992, p 813
3. R.E. Skelton, Dynamic Systems Control, 15. I. Prigogine, From Being to Becoming: Time
John Wiley and Sons, Inc., 1988 Complexity in the Physical Sciences, W.H.
4. K. Chung and O. Richmond, A Deformation Freeman and Company, New York, 1980
Theory of Plasticity Based Upon Minimum 16. T. Altan, S. Oh, and H. Gegel, Metal Form-
Work Paths, Int. J. Plast., Vol 9, 1993, p ing: Fundamentals and Applications, ASM
907920 International, 1983
5. K. Chung and O. Richmond, The Mechanics 17. H.L. Gegel, Synthesis of Atomistics and
of Ideal Forming, Trans. ASME, Vol 61, Continuum Mechanics, Computer Simula-
1994, p 176181 tions in Materials Science, R.J. Arsenault,
6. W.G. Frazier et al., Modeling and Simula- J.R. Beeler, Jr., and D.M. Esterling, Ed.,
Fig. 6 Optimal die profile for achieving the final grain tion of Metal Forming Equipment, J. Mater. ASM International, 1988, p 291344
size of 26, 30, and 15 mm
Eng. Perform., Vol 6 (No. 2), 1997, p 153 18. P. Dadras and J.F. Thomas, Jr., Metallurgi-
156 cal Transactions, Characterization and
The die shape can be described by the radius r 7. R.H. Wagoner and J.-L. Chenot, Metal Modeling for Forging Deformation of
and axial distance (die throat length) y, where Forming Analysis, Cambridge University Ti-6242, Metall. Trans. A, Vol 12A, 1981,
r(t) r0ee(t)/2 and Press, Cambridge, UK, 2001, p 206255 p 1867
8. Y.V.R.K. Prasad and S. Sasidhara, Hot 19. J.F. Thomas, Jr. and R. Srinivasan, Constitu-
t
Working Guide: A Compendium of Process- tive Equations for High-Temperature Defor-

y(t ) = Vram e e ( t ) dt
0 (Eq 13)
ing Maps, ASM International, 1997 mation, Computer Simulations in Materials
9. Y.V.R.K. Prasad, et al., Modeling of Science, R.J. Arsenault, J.R. Beeler, Jr. and
Dynamic Material Behavior in Hot Defor- D.M. Esterling, Ed., ASM International,
Figure 6 gives the optimal die profile for mation: Forging of Ti-6242, Metall. Trans. 1988, p 269290
achieving final grain sizes of 26, 30, and 15 mm A, Vol 15A, 1984, p 18831891 20. W.G. Frazier et al., Application of Control
obtained using this approach. The optimum ram 10. H.L. Gegel, Synthesis of Atomistics and Theory Principles to the Optimization of
velocities for achieving these grain sizes were Continuum Mechanics, Computer Simula- Grain Size During Hot Extrusion, J.
8.43, 5, and 25.1 mm/s, respectively. Note that tion in Materials Science, R.J. Arsenault Mater. Sci. Technol., Vol 14, 1998, p
the die shape is almost the same for the three op- et al., Ed., ASM International, 1988, p 291 2531
timization cases. This means that the same die 344 21. J.C. Malas et al., Optimization of Micro-
can be used to achieve the different recrystal- 11. Y.V.R.K. Prasad, et al., Titanium Alloy structure Development during Hot Working
lized grain sizes by changing only the velocity Processing, Adv. Mater. Process., Vol 157 using Control Theory, Metall. Mater. Trans.
of the extrusion press. (No. 6), 2000, p 8589 A, Vol 28A, Sept 1997, p 19211930
12. D. Cebon and M.F. Ashby, Information Sys- 22. D.E. Kirk, Optimal Control Theory: An
REFERENCES tems for Material and Process Selection, Introduction, Prentice-Hall, 1970, p 2946,
Adv. Mater. Process., Vol 157 (No. 6), 2000, 184309
1. J.C. Malas and W.G. Frazier, Optimal p 4448 23. H. Yada, Proc. Int. Symp. Accelerated
Design of Thermomechanical Processes 13. J.C. Malas, A Thermodynamic and Con- Cooling of Rolled Steel, Conf of Metallur-
Using Ideal Forming Concepts, The Integra- tinuum Approach to the Design and Control gists, CIM, Pergamon Press, 1987, p
tion of Material, Processes and Product of Precision Forging Processes, masters 105120
Design, Zabaras et al., Ed., A.A. Balkema, thesis, Wright State University, Dayton, OH, 24. C. Devadas et al., Metall. Trans A, Vol 22A,
1999, p 229236 1985 1991, p 335349
Chapter 24
Application of Multidisciplinary
Optimization (MDO) Techniques to the
Manufacturing of Aircraft
Engine Components
Shesh K. Srivatsa, GE Aircraft Engines
MANUFACTURING INDUSTRIES face ment, and machining. The individual steps of the analysis and reviewed upfront with the engine
great challenges in the current dynamic and glob- disk design process, broken down into the me- manufacturer prior to committing to production.
ally competitive market environment to develop chanical design and the process/manufacturing A detailed simulation of the various steps in the
new products and processes to satisfy market de- aspect, are shown in Fig. 1. The thermomechan- manufacturing process is conducted to deter-
mands in a timely, efficient, and cost-effective ical design, for example, starts with a simple mine both residual stresses and final distortions
manner. Both original equipment manufacturers analysis to obtain a rough thickness distribution of the finished part after machining operations.
(OEMs) and their supplier organizations face of the disk. As knowledge about the design in- These residual stresses, in turn, are used in the
several design and manufacturing challenges that creases, more complex analysis models are cre- subsequent life prediction of the part.
cannot be met by traditional methods. There is a ated up to a full three-dimensional (3D) finite- After the initial die design is completed, the
strong need to dramatically reduce product de- element analysis with tens of thousands of part is forged, heat treated, inspected, and ma-
velopment cycles and product design and manu- elements. The objective during the mechanical chined. If test or analysis results show that the
facturing costs. New methodologies such as mul- design phase is the determination of the final disk disk does not meet its life requirement, or ma-
tidisciplinary optimization (MDO) are needed to shape, as early as possible in the design timeline chining distortion exceeds the allowable limit, or
overcome major obstacles to producing good in order to be able to release the forgings, which forging and heat treatment do not produce the
product and process designs quickly. tend to require a long lead time. A shape is to be desired material microstructure and mechanical
This chapter discusses how MDO technolo- determined that meets mission requirements at properties, or the press load required is beyond
gies can help reduce cost, cycle time, and deliv- minimum weight and/or minimum cost subject to the capacity of the forging press, or the material
ery time. The focus in this chapter is on process other technical and business constraints. does not fill the dies completely and without de-
optimization of aircraft engine components, Historically, a sequential and a largely empir- fects, or any other requirement is not fulfilled,
specifically those manufactured by bulk form- ical process had been employed where the shape then part of or all of the process has to be re-
ing processes. However, the MDO methods are of the disk as created by the engine manufac- peated. If that is not possible, the finished disk
equally applicable to other components in other turer was handed to the forging supplier. The shape has to be changed and the mechanical de-
industries. Multidisciplinary optimization tech- engine manufacturer determined the final shape signat least in parthas to be revised. The
nologies are becoming key design techniques of the disk, added a protective layer around it same applies if the design does not meet life re-
for the new-economy businesses to ensure de- (resulting in the so-called ship shape), and quirements. This process of iterating between
livery speed and reliability. This has resulted in passed it to the forging supplier. The protective design and manufacturing to ensure that the parts
productivity enhancements by increasing design layer is added to account for variation and un- meet all their design requirements is a very ex-
speed, minimizing human error, and improving certainty inherent in any manufacturing process pensive and time-consuming process.
designs, a critical requirement. and to ensure that the geometry will clean up to A basic requirement of any forging shape and
the final required shape. The supplier then de- process design methodology is an accurate de-
signed a forging die based on empirical and test scription of the deformation mechanics involved
Turbine Disk Design and results and prior experience to produce a disk to in the process. The deformation analysis tech-
Manufacturing fulfill the engine manufacturers requirements. nology has developed over the last couple of
Starting in the 1980s, the largely empirical pro- decades to a mature level and meets the basic
The design and manufacturing of gas turbine cess of die design was gradually replaced by ma- design requirements. Process analysis for cost
engine components is a complicated, highly cou- terial modeling methods and finite-element effectiveness and rapid product development con-
pled, and iterative multidisciplinary process. For modeling (FEM) techniques. Two decades later, tinues to advance in sophistication through the
life-critical rotating parts and disks, the focus of at least in the aerospace and automotive indus- use of modern design methods and tools. The
this chapter, it involves both thermomechanical tries, these analysis techniques for metalworking role of computers in process development con-
design for life prediction and manufacturing processes are accepted, routine practices. Die tinues to expand through applications of data
process simulation including forging, heat treat- designs are usually verified with finite-element acquisition and management, modeling and sim-
Chapter 24: Application of MDO Techniques to the Manufacturing of Aircraft Engine Components / 347
Fig. 1 Key activities in producing a forged component demonstrating the application of geometry and finite-element-based multidisciplinary optimization for complex product and
process development
ulation, and (model-based) process control. The the individual steps in this process are not fully cal operations that have an interacting effect on
material modeling methods and finite-element utilized. Engineering analysis and process simu- the finished engine component. Many metal al-
modeling techniques developed in the 1980s and lation tools have to be integrated into an opti- loys are currently employed in the manufacture
1990s for metalworking processes are accepted, mization environment. An integrated procedure of automotive, aerospace, and other hardware
routine practices in well-managed part supplier that addresses both mechanical design and man- components. The high cost of manufacturing
companies. However, in spite of the increased ufacturing processes is almost mandatory to re- critical components can be significantly reduced
use of modeling technology, actual process deduce design cycle time and eliminate wasteful it- with the development of mathematically and
sign variables (number of dies, shapes of dies, erations because of the multidisciplinary nature physically sound computational methodologies
etc.) still require a significant amount of expert of the process and the prohibitive costs involved for process design, optimization, and control.
knowledge. Design rules are secrets closely if changes become necessary once actual parts Forging is one of the most common processes
guarded by the forgers and based to a consider- are being produced. By automating the design for manufacturing rotating disks in gas turbine
able extent on previous experience. Simulation and analysis tools involved and also incorporat- engines that can sustain the high stresses in an air-
methods have just moved some of the trial-and- ing optimization techniques, it is possible to sig- craft engine or power turbine. As a result, it is one
error design methodologies from the shop floor nificantly reduce the lead time and engineering of the most critical procedures in the turbine disk
into the office. To overcome this dependency on cost as well as to achieve an optimal design. manufacturing process. During the forging step, a
a small group of industry experts and to reach Despite the emergence of new materials and block of metal, which may be cylindrical or an-
designs that are more robust and ultimately less processes, bulk forming process continue to be other shape, is transformed to a shape close to that
costly, optimization-based design techniques need critical to the production of parts for high- of a disk in one or several operations. The objec-
to be employed. performance equipment. Many of the new alloys tive during the simulation of the forging process is
In addition, the process just described results for critical applications are hard to work and the determination of the die shape on the one hand
in merely an acceptable design, that is, one that require sophisticated workability and design and of an optimal forging process on the other that
meets the specifications, but not an optimal one methods for successful processing. Forged com- ensures proper die filling without compromising
with respect to user-defined criteria such as ponents are the backbone of the modern gas tur- mechanical properties of the disk through the vi-
weight, cost, and performance. In todays cost- bine engine; the entire rotor structure is com- olation of stress, strain, strain rate, or temperature
competitive environment, it is imperative that posed of disks, spools, shafts, blisks, and seals limits. The subsequent heat treatment process is
optimal solutions and not just acceptable solu- that are machined from cast and wrought or designed to generate acceptable mechanical prop-
tions be obtained and be obtained rapidly. powder materials. These components experience erties in the forged disk. Following heat treat-
Manual optimization based on experience is high centrifugal and torsional loads at elevated ment, the disk is machined to its final shape for
time consuming, expensive, and tedious due to temperatures during service and constitute the assembly. A simulation of the machining process
the multidisciplinary nature of the task and most demanding structural applications for results in the final disk shape with accurate resid-
because of the complexity of computer-aided todays advanced aerospace materials. These ual stresses and distortions. Each step has special
design (CAD) and computer-aided engineering components represent more than 40% of the process requirements involving different physical
(CAE) tools. Accurate simulation tools for indi- weight and cost of a gas turbine engine and un- phenomena. All the conditions must be satisfied
vidual stages of the disk design and manufactur- dergo a sophisticated series of manufacturing simultaneously for each step. Therefore, the disk
ing processes are widely available and routinely operations. Of these manufacturing processes, manufacturing process is highly complicated and
used. However, opportunities for optimization of forging, heat treatment, and machining are criti- multidisciplinary.
Die designers in todays cost-competitive en- freeing engineers to do other engineering duties faction of other constraints, the weight saving is
vironment have to produce a design as near net is now a priority task for the parts supplier indus- 50 to 80%. Clearly, this is a significant material
shape as possible, that is, a design that requires tries. Proven software optimization tools, such savings over the manually optimized forging
the least amount of forging material and yet as iSIGHT (Engineous Software, Inc.) described designs for jet engine disks. Die designers are
simultaneously meets all of the design specifi- in this chapter, can drive simulation codes from now able to explore, with the help of a fully au-
cations. For aerospace components made from multiple disciplines toward achieving superior tomated software system, a bigger design space
high-cost superalloys, material cost is the most designs in less time. The iSIGHT system is a and find a feasible design of the least forging
significant cost driver and is approximately 50% generic software shell that can be applied to the weight. In addition, a shape optimized with
of the forging cost. Therefore, weight reduction design of many different products and processes. iSIGHT can have better properties than the man-
is the most effective cost-reduction strategy. Also, iSIGHT is a comprehensive application ually designed shape, thus allowing for addi-
Significant amounts of extra material are in- with a user-friendly graphical user interface tional weight reductions for equivalent mechan-
cluded in a forged part, which is subsequently (GUI) that allows end users to easily use it. ical performance as the manual design.
machined to get the final disk shape. This is the This chapter addresses one of the important By utilizing this technology, the quality of
preferred way to ensure good materials proper- aspects of the turbine disk forging process turbine disks is improved, the design process
ties and avoid having to scrap the part due to the design of die geometry to achieve near-net- variation is minimized, and the manufacturing
process variability. However, decreasing the ma- shape forging. Using numerical optimization cost is reduced. Several case histories and differ-
terial consumption by even a modest percentage tools, the capability to achieve significant mate- ent ways of using the optimization tools show
can drastically reduce the manufacturing cost of rial savings on top of manually optimized forg- the benefits of using MDO for solving product
turbine disks. Previous attempts to manually ing designs for jet engine disks has been demon- and process design problems by exploring more
achieve a balance between these two conflicting strated. The design process is formulated as a design options and parameter trade-offs in a
objectives have had limited success due to the parametric geometry and high-fidelity analysis- given time than is possible by a human designer.
complexity of the problem. With the advance of based shape optimization problem. The objec- This chapter is not intended to serve as a re-
optimization, numerical simulation, and geome- tive is to minimize the forging weight by chang- view of optimization methods. There are several
try modeling, it is now possible to attack this ing the die geometry parameters. The forging such review articles, for example, Ref 1. The
difficult problem. Optimization of the forging constraints include producibility, materials prop- focus here is on the application of MDO tech-
process results in a minimum-weight, forgeable erties, geometry, and forging time conditions. niques to solve real-life manufacturing prob-
disk that meets all constraints in terms of The forging weight and cost are minimized lems. Previous work related to the topic can be
process parameters within specified processing within prescribed processing windows and forg- found in the literature. Kodiyalam, Kumar, and
windows and rules while maintaining properties. ing rules including bounds on strain, tempera- Finnigan (Ref 2) proposed a constructive solid
The optimization of the heat treatment process ture, strain rate, press capacity, dwell time, sonic geometry approach to three-dimensional struc-
reduces residual stresses while maintaining re- coverage, fillet radius, draft angles, die fill, and tural shape optimization. Hardee and coworkers
quired cooling rates through the control of sur- so forth. A generic and noninvasive procedure (Ref 3) demonstrated how to use ProEngineer as
face heat transfer coefficients. Current industry was developed to integrate CAD and CAE tools the CAD/CAE system to perform structural de-
approaches involve at least a few (25) itera- for forging design optimization. A fully auto- sign optimization based on continuous sensi-
tions of the process model to yield an acceptable mated analysis and optimization system that tivity analysis. Fourment and Chenot (Ref 4, 5)
solution. It is a challenge for a manual design works in a heterogeneous and networked com- presented their work on a metalforming prob-
process to generate a good forging design, let puting environment was built on the top of three lem, where they optimized the initial shape of
alone an optimal one. commercial software packages: DEFORM for the part and the shape of the preform tool in a
simulating metalforming process, a commercial two-step forging operation. Wright and Grandhi
CAD tool Unigraphics for defining and manipu- (Ref 6) as well as Gao and Grandhi (Ref 7) pub-
Optimization Techniques lating geometry, and iSIGHT for software inte- lished their studies on forging preform shape
gration and optimization. The analysis involves optimization. All the aforementioned research
Numerous optimization techniques have been thermal-mechanical coupling, nonlinear mate- related to forging optimization has utilized inter-
developed and demonstrated successfully on rial behavior, friction modeling, very large de- nally developed CAD/CAE systems. Thus it is
relatively simple problems with few parameters formation, time-dependency, and multiple oper- possible to develop analytical design sensitivi-
and simple analytical models. However, design ations. The iSIGHT system replaces the manual ties and solve the optimization problem effi-
optimization is not used as widely today because iterative portion of the traditional design process ciently. This chapter presents a method of in-
of many deployment difficulties, including the with an automated, computer-controlled proce- tegrating commercial CAD/CAE systems and
need to educate designers on optimization tech- dure. The software helps link all of the relevant addressing the interprocess communication prob-
nologies and the need to modify simulation codes simulation tools, then automatically changes the lem. The approach is generic and noninvasive
for optimization. Furthermore, optimization of design parameters, runs the analysis codes, as- and readily lends itself to more complicated
an individual discipline does not necessarily sesses the output against the target objectives, applications. The method is demonstrated with
produce an overall optimal design. To achieve and changes the design variables based on in- several examples.
the goal of evaluating many more complete structions from an optimization algorithm cho- Metalforming Process Optimization. Opti-
design alternatives in less time, one must ad- sen for the specific problem. mization is the process of searching for the best
dress the problem of how to perform optimiza- Through a series of iterations among the dif- solution, which includes finding the best value
tion across disciplinary, organization, and com- ferent software tools, a user-defined optimal sce- for the objective function while satisfying the
plexity boundaries. nario is established. The whole process iterates design constraints. In metalforming, the objec-
Circumstances frequently prevail in the design without human intervention until an optimal so- tives of process optimization may include one or
process for a team of engineers from diversified lution is achieved. The optimization system has more of the following:
disciplines to combine their knowledge to pro- been applied to selected disk-design problems
duce an overall optimal solution. Additionally, and achieved significant weight reductions. The Forging cost
the requirement exists to combine an understand- total forging weight was reduced by 5 to 10% Forging weight
ing of process design with the ability to use sim- over manual design optimization. If the saving Press load
ulation codes correctly and to overcome the te- in weight is expressed in terms of the excess Total number of operations
dious process of manual iteration to obtain an material added over the smallest geometrically Die cost
optimal design. Overcoming these obstacles and feasible shape to ensure forgeability and satis- Lead time
The primary objective here is to use optimiza- the time of publication, the industry had been Unigraphics: CAD code for geometry ma-
tion methods to design the necessary sequences moving toward adopting off-the-shelf commer- nipulation
of forming processes, select appropriate dies and cial software because of the high cost associated
preforms for each process, and control/design with software development and maintenance. The three software tools are briefly described in
process parameters for each process such that, The drawback of this approach is that the soft- the sections that follow.
for a given raw material with a given initial geom- ware capability becomes one of the constraints. iSIGHT for Optimization and Program
etry, one can obtain a final product with the de- For instance, analytical sensitivities are gener- Control. The commercial software package
sired microstructure and shape. Desired objec- ally not available with commercial software, and iSIGHT (Ref 9) is used as the integration and
tives may include one or more of the following thus other methods such as finite difference for optimization framework. The system provides
criteria and may represent, for example, the de- gradient evaluations have to be employed for a holistic approach to optimization by embed-
viation of the resulting material state of the final the sensitivity analysis. Software licensing agree- ding the concept of interdigitation (Ref 811),
product from a desired state for the process de- ments (multiple versus single user/computer li- which was originally developed at GE Cor-
fined by design parameters: censing costs) can prevent parallelism for gradi- porate Research and Development Center as
ent evaluations. Even when it is possible to use part of the Engineous system (Ref 10). Users
Uniform deformation in the final product
analytical sensitivities, it is complicated and en- have access to a suite of optimization tech-
Desired mechanical properties in the final
gineering labor intensive to develop these for niques including gradient-based, exploratory,
product
general shape optimization problems. Barriers and AI/expert system based algorithms within
Desired microstructure in the final product
between various engineering and manufacturing iSIGHT. These techniques may be used alone
Required shape of the final product
departments are also a factor and are often diffi- or collectively to efficiently solve complicated
Residual-stress distribution to minimize dis-
cult to break down. optimization problems. Another characteristic
tortion during manufacturing and/or during
Despite all these obstacles, though, progress of iSIGHT is its ability to nonintrusively inte-
service
toward a comprehensive optimization is apparent. grate external programs. iSIGHT provides ex-
Minimum deformation and wear of the die
cellent utilities to manage data flow between
Minimum energy or load spent in deform-
simulation codes, automate program execution,
ing the material
Minimum number of forging steps Collaborative Optimization and allow external codes to reside on their
Environment respective platforms irrespective of where
No defects such as forging laps or die non-
iSIGHT is installedan important point con-
fill
cerning leasing and maintenance cost for those
The objective is to develop a flexible design
These objectives can be satisfied by controlling software tools that may be licensed only on one
environment that can be used to design complex
design variables that can represent design of specific workstation.
products and processes to achieve affordable
the preform: design of the material state (micro- DEFORM: CAE Code for Simulating
globally optimized designs. This is achieved
structure) in the initial billet, appropriate selec- Manufacturing Process. Advanced-process
through the creation of a collaborative optimiza-
tion of the process parameters (ram speed and simulation tools are becoming more and more
tion environment (COE). The COE provides a
pressure history, operating temperature, etc.), available for all stages of the disk design and
generic framework for design automation, sys-
the initial shape of the workpiece, shape of the manufacturing process. Simulation tools such
tem integration, and design optimization. There
dies, thermal boundary conditions, and so forth. as DEFORM (Ref 12), ABAQUS (Ref 13), or
is collaboration between optimization algo-
In most industrial forming applications, the de- MARC (Ref 14) can accurately predict the
rithms, between disciplines, between computing
sired objectives indicated previously are seldom mechanical behavior and properties during the
systems, and between organizations. The key
simple enough to be achieved in a single form- manufacturing process. Therefore, these tools
concepts are:
ing operation. As a result, intermediate deforma- have become the state-of-the-art and are widely
tion or performing steps are used to efficiently Address multidisciplinary and multilevel used. In combination with numerical optimiza-
transform the initial geometry into a final shape optimization tion techniques, these tools offer the opportu-
and/or with desired materials properties. Thermal Interdigitationinnovative approach to nity to improve individual steps in the overall
processing is also used in between deformation leveraging and combining multiple opti- process (Ref 15). DEFORM was chosen as the
stages to control the microstructure or product mization techniques (Ref 811) such that tool to be applied in forging and heat treatment
quality. Various examples of optimization for- the user has a suite of optimization tools optimization procedures described in this chap-
mulations are described in this chapter. available, including gradient based and ter. The finite-element-based package DEFORM
Challenges to Forging Optimization. Opti- heuristic search techniques, genetic (adap- has been demonstrated to be able to capture the
mization of both the forging and the heat treat- tive) algorithms, and simulated annealing, physics of forging processes and predict the
ment process individually has been successful, which can be used in any combination dur- macrobehavior of the material. It is, therefore,
but the complete MDO scenario still faces a ing the optimization process widely accepted and used in the forging indus-
number of obstacles. Parametric CAD tools are Noninvasive coupling of commercially try. It provides all the forging data, such as press
not as robust for complicated geometries, espe- available CAD and CAE tools load, strain, strain rate, stress, temperature, and
cially 3D, as would be necessary in an automatic Multidisciplinary design optimization lan- so forth, which may constitute constraints and
optimization environment. The same applies to guage (MDOL) to enable formulation and objectives for the optimization.
the interface between and integration of CAD solution of complex problems Unigraphics: CAD Code for Geometry
and CAE tools. Automatic 3D meshing (hexa- Manipulation. General-purpose commercial
hedrons) capability is still somewhat limited. CAD systems have evolved considerably and are
Computational resources constitute another bot- System Components becoming mature and widely used in the manu-
tleneck; for industrial applications, fast turn- facturing industry. One of the main attractions of
The forging and heat treatment optimization
around (generally less than 8 h) is a requirement. these systems is that they house a rich collection
tools discussed here in detail involve the integra-
For a complicated structural optimization prob- of user functions or application interfaces (APIs)
tion of commercially available software to reduce
lem, the effort of getting the objective function, for geometry queries, manipulation, display,
development and maintenance cost:
constraints, and corresponding sensitivities is checking, and so forth. These utilities make the
computationally expensive. Therefore, a variety iSIGHT: optimization and program control difficult task of geometry handling easier. For
of approximation methodologies have been de- DEFORM: CAE code for simulating manu- the present application, unigraphics (UG) (Ref
veloped for reducing the computational cost. At facturing process 16) has been chosen as the CAD system, but the
method of integration outlined here is generic through interprocess communication. The inte- in parametric form to enable easy manipulation
and applicable to other CAD systems. grated system involving three software tools is and updating. After completing the problem spe-
A number of modules were developed to sup- shown in Fig. 2 cification, the user may start the iSIGHT run.
port design optimization on top of the UG Open iSIGHT uses DEFORM as its simulation code The routines within iSIGHT will automatically
API library (Ref 16), including: that supplies the objective and constraint infor- launch a noninteractive UG session and establish
mation. iSIGHT controls a number of DEFORM the interprocess connection, which is based on
Graphically select feature and sketch pa-
parameters, mainly the design variables. Expect (Ref 17), a Tcl extension. iSIGHT
rameters as design variables
Because DEFORM does not directly calculate changes the design variable values and writes
Update geometry model from an external
constraint and objective information for the opti- them into the file storing the selected geometry
process
mization procedure, a specific DEFORM post- parameters. The noninteractive UG process reads
Monitor shape change
processing function has been developed for that the files and updates the geometry models. It is
Evaluate geometry properties such as vol-
purpose. DEFORM is executed by iSIGHT fully also possible to monitor the geometry change in
ume, which is the objective function for
automatically each time iSIGHT needs the solu- real time through an interactive UG session.
disk forging optimization
tion for a new data point. The procedure is The system has been applied to the forging
Automatically convert UG geometry to the
implemented in iSIGHT in the form of three sep- shape optimization of turbine disks, where the
format accepted by the CAE code
arate tasks. The first task, preprocessing, exe- weight of the disks is minimized under maxi-
There are some geometry manipulation features cutes the DEFORM preprocessor, which reads mum press load and geometric constraints. In
built into DEFORM. However, CAD systems the respective DEFORM keyword file and cre- the present study, the forging process is modeled
come with a number of standard features such as ates the DEFORM analysis database. The pre- as a time-dependent, plastic-deforming, and
checking for geometry errors, invalid geome- processor also starts the DEFORM simulation either isothermal or nonisothermal process.
tries, use of parametric modeling, feature-based run, which is executed in batch mode. Once this Since the forging simulation is conducted in an
modeling, creating offset geometries automati- has been accomplished, the preprocessing task optimization environment, some of the process
cally (forging covers over ship or heat treat exits and passes the process ID of the DEFORM and geometry parameters are modified in each
shapes). Building these features into DEFORM batch job back to the main task. The second DEFORM run. Therefore, it is necessary to re-
involve much more effort than using an off-the- task, run control, then checks on the execution of generate the mesh and redefine the boundary
shelf CAD package. Therefore, the decision here this batch job in 30 s intervals. Once the batch conditions. Furthermore, the user needs to post-
was to demonstrate the optimization system process has terminated, the run-control task exits, process the analysis results and extract informa-
using Unigraphics. and the third task, postprocessing, is started. It tion on the optimization objective and constraint
executes the modified DEFORM postprocessor, functions. Several modules based on Expect
System Integration and Operation which determines the global response quantities have been developed that drive DEFORM to ac-
that can be used as constraints and objectives for complish the following tasks:
System Integration. A general procedure is the optimization. The DEFORM postprocessor
introduced to generically and noninvasively in- extracts the relevant outputs from the DEFORM Import geometry and then regenerate die and
tegrate commercial CAD and CAE systems to database. billet meshes
support geometry and detailed analysis-based System Application for Geometry-Based Create appropriate boundary conditions
optimization. The integration and optimization Optimization. This section describes the use of Start DEFORM simulation in batch mode
environment iSIGHT has been adopted as the the iSIGHT system for geometry-based optimiza- Monitor DEFORM runs
framework of the system. A client-server archi- tion, followed by a description of its GUI. To use Postprocess simulation results to extract
tecture is established to allow iSIGHT to drive the system for geometry-based optimization, the maximum press load, strain, temperature
the CAD system for geometry manipulation. user first opens a UG interactive session, loads and so forth
The CAE system is wrapped and called upon relevant part files, graphically selects geometry Each of these modules acts like a separate exe-
within iSIGHT to perform detailed simulation. parameters that are design variables, and chooses cutable, or simcode in iSIGHT terminology.
Data passing between the optimization environ- objects whose volumes and other geometric iSIGHT executes these simcodes in a prede-
ment and the CAD and CAE tools are handled properties are desired. The geometry is defined fined sequence, including potential looping and
branching.
A summary of the tasks to perform forging
optimization and the typical engineering hours
and skill sets needed are summarized, providing
a checklist for implementing MDO techniques
in an industrial environment:
1. Parameterize die geometry and select design
variables.
8 to 16 h
Experience with UG parametric model-
ing
Forging expertise to identify design para-
meters
2. Create a standard DEFORM keyword file as
template.
4 to 8 h
Familiarity with DEFORM
3. Formulate optimization problem, and define
forging operation sequence and control para-
meters.
8 to 16 h
Fig. 2 Integrated forging optimization system Knowledge of iSIGHT
4. Start forging optimization run. iSIGHT, the user has the option to change the 2D geometry is then incorporated into the
25 to 50 h (computer elapse/cycle time) control of the problem, switching between sin- DEFORM master file.
on typical HP or SGI workstations. The gle run and optimization, to change any of 3. iSIGHT then executes DEFORM 2D to per-
computer times refer to year 2000 state- the input variables (especially the design vari- form the forging simulation on this design.
of-the-art and are continually decreas- ables), to activate or deactivate design variables, 4. Postprocessing information is extracted from
ing with improvements in computer to change objectives and constraints, or to mod- the DEFORM 2D database, that assesses the
technology ify the optimization plan. The default for the quality of the forged part in terms of user-
Knowledge of iSIGHT forging and heat treatment optimization is defined constraints. This data is retrieved by
5. Postprocess optimization results and review. ADSModified Method of Feasible Direc- iSIGHT.
8 to 16 h tions. This has mostly produced satisfactory 5. The output of the forging process is com-
Knowledge of iSIGHT and DEFORM results so far, and it is not recommended to re- pared against requirements that have been
place it with another technique unless the user is specified by the engineer with iSIGHT. This
It is thus possible to complete one optimization familiar with the various optimization tech- gives a numerical assessment of the current
problem with about 2 to 5 days of engineering niques iSIGHT offers. The technique ADS design.
effort and within an elapse time of one week. As Sequential Linear Programming has also been 6. Now, iSIGHT again changes the dimensions
computer hardware and software technology successfully applied. The user has the option to of the top and bottom dies. The iSIGHT soft-
continually improve, this time is expected to access the control parameters for each optimiza- ware has numerous exploration methods that
drop to less than one day within a few years. tion technique, which are explained in detail in can be used to intelligently make the deci-
Graphical User Interface (GUI). A GUI has the respective manuals of the individual opti- sions on which die dimensions to change,
been developed that helps the user set up the mization packages inside iSIGHT, for example, and by how much, to improve the forging
specific optimization case as shown in Fig. 3. ADS (Ref 18). Inside the overSIGHT module of process.
Via the iSIGHT GUI, the user has complete con- iSIGHT the user has the option to attach tables
trol over the simulation of the forging or heat and/or graphs to the iSIGHT run to permit inter-
treatment process and the selection of the design active tracking of the changes in the design vari-
variables. In principle, any number of design ables, constraints, and objectives on the screen. Formulation of Forging Shape
variables can be chosen, but both in terms of run The use of the corresponding buttons in the task Optimization
time for the optimization procedure and in terms manager window is pretty straightforward.
of actual process and geometry control, about 15 The steps performed during one iteration of This section describes some methods and ex-
can be considered the upper limit. forging optimization are: amples for shape optimization of forgings. Real
All the parameters describing the process and disk shapes have been optimized, but due to the
optimization sequence are set in the description 1. The dimensions of the top and bottom dies sensitive nature of that information, only simpli-
file The GUI allows the creation of a fresh de- are sent to the UG CAD software from fied examples that demonstrate the principles
scription file, or as is more often the case, the iSIGHT, and the CAD geometry is updated. and techniques can be presented here. The vari-
user may want to modify an existing file to save The size of the billet is automatically updated ous optimization features of the methods using
time. When the file is saved, some error check- to correspond to the necessary volume to fill iSIGHT include:
ing is performed. Once the forging or heat treat- the die.
ment applicationwhich resides in its descrip- 2. The 3D axisymmetric CAD geometry is cut Utilized feature-based modeling in Unigra-
tion fileis loaded into iSIGHT, the procedure to produce the two-dimensional (2D) cross phics
is ready to run. Inside the foreSIGHT module of section that is used by DEFORM 2D. This Customized UG utilities including geome-
try error checking
Developed modules on the DEFORM side
to help with data processing
Started from most aggressive shape (geo-
metrically minimum weight) and added ma-
terial to satisfy constraints
Used modified feasible direction method
with finite-difference gradients
Conducted computations in a heterogeneous
environment over the network
Modeled forging process as time-dependent,
plastic-deforming, and isothermal/noisother-
mal, rigid dies, and 2D axisymmetric
Simulated all stages of forging process in-
cluding the operations of:
a. Transfer from furnace to dies
b. Resting on dies prior to deformation
c. Forge under strain-rate control to maxi-
mum tonnage
d. Forge under load control
e. Dwell at maximum tonnage
Incorporated:
a. Equipment constraints: press capacity,
transfer, and dwell times
b. Material-related constraints: strain, strain
rate, and temperature-processing windows
Fig. 3 Graphical user interface of iSIGHT optimization system. The tabs on the right are used to select the process
c. Geometrical constraints: forging rules
(forging or heat treatment), and the tabs at the top are used to define the process and solution-control
parameters. defining the sonic and final machine part
coverage, minimum draft angle as well are used. Additional equipment constraints can excessive stress, which leads to costly repair and
as corner and fillet radii be imposed on transfer and dwell times. even complete replacement of the die. The re-
d. Invalid geometry and topological checks Material Constraints. The forged part must quirement on die life and wear is represented as:
e. Die-nonfill check and lap-formation possess the right materials properties. Close cor-
checks relation has been observed between materials s sa (Eq 6)
properties and forging and heat treat processing
In summary, the optimization procedure ties The time spent in forging a part must be min-
conditions. Materials properties are dependent on
the complex, multidisciplinary aspects of good imalthat is, as quick as possiblewhich is an
the thermomechanical processing path character-
process design together. This shows the link be- important productivity consideration. The forg-
ized by equivalent strain rate e, equivalent strain
tween the metalworking equipment, the mate- ing time limitation is represented as:
e, and temperature T. When working with diffi-
rial system (tooling, lubricant, and workpiece),
cult-to-process alloys, as is common with aircraft t ta (Eq 7)
and the equipment controls. MDO techniques
engine materials, the thermomechanical process-
are shown to optimize the process and the prod-
ing path must be tightly controlled in order to In Eq 1 through 7 superscripts lb and ub stand
uct performance. The overall objectives and
achieve acceptable microstructures and mechani- for lower and upper bounds, respectively, and
benefits for the example of turbine-disk forg-
cal properties in the final product. The forging the subscript a represents a prescribed allowable
ings include:
can be partitioned into various zones with each value.
Forging cost reduction is key to reducing zone having its own requirement on materials Many of these design variables are dependent
cost of aircraft engines properties. For example, the bore of a disk may on each other. These dependencies are impossi-
Use of optimization tools is a key enabler in be strength limited whereas the rim may be creep ble to formulate in a closed form. This consider-
achieving this goal limited, requiring different zones of the forging to ably increases the complexity of the optimiza-
Developed an integrated system for forging have the appropriate thermomechanical process- tion process.
die-shape optimization ing path to obtain the needed mechanical proper- Implementation of Constraints. Here are
Formulated forging shape optimization ties. These thermomechanical rules or processing some additional comments on the implementa-
problem for turbine disk forging windows are material dependent. They are repre- tion of the various constraints within the opti-
Developed a generic procedure to noninva- sented in the optimization framework as max/min mization framework:
sively integrate external CAD and CAE limits on strain, strain-rate, temperature, adiabatic
Press load and strain-rate conditions are
systems for geometry and detailed analysis- heating, and postforge cooling rate and define the
implemented in the forging software, that
based optimization processing window for the material to meet mi-
is, DEFORM. The strain rate is controlled
Formulated ways to handle the optimization crostructural and property requirements.
by adjusting the forging die speed as the
and constraints for a highly nonlinear prob-
e ea
calculation proceeds so that Eq 2 is implic-
lem with a very irregular response surface
itly satisfied. The maximum strain-rate
Applied the system to both isothermal and
e ea (Eq 2) constraint is enforced throughout the forg-
nonisothermal forging shape optimization
ing process.
of turbine disks
T lba T T ub (Eq 3) The strain and temperature constraints are
Demonstrated feasibility and success of a
enforced within the sonic geometry, which
forging and heat treatment process optimiza- Geometrical Constraints. Based on experi- is the disk shape for sonic inspection, and
tion ence, die designers have developed several not within the entire forging volume. There
geometry-related forging rules. Geometric con- is no need to enforce the constraints in the
Formulation of Objectives and straints are defined to ensure adequate cover over material that is machined away.
Constraints the finished shape to ensure the forging will The die-nonfill check is also used to detect
clean up in spite of process variability (process lap formation and ensure a successful forg-
Objective and Design Variables. The opti- parameters, die stack concentricity, die wear/ ing process. The objective is to completely
mization problem is formulated to minimize the deflection, etc.) and to account for all subsequent fill the finishing dies during the last opera-
weight of the forging subject to several process manufacturing operations such as heat treatment, tion. A numerical measure of this criterion
constraints that are aligned with standard forg- machining, inspection, and joining. These are is obtained by comparing the outline of the
ing practices. The shape of the top and bottom represented as bounds on design variables, pro- desired part and that of the actual forged
dies determine the weight of the forging and are tective cover h over the finished part shape for part.
defined by a set of geometrical parameters. The sonic inspection, the extent of die-nonfill d, any Some of the constraints may not be needed
design variable set {pi, i 1, , N} is a subset lap formation, minimum fillet and corner radii, for certain forging processes and/or materi-
of geometry parameters that defines the die minimum forging thickness, and minimum draft als. For instance, there is no temperature
shape, where N is the number of design vari- angles. These geometrical constraints are de- constraint for isothermal forging. In such
ables. The forging volume, V, which is a smooth pendent on the material being forged and ensure cases, the appropriate constraints are ren-
function of the shape design variables, is the manufacturability and inspectability of the forg- dered inactive.
objective function. ing. Proper bounds on geometrical design vari-
Optimization Constraints. There are five ables avoid the creation of invalid geometries and The strain, temperature, sonic coverage, die-
types of constraints. prevent the occurrence of topological change. nonfill, and stress in Eq 2 to 6 may be func-
Equipment Constraints. The forging press The utilities in the UG CAD system are used for tions of space and time. To simplify the prob-
must have the capacity to forge the part. This is invalid geometry and topological checks. lem, these conditions are enforced only at the
a producibility condition. The load P required to end of the forging operation and, therefore, are
functions of space and not of time. Despite this
a pi (pi) a
(pi)lb (i 1, . . ., N )
forge the part should be less than the maximum ub (Eq 4)
tonnage available on the press (Pa) on which it is simplification, Eq 2 to 6 still need to be satis-
to be forged: h ha fied at every node or element of a finite-ele-
ment model. A further complication is that
P Pa (Eq 1) d da (Eq 5) some type of automatic unstructured mesh
generation has to be employed in the analysis,
Alternatively, this can be formulated as an en- Die stress cannot exceed a critical threshold due to the large deformation involved in the
ergy constraint when hammer or screw presses value. Otherwise, a die can be damaged due to forging process. The introduction of unstruc-
tured meshes causes the number of nodes and T ub Numerical Examples

elements as well as node and element locations Tmax = lim T g a dV + lTaub
T
W
l0
to be different from one design to another. Consider the forging shape optimization of a
Therefore, it is impractical and also impossi- T ub generic turbine disk. A cylindrical billet is forged
ble to impose constraints of a field function
form.

g a dV + l
W T (Eq 12)
into a disk of the shape shown in Fig. 4 (top). The
die geometry is captured in a CAD Unigraphics
One solution is to introduce minimum and parametric model (Fig. 4, middle), and the prob-
maximum types of constraint functions. They h h lem is to design the die shape so that the forging
are global functions defined for the entire model l0
G

h = lim h g d G + l ha
ha
g d G + l
G ha weight is minimized. It is required that the max-
and thus can be uniquely computed even when imum press load not exceed a prescribed value
an unstructured mesh is used. The global strain, (Eq 13) and that the forged part satisfy the condition of
temperature, sonic coverage, and die-nonfill are minimum cover over the sonic shape. Because
defined as: the disk is axisymmetric, only its radial-axial
d d
cross section needs to be considered (Fig. 4, bot-
e = min e l0
G

d = lim d g a d G + ld a
d
g a d G + l
G d tom). Several fillet radii R1 to R6 have been cho-
xW
sen as design variables. Both invalid geometry
Tmin = min T (Eq 14)
xW and intrusion into the minimum coverage over
Tmax = max T
the sonic shape can be prevented by putting sim-
where g(x) may be defined in different ways and
xW ple bounds on the design variables. It should be
l is a nonnegative parameter. For example:
noted that simple bounds may not be sufficient to
h = min h guarantee geometry validity in a more general
xG
1 if x 1 situation. They work here because there is no
d = max d g( x ) =
xG 0 if x > 1 (Eq 15) coupling among the selected design variables.
s = max s (Eq 8) Thus, the optimization problem is formulated
xW and as:
where W and G are the workpiece and its bound- Minimize: V
(1 - x )2 if x 1
ary, respectively, W represents the dies, and x is g( x ) =
the spatial coordinates. Thus, the strain, temper- 0 if x > 1 (Eq 16) Subject to:
ature, sonic coverage, die-nonfill, and stress
Rilb Ri Riub (i 1, . . . , 6)
constraints can be rewritten as: Equation 15 is a step function that is discon-
e e tinuous and tries to weight strain, temperature, P Pub
a sonic cover, and die-nonfill violation equally.
The averaged quantities computed using this
T min T lb
a function are not going to be smooth, which can
again be problematic for the gradient-based op-
T max T ub
a
timization algorithms. Equation 16 is able to
h ha overcome these difficulties. It is a smooth func-
tion and tends to weight more on the more se-

d da verely violated portion of the strain, tempera-
s
ture, sonic cover, and die-nonfill.
s a (Eq 9) By carefully examining the definition of these
global functions, one may conclude that the
One problem with this approach is that the
global functions have the following properties:
constraint functions may not be smooth func-
Equations 3 and 5 are equivalent to Eq 9, where
tions of the design variables, and this can cause
the global functions e , T min, T max, h , and d may
difficulties with the gradient-based algorithms.
be computed using min/max formulation (Eq 8)
Another problem is that this formulation de-
or the averaged formulation (Eq 1014) with
pends on point (nodal or elemental) information
either step function (Eq 15) or smooth function
that is dependent on the finite-element mesh
(Eq 16).
used and is not always a unique quantity. To
The following relations hold when constraints
resolve this issue, the following averaging
are violated:
schemes are proposed to compute the global
strain, temperature, sonic coverage, and die-
nonfill: (e)min/max (e)average1 (e)average2

(T min)min/max (T min)average1 (T min)average2
e e
(T max)min/max (T max)average1 (T max)average2
l0
W

e = lim e g dV + le a
ea
g dV + l
W e a

(h )min/max (h )average1 (h )average2
(Eq 10)

(d)min/max (d)average1 (d)average2 (Eq 17)
T
Tmin = lim T g lb dV + lTalb
where ( )min/max represents min/max formula- Fig. 4 Shape of generic turbine disk (top) and a CAD
T
W a
l0 tion (Eq 6), while ( )average1 and ( )average2 are parametric model of a cylindrical billet and die
shape (middle). Several fillet radii R1 to R6 are chosen as
T averaged formulation (Eq 1014) computed design variables. Zones can also be defined for regions

g lb dV + l using step function (Eq 15) and smooth func- with different constraints and property requirements (e.g.,
W Ta (Eq 11)
tion (Eq 16), respectively. see Fig. 12).
where V is the volume of the workpiece, P and isothermal, closed-die forging process is con- result of the optimization, the normalized press
Pub are the maximum press load and its upper sidered. The top and bottom dies are assumed to load drops from 96.9 for the initial design to
bound, respectively, and Rilb and Riub are the be rigid. The maximum load P normally occurs 77.8, which is within the available press capac-
given lower and upper bounds of the fillet radii, at the end of the forging stroke as the dies fill ity, a 19.7% reduction. The normalized volume,
respectively. The most aggressive shape, which out and the material starts to move into the flash however, has increased by 12.4% from 82.4 of
corresponds to the lowest volume V, has been region. The load changes rapidly with the stroke the initial minimum-weight shape to 92.6 of the
chosen as the initial design. It is relatively easy to at this stage of the process. Therefore, it is dif- final optimized shape. Intuitively, the result
get this shape from the specified disk design by ficult to accurately compare the loads at the end makes sense: the press load is reduced by de-
adding a minimum cover. This shape will have of the stroke from different die designs due to signing a set of dies with smoother boundary
small corner radii (all the design variables at their the inherent noise in the load predictions. For shapes to facilitate streamlined metal flow, but
lower bounds) making it difficult for the metal to this reason, the load P is arbitrarily set to be the the cost is a larger volume of the forged part,
flow around such sharp corners. As a result, the stroke-averaged load in between 98 and 99% that is, more material is used. The optimization
press load constraint may be violated for this de- of the final stroke. This enables a truer compar- and simulation took about 25 h on a single-
sign, and thus the fillet radii Ri, i 1, . . . , 6, ison of the load requirements for the various processor HP3600 workstation.
have to be increased, which results in a larger designs rather than considering the load at a Nonisothermal Forging Optimization. In
volume. Subject to the press load constraint Pub, unique value of the forging stroke. A good esti- the next application, the single-step isothermal
the optimizer should choose the optimal values mate on the real maximum press load may be forging process of the previous example is
of design variables Ri. As is pointed out in the obtained by multiplying P with a correction fac- replaced with a multistep, in particular, three-
previous section, iSIGHT provides a suite of op- tor. Automatic mesh regeneration is enabled to step nonisothermal one. Many of the forging
timization algorithms from ADS (Ref 18) and accommodate the large deformation that is in- processes are nonisothermal. The temperature
other packages. The modified method of feasible evitable in the forging process. Four design of the workpiece may change significantly due
directions is employed in this study. Since ana- variables R1 to R4 are used in this application. to heat exchange with the environment and the
lytical design sensitivities are not available, the Due to repeated remeshing during the forging dies. The temperature change will, in turn,
gradient information has to be obtained through simulation, nonsmoothness is introduced in the alter materials properties of the workpiece and
finite differences. finite-element solution. Therefore, a 10% per- elevate the press load requirement. Multistep
The numerical computation involved three turbation on the design variables has to be used forging whereby both the workpiece and dies
different workstations: a SUN Ultra-1 running during sensitivity analysis using finite differ- are reheated in between forging steps are in-
iSIGHT v3.0, a HP9000 running Unigraphics ences to smooth out the design space. Although troduced to compensate for the heat loss and to
v11.1, and a 16-processor Silicon Graphics the design sensitivities so calculated may not obtain workpiece and die temperatures close
Origin 2000 running DEFORM. This demon- be very accurate locally, they provide the opti- to the desired values. Here it is assumed that
strates that the integrated system can run in a mizer with the right search directions in a global 75% of the total stroke is reached in the first
heterogeneous environment over the network. perspective. forging step, while 90% and 100% of the
Both isothermal and nonisothermal forging opti- The initial and final shapes of the disk are stroke are achieved in the second and third
mization are presented for this shape. More de- shown in Fig. 5. The objective (volume) and strokes, respectively. Six design variables R1 to
tails of the optimization results are presented in constraint (press-load) function values versus R6 are used.
Ref 19 and 20. simulation runs are shown in Fig. 6. All the data Optimization results are shown in Fig. 7 and
Isothermal Forging Optimization. In one except the number of forging simulations have 8. The comparison between the initial and final
example, a time-dependent, plastic-deforming, been normalized. The results suggest that the shapes are shown in Fig. 7, while the disk vol-
optimization is close to convergence after 15 ume and press load histories against simula-
simulation runs. The smaller fluctuations are tion runs are shown in the left and right of
the result of finite-difference perturbation, Fig. 8. The optimization has converged after
while the larger ones are due to line search of about 30 simulations. The press load, which vi-
the optimizer. Since the abscissa shows the olates the constraint severely at the beginning,
number of simulation runs as opposed to the has been reduced from 99.4 to 79.6, a 19.9%
number of optimization iterations, the results of decrease. The value 79.6 is the load capacity of
both finite differencing and line search have the forging press and is shown in Fig. 8 (right)
been included. The upper bound of the press as a dashed line. The volume is increased from
load Pub 77.8 is shown as a dashed line in 82.4 of the initial minimum-weight design to
Fig. 5 Initial and final shapes for isothermal forging Fig. 6 (right). It is apparent that the forging 97.6 of the final optimized shape, a 18.4%
optimization press load far exceeds this limit initially. As a change. These results are qualitatively similar
in trend to the isothermal simulations described
earlier. The simulation and optimization took
about 35 h on a single-processor HP3600 work-
station.
Fig. 7 Initial and final shapes for nonisothermal forging

Fig. 6 Disk volume and press load versus the number of forging simulations for isothermal forging optimization optimization
Fig. 9 Influence of model resolution: the noise makes

it difficult to differentiate Designs 1 and 2.
Fig. 8 Designs 1 and 3, which are farther apart from each other,
Disk volume and press load versus the number of forging simulations for nonisothermal forging optimization
can be easily distinguished
Discretization, Approximation, and and development cost, it creates some limita- cations. DASA has been incorporated into the
Searching (DASA) Optimization tion for the application of first-order opti- iSIGHT Forging Optimization System.
mization methods since the analytical and Features of DASA. The basic ideas of DASA
In the methodology described previously, the semianalytical sensitivity information is are:
shape optimization problem for die design was often unavailable. Difficulties with field con-
Discretize the design space to the extent that
formulated in terms of global field constraint straint functions and enforcing constraints at
the analysis tool is able to resolve differences
functions, and commercial CAD and CAE tools every node or element is impractical.
between two designs. Then represent the de-
for forging design were integrated into a generic An averaging technique is needed to define
sign space with these discrete design points
and noninvasive procedure for shape optimiza- global or domain-based constraint functions.
only. This is consistent with the actual design
tion. However, there are several issues with the A practical discrete approximation search-
practice where die designers treat a geometry
application of this methodology to forging opti- ing method is needed to differentiate designs
parameter as a series of discrete values with
mization. One way to overcome these difficul- and reduce/eliminate the influence of the
certain spacing, say a quarter of an inch,
ties is a new method known as discretization, noise in the finite-element numerical results.
rather than a single continuous variable.
approximation, and searching (DASA) (Ref 21).
Gradient-based methods are probably the most Conduct a sequence of local searches in the
It has three essential ingredients: discretized de-
efficient ones for many types of optimization neighborhood of a base design in each de-
sign space, approximation in the neighborhood
problems. However, there are two big issues with sign iteration or cycle. This is to avoid the
of a base design, and searching in the discretized
this approach. First, there is no reliable and effi- expensive exhausted search, but there is a
neighborhood for a better design in each of the
cient way to obtain the sensitivity information danger of being trapped at the local opti-
optimization iterations.
since generally one does not have access to the mum similar to other numerical optimiza-
Issues with applying the conventional numer-
source code of the commercial software used to tion methods. One strategy for reducing the
ical optimization methods, which are designed
conduct the simulations. Second, and more im- chance of a local optimum is to start from
for a continuous design space as well as smooth
portant, the response surface for this complex different initial designs. It is desirable in
objective and constraint functions, to this appli-
process is inherently nonsmooth, which creates forging optimization to start optimization at
cation include:
difficulties for algorithms based on the assump- the most aggressive designa design of
There are a few important elements in the tion of smooth objective and constraint functions. least forging weight after satisfying geome-
forging process simulation that may cause Another possible approach to this nonsmooth try conditions including sonic cover, draft
nonsmoothness (noise) and even disconti- forging optimization problem is to build several angle, as well as fillet and corner radii. Such
nuity in the objective and/or some of the global response surfaces for the objective and a starting design is closest to the global op-
constraint functions. Examples include rigid constraint functions using design of experiment timum and thus the possibility of getting the
plastic material assumption, friction model, (DOE) type of methods. Then gradient-based best design is improved.
repeated remeshing due to large deforma- optimization algorithms can be applied to the Build a local approximation or response sur-
tion, and so forth. global response surfaces. However, a detailed face in the neighborhood of each base de-
Data mapping following each mesh regen- study showed that these response surfaces are sign. Because each forging simulation may
eration introduces a differing amount of of poor quality and computationally very time- take hours to complete depending on the size
error in each iteration of the optimization consuming to obtain. Therefore, building local of a model, some type of response surface
process. Optimization works by comparing response surfaces along the optimization path is has to be introduced to reduce analysis
different designs and identifying the better the more reliable and efficient approach. time. A local approximation strikes the bal-
one. If designs under evaluation are beyond DASA takes advantage of the response surface ance between speed and accuracy. The local
the resolution of the analysis tool, it will be methodology, incorporates the notion of neigh- search will exhaust all the design points sur-
impossible to choose the better design since borhood searching common to most numerical rounding a base design where the function
differences between various designs are lost optimization algorithms, and controls the influ- values at these points are computed using
in the noise of the calculations (Fig. 9). ence of inaccuracies in modeling through dis- the local response surface as opposed to the
Commercial software packages are em- cretizations of the design space. It proves to be actual finite-element analysis. However, the
ployed in this work and the finite-difference effective for noisy and nonsmooth optimization design selected in each design iteration will
method, which is expensive and prone to problems. No rigorous convergence study similar be validated with the more detailed model.
truncation and discretization errors, is the to those in gradient-based algorithms has been
The DASA algorithm involves the following
only resort to obtain design-sensitivity infor- completed, and DASA is not guaranteed to yield
steps and is summarized in a flow chart (Fig. 10):
mation. As a side note, the industry is moving a solution that is optimal in the stringent mathe-
toward using off-the-shelf code. Although matical sense, but it secures an improved solu- 1. Discretize the design space, choose initial
this greatly reduces the software maintenance tion, which is often sufficient in industrial appli- design Xb, and read algorithm data.
hood of Xb. Note that in this step Xmin is only

checked against all points whose objective and
constraint functions have been calculated through
the exact analysis.
Example Application of DASA. Consider the
forging design of a generic gas turbine disk.
Because of axisymmetry of the disk, a 2D analy-
sis is needed. There are 10 design variables, and
the section is divided into four zones (Fig. 12).
Design constraints on strain, temperature, sonic
cover, die nonfill, corner and fillet radii, and draft
angles are imposed within each zone. In addition,
there is a limitation on the forging time. Therefore,
there are a total of 21 optimization constraints.
Some additional details of the problem include:
Limitation of the forging load is maintained

within DEFORM.
Top and bottom dies are assumed to be rigid.
The forging simulation is time dependent
and nonisothermal.
Global constraints are computed using the
Fig. 10 Flow chart of discretization, approximation, and searching (DASA) optimization step-function formulation (Eq 16).
Initial design starts at the most aggressive
geometrythe minimum weight design
2. Select the initial interpolation point set fitting of second-order polynomials has been used after satisfying sonic cover, corner and fillet
{Xi, i 1, 2, , K} in the neighborhood of here. The criteria for choosing the best design radii, and draft angle constraints.
the base design Xb. Xmin in the neighborhood of Xb in step 5 are: The numerical computation involves two
3. Evaluate the objective F(X) and constraint
If there is no constraint violation at all, then different workstations: a SUN Ultra-1 run-
functions Vj(X), j 1, 2, , M at the inter- ning iSIGHT v4.0 as well as UG v15, and a
Xmin is the point where F(X) takes the small-
polation points in the neighborhood of Xb. 16-processor Silicon Graphics Origin 2000
est value.
4. Fit local response surfaces for the objective
If there is a constraint violation, then Xmin running DEFORM.
function F(X) and each of the constraint Vj A typical optimization run takes about 50 h
is the point where the sum of all constraint
(X), j 1, 2, , M in the neighborhood of Xb on a single-processor HP3600 workstation.
using function values obtained at the inter- violations [V ( X
l min
2
)] is the smallest.
polation points. l
Figure 13 shows the initial and final cross sec-
5. Select Xmin among all points in the neighbor- Similar conditions are used in step 6 to check if tions of the forging design as well as the sonic
hood of Xb such that it represents the best Xmin is indeed the best design in the neighbor- shape. Figure 14 depicts the optimization history
design. Here, all unknown objective and con-
straint functions are evaluated using the local
response surfaces.
6. Conduct exact objective and constraint func-
tion evaluations to validate that Xmin is indeed
the best design among points in the neighbor-
hood of Xb whose objective and constraint
functions have been obtained through the
exact analysis.
7. If step 6 confirms that Xmin is the best design
in neighborhood of Xb then go to step 8.
Otherwise, add this point to the interpolation
point set and go to step 4 to reconstruct the Fig. 11 Different strategies for selecting the initial interpolation point set in the two-dimensional space (N 2): left
local response surfaces. (N 3) points, middle (2N 3) points, and right (2N 1) points
8. If Xmin Xb, it is the optimal solution and stop.
Otherwise, set Xb Xmin and go to step 2.
DASA Implementation. The following are a
few additional details of the implementation of
DASA. The design space is discretized using a
uniform rectangular grid in step 1. Thus, the
neighborhood of a base design is a hypercube.
There are several strategies to select the initial in-
terpolation point set {Xi, i 1, 2, , K} in step 2.
Figure 11 shows three examples in the 2D case.
The first method is used in the numerical example
given in the next section. Different fitting meth-
ods and type of response surfaces may be used in
step 4 to create the local approximation of the ob-
jective and constraint functions. The least-square Fig. 12 Example of 10 design variables and four zones for design constraints
while keeping the maximum instantaneous Some examples of preform die design for
strain rate less than a specified limit and the axisymmetric forging problems are described in
forging temperature between specified max- the paragraphs that follow. Figure 15 shows that
imum/minimum limits? What is the com- without proper preform shape definition, the
bination of maximum load and dwell time material may not fill the finisher dies com-
under maximum load that minimizes the pletely or may result in the formation of folds
total forge time and minimizes the creep in or laps. An optimization system has been built
the dies at high loads? into the commercial metalforming software
What are the forging parameters (die speed, DEFORM (Ref 22). This was used to design an
Fig. 13 Discretization, approximation, and searching
forge temperature, etc.) to minimize the load optimal preform die, with which the material
(DASA) results: initial and optimal designs sur-
rounding the sonic shape; the initial shape is inside the and/or energy required (which in turn can filled the finisher die and the distribution of ef-
optimal design shape minimize die stresses and wear)? fective strain was made more uniform (Fig. 16).
What are the forging parameters to obtain It was also applied to the preform design prob-
the desired microstructure in the final lem (Fig. 17), and the distribution of effective
1
product or the desired shape of the final strain was more uniform with the optimal
0.9
0.8
product? shape. The methodology has been demonstrated
Normalized disk volume
for a few different shapes including more com-

0.7 The following examples illustrate the power and
0.6 plicated part geometries (Ref 22). In these ex-
versatility of the optimization tools with alterna-
0.5 amples, the shape of the preform was repre-
tive optimization formulations instead of mini-
0.4 sented by B-spline curves or piecewise linear
mizing the forging weight.
0.3 curves. The design variables are the control
0.2 points of the B-spline or piecewise linear curve.
0.1 Optimal Preform Design in Forging For example, if the diameter and height of billet
0 are design variables, billet shape can be repre-
1 1.5 2 2.5 3 3.5 4 4.5 5
An important parameter that characterizes sented by a rectangle with four control points,
Number of optimization iteration
metalforming processes is the initial shape of the and the x and y coordinates of corner point can
Fig. 14 Objective function (forging volume) versus workpiece or the shape of the preform. The ma- be design variables. Alternatively, it is also pos-
design iterations with the discretization, ap- terial microstructure as well as the geometry of sible to use Bezier representations of the geom-
proximation, and searching (DASA) optimization method
the final product are strongly dependent on the etry, and the control points of the Bezier curve
shape of the initial workpiece as well as on the are the design parameters.
of the forging volume. Since the optimizer is preform shapes at each of the subsequent form- Shape optimization with constraints requires
started from the most aggressive geometry, it is ing stages. It is therefore possible to envision a the evaluation of the gradients of all objective
expected the optimizer will add material to meet design procedure by which one can control the and constraint functions with respect to each
the design requirements since the initial design shape of the initial workpiece so as to achieve shape control parameter. The sensitivities of the
is infeasible. The weight added to the starting a desired geometry and/or material state in the nodal velocity with respect to the design vari-
design is at the appropriate locations and is the final product. A formal methodology for this ables are provided by direct differentiation of the
minimum needed to satisfy the various con- would be to pose the preform design as an opti- element stiffness equations. By using the sensi-
straints. Histories, which the user can look at to mization problem where the objective function tivities of the nodal velocity, the sensitivities of
monitor the progress of the optimization run, are would be a measure of the difference between the nodal coordinate, effective strain rate and ef-
also generated for all the constraints. the desired material microstructure/geometry fective strain are calculated and updated at every
for the final product and the computed material step. At the end of forging simulation, the objec-
microstructure/geometry corresponding to a tive and constraint function values and their gra-
Alternative Forging Optimization given initial shape of the workpiece. The pre- dients are calculated. In the optimization pro-
Formulations form design problem may also be subjected to gram, the gradients of objective and constraint
certain other geometrical or processing con- functions are used to determine the new search
straints by the user. direction.
The preceding sections focus on minimizing
the weight of the forging. However, optimization
problems can be formulated in several different
ways besides the minimization of forging weight.
Indeed, the goal of a general-purpose optimiza-
tion program is to systematically and automati-
cally evaluate various input parameters or design
variables that fulfill user-defined objectives.
Some examples of practical interest are:
What is the preform shape required to make
the strain uniform and greater than a speci-
fied minimum value in the final product?
The uniformity of deformation and grain
size in the part serves to minimize the vari-
ation of mechanical properties through the
part.
What are the forging parameters to keep the
temperature during forging between speci-
fied maximum/minimum limits?
What are the forging parameters (die speed/
energy/load) to minimize the forge time Fig. 15 Fold and underfill defects due to improper preform shape definition
leads to inaccurate sensitivity results and other

numerical problems due to the fact that the new
mesh topology could be very different from the
old one, invalidating the design-sensitivity defi-
nition. For large shape changes during optimiza-
tion iterations, it may be necessary to generate
an entirely new mesh. Shape changes should
avoid kinks and other geometric irregularities.
One way to ensure this is to impose additional
geometric constraints, for example, the continu-
ity of slopes and curvature signs at boundary
points, or to smooth out the irregularities in the
geometry at each stage.
Another example of making the strain uniform
in a more complicated part geometry is shown in
Fig. 18 (Ref 23). Here the preform shapes were
defined as simple third-order polynomials and
Fig. 16 (a) Optimal preform die shape. (b) Effective plastic-strain distribution without preform die. (c) Effective
the problem converged in about 25 steps in less
plastic-strain distribution with optimal preform die
than 2 h of HP 3600 computer time using a
relatively coarse mesh. For more complicated
conditions, preform shapes can be defined by B-
splines or Bezier curves rather than lower-order
polynomials. In this example, the initial simula-
tions were done with a relatively coarse mesh.
The coarse finite-element mesh solution quickly
gets to the optimum, and this can serve as the
starting point for a fine-mesh solution. This two-
step procedure minimizes the number of fine-
mesh iterations and enables the fine-mesh opti-
mum solution to be obtained rather quickly. Also,
because the user is not concerned with material
that is machined off, modeling of complex con-
ditions also may be simplified by not interrogat-
ing field variables in the whole volume of forg-
ing, but just in the finished part shape. The
finished shape is defined as a separate object, and
the field variables are remeshed and remapped to
this shape after the simulation is completed. The
advantage of this procedure is that the field vari-
Fig. 17 (a) Initial preform shape. (b) Effective plastic-strain distribution with initial preform shape. (c) Effective plastic-
ables from one design to another are evaluated
strain distribution with optimal preform shape
on the same mesh, and this considerably reduces
the noise in the gradient calculations. Another
advantage is that localized peak values at die/
A = 0.1
H B = 0.2
forging corners are removed, and it is easier to
E
G C = 0.3 make the strain uniform. In the limit, the user can
D = 0.4 make the forging and finished shapes to be the
N
M LKJ I F F E = 0.5 same, but remeshing to the same final mesh does
H G G
F H H G F = 0.6 have an advantage.
G G = 0.7
H G H = 0.8
I
I = 0.9 Multistep Forging Optimization
J J J = 1.0
I G K = 1.1 Optimization of multistep processes may be
L = 1.2 viewed as the design of the forming sequence
H
M = 1.3
that converts the initial workpiece to the final
F N = 1.4
G F F G
product while meeting the desired manufactur-
O = 1.5
P = 1.6
ing objectives and satisfying various process,
material, equipment, and geometrical constraints.
A forming sequence can be viewed at two levels
Fig. 18 (a) Effective plastic-strain distribution with initial preform shape. (b) Effective plastic-strain distribution with
optimal preform shape for the purpose of optimization (a) the broad
identification of the number, type, and order of
forming/heat-treatment operations that make up
Shape optimization should be carried out at compute the sensitivity gradients, the boundary the sequence (e.g., forward extrusion, open-die
the geometry level in that the constraints are nodes can simply be moved to the perturbed forging, etc.) and (b) the specific identification/
specified in terms of boundary points and geo- boundary without regenerating the entire mesh. selection of design variables in each of the form-
metric entities rather than individual finite ele- In fact, generating a new mesh for the gradient ing operations (e.g., die speed/load/energy, die
ments or nodes. When the shape is perturbed to calculation is not recommended because it often shapes in a preforming stage, etc.).
Fig. 19 Typical weight savings between manual (cur-

rent forging shape) and optimized designs
Multistep forging optimization involves the Fig. 20 Various generic optimized disk shapes
following considerations:
Minimization of number of forging steps
components such as disks are machined alter- the airflow on individual sections of the part and
Creation of optimization constraints for
nately on either side in an attempt to stepwise thus influence the local surface heat transfer coef-
each forging step
balance the distortion. The time spent flipping ficients. Obviously, the heat transfer coefficients
Development of criteria to decide when to
parts erodes productivity for thick, stiff compo- that can be achieved with fan cooling are lower
switch to a new forging step
nents; for thin components the strategy may be than those for oil quenching, so that for thick
Use of standardized preform shapes to keep
inadequate. Third, residual stresses and asso- parts it may not be possible to satisfy cooling-rate
production costs down
ciated distortion add complexity to machining requirements, but for moderately thick parts fan
Program control to simulate all forging
process development and shop operations. cooling offers clear advantages. For very thin
steps and coordinate data transfer between
Distortion affects the details of the machining parts such as engine seals where machining dis-
them
plan and the way the component interfaces with tortions due to residual stresses are especially
Other examples of both single and multistep machining fixtures. These effects generally vary critical, fan cooling may be the only process that
metalforming optimization can be found in Ref between material suppliers and from lot to lot. produces acceptable parts.
24 to 33. Figure 19 shows the typical weight Distortion thereby not only influences the effort Previous work related to heat treat optimiza-
savings between manual and optimized designs, incurred during initial development of machin- tion can be found in Ref 35 to 37, Batista and
with the final machined shape also shown for ing plans, but may require adjustments after the Kosel (Ref 35) developed the analytical sensi-
reference. Finally, Fig. 20 shows various generic initial plan has been set. tivities of the residual stresses after quenching.
optimized disk shapes. The buildup of residual stresses during heat The sensitivities of the relative error of residual
treatment (oil quench or fan cool) and subsequent stresses from the estimated errors of material
relaxation following metal removal are impossi- data during the quenching process were re-
Heat Treatment Optimization ble to assess using intuition, engineering judg- ported. Karthikeyan et al. (Ref 36) developed
ment, or empirical methods. The physical inter- mathematical models to optimize the heat treat-
Of the manufacturing processes to produce play of quench heat transfer, elevated-temperature ment conditions for maximum yield strength and
gas turbine engine components, heat treatment mechanical behavior, and plastic-deformation ductility of aluminum/silicon-carbide particulate
and machining are two critical operations that localization is complex. Subtle changes in pro- composites. Saigal and Leisk (Ref 37) optimized
have an interacting effect on the finished engine cessing conditions and component geometry can the tensile properties of alumina/aluminum
component. For example, nickel-base superal- significantly affect the magnitude and pattern of metal-matrix composites using Taguchi analysis.
loys are heat treated following forging to precip- residual stresses. Determining residual stresses Formulation of Heat Treatment Optimiza-
itate the gamma-prime (g) strengthening phase; and subsequent distortion requires modeling tion. The heat treatment optimization tool (Ref
a process that involves quenching from near the using finite-element method. This method has 38) optimizes the heat treatment of engine disks
solvus temperature. Thermal gradients during been used to evaluate the effect of processing by changing the heat transfer coefficients (the de-
quenching cause thermal stresses, which drive conditions on residual-stress development and sign variables) around the disk surface in order to:
localized plastic deformation and residual-stress the effect of residual stresses upon distortion
Achieve a residual-stress distribution for
buildup. Upon cooling to room temperature, during machining.
minimum distortion during machining
residual stresses can exceed half of the alloy ten- The purpose of the heat treatment process is
Minimize cooling-rate nonuniformity and
sile strength, which leads to distortion following to develop the necessary mechanical properties
hence residual stresses
quench and later during machining. in the forged part. This is achieved by heating
Achieve cooling rates to meet mechanical
Residual stresses and associated distortion the part to solution temperature and then cooling
property requirements
have a significant effect on manufacturing cost it rapidly. During the cooling phase residual
Improve part-dimensional stability during
in three distinct ways. First, the forging and in- stresses are introduced. In the case of nickel-
machining and during service
termediate heat treat shapes contain additional base superalloys, a certain minimum cooling
material to account for expected distortion. This rate has to be maintained to generate the needed As in the forging optimization software de-
material, added to ensure a positive material en- creep and tensile properties. On the other hand, scribed earlier, the heat treat optimizer is based
velope over the finished part shape, represents a the faster the cooling process is, the higher are on a general procedure to generically and non-
raw material cost and increases the machining the resulting residual stresses. invasively integrate commercially available soft-
cost. This also imposes a limit on the benefit of A description of the various heat treat proce- ware. The integration and optimization environ-
near-net-shape forging, which is being pursued dures is given in Ref 34. Traditionally, an oil- ment is similar to that described for forging
vigorously by the industry. Second, part move- quenching process has been employed, which en- optimization except that the UG CAD system is
ment during machining requires that the ma- sures fast cooling and thus a high cooling rate, but not used since the geometry is not part of the
chining process engineer plan machining opera- the oil-quenching process introduces high resid- heat treat optimization process. iSIGHT has been
tions and fixtures so that distortion does not ual stresses, and, from a process optimization adopted as the framework of the system and
compromise the finished part shape. Movement point of view, offers very little room for improve- DEFORM as its simulation code that supplies
during machining or excessive distortion from ment as there are very few parameters that can be the objective and constraint information. The
an inappropriate metal-removal plan can cause controlled. Therefore, fan cooling is gaining heat treat module (Ref 39) of DEFORM is used
machining scrap. To combat this possibility, larger acceptance where it is possible to control to simulate the heat treat process.
The challenge here is to formulate an opti- The procedure gives the user a choice in terms 11 to 15 min on an HP-715 workstation, execu-
mization problem without actually having to ex- of optimization constraints. A hard constraint ccr tion time per iteration was in the range of 3 to
ecute a combined heat-transfer/stress-analysis on the cooling rate could be imposed such that: 4 h, including computational overhead, using fi-
each time the optimizer needs a new design nite differencing to obtain gradients for the opti-

point. An accurate heat transfer analysis requires 1 t target - t if t < t target mizer. In order to cut down on these times, the
small time steps in the simulation, and a stress
ccr =
nodes nodes 0 if

t t target (Eq 19) optimization was started with all heat transfer
analysis, in turn, requires a fine finite-element coefficients linked to only one design variable.
mesh; therefore the combination of both is the This constraint has a discontinuity at 0, exactly This problem was executed for six iterations,
most computationally expensive analysis pos- where it is active, and will never assume a value using the sequential linear programming tech-
sible. In general, though, the stress analysis is less than 0, that is satisfied and not active. This nique from ADS (Ref 18), which is one of the
much more time consuming than the heat trans- discontinuity leads to problems with gradient- packages in iSIGHT, until both constraints were
fer analysis alone. Since it is known that spa- based optimizers, which will always see a zero active. The full convergence history of the ob-
tially uniform cooling reduces residual stresses, constraint gradient for a satisfied or active con- jective function is depicted in Fig. 22.
the idea is to formulate an objective function that straint; therefore, in the case of constraint satis- For this problem, the modified method of fea-
penalizes nonuniform cooling and at the same faction the constraint value of 0 is replaced with sible directions, also from ADS, was chosen as
time ensures cooling at or above the target cool- the difference of the target cooling rate ttarget and the optimization technique. The deviation func-
ing rate. These are obviously two conflicting the minimum tmin of all nodal cooling rates: tion was initially reduced from a value of 1.4 to
objectives since fast cooling always means un- about 0.6 and then further down to under 0.2.
even cooling as the heat can only be extracted at ccr ttarget tmin (Eq 20) These numbers as such have no physical mean-
the surface of the part. Therefore, the objective ing, but the significance can be seen in the re-
function for the heat treatment optimization Thus, at least the sign of the constraint gradi- duction of the standard deviation of the nodal
problem is formulated as a quadratic that penal- ent that the optimizer sees above and below a cooling rates, indicating a much more uniform
izes the deviation from the cooling-rate target, constraint value of 0 will be equal. An additional cooling than at the starting point. This is also ev-
which is a material-dependent value: constraint can be placed on the nodal fraction idenced by a comparison of the initial cooling-
that fulfills the cooling-rate target, which has to rate distribution (Fig. 23) and the optimized
2 be equal to 1.0 if the target is met everywhere. cooling-rate distribution (Fig. 24), all normal-
w t
(
target -t )
if t < t target The two constraints may seem somewhat redun- ized with respect to the target value.
obj = (1 - w) (t
nodes

target -t
2
)
if t t target (Eq 18) dant, but depending on the optimization strategy Figure 25 shows the history of the design vari-

used, one or the other or a combination of both ables for the second segment in the optimization
leads to the best convergence. process where all design variables are active,
where w is a user-defined weighting factor be- Heat Treatment Optimization of a Turbine normalized with respect to the starting value.
tween 0 and 1 that penalizes under- and over- Disk. The heat treatment optimization procedure The result is large heat transfer coefficients
achievement of the target cooling rate differ- described previously was applied to a generic around the thick portions of the disk (h1, h2, h3,
ently, t is the cooling rate, and ttarget is the target turbine disk. Figure 21 shows the heat treatment h7, h8) and significantly lower coefficients for
cooling rate to achieve the desired properties. A geometry and the distribution of the nine design the thinner sections, leading to a more uniform
value w of close to 1 seems to give the best re- variables employed. Since each finite-element cooling while still maintaining the target cooling
sults. The objective function is really an integral analysis of the heat transfer problem took about rate.
over the volume of the part or the cross-sectional
area for the 2D analysis for the axisymmetric
parts considered here, but since temperature
values are only available at the finite-element
nodes, it is formulated here as a sum over all
nodes.
This quadratic formulation of the objective
ensures a smooth convergence behavior. It cap-
tures both the desire for uniform cooling and the
fulfillment of the cooling-rate target by penaliz-
ing the objective with the square of the deviation
from the target. Additionally, with the weight w Fig. 21 Generic turbine disk geometry and heat transfer coefficient distribution
that is to be chosen between 0 and 1, the user has
the option to differentiate between over- and un-
derachievement of the cooling-rate target. Since
the objective of the heat treatment optimization
is to at least reach the cooling-rate target, the
user usually wants to penalize underachieve-
ment far more severely than overachievement.
A weighting factor w of 0.98 (98% penalty on
cooling-rate underachievement) has been suc-
cessfully applied.
The design variables are the surface heat trans-
fer coefficients, hi, which can be related back to
a certain airflow produced by the fan-cooling
apparatus. Usually, a breakpoint between two
design variables should be chosen wherever the
geometry of the part changes, for example, as the
thickness changes from bore to web or web to
rim, and so forth, as shown in Fig. 21. Fig. 22 Objective function history
Fig. 23 Initial cooling-rate distribution: note the large surface/center non- Fig. 24 Optimized cooling-rate distribution: note uniformity in cooling
uniformity rate compared to Fig. 23
The cooling-rate target is met throughout

the part
Maximum stresses reduced by 89% (ten-
sion) and 84 % (compression) as compared
to a traditional oil-quenched part
An integrated engine disk design and manu-
facturing simulation and optimization scenario
has been formulated. As an initial step, a heat
treatment optimization procedure was developed,
taking advantage of the lower computational cost
of a pure heat transfer analysis compared to a
combined heat-transfer/stress-analysis. The for-
mulation of the objective as a quadratic cooling-
rate deviation function has shown to produce con-
siderably lower residual stresses in the generic
Fig. 25 Normalized design variable history turbine disk analyzed in this chapter. In other
words, with this formulation it is possible to
The question still to be answered is what effect process. These results clearly show the advantage lower residual stresses in forged parts without the
this optimization procedure, which is based on of a numerically optimized fan-cooling process cost of a coupled thermal-stress analysis during
heat transfer analysis only, has on the residual compared to the traditional oil quenching. One the actual optimization.
stresses of the part which is of ultimate inter- can claim, however, that the starting point for the Alternative Heat Treatment Optimization
est. Therefore, a combined heat-transfer/stress- optimization was not very realistic with almost Formulations. The previous examples have fo-
analysis was performed on both the starting con- maximum fan cooling all around the disk, and an cused on minimizing the cooling-rate variations
figuration and on a disk with the optimized heat experienced person would have been able to set and the residual stresses in the forging. The
transfer coefficient distribution. For comparison up a better fan-cooling process manually. How- general framework of iSIGHT permits the opti-
ever, this argument misses the point since the mization problem to be formulated in several
purposes, an analysis of a typical oil-quenching
starting point for the optimization was deliber- different ways. The goal of a general-purpose
process was also performed.
ately chosen to be far away from the optimum in optimization program is to systematically and
Figures 26 and 27 show the resulting hoop and
order to show that the optimizer will find the op- automatically determine the input parameters or
radial residual stresses, respectively, normalized
timal processwithin convergence margins, of design variables that fulfill the user-defined ob-
with respect to the maximum tensile stress of the
course, no matter where it starts. This reduces the jectives. Examples of alternative heat treatment
oil-quenched part. Residual stresses are highest
need for prior experience in arriving at the opti- optimization formulations of practical interest
for the oil-quenched disk, which was closely fol-
mized solution, which was confirmed during that illustrate the power and versatility of these
lowed by the nonoptimized fan-cooled disk with
multiple runs with different starting points on ac- tools include:
uniform high fan blowing all around. The resid-
tual geometries, which are of proprietary nature
ual stresses for the optimized disk, in turn, are The inverse problem of determining heat
and cannot be shown here. The formulation of
considerably lower, almost by one order of mag- transfer coefficients from measured temper-
the objective function as a quadratic clearly aids
nitude compared to the oil-quenched part in ature-time data
in this behavior.
terms of tensile stresses (Fig. 26, 27). The reduc- The problem of not minimizing residual
Therefore, to summarize, the optimized solu-
tions in compressive stresses are not quite that stresses but of creating a residual-stress dis-
tion results in:
large, but still by a factor of between six and tribution that minimizes the subsequent
seven. Table 1 shows the stress reductions of A cooling-rate distribution that is much machining distortions and ensures optimal
the maximum tensile and compressive hoop and more uniform, especially in the thinner disk dimensional stability during service of the
radial stresses compared to the oil-quenching sections part
Fig. 26 Hoop stress. (a) Oil-quench process. (b) Optimized process
Fig. 27 Radial stress. (a) Oil-quench process. (b) Optimized process
Table 1 Stress reductions in tension and Of these, the inverse problem is described in mized to obtain minimum distortion satisfying
compression after heat treatment some detail. The inverse problem of obtaining the two constraints of maximum stress (to
optimization surface heat transfer coefficients from meas- avoid quench cracking) and the average surface
Initial Final
ured temperature-time data during the heat hardness requirements of the final product (Ref
Traditional optimizer optimized treatment of a part has received considerable 41). The design variables are the heat transfer
oil quench guess solution attention. Several inverse methods (for exam- coefficients that are functions of the surface
Hoop stress ple, Ref 40) have been developed and used to and ambient temperatures or time. This func-
Tension 1.000 0.861 0.112 solve this problem and have met with varying tionality is represented by spline functions
Compression 1.007 0.854 0.159 degrees of success. The inverse problem can be making the coefficients defining the splines as
Radial stress cast as an optimization problem. The objective the design variables First, the response surface
Tension 1.000 0.850 0.093 function is the sum of the squares of the devia- method is used to obtain the analytical models
Compression 1.254 0.908 0.178 tion of predicted and measured temperatures of the objective function and constraints in
with the surface heat transfer coefficients as terms of the design variables. Then, some of
the design variables. The optimization problem the less important terms are eliminated to im-
The problem of obtaining a cooling-rate dis- then is to determine the set of design vari- prove the fit of the response models. Next, with
tribution that in turn results in an optimal ables that minimize the objective function. Two the closed-form response surface equations, the
distribution of mechanical properties (e.g., examples of the solution of this problem are sequential quadratic programming method in
high tensile strength in the disk bore and described. the design optimization tool DOT (Ref 18) is
high creep strength in the disk rim) and a In the first example, the heat transfer coeffi- used to obtain the optimal design point. The
minimum weight of the finished disk shape cient during a gas-quenching process is opti- finite-element package, DEFORM is used to
predict the material responses during the the heat-transfer coefficients are strongly de- mization problems. This method iteratively finds
quenching process. pendent on the number and location of the ther- a search direction and performs a one-dimen-
Phase transformations, deformation, and heat mocouples. In particular it is shown that the sional search along this direction. The emphasis
transfer are integrated during the simulation placement of the thermocouples strongly influ- is to reduce the objective while maintaining a
process. Due to the fitting error of the response ences the confidence bounds for the estimated feasible design. This technique attempts to rap-
surfaces, the optimal design point obtained by parameters. The procedure has been success- idly obtain an optimal design, handles inequality
the response surfaces may not satisfy the con- fully applied to selected examples. and equality constraints, and satisfies constraints
straint with the finite-element method even These examples have served to illustrate that with high precision at the optimal point.
though the constraints are satisfied by the re- with general-purpose optimization tools, the Successive Approximation Method (SAM).
sponse surface method. Therefore, the optimal user has a wide variety of options of selecting This method lets one specify a nonlinear prob-
design point obtained by the response surface the objective function and constraints for any lem as a linearized problem. The SAM method
method is added to the design of experimental particular situation to obtain the optimal solution is a general one, which uses a simplex algorithm
points to improve the fitting accuracy of the quickly and effectively. in addition to sparse matrix methods for lin-
response surfaces until the relative errors of the earized problems. If one of the variables is
objective and constraints obtained by the re- declared an integer, the simplex algorithm is it-
sponse surface method and finite-element method Summary of Optimization erated with a branch and bound algorithm until
are in good agreement. The response surface Techniques the desired optimal solution is found.
method provides an effective way to optimize Response surface methodology (RSM) is
this process. one of the approaches for function approxima-
The second example described here deals Optimization Methods tion in design optimization and is especially
with the more conventional inverse problem of suitable when the sensitivity information is not
obtaining heat surface transfer coefficients No one optimization approach works well for available or difficult to obtain. The response sur-
from temperature-time data (Ref 42). In this all problems, particularly for MDO problems. face model is a simplified Taylor series with
case, the disk geometry shown in Fig. 28 was To avoid convergence or other problems, special only lower-order terms. The advantage of RSM
used. The temperature-time data during quench- techniques have to be developed on a case-by- is its wide applicability since any real function
ing is prescribed at two points, and the bound- case basis. All the techniques described in this can be expanded as a Taylor series. Since the ac-
ary of the disk is divided into segments, each section are numerical optimization techniques. curacy of Taylor series is usually restrained by
with its own heat transfer coefficient, which is They generally assume the design space is uni- the truncation error in highly nonlinear cases or
specified as a function of surface temperature. modal, convex, and continuous in nature. For the in approximating a large domain, the RSM faces
The control points in the functional depend- forging optimization problem, all these schemes an accuracy and complexity challenge. If too
ence of heat transfer coefficient on temperature produced results (optimized forging weight) many terms are included, then the process of
are the design variables. The optimization pro- within 1 to 2% of each other, but the number of model identification is complicated and compu-
cedure is incorporated into the heat treat mod- iterations required varied significantly, depend- tationally expensive. To find a balance between
ule of DEFORM, and a converged solution is ing on the optimization scheme used. The vari- the accuracy and complexity, in practical appli-
obtained in 20 iterations matching the predicted ous methods are briefly described in this section; cations, a second-order polynomial is a widely
and measured temperature profiles very closely. more details may be found in the iSIGHT man- used formulation for RSM. Higher-order terms
For a more realistic problem, there would be uals (Ref 9). can be selectively dropped, depending on their
several temperature-measurement points, and Exterior Penalty (EP). This method is widely importance as determined from initial analyses.
the surface of the disk would be divided into used for constrained optimization. It is usually Unlike gradient-based approximations, the RSM
several segments, each with its own heat trans- reliable and has a relatively good chance of find- usually does not match the function values at
fer coefficient. ing the true optimal solution if a relative min- trial points, and model parameters change grad-
Another optimization-based inverse method, imum exists. The EP method approaches the ually with new points added into the data set.
which determines the thermal boundary condi- optimum from an infeasible region, becoming Therefore, the use of RSM in design optimiza-
tions (heat transfer coefficients) during the heat feasible in the limit as the penalty parameter ap- tion (especially with discrete variables) could
treatment of aircraft engine disks has been de- proaches infinity. create traps or oscillations in convergence.
veloped (Ref 55). A nonlinear optimization DONLPSequential Quadratic Program- Further details of the RSM method and its appli-
methodology has been developed that provides ming. This technique uses a slightly modified cation are provided in Ref 43 and 44.
the mean and confidence bounds for the pre- version of Pantoja-Mayne update for the Hessian
dicted heat transfer coefficients. It is shown that of the Lagrangian, variable scaling and an im- Optimization Exploration Techniques
proved step-size algorithm. With this technique,
bounds on the variables are treated in a projected The use of MDO optimization systems such
gradientlike fashion. as iSIGHT can combine multiple exploration
Hook-Jeeves Direct-Search Method. This capabilities. The use of these design exploration
algorithm begins with a starting guess and methods in conjunction with approximation
searches for a local minimum. This optimization models can be used to greatly reduce the compu-
technique does not require the objective function tations required to obtain an optimized solution.
to be continuous. Because the algorithm does Design of experiments (DOE) allows for a
not use derivatives of the objective function, the larger number of design parameters to be used to
function does not need to be differentiable. examine the effects on the objective. DOE is
Sequential Linear Programming (SLP). This then able to estimate a near-optimal design from
is a strategy used to solve constrained optimi- these data. Engineers can gain a lot of insight
zation problems. This technique is easily coded from a minimum number of simulations. This
and applicable to many practical engineering insight can be used in subsequent optimizations.
problems. Exploitive optimization techniques are
Modified Method of Feasible Directions very good at improving a design in a local area.
Fig. 28 The determination of heat transfer coefficients (MMFD). This is a direct numerical optimiza- These techniques are commonly known as hill-
from temperature-time data tion technique used to solve constrained opti- climbing techniques since they require gradient
calculations to learn which way to improve the tions to include the effects of contact and fric- models have reduced shop-floor trial and error.
design. tion for both steady-state and non-steady-state The next step is to reduce modeling trial and
Explorative Optimization Techniques. These metalforming processes. This publication also error and use optimization tools to obtain a pro-
optimization techniques are very good at ex- contains references to other work in the area of cess and product that is not just acceptable but
ploring the complete design space. They usually sensitivity analysis for various forming pro- optimal. In this chapter, the application of MDO
require 500 designs to begin to find optimal cesses with respect to shape and material param- techniques to two manufacturing processes,
designs. eters for isothermal as well as nonisothermal forging and heat treatment, has been described
Approximation Models. With a few well- deformations, steady and non-steady processes. in detail. The forging optimization system mini-
chosen design points, an approximation model One needs to calculate the sensitivity of the ma- mizes the total forging cost subject to the con-
of the real simulation can be created. Any of terial state and geometry at various stages of de- straints of satisfying property requirements and
iSIGHTs optimization techniques can then use formation with respect to infinitesimal changes equipment capability. The heat treatment opti-
this approximation model for calculating an in each of the design variables (process para- mizer results in a part optimized for properties
optimal design. These approximation models are meters and die shape). The formulation is based and stable dimensions both during machining
automatically updated as more simulations are on the differentiation of the governing field and during engine service.
performed. equations of the direct problem and develop- The optimization method was formulated as a
ment of weak forms for the corresponding field- parametric geometry and detailed forging and
sensitivity equations that are consistent with the heat treatment analysis based problem. An aver-
Sensitivity Analysis kinematic analysis, material behavior, and the aging scheme was introduced to compute global
contact/friction subproblem, and the transfer of constraint functions that are smooth and work
As a by-product of optimization, a sensitiv- design sensitivities between meshes during with unstructured finite-element meshes. The
ity analysis showing how sensitive the outputs remeshing. Different schemes of design differ- methodology was demonstrated on the isother-
are to the process variables is useful to help entiation can be envisioned depending on the mal and nonisothermal forging shape and heat
identify which variables to control and how level at which the design differentiation is per- treatment optimization of turbine disks. The
tightly. The results of this analysis can help set formed. The field-sensitivity equations are linear results showed both significant process and cost
tolerances and controls on the significant and can be efficiently solved simultaneously improvements. The results were obtained on
process variables and lead to an effective pro- with the solution of the direct-deformation prob- workstations in a reasonable period of time and
cess control strategy and ensure a robust pro- lem. To avoid issues of nondifferentiability of with minimum user interaction. The upfront op-
cess and product by reducing the effects of the contact conditions, appropriate regularizing timization effort is small compared to the manu-
variability that is inherent in any process. assumptions are introduced. Finite-dimensional facturing savings it results in.
Several papers have been published on sensitiv- gradients of objective functions are then com- The optimization formulation was incorpo-
ity analysis, and examples may be found in Ref puted using the results of the shape-sensitivity rated into an integrated system involving three
43 and Ref 45 to 52. These describe shape and analysis. The accuracy and effectiveness of the commercial software packages: iSIGHT, Uni-
process optimization procedures to design pre- method are demonstrated with representative graphics, and DEFORM. A general procedure
forms based on sensitivity analysis and genetic extrusion and forging problems. The results of was developed to noninvasively integrate exter-
(adaptive) algorithm techniques. the continuum sensitivity analysis are validated nal CAD and CAE systems for geometry and
A gradient-based approach can be used to by a comparison with those obtained by finite- detailed analysis-based optimization. iSIGHT
solve the constrained optimization problem ef- difference approximations (i.e., using the solu- was used as the integration framework. A client-
ficiently. This necessitates the evaluation of tion of a perturbed deformation problem). server architecture was established for the
the gradient of the objective function and con- The procedure has been extended to multi- iSIGHT and CAD integration so that iSIGHT
straints with respect to the design variables. stage forming processes that involve the com- can drive the CAD system through interprocess
These gradients or sensitivities are quantitative putation of both shape as well as parameter communication to perform geometry operations
measures of changes in the objective function sensitivities (Ref 46). However, unlike single- such as updating models, computing geometry
and constraints as a result of perturbations in the stage, shape-sensitivity analysis, where the ini- properties, exporting geometry for mesh gen-
design variables. Gradient computations can be tial workpiece shape depends explicitly on shape eration, monitoring shape modification, and so
performed using finite-difference approxima- design variables, the intermediate preform shape forth. The noninvasive coupling of a CAE tool
tions and the results of the direct analysis for in a generic forming stage of a multistage such as DEFORM with a general-purpose op-
two nearby design variables. In addition to sig- process depends implicitly on the design vari- timization software like iSIGHT permits the
nificant computer resources required for solving ables (nonshape parameters) that define the pro- user to construct a variety of objective functions
the direct problem multiple times, difficulties cessing history of the intermediate preform. and define the appropriate constraints and de-
arise in such calculations from the fact that Using analytical gradients instead of numerical sign variables to optimize special-purpose cus-
many direct-analysis tools are insensitive to in- derivatives, ANSYS (Ref 53) has currently in- tomized processes. The examples given here
finitesimal changes in the design variables (e.g., troduced the calculation of the high-order deriv- merely serve as illustrations of the power of
the die surface) and cannot provide accurate sen- atives of displacement, stresses, and reaction optimizing tools.
sitivity fields. In addition, the calculated sensi- forces in a product called FEM Explorer. It pro- Material models for microstructure and prop-
tivities are dependent on the step size used to vides the first-order sensitivities, but also the erty evolution and material behavior need to
compute the perturbed direct-deformation prob- full Taylor expansion. The main issue is to re- be used in conjunction with optimization algo-
lem. This is particularly true when the computed duce computational time by using analytical de- rithms to optimize workability, mechanical prop-
sensitivity fields are of the same order of magni- rivatives as opposed to finite differencing. erties, achieve efficient material flow, and precise
tude as the numerical error in the solution of the dimensional control. MDO can overcome the mul-
direct analysis. However, an efficient design titudinous challenges associated with achieving si-
methodology necessitates the accurate computa- Conclusions multaneous microstructure/property control, part
tion of the design derivatives of various defor- quality (durability), shorter delivery time, and re-
mation-related parameters. Sensitivity analysis The material modeling methods and finite- duced manufacturing costs. It is in the interest of
is a method that is widely used to evaluate these element modeling techniques developed in the the component supplier industries to adopt MDO
gradients. 1980s and 1990s for metalworking processes are to help them increase their productivity.
A continuum sensitivity analysis was devel- accepted, routine practices in the automotive, Even after a process design has gone through
oped (Ref 45, 46) for large inelastic deforma- aerospace, and other industries. These process computer analysis and optimization, some shop
floor tweaking of the process is often necessary. tion constraints, for example, material- mization compatibility. The die designer
Thermal shrinkage/expansion, lubrication, die dependent optimization strategy; develop should use feature-based parametric
deformations, material property variability, and processing windows and microstructural modeling, keeping in mind the optimiza-
other uncertainties that cannot be incorporated models for various materials in terms of tion requirements when defining the die
into a deterministic process model require fine strain, strain-rate, temperature con- geometries. This will ensure smooth
tuning the process on the shop floor. Computer straints used by the optimizer geometry transition to optimization and
analysis is a cost-effective way to weed out poor e. Capture manufacturing equipment con- simplify the task of setting up the opti-
designs that will not pass a manufacturing trial straints more realistically and optimize mizer. The way geometries are normally
or will pass only after a great deal of effort. This parameters only to the tolerance to which created, some of the effort has to be re-
is because the costs of computer analysis have they can be controlled and integrate with done to convert the geometry from the die
dropped several orders of magnitude in the past process control designer into optimization usable format.
decades, while the costs of labor, plant space, f. Incorporate soft versus hard constraint b. Check for geometry errors in CAD-FEM
and other factors have increased. The costs of differentiation: some constraints can translation.
computer analysis will continue to drop an order be approximately satisfied (soft) as op- c. Use robust parametric CAD representa-
of magnitude every five years, while the costs of posed to others that have to be satisfied tion of geometries (especially in 3D).
shop-floor trials will continue to rise. Someday, (hard)
Programmatic:
it will be cost effective to optimize die designs g. Heat treatment: include shape optimiza-
a. Perform seamless integration of various
so that they are less sensitive to the process vari- tion, more generalized heat transfer coef-
tools.
ables over which one has limited control. After ficient dependence on temperature/time,
b. Improve user-friendliness (graphical user
this happens, when the process is put into pro- quench crack criteria, and coupled ther-
interface).
duction, it will result in one good part after an- mal/stress analyses optimization
c. Keep updated with latest releases of op-
other, even with the inherent variations in the h. Conduct more extensive applications and
timization, CAD and CAE tools.
process parameters over which one has little or validation
Communication:
no control. The potential cost savings will be Methods to reduce computational effort:
a. Incorporate new interprocess communi-
large. Deterministic models current at the time a. Subgroup design variables: establish
cation mechanisms to improve data pass-
of publication assume that the material data, op- parametric relationships in CAD system
ing between processes.
erating conditions, and other parameters are well to reduce the total number of design vari-
b. Develop systems to permit distributed
specified. In reality, many defects and part rejec- ables to be optimized.
computing on a network using diverse
tions occur due to uncertainties in some or all of b. Screen shapes in CAD system: identify
computer platforms using the latest web-
these parameters, and the computer models need the shape design variables that have the
based technology.
to incorporate these statistical data variations in most significant influence on forging
Extend to other manufacturing processes:
order to provide confidence levels on the results. weight and ignore the low-influence
a. Machining: minimize number of ma-
Most of the real problems are inverse in na- variables for optimization.
chining passes for acceptable distortion.
ture; that is, they involve the determination of c. Use approximate models and response
b. Inertia welding: minimize energy and
conditions that will result in a given final situa- surfaces to speed up computationsas
material requirements.
tion. Analysis tools are direct in nature; that is, optimization problems grow in size,
c. Inspection: minimize time for sonic and
they tell the user what the final outcome will be algorithms such as sequential quadratic
number of scans; include sonicability as
for a given set of conditions. Instead of manually programming require large amounts of
a geometric constraint during forging
running direct simulations for a range of given computer memory. The optimizer itself
optimization.
conditions, the optimization tools provide an begins to use considerable computer
d. Determination of hard to measure materi-
alternative to determine the conditions that will time, relative to the function and gradient
als properties. Use the materials proper-
provide not just an acceptable but an optimal computations. Several methods are in the
ties as the design variables and their opti-
product. The quantity to be optimized (the ob- research stage to overcome such prob-
mal values are obtained by minimizing the
jective function), the independent parameters lems, and these need to be more robust.
deviations of modeling predictions from
(the design variables), and the bounds within One strategy is to identify potential de-
the measured values of selected response
which the user can operate (optimization con- signs with screening DOE runs in a
variables.
straints) can all be decided by the user. These are large design space and then zoom into
Develop fast-acting low-fidelity optimiza-
the advantages of MDO. smaller design spaces, rerun DOEs, con-
tion models for preliminary design trade-
struct response surfaces, validate results
offs for new parts.
with extra runs, and conduct optimiza-
Future Work tion on response surfaces. As these problems and issues are addressed,
d. Develop adaptive meshing to reduce one should see multidisciplinary optimization
number of finite-elements without sacri- being applied to an increasingly broader range of
Future work should be directed at transition-
ficing accuracy. processes, first in isolation to each process and
ing a robust, user-friendly, fast-acting tool to a
e. Improve optimization robustness: discrete then in a coupled manner to interacting processes.
production environment for routine application.
optimization, other optimization tech- The ultimate goal is integrating the simultaneous
Some of the items to be considered are:
niques. optimization of all the steps involved in product
More comprehensive optimization of manu- f. Conduct rapid optimization on a coarse design and manufacturing as a multidisciplinary
facturing processes: finite-element mesh followed by fine optimization problem. This will result in the over-
a. Inclusion of coupled die-stress analysis tuning on a fine mesh using the coarse all cost and performance optimization within an
with die life/wear as a constraint mesh results as a good starting point. acceptable turnaround time.
b. Generalized formulation of multiple-step g. Use response surface mapping, sensitiv-
forging operations ity coefficients in conjunction with de-
c. Exploration of a larger design space by sign of experiments and design for six ACKNOWLEDGMENT
incorporating more geometric parame- sigma methodologies.
ters as design variables CAD related: The development of forging and heat treat-
d. Incorporation of additional optimiza- a. Set up initial geometry to ensure opti- ment optimization techniques was jointly funded
by DARPA through the RaDEO (Rapid Design 12. DEFORM Users Manual, Scientific Form- a Genetic Algorithm to the Optimal Design
Exploration and Optimization) project and the ing Technologies Corp., Columbus, OH, of the Die Shape in Extrusion, J. Mater.
General Electric Company. Engineous Software 1999 Process. Technol., Vol 72, 1998, p 6977
Inc. (ESI) was a team member of this project. 13. ABAQUS/Standard Users Manual, Hibbit, 28. G. Zhao, Z. Zhao, T. Wang, and R.V.
The author would like to thank Peter Finnigan, Karlsson, and Sorensen, Inc., Pawtucket, Grandhi, Preform Design of a Generic
Ramesh Gambheera, Beichang He, Peter Rhl, RI, 1997 Turbine Disk Forging Process, J. Mater.
and Youdong Zhou at the GE Global Research 14. MARC manuals, MARC Analysis Research Process. Technol., Vol 84, 1998, p 193201
Center for many helpful discussions and their Corporation, Palo Alto, CA, 1994 29. J.S. Chung and S.M. Hwang, Application of
contributions to this project. 15. V. Kumar, M.D. German, and S. Srivatsa, a Genetic Algorithm to Process Optimal
Design Optimization of Thermomechanical Design in Non-Isothermal Metal Forming,
Processes with Application to Heat Treat- J. Mater. Process. Technol., Vol 8081,
REFERENCES ment for Turbine Disks, presented at the 1998, p 136143
Manufacturing International Conference 30. M.S. Joun and S.M. Hwang, Die Shape
1. Proceedings of AIAA/NASA/USAF/ISSMO (Atlanta, GA), 1990 Optimal Design in Three Dimensional
Symposium on Multidisciplinary Optimiza- 16. Unigraphics User Function Programming Shape Metal Extrusion by the Finite Element
tion, 68 Sept 2000 Manual Version 11, Electronic Data Sys- Method, Int. J. Numer. Methods Eng., Vol
2. S. Kodiyalam, V. Kumar, and P.M. tems Corp., Unigraphics Div., St. Louis, 41, 1998, p 311335
Finnigan. Constructive Solid Geometry to MO, 1996 31. S. H. Chung and S.M. Hwang, Optimal
Three-Dimensional Structural Shape Opti- 17. D. Libes, Exploring Expect: A Tcl-Based Process Design in Non-Isothermal Non-
mization. AIAA J., Vol 30 (No. 5), 1992, Toolkit for Automating Interactive Prog- steady Metal Forming by the Finite Element
p 14081415 rams, OReilly and Associates, 1995 Method, Int. J. Numer. Methods Eng., Vol
3. E. Hardee, K.-H. Chang, K.K. Choi, X. Yu, 18. G.N. Vanderplaats, ADSA FORTRAN 42, 1998, p 13431390
and I. Grindeanu, A CAD-Based Design Program for Automated Design Synthesis 32. S.M. Byon and S.M. Hwang, Process
Sensitivity Analysis and Optimization for Version 3.00, VMA Engineering, March 1988 Optimal Design in Non-Isothermal Steady
Structural Shape Design Applications, Proc. 19. B. He, P.J. Rhl, R. Irani, S. Thamboo, and State Metal Forming by the Finite Element
Sixth AIAA/NASA/ISSMO Symposium on S. Srivatsa, CAD/CAE Integration with Method. Int. J. Numer. Methods Eng., Vol
Multidisciplinary Analysis and Optimization Application to the Optimal Design of 46, 1999, p 10751100
(Bellevue, WA), 46 Sept 1996 Turbine Disks, Proc. 39th AIAA/ASME/ 33. S. Roy, S. Ghosh, and R. Shivpuri, Optimal
4. L. Fourment and J.L. Chenot, Optimal ASCE/AHS/ASC Structure, Structural Dy- Design of Process Variables in Multipass
Design for Non-Steady-State Metal Forming namics and Materials Conference (Long Wire Drawing by Genetic Algorithms, J.
ProcessesI. Shape Optimization Method, Beach, CA), April 1998, p 27422749 Manuf. Sci. Eng., Vol 118, 1996, p 244251
Int. J. Numer. Methods Eng., Vol 39 (No. 1), 20. B. He, Y. Zhou, R. Gambheera, and S. 34. G.E. Totten, C.E. Bates, and N.A. Clinton,
1996, p 3350 Srivatsa. Turbine Disk Forging Process Handbook of Quenchants and Quenching
5. L. Fourment and J.L. Chenot, Optimal Optimization, 1999 ASME Design Engi- Technology, ASM International, 1993, p
Design for Non-Steady-State Metal Form- neering Technical Conf. (Las Vegas, NV), 3568
ing ProcessesII. Application of Shape 1216 Sept 1999 35. M. Batista and F. Kosel, Sensitivity
Optimization in Forging, Int. J. Numer. 21. B. He, Y. Zhou, R. Gambheera, and S. Analysis of Heat Treatment of Steel,
Methods Eng., Vol 39 (No. 1), 1996, Srivatsa. A Practical Approach to Forging AIAA-96-4152, 1996, p 14521460
5166 Optimization, paper AIAA-2000-4940, 36. R. Karthikeyan, P.R.L. Narayanan, and R.S.
6. E. Wright and R.V. Grandhi, A Shape AIAA/NASA/USAF/ISSMO Symposium Naagarazan, Heat Treatment Optimization
Optimization Technique for Controlling on Multidisciplinary Optimization, 68 for Tensile Properties of Al/SiCp Metal
Deformation Parameters in Forging, Proc. Sept 2000 Matrix Composites Using Design of Experi-
38th AIAA/ASME/ASCE/AHS/ASC Struc- 22. J.B. Yang and W.T. Wu, Scientific Forming ments, Process, Fabr. Adv. Mater., Vol 5,
tural, Dynamics, and Materials Conference Technology Corporation, personal commu- p 703711
(Kissimmee, FL), 710 April 1997, p nication of unpublished work 37. A. Saigal and G. Leisk, Taguchi Analysis of
14041414 23. S.K. Srivatsa, GE Aircraft Engines, unpub- Heat Treatment Variables on the Mechanical
7. Z. Gao, E. Wright, and R.V. Grandhi. lished work Behavior of Alumina/Aluminum Metal
Thermo-Mechanical Sensitivity Analysis 24. A. Srikanth and N. Zabaras, Preform Matrix Composites, Compos. Eng., Vol 5
for Preform Design in Metal Forming, Proc. Design and Shape Optimization in Metal (No. 2), 1995, p 129142
7th AIAA/USAF/NASA/ISSMO Symposium Forming Processes, Comput. Methods Appl. 38. P.J. Rhl and S.K. Srivatsa. A Comprehensive
on Multidisciplinary Analysis and Opti- Mech. Eng., Vol 190, 2000, p 18591901 Approach to Engine Disk IPPD, Proc. 38th
mization (St. Louis, MO), 24 Sept 1998, 25. A. Srikanth and N. Zabaras, A Gradient AIAA/ASME/ASCE/AHS/ASC Structural, Dy-
p 10191029 Based Optimization Approach for the namics, and Materials Conference and Ex-
8. D.J. Powell, Inter-GEN, A Hybrid Ap- Design of Single and Multi-Stage Metal hibit and AIAA/ASME/AHS Adaptive Struc-
proach to Engineering Optimization, Forming Processes, Proc. ASME Manufac- tures Forum (Kissimmee, FL), 710 April
Technical Report, General Electric Co., Feb turing Engineering Division, R.J. Furness, 1997, p 12501257
1991 Ed., MED-Vol 11, 2000, p 495507; pre- 39. T.C. Tszeng, W.T. Wu, and J.P. Tang,
9. iSIGHT Developers Guide Version 2.2., sented in the Symposium on Advances in Prediction of Distortion During Heat
Engineous Software, Inc., Raleigh, NC, 1997 Metal Forming, International Mechanical Treating and Machining Processes, Proc.
10. Engineous User Manual, General Electric Engineering Congress and Exposition 16th ASM Heat Treating Society Conf.
Corporate Research and Development, (Orlando, FL), 510 Nov 2000 Exposition (Cincinnati, OH) 1998, p 915
Schenectady, NY, 1995 26. G. Zhao, E. Wright, and R.V Grandhi, 40. J.V. Beck, B. Blackwell, and C.R. St. Clair,
11. H. Lee, S. Goel, et al., Toward Modeling Preform Die Shape Design in Metal Form- Jr., Inverse Heat Conduction: Ill-Posed
the Concurrent Design of Aircraft Engine ing Using an Optimization Method, Int. Problems, John Wiley & Sons, 1985
Turbines, presented at the International Gas J. Numer. Methods Eng., Vol 40, 1997, 41. Z. Li, R.V. Grandhi, and R Shivpuri,
Turbine and Aeroengine Congress and Ex- p 12131230 Optimum Design of Heat Transfer Coef-
position (Cincinnati, OH), May 1993 27. J.S. Chung and S.M. Hwang, Application of ficient During Gas Quenching Process
by Using Response Surface Method, Int. Frazier, A Continuum Lagrangian Sensitiv- 50. I. Doltsinis and T. Rodic, Process Design and
J. Mach. Tools Manuf., Vol 42, 2002, ity Analysis for Metal Forming Processes Sensitivity Analysis in Metal Forming, Int. J.
p 549558 with Application to Die Design Problems, Numer. Meth. Eng., Vol 45, 1999, p 661692
42. J.B. Yang and W.T. Wu, Scientific Forming Int. J. Numer. Meth. Eng., Vol 48, 2000, p 51. M. Kleiber and H.J. Antunez, T.D. Hien,
Technologies Corp., personal communica- 679720 and P. Kowalczyk, Parameter Sensitivity in
tion of unpublished work 47. Z.Y. Gao and R.V. Grandhi, Sensitivity Non-Linear MechanicsTheory and Finite
43. J.J. Tsay, J.E.B. Cardosa, and J.S. Arora, Analysis and Shape Optimization for Pre- Element Computations, John Wiley & Sons,
Nonlinear Structural Design Sensitivity form Design in Thermo-Mechanical Cou- 1997
Analysis for Path Dependent Problems, Part pled Analysis, Int. J. Numer. Meth. Eng., Vol 52. J.J. Tsay and J.S. Arora, Nonlinear
2: Analytical Examples, Comput. Meth. 45, 1999, p 13491373 Structural Design Sensitivity Analysis for
Appl. Mech. Eng., Vol 81, 1990, 209228 48. H.J. Antunez and M. Kleiber, Sensitivity Path Dependent Problems. Part 1: General
44. R.H. Myers and D.C. Montgomery, Response Analysis of Metal Forming Processes Theory, Comput. Meth. Appl. Mech. Eng.,
Surface Methodology: Process and Product Involving Frictional Contact in Steady Vol 81, 1990, p 183208
Optimization Using Designed Experiments, State, J. Mater. Process. Technol., Vol 60, 53. G. Venter, R.T. Haftka, and J.H. Starnes,
John Wiley & Sons, 1995, p 641653 1996, p 485491 Construction of Response Surfaces Approx-
45. S. Badrinarayanan and N. Zabaras, A 49. M.A. Ghouali, G. Duvaut, S. Ortola, and A. imations for Design Optimization, AIAA J.,
Sensitivity Analysis for the Optimal Design Oster, Local Analytical Design Sensitivity 36 (No. 12), 1998, p 22422249
of Metal Forming Processes, Comput. Meth. Analysis of the Forging Problem Using 54. Ansys 7.0 Manuals, Ansys Inc., 2002
Appl. Mech. Eng., Vol 129, 1996, p 319348 FEM, Comput. Meth. Appl. Mech. Eng., Vol 55. S. Akkaram, A. Makinde, Y. Zhou, and S.K.
46. N. Zabaras, Y. Bao, A. Srikanth, and W.G. 163, 1998, p 5570 Srivatsa, personal communication
1119 Index 7/29/03 4:16 PM Page 389

Index
A heating and temperature control, 160 section thickness, of rib-web forgings, 205 (T)
high-definition near-net-shape, 158 2017
ABAQUS (software), 349 high stacking-fault energy, 35, 148 extrusion rate vs. flow stress, 308 (F)
Abrasive wear, 288, 289 homogenization studies of, 123127 2024
Absolute temperature, 77 hot ductility and specimen geometry, 103 bar rolling of, 180
Accuracy, 230 for hot extrusion, 308 centerburst criteria, 318 (F)
Acetyl resins, 287 hot forging temperatures, 209 (T) Cockroft-Latham fracture criterion, 32 (F)
Activation energy, 26, 38, 39, 106 hydraulic presses for, 158 cold rolling, 181
Adhesive wear, 288 hydrogen pickup, 160 compression tests of, 180
Adiabatic heating, 96 impact-extrusion, 305, 306 compression tests on, 175 (F)
Adiabatic shearbands, 12, 19, 87 isothermal forging for, 147 ductility tests, 266
Adiabatic strain rate, 96 linear strain, 59, 60 (F) extrudability rating, 308 (F)
ADS module, 360 lubricants and friction coefficients, 237 (T), 314 extrusion rate vs. flow stress, 308 (F)
Advancing front triangulation methods, 328 (T) extrusion temperature and speed, 308 (F)
Algorithm steps, 355356 lubricants and lubrication, 159, 160 fracture limits, 54 (F), 56 (F), 59
Alligatoring mandrel forging, 160 fracture line limit, 182 (F)
of aluminum alloys, 267 (F), 268 (F) maximum dimension of web, 146 (F) fracture locus for, 176, 176 (F)
shape factor influence on, 267 open-die forging, 158 fracture strain lines, 181 (F)
shear stress at roll/metal interface, 268 orbital (rotary) forging, 159 in hot rolling, 55
studies of, 267 phosphate coating, 287 hot rolling, 181
susceptible alloys, 265 precision, 158 linear strain, 60 (F)
Alloys preforms, design guidelines for, 213 upset test representation, 60 (F)
temperature dependence of workability, 104 (F) processing map, composite, 17 (F) workability predictions for, 180181
workability dependence on a wide range of, 103 radiant heating for testing, 93 2024-T351
Alloys with an intermetallic second phase, 192 reverse or forward extrusion, 160 reduction ratio/energy capacity of, 32
Alloys with low melting point phases, 192 ring rolling, 160 2024-TO
Alpha-iron, 116 roll forging, 159 reduction ratio/energy capacity of, 32
Alpha parameter, 110, 111 second phase particles in static recrystallization, 2025
ALPID program (software), 16 36 forgeability and forging temperature of, 157 (F)
Aluminum and aluminum alloys, general shear friction factor, 210 forging temperatures for, 157
aluminum-lithium, 35, 35 (F) spin forging, 159160 2219
aluminum-magnesium, 105, 105 (F) structure of extruded parts, 305 forging temperatures for, 157
blocker-type, 158 torsion and plane-strain compression, 102 (F) 2618
cavitation affected, 80 torsion testing, 102 forgeability and forging temperature of, 157 (F)
closed-die forgings for, 158159, 159 (F) torsional ductility of, 105, 105 (F) forging temperatures for, 157
compatibility between materials and manufacturing typical defects in, 297 (F) 3003
processes, 141 (T) upset-forging, 159 extrudability rating, 308 (F)
composite processing map for, 32 (F) wire drawing die materials for, 289 (T) extrusion pressure requirements, 305 (F)
die-heating equipment for, 160 Aluminum and aluminum alloys, specific types extrusion rate vs. flow stress, 308 (F)
die-lubrication for, 160 pure forging temperatures for, 157
die temperature ranges, 158, 158 (T) flow behavior, 117, 117 (F) 4032
die types used for, 305 (F) grain size, 39 (F) forgeability and forging temperature of, 157 (F)
dimensions of preform and finished dimensions, high-stacking-fault-energy, 112 forging temperatures for, 157
205 (T) twist reversal on, 113, 113 (F) 4543
drawing lubricants and coefficient of friction val- 1050 extrusion rate vs. flow stress, 308 (F)
ues, 287 (T) extrusion rate vs. flow stress, 308 (F) 5052
equivalent stress-strain curves of, 88 1060 flow stress and strain to failure as a function of
as exception to deformation response, 99 extrudability rating, 308 (F) deformation, 126
extruded workpiece dimensions, 306 1100 ram displacement vs. surface temperature, 309
extrusion of, 124 extrudability rating, 308 (F) (F)
extrusion of, without lubrication, 146 (F) extrusion pressure requirements, 305 (F) rolling temperature for, 125
extrusion temperature classification, 292 (F) forging temperatures for, 157 5056
flow curves, 102 (F) ram displacement vs. surface temperature, 309 extrusion rate vs. flow stress, 308 (F)
flow-stress data for, 117, 117 (F), 124 (F) (F) 5083
flow stress-strain data of various metals, C and m 1350 deformation and failure of, 84 (T)
values, 383 (T) extrudability rating, 308 (F) edge cracking, 126 (F)
flow stress-strain data of various metals, K and n 2011 extrudability rating, 308 (F)
values, 382 (T) extrudability rating, 308 (F) extrusion temperature and speed, 308 (F)
forgeability behaviors of, 193 (F) 2014 flow stress and strain to failure as a function of
forgeability of, 157 extrudability rating, 308 (F) deformation, 127 (F)
forgeability ratings for, 157 (F) extrusion pressure requirements, 305 (F) forgeability and forging temperature of, 157 (F)
forging design limits, 209 (F) extrusion temperature and speed, 308 (F) forging temperatures for, 157
forging methods for, 157160 fillet and corner radii for, 204 (T) homogenization study, 126
forging temperatures for, 157, 157 (F), 157 (T), forgeability and forging temperature of, 157 (F) purity level study, 126
160 forging temperatures for, 157 soak period, 270
1119 Index 7/29/03 4:16 PM Page 390

390 / Reference Information
Aluminum and aluminum alloys, specific types surface cracking and alligatoring, 267 Backward (indirect) extrusion, 302, 302 (F)
(continued) tests on, 265 Backward slip, 233
strain to failure, as a function of homogenization 7079 Bainite-start temperature (Bs), 249
time, 127 (F) extrusion temperature and speed, 308 (F) Bainitic microstructure, 272273
studies for, 126 (F) forging temperatures for, 157 Bar rolling
5086 7150 of 2024-T351 aluminum, 180
extrudability rating, 308 (F) extrusion temperature and speed, 308 (F) alligatoring zones, 267 (F)
extrusion temperature and speed, 308 (F) 7178 centerburst formation zones, 267 (F)
5154 extrudability rating, 308 (F) controlled rolling, 255 (F)
extrusion rate vs. flow stress, 308 (F) 7475 cooling rate for grain refinement, 256
5182 cavitation failure in, 129 dynamic recrystallization, 255256
alligatoring, 268 (F) cavitation in, 81 (F) flow-stress behavior and torsion testing, 117
DC ingots, composition of, 125 (T) cavity radius, 82 (F) grain size, 255
ductility tests, 266 strain-rate sensitivity, 129 lead, 180
EMC ingots, composition of, 125 (T) superplastic forming, 132 recrystallization controlled rolling (RCR) in, 255
flow stress of, as a function of homogenization 8090 reheat temperatures, 255
time, 126 (F) flow curves of, 75 safe zones, 267 (F)
homogenization study, 125 true-stress/true-strain curve for, 76 (F) split ends, 269 (F)
no ductility problems, 266 (F) Al-5Mg time-temperature profile of, 255 (F)
secondary-tension test (STS) for, 52 reduction percentage at fracture, 268 (T) workability applications, 180181
strain to failure, as a function of homogenization Al-7Mg Barreling, 106
time, 126 (F) reduction percentage at fracture, 268 (T) coefficient, 63
surface cracking, 265 (F) Al-8Mg compression test, 48
surface cracking and alligatoring, 267 alligatoring, 267 control of, 48
susceptible to alligatoring, 265 Al-Mg alloy deformation process in, 49 (F)
torsion testing, 125 (T) strain to failure as a function of deformation mean effective strain for, 63
5252 temperature, 124 (F) in nonisothermal testing, 49 (F)
FEM for shear-strain, 97, 97 (F) Aluminum coatings, for tantalum and tantalum al- plane-strain compression test (PSC), 65
5456 loys, 157 in ring compression test, 53
extrusion temperature and speed, 308 (F) Aluminum killed steels, fine grain size of, 149 bcc (body-centered cubic) metals, 98, 99 (T),
6051 Analytic models, 13 102103, 103 (F)
extrusion rate vs. flow stress, 308 (F) Analytical equations (closed form), 10 Bend testing, 51, 321
6061 Analytical models, 325, 326 Bending process, 144
centerburst analysis in, 185 Anisotropy Bertella-Oliver equation, 24
disk compression test of, 185 effects of, 30, 175 Beryllium, 193 (F)
disk cracked in transition regime, 186 (F) parameter vs. local axial true strain, 77 (F) Bingham substance, 290
double extrusion forging of, 185 (F) in wrought alloys, 31 (F) Bishop and Hill approach, 42
extrudability rating, 308 (F) Annealing, 153, 278, 305, 317. See also heat treatment Blanking, 306
extrusion pressure requirements, 305 (F) ANSYS (software), 364 Blocker dies, 208
forgeability and forging temperature of, 157 (F) Apparent activation energy, 17, 77, 101 for aluminum alloys, 158
forging temperatures for, 157 Apparent cavity growth rate, 81, 82, 83 cavity of, 213
relative workability in forging and die-filling, Approach angle, 279, 280 design of, 216
194 (F), 210 (F) Approach semiangle, 280 die-filling, 218 (F)
6063 Approximate methods, 14 result and simulations of, 229 (F)
extrudability rating, 308 (F) Approximation, 355 section from, 219 (F)
extrusion rate vs. flow stress, 308 (F) Ar3 temperature, 249, 253 Body-centered cubic (bcc) metals, 98
7xxx series, 124, 125 (F) Arbitrary Lagrangian Eulerian (ALE) method, 223 textures and axial forces, 99 (T)
7001 Artificial intelligence techniques, 258 torsion and tension testing, 102103, 103 (F)
extrusion temperature and speed, 308 (F) Artificial neural network, 271 Bolt-heading process, 55, 179
7005 Ashby map (fracture mechanisms), 27 Boltzmann constant, 40
extrusion rate vs. flow stress, 308 (F) Ashby process selection chart, 141 (F) Boron, 272
7010 Aspect ratio, 142, 179 Boundary conditions, 13, 13 (F), 16, 204, 327
forgeability and forging temperature of, 157 (F) Atmosphere, 70 Boundary element method (BEM), 14, 15
forging temperatures for, 157 Austenite Boundary migration, 28, 41
7039 conditioning, 249 Boundary pinning, 270
forging temperatures for, 157 grain coarsening, 250 (F) Boundary value problems (BVP), 224
7049 grain diameter, 250 (F) Brasses, 147, 148
extrusion temperature and speed, 308 (F) grain-growth inhibitors, 9 Bridge-type dies, 292, 301, 307, 314
forgeability and forging temperature of, 157 (F) grain size, 254 (F) Bridgman correction factor, 24, 24 (F)
forging temperatures for, 157 interface area, 249 Brittle fractures, 27
7050 interfacial area, 250 (F) Bubble node, 225
forgeability and forging temperature of, 157 (F) metallurgy, 249 Bulging, 173
forging temperatures for, 157 microalloys and recrystallization stop temperature Bulk deformation process
7075 of, 251 (F) advantages of, 139
alligatoring, 267, 267 (F), 268 (F) recrystallization, 9 defects from, 3
cold open-die forging of recrystallized struc- recrystallization stop temperature of, 251 (F) defined, 3
tures, 200 rolling reduction on grains, 250 (F) described, 144145
ductile fractures, 200 temperatures, thermomechanical treatments on, free-surface fracture in, 59
ductility tests, 266 248 (F) general classification of, 3 (F)
edge cracking, 266 (F) Austenite to ferrite temperature (Ar3), 249, 253 workability diagram, 5 (F)
extrudability rating, 308 (F) Autocatalytic process, 19 Bulk workability testing. See also compression test-
extrusion pressure requirements, 305 (F) Average strain, 294 ing; hot tension testing; plane-strain compres-
extrusion rate vs. flow stress, 308 (F) Averaging functions, 353 sion testing (PSC); ring compression testing;
extrusion temperature and speed, 308 (F) Avitzur analysis, 6465 tension testing; torsion testing
finite element analysis (FEM) modeling, 200 Avrami-exponent, 38, 39 advanced analysis in, 25, 50
flow softening during hot working, 148 Axial-strain distribution, 79 bend test, 51, 321
forgeability and forging temperature of, 157 (F) Axial true strains, 58 compression testing, 48
forging temperatures for, 157 ductility testing, 50
free surface cracking of, 266 flow stress in compression, 49
relative workability in forging and die-filling, B plastic instability in, 50
194 (F), 210 (F) primary testing, 47
secondary-tension test (STS) for, 52, 52 (F) Back extrusion, 181183 specialized tests
surface and edge cracking, 265 (F) Back tension, 234 partial-width indentation test, 52
1119 Index 7/29/03 4:16 PM Page 391

Index / 391
secondary-tension test, 51, 52, 191 sulfides in, 149 forging temperature for, 150 (T)
test conditions for, 48 titanium in, 252 Hydrafilm process, 313
torsional rotation rates, 47 (T) transformation induced plasticity steels (TRIP), 4320
workability analysis, 5556 273, 274 forging temperature for, 150 (T)
Bulk working operations. See specific type of opera- vanadium as microalloy, 255 4340
tion void nucleation, 105 die housings, 305
Burgers (slip) vector, 37, 97, 97 (F), 251 volume fraction effect of, 28 (F) fillet and corner radii for, 204 (T)
Business models, 325 warm extrusion, 218 forging temperature for, 150 (T)
Buster dies, 208 wire drawing die materials for, 289 (T) forging temperature/pressure curve, 155 (F)
result and simulations of, 228 (F) workability limitations of, 58 forging temperature/upset reduction curve, 155
Carbon and low alloy steels, specific types (F)
0.05% C steel in hot torsion testing, 49 (F)
C work hardening exponent and strength coeffi- relative workability in forging and die-filling,
cient, 147 (T) 194 (F), 210 (F)
CA (cellular automata), 40 0.16 carbon section thickness, of rib-web forgings, 205 (T)
CAD tools (software), 348, 349, 353 (F) grain size, 114 workability of, 104, 104 (F)
CAE tools (software), 348, 349 0.25% low carbon 4340 (A)
Cahn-Hagel interface-averaged migration rate, 40 flow curves for, 113 (F) work hardening exponent and strength coeffi-
Caliber rolling, 242 0.6% C steel cient, 147 (T)
Cam plastometer work hardening exponent and strength coeffi- 4600 grade powder, 181
comparison with other hot working simulators, 122 cient, 147 (T) 4615
(T) 0.85% C (pearlitic, near-eutectoid) high carbon forging temperature for, 150 (T)
in compressive tests, 4849 flow curves for torsion and tension, 103 (F) 4620
for testing very high strain rates, 122 1006 cold forging of, 182 (F)
Canning, 299 truncated-cone indentation test, 201 (F) 5160
Carbides, 149, 165 1010 forging temperature for, 150 (T)
Carbon and low alloy steels, general forging temperature for, 150 (T) 6150
advanced high-strength steels (AHSS), 273 1015 forging temperature for, 150 (T)
Bridgman correction factor for, 24 forging temperature for, 150 (T) 8620
carbon content in, 255 truncated-cone indentation test, 201 (F) forging temperature for, 150 (T)
carbon content influence on, 149 1020 9310
cold worked, 147 forging temperature for, 150 (T) forging temperature for, 150 (T)
compatibility between materials and manufacturing forging temperature/pressure curve, 155 (F) AF 1410
processes, 141 (T) forging temperature/upset reduction curve, 155 deformation resistance, 73
complex phase steels (CP), 273 (F) ductility, 73
deformation-time-temperature schedule, 118 (F) fracture limits, 59 dynamic recrystallization, 75 (F)
development of, 273274 hot-forming pressures, 163 (T) Gleeble curve for, 75 (F)
drawing lubricants and coefficient of friction val- linear strain, 59 (F) hot working curves, 73
ues, 287 (T) 1030 microstructure of, 74
dual-phase steels (DP), 273, 274 forging temperature for, 150 (T) on-cooling Gleeble curves, 74 (F)
ductility influenced by carbon content, 105, 105 relative workability in forging and die-filling, HY-100 steel
(F) 194 (F), 210 (F) true-stress/true-strain rate curves for, 63 (F)
dynamic recrystallization, 343 (F) 1040 Cast ingots, 29
extrusion temperature classification, 292 (F) forging temperature for, 150 (T) Cast iron, 141 (T)
and flow stress, 147 in hot torsion testing, 49 (F) Casting process, 125
flow stress-strain data of various metals, 379 (T) workability of, 104, 104 (F) Cauchy stress tensor, 223
flow stress-strain data of various metals, C and m 1045 Cavitating materials, 84 (T)
values, 380382 (T) bilinear strain, 60 (F) Cavitation map, 129
forgeability behaviors of, 193 (F) cold finished steel, bilinear strain, 60 (F) Cavities and cavitation
forgeability of, 149 forging simulation, 228 architecture, 83
forging flash and width-thickness ratio by forging fracture limits, 59 cavity shape and distribution, 83
weight, 212 (F) fracture locus for, 176, 176 (F) in certain metals, 80 (F)
forging lubricants, 150 model study, 228 coalescence, 80, 8182
forging, pressure vs. forging height, 215 (F) truncated-cone indentation test, 201 (F) continuous cavity nucleation, 82
forging temperature, 149, 150 (T) upset test representation, 60 (F) curves for, 83
hot forging, 209 (T) 1050 described, 86, 129
upset reduction curve, 155 (F) forging temperature for, 150 (T) effect of stress state on, 197 (F)
grain size in, 255 1055 effects on stress-strain behavior, 82
high-strength low alloy, torsion testing, 117 forging simulation, 228 effects on tensile ductility, cavitation, 83
hot ductility, in torsion, 108 model study, 228 finite element analysis (FEM), 43
hot ductility in torsion, 108 1060 formation
hot ductility, in torsion, 108 (F) forging temperature for, 150 (T) as a function of temperature and strain rate, 68,
for hot extrusion, 308 1070 80
hot forging temperatures, 209 (T) forging temperature for, 150 (T) from hot-tension testing, 71, 80
intrinsic ductility material constant, 108 1080 growth mechanisms, 81
low carbon forging temperature for, 150 (T) growth of individual cavities, 80
flow behavior under hot working conditions, 37 1095 growth rate, 81 (F), 83, 83 (F)
(F) forging temperature for, 150 (T) in hot-tension testing, 79
flow curves for torsion and tension, 103 (F) 1130 during hot-tension testing, 8083
lubricants for, 58, 150, 237 (T), 314 (T) relative workability in forging and die-filling, in hot-working, 68
martensitic steels (MART), 273 194 (F), 210 (F) materials affected by, 80
maximum dimension of web, 146 (F) 1.3%C nucleation, 35, 80, 81
medium carbon, 8 (F), 59 reduction percentage at fracture, 268 (T) nucleation rate, 83
microalloyed, 118 (F), 150151 1541, 151 radius, predictions for, 82 (F)
microstructures, true stress of, 147 (F) 1.5%W steel reduction of
microstructures, work-hardening exponent and reduction percentage at fracture, 268 (T) by homogeneous compression, 80
yield strength, 147 (F) 4130 by isostatic pressing, 80
niobium-vanadium microalloy, 118 (F), 119 for aluminum extrusion, 304 (F) stationary cavity array, 82
(F) forging temperature for, 150 (T) upper and lower limits of, 82
powder preform design, 181 micrograph of seam with oxide, 9 (F) Cavity volume fraction, 80, 81
preforms, design guidelines for, 213 4140 CDRX (continuous dynamic recrystallization), 35
resulfurized steel, 193 (F), 194 (F), 210 (F) for aluminum extrusion, 304 (F) Cellular automata (CA), 40
shear friction factor, 210 flow lines in, 31 (F) Cementite (Fe3C) lamellae, 115
1119 Index 7/29/03 4:16 PM Page 392

Centerbursts defined, 291 backward tracing methods, 368

analysis, 178, 185 described, 303, 306 boundary conditions, 371
causes of, 6 extrusion lengths of, 306 business-engineering integration, 329
centerline-porosity, 318 extrusion pressure, 306 costs, 369
Cockroft and Latham criterion, 318 lubricants and lubrication, 146, 306 cross-functionality of, 329
cold extrusion, 317 (F) problems and potential causes, 303 (F) design objectives and constraints, 369
cuppy-core fractures, 318, 319 quality of, 305 design parameters for forging, 374 (F)
defects from, 6 stock for slugs, 306 design sensitivity analysis, 368
in drawing, 282 (F) surface preparation, 306307 design verification, 375
electrolytic tough pitch copper, 318 three-step, 303 (F) die-velocity, design, 372374
during extrusion, 183184 tooling for, 303304 die velocity, flash cost, 374375
in extrusion, 282 (F) Cold forging, 218 effective strain-rate error, 375 (F)
in forgings, 185 analysis of, 371 finite-difference scheme, 368
formation, 178 (F), 179 (F), 281282 defined, 191 finite element-based sensitivity analysis, 368
fractures, 19, 32 described, 147 finite element method (FEM), 368
height-to-length (h/l) parameter, 317 lubricants for, 147 forging, 374 (F)
lower-bound method, 318 preform design, 181 friction model, 371
mapping functions, 185 (F) temperatures, 191 global-optimum solutions, 329
matrix method, 318 Cold heading, 303 independent parameters, 369
shape factor influence on, 267 Cold (impact) extrusion, 305, 306, 307 integral blade and rotor forging (IBR), 329
slip-line method for plastic flow, 318 Cold open-die forging, 199, 200, 201 iterative analyses, 329
studies of, 267, 318 Cold rolling, 232 mapping functions, 370
upper bound method, 318 2024 aluminum alloy, 181 material response function, 369
Central composite rotatable design, 133 edge preparation and stress levels, 268 merit functions, 369
Ceramic coatings, 289 elastic-plastic finite element method (FEM), near-net-shape forming, 368
Ceramic glass precoats, 169 262 nonlinear finite element analysis, 368
Chevron cracking. See centerbursts spalling, 237 optimal die-velocity schedule for, 374 (F)
Chilling during rolling, 191 Cold shut (term), 7 optimal shape compared with forging, 375 (F)
Chromating, 287 Cold upset testing, 5, 57, 5859 parametric blending functions, 370
Circumferential true strains, 58 Cold working operations performance constraints, 369
Closed-die forging bend test for, 51 preform shape description, 369
for aluminum alloys, 158159, 159 (F) Cockroft criterion, 178 preform shape sensitivity, 371372
applicable cold die forging tests, 201 deformation processes, 147148 process design automation, 329
axisymmetric deformation, 188 described, 145 Simpsons method, 371
defects, 201, 202 (F) and grain size, 147 as software robot, 329
deformation sequence, 202 (F) process of, defined, 3031 upstream process design, 329
described, 146147 strain-hardening in, 31 Computer processing time, 327, 328
distribute gage volume (DGV) test, 202 Cold working temperature, 61, 101102 Computer run times, 356357
guide to methods for, 128 (T) Collaborative optimization environment (COE), 349 Conical dies, 280281, 292
impression dies, 188 Collar test, 200 (F) Conical plug, 283
investigative methods, 202203 Complex behaviors, analysis for, 35, 95 Considres construction, 63, 64 (F)
laps from, 202 (F) Composite rotatable design, 133 Considres criterion, 82
load-stroke curve for, 215 (F) in thermomechanical testing, 133 (T) Constant shear friction model, 220
near-net-shape forming in, 158 Compression, frictionless (ideal), 55 Constant-shearing friction factor, 371
of nickel-base alloys, 163 Compression stress-strain curves, 63 Constant-strain-rate isothermal compression testing,
plane-strain deformation, 188 Compression testing. See also ring compression test- 199
stainless steel, 151 ing; upset testing Constant stress/strain elements (CST), 225
stress distribution in, 216 (F) advantages and disadvantages of, 48 Constant true-strain rate, 49
Closed-form analysis, 1314 barreling, 48 Constitutive equations, 26
Coalescence stage of ductile fractures, 28 of certain metals Constitutive relations, 33 (F)
Cobalt and cobalt alloys, general 2024 aluminum alloy, 180 Constrained flow stress, 27
extrusion temperature classification, 292 (F) 2024-T35 aluminum alloy, 175 (F) Constraints, 337, 342
Cobalt and cobalt alloys, specific types JBK-75 specimen, 204 (F) averaging functions, 353
HS 188 deformation to fracture vs. mean hydrostatic stress for continuity, 358
deformation resistance of, 73 and flow stress, 293 (F) design objectives and constraints, 369
ductility of, 73 die chilling in, 49 design variables, zones and constraints, 356 (F)
hot working temperature range of, 73 engineering strain rate, 48 discontinuous step functions, 353
on-cooling Gleeble curves, 74 (F), 75 (F) on flanges and prebulged specimens, 175 (F) equipment constraints, 352
Cockcroft and Latham criterion, 106108 free surface strains, 176 (F) geometric, 352, 358
centerburst failure, 32 (F), 318 free-surface fracture, 320 lateral constraints, 205
described, 12 friction effects of, 175 material constraints, 352
for fracture, 319 friction hill, 320 minimum and maximum functions, 352
matrix method, 318 friction reduction in, 48 multidisciplinary optimization (MDO), 349
slip-line method for plastic flow, 318 frictionless, 175 nonholonomic constraints, 333
Cockcroft criterion, 177, 178, 178 (F) geometry effects, 175 in optimal design, 352
Coefficient of friction for hot forging operations, 48 optimization, 360
Coulombs coefficient of friction, 31, 50, 64 instability in, 63 performance constraints, 369
drawing, 286, 287 (T) isothermal conditions in, 49 relaxed constraints, 4243
extrusion, 314 (T) machine control for constant strain in, 62 smoothness, 353
shear stress, 298 (F) for measuring flow stress, 48 Contact condition, 223224
Coefficient of thermal expansion (CTE), 128, 129, reduced-section test, 204 (F) Contact surface fractures, 19, 183
130, 130 (F) strain softening in, 63 Contact surfaces and internal fracture, 183
Cogging, 132 tapered prebulged, 54 (F) Contact time, of forging equipment, 210
Coilers, 272 test specimens, 53 (F), 54 (F) Continuous dynamic recrystallization (CDRX), 35
Cold and warm forging, 218219 true strain-rate in, 48 Continuum mechanics, 10, 11
Cold and worm working temperatures, 101 true strains in, 62 Continuum plasticity, 12
Cold backward extrusion, 302 (F) Compression testing system, 61 (F) Control theory, 341
Cold extrusion, 218, 219 (F) Compressive loading, 50 (F) Controlled rolling
annealing, 305 Compressive stress state, 6 austenite conditioning, 249
blanking, 306 Compressive tests, 4849 defined, 248
centerbursts, 317 (F) Computational efficiency, 230 fundamentals of, 248
and cold heading, 303 Computer aided optimization (CAO) grain coarsening, 250251
of copper alloys, 306 activity based models, 329 microalloying, 249
1119 Index 7/29/03 4:16 PM Page 393

Index / 393
recovery and recrystallization, 251252 forgeability of, 160 Cyclic flow curves, 114
restoration processes, 250 heating time, 161 Cylindrical compression testing
Convection heat, 226 brass interface friction factor in, 63
Conventional controlled rolling (CCR), 248 forgeability of, 160 isothermal stress-strain curve, 62
described, 253 C10200 lubrication for, 61
development of, 252253 cold extrusion, 306 testing apparatus, 61
intensified controlled rolling, 253254 forgeability of, 160 Cylindrical plug, 283
niobium, effect of, on Ar3 temperature, 253 forgeability ratings of, 161 (T)
suppression of recrystallization, 253 C18200
Conventional extrusion, 291 forgeability ratings of, 161 (T) D
Conventional hot rolling (CHR), 248 forging temperatures, 161 (T)
Conversion coatings, 147, 284, 287 C26000 Damage model, 271
Conversions cold extrusion, 306 Damage modeling, 134
shear-stress/shear-strain conversion to effective C37700 Damage parameter, 134
stress-effective strain, 99, 101, 102 forgeability of, 160 Dead-metal zones (DMZs), 2829, 50, 58, 260
torsion data to effective stress and effective strain, forgeability ratings of, 161 (T) Dead zone, 178 (F)
103 forging temperatures, 161 (T) Defects. See also cavities and cavitation; center-
Cooling history, 116 heating time, 161 bursts; surface fracture
Cooling rate, 151 C46400 in aluminum and aluminum alloys, 297 (F)
Cooling tables, 272 forgeability ratings of, 161 (T) avoidance in dynamic material modeling (DMM),
Copper and copper alloys, general forging temperatures, 161 (T) 340341
Burgers (slip) vector, 97 C64200 from bulk deformation process, 3
cavitation affected, 80 forgeability ratings of, 161 (T) causes of, 201, 210
closed-die forging, 161 forging temperatures, 161 (T) deformation bands, 250
cold extrusion, 306 C67500 from deformation processing, 19
cold worked, 147 forgeability ratings of, 161 (T) from die-contact cracking, 6
compatibility between materials and manufacturing forging temperatures, 161 (T) example of analysis for, 216
processes, 141 (T) Coronze 638 extrusion defect in rib, 8 (F)
copper-nickel alloys, 161 deformation and failure of, 84 (T) finite element method (FEM) modeling of, 227
die materials for, 161, 162 (T) Cu-30Zn brass flow-related, causes of, 7
die temperature ranges for, 161 lead content in, 161 forging, 7 (F), 8 (F)
dies, 161 Cu-40Zn brass during forging-temperature selection, 191
drawing lubricants and coefficient of friction val- lead content in, 161 in forming electrode, 227 (F)
ues, 287 (T) Corrosion resistance, 9 in hot-working, 68
effects of temperature and strain rate history, Corrosive wear, 288, 289 non-metallic inclusions, aluminum-killed steels,
116 Cost control and savings 305
electrolytic tough pitch, 95, 95 (F), 96 on material, 348 plasticity analysis, 178
extrusion temperature classification, 292 (F) from modeling techniques, 220, 346347, 359 (F) rolled bar imperfections, 4 (F)
flow curves, 102 (F) reduction in manufacturing, 151 in specimen surfaces, 175
and flow stress, 147 weight reduction strategy, 348 in specimens, 175
flow stress-strain data of various metals, K and n Costs statistical uncertainties of, 376
values, 384 (T) computer aided optimization (CAO), 369 types of
forgeability ratings of, 161 (T) of computer analysis, 364, 368 bamboo defects, 297, 305, 311
forging methods for, 161162 control of, by optimal design, 337 blistering, 300
forging of, 161 (F) die life, 217 central discontinuity in, 6 (F)
forging severity, 161 of dies, in drawing, 146 circumferential, 316
forging temperatures for, 161 of dimensional tolerances, 143 cold shuts, 201
grain size, 114 of distortion, 359 crows feet, 316, 317, 317 (F)
heating practices, 161 equipment utilization, 143144 dislocations, 35
heating time, 161 of flash, in design, 374 edge cracking, 264
for hot extrusion, 308 of heat treatment, 151 extrusion-type, 7, 204 (T)
hot forging temperatures, 209 (T) labor, 144 in extrusions, 201
leaded-copper zinc alloys, forgeability of, 160 from material weight savings, 359 (F) fir-tree cracking, 173, 186187, 297, 316
low stacking fault energy, 112 operating, 143 flow-through, 7, 8 (F), 201, 204 (F)
lubricants and friction coefficients, 237 (T) and ring rolling, 162 fold and underfill, 357 (F)
lubricants and lubrication, 161 of shop-floor trials, 364 forging burst, 6 (F)
lubricants for extrusion, with coefficient of friction of tooling, fixtures and equipment, 144 intragranular defects, 250
values, 314 (T) Coulombs coefficient of friction, 31, 50, 64 laps, 7, 7 (F)
maximum dimension of web, 146 (F) Coulombs friction model, 220, 371 laps and folds, 201
naval brass, heating time, 161 Coulombs law of friction, 3132, 224 line, 35
oxygen-free high conductivity, flow curves for, Cracks and cracking longitudinal, 316
102, 102 (F) detection of, 58 orange peel, 58, 297, 317
radiant heating for testing, 93 and net-shape forging, 149 pipe/piping, 4 (F), 6, 297, 300
ring rolling, 162 nucleation, 19 seams, 8 (F)
shear friction factor, 210 shear cracking and flow softening, 148 silicate inclusions, 305
silicon bronzes, 161 stability criterion, 19 splitting, 316, 317 (F)
temperature dependence of axial forces, 98 (F) of stainless steels, martensitic, 152 suck-in, 7
tensile fracture in torsionally prestrained speci- temperature limits, 167 transverse cracking, 317
mens, 87 (F) from thermal cycling, 217 (F) twins, 250
textures, 98, 98 (F), 99 (F) Crankshaft hot forging process, 228 DEFORM (commercial FEM code), 217, 218
torsion tested, 116 (F) Creep forming, 149 DEFORM (software), 262
upset forging, 162 Critical roll speeds, 241 described, 349
wire drawing die materials for, 289 (T) Cross-section area reduction, 30 multidisciplinary optimization (MDO) techniques,
work hardening exponent and strength coefficient, Cross slip climb, 28 348, 349, 350, 362
147 (T) Crosshead, 7778 preform design, 357
Zener-Holloman parameter (Z) for, 114 Crosshead velocity, 25 Deformation
Copper and copper alloys, specific types Crystal-plasticity FEM, 43, 98 tension and compression, 172 (F)
70/30 brass (A) Crystal structure, 192 time-to-peak stress, temperature of, 252 (F)
work hardening exponent and strength coeffi- Crystallographic texture, 87 Deformation heating, 87, 108
cient, 147 (T) Current shear stress, 95 adiabatic strain rate, 96
alpha/beta brass Customer pulling value, 326 causes of, 112
deformation and failure of, 84 (T) Cutting modeling, 228229 correction factor for, 62
aluminum bronze Cycle time, 143 ductility impacts of, 104
1119 Index 7/29/03 4:16 PM Page 394

Deformation heating (continued) closed-die forging, 19 preheated, 198

flow softening from, 61 considerations, 293294 roller dies and rolling, 282285
softening causes, 93 die-bearing lengths, 294 roller dies configuration, 282 (F)
Deformation mechanisms in extrusion, 295 shaped dies, 307
grain-boundary sliding, 76 extrusion force, 295 shaving dies, 282
power-law creep, 148 hollow sections, 294 shaving/drawing, 282 (F)
slip, 76 hot extrusion, 308 shear face dies, 310
Deformation modeling, 1314 hydrostatic extrusion, 19 spider, 301, 314
Deformation parameters, 114 for magnesium alloys, 307 sticking friction, 310
Deformation patterns, 201 (F) materials for, 289 testing of, 208
Deformation processes multidisciplinary optimization (MDO), 346 types of, 208
advantages of, 139 (T) optimal die design, 293 worn, section of, 288 (F)
analysis methods for, 326 (T) pack rolling, 19 Differential equations, 11
bulk, characteristics of, 146 (T) relief, 294 Differential scanning calorimetry (DSC), 130
costs of, 139 shrinkage allowance in, 308 Differential testing, 95
design capability ratings for analysis, 326 (T) split die principle, 295 Differential thermal analysis (DTA), 130
design decisions, 139, 139 (T), 140 thin-wall sections, 294 Diffusion equations, 11
inhomogeneity, 235 (F) for titanium forgings, 217218 Diffusion flow, 148
in plane strain, 232 tool strength analysis, 294 Diffusion growth, 81 (F)
potential disadvantages with, 139 (T) Die forging. See also closed-die forging; open-die Digital-to-analog (D/A) converter, 92
primary working processes forging Dimensional tolerances, 143
bar rolling, 117 defect prediction techniques, 205 Dimensions
drawing, 146 die fill determination, 369 (F) of preforms, 213
extrusion, 144 laps in, 7 test specimens, 189
forging, 146 lateral constraints, 205 Direct-equilibrium method
open-die forging, 144 schematic of, 209 (F) algorithms for cavity volume fraction, 80
plate rolling, 117 tests for parameters of, 205 analysis of cavitation effects, 83
rolling, 144, 145, 232 Die friction effective strain in, 79
sheet rolling, 117 in extrusion, 178 effective strain rate in, 79
tube piercing, 144 measurement of, 296 (F) effective stress in, 79
process selection charts, 140 in wire drawing, 178 engineering stress-strain curves, 80 (F)
process variables, 341 Die friction factors, 57 formulation of, 79
reduction percentage at fracture, 268 (F) Die geometry. See also die angles; die design hot-tension testing, parameters predicted, 79
simulation results, 328 (F) as constraint in optimal design, 352 normal plastic anisotropy parameter, 80 (F)
of superalloys, 154 drawing, 278, 281 for partial differential equations, 14
temperature variations in, 341 land length, 286 specimen geometry in, 79
types of, 144145 multidisciplinary optimization (MDO), 348 strain-hardening exponent in, 79
workability limit, 150 (F) secondary tensile stress, 189 strain-rate sensitivity, 80 (F)
Deformation-processing map, 86 (F) Die-heating equipment, 160 strain-rate sensitivity index in, 79
Deformation rate, 68 Die life and wear, 217, 288 (F), 288289 strength coefficient in, 79
Deformation resistance, 73 Die lubrication uniaxial testing of sheet materials, 80
Deformation textures, 98 for aluminum alloys, 160 Direct (forward) extrusion. See forward (direct)
Deformation-time-temperature schedule, 118 (F) for copper and copper alloys, 161 extrusion
Deformation values, 76 for stainless steels, 153 Direct resistance, 68
Deformation work per unit volume (h), 96 Die materials, 288 (T) Direction of grain flow, 9
Delaunay triangulation method, 327 for copper alloys, 162 (T) Discontinuities, 3, 6
Density, 128, 129, 130 for copper and copper alloys, 161, 162 (T) Discontinuous dynamic recrystallization process
Descaling and rough rolling, 272 die inserts, 304 (DDRX), 36, 36 (F), 37, 113114
Design and tolerances, 159 (T) punches, 304 Discontinuous step functions, 353
Design approach, 341 shrink rings, 305 Discretization, approximation, and searching opti-
Design considerations, 10 Die pickup, 290, 310 mization (DASA)
Design for manufacturing (DFM), 140 Die temperature ranges algorithm steps, 355356
Design optimization tool (DOT), 362, 373 for aluminum alloys, 158, 158 (T) approximation, 355
Design variables, 362 for copper and copper alloys, 161 computer run times, 356357
Development schedule, 350351 for titanium alloys, 169 (T) design variables, zones and constraints, 356 (F)
Die angles Die tryout, 208 discretized design space, 355
approach angle, 279, 280 Die underfill, 19 flow chart of, 356 (F)
approach semiangle, 280 Dies, 161 gradient-based methods, 355
centerbursts, 178, 183184 bleeder hole, 301 initial and optimal designs, 357 (F)
critical die angles, 318 bridge-type, 292, 301, 307, 314 interpolation point, strategies for selection of, 356
dead metal zones, 178 ceramic coatings, 314 (F)
die angle selection, 304 choke and relief in bearing, 294 (F) local optimums, 355
draft angle for, 161 conical modeling methods, 356357
entry angle, 286 drawing stress, 281 neighborhood searching, 355
in extrusion, 178 flow through, 280281 Discretized design space, 355
optimal half angle, 279, 280 power consumption, 281 Disk compression testing, 185
upper-bound method, 184 conical die entry, 294 (F), 295 (F) Dislocation climb, 28
variation in mode of flow, 178 (F) double-reduction dies, 313 (F), 313314 Dislocation density, 37
in wire drawing, 178 extrusion, 294 (F) Dislocation glide, 99
Die-chill casting (DC), 125 extrusion, for aluminum, 305 (F) Dislocation glide and climb, 148
Die chilling finisher Dislocations, 35
in compression test, 49 choosing flash width and thickness, 211212 Dissipator power co-content, 33
effects of, 50 empirical formulae, 212 Distortion, 359
simulation of, in multiphase alloys, 117 flash design in closed-die forging, 211 Double-cone test
Die compression flash dimensions, 211 establishes microstructure evolution during hot
disk cracked in transition regime, 186 (F) finite element method (FEM) analysis of, 208 working, 195
surface strains vs. surface conditions, 186 (F) flat-face dies, 310 samples, 191
Die contact fracture, 6, 185186 heating of, 161 secondary tensile stresses, 195
Die contact surfaces, 173 hot-die forging, 147 specimen geometry for, 191 (F)
Die design, 147. See also die angles; die geometry hot extrusion, 310 Double extrusion forging, 185 (F)
approach semiangle, 280 nitriding dies, 314 Double indentation, slip line fields, 178 (F)
aspect ratio in, 179 physical modeling of, 208 Double reduction die, hydrostatic extrusion, 313 (F),
choke, 294 porthole, 301, 314 313314
1119 Index 7/29/03 4:16 PM Page 395

Index / 395
Draft angles for dies, 161 collar test, 200 Dynamic recovery (DRV), 35 (F), 38, 61, 87, 108
Draft schedule, 271 defined, 27, 199 causes of, 28
Drawing die described, 86 cavity nucleation in, 35
back pull (back tension), 280 dimpled rupture mode, 28 (F) constitutive equations for, 12
coefficient of friction in, 286 finite element analysis (FEM) modeling, 200 cross slip climb in, 28
die pickup, 290 grooved compression test, 200 described, 112
die/wire interface pressure, 279, 280 gross centerbursts, 200 dislocation density, 37
differential strain hardening, 279 hydrostatic stress, 172 ductility in, 35
reduction, limits to, 279 intergranular fracture, 270 dynamic spheroidization, 112
secondary tensile stresses, 279 locations of, 263 flow curves, 37
use of, 278 mechanical anisotropy effect of, 58 flow-stress curves of, 148 (F)
Drawing process mechanism at cold various temperatures, flow-stress levels with, 75
annealing, 278 103104 healing of, 104
break failures, causes of, 319 mechanisms of, 57 during hot deformation, 148
bulge formation conditions, 282 (F) micromechanisms of, 172 in hot-working, 35, 145
centerburst, 185, 282 (F) secondary tensile stress, 200 materials experiencing, 28
deformation processes, 146 surface strains, 200 softening causes, 93
described, 146, 278 truncated cone test, 200 strain hardening balances with, 75
die costs of, 146 upset test for, 199 strain-induced precipitation, 112
die geometry, 278, 281 Ductile materials, 52, 58 strain-rate sensitivity (m) and, 19
draw forces in, 279 Ductile rupture, 28 stress-strain curve for, 28 (F)
drawing of bar and wire Ductility torsion flow curves with hot working materials,
approach angle, 280 alloying effect on, 103104 113
centerbursts, 281282 as basis of optimal working temperatures, 108, 108 and work hardening, 148
determination of friction, 280 (F) Dynamic recrystallization controlled rolling (DRCR),
drawing of profiles, 282 of certain metals, 73 254
external deformation and shaving, 282 changes to, causes by, 104 Dynamic recrystallization (DRX), 108. See also con-
flow through conical dies, 280281 defined, 22, 24 tinuous dynamic recrystallization (CDRX); dis-
relative drawing stress and power consumption, deformation heating, 104 continuous dynamic recrystallization process
280281 dependence upon temperature, 105 (DDRX)
residual stress, 281 determinants of, 19 AF 1410, 75 (F)
strain distribution, 281 determined in hot torsion tests, 104 (F), 105 in bar rolling, 255256
stresses and forces, 279 in dynamic recovery, 35 boundary migration from, 28
engineering compressive strain (reduction) in, effect of bore diameter and wall thickness, 103 (F) constitutive equations for, 12
279 engineering strain to fracture, 24 dynamic spheroidization, 112
in fracture models, 177 of ferrous alloys, 103 flow curves of, 37
friction, determination of, 280 as fracture resistance, 4 flow-stress curves of, 148 (F)
friction in, 278, 279 as a function of sulfide content, 105 forms of, 253
heat generation in, 280 grain growth, 104 grain size, 114
height-to-length (h/l) ratio, 282, 321 hot, in torsion, 108, 108 (F) healing of, 104
interface pressure, pressure-multiplying factor, 279 hot shortness, 104 during hot deformation, 148
interface shear strength, 280 hot-tension testing for low to moderate, 54 in hot-working, 145
limiting reduction per pass, 321 in hot torsion testing, 49 (F) initiation level of, 37
lubrication influence of axial stress on, 107, 108 materials experiencing, 28
dry and wet drawing, 284 intergranular fracture, 104 microstructure development, 112
dry drawing, 284 intrinsic or in simple shear, 108 models of, 40, 41
hydrodynamic, 285286 loss of, 265 retardation of, 252
mixed-film lubrication, 286 measure of, 24, 104 and rolling load, 254
plastohydrodynamic (full-fluid-film) lubrication, mechanical anisotropy, 9 softening causes, 93
284 optimum temperature of, 104 from stores of dislocations, 112113
solid-film lubrication, 286 parameters predicted by FEM, 79 strain-hardening rate, 28
surface preparation, 287288 temperature effect, 103104, 105 strain-induced precipitation, 112
tube drawing, 286 from tension testing, 24 strain-rate sensitivity (m) and, 19
material during, 175 (F) in torsion test, 24 stress-strain curve for, 28 (F)
minimum drawing stress, 280 Ductility data, effects of specimen geometry variation temperature-strain rate conditions and, 37
multipass-drawing, 321 on, 88 texture effects of, 99
profile drawing, 282 Ductility peak, 105 torsion flow curves with hot working materials, 113
redundant work in, 280 Ductility ratings, 72 Zener-Holloman parameter, 114
roller dies and rolling, 282 Ductility trough, 264, 270 (F) Dynamic restoration process, 104
shaving, 282 Dwell time, 132, 133 Dynamic spheroidization, 112
sliding velocity in, 279 Dyna (software), 262 Dynamic strain aging, 18
slip line fields, 179 (F) Dynamic and preheat temperature influence, 190
speed limits of, 284 Dynamic material modeling (DMM)
temperature of, 278 about, 17 E
true strains in, 279 defect avoidance, 333
tube drawing, 283284 described, 3334, 340 Economics. See costs
wear development of, 17 Edge cracking
die wear, 289 die design, 334 of 7075 Al alloy, 266 (F)
wear mechanisms, 288289 equipment, choices of, 333 of low carbon steels, 267
wire surface condition, 284 of equipment systems, 338 during rolling, 173
Drawing stress flow stress, 17 from secondary tensile strains, 265266
calculations of, 319 forging, 334 and strain-to-failure, 127
limits to, 279 manufacturing plan, process and inspection trav- testing for resistance to, 126
DRV (dynamic recovery). See dynamic recovery eler, 335 Edge preparation and stress levels, 264, 265, 266,
Dry drawing, 284, 285 metallurgical interpretation of, 17 267, 268
DSC cell (differential scanning calorimetry), 130, microstructure, 17 Effect of strain rate and temperature, 25
130 (F) microstructure based models, 333 Effective combined stress. See equivalent combined
DTA cell (differential thermal analysis), 130, 130 modeling techniques, 340 stress
(F) nonholonomic constraints, 333 Effective gage, 95
Dual-fracture loci, 177, 178 process variables definition by, 333 Effective gage length, 95
Ductile fracture map, 16 (F) processing maps, 17, 33, 33 (F) Effective radius, 112
Ductile fractures, 57 stress analysis, 334 Effective strain
causes of, 269 Dynamic processes, 28 defined, 26, 32
1119 Index 7/29/03 4:16 PM Page 396

Effective strain (continued) Erhardt-Push-Bench-Method, 283 temperature for, 291, 317

in direct-equilibrium method, 79 Etching, 9 temperature increase of billet, 296
von Mises, 88, 99 Eulerian calculation, for chip formation, 230 (F) temperature vs. ram speed, 309 (F)
Effective strain distribution, diagrams of, 262 (F), Eulerian method, 223, 229 tolerances, 307
263 (F), 264 (F) Expect (software), 350 tool steels for extrusion of, 304 (F)
Effective strain-rate, 79, 108 Explicit modeling, 222 tooling components, 304
Effective strain-rate distribution, diagrams of, 262 Exterior penalty method (EP), 363 tooling for, 292
(F), 263 (F), 264 (F) Extrapolation method tube extrusion, 301
Effective stress, 26, 79, 99 for end effects, 50 (F) workability concepts for, 172
Effective stress and strain, 26 for H/D ratios >2, 50 Extrusion dies, 292, 304, 306
Effective stress distribution, diagrams of, 262 (F), Extrema values, 133 Extrusion operations, 108
263 (F), 264 (F) Extrusion. See also backward (indirect) extrusion; Extrusion process, 294, 321, 344
Effective stress-effective strain, conversion from cold extrusion; forward (direct) extrusion; hot Extrusion punch pressure, 300
shear-stress/shear-strain, 99 extrusion Extrusion ratio, 294, 295, 296, 313
Effective stress-strain curves, 49 (F) advantages of, 219 Extrusion/wire drawing, 128 (T)
Effective uniaxial stress, 78 alloys classified on temperature of, 292 (T)
Efficiency of power dissipation, 33 of aluminum alloys, 146 (F)
Elastic modulus, 12 billet-on-billet, 301 F
Elastic-plastic behavior, 15 butt size, 300
Elastic-plastic boundary, 259 canning equipment with composite billet, 299 (F) Faced-centered cubic (fcc) metals, 98, 102 (F)
Elastic-plastic finite element method (FEM), 262 centerburst analysis, 185 Factorial design testing, 131, 131 (T)
Elastic-plastic methods, 16 centerburst during, 183 Failure mode, 79
Elastic solid, 75 of certain metals, 299 Failure modes determinants, 83, 268
Elastic-viscoplastic behavior, 15 comparison of conventional to hydrostatic, 293 (F) Fan cooling, heat treatment, 359, 361
Elastic (Youngs) modulus, 23 components of, 292 Fatigue wear, 288
Electric discharge machining (EDM), 268 conditions and considerations in, 291 fcc (face-centered cubic) metals. See faced-centered
Electrode cold forging process, 227228 continuous, 301 (F), 303 cubic (fcc) metals
Electrolytic tough pitch copper, and centerbursts, 318 defects FEA method (FEM). See finite element analysis
Electromagnetic casting (EMC), 125 bambooing, 297 method (FEA method/FEM)
Elongated primary grains, 35 centerburst, 297 FEM Explorer, 364
Elongation at fracture, 2425 centerbursts, 282 (F), 295 (F), 317 Ferrite, grain size, 250 (F), 274 (F)
Elongation rate, 173 circumferential cracking, 317 Ferrite nucleation, 249
Elongations, 83 (F) crater (pipe), 297 Ferrous alloys, 93, 103
Elroll (software), 262 crows-feet cracking, 297 Fiber, 87
Empirical criterion of fracture, 173, 175176 fir tree defect, 186, 297 Fiberlike structure, development of, 19
Empirical guidelines hot-shortness, 317 Fibrous tearing mode, 28
for die design, 202 laminations, 297 Fields and Backofen method, 94, 95, 100
for preform design, 203 metallurgical deficiencies, 297 Fillet and corner radii, 204 (T)
End effects, 50 (F), 89 orange peel, 297 Fine development, 321
Engineering compressive strain (reduction), 279 in rib, 8 (F) Finish-forging, 213
Engineering normal strain, 22, 23 scoring, 297 Finisher dies, 208, 211
Engineering normal stress, 23 shear cracks, 297 Finishing train, 272
Engineering power law formula, 79 splitting, 297 Finite-difference method (FDM). See direct-equilib-
Engineering process design suck-in defect, 7 rium method
analytical models, 325 defined, 291 Finite element analysis (FEA) software, 16
business models, 325 deformation processes, 144, 146 Finite element analysis method (FEA method/FEM).
lean manufacturing, 326 described, 146 See also process design methods
methods of, 325 in die angles advantages of, 326
performance measurement, 326 centerbursts, 178 boundary conditions, 327
process design methods, 326, 327, 328, 329 dead metal zones, 178 calculations compared with test results, 196 (F)
Engineering shear strain, 47, 88 in die friction, 178 for certain processes
Engineering shear-strain rate, 88 die wash, 298 backward extrusion, 227
Engineering strain, 73 (F) direct and indirect, 303 for forging, 15
Engineering stress, 74 effective strain rates for, 108 for forging load analysis, 230
Engineering stress-strain curves, 23 (F), 74 extrusion rate for, 301 forging process, 347 (F)
cavitation effects on, 82 flow patterns, 296, 296 (F) of hot rolling, 55
for non-hardening samples, 80 (F) in fracture models, 177 for rolling, 262
of orthorhombic titanium aluminide alloy, 74, 76 friction assisted, 301 for shape rolling, 242243
(F) heat generation in, 296 strip rolling, abstracts on, 238
predictions for, 82 (F) heat insulation, 298 for certain testing
using direct-equilibrium approach, 80 (F) internal pressure limit, 296 of hot compression testing, 63
Engineous System, 349 limit diagram for speed, 307 (F), 321 (F) of hot ring compression tests, 65
Enthalpy, 129 lubricants and lubrication, 146, 297, 298 hot-tension testing, 79
Entrained film thickness, 285 material during, 175 (F) of thermomechanical testing, 132
Environmental chamber, 71 (F) material flow in, 300 (F) of Cockroft and Latham, 196 (F)
Environmental chambers, 70 metal flow during, 212 commercial software products, 217
Equal channel angular extrusion (ECAE), 35 microstructure of, 115 compared with SFEM, 328
Equiaxed subgrains, 35, 35 (F) non-steady state, 291, 297 compared with upper bound methods, 328
Equicohesive temperatures, 149 oxides, 298 for complex strain behaviors, 95
Equilibrium and compatibility equations, 12 pressure requirements for aluminum alloys, 305 computational intensity of, 230
Equipment systems, dynamic models, 338 (F) computer processing time, 327, 328
Equipment testing procedures, 68 processing parameters, 123 constant stress/strain elements (CST), 225
Equivalent combined stress, 12 ram displacement vs. surface temperature, 309 (F) for cracks and voids, 15
Equivalent strain, 123, 134 rate vs. flow stress, 308 (F) defect modeling, 227
Equivalent strain at fracture, 12 ratings for aluminum and aluminum alloys, 308 (F) degenerated elements, 225
Equivalent strain increment, 59 schematic, 281 (F) description of the problem, 223
Equivalent strain rate, 12 sections of, 307 (F) development of, 220
Equivalent stress, 59, 134 of shapes, 301 for die/model design, 230
Equivalent stress-strain curves, 88 stem speed for various materials, 310 distortion, 359
Equivalent tensile stress, and shear yield, 123 sticking friction, 298 dynamic effects, 222
Equivalent true strain, 12 stress and strain states of, 172 for effective gage length, 95
Equivalent true strain rate, 101 summary of methods, 298 (F) elastic-plastic methods, 16
Erhardt process, 283 surface cracking from, 317 (F) engineering problems addressed by, 15 (T)
1119 Index 7/29/03 4:16 PM Page 397

Index / 397
expertise levels, 329 Flexibility, 143 Flow-related defects, 7

finite element formulation, 224 Flexible boundary model, 252 Flow softening
friction and contact conditions, 223224 Floating plugs, 286 (F) from deformation heating, 61, 62
Gaussian quadrature formula, 225 Flow-behavior, 118 (F) deformation heating, 87
hardware and software development in, 230 Flow curves. See also true-stress/true-strain curves dynamic recovery, 64
for heat losses, 97 (flow curves) dynamic recrystallization, 64
for insufficient die filling, 15 of certain metals effects of, 33
Jacobin matrix mapping, 225 0.25% low carbon, 113 (F) and flow localization, 50
Lagrangian multiplier method, 225 304L stainless, 100, 100 (F) during hot working, 148
laps, 227 316 stainless, 38 (F) from microstructural effects, 112
for local deformation, 191. See also modeling 317 stainless, 38 (F) from microstructural instabilities, 62
methods and techniques Al-8090 alloy, 75 microstructural instabilities, 87
material and process assumptions, 230 Armco iron, 37 (F) negative strain hardening, 29
mesh-based, unstructured, 329 austenitic stainless steels, 117 (F) during plane-strain forging, 112
metal forming applications using, 15 ferritic low alloy, 117 (F) rate of, 50
for microstructural variations, 15 high-purity 0.8% C pearlitic iron-carbon alloy, recrystallization, 148
for microstructure development, 230 115 (F) shear cracking, 148
MINI element for three dimensional incompress- nickel, 113 spheroidization
ible simulations, 225 Waspaloy, 48 (F), 100 of lamellar microstructures, 87
mixed variation formulation, 224 at cold working temperatures, 62 (F) pearlitic, 115 (F)
modeling techniques, 222223 defined, 37, 74 rapid spheroidization, 64
multidisciplinary optimization (MDO), 346 determination of, 25 strain localization, 148
nodal parameters, 327 discontinuous dynamic recrystallization process textural softening, 87
node-to segment contact, 224 (DDRX), 3738 from thermal softening, 112
for non uniform temperature, 96 dynamic recovery, 37 without chilling, 192
parametric geometry, 327 of dynamic recrystallization, 37 Flow softening parameter, 29
for partial differential equations, 14 initial grain size, effects of, 40 Flow stress, 4
penalty method, 225 isothermal hot-tension test data, 96 behavior and torsion testing, 100
physical modeling, 134 multiple-peak, 114 cavitation, 75
plasticity theory, 178 single-peak, 114 of certain metals
for poor flow of material, 15 slope of, 24 copper and copper alloys, 147
for poor shape control, 15 typical, 100101 low-carbon steels, 147
preliminary assumptions, 222223 Flow-dividing plane, diagrams of, 234 (F) Ti-6Al-4V, 77
problems addressed by, 15 Flow during deformation, 12 titanium alloys, 168 (F)
process modeling by, 3 Flow lines, 31 (F) defined, 10, 122
for process problem solving, 128 (T) Flow localization equations for, 147, 148
quasi-static analysis, 222 alpha parameter use as gage to, 111 of extruded material, 295
residual stresses, 359 defined, 28 and forging temperatures, 168 (F)
rigid-plastic analysis, 222223 flow softening, 29, 50 as function of strain and grain size, 77 (F)
rigid-viscoplastic methods, 16 during hot-tension testing, 82 as a function of temperature and strain rate, 33, 68
ring compression test, 227 in hot working, 29 Gleeble testing and, 73
simultaneous linear equations, 327 indication of, 100 grain growth, 75
software for, 128 micrographs of, 83 (F) hot-tension testing, 73
system integration of, 230 microstructure variations, 108 under hot working conditions, 47
for temperature nonuniformity, 97 at minimum, with torque instability, 110 levels with dynamic recovery, 75
for texture evolution modeling, 43 negative strain hardening, 50 measurement of, 48
three dimensional analysis, 222 of orthorhombic titanium aluminide alloy (Ti- microstructural changes with, 75
three-dimensional analysis, 225 21Al-22Nb), 82 single phase alloys, 105
for tool stress, 230 parameter, 109 strain-enhanced grain growth, 75
updated-Lagrangian (UL) formulation, 223 in plane-strain compression, 29 with superplastic materials, 75
uses of, 15 predictor of, 110 temperature sensitivity, 93
volumetric locking, 225 scribe-line measurements, 108 with varying temperatures and strain rates, 77
Finite-volume methods, 15 shear bands, 109 Zener-Holloman parameter (Z), 105
Fir tree defect, 173, 186187, 297, 316 strain-rate sensitivity, 50, 109 Flow stress curves, 75, 148 (F)
Fixed-end testing, 99 torque twist behavior, 108 Flow stress data, 9394
Fixed-strain rate, torsion testing at decreasing tem- in torsion, 108 for aluminum, 117, 117 (F)
peratures, 117 (F) during torsion testing, 112 torsion testing at decreasing temperatures, 117
Flash Flow-localization analysis torsion testing for, 86, 87
in closed-die forging, 146, 211 elements of, 110 Flow stress data analysis, 100, 101
design of, 211 as function of torque softening rate and rate-sensi- Flow stress data comparison, 101102
dimensions, 211, 214 tivity, 110 Flow stress, measured, with friction and barreling, 63
flash allowance, 208 indication of, 110 (F) Flow stress-strain data
flash zone, 208 in presence of heat transfer, 112 of various aluminum alloys, C and m values, 383
thickness, 228 during torsion testing, 109 (T)
width-thickness ratio, 211212, 212 (F) Flow-localization-controlled failure, 192 of various aluminum alloys, K and n values, 382
Flat-face dies, 292 Flow-localization-controlled failure test (T)
Flat rolling causes of, 196197 of various copper alloys, K and n values, 384 (T)
elastic-plastic boundary, 259 defects of, 196 of various nonferrous alloys, C and m values, 384
entry of strip into, 258259 isothermal compression and side pressing tests, (T)
forces, torques and speeds in, 258260 198 for various steels, 379 (T)
friction hill, 233234 nonisothermal sidepressing test, 197, 198 for various steels, C and m values, 380382 (T)
homogeneous deformation, 233 nonisothermal upset test, 196, 197 for various steels, K and n values, 379 (T)
lubrication, 237238 torsion gage section marking, 89 Flow stress values, error corrections to, 63
plane strain, 233 Flow-localization-controlled workability, 108 Flow stress values, from ring compression tests, of
process of, 258 Flow-localization defects, and temperature buildup, various alloy systems, 385 (T)
relative slip, 234235 31 Flow stress values, from uniform compression tests,
roll separating force, 259 Flow-localization during hot forging of various alloy systems, 385 (T)
roll torque, 259 constant-strain-rate isothermal compression test, Flow-through defects, 7, 8 (F), 19
schematic of process, 259 (F) 198, 199 Folding. See laps
speed stress and distribution of, 233 with preheated dies (i.e., isothermal forging), Ford and Alexander model, 262
strain distribution, 235236 198 Forge (software), 262
strip shape, 236237 strain rate effects on, 199 Forgeability
two-dimensional models, 262 temperature effect on, 199 of alloys, 149
1119 Index 7/29/03 4:16 PM Page 398

Forgeability (continued) on die and preform design, 205 (F) loads during, 191
of alloys with an intermetallic second phase, 192 precision design, 332 (F) Forming-limit diagrams, 32
of alloys with low melting point phases, 192 process-parameter design, 335 (F) 301 stainless steel, 129
of pure metals, 192 process sequence design, 335 (F) for bolt-heading process, 179180
ratings Forging equipment fracture models, 177
for aluminum alloys, 157 (F) characteristics of, 209 strain path design to avoid failure, 128, 129 (F)
for copper alloys, 161 (T) contact time of, 210 workability graphic design, 60
for heat-resistant alloys, 155 (T) hammers (high speed), 188 Formability, 5, 76
of single phase alloys, 192 hydraulic presses (slow), 188 Forming, 11 (F)
and solute content, 194 (F) influence of, 210 Forming loads, 229 (F)
Forgeability behaviors, of various alloy systems, 193 production rate, 210 Forward (direct) extrusion, 291, 294 (F)
(F) ram speed and strain rate and flow stress, 208 with conical die entry, 295 (F)
Forging. See also die forging rate of deformation, 210 container friction, 299
backward/forward process, 217 (F), 219 (F) screw presses (medium), 188 defects, 300
blocker design, 217 (F) Forging load described, 299
center of loading, 211 hot forging process, modeling complexities of, 228 extrusion pressure variations, 300 (F)
central discontinuity in, 6 (F) load and forging height, 215 friction-assisted, 302 (F)
cogging, 132 load estimation, 214 incipient dead metal zones, 300
common metallurgical defects in, 188 (T) load estimation by slab method, 215 lubricant or contamination, 300
defined, 208 load estimation by slab method finite element material flow, 299, 300 (F)
deformation processes, 146147 method (FEM), 216 process characteristics, 299
described, 146147 load vs. displacement curves, 210 (F) container friction, 299
design and tolerances, 159 (T) Forging lubricants. See lubricants and lubrication incipient dead metal zones, 300
design guides, 154 (T) Forging methods for aluminum alloys, 157160 material flow, 299
design limits of, 209 (F) Forging operations, optimal working temperature for, with sticking friction, 300 (F)
design of, 208 108 surface lamination, 300
design parameters for optimization, 374 (F) Forging pressure, 155 (F), 156 of a tube, 301 (F)
die-velocity schedule, 374 (F) Forging process Forward rod backward cup extrusion, 218
ductile fracture, 16 activities in, 347 (F) Forward slip, 233, 235, 241
electrode, progression of, 227 (F) activity-based model for, 331 (F) Fractional factorial method, 131
finite element method (FEM) of defects, 216 (F) block diagram of, 338 (F) Fracture-controlled defect formation, 9394, 192
flash and width-thickness ratio by forging weight, finite element method (FEM), 347 (F) Fracture-controlled failure
212 (F) inspection and process traveler, 334 (F) of gamma titanium aluminide alloy, 82
flow-through crack, 8 (F) integrated optimization system, 350 (F) indication of, 100
flow-through defects, 8 (F) interdependence of parameters, 332 (F) torsion gage section marking, 89
forging burst, 6 (F) multidisciplinary optimization (MDO), 347 (F), Fracture criteria
gas turbine disks, 347348 351 for bolt-heading process, 179
grain flow pattern, 9 (F) non-isothermal and non-steady-state physics, 331 Cockroft criterion, 177
guide to methods for, 128 (T) (F) Cockroft-Latham fracture criterion, 32
hydrostatic stress during, 16 (F) optimization of, 348 in material selection, 180
lap caused by improper radius, 7 (F) shape optimization, 351352 for metalworking processes, 177
lap with oxide, 7 (F) Forging process parameters, 339 in process selection, 180
material during, 175 (F) Forging process variables Fracture limit curves, 55
metal flow during, 212 die temperature, 188 Fracture limits
optimal design process, 358359 forging equipment, 210 for 2024 aluminum alloy, 54 (F)
optimization alternates, preform design, 357 forging materials, 209210 during rolling, 56 (F)
physical phenomena during, 10 (F) friction and lubrication, 210 with servohydraulic test equipment, 183
planes and directions of metal flow, 213 (F) heat transfer and temperatures, 210 upset test representation, 59, 59 (F), 60 (F)
preforming, blocking and finish operations, 213 (F) lubrication, 188 Fracture line limit, 55
pressure vs. forging height, 215 (F) production lot size and tolerances, 210211 Fracture locations, 173
process design methods, 208 (F) test techniques for, 188 Fracture loci, 176, 184
process development, 347348 working speed, 188 Fracture mechanisms. See also cavities and cavita-
process of, for crankshaft, 228 (F) hammers (high speed), 188 tion; ductile fractures
in secondary tensile stress state, 173 hydraulic presses (slow), 188 brittle fractures, 27
secondary tensile stress state in, 173 screw presses (medium), 188 centerline-porosity, 319
sequence of, 218 (F) workpiece preheat temperature, 188 coalescence stage, 28
shape classification, 211 (F) Forging rate, 163 Cockroft and Latham criterion, 319
starting shape for, 374 (F) Forging sequence design, 205 (F) defined, 27
stress and strain states of, 172 Forging severity described, 319
stress state in, effects of billet shape and enclosure, copper and copper alloys, 161, 161 (F) dimpled rupture mode, 27
192 (F) levels of, 151152, 152 (F) ductile fractures, 27, 28
of superalloys, 154 (T) Forging stresses and loads, prediction of, 213216 fibrous tearing mode, 28
thermal gradients and stress, 217 empirical methods for, 214215 map of, 27, 27 (F)
titanium alloys, 167 load-stroke curves, 214 material structure, 319
types of simplified slab method, 214215 origins of, 19
alpha-beta forging, 167168 Forging temperatures probable plane of fracture, 269 (F)
beta forging, 167, 168 for certain metals shear band tearing mode, 28
closed-die forging, 146 aluminum alloys, 157 (F), 157 (T), 160 stages of, 27
cold forging, 147 carbon steels, 149 strain rates of, 86
isothermal, 354 of copper alloys, 161 (T) void growth stage, 28
isothermal (hot-die) forging, 147 copper and copper alloys, 161 void initiation stage, 28
multi-step, 358359 for heat-resistant alloys, 155 (T) wedge cracking failure, 86
nonisothermal, 354 for high-temperature alloys, 153 (T) Fracture models, 177
open-die forging, 146 for magnesium alloys, 162 (T) Fracture prediction, 180
rotary forging and swaging, 146 nickel-base alloys, 164 (T) Fracture resistance. See ductility
using optimal preform shape, 374 (F) titanium alloys, 167 (T) Fracture strain
using stage 1 preform, 375 (F) for various carbon and alloy steels, 150 (T) locus predictions, 177 (F), 178 (F)
variables in, 209 (F) for various metals, 155 (F) torsional and tensile correlation, various alloys,
weight reduction strategy, 348 cold-forging, defined, 191 106, 107 (F)
workability concepts for, 172 defects during, 191 Fracture strain diagram, 187
Forging brass, 160 equipment for monitoring, 160 Fracture strain lines, 181 (F), 184 (F), 185 (F)
Forging characteristics, of refractory metals, 156 (T) and flow stress, 168 (F) Fracture types. See defects
Forging design vs. forging pressure, 155 (F) Frame furnace tension-testing equipment, 70, 7172
1119 Index 7/29/03 4:16 PM Page 399

Index / 399
Fredholm integral equation, 15 on-cooling specimen, 73 (F) ductile fractures, 270

Free-end testing, 99 on-heating curve, 69 dynamic recrystallization, 114
Free-surface cracking, 149, 266 reduction of area, 69, 72 dynamically-recrystallized, 37
Free-surface fracture, 5, 59, 173 ultimate tensile strength, 69 effects on workability, 29
Friction yield, 69 at equilibrium, 39 (F)
coefficient of, in rolling, 233 Gleeble testing, 7172 fatigue crack resistance, 129
combined, in transition, 65 flow stress and, 73 of fully recrystallized austenite, 249
Coulombs coefficient of, 31 hot ductility curves from, 73 high-temperature creep resistance, 129
Coulombs law of friction, 31 hot-tensile strength curves from, 73 initial, 189
determination of, 234, 235 for hot-working, 68 of optimal design process, 344
at die contact surfaces, 173 on-cooling data, 70 optimal die profile for, 345 (F)
in drawing, 278, 279 on-heating data, 73 (F) room-temperature tensile strength, 129
edge effect, in rolling, 236 sensitivity of, 74 thermomechanical processing and, 9
effect on extrusion forces, 300 water quenching, 70 of thermomechanical testing, 131
effects of, 50 Gleeble testing equipment, 69 (F) torsion specimen size, 87
during extrusion, centerburst control, 297 grips for, 68 trajectories of, 344 (F)
from forward slip, 235 heating systems, 6869 variations in, 9
in hot plane-strain compression test (HPSC), 65 for high-strain high-temperature conditions, 54 workability and, 9
indicators of, 234 linear variable differential transformer (LVDT), 53 Grain structures, 9, 29, 72
interface factor of, 32 (F) Graphical user interface (GUI), 351
interface values, from ring compression test, 52 procedures for, 69 Greens theorem, 369
modeling techniques, 230 sample design, 6870 Grip ends, 68, 89, 93
as process variables affecting workability, 31 specimen geometry, 68, 69 Group technology (GT), 142143
in rolling interface pressure, 236 thermocouples, 68, 69 Growth of individual cavities, 80, 81
at rolling point entry, 233 Gradient-based methods, 355 Guard ring box, 62, 62 (F)
from skidding, 235 Grain boundary
sliding friction, 65 after torsion testing, 115
sticking friction, 65, 298 (F), 299, 310 cavitation, fracture-controlled defect formation, H
sticking zone, 235 192
Friction condition, finite element method (FEM), 224 cracks in, 9 Hall-Petch equation, 147
Friction effects critical particle size, 251 Hamiltonian, 41
barreling, 175 energy, solute drag, 250 Hardenability, 10
cracking failure, 175 pinning by particles, 250251 Hardness state, 95
strain path, upset-tests, 59 (F) sliding, 103 Hardware requirements, 340341
tension failure, 175 causing wedge cracking, 28 Heat checking, 10
in upset testing, 29 (F) creep forming, 149 Heat generation, 225, 296
Friction factors from deformation at low strain rate, 149 Heat loss parameter, 130
barreling coefficient, 63, 63 (F) plastic anisotropy parameter, 76 Heat-transfer coefficients, 129
for common metals, 210 thermal activation of, 104 as design variables, 362
determination of, 53 triple points, 32 determination of, from time-temperature data, 363
Male and Cockroft ring compression test, 57 voids, nucleation of, 192 (F) (F)
specimen geometry, 63 (F) Grain boundary working, 191 hot forging process, modeling complexities of, 228
Friction hill, 260 Grain coarsening, 114 measurement test for, inverse method, 131
cold hardening, 234 of alpha-beta titanium, 115 in optimal design process, 362
critical roll speeds, 241 development of, 250 testing for, 131
described, 233234 grain boundary pinning, 250251 at tool/workpiece interface, 217
diagrams of, 234 (F) by secondary recrystallization, 251 turbine disk, 360 (F)
variations, causes of, 241 solute drag, 250 Heat transfer, governing equations for, 225226
Friction reduction, 48 Grain-coarsening temperatures, 249, 250251 Heat treatment
Friction work, 225 Grain flow, 9 and cooling rate, 151
Frictional force, in rolling direction, 240 Grain flow pattern, in wrought materials, 30 costs of, 151
Frictionless compression, strain ratio for, 55 Grain growth fan cooling, 359, 361
Front tension, 235 analytical models of, 40, 4142 of nickel-base alloys, 164
Furnace types, 93 development of, 250 oil-quenching, 359, 361
ductility impacts of, 104 optimization, 348
factors retarding, 250 alternative experimental methods, 361362
G in gamma titanium aluminide alloy, 42 constraints, 360
grain boundary energy, 39 cooling-rate distribution, 361 (F)
Gage length Hillert model of, 42 DEFORM (software), 359360
ductility from, measurement of, 2425 during hot-working, 148 formulation of, 359360
effect of, on strain distribution, 78 inhibition of, 254 hoop stress, 362 (F)
hot-tension testing, 71 key points on, 149 iSIGHT (software), 359360
strain gradients, 78 models of, 40 limitations of, 362
Unwins equation, 71 temperature and, 9 objective functions, 360, 360 (F)
Gage length specimens, 78 (F) Grain orientation, 189 radial stress, 362 (F)
Gage length-to-radius ratio on effective strain to fail- Grain shapes, 189 residual stresses, 361
ure, 103 (F) Grain size sequential linear programming, 360
Gage marks, 57, 89, 9394 about, 9 of turbine disk, 360
Gage width, 78 in certain metals purpose of, 359
Gas turbine disks, 347348 0.16 carbon steel, 114 of stainless steels, 152, 153
Geometric constraints, 358 304L, 114 Heating and temperature control of treatment for alu-
Geometric tolerances, 143 of aluminum killed steels, 149 minum alloys, 160
Geometry effects, 175 control of, in nickel-base alloys, 165 Heating, of dies, 161
Glass coatings. See lubricants and lubrication copper, 114 Heating systems for hot-tension testing, 70
Gleeble curves ferrite, 249 Height-to-length (h/l) ratio
for certain metals in iron, 149 and centerbursts, 317
AF 1410, 75 (F) magnesium alloys and, 162 deformation zones in, 261 (F)
Alloy 901, 76 (F) nickel, 114 in drawing, 279, 280
cobalt-base superalloy, 75 (F) nickel-base alloys, 165 in extrusion, 295
lanthanum-bearing alloy, 76 (F) ultra-fine ferrite, 274 homogeneity lower values, 295
on heating sample, 70(F) cold working, 147 pressure multiplying factor, 261 (F)
hot-workability ratings, 73 (F) control of, 9, 149 as reduction limit, 279
on-cooling, for commercial alloys, 74 (F) distribution of, 251 redundant work in, 280
1119 Index 7/29/03 4:16 PM Page 400

Helmholtz equations, 15 constant-shearing friction factor, 371 on cooling Gleeble curve, 73 (F)
High-definition forgings. See near-net-shape forming flow-related defects, 7 ductility ratings, 72
High speed isothermal compression testing, 192 temperatures of, 209 (T) grain structure, 72
High speed isothermal tension test, 192 Hot forging operations, 48 Hot working
High stacking-fault energy Hot forging process, 228 characteristics of
aluminum alloys, 35, 148 Hot-forming pressures, 163 (T) deformation rate, 68
aluminum, pure, 35, 112 Hot hydrostatic extrusion, 312, 313 temperature, 68
iron, in the ferrite phase field, 35 Hot-impression-die forging. See closed-die forging defects, 68
metals of, 112 Hot plane-strain compression testing (HPSC), 6566. defined, 5, 28
microstructure in, 35 See also plane-strain compression testing deformation levels, 68
nickel alloys, 148 Hot-reduction temperatures, ductility ratings, 72 deformation processes, 148149
titanium, in the beta phase field, 35 Hot ring compression testing, finite element analysis dynamic recovery, 148
High-strain conditions, 54 (FEA) of, 65 dynamic recovery and recrystallization in, 35
High-strain rates and hot shortness, 124 Hot rolling. See also controlled rolling dynamic recrystallization, 148
High-temperature alloys, 210 of certain metals effects on microstructures, 148 (F)
High-temperature conditions, 54 2024 aluminum alloy, 181 flow stress vs. temperature and strain rate, 68
Hillert model of grain growth, 42 aluminum-magnesium alloy 5182, 125 force levels requirements, 68
Hills general method, 220 steel, 265 grain growth during, 148
Hills variation principle, 50 finite element analysis (FEA) of, 55 high surface tensile stresses, causes of, 68
History effects, 116 homogenization sensitivity, 124 homologous temperature, 148
Hoffmanner model, 108 (T) lead modeling of, 55 internal tensile stresses, causes of, 68
Homogeneous compression, 55, 173 lubrication for wear, 237 mechanisms of, 28
Homogeneous deformation, 233, 260 models of, 17 microstructure, 35, 140
Homogenization, 6, 124, 126, 149 press stalling from high flow stress, 124 physical simulation of, 122123
Homogenization heat treatment, 29, 123127 rigid-plastic finite element method (FEM), 262 prestrain effects of, on pure nickel, 36 (F)
Homologous temperature, 62, 148 slip-line fields for, 261 (F) process of, defined, 3031
Hook-Jeeves direct search method, 363 surface cracking, 236 processing parameters requirements, 68
Hoop stress, 59 temperature selection for, 125 single phase materials, 114
Hopkinson (Kolsky)-bar method, 77 Hot shortness temperatures of, 148
Hot-compression testing cause of vanishing flow stress, 123 tensile ductility (fracture strain) vs. temperature
cylindrical compression test, 6163 defined, 29 and strain, 68
determination of the flow curve, 63 ductility impacts of, 104 workability testing, 68
effect of deformation heating, 6263 in extrusion, 317 Hot working conditions
effect of strain rate, 62 fracture-controlled defect formation, 192 flow stress under, 47
effect of temperature, 61 and high-strain rates, 124 stress-strain curve under, 47
fracture-controlled defect formation, 192 Hot tensile strength curves, 73 torsion testing under, 47
guard ring box for, 62, 62 (F) Hot tension testing, 6885 Hot working operations, strain rates in, 18
hot plane-strain compression test, testing condi- apparent cavity growth rate, 82 Hot working simulation by hot torsion testing
tions, 6566 atmosphere for, 70 cam plastometer, 122
incremental loading in, 61 cavities/cavitation, 71, 7980, 82 commercial rolling parameters, 126127
plane-strain compression test, testing conditions, closed-loop temperature controllers, 68 cost-benefit analysis of EMC ingot, 125
65 for crack propagation, 54 effect of alloy variation within specification,
plastic instability in, 61, 63, 65 crosshead speed effects, 7778 124125
ring compression test, 6465 design parameters for, 68 extrusion of Al-Mg Alloy, 124
test types, 61 direct-equilibrium method, 79 hot torsion testing, 122124
testing apparatus, 61 for ductility measurement, 54 rolling temperatures, 125
tests for initial forging temperature and strain rate, engineering stress-strain curves, 82 (F) uniaxial compression of cylindrical specimens, 122
192 equipment for, 68 Hot working simulators, 122 (T)
thermocouples, 62 failure modes, 82 Hot working temperatures
true strain in, 63 finite element method (FEM), 79 defined, 145
Hot-die forging, grain-boundary sliding, 147, 149 flow localization, 82 ductility ratings, 72
Hot ductility flow stress, 73 flow-stress data comparison, 101102
nonmonotonic increase with deformation tempera- for hot-working, 68 of HSS 188, 73
ture, 125 for low ductile materials, 54 maximum, from on-cooling data, 72
recovery of, 7273 maximum and minimum temperatures for, 54 torsion testing at, 101
specimen geometry impact on, 103 numerical techniques, 79 Hydrafilm process, 313
and strength data from the Gleeble test, 7274 plastic anisotropy parameter, 71 Hydraulic presses, 182
in torsion, 108 reduction of area, 71 Hydraw process, 311
Hot extrusion results from, 7071 Hydrodynamic lubrication
billet preparation, 309 to simulate hot working operations at low strain in drawing, 280
bridge-type die, 307 rates, 122 in dry drawing, 285
defined, 291 specimen geometry, 68, 7172 inlet tube for, 284 (F)
dies and tooling for, 310 strain-rate, 68 Sommerfeld number, defined, 240
dummy block, 309 strain-rate sensitivity, 82 Hydrostatic extrusion
exit temperature, 309 strain-rate variation in, 69 advantages of, 311
flow stress, 309 temperature control, 68 augmented process, 311
isothermal extrusion, 309 tension machine characteristics, 68 bambooing defects, 311
lubricants and lubrication, 307, 310 tension-test frames for, 70 basic process of, 311 (F)
material flow in, 302 (F) total tensile elongation, 7071 billet augmentation, 312
materials for, 308, 309 universal testing machines for, 70 of brittle materials, 313 (F), 313314
near-net-shape forming, 146 yield/ultimate tensile strength, 71 castor oil, 312
optimal extrusion temperature, 307 Hot torsion testing ceramic die coatings, 314
pressure requirements, 309310 of aluminum alloys, 123127 defined, 291
process limitations of, 296 (F) described, 122123 described, 310311
products of, 291 ductility in, 49 (F) equipment design, 312
ram speeds, 310 ductility of certain metals, 49 (F) fluid handling time, 313
schematic of, 293 (F) for materials that strain or strain-rate harden, 123 hydraulic medium and lubrication, 312
shaped dies, 307 maximum shear strain in, 123 limitations of, 311312
skull, 310 rate-sensitivity in, 94 lubrication pressure limits, 312
taper heating, 309 shear stress in, 123 methods of, 311
Ugine-Sjournet process, 310 Hot upset testing. See cylindrical compression test mixed-film lubrication, 312
unlubricated, of aluminum alloys, 310 Hot workability ratings nitriding, 314
Hot forging of certain metals, 52 (T), 72 (T) solid lubricant entrainment, 298
1119 Index 7/29/03 4:16 PM Page 401

Index / 401
stick-slip phenomena, 312 Ti-10V-2Fe-3Al, 199 (F) Loading history, 87

tool materials for, 312 Ti-10V-2Fe-3Al in, 50 (F) Loads during forging-temperature selection, 191
Hydrostatic stress, 173 Ti-6Al-2Sn-4Zr-0.1Si (Ti-6242Si), 199 (F) Local optimums, 355
defined, 16, 26, 78 heated subassembly in, 49 (F) Local softening from adiabatic temperature increases,
ductile fracture, 172 Isothermal (hot-die) forging 28
effects of, on fracture loci, 184 for aluminum and aluminum alloys, 147 Localized conditions, 172
during forging, 16 (F) disk volume and press load, 354 (F) Localized melting. See hot shortness
Hydrostatic tension, 6, 76 grain-boundary sliding, 149 Localized necking. See flow localization
Hyperbolic wave equations, 11 initial and final shapes for, 354 (F) Localized strains
mesh regeneration, 354 during rolling, 174 (F)
of nickel based alloys, 10 in upset test representation, 173 (F)
I optimal design process, 354 variation by aspect ration and friction, 173 (F)
optimization, 354 Localized thinning models, 177, 177 (F)
Ideal compression, strain ratio for, 55 stroke-averaged load, 354 Low stacking-fault energy, 148
Implicit method, 222 of titanium, 10 Lower-bound method, 14, 42, 318
Impression die-forging, 208, 212213 Isothermal hot-tension test data, 75 Lubricants and lubrication. See also coefficient of
Incipient melting (hot shortness). See hot shortness effect of test conditions on flow behavior, 7677 friction; die lubrication
Inclusions, 6 material coefficients from, 7576 alloys and, in hot and cold rolling, 237
Incompressible flow, types of, 225 stress-strain curves, 7475 approximation of, in rolling, 233
Indentation pressure, 179 (F) Isothermal hot-tension testing, 74, 78 boron nitride solution, 61
Indirect (backward) extrusion, 291 Isothermal plane-strain sidepressing, 111 for certain metals
Indirect-compression processes, 144 Isothermal recrystallization and precipitation, 253 (F) for aluminum and aluminum alloys, 306
Induction heating Isothermal rolling, 149 for carbon and alloy steels, 150
for hot-tension testing, 68 Isothermal superplastic forming, 154 for copper and copper alloys, 161
for load-frame testing, 71 (F) for hot forging of steel, 150 (T)
metals suited for, 93 for low chromium metals, 287
Information content of a part, 142 J for molybdenum and molybdenum alloys, 157
Ingots, 4 (F), 6 for stainless steels, 153
Inhomogeneities, effect on fracture, 173 J2 flow rule, 223 for superalloys, 154
Inhomogeneous deformation, 233, 234 Johnson-Cook equation, 26 for tantalum and tantalum alloys, 157
Instability, 339 for titanium alloys, 169170
Instability parameter, 33 for tungsten and tungsten alloys, 157
Instantaneous breadth, 65 K chromating, 287
Instantaneous thickness, 65 for cold forging, 147
Instantaneous value of pressure, 65 Kinematically admissible velocity field, 14 colloidal graphite, 153, 154, 157, 161
Insulation, 170 Kronecker delta, 41, 223 considerations in rolling, 272
Intensified controlled rolling, 253254 conversion coatings, 147, 284, 287
Interface friction factor, 32, 63, 64 cooling capacity, 290
Interface-migration rate, 40 L for cylindrical compression test, 61
Interface pressure, 260, 279 deformation processes, effects on, 235 (F)
Interfacial shear (friction) component, 50 Laasoui-Jonas model, 12 delivery failure and cracking, 265
Intergranular fracture, 104 Lagrangian analysis, 229 (F) for dies, 161
Internal cavities (pores), 68 Lagrangian multiplier method, 224 for drawing, 287 (F)
Internal deformation, 281 Lagrangian transient approach, 229 dry drawing, 284
Internal fractures, 19, 185 Lanthanum, 74 effects of, 149
Internal variable model, 12 Laplace equations, 11, 15 elastohydrodynamic lubrication, 285
Interpolation functions, 16 LApp (polycrystal plasticity program), 42 entrained film thickness, 285
Intragranular slip, 81 Laps explosion risks of, 153
Intrinsic ductility, 52 bulkhead forging, 203 (F) extreme pressure additives, 237
Intrinsic workability, 340 causes of, 7 extreme pressure (EP) lubrication, 287
Inverse method, for determining heat transfer coeffi- from closed-die forging, 202 (F) extrusion, 146
cients, 131 defined, 22 with coefficient of friction values for various
Iron, general forging sequence design on, 205 (F) metals, 314 (T)
ductility vs. temperature, 104 modeling of, 227228 comparative evaluation, 301
ferrite, flow curves, 37 (F) result and simulations of, 229 (F) glasslike, 298
forgeability behaviors of, 193 (F) specimen defects, 175 greaselike, 298
phase field, 35 Laps and folds, 19 ranking for, 295
SFE (stacking-fault-energy), 35, 36, 112 Large grained cast structures, 191 solid-film lubrication, 298
steady flow stress, 116 (F) Laser flash technique, 130 forging oils, 153, 154, 161
Swedish, 104 Lateral inhomogeneity, 236 friction factors, 210, 314 (T)
textures, 98 Law of constancy of volume, 49 glass and glass coatings, 48, 153
in torsion, 98 Layer-lattice compounds, 286287, 299 borosilicate, 157
vacuum melted, 116 (F) Lead and lead alloys, general borosilicate glass, 299
Iron, specific types bars during rolling, 55 (F) ceramic glass precoats, 169
Armco cavitation affected, 80 glass lubricant, 310
flow curves, 37 (F) and cracking in copper, 160 glass lubrication, 299
stress-strain curves, 113 (F) for hot extrusion, 308 low melting, 61
Ironing, tube drawing, 283 as simulation material, 180 graphite, 157, 160, 170
ISIGHT (software) Lead and lead alloys, specific types for high temperatures, 286
ADS module sequential linear programming, 360 Pb-Sn graphite/mineral oil, 159
described, 349 tensile elongation as a function of strain-rate graphite sheets, 61
Engineous System, 349 sensitivity, 77 (F) graphitic lubricants, 299
features of, 351352 Lean manufacturing, 326 heat buildup in drawing die, 280
as framework for integration, 350 Levy-Mises equations, 59 hot extrusion, 307, 310
multidisciplinary optimization (MDO), 348, 349, Levy-von Mises criterion, 12 hot forging process, modeling complexities of, 228
350, 351 Liapunov function, 340 hydrodynamic lubrication, 237238
optimization procedure, 352 Liapunov material stability criteria, 340 in drawing, 280
Isothermal compression and side pressing tests, 198 Liapunov stability analysis, 17 in dry drawing, 285
Isothermal conditions, 49 Linear elliptic partial differential equations, 15 inlet tube for, 284 (F)
Isothermal high-strain-rate flow curves, 100 Linear variable differential transformer (LVDT), 130 Sommerfeld number, defined, 240
Isothermal hot compression testing (F) viscosity effects, 241
alpha parameter in, 111 Load-frame testing, 71 (F) layer-lattice compounds, 286287, 299
of certain metals Load-stroke curve, 215 (F) liquid, 147
1119 Index 7/29/03 4:16 PM Page 402

Lubricants and lubrication (continued) grain growth, 162 Mean strain rate, 49
low-viscosity mineral oil, 153 hammer forging and, 162 Measurement
mixed-film, 286, 290 AZ61A of flow-localization-controlled workability, 108
modeling techniques, 230 forging temperatures for, 162 (T) of work-hardening rates, 94
molybdenum disulfide, 57, 61, 157, 286 AZ80 of yield stress, 94
oil-graphite mixtures, 150 relative workability in forging and die-filling, Mechanical anisotropy, 9, 58
organic and inorganic compounds, 159, 160 194 (F), 210 (F) Mechanical stability, 340
oxalted coatings, 287 AZ80A Mechanical tests, 10
oxide coatings, 287 forging temperatures for, 162 (T) Mechanical texture (fiber), 87
phosphate coating, 287, 306 grain growth, 162 Mechanistic modeling of cavitation, 195196. See
for plane-strain compression test (PSC), 65 hammer forging and, 162 also cavities and cavitation
plastohydrodynamic (full-fluid-film) lubrication, EK31A Mesh regeneration
284285 forging temperatures for, 162 (T) isothermal forging optimization, 354
polymer coatings, 287 HM21A shape optimization, 358
polymers, 284, 299 forging temperatures for, 162 (T) Meshed-solution methods. See also finite element
pressurized, 286 grain growth, 162 analysis method (FEA method/FEM)
in rolling, 237 (T), 237238, 240 (F), 241 (F) hammer forging and, 162 advantages of, 13
in rolling interface pressure, 236 Mg-0.5% Zr boundary element method (BEM), 13
selection factors, 150 tensile elongation as a function of strain-rate defined, 13
soaps sensitivity, 77 (F) direct-equilibrium method, 13
additives to, 286 QE22A models, 326
characteristics of, 286 forging temperatures for, 162 (T) uses of, 15
dry drawing, 284 ZE62 Meshes
in hydrodynamic lubrication, 285286 forging temperatures for, 162 (T) Advancing Front triangulation methods, 328
lubricating powdered soaps, 285 ZEA42A Delaunay triangulation method, 327
rheology of, 286 forging temperatures for, 162 (T) Metadynamic recrystallization (MRDX), 36, 3839,
sodium tallow soap for aluminum alloys, 306 ZK21A 253, 256
viscosity of, 285286, 286 (F) forging temperatures for, 162 (T) Metal coatings, 287. See also lubricants and lubrica-
solid film, 286287, 290 ZK60A tion
for specimen protection, 93 forging temperatures for, 162 (T) Metal flow, 232
sulfuration, 287 grain growth, 162 during extrusion, 212
sull coatings, 287 hammer forging and, 162 planes of, 213
thick-film lubrication, 317 (F) ZK61A during upsetting, 212
thickness film, geometric factors for, 285 (F) grain growth, 162 Metal forming analysis, 11
trapped, 7, 8 (F) Male and Cockroft ring compression testing, 57 Metal powders, 309
Ugine-Sjournet process, 299 Mandrel forging, 160 Metalforming process optimization, 348349
upset forging, 159 Mandrels, 283 Metallography, 128 (T)
visco-pressure coefficient, 285 Manual design process, 329, 331, 332, 333 Metallurgical considerations, 29
viscosity effects, 241 Manufacturing methods, 142 (F) Metallurgical phenomena affecting workability, 269
abietic acid, 305 Manufacturing processes Metalworking processes, 3031, 177
water-base graphite, 61 characteristics of, 140 Microalloyed bar, 254255
water-base synthetic lubricants, 150 characteristics rating of, 145 (T) Microalloyed steel forgings, 150, 151, 272273
water-graphite mixtures, 150 classes of, 140 (F) Microalloying, 249
wet drawing, 284 dimensional tolerances, 143 Microalloying elements, 249
of workpiece-tool interface, 29 material aspects of, 140141 Microcracks, 58
zinc stearate, 306 materials compatibility with, 141 (T) Micromechanical approach, 83
Ludwik equation, 25 part shape and complexity, 142 Micromechanisms of ductile fracture, 172
Lumped parameter methods, 13 part size, 140141 Microstructural instabilities, 87
LVDT (linear variable differential transformer), 130 process factors, 143144 Microstructural modeling, 10, 35, 3943, 339
(F) production volume, 144 Microstructural softening, 93
rating scale for, 144 (T) Microstructural trajectories, 344
surface roughness and tolerances from, 144 (F) Microstructure
M and their attributes, 145 (T) acicular-ferrite, 151
web thickness and material movement, 146 (F) defined, 10
Machinability, 10 MARC (software), 262, 349 deformation parameters, 114
Macroetching methods, 6 Mass scaling technique, 222 delta-ferrite formation in stainless steels, 152
Macroscopic fracture models, 177 Material attributes, 10 development of, 35, 35 (F), 343 (F)
Magnesium and magnesium alloys, general, 162163 Material behavior models, 18 (F) effects of specimen geometry variation on, 88
in aluminum, 125 Material characterization, 230 evolution of, 343 (F)
coefficient of thermal expansion (CTE), 307 Material coefficients, 75 ferrite-pearlite, 151
compatibility between materials and manufacturing Material control, 10 fibering, 147
processes, 141 (T) Material data, 128 in high-stacking-fault-energy materials, 35
containing zinc, forgeability behaviors of, 193 (F) Material density, 96 hot-working effects on, 148 (F)
die heating, 162 Material factors affecting workability, 27 lath martensite, 151
drawing lubricants and coefficient of friction val- Material flow, 10 mechanisms of, 3536, 172
ues, 287 (T) Material flow analysis, 339 phases present in, 11
extrusion of, 307 Material inhomogeneities of steel, 58 phenomenology, 3839
extrusion temperature classification, 292 (F) Material properties discontinuous dynamic recrystallization process
flow stress effects in aluminum, 125 flow-stress variability, 189 (DDRX), 38
forgeability behaviors of, 193 (F) physical properties, 189 dynamic recovery, 38
forgeability factors, 162 sensitivity analysis of, 13 of plastic flow, 37, 38
forging temperatures for, 162, 162 (T) Material trajectory optimization, 341 work-hardening exponent and yield strength, 147
grain size and, 162 Materials modeling, 338339 (F)
hammer forging, 162 Materials selection, 140, 141 Microstructure development, 112. See also dynamic
for hot extrusion, 308 Materials stability criteria, 339 recovery; dynamic recrystallization
hot forging temperatures, 209 (T) Materials utilization, 143 deformation processing, 112
hydraulic presses for, 162 Mathematical expressions for the flow curve, 25 multiphase alloys, 114
lubricants and friction coefficients, 237 (T), 314 Maximum hot-working temperatures, 72 in single phase materials, 112
(T) Maximum principal stress, 12 torsion testing as means of determining, 115116
wire drawing die materials for, 289 (T) Maximum shear strain in hot torsion testing, 123 two phase alloys, 114
Magnesium and magnesium alloys, specific types Maximum shear stress (Tresca yield criterion), 26 Microstructure evolution
AZ31B Maximum tensile stress, 134 cellular automata (CA), 40
forging temperatures for, 162 (T) McClintock model, 177, 177 (F) elements affecting, 132
1119 Index 7/29/03 4:16 PM Page 403

Index / 403
by finite element analysis (FEA) modeling, 10 slip-line method, 63, 220221, 227 process analysis, 346
grain growth, models of, 40, 4142 state-space (lumped parameter) method, 341 ProEngineer (software), 348
hybrid MC-CA models, 41 states, time evolution of, 341 response surface method, 363
Monte-Carlo techniques, 40 thick ring compression test, 226 (F) sensitivity analysis, 349
texture evolution, models of, 40 thin ring compression test, 226 (F) shape optimization, 357, 358
titanium alloys, 115 three-dimensional finite element method (FEM), ship shape, 346
torsion testing, 115, 117 267 system integration, 350
Widmansttten alpha microstructure, 115 three-dimensional Mohrs circle, 264, 265 techniques checklist, 350
Microstructure optimization, 341 Total Lagrangian (TL), 223 thermomechanical design, 346
Microstructure variations, 108 trajectory optimization, 341 Unigraphics (software), 348, 349
Mill-simulation tests, 91 two-dimensional buckling, 226 (F) Multidisciplinary process design and optimization
Miller indices, 98 (T), 99 (T) unstable flow (buckling mode), 226 (MPDO). See also finite element analysis
Minimum and maximum functions, 352 updated Lagrangian (UL), 223 method (FEA method/FEM); modeling methods
Mobility, 41 upper-bound method, 220222, 327 and techniques; optimal design process
Modeling methods and techniques. See also finite el- energy principle basis, 221 about, 325
ement analysis method (FEA method/FEM); friction effect on neutral point location, 226 (F) benefits of, 337
multidisciplinary optimization (MDO); optimal jump condition, 221 bubble chart, 331 (F)
design process; physical modeling; ring com- upper-bound elemental technique (UBET), 221 engineering design, 325
pression test; Sachs (slab) method (SM) upper-bound theorem, 221 process design methods, 326327
advances in, 346347 workability, 10 software robot, 331
analytical techniques, 220 Modified empirical criterion, 184185 upper-bound analysis elemental technique, 327
arbitrary Lagrangian Eulerian (ALE) method, 223 Modified method of feasible directions (MMFD), use of, 325326
boundary value problems (BVP) 363 Multiphase alloys, 115, 117
constraint conditions for, 224 Molybdenum and molybdenum alloys, general Multiple test piece method, 95
Lagrangian multiplier method, 224 forgeability behaviors of, 193 (F) Multiple testpiece method, 95
penalty method, 224 forging characteristics of, 157
principle of virtual work, 224 hot forging temperatures, 209 (T)
strong solution, 224 lubricants for, 157 N
variational form of, 224 microalloy benefits of, 272
weak solution, 224 molybdenum alloys containing oxides, 193 (F) Nadais approach, 94
Cauchy stress tensor, 223 molybdenum, soluble and solute drag, 254 Near-net-shape forming, 145
comparison of, 226 relative workability in forging and die-filling, 194 in closed-die forging, 158
constant shear friction model, 220 (F), 210 (F) hot extrusion, 146
continuum mechanics, 10 wire drawing die materials for, 289 (T) materials selection, 141
control theory, 341 Molybdenum and molybdenum alloys, specific types multidisciplinary optimization (MDO), 348
cost control, 346347 unalloyed Neck profile, 79
cost savings, 220, 346347 fillet and corner radii for, 204 (T) Necking instability, 23
Coulombs friction model, 220 forging characteristics of, 156 (T) Necking mechanism, 23, 129
Coulombs law of friction, 224 section thickness, of rib-web forgings, 205 (T) Necklace-recrystallization behavior, 40
cutting modeling, 228229 Mo-0.5Ti Negative strain hardening. See flow softening
described, 340 forging characteristics of, 156 (T) Neighborhood, 40
design considerations of bulk forming, 10 Mo-0.5Ti-0.08Zr Neighborhood searching, 355
discretization, approximation, and searching opti- forging characteristics of, 156 (T) Net-shape forging, 149, 159
mization (DASA), 356357 Mo-25W-0.1Zr Net-shape forming, 145
dynamic material modeling (DMM), 340 forging characteristics of, 156 (T) Neural networks, 325
electrode cold forging process, 227 (F), 228 (F) Mo-30W Neutral angle, 241
Eulerian method, 223 forging characteristics of, 156 (T) Neutral plane, 233, 234 (F), 235
example simulation TZM Neutral point (neutral region)
crankshaft hot forging process, 228 for hot extrusion dies, 310 in flat rolling, 259
cutting modeling, 228229 Monte Carlo Step (MCS), 41 in strip rolling, 239
electrode cold forging process, 227228 Monte Carlo techniques, 40, 41 Neutral zone, 234
ring compression, 226227 Montmitonnet model, 270 Newton-Raphson method, 229, 370, 371
flow stress, 10 MoS2 lubricant, 57, 61, 157, 286 Nickel and nickel alloys, general. See also
forging behavior of, 157 Multi-axial fractures, 12 superalloys, general nickel-base.
in forming processes, 220231 Multi-step forging, 358359 carbides in, 165
friction, 230 Multiaxial stress states, 26 closed-die forging of, 163
goals of, 10 Multidisciplinary optimization (MDO) compatibility between materials and manufacturing
Hills general method, 220 Bezier curves, 358 processes, 141 (T)
the hot-tension test, 7980 CAD tools (software), 348, 349 drawing lubricants and coefficient of friction val-
isothermal hot-tension test, 7879 CAE tools (software), 348, 349 ues, 287 (T)
J2 flow rule, 223 collaborative optimization environment (COE), extrusion temperature classification, 292 (F)
Kronecker delta, 223 349 finish forging guidelines, 163
length-scales in, 11 components of, 349 flow softening during hot working, 148
lubrication, 230 constraints, 349 flow stress testing, 93
material characterization, 230 DEFORM (software), 348, 349, 350, 362 forging design limits, 209 (F)
material flow, 10 described, 346 forging for specific alloys, 163164
microstructural modeling, 10, 35, 3943, 339 development schedule, 350351 forging rate of, 163
microstructure evolution, 10 die designs, 346 forging reduction of, 163
microstructure optimization, 341 die geometry, 348 forging temperatures, 164 (T)
model resolution, 355 (F) examples of, 353 fracture mechanisms map, 27 (F)
modeling techniques, 340 finite element method (FEM), 346 grain size, 114, 165
Monte Carlo techniques, 40, 41 forging process, 347 (F), 351 guidelines for breakdown, 163
numerical techniques, 10 future work on, 365 guidelines for finish forging, 163
Piola-Kirchhoff stress tensor, 223 graphical user interface (GUI), 351 heat treatment of, 164
process conditions, 230 hardware requirements, 340341 high stacking-fault energy, 148
process optimization, 341 iSIGHT (software), 348, 349, 350, 351 hot forming pressures, 163 (T)
remeshing of elements, 223 isothermal forging optimization, 354 induction heating for, 93
research in, 230231 life prediction, 346 isothermal forging, 163, 166
rigid-plastic analysis, 223 manufacturing process simulation, 346 localized heating and, 163
rigid-viscoplastic analysis, 223 metalforming process optimization, 348349 lubricants and friction coefficients, 237 (T), 314
ring compression problem, 221 (F) near-net-shape forming, 348 (T)
shear friction law, 224 nonsmooth algorithms, 355 precipitates in, 165
slip-line field (SLF), 327 objective functions, 349, 357 recordkeeping for, 155
1119 Index 7/29/03 4:16 PM Page 404

Nickel and nickel alloys, general (continued) structure control phases and working tempera- M-252
static softening process, 36, 36 (F) ture ranges, 166 (T) critical melting and precipitation temperatures,
temperatures thermomechanical processing (TMP), 166 165 (T)
for melting, 165 (T) Alloy 903 (UNS N19903) forging temperatures and ratings, 155 (T)
for precipitation, 165 (T) forging practice for, 165 N 155
tensile elongation as a function of strain-rate sensi- Alloy 907 (UNS N19907) relative workability in forging and die-filling,
tivity, 77 (F) forging practice for, 165 194 (F), 210 (F)
thermal cracking of, 163 Alloy 909 (UNS N19909) Nimonic AP1
thermomechanical processing, 165166 forging practice for, 165 processing maps, 33 (F), 34 (F)
thermomechanical processing (TMP), 165 Alloy 925 Pyromet CTX-1
torsion testing, 115 forging practice for, 165 structure control phases and working tempera-
wire drawing die materials for, 289 (T) Alloy K-500 (UNS N05500) ture ranges, 166 (T)
Zener-Holloman parameter (Z) for, 114 cooling after forging, 163 R-235
Nickel and nickel alloys, specific types. See also ductile fracture map, 16 (F) forging temperatures and ratings, 155 (T)
Superalloys, specific types. forging practice for, 164 Ren 41
pure forging temperature, 164 (T) critical melting and precipitation temperatures,
SFE (stacking-fault-energy), low, 36 Alloy R-235 165 (T)
Alloy 200 critical melting and precipitation temperatures, design guide for, 154
cooling after forging, 163 165 (T) forging temperatures and ratings, 155 (T)
forging practice for, 163 Alloy U-700 hot working temperature range of, 73
forging temperature, 164 (T) forging difficulty of, 155 on-cooling Gleeble curves, 74 (F)
Alloy 214 Alloy X relative workability in forging and die-filling,
forging temperatures and ratings, 155 (T) critical melting and precipitation temperatures, 194 (F), 210 (F)
Alloy 230 165 (T) Ren 75
forging temperatures and ratings, 155 (T) Alloy X-750 (UNS NO7750) structure control phases and working tempera-
Alloy 301 (UNS N03301) critical melting and precipitation temperatures, ture ranges, 166 (T)
cooling after forging, 163 165 (T) superplasticity of, 166
forging practice for, 164 forging practice for, 164 thermomechanical processing (TMP), 165166
forging temperature, 164 (T) forging temperature, 164 (T) Ren 95 (Ni-14Cr-8Co-3.5Mo-3.5W-3.5Nb-3.5Af-
Alloy 400 hot deformation resistance, 163 2.5Ti)
cooling after forging, 163 hot-forming pressures, 163 (T) forging rate, 163
forging practice for, 164 Astroloy (UNS N13017) Superalloy 718, 132
forging temperature, 164 (T) critical melting and precipitation temperatures, U-500
hot-forming pressures, 163 (T) 165 (T) critical melting and precipitation temperatures,
Alloy 600 (UNS N06600) forging difficulty of, 155 165 (T)
cooling after forging, 163 forging rate, 163 forging temperatures and ratings, 155 (T)
forging practice for, 164 forging temperatures and ratings, 155 (T) U-700
forging temperature, 164 (T) structure control phases and working tempera- critical melting and precipitation temperatures,
hot deformation resistance, 163 ture ranges, 166 (T) 165 (T)
hot-forming pressures, 163 (T) superplasticity of, 166 forging temperatures and ratings, 155 (T)
Alloy 625 (UNS N06625) Hastelloy B NIKE2D program (software), 16
cooling after forging, 163 relative workability in forging and die-filling, Niobium and niobium alloys, general
forging practice for, 164 194 (F), 210 (F) in conventional controlled rolling, 253
forging temperature, 164 (T) Hastelloy C effect of, on Ar3 temperature, 253
hot deformation resistance, 163 relative workability in forging and die-filling, effects of, on recrystallization stop temperature
hot-forming pressures, 163 (T) 194 (F), 210 (F) (TRXN), 251252
Alloy 700 Hastelloy W forgeability behaviors of, 193 (F)
constant-strain-rate isothermal compression test, forging temperatures and ratings, 155 (T) forging characteristics, 156 (T)
198 (F) Hastelloy X grain refinement alloy, 253
flow-localization during hot forging, 198 (F) forging temperatures and ratings, 155 (T) and grain size, 9
Alloy 706 IN-100 for grain size control, 151
forging practice for, 164 billet temperature, 166 hot forging temperatures, 209 (T)
Alloy 718 (UNS N07718) for high temperatures, 93 microalloy benefits of, 272, 273
critical melting and precipitation temperatures, structure control phases and working tempera- in microalloyed steel forgings, 151
165 (T) ture ranges, 166 (T) precipitates, 249
forging practice for, 164 superplasticity of, 166 soluble
forging temperature, 164 (T) INCO 901 solute drag, 254
hot deformation resistance, 163 relative workability in forging and die-filling, solute drag, 253
hot-forming pressures, 163 (T) 194 (F), 210 (F) solute temperature of, 251
notched-bar upset testing, 195 (F) Inconel 600 (UNS N06600) Nitrides, 149
precipitate levels, 166 ductility influenced by axial stress during tor- No-slip point, 260
structure control phases and working tempera- sion, 107 Nodal parameters, 327
ture ranges, 166 (T) forging difficulty of, 155 Non-holomonic constraint, 339
Alloy 722 (UNS N07722) forging temperatures and ratings, 155 (T) Non-homogeneous (inhomogeneous) deformation
forging practice for, 165 intrinsic ductility material constant, 108 vs. homogeneous deformation, 260
forging temperature, 164 (T) rates of maximum axial stress to maximum shape factor, 263
Alloy 751 (UNS N07751) shear stress vs. values of shear strain at fail- Non-slip continuous machines, 284
forging practice for, 165 ure, 107 (F) Nonisothermal forging
forging temperature, 164 (T) Inconel 700 disk volume and press load, 355 (F)
Alloy 800 (UNS N08800) forging temperatures and ratings, 155 (T) initial and final shapes for, 354 (F)
cooling after forging, 163 Inconel 718 modeling complexities of, 228
forging practice for, 165 forging temperatures and ratings, 155 (T) optimal design process, 354
forging temperature, 164 (T) for high temperatures, 93 Nonisothermal sidepressing test, 198
hot deformation resistance, 163 stress-concentrations factor, 195 Nonisothermal testing, 49 (F)
Alloy 825 (UNS N08825) Inconel 751 Nonlinear viscoplastic FEM, 325
cooling after forging, 163 forging temperatures and ratings, 155 (T) Nonsmooth algorithms, 355
forging practice for, 165 Inconel 901 (UNS N09901) Normal and shear stresses, 22 (F)
forging temperature, 164 (T) forging difficulty of, 155 Normal plastic anisotropy parameter, 80 (F)
Alloy 901 forging temperatures and ratings, 155 (T) Normal reductions, 72
critical melting and precipitation temperatures, Inconel X-750 Normal strain, 12
165 (T) forging temperatures and ratings, 155 (T) Normalized flow-softening rate, 29
Gleeble curve for, 76 (F) K-Monel Normalized torque hardening (or softening) rate, 110
precipitate levels, 166 damage modeling of, 134 Norton-Hoff visco-plastic description, 267
1119 Index 7/29/03 4:16 PM Page 405

Index / 405
Notched-bar upset testing sensitivity analysis, 364 plasticine modeling, 134

effects of secondary tensile stresses on forgeability, sequential linear programming (SLP), 363 for process problem solving, 128 (T)
193 sequential quadratic programming (DONLP), 363 scale proportions, 203
nickel-base alloys, rupture sensitivity of, 195 (F) shape-sensitivity analysis, 364 similarity principle, 203
rating system for, 194 (F) software for, 348 Piercing operations, 108
stress-concentration factor, 194195 stability map, 340 Pilger mill, 283
Nucleation, probability of, 40 stable processing space, 339 Pinning force, 251, 254
Numbers, of test specimens, 189 state equations, 338 Piola-Kirchhoff stress tensor, 223
Numerical methods of solution, 1314 state-space (lumped parameter) method, 337, Pipe/piping, 4 (F), 6, 297, 300
Numerical techniques, 10. See also modeling meth- 338 Plane strain, 27, 51, 232
ods and techniques state-space models of dynamic crystallization, 343 Plane-strain compression, 29
continuum mechanics, 15 state-variable formulations, 338 Plane-strain compression testing (PSC), 51 (F)
dynamic problems, 13 subsystems, 338 before and after, 64 (F)
economics of, 17 successive approximation method (SAM), 363 barreling, 65
fundamental principles of, 17 system models, 342 friction at platen/specimen interface, 65
statically indeterminate system, analysis of, 13 techniques for, 348 lubrication for, 65
thermodynamic stability parameters, 339 for microstructural development, 65
trajectory optimization, 337 for rolling operations, 51
O trial trajectories, 342 for testing sheet or thin plate, 65
two-stage approach, 344 for thermomechanical processing, 51
Objective functions, 349, 357 weight factors, 342 true strain from, 65
Objectives, 337 Optimal die angle, 281, 289, 295, 295 (F) true stress from, 65
Oil-quenching, 359, 361 Optimal die half angle, 279 (F), 283 Plane-strain forging, 112
On-cooling data, 70, 72 Optimal die profile, 345, 345 (F) Plane-stress, 26, 173
On-cooling tension test, 192 Optimal half angle, 279, 280 Plane-stress and plain-strain, 26
On-heating curve, of Gleeble testing equipment, 69 Optimal ram velocities, 345 Plastic anisotropy parameter, 71, 76
One-dimensional model of roll separating forces, 262 Optimal working temperature, 105 Plastic deformation
Open-die forging Optimization Cockroft criterion, 177
for aluminum alloys, 158 of forging process, 348 heat from, 225
deformation processes, 144 of heat treatment, 348, 359360, 361362 Johnson-Cook equation for, 26
described, 146 of microstructures, 341 as material attributes, 10
plane-strain deformation, 188 two-stage approach, 341 (F) Plastic deformation processes, 30
side-pressing operations, 188 Optimization alternates, 357 Plastic flow, 26, 37
stainless steel, 151 Optimization procedure, 352 Plastic instabilities, 50, 61, 65
upsetting-operations, 188 Optimum temperature of ductility, 104 Plastic solid, 75
Open-die forging tests, 189193 Orange peel, 58, 297, 317 Plastic work, 225
Operating costs, 143 Orbital (rotary) forging, 159 Plasticine modeling, 129, 134, 134 (F)
Optical encoders, 92 Orientation distribution functions (ODFs), 99 (F) Plasticity analysis, 178, 340
Optical microscope, 131 Orientation of inclusions, 175 Plasticity-controlled growth, 81 (F)
Optimal control problem, 342 (F) Orowan and Pascoe equations, 66 Plasticity equations, 53
Optimal design process. See also modeling methods Orowan model, 262 Plasticity theory
and techniques; multidisciplinary optimization Ostwald ripening, 254 analysis methods for, 178
(MDO); multidisciplinary process design and Oxalted coatings, 287 finite element analysis (FEA), 178
optimization (MPDO) Oxide coatings, 287 modifications of, 271
ANSYS (software), 364 Oxides, grain size, control of by, 149 slip line method, 178
constraints, 337, 342 Oyane criterion, 271 von Mises yield criterion, 172
continuum sensitivity analysis, 364 Oyane ductile fracture criteria, 16, 16 (F) Plate rolling
cost control, 337 C-Mn plate workability test, 263
criteria, 337 calculated stress, three dimensional, 243 (F)
described, 337 P coefficient of heat transfer, 263
design approach, 341 deformation in, 260
design optimization tool (DOT), 362 Pancake, 195, 248, 253 deformation processes, 117
dynamic modeling, 338 Parabolic (power) law, 95 flow-stress behavior and torsion testing, 117
examples of, 343, 344 Parametric geometry, 327 nonhomogeneity of deformation, 263
exterior penalty method (EP), 363 Part complexity, 142 prediction of stresses and roll-separating forces,
FEM Explorer, 364 Partial differential equations, 14 242
forging, 358359 Partial-width indentation test, 51 (F), 52 shape factor, 263
forging process parameters, 339 Particle stimulated nucleation (PSN), 36 spread, 241, 242
geometry-based system application for, 350 Pass geometry, 233 strain distribution, 263
gradient-based approach, 364 Peak-ductility temperature (PDT), 69, 72 stress analysis of, 243 (F)
of grain size, 344 Pearlitic spheroidization, 115 (F) time-temperature profile of, 255 (F)
in heat transfer coefficients, 362 Penalty method, 224 Plutonium, 77 (F)
Hook-Jeeves direct search method, 363 Perfection as goal, 326 Pohlandt equation, 95
inputs as trajectories, 341 Performance measurement, 326 Point of instability, 64 (F)
iSIGHT (software), GUI interface, 351 (F) Phase changes, 192 Poisson equations, 11
Liapunov material stability criteria, 340 Phenomenology, of microstructure, 38 Poissons ratio, 240
material flow analysis, 339 Phosphate coating, 285, 287 Pole figures, 99 (F)
material process models, 338 Physical metallurgy, 10, 17 Polycrystal plasticity program (LApp), 42
material trajectory, 338 Physical modeling. See also modeling methods and Polycrystalline materials, microstructure changes to,
materials stability criteria, during forced dissipative techniques 77
flow, 339 boundary conditions, 203 Polyethylene, 287
microstructural trajectories, 344 for conventional hot forging, 204 Polymer coatings, 287
microstructures, modeling of, 339 of dead metal zones, 204 Polymethylmethacrylate (PMMA), 287
modified method of feasible directions (MMFD), of ferrous-powder compacts, 204 Polytetrafluoroethylene (PTFE), 287
363 finite element method (FEM), 134 Polyvinylchloride, 287
non-holomonic constraint, 339 friction, 203 Porosity, 6
objectives, 337 geometry, 203 Porous plasticity model, 16
optimality criteria, 342 grid techniques, 203 Post-dynamic static recrystallization (PDSR), 256
optimization exploration techniques, 363 heat-transfer, 203 Post recrystallization. See metadynamic recrystalliza-
penalty functions, 342, 343 (F) for isothermal hot forging, 204 tion (MRDX)
plasticity analysis, 340 of lap defects, 204 Power curve, 25 (F)
process parameters, 344 limits to, 129 Power dissipation, 18, 79
response surface methodology (RSM), 363 material selection, 203 Power-law relationship, 50, 94
1119 Index 7/29/03 4:16 PM Page 406

Power per unit volume, 33 temperature, 30 Reheat temperatures, 272

Precipitates, 165 upper temperature limit, 189 Relative density, 79
Precipitation, 40, 191, 252 working speeds, 197 Relative slip, 233, 234
Precision forging, 158, 211 workpiece preheat temperature, 189 Remeshing of elements, 16
Preforms workpiece temperatures, 197 Residual compressive stresses, 281
before and after finish, 215 (F) Processing history effects. See history effects Residual stresses, 281, 359
back extrusion part, 183 (F), 184 (F) Processing maps Resistance to wear, 10
blocker design, 212213 for certain metals, 17 (F), 18 (F), 32 (F), 33 (F), 34 Response surface, 355
deformation of, 182 (F) Response surface methodology (RSM), 363
design, 182 (F), 183 (F), 214 (F) dynamic material modeling (DMM), 17 Restoration processes, 250
for ball bearing race, 181 failure mechanisms of, 32 Reverse or forward extrusion, 160
cold-forging trials, 181 safe region in, 32 Reversing mill, 125
DEFORM (software), 357 using dynamic material models, 33 Rheostats (variable resistors), 92
for ferrous alloys, 212 Processing temperature, 12 Rib walls, 186187
guidelines for various metals, 213 Product yield losses, 68 Rib-web structural parts, dimensions of preform and
low alloy steel powder, 181 Production rate, 210 finished dimensions, 205 (T)
sintered 601AB aluminum alloy, 181 Production volume, 144 Rigid boundary model, 252
development of, 212 ProEngineer (software), 348 Rigid-plastic analysis, 223
dimensions for Proeutectoid ferrite, 254 Rigid-plastic finite element method (FEM), 262
aluminum alloys, 215 (T) Proportionality factor, 172 Rigid-viscoplastic analysis, 223
carbon steels, 214 (T) PSN (particle stimulated nucleation), 36 Ring compression modeling techniques, 226227
low-alloy steels, 214 (T) Punch design, 304 Ring compression testing. See also Sachs (slab)
titanium alloys, 215 (T) Pure metals method (SM)
dimensions of, 213 forgeability of, 192 analysis of, 5253
fir tree defect, 186 single-phase, ductility of, 103 bulging error, 226
geometry, 202 (F), 205 (F) Pure shear, 27 calibration curve for, 65 (F)
for H-shaped forgings, 214 (F) coefficients of friction, 201
metal distribution in, 212 critical friction value, 52
plastic-strain distribution, 358 (F) Q finite element method (FEM), 227
shape definition and defects, 357 (F) flow curve data, 201
sintered aluminum alloy powder, 186 Quality, 143 for flow stress in compression, 64
strains during forging of, 183 (F) Quartz dilatometer, 130, 130 (F) and flow stress measurement, 65
for superalloys, 154 Quenching, 70, 93 friction factors, 57, 201, 226
testing of, 182 geometry of, 53
Preheat temperature, 6970 lubricant varied, 201
Primary working process, 260 R for measuring interface friction, 64
Principal stresses specimens for, 64 (F)
defined, 22 R-model, 177 upper-bound method, 226227
as definition of state of stress, 26 Radiation heat, 226 Ring geometry, interface friction factor, 64
directions of, and yield criteria, 27 (F) Rate-sensitivity, 94 Ring rolling, 151, 160
Principal stresses and strains, 51 (F) Recovery Ring test, 52 (F)
Principle of maximum entropy production, 33 factors retarding, 250 Robustness and ease of use, in modeling techniques,
Principle of virtual work, 224 key points on, 149 230
Process-activity diagnosis, 326 in multiple hit tests, 133 Roll cladding, 236
Process analysis, 346 Recovery and recrystallization Roll coating buildup (roll pickup), 237, 238
Process conditions, 230 controlled rolling, 251252 Roll forging, 151, 159
Process controls, 3 models of recrystallization suppression, 251252 Roll interface limits, 264
Process design, 208219 Recrystallization, 35, 100 Roll/metal interface, 268
Process design methods, 3. See also finite element after deformation, 148 Roll pass design, 244, 245, 272
analysis method (FEA method/FEM); modeling factors retarding, 250 Roll pass schedules, 245, 246
methods and techniques; optimal design process; and flow softening, 148 Roll pressure, 240
Sachs (slab) method (SM) forms of, 253 Roll-separating forces, 117
analytical models, 326 key points on, 149 approximation of, 238
mesh-based models, 326 percent, 133 for carbon steels, 261
meshes in single hit tests, 133 in flat rolling, 259
Advancing Front triangulation methods, 328 suppression of, 253 one-dimensional model, 262
Delaunay triangulation method, 327 Recrystallization controlled rolling (RCR), 249, predicted vs. measured, 118 (F)
other modeling methods 251252, 254, 255 simplified method for estimating, 238239
slip-line field (SLF), 327 Recrystallization of specimens, 190 in strip rolling, 240
upper bound method, 327 Recrystallization percent, 131 upper bound method for, 242
simplified finite element method (FEM) models, Recrystallization stop temperature (TRXN), 248, 249, Roll stand, 244 (F)
326, 327 251252 Roll stress distribution, 239240
upper bound elemental technique analysis (UBET) Recrystallization temperature, 192 Roll torque
(two-dimensional analysis), 327 Reduction-anneal sequence, 131 approximation of, 238
Process measurements, 172 Reduction in area vs. temperature test, 53 (F) for carbon steels, 261
Process modeling, 3 Reduction of area in flat rolling, 259
Process modeling concepts, 10 ductility measure, from tension testing, 2526 predicted vs. measured, 118 (F)
Process optimization, 341 fracture-controlled defect formation, 192 two-dimensional models, 262
Process parameters, 344 from hot-tension testing, 71 upper bound method for, 242
Process problem solving, 128 (T) for low ductility materials, 72 Roller dies and rolling, 282
Process simulation to predict metal flow and forging Redundant work, 14, 280 Rolling, 232246. See also finite element analysis
stresses Refractory metals method (FEA method/FEM); flat rolling; spe-
investigation of defect formation in ring gear forg- compatibility between materials and manufacturing cific rolling process
ing, 216217 processes, 141 (T) of 2024 T351 aluminum alloy, 56 (F)
investigation of tool failure, 217 forgeability of, 156 of airfoil shapes, 244 (F), 245 (F)
multistage forging simulations of aircraft compo- forging characteristics of, 156 (T) back tension, 234
nents, 217218 lubricants and friction coefficients, 237 (T), 287 backward slip, 233
Process variables (T), 314 (T) of bimetallic strip, 236
die temperatures, 197 molybdenum and molybdenum alloys, 157 critical roll speeds, 241
friction, 31 niobium and niobium alloys, 156157 defects
lower temperature limit, 189 tantalum and tantalum alloys, 157 alligatoring, 236
strain, 30 tungsten and tungsten alloys, 157 barreling, 236
strain rate, 30 Reheat furnace, 272 crocodiling, 236
1119 Index 7/29/03 4:16 PM Page 407

Index / 407
ductile fractures, 263264 Rolling mills nonlinear material behavior, 369

during edge cracking, 173 components of, 258 shape design, 370 (F), 372
edge cracking during, 173 equipment review sheet, 271 shape optimization, 370
heat streak defect, 238 limits of, 264 state-space models, 369
laps in, 7 roll chilling, 265 strain-rate error with constant velocity die stroke,
roll coating buildup (roll pickup), 238 roll forces, 261 373 (F)
seams, 8 (F), 9 (F) roll interface limits, 264 test configuration for, 372 (F)
spalling, 237 roll stand, 243 updated Lagrangian mesh system, 369
surface cracking, 236 rolling torque, 117 Sequential linear programming (SLP), 363
deformation and grid system, 242 (F) schematic of hot strip mill, 272 (F) Sequential quadratic programming (DONLP), 363
deformation zone configuration, 244 (F) two-high mill, 258 Servohydraulic test equipment, 183
described, 145 work rolls Servohydraulic universal testing machine, 71 (F)
edge preparation and stress levels, 266267 characteristics of, 258 Severe cracking, 68
edge profiles of, 236 (F) cross-sections of, 245 Severe plastic deformation processes (SPD), 35
flow-dividing plane, 235 deformation of, 260 SFE (stacking-fault-energy), 35, 36
foil, 232 elastic deflection of, 240 Shape classification, 142 (F), 143 (F)
forward slip, 233, 235 elastic deformation of, 237 (F) Shape factor
fracture analysis, 264 electric discharge machining (EDM), 268 deformation type driven by, 260
friction, 234, 235, 236 schematic of, 259 (F) effect on field variable distribution, 274
friction hill, 236 surface preparation, 268 influence on defects, 267
cold hardening, 234 surface roughness effect, 268 workability effect, 274
described, 233234 Rolling operations Shape optimization, 351352, 357, 358
vs. speed, 242 (F) plane strain compression test for, 51 Shape rolling
variations, causes of, 241 strain-rate testing for, 92 caliber rolling, 242
front tension, 235 Rolling point entry, 233 cold corners, 269
geometric representation of rolled section, 245 (F) ROLPAS (software), 244, 245 computer-aided graphics for caliper shapes,
groove geometry for, 180 (F) Rotary forging and swaging, 146 245
guide to methods for, 128 (T) Rotary piercing, 108 elongation estimates, 243
homogeneous deformation, 233 Rotary variable-differential transformer, 92 finite element method (FEM) for, 242
homogeneous plane-strain, 260 (F) Round-bar specimens, 79 (F), 103 of flanged beams, 243 (F)
in hot-working during, 68 RPDPROD (software), 245 nonuniform deformation, 244 (F)
imperfections in rolled bar, 4 (F) roll pass design, 242, 243244, 245, 246
inhomogeneous deformation, 234, 235, 236 simplified-finite-element-slab-method (FSEM),
inhomogeneous plane-strain, 260 (F) S 245, 246
interface pressure, 236 (F) three-dimensional finite element method (FEM),
intermediate angle pass design, 245 (F) Sachs (slab) method (SM), 327 262
localized strains, 55 (F), 181 (F) approximate and closed form solutions, 14 uneven cooling, 269
lubricants and lubrication load estimation by slab method, 215 universal rolling, 242
functions of, 237238 load estimation by the simplified-finite-element- workability, 269
hydrodynamic lubrication, 241 (F) slab-method (FSEM), 216 Shaped dies, 292
lubrication entry zone, 240 (F) process design methods, 327 Shapes, 142, 232
material during, 175 (F) ring compression testing, 220221, 227 Shaving dies, 282, 282 (F)
material flow in, 236 (F) roll pass design, 245 Shear, 65
maximum shear stress (Tresca yield criterion), simplified-finite-element-slab-method (FSEM), Shear band tearing mode, 28
268 245, 246 Shear banding, with chilling, 192
metal flow in, 232 simplified slab method, 214215 Shear bands
models of, 261, 262 slab method of plasticity analysis, 63 adiabatic, 87, 112
neutral plane, 233, 235 TASKS (software), 246 equation for predicting, 111
neutral zone, 234 texture evolution modeling, 42 formation of, in compression test, 64 (F)
plane strain, 233 Safe region, 3233 as instability condition, 109
plane-strain assumption, 260 Sample heating, 68 in isothermal sidepressing, 200 (F)
process design, 271274 Schmids law, 22 material inhomogeneity, 8687
processes of, 232, 233, 241242, 258264 Schmidt trigger circuit, 92 occurrence of, 109
product thickness, 232 Scribe-line measurements, 100, 108 and strain softening, 64
products from, 232 fracture-controlled failure, 100 temperature nonuniformity, 87
reduction in thickness and tensile fracture strain, Seamless tubes, 283 tubular type specimen, 108
266 Seams, 8, 8 (F), 9 (F), 175 Shear flow stress, 27, 234
relative slip, 233, 234 Second phase conditions, 131 Shear friction factor, 210
representation of, 239 (F) Second-phase particles, 40 Shear friction law, 224
roll camber, 236237 Secondary deformation processes, 145 Shear modulus, 37
roll coating buildup (roll pickup), 237 Secondary operations, 7 Shear-related defects, 7, 8, 9 (F)
roll groove geometry, 55 (F) Secondary tensile and hydrostatic tests, 191 Shear strain, 12, 22 (F), 2223
roll pass designs, 243 (F) Secondary tensile stress state, 173 Shear strain rate, 47
roll schedule design, 271272 Secondary-tension test (STS), 51 (F), 52 (F) Shear strain to failure, 192
schematic of conditions, 233 (F) Section thickness of rib-web forgings, 205 (T) Shear stress
in secondary tensile stress state, 173 Segregation, 6 and coefficient of friction, 298 (F)
secondary tensile stress state in, 173 Sjournet process, 299 defined, 22, 47
sequence for various materials, 232 (F) Selenium, 152 in hot torsion testing, 123
simulation of, 117 (F) Sensitivity analysis Shear stress derivation for arbitrary flow laws,
slip line fields, 179 (F) adjoint variable formulation method, 368369 9596
state of stress diagram, 268 (F) analytical coordinate derivatives compared with Shear-stress/shear-strain, 94, 94 (F), 95, 99
strain distribution, 235 central difference approximations, 372 (F), Shear/torsion analogy, 86 (F)
stress and strain states of, 172 373 (F) Shear yield, 123
stress distribution calculation by elements, 240 (F) computational, 370 Shearing process, 144
stress distribution in, 239 (F) direct methods, 368 Sheet forming operations, 3, 222
stresses and forces, 238 (F) from equilibrium equations, 375 Sheet forming process, 4 (F)
strip shape, roll camber, 236237 flow formulations, 369, 371 Sheet metal, 76
temperature and surface cracking, 265 infinitesimal strain theory, 369 Sheet metal forming, 128 (T)
three-dimensional finite element method (FEM), of material properties, 13 Sheet rolling, 117
267 minimum work rate potential, 371 Sheet specimens, 79 (F), 82 (F)
workability concepts for, 172 mixed-boundary conditions, 369 Ship shape, 346
workability in, 264271 multidisciplinary optimization (MDO), 349 Shock loading, 10
Rolling load, 254 Newton-Raphson method, 370, 371 SHPROL (software), 242, 244245
1119 Index 7/29/03 4:16 PM Page 408

Side pressing, 179 (F) prebulging, 55 forging temperatures of, 152

Side-pressing test, 191, 192 (F) with shoulder deformation, 72 (F) high-energy forged extrusion, 29 (F)
Silicon steel, 88 shoulder design, 78, 89 low stacking-fault energy, 148
Similarity principle, 204 with shoulder undeformed, 72 (F) SFE (stacking-fault-energy), 36
Simplified finite element method (FEM) models, 326, for specific tests ultrasound testing of, 152
327, 328, 341 collar test, 200 (F) Stainless steels, general ferritic, 152
Simplified-finite-element-slab-method (FSEM), 245, compression test, 204 (F) annealing of, 153
246 double-cone test, 191 (F) finishing temperature limitation, 153
Simultaneous linear equations, 327 hot-compression testing, 63 (F), 64 (F) Stainless steels, general martensitic
Single phase alloys, 192 hot-tension testing, 68, 7172 cracking, 152
Skull, 310 notched-bar upset testing of, 194 (F) delta-ferrite formation in, 152
Slab method. See Sachs (slab) method (SM) ring compression test, 64, 64 (F) heat treatment of, 152
Sliding friction, 298 (F) torsion testing, 63 (F), 123 (F) hot forging temperatures, 209 (T)
Sliding of grains, 81 torsion testing, high-rate, 96 selenium added to, 152
Sliding velocity, 279 wedge-forging test, 189 (F) sulfur added to, 152
Sliding-wear coefficients, 288289 specimen size, 103, 103 (F) Stainless steels, general precipitation hardening,
Slimness ratio, 24 specimen surfaces, defects in. See also defects 152153, 193 (F)
Slip, 76. See shear on stress-strain curves, 88 hot forging temperatures, 209 (T)
Slip-line field (SLF), 177, 178 (F), 179 (F), 327 surface defects, 175 Stainless steels, specific types
Slip-line method thin-wall tubes, 110 13-8 Mo
for double indentation, 178, 179 thin-wall tubular compared to solid-bar, 88 compositions and forging temperatures of, 153
modeling techniques, 220221 variable sized, 191 (T)
for plane-strain analysis, 14, 221 variations of, 88 15-5 PH
plasticity theory, 178 on wrought materials, 88 compositions and forging temperatures of, 153
Slip (shear), 97 Spectral-methods, 15 (T)
Slip (twinning) system, 42 Spheroidization percent, 133, 134 (F) 15-7Mo
Slip vector (Burgers), 37 Spider, 292 relative workability in forging and die-filling,
Smoothness, 353 Spin forging, 159160 194 (F), 210 (F)
Software Split-furnace, 71 (F) 1.5%W steel
ABAQUS, 349 Springback, 3 reduction percentage at fracture, 268 (T)
ALPID program, 16 Square dies. See flat-face dies 16-25-6
ANSYS, 364 St. Venants principle, 89 relative workability in forging and die-filling,
CAD tools, 348, 349, 353 (F) Stability criterion, 18, 19 194 (F), 210 (F)
CAE tools, 348, 349 Stability map, 340 17-4 PH
DEFORM, 262, 348, 349, 350, 357, 362 Stability modeling, 17 compositions and forging temperatures of, 153
Dyna, 262 Stable grain diameter, 39 (T)
Elroll, 262 Stable processing space, 339 17-7PH
Expect, 350 Stacking-fault-energy (SFE), 35, 148 fillet and corner radii for, 204 (T)
finite element analysis (FEA) software, 16 Stainless steels, general relative workability in forging and die-filling,
Forge, 262 for anvils, 61 194 (F), 210 (F)
iSIGHT, 348, 349, 350, 351352, 360 closed-die forging, 151 section thickness, of rib-web forgings, 205 (T)
MARC, 262, 349 containing soluble carbides or nitrides 19-9DL/19DX
NIKE2D program, 16 forgeability behaviors of, 193 (F) compositions and forging temperatures of, 153
for optimal design process, 348 die-life for, 151 (T)
ProEngineer, 348 die lubrication, 153 201
ROLPAS, 244, 245 drawing lubricants and coefficient of friction val- forging difficulty level, 151
RPDPROD, 245 ues, 287 (T) 301
SHPROL, 242, 244245 flow stress-strain data of various metals, C and m forging difficulty level, 151
TASKS, 246 values, 381382 (T) form limiting diagram, 129
Unigraphics, 348, 349 flow stress-strain data of various metals, K and n 302
Solidus temperature, 192 values, 379 (T) compositions and forging temperatures of, 153
Solubility, 249 flow stress values of various metals, from ring (T)
Solute content, 29, 31 (F) compression tests of, 385 (T) forging difficulty level, 151
Solute drag flow stress values of various metals, from uniform hot-forming pressures, 163 (T)
effects of, 250 compression tests of, 385 (T) 303
grain boundary energy, 250 forging methods, 151152 compositions and forging temperatures of, 153
inhibition of grain growth, 254 forging methods for, 151 (T)
soluble molybdenum, 254 microstructure of, 151 forging difficulty level, 151
soluble niobium, 253, 254 heat treatment of, 153 fracture limits, 59
Solutioning temperature, 192 high chromium 303Se
Sound flow, 178 forgeability behaviors of, 193 (F) compositions and forging temperatures of, 153
Specialty steels, hot-workability ratings, 72 (T) high-nitrogen stainless steel (T)
Specific heat, 96, 128, 129 recrystallization dependence on rolling tempera- 304
Specimen reduction of area as a function of test tem- ture, 190 (F) ductility determined in hot torsion tests, 104 (F),
perature. See Gleeble curve recrystallized structure percentage, 190 (F) 105
Specimens and specimen geometry side pressed test, 192 (F) flow localization effects of heat transfer, 112
alpha parameter values, 110 lubricants and friction coefficients, 237 (T), 314 (T) forging difficulty level, 151
aspect ratios open-die forging, 151 forging temperature/pressure curve, 155 (F)
in direct-equilibrium method, 79 ring rolling, 151 forging temperature/upset reduction curve, 155
on ductility data, 88 roll forging, 151 (F)
fracture strain line for, 184 (F) specimen geometry, 103 (F) gage length-to-radius ratio on effective strain to
for Gleeble testing, 69 (F) stainless steel including selenium, forgeability be- failure, 103 (F)
Gleeble testing equipment haviors of, 193 (F) hot ductility and specimen geometry, 103
buttonhead, 68, 69 and strength at elevated temperatures, 151 in hot torsion testing, 49 (F)
grain size, 69 surface defects of, 152 relative workability in forging and die-filling,
grain width, 131 upset forging, 151 194 (F), 210 (F)
grip end design, 89 wire drawing die materials for, 289 (T) specimen geometry, 88
impact on hot ductility, 103 Stainless steels, general austenitic, 117 (F) truncated-cone indentation test, 201 (F)
impact on temperature profile, 103 deformation-processing map for, 86 (F) workability of, 104, 104 (F)
length-to-radius ratio, 88 delta-ferrite formation in, 152 304L
on microstructure, 88 ferrous alloys, 93 activation energy values, 106
minimum size, in grains, 131 finishing temperatures for, 152 ductility, 105
neck stress, 7879 flow curves, 37, 117 (F) flow curves, 100, 100 (F), 102
1119 Index 7/29/03 4:16 PM Page 409

Index / 409
flow curves for, 100 forging difficulty level, 151 No. 10

flow localization, 109 (F) in hot torsion testing, 49 (F) compositions and forging temperatures of, 153
flow-stress behavior and torsion testing, 100 selenium added to, 152 (T)
flow-stress behavior and torsion testing for, 100 sulfur added to, 152 Pyromet 355
grain size change, 114 workability of, 104, 104 (F) compositions and forging temperatures of, 153
high strain rates, 106 (F) 414 (T)
isothermal high-strain-rate flow curves, 100 compositions and forging temperatures of, 153 UNS S21800
shear bands, 109 (F) (T) compositions and forging temperatures of, 153
strain rate effect on torsional ductility, 105, 106 416 (T)
(F) compositions and forging temperatures of, 153 V57
strain-rate sensitivity, 100 (T) compositions and forging temperatures of, 153
strain-rate sensitivity in, 100 forging difficulty level, 151 (T)
torsional and tensile correlation, 107, 108 (T) 420 Star trials, 133
Zener-Holloman parameter, 105 forging difficulty level, 151 State of stress, 26
305 420F State-space (lumped parameter) method, 337, 341
compositions and forging temperatures of, 153 compositions and forging temperatures of, 153 State-space models, 343
(T) (T) States, time evolution of, 341
309 430 Static processes, 28
compositions and forging temperatures of, 153 compositions and forging temperatures of, 153 Static recrystallization (SRX)
(T) (T) after deformation, 148
finishing temperatures for, 152 forging difficulty level, 151 in aluminum, 36
forging difficulty level, 151 430F forms of, 253
309S compositions and forging temperatures of, 153 grain refinement and, 148
compositions and forging temperatures of, 153 (T) kinetics of, 3839
(T) 431 models of, 40, 43
delta-ferrite in, 152 compositions and forging temperatures of, 153 Static softening process, 36 (F)
310 (T) Statically indeterminate system, 13
compositions and forging temperatures of, 153 forging difficulty level, 151 Statistical design of thermomechanical testing, 133
(T) 440 Statistical mechanics in Monte-Carlo techniques, 40
finishing temperatures for, 152 forging difficulty level, 151 Steady-state, 11
forging difficulty level, 151 440A Steady-state dislocation density, 37
310S compositions and forging temperatures of, 153 Steady-state flow stress, 37
compositions and forging temperatures of, 153 (T) Sticking friction, 65, 298 (F), 299, 310
(T) 440B Strain
delta-ferrite in, 152 compositions and forging temperatures of, 153 affected by specimen geometry, 88
314 (T) axial, 58
delta-ferrite in, 152 440C defined, 22, 172
finishing temperatures for, 152 compositions and forging temperatures of, 153 deformation zones, 58 (F)
forging difficulty level, 151 (T) as free edge of forgings, 174 (F)
316 440F free-surface, 59
compositions and forging temperatures of, 153 compositions and forging temperatures of, 153 hoop, 58
(T) (T) increase of, with increase in die contact pressure,
flow curves, 38 (F) 443 187 (F)
forging difficulty level, 151 finishing temperature limitation, 153 levels achievable in torsion testing, 86
Hydrafilm process, 313 446 in multiple testpiece method, 95
317 finishing temperature limitation, 153 non-uniform, 189
compositions and forging temperatures of, 153 501 at peak stress, 38
(T) compositions and forging temperatures of, 153 as process variables affecting workability, 30
flow curves, 38 (F) (T) progression of, and decline of fracture line, 187 (F)
forging difficulty level, 151 502 and strain-rate hardening materials, 123
321 compositions and forging temperatures of, 153 torsional and tensile correlation, 106108
compositions and forging temperatures of, 153 (T) trajectories of, 344 (F)
(T) 636 upset cylinder, 58 (F)
finishing temperatures for, 152 compositions and forging temperatures of, 153 upset test measurements, 5758
forging difficulty level, 151 (T) Strain contours, 133 (F)
relative workability in forging and die-filling, A286 Strain distribution, 78 (F), 235
194 (F), 210 (F) relative workability in forging and die-filling, Strain fracture, 26 (F)
specimen geometry, 88 194 (F), 210 (F) Strain gages, 93
329 (Carpenter 7-Mo) AMS 5616 (Greek Ascoloy) Strain-hardening, 25
finishing temperature limitation, 153 compositions and forging temperatures of, 153 balances with dynamic recovery, 75
347 (T) in cold working, 31
compositions and forging temperatures of, 153 AMS 5700, 153 defined, 23
(T) compositions and forging temperatures of, 153 power-law relationship for, 50
finishing temperatures for, 152 (T) Strain-hardening coefficient, 2526
forging difficulty level, 151 Carpenter 20Cb-3 Strain-hardening exponent
348 compositions and forging temperatures of, 153 defined, 12, 25
compositions and forging temperatures of, 153 (T) in direct-equilibrium method, 79
(T) Carpenter 41 from torsion testing, 47, 102, 103
finishing temperatures for, 152 compositions and forging temperatures of, 153 true-stress/true-strain curves, 25 (F)
403 (T) Strain-hardening rate, 25, 37
forging difficulty level, 151 Carpenter 7-Mo (Type 329) Strain-induced precipitation of microalloyed austen-
notched-bar upset testing of, 194 finishing temperature limitation, 153 ite, 253
404 Carpenter 901 Strain localization, 19, 148
compositions and forging temperatures of, 153 compositions and forging temperatures of, 153 Strain-path slope, 59
(T) (T) Strain paths, 59 (F)
405 HiMark 250, 153 Strain rate, 77, 86
finishing temperature limitation, 151 finishing temperature limitation, 153 Strain-rate
forging difficulty level, 151 HiMark 300 defined, 25
410 finishing temperature limitation, 153 dependence upon temperature and flow stress at
compositions and forging temperatures of, 153 Lapelloy constant strain, 25
(T) compositions and forging temperatures of, 153 (T) effect on workability, 105
ductility determined in hot torsion tests, 104 (F), Lapelloy C fracture mechanism origins, 19
105 compositions and forging temperatures of, 153 (T) for hot-tension testing, 68
1119 Index 7/29/03 4:16 PM Page 410

Strain-rate (continued) finite element method (FEM), 238 Alloy 556 (UNS R30556 or N-155)
in multiple testpiece method, 95 friction directions during, 233 (F) forging description for, 155
as process variables affecting workability, 30 homogeneous deformation, 238 forging temperatures and ratings, 155 (T)
in strip rolling, 239 hydrodynamic lubrication, 240241 Alloy 718 (UNS N07718)
trajectories of, 344 (F) modeling of, 238241 on-cooling Gleeble curves, 74 (F)
Strain-rate change test, for strain-rate sensitivity, 26 neutral-point in, 239 Alloy 800 (UNS N08800)
(F) roll pressure in, 240 forging description for, 155
Strain-rate effect, 62 (F) roll-separating force and torque, 240 forging temperatures and ratings, 155 (T)
Strain-rate hardening, 76 roll-separating forces, 238, 240 Astrology
Strain-rate hardening exponent, 75 roll-separating torque, 240 design guide for, 154
Strain-rate history, 116, 116 (F), 117 roll stress distribution in, 239240 B-1900
Strain-rate sensitivity roll-torques, 238 design guide for, 154
in 304L alloy, 100 strain-rate in, 239 Hastelloy X
above homologous temperature, 62 stress distribution in, 242 design guide for, 154
defined, 12, 17, 25, 62 visco-plastic analysis, 238 Haynes 188
direct-equilibrium method, 80 (F) von Karman equations, 238 forging temperatures and ratings, 155 (T)
dynamic propagation of microcracks, 18 Strong solution, 224 HS 188, 74 (F)
dynamic recovery and, 19 Structural stability, 340 HS-25 (L-605)
dynamic recrystallization and, 19 Subgrain growth, 40, 4142 forging temperatures and ratings, 155 (T)
failure modes and, 83 Successive approximation method (SAM), 363 Inco 178
flow localization, 50, 109 (F) Sulfides design guide for, 154
as function of strain rate, 77 (F) effect on workability, 105, 105 (F) Inco 901
hot-tension testing failure mode determinant, 82 as machining additives, 149 design guide for, 154
inertia effects on, 117 Sulfur, 152 J-1570
localized deformation and, 18 Sulfuration, 287 forging temperatures and ratings, 155 (T)
necking and, 18 Sull coatings, 287 J-1650
shear band formations and, 18 Superalloys, general forging temperatures and ratings, 155 (T)
of shear stress, 94 deformation processes, control of, 154 J6-25-6
strain-aging, 117 die heating for, 154 forging temperatures and ratings, 155 (T)
strain-rate change test for, 26 (F) flow stress values of various metals, from ring M252
superplastic flow, 80 compression tests of, 385 (T) compositions and forging temperatures of, 153
of taper on specimen geometry, 80 flow stress values of various metals, from uniform (T)
titanium alloys, 168, 168 (T) compression tests of, 385 (T) N155
in torsion testing, 47 forgeability behaviors of, 193 (F) compositions and forging temperatures of, 153
values of, 26, 101 forgeability ratings, 154 (T)
in Waspaloy, 101 forging methods for, 153154 Pyromet 600
Strain-rate sensitivity exponent, 75, 76 hot-workability ratings, 72 (T) finishing temperature limitation, 153
Strain-rate sensitivity index in direct-equilibrium isothermal superplastic forming for, 154 Pyromet 718
method, 79 lubricants for, 154 compositions and forging temperatures of, 153
Strain-rate sensitivity map, 128, 129, 129 (F) methods used on, 153154 (T)
Strain-rate variation, 69 near-net-shape forming, 153 Pyromet 860
Strain rate vs. flow stress, 168 (F) preforms for, 154 compositions and forging temperatures of, 153
Strain softening, 50 (F) recrystallization requirements for, 154 (T)
Strain-softening, 64, 168 testing of, 154 S-186
Strain-softening material, 64 Superalloys, general cobalt-base, 155156 forging temperatures and ratings, 155 (T)
Strain-to-failure, 122, 123, 127 forging considerations for, 156 TD-nickel
Strength, 9, 22 forging pressure for, 156 design guide for, 154
Strength coefficient Gleeble curve for, 75 (F) U-57
in cold working, 147 grain growth of, 156 forging temperatures and ratings, 155 (T)
defined, 25, 25 (F) hot forging temperatures, 209 (T) Udimet 630
in direct-equilibrium method, 79 hot-tensile strength curves of, 73 design guide for, 154
for selected metals, 147 (T) Superalloys, general iron-base, 154155 Udimet 700, 115
Stress double-vacuum induction melting, 155 billet material, specimen microstructure, 115 (F)
cylinder, 57 (F) grain-boundary fracture, 154155 ductility determined in hot torsion tests, 104 (F),
defined, 22 hot forging temperatures, 209 (T) 105
states, 59 ordered intermetallics, 154 extrusion, specimen microstructure, 115
Stress and strain states, 172173 stringers from inclusions, 155 in hot torsion testing, 49 (F)
Stress concentrations, 58 vacuum melting, 155 microstructure of, 115
Stress-corrosion-cracking, 126 Superalloys, general nickel-base, 155156 optimal working temperature, 105
Stress distribution, 23 (F) for anvils, 61 strain rate effect on, 105, 105 (F)
Stress intensification factor, 7980 ductility determined in hot torsion tests, 104 (F), torsion testing, specimen microstructure, 115
Stress ratio, 59 105 torsion vs. extrusion, specimen microstructure,
Stress states hot forging temperatures, 209 (T) 116 (F)
determination of, 19 hot tensile strength curves of, 73 Unitemp HN(ESR)
influence of, on strain fracture, 26 (F) Superalloys, specific types. See also cobalt and cobalt reduction in area vs. temperature test on, 53 (F)
typical, 27 (F) alloys, specific types; nickel and nickel alloys, Unitemp HN(VSR)
Stress-strain curves. See also flow curves specific types reduction in area vs. temperature test on, 53 (F)
construction of, 54 16-25-6 Waspaloy (UNS N07001)
defined, 12, 23 relative workability in forging and die-filling, compositions and forging temperatures of, 153
for dynamic recovery, 28 (F) 194 (F), 210 (F) (T)
for dynamic recrystallization, 28 (F) A-286 (UNS S66286) critical melting and precipitation temperatures,
effects of specimen geometry variation on, 88 compositions and forging temperatures of, 153 (T) 165 (T)
ferritic low alloy steel, 117 (F) design guide for, 154 design guide for, 154
four hit tests, 132 (F) fillet and corner radii for, 204 (T) ductility determined in hot torsion tests, 104 (F),
under hot working conditions, 47 forging description for, 155 105
single hit tests, 132 (F) forging difficulty of, 155 flow curves, 48 (F), 100
Stress triaxiality factor, 78, 78 (F), 79 forging temperature/pressure curve, 155 (F) flow stress data analysis, 100101
Stress vs. strain, 78 (F) forging temperatures and ratings, 155 (T) forging temperatures and ratings, 155 (T)
Stringers, 155 relative workability in forging and die-filling, for high temperatures, 93
Strip rolling, 232 194 (F), 210 (F) in hot torsion testing, 49 (F)
critical roll speed, 241 section thickness, of rib-web forgings, 205 (T) relative workability in forging and die-filling,
deformation in, 260 Alloy-25 194 (F), 210 (F)
elastic deflection of rolls, 240 forging considerations for, 156 strain-rate sensitivity in, 101
1119 Index 7/29/03 4:16 PM Page 411

Index / 411
structure control phases and working tempera- melting point, 61, 165 (T), 189 Thermal conductivity, 130, 189
ture ranges, 166 (T) optimal, for hot extrusion, 307 Thermal conductivity/diffusivity, 128, 129, 130
superplasticity of, 166 optimal working temperature for forgings and ex- Thermal diffusivity, 130
thermomechanical processing (TMP), 165 trusion, 108 Thermal fatigue wear, 289
X-1900 optimal working temperature for rotary piercing Thermocouples, 62, 68, 69
design guide for, 154 operations, 108 Thermodynamic stability parameters, 339
Superplastic behavior, 149 at peak, in drawing, 280 Thermomechanical design, 346
Superplastic deformation, 81 (F) phase change, 189 Thermomechanical history, 116
Superplastic diffusions growth, 81 (F) for precipitation, 165 (T) Thermomechanical process simulations, 15
Superplastic flow, 80 as process variables affecting workability, 30 Thermomechanical processing (TMP)
Superplastic forming, 132 recrystallization, 189 defined, 248
Superplastic materials, 75, 129, 132 recrystallization stop temperature (TRXN), 249 effect of, 35
Superplastic region, 77 of solute niobium, 251 grain size and, 9
Superplasticity, 76 trajectories of, 344 (F) of microalloyed bar, 254255
Surface and edge cracking, 265 (F) transformation temperatures, 249 nickel-base alloys, 165
Surface-condition factors, 58 workability and, 9 plane-strain compression test for, 51
Surface cracking Temperature buildup and flow localization defects, references to, 272273
of 5182 Al alloy, 265 (F) relationship of, 31 Thermomechanical testing
from die contact, 186 (F) Temperature compensated strain-rate parameter. See composite rotatable design, 133
from drawing, 317 (F) Zener-Holloman parameter (Z) compression test, 133
from extrusion, 317 (F) Temperature distribution, diagrams of, 262 (F), 263 damage modeling in, 134, 134 (F)
fracture line limit, 55 (F), 264 (F) defects observed in, 128
hot steel strip, 266 (F) Temperature limits, 167 design of tests, 131
lubricant failure, 265 Temperature sensitivity, 17 effects of testing on material tested, 128
roll chilling, 265 Temperature-strain rate conditions, 37 finite element method (FEM) analysis of, 132
and rolling temperature, 265 Temperature variations, 341 grain size, 131
testing for, 265 Tensile deformation energy condition, 177 guide to process tests, 128 (T)
from thick-film lubrication, 317 (F) Tensile ductility, 68, 83 parameters of, 133
transition zones, 186 (F) Tensile fracture, modes of, as function of tempera- process tests, 128
Surface defects, 152. See also defects ture, 27 (F) purpose of, 128
Surface fatigue wear, 289 Tensile hydrostatic stress, 179 recrystallization equation, 131
Surface fracture strain, 104 Tensile specimen neck, 23 (F) recrystallization percent of, 131
Surface fractures, 185 Tensile strain, 134, 184 (F). See also strain results analysis, 131
Surface scribe line, 89, 9394 Tensile strain energy per unit volume, 32 second phase conditions, 131
Surface-to-volume ratio, 142 Tensile stress, 68, 177. See also stress single and multiple hit tests, 132133
Symmetry plane, 78 Tensile stress criterion, 178179, 183 source of problems, 128, 129
System models, 342 Tension-based processes, 144 statistical design of, 133
Tension machine characteristics, 68 test efficiency, 131
Tension-test frames, 70 texture, 131
T Tension testing Thermomechanical treatments, 248 (F), 273 (F)
area reduction at fracture, 320 Thermoplastic plastics, 146 (F)
Taguchi method, 271 Cockroft and Latham analysis, 320 Thermoplastics, 141 (T)
Tantalum and tantalum alloys, general comparison with other hot working simulators, 122 Thermoset plastics, 141 (T), 146 (F)
coatings for, 157 (T) Thick-wall torsion testing, 110
forgeability behaviors of, 193 (F) deformation to fracture vs. mean hydrostatic stress Thin-wall tube testing, 95
forging characteristics of, 156 (T), 157 and flow stress, 293 (F) Thin-wall tubes
hot forging temperatures, 209 (T) described, 320 shear stress derivation for arbitrary flow laws,
lubricants and lubrication for, 157 ductility from, measurement of, 24 95
Tantalum and tantalum alloys, specific flow stress level, 320 shear stress vs. shear strain, 94
99.8% Ta fracture-controlled defect formation, 192 specimen design, 88
hot tensile strength curves of, 73 necking instability in, 23 Three-dimensional finite element method (FEM),
TASKS (software), 246 plastic instability in, 63 262, 267
Taylor approach, 42 tests for initial forging temperature and strain rate, Three-dimensional Mohrs circle, 264, 265
Techniques checklist, 350 192 Three dimensional stress states, 173
Techniques for optimal design process, 348 Tertiary scales, 265 Three point bend test, 50 (F)
Teflon lubricant, 186 Test equipment. See also Gleeble testing equipment; Titanium aluminides, general
Temperature. See also forging temperatures specific testing cavitation affected, 80
of austenite recrystallization, 9 closed-loop temperature controllers, 68 deformation and failure of, 84 (T)
austenite to ferrite temperature (Ar3), 249, 253 components, 8990 engineering stress-strain curves of, 74
bainite-start temperature (Bs), 249 digital-to-analog (D/A) converter, 92 failure modes of, 82
control for hot-tension testing, 68 grips for, 93 flow localization, micrographs of, 83 (F)
decrease in ductility at, 105 heating systems, 93 flow localization of, 82
derivative of stress in terms of, 62 (F) hydraulic motors and actuators, 9192 fracture controlled failure of, 82
of drawing process, 278 load cells, 9293 grain growth in, 42
ductile-to-brittle transition temperature (DBTT), motors, electric, 9091 plastic anisotropy parameter, 76
270 protective coatings, 93 width versus thickness strain, 77 (F)
effect on compressive flow curve, 61 pyrometers, 68 Titanium aluminides, specific types
effect on ductility, 103104 rheostats (variable resistors), 92 super alpha2 titanium aluminide
effect on flow stress, 77, 168 rotary variable-differential transformer, 92 flow-softening during hot working, 148
effect on microstructural properties, 77 servohydraulic, 183 Ti-21Al-22Nb
effect on workability, 116 test frame design, 90 anisotropy parameter vs. local axial true strain,
effects of, in hot-compression testing, 61 thermocouples, 68 77 (F)
equicohesive, 149 twist sensors, 9293 engineering stress-strain curves, 76 (F)
of forged material, 210 water cooled grips, 93, 93 (F) engineering stress-strain curves for, 76 (F)
fracture mechanisms, 19, 86 Tests for workability, 189201. See also specific test- plastic anisotropy parameter, 76
grain-coarsening temperature (TGC), 249 ing Titanium and titanium alloys, general, 166170
grain growth and, 9 Textural softening, 87 cavitation affected, 80
homologous, effect on strain rate, 62 (F) Texture, 87, 98 (T), 131 classes of, 166167
hot rolling, selection for, 125 Texture evolution, 4243 compatibility between materials and manufacturing
for hot-tension testing, 54, 70 Theoretical calibration curve, 52 (F) processes, 141 (T)
hot working, 68, 148 Theoretical fracture models and criteria, 177179 cracking, 167
incipient melting problems, 105 Theoretical fracture stress criteria, 185 deformation of, 51
of insert, after cutting, 230 (F) Thermal conduction, 96 die cooling, 169
1119 Index 7/29/03 4:16 PM Page 412

Titanium and titanium alloys, general (continued) in hot-compression testing, 61 relative workability in forging and die-filling,
die preheating, 169 in isothermal hot-compression test, 50 (F) 194 (F), 210 (F)
die temperature ranges, 169, 169 (T) isothermally sidepressed, 199 (F) section thickness, of rib-web forgings, 205 (T)
drawing lubricants and coefficient of friction val- recommended forging temperatures, 167 (T) strain-rate sensitivity, 129, 168, 168 (T)
ues, 287 (T) strain-rate sensitivity, 168, 168 (T) superplastic region of, 77
effect of deformation rate, 168169 temperature effect on flow stress, 168 temperature effect on flow stress, 168
effect of temperature on flow stress, 168 transverse metallographic sections, 199 (F) tensile elongation as a function of strain-rate
effects of, in preventing grain growth, 252 Ti-13V-11Cr-3Al sensitivity, 77 (F)
extrusion temperature classification, 292 (F) recommended forging temperatures, 167 (T) tension test results, 196 (F)
and flow stress, 168 (F) relative workability in forging and die-filling, Ti-6Al-4V ELI
flow stress testing, 93 194 (F), 210 (F) recommended forging temperatures, 167 (T)
flow stress values of various metals, from ring Ti-15V-3Cr-3Al-3Sn Ti-6Al-6V-2Sn
compression tests of, 385 (T) recommended forging temperatures, 167 (T) recommended forging temperatures, 167 (T)
flow stress values of various metals, from uniform Ti-17 (T1-5Al-2Sn-2Zr-4Cr-4Mo) Ti-8Al-1Mo-1V
compression tests of, 385 (T) recommended forging temperatures, 167 (T) recommended forging temperatures, 167 (T)
forging of, 166 Ti-3Al-8V-6Cr-4Mo-4Zr (Beta C) Ti-8Al-1Mo-!V
forging processes, 167168 recommended forging temperatures, 167 (T) strain-rate sensitivity, 168, 168 (T)
forging temperatures, 167 (T), 168 (F) Ti-4.5Sn-6Zr-11.5Mo (Beta III) temperature effect on flow stress, 168
hot forging temperatures, 209 (T) recommended forging temperatures, 167 (T) Ti-8Al-8V-2Fe-3Al
induction heating for, 93 Ti-49Al-2V recommended forging temperatures, 167 (T)
insulation, 170 effect of strain and strain rate on percent spher- Ti-Al samples
lubricants, 169170 oidization, 134 (F) true-stress/true-strain curve for, 76 (F)
lubricants and friction coefficients, 237 (T), 314 Ti-5Al-2.25Sn Ti-CP
(T) tensile elongation as a function of strain-rate recommended forging temperatures, 167 (T)
microalloy benefits of, 272 sensitivity, 77 (F) Ti-Zr-Mo (TZM) alloy
microstructure of, 166 Ti-5Al-2.5Sn for anvils, 61
oxidation levels, 166 recommended forging temperatures, 167 (T) Transage 129 (Ti-2Al-11.5V-2Sn-11Zr)
phosphate coating, 287 relative workability in forging and die-filling, recommended forging temperatures, 167 (T)
preform (blocker) design, 213 194 (F), 210 (F) Transage 175 (Ti-2.7Al-13V-7Sn-2Zr)
preforms, design guidelines for, 213 Ti-5Al-6Sn-2Zr-1Mo-0.1Si recommended forging temperatures, 167 (T)
rib-web structural parts, 205 (T) recommended forging temperatures, 167 (T) Titanium-nitride, 9
shear friction factor, 210 Ti-6242Si (Ti-6Al-2Sn-4Zr-2Mo-0.1Si) Tolerances
strain-softening, 168 acicular Widmansttten alpha microstructure, costs of, 143
temperature control and heating, 169 109 (F) dimensional, 143
titanium precipitates, 249 below transus temperature preheat, flow stress forging equipment impact on, 210
wire drawing die materials for, 289 (T) of, 198 (F) geometric, 143
Titanium and titanium alloys, general alpha below transus temperature preheat, shear-band- guidelines for, 143
described, 166 ing of, 198 (F) specification of, 140
recommended forging temperatures, 167 (T) at beta transus temperature, 198 (F) Tool and design rules, for process problem solving,
Titanium and titanium alloys, general alpha-beta constant-strain-rate isothermal compression test, 128 (T)
alpha-beta forging, 167168 198 (F) Tool steels, general
described, 166 equiaxed alpha microstructure, 109 (F) for platens, 61
grain coarsening of, 115 flow-localization-controlled failure test, 198 Tool steels, specific types
torsion testing, 115 (F) D2
Widmansttten alpha microstructure of, 115 flow-localization during hot forging, 198 (F) for aluminum extrusion, 304 (F)
Titanium and titanium alloys, general beta flow stress, 198 (F) die inserts, 304
all-beta titanium alloys isothermally sidepressed, 199 (F) H11
forgeability behaviors of, 193 (F) nonisothermal sidepressing test, 198 (F) for aluminum extrusion, 304 (F)
alloys in beta phase preheat path, effects of, 198 (F) die material, 216
high-stacking-fault-energy (SFE), 35 shear banding, 198 (F) fillet and corner radii for, 204 (T)
beta forging, 167 shear bands in isothermal sidepressing, 200 (F) relative workability in forging and die-filling,
acicular microstructure, 168 strain-rate sensitivity of, 109 194 (F), 210 (F)
compared to alpha-beta forging, 168 thick-wall torsion tests, 110 section thickness, of rib-web forgings, 205
Widmansttten microstructure, 168 torque-twist curve, 111, 111 (F) (T)
described, 167 transverse metallographic sections, 199 (F) H12
high-stacking-fault-energy, 35 upset tested axial cross section of, 197 (F) in dies, 307
Titanium and titanium alloys, general gamma workability diagrams, for colony-alpha mi- in punches, 307
compression test, 133 crostructure, 199 H13
Titanium and titanium alloys, specific types workability maps, 200 (F) for aluminum extrusion, 304 (F)
commercially pure working speed, effects of, 198 (F) in dies, 307
flow softening during hot working, 148 Ti-6Al-2Nb-1Ta-0.8Mo in punches, 307
Corona 5 (Ti-4.5Al-5Mo-1.5Cr) recommended forging temperatures, 167 (T) H21
recommended forging temperatures, 167 (T) Ti-6Al-2Sn-2Zr-2Mo-2Cr for aluminum extrusion, 304 (F)
IMI 550 (Ti-4Al-4Mo-2Sn) recommended forging temperatures, 167 (T) H46
recommended forging temperatures, 167 (T) Ti-6Al-2Sn-4Zr-2Mo(+0.2Si) compositions and forging temperatures of, 153
IMI 679 (Ti-2Al-11Sn-4Zr-1Mo-.025Si) recommended forging temperatures, 167 (T) (T)
recommended forging temperatures, 167 (T) Ti-6Al-2Sn-4Zr-6Mo L6
IMI 685 recommended forging temperatures, 167 (T) for aluminum extrusion, 304 (F)
recommended forging temperatures, 167 (T) Ti-6Al-2Zn-4Zr-2Mo-0.1Si M2
IMI 700 (Ti-6Al-5Zr-4Mo-1Cu-.02Si) grain-boundary voids, 192 (F) die inserts, 304
recommended forging temperatures, 167 (T) Ti-6Al-4V, 129, 166 punch material, 304
IMI 829 cavitation in, 81 (F) M4
recommended forging temperatures, 167 (T) cavity growth parameter in, 197 (F) die inserts, 304
IMI 834 equiaxed microstructure, 77 punch material, 304
recommended forging temperatures, 167 (T) fillet and corner radii for, 204 (T) M42
IMI685 (Ti-6Al-5Zr-4Mo-0.1Si) flow softening during hot working, 148 hot workability of, 73
processing maps, 271 flow stress of, 77 (F) on-cooling Gleeble curves, 74 (F)
T1-6242 beta flow-through defect, 204 (F) S1
processing map for, 18 (F) flow-through defects, 8 (F) for aluminum extrusion, 304 (F)
Ti-10V-2Fe-3Al hot-tension testing, 78 T1
constant-strain-rate isothermal compression test, Hydrafilm process, 313 for aluminum extrusion, 304 (F)
198 (F) laps in bulkhead forging, 203 (F) W1
flow-localization during hot forging, 198 (F) recommended forging temperatures, 167 (T) in aluminum extrusion, 304 (F)
1119 Index 7/29/03 4:16 PM Page 413

Index / 413
Tool wear, 229 test frame design, 90 Uniaxial compressive flow stress, 50
Tooling, 292, 304, 306, 307 torque sensors, 9293 Uniaxial compressive stress, 4950
Torque and angle of elastic unloading, 96 torque transducers, 9293 Uniaxial flow curve, 99
Torque cells, 93 twist sensors, 92 Uniaxial tensile strains, 52
Torque-equilibrium equation, 94 types of, 9293 Uniform elongation, 23
Torque maximum (torque instability), 110 water cooled grips, 93, 93 (F) Unigraphics (software), 348, 349
Torque softening, 109 Torsional ductility, 117 Universal gas constant, 17, 26, 77, 101
Torque twist behavior, 108109 Total factor productivity (TFP), 326 Universal rolling, 242
Torque-twist curve, 96, 109 Total Lagrangian (TL), 223 Universal testing machines, 70
Torque-twist data, 94 Total tensile elongation, 7071 Unstable flow (buckling mode), 226
Torsion analysis, 94 Total tension elongations, 8384 Unwins equation, 71
Torsion flow curves, 113 Trajectories, 341, 344 (F) Updated Lagrangian method (UL), 223, 229
Torsion fracture data, 103, 103 (F) Trajectory optimization, 325, 337, 341 Upper and lower bounds, 131
Torsion/shear analogy, 86 (F) Transfer table/coil box, 272 Upper bound elemental technique analysis (UBET),
Torsion testing Transformation temperatures, 249 327
benefits of, 86 Transport phenomena, 10 Upper bound method, 220, 327
at cold and worm working temperatures, 101 Transverse metallographic sections, 199 (F) axial flow stress, 63
at cold working temperature, 101102 Treatment matrix, in central composite rotatable de- centerbursts, 178, 318
comparison with other hot working simulators, 122 sign, 133 contact surface fracture with internal fracture, 183
(T) Tresca criterion, 12, 103 deficiency of, 43
cooling history in, 116 Tresca yield criterion, 2627, 268 described, 221222
at decreasing temperatures, 117 Triaxial stresses, 23, 24 energy principle basis, 221
deformation to fracture vs. mean hydrostatic stress Triple-point cracking, 192 external work required, 178
and flow stress, 293 (F) True fracture strain, 24 in extrusion, 178
described, 320321 True fracture stress, 24 finite element method (FEM) compared with, 328
ductility data compared to other workability data, True normal stress, 23 flow mode for defect formation, 179
106108 True strain jump condition, 221
for effects of strain rate-history, 117 axial, 58 kinematically admissible velocity field, 14
failure modes, 86 circumferential, 58 modeling techniques, 221222
fixed-strain rate, 117 in compression testing, 62 of plasticity analysis, 178
flow-stress data, 86, 100, 101 constant, in torsion testing, 86 for predictions of spread, metal flow and roll
flow-stresses at decreasing temperatures, 117 defined, 23, 74 torque, 242
flow tests data comparison with other workability determination of, 24 ring compression test, 226227
tests, 101 in drawing, 279 software for, 242
fracture-controlled defect formation, 192 in extrusion, 295 studies of alligatoring, 267
under hot working conditions, 47 from plane-strain compression test (PSC), 65 texture evolution modeling, 42
at hot working temperature, 101102 True strain-rate, 25, 48 theoretical fracture models and criteria, 178
interpretation of, 86 True stress upper-bound elemental technique (UBET), 221
material considerations, 87 defined, 74 upper-bound theorem, 221
as means of determining microstructure develop- determination of, 24 in wire drawing, 178
ment, 116 from plane-strain compression test (PSC), 65 Upset forging
microstructure deformation parameters, 114 True stress-strain curve of aluminum alloys, 159
multiphase alloys, 115 plastic instabilities in, 50 forging pressure vs. forging temperature, 155 (F)
procedures, 87 in uniaxial compression, 51 forging severity, 152 (F)
to replicate thermo-mechanical history, 116 True-stress/true-strain curves (flow curves), 23 (F) of various metals, 151, 152, 155 (F)
results compared to production experience, for Al-8090 alloy, 76 (F) Upset-test reductions at fracture, 192
117 defined, 23, 24 Upset test representation, 173 (F), 180 (F)
shear bands, 86 and strain hardening coefficient, 25 (F) Upset testing. See also cold upset testing; cylindrical
specimen design for, 8788 for Ti-Al samples, 76 (F) compression test
specimen geometry, 103, 103 (F), 123 (F) True-stress/true-strain rate curves, 63 (F) axial cross section, 197 (F)
specimen with inclusions, 87 (F) True tensile strain, 24 (F) friction effects in, 29 (F)
strain-hardening exponent in, 47 Truncated-cone indentation test, 201 (F) specimen geometry and preparation for, 194 (F)
strain-rate history, 116 Tube drawing, 282283, 283 (F) as workability measurement, 59
strain-rate sensitivity, 47 Tube extrusion, 302 (F), 307 (F) Upsetting-operations, 212, 301 (F), 373 (F)
surface-region fracture, 320321 Tube piercing, 144
for temperature nonuniformity in finite element Tube sinking, 283
method (FEM), 97 Tubular specimens, 96, 103 V
tests for initial forging temperature and strain rate, Tungsten and tungsten alloys, general
192 in composite extrusion billet, 299 (F) Vacancy condensation, 81
torsion flow curves from, 113 forgeability, 157, 193 (F) Vacuum chambers, 70
tubular specimens, 96 forging characteristics of, 156 (T) Value, 326
typical flow-stress and strain to failure as a func- hot forging temperatures, 209 (T) Value flow, uninterrupted, 326
tion of deformation, 123 (F) lubricants for, 157 Vanadium and vanadium alloys, general
typical stress-strain curve, 123 (F) wire drawing die materials for, 289 (T) bainitic microstructure, 272273
work-hardening rates impact on, 101102 Tungsten carbide, 61, 304305 effects of, on recrystallization stop temperature, 252
Torsion testing equipment Turbine disks, 353 (F), 360 (F) vanadium as microalloy, 151, 255
axial load control, 90 Twist rate, 88 vanadium precipitates, 249
clutch/brake assembly, 91 Twist sensors, 92 Variable resistors, 92
digital-to-analog (D/A) converter, 92 Twists to failure, as measure of ductility, 104 Visco-plastic analysis, 238
for fatigue, 91 Two-dimensional analysis, 327 Visioplasticity analysis, 185, 186
grips for, 93 Two-dimensional models, 262 Void coalescence phenomena, 57, 177
heating systems, 93, 93 (F) Two-phase alloys, 115 Void growth, 28 (F), 177
hydraulic motors and actuators, 9192 Two-slice approach, 83 Void growth model, 177
load cells, 90, 9293 Two-stage approach, 344 Void growth stage, 28
motors, electric, 9091 Void initiation, 28, 28 (F)
motors, hydraulic, 91 Void nucleation, 105
protective coatings, 93 U Voids. See cavities and cavitation
rheostats (variable resistors), 92 Volume fraction, 28 (F)
ringing, 90, 91 Ugine-Sjournet process, 299, 310 von Karman equations, 238
rotary variable-differential transformer, 92 Ultimate tensile strength, 23, 24 von Mises
rotation transducers, 92 Ultrasonic methods, 6 effective strain, 88
shear-strain rates, 91 Uniaxial compression, 51, 122 normalized torque hardening (or softening) rate,
strain rates, 90 Uniaxial compression testing, 57, 122 (T), 149 110
1119 Index 7/29/03 4:16 PM Page 414

von Mises (continued) Wedge test specimens, 131, 131 (F) Workability parameter, 26, 26 (F)
shear-stress/shear-strain conversion to effective Weight reduction strategy, 348 Working temperatures, 166 (T)
stress-effective strain, 99, 101, 102 Wet drawing, 284 Workpiece-tool interface, 29
surface effective strain rates plotted, 105, 106 (F) Widmansttten alpha microstructure, 115 Wrought alloys, 31 (F)
von Mises criterion, for shear flow stress, 234 Widmansttten microstructure, 168 Wrought forms, 3
von Mises yield criterion, 47 Width versus thickness strain, 77 (F) Wrought materials, 30
defined, 12, 27 Wire drawing
plasticity theory, 172 carbide die cross section, 278 (F)
centerburst formation predictions, 179 (F) Y
descaling, 288
W die angles, 178 Yada equations, 343 (F)
die friction, 178 Yield criteria (flow rules), 12, 2728
Warm extrusion, 218 friction effects, 289 Yield strength, 147 (F)
Warm forging, 151 pickling, 288 Yield stress, 94
Waste reduction, 326 pressurized chamber, 285 (F) Yield/ultimate tensile strength, 71
Water quenching, 70 process of, 278 (F) Yielding criteria, 26
Wave equations, 11 relative draw stress, 281 (F) Yielding (plastic flow), 27
Weak solution, 224 schematic, 281 (F) Youngs modulus, 23
Wear resistance, 10 stress and strain states of, 172
Web-thickness, 145 surface preparation, 287288
Web thickness-to-depth, 142, 147 Wire-drawing machines, 284 Z
Wedge cracking failure, 28, 86 Wire/rod drawing, 128 (T)
Wedge-forging test Wiredrawing steels, 287 (T) Z-model, 177
crystal structure, 192 Wolfs ear, 87 Zener-Holloman parameter (Z)
effects of secondary tensile stresses on forgeability, Woodfords trend line, 83 for certain metals, 101, 114
193 Work, 42, 106 defined, 26
finite element method (FEM) predictions for, 190 Work-hardening, 3839, 148 dynamic recovery, 37, 38
(F) Work-hardening exponent, 147, 147 (F), 147 (T) flow stress, effect of strain rate and temperature
flow-localization-controlled failure, 192 Work-hardening rates, 94, 101102 on, 77
forgeability, 192 Work-in-progress, 326 flow stress, of single phase alloys, 105
forging-temperature selection, 191 Workability. See also bulk workability testing; de- strain rate and temperature dependence, 115
fracture-controlled defect formation, 192 fects; modeling methods and techniques; specific Zener pinning, 39, 42
material and physical properties, 192 deformation processes; specific testing Zero-ductility temperature (ZDT), 69, 7273
material aspects, interaction between, 192 concepts, 172 Zinc and zinc alloys, general
on-cooling tension test, 193 defined, 4, 10, 22, 35, 5455, 86, 188 compatibility between materials and manufacturing
phase changes, 192 described, 149 processes, 141 (T)
physical modeling of, 205 (F) determinants of, 19 extrusion temperature classification, 292 (F)
recrystallization temperature, 192 projecting workability from mechanical tests, for hot extrusion, 308
side-pressing test, 191 319321 maximum dimension of web, 146 (F)
solidus temperature, 192 summary, 205 phosphate coating, 287
solutioning temperature, 192 theory and application, 172187 Zinc and zinc alloys, specific types
specimen geometry for, 189 (F) Workability analysis, 54 (F) Zn-22AL
test results, 190, 191 Workability applications, 179187 deformation and failure of, 84 (T)
test specimens, 189, 190 Workability constant, 12 Zirconium alloys, specific types
tests for initial forging temperature and strain rate, Workability fracture criteria, 32 Zircaloy-2
192 Workability in forging, 188205 Hydrafilm process, 313
workability tests for establishing effects of process Workability limits, 150 (F), 264, 267 Zircaloy-4
variables, 192, 193 Workability maps, 200 (F) tensile elongation as a function of strain-rate
Wedge preheat temperatures, 190 Workability measurement, 59 sensitivity, 77 (F)
ASM International is the society for materials
engineers and scientists, a worldwide network
dedicated to advancing industry, technology, and
applications of metals and materials.
ASM International, Materials Park, Ohio, USA

This publication is copyright ASM International. All rights reserved.
Publication title Product code
Handbook of Workability and Process Design #06701G
To order products from ASM International:

Online Visit www.asminternational.org/bookstore
Telephone 1-800-336-5152 (US) or 1-440-338-5151 (Outside US)

Fax 1-440-338-4634
Customer Service, ASM International
Mail
9639 Kinsman Rd, Materials Park, Ohio 44073-0002, USA
Email CustomerService@asminternational.org
American Technical Publishers Ltd.
27-29 Knowl Piece, Wilbury Way, Hitchin Hertfordshire SG4 0SX,
In Europe United Kingdom
Telephone: 01462 437933 (account holders), 01462 431525 (credit card)
www.ameritech.co.uk
Neutrino Inc.
In Japan Takahashi Bldg., 44-3 Fuda 1-chome, Chofu-Shi, Tokyo 182 Japan
Telephone: 81 (0) 424 84 5550
Terms of Use. This publication is being made available in PDF format as a benefit to members and
customers of ASM International. You may download and print a copy of this publication for your
personal use only. Other use and distribution is prohibited without the express written permission of
ASM International.
No warranties, express or implied, including, without limitation, warranties of merchantability or
fitness for a particular purpose, are given in connection with this publication. Although this
information is believed to be accurate by ASM, ASM cannot guarantee that favorable results will be
obtained from the use of this publication alone. This publication is intended for use by persons having
technical skill, at their sole discretion and risk. Since the conditions of product or material use are
outside of ASM's control, ASM assumes no liability or obligation in connection with any use of this
information. As with any material, evaluation of the material under end-use conditions prior to
specification is essential. Therefore, specific testing under actual conditions is recommended.
Nothing contained in this publication shall be construed as a grant of any right of manufacture, sale,
use, or reproduction, in connection with any method, process, apparatus, product, composition, or
system, whether or not covered by letters patent, copyright, or trademark, and nothing contained in this
publication shall be construed as a defense against any alleged infringement of letters patent,
copyright, or trademark, or as a defense against liability for such infringement.

Workability and Process Design PDF

Uploaded by

Copyright:

Available Formats

Workability and Process Design PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Workability and Process Design PDF

Uploaded by

Copyright:

Available Formats

What is the purpose of the publication?

What is the purpose of the publication?

What topics are covered in the publication?

What topics are covered in the publication?

1119 FM 7/29/03 3:39 PM Page i

2003 ASM International. All Rights Reserved. www.asminternational.org

2003 ASM International. All Rights Reserved. www.asminternational.org

First printing, September 2003

Library of Congress Cataloging-in-Publication Data

Handbook of workability and process design / edited by George E. Dieter, Howard A.

Printed in the United States of America

2003 ASM International. All Rights Reserved. www.asminternational.org

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi Chapter 7 Hot-Tension Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

2003 ASM International. All Rights Reserved. www.asminternational.org

Chapter 13 Workability in Forging. . . . . . . . . . . . . . . . . . . . . . . . . . 188 Forward (Direct) Extrusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299

2003 ASM International. All Rights Reserved. www.asminternational.org

2003 ASM International. All Rights Reserved. www.asminternational.org

George E. Dieter Howard A. Kuhn S. Lee Semiatin

ASM International, Materials Park, Ohio, USA

To order products from ASM International:

Telephone 1-800-336-5152 (US) or 1-440-338-5151 (Outside US)

Workability and Process Design

Fig. 4 Longitudinal sections of two types of ingots

Fig. 6 Section through a heat resistant alloy forging

combined with higher than normal forging tem-

Seams may be continuous or intermittent, as

Fig. 13 Micrograph of a seam in a cross section of a 19

ucts is almost always greatest in the longitudinal

Process Design and Modeling

A considerable array of mechanical tests and

ing operation (Fig. 16), the major process vari-

ing or newly formed microcracks in the work-

Note that the normal stress given by Eq 1 is

s = P/A0 (Eq 8) Fig. 5 Stress distribution at the neck of a tensile spec-

Q Tensile stresses are positive, and compressive

1 s occurs because a die element keeps one dimen-

Dynamic recovery results from the annihila-

presence of low-melting-point compounds that

where g, the normalized flow-softening rate, is

V = A0L0 = A1L1 (Eq 43)

If the plastic deformation is expressed as true

Deformation in metalworking is often ex-

From constancy of volume, Eq 43, and the

Strain rate has three chief effects in metal de-

where dh/dt is the deformation velocity, v, and h

Fig. 22 Flow lines in a forged 4140 steel hook.

ously with deformation. Therefore, the total de-

Workability = f1 (material) f2(process) (Eq 51)

where f1 is a function of the basic ductility of the

(227 C, or 440 F), ductility should be at a max-

Evolution of Microstructure during

Flow Curves in which denotes a constant whose magnitude

Z e exp(Qdrv/RT ) C1 sinh(C2ss)ndrv (Eq 1) There are a number of alternate approaches to

in which k1 and k2 are constants. Similarly, for a

in which k3 and k4 are constants.

Fig. 7 Schematic illustration of work hardening behav-

Bulk Workability Testing

conditions in hot working. The principal disad-

Fig. 5 Heated subassembly with specimen in position

in which d denotes any of the dimensions (l,

es (1/ ) e (Eq 15)

rs/L). The corresponding shear stress is ob-

(d)/c. The parameter is the fraction of