NASA Conference Publication 3034

Part 1

Computational
Methods for
Structural
Mechanics and
Dynamics
Edited by
W. Jefferson Stroud
Jerrold M. Housner
John A. Tanner
and Robert J. Hayduk
NASA Langley Research Center
Hampton, Virginia

Proceedings of a workshop sponsored by

the National Aeronautics and Space
Administration and held at
NASA Langley Research Center
Hampton, Virginia
June 19-21, 1985

National Aeronautics and

Space Administration
Office of Management
Scientific and Technical
Information Division
1989
 PREFACE

This document contains the proceedings of the Workshop on Computational

Methods for Structural Mechanics and Dynamics held at NASA Langley Research
Center, June 19-21, 1985. The workshop was sponsored by NASA Langley
Research Center.

The workshop had two objectives. The first objective was to introduce to the
structural analysis technical community a new Langley research activity in
structural analysis called Computational Structural Mechanics, or CSM. The
second objective was to hear experts discuss important structural analysis
problems and methods for solving those problems.

The workshop was organized into the following four sessions:

1. Local/Global Nonlinear Stress Analysis - Full day - June 19

2. Tire Modeling Half day - June 20

3. Transient Dynamics Half day - June 20

4. Multi-Body Dynamics Full day - June 21

Each session closed with a panel discussion.

Papers in these proceedings are grouped by session and identified in the contents.
The order of the papers is the order of the presentations at the workshop. The
proceedings also include any transcription of questions and answers that followed
each paper and panel discussions that followed each session.

The use of trade names or names of manufacturers in this publication does not
constitute an official endorsement of such products or manufacturers, either
expressed or implied, by the National Aeronautics and Space Administration.

W. Jefferson Stroud
Jerrold M. Housner
John A. Tanner
Robert J. Hayduk
Workshop Co-Chairmen

PRECEDiNG PAGE BLANK NOT NUllrED

iii
 CONTENTS

PREFACE ......................................................................... iii

CONTENTS ....................................................................... v

INTRODUCTION: Computational Structural Mechanics Activity . . . . . . . . . . . . . . . . . . . . . . 1

PART 1

LOCAL/GLOBAL NONLINEAR STRESS ANALYSIS

SOME ISSUES IN NUMERICAL SIMULATION OF NONLINEAR STRUCTURAL RESPONSE ............ 7

H. D. Hibbitt

COMPUTERIZED STRUCTURAL MECHANICS FOR 1990's: ADVANCED AIRCRAFT NEEDS . . . . . . . . . . 25

A. V. Viswanathan and B. F. Backman

NONLINEAR RESPONSE OF A BLADE-STIFFENED GRAPHITE-EPOXY PANEL WITH A

DISCONTINUOUS STIFFENER: WORK IN PROGRESS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Norman F. Knight, Jr., William H. Greene, and W. Jefferson Stroud

COMPUTATIONAL PROCEDURES FOR POSTBUCKLING OF COMPOSITE SHELLS . . . . . . . . . . . . . . . . . . . 67

G. M. Stanley and C. A. Felippa

A REVIEW OF SOME PROBLEMS IN GLOBAL-LOCAL STRESS ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . 89

Richard B. Nelson

SOME COMMENTS ON GLOBAL LOCAL ANALYSES .......................................... 103

Satya N. Atluri

ON COMPUTATIONAL SCHEMES FOR GLOBAL-LOCAL STRESS ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . 123

J. N. Reddy

GLOBAL FUNCTIONS IN GLOBAL-LOCAL FINITE-ELEMENT ANALYSIS OF LOCALIZED

STRESSES IN PRISMATIC STRUCTURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Stanley B. Dong

GLOBAL-LOCAL METHODOLOGIES AND THEIR APPLICATION TO NONLINEAR ANALYSIS . . . . . . . . . . 151

Ahmed K. Noor

APPLICATION OF THE P-VERSION OF THE FINITE-ELEMENT METHOD TO

GLOBAL-LOCAL PROBLEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Barna A. Szabo'

QUESTION AND ANSWER SESSION ..................................................... 179

PANEL DISCUSSION ................................................................. 190

pRECEDfNG PAGE BLAFiK NOT FLMfD

V
 TIRE MODELING

FEATURES AND CHARACTERIZATION NEEDS OF RUBBER COMPOSITE STRUCTURES .............. 211

Farhad Tabaddor

NONLINEAR HIERARCHICAL SUBSTRUCTURAL PARALLELISM AND COMPUTER ARCHITECTURE ...... 249

Joe Padovan

ADAPTIVE METHODS, ROLLING CONTACT, AND NONCLASSICAL FRICTION LAWS ............... 269
J. T. Oden

CONTACT SOLUTION ALGORITHMS ..................................................... 291

John T. Tielking

EXPLOITING SYMMETRIES IN THE MODELING AND ANALYSIS OF TIRES ..................... 317

Ahmed K. Noor, C. M. Andersen and John A. Tanner

PANEL DISCUSSION ................................................................ 331

PART 2*

TRANSIENT DYNAMICS

SUMMARY: TRANSIENT DYNAMICS .................................................... 341

R o b e r t J . Hayduk

IMPROVING TRANSIENT ANALYSIS TECHNOLOGY FOR AIRCRAFT STRUCTURES . . . . . . . . . . . . . . . . . 3 4 3

R. J. Melosh and Mladen Chargin

EXPLICIT, IMPLICIT, AND HYBRID METHODS .......................................... 355

T. Belytschko

UNCONDITIONALLY STABLE CONCURRENT PROCEDURES FOR TRANSIENT

FINITE-ELEMENT ANALYSIS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
Michael Ortiz and Bahram Nour-Omid

TRANSIENT ANALYSIS TECHNIQUES IN PERFORMING IMPACT AND CRASH DYNAMIC STUDIES . . . . 3 8 3

A. B. Pifko and R. Winter

APPLICATION AND IMPLEMENTATION OF TRANSIENT ALGORITHMS IN COMPUTER PROGRAMS ..... 407

David J. Benson

QUESTION AND ANSWER SESSION ..................................................... 419

PANEL DISCUSSION ................................................................ 425

*Presented under separate cover.

vi
 MULTI-BODY DYNAMICS

RECENT DEVELOPMENTS IN DEPLOYMENT ANALYSIS SIMULATION USING A MULTI-BODY

COMPUTER CODE ................................................................. 435
Jerrold M. Housner

DYNAMIC ANALYSIS OF SPACE STRUCTURES INCLUDING ELASTIC, MULTIBODY, AND

CONTROL BEHAVIOR .............................................................. 443
Larry Pinson and Keto Soosaar

Part 1: DYNAMICS OF FLEXIBLE MULTI-BODY MECHANISMS AND MANIPULATORS:

AN OVERVIEW ................................................................... 455
Steven Dubowsky

Part 2 : APPLICATION OF FINITE-ELEMENT METHODS TO DYNAMIC ANALYSIS OF

FLEXIBLE SPATIAL AND CO-PLANAR LINKAGE SYSTEMS ................................ 459
Steven Dubowsky
DYNAMICS OF ARTICULATED STRUCTURES .............................................. 491
Edward J. Haug

MODAL REDUCTION STRATEGIES FOR INTERCONNECTED FLEXIBLE BODIES SIMULATION ........ 517
F. 0. Eke and G. K. Man

COMPUTATIONAL ASPECTS OF MULTI-BODY DYNAMICS .................................... 527

K. C. Park

CONSTRAINT ELIMINATION IN DYNAMICAL SYSTEMS ..................................... 537

R. P. Singh and P. W. Likins

NONLINEAR CHARACTERISTICS OF JOINTS AS ELEMENTS OF MULTI-BODY DYNAMIC

SYSTEMS ....................................................................... 543
Edward F. Crawley

QUESTION AND ANSWER SESSION ..................................................... 571

PANEL DISCUSSION ................................................................ 589

vii
 PART 1

INTRODUCTION

Computational S t r u c t u r a l Mechanics A c t i v i t y

NASA Langley h a s i n i t i a t e d a new r e s e a r c h a c t i v i t y i n s t r u c t u r a l a n a l y s i s c a l l e d

Computational S t r u c t u r a l Mechanics (CSM). The b r o a d o b j e c t i v e o f t h e CSM a c t i v i t y
i s t o d e v e l o p advanced s t r u c t u r a l a n a l y s i s technology t h a t w i l l e x p l o i t modern and
emerging computers, such as computers w i t h v e c t o r and p a r a l l e l p r o c e s s i n g c a p a b i l i -
t i e s . There a r e t h r e e main r e s e a r c h a c t i v i t i e s under way i n CSM: (1) p a r a l l e l pro-
c e s s i n g , ( 2 ) s o f t w a r e t e s t bed, and ( 3 ) s t r u c t u r a l a n a l y s i s methods. The f o l l o w i n g
t h r e e p a r a g r a p h s p r o v i d e a b r i e f d e s c r i p t i o n o f t h e t h r e e main CSM r e s e a r c h
activities.

P a r a l l e l p r o c e s s i n g . - Within t h e n e x t 5 y e a r s , a l l new high-performance computers

w i l l have m u l t i p l e p r o c e s s o r s . CSM r e s e a r c h e r s a r e d e v e l o p i n g s o f t w a r e a p p r o a c h e s
and s t r u c t u r a l a n a l y s i s a l g o r i t h m s t h a t w i l l e x p l o i t t h i s new c a p a b i l i t y . This
r e s e a r c h a c t i v i t y h a s been under way f o r s e v e r a l y e a r s and has used an e x p e r i m e n t a l ,
in-house-developed m u l t i p r o c e s s o r computer known a s t h e f i n i t e - e l e m e n t machine
(FEM). A t t h i s w r i t i n g , FEM has 1 6 p r o c e s s o r s . A new commercial m u l t i p r o c e s s o r
computer w i l l b e d e l i v e r e d t o Langley f o r p a r a l l e l p r o c e s s i n g r e s e a r c h . The
p a r a l l e l p r o c e s s i n g r e s e a r c h a c t i v i t y i n v o l v e s several in-house r e s e a r c h e r s and
universities.

Software t e s t bed.- The t e s t bed i s t o be a modern, modular s y s t e m t h a t h a n d l e s d a t a

e f f i c i e n t l y , t h a t c o n t a i n s a command language which i s powerful and e a s y t o l e a r n
and u s e , and t h a t h a s an a r c h i t e c t u r e which a l l o w s u s e r s t o add s o f t w a r e w i t h minimal
d i f f i c u l t y . The t e s t bed w i l l b e u s e d t o s t u d y t h e i n g r e d i e n t s o f modern s o f t w a r e
and how t h o s e i n g r e d i e n t s s h o u l d f i t t o g e t h e r . More i m p o r t a n t l y , t h e t e s t bed w i l l
b e used t o s t u d y and e v a l u a t e s t r u c t u r a l a n a l y s i s methods on p r a c t i c a l a p p l i c a t i o n s
problems. The t e s t bed r e s e a r c h i s j u s t now g e t t i n g under way. I t i n v o l v e s in-house
r e s e a r c h e r s and c o n t r a c t o r s .

S t r u c t u r a l a n a l y s i s methods.- The s t r u c t u r a l a n a l y s i s methods r e s e a r c h h a s s e v e r a l

goals. One g o a l i s t o d e v e l o p a n a l y s i s methods t h a t are g e n e r a l . T h i s g o a l o f
g e n e r a l i t y l e a d s n a t u r a l l y t o f i n i t e - e l e m e n t methods, b u t t h e r e s e a r c h w i l l a l s o
include o t h e r s t r u c t u r a l a n a l y s i s methods. Another g o a l i s t h a t t h e methods b e
amenable t o e r r o r a n a l y s i s ; t h a t i s , g i v e n a p h y s i c a l problem and a m a t h e m a t i c a l
model of t h a t problem, an a n a l y s t would l i k e t o know t h e p r o b a b l e e r r o r i n p r e d i c t i n g
a g i v e n r e s p o n s e q u a n t i t y . The u l t i m a t e o b j e c t i v e i s t o s p e c i f y t h e e r r o r t o l e r a n c e s
and t o u s e automated l o g i c t o a d j u s t t h e m a t h e m a t i c a l model o r s o l u t i o n s t r a t e g y t o
o b t a i n t h a t a c c u r a c y . A t h i r d g o a l i s t o d e v e l o p s t r u c t u r a l a n a l y s i s methods t h a t
can e x p l o i t p a r a l l e l p r o c e s s i n g computers. Our s t r u c t u r a l a n a l y s i s methods r e s e a r c h
w i l l f o c u s i n i t i a l l y on t h r e e t y p e s o f problems: l o c a l / g l o b a l n o n l i n e a r stress
a n a l y s i s , n o n l i n e a r t r a n s i e n t dynamics, and t i r e modeling. These t h r e e t y p e s of
problems a r e t h e t o p i c s o f t h e workshop and a r e d i s c u s s e d i n d e t a i l s u b s e q u e n t l y .

Workshop D e s c r i p t i o n

T h i s workshop had s e v e r a l o b j e c t i v e s , among which w e r e

1. To i n t r o d u c e CSM t o t h e s t r u c t u r a l a n a l y s i s t e c h n i c a l community,
p a r t i c u l a r l y t h e u n i v e r s i t y community

1
 2. To h e a r e x p e r t s d i s c u s s i m p o r t a n t s t r u c t u r a l a n a l y s i s problems and
methods f o r s o l v i n g t h o s e problems

3. TO h e l p make d e c i s i o n s r e g a r d i n g r e s e a r c h t h r u s t s i n s t r u c t u r a l
a n a l y s i s methods development

Three t y p e s o f problems were a d d r e s s e d a t t h e workshop: iocal/global nonlinear

stress a n a l y s i s , n o n l i n e a r t r a n s i e n t dynamics, and t i r e modeling. The f o l l o w i n g
i s a d e s c r i p t i o n o f t h e s e t h r e e t y p e s o f problems.

L o c a l / g l o b a l n o n l i n e a r stress a n a l y s i s . - A l o c a l / g l o b a l stress a n a l y s i s i s a l o c a l ,
d e t a i l e d stress a n a l y s i s w i t h i n a l a r g e r , l e s s - r e f i n e d a n a l y s i s model. Both t h e
a p p l i c a t i o n and t h e s o l u t i o n p r o c e d u r e a r e t a k e n t o b e a r b i t r a r y . The d e f i n i t i o n
o f l o c a l / g l o b a l i s n o t precise. Because o f t h i s a m b i g u i t y , two f o c u s problems a r e
used t o d e f i n e ( f o r t h e workshop) what i s meant by t h e words l o c a l and g l o b a l . The
f i r s t f o c u s problem i s a s t i f f e n e d f l a t composite p a n e l w i t h a c i r c u l a r h o l e t h a t
c a u s e s a s t i f f e n e r t o be d i s c o n t i n u o u s . The p a n e l i s s u b j e c t e d t o a n i n - p l a n e com-
p r e s s i v e l o a d i n t h e d i r e c t i o n o f t h e s t i f f e n e r s . The second f o c u s problem i s
s i m i l a r , e x c e p t t h a t i t i s curved and u n s t i f f e n e d . I n each case, t h e o v e r a l l
r e s p o n s e o f t h e p a n e l i s t h e g l o b a l problem; t h e r e s p o n s e n e a r t h e h o l e i s t h e
l o c a l problem. I n t h e i r t a l k s , s p e a k e r s might n o t have p r e s e n t e d s o l u t i o n s t o t h e
workshop f o c u s problems, b u t t h e i r comments r e g a r d i n g a n a l y s i s methods r e f e r r e d t o
t h e f o c u s problems.

S t r u c t u r a l problems r e q u i r i n g a l o c a l / g l o b a l stress a n a l y s i s g e n e r a l l y i n v o l v e d i s -
c o n t i n u i t i e s which c a u s e r a p i d changes i n stress. The h i g h stress g r a d i e n t s a r e a
l o c a l phenomenon. A n a l y s i s p r o c e d u r e s t h a t m u s t be used t o p r e d i c t t h e s e stress
g r a d i e n t s are n o t r e q u i r e d away from t h e d i s c o n t i n u i t y . I n r e g i o n s i n which stresses
v a r y s l o w l y , l e s s - r e f i n e d a n a l y s e s are a d e q u a t e . P r a c t i c a l c o n s i d e r a t i o n s demand a
m u l t i l e v e l approach. I t is t h e s e various l e v e l s of a n a l y s i s t h a t cause d i f f i c u l t i e s
i n t h e l o c a l / g l o b a l problem. I n t h e v i c i n i t y o f t h e d i s c o n t i n u i t y , an a n a l y s t may
r e f i n e a f i n i t e - e l e m e n t g r i d , i n t r o d u c e a t h r e e - d i m e n s i o n a l g r i d , u s e more p o w e r f u l
e l e m e n t s , a p p l y a c l a s s i c a l s o l u t i o n , and u s e o t h e r a n a l y s i s t o o l s and a p p r o a c h e s .
U s u a l l y , t h e a n a l y s t w i l l t a k e a d d i t i o n a l steps t o i n s u r e t h a t h i s a n a l y s i s i s
adequate. H e may rework t h e problem w i t h a m o d i f i e d model, o r h e may make compari-
s o n s w i t h t h e r e s u l t s from o t h e r a n a l y s i s p r o c e d u r e s .

To h e l p t h e s t r u c t u r a l a n a l y s t overcome t h e d i f f i c u l t i e s j u s t d e s c r i b e d , a s y s t e -
matic a n a l y s i s p r o c e d u r e i s needed t h a t i s amenable t o e r r o r a n a l y s i s . The a n a l y s t
would l i k e t o prescribe t h e e r r o r t o l e r a n c e s i n p r e d i c t i n g t h e stress a t s p e c i f i e d
l o c a t i o n s and u s e automated l o g i c t o a d j u s t t h e m a t h e m a t i c a l model o r s o l u t i o n
s t r a t e g y t o o b t a i n t h a t accuracy. I n a d d i t i o n , a n a n a l y s t needs t h e c a p a b i l i t y t o
c a l c u l a t e d e r i v a t i v e s o f r e s p o n s e q u a n t i t i e s w i t h r e s p e c t t o parameters t h a t d e f i n e
t h e problem. Such a s e n s i t i v i t y a n a l y s i s i s n e c e s s a r y f o r d e s i g n p u r p o s e s and i s
h e l p f u l i n determining t h e v a l i d i t y of an a n a l y s i s . Finally, t h i s systematic
a n a l y s i s p r o c e d u r e t h a t i n c l u d e s e r r o r a n a l y s i s , a d a p t i v e s o l u t i o n r e f i n e m e n t , and
s e n s i t i v i t y a n a l y s i s s h o u l d b e e a s y t o use. The g o a l i s f o r s o l u t i o n s t o l o c a l /
g l o b a l problems, as w e l l as o t h e r s t r u c t u r a l a n a l y s i s problems, t o become r o u t i n e .

T i r e modeling.- One o f t h e most c h a l l e n g i n g problems i n t h e f i e l d o f Computational

S t r u c t u r a l Mechanics is p r e d i c t i n g t h e r e s p o n s e of a u t o m o t i v e and a i r c r a f t t i r e s
d u r i n g ground h a n d l i n g o p e r a t i o n s . The pneumatic t i r e e x h i b i t s a complex geometry
which can g e n e r a l l y be d e s c r i b e d as a n o n c i r c u l a r , i n c o m p l e t e t o r u s . Under normal
o p e r a t i o n a l c o n d i t i o n s , t h e s e t i r e s are s u b j e c t e d t o l a r g e d e f o r m a t i o n s and

2
 structural rotations. The heavy w e i g h t l o a d i n g r e q u i r e m e n t s imposed on many a i r -
c r a f t and o f f - r o a d v e h i c l e t i r e s require t h e i r carcasses t o be s o t h i c k t h a t a s i g -
I n i f i c a n t p o r t i o n o f t h e r e s u l t i n g s t r a i n energy i s a t t r i b u t e d t o t r a n s v e r s e s h e a r
d e f o r m a t i o n . F u r t h e r m o r e , t h e most i n t e r e s t i n g l o a d i n g c o n d i t i o n s associated with
b r a k i n g and s t e e r i n g o p e r a t i o n s are nonsymmetric.

The t i r e i s a l a m i n a t e d composite s t r u c t u r e c o n s i s t i n g o f r u b b e r and t e x t i l e con-

stituents. The material p r o p e r t i e s of t h e c o n s t i t u e n t s may d i f f e r by a s much as
2 or 3 o r d e r s o f magnitude, t h e r e b y p u s h i n g t h e l i m i t s o f c u r r e n t composite t h e o r y .
The r u b b e r e x h i b i t s n e a r l y i n c o m p r e s s i b l e c h a r a c t e r i s t i c s and i s s u b j e c t t o l a r g e
hysteresis effects. Under dynamic l o a d i n g c o n d i t i o n s , t h e s e h y s t e r e s i s e f f e c t s can
l e a d t o l a r g e t h e r m a l g r a d i e n t s i n t h e t i r e carcass and, hence, s i g n i f i c a n t l y a f f e c t
t h e material p r o p e r t i e s o f t h e c o n s t i t u e n t s . Furthermore, t h e material p r o p e r t i e s
o f t h e t i r e carcass c a n b e h i g h l y a n i s o t r o p i c and nonhomogeneous.

A c e n t r a l f e a t u r e of any t i r e model must b e a s o p h i s t i c a t e d c o n t a c t a l g o r i t h m which

i s capable of h a n d l i n g c o n t a c t f r i c t i o n f o r c e s . The t i r e d e s i g n e r w i l l a l s o r e q u i r e
t h e model t o p r e d i c t stress d i s t r i b u t i o n s i n c r i t i c a l r e g i o n s o f t h e t i r e carcass.
T h i s r e q u i r e m e n t might n e c e s s i t a t e t h e l o c a l i n t e g r a t i o n o f t h r e e - d i m e n s i o n a l f i n i t e
e l e m e n t s i n t o an o t h e r w i s e two-dimensional model. F i n a l l y , t h e model must demon-
s t r a t e i t s c o s t e f f e c t i v e n e s s b e f o r e i t i s g e n e r a l l y a c c e p t e d by t h e United S t a t e s
t i r e industry.

N o n l i n e a r t r a n s i e n t dynamics.- During t h e l a s t 2 decades t h e r e h a s been a growing

i n t e r e s t i n t h e numerical s o l u t i o n o f t i m e v a r y i n g e q u a t i o n s o f s t r u c t u r a l and
mechanical systems. Such i n t e r e s t comes from such d i v e r s e f i e l d s o f a p p l i c a t i o n as
impact mechanisms, robotic m a n i p u l a t i o n , a i r c r a f t s t r u c t u r e s and space s t r u c t u r e s .
These a p p l i c a t i o n s o f t e n i n v o l v e t h e r e s p o n s e o f systems composed o f many s t r u c t u r a l
o r mechanical components.

Dynamic a n a l y s e s o f t h e s e systems i n v o l v e b o t h t h e f o r m u l a t i o n o f t h e g o v e r n i n g
e q u a t i o n s and t h e i r s o l u t i o n . Usually, t h e governing equations are d i s c r e t i z e d
s p a t i a l l y and t h e n i n t e g r a t e d t e m p o r a l l y u s i n g one o f a wide v a r i e t y o f t i m e i n t e -
g r a t i o n procedures. I t i s t h e r e f o r e r e a s o n a b l e t o c o n s i d e r t h e dynamic a n a l y s i s as
c o n s i s t i n g o f two a c t i v i t i e s : a f o r m u l a t i o n and s p a t i a l modeling a c t i v i t y and a
temporal s o l u t i o n a c t i v i t y .

The f o r m u l a t i o n and s p a t i a l modeling a c t i v i t y i n v o l v e s t r e a t m e n t o f c o n s t r a i n t s ,

c o n t r o l s , p o s s i b l e l a r g e d e f o r m a t i o n s of components and l a r g e r e l a t i v e a n g u l a r motion
between components. I t i s n o t u n u s u a l t h a t t h e f o r m u l a t i o n and modeling r e s u l t s i n
l a r g e s e t s of n o n l i n e a r g o v e r n i n g e q u a t i o n s . Consequently, t h e t e m p o r a l s o l u t i o n o f
such s y s t e m s can be c o m p u t a t i o n a l l y i n t e n s i v e , and it becomes t h e a i m of t h e formula-
tion and modeling a c t i v i t y t o improve c o m p u t a t i o n a l e f f i c i e n c y and computer
implementation.

The s o l u t i o n o f t h e g o v e r n i n g e q u a t i o n s i n v o l v e s t h e temporal i n t e g r a t i o n o f equa-

t i o n s of motion s u b j e c t t o i n i t i a l c o n d i t i o n s and c o n s t r a i n t s . The i n t e g r a t i o n i s
u s u a l l y c a r r i e d o u t u s i n g e i t h e r some t y p e o f modal s u p e r p o s i t i o n p r o c e d u r e o r a
d i r e c t t i m e i n t e g r a t i o n p r o c e d u r e . The modal s u p e r p o s i t i o n p r o c e d u r e , though v e r y
p o w e r f u l f o r t h e s o l u t i o n o f l i n e a r e q u a t i o n s , l o s e s much of i t s e f f e c t i v e n e s s i n
d e a l i n g w i t h n o n l i n e a r e q u a t i o n s and f o r t h i s r e a s o n i s n o t i n c l u d e d i n t h i s volume.
G e n e r a l l y , d i r e c t t i m e i n t e g r a t i o n p r o c e d u r e s are used f o r dynamic a n a l y s i s o f
m e c h a n i c a l and s t r u c t u r a l systems governed by n o n l i n e a r e q u a t i o n s .

3
D i r e c t t i m e i n t e g r a t i o n p r o c e d u r e s a r e u s u a l l y r e f e r r e d t o as a l g o r i t h m s s i n c e i t i s
t h e computation o f t h e r e s p o n s e q u a n t i t i e s which are s o u g h t by t h e s e p r o c e d u r e s
r a t h e r t h a n theorems r e l a t i n g t o s o l u t i o n e x i s t e n c e , u n i q u e n e s s , e t c . A v e r y l a r g e
number of t h e s e a l g o r i t h m s e x i s t . T h e i r p o p u l a r i t y i s a t t e s t e d t o by t h e i n c l u s i o n
of such a l g o r i t h m s i n n e a r l y a l l g e n e r a l - p u r p o s e computer programs f o r a n a l y z i n g
s t r u c t u r e s , mechanisms, and s a t e l l i t e s . Many such programs o f f e r t h e u s e r a selec-
t i o n of a l g o r i t h m s . However, f o r t h e s o l u t i o n o f problems i n v o l v i n g t h e temporal
i n t e g r a t i o n o f a l a r g e number o f e q u a t i o n s , t h e u s e o f some o f t h e s e a l g o r i t h m s
l e a d s t o very i n t e n s i v e computer usage due t o a l a c k o f c o m p u t a t i o n a l e f f i c i e n c y .

Not o n l y i s s o l u t i o n e f f i c i e n c y i n need of improvement, b u t so i s s o l u t i o n r e l i a -

bility. The u s e r d e s i r e s a c e r t a i n l e v e l o f c o n f i d e n c e t h a t t h e computed r e s u l t s
a r e r e a s o n a b l y a c c u r a t e . U n l i k e problems o f s t a t i c s , s o l u t i o n s a t i s f a c t i o n o f t h e
g o v e r n i n g e q u a t i o n s does n o t p r o v i d e a g u a r a n t e e t h a t t h e s o l u t i o n i s c o r r e c t , f o r
t h e s o l u t i o n must come from a p r e s c r i b e d s e t o f i n i t i a l c o n d i t i o n s . Conservation of
energy a f f o r d s a l e v e l o f c o n f i d e n c e i n t h e s o l u t i o n , i f such i s a p p l i c a b l e , b u t i n
many problems i t i s n o t a p p l i c a b l e . Thus, b o t h e f f i c i e n c y and r e l i a b i l i t y are i t e m s
which need improvement.

Research i n t o d e v e l o p i n g improved f o r m u l a t i o n s and modeling and t h e i r s o l u t i o n i s

p r e s e n t l y b e i n g performed. However, much remains t o b e done. Furthermore, i t t a k e s
c o n s i d e r a b l e t i m e for such developments t o f i n d t h e i r way i n t o computer programs
which are used i n r o u t i n e p r a c t i c e . Y e t t h e a p p l i c a t i o n needs t o c o n t i n u e t o grow.

Two s e s s i o n s i n t h i s r e s e a r c h a r e a were h e l d . One h a l f - d a y s e s s i o n d e a l t e x c l u s i v e l y

w i t h s o l u t i o n p r o c e d u r e s , w h i l e one f u l l - d a y s e s s i o n d e a l t w i t h development and
t r e a t m e n t o f i t e m s c r i t i c a l t o p r o p e r f o r m u l a t i o n and modeling o f multi-body systems.

T h i r t y papers d e a l i n g w i t h t h e t h r e e classes of problems j u s t d i s c u s s e d w e r e pre-

s e n t e d a t t h e workshop. I n t h e s e papers, s o l u t i o n c o n c e p t s are reviewed, computa-
t i o n a l a l g o r i t h m s a r e o u t l i n e d , an a s s e s s m e n t of v a r i o u s m e t h o d o l o g i e s i s made, and
problem areas a r e i d e n t i f i e d . I n a d d i t i o n , d i r e c t i o n s f o r f u t u r e r e s e a r c h are
recommended .

4
LOCAL/GLOBAL NONLINEAR
STRESS ANALYSIS

5
 Some Issues in Numerical Simulation of Nonlinear Structural Response

H. D. Hibbitt .
Hibbitt, Karlsson and Sorensen
Providence, R.I.

1. Introduction
For some considerable time the author has been involved in the development of commer-
cial finite element software, and this paper is written from that perspective. One of the
remarkable features of the finite element method is its generality, and there is no better
reflection of this than the observation that our company and several of our competitors
each market one product only-a single, “general purpose” finite element code (our par-
ticular code is called Abaqus). These codes provide practical tools that are used in an
astonishingly wide range of engineering applications that include critical aspects of the
safety evaluation of nuclear power plants or of heavily loaded offshore structures in the
hostile environments of the Korth Sea or the Arctic, major design activities associated with
the development of airframes for high strength and minimum weight, thermal analysis of
electronic components, and the design of sports equipment. For various reasons, the code
that my company develops and markets is generally used in the area of more advanced
applications. These applications almost always involve nonlinear effects. There is little
doubt in my mind that the need for such analysis will continue to grow-it is very easy to
identify problems which should be reachable in the next generation or two of computers
and software and which have substantial economic importance.
T h e development, maintenance and support of production software involves many ac-
tivities, but-at least in the more advanced application areas where we try to contribute-
the effectiveness of our product depends critically on the quality of the mechanics and
mechanics related algorithms that we implement. It is generally true that the end users
are not sophisticated with respect to what is now being called “computational mechanics.”
They have other interests and motivations. not the least of which is the need to complete
work successfully and to a schedule. Thus. ”algorithmic robustness” is of primary con-
cern t o us: we should choose those methods that we believe will maximize reliability with
minimal understanding on the part of the user. Computational efficiencj- is important
because there are always limited resources. and hence problems that we would like to do
but which are too time consuming or costly. In reality. we compromise: for example. we

PRECEDMG PAGE BLANK NOT FILMED

7
 knowingly commit what Strang and Fix (1973) call “variational crimes’‘ because we get
awa\ Mith therri often enough for it to be worthwhile. But robustness implies a need for
thorough understanding of the algorithms: we should at least know where the limitations
of an algorithm are likely to be. This is not a simple task: for example, we still do not
have practical ways or assessing local error, even in a linear numerical solution.
It is easy to identify important practical problems that. presently, are computationally
rather difficult. but which will become relatively routine in the not too distance future.
Two that are taking up much of our time just now are the problem of simulating vehicle
(especially automobile) crashes. and the simulation of rather complicated contact situa-
tions, such as the analysis of threaded connectors in drill pipe or casing which is subjected
t o very large axial forces, causing possible thread jump and quite substantial strains in the
pipe. Both problems are modeled today on current generation computers (the Cray-1 and
X-MP) with Abaqus and other codes. They challenge the limits of the algorithms in our
code. and are computationally intensive-typical run times are several hours per case. The
observations made in this paper are based on our experience to date with such problems.
Large scale general purpose codes have a rather long life span: two codes that are
widely used a t this time (Ansys and Kastran) were begun in the mid 1960’s. Thus, in
designing such a code, it is important to try to anticipate what sort of computers will be
in general use for such applications in 10-15 years. Based on past history, this is a difficult
extrapolation-computers are still developing at a rapid rate. We can make some guesses.
For example, a t Cray Research’s Science and Engineering Symposium held in Minneapolis
in 1985 Seymore Cray discussed the specification of the Cray-3. which he expects will be
available in three years. The most relevant parts of that specification from our point of
view are the size of the high-speed, directly addressable, memory (10” words), and the
machine‘s parallelism. Consider a problem of order 50000 unknowns: many important
cases of the types mentioned above are of this size or smaller. The rms half-bandwidth in
such a case would be about 5000, so t h a t the assembled symmetric part of the Jacobian
(stiffness) matrix will occupy 250 x loG words-a quarter of the available memory; the
complete matrix will occupy half the memory. (More and more problems are likely to need
the full Jacobian matrix in the context of Newton methods, because of the use of more
realistic constitutive models, such as damage mechanics models for brittle or composite
materials, which have a non-symmetric Jacobian). Element matrices, state variables for
constitutive calculations, etc. are unlikely t o occupy more than 5 x 10‘ words in such
a case. Thus, it would seem that, for such applications. we need not worry about “I/O”
problems-we can assume the model can run “in-core.” This means that, for our purposes,
the practical computational limitation will be the time taken to do the arithmetic. Cray
Research and others seem t o be moving rapidly toward parallelism to provide arithmetic
speed. This should fit well with that part of our finite element codes that perform element
and constitutive calculations: there we process a large number of such calculation points
(perhaps 4000 elements, 30000 integration points in the 50000 degree-of-freedom case) so
that, provided the code is designed to allow inner loops to spread over the processors, load
balancing should not be much of a problem. It is interesting to note that. in this part
of the code. typical vectors are of order 10-100, so that any vector processor with a long
start-up time (like the Cyber 205) is not very suitable. The solution of the linear equations
7

8
is not such a natural fit into parallel architecture. but the problem is such an important
one that it is likely that very effective modules will become available (such as Floating
Point Systems now provide for their processors). Since we in fact are solving a nonlinear
system, it may be that non-fiewton methods. not requiring the resolution of a Jacobian
matrix, may be preferable in such an environment. I assume that load balancing is the
critical issue in a multi-processor machine. and that the most effective approach to that
issue is a t the inner loop level, rather than the macro (finite element) level.
The remainder of the paper discusses three principal areas. First, the paper discusses
our experiences with the approaches we use to the two initial value problems of primary
interest here-static and dynamic nonlinear response of structures. It then has a brief
section on shell modeling. Then it discusses our current approach t o the constitutive
integration problem, in the context of conventional plasticity models. Finally, the paper
lists some areas where we hope t h a t research work will provide us with new methods, or
with improvements t o our present approaches.

2. Experience with initial value problems

2.1 Structural dynamics

T h e nonlinear dynamic problems that we most often see involve globally nonlinear
response-most of the structure yields and undergoes finite rotations and, possibly. finite
strains. However. locally nonlinear dynamic problems are not uncommon: in fact they
represent a sufficiently important application area that we have found it worthwhile to
provide methods to model them with some efficiency. A good example of such a case is
the flow induced vibration of a steam generator tube in a nuclear plant. The tube rattles
in its support plates due to vibration excited by instabilities in the coolant flow, and this
intermittent contact causes wear and, eventually, may lead to leakage. The tube itself may
be ”sprung”-that is, the initial stress in the tube may be large enough that it forms a
significant contribution to the tube’s stiffness-but the vibration amplitude is not large
enough t h a t the stiffness of the tube changes significantly during the motion. The nonlin-
earities are therefore confined t o the support interaction, where they involve intermittent,
“chattering” contact and friction. Our approach to such problems has been through con-
ventional Guyan reduction. retaining the transverse displacements at the support points
and enough other degrees of freedom in the tube t o model its response accurately in the
frequency range of interest (up t o about 150 Hz in this case). The Guyan method has
two drawbacks-it is based on guesswork, and the reduced model may be much bigger
than necessary. Since the majority of the applications that have come to our attention
involve relatively small models (typically less than 1000 degrees of freedom, with 30-80
degrees of freedom directly involved in local nonlinearities). these limitations are not very
critical: it is not much effort to extract the modes of interest on the full model and then
to experiment with reduced models until a reasonably small one is found that provides
adequate matching.
Nonlinear dynamic problems are integrated explicitly or implicitly. The explicit ap-
proach brings with it several advantages because it is always used with a lumped mass

9
matrix and so no equations need be solved: this leads t o great simplification in the com-
puter code. But conditional stability is a serious limitation, especially in structural models
where the thickness of the shell is usually the determining factor with respect to the time
step. Many of the problems t h a t we see are ”event and response” cases, in which an initial
input of energy is mostly dissipated as inelastic response in the structure. In these cases
the response usually damps out fairly quickly into a pattern of plastic hinges, until some
later disturbance-perhaps a secondary impact, or a sudden effect associated with geome-
try changes-causes this pattern t o undergo a redistribution. Indeed, this phenomenon of
plastic hinge formation is so much a part of such problems that it has given rise t o the suc-
cessful mode form solution approach of P. S. Symonds, J. B. Martin, and others (Symonds,
1967). I t is difficult t o accept t h a t a conditionally stable integration method which can
never expand the time step beyond some fraction of the shortest period exhibited by t h e
model provides the optimal approach for such an application. My own experience with
explicit methods is limited, so t h a t I do not know how one deals with constraints (such as
arise in mixed element formulations) within these methods. But if the methods are only
useful when we do not solve equations, it seems that the generality of their application is
limited.
Implicit methods are chosen for their numerical stability. Stability is usually discussed
in the context of linear systems, and 1 am not sure of what stability proofs exist for
the nonlinear problems we are trying t o model. My own practical experience has been
that numerical stability of t h e operator has never been a limitation: except for accuracy
considerations, the time step is most usually limited by our ability to solve the equations.
I will return to the equation solving problem later, because it is. in our experience, a
key issue. lmplicit methods have the great advantage of generality-the time step can be
chosen for modeling reasons. And since we have t o solve equations anyway, we are a t liberty
to introduce whatever additional equations we care to, including, for example, Lagrange
multipliers t o impose constraints. In Abaqus we have for some time been using the Hilber-
Hughes-Taylor (1978) operator (an extension of the trapezoidal rule that provides the
ability to introduce some numerical damping) with a simple “automatic” time step selection
scheme, and this approach has been of great practical benefit. Again, consider the car
crashworthiness problem. A front end collision test case for a car design usually involves
about 10-15 milliseconds of response. With the shell elements in Abaqus and a mesh
t h a t is adequate t o model the response usefully, the stability limit of the central difference
operator is in the microsecond range. Using the implicit method, the analysis is usually
completed in 150-300 time increments, which range from a few microseconds just after a
major impact u p t,o a significant fraction of a millisecond during t h e fixed plastic hinge
rkgime. T h e utility of the method then depends on whether we can solve the nonlinear
equations 150-300 times in an acceptable amount of computer time.

2.2 Statics
Smoot.hly nonlinear static problems are not uncommon, but it seems that many impor-
tant applications involve abrupt changes in the response. which is often unstable as well.
The car crash problem again serves to provide an example. During a front end crash, most
conventional designs of front rails in cars are responding unstably during about 80% of
their usable deformation, and t h e switch into this collapsing response occurs quite sharply,
as a buckling of the structure (which is a shell), after it has undergone considerable plastic
deformation. Most shell buckling cases involve relatively thin shells, so t h a t the equilib-
rium equations are not well conditioned. Elastic-plastic buckling often involves sudden
“localization” of the plastic deformation into narrow regions, while major parts of the
structure, which were yielding in the pre-collapse phase, unload elastically. Such problems
make demands on the solution strategy-it must be able to detect, and to switch into,
the alternate equilibrium path, and the instability of the collapse phase of the response
must be handled. For the latter purpose we have found that our version of a “Riks” algo-
rithm (Riks, 1979) has been most valuable. We use it with an L‘automati”’ incrementation
scheme, and find t h a t it can often march right through into the collapse as far as we want
to go, without too much stuttering. The most common complaint that we receive is that
the solution sometimes turns back on itself. When this occurs it is at critical points where,
presumably, the equilibrium path has very high curvature. For the present we ascribe this
t o a weakness in our implementation only: the method finds an equilibrium path, but it
is not the path of interest. We have nothing in our code t h a t can detect t h e possibility of
alternative equilibrium paths: for example, we know we cannot obtain sensible results for
a round bar in compression (Euler buckling) or tension (necking), without knowing that
the switch may occur and seeding the problem definition with a suitable imperfection. A
robust algorithm would not need this.
Rate form constitutive models (such as conventional plasticity or visco-plasticity theo-
ries) still must be integrated, even though the overall problem is quasi-static. There seems
t o be relatively little motivation for choosing explicit methods in general for this aspect of
the problem, although there are some particular cases where an explicit approach makes
sense: for example, many high temperature creep problems associated with metal struc-
tures can be treated efficiently with explicit integration of the creep model, because in these
cases the response times of interest are usually not very long compared t o characteristic
relaxation times for the material subjected t o the stresses that arise in the structure-
otherwise the design would be unacceptable anyway. In Abaqus, except for this particular
case, we use implicit integration for rate plasticity models, as discussed below.

3. Equation solving
Almost all of the procedures we use in Abaqus are based on a “full” Newton method.
We have tried a few alternatives (modified Newton and quasi-Newton methods), but so
many problems of interest are not very well conditioned and exhibit such knotty response
t h a t we have not looked a t many of our standard test cases before we have rejected the
alternatives t h a t we have tried, and returned to the quadratic convergence of the full
Newton method. However. our work is hardly rigorous or complete.
A significant part of the usage of our code involves systems for which the Jacobian
matrix of the Newton method is not symmetric. Examples are the non-associated flow
plasticity models t h a t are often used in soil mechanics applications, and loading cases with
“follower forces”, like Morison drag on offshore pipes and risers. In these cases we usually
form and solve the non-symmetric system. There are some problems where we offer the

11
 possibility of approximating the Jacobian with its symmetric part because our experience
has suggested that many typical cases of t h a t class are analysed with less computational
expense in that way. A n example is mentioned later in this paper in connection with
integration of plasticity models. But we always provide the possibility of invoking the
non-symmetric capability if needed, typically by restarting the simulation at soTle point,
because it often happens that the symmetric approximation becomes less effective as the
solution develops.
Newton’s method introduces difficulties of its own. The most expensive is the need to
form and solve a system of linear equations at each iteration, and the most awkward is the
need to define the Jacobian matrix. The algebraic manipulations involved in defining this
matrix can be formidable-and there are obviously cases when it cannot be defined, except
perhaps numerically. For example, a recent paper on numerical methods in plasticity (Simo
and Ortiz, 1984) contains the remark:
“In general, however, the task of evaluating the consistent tangent
moduli in closed form may prove exceedingly laborious. It would appear,
therefore, that a general purpose implementation of the physically more
compelling algorithm.. . may require the use of quasi-Newton or secant-
Newton methods . . .”
This difficulty should not be underestimated.

The straightforward approach we have been using for conventional plasticity models is
discussed in Section 5 . One further comment concerning t h a t method is appropriate
here: even in cases when the rate plasticity model exhibits the usual symmetry property
obtained from assuming associated flow and a smooth yield surface, the exact Jacobian of
the integrated model-using the integration operator that we have chosen-is not always
symmetric.
The full Newton method is expensive computationally. We can provide some numbers
to quantify this on computers that are available today. Abaqus is a general purpose code
and is used on many different computer systems, presently ranging from the Apollo 300
t o the Cray X-MP. The code is not particularly “tuned” t o any system, and therefore
should be representative of typical straightforward finite element codes (we know from
benchmarks t h a t this is generally true). We use some standard benchmark problems to
estimate performance of o u r code on various computers: experience has shown that these
benchmarks are reliable. One of these is a shell model. with 1000 eight-node elements.
It has about 17000 degrees of freedom and an rms wavefront of 400 degrees of freedom.
Such a mesh would be adequate for the typical front-end collision analysis that I used
as a n example above. On a CrayX-MP this problem takes 48 cp seconds per iteration:
60% of t h a t time is in the linear equation solver. and most of the rest is associated with
element and constitutive calculations. A typical dynamic analysis of a front end collision
needs 150-300 time increments w i t h the implicit operator that we use, and between three
and four iterations per increment with the “almost full Newton’‘ implementation of the
shell in Abaqus (the initial stress terms associated with bending are not defined exactly
in the Jacobian--there is no basic problem, we have simply not completed some lengthy
manipulations. I do not think the terms we omit are very significant). Thus, such a crash

12
simulation can be expected t o pass through the basic loop about 500-1200 times if our
algorithms for time stepping, impact, etc. all are working well. This implies about 6; to
16 hours on the Cray for the job.

4. Shells
A substantial part of the modeling for which Abaqus is used involves shells, much of it
in cases where geometric a n d material nonlinearities dominate the response. We have
tried t o provide useful capabilities for shell modeling, but we know t h a t there are serious
deficiencies in what we offer. Irons wrote often about shells: his definitive account of
SemiLoof (Irons, 1976) and his textbook (Irons and Ahmad, 1980) both contain succinct
statements of the difficulties. My impression is t h a t things are improving rapidly. I hope
so: I think we are all aware of the need.
In Abaqus we always treat shells as shells-we do not have any numerically degener-
ated solid elements acting as shells. T h e elements are formulated in terms of forces and
moments a t integration points. T h e behavior of the cross section is defined in closed form,
or by numerical integration through the thickness. In most cases we use shear deformable
elements, b u t the transverse shears are not considered in the constitutive calculations-we
assume the shell is thin enough t h a t transverse shear is not very important: it is simply
a numerical technique t h a t allows us t o manage with low order interpolation. Low order
elements are often desirable in practical cases.
Abaqus includes three types of shell model: axi-symmetric shells with axi-symmetric
deformations. general shells. and pipes and elbows with deforming sections (ovalization and
warping). These last elements are a rather special case which turns out to be important
in certain piping applications t h a t arise in the nuclear power industry and in the offshore
oil industry. We often see problems where a mixture of shell and solid modeling is needed
(li-joints in offshore platforms; Tees in pressure vessels). We have a standard technique for
joining the shells to the solids, based on kinematic assumptions introduced as constraints.
T h e approach appears to be satisfactory for the small strain cases where we have seen
results.
T h e purely axi-symmetric case is rather simple because the problem is one dimen-
sional. For this case we use a linear interpolation element with one integration point and a
quadratic interpolation element with two. Our impression is that these elements are quite
effective. O u r implementation of these elements is based on a simplified finite membrane
strain theory. T h e theory is similar in concept to Rodal’s thesis (Rodal and Witmer, 1979),
except t h a t we use somewhat different strain measures because our applications involve
material models for which logarithmic strain seems t o be appropriate, a n d , in a case like
this where the principal directions do not rotate. it is easy to work directly in terms of
this strain. T h e main simplifying assumptions in the theory are that the thinning of the
shell is uniform through the thickness and is defined by an incompressibility assumption
on the reference surface of the shell. and that the thinning occurs smoothly, so that we
can neglect gradients of thickness change with respect to position.
There are many applications for which a finite membrane strain model should be
useful-obvious examples arise in sheet metal forming processes. T h e axi-symmetric case

13
 is particularly simple and our version of a finite membrane strain theory is quite basic.
Nevertheless. it takes two pages of the Abaqus theory manual just to write out the definition
of the initial stress matrix. and the manipulations involved in reaching t h a t definition are
lengthy. The extension of the same formulation t o general shell deformations will involve
substantial algebraic manipulation.
The general shell elements that we use are the Batoz triangle, and the four and eight
node “Ahmad quadrilaterals” (shear flexible elements using reduced integration). Our
implementation of the Batoz triangle uses Batoz’ interpolation functions in a co-rotational
frame for bending, together with constant membrane strain. The element is in Abaqus
because it is essential for us to offer a three-node triangular element for shell problems,
and we have been advised that this element is among the better elements of this category.
The element appears t o behave well in bending-it converges rapidly and is relatively
insensitive to distortion. The shortcomings of our implementation of the element are the
constant membrane strain assumption, the faceted geometric approximation, and the need
t o use three integration points, this last because it makes the element more expensive than
it should be, given the constant membrane strain assumption.
The Ahmad quadrilaterals are very attractive for cases involving material nonlinearity
because the use of reduced integration minimizes the constitutive evaluations needed t o
form the element. The basis functions are also very simple. The four node element is
of limited value without hourglass control: we use the hourglass controls proposed by
Belytschko (Belytschko and Tsay. 1983) for this element. This improves the situation,
but it is still not entirely satisfactory-we still find it necessary to change the hourglass
stiffnesses from time to time. without being too sure of the reasons for the values we choose.
Abaqus includes a print option which summarizes the energy content of the solution. It
is not uncommon to see relatively significant “artificial strain energy’‘ associated with the
hourglass control: in typical high energy dynamic events this artificial strain energy may
even exceed the residual “physical” energy (strain energy plus kinetic energy) at the end
of the event. Yet the overall predictions of the significant aspects of the response are
generally usable.
As Irons pointed out in his last paper (Irons and Loikkanen, 1983), the eight node
Ahmad element does not pass the patch test except as a rectangle. We offer this element
for historical reasons-it seemed t o be a good choice at the time. In fact it does respectably
well on shells that are not too thin, even if it is somewhat distorted. But there is another
practical objection to the eight node interpolation scheme: contact problems are awkward
with such functions. Contact over the entire element does not present serious difficulties,
but. with the contact algorithms that we use (which have Lagrange multipliers to represent
the contact pressure), we do not know how to deal with partial contact over an element-
that is. with the possibility that contact or separation may occur over part of the surface
represented b> an element. We plan to add the nine-node version of the same element t o
overcome this last objection, even though Irons has told us that this form of the element
onl> passes the patch test when it is geometrically bilinear (which would soon not be the
case in a large displacement analysis. even if it were true to start with).
Both of the A4hmadelements, as we have implemented them, perform poorly if they
are distorted and the shell is thin. This is a serious objection, but it would appear that

14
the element of K . C. Park (Park and Stanley, 1 ~ 8 4 )alleviates this problem. I think it
is important. for practical cases, that we retain the simplicity of the low order elements
as well as the cost effectiveness of reduced integration, especially when relatively severe
nonlinearities (requiring, for example, modeling of finite strain effects) are present.

5. Integration of Conventional Plasticity Models

Conventional plasticity models present an interesting integration problem: the mate-
rial behavior changes drastically a t yield; the plastic strain is defined only as a rate, and
the stress measures usually used are defined on the current configuration, so that the rota-
tion of the material must be accounted for in some way. The problem has been receiving
attention recently (for example Ortiz and Popov, 1984), and new approaches have been
suggested. I would like to describe the approach that we have been using in Abaqus: over-
all: we are reasonably satisfied with it, but it raises some questions t h a t we have not been
able t o answer, and which may also be relevant to the more recently proposed approaches.
Abaqus is primarily an implicit code, so t h a t it is desirable t o use an algorithm that
is unconditionally stable. This is not strictly essential. In typical cases involving metals
under ordinary conditions, our experience has been that we are unlikely to succeed in
using increments in which the rotations exceed 10-15 degrees, or the strains exceed a
few percent over important parts of the mesh (these limitations are associated with our
Newton approach to solving the equations). Thus, any method that is stable for this size of
increment should be satisfactory. However, I would be surprised to see conditionally stable
algorithms that can handle such increments: typical yield strains in metals are about
so t h a t the increment size is about ten times the size of the yield surface in strain space,
and I assume that the stability limit of a conditionally stable operator would not be larger
than the yield strain. Because we have found the full Newton method to be effective, we
want an algorithm that is sufficiently simple so t h a t the Jacobian matrix, (dT/dE)t+nt,
where T is the stress , E is the strain, and t i At is the time at the end of the increment
(where we write the equilibrium equations in an implicit code), can be obtained exactly
for common plasticity models. It is desirable t h a t this matrix have an additional property:
several plasticity models of practical importance are derived from a locally smooth potential
which is also the yield surface, so that the rate equations of the model give a symmetric
form dT/aE. We would like the integration operator to preserve t h i s symmetry-this is of
practical significance with respect to computational cost. The operator we use in Abaqus
does not do so, except in a restricted class of plasticity models (which does not include
some of those t h a t are available the code and which have a symmetric rate form).
We first integrate the kinematic aspect of the model by using the algorithm of Hughes
and Winget (1980) to define an increment of rotation. which we apply to all vector and
tensor valued functions at the material point; we also follow t h e Hughes-Winget suggestion
and define the strain increment by the central difference approximation. Various authors
have commented on the inadequacies of this class of method in the context of theories that
involve tensor valued hardening variables, such as classical kinematic hardening theory. I
do not see much utility in simple kinematic hardening with respect t o finite strain appli-
cations. and the approach does not suffer the same obvious deficiency when it is used with

15
 the isotropic yield models t h a t we generally use for metals and soils at finite strains. The
algorithm leads t o a symmetric initial stress matrix. which is tedious to derive.
We are left with the integration of the change of state associated with deforma-
tion. The following discussion considers only isothermal, rate independent behavior with
isotropic hardening. In Abaqus we use the same approach for rate dependent models (ex-
cept where we integrate explicitly. as mentioned in Section 2.2 above), for non-isothermal
cases (including fully coupled temperature-displacement calculations) and for more com-
plicated hardening models. such as kinematic hardening.
T h e isotropic hardening plasticity models in Abaqus all have the following structure.
We assume a strain rate decomposition,

d E = dE"' + dEp'
where d E is a differential change in total mechanical strain, dE"' is a differential change
in the elastic strain. and dE'" is a differential change in the inelastic (plastic) strain. We
have introduced the Hughes- Winget approach t o account, in an approximate way, for the
rigid-body rotation of the material during the increment, and t o define a finite increment of
total strain during the increment. This allows us t o write this decomposition in integrated
form as

E ~ E" - E?'' (1)

where E. E"' a n d ET'' are defined as summations of the corresponding rotated values a t
the start of each increment and the incremental values associated with that increment.
During the constitutive calculations E is known, except in the case of plane stress.
We assume an elastic strain energy potential so that the stress. T. is defined as

where II' = W(E"'.O) is the strain energy potential, and 8 is the temperature.
Some of the plasticity models t h a t we use in Abaqus assume linear elasticity. while
others (soils models) use a nonlinear elasticity in which the logarithm of the pressure
stress is proportional to the volumetric strain. None of these elasticity models has internal
constraints such as incompressibility, so t h a t (2) defines the stress completely. It can be
differentiated to give

which we write as

dT ' D " : dE" (3)

A simple isotropic hardening model has the yield function

f = O (4)

I 16
where f = f ( T . O , H " ) , with H a being a set of hardening parameters; f is defined such
t h a t . whenever f < 0, the response is purely elastic. The models we use in Abaqus all
have a smooth yield function. so that a j aT. a 2 j aT 3T. and d f / d H " are well defined
I

everywhere on f.
T h e flow rule is written

dE"' = dX- (5)

dT
where dX is a positive scalar and g ( T ,8 , H " ) is the flow potential. Since the material
models being discussed are rate independent, dX is determined only by the kinematic
solution a t the material point being considered.
T h e hardening parameters evolve with plastic strain:

dH" = h"(dEP',T ,8 , H P )
where h" defines t h e hardening: h" must be homogeneous of degree one in dEP" for the
model t>obe rate independent. Therefore

T h e plasticity model is now defined. except for choosing particular forms for the elastic
strain energy potential, K', the yield function, f, the flow potential. g , and the hardening
rules, h".

T h e only rate equations in the formal definition of the material model are the evo-
lutionary rule for the hardening and the flow rule. The simplest operator which may a t
least fulfill the requirement of unconditional stability mentioned a t the beginning of this
section is the backward Euler method, which can be introduced into (5) t o give

while (6) is writt,en

ag
AH" = AX h"( -, T ,8 , H ' )
aT
In these equations and in all of t,he following. all quantities are evaluated at the end
of the time increment.
Remark: Ortiz and Popov (1984) propose the midpoint rule as a more accurate oper-
ator. based on an error analysis with small strain increments. Their analysis
appears t o assume t h a t the material is yielding during the entire increment.
This is often not the case-in most calculations there are always material
points where yielding begins part way through the increment. At such points
the mid-point rule requires solution for the initial intersection with the yield
surface during the increment. and application of the rule only within the

17
 yielding part of the increment. This creates some formidable algebra. espe-
cially if one wishes to derive the Jacobian. O u r analysis of this problem leads
us t o the conclusion that, even for a n associated flow material, the Jacobian
will, in this case. not be symmetric, and that its non-symmetric part may
not be very small. We choose the backward Euler method partly because it
does not introduce this difficulty, and so retains sufficient tractability that
we can complete the algebra rather easily. It is interesting t o note that Ortiz
and Popov confirm the conclusion of Schreyer et al. (1979): that, for strain
increments that are not small compared to the size of the yield surface in
strain space, the backward Euler method exhibits the highest accuracy of
any of the simple methods that they tested.
From a computational viewpoint the problem is now algebraic: we must solve the
nonlinear equations ( l ) , (2), ( 4 ) , ( 7 ) and ( 8 ) , and thus define the state of the material at
the end of the increment. The "material stiffness matrix",

must also be defined for the overall Newton scheme that we use for the equilibrium equa-
tions. Simple manipulations lead t o the definition

ID] = [II]+ AA[H]: [L]]-'[H]

where 11; is the fourth order identity tensor, and

[HI = IDJe': IZ]

1 -
iZ;= [I]- - N M :
d

18
and

For a non-associated flow material, af/dT and N are not related, so that [D]cannot
be symmetric. For an associated flow material, 13fjdT and N (and hence N) will be
co-linear. Therefore ID] will be symmetric if

which is the case, for example, for the simple Prandtl-Reuss model t h a t is commonly used
for metals, but is not true for the modified Cam-clay model t h a t is sometimes used for
clays. However, in associated flow cases when it is not true, the non-symmetric contri-
butions to ID] appear only in the terms multiplied by (AA)2, so that they are of the
order of (the plastic strain increment)2 compared to unity. This suggests t h a t , for prac-
tical purposes, the lack of symmetry should not degrade the convergence of the Newton
iterations for equilibrium if we approximate ID] with its symmetric part, and our experi-
ence t o date confirms this-at least, the performance of the algorithm with the symmetric
approximation has been satisfactory, by our standards.

There remains the problem of solving the algebraic equations for the state at the end
of the increment. In the general case, this is not a simple matter: whatever method is used
should not be expensive on the computer. but it must work for the tightly curved yield
functions that appear in some models of practical interest, even when the strain increment
is many times the typical size of the yield surface in strain space. The problem is made
more awkward in cases such as Cam-clay, where exponential terms appear in the elasticity
and in the hardening. Our approach has been as follows. A Newton scheme should work,
provided we choose suitable variables as a basis, and provided we start with a good guess.
T h e plastic strain increments, AEp', should be a suitable set of variables. Then Kewton's
method for ( I ) , ( Z ) , (4), (7) and (8) is the linear equations
- 1
[!I]t AA[Z] : [L] : C p ' = [ Z ] : (AAN - AE"')
: [D]"'] tN- f
d
which are solved for CP' , the improvement to the solution for GEP':

The elastic strain is then obtained from ( l ) ,the stress from ( 2 ) . L A from the projection
of ( 7 ) onto N:

and AH" from ( 8 ) .

19
 This loop is repeated until the components of LE"' converge. A tight convergence
criterion is necessar), so that the solution is accurately defined: this is essential for the
overall Newton scheme for the equilibrium equations to converge quadratically.
The Newton method requires solution of the linear system a t each iteration. Although
the system of equations is not large (at most six), these computations are done at each
integration point and at each iteration of the overall equilibrium solution, so that it is
desirable to solve the problem more efficiently than by direct application of the Newton
method. In addition, when the strain increment is large, a reasonable starting guess is
necessary for the method to converge. For these reasons we have been using a projection
of the problem onto a smaller number of variables t o start the solution. If this subspace
is chosen appropriately, we should be able to develop a useful estimate of the solution a t
low cost. The general idea is as follows.
Let K O , (Y .
= I , . . . n, be a set of tensors that are orthonormal:

and choose n t o be less than the number of stress components in the actual problem.
The K u are chosen for a particular guess and are fixed during the solution for that
guess.
Assume that the plastic strain increment is

= ne='Ka
,%-E"' (9)
The elasticity, (2), and the strain rate decomposition. ( 1 ) . then define the stress,
which we require to satisfy yield, ( 4 ) . The integrated flow equation, (7), is imposed in the
sub-space:

where

Since the K a are orthonormal, (9) and (10) then define

Aep$ = AxNa
where

The Newton scheme described above for the full stress space projects directly:

20
where

and

These linear equations provide the c a , the improvements t o the Aeg':

At each iteration the solution is updated as described for the full stress-space solution,
except t h a t Ax is calculated by projecting (11) onto N a :

T h e utility of this technique depends on the choice of the Ka, and on n. Clearly it is
unlikely t h a t it would be worthwhile t o use n > 2. Most of the plasticity models in Abaqus
are isotropic. in the sense t,hat the yield function, f, and the flow potential, g , are defined
in terms of the stress invariants. An obvious choice for the K O in these cases is n = 2,
with

(here I is the identity matrix),

and

where S" is the deviatoric part of To,the stress that would occur at the end of the
increment if there were no plasticity occurring in the increment. and

If the only st,ress dependencies in f and g are the effective pressure stress,
1
p = --I : T
3
and the deviatoric stress magnitude, q; and the elasticity is isotropic, (and the hardening
is isotropic, as has been assumed here), the subspace solution is the exact answer to the

21
 problem, except for the plane stress case. For some simple yield surfaces and hardening
definitions the subspace problem can be solved in cloyed form, without the need for iter-
ation. For the simplest case of a Mises material. where q is the only stress term in f and
g . the one-dimensional sub-space defined by K 2 above provides the exact solution. For
perfect plasticity the method is then precisely Wilkins‘ “radial return” algorithm.
Our experience with more general yield functions and flow potentials, where the third
stress invariant is also used, is t h a t the two-dimensional sub-space provides a satisfactory
guess, from which the solution can be completed usually with no more than two iterations
of the full stress-space problem.

6. Closure
The paper has presented a brief review of our experience in the recent past in the
general area of nonlinear structural analysis of metal shells. It is not difficult t o identify
the most severe limitations in the methods we have used so far, t o write down a “wish
list” of areas where we would like to be using better methods than those we have used to
date. The list is as follows.
0 We do not like the rapid deterioration in the accuracy of the isoparametric shell
elements that we currently use when they are not rectangles and the shell is thin.
0 Reduced integration elements are attractive because they minimize constitutive calcu-
lations, which are often a significant part of t h e computational cost when the material
model is not simple. Numerical artifices, such as hourglass control, are not objection-
able, when they are effective and are well understood. Our own understanding so far
is lacking.
0 The characteristic hinging of thin metal shells under compressive loads raises difficul-
ties and opportunities. The difficulties are associated with concerns about capturing
this behavior with smooth discretizations. The opportunities are indicated by the re-
markable success of the rigid-plastic models of Wierzbicki. Perhaps it is true that, in
practical cases, load bearing members are too thick for this to be important, although
movies of front-end collision experiments on cars shown very pronounced hinging.
0 A straightforward finite membrane strain shell formulation would have wide applica-
tion in several problem areas that we often encounter. The axi-symmetric formulation
we have been using is based substantially on the assumptions introduced by Rodal in
his thesis, although we have taken a rather different approach in detail. We do not
have a good appreciation of the limitations of the formulation.
0 We expect to continue to work with implicit methods in many problem areas, including
a large part of the analysis work associated with structural design. If this is a correct
assessment. it would be highly desirable t o have equation solution methods t h a t are
more effective and less difficult t o use than Kewton’s method. This is, from our point
of view. one of the most severe limitat,ions t h a t we face in practical applications.
0 So much shell behavior involves branching on the solution path t h a t it would be very
satisfying to have methods t h a t can handle this unassisted.

22
 0 Finally, we have not mentioned rezoning, or the treatment of localization, but these
are. becoming important issues in practical cases.

7. References

T. Belytschko and C.-S. Tsay, “A Stabilization Procedure for the Quadrilateral Plate
Element with One-Point Quadrature”, Int. J . Num. Meth. Eng., Vol. 19, pp. 405-419,
1983.
H.M. Hilber, T.J.R. Hughes and R.L. Taylor. ‘bCollocation,Dissipation and Overshoot
for Time Integration Schemes in Structural Dynamics”, Earthquake Eng. and Struc. Dyn.,
. pp. 99-117, 1978.
V O ~16,
T. J . R. Hughes and J. Winget, “Finite Rotation Effects in Numerical Integration of
Rate Constitutive Equations Arising in Large Deformation Analysis”, Int. J . Num. Meth.
Eng., Vol. 15. pp. 1862-1867, 1980.
B. M. Irons, “The Semi-Loof Shell Elements”, Chapter 11 of Finite Elements for Thin
Shells & Curved Members, edited by D. G. Ashwell and R. H. Gallagher, John Wiley &
Sons, London, 1976.
B. M. Irons and S. Ahmad. Techniques of Finite Elements, Ellis Horwood Ltd., Chich-
ester, England, 1980.
B. M. Irons and M. Loikkanen, “An Engineers’ Defense of the Patch Test”, Int. J.
Kum. Meth. Eng., Vol. 19, pp. 1391-1401, 1983.
M. Ortiz and E. P. Popov, “Accuracy and Stability of Integration Algorithms for
Elastoplastic Constitutive Relations”, Brown University. Division of Engineering Report,
1984.
K . C. Park and G . M. Stanley, “A Curved C* Shell Element Based On Assumed
Natural-Coordinate Strains”, Applied Mechanics Lab.. Lockheed Palo Alto Research Lab.,
Palo Alto, Cal., 1984.
E. Riks, “An Incremental Approach to the Solution of Snapping and Buckling Prob-
lems”, Int. J . Solids and Struc.. Vol. 15, pp. 329-551. 1979.
J . J . -4. Rodal and E. A. Witmer, “Finite-Strain Large-Deflection Elastic-Viscoplastic
Finite-Element Transient Response Analysis of S t r u c t u r e s “ . c o n t r a c t o r r e p o r t 159874,
NASA Lewis, 1979.
J . C. Simo and M. Oritz, “A Unified Approach to Finite Deformation Elastoplastic
Analysis Based o n the Use of Hyperelastic Constitutive Equations”, Brown University
Division of Engineering Report, 1984.
G . Strang, and G . J . Fix, An Analysis of the Finite Element hlethod, Prentice-Hall,
New Jersey, 1973.
H. L. Schreyer. R. F. Kulak and J. M. Kramer, “.4ccurate Numerical Solutions for
Elastic-Plastic Models”, Trans. ASME, J. Pressure Vessel Technology. Vol. 101, 1979.
P. S. Symonds, “Survey of Methods of Analysis for Plastic Deformation of Structures
under Dynamic Loading“, Brown University Division of Engineering Report. 1967.

23
 N89-24640
COMPUTER IZED STRU TU AL ME i A N I C S FOR 1990's: ADV N ED AIRCRAFT NEEDS

A. V. Viswanathan and B. F. Backman

Boei ng M i 1 it a r y A i r p l ane Company
S e a t t l e , Washington

Abstract

This paper describes t h e needs f o r computerized s t r u c t u r a l mechanics (CSM) as

seen from t h e s t a n d p o i n t o f t h e a i r c r a f t i n d u s t r y . I t p r o j e c t s these needs
into t h e 1990's w i t h s p e c i a l focus on t h e new advanced m a t e r i a l s . It i d e n t i -
fies t h e major areas:

0 preliminary design/analysis
research
0 d e t a i l design/analysis

and e l a b o r a t e s on t h e r o l e o f l o c a l / g l o b a l analyses i n these d i f f e r e n t areas.

The lessons l e a r n e d i n t h e past a r e used as a b a s i s f o r t h e design o f a CSM

framework t h a t c o u l d modi f y and consol id a t e e x i s t i ng techno1 ogy and in c l ude
f u t u r e developments i n a r a t i o n a l and u s e f u l way.

A philosophy i s s t a t e d , and a s e t o f analyses needs d r i v e n by t h e emerging

advanced composites i s enumerated. The r o l e s o f NASA, t h e u n i v e r s i t i e s , and
t h e i n d u s t r y are i d e n t i f i e d .

F i n a l l y , a s e t o f r a t i o n a l research t a r g e t s i s recommended based on both t h e

new types o f computers and t h e increased complexity t h e " i n d u s t r y " faces.

Computerized s t r u c t u r a l mechanics should be more than new methods i n

s t r u c t u r a l mechanics and numerical analyses. It should be a s e t o f engineer-
i n g a p p l i c a t i o n s s o f t w a r e products t h a t combines i n n o v a t i o n s i n s t r u c t u r a l
mechanics, numerical a n a l y s i s , data processing, search and d i s p l a y f e a t u r e s ,
and recent hardware advances and i s organized i n a framework t h a t d i r e c t l y
supports t h e desi gn process.

BRECEDiNG PAGE B L M K NOT FlLMED

25
 There a r e two aspects t o t h e development o f engi n e e r i ng appl ic a t i o n s s o f t w a r e :
(1) in n o v a t i ons and ( 2 ) improvements i n t h e p r o d u c t i v i t y o f engineers .
While t h i s paper c o n c e n t r a t e s on t h e f i r s t aspect, i t i s s t i l l i m p o r t a n t t o
c o n s i d e r t h e p r o d u c t i v i t y aspect i n t h e development o f new methods because t h e
s o f t w a r e u l t i m a t e l y w i l l be p a r t o f t h e s e t o f t o o l s used i n t h e design o f
a i r c r a f t structures.

Structural Analysis Needs

for the Aircraft Industry

0 New capabilities
Concepts
Materials
Details
0 Improved productivity
Engineering workstations
Tutorial software
CAD/CAM interfaces

I 26
New n o n - d u c t i l e m a t e r i a l s have entered t h e scene and w i l l be p a r t o f t h e
p i c t u r e f o r a l o n g t i m e t o come. These m a t e r i a l s r e q u i r e more d e t a i l e d analyses
n o t o n l y because of m i c r o - c o n s i d e r a t i o n s and i n t e r l a m i n a r e f f e c t s b u t a l s o
because o f t h e i r u n f o r g i v i n g n a t u r e which e n t a i l s a s i g n i f i c a n t p a r t i c i p a t i o n
o f "secondary" e f f e c t s i n t h e f a i l u r e modes. The complex d e f i n i t i o n of these
m a t e r i a l s and t h e i r responses r e q u i r e d t o i d e n t i f y behavior has r e s u l t e d i n a
l a r g e i n c r e a s e i n t h e i n f o r m a t i o n volumes f o r processing of a t y p i c a l a i r c r a f t
problem i n t h e s t r u c t u r e s f i e l d .

A t t h e same time, we have seen t i g h t e n i n g up o f requirements n o t o n l y i n terms

o f adverse environment b u t a l s o i n terms o f improved q u a l i t y . I n both cases
t h i s t i g h t e n i n g r e s u l t s i n a need f o r more s o p h i s t i c a t e d analyses.

The general areas o f computing technology and i n f o r m a t i o n science have seen

dramatic changes i n both hardware and software. There have been hardware
changes t h a t c o u l d be p a r l a y e d i n t o : (1) o p t i m i z a t i o n o f s t r u c t u r e s w i t h
p r a c t i c a l c o n s t r a i n t s , ( 2 ) n o n l i n e a r analyses a t reasonable cost, and
( 3 ) m i cro-analyses a t acceptable storage r e q u i rements. Developments have
o c c u r r e d i n t h e f i e l d o f data processing t h a t make data base management a
n a t u r a l c o r n e r s t o n e i n t h e f u t u r e o f CSM. F i n a l l y , t h e r e have been develop-
ments i n t h e f i e l d s o f graphics and CAD/CAM which would make t h e establishment
o f standardized user i n t e r f a c e s e x t r a o r d i n a r i l y u s e f u l both from t h e stand-
p o i n t o f techno1 ogy and management.

Motivation for Increased Development

Advanced composites
Higher complexity
More unforgiving
Larger information volumes
More stringent requirements
Lighter weights
Higher temperatures
Longer lives
Improved computing technology
Numerical methods
Data processing/search and display
CAD/CAM

27
 The o n l y r e s u l t o f s i g n i f i c a n c e i n t h e a i r c r a f t i n d u s t r y i s t o produce a new
" b e t t e r " a i r c r a f t . Any o t h e r r e s u l t i s i n t e r m e d i a t e and w i l l o n l y be accept-
a b l e i f i t c o n t r i b u t e s t o t h i s improvement. The success o f CSM i s t h e r e f o r e
c o n t i ngent on t h r e e a b i 1 it i es :

e To promote and support new technology i n t h e f i e l d s o f s t r u c t u r e s ,

numerical methods, and data processing

e To i n c l u d e and improve methods i n t h e p r e l i m i n a r y design and a n a l y s i s o f

new a i r c r a f t s ( t h e need t o develop and r e f i n e conceptual methods f o r
t h e s e l e c t i o n and comparisons o f s o p h i s t i c a t e d candidates w i t h no
experience b a s i s should be a s t r o n g d r i v e r i n goal s e t t i n g )

To i n t r o d u c e new rnechods and i n t e g r a t e these methods i n a framework t h a t

can produce t h e v i s i b i l i t y , data reduction, and s o p h i s t i c a t i o n s t h a t
a r e necessary t o support a p r o d u c t i o n e f f o r t i n a new era

What Constitutes Success?

e Better aerospace vehicles

I)Therefore "CSM" must

Support technology development

Provide NEW preliminary design and analysis methods

Automate detail design and analysis methods

28
The successful development o f new methods i n computerized s t r u c t u r a l mechanics
r e q u i r e s an understanding o f t h e psychology o f t h e s i t u a t i o n .

1. An e f f i c i e n t method t h a t supports t h e needs o f t h e i n d u s t r y must be

established.

2. A p r a c t i c a l i n p u t and o u t p u t language must be used.

3. A s e t o f u s e f u l user i n t e r f a c e s must be a v a i l a b l e .

4. An a p p r o p r i a t e amount o f v i s i b i l i t y and data r e d u c t i o n f e a t u r e s should

be provided.

5. A r e a l i s t i c m a r k e t i n g e f f o r t should be launched.

6. A developer and user i n t e r a c t i o n i s required.

Yardstick for Success for CSM

Computerized methods development
will not be successful unless
the new “product” is to tally acceptable
to the users

29
 The CSM framework should be designed and implemented w i t h common design ques-
t i o n s i n mind. O f these, t h e q u e s t i o n o f m a t e r i a l choice and mix i s v e r y
complex. An optimum d i s t r i b u t i o n o f m a t e r i a l s i n t h e p r i m a r y s t r u c t u r e o f an
a i r p l a n e i s dependent on t h e c r i t i c a l f a i l u r e modes. Is i t s t r e n g t h ( i n
t e n s i o n ) , s t a b i l i t y , f a t i g u e , damage t o l e r a n c e , o r s t i f f n e s s requirements t h a t
s i z e a l o c a l d e t a i l ? Again, i t can be seen how a g l o b a l / l o c a l / g l o b a l c y c l e
must be used t o come t o g r i p s w i t h t h i s design problem. A l l t h e design
d r i v e r s a r e c l o s e l y r e l a t e d , and t h e q u e s t i o n of environment i n v o l v e s t h e
d e t e r m i n a t i o n o f temperature, m o i s t u r e , presence of chemical, c o r r o s i o n , and
r i s k s o f FOD*. These e f f e c t s need t o be assessed a t l e a s t on a p a r a l l e l t r a c k .
The performance requirements, such as speed, l o a d f a c t o r s , r o l l r a t e s , and
l a n d i n g speeds, a l l a r e very b a s i c and need t o be addressed again i n a l o c a l /
g l o b a l f a s h i o n . F i n a l l y , design c r i t e r i a a r e l l s o f t " e a r l y i n t h e design
process and e v o l v e i n a c y c l i c manner as t h e l o c a l / g l o b a l analyses o f t h e
s t r u c t u r e mature.

What are the Structural Design Drivers?

Material choice and mix

Strength

Stiffness

Damage tolerance

Fatigue performance

Environment

Performance requirements

Design criteria

* f o r e i g n o b j e c t damage

30
There a r e t h r e e f i e l d s o f s t r u c t u r e s t h a t have t o be i n c l u d e d i n order t o g i v e
a t t e n t i o n t o a l l t h e aspects o f e n g i n e e r i n g design i n t h e a i r c r a f t i n d u s t r y .
The f i r s t o f these f i e l d s i s t h e research a c t i v i t y t h a t leads t o a new tech-
nology and a b e t t e r understanding o f t h e i n g r e d i e n t s i n t h e design process.
The second f i e l d i s t h e p r e l i m i n a r y design and a n a l y s i s t h a t leads t o t h e
s e l e c t i o n o f t h e a p p r o p r i a t e f a m i l y o f v e h i c l e s o r products. Both syntheses
and analyses on one s i d e and parameters and v a r i a b l e s on t h e o t h e r a r e i n -
volved i n t h i s design and a n a l y s i s .

The t h i r d f i e l d i s t h e d e t a i l design and a n a l y s i s , o f which perhaps t h e l o c a l l

g l o b a l concept i s more a p p l i c a b l e than anywhere else. Here t h e o b j e c t i v e i s
t o l e a r n as much as p o s s i b l e about a s p e c i f i c design candidate. The demands
a r e e s p e c i a l l y s t r i n g e n t on CSM t o i n c l u d e f e a t u r e s t h a t f a c i l i t a t e d a t a
r e d u c t i o n and d i s p l a y as w e l l as s o p h i s t i c a t e d l o c a l analyses methods.

CSM, What Should be Included?

Research Production
Preliminary Design (Pilots) (Detail Design)

Optimization Trade-Studies Synthesis Detail Design

Local-Global Sensitivities Analysis Optimization

Parameter
Local-Global
Evaluation

LocaCGlobal Error Estimation

31
 The overwhelming requi rement i n t h e p r e l i m i n a r y design and a n a l y s i s phase
deals w i t h s y n t h e s i s methods t h a t d i s c r i m i n a t e between parameters and v a r i -
ables i n t h e design process and makes i t p o s s i b l e t o study a l t e r n a t i v e
formul a t i ons.

Preliminary Design and Analysis

.Access to expert system data base (trade study

results)

Parametric results from local-global analyses as

baseline for optimization

Determination of sensitivities (data base)

Algorithms for parameter determinations and

variable selections

Strategy algorithms, history function

32
The r o l e o f research i n t h e CSM arena i s c e n t r a l and i t s p r i m a r y purpose i s t o
promote b e t t e r understanding o f t h e m a t e r i a l s and s t r u c t u r e s t e c h n o l o g i e s
r e q u i r e d t o produce b e t t e r , v e h i c l e s . To t h a t end, i t i s e s s e n t i a l t o have
s i g n i f i c a n t development i n (1) s t r u c t u r a l and continuum mechanics,
( 2 ) numerical methods, and ( 3 ) f a i l u r e p r e d i c t i o n and t e s t e v a l u a t i o n . T h i s
must be done i n a framework t h a t a l l o w s f o r e m p i r i c i s m t h a t i s s e n s i t i v e t o
user needs, aims a t s y n t h e s i s , and produces p i l o t c a p a b i l i t i e s w i t h w e l l -
defined interfaces.

Research

Phenomena evaluation for better understanding

Special purpose detail analyses

New numerical methods

“Pilot ” development

Test data evaluation methods

Empirical corrections methods

33
 The d e t a i l design and a n a l y s i s f i e l d s have always been c h a r a c t e r i z e d by l a r g e
volumes o f data. T h i s i s becoming more and more t h e case as t h e new advanced
composite m a t e r i a l s a r e i n t r o d u c e d i n t o t h e p r o d u c t i o n environment. ( J u s t
c o n s i d e r t h e simple problem o f c a l c u l a t i n g margins o f safety.) This t o g e t h e r
w i t h more complex requirements and t h e l e s s f o r g i v i n g n a t u r e o f t h e m a t e r i a l s
has r e s u l t e d i n order-of-magnitude increases i n data volumes i n a d d i t i o n t o a
myriad o f l o c a l analyses methods demands.

Production (Detail Design and Analysis)

Postprocessing, search and display (graphics)

Strength checking (software for detail analyses)

Management visibility and access

Data base access (parametric representation ...allowables)

Automated resize. (optimization), baseline updates

Strategy for local-global analyses

Error estimation

34
The CSM framework has t o have an i n f o r m a t i o n c e n t e r t h a t supports t h e l o c a l /
g l o b a l analyses and makes i t p o s s i b l e t o combine a l l aspects of theory,
empiricism, t e s t data, and c r i t e r i a .

T h i s data base should be such t h a t i t supports t h e p r e l i m i n a r y design, d e t a i l

design, and research.

Information Center (Data Base)

Self educating expert system (better modeling, improved

parameter evaluation, trade projections)

Test data evaluation and empirical correction

Sensitivities

History writing

Search and display

35
 The b u i l d i n g b l o c k s f o r CSM are n a t u r a l l y many and o f very d i f f e r e n t nature,
b u t t h e c e n t r a l core i n a l o c a l / g l o b a l system w i l l have t o be a s e t o f s p e c i a l
purpose methods t h a t i n a very f l e x i b l e way can be included, updated, and
rep1 aced.

What are the Building Blocks?

interfaces

36
Without t h e proper d i s c i p l i n e i n t h e s t r u c t u r i n g and t a r g e t i n g f o r CSM, t h e r e
i s a s i g n i f i c a n t r i s k t h a t spur-of-the-moment CSM designs w i l l p r e v a i l . Such
designs w i l l be avoided by a c e n t r a l NASA l e a d e r s h i p .

Should the Philosophy be Uncontrolled Growth?

THE ANSWER IS NO!

37
 From t h e s t a n d p o i n t o f t h e a i r c r a f t i n d u s t r y , t h e l a s t 20 y e a r s have seen an
i n c r e a s e d r e 1 iance on h i g h l y s p e c i a l i z e d e x p e r t s t o p e r f o r m s p e c i a l purpose
( 1 oca1 ) analyses. T h i s i s an u n d e s i r a b l e development, and t h e CSM framework
should p r o v i d e t h e r e q u i rements necessary t o c u r t a i l and r e v e r s e t h i s s o f t w a r e
e n a i n e e r i n a t r e n d . A t t h e same t i m e a number o f qeneral systems have emerged,
r e i c h e d m a h r i t y , expanded t o a h i g h l e v e l o f maintenance budget needs, and
t h e n stagnated. F i n a l l y , t h e framework must p r o v i d e l o c a /g1 oba communi ca-
t i o n s and u s e r i n t e r f a c e s .

Need from Users

Present situation

Increasing dependence on highly specialized

experts

General purpose systems in use require more

user driven long term plans

*As the present trend shows high main-

tenance burden and early stagnation

Proper framework for global to multi-local

communications is missing

38
The s t r a t e y y f o r t h e 1ong-term CSM development c o u l d in c l ude: (1) ident if ica-
t i o n o f e x i s t i n g s o f t w a r e t h a t can be m o d i f i e d f o r a new environment, ( 2 ) a
p l a n t h a t i s based on an e v a l u a t i o n o f i n d u s t r y needs and experience w i t h new
o p e r a t i n g systems, compilers, hardware, and methods, (3) t h e new hardware poten-
t i a l t h a t must be p a r t o f t h e p i c t u r e when t a r g e t i n g s p e c i a l f i e l d s ; o p t i m i z a -
t i o n i s s t i l l a cost/hardware c o n s t r a i n e d a c t i v i t y , and ( 4 ) emphasis t h a t
should be based on t h e p r i o r i t i e s t h a t come out o f i n d u s t r y needs. NASA has a
key r o l e t o p l a y i n p u l l i n g a l l t h e s e t o g e t h e r .

Strategy

Plan that modifies and uses existing software where feasible

Plans driven by what the industry presently knows

Plan that recognizes new hardware potentials

Plan that encourages more “applied research” into fields

driven by industry needs

?9
 The requirements mentioned p r e v i o u s l y t r a n s l a t e i n t o a s e t o f o r g a n i z a t i o n a l
i s s u e s t h a t a r e e s s e n t i a l t o CSM development follow-through. F i r s t , t h e s i z e
o f t h e s o f t w a r e demand i s such t h a t a n a t i o n a l commitment i s necessary i f t h e
technology b e n e f i t s a r e t o be r e a l i z e d . Second, a number o f general-purpose
s c i e n t i f ic systems and an assortment o f speci a1 -purpose s o f t w a r e have demon-
s t r a t e d t h e need f o r a framework designed t o draw upon t h e b e n e f i t s o f both.
F i n a l l y , u s e r acceptance w i l l depend on good v i s i b i l i t y and easy access t o
s o l u t i o n s and r e s u l t s .

Fundamental Organizational Issues

The successful development of

CSM requires central leadership

CSM should provide the framework

for unification of general/global and
special/local methods

Search and display algorithms and

standardized user interfaces are as
important as solutions

40
The backbone o f t h e CSM design approach i s t h e d a t a management, b u t t h e foun-
d a t i o n f o r success can be found i n t h e q u a l i t y o f t h e special-purpose methods
and t h e associated software. Experience has shown t h a t t h e success and accep-
tance o f these methods a r e very dependent on a v a i l a b i l i t y o f r e q u i r e d data,
ease o f i n t e r p r e t a t i o n o f r e s u l t s , and v i s i b i l i t y of t h e steps l e a d i n g t o t h e
f i n a l solution.

The approach should a l s o deal w i t h t h e present t r e n d s i n s t r u c t u r e s , i n which

on one hand we a r e moving i n t o f i e l d s w i t h very l i m i t e d experience bases and
on t h e o t h e r hand we produce i n n o v a t i o n through p o i n t design t e s t i n g . We can
address both s i t u a t i o n s i n a framework t h a t not o n l y accepts new methods w i t h
ease b u t a l s o uses these methods f o r experience development and e m p i r i c a l
evolution.

Could This be a “Better” Approach?

Data communication
Problem definition
(analyseddesignstrategy)

Structural Preliminary Production

Research
analysis design (detail design)
I I I I

Search and General purpose Ad hoc

display algorithms structural analysis software
system

data bases
Requested
results Learning module
Parameter selection
Empirical rules
History and
management
visibility

History and managementvisibility

41
 The t y p i c a l s i t u a t i o n i n t h e s t r u c t u r e s f i e l d i n v o l v e s a number o f analyses
t h a t a r e d i r e c t e d a t d i f f e r e n t l e v e l s o f r e s o l u t i o n . These l e v e l s w i l l have
t o be r e v i s i t e d a number o f t i m e s d u r i n g t h e design process. The e v o l u t i o n o f
t h e d e s i gn, t h e r e f o r e , o b v i o u s l y r e q u i r e s a number o f g l o b a l - t o - 1 oca1 - t o -
g l o b a l t r a n s i t i o n s i n v o l v i n g huge d a t a volumes. The e f f i c i e n c y o f t h e s e
t r a n s i t i o n s r e q u i r e s as much a t t e n t i o n as t h e methods development.

Local-Global Analyses

Internal loads/global, overall FE analyses, stability,

aeroelastic effects, flutter, ...

Detail stress analysis (linear/nonlinear), “local”

buckling, all0wables processing, postbuckling
analyses, interlaminar analyses, residual strength,
damage tolerance analyses, fatigue analysis

Local optimization, multilevel optimization,

automated remodeling, ...

Error estimation

Microdetail-macro strategy (transitions)

42
An understanding o f t h e design d r i v e r s leads t o t h e i d e n t i f i c a t i o n o f a number
o f f i e l d s i n which work should be done i n o r d e r t o e f f i c i e n t l y produce pre-
l i m i n a r y and f i n a l designs. Here t h e p r e l i m i n a r y and f i n a l design processes
a r e d i f f e r e n t i n focus, b u t s i m i l a r p r i n c i p l e s c o u l d serve both e n t e r p r i s e s
and emphasis should be g i v e n t o synthesis-1 ik e f e a t u r e s .

Development Support

Determination of material development targets

Selection of “best” structural materials (based on most important

design drivers, and determine failure progression)

Establishing environmental effects

Determining an optimum set of performance requirements

(temperature/speed, weight, load factor, ...)

Exploration of sensitivities to variation in design criteria

Multilevel strength, stress/stability analyses

Multilevel optimization

Pre- and postprocessing with search and display features

43
 The framework f o r CSM must be such t h a t both c a t e g o r i e s o f users/developers
( t y p i c a l engineers and e x p e r t s ) can be allowed t h e i r proper i n f l u e n c e i n a
manner t h a t promotes n a t u r a l r o l e s and a1 1ows f o r organized communications o f
both needs and r e s u l t s .

What are We Aiming For?

.A system for technology development and improved
scientific understanding for experts

.A system for preliminary design, detail design and

analyses for typical engineers in production environments

44
The l o c a l / g l o b a l a n a l y s i s development i s seen as a three-pronged e f f o r t t h a t
in c l udes : ( 1) advanced methods, (2) t y p i c a l engi n e e r i ng analyses, and
( 3 ) data base and a s s o c i a t e d methods f o r experience development. The advanced
methods would p r i m a r i l y be intended f o r e x p e r t e v a l u a t i o n s , b u t could, i f
p r o p e r l y packaged, be i n c l u d e d among t h e t y p i c a l engi n e e r i ng analyses. The
e x p e r t e v a l u a t i o n s would produce d i r e c t i n p u t t o t h e o v e r a l l design e f f o r t ,
b u t would a l s o feed t h e data base and i n d i r e c t l y support both p a r a m e t r i c
e v a l u a t i o n s and p r e l i m i n a r y design approaches. The example problem represents
a number o f analyses t h a t belong on t h e l o c a l l e v e l and should be considered
from t h e desi gn standpoi n t .

Local-Global Analysis

Visibility

data
base I -
I Test-empiricism-theory

system
U
quality design

45
 The example problem i n t h e c o n t e x t shown i n t h e previous f i g u r e represents a
number o f analyses and syntheses. A l l o f these belong t o t h e l o c a l o r m i c r o
l e v e l when seen from t h e s t a n d p o i n t o f t h e design process. A l l o f these, and
a number o f others, are n a t u r a l development (research) t a r g e t s t h a t a f t e r
implementation should e i t h e r d i r e c t l y feed r e s u l t s i n t o t h e " t y p i c a l "
e n g i n e e r i n g a c t i v i t y o r produce r e s u l t s t o t h e l o c a l data base f o r f u t u r e use
o r paramet r i c access. A1 1 these devel opments should r e s u l t i n p i 1o t capabi 1i-
t i e s o r products t h a t i n an ad hoc f a s h i o n would s a t i s f y one o f t h e above two
requirements.

What are the Analyses for the Example Problem?

0 Local failure modes and progression

based on test, for correlation and
empirical inputs
0 Multilevel analyses: interlaminar,
local damage growth, local influence
of postbuckling
Stability analyses
Postbuckling
analyses
5"
0 Nonlinear material
response,...
0 Flaw-growthfor a number of
damages
Local optinnization
0 Allowables determination
0 Automated "remodeling" ' and
Parametric representation and transition regions
empirical input to data base
0 Screening for criticalities, critical
requirements
0 "Gross" FE properties
0 Residual stresses

46
The example problem i s one o f many analogous l o c a l problems t h a t must be
solved as p a r t o f t h e design process. Each one o f these problems i s solved a
number o f times d u r i n g t h e e v o l u t i o n o f t h e design. We a r e t h e r e f o r e f o r c e d
t o minimize t h e number o f m i c r o e v a l u a t i o n s and e x p e r t involvement i n o r d e r t o
produce an e f f i c i e n t design process. T h i s can be done by i n c l u d i n g t h e
r e s u l t s on a communal experience b a s i s i n a way t h a t supports t h e p r a c t i c a l
e n g i n e e r i n g tasks. Many o f t h e l o c a l a n a l y s i s packages must be designed f o r
minimum involvement by experts. I n many types of problems, i t w i l l be
necessary t o i d e n t i f y i m p o r t a n t parameters and produce s o l u t i o n s i n advance i n
a format t h a t can be accessed i n a data base management environment. Very
s i m i l a r c o n s i d e r a t i o n s apply t o t h e a l l o w a b l e s question, w h e t h e r ‘ i t i s pro-
cessing o f m a t e r i a l s t r u c t u r a l allowables. F i n a l l y , i n t h e f a i l u r e p r e d i c t i o n
development, one can see elements o f t h e o t h e r types and here one can h a r d l y
expect p r a c t i c a l support o f t h e design process w i t h o u t a t w o - l e v e l s o f t w a r e
development t h a t u l it m a t e l y depends on empi r i c a l m o d i f i c a t i o n s .

The Example Problem

Local with “micro” requirements

Local analyses with minimum expert requirements

Basis for parameter selection

Panel for allowables generation

Candidate for failure prediction development

47
 I n conclusion, we f i n d t h a t t h e r e a r e a number o f research t a r g e t s w a i t i n g f o r
i n i t i a t i v e from t h e e n g i n e e r i n g community. It has become c l e a r , however, t h a t
t h e m a j o r i t y o f these t a r g e t s w i l l be missed i f not approached from t h e stand-
p o i n t o f t h e i r o v e r a l l r o l e i n t h e design process. It i s a l s o f e l t t h a t t h e
o b j e c t i v e s w i l l n o t be met w i t h o u t a proper technology t r a n s f e r t o t h e users.
This n a t u r a l l y i n v o l v e s b o t h t h e s o f t w a r e packaging and t h e promotional
a c t iv i t y necessary f o r d i ssemi n a t ion.

Conclusions
Research should include

Analysis methods (nonlinear, advanced materials,...

structures)

Numerical methods

Artificial intelligence

Search and display/data processing

Emphasis on technology transfer to industry

48
NASA has a key r o l e i n t h e development o f CSM f o r t h e 1990's and beyond. A
n a t i o n a l e f f o r t i s r e q u i r e d i f new m a t e r i a l s (composites), new computers, new
methods, and new requirements are t o be addressed i n a manner t h a t e s t a b l i s h e s
economic advances' and preserves t h e s u p e r i o r s a f e t y r e c o r d e s t a b l i s h e d by t h e
a i r c r a f t i n d u s t r y . NASA has a r o l e not o n l y i n l e a d e r s h i p b u t a l s o as com-
municator a s s u r i n g technology t r a n s f e r and promoting user acceptance.

Recommendations

Clear interfaces
0 Standardization

.Standing committee

49
 NONLINEAR RESPONSE OF A BLADE-STIFFENED GRAPHITE-EPOXY PANEL
WITH A DISCONTINUOUS STIFFENER: WORK I N PROGRESS

Norman F. Knight, J r . , William H. Greene,

and W. J e f f e r s o n S t r o u d
NASA Langley Research C e n t e r
Hampton, V i r g i n i a

INTRODUCTION

D i s c o n t i n u i t i e s and e c c e n t r i c i t i e s are u s u a l l y p r e s e n t i n p r a c t i c a l
s t r u c t u r e s . I n a d d i t i o n , p o t e n t i a l damage of o t h e r w i s e p e r f e c t s t r u c t u r e s i s o f t e n
an important design consideration. Predicting the s t r u c t u r a l response i n the
p r e s e n c e o f d i s c o n t i n u i t i e s , e c c e n t r i c i t i e s , and damage is p a r t i c u l a r l y d i f f i c u l t
when t h e component is b u i l t from graphite-epoxy materials o r is l o a d e d i n t o t h e
n o n l i n e a r range. Recent i n t e r e s t i n a p p l y i n g graphite-epoxy materials t o a i r c r a f t
p r i m a r y s t r u c t u r e s has l e d t o s e v e r a l s t u d i e s of p o s t b u c k l i n g b e h a v i o r and f a i l u r e
c h a r a c t e r i s t i c s o f graphite-epoxy s t r u c t u r a l components ( e . g . , refs. 1-31.
However, these s t u d i e s c o n c e n t r a t e d on two t o p i c s : p r e d i c t i o n o f t h e o v e r a l l
r e s p o n s e of composite s t r u c t u r a l components i n the p o s t b u c k l i n g r a n g e or f a i l u r e
mechanisms and a n a l y t i c a l f a i l u r e p r e d i c t i o n t e c h n i q u e s f o r f i b r o u s composite
materials. The problem o f c a l c u l a t i n g d e t a i l e d stress d i s t r i b u t i o n s around
d i s c o n t i n u i t i e s i n b u c k l e d , composite s t r u c t u r a l components f o r u s e w i t h t h e
v a r i o u s a n a l y t i c a l f a i l u r e p r e d i c t i o n t e c h n i q u e s h a s n o t been t h o r o u g h l y e x p l o r e d .

The p u r p o s e of t h i s p a p e r i s t h e a p p l i c a t i o n of c o m p u t a t i o n a l methods t o t h e
d e t a i l e d stress a n a l y s i s problem which is t h e f o c u s o f t h i s s e s s i o n o f t h e
workshop. One approach t o u n c o v e r i n g t h e d i f f i c u l t i e s o f t h i s t y p e of a n a l y s i s and
t o p r o v i d i n g s p e c i f i c d i r e c t i o n s f o r f u t u r e research i n t h i s area is a d i r e c t
a t t a c k on t h e problem u s i n g c u r r e n t l y a v a i l a b l e a n a l y s i s tools. A c a n d i d a t e
problem has been selected and t h e remainder o f t h e p a p e r describes e x p e r i e n c e s from
c a l c u l a t i n g its s t r u c t u r a l response.

PRECEDW PAGE R A M K NOT FILMED

 BLADE-STIFFENED GRAPHITE-EPOXY PANEL WITH A DISCONTINUOUS STIFFENER:
FOCUS PROBLEM

The f o c u s problem f o r the l o c a l / g l o b a l stress a n a l y s i s s e s s i o n of t h i s

workshop i s t o d e t e r m i n e t h e n o n l i n e a r r e s p o n s e o f a f l a t b l a d e - s t i f f e n e d g r a p h i t e -
epoxy p a n e l w i t h a d i s c o n t i n u o u s s t i f f e n e r . The material s y s t e m f o r t h e p a n e l is
T300/5208 graphite-epoxy w i t h a nominal p l y t h i c k n e s s o f 0.0055 i n . T y p i c a l l a m i n a
p r o p e r t i e s f o r t h i s graphite-epoxy system are 19,000 k s i for t h e l o n g i t u d i n a l
Young's modulus, 1 , 8 9 0 k s i f o r t h e t r a n s v e r s e Young's modulus, 930 k s i f o r t h e
shear modulus, and 0.38 f o r t h e major P o i s s o n ' s r a t i o . The p a n e l s k i n has 25 p l i e s
[445/0, / 5 4 5 / 0 /?45/0, / T 4 5 / 0 ,/+45/0,/545 1) and t h e b l a d e s t i f f e n e r s have 2 4 p l i e s
[ [:f 45 1 0 , I T 4 5 1 . The o v e r a l l l e n g t h of t h e p a n e l is 30 i n . , t h e o v e r a l l w i d t h is
11.5 i n . , s t i f f e n e r s p a c i n g is 4.5 i n . , s t i f f e n e r h e i g h t is 1.4 i n . , and t h e h o l e
diameter i s 2 i n . The l o a d i n g i s uniform a x i a l compression. The l o a d e d e n d s of
t h e p a n e l a r e clamped and t h e s i d e s a r e f r e e .

T h i s problem was selected a s t h e f o c u s problem b e c a u s e e x p e r i m e n t a l r e s u l t s

a r e a v a i l a b l e and because i t has c h a r a c t e r i s t i c s which o f t e n r e q u i r e a l o c a l / g l o b a l
a n a l y s i s . These c h a r a c t e r i s t i c s i n c l u d e a d i s c o n t i n u i t y , e c c e n t r i c l o a d i n g , l a r g e
d i s p l a c e m e n t s , l a r g e stress g r a d i e n t s , h i g h i n p l a n e l o a d i n g , and a b r i t t l e material
s y s t e m . T h i s problem r e p r e s e n t s a g e n e r i c c l a s s of l a m i n a t e d composite s t r u c t u r e s
w i t h d i s c o n t i n u i t i e s i n which t h e i n t e r l a m i n a r stress s t a t e becomes i m p o r t a n t .

* G ~ a p h j t ~ - e p ~ (xTy3 0 0 f 5 2 0 8 )
* Flat panel with t
e 3 0 in. long
0 1 1 . 5 in. wide
0 Stiffener spacing of 4.5 in.
0 Stiffener height of 1.4 in.
0 2,0--in,-diameter hole
* 25-ply panel skin
I! 24-ply blade stiffeners
0 A x i a ~ l yloaded with loaded ends clamped and sides free

52
 --

LOCAL/GLOBAL TERMINOLOGY

The d e f i n i t i o n of a local/global s t r u c t u r a l analysis is not precise. For

example, several l e v e l s of d e t a i l a r e considered i n t h e a n a l y s i s of an a i r c r a f t
s t r u c t u r e , and t h e concepts of l o c a l and global can change with every change i n
a n a l y s i s l e v e l . To some, the e n t i r e a i r c r a f t is the global model, and a fuselage
s e c t i o n is the l o c a l model. To others, lamination theory represents the global
model, and micromechanics i s used f o r the l o c a l model. For t h i s workshop, a
local/global a n a l y s i s is a l o c a l , d e t a i l e d s t r e s s a n a l y s i s w i t h i n a l a r g e r , l e s s -
refined global analysis model. The overall response of t h e panel is the global
problem; the response near the hole is t h e l o c a l problem.

Current areas of research associated w i t h local/global methodologies a r e

described i n t h e l i t e r a t u r e (e.g., r e f s . 4 and 5 ) . One research area
( d i s c r e t i z a t i o n procedures) includes finite-element methods, f i n i t e - d i f f e r e n c e
methods, and boundary element o r boundary i n t e g r a l methods. Adaptive mesh
refinement, h- and p-convergence, and e r r o r analysis a r e current research topics i n
t h i s area which address d i s c r e t i z a t i o n e f f e c t s i n t h e presence of a l a r g e stress
gradient. A second area ( r e f i n e d t h e o r i e s ) includes research i n transverse shear
formulations and three-dimensional e l a s t i c i t y solutions. These topics focus on the
mathematical representation of t h e mechanics of the problem. A t h i r d area includes
c l a s s i c a l and closed-form solutions'which a r e often r e s t r i c t e d t o simple
geometries, s p e c i f i c boundary conditions and ,material systems, and often t o a
l i n e a r response prediction. A f o u r t h area is h y b r i d techniques i n which two o r
more methods a r e used simultaneously b u t i n d i f f e r e n t domains of the s t r u c t u r e .
A l l f o u r of thes'e a r e a s of research a r e addressed i n the local/global session of
t h i s workshop.

0 Concept of IocaVglobal changes with analysis level

0 Definitions

--- Global means overall panel response

--- Local means response near the hole

0 LocaVglobal methodologies

--- Discretization procedures

--- Refined theories

--- Classical and closed-form solutions

--- Hybrid techniques

53
 APPLICATION OF THE FINITE-ELEMENT METHOD
TO LOCAL/GLOBAL STRESS ANALYSIS

The l o c a l / g l o b a l methodology adopted f o r t h i s paper is t h e f i n i t e - e l e m e n t

method because o f i t s g e n e r a l i t y . The f i r s t s t e p i n a p p l y i n g t h e method was t o
d e v e l o p and v e r i f y a f i n i t e - e l e m e n t model of t h e f o c u s problem. Model v e r i f i c a t i o n
i n v o l v e d s o l v i n g s i m p l e r example cases and comparing t h e results w i t h other
a n a l y t i c a l r e s u l t s . T h i s model v e r i f i c a t i o n p r o c e s s was a i d e d by t h e development
o f a f l e x i b l e mesh g e n e r a t i o n c a p a b i l i t y which a l l o w e d v a r i o u s f i n i t e - e l e m e n t
d i s c r e t i z a t i o n s t o b e e v a l u a t e d r a p i d l y and s y s t e m a t i c a l l y . The mesh g e n e r a t i o n
c a p a b i l i t y a l s o p r o v i d e d a n easy way t o c o n s t r u c t and s t u d y s e v e r a l i d e a l i z e d
example cases.

Once a n a d e q u a t e f i n i t e - e l e m e n t model f o r t h e g l o b a l r e s p o n s e was v e r i f i e d ,

t h e n o n l i n e a r s t r u c t u r a l r e s p o n s e was c a l c u l a t e d . To i d e n t i f y the l o c a l modeling
d e t a i l r e q u i r e d t o p r e d i c t a c c u r a t e l y the stress d i s t r i b u t i o n n e a r t h e h o l e , l i n e a r
stress a n a l y s e s were performed on a r e c t a n g u l a r p l a t e w i t h a c i r c u l a r h o l e u s i n g
s e v e r a l r e f i n e d 2-D models n e a r t h e h o l e .

0 Finite-element model development

0 Finite-element model verification

0 Global nonlinear response prediction

0 Local linear stress analysis

54
 FINITE-ELEMENT MODEL DEVELOPMENT

The model development s t r a t e g y is t o p r e d i c t t h e g l o b a l n o n l i n e a r r e s p o n s e

u s i n g t h e complete model and t h e n t o c o n s t r u c t a r e f i n e d , l o c a l 2-D model f o r a
small d i s t a n c e away from the hole t o p r e d i c t a c c u r a t e l y t h e large stress g r a d i e n t .
Displacements and r o t a t i o n s from t h e g l o b a l n o n l i n e a r s o l u t i o n o b t a i n e d u s i n g t h e
complete model w i l l b e a p p l i e d t o the r e f i n e d model and t h e s t a t e o f stress w i l l b e
determined. T h i s s t r a t e g y w i l l be referred t o as a m u l t i - l e v e l o r ffzoom-inff
approach.

The automated mesh g e n e r a t i o n c a p a b i l i t y allowed v e r s a t i l e modeling of t h e

complete problem a s well a s l o c a l r e g i o n s n e a r t h e h o l e . The a n a l y s t c o u l d s p e c i f y
t h e number of e l e m e n t s a c r o s s t h e s t i f f e n e r d e p t h and down the l e n g t h of t h e p a n e l ,
the number of r i n g s of e l e m e n t s around t h e h o l e , a n d t h e number of e l e m e n t s around t h e
h o l e , and c o u l d c o n t r o l t h e element s p a c i n g i n t h e v i c i n i t y of the h o l e . Models could
a l s o be g e n e r a t e d w i t h the h o l e and d i s c o n t i n u o u s s t i f f e n e r f i l l e d - i n o r w i t h no
stiffeners.

Complete global model

Global model
of panel skin
model

55
 FINITE-ELEMENT MODEL VERIFICATION:
BLADE STIFFENED PANEL

I n t h e f i r s t s t e p o f t h e v e r i f i c a t i o n p r o c e s s , a s i m p l i f i e d v e r s i o n of t h e
f o c u s problem was s t u d i e d . T h i s s i m p l e r problem was i d e n t i c a l t o t h e f o c u s problem
e x c e p t t h a t t h e h o l e and d i s c o n t i n u o u s s t i f f e n e r a r e f i l l e d - i n and t h e end boundary
c o n d i t i o n s are now s i m p l e s u p p o r t c o n d i t i o n s . For t h i s p r i s m a t i c p a n e l , a n e x a c t
s o l u t i o n was o b t a i n e d u s i n g t h e PASCO computer code ( P a n e l A n a l y s i s and S i z i n g
Code, ref. 6 ) . The f i n i t e - e l e m e n t a n a l y s i s system EAL ( E n g i n e e r i n g A n a l y s i s
Language;ref. 7 ) was used f o r t h e f i n i t e - e l e m e n t a n a l y s i s . The f i n i t e - e l e m e n t
model used i n t h e v e r i f i c a t i o n was developed from t h a t of t h e f o c u s problem t o
d e t e r m i n e i f any problems r e l a t e d t o element d i s t o r t i o n o r a s p e c t r a t i o were
p r e s e n t . The p r e b u c k l i n g boundary c o n d i t i o n s and end l o a d i n g a r e s u c h t h a t a
uniform stress s t a t e is p r e s e n t i n t h e s k i n and t h e b l a d e s t i f f e n e r s . The three
l o w e s t b u c k l i n g e i g e n v a l u e s o b t a i n e d u s i n g EAL are very c l o s e t o t h e PASCO
solutions.

Buckling mode

----

Eigenvalues
1 2 3
PASCO 44536 51063 61601
EAL 44652 51182 60975
Difference +0.26% +0.23% - 1 .O2%

56
 FINITE-ELEMENT MODEL VERIFICATION:
FOCUS PROBLEM

The next s t e p of the v e r i f i c a t i o n process was t o define an adequate f i n i t e -

element model f o r t h e global response of the focus problem. Finite-element model
v e r i f i c a t i o n f o r the focus problem s t a r t e d w i t h a "reasonable g r i d " of 376 4-node
assumed-stress q u a d r i l a t e r a l elements and 422 nodes. T h i s d i s c r e t i z a t i o n i s
referred t o a s Mesh 1 . A second, refined g r i d of 1088 4-node q u a d r i l a t e r a l
elements and 1168 nodes (Mesh 2 ) was generated f o r model v e r i f i c a t i o n . Linear
bifurcation buckling s o l u t i o n s f o r Mesh 1 and Mesh 2 were compared t o e s t a b l i s h the
adequacy of the models. The three lowest eigenvalues from both d i s c r e t i z a t i o n s
agree w i t h i n approximately one percent.

Mesh 1 Mesh 2
3 7 6 elements 1 0 8 8 elements
4 2 2 nodes

Eigenvalues
1 2 3 4
Mesh 1 4 1 3 7 8 5 2 7 5 4 5 4 2 8 8 5 5 3 4 4
Mesh 2 4 1 8 2 9 5 2 5 3 3 5 4 2 5 9 5 6 8 9 5
Change -1.08% +0.42% +0.05% - 2 . 7 3 %

57
 LINEAR BIFURCATION BUCKLING SOLUTION

Oblique views o f t h e p r e b u c k l i n g d e f o r m a t i o n p a t t e r n and the eigenmodes

c o r r e s p o n d i n g t o t h e f o u r l o w e s t e i g e n v a l u e s a r e shown i n t h i s f i g u r e . The
I
d i s c o n t i n u o u s s t i f f e n e r l e a d s t o an e c c e n t r i c l o a d i n g c o n d i t i o n which c a u s e s l a r g e
out-of-plane d i s p l a c e m e n t s t o d e v e l o p n e a r the h o l e from t h e o n s e t of l o a d i n g .
Because of t h i s c o u p l i n g between i n p l a n e and out-of-plane d i s p l a c e m e n t s , no l i n e a r
I
e q u i l i b r i u m p a t h e x i s t s and t h e l i n e a r b i f u r c a t i o n b u c k l i n g r e s u l t s do n o t have t h e
t r a d i t i o n a l meaning. The l i n e a r b u c k l i n g s o l u t i o n s may b e used a s a g u i d e i n
s e l e c t i n g t h e i n i t i a l l o a d f o r t h e n o n l i n e a r a n a l y s i s and i n c h o o s i n g a l o a d s t e p
s i z e . However, t h e i r main use is i n s t u d y i n g t h e e f f e c t s of s p a t i a l
I discretization.

Linear prebuckling deformation pattern

Mode 3

Mode 2

58
 GLOBAL NONLINEAR RESPONSE PREDICTION

The global n o n l i n e a r r e s p o n s e p r e d i c t e d f o r the f o c u s problem was o b t a i n e d

u s i n g a new release o f EAL. T h i s new release h a s a n o n l i n e a r a n a l y s i s c a p a b i l i t y
using a corotational formulation with l i n e a r strain-displacement r e l a t i o n s within
t h e e l e m e n t s . F o r t h i s problem, t h e l o a d i n g was a p p l i e d i n i n c r e m e n t s w i t h a f u l l
Newton-Raphson algorithm. Convergence was based on t h e maximum e r r o r i n t h e
residual force vector.

An o b l i q u e view o f t h e deformed geometry f o r t h e l a s t c a l c u l a t e d s o l u t i o n is

similar t o t h e l i n e a r s o l u t i o n shown p r e v i o u s l y , i n d i c a t i n g t h a t t h e p r i m a r y
e q u i l i b r i u m p a t h is b e i n g f o l l o w e d . A g l o b a l r e s p o n s e q u a n t i t y , end s h o r t e n i n g , is
n e a r l y a l i n e a r f u n c t i o n of t h e a p p l i e d l o a d . A l o c a l r e s p o n s e q u a n t i t y , out-of-
p l a n e d i s p l a c e m e n t a t t h e edge of the hole and b l a d e s t i f f e n e r , i n d i c a t e s l a r g e
d i s p l a c e m e n t s from t h e o n s e t of l o a d i n g . L o n g i t u d i n a l i n p l a n e s t r e s s - r e s u l t a n t
d i s t r i b u t i o n s f o r two v a l u e s o f t h e a p p l i e d l o a d , as a f u n c t i o n o f d i s t a n c e from
t h e h o l e , i n d i c a t e h i g h i n p l a n e stresses and a h i g h stress g r a d i e n t n e a r t h e hole.

These h i g h i n p l a n e stresses and stress g r a d i e n t s coupled w i t h t h e l a r g e o u t -

o f - p l a n e d i s p l a c e m e n t s and t h e f r e e edge of t h e h o l e may c a u s e material
n o n l i n e a r i t i e s , l o c a l f a i l u r e s , a n d / o r d e l a m i n a t i o n s t o d e v e l o p i n order t o p r o v i d e
l o c a l stress r e l i e f mechanisms ( l i k e p l a s t i c i t y i n metal s t r u c t u r e s ) n e a r t h e h o l e
and b l a d e s t i f f e n e r . However, an a c c u r a t e p r e d i c t i o n of the e f f e c t s of these
mechanisms on t h e g l o b a l n o n l i n e a r r e s p o n s e is beyond t h e c u r r e n t a n a l y s i s
c a p a b i l i t i e s . Stress f r a c t u r e c r i t e r i a a r e developed i n r e f e r e n c e 8 f o r a n i n p l a n e
l o a d i n g c o n d i t i o n i n which t h e i n f l u e n c e o f these stress r e l i e f mechanisms c a n b e
a c c o u n t e d f o r i n f a i l u r e s t u d i e s w i t h o u t knowing e x a c t l y what is happening l o c a l l y
n e a r t h e h o l e . The p o i n t stress f a i l u r e c r i t e r i o n developed i n r e f e r e n c e 8 and
a p p l i e d i n r e f e r e n c e 9 t o a broad c l a s s o f l a m i n a t e d composite p l a t e s w i t h h o l e s
w i l l b e used i n t h i s s t u d y a s a g u i d e f o r e s t a b l i s h i n g a n a d e q u a t e f i n i t e - e l e m e n t
model f o r p r e d i c t i n g t h e stress d i s t r i b u t i o n s n e a r t h e h o l e .

Out-of-plane
50
End shortening
" 5c
displacement

25 25
P P
(KIPS) --- Extension of (KIPS)
linear solution
Linear buckling
solution
0 .02 .04 -06 0 .1 .2 .3
U (in.) W (in.)

59
 GLOBAL NONLINEAR RESPONSE PREDICTION
lnplane stress resultant
at panel midlength
I
I
I
FEdge of hole I

\\\
I

I
I

Deformed geometry NX
(lb/in.)
4000
Load 6 1 I

. \ Load A I
I
I
I I I l l
1.0 2.0 3.0 4.0
Y (in.)

60
 POINT STRESS FAILURE C R I T E R I O N

The point s t r e s s f a i l u r e c r i t e r i o n assumes t h a t f a i l u r e occurs when the s t r e s s

a t a distance do away from the edge of the hole reaches ultimate. The d i s t a n c e d o
is a c h a r a c t e r i s t i c dimension which takes the form of a material property. A
consequence of u s i n g t h i s c r i t e r i o n is t h a t an accurate prediction of the s t a t e of
s t r e s s p r e c i s e l y a t t h e f r e e edge of t h e hole is not required. I t i s only
necessary t o have an accurate stress prediction a t a distance d o from the edge of
t h e hole. Based on r e s u l t s of reference 9 w i t h a s i m i l a r graphite-epoxy material
system, a value of d o = 0.05 inches was assumed. Model A corresponds t o the global
model near t h e hole f o r Mesh 1 . Model B is a refined model which more accurately
p r e d i c t s the s t r e s s gradient near t h e hole.

Assumed value of
do is 0.05 in.

Model A Model B

61
 ~~

LOCAL LINEAR STRESS N LYSIS

To u n d e r s t a n d how t h e s p a t i a l d i s c r e t i z a t i o n n e a r t h e h o l e a f f e c t s t h e
p r e d i c t i o n o f t h e stress a t d o , a s i m p l e r s t r u c t u r a l c o n f i g u r a t i o n was c o n s i d e r e d
which d i d n o t i n c l u d e t h e s t i f f e n e r s . Using t h i s p l a n a r s t r u c t u r e , a n a d e q u a t e 2 - D
f i n i t e - e l e m e n t model was i d e n t i f i e d f o r the l o c a l stress a n a l y s i s . T h i s a p p r o a c h
p r o v i d e d t h e n e c e s s a r y i n s i g h t r e q u i r e d f o r a m u l t i - l e v e l model o f t h e f o c u s
problem. An a l t e r n a t e approach would have been t o use an a d a p t i v e mesh r e f i n e m e n t
procedure. However, no s u c h p r o c e d u r e was a v a i l a b l e . The l o n g i t u d i n a l i n p l a n e
stress r e s u l t a n t d i s t r i b u t i o n s a s a f u n c t i o n o f d i s t a n c e away from t h e edge o f t h e
h o l e are shown f o r two f i n i t e - e l e m e n t models. The r e s u l t s from b o t h models
approach one a n o t h e r away from t h e h o l e . However, a t a d i s t a n c e d o from t h e edge
o f t h e h o l e , t h e s o l u t i o n s f o r Models A and B d i f f e r by 12.5 p e r c e n t . The f i n i t e -
element model i n t h e v i c i n i t y o f t h e h o l e was r e f i n e d by d o u b l i n g t h e number of
elements. The i n p l a n e stress r e s u l t a n t a t d o changed by o n l y 2.2 p e r c e n t between
Model B and a model w i t h h a l f a s many elements.

Unstiffened Flat Panel

V
NX I i\\
(Iblin.1 I j \\

0 1.0 2.0

62
 STATUS AND A D D I T I O N A L TASKS

The o v e r a l l s t r a t e g y f o r t h i s s t u d y i s t o p r e d i c t t h e global n o n l i n e a r
r e s p o n s e u s i n g t h e complete g l o b a l model and t h e n t o c o n s t r u c t a r e f i n e d , l o c a l 2-D
model f o r a small d i s t a n c e away from t h e hole. The g l o b a l n o n l i n e a r r e s p o n s e has
been p r e d i c t e d f o r t h e f o c u s problem and t h e l o c a l modeling d e t a i l r e q u i r e d f o r a n
a c c u r a t e l o c a l stress a n a l y s i s n e a r t h e h o l e o f a n u n s t i f f e n e d p a n e l has been
i d e n t i f i e d . The t a s k s t h a t remain t o be completed f o r t h e f o c u s problem i n c l u d e
p e r f o r m i n g t h e m u l t i - l e v e l a n a l y s i s and a p p l y i n g a f a i l u r e c r i t e r i o n . The m u l t i -
l e v e l a n a l y s i s w i l l i n v o l v e a p p l y i n g t h e d i s p l a c e m e n t s and r o t a t i o n s from t h e
g l o b a l n o n l i n e a r s o l u t i o n on t h e r e f i n e d l o c a l 2-D model and d e t e r m i n i n g t h e s t a t e
of stress a t d o . I n a d d i t i o n , a t h r e e - d i m e n s i o n a l model n e a r t h e d i s c o n t i n u i t y
w i l l be required for an accurate determination of the through-the-thickness state
o f stress ( i . e . , normal and t r a n s v e r s e s h e a r i n g stress d i s t r i b u t i o n s ) . The u s e o f
3-D e l e m e n t s w i t h i n a 2-D model w i l l a l s o r e q u i r e a s t r a t e g y f o r the t r a n s i t i o n or
b l e n d i n g of the two models.

Status

0 Finite element model developed and verified

0 Global nonlinear response predicted

0 Required modeling detail identified for stress gradient

near the hole for an unstiffened panel

Additional tasks for focus problem

0 Perform multi-level 2-D analysis (refined, local 2-D model)

0 Apply point stress failure criterion

0 Perform multi-level 3-0 analysis (refined, local 3-D model)

63
 SUMMARY

The l o c a l / g l o b a l n o n l i n e a r stress a n a l y s i s o f a b l a d e - s t i f f e n e d graphite-epoxy

p a n e l w i t h a d i s c o n t i n u o u s s t i f f e n e r is indeed a c o m p u t a t i o n a l c h a l l e n g e .
S u b s t a n t i a l e n g i n e e r i n g e f f o r t is r e q u i r e d i n modeling t h e s t r u c t u r e , i n v e r i f y i n g
t h a t t h e p h y s i c s of t h e problem a r e modeled, and i n i n t e r p r e t i n g t h e p r e d i c t e d
n o n l i n e a r s o l u t i o n s . Approximately f i f t y p e r c e n t o f t h e a n a l y s i s e f f o r t t o d a t e was
devoted t o model development and v e r i f i c a t i o n . The development o f a f l e x i b l e mesh
g e n e r a t i o n c a p a b i l i t y was e s s e n t i a l f o r model v e r i f i c a t i o n . S e v e r a l models o f
similar b u t s i m p l e r s t r u c t u r e s were r e q u i r e d and e a s i l y g e n e r a t e d u s i n g t h e
automated mesh g e n e r a t o r .

To complete t h e a n a l y s i s e f f o r t f o r t h e f o c u s problem s e v e r a l i s s u e s need t o be

addressed. The t r a n s i t i o n o r i n t e r f a c e between t h e v a r i o u s l e v e l s o f t h e m u l t i -
l e v e l model needs t o b e d e f i n e d . An a d a p t i v e mesh r e f i n e m e n t p r o c e d u r e is needed
t o automate t h e d e f i n i t i o n o f t h e f i n i t e - e l e m e n t models a t each stage o f t h e m u l t i -
l e v e l approach. To o b t a i n a d e t a i l e d t h r o u g h - t h e - t h i c k n e s s stress d i s t r i b u t i o n , a
three-dimensional a n a l y s i s w i l l be r e q u i r e d and t h e number of three-D e l e m e n t s through-
t h e - t h i c k n e s s of t h e l a m i n a t e needs t o be determined. I n a d d i t i o n , t o p r e d i c t t h e
r e s p o n s e of t h e s t r u c t u r e up t o o v e r a l l s t r u c t u r a l f a i l u r e , a p r o g r e s s i v e f a i l u r e
a n a l y s i s c a p a b i l i t y would b e r e q u i r e d i n which v a r i o u s f a i l u r e mechanism and
f a i l u r e c r i t e r i a are i n c o r p o r a t e d .

0 Substantial engineering effort required in modeling,

model verification, and response interpretation.

0 Flexible mesh generation capability essential to model

verification.

0 Definition of transitionlinterface region between

multi-level models required.

0 Required number of 3-D elements through-the-thickness

to be determined.

0 Nonlinear analysis procedure with progressive failure

analysis capability needed.

64
 REFERENCES

1. Starnes, James H . b J r . , Dickson, John N . and Rouse, Marshall, "Postbuckling of

Graphite-Epoxy Panels;" ACEE Composite S t r u c t u r e s Technology: Review of
Selected NASA Research on Composite Materials and S t r u c t u- r e s , NASA CP-2321,
1984, pp. 137-160.

2. Starnes, James H., Jr. and Williams, J e r r y C., "Failure C h a r a c t e r i s t i c s of

Graphite-Epoxy S t r u c t u r a l Components Loaded i n Compression," NASA TM-84552,
September 1982. Also, i n Mechanics of Composite Materials; Recent Advances,
Proceedings of the I U T A M Symposium on Mechanics of Composite Materials, Z.
Hashin and C. T. Herakovich ( e d i t o r s ) , Pergamon Press, Inc., 1983, pp. 283-
306.

3. Noor, A. K., Shuart, M. J . , S t a r n e s , J. H., Jr., and Williams, J. G.

(Compilers), F a i l u r e Analysis and Mechanisms of F a i l u r e of Fibrous Composite
S t r u c t u r e s , NASA CP-2278, 1983.

4. Zienkiewicz, 0. C., "The Generalized F i n i t e Element Method---State of the A r t

and Future Directions," Journal of Applied Mechanics, ASME, Vol. 50, December
1983, pp. 1210-1217.

5. Noor, A. K. and Pilkey, W. D. ( e d i t o r s ) , State-of-the-Art Surveys on F i n i t e

Element Technology, ASME, New York, 1983.

6. Anderson, Melvin S. and Stroud, W. J e f f e r s o n , "A General Panel Sizing Computer

Code and I t s Application t o Composite S t r u c t u r a l Panels," A I A A Journal, Vol.
17, No. 8 , August 1979, pp. 892-897.

7. Whetstone, W. D . , E I S I - E A L Engineering Analysis Language Reference Manual -

EISI-EAL System'Level 2091, Engineering Information Systems, Inc., S m ; o s € ,
C k , J u l y 1983.

8. Whitney, J . M. and N u i s m e r , R. J . , "Stress Fracture C r i t e r i a f o r Laminated

Composites Containing S t r e s s ConCentrations," Journal of Composite Materials,
Vol. 8 , J u l y 1974, p p . 253-265.

9. M i k u l a s , Martin M . , Jr., "Failure Prediction Techniques f o r Compression Loaded

Composite Laminates w i t h Holes," Selected NASA Research i n Composite Materials-
and S t r u c t u r e s , NASA CP-2142, 1980, pp. 1-34.

65
 N89-24642 ~

C O M P UTATI0NA L P R 0C E DU R ES FOR
POSTBUCKLING OF COMPOSITE SHELLS

G. M. Stanley and C. A. Felippa

Lockheed Palo Alto Research Laboratory
Palo Alto, California

SUMMARY

A recently developed finite-element capability for general nonlinear shell analysis. fea-
turing the use of three-dimensional constitutive equations within an efficient resultant-
oriented framework, is employed t o simulate the postbuckling response o f an axially
compressed composite cylindrical panel with a circular cutout. The problem is a
generic example of modern composite aircraft components for which postbuckling
strength (Le., fail-safety) is desired in the presence of local discontinuities such as
holes and cracked stiffeners. While the computational software does a reasonable
j o b of predicting both the buckling load and the qualitative aspects o f postbuckling
(compared both with experiment and another code) there are some discrepancies due
t o (1) uncertainties in the nominal layer material properties. (2) structural sensitivity
t o initial imperfections, and (3) the neglect of dynamic and local material delamina-
tion effects in the numerical model. Corresponding refinements are suggested for the
realistic continuation o f this type of analysis.

31. INTRODUCTION
Advanced composite materials, due to their superior strength-to-weight ratios
and stiffness tailorability, have become key ingredients in the design of modern
aerospace vehicles. However, the complex structural response associated with such
materials coupled with the intricacy of their fabrication creates harsh requirements
for numerical simulation.
Specifically, a problem that is of current interest to NASA/LaRC is the determina-
tion of the postbuckling strength of thin laminated composite shells comprising
the “skin” of stiffened air-transport fuselages [I]. These shells are required t o
maintain safe load-carrying capability substantially beyond the point at which
skin buckling (Le., wrinkling) occurs. To complicate matters, aircraft fuselages
typically feature local discontinuites, such as fasteners, stiffeners and cutouts,
which can induce high local stress gradients t h a t tend t o delaminate t h e compos-
ite material. In the presence of buckling, these local delaminations can propagate
throughout - and hence fail - a composite structure.
It is NASA’s ultimate goal to be able t o predict such phenomena analytically.
To this end, they have asked us to match some experimental data obtained for a
.--- 67
PRECEDING PAGE BLANK NOT FILMED
 representative test specimen [l].*We have completed the first phase of t h e global
analysis, which is described in t h e following sections. It employs a recently devel-
oped finite-element shell analysis capability, featuring the use of three-dimensional
constitutive equations within an efficient resultant-oriented framework, t o deter-
mine the buckling load and explore the postbuckling regime of an elastically in-tact
composite model. T h e more ambitious global/local analysis, which involves the
prediction of local material delaminat,ion and it,s interaction with global structural
response t o determine postbuckling strength, is still in the planning phase.

$2.PR,OBLEM DESCRIPTION

$2.1 Setup

T h e focal problem mentioned above is depicted in Figure 1 . Shown is a moder-

ately deep (55.6 deg) cylindrical panel with a circular cutout. T h e panel has a
14 in. square planform, a 15 in. radius of curvature a n d a radius-to-thickness ratio
of 150. T h e hole is centrally located and measures 2 in. in diameter. T h e compos-
ite shell wall consists of 16 layers of unidirectional graphite fibers in epoxy-resin.
Each layer is .0056 in. thick (for a total of .086 in.) a n d the layers are arranged
in the symmetric, quasi-isotropic *’ stacking sequence: { * 4 5 / 9 0 / 0 / 0 / 9 0 /
45)
degrees - repeated twice. T h e orthotropic-elastic material properties for each
layer are listed in the figure. Note t h a t these properties represent nominal values
a n d will require some adjustment in the sequel (see $ 5 ) .
Surrounded by a metallic test frame, the appropriate boundary conditions for the
cylindrical panel are (i) fully clamped on the bottom edge, (ii) clamped except for
axial motion on the top edge and ( i i i ) simply supported on the vertical edges.

§ 2.2 Experimental Results

T h e test conducted by NASA consisted of statically imposing a uniform end-
shortening, 6, t o determine t h e load-carrying capability of t h e panel beyond the
initial buckling load. Experimental results are shown in Figure 2.
Figure 2a represents a normalized “load versus end-shortening” curve. Note t h a t
buckling occurs abruptly, followed by a rapid drop in t h e axial load. Somewhere

* This problem was first suggested by Dr. Norman F. Knight, J r . a t NASA Langley
Research Center. We amusingly refer to it as “Knight’s problem” both as an ac-
knowledgment to the originator and as a reminder of the many pitfalls obstructing
its numerical solution.
** The term quasi-istropic refers to the fact that the resultant constitutive matrix is es-
sentially isotropic, due to a balanced sequence of fiber angles through the thickness.
However, in contrast to truly isotropic materials, there is some additional coupling
between bending and twisting deformations.

68
between the top and bottom of this vertical branch, which spans a period of rnilli-
seconds in the experiment, delamination occurs near the hole (Fig. 2b). Although
the delamination gets progressively worse and eventually distorts the hole, there
is a secondary (postbuckling) stiffening branch in the load/displacement curve.
T h e test was stopped a t the point on this secondary branch labeled bbco11apse77,
at
which point extensive delamination is evident (see [ 11 for details).

$3.COMPUTATIONAL APPROACH

$3.1 Formulation

T h e formulation of the governing equations and associated solution algorithms

used t o analyze the above problem is outlined in Figure 3. A detailed description
of this approach may be found in [a].Briefly, it employs continuurn-based (CB)
shell elements, similar to the Ahmad element 131, but extended to the nonlinear
regime and reduced t o an economical resultant form in which element stiffness and
force operators are pre-integrated through the thickness.

Shell Equations

Thickness pre-inbegration of the CB shell equations is achieved by augmenting the

standard, Mindlin-type [ 4 ] hypotheses of straight normals and zero normal stress
with two additional hypotheses, namely: small trunsverse-shear strains and mild
taper. As argued in [2],these additional hypotheses do not significantly alter the
range of applicability of the original formulation. The resulting theory, which is
expressed in terms of stress-resultants rather than pointwise (continuum) stresses,
is referred to as a continuum-bused resultant (CBR)shell formulation.
Note t h a t unlike earlier efforts to pre-integrate the CB shell equations, which typi-
cally assume a constant through-thickness variation of the surface metric (e.g., [ 5 ] ) ,
t h e present CBR formulation bypasses this assumption and hence is not restricted
to very thin shells.

Shell Elements

To spatially discretize the CBR shell equations, a variety of shell finite elements
have been implemented within the above framework. However, on the basis of the
numerical evaluation conducted in [2],only the following two shell elements were
considered for the present analysis: (i) the nine-node Heterosis (HET) element [6],
and (ii) a new nine-node assumed natural strain (ANS) shell-element [7]. While
both elements are parabolically curved (Fig. 3) and use standard isoparametric
interpolation as a starting point, each departs from the basic recipe in order to
properly represent inextensional bending deformation for thin shells.
69
 In particular, the 9 HET element selectively under-integrates all stiffness and force
terms involving membrane strains to avoid membrane “locking”, and uses a mix-
ture of Lagrange shape functions (for rotations) and Serendipity shape functions
(for translations) t o avoid “spurious modes” otherwise evoked by reduced integra-
tion.
In contrast, the 9-ANS element assumes an appropriate (inextensionally accurate)
strain field frorn the outset, using a modified set of Lagrange shape functions and
employing full numerical integration throughout. Due to the fact that the strains
are assumed in the generally non-orthogonal isoparametric coordinate basis, a n
apparent advantage of the ANS approach is its decreased sensitivity t o element
mesh distortion.

Nonlinear Solution Procedures

After performing an updated-lagrange linearization of the equilibrium equations,

a modified Newton-Raphson (NR) version of the Riks-Crisfield (RC) arc-length
control algorithm IS] is used t o trace incrementally the load-displacement curve.
T h e RC procedure was adopted as a convenient means of statically traversing the
bifurcation points that arise in shell postbuckling analysis. However, such methods
are not foolproof, and special attention by the analyst (e.g., in the selection of
imperfections, step-sizes and error tolerances) is often required in the vicinity of
closely spaced or multiple bifurcation points.
Two kinds of update procedure are required to advance the solution of the discrete
nonlinear shell equations from one load-step t o the next: (i) a kinematic update
t h a t “accumulates” incremental nodal displacement - both translational and ro-
tational components; and ( i i ) a constitutive update t h a t generates new element
stresses from the corresponding displacement field.
Only the rotational part, of the kinematic update is non-trivial.* In the present
approach, the rotation increments are used to update orthogonal triads defined at
each shell node. The triad-update algorithm involves no trigonometric functions,
maintains orthogonality at each step and provides a shell-oriented coordinate sys-
tem in which the normal rotational degree of freedom may be eliminated at all shell
nodes (except at junctures). Furthermore, once the nodal triads and reference-
surface coordinates have been so updated, the current element configuration is
completely and uniquely defined (see (21 and [9]for details).
T h e stress update is handled via an incrementally objective algorithm 1101 that

* Since the translational components of displacement are vectorial, translation incre-

ments are simply added to obtain total displacements and hence update the nodal
coordinates.

70
features a midpoint-rule numerical integration of rate-type constitutive equations.
For finite-strain analysis, the constitutive algorithm additionally involves shell
thickness updates t h a t account for large Poisson effects. These are computed (as
in [ 1 1 ) ) by recovering the normal strain increments from the constitutive equations
via the zero normal stress (ZNS) hypothesis.

Summary of Computational Features

T h e computational features of the above approach may be summarized as follows:

0 Applicable t o Both T h i n and Moderately Thick Shells

0 Rotations M a y Be Arbitrarily Large

0 Strains M a y Be Large ( Except Transverse Shears )
0 I
Resultant Format Yields Cost Savings c( Number of Layers I
0 Shell Elements Are Fairly Robust

- N o Locking
- No Spurious Mechanisms
- Low Sensitivity to Mesh Distortion (Especially ANS Elements)

Finally, note that while the formulation allows for finite strains and inelastic ma-
terial behavior, the present analysis simply employs orthotropic linear-elasticity
within each layer and does not account for the material delamination (z.e., dam-
age) observed in the N A S A experiment.

53.2 Implement,ation (NICE Software Architecture)

T h e shell-element capabilities mentioned above have been implemented in a mod-

ular fashion t o facilitate research and transferral to other finite element codes.
Some of the specific shell-element functions that are available via independent
FORTRAN77 subroutine calls are shown in Figure 4a. A complete description of
the shell-element software (including listings) is given in Appendix S of reference
PI.
T h e host finite-element program used by the authors is actually a network of in-
dependently executable programs (or processors) which are coordinated via high-
level procedures written in a mathematically oriented command-language (Fig.
4b). In such an environment, the shell-element software is embedded in a sin-
gle processor and the global solution algorithms are implemented as procedures;
examples are given in [12].

71
 The software architecture (i.e., utilities) used to construct this particular analysis
system is known as NICE (for Network of Interactive Computational E l e m e n t s ,
11 2,131). Due t o the flexibility provided by the NICE archit,ecture and its suitability
for nonlinear and coupled-field problems, it is currently being explored by N A S A
as the basis for a standard generic testbed system for Computational Structural
Mechanics (see [ 1 4 ] and other presentations therein). One of the motivating factors
for developing such a system is the implementational complexity associated with
a comprehensive global/local analysis of the present composite-shell problem.

54. FINITE-ELEMENT MODELS

Due to the physics of the problem, a full numerical model of the test specimen is
required (Fig. 5 ) . The slight anisotropy emanating from the composite material
stacking sequence is only partly responsible for the lack of available symmetry."
A s will become apparent, the nonlinear postbuckling response is inherently non-
symmetric due t,o the presence of the hole and the participation of many diverse
mode shapes.
Several combinations of shell-element type and mesh density were employed during
the course of the linear (pre-buckling), stability (buckling eigenvalue) and nonlin-
ear (postbuckling) analyses. Figure 5 shows three representative grids, involving
300, 1500 and 5000 degrees-of-freedom, respectively. These grids correspond to
16, 80 and 256 nine-node elements (or alternatively to 64, 320 and 1024 four-node
elements), respectively. Note that there is intrinsic element mesh-distortion in
these models - bot,h in-plane and out,-of-pla.ne - due to t,he focus on the hole
and the curvature of the shell. However, the elernents nearest the hole (where it
counts most) have the most regular shapes.
The coarsest grid ( 1 ) was used to verify the modeling procedure, the finest grid
(3) was used exclusively to check convergence of the linear and eigen solutions,
and the intermediate grid (2) became the workhorse for nonlinear analysis. Fur-
thermore, as little difference was observed between the 9-HET and 9-ANS elements
(53.1) during the early stages of analysis, the 9-HET element (which is slightly less
expensive) is fea.tured in the analytical results t h a t follow.
Boundary coiiditions were imposed as described in Section 2 and illustrated in
Figure 5. To simulate end-shortening, an axial f o r c e was applied in conjunction
with a degree-of-freedom equivalence among all axial displacements on t h e loaded
edge. This was done t o avoid the use of specified displacements, which tend to
complicate the adaptive, arc-length-based nonlinear solution algorithm.

* We have confirmed this via numerical experiments with an isotropic model.

72
55. LINEAR (PRE-BUCKLING) ANALYSIS

Results for the linear pre-buckling analysis are shown in Figure 6. The defor-
nia.tions d u e to an applied axial compressive load of 22480 Ibs ( 1 k N ) are shown
rnagriified by a factor of 10 in the top half of the figure ~~ for grid 2 . The corre-
sponding distribution of the axial stress resultant, N , , along the panel circumfer-
ence a t mid-span is shown in the bottom half of the figure - for both grids 2 and
3.
R,egarding the displacement solution, convergence of the axial end-shortening,
6, was achieved with grid 2. (Grid 3 yielded less than a 1% increase in end-
shortening.) However, the converged end-shortening solution, 6 = .0316, is ap-
proximately 15% lowcr than thc cxpcrimcntal valuc; as deduced from the linear
portion of the experimental load-displacement curve (Fig. 2 ) . It is presumed that
this over-estimation of the axial stiffness is due to the uncertainty in the nominal
material propert,ies, a conclusion that is reinforced in 58. Thus, to compensate
for the rriisrnatcli, the lamina principle elastic modulus ( E , ) is reduced by a cor-
responding fact,or i n the subsequent nonlinear analysis ( $ 7 ) .
From a qualitative perspective, the solution shows substantial bending deformation
in the vicinity of the hole (Fig. 6a). This suggests that geometric nonlinearity
may be important even a t relatively low load levels and diminishes the credibility
of linear response and buckling-eigenvalue analyses.
Regarding the linear stress solution, note that the compressive axial resultant,
N , , is distributed evenly along most of the panel circumference except for a very
localized region near the hole. While grid 2 was adequate for the displacement
solution, grid 3 provides much more accurate resolution of this stress concentration.
In particular, grid 2 yielded a peak stress concentration factor (SCF) of 2.8; about
14% less than the SCF obtained with grid 3.
As experimental d a t a were not available for verification of the computed stresses,
the convergence of the grid 3 stress solution was inferred by comparison with
a closed-form (asymptotic) solution due to C.R. Steele (private communication,
Stanford University, 1985). While the closed-form solution pertained t o a purely
isotropic panel, the linear finite-element stress solutions for isotropic and quasi-
isotropic panels were found to be quite similar. It is also interesting that the SCFs
for both the isotropic model (3.1) and the quasi-isotropic model (3.25) are not
very different from the classical SCF for a flat plate with a circular hole (3.0).
Finally, note that by strongly biasing the mesh towards the hole, it was possible
to obtain grid-3 accuracy with grid 2 for the local stress gradients. Such biasing,
however, was found to be unnecessary in the subsequent, globally oriented buckling
and postbuckling analyses.

73
 $6. STABILITY (BUCKLING) ANALYSIS

Figure 7 shows the first 5 buckling eigenmodes for grid 2. These results represent
perturbations about the linear pre-buckling solution described. The eigenvalues,
X 1.084,1.10ci,1.181,2.432,1.582, are the ratios of the corresponding buckling
::~

loads t o the axial load applied in the linear pre-buckling analysis. As before, grid
2 seems to provide adequate resolution, with grid 3 giving only a 2% reduction in
the first two eigenvalues, and a 4% reduction in the remaining three.
T h e following observations are important for subsequent computational purposes:
( i ) the eigenvalues are closely spaced; ( i i ) t,he eigenmodes are vastly different in
character; (iii) therc is no single form of symmetry to be exploited computation-
ally; (iv) the first, buckling mode is symmetric and bears the most resemblance
to the linear pre-buckling solution; (iv) the second and third modes possess skew
symmetries; (v) the fourth and fifth modes are symmetric; the latter mode fea-
turing pra.ctically no distortion of the circular hole; and (vi) higher modes (not
shown) look much like those for a cylindrical panel without a hole, though the
values remain closely spaced.
Finally, it, was f o u n d that the first (z.e., critical) buckling load is approximately
25% lower than t h a t of an identical cylindrical panel without a hole. Hence, while
t h e influence of the hole on the buckling loud is only moderate (i.e., relative to the
stress concentration fact>or),its influence on the buckling modes is profound. A s
we shall see, the hole has an even stronger influence on the postbuckling response.

$7. NONLINEAR (POSTB‘IJCKLINC,)ANALYSIS

In practice), more than just a linearly converged model and an adaptive solution
strategy were necessary to obtain a reasonable nonlinear solution. One additional
pre-requisite was a 15% reduction in the nominal elastic modulus, El (from 19.6 x
I O G to 17.1 x lo”), to match the linear branch of the load versus end-shortening
curve (as explained in $5). Another important ingredient was the specification of
initial imperfections. In this regard, three versions of the analysis were run: (1)
one involving no imperfections; (2) one with an imperfection amplitude of 1% of
t,he thickness applied to each of t h e first 4 buckling modes (see Fig. 7); and (3)
one with an imperfection arnp1it)udeof 10% of the thickness applied to each of the
first 4 buckling modes. A discussion of the results for these three cases follows.

07.1 No Imperfect ions

One would think that the out-of-plane “imperfections” introduced by the linear
pre-buckling solution (Fig. 6 ) would be sufficient to trigger a realistic buckling
response. However, this was not the case. With no imperfections, the computed
solution path resembled the experiment only up to the descending part of the

74
load-displacement curve (Fig. sa). The computed curve then rolled back onto
itself with t h e stiffening branch of the postbuckling curve practically aligned with
t h e pre-buckling curve.
To gain further insight, it is useful to look at the deformation and stress histories
portrayed in Figure 8b. Shown is a sequence of computational “snapshots” taken
a t various points (i.e., load steps) on the nonlinear load-displacement curve.* The
variation of the axial stress resultant along the mid-span circumference is plotted
below each frame.
Note t h a t the deformation starts out (at load step 10) much like the first linear
buckling mode (Fig. 7 ) , with inward dimples both fore and aft of the hole, then
articulates through the second and third modes during the initial postbuckling
phase (steps 15-20). This rotation of the two dimples about the hole is probably
triggered by the bending/twisting coupling inherent in the composite stacking
sequence. T h e dimples continue t o rotate and broaden until, a t step 40, the
pattern begins t o resemble the fifth linear buckling mode. Evidently, it is this
“locking” into mode 5 that is responsible for the excessive secondary stiffening
in the load-displacement curve (Fig. Sa). Clearly, mode 5 is an unrealistically
s t i f l one, resembling what might occur if a ring sliflener had been placed around
t h e hole. This is also evident in the axial stress distribution, where the stress
concentration has practically been shed by step 40.

57.2 Large Imperfections

To avoid the unrealistic mode-5 locking observed in the preceding analysis, a fairly
large initial imperfection was introduced in the numerical model. This was accom-
plished by adding an equal measure of each of the first 4 buckling modes t o the
initial geometry, such that the maximum radial (i.e., shell-normal) displacement
in each mode was equal to 10% of the panel thickness. Thus, the magnitude of
t h e combined radial imperfection approached 40% of the thickness at some points.
It is emphasized that this rather arbitrary choice of imperfections was designed
primarily t o minimize the influence of mode 5.
T h e computed nonlinear response for the imperfect panel is shown in Figure 9.
Note t h a t the secondary stiffening branch bears more resemblance to the exper-
iment than did the imperfection-less analysis. Unfortunately, it is also true t h a t
t h e buckling (i.e., peak) load has dropped by about 15%, and now underestimates
t h e experimental buckling load by more than 20%. A heuristic explanation is

* Note that the pre-buckling phase of the analysis appears linear with respect to the
axial displacement, 6. However, due to the rapid growth in radial displacements
(not shown), the analysis is actually quite nonlinear from the outset; which explains
the relatively large number of load-steps required on the ‘‘linear” branch of Fig. 8 .

75
 provided by considering the deformation and stress histories.
In Figure 9b, we see that with the lO%-h imperfections, the buckling pattern leaves
mode 1 almost immediately and develops an intensified inward dimple on one side
of the panel. Consequently, the full axial load is re-distributed to the other side
of the panel (see N , plots in Fig. 9 b ) , accounting for the reduction in both the
buckling load and the postbuckling stiffness (fig. s a ) .
57.3 Small Imperfections (the &BottomLine”)
Finally, the “best of both worlds” was obtained with a 1%-h imperfection in each
of the first 4 modes. Compared with the previous analysis, the computed load-
displacement curve (Fig. 10) shows both a n increase in the maximum load and
a decrease in the minimum load, thus bringing the solution more in line with
experiment.
The improved performance, obtained by reducing the imperfections, may again
be related t o the deformation history (Fig. lob). Here, as in the case without
imperfections, two inward dimples develop and proceed to rotate about the hole.
Just after buckling, however, one dimple tends t o deepen while the other dimin-
ishes, and eventually there is a double snap-through. This accounts for the double
dip in t h e load-displacement curve (Fig. l o a ) and seems to explain why a lower
minimum load is obtained with the smaller imperfection.
Still, there are some serious descrepancies between analysis and experiment,
namely: (i) a 7% under-estimation of the buckling load, (ii) a 25% over-estimation
of the minimum load, (iii) a 30% under-estimation of the postbuckling end-
shortening, and (iv) a 10% over-estimation in the postbuckling stiffness. These
will be addressed in $9.

$8. CORROBORATION WITH ANOTHER CODE

To support the above results, obtained via the computational procedures described
in Section 3, parallel analyses were performed with another finite-element com-
puter code. For this purpose, we employed the STAGS code [15],which has been
used for more than a decade by varous government agencies and industrial firms
(e.g., NASA and Lockheed) to analyze difficult nonlinear shell problems. Another
reason for using STAGS is t h a t it features finite-element computational proce-
dures t h a t are substantially different from those used in the present approach,
thus adding strength to an analytical comparison.
For the linear (pre-buckling and buckling) analysis, excellent agreement was ob-
tained between the STAGS and NICE-based solutions. For completeness, the
STAGS runs were performed with two radically different shell-element types, which
both converged t o the same solution as the NICE/g-HET element, albeit at slower

76
rates and from below in stiffness. (Note t h a t while the axial pre-buckling stiffness
of t h e STAGS elements converged from below, the buckling loads converged from
above).
In particular, t h e comparison included: (i) the commonly used STAGS/410 el-
ement ~~a flat quadrilateral plate element based on Kirchhoff-Love theory and
cubic membrane/bending displacement interpolation; a n d (ii) the less frequently
used STAGS/422 element - a quadrilateral composed of two Kirchhoff-based tri-
angular plate sub-elements with cubic bending interpolation a n d quadratic mem-
brane interpolation. It is believed t h a t the relatively slow convergence “from
below” of t h e STAGS/410 element is due t o warping sensitivity, while t h a t of the
STAGS/422 element is probably due to the incornputibility between membrane
a n d bending displacement fields for non-flat quadrilateral element shapes.*
For the nonlinear comparisons with NICE/S-HET, the STAGS/410 element was
used exclusively. T h e resulting load-displacement curves (for 1%-h and 10%-h
imperfections) a r e summarized in Figure 1 1 . T h e dashed curves represent the
NICE/9_HET solutions and the dotted curves represent the STAGS/410 solutions;
both were obtained with Grid 2 (Fig. 5 ) .
Note t h a t although the STAGS and NICE-based solutions use different finite-
element types, large-rotation update procedures and nonlinear solution strategies,
t h e correlation is remarkable especially during the postbuckling phase. Even
t h e STAGS zero-imperfection analysis (not shown) resulted in the same exces-
sive postbuckling stiffness as displayed in Figure 8a. One other point: While
t h e STAGS/410 element consistently shows about a 5% higher buckling load
t h a n t h e NICE/9_HET element, thus coming closer t o the experimental peak;
STAGS/410 is actually less accurate - with respect t o discretization errors
- t h a n NICE/g-HET. This follows from t h c fact t h a t both STAGS/410 and
NICE/g-HET converge from above in t h e buckling load. This was confirmed by
running t h e NlCE/9_HET element with a coarser grid, for which it too showed a
5% higher peak. That non-converged solut,ions compare better with experiment
t h a n converged ones suggests t h a t spatial discretization is not the only source of
error here (e.g., see $9).

* It is interesting to note that the STAGS/422 element was used in the related study
conducted in [ I ] ,where it yielded a 17% more flexible linear solution, and thus agreed
better with the linear portion of the experiment. Nevertheless, it has since been
found that the boundary conditions were not consistently applied to the element’s
mid-side freedoms in that analysis. By correcting this implementation error, the
17% discrepancy with the other elements has been completely eliminated. Thus, it
appears that the “accuracy” obtained in [ l ]with the STAGS/422 element is due to
compensating errors in the nominal material properties.

77
 $9. CONCLUSIONS

59.1 Summary

The present study may be summarized as follows:

0 PURPOSE:
- Validate continuum-based resultant (CBR) shell formulation
- Evaluate new shell elements
- Gain experience with composite-shell postbuckling analysis
0 RESULTS:
- Good agreement in pre-buckling/buckling range
- Good qualitutive agreement in postbuckling range
- Discrepancies due to:
-_ Material properties
-_ Imperfection sensitivity
__ Dynamic effects
-_ Delamination

T h e “good” agreemerit obtained between the present shell-element formulation

and experiment in the pre-buckling and buckling range was possible only after
adjusting the nominal material properties so that the linear axial stiffnesses coa-
lesced. T h e material-property modification was further justified via corroboration
with the STAGS finite-element code, which features a substantially different com-
putational approach.
The “best” solution for the nonlinear response was obtained by introducing small
( 1 %-thickness) imperfections corresponding to each of the first four buckling
modes. T h e computed load-displacement curve (Fig. loa), which again compared
well with STAGS (Fig. l l ) , still showed major discrepancies with the experi-
ment: T h e discrepancy in the buckling load (which is relatively small) may be
due to the inadequacy of adjusting only the principal layer elastic modulus, E l ,
rather than the complete set of orthotropic rriaterial constants. The discrepancy in
the unloading phase of postbuckling (i. e., the computed end-shortening reversal)
is attributed to the quasi-static approximation of what is, in reality, a dynamic
phenomenon. Finally, the discrepancy in the stiflening phase of postbuckling is
clearly dominated by damage, i e . , delamination observed in the experiment but
not represented in t h e model.

78
59.2 Recommendations

T h e goal is t o eliminate the discrepancies listed with a minimum of computational

cost and complexity. To this end, the following steps are recommended:

1 ) PERFORM ADDITIONAL EXPERIMENTS. First, more experimental d a t a

are required to verify existing computational capabilities for composite shell post-
buckling. For example, panel imperfections should be carefully measured and
selected strains and overall deformation patterns should be monitored a t frequent
intervals. Additionally, a n isotropic panel should be tested in order to eliminate
t h e material identification problem and also to provide a standard benchmark for
shell-element evaluation. The isotropic problem would be valuable for screening
out geometrically sensitive elements.

2) REFINE NONLINEAR GRID. The nonlinear analysis should be repeated

with a finer grid (e.g., Grid 3 ) , as convergence in the linear regime is no guarantee
of convergence in the nonlinear regime. Moreover, a study of modal participa-
tion both in the imperfections and in the nonlinear response may help establish
modeling guidelines for future analysis.

3) INCLUDE D Y N A M I C EFFECTS. Dynamic effects, which are relatively

straightforward to incorporate, should be assessed a t the first opportunity. The
analysis could be started in a quasi-static mode, switched to an explicit transient
response algorithm during the unstable phase, and switched again to an implicit
algorithm during the stable postbuckling phase.

4 ) R E P R E S E N T LOCAL FAILURE: (DELAMINATION). To account for the

composite delamination mechanism, appropriate failure criteria and progressive-
failure modeling procedures need to be developed, implemented and evaluated.
For shell-based failure criteria, accurate stress recovery is essential. Improved
stress resolution is required both along the surface (e.g., via stress/displacement
iterat,ions and adapt,ive refinement,) and thrOiigh-th,e-thl.ckncs.s, as both normal and
transverse-shear stresses can play a dominant role in delamination. Progressive
failure may then be simulatcd by methods ranging from a simplified shell model
that selectively degrades luyer yroperlies, to a full 3l) uridysis riear t h e hole with ail
evolving 2D/3D transition. An intermediate approach is to “split” shell elements
along delaminating boundaries. (See Fig. 12.) The simplest approach, however,
has obvious implementational advantages.

5) DEVELOP EFFICIENT GLOBAL/LOCAL ALGORITHMS. Finally, there is

a need t o reduce the cost of nonlinear analysis for such problems. The cost of the
global analysis is dominated by the large number of iterations/steps in the linear-
to-postbuckling transition regime. Possible approaches include the reduced-basis

79
 technique 116,171, Thurston’s method 1181, a n d improved extrapolators. For the
combined global/local problem, where material “properties’) a r e changing rapidly
d u r i n g postbuckling. additional feat,ures such as line-searches [ 191, quasi-Newton
stiffness u p d a t e s (201 a n d nonlinear s u b s t r u c t u r i n g may greatly improve effi-
c ienc y.
We a r c presently acting on rccomrnendations (2)-(3) a n d will report t h e outcome
in a forthcoming paper.

ACKNOWLEDGMENTS
The preparation of this paper was supported by the Independent Research Program of
Lockheed Missiles and Space Company, lnc. Ongoing research is being supported by
NASA/Langley Research Center as part of their newly launched program in Computa-
tional Structural Mechanics.

REFERENCES

Knight, N.F, J r . and Starnes, J . H . , J r . (1984), “Postbuckling Behavior of Axially

Compressed Graphitc-Epoxy Cylindrical I’ancls with Circular Holes”, prescnted a t
t,hc 1984 A S M E J o i n t Pressrirc Vcsscls and Piping / Applied Mechanics Converence,
San Antonio, Texas.

Stanley, G.M. (1985), Continuum-Based Shell Elements, Stanford University Ph.D.

thesis, also Lockheed report’ LMSC-F035839.

Ahrriad, S., Irons, B.M. and Zienkiewiecz, O.C. (1970), “Analysis of Thick and Thin
Shell St,ruct,ures by Curved Finit,e Elerrients” , I n t e r n a t i o n a l J o u r n a l for N u m e r i c a l
Melhods in Enyineer1rig, 2 , 419-451.

Mindlin, R.D. (1951), "influence of Rotat,ory lnertia and Shear on Flexural Motions
of Isotropic, Elastic Plates”, J o u r n a l of Applied M e c h a n i c s , 18, 31-38.

Parisch, H . (1981), “Nonlinear Analysis of Shells [Jsing lsoparametric Elements”,

pp. 47- 6 3 iri Nonlinear Finite Element Analysis of Plates a n d Shells, (eds. T.J.R.
Hughes et al.), ASME, N e w York.

Hughes, T . J . R . , and Cohen! M. (1979), “The ‘Heterosis’ Finite Element for Plate
Bendirig” , C o m p u t e r s a n d S t r u c t u r e s , 9, 445-450.

Park, K.C. and Stanley, G.M. (1986)) “A Curved Co Shell Element based on As-
sumed Natural-Coordinate Strains”, in J o u r n a l of Applied ,l/lechanics, June
1986.

Crisfield, M.A. (1983), “A Fast Incremental/lterative Solution Procedure that Han-

dles Snap-Through” , C o m p u t e r s a n d S t r u c t u r e s , 13, 55-62.

Rankin, C.C. and Brogan, F.A. (1984), “An Element-Independent Corotational

Procedure for the Treatment of Large Rotations”, pp. 85-100 in Collapse Analysis
of Structures (eds. L.H. Sobel and K . Thomas), ASME, New York.

80
[ l o ] Hughes, T . J . R . and Winget, 3 . (1980), “Finite Rotation Effects in Numerical ln-
tegration of Rate Constitutive Equations Arising in Large-Deformation Analysis”,
International J o u r n a l f o r N u m e r i c a l M e t h o d s in Engineering, 15, 1862-1867.

[ I l ] Hughes, T.J.R. and Carnoy, E. (1981)) “Nonlinear Finite Element Shell Formulation
Accounting for Large Membrane Strains” , pp. 193-208 i n Nonlinear Finite ElerneriL
Analysis of Plates and Shells (eds. T . J . R . Hughes el al.), ASME, New York.

1121 Felippa, C.A. and Stanley, G.M. (1985), “NICE: A Utility Architecture for Com-
putational Mechanics”, proceedings of US-Europe Symposium on Finite-Element
Methods for Nonlinear Problems, Univ. of Trondheim, Norway (ed. by P. Bergan).

1131 Felippa, C.A. (1981), “Architecture of a Distributed Analysis Network for Compu-
tational Mechanics”, C o m p u t e r s and Structures, 13, 405-413.

[ 141 Geers, T.L. (1984)) “Comments on Proposed Research Program in Computational

Structural Mcchanics” proceedings of Workshop on Computational Structural Me-
)

chanics, held 5-7 Sep 1984 a t NASA/LaRC, edited by W.J. Stroud.

1151 Almroth, B.O., Brogan, F.A. and Stanley, G.M. (1979), “Structural Analysis of
General Slrells, Vol. IT: User Instructions for the STAGS(C-I) Computer Code”,
Lockheed Report LMSC-D33873.

116) Noor, A.K. and Peters, J.M. (1983), “Recent Advances in Reduction Methods for
Instability Analysis of Structures”, C o m p u t e r s a n d Structures, 16, 67-80.

1171 Stehlin, P. and Brogan, F . A . (1984), “Analysis of Stjructjural Collapse by the Re-
duced Basis Technique”, pp. 69-84 in Collapse Analysis of Structures (eds. L . H .
Sobel and I(.Thomas), ASME, New York.

(181 Thurston, G.A., Brogan, F . A . and Stehlin, P. (1985), “Postbuckling Analysis Using
a General Purpose Code”, presented at AIAA 26th Structures, Structural Dynamics
and Materials Conference) Orlando, Florida, AIAA Paper N o . 85-07 19-CP.

1191 Crisfield, M.A. (1982), “An Arc-Length Method Including Line Searches and Ac-
celeration”, International Journal for N u m e r i c a l Methods in Engineering, 19, 1269-
1289.

120) Dennis, J .E. and More, J . J . (1977)) “Quasi-Newton Methods, Motivation and The-
ory”, SIA M R e v i e w , 19, 46-89.

81
 COMPOSITE CYLINDRICAL PANEL WITH CIRCULAR HOLE

R = 15 in.
Metallic
Test
Fixture
7 NOMINAL COMPOS1TE
-MATERIAL
-. PROPERTIES:
I

v 16 Layers

C ra ph i te- Epoxy
hlayer = 0 . 0 0 5 6 in.

=
1 14 in.
htotal = 0.09 in.

E
1
6
= 19.6 x 10 psi

E 2 = 1.89 x 106 psi

vz1 = 0.38, v 1 2 = 0.037

I C t Z = 0.93 x 106 psi
c
{t45/90/0/0/90r45}osyrn

....
------I
Figure 1. Knight’s problem, NASA test specimen.

0 2 4 6 x lo-=

END-SHORTEN ING, d / L

Figure 2. Experimental results.

82
0 Continuum-based ( C B ) shell equations
- 3-D C o n t i n u u m equations ( e q u i l i b r i u m /con s t itut ive)
- Embed sheil hypotheses ( s t r a i g h t normals, zero normal stress)

0 Reduce t o " r e s u l t a n t " ( C B R ) form

-

-
Assume: small transverse-shear
Preintegrate through-the thickness
strains, mild taper
<\
a__---
-1

0 Discretize v i a c u r v e d "isoparametrlc" elements

- Selective/reduced integration (SRI)
- Assumed n a t u r a l s t r a i n ( A N S I

I
1

0 Solve nonlinear m a t r i x equations v i a :

- Linearization w.r.t. c u r r r e n t configuration (UL)
- Modified NR a l g o r i t h m w i t h a d a p t i v e (RC) s t r a t e g y
- Nodal t r i a d updates f o r l a r g e r o t a t i o n s

0 Solve r a t e c o n s t i t u t i v e equations v i a :
- "Midpoint r u l e " incremental a l g o r i t h m
- ZNS r e c o v e r y of normal s t r a i n s ( t h i c k n e s s updates)
I

Figure 3. Computational approach.

USER

I I I I
...
GLOBAL D A T A B A S E

a. NICE (Network of Interactive Computational Elements)

Figure 4. Implementation.

83
 I x 16 NODES 300 DOF

a. Grid I IColrrcI

I.*

C. Grid 3 (Refined)

Figure 5. Finite-element models.

a. Linear Displacement Solution

(Magnified by 10)

01

Figure 6. Linear (prebuckling) analysis.

84
 /BUCKLING MODE 41

BUCKLING MODE 2 1

[BUCKLING MODE 5 1

Figure 7. Stability (buckling eigenvalue) analysis.

3
STEP 10 STEP 30
x10-

2.
h
P
X ' I
STEP 35
0
<
s 1.

0.
/ ,
I
1.0 2.0
I I
3.0 4
STEP 20

END-SHORTEN1 N C , 6/L x1~-3

- EXPERIMENTAL
-...--
0
F I N IT€-ELEMENT
LOAD-STEP NUMBER Nx 1

a. Load versus End-Shortening b. Deformations and Stresses

Figure 8. Nonlinear (postbuckling) analysis; no imperfections.

85
 1o -~

LOAD

2.0 -
.
5
n. F.E.
d
4
s 1.0 -

EXPERIMENTAL
+-t* FINITE-ELEMENT
0 LOAD-STEP NUMBER

0.0 I I 1 I
0.0 1.0 2.0 3.0 4.0 x 1 0 - 3
END-SHORTENING, d/L

a. Load Versus End Shortening Nx 1 &-+--I

b. Deformations and Stresses

Figure 9. Nonlinear (postbuckling) ana lysis; 10%-h imperfections.

3.0 STEP 1 0 STEP 3 0

x10 -

.INEAR BUCKLING LOAD

2.0
a
wa
DAMAGE STEP 40
0
a
s
1.0

STEP 2 0

0.0

a. Load versus End-Shortening b. Deformations and Stresses

Figure 10. Nonlinear (postbuckling) analysis; 1%-h imperfections.

86
 3.0

2.0
4
\
n
d
<
0
J 1.0
EXPERIMENT EXPERIMENT
--- STAGS / 41 0 ELEMENT --- S T A G S / 4 l O ELEMENT
--- NICE/S-HET ELEMENT --- N I C E / g - H E T ELEMENT

0.0
/ I 1 I I
0.0 1.0 2.0 3.0 4.0 5.
END SHORTENING, 6/L

a. 10%-h Imperfections

Figure 11. Corroboration with the STAGS code, postbuckline analyses.

G = G,
=- G2

SHELL ELEMENT SHELL ELEMENT

a. Simple: Shell-Element Material-Property Degradation

SHELL ELEMENTS SHELL ELEMENT

b. Intermediate: Shell-Element "Splitting"

SOLID ELEMENTS t SHELL ELEMENT

CONSTRAINED
INTERFACE

C. Complex: 3 0 / 2 0 Transition

Figure 12. Composite failure-progression models.

87
 N89-24643 *

A REVIEW OF SOME PROBLEMS I N GLOBAL-LOCAL

STRESS ANALYSIS

R i c h a r d B. Nelson
UCLA
L o s Angeles, C a l i f o r n i a

INTRODUCTION

The c o n t i n u a l l y i n c r e a s i n g power and economy of computers provides t h e

s t r u c t u r a l e n g i n e e r i n g and mechanics community w i t h a n o p p o r t u n i t y and c h a l -
l e n g e t o make major a d v a n c e s i n computer i n t e n s i v e areas of a n a l y s i s , d e s i g n
and n o n d e s t r u c t i v e e v a l u a t i o n of complex s t r u c t u r a l s y s t e m s . C e r t a i n l y t h e
a v a i l a b i l i t y of modern computers i s making i t possible t o c o n s i d e r i n c r e a s -
i n g l y l a r g e r and more complex s t r u c t u r a l a n a l y s e s . S t a t e - o f - t h e - a r t commer-
c i a l g r a d e s o f t w a r e i s g e n e r a l l y a v a i l a b l e t o u s e f o r a n a l y z i n g a v a r i e t y of
l i n e a r and n o n l i n e a r problems on l a r g e mini o r mainframe computers. A t t h e
same t i m e , s t r u c t u r a l a n a l y s i s programs are b e i n g "down s i z e d " for u s e on per-
s o n a l computers.

Y e t , g i v e n a l l of t h e s e a d v a n c e s , some i m p o r t a n t and p e r h a p s even c r i t i -

c a l problems are d e v e l o p i n g which must be r e s o l v e d if t h e r e m a r k a b l e improve-
ments i n computer-based a n a l y s i s and d e s i g n over t h e past 10 y e a r s are t o
c o n t i n u e and t h e s t r u c t u r a l e n g i n e e r i n g community is t o t a k e f u l l a d v a n t a g e of
t h e new computing power. S e v e r a l problems s h o u l d be b r i e f l y mentioned.

F i r s t , modern s t r u c t u r a l a n a l y s i s s o f t w a r e is g e n e r a l l y a p r o p r i e t a r y
p r o d u c t of a n a c t i v e and v e r y c o m p e t i t i v e commercial s o f t w a r e i n d u s t r y . A s
s u c h t h e software is beyond t h e c o n t r o l o f t h e e n g i n e e r s who are almost com-
p l e t e l y d e p e n d e n t on i t €or p e r f o r m i n g s t r u c t u r a l a n a l y s i s and d e s i g n . m u s ,
t h e y do n o t d i r e c t l y c o n t r o l t h e computer a n a l y s i s and t h e r e f o r e are n o t able
t o f u l l y u n d e r s t a n d t h e r e s u l t s t h e y o b t a i n . A s a consequence e n g i n e e r s must
r e l y on faith and e a r l i e r e x p e r i e n c e w i t h g i v e n s o f t w a r e t o j u s t i f y t h e i r
a n a l y s e s and s u b s e q u e n t d e s i g n s .

Second, t h e software packages a v a i l a b l e t o t h e e n g i n e e r i n g community are

enormously complex, so t h a t even i f s o u r c e l i s t i n g of t h e programs were a v a i l -
a b l e , f e w e n g i n e e r s would be able, much less w i l l i n g , t o l e a r n how t h e program
works. Thus, t h e s h e a r s i z e and c o m p l e x i t y of t h e software e n c o u r a g e s b o t h
t h e u s e r and t h e s o f t w a r e vendor t o l e t w e l l enough a l o n e . Incremental
changes i n s o f t w a r e are s u r p r i s i n g l y d i f f i c u l t and most software r e q u i r e s
almost c o n t i n u o u s s u p p o r t by a t e c h n i c a l s t a f f .

A s a r e s u l t l i t t l e i n c e n t i v e e x i s t s f o r e i t h e r e n g i n e e r s or software
f i r m s t o push f o r major m o d i f i c a t i o n s i n e x i s t i n g s o f t w a r e or t o d e v e l o p fun-
d a m e n t a l l y new and more p o w e r f u l software. U n f o r t u n a t e l y , i n a d d i t i o n t o
b e i n g hampered by s i z e and c o m p l e x i t y , t o d a y ' s software is a p r o d u c t d e v e l o p e d
f o r y e s t e r d a y ' s computers. For example, most programs are w r i t t e n u s i n g l o g i c
d e s i g n e d around one-dimensional a r r a y s t o store compacted s t i f f n e s s e s and
column s o l u t i o n t e c h n i q u e s t o solve e q u a t i o n s i n o r d e r t o minimize s t o r a g e
requirements. However, modern computers have almost u n l i m i t e d ( v i r t u a l )

PRECEDING PAGE BLANK NOT FlLMED

 memory and a r r a y processors which would be more e f f e c t i v e l y used i f t h e s o f t -
ware had a d i f f e r e n t program a r c h i t e c t u r e . Thus, much of t o d a y ' s s t r u c t u r a l
a n a l y s i s s o f t w a r e is u n a b l e t o t a k e E u l l a d v a n t a g e of t h e most advanced com-
p u t i n g machinery. A major r e v i s i o n of a n e x i s t i n g program i s a tremendous
u n d e r t a k i n g and, i f d e v e l o p e d , may have s i g n i f i c a n t short-term costs t o b o t h
t h e s o f t w a r e d e v e l o p e r and t h e u s e r .

F i n a l l y , it s h o u l d be r e c o g n i z e d t h a t a number of " r e s e a r c h " areas i n

s t r u c t u r a l mechanics have r e a c h e d t h e development stage and are s t a r t i n g t o be
used by t h e s t r u c t u r a l e n g i n e e r i n g p r o f e s s i o n . Examples are s t r u c t u r a l l i f e
p r e d i c t i o n s u s i n g f r a c t u r e mechanics, s t r u c t u r a l system i d e n t i f i c a t i o n , s t r u c -
t u r a l d e s i g n o p t i m i z a t i o n , semi-automated, a d a p t i v e computer modeling, and t h e
a n a l y s i s of i n f i n i t e domain problems. The development of t h e s e areas has
been and w i l l c o n t i n u e t o be slowed u n t i l t h e t i m e when t h e n e c e s s a r y a n a l y s i s
t e c h n i q u e s become g e n e r a l l y a v a i l a b l e i n commercial g r a d e programs.

I f new g e n e r a t i o n s o f t w a r e i s t o have maximum impact on t h e s t r u c t u r a l

e n g i n e e r i n g p r o f e s s i o n i t must be c o n f i g u r e d so t h a t basic a n a l y s i s c a n be
u s e d as a t o o l i n a more g e n e r a l comprehensive e n g i n e e r i n g program. The s o f t -
ware a l s o s h o u l d be as adaptable as p o s s i b l e t o t h e e v o l u t i o n of computing
machinery.

I t i s t h e p u r p o s e of t h i s p r e s e n t a t i o n t o provide some areas of s t r u c -

t u r a l e n g i n e e r i n g which a r e n o t w e l l s e r v e d by t o d a y ' s software and which
s h o u l d be g i v e n s e r i o u s a t t e n t i o n by d e v e l o p e r s of f u t u r e s t r u c t u r a l a n a l y s i s
programs. I n k e e p i n g w i t h t h e theme of t h i s workshop s e s s i o n , several aspects
o f g l o h a l - l o c a l stress a n a l y s i s w i l l be d i s c u s s e d , w i t h a t t e n t i o n drawn t o
b o t h t h e n a t u r e of t h e problem and t h e t y p e of c o m p u t a t i o n a l s o f t w a r e which
s h o u l d be d e v e l o p e d t o i n v e s t i g a t e t h e problem.

BASELINE MODELING CONCEPTS

P e r h a p s t h e most d i f f i c u l t d e c i s i o n the e n g i n e e r a n a l y s t must make i s t h e

c h o i c e of t h e p r o p e r m a t h e m a t i c a l model f o r u s e i n i n v e s t i g a t i n g s t r u c t u r a l
b e h a v i o r ( f i g . 1 ) . G e n e r a l l y t h e a n a l y s t inust choose f i r s t t h e d i m e n s i o n a l
l e v e l of t h e model, e.g., l-D ( t r u s s / b e a m ) t e c h n i c a l t h e o r y , 2-D t e c h n i c a l
t h e o r y , 2-D continuum modeling, and 3-D continuum modeling or a model composed
of a m i x t u r e of some or a l l o f t h e above. A major c o n s i d e r a t i o n i n t h i s
c h o i c e i s t h e a v a i l a b l e computer e l e m e n t l i b r a r y (which, f o r most programs i s
r e a s o n a b l y complete) and t h e performance of g i v e n f i n i t e e l e m e n t s . A l s o ,
elements and/or procedures f o r i n t e r f a c i n g d i f f e r e n t t y p e s of element t y p e s
a r e v e r y i m p o r t a n t f o r problems where mixed e l e m e n t t y p e s a r e t o be u s e d .

The a n a l y s t must also s p e c i f y t h e l e v e l of p h y s i c s t o be treated by t h e

computer program, e s p e c i a l l y t h e material c o n s t i t u e n c y , t h e e f f e c t of i n i t i a l
stress, l a r g e motion ( s t a b i l i t y ) , dynamical r e s p o n s e and n o n l i n e a r e f f e c t s .
The a n a l y s t s h o u l d be f r e e t o i n v e s t i g a t e a l l t h e s e e f f e c t s i f n e c e s s a r y and
n o t be c o n s t r a i n e d by l i m i t s on t h e f e a t u r e s of a g i v e n software package
(e.g., l a c k of a g e o m e t r i c s t i f f n e s s m a t r i x f o r a 3-D continuum f i n i t e ele-
m e n t ) . As may be shown w i t h a number of s y s t e m s , s t r u c t u r a l b e h a v i o r may
i n i t i a l l y be l i n e a r , e l a s t i c and even q u a s i s t a t i c i n n a t u r e under g i v e n l o a d s ,
b u t g i v e n " s n a l l " changes i n c o n f i g u r a t i o n , s u c h as a h o l e or n o t c h i n a
c r i t i c a l s e c t i o n , t h e problem may be f u n d a m e n t a l l y d i f f e r e n t i n v o l v i n g

90
n o n l i n e a r e l a s t o p l a s t i c r e s p o n s e , large d e f o r m a t i o n and l e a d i n g t o large-scale
failure.

Thus t o i n v e s t i g a t e a s y s t e m t h o r o u g h l y t h e a n a l y s t needs t o have access

t o a v e r y complete a n a l y s i s package i n v o l v i n g comprehensive p h y s i c s and numer-
i c a l modeling.

The e f f e c t i v e use o f modern f i n i t e - e l e m e n t s o f t w a r e depends on t h e basic

s k i l l of t h e a n a l y s t / e n g i n e e r , who may be tempted t o replace i n s i g h t and basic
knowledge of s t r u c t u r a l a n a l y s i s w i t h d e g r e e s of freedom (DOF'S). Such a n a l -
y s e s f i r s t of a l l cost more t h a n i s n e c e s s a r y . To make matters worse large
DOF models w i t h e x c e l l e n t g r a p h i c a l c h a r a c t e r i s t i c s may c o n t a i n l o w r e s o l u t i o n
physics. (However, t h r e e d i m e n s i o n a l models may be much easier t o present t o
c l i e n t s or company e x e c u t i v e s ! )

C o n s i d e r t h e c a n t i l e v e r beam shown i n f i g u r e 2. The simple 2 DOF model

gives e x c e l l e n t beam d e f l e c t i o n s from which t h e t r a i n e d e n g i n e e r c a n o b t a i n a
w e a l t h o f a c c u r a t e stress d a t a . The p l a n a r 96 DOF model i s i n many respects
less accurate. For example, t h e ( e x a c t ) c u b i c l a t e r a l d i s p l a c e m e n t i s o n l y
a p p r o x i m a t e l y modeled by a series of parabolas. The stresses n e a r t h e t i p and
n e a r t h e l e f t end are c a p t u r e d w i t h l i t t l e , i f any, a d d i t i o n a l p r e c i s i o n . The
192 DOF p l a n a r model i s , i f a n y t h i n g , l e s s precise t h a n t h e 96 DOF model s i n c e
lower p r e c i s i o n 4 node e l e m e n t s are b e i n g used. F i n a l l y , t h e complex 3-D
model w i t h 768 DOF i s s t i l l less a c c u r a t e s i n c e t h e p l a n e stress a s s u m p t i o n i s
no l o n g e r b u i l t i n t o t h e model and now must be o b t a i n e d ( a p p r o x i m a t e l y )
t h r o u g h t h e s o l u t i o n of a large s y s t e m of e q u a t i o n s .

T h i s example shows t h e e f f i c i e n c y t h a t c a n be a c h i e v e d u s i n g a w e l l -
t h o u g h t - o u t f i n i t e - e l e m e n t model b u t i t a l s o shows some d i f f i c u l t i e s t h a t c a n
r e s u l t . The d e c e p t i v e ease w i t h which s o f t w a r e can be u s e d i n v i t e s abuse by
unwary or p o o r l y t r a i n e d u s e r s . I t i s t h e a u t h o r ' s o p i n i o n t h a t some
( o p t i o n a l ) d i a g n o s t i c s s h o u l d be a v a i l a b l e t o warn or g u i d e u s e r s i n t h e
g e n e r a t i o n of f i n i t e - e l e m e n t models.

o n e o f t h e most common u s e s of f i n i t e - e l e m e n t a n a l y s i s is t o i n v e s t i g a t e
t h e b e h a v i o r of s t r u c t u r e s i n t h e v i c i n i t y of sudden v a r i a t i o n s i n s t r u c t u r a l
geometry or c o n f i g u r a t i o n i n which s i n q u l a r i t i e s i n t h e stress f i e l d or stress
c o n c e n t r a t i o n s may o c c u r . This area o f a n a l y s i s w i l l be termed n e a r - f i e l d
modeling ( f i g . 3 ) . Two d i f f e r e n t a p p r o a c h e s for i n v e s t i g a t i n g t h e b e h a v i o r of
s t r u c t u r e s i n t h e v i c i n i t y of stress c o n c e n t r a t i o n s r e f l e c t basic c o n c e p t s
g l o b a l - l o c a l stress a n a l y s i s . I n t h e f i r s t a p p r o a c h , t h e f i n i t e - e l e m e n t model
of the g e n e r a l s t r u c t u r e away from the area of stress c o n c e n t r a t i o n ( a h o l e i n
f i g . 3 ) i s c o u p l e d w i t h a n a n a l y t i c a l s o l u t i o n [ l l . The f i n i t e - e l e m e n t model
may or may n o t e x t e n d i n t o t h e r e g i o n R < RL i n f l u e n c e d by t h e a n a l y t i c a l
-
s o l u t i o n . For t h e problem shown t h e a n a l y t i c stress f i e l d i s i n t h e form of
K i r s h ' s s o l u t i o n [ 2 1 w i t h a n unknown stress f a c t o r u # uo. The i n t e r a c t i o n
between t h e e x-t e r i o r f i n i t e - e l e m e n t model and t h e i n t e r i o r a n a l y t i c a l model i s
used t o f i n d u as w e l l as t h e nodal d i s p l a c e m e n t s and i n t e r n a l stresses i n
the exterior grid.

The a p p r o a c h r e q u i r e s a n a n a l y t i c a l s o l u t i o n simple enough t o be e f f e c -

t i v e l y u s e d i n a computer a n a l y s i s , This a p p r o a c h appears t o be l i m i t e d t o
p r i m a r i l y i s o t r o p i c homogeneous e l a s t i c s t r u c t u r e s w i t h v e r y simple stress
c o n c e n t r a t i o n s , and f o r simple l o a d i n g s .

91
 C u r r e n t software p a c k a g e s g e n e r a l l y d o n o t c o n t a i n a l i b r a r y of a n a l y t i -
c a l s o l u t i o n s and, i f t h e n u m e r i c a l and a n a l y t i c a l r e g i o n s o v e r l a p , may n o t
have t h e c a p a b i l i t y f o r g e n e r a t i n g and s o l v i n g t h e n e c e s s a r y e q u a t i o n s of
equilibrium.

A d i f f e r e n t a p p r o a c h t o s o l v i n g t h e same problem i s t o s i m p l y model t h e

e n t i r e s t r u c t u r e w i t h a f i n i t e - e l e m e n t model u s i n g a mesh which is s u f f i -
c i e n t l y f i n e l y zoned n e a r t h e h o l e t o c a p t u r e t h e stress c o n c e n t r a t i o n
( f i g . 4 ) . The g e n e r a t i o n of such a mesh r e q u i r e s a basic u n d e r s t a n d i n g of t h e
p h y s i c s of t h e stress c o n c e n t r a t i o n , e s p e c i a l l y t h e c h a r a c t e r i s t i c l e n g t h s of
t h e d e c a y of t h e stress c o n c e n t r a t i o n , and a l s o a good u n d e r s t a n d i n g of t h e
c a p a b i l i t i e s of t h e f i n i t e e l e m e n t s c h o s e n t o model t h e s t r u c t u r e .

An a s s e s s m e n t of t h e q u a l i t y of t h e f i n i t e - e l e m e n t model may be d i f f i c u l t
i f t h e a n a l y s t h a s a l i m i t e d understanding of t h e f i n i t e elements b e i n g used
t o r e p r e s e n t t h e p h y s i c s , i . e . , a problem i f t h e a n a l y s t is u s i n g a program as
a " b l a c k box. 'I

I n o r d e r t o check f o r a v a l i d s o l u t i o n t h e e n g i n e e r o f t e n u s e s a f i n e r
mesh t o r e - a n a l y z e t h e problem e s p e c i a l l y i n t h e v i c i n i t y of t h e a p p a r e n t
s i n g u l a r i t y i n d i c a t e d from t h e p r e v i o u s s o l u t i o n . T h i s " f i e l d " a p p l i c a t i o n of
t h e p a t c h t e s t is commonly used t o check t h e convergence c h a r a c t e r i s t i c s of a
computer s i m u l a t i o n . I t would be h e l p f u l i f it were n e c e s s a r y t o g e n e r a t e
only t h e data i n t h e r e v i s e d p a r t of t h e s t r u c t u r e and the data g e n e r a t e d f o r
t h e unchanged p a r t c o u l d be r e u s e d . T h i s is a simple t a s k and y e t one which
i s n o t commonly a v a i l a b l e i n commercial computer programs. T h i s f e a t u r e may
become q u i t e i m p o r t a n t i f t h e p h y s i c s i s more c o m p l i c a t e d t h a n f i r s t believed,
s u c h as might be t h e case i f a c e n t r a l s t i f f e n e r were p r e s e n t , which would
force t h e a n a l y s t t o c o n s i d e r t h e stress c o n c e n t r a t i o n due t o t h e h o l e and t h e
n e a r b y s h e a r l a g problem due t o t h e i n t e r r u p t e d s t i f f e n e r .

The problem of a n a l y z i n g s y s t e m s i n which a p a r t of t h e s y s t e m i n v o l v e s

a n i n f i n i t e or s e m i - i n f i n i t e continuum i s a p e r p l e x i n g and d i f f i c u l t problem,
s i n c e a n u n l i m i t e d number of f i n i t e e l e m e n t s may be needed t o model t h e com-
p l e t e system. The idea o f u s i n g a n a n a l y t i c a l s o l u t i o n f o r f a r - f i e l d b e h a v i o r
t o g e t h e r w i t h a f i n i t e - e l e m e n t model of t h e n e a r - f i e l d s t r u c t u r e is a compel-
l i n g one and h a s been used by a number o f a u t h o r s [3,41. I n f i g u r e 5, RM
d e n o t e s t h e o u t e r r a d i u s of t h e f i n i t e - e l e m e n t model.

As i n t h e n e a r - f i e l d problem, a n a n a l y t i c a l s o l u t i o n t h a t c a n be e f f e c -
t i v e l y u t i l i z e d i n t h e c o n t e x t of a f i n i t e - e l e m e n t a n a l y s i s i s r e q u i r e d . The
need €or u s i n g a n a n a l y t i c a l s o l u t i o n h a s r e s t r i c t e d t h e method t o problems
i n v o l v i n g e l a s t i c i s o t r o p i c homogeneous media and t o a r e l a t i v e l y small class
of s t a t i c or f o r c e d v i b r a t i o n problems. R e c e n t l y e f f o r t s have been u n d e r t a k e n
t o g e n e r a l i z e t h e method t o problems where the medium was o r t h o t r o p i c or
l a y e r e d u s i n g f i n i t e - e l e m e n t s o l u t i o n s €or t h e f a r - f i e l d r e s p o n s e o f l a y e r e d
media i n place oE a n a l y t i c a l f u n c t i o n s .

I n t h i s a p p r o a c h , a n a l y t i c a l s o l u t i o n s may have a number of terms, e a c h

w i t h some c h a r a c t e r i s t i c f a c t o r Fi which must be related t o t h e a p p l i e d
l o a d . A s i n t h e n e a r - f i e l d case, t h e r e g i o n o f t h e a n a l y t i c a l s o l u t i o n may or
may n o t e x t e n d i n t o t h e r e g i o n of t h e f i n i t e - e l e m e n t model.

92
 The t e c h n i q u e i s n o t e a s i l y a p p l i e d u s i n g c o n v e n t i o n a l f i n i t e - e l e m e n t
a n a l y s i s packages s i n c e n e i t h e r t h e d e f i n i t i o n of t h e f a r - f i e l d s o l u t i o n n o r
t h e t e c h n i q u e s f o r matching t h e E i n i t e - e l e m e n t g r i d w i t h t h e f a r - f i e l d a n a l y t -
i c a l s o l u t i o n s are c o n t a i n e d i n t h e programs.

The f a r - f i e l d a n a l y s i s problem i s o f t e n i n v e s t i g a t e d by u s i n g a s o - c a l l e d
media i s l a n d t o t r e a t s t r u c t u r e media i n t e r a c t i o n ( f i g . 6). The f i n i t e -
e l e m e n t model t y p i c a l l y e x t e n d s as f a r as i s e c o n o m i c a l l y practical from t h e
s i t e of i n t e r e s t . On t h e o u t e r boundary, some special p r o c e d u r e is used t o
make t h e boundary a t r a n s m i t t i n g o n e , i . e . , t o permit o u t g o i n g waves t o p a s s o u t
of t h e c o m p u t a t i o n a l g r i d and e l i m i n a t e s p u r i o u s r e f l e c t i o n s t h a t might
c o n t a m i n a t e t h e s o l u t i o n . T y p i c a l boundary t r e a t m e n t s are t o u s e dampers, o r
special p a r a x i a l boundary e l e m e n t s [51, or r e c e n t l y , t o u s e a s o - c a l l e d boun-
d a r y zone s u p e r p o s i t i o n zone [6,71 t o t r a p and c a n c e l s p u r i o u s waves.

The l a t t e r method appears q u i t e p r o m i s i n g and r e q u i r e s o n l y some basic

knowledge o f wave s p e e d s i n t h e boundary zone. The r e g i o n i n t h e i n t e r i o r may
behave i n a l i n e a r or n o n l i n e a r f a s h i o n . An i m p o r t a n t c h a r a c t e r i s t i c o f t h e
boundary zone s u p e r p o s i t i o n method i s t h a t it is v e r y simple t o program and
may be u s e d i n p r i n c i p l e w i t h any f i n i t e - e l e m e n t s o f t w a r e package. Unfor-
t u n a t e l y , i n practice t h i s i s n o t t h e case s i n c e t h e a n a l y s t may be u s i n g
" b l a c k box" s o f t w a r e over which h e h a s no c o n t r o l .

The development of a p r a c t i c a l t o o l €or n e a r - f i e l d / € a r - f i e l d a n a l y s i s h a s

major i m p l i c a t i o n s €or s u c h problems a s , u s i n g u l t r a s o n i c s f o r n o n d e s t r u c t i v e
t e s t i n g , d e v e l o p i n g s e n s o r / c o n t r o l s y s t e m s on v e r y l a r g e s p a c e c r a f t , and
s t u d y i n g impact o f l a r g e b o d i e s , s u c h as s p a c e c r a f t and S h u t t l e - t y p e t r a n s p o r t
vehicles.

The b e h a v i o r o f c o n n e c t i o n s i n s t r u c t u r e s i s a p e r s i s t e n t problem f o r
s t r u c t u r a l e n g i n e e r s , e s p e c i a l l y on s t r u c t u r e s such as a space s t a t i o n which
may have h u n d r e d s or even t h o u s a n d s o f c o n n e c t i o n s . U n l i k e t e r r e s t r i a l
s y s t e m s the c o n n e c t i o n s on spacecraft may be v e r y l i g h t l y l o a d e d and t h e r e f o r e
p l a y a v e r y important role i n d e t e r m i n i n g s t r u c t u r a l r e s p o n s e of the o v e r a l l
system. The c o n n e c t i o n i s a p h y s i c a l stress c o n c e n t r a t i o n ( o r s u b s t r u c t u r e ) i n
which s t r u c t u r a l c h a r a c t e r i s t i c s may be q u i t e complex ( f i g . 7 ) . Y e t for
p u r p o s e s of a n a l y z i n g a v e r y l a r g e system t h e h i g h l y r e s o l v e d b e h a v i o r o f a
c o n n e c t i o n must be c o n s i s t e n t l y and a p p r o p r i a t e l y reduced t o a l e v e l usable
€or t h e a n a l y s i s of t h e large-scale system.

The development of a simple s u b s t r u c t u r e model which g i v e s t h e e s s e n t i a l

b e h a v i o r of t h e s u b s t r u c t u r e i n a global s y s t e m is a n o n - t r i v i a l t a s k . C e r -
t a i n l y "downsizing" a h i g h l y r e s o l v e d model o f a c o n n e c t i o n t o a much simpler
model s u i t a b l e f o r u s e i n a global a n a l y s i s s h o u l d be based on a c o n s i s t e n t
f o r m u l a t i o n i n which o v e r a l l i n t e r n a l e n e r g y under s p e c i f i c d e f o r m a t i o n pat-
t e r n s is m a i n t a i n e d , and t h e model s h o u l d a c c o u n t f o r appropriate r i g i d - b o d y
behavior.

The u s e of c o n n e c t i o n e l e m e n t s h a s been i n c o r p o r a t e d i n a number of

f i n i t e - e l e m e n t programs, e s p e c i a l l y €or u s e i n p i p i n g a n a l y s e s i n t h e n u c l e a r
power i n d u s t r y . However, t h e s o f t w a r e for a n a n a l y s i s s u c h as d e s c r i b e d f o r
t h e c o n n e c t i o n w i l l have t o have t h e c a p a b i l i t y of g i v i n g t h e u s e r c o n t r o l
o v e r t h e d e f i n i t i o n o f t h e e l e m e n t (as compared t o h a v i n g access t o one o f
several basic c o n n e c t i o n e l e m e n t s i n a g e n e r a l e l e m e n t l i b r a r y ) .

93
 A d i f f e r e n t t y p e o f s u b s t r u c t u r e problem which is e n c o u n t e r e d i n g l o b a l -
local f i n i t e - e l e m e n t a n a l y s i s i s t h e case where t h e i m p o r t a n t p h y s i c s i s i n a
small r e g i o n of a l a r g e s t r u c t u r e ( f i g . 8). T h i s is t h e t y p i c a l case i n a
problem i n v o l v i n g t h e a n a l y s i s of crack growth u s i n g f r a c t u r e mechanics. In
order t o d e t e r m i n e t h e rate a t which t h e c r a c k shown on t h e s t r u c t u r e w i l l
grow g i v e n a p p l i e d c y c l i c l o a d i n g s , t h e stress i n t e n s i t y f a c t o r a t t h e f r o n t
of t h e c r a c k must be d e t e r m i n e d f o r t h e c r a c k as it grows d u r i n g l o a d i n g , or
a l t e r n a t i v e l y , t h e s t r a i n e n e r g y release rate. This is accomplished by
a n a l y z i n g t h e s t r u c t u r e w i t h a g i v e n l e n g t h o f c r a c k , and t h e n r e l e a s i n g t h e
c o n n e c t i o n between e l e m e n t s a t t h e t i p of t h e c r a c k ( a l l o w i n g i t t o advance
f o r one e l e m e n t ) and r e a n a l y z i n g t h e s t r u c t u r e .

T h i s p r o c e d u r e amounts t o a model r e v i s i o n ; t h u s , t h e e n t i r e s t r u c t u r a l
a n a l y s i s problem must be r e e s t a b l i s h e d and r e s o l v e d . T h i s is o n l y p r a c t i c a l
on a l a r g e system i n v o l v i n g many t h o u s a n d s of D O F ' s i f t h e s u r r o u n d i n g s t r u c -
t u r e is t r e a t e d a s a s u b s t r u c t u r e and t h e c r a c k growth r e g i o n as t h e p r i m a r y
s t r u c t u r e (which may be r e p e a t e d l y m o d i f i e d t o p e r f o r m t h e s t r a i n e n e r g y
release r a t e c a l c u l a t i o n ) . T h i s t y p e of a n a l y s i s c a n be done u s i n g a v a i l a b l e
commercial software, b u t o n l y i n a o n e - s o l u t i o n - a t - a - t i m e mode. I t would be
v e r y h e l p f u l i f t h e p r o c e d u r e c o u l d be c a r r i e d forward i n a semiautomatic
manner t h a t would r e q u i r e s u b s t a n t i a l software development.

T h i s a n a l y s i s i s v e r y i m p o r t a n t i n making s a f e - l i f e p r e d i c t i o n s f o r
c r i t i c a l components i n a i r c r a f t and spacecraft. O f c o u r s e , the problem i s
much more c o m p l i c a t e d i f t h e d i r e c t i o n of c r a c k growth i s unknown, since t h e
f i n i t e - e l e m e n t models of t h e s u b s t r u c t u r e and s t r u c t u r e c o u l d n o t be d e t e r -
mined p r i o r t o a n a l y s i s . I n s h o r t t h e f i n i t e - e l e m e n t model would have t o be
adaptive.

Based on t h e comments i n f i g u r e 8 it i s e v i d e n t t h a t f i n i t e - e l e m e n t
modeling must be a d a p t i v e i n order t o make s a f e - l i f e p r e d i c t i o n s , a process
which now i n v o l v e s t h e e n g i n e e r a n a l y s t d i r e c t l y . I n f a c t , c o n s i d e r a b l e
r e s e a r c h h a s been done t o d e v e l o p s e m i - a u t o m a t i c , a d a p t i v e f i n i t e - e l e m e n t mesh
g e n e r a t o r s [8-101. These p r o c e d u r e s operate i n b a s i c a l l y one o f t w o ways,
r e f i n e m e n t of t h e mesh i t s e l f , u s i n g s i m i l a r f i n i t e e l e m e n t s the same o r d e r of
a p p r o x i m a t i o n w i t h i n e a c h e l e m e n t (H-convergence), o r l e a v i n g t h e g r i d f i x e d
b u t r e f i n i n g t h e p h y s i c s w i t h i n e a c h e l e m e n t (P-convergence). Different
s t r a t e g i e s are u s e d t o assess t h e q u a l i t y o f a s o l u t i o n f o r a g i v e n f i n i t e -
e l e m e n t g r i d . The same i n f o r m a t i o n i s t h e n used t o r e v i s e t h e model and
improve t h e s o l u t i o n .

I n r e s p o n s e t o a t e s t problem proposed by NASA as a v e h i c l e f o r d i s c u s -

s i o n a t t h e workshop ( f i g . 9) a simple h i g h l y i d e a l i z e d model of t h e s t r u c t u r e
w a s prepared by D r . P a o l o Roberti and a n a l y z e d u s i n g h i s a l g o r i t h m [ l o ] . T h i s
a l g o r i t h m u s e s t r i a n g l e c o n s t a n t s t r a i n f i n i t e e l e m e n t s and H-convergence.
The r e s u l t s are r e m a r k a b l e , g i v i n g almost a map of t h e stress c o n c e n t r a t i o n i n
t h e v i c i n i t y of t h e h o l e i n t h e s t i f f e n e d p a n e l , f i g u r e 10a-f. O f course,
t h i s a n a l y s i s was conducted o n l y f o r a l i n e a r s t a t i c s o l u t i o n . I n r e a l i t y ,
t h e p r e s e n c e of the h o l e i n t h e p a n e l may l e a d t o i n s t a b i l i t y or even f a i l u r e .
N e v e r t h e l e s s , t h e a n a l y s i s i s i m p o r t a n t i n d e v e l o p i n g a f i n i t e - e l e m e n t model
with a s p e c i f i e d p r e c i s i o n .

Approaches s u c h as t h i s o n l y h i n t a t t h e tremendous problem s o l v i n g power

t h a t c a n be b r o u g h t t o bear on s t r u c t u r a l e n g i n e e r i n g problems i f t h e software

94
a v a i l a b l e c a n be d e s i g n e d t o be f l e x i b l e enough t o adopt new and d i f f e r e n t
concepts i n a n a l y s i s .

CONCLUDING REMARKS

The v a r i o u s t y p e s o f local-global f i n i t e - e l e m e n t problems p o i n t o u t t h e

need t o d e v e l o p a new g e n e r a t i o n of software. F i r s t , t h i s new s o f t w a r e needs
t o have a complete a n a l y s i s c a p a b i l i t y , encompassing l i n e a r and n o n l i n e a r
a n a l y s i s o f 1-, 2-, and 3-dimensional f i n i t e - e l e m e n t models, as w e l l a s mixed
d i m e n s i o n a l models. The software must be c a p a b l e of t r e a t i n g s t a t i c a n d
dynamic ( v i b r a t i o n and t r a n s i e n t r e s p o n s e ) problems, i n c l u d i n g t h e s t a b i l i t y
e f f e c t s of i n i t i a l stress, and t h e software s h o u l d be a b l e t o t r e a t b o t h
e l a s t i c and e l a s t o - p l a s t i c materials.

The software s h o u l d c a r r y a s e t of o p t i o n a l d i a g n o s t i c s t o assist t h e

program u s e r d u r i n g model g e n e r a t i o n i n order t o h e l p a v o i d o b v i o u s s t r u c t u r a l
modeling errors. I n a d d i t i o n , t h e program software s h o u l d be w e l l documented
so t h e u s e r h a s a complete t e c h n i c a l r e f e r e n c e f o r e a c h t y p e o f e l e m e n t con-
t a i n e d i n t h e program l i b r a r y , i n c l u d i n g i n f o r m a t i o n on s u c h topics as t h e
t y p e of n u m e r i c a l i n t e g r a t i o n , u s e of u n d e r i n t e g r a t i o n , and i n c l u s i o n o f
i n c o m p a t i b l e modes, etc. Some packaged i n f o r m a t i o n s h o u l d a l s o be a v a i l a b l e
t o assist t h e u s e r i n b u i l d i n g mixed-dimensional models.

An i m p o r t a n t advancement i n f i n i t e - e l e m e n t s o f t w a r e s h o u l d be i n t h e
development of program m o d u l a r i t y , so t h a t t h e u s e r c a n select from a menu
v a r i o u s basic o p e r a t i o n s i n m a t r i x s t r u c t u r a l a n a l y s i s , i n c l u d i n g m a t r i x for-
m u l a t i o n and storage, assembly ( b y r o w or column), s o l u t i o n ( b y r o w , column or
wave f r o n t ) , and method o f t i m e i n t e g r a t i o n . Most i m p o r t a n t , t h e s o f t w a r e
s h o u l d permit t h e u s e r / a n a l y s t t o l i n k t o t h e computer program h i s own spe-
c i a l i z e d software. U s e r programs might i n c l u d e f o r m u l a t i o n o f ( s u b s t r u c t u r e )
s t i f f n e s s matrices, s p e c i a l i z e d s o l u t i o n packages ( m a t r i x i n v e r s i o n , p a r t i a l
i n v e r s i o n ) , t i m e i n t e g r a t i o n and, f o r n o n l i n e a r problems, i n p u t of d i f f e r e n t
t y p e s of materials.

The n e x t g e n e r a t i o n o f f i n i t e - e l e m e n t s o f t w a r e a l s o s h o u l d be developed
w i t h t h e idea of a n a l y s i s s e r v i n g as a basic t o o l i n d e s i g n , s y s t e m i d e n t i -
f i c a t i o n and o p t i m i z a t i o n .

The i m p l e m e n t a t i o n of a d a p t i v e f i n i t e - e l e m e n t modeling t e c h n i q u e s i n com-

mercial g r a d e software w i l l have a major impact on t h e s t r u c t u r a l e n g i n e e r i n g
community which now i n v e s t s a s i g n i f i c a n t e f f o r t on basic a n a l y s i s , e s p e c i a l l y
i n t h e modeling, s o l u t i o n , and r e m o d e l i n g c y c l e . A number of problems i n non-
l i n e a r s t r u c t u r a l a n a l y s i s w i l l a l s o b e n e f i t from a d a p t i v e computer modeling,
s u c h as making s a f e - l i f e p r e d i c t i o n s €or s t r u c t u r e s u s i n g f r a c t u r e mechanics
concepts.

H o p e f u l l y , a new g e n e r a t i o n o f s o f t w a r e c a n be d e v e l o p e d w i t h many, i f
n o t a l l , of t h e f e a t u r e s d e s c r i b e d . I f it i s possible t o do so, t h e n s t r u c -
t u r a l a n a l y s i s s o f t w a r e w i l l become a much more complete, v e r s a t i l e and
reliable tool f o r t h e s t r u c t u r a l engineer.

95
 REFERENCES

1 . B r a d f o r d , L. G . , Dong, S. B., Nicol, D. A. C., and Westmann, R. A.,

" A p p l i c a t i o n o f G l o b a l - L o c a l F i n i t e E l e m e n t Method t o F r a c t u r e Mechanics,"
E l e c t r i c Power R e s e a r c h I n s t i t u t e Report #EPRI NP-239 ( R e s e a r c h P r o j e c t
299-1 1 T e c h n i c a l Report 1 , September 1976.

2. Timoshenko, S. e . , a n d G o o d i e r , J. N., TheoryofElasticity, Third Edi-

t i o n , McGraw H i l l , N e w Y o r k , 1970.

3. Muki, R., and Dong, S. B., "Some Remarks on t h e U s e of Asymptotic Solu-

t i o n s i n Global Local F i n i t e Element A n a l y s i s €or a n E l a s t i c Half S p a c e , "
on M e c h a n i c a l B e h a v i o r -
Recent R e s e a r c h - of S o l i d s , U n i v e r s i t y o f Tokyo
P r e s s , 1979, pp. 55-78.

4. Medina, F., and T a y l o r , R. L., " F i n i t e E l e m e n t T e c h n i q u e s f o r Problems of

Unbounded Domains," I n t e r n a t i o n a l J o u r n a l f o r Numerical Methods -~ i n Engi-
n e e r i n g , V o l . 19, N o . 8, August 1983, pp. 1209-1226.

5. Cohen, M., and J e n n i n g s , P. C., " S i l e n t Boundary Methods €or A n a l y s i s , "

C o m p u t a t i o n a l Methods f o r I n f i n i t e Domain M e d i a - S t r u c t u r e I n t e r a c t i o n ,
ASME, 1982, pp. 183-204.

6. Kumar, R. R., and Marti, J., "A N o n - R e f l e c t i n g Boundary for E x p l i c i t

C a l c u l a t i o n s , " C o m p u t a t i o n a l Methods f o r I n f i n i t e Domain Media - S t r u c t u r e
I n t e r a c t i o n , ASME, 1981 , pp. 183-204.

7. Muki, Y . , "A R a d i a t i n g Boundary Method f o r L i n e a r and N o n l i n e a r T r a n s i e n t

F i n i t e Element A n a l y s i s , " M.S. T h e s i s , UCLA S c h o o l of E n g i n e e r i n g and
A p p l i e d S c i e n c e , 1985.

8. d e S. R. Gago, K e l l y , D. W., Z i e n k i e w i c z , 0. C., and Babuska, I., "A

P o s t e r i o r i Error A n a l y s i s and Adaptive P r o c e s s e s i n t h e F i n i t e Element
Method: Part I - E r r o r A n a l y s i s , " I n t e r n a t i o n a l J o u r n a l f o r Numerical
Methods - i n E n g i n e e r i n g , V o l . 1 9 , N o . 8, August 1983, pp. 1593-1620.

9. d e S. R. Gago, K e l l y , D. W., Z i e n k i e w i c z , 0. C., and Babuska, I., " A

P o s t e r i o r i Error A n a l y s i s and A d a p t i v e P r o c e s s e s i n t h e F i n i t e Element
Method: P a r t I1 - A d a p t i v e Mesh R e f i n e m e n t , " I n t e r n a t i o n a l J o u r n a l f o r
Numerical --- Methods i n E n g i n e e r i n g , V o l . 1 9 , No. 8, August 1983,
pp. 1621-1656.

10. Roberti, P. , " S e l f A d a p t i v e Mesh i n S t r e s s A n a l y s i s , " Ph.D. Dissertation,

IJCLA S c h o o l of E n g i n e e r i n g and Applied S c i e n c e , 1984.

96
 BASELINE MODELING CONCEPTS FINITE-ELEMENT MODELING

I - Which Model is Better?

Dimensional Level of Model

d p
- Technical Theory
1-'D Truss/Beam
21- D Panel/PIate/Shell
- Continuum Representation BEAM
2-D Plane Stress, Plane Strain PLANAR 2-D 1
*, 2 D O F

nth order Symmetry

31- D

Physical Requirements of Model

rITIiIiIIIiIT-11
PLANAR 2-D
4P
96D0F

- Material (Elastic, Plastic, Anisotropic) 192 DOF

- Initial Stress (Stability) P

- Dynamics (Vibration, Transient Analysis) 768 DOF

- Nonlinear (Large Deformation, Separation)
P

Figure 1 Figure 2

97
 NEAR-FIELD MODELING - Finite-Element Modeling with Mesh
Refinement
Singularities/Stress Cmcentrations
- Finite-Element Model with Near-Field
Analytical Solution

- Analytical Solution(s1 Required Understanding of Characteristic Lengths of

Capable of Being Evaluated and Utilized Physical Processes Required
- Restricts Approach t o Evaluation of Model, Solution Difficult
Isotropic, Homogeneous Elastic Systems,
Very Simple Regions, Simple Loadings, Model Refinement Capability Important
Statical Problems, or Forced Vibrations

Figure 3 Figure 4

98
 FAR-FIELD MODELING
-Finite-Element Model In Continuum Island
Finite Structure Imbedded in Infinite Continuum
- Finite Element Model with Far Field
Analytical Solution

\
J
RADIATION
B. C.

Formulation Requires Special Boundary

Analytical Solution(s) Required Conditions to Insure Radiation

Capable of Being Evaluated and Utilized Dampers

Reflecting Zones
Restricted t o Isotropic Homogeneous Elastic
Far Field, Static or Forced Vibration Knowledge of Wave Transmitting
Characteristics of Medium Required

Figure 5 Figure 6

99
 SUBSTRUCTURE CONCEPTS - Surrounding Substructure

Use of Highly Resolved Models

- Imbedded Substructures

I *' CRACKGROWTH
REGION

Figure 7 Figure 8

100
 ADAPTIVE FINITE ELEMENT MODELING

BLADE-STIFFENED PANEL WITH DISCONTINUOUS STIFFENER

UNIFORM END SHORTENING

\ IP

Figure 9

RESULTS OF ROBERTI'S ALGORITHM FOR

SUCCESSIVELY REFINED MESHES TO
CAPTURE STRESS CONCENTRATIONS
IN NASA TEST PROBLEM

b e

C f

F i g u r e 10

101
 N89- 2 4 6 4 4
SOME COMMENTS ON GLOBAL-LOCAL ANALYSES

Satya N. Atluri
Georgia Institute of Technology
Atlanta, Georgia

ABSTRACT

The maln theme of this paper concerns methods that may be classified as global
(approximate) and local (exact). Some specific applications of these methods are
found in:

(1) Fracture and fatigue analysis of structures with 3-D surface flaws

(2) Large-deformation, post-buckling analysis of large space trusses and space

frames, and their control

(3) Stresses around holes in composite laminates

PRECEDING PAGE U N K NOT FKMED 103

 A typical engineering problem is illustrated in figure 1, which shows a corner
flaw at the intersection of a nozzle and a pressure vessel. The shape of the
surface flaw may often be approximated mathematically as quarter-elliptical or
quarter-circular. For the problem shown in figure 1, wherein the crack is located
in the longitudinal plane of symmetry of the structure, only the so-called Mode I
conditions prevail. In figure 1, the presence of a traction-free crack, in an
otherwise unflawed solid, alters the stress-state only locally. From a viewpoint of
fracture mechanics, however, the main quantities of interest are only the stress-
intensity factors (strengths of asymptotic stress singularities) near the crack
front. For analyzing fatigue crack growth and crack instability under thermal shock
various flaw sizes and shapes need to be considered. The primary objective of
analysis is to determine the variation of the Mode I stress-intensity factor along
the border on the surface flaw.

Figure 1. Corner surface-flaw at the pressure-vessel-nozzle intersection.

104
I Figure 2 shows the schematic of a 12-bay space frame. The equations of dynamic
motion of the frame, assuming large deformations and plasticity, may be written as:

t
where M, is the mass matrix, the matrix of passive damping, K, the tangent stiff-
ness matrix (which includes the effect of large defgrmation and plasticity), 4f the
control-actuator force, gE the external load, (N+l)X_ the acceleration vector at
time tN+1, (N+l)?N the velocity vector at tN+1, the incremental displacement be-
tween tN and tN+1, and “1% the internal-force vector at tN. In order to implement
the control algorithms in an efficient manner, the order of the above system of
equations must be as small as possible (i.e., each frame member must be modeled by
no more than one finite element). Further, the control must be implemented for
pulse-type loading of high intensity, such that the above system of equations must
be integrated directly rather than using a modal-decomposition. Also requirements
of on-line control may necessitate that tK_, C,, and M, be known e x p l i c i a (in closed
form) for arbitrary values of deformation, without the need for introducing approxi-
mate shape functions for deformation of each element and without the need €or any
numerical integrations over each element. In figure 2, the object of inquiry is
what effect does local (member) instability have on global (system) stability? How
can we control the dynamic deformations locally to improve global behavior? Each
member may be treated as a truss member, o r a 3-D beam-type member, depending on
joint design. How can local effects be accounted for simply and efficiently, so
that algorithms for control of dynamic motion may be implemented, on line, using
on-board computers in a large space structure?

4 4
72 8 3

Figure 2. Schematic of a 12-bay space frame.

105
, An appraisal of the computational mechanics methods is given in figure 3.
These methods include the finite-element, boundary-element, and edge function
methods (fig. 3).

F: IN ITE ELEMENTS:
TRIAL AND TEST F U N C T I O N S ARE B O T H A P P R O X I M A T E
TRIAL AND TEST F U N C T I O N S ARE, I N GENERAL, ALIKE - GALERKIN
APPROACH
I N SOME I N S T A N C E S I T I S B E S T TO HAVE T E S T F U N C T I O N S D I F F E R E N T
FROM TRIAL - PETROV-GALERKIN APPROACH
FUNCTIONS
-GLOBALLY AND LOCALLY APPROXIMATE
THE SOLUTION I S BOTH
VERSATILE OR ARBITRARY GEOMETFIY, BOUNDARY CONDITIONS, SUITED
FOR G L O B A L L Y A P P R O X I M A T E NCINLINEAR S O L U T I O N S

BOUNDARYELEMENTS:

TEST FUNCTIONS ARE GLOBALLY EXACT FOR THE GIVEN LINEAR PROBLEM,
OR A T L E A S T FOR T H E HIGHEST-ORDER D I F F E R E N T I A L OPERATOR O F
T H E PROBLEM
TRIAL FUNCTIONS ARE APPROXIMATE ( A T BOUNDARY ONLY FOR LINEAR
PROBLEMS, AND IN INTERIOR ALSO FOR NONLINEAR PROBLEMS)
THE SOLUTION SI BOTH LOCALLY AND GLOBALLY APPROXIMATE
NOT A S VERSATILE AS THE FINITE-ELEMENT METHOD, BUT EXCELLENT
FOR SOME S P E C I F I C PROBLEMS

EDGE FUNCTION METHOD:

TRIAL FUNCTIQNS ARE GLOBALLY EXACT
TEST FUNCTIONS ASSUMED ONLY AT BOUNDARY
LIMITED TO LINEAR PROBLEMS POSED BY CLASSICAL DIFFERENTIAL
EQUATIONS

Figure 3. Appraisal of computational mechanics methods.

106
 In the most commonly used Galerkin finite-element approach in computational
solid mechanics, the trial and test function spaces are identical and consist of
simple piecewise continuous algebraic polynomials over each finite element, such
that these functions and their appropriate-order derivatives (as dictated by the
problem on hand) are continuous at the interelement boundaries. For problems of
fourth or higher order, such as those of plates and shells, the development of
finite elements has long been, and continues to be, plagued by the need for C1
(or higher order) continuity at the interelement boundaries. However, the success
of the finite-element method in structural mechanics is unparalleled and is mainly
due to the intuitive and 'geometric' interpretation of the method. The method is
versatile in its ability to deal with complicated structural assemblies, such as of
beams, plates, and shells, of the type used in aerospace applications. The solu-
tions obtained through the finite-element method may be classified, in general, as
being both globally as well as locally approximate.

On the other hand, in linear and nonlinear solid mechanics, it is often pos-
sible to derive certain integral representations for displacements. A key ingre-
dient which makes such derivations possible is the singular solution, in an infinite
space, of the corresponding differential equation (in certain linear problems) or of
the highest-order differential operator (in the nonlinear case, or even in the
linear case when the full linear equation cannot be conveniently solved), for a
'unit' load applied at a generic point in the infinite space. When the problem is
linear and the singular solution can be established for the complete linear dif-
ferential equation of the problem, the aforementioned integral representations for
displacements involve only boundary integrals of unknown trial functions and their
appropriate derivatives. Such an integral representation, when discretized, leads
to the so-called boundary-element method. Such pure boundary-element methods are
possible in linear, isotropic, elastostatics, and in problems of static bending of
linear elastic isotropic plates. On the other hand, as in such cases as (i) linear
problems wherein the singular solutions cannot be established for the entire dif-
ferential equations, (ii) anisotropic materials, and (iii) problems of large
deformation and material inelasticity, the integral representations (if any) for
displacements would involve not only boundary integrals but also interior-domain
integrals of the trial functions and/or their derivatives. A discretization of such
integral equations would lead not only to a simple boundary-element method but also
to a sort of hybrid boundary/interior element method.

When asymptotic solutions to the governing differential equations of the prob-

lem are used as assumed trial functions, the interior residual error is zero, and
only the boundary conditions need to be satisfied in a weighted residual method.
Such an approach is called the edge-function method, but is limited mostly to linear
problems. For further details, see references 1 through 4.

107
 The s u r f a c e - f l o w problems f o r c u r r e n t methods are n o t e d i n f i g u r e 4 . Problems
f o r t h e proposed method a r e a l s o shown.

A1 1 Present Methods: S i n g u l a r s t r e s s - s t a t e near t h e f l a w

border i s model ed by 1oca1 l y approximate methods

( I ) F i n i t e - E l e m e n t Methods ( s i n g u l a r elements)
A t l u r i & K a t h i r e s a n ( r e f s . 5-10)
( H y b r i d crack elements) 3-6,000 d.0.f.
Tracy, Barsoum, Newman & Raju ( r e f s . 11-13)
( D i s t o r t e d isoparametric elements and
s i n g u l a r shape fn.) 5-10,000 d.0.f.
These a r e very expensive, b u t accommodate
a r b i t r a r y geometries o f s t r u c t u r e and flaw.

Boundary-El ement Methods ( f o r l o c a l l y approxi-

( II -
mate s t r e s s a n a l y s i s and K - e s t i m a t i o n from
stress extrapol i t i o n )
Cruse ( r e f . 141, H e l i o t e t a l . ( r e f . 1 5 ) .
Not s u i t a b l e f o r ' t h i n ' s h e l l s w i t h flaws.
S t i 11 very expensive.

( III 1 L i ne-Spri ng Method

- o f s t r u c t u r e and
E m i t e d t o simple geometries
flaw.

Proposed Method:
---------
It i s a GLOBALLY APPROXIMATE, b u t -- LOCALLY EXACT
METHOD
S i n g u l a r s t r e s s - s t a t e near t h e f l a w i s NOT MODELED
NUMERICALLY
It i s about 30 times cheaper than t h e s i n g u l a r
f i n i t e - e l e m e n t method
' -
D e t a i l s ( A t l u r i & N i s h i o k a ( r e f s . 16-21) several
papers w i t h v a r i e d examples)

F i g u r e 4 . S u r f a c e - f l a w problems.

108
 solution for
uncracked body

Locally exact solution

for pressurized crack

Remarks: 1. S o l u t i o n D: A r a t h e r complicated a n a l y t i c a l
s o l u t i o n ( A t l u r i & V i jayakumar, Journal o f
Appl i e d Mechanics, 19811 ( r e f . 2 2)
2. Local s o l u t i o n due t o crack-face t r a c t i o n alone
i s (i.e., t h e S o l u t i o n C) t h e source o f singu-
l a r i t y . The stresses due to t h i s l o c a l s o l u t i o n
decay- very r a p i d l y . Only one o r two 1t e r a t i o n- s
a r e s u f f i c i e n t t o o b t a i n K - s o l u t i o n s w i t h 1%
accuracy.

About 30 times cheaper than t h e usual f i n i t e -

element method f o r a t y p i c a l problem such as
flawed BWR nozzle.

Figure 5. Global (approximate) and local (exact) analyses of embedded flaws.

109
 Some comments concerning the solution of surface flaws in finite bodies using
the present procedure are in order (fig. 4 ) . Since the analytical solution OF an
elliptical flaw embedded in an infinite solid is used as solution D, it is necessary
to define the residual stresses over the entire crack-plane including the fictitious
portion of the crack which lies outside of the finite body containing only a surface
flaw (fig. 5). Moreover it is well known that the accuracy of the ‘least-squares’
type funct€on-interpolation inside the interpolated region can be increased with the
number of polynomial terms; however, the interpolating curve may change drastically
outside of the region of interpolation. The optimum variation of pressure on the
crack surface extended into the fictitious region should be as shown in figure 6 .
For in-depth discussions of a variety of surface problems and their solutions, see
refs. 19-21, 23-28.

ASSUMED
\ / ACTUAL

ICAL

Figure 6. Postulated residual stress distributions on fictitious portions of an

elliptical flaw, in the case of semi-elliptical or quarter-elliptical
surface flaws.

110
 For both 3-dimensional truss and frame members, explicit (locally exact)
tangent stiffness matrices have been derived (fig. 7). Some effects of local
(member) buckling on global (structural) behavior are illustrated in figures 8
and 9.

Truss Member:
t a c h member undergoes l a r g e displacement and l a r g e
r i g i d rotation
Member m a t e r i a l i s n o n l i n e a r
Each member may buckle and become curved (what
e f f e c t does i t have on global s t a b i l i t y ? )

Frame Member:
'-Concepts o f 3-dimensional semi-tangential r o t a t i o n s
employed
Each member undergoes a r b i t r a r i l y l a r g e r i g i d r o t a -
ti ons and r ig i d d i s p l acements
Bending-stretching c o u p l i n g i n c o r p o r a t e d i n each
member
P l a s t i c - h i n g e method used t o account f o r p l a s t i c i t y
i n each member
Member forces: a x i a1 , shear, and bending- t w i s ti ng
moments

Kondoh and A t l u r i ( r e f s . 29-30) and Kondoh e t a1 . (refs. 31-32)

Figure 7. Space trusses and space frames.

111
 L = 66.04(cm)
( a ) Thompson's Strut

system i m p e r f e c t i o n considered

no system i m p e r f e c t i o n

5 0

Bot h member
buckling and
considered (two d i f f e r e n t
30:: ;/ magnitudes o f i m p e r f e c t i o n s )

~ oj$
1,
Dlsplocement of Node No 10 In the X Direction
00 1
03 2 0 $0 6 0 6 0 '00 ' 2 0 '40 I60
(CU)

( b ) Effect of local (member) buckling on global (struc-

tural 1 behavior.
Tangent s t i f f n e s s of each member i s exact in both the pre-buckled
and post-buckled states o f member ( r e f . 29)

Figure 8. Thompson's s t r u t and e f f e c t of l o c a l buckling on g l o b a l behavior.

112
 10-5
9
\
-9,
N
n
I1
-e
n

-7

-6

-5 EA = 4.59Kg;
Member Buckling E1 = 250 Kgm2

-4

-7

EA = 4.59Kq; E I = 123
Kgm2

I t x3 i
I
- 10 - 20
I I
-5 -I5
pI -
-25 I
PI /EA
104

(a) Load System: ( i ) PI: Vertical (b) Stabi 1 i ty boundary under 1 oads
point loads as all nodes; PI and P2
( i i ) Pz: V e r t i c r p o i n t loads a t
nodes i n quadrants XI, x2 > 0

Figure 9. Study of the effect of member buckling on global (system) stability.

113
 Examples of the efficiency of the global/local approach in analyzing frames are
illustrated in figures 10 and 11. In figure 10, the classical problem of a two-bar
frame is schematically illustrated. In the present approach, the tangent stiffness
matrix of each member (respresented by a single finite element) is derived from
exact solutions of governing differential equations which account for the bending-
stretching coupling. Thus, no "shape functions" are assumed in each element, and no
numerical integrations are performed in forming the tangent stiffness matrix. The
present numerical integrations are performed in forming the tangent stiffness
matrix. The present numerical results are shown to agree excellently with those of
Wood and Zienkiewicz (ref. 3 4 ) , as well as the experimental results of Williams
(ref. 35). However, Wood and Zienkiewicz use five finite elements to model each
member of the frame.

In figure 11, the problem of plastic collapse of a frame is €llustrated. Here

again, the tangent stiffness matrix of each member (represented by a single finite
element) is derived in closed form, accounting for large deformations, large rota-
tions, and plasticity. A plastic-hinge method is used, and the progressive develop-
ment of plastic hinges, at various load levels, Is indicated in figure 11.

E = I 03r107~1b/~n~

00
y
00
1000

01
2000

02
2000 4000

03 04
5COO

05
6000

06
-
1000 9"h)

07 Slnl

Figure 10. Variation of load-point displacement and support reaction with applied
load in a two-bar frame. Tangent stiffness of each 3-D beam member
undergoing large deformation, large rotation, and plasticity is exact.
Locally exact solution (ref. 30).

114
 PL ~b

3 ~1 1
&I4
2.2198

2.667 (Limit Analysis, U. B 1

2.51 ~ $ f2.500
- (Llmlt- ~
Analyslsi L. 8.1 -d=

2.3962

1
iI I 2.5454 2 6186

R
I.Oi

0.5ll/
Iii// M; 0 Fully Plostic Bendin
of Sectlon 6 Moment

1'. Moment of lnertla of Sectlon @

12
0 PIos11c Hinge

05 10 15 2.0

' TANGENT S T I F F N E S S O F EACH MEMBER UNDERGOING LARGE DEFORMATION, LARGE R O T A T I O N AND

PLASTICITY IS EXPLICIT AND EXACT, PLASTIC-HINGE METHOD USED,
EXTENSION TO CRASH ANALYSIS OF FRAMES BEING STUDIED,

F i g u r e 11. P l a s t i c collapse of frame.

115
 Figure 12 shows a problem of current interest in the analysis of stiffened
composite plates. Issues involve the following: (1) stress concentrations near the
hole in a composite laminate, ( 2 ) local buckling of stiffeners, ( 3 ) effect of geo-
metric imperfections, ( 4 ) effect of discontinuities, and ( 5 ) three-dimensional
effects and delaminations near the hole. An efficient globally approximate and
locally exact approach could possibly include: (1) use of locally exact, laminated
hole elements with embedded three-dimensional stress state (refs. 36 and 371,
(2) use of locally exact stiffener elements as described earlier (ref. 32),
( 3 ) techniques for proper interacting of various elements, and ( 4 ) hole elements
that can be improved by incorporating possible free-edge singularities in u
3i'

Focus PROBLEM

(GLOBALLY 8 LOCALLY APPROXIMATE) F I N I T E ELEMENTS]

Figure 12. A stiffened laminated-composite panel with a hole.

116
I Another example of the advantages of using a global/local approach is illus-
trated here in the problem of analysis of stresses near a hole in a laminated
I composite [two cases of (-45/+45), and (90/0), laminates are discussed]. Figure 13a
shows a typical finite-element model with "special-hole elements" in which a 3-D
asymptotic hole solution is embedded. Figures 13b and 13c illustrate the excellent
accuracy obtained from the present approach, in comparison with a fully 3-D finite-
element solution of Rybicki and Hopper (ref. 38). The present solution is, however,
an order of magnitude less expensive. Further details are given in references 36
and 37.

TYPICALFUI MODEL OF A LAMINATE STRESS AROUND A HOLE STRESS AROUND A HOLE

WITH HOLE, 3-0 ASYMPTOTIC "HOLE- IN (-45/+45), LAMINATE IN (90/0), LAMINATE
SOLUTIONS" EMBEDDED I N ELEMENTS
NEAR THE HOLD(Ref. 36)

Figure 13. Analysis of stress state near a hole in laminated composites.

117
The following conclusions and recommendations are given.
.. Hybrid analytical/numerical methodologies should be explored
Simplified analysis procedures for elasto-plastic should be considered
(Dynamic response calculations should be studied (some benchmark problems
essential))
Constitutive models badly need improvement
Methods of coupling of problem-specific methodologies for use in general
purpose programs should be explored
Trends to treat structural mechanics problems as continuum mechanics
problems should be critically reviewed; the knowledge base in structural
mechanics should be fruitfully utilized
Attempts to bridge the gap between micromechanics and macromechanics of
heterogeneous (composite) media through computational mechanics should be
explored
Computational stochastic structural analysis methods should be developed
Algorithms for new computing systems (MIMD) should be explored
Expert systems, .... (?)

NASA’s role should be to provide:

Increased research support to universities

Predoctoral NASA fellowships (up to 20K per year, tax-free) that could be
awarded to attract the best students
Long-range funding to properly plan and sustain high-quality research
efforts
Increased access to supercomputers
Frequent visits to NASAfacilitiesby graduate students to participate in
laboratory testing. University facilities in this area are scarce;
students in computational mechanics should get some first-hand experience
in experimental mechanics

118
 REFERENCES

1. Atluri, S.N. and Kathiresan, K., "Influence of Flaw Shapes on Stress Intensity
Factors for Pressure Vessel Surface Flaws and Nozzle Corner Cracks", Journal of
Pressure Vessel Technology, Vol. 102, August 1980, pp. 78-286.

2. Atluri, S.N. (Editor and Contributor to Part 11, Finite Element Methods), Finite
Element Handbook, McGraw-Hill Book Co., 1200 pp. (In Press).

3. Atluri, S.N., "Computational Solid Mechanics (Finite Elements and Boundary

Elements): Present Status and Future Directions", (Keynote Paper) in Z o c . 4th
-- Modeling (Eds: H.M. Hsia, et al.) ,
International Conference on Applied Numerical
pp. 19-37, 1984-

4. O'Donoghue, P.E. and Atluri, S.N., "Control of Dynamic Response of a Continuum

Model of a Large Space Structure", AIAA 26th Structures, Structural Dynamics &
Materials Conference, Orlando, Florida, A G i 1 15-17, 1985.

5. Atluri, S.N., Kathiresan, K., and Kobayashi, A.S., "Three-Dimensional Linear

Fracture Mechanics Analysis by a Displacement-Hybrid Finite-Element Model",
I_ --
Trans. of the 3rd International Conference on Structural Mechanics in Reactor
-
Technology, London, Vol. 5, Part L, Paper No. L-7/3, September 1975, 13 pp.

6. Atluri, S.N. and Kathiresan, K., "On a 3-D 'Singularity Element' for Computation
of Combined Mode Stress Intensities", Advances in Engineering Science, Vol. 1,
November 1976, pp. 267-274.

7. Atluri, S.N., Kathiresan, K., Kobayashi, A.S., and Nakagaki, M., "Inner
Surface Cracks in an Internally Pressurized Cylinder Analyzed by a Three-
Dimensional Dis p 1acement-Hybrid F inite E1ement Me thod", Pressure Vess e1
Technology, Part I1 - Materials and Fabrication, ASME, 1977, pp. 527-533.

8. Atluri, S.N. and Kathiresan, K., "Outer and Inner Surface Flaws in Thick-Walled
Pressure Vessels", Trans. of 4th International Conference on Structural
Mechanics in Reactor Technology, Paper G 5/4, 1977, 13 pp.

9. Atluri, S.N. and Kathiresan, K., "Stress Analysis of Typical Flaws in Aerospace
Structural Components Using 3-D Hybrid Displacement Finite Element Method",
Proceedings of 19th AIAA/ASME/SAE Structures, Structural Dynamics and Materials
Conference, Bethesda, Maryland, 1978, pp. 340-351.

10. Atluri, S.N. and Kathiresan, K., "3-D Analysis of Surface Flaws in Thick-Walled
Reactor Pressure Vessels Using Displacement-Hybrid Finite Element Method",
Nuclear Engineering & Design,"Vol. 51, No. 2, January 1979, pp. 163-176.

11. Tracy, D.M., "Finite Elements for Three-Dimensional Elastic Crack Analysis",
-
Nuclear Engineering 6 Design, Vol. 16, 1977, pp. 282-290.

12. Barsoum, R.S., "On the Use of Isoparametric Finite Elements in Linear Fracture
Mechanics", International Journal-for Numerical Methods-in Engineering, Vol. 10,
NO. 1, 1976, pp. 25-37.

119
13. Newman, J.C., Jr. and Raju, I.S., "Stress-Intensity Factor Equations for Cracks
in Three-Dimensional Finite Bodies Subjected to Tension and Bending Loads''9
(S.N. Atluri, Ed.) Computational Method in the Mechanics of Fracture, North-
Holland Publishing Co. (In Press).

14. Cruse, T.A., "Boundary-Integral Equation Method for Three-Dimensional Elastic

Fracture Mechanics", AFOSR-TR-75-0813, U.S. Air Force, 1975, 52 pp.

15. Heliot, J., Labbens, R.C., and Pellissier-Tannon, A., "Semi-Elliptical Cracks in
a Cylinder Subjected to Stress Gradients", in Fracture Mechanics, ASTM STP 677,
1979, pp. 341-364.

16. Atluri, S.N. and Nishioka, T., "Analytical Solution for Embedded Elliptical
Cracks, and Finite Alternating Method for Elliptical Surface Cracks, Subjected
--
to Arbitrary Loadings", Engineering Fracture Mechanics, Vol. 17, No. 3, 1982,
pp. 247-268.

17. Atluri, S.N. and Nishioka, T., "Integrity Analyses of Surface-Flawed Aircraft
Attachment Lugs: A New, Inexpensive, 3-D Alternating Method", Proceedings of
22nd AIAA/ASME/ASCE/AHS Structures and Material Conference, New Orleans,
Louisiana, 1982, pp. 287-301.

S.N. and Nishioka, T., "Analysis of Surface Flaw in Pressure Vessels by

18. A t l u r i ,
a New 3-Dimensional Alternating Method", Aspects of Fracture Mechanics in
Pressure Vessels and Piping, PVP, Vol. 58, ASME, New York, 1982, pp. 17-37.
Also Journal of Pressure Vessel Technology, ASME Val. 104, 1982, pp. 299-307.

19. Atluri, S.N. and Nishioka, T., "Stress Intensity Factor Variation with Loading
and Crack Depth for Long Flaws of Semi-Elliptical Shape", Advances in Aerospace
Structures, AD Vol. 2, ASME, 1982, pp. 47-50.

20. Atluri, S.N. and Nishioka, T., "An Inexpensive 3-D Finite Element-Alternating
Method €or the Analysis of Surface Flawed Aircraft Structural Components", AIkA
Journal,
- - Vol. 21, No. 5, 1983, pp. 749-758.

21. Atluri, S.N. and Nishioka, T., "Multiple Semi-Elliptical Surface Cracks in
Pressure Vessels", Journal of PressureVessel Technology, Trans. ASME, Vol. 54,
1982, pp. 612-619.

22. Atluri. S.N. and Vijayakumar, K., "An Embedded Elliptical Crack, in an Infinite
Solid,.Subject to A;bi trary Crack-Face Tractions", Journal of Applied Mechanics,
Vol. 103, No. 1, 1981, pp. 88-96.

23. O'Donoghue, P.E., Nishioka, T., and Atluri, S.N., '$Multiple, Coplanar, Embedded
Elliptical Cracks, in an Infinite Solid Subject to Arbitrary Crack Face
Tractions", International Journal of Numerical Methods in Engineering, Vol. 21,
1985, pp. 437-449.

24. Atluri, S.N. and Nishioka, T., "Computational Methods for Three-Dimensional
-
Problems of Fracture",in Computational Methods in the Mechanics of Fracture
(S.N. Atluri, Editor), North-Holland (In Press).

25. O'Donoghue, P.E., Nishioka, T., and Atluri, S.N., "Multiplie Surface Cracks in
Pressure Vessels", Engineering Fracture Mechanics, Vol. 20, No. 3, 1984,
pp. 545-560.

120
I
26. Atluri, S.N., "Computational and Theoretical Studies on Dynamic Fracture Mech-
anics and Three-Dimensional Crack Problems", Keynote Lecture, Proceedings of ICF
I
-

International Symposium on Fracture Mechanics, Science Press, Beijing, 1983,

pp. 1-17.

27. O'Donoghue, P.E., Nishioka, T., and Atluri, S.N., "Analysis of Interaction
Behavior of Surface Flaws in Pressure Vessels", Journal of Pressure Vessels
Technology, Trans. ASME, 1984.

28. Punch, E.F. and Atluri, S.N., Development and Testing of Stable, Invariant,
Isoparametric Curvilinear 2- and 3-D Hybrid Stress Elements", Computer Methods
in Applied Mechanics and Engineering, Vol. 47, 1984, pp. 331-356.

29. Kondoh, K. and Atluri, S.N., "Influence of Local Buckling on Global

Instability: Simplified, Large Deformation, Post-Buckling Analyses of Plane
Trusses", -
Computers and Structures, Vol. 21, No. 4, 1985, pp. 613-628.

30. Kondoh, K. and Atluri, S.N., "A Simplified Finite Element Yethod for Large
Deformation, Post-Buckling Analyses of Large Frame Structures, Using Explicitly
Derived Tangent StifEness Matrices", International Journal of Numerical Methods
-
in Engineering, Vol. 23, No. 1, 1986, pp. 69-90.

31. Kondoh, K., Tanaka, K., and Atluri, S.N., "LSS-NLAP: A Program for Simplified
Nonlinear Analysis of Large Space-Trusses and Frames, Using Explicity Derived
Tangent Stiffnesses, and Accounting for Local Buckling", Report to Air Force
Wright Aeronautical Laborato.ries, 1985, 171 pp, USAF, WPAFB.

32. Tanaka, K., Kondoh, K., and Atluri, S.N., "Instability Analysis of Space Trusses
Using Exact Tangent-Stiffness Matrices" in Finite Element Analysis & Design,
North-Holland, Vol. 1, No. 4, 1985, pp. 291-312.

33. Noor, A.K. and Peters, J.M., "Instability Analysis of Space TL-usseS", Computer
Methods in Applied Mechanical Engineering", Vol. 40, 1983, pp. 199-218.

34. Wood, R.D. and Zienkiewicz, O.C., "Geometrically Nonlinear Finite Element
Analysis of Beams, Frames, Arches, and Axisymmetric Shells", Computers and
Structures, V o l . 7 , 1977, pp. 725-735.
-
35. Williams, F.W., "An Approach t o the Nonlinear Behavior of Members of a Riprid
--Appl. Math.,
Jointed Framework With Finite Deflections", Quant. J. Mech.
Vel. 17, NO. 4, 1964, pp. 451-469.

36. Atluri, S.N. and Nishioka, T., "Stress Analysis of Holes in Angle-Ply Laminates:
An Efficient Assumed Stress 'Special-Hole-Element", Approach and a Simple Esti-
mation Method", Computers and Structures, Vol. 15, November 1982, pp. 135-147.

37. Atluri, S.N. and Nishioka, T., "A Simple Estimation Method of Stress Intensity
Factors for Through Cracks in Angle-Ply Laminates", Engineering Fracture
Mechanics, Vol. 16, No. 4, 1982, pp. 573-583.
38. Rybicki, E.F. and Hopper, A.T., "Analytical Investigation of Stress Concentra-
tions Due to Holes in Fiber Reinforced Plastic Laminated Plates: Three-
Dimensional Models", AFML-TR-73-100, Battelle Columbus Labs, 1973.

121
 ON COMPUTATIONAL SCHEMES FOR GLOBAL-LOCAL
STRESS ANALYSIS

J . N. Reddy
Virginia Polytechnic I n s t i t u t e and S t a t e University
B1 acksburg, V i rgi n i a

1. INTRODUCTION
T h i s paper primarily deals w i t h an overview of global-local s t r e s s analysis
methods and associated d i f f i c u l t i e s and recommendations f o r future research. The
phrase global-local analysis i s understood t o be an analysis i n which some p a r t s of
the domain o r s t r u c t u r e a r e identified, f o r reasons of accurate determination of
s t r e s s e s and displacements or f o r more refined analysis t h a n i n the remaining p a r t s .
The p a r t s o f refined analysis a r e termed local and the remaining p a r t s a r e called
global. Typically local regions a r e s m a l l s i z e compared t o global regions, while
t h e m p u t a t i o n a l e f f o r t can be l a r g e r i n local regions t h a n i n global regions.

2. CONTENTS

T h i s paper i s divided i n t o the following parts:

Mo t i va t i on-P rob1 ems ( p rob ems t h a t motivated global -1 ocal ana y s i s )
Common Features
Focus Problem
Analysis Methods
G1 obal -Local Approaches
Example Problem
Conclusions and Recommendations

3. MOTIVATION PROBLEMS
The following s t r e s s analysis problems, among many others, motivate us t o use
gl obal-1 ocal aproaches:
Free-Edge S t r e s s Concentration i n Laminates
Contact Stress Problems
Impact
Fracture Mechanics
Unbounded-Domai n Problems

123
PRECEDING PAGE B U N K NOT FILMED
 Almost a l l laminated composite s t r u c t u r a l elements have f r e e edges a t t h e
boundaries ( i n c l u d i n g h o l e s ) o f t h e elements. It i s w e l l known t h a t t h e t r a n s v e r s e
normal and shear stresses a r e very l a r g e a t t h e edges (more p r e c i s e l y , t h e s t r e s s e s
a r e l a r g e w i t h i n a d i s t a n c e o f t h e o r d e r o f t h i c k n e s s o f t h e l a m i n a t e from t h e f r e e
edge). While t h e c l a s s i c a l laminate t h e o r y i s adequate t o describe t h e behavior o f
t h e laminate everywhere except i n t h e "boundary l a y e r " i n which t h e t r a n s v e r s e normal
and shear stresses a r e l a r g e , a r e f i n e d theory, o f t e n t h e 3-D e l a s t i c i t y theory, i s
needed t o describe t h e s t a t e o f s t r e s s near t h e edges.

Contact s t r e s s problems ( f o r example, b o l t e d and bonded j o i n t s , t i r e c o n t a c t ,

and metal-forming problems) r e q u i r e t h e use o f a s p e c i a l theory t h a t accounts f o r
a p p r o p r i a t e c o n s t i t u t i v e laws and f r i c t i o n and a l l o w s f o r s l i p , s l i d e , and s e p a r a t i o n
o f t h e mating p a r t s i n t h e c o n t a c t regions. Elsewhere, a p p r o p r i a t e e l a s t i c i t y t h e o r y
can be used.

Impact o f two s o l i d bodies can be modelled by t h e use o f one t h e o r y i n t h e

v i c i n i t y o f c o n t a c t and by another theory elsewhere. Since t h e s t r e s s e s a r e much
l a r g e r i n t h e c o n t a c t r e g i o n than elsewhere, more r e f i n e d t h e o r y and a n a l y s i s a r e
r e q u i r e d i n c o n t a c t regions. Of course, t h e theory and a n a l y s i s used depend on t h e
t y p e o f s t r u c t u r e , l o a d i ng and deformation.

S t r u c t u r e s c o n t a i n i n g cracks, whether formed d u r i n g manufacturing o r s e r v i c e ,

r e q u i r e s p e c i a l t r e a t m e n t o f s t r e s s f i e l d s around cracks, o f t e n u s i n g 3-D s t r e s s
analyses and/or n o n l i n e a r f r a c t u r e mechanics t h e o r i e s , w h i l e t h e l i n e a r e l a s t i c
f r a c t u r e mechanics t h e o r y i s adequate away from t h e cracks.

Problems i n v o l v i n g unbounded regions ( f o r example, s o i l mechanics and earthquake

engineering) a r e by t h e i r n a t u r e d i v i d e d i n t o l o c a l and g l o b a l regions. Global
regions, i n theory, can be i n f i n i t e b u t i n p r a c t i c e they are f i n i t e l y l a r g e , and l e s s
r e f i n e d theory and/or a n a l y s i s i s used t o determine t h e s t r e s s f i e l d and o t h e r
p e r t i n e n t information.

4. COMMON FEATURES

The m o t i v a t i n g problems 1i s t e d p r e v i o u s l y share c e r t a i n common p h y s i c a l f e a t u r e s

t h a t a r e s i g n i f i c a n t from t h e m o d e l l i n g and a n a l y s i s p o i n t s o f view. The f o l l o w i n g
l i s t p r o v i d e s some o f these f e a t u r e s :

S t r e s s Concentration ( 1arge 1oca1 g r a d i e n t s )

Three-Dimensional S t a t e o f S t r e s s

Large R o t a t i o n s / S t r a i n s

. Local D i s c o n t i n u i t i e s ( h o l e s , d i s c o n t i nuous f i b e r s , e t c .

. Material Nonlinearities (nonlinear e l a s t i c i t y , p l a s t i c i t y , etc.)

A g l o b a l - l o c a l s t r e s s a n a l y s i s should account f o r a l l f e a t u r e s t h a t a r e p r e s e n t
i n t h e problem. O f course, some of those f e a t u r e s are n o t t o be i n c l u d e d i n t h e
g l o b a l model.

124
 5. FOCUS PROBLEM

The focus problem i d e n t i f i e d by NASA Langley Research Center i s a blade-

s t i f f e n e d panel w i t h a discontinuous s t i f f e n e r . The problem has the following
p hy s i ca 1 f ea t u res :
Geometric Discontinuities
Local S t r e s s Gradients
Eccentric Loading
Large D i spl acements
Free Edges

We shall return t o this problem l a t e r t o discuss the global-local analysis

approach.

6. ANALYSIS METHODS

6.1 COMMON APPROACHES

The commonly used analysis methods f o r structural problems a r e

Classical ( o r Analytical) Methods
C1 assical Variational Methods
F i n i te-Di fference Method
F i n i te-El ement Method

Boundary Element Method

Some noted advantages and disadvantages of these methods a r e outlined next.

6.2 CLASSICAL AND VARIATIONAL METHODS

The c l a s s i c a l method of solving problems exactly i s the best there is. However,
most practical problems ( w h i c h have irregular geometries, anisotropic materials, d i s -
c o n t i n u i t i e s , geometric and/or material nonlinearities, etc.) do not admit exact
solutions by the c l a s s i c a l approach.
The c l a s s i c a l variational (e.g., Ritz, Galerkin and weighted-residual ) methods
y i e l d continuous solutions throughout the domain, g i v i n g h i g h resolution of displace-
ments and s t r e s s e s . They a r e computationally e f f i c i e n t f o r a given problem. For a
given order of approximation, previously computed ( f o r lower order approximation)
matrix c o e f f i c i e n t s can be used. These methods, however, have two major short-
comings: ( i ) the approximation functions a r e not easy and a r e often impossible t o

125
 c o n s t r u c t f o r most p r a c t i c a l s t r u c t u r a l problems; and ( i i ) t h e v a r i a t i o n a l methods
cannot be implemented on a computer f o r t h e a n a l y s i s o f a c l a s s o f problems because
t h e r e s u l t i n g a1 gebraic equations depend on t h e approximat5%-?h3ionsy which i n
t u r n depend on a s p e c i f i c problem.

6.3 FINITE-DIFFERENCE METHOD (FDM)

The f i n i t e d i f f e r e n c e method i s simple i n f o r m u l a t i o n (based on t h e representa-

. t i o n o f d e r i v a t i v e s o f a f u n c t i o n i n terms o f a f i n i t e T a y l o r ' s s e r i e s expansion) and
easy t o implement on t h e computer. The method dominated t h e f i e l d o f numerical
methods u n t i l t h e s i x t i e s , when t h e f i n i t e - e l e m e n t method gained p o p u l a r i t y , espe-
c i a l l y i n s o l i d and s t r u c t u r a l mechanics. The disadvantages o f t h e f i n i t e - d i f f e r e n c e
method include: d i f f i c u l t y i n r e p r e s e n t i n g complex geometries, i n e x a c t r e p r e s e n t a t i o n
o f boundary c o n d i t i o n s on n o n - s t r a i g h t boundaries, and d i f f i c u l t y i n developing
h i g h e r - o r d e r approximations. Because o f these d i f f i c u l t i e s the method does n o t l e n d
i t s e l f f o r general-purpose code development. The method i s seldom used f o r s p a t i a l
approximations i n s t r u c t u r a l mechanics problems.

6.4 FINITE-ELEMENT METHOD

The f in i t e - e l ement method overcomes t h e shortcomings o f t h e c l a s s i c a l v a r i a-

t i o n a l methods. T h i s approach i s systematic (modular) and n a t u r a l and a l l o w s an
a c c u r a t e r e p r e s e n t a t i o n o f complex geometries. Higher-order approximations a r e easy
t o use w i t h o u t changing t h e modular s t r u c t u r e o f t h e approach. The method i s i d e a l l y
s u i t e d f o r general purpose and computer program development. The disadvantages,
compared t o o t h e r computing methods, a r e t h e l a r g e f o r m u l a t i v e and computational
e f f o r t s . The f i n i t e - e l e m e n t method i s t h e most f r e q u e n t l y used numerical method i n
s t r u c t u r a l a n a l y s i s . It i s now a key component o f any mechanical CAD/CAM system.

6.5 BOUNDARY ELEMENT METHOD (BEM)

The f i n i t e - d i f f e r e n c e and f i n i t e - e l e m e n t methods can be c l a s s i f i e d as domain

methods because they i n v o l v e approximations o f t h e e n t i r e domain. The boundary
element method, a l s o known as t h e boundary i n t e g r a l method (BIE), seeks approxima-
t i o n s o n l y on t h e boundary o f t h e domain by c o n v e r t i n g t h e governing d i f f e r e n t i a l
equations t o i n t e g r a l s over t h e boundary o f t h e domain. The d i m e n s i o n a l i t y o f t h e
problem i s thereby e f f e c t i v e l y reduced by one. Because t h e i n t e r i o r o f t h e domain i s
1
n o t approximated, t h e computational t i m e i n v o l v e d i s l e s s (BEM/FEM - where nxn i s
PI
n'
t h e f i n i t e - e l e m e n t mesh). The BEM o f f e r s continuous i n t e r i o r m o d e l l i n g w i t h i n t h e
s o l u t i on domain, g i v i ng h i g h r e s o l u t i o n o f d i s p l acements and stresses. The method i s
unsui tab1 e f o r p r o b l ems r e q u i r i ng i n f o r m a t i o n a t a 1arge number o f i n t e r n a l p o i n t s .
A p p l i c a t i o n o f BEM t o n o n l i n e a r problems and problems w i t h d i s c o n t i n u i t i e s i s n o t
f u l l y established.

126
 6.6 CONCLUSIONS

I n conclusion, t h e general a p p l i c a t i o n o f t h e f i n i t e - e ,?merit metl,oc t o s t r u c -

t u r a l problems i s unmatched t o date. S t r u c t u r a l problems, because o f t h e modular
n a t u r e o f FEM, a r e b e t t e r modelled by FEM. A s u i t a b l e combination o f FEM and REM can
be advantageous i n some problems (e.g. , unbounded-domain problems).

7. GLOBAL-LOCAL APPROACHES

7.1 MODELS AND METHODS

I n f o r m u l a t i n g a g i v e n problem, e i t h e r t h e same t h e o r y i s used throughout or a

r e f i n e d t h e o r y i s used l o c a l l y and a l e s s r e f i n e d t h e o r y g l o b a l l y .

The g l o b a l - l o c a l a n a l y s i s methods can be FEM throughout, FEM and c l a s s i c a l

s o l u t i o n , o r BEM and FEM. When t h e same theory and FEM a r e used throughout, i t i s
understood t h a t l o c a l l y s p e c i a l elements are used (e.g., f r i c t i o n element, i n t e r f a c e
element w i t h s l i d i n g , o r elements which a l l o w opening and c l o s i n g along element
i n t e r f a c e s ) . C l a s s i c a l s o l u t i o n s are a v a i l a b l e , f o r example, f o r i n f i n i t e p l a t e s
w i t h holes. The s o l u t i o n i s n o t v a l i d f a r away from t h e h o l e i f t h e p l a t e i s l o n g
b u t n o t i n f i n i t e . I n such cases, t h e FEM can be used g l o b a l l y and t h e c l a s s i c a l
s o l u t i o n can be used l o c a l l y . F o r s o i l mechanics and earthquake e n g i n e e r i n g prob-
lems, a combination o f BEM and FEM proves c o m p u t a t i o n a l l y e f f i c i e n t . In some s i t u a -
t i o n s experimental methods g l o b a l l y and computational methods l o c a l l y a r e
recommended.

7.2 SOME EXAMPLES

Some exampl e p r o b l ems t h a t r e q u i r e g1 obal-1 oca1 a n a l y s i s are 1is t e d here.

Free-edge s t r e s s a n a l y s i s o f laminates

Contact s t r e s s problems
Stress analysis o f structures w i t h d i s c o n t i n u i t i e s

A b l a d e - s t i f f e n e d panel w i t h a discontinuous s t i f f e n e r - t h e focus

p rob1em

As mentioned e a r l i e r , free-edge s t r e s s problem r e q u i r e s a r e f i n e d t h e o r y near

f r e e edges. F o r example, t h e c l a s s i c a l l a m i n a t e theory g l o b a l l y and . e i t h e r quasi-30
o r f u l l 3-0 t h e o r y l o c a l l y (depending on t h e l a m i n a t i o n scheme, geometry and l o a d i n g )
can be used t o analyze t h e problem. The problem w i l l be discussed i n more d e t a i l
.
1a t e r

I n b o l t e d j o i n t problems, an experimental technique such as t h e M o i r 6

in t e r f e r o m e t r y can be used t o determi ne t h e s u r f a c e displacements (and hence s t r a i n s
and s t r e s s e s ) and t h e f i n i t e - e l e m e n t method can be used t o determine t h e i n t e r i o r
d i s p l acement and s t r e s s f i e l ds.

127
 In the s t r e s s analysis of plates w i t h holes, one can use a laminate theory away
from holes and 3-D e l a s t i c i t y theory around the hole, and use the finite-element
method t o model the e n t i r e problem.
In the case of the focus problem, which has a l l features t h a t a r e present i n the
examples discussed above, the global theory should be a shear deformation p l a t e
theory (2D) w i t h the Von Karmen geometric non-linearity and the local theory can be
the f u l l y 3D laminate theory. The finite-element method should be used throughout.
When FEM i s used l o c a l l y , the f u l l y 3D elements o r 3D degenerate elements can be
used.

7.3 DIFFICULTIES

In using global-local approaches, we face some d i f f i c u l t i e s . Some of these a r e

Interfacing between regions
Interfacing between methods
Selection of regions
Changing regions and interfaces

When the finite-element method i s used, the elements used globally and l o c a l l y
can be d i f f e r e n t . Then i t i s important t o have compatibility of the nodal degrees of
freedom a t the i n t e r f a c e of the elements. A special i n t e r f a c e element m i g h t be
needed i n some s i t u a t i o n s . When two d i f f e r e n t methods a r e used, the unknowns i n the
two methods should be the same. Selection of the local and global regions depends on
the physical features and accuracy desired. In some cases, the regions m i g h t have t o
be determined only a f t e r a preliminary analysis. The global and local regions can
change d u r i n g the history of deformation/ loading. For example, i n e l a s t i c - p l a s t i c
analysis, the p l a s t i c zones a r e unknown a p r i o r i and they change w i t h loading.

8. EXAMPLE PROBLEM
Here we b r i e f l y discuss the free-edge s t r e s s problem i n symmetric laminates.
Figure 1 shows the laminate geometry, loading, and the domain modelled. Because of
the assumed symmetry of the lamination about the midplane and the constant s t r a i n i n g
along the x-axis, the displacement f i e l d can be approximated (ref. 1) as

I u = uox + u ( y , z )
v = V(y,z)
w = W(y,z)
where Uo i s a constant, K .

128
 The displacement f i e l d i s t h r e e dimensional b u t i t leads, when s u b s t i t u t e d i n t o
t h e N a v i e r equations o f e q u i l i b r i u m , t o t h r e e p a r t i a l d i f f e r e n t i a l equations i n two
independent v a r i a b l e s , y and z.

I t i s w e l l known t h a t t h e t r a n s v e r s e normal s t r e s s , uZ, i s very l a r g e (un-

bounded) near t h e f r e e edge. To reduce i t s magnitude a cap i s used on t h e f r e e
edge. The e f f e c t o f t h e cap on t h e s t r e s s d i s t r i b u t i o n uZ i s i n v e s t i g a t e d . The
f in i t e - e l ement method w i t h t h e four-node b i 1inear r e c t a n g u l a r e l ement is used t o
model t h e computational domain. A r e f i n e d mesh i s used near t h e free edge and i n t h e
cap.

F i g u r e s 2 and 3 show t h e d i s t r i b u t i o n of t h e t r a n s v e r s e normal s t r e s s uZ a l o n g

t h e w i d t h o f t h e l a m i n a t e f o r [Oo/9OoIsand [45"/-45"Is laminates, r e s p e c t i v e l y ,
( E 1 = 137.89 GPa, E2 = E3 = 14.48 GPa, 612 = 613 = 623 = 5.86 GPa,
~ 1 2= v i 3 = 9 3 = 0.21). R e s u l t s f o r b o t h capped and uncapped laminates a r e
presented ( f o r k = 0.001, b = 25.4 cm, h = 2.54 cm and t h i c k n e s s o f t h e cap,
t = 0.08 cm). We observe t h a t t h e s t r e s s i s e s s e n t i a l l y zero i n s i d e t h e l a m i n a t e b u t
has q u i t e a l a r g e magnitude w i t h i n a d i s t a n c e o f y / b = 0.1 (one-tenth o f t h e w i d t h )
from t h e f r e e edge. Hence a l a m i n a t e t h e o r y i s s u f f i c i e n t t o model t h e i n t e r i o r ,
w h i l e t h e quasi-3D can be used t o model t h e free-edge s t r e s s f i e l d . The e f f e c t o f
t h e free-edge r e i n f o r c e m e n t ( i .e. , cap) on t h e s t r e s s magnitude i s s i g n i f i c a n t ; t h e
magnitude i s reduced t o l e s s than o n e - t h i r d o f t h a t w i t h o u t cap.

F o r a more d e t a i l e d and complete s t r e s s d i s t r i b u t i o n near t h e f r e e edge o f a

more general l a m i n a t e (e.g., w i t h o u t symmetry about t h e midplane), a three-
dimensional model i s needed.

9. CONCLUSIONS AND RECOMMENDATIONS

9.1 AREAS NEEDING SUPPORT

A r e v i e w o f t h e l i t e r a t u r e shows t h e r e a r e very few cases o f g l o b a l - l o c a l

analyses o f s t r u c t u r a l problems i n v o l v i n g t h e " p h y s i c a l features'' discussed e a r l i e r .
It i s recommended that the f o l l o w i n g areas o f g l o b a l - l o c a l approaches be
investigated:

G1 obal-1 oca1 a n a l y s i s o f problems w i t h "common f e a t u r e s " out1 ined

e a r l ie r
I n v e s t i g a t i o n and development o f in t e r f a c e elements

F e a s i b i l i t y o f BEM as a computational t o o l f o r n o n l i n e a r problems and

i t s i n t e r f a c e w i t h FEM

Development o f adaptive mesh refinements and time-stepping a l g o r i t h m s

E x p l o i t a t i o n of t h e v e c t o r and p a r a l l e l processor computers f o r

e f f i c i e n t structural analysis

129
 0 Finite-element calculations

S o l u t i o n o f equations

E i genval ue computations

The use o f p a r a l l e l processors can d i c t a t e t h e s o l u t i o n procedures, f o r

example, i t e r a t i v e methods over one-step methods.

9.2 NASA's INVOLVEMENT

NASA (CSM) shoul d be invol ved i n t h e g l obal-1 ocal a n a l y s i s devel opment because
o f the tremendous impact t h i s f i e l d has on computational mechanics a p p l i e d t o space
s t r u c t u r e s . I n p a r t i c u l ar, NASA shoul d undertake t h e f o l l owi ng t a s k s i n t h e g l o b a l -
1ocal a n a l y s i s area:

Support i n d i v i d u a l grants (as opposed t o l a r g e group g r a n t s )

C o l l a b o r a t e w i t h u n i v e r s i t y f a c u l t y and graduate students by i d e n t i f y i n g

s p e c i f i c problem areas and p r o v i d i n g computational time and s c i e n t i f i c
advice
Give graduate s t u d e n t r e s i d e n t s h i p s , d u r i n g which students spend a few
weeks (perhaps t h e summers) a t NASA

Conduct workshops (say, once i n two y e a r s ) t o b r i n g t h e l a t e s t

developments f o r c r i t i c a l e v a l u a t i o n and t o s e t f u t u r e d i r e c t i o n s .

ACKNOWLEDGMENTS

T h i s paper was prepared v i a t h e support o f t h e S t r u c t u r a l Mechanics Branch o f

NASA Langley Research Center (NAG-1-459 1.

REFERENCE

H e y l i g e r , P. R. and Reddy, J. N., "Reduction o f Free-Edge S t r e s s C o n c e n t r a t i o n i n

Symmetric L a m i n a t e s , " V i r g i n i a P o l y t e c h n i c I n s t i t u t e and S t a t e U n i v e r s i t y ,
Blacksburg, VA, CCMS Report 85-01, January 1985.

BIBLIOGRAPHY

Arnold, D. N., Babuska, I . and Osborn, J . , " F i n i t e Element Methods: P r i n c i p l e s f o r

T h e i r Selection," Computer Methods i n Appl. Mech. ll Engng., Vol. 45, pp. 57-96,
1984.

A t l u r i , S. N. GALLAGHER, R. H., and Zienkiewicz, 0. C., H y b r i d and Mixed F i n i t e

Element Methods, Wiley, Chichester (U.K.), 1983.

130
 Banerjee, P. K . and B u t t e r f i e l d , R . , Boundary Element Methods .in Engineering Science,
McGraw-Hill, London, 1981.
Brebbia, C. A., The Boundary Element Method f o r Enqineers, Pentech Press, London,
1980.
Buzbee, B. L. and McKnight, A. L . , "Impact of Vector Processors on Algorithms," Los
Alamos S c i e n t i f i c Laboratory, Los Alamos, NM 87545, LA-8123-MS, 1979.
Calahan, D. A , , " E x p l o i t i n g P a r a l l e l i s m i n Designing A1 gorithms," Basic Quest. Des.
Theory, (Symposium, New York, 19741, pp. 285-302, 1974.
Chen, S. C., Kuck, D . J . and Sameh, A. H . , "Practical P a r a l l e l Band T r i a n g u l a r System
S o l v e r s , " ACM Trans. Math. Software, Vol. 4, pp. 270-77, 1978.
Conrad, V. and Wallach, Y . , " I t e r a t i v e S o l u t i o n of L i n e a r Equations on a P a r a l l e l
Processor System," IEEE Trans. Computers, Vol. 26, pp. 838-47, 1977.
Csanky, L., "Fast P a r a l l e l Matrix Inversion Algorithms," SIAM J . Comput., Vol. 5,
pp. 618-23, 1976.

Durnier, A . , "A Theory of Friction," I n t . J . S o l i d s 81 S t r u c t . , Vol. 20, No. 7,

pp. 637-647, 1984.
Fenner, R. T., "The Boundary I n t e g r a l Equation (Boundary E l ement) Method i n
Engineering S t r e s s Analysis," J . S t r a i n Analysis, Vol. 18, No. 4, 1983.

Feng, T. , ( e d . ) , P a r a l l e l Processing, Springer-Verlag, New York, 1975.

Gadala, M. S. and Oyavas, G. A E . , "Numerical S o l u t i o n s of Nonlinear Problems of
Continua-I (Survey of Formulation Methods and S o l u t i o n Techniques,") Computers and
S t r u c t u r e s , Vol. 19, No. 5/6, pp. 865-877, 1984.
Heller, D., "A Survey of Parallel Algorithms i n Numerical Linear Algebra," SIAM Rev.,
Vol. 20, pp. 740-77, 1978.
Knight, J r . , N . F. and S t a r n e s , J r . , J . H . , 'Postbuckling Behavior o f S e l e c t e d Curved
S t i f f e n e d Graphite-Epoxy Panels Loaded i n Axial Compression," 26th
AIAA/ASME/ASCE/AHS S t r u c t u r e s , S t r u c t u r a l Dynamics and Materiamonference, AIAA
P a p e r No. 85-0168 -CP , Orlando, Fl., 1985.

Matthies, H . and S t r a n g , G . , "The S o l u t i o n of Nonlinear Finite Element Equations,"

I n t . J . Numer. Meth. i n Enqnq., Vol. 14, pp. 1613-1626, 1979.

O r t e g a , J . M. and Voigt, R. G . , "Solution o f P a r t i a l D i f f e r e n t i a l Equations on Vector

and P a r a l l e l Computers," NASA CR-172500, Jan. 1985.

Pagano, N . J . and Soni, S. R . , "Global-Local Laminate Variational Model , ' I I n t . J .

S o l i d s and S t r u c t . , Vol. 19, No. 1, pp. 207-228, 1981.

P i p e s , R. B. and Pagano, N . J . , " I n t e r l a m i n a r S t r e s s e s Composite Laminates Under

Uniform Axial Extension," J . Comp. Materials, Vol. 4, pp. 538-548, 1970.

131
 Reddy, J . N . , An Introduction t o the Finite Element Method, McGraw-Hill, New York,
1984.
Reddy, J . N . , Energy and Variational Methods in Applied Mechanics, Wiley, New York,
1984.
Rybicki, E. F., "Approximate Three-Dimensional Solutions for Symmetric Laminates
Under Inplane Loading," J . Comp. Materials, Vol. 5, pp. 354-360, 1971.
Traub, J . F. ( e d . ) , Complexity o f Sequential and Parallel Numerical Algorithms,
Academic Press, New York, 1973.
Whitcomb, J . D., Raju, L . S. and Goree, J . G . , "Reliability o f the F i n i t e Element
Method f o r Calculating Free-Edge Stresses i n Composite Lami nates ,'I Computers and
Structures, Vol. 15, No. 1, pp. 23-37, 1982.
I Zienkiewicz, 0. C . , Gago, J . P . , De, S. R . , and Kelly, D. W., "The Hierarchical
Concept in Finite Element Analysis," Computers and Structures, Vol. 16, No. 1-4,
pp. 53-65, 1983.

132
 t Domain modelled

2h Y

?2b - 4-
Figure 1. Laminate geometry and loading (cap i s not shown).

b"
8
v)
1.0-
c
L

-
v)

Y
(d
-b-
E0
capped
Q,
2 0.0-
Q,
>
v)
c
2 uncapped
!-
\ "
0,8 0.85
I
0.9
" " " " '
0,s 1
YIb

Figure 2. D i s t r i b u t i o n o f t h e transverse normal

s t r e s s along the width o f a cross-ply laminate
co 0/90° 1s.

133
 1.04

I ' " ' I ' ' I ' Z "

0,8 0.85 r3,9 0,95 1,o

Y/b
F i g u r e 3. D i s t r i b u t i o n o f the transverse normal stress (uz)
along the width ( y / b ) o f the laminate C45O/-45"Is.

134
 GLOBAL FUNCTIONS I N GLOBAL-LOCAL FINITE-ELEMENT ANALYSIS

OF
LOCALIZED STRESSES IN PRISMATIC STRUCTURES

Stanley B. Dong
University o f California
Los Angeles, C a l i f o r n i a

Abstract

An i m p o r t a n t c o n s i d e r a t i o n i n t h e global l o c a l f i n i t e - e l ement method

(GLFEM) i s t h e a v a i l a b i l i t y o f global f u n c t i o n s f o r t h e g i v e n problem.
The r o l e and mathematical requi rements o f these g l obal f u n c t i o n s i n a GLFEM
a n a l y s i s o f 1oca1 i z e d s t r e s s s t a t e s i n p r i s n a t i c s t r u c t u r e s a r e discussed.
A method i s described f o r determining these global functions. Underlying
t h i s method a r e theorems due t o Toupin and Knowles on s t r a i n energy decay
rates, which a r e re1 ated t o a q u a n t i t a t i v e expression o f Sai nt-Venantf s
prlnciple. It i s mentioned t h a t a mathematically complete s e t o f global
f u n c t i o n s can be generated, so t h a t any a r b i t r a r y i n t e r f a c e c o n d i t i o n
between t h e f i n i t e element and global subregions can be represented.
Convergence t o t h e t r u e behavior can be achieved w i t h i n c r e a s i n g global
f u n c t i o n s and f i n i t e - e l ement degrees o f freedom. Speci f i c a t t e n t i on i s
devoted t o mathematically two-dimensional and three-dimensional p r i s m a t i c
structures. Comments a r e o f f e r e d on t h e GLFEM a n a l y s i s o f NASA f l a t panel
w i t h a discontinuous s t i f f e n e r . Methods f o r determining global f u n c t i o n s
f o r o t h e r e f f e c t s a r e a l s o i n d i c a t e d , such as steady-state dynamics and
bodies under i n i t i a l s t r e s s .

135
 Introduction

The f i n i t e - e l ement method (FEM) has r e v o l uti onized s t r u c t u r a l and

s t r e s s analyses i n t h e l a s t q u a r t e r o f t h i s century. I t s impact has been
widespread, even extending beyond t h e preserve o f s t r u c t u r a l engineers t o
other fields. Although E M i s acknowledged a s an extremely powerful
model i n g technique, t h e a n a l y s i s community w i t h i t s c o l l e c t i v e experience
w i l l a d m i t t h a t i t i s n o t t h e q u i n t e s s e n t i a l technique. There a r e problems
n o t w e l l s u i t e d t o FEM t h a t r e s u l t i n clunsy, i n e f f e c t i v e and c o s t l y
mathematical models. Examples can be c i t e d fran problems i n v o l v i n g s t r e s s
s i ngul a r i ti es and in f i n i t e domai ns. To o b v i a t e t h e d i f f i c u l ti es,
m o d i f i c a t i o n s t o FEM have been explored. One a l t e r n a t e approach which
bodes considerable pranise i s t h e so-called Global-Local Finite-Element
Method (GLFEM).

GLFEM u t i l i z e s both conventi onal f i n i t e e l ements and c l assi c a l R i t z

f u n c t i o n s i n t h e modeling process. T h e i r r e s p e c t i v e r o l e s a r e r e a d i l y
apparent; f i n i t e elements work w e l l i n r e g i o n s where complicated geanetry
and inhomogeneous m a t e r i a l c h a r a c t e r i z a t i o n s p r e v a i l , and R i t z f u n c t i ons,
h e r e i n a f t e r r e f e r r e d t o as global f u n c t i o n s i n GLFEM, enable t h e behavior
o t h e r r e g i o n s t o be represented a c c u r a t e l y and e f f i c i e n t l y . A t t h i s stage
o f development, GLFEM can be assessed t o be i n i t s m a t u r i n g phase. It i s
o f good lineage, has already e x h i b i t e d an enhanced c a p a b i l i t y above E M i n
c e r t a i n problems, and p r a n i s e s e f f e c t i v e n e s s i n o t h e r classes o f problems
upon i t s f u l l development.

Herein, GLFEM as appl i e d t o t h e a n a l y s i s o f l o c a l i z e d s t r e s s e s i n

p r i s m a t i c s t r u c t u r e s i s discussed. F i rst, t h e essence o f GLFEM and
v a r i o u s GLFEM modeling l a y o u t s a r e sunmarized. A b r i e f review o f some
problems t h a t have been s u c c e s s f u l l y analyzed by GLFEM i s given. Then,
t h e main theme re1 a t i n g t o GLFEM a n a l y s i s o f l o c a l i z e d s t r e s s s t a t e s i s
addressed. P r i s m a t i c s t r u c t u r e s t h a t can be described mathematically
by two s p a t i a l v a r i a b l e s a r e discussed first. A t t e n t i o n i s devoted t o
t h e g1 obal f u n c t i o n s r t h e i r devel opment and the1 r r o l e s i n t h e present
s e t t i n g . Then, three-dimensional s t r u c t u r e s a r e considered, w f t h
reference t o t h e NASA example problem, where an o u t l i n e o f a method o f
a t t a c k i s given. Last, comments on t h e a n a l y s i s o f l o c a l i z e d s t r e s s e s
i n v o l v i n g steady-state dynamic e f f e c t s as w e l l as o t h e r c o n d i t i o n s a r e
given.

Basi c Concepts o f GLFEM and t h e Various Mesh Conf i g u r a t i ons

H a m i l t o n ' s p r i nci ple, or a1 t e r n a t i v e l y t h e theorem o f minimun

p o t e n t i a l energy when no i n e r t i a l e f f e c t s a r e present, may be considered
as t h e b a s i s f o r generating GLFEM equations. The theory and v a r i a t i o n a l
d e r i v a t i o n of these equations may be found i n Ref. Cl,pp.451-4741. Also
i n c l u d e d t h e r e i n i s a survey of GLFEM c o n t r i b u t i o n s t o t h e l i t e r a t u r e up
t o 1982.

As noted e a r l i e r , t h e technique u t i l i z e s f i n i t e - e l e m e n t model i n g w i t h

c l a s s i c a l R i t z approximations simultaneously. It enjoys t h e advantages o f

136
more v e r s a t i l e model Ing capabil I t i e s with s u b s t a n t i a l l y fewer degrees o f
freedom. V a r i ous gl obal / l o c a l model i n g conf i g u r a t i ons are 111ustrated 1n
Fig. 1. Figs. l a and I f represent, respectively, t h e c l a s s i c a l R l t z and
f i n 1te-el ement conf I g u r a t i ons. The others are possi b l e GLFEM mesh 1ayouts.
I n a given problem, t h e model i n g may take t h e form of any one of these con-
f i g u r a t i o n s or a combination o f two or more of them f o r various subregions.
An important key i s t h e enforcement of k i nemati c 1nter-reg1 onal c o n t i nul t y
between various global and 1oca1 subregions by means o f constrai n t equa-
tions. I n problems on l o c a l ized s t r e s s states, only t h e Fig. IC conf igura-
t i o n w i l l be usedr where f i n i t e elements exclusively are used i n one sub-
region and global functions i n t h e other. MOrewer, t h e global subregion
may be i n f i n i t e i n extent.

The governing m a t r i x equations i n a GLFEM analysis have t h e form:

where 161 denotes t h e finite-element degrees o f freedom and {SI contains

t h e array of general ized coordi nates associ ated w i t h t h e g1 obal f u n c t i ons.
I n Eq. (11, CKgg3 CMg 1 8 CK,I and CMgalrefer t o t h e global and l o c 1
9
s t i f f n e s s and mass m a t r ces o f t h e system. The matrices CKagl = CKggl ?
= CMgg IT represent g l obal -local coup1 i n g f ran imposi ng k i nemati c
a t 1n t e r f ace (s 1 between subregi on(s 1. D e t a i l s on t h e formati on
o f these matrices may be found i n Ref. C11.

It i s mentioned t h a t GLFEM v a r i a n t s are possible, which do not lead

t o t h e same s e t o f governing equations as Eq. (1). These v a r i a n t s contain
t h e spi r i t o f GLFEM and employ t h e model i n g conf i g u r a t i ons shown i n Fi g. 1;
however, t h e method o f enforcing inter-regional c o n t i n u i t y may d i f f e r . An
appl i c a t i on concerned w i t h e l a s t i c wave scatter1 ng w i l l I 1 1u s t r a t e one such
v a r i ant.

Another key p o i n t I n GLFEM i s t h e a v a i l a b i l i t y o f an appropriate set

of global functions f o r a given problem or a class o f problems. The
accuracy and effectiveness o f t h e method a r e dependent upon t h e q u a l i t y
of t h e global functions. The cfioice of these global functions f o r t h e
analysis of l c c a l i z e d stresses i n p r i m a t i c s t r u c t u r e s and t h e i r method
o f d e r i v a t i o n w i l l be discussed i n what follows. It w i l l become.apparent
why these global functions, together w i t t i the f i n l t e - e l e n e n t model o f

.
the subregion t h a t contains t h e l o c a l i z e d stresses, w i l l lead t o a
super1 or model

Sane Examples o f Global Functions f o r GLFEM

Two areas i d e a l l y suited t o GLFEM a r e f r a c t u r e mechanics and I n f i n i t e

137
and/or s e m i - i n f i n i t e domain problems. Much has been published on v a r i o u s
aspects o f f r a c t u r e mechanics problems. Many nunerical methods have been
used, many f a l l i n g w i t h i n a G L E M c l a s s i f i c a t i o n or i t s v a r i a n t s . The
global subregion model u s u a l l y takes t h e form o f special crack t i p
elements, where t h e s i n g u l a r s t r e s s f i e l d i s i n c o r p o r a t e d i n t o t h e
s t i f f n e s s m a t r i x . These e l m e n t s a r e w e l l known. No f u r t h e r e l a b o r a t i o n
on t h i s s u b j e c t w i l l be g i v e n here. E l a s t o s t a t i c analyses o f half-space
problems, wherein t h e f a r f i e l d behavior i s represented by global f u n c t i o n s
( f o r example, Boussinesq or C e r r u t i s o l u t i o n s ) , have a l s o met w i t h
considerable success. A nunber o f references on both o f these s u b j e c t s
may be found i n Ref. C11.

Herei n, two r e c e n t GLFEM appl i c a t i o n s a r e mentioned t o emphasize t h e

r o l e s o f t h e global f u n c t i o n s and t h e i r m a t h m a t i c a l s u i t a b i l i t y . They a r e
concerned w i t h (1) steady-state e l a s t i c wave s c a t t e r i n g by axisymmetric
o b j e c t s embedded i n an i n f i n i t e i s o t r o p i c medium and (2) steady-state
s o i l - s t r u c t u r e i n t e r a c t i o n i n v o l v i n g an axisymmetric s t r u c t u r e occupying
some l o c a l e i n a s e m i - i n f i n i t e mediun. The f e a t u r e o f note i s t h a t these
global f u n c t i o n s c o n s t i t u t e a complete s e t o f e i g e n f u n c t i o n s and have t h e
capabil i t y o f mathematically r e p r e s e n t i ng an a r b i t r a r y s c a t t e r e d f i e l d t o
any g i v e n accuracy. Hence, t h e t r u e behavior i n t h e f a r f i e l d can be
ach iwed.

I n Fig. 2 i s shown an el a s t i c , axisymmetric i n c l u s l o n embedded i n an

e l a s t i c , i s o t r o p i c mediun. Because f i n i t e elements a r e used f o r the o b j e c t
i t may have inhanogeneous, o r t h o t r o p i c p r o p e r t i e s . The f i n i t e - e l ement
subregion i n c l u d e s t h i s o b j e c t and a p o r t i o n o f t h e surrounding mediun.
For convenience i n t h e analysis, t h e i n t e r f a c e i s t a k e n t o be s p h e r i c a l .
Outside o f t h e f i n i t e - e l e m e n t subregion i s t h e o u t e r f i e l d , where a
complete s e t o f outgoing s p h e r f c a l harmonics i s used t o model t h e s c a t t e r e d
f i e 1 d. Each component s a t i s f i e s t h e equations o f motion and t h e Sanmerfel d
r a d i a t i on c o n d i t i o n s . The g l obal f u n c t i o n s have speci f i c s t r e s s and
d i spl acement d i s t r i b u t i ons a t t h e i n t e r f a c e , and t h e i r undetermined
s t r e n g t h s a r e t h e g l o b a l f u n c t i o n c o e f f i c i e n t s or t h e terms i n {SI. A
g i v e n i n c i d e n t wave i l l u n i n a t e s t h i s o b j e c t . The s c a t t e r e d f i e l d i s
determined by s o l v i n g t h e f i n i t e - e l e m e n t equations and r e q u i r i n g t h a t t h e
sun of i n c i d e n t and s c a t t e r e d wave f i e l d s based on t h e global f u n c t i o n s
have both t r a c t i o n and d i spl acement c o n t i nul t y w i t h the f i n i t e - e l ement
data a t t h e i n t e r f a c e . D e t a i l s o f t h i s a n a l y s i s may be found i n Ref. C21.
Here, a t t e n t i on i s c a l l e d t o t h e mathematical f l e x i b i l i t y o f t h e g l o b a l
f u n c t i o n s f o r accommodating i n t e r f a c e c o n t i n u i t y t o any p r e c i s i o n w i t h a
s u f f i c i e n t nunber o f terms.

The dynamic s o i l - s t r u c t u r e in t e r a t i on problem under steady-state

c o n d i t i o n s i s shown i n Fig. 3. The approach used h e r e i s s i m i l a r t o t h a t
f o r e l a s t i c wave s c a t t e r i n g by an o b j e c t embedded i n t h e e n t i r e space.
I n f a c t , t h e same set o f s p h e r i c a l harmonics f o r t h e e n t i r e space may be
a p p l i e d t o t h i s half-space problem. However, t r a c t i o n - f r e e surface
c o n d i t i o n s a r e n o t s a t i s f i e d by t h e spherical harmonics. Thus, i n
a d d i t i o n t o t r a c t i o n and d i spl acement c o n t i nui t y a t t h e hemispherical
i n t e r f a c e , i t i s necessary t o enforce t h e t r a c t i o n - f r e e surface i n t h e
g l obal subregion. I n Refs. C3,41, d e t a i l s concerni ng an i n t e g r a l
c o n s t r a i n t c o n d i t i o n t o meet t h i s t r a c t i o n - f r e e surface c o n d i t i o n a r e
given. Again, i t i s noted t h a t because a complete set o f g l o b a l
f u n c t i o n s i s used ( t h a t i s , a s e t capable o f modeling any a r b i t r a r y t r a c t i o n
and d i spl acement condi ti ons between v a r i o u s subregi ons in a GLFEM 1ayout ),
t h e a n a l y s i s procedure enjoys t h e o p p o r t u n i t y o f converging o n t o t h e
t r u e behavior w i t h i n c r e a s i n g FEM and global degrees o f freedom.

Mathematical l y Two-Dimensi onal S t r u c t u r e s

The choice o f g l obal f u n c t i ons f o r mathematical l y two-dimensional

s t r u c t u r e s w i l l now be discussed. As i l l u s t r a t i o n s o f t h i s c l a s s o f
problems and t h e i r GLFEM layouts, r e f e r t o Fig. 4 , where examples o f a
laminated composite p l a t e and c y l i n d e r s a r e given. The double l a p j o i n t
may be considered as a plane s t r a i n problem herein. The s c a r f j o i n t
j o i n i n g t w o c y l i n d e r s may be taken as an axisymmetric s t r u c t u r e under
axisymmetric or asymmetric loads. The purpose i s t o study t h e s t r e s s e s
i n t h e s e j o i nts.

Uniform s t r e s s s t a t e s e x i s t a t p o i n t s w e l l away fran these l o c a l i z e d

s t r e s s regions. I f FEM were used, it i s obvious t h a t an awkward model
woul d r e s u l t . I n GLFEM, two-dimensional f i n i t e e l ements ( p l anar or
a x i symnetri c t o r o i dal e l ements 1 a r e used f o r t h e subregi on c o n t a i n i ng t h e
1ocal i z e d stresses. I f t h e 1ocal i z e d s t r e s s s t a t e c o n t a i ns a s i ngul a r i t y ,
a g l o b a l subregion w i t h i n t h e f i n i t e - e l e m e n t subregion may be added. The
in t e r f ace 1o c a t i on i s dependent on t h e g l obal f u n c t i ons' mathemati c a l
capabil i t y f o r c a p t u r i n g t h e t r a n s i t i o n a l s t r e s s and d i spl acement f l e 1 ds
accurately. For g l obal f u n c t i o n s capable of r e p r e s e n t i ng t h e t r u e
behavior, t h e f i n i t e - e l e m e n t subregion can be q u i t e small w i t h t h e
i n t e r f a c e k 1 near t o t h e l o c a l i z e d s t r e s s area. An independent set of
global f u n c t i o n s must be adopted a t each i n t e r f a c e . For t h e l a p j o i n t
i n Fig. 4 , two or t h r e e d i s t i n c t systems o f global f u n c t i o n s may be
needed depending on t h e t h i c k n e s s and m a t e r i a l p r o p e r t i e s o f t h e p l a t e
components. Each set o f global f u n c t i o n s i s associated w i t h i t s own
s e t o f general i z e d c o o r d i n a t e s or global c o e f f i c i e n t s . For t h e
c y l i n d r i c a l s c a r f j o i n t , two independent s e t s a r e needed.

The g l o b a l f u n c t i o n s i n these casesare based on theorems r e l a t i n g t o

a q u a n t i t a t i v e expression o f St. Venant's p r i n c i p l e . Toupin C51 and
Knowles C61 presented upper bound e s t i m a t e s o f s t r a i n energy decay r a t e s
i n terms o f distance f ran a sel f - q u i 1 i b r a t e d s t r e s s state. Thei r
r e s u l t s can be s t a t e d i n t h e form o f a s t r a i n energy i n e q u a l i t y :

d V(0) e '2YX

where Y i s t h e i n v e r s e o f t h e c h a r a c t e r i s t i c decay length, V(0) i s t h e

t o t a l s t r a i n energy and V ( x ) i s t h a t p o r t i o n o f V ( 0 ) i n t h e body beyond
x. Since t h e s t r a i n energy i s quadratic, t h e mechanical v a r i a b l e s such
as stress, s t r a i n and displacement a r e of t h e forms:

139
 where Ki (i=lr2,3) a r e constants.

Based on these theorems, a boundary-Val ue problem can be formulated

f o r a prismatic structure. Using t h e s o l u t i o n form i n t h e p r i s m a t i c
d i r e c t i o n as e-Yx, t h e a n a l y s i s leads t o an efgenvalue problem. The
eigenvalues Y ' S a r e t h e c h a r a c t e r i s t i c i n v e r s e decay l e n g t h s and t h e
e i genf u n c t i ons a r e d i stri b u t i ons o f sel f - q u i 1 ib r a t e d s t r e s s states.
These eigendata comprise a complete set fran which any a r b i t r a r y sel f-
e q u i l i b r a t e d s t r e s s s t a t e may be represented. These eigendata may be
used as global f u n c t i o n s f o r d e s c r i b t n g t h e f a r - f i e l d behavior i n a
p r i s m a t i c s t r u c t u r e . Horgan and h i s colleagues have solved a nunber
o f problems on hanogeneous and sandwich p l a t e s under plane s t r a i n using
t h e A i r y s t r e s s f u n c t i o n as t h e primary dependent v a r i a b l e (see, f o r
exampl es, Refs. C7,81).

For a l a n i n a t e d composite s t r u c t u r e , it i s more convenient t o

determine t h e eigendata n u n e r i c a l l y . Dong and Goetschel C91 developed
a one-dimenslonal f i n i t e - e l e m e n t a n a l y s i s f o r e x t r a c t i n g eigendata f o r
a laminated composite p l a t e w i t h an a r b i t r a r y nunber o f bonded, e l a s t i c
1aminates. F i n i t e - e l ement d i s c r e t i z a t i o n occurs i n t h e t h i c k n e s s
d i r e c t i o n , see Fig. 5. Applying t h e theorem of minimun p o t e n t i a l energy,
a system o f second-order o r d i n a r d i f f e r e n t i a l equations i s obtained.
By i n v o k i ng exponenti a1 decay e ,'-; t h e f o l l owing second-order a1 gebrai c
e i genval ue probl em r e s u l t s :

where (0)i s an ordered s e t o f t h e p l a t e ' s nodal displacements. This

e q u a t i o n i s r e d u c i b l e t o first o r d e r w i t h a non-symmetric m a t r i x . If a
1arge nunber o f degrees o f freedom a r e involved, a Block-Stodol a i t e r a t i on
technique E101 can be used t o e x t r a c t t h e eigendata e f f i c i e n t l y . The
sol u t i o n c o n s i s t s o f a complete set of eigenval ues and corresponding
eigenvectors, which are t h e sel f - q u i 1 i b r a t e d displacement s t a t e s f o r t h e
g i v e n composite p l a t e . Stresses can be computed fran these displacements.

Laminated c y l i n d e r s may a l s o be solved u s i n g t h e same f i n i t e - e l e m e n t

scheme, see Ref. C113. The mechanical v a r i a b l e s have c i r c u n f e r e n t i a1
dependence, which may be expressed a n a l y t i c a l l y by F o u r i e r series. As
a c i r c m f e r e n t i a l mode nunber m occurs i n t h i s case, t h e c o u n t e r p a r t t o
Eq. ( 4 ) f o r each c i r c u n f e r e n t i a l mode has t h e form:

The s o l u t i o n t o Eq. ( 4 ) or (5) p r w i d e s t h e g l o b a l f u n c t i o n data

base f o r t h e nunerical e v a l u a t i o n o f t h e global s t i f f n e s s m a t r i x and t h e

140
g l obal-local coup1 i n g as a p r e l ude t o a mathematically two-dimensional
GLFEM a n a l y s i s o f t h e l o c a l i z e d s t r e s s zones. S a e p r e l iminary r e s u l t s
of t h i s t y p e have been obtained, which a r e contained i n Refs. C12,133.
These 1i m i t e d scope s t u d i e s i n d i c a t e an o v e r a l l f e a s i b i l i t y f o r t h i s
approach.

The accuracy o f t h e global f u n c t i o n s depends on t h e f i n e n e s s o f t h e

one-dimensi onal f i n i t e - e l ement model adopted f o r t h e e i genproblem. Since
only one-dimensional f i n i t e elements a r e used, a l a r g e model does n o t i n c u r
an i n o r d i n a t e computational e f f o r t because o f a very small bandwidth.

The nunber of g l o b a l f u n c t i o n s r e q u i r e d i n a GLFEM a n a l y s i s depends

on both t h e n a t u r e of t h e l o c a l i z e d s t r e s s and t h e l o c a t i o n o f t h e
i n t e r f a c e . Having an i n t e r f a c e near t h e l o c a l i z e d s t r e s s zone w i l l
r e q u i r e a l a r g e r nunber of g l o b a l functions, b u t w i t h a decrease i n t h e
f in i te-el ement coordi nates. Converse1 y, an in t e r f ace f a r removed f ran
t h e l o c a l i z e d zone needs fewer global f u n c t i o n s t but i s counteracted by
a g r e a t e r nunber of f i n i t e - e l e m e n t degrees o f freedom.

Three-Dimensi onal S t r u c t u r e s and t h e NASA Probl em

A schematic o f a three-dimensional p r i s m a t i c s t r u c t u r e and t h e NASA

problem o f a f l a t s t i f f e n e d composite panel w i t h a discontinuous s t i f f e n e r
a r e shown i n Fig. 6. I n t h i s c l a s s of problems, three-dimensional f i n i t e
elements must be employed i n t h e l o c a l i z e d s t r e s s region. Global f u n c t i o n s
must be used a t t h e i n t e r f a c e . They can be obtained fran a two-dimensional
f i n i t e - e l e m e n t a n a l y s i s o f t h e i n v e r s e c h a r a c t e r i s t i c decay lengths.

The a n a l y s i s t o determine t h e global f u n c t i o n s f o l l o w s t h e same

methodology as t h a t f o r mathematical l y two-dimensi onal s t r u c t u r e s . The
p r i smati c cross-secti on is model ed by two-dimensi onal f i n i t e e l ements.
With t h e dependence i n t h e p r i s m a t i c d i r e c t i o n t a k e n as e-yx, an
eigenproblem emerges f o r t h e e x t r a c t i o n o f eigendata t h a t form t h e
g l o b a l f u n c t i o n data base f o r t h e g i v e n cross section. The o t h e r
aspects a r e t h e same as t h a t described i n t h e previous section. It i s
obvious t h a t s i n t h i s case, t h e computational e f f o r t i s greater.

Sane comments can be g i v e n on a GLFEM a n a l y s i s o f t h e NASA f l a t panel.

The s e t o f two-dimensional global f u n c t i o n s c o n s t i t u t e s a complete system
o f e l genf u n c t i ons, w i t h t h e non-zero e i genval ues associ ated w i t h i n v e r s e
c h a r a c t e r i s t i c decay l e n g t h s o f sel f - q u i 1 i b r a t e d s t r e s s states. There
a r e two z e r o eigenvalues f o r two s t r e s s d i s t r i b u t i o n s e x h i b i t i n g no decay.
They are t h e uniform a x i a l deformation and p u r e bending states. These two
global f u n c t i o n s a r e needed i n a GLFEM a n a l y s i s o f t h e NASA f l a t panel#
since t h e discontinuous s t i f f e n e r may produce bending i n a d d i t i o n t o i t s
uniform end shortening. A s e t o f global f u n c t i o n s w i t h a l l o f these
members present should p e r m i t a three-dimensional f i n i t e - e l e m e n t model
t o be concentrated on t h e d e t a i l s of t h e discontinuous s t i f f e n e r region.
Parametric s t u d i e s wherein t h e h o l e and t h e gap l e n g t h i n t h e
d i s c o n t i nuous s t i f f e n e r a r e v a r i e d may be conducted. Each c o n f i g u r a t i o n
w i l l r e q u i r e a change of t h e three-dimensional f i n i t e - e l e m e n t mesh, but
the same s e t o f g l o b a l f u n c t i o n s may be used i n a l l cases.

141
 Appl i c a t i ons t o Steady-State Probl ems

The d i scussion o f g l o b a l f u n c t i o n s i n t h e two e a r l i e r s e c t i o n s

p e r t a i n e d t o e l a s t o s t a t i c a n a l y s i s o f t h e l o c a l i z e d s t r e s s zones. Here
some ranarks on steady-state dynamic e f f e c t s a r e made. The one-dimensional
f i n i t e - e l e m e n t method f o r g e n e r a t i n g global f u n c t i o n data bases can be
m o d i f i e d f o r steady-state i n e r t i a l e f f e c t s by i n c l u d i n g k i n e t i c energy i n
t h e problem formulation. I n s t e a d o f Eqs. ( 4 ) and (51, those equations
become, r e s p e c t i v e l y :

where w i s t h e steady-state f o r c i n g frequency. The d e r i v a t i o n s o f these

equations a r e g i v e n i n Refs. C11,143.

With these g l o b a l functions, it i s p o s s i b l e t o study e l a s t i c wave

s c a t t e r 1 ng 1n p r i m a t i c s t r u c t u r e s by d i s c o n t i nut t l e s d u r l ng v i b r a t i on
or by some o t h e r steady-state dynamic i n p u t . G L E M a n a l y s i s o f t h i s
t y p e of p r i s m a t i c s t r u c t u r e s w i l l be s i m i l a r t o problems o f e l a s t i c wave
s c a t t e r i n g by an o b j e c t embedded i n an i n f l n l t e mediun or s o i l - s t r u c t u r e
1n t e r a c t i on.

E f f e c t s o f I n i t i a l Stress

Us1 ng t h e same methodol ogy, p r i smati c s t r u c t u r e s under in i ti a1

s t r e s s may a l s o be analyzed. In this t h e global f u n c t i o n s must
1n c l ude t h e prestress1 ng e f f e c t . One-dimensi onal f i n i t e - e l e m e n t
a n a l y s i s o f wave propagation i n l a n i n a t e d composite p l a t e s and c y l i n d e r s
under i n i t i a l s t r e s s have been explored, see Refs. C15,161. It i s a
s t r a i g h t f o r w a r d t a s k t o adapt these f o r m u l a t i o n s t o generate an
e i g e n p r o b l a n f o r t h e g l o b a l f u n c t i o n s f o r a p r i s m a t i c s t r u c t u r e under
i n i t i a l s t r e s s . Also, no conceptual d i f f i c u l t i e s a r e seen i n an
e x t e n s i on t o three-dimensi onal p r i s m a t i c s t r u c t u r e s under prestress.

Concl udi ng Remarks

Considerable d i s c u s s i o n has been devoted t o t h e s t r a t e g i e s o f GLFEM

analyses of p r i s m a t i c s t r u c t u r e s w i t h l o c a l i z e d s t r e s s r e g i o n s and o t h e r
d i s c o n t i n u i t i e s . The r o l e o f t h e g l o b a l f u n c t i o n s has been c l e a r l y
out1 i n e d and t h e 1 r mathematical requi r m e n t s I n d i c a t e d . The method
f o r d e r l v i n g these global f u n c t i o n s f o r p r i s m a t i c s t r u c t u r e s , whose
cross-sect1 onal g e u n e t r i e s a r e compl i c a t e d by l a n i n a t e d c o n s t r u c t 1on,
has been discussed. Frun t h e d i s c u s s i o n o f GLFEM a n a l y s i s strategy,
i t should be c l e a r t h a t GLEM i s f e a s i b l e and e f f e c t i v e . Considerable
economy o f computational e f f o r t s over a s t r i c t l y FEM approach should
be r e a l ized.

142
References

1. Dong, S. B. , "G1 obal -Local F i n i t e E l ement Methods, State-of-the-Art

Surveys on F i n i t e Element Techno1 ogy, e d i t e d by A.K. Noor and W.D.
Pi1 key, Chap. 14, pp.451-474, h e r i c a n Society o f Mechanical
Engi neers, 1983.

2. Goetschel, D.B.8 Dong, S.B. and Muki, R.8 "A Global-Local F i n i t e

Element A n a l y s i s o f Axisymmetric S c a t t e r i n g o f E l a s t i c Waves,"

.
Journal of Appl i e d Mechani cs, h e r i can S o c i e t y o f Mechanical
Engineers, Vol 49, pp. 816-820, 1982.

3. Avanessian, V., Muki, R. and Dong, S.B.8 *'Forced O s c i l l a t i o n o f an

Axisymmetric S t r u c t u r e i n Contact w i t h an E l a s t i c Half-space by a
V e r s i o n of Global Local F i n i t e Element,11 Journal o f Sound and
.
V i b r a t i o n , Vol 104, pp. 449-463, 1986.
4. ,
Avanessi an, V. , Muki R. and Dong, S. 8. , "Axi symmetric Sol 1- S t r u c t u r e
I n t e r a c t i o n by G1 obal -Local F i n i t e Elements, I n t e r n a t i onal Journal
of Earthquake Engineering and S t r u c t u r a l Dynamics, i n press.

5. Toupi n, R. A. "Sai nt-Venant' s P r i n c i p l e, A r c h i v e f o r R a t i onal

Mechanics and Analysis, Vol. 18, pp. 83-96, 1965.

6. Knowles, J.K., "On Saint-Venant's P r i n c i p l e i n Two-Dimensional L i n e a r

Theory of E l a s t i c i t y r t l A r c h i v e f o r R a t i o n a l Mechanics and Analysis,
Vol. 21, pp. 1-22, 1966.

7. Choir I.and Horgan, C.O., "Saint-Venant's P r i n c i p l e and End E f f e c t s

i n A n i s o t r o p i c E l a s t i c i t y , " Journal o f Appl l e d Mechanics, Anerican
.
Soci e t y o f Mechanical Engi n e e m Vol 44, pp. 424-430, 1977.

8. Horgan, C.O., "Sane Remarks on Saint-Venant's P r i n c i p l e f o r

Transversely I s o t r o p i c Composites, 1) Journal o f E l a s t i c i ty,
Vole 2, pp. 335-339, 1972.

9. Dong, S.B. and Goetschel, D.B., !'Edge E f f e c t s i n Laminated Composite

Journal of Appl l e d Mechanics, Anerican Society of Mechanical
Engineersr Vol. 49, pp. 129-135, 1982.

10 0 Dong, S.B.8 "A Block-Stodol a Eigensol & i o n Technique f o r Large

A1 g e b r a i c Systems w i t h Non-Symmetrical Matricesrtl I n t e r n a t i o n a l
Journal f o r Nunerical Methods i n Engineering, Vol. 11, pp. 247-267,
1977.
11. Huang, K.H. and Dong, S.B., "Propagating Waves and Standing V i b r a t i o n s
i n a Composite Journal o f Sound and Vibration, Vol. 96,
pp. 363-379, 1984.

12. Chen, W.T. , 9 t u d i es Re1 a t i ng t o Appl i c a t i on o f t h e Global-Local F i n i t e

Element Method t o E l a s t o s t a t i c A n a l y s i s o f End E f f e c t s i n C y l i n d e r s r f f
M.S. Research P r o j e c t Report, Dept. o f C i v i l Engineering, U n i v e r s i t y
o f C a l i f o r n i a , Los Angeles, March 1985.

143
13. Nystran, D.M., TJunerical E v a l u a t i o n o f Global S t i f f n e s s M a t r i x f o r
A n a l y s i s o f Sel f-Equil ib r a t e d S t r e s s States, M. S. Thesi s, Graduate
D i v i s i on, U n i v e r s i t y o f Cal i f o r n i a, Los Angel es, June 1985.

14. Dong, S.B. and Huang, K.H., '!Edge V i b r a t i o n s i n Laminated Composite

P1 ates,", Journal o f Appl i e d Mechanics, h e r i c a n Society o f
.
Mechanical Engineers, Vol 52, pp. 433-438, 1985.

15. Bradford, L.G. and Dong, S.B., V a t u r a l Vibrations o f Orthotropic

P1 ates under I n i ti a1 Stress, I n t e r n a t i o n a l Journal o f Sol i d s and
Structures, Vol. 11, pp. 213-230, 1975.

16. Bradford, L.G. and Dong, S.B., "Natural Vibrations o f Orthotropic

Cy1i n d e r s under I n i ti a1 Stress," Journal o f Sound and V i b r a t i on,
Vol. 60, pp. 157-175, 1978.

144
(a) GLOBAL FUNCTIONS (b) GLOBAL FUNCTIONS
ONLY OVER ENTIRE OVER (VOLjg
DOMAIN GLOBAL &
LOCAL FUNCTIONS
OVER (VOL)eg

(c) GLOBAL FUNCTIONS (d) GLOBAL & LOCAL

OVER (VOL)g FUNCTIONS OVER
LOCAL FUNCTIONS ENTIRE DOMAIN
OVER ( V O L ) i

(e) GLOBAL & LOCAL (f) LOCAL FUNCTIONS

FUNCTIONS OVER OVER ENTIRE DOMAIN
(VOL) e,
LOCAL FUNCTIONS
OVER ( V 0 L ) i

1
- GLOBAL REGION {g}= [Bg]{S}
FffR LOCAL (FINITE ELEMENT) REGION
{E}= [Nil{&}
COMBINED GLOBAL-LOCAL REGION

(.} = [$, I {4} +[Ngl{ s}

F i g u r e 1. B a s i c g l o b a l - l o c a l mesh c o n f i g u r a t i o n s .

145
 SC ATT ERED

finite elements

MESH BOUNDARY

INCIDENT WAVES
global functions

SCATTERED WAVE F I E L D - S P H E R I C A L HARMONICS

n=o J

F i g u r e 2. E l a s t i c wave s c a t t e r i n g by axisymmetric i n c l u s i o n embedded i n

an i n f i n i t e i s o t r o p i c medium.

146
 TRRCTION-FREE

Figure 3. Soil-structure interaction problem.

F I N I T E ELEMENT SUBREGION
(TWO-OIMENSIONAL PLANAR ELEMENTS)

ALUM I NUM

-
DOUBLE L A P JOINT OF ALUMINUM AND COMPOSITE PLATES
I’ COMPOS I TE

UNDER PLANE S T R A I N

I
SCARF J O I N T OF ALUM I NUM AN0 COMPOS I TE CYL I NDERS

Figure 4. Two-dimensional prismatic structures.

147
 STRAlN ENERGY WCAY RATE

self-equilibrated
tractions on this
surface
I-.-. --.
&

FORMULATION OF TWO-DIMENSIONAL PROBLEM FOR

PLATE WITH PLANE ANISOTROPIC flATERlALS

' [TYPICAL LAMINATE

f= (Y - Y b 1 (Y, -Yb 1
SOLUTION FORM IN X-DIRECTION SECOND ORDER ALGEBRAIC EIGENVALUE PROB.

Figure 5. Finite-element analysis of self-equil ibrated edge effects

in a composite plate.
148
 EXAMPLE OF A

-
THREE-DIMENSIONAL P R I S M A T I C STRUCTURE

P
UNIFORM END SHORTENING-

-j-L/2- -
I
THE NASA FLAT PANEL WITH
'tee
DISCONTINUOUS STIFFENER I
I
I
I

F I N I T E ELEMENT SUBREGION
(SOL I D , PLATE AND SHELL ELEMENTS)

Figure 6. Three-dimensional prismatic structures.

149
 GLOBAL-LOCAL METHODOLOGIES AND T H E I R APPLICATION
TO NONLINEAR ANALYSIS

Ahmed K. Noor
George Washington U n i v e r s i t y
N A S A Langley Research Center
Hampton, V i r g i n i a

ABSTRACT

An assessment is made of t h e p o t e n t i a l of d i f f e r e n t g l o b a l - l o c a l a n a l y s i s
s t r a t e g i e s f o r p r e d i c t i n g t h e n o n l i n e a r and p o s t b u c k l i n g r e s p o n s e s of
s t r u c t u r e s . Two p o s t b u c k l i n g problems of composite p a n e l s are u s e d a s bench-
m a r k s and t h e a p p l i c a t i o n of d i f f e r e n t g l o b a l - l o c a l methodologies t o these
benchmarks is o u t l i n e d . The key e l e m e n t s of each o f t h e g l o b a l - l o c a l
s t r a t e g i e s are d i s c u s s e d and f u t u r e research areas needed t o r e a l i z e t h e f u l l
p o t e n t i a l of g l o b a l - l o c a l methodologies are i d e n t i f i e d .

NOMENCLATURE

E L J ET E l a s t i c moduli of t h e i n d i v i d u a l layers i n t h e
d i r e c t i o n o f fibers and normal t o i t , r e s p e c t i v e l y
GLTJ GTT Shear moduli i n p l a n e o f f i b e r s and normal t o i t ,
respectively
h Thickness o f p a n e l
L 1 ' L2 S i d e l e n g t h s of p a n e l
N T o t a l a x i a l f o r c e a c t i n g on t h e edge of t h e p a n e l
9 Edge d i s p l a c e m e n t
R Radius o f c u r v a t u r e of t h e panel m i d d l e s u r f a c e
r Number of g l o b a l approximation v e c t o r s
U T o t a l s t r a i n energy o f the p a n e l
u w Displacement components i n t h e c o o r d i n a t e d i r e c t i o n s
x a' x Orthogonal c u r v i 1i n e a r c o o r d i n a t e s y s tem
a' 3 Major P o i s s o n ' s r a t i o of t h e i n d i v i d u a l layers
V
LT
E Axial s t r a i n
1
The r a n g e of t h e s u b s c r i p t a i s 1.2.

1. INTRODUCTION

C o n s i d e r a b l e p r o g r e s s has r e c e n t l y been made i n c o m p u t a t i o n a l mechanics

which is m a n i f e s t e d by t h e development o f v e r s a t i l e and powerful f i n i t e -
element d i s c r e t i z a t i o n methods, improved n u m e r i c a l a l g o r i t h m s and programming
t e c h n i q u e s ( s e e , f o r example, BELYTSCHKO & HUGHES C19831; NOOR & PILKEY
C19831; NOOR C19831; L I U , BELYTSCHKO & PARK C19841; and KARDESTUNCER
C19851). Also, an e x p l o s i v e growth has t a k e n p l a c e i n computer t e c h n o l o g y .
I n p a r t i c u l a r , t h e i n t r o d u c t i o n of large e x p e n s i v e computer s y s t e m s , u s u a l l y
referred t o as s u p e r s y s t e m s , s u c h a s CRAY X-MP, CDC CYBER 205 and Denelcor
HEP1 has made p o s s i b l e new l e v e l s of s o p h i s t i c a t i o n i n t h e modeling of complex
s t r u c t u r e s which were not p o s s i b l e b e f o r e ( N O O R , STORAASLI & FULTON C19841).
I n s p i t e o f these advances t h e d e t a i l e d stress a n a l y s i s o f complex s t r u c t u r e s
is very time consuming a n d , t h e r e f o r e , is n o t e c o n o m i c a l l y f e a s i b l e . To d a t e
t h e o n l y r e a l i s t i c s t r u c t u r a l r e s p o n s e s i m u l a t i o n s t h a t have been o b t a i n e d

151
PRECEDING PAGE BLANK NOT FlLMED
 involve either simple structural configurations or components of complicated
configurations. Prediction of the response of future complex structures such
as those of flight vehicles is likely to require more sophisticated analysis
models than has heretcfore been done. This is because of the requirements of
high performance, light weight and economy and the associated stringent design
criteria. Also, analysis may of necessity replace tests in some mission-
critical areas.
Among the different analysis methodologies that have high potential for
the accurate prediction of detailed stress distribution in structures without
overtaxing the available computational resources are the global-local
methodologies which are basically hybrid modeling and/or analysis techniques.
In order to put these methodologies in proper perspective, a brief summary is
given of the different approaches for reducing the cost and/or time for
solving nonlinear problems. The efforts devoted to this activity can be
grouped into three different levels.
The first level is modeling. Reductions in cost of analysis can be
achieved by using simple models that capture major effects in the responses
and by exploiting all the symmetries and quasi-symmetries in the problem (see
NOOR & PETERS C19851; and NOOR, ANDERSEN & TANNER C19851).
The second level is that of computational strategies. Significant re-
ductions in time can be achieved by incorporating the known physical behavior
into the computational model of the structure and by using global-local
methodologies in which different analysis methods and/or models are coupled
for predicting the nonlinear response of the structure.
The third level is that of - --algorithms. These include fast
numerical
algorithms for solution of equations (e.g., multigrid methods, operator
splitting techniques, dynamic relaxation, and element-by-element techniques
-HACKBUSCH & TROTTENBERG C19821; UNDERWOOD C19831; and HUGHES, RAEFSKY,
MULLER, WINGET & LEVIT C19841); as well as the vectorized and parallel
numerical algorithms for use on pipeline and parallel processors (SCHENDEL
C19841; MIKLOSKO & KOTOV C19841; and PADDON C19841).
The present study deals with global-local methodologies which belong to
the second category. Specifically, the objectives of this paper are:
1 ) To review and assess the potential of a number of different global-
local analysis strategies for predicting the nonlinear and postbuckling
responses of structures
2) To identify the future directions for research required to realize
their full potential
Discussion of global-local methodologies is primarily focused on
nonlinear analysis of composite panels with discontinuities (e.g., stiffeners
and cutouts). Two benchmark problems of composite cylindrical panels with
cutouts typical of those used in modern aircraft structures are selected to
provide a focus for the discussion. However, many of the conclusions apply to
other complex structural configurations.
The paper is divided into three parts. The first part describes the two
benchmark problems, identifies their major characteristics, and lists the
difficulties encountered in analyzing them using conventional finite-element
methods. In the second part of the paper four global-local analysis
strategies are reviewed and the potential for using these strategies in
analyzing the benchmark problems is assessed.
The third part of the paper identifies the items that pace the progress
of global-local methodologies and their application t o nonlinear analysis.
These are the recommended future directions of research.

152
 2. BENCHMARK PROBLEMS AND THEIR CHARACTERISTICS

The two benchmark problems selected for this study are shown in Fig. 1.
They are postbuckling problems of laminated composite cylindrical panels with
central circular cutouts. The loading consists of applied axial end
displacements. One of the panels is unstiffened and the other has discrete
blade stiffeners. The two panels are generic examples of modern composite
aircraft components for which postbuckling strength is desired in the presence
of local discontinuities such as holes and cracked stiffeners. In the
conventional finite-element approach the panels are modeled using two-
dimensional shell elements and the stiffeners are modeled using two-
dimensional plate elements. The unstiffened panel was analyzed using an in-
house research program. Shear-flexible mixed finite elements were used in the
modeling (see NOOR & ANDERSEN C19821). Also, extensive numerical solutions to
this problem using continuum-based shell elements are presented in STANLEY
C19851 in which the imperfection sensitivity of the panel is assessed. The
stiffened panel was analyzed using the EISI-EAL program and the analysis
results are given in KNIGHT, GREENE & STROUD C19851. The characteristics of
the finite-element models used in the present study and those used in KNIGHT,
GREENE & STROUD C19851 are summarized in Table 1. The response of both panels
exhibits inversion symmetry, and therefore, only one half of each panel needs
to be analyzed (see NOOR, MATHERS & ANDERSON C19771). The numbers between
parentheses in Table 1 refer to the number of elements and degrees of freedom
required for analyzing the full panel.
The response of the unstiffened panel is shown in Figs. 2 and 3. The
postbuckling response of the panel exhibits a sudden drop in the loading due
to delamination in the neighborhood of the cutout (see KNIGHT & STARNES
C19841). Figure 2 shows plots of the total axial load versus axial and normal
displacements up to the maximum load reached. For the range of loading
considered, the results shown in Fig. 2 agree reasonably well with the
experimental and numerical results presented in KNIGHT & STARNES C19841.
Figure 3 shows normalized contour plots of the axial and normal displace-
ments as well as the axial strains on the top and bottom surfaces. Note the
high strain concentration at the cutout. A finer model near the cutout is
required for the accurate prediction of the strain in that region. On the
basis of the studies made, the following characteristics of the two benchmark
problems can be identified:
1. The presence of discontinuities (cutouts and stiffeners) results in
large numbers of degrees of freedom in the finite-element models.
2. The detailed stress analysis (including determination of interlaminar
stresses) near the cutout requires either a higher order two-dimensional
theory or a three-dimensional theory. Failure analysis (including prediction
of delamination) requires even more sophistication in the modeling and
analysis.
3. The postbuckling response exhibits large rotations in certain
regions.
4. The postbuckling response of the unstiffened panel is highly
sensitive to initial imperfections. Tracing the postbuckling response past
the maximum load point requires the inclusion of initial imperfections in the
model (see STANLEY C19851).

3. GLOBAL-LOCAL ANALYSIS STRATEGIES

In this section, the application of four different global-local analysis

153
 s t r a t e g i e s t o t h e p r e d i c t i o n of t h e n o n l i n e a r r e s p o n s e of benchmark composite
p a n e l s is d i s c u s s e d . I n each case, t h e k e y f e a t u r e s of t h e g l o b a l - l o c a l
approach and t h e r e q u i r e m e n t s f o r i t s e f f e c t i v e implementation are i d e n t i f i e d .

3.1 Zooming Technique

The f i r s t g l o b a l - l o c a l approach c o n s i d e r e d is t h e zooming t e c h n i q u e i n

which a g l o b a l s o l u t i o n is o b t a i n e d u s i n g a c o a r s e g r i d . Then t h e d e t a i l e d
stress d i s t r i b u t i o n n e a r t h e c u t o u t ( l o c a l s o l u t i o n ) i s o b t a i n e d by zooming on
t h a t area, r e f i n i n g t h e model, and u s i n g t h e d i s p l a c e m e n t s from t h e c o a r s e r
model as i n p u t f o r t h e r e f i n e d model (see F i g . 4 ) . S i n g l e o r m u l t i p l e l e v e l s
of zooming can be made c o r r e s p o n d i n g t o m u l t i p l e passes from c o a r s e t o f i n e
s u b d i v i s i o n s ( W I L K I N S C19831, and H I R A I , UCHIYAMA, M I Z U T A & PILKEY C19851).
I n t e r e l e m e n t c o m p a t i b i l i t y can be m a i n t a i n e d by e i t h e r embedding t h e r e g i o n of
t h e f i n e g r i d i n t o a s u p e r e l e m e n t o r by u s i n g t r a n s i t i o n e l e m e n t s a t t h e
i n t e r f a c e between t h e c o a r s e and t h e f i n e g r i d s ( A R M E N , Grumman Aircraft
Systems, Bethpage, N Y , P r i v a t e Communication, 1985).
A number of q u e s t i o n s remain i n c o n n e c t i o n w i t h t h i s t e c h n i q u e , namely:
1 . L i m i t a t i o n s of t h e t e c h n i q u e f o r n o n l i n e a r problems
2 . C r i t e r i a f o r s e l e c t i n g t h e e x t e n t of t h e r e g i o n f o r r e f i n i n g t h e
model and s e l e c t i o n of the r e f i n e d model
3. Treatment of t h e i n t e r f a c e s between t h e c o a r s e and t h e f i n e g r i d s and
t h e e f f e c t of t h e e r r o r on t h e boundary d a t a f o r t h e r e f i n e d model on t h e
accuracy of t h e stresses i n t h a t model
The e a r l y work on zooming t e c h n i q u e s was based on h e u r i s t i c and i n t u i t i v e
approaches f o r s e l e c t i n g t h e f i n e model. More r e c e n t work is based on t h e
d i s t r i b u t i o n of t h e s t r a i n e n e r g y d e n s i t y f u n c t i o n . C u r r e n t l y , a v a r i e t y o f
a d a p t i v e r e f i n e m e n t t e c h n i q u e s are a v a i l a b l e . Some of t h e s e t e c h n i q u e s w i l l
be d i s c u s s e d i n s u b s e q u e n t s e c t i o n s .

3.2 -
S i m u l t a n e o u s A p p l i c a t i o n o f Two D i s c r e t i z a t i o n Techniques

The early a p p l i c a t i o n s of t h i s approach c o n s i s t e d of t h e s i m u l t a n e o u s u s e

of t h e f i n i t e - e l e m e n t method and t h e g l o b a l ( c l a s s i c a l ) v a r i a t i o n a l t e c h n i q u e
(see MOTE C19711). A v a r i e t y o f o p t i o n s a r e a v a i l a b l e depending on t h e e x t e n t
of u s i n g each of t h e two t e c h n i q u e s i n t h e model (DONG C19831). The concept
has been l a t e r g e n e r a l i z e d t o cover o t h e r t e c h n i q u e s f o r t h e g l o b a l
( a p p r o x i m a t e ) s o l u t i o n and t h e l o c a l ( d e t a i l e d ) s o l u t i o n . A l i s t of t h e
common t e c h n i q u e s u s e d f o r g e n e r a t i n g these s o l u t i o n s is shown i n T a b l e 2.
The g l o b a l s o l u t i o n can be o b t a i n e d by u s i n g c l a s s i c a l v a r i a t i o n a l
methods ( e . g . , Rayleigh-Ritz o r w e i g h t e d - r e s i d u a l a p p r o a c h e s ) , or any o f t h e
d i s c r e t e element methods ( c o n v e n t i o n a l f i n i t e e l e m e n t s , g l o b a l e l e m e n t s ,
boundary element method) o r t h e i r c o m b i n a t i o n s . I n t h e g l o b a l ( o r macro)
element method t h e s t r u c t u r e is d i v i d e d i n t o a small number of e l e m e n t s and a
s u i t a b l e ( u s u a l l y nonpolynomial) a p p r o x i m a t i o n is made w i t h i n each e l e m e n t .
The c o n t i n u i t y of t h e f i e l d v a r i a b l e s a c r o s s t h e i n t e r f a c e s i s imposed
i m p l i c i t l y i n t h e v a r i a t i o n a l f u n c t i o n a l (DELVES & HALL C19791, and DELVES &
P H I L L I P S C19801). The g l o b a l element method is a compromise between c l a s s i c a l
v a r i a t i o n a l t e c h n i q u e s and f i n i t e - e l e m e n t method. It combines t h e r a p i d
convergence of g l o b a l v a r i a t i o n a l methods w i t h t h e a b i l i t y t o h a n d l e
complicated geometries.
The boundary element method is an e f f e c t i v e t e c h n i q u e f o r s o l u t i o n of
l i n e a r and materially n o n l i n e a r s t r u c t u r a l problems w i t h h i g h stress g r a d i e n t s
( o r s i n g u l a r i t i e s ) . However, i t is n o t c o m p e t i t i v e w i t h e i t h e r t h e f i n i t e -

154
element o r t h e g l o b a l element method f o r g e o m e t r i c a l l y n o n l i n e a r problems (see
K A M I Y A & S A W A K I C19821; and K A M I Y A , S A W A K I & NAKAMURA C19841).
Among t h e d i f f e r e n t t e c h n i q u e s f o r l o c a l a n a l y s i s are t h e d i s c r e t e ele-
ment methods and a n a l y t i c a l s o l u t i o n s ( e . g . , p o l a r and/or e d g e f u n c t i o n s
PATTIBIRAMAN, RAMAMURTI & REDDY [1974]). F i g u r e 5 shows three d i f f e r e n t
combinations of d i s c r e t e element methods f o r a n a l y z i n g !he u n s t i f f e n e d
composite p a n e l .
The e f f e c t i v e implementation of t h i s g l o b a l - l o c a l s t r a t e g y r e q u i r e s :
a ) c r i t e r i a f o r s e l e c t i n g t h e p r o p e r g l o b a l and l o c a l a n a l y s i s t e c h n i q u e s and
b ) problem-adaptive s t r a t e g i e s f o r g e n e r a t i n g g l o b a l s o l u t i o n s and t r e a t m e n t
of i n t e r f a c e s .

3.3 -
Reduction Methods

These are h y b r i d two-step t e c h n i q u e s which a r e based on t h e s u c c e s s i v e

a p p l i c a t i o n of a d i s c r e t e element method ( f i n i t e e l e m e n t s , boundary e l e m e n t s
or combination o f f i n i t e e l e m e n t s and boundary e l e m e n t s ) and c l a s s i c a l
v a r i a t i o n a l t e c h n i q u e s (see, f o r example, NOOR C19821; and NOOR & PETERS
L19831). The d i s c r e t e element method is used t o g e n e r a t e - few g l o b a l
a p p r o x i m a t i o n v e c t o r s ( o r modes). The c l a s s i c a l v a r i a t i o n a l t e c h n i q u e is t h e n
used t o compute t h e a m p l i t u d e s of these modes. The primary o b j e c t i v e o f u s i n g
r e d u c t i o n methods is t o r e d u c e c o n s i d e r a b l y t h e number o f d e g r e e s o f freedom
i n t h e i n i t i a l d i s c r e t i z a t i o n , and hence, r e d u c e t h e c o m p u t a t i o n a l e f f o r t
i n v o l v e d i n t h e s o l u t i o n of t h e n o n l i n e a r problem.
The a p p l i c a t i o n o f r e d u c t i o n methods t o t h e u n s t i f f e n e d composite p a n e l
problem is d e p i c t e d i n F i g s . 6 and 7. F i g u r e 6 shows t h e a c c u r a c y of t h e
normal d i s p l a c e m e n t s and t o t a l s t r a i n e n e r g y o b t a i n e d by u s i n g f o u r g l o b a l
a p p r o x i m a t i o n v e c t o r s ( g e n e r a t e d a t z e r o l o a d i n g ) . F i g u r e 7 shows c o n t o u r s o f
t h e normal d i s p l a c e m e n t w of t h e f i r s t three g l o b a l a p p r o x i m a t i o n v e c t o r s .
Two r e c e n t a p p l i c a t i o n s of r e d u c t i o n methods d e s e r v e f u r t h e r e x a m i n a t i o n .
I n t h e f i r s t , o n l y p a r t i a l r e d u c t i o n is made. The d e g r e e s of freedom i n t h e
r e g i o n of s t r o n g n o n l i n e a r i t y ( e . g . , n e a r t h e c u t o u t ) are r e t a i n e d ; t h e o t h e r
degrees of freedom are r e d u c e d . I n t h e second a p p l i c a t i o n , t h e r e s p o n s e of a
complex s t r u c t u r e ( e . g . , s t i f f e n e d p a n e l ) is g e n e r a t e d u s i n g small ( o r l a r g e )
p e r t u r b a t i o n s from t h e r e s p o n s e of a s i m p l e r s y s t e m ( e . g . , u n s t i f f e n e d
p a n e l ) . I t is a l s o p o s s i b l e t o use a h i e r a r c h y of s i m p l e r s t r u c t u r a l s y s t e m s
i n g e n e r a t i n g t h e r e s p o n s e of t h e o r i g i n a l complex s t r u c t u r e . T h i s is
accomplished by c h o o s i n g a number of p e r t u r b a t i o n parameters, and s u c c e s s i v e l y
a p p l y i n g a s i n g l e - p a r a m e t e r r e d u c t i o n method w i t h each of t h e p a r a m e t e r s .
A p p l i c a t i o n of t h i s s t r a t e g y t o t h e n o n l i n e a r a n a l y s i s of a n i s o t r o p i c p a n e l s
is described i n NOOR c19851.
The w i d e a c c e p t a n c e o f r e d u c t i o n methods and t h e i r i n c o r p o r a t i o n i n t o
commercial programs r e q u i r e s : a > t h e s e l e c t i o n of a s i m p l e s e t of g l o b a l
a p p r o x i m a t i o n v e c t o r s and b ) t h e development of a problem-adaptive.strategy
f o r e r r o r s e n s i n g and c o n t r o l .

._I
-
3.4 H i e r a r c h y o f Mathematical Models and/or Numerical Approximation
-

Techniques

The l a s t g l o b a l - l o c a l approach c o n s i d e r e d i s t h a t based on a h i e r a r c h y of

mathematical models f o r d i f f e r e n t p a r t s of t h e s t r u c t u r e . The a p p l i c a t i o n o f
t h i s approach t o the s t i f f e n e d composite p a n e l is d e p i c t e d i n F i g . 8 where a
h e u r i s t i c choice is made of t h e mathematical models. For t h e p a n e l a
boundary-layer ( o r a h i g h e r o r d e r two-dimensional) t h e o r y i s used n e a r t h e

155
 c u t o u t , f o l l o w e d by a f i r s t - o r d e r shear d e f o r m a t i o n t h e o r y , and t h e n a
c l a s s i c a l s h e l l t h e o r y . For t h e s t i f f e n e r s , a p l a t e t h e o r y is used n e a r the
c u t o u t , f o l l o w e d by a t h i n - w a l l e d beam t h e o r y and t h e n a shear d e f o r m a t i o n o r
a c l a s s i c a l beam t h e o r y . The e f f e c t i v e implementation of t h i s approach
r e q u i r e s the following:
1 . Systematic p r o c e d u r e f o r g e n e r a t i n g t h e h i e r a r c h y of mathematical
models ( e . g . , t h e method of i n i t i a l f u n c t i o n s o f V. Z . Vlasov - VLASOV &
LEONTEV C19661; and I Y E N G A R , CHANDRASHEKHARAN & SEBASTIAN C19741); o r t h e
a s y m p t o t i c i n t e g r a t i o n t e c h n i q u e - GOLDENVEIZER C19761)
2 . C r i t e r i a f o r t h e a d a p t i v e r e f i n e m e n t of t h e mathematical model
3. Treatment of t h e i n t e r f a c e s between t h e d i f f e r e n t r e g i o n s
4. TREATMENT OF INTERFACES

The t r e a t m e n t of i n t e r f a c e s is one of t h e k e y elements of the g l o b a l -

l o c a l a n a l y s i s . The two commonly u s e d a p p r o a c h e s f o r m a i n t a i n i n g d i s p l a c e m e n t
c o m p a t i b i l i t y and t r a c t i o n r e c i p r o c i t y a t t h e i n t e r f a c e s are: 1 ) Lagrange
m u l t i p l i e r method; and 2 ) p e n a l t y f u n c t i o n method. The second approach has
t h e a d v a n t a g e t h a t i t does n o t lead t o any e x t r a unknowns o r e q u a t i o n s (DELVES
& HALL C19791). The n u m e r i c a l problem a s s o c i a t e d w i t h i n c r e a s i n g t h e p e n a l t y
w e i g h t , t o meet c o n s t r a i n t s a t i s f a c t i o n t o l e r a n c e s can be overcome by u s i n g
t h e i t e r a t i v e procedure d e s c r i b e d i n FELIPPA C19781.

5. Q U A L I T Y C O N T R O L OF N U M E R I C A L S O L U T I O N S

One of t h e most d i f f i c u l t a s p e c t s of numerical modeling i s t h e v a l i d a t i o n

of t h e r e s u l t s and e n s u r i n g t h a t a g i v e n model is a d e q u a t e f o r t h e p a r t i c u l a r
problem a t hand. I n g e n e r a l , t h e r e a r e t h r e e t y p e s of e r r o r s i n t h e n u m e r i c a l
s o l u t i o n . These e r r o r s are (see UTKU & MELOSH C19841):
1 . Mathematical modeling e r r o r s , which r e s u l t from t h e s i m p l i f i c a t i o n s
--I_---

made i n a b s t r a c t i n g t h e mathematical model from t h e r e a l s t r u c t u r e .

2. D i s c r e t i z a t i o n errors, which a r e caused by t h e numerical d i s c r e t i -
z a t i o n of t h e c o n t i n u o u s mathematical model.
3. Mza>ipulation-errors, which are c a u s e d by: a ) t h e f i n i t e p r e c i s i o n
of t h e computers ( l i m i t a t i o n i n r e p r e s e n t i n g r e a l numbers due t o t h e f i n i t e -
n e s s of t h e computer word l e n g t h ) ; and b ) t h e e r r o r s r e s u l t i n g i n t h e p r o c e s s
of s o l v i n g t h e e q u a t i o n s of t h e d i s c r e t e model ( e . g . , u s i n g i t e r a t i v e
methods).
I n t h i s paper o n l y t h e second t y p e , namely: d i s c r e t i z a t i o n e r r o r s , i s
c o n s i d e r e d . There a r e two c l a s s i c a l a p p r o a c h e s f o r e s t i m a t i n g t h e s e e r r o r s
(KELLY, GAG0 & Z I E N K I E W I C Z C19831).
1 . E x t e n s i o n methods - based on r e a n a l y s i s of t h e s t r u c t u r e on a
sequence of meshes of i n c r e a s i n g r e f i n e m e n t s ( h e x t e n s i o n ) ; w i t h a h i e r a r c h i c
s e t of i n t e r p o l a t i o n polynomials ( p e x t e n s i o n ) ; o r u s i n g a combination o f t h e
two (h-p e x t e n s i o n ) .
2. Dual ( o r complementary) -- p r o c e d u r e - based on o b t a i n i n g two s o l u t i o n s
w i t h two d i f f e r e n t computer programs t o p r o v i d e bounds on g l o b a l r e s p o n s e
characteristics.
Both o f t h e s e approaches are t o o e x p e n s i v e f o r p r a c t i c a l implementation.
I n r e c e n t years, c o n s i d e r a b l e e f f o r t has been devoted t o t h e development
of - a p o s t e r i o r i e r r o r estimates t h a t are based on i n f o r m a t i o n o b t a i n e d d u r i n g
t h e s o l u t i o n p r o c e s s i t s e l f (KELLY, GAG0 & Z I E N K I E W I C Z C19831; BABUSKA & CUI
C19851; SZABO C19841; and SPECHT C19841). For s t r u c t u r a l mechanics problems,
a l l these e r r o r estimates were developed f o r c o m p a t i b l e d i s p l a c e m e n t models.

156
 Among t h e e r r o r e s t i m a t o r s developed t o date are t h e f o l l o w i n g two:
1 . Local energy norm e r r o r . T h i s is t h e s q u a r e r o o t of t h e s t r a i n
e n e r g y o f T h e e r r o r . T h i s is a l o c a l - g l o b a l measure i n t h e s e n s e t h a t i t
measures a g l o b a l r e s p o n s e c h a r a c t e r i s t i c , l o c a l l y ( w i t h i n a n i n d i v i d u a l
e l e m e n t ) . I n n o n l i n e a r problems, t h e measure can be used by l i n e a r i z a t i o n
around a n o n l i n e a r s o l u t i o n and e v a l u a t i n g t h e e n e r g y norm of t h e l i n e a r i z e d
problem.
2. I n t e r i o r and boundary r e s i d u a l s . These r e p r e s e n t t h e e q u i l i b r i u m
defects i n t h e i n t e r i o r on t h e p o r t i o n of t h e boundary where t r a c t i o n s are
p r e s c r i b e d as well as t h e jumps i n t h e t r a c t i o n s a t i n t e r e l e m e n t b o u n d a r i e s .
For uniform g r i d s w i t h l i n e a r , b i l i n e a r and t r i l i n e a r s h a p e f u n c t i o n s t h e
c o n t r i b u t i o n s of t h e jumps dominate t h e r e s i d u a l a n d , t h e r e f o r e , t h e r e s i d u a l
can be approximated by t h e t r a c t i o n jumps. A s i m p l e approximate method o f
e v a l u a t i n g these r e s i d u a l s f o r e l e m e n t s w i t h h i e r a r c h i c shape f u n c t i o n s was
g i v e n i n KELLY, GAG0 & Z I E N K I E W I C Z C19831.
The e r r o r e s t i m a t o r s , i n a d d i t i o n t o p r o v i d i n g i n f o r m a t i o n a b o u t t h e
q u a l i t y of s o l u t i o n s , form t h e b a s i s f o r a d a p t i v e improvement of t h e f i n i t e -
element s o l u t i o n . T h i s can be accomplished by e n r i c h i n g o r improving t h e
a p p r o x i m a t i o n u s i n g one of t h e f o l l o w i n g approaches ( o r p o s s i b l y , t h e i r
combinations).
1 . R e f i n i n g t h e mesh
2. Moving t h e nodes (node r e l o c a t i o n )
3. I n c r e a s i n g t h e l o c a l o r d e r of t h e approximation
4. Using t h e i t e r a t e d defect c o r r e c t i o n method
The t h i r d approach has t h e a d v a n t a g e s o v e r t h e f i r s t two of b e i n g easy t o
implement and o f p r o v i d i n g a s i m p l e formula f o r t h e e r r o r e s t i m a t o r . The
f o u r t h approach is based on u s i n g t h e numerical s o l u t i o n o b t a i n e d t o con-
s t r u c t a pseudo o r n e i g h b o r i n g problem whose e x a c t s o l u t i o n i s known ( e . g . ,
polynomial o r s p l i n e i n t e r p o l a t i o n of t h e d i s c r e t e n u m e r i c a l s o l u t i o n ) . The
pseudo problem i s t h e n s o l v e d u s i n g t h e same f i n i t e - e l e m e n t model as t h a t used
f o r t h e o r i g i n a l problem. The e r r o r i n t h e pseudo problem is assumed t o be a
c l o s e a p p r o x i m a t i o n o f t h e e r r o r i n t h e o r i g i n a l problem and i s used a s a
c o r r e c t i o n t o t h a t s o l u t i o n . The t e c h n i q u e has been s u c c e s s f u l l y a p p l i e d t o
t h e n u m e r i c a l s o l u t i o n o f s t i f f systems of o r d i n a r y d i f f e r e n t i a l e q u a t i o n s and
a p p e a r s t o have h i g h p o t e n t i a l f o r a p p l i c a t i o n t o f i n i t e - e l e m e n t boundary
value-problems (ZADUNAISKY C19761; FRANK, HERTLING & MONNET [ 19831 ; and BOHMER
& STETTER [ I 9841).

6. POSTPROCESSING AND STRESS CALCULATION

I n d i s p l a c e m e n t f i n i t e - e l e m e n t models, t h e s t r a i n energy of t h e s t r u c t u r e
is t h e h i g h e s t q u a l i t y i n f o r m a t i o n t h a t can be e x t r a c t e d from t h e f i n i t e -
element s o l u t i o n . The a c c u r a c y and r a t e of convergence of stresses depend on
how (and where) t h e y are computed. S e v e r a l approaches have been s u g g e s t e d f o r
improving t h e a c c u r a c y of stress c a l c u l a t i o n s (see, f o r example, HINTON &
CAMPBELL [ 19743 ; CAREY [ 19821 ; Z I E N K I E W I C Z , X I - K U I & NAKAZAWA [19851; and
BABUSKA & MILLER C19841). Among these are:
1 . E v a l u a t i n g t h e stresses a t numerical q u a d r a t u r e p o i n t s and deter-
mining t h e i r v a l u e s a t t h e nodes by e x t r a p o l a t i o n
2. Computing t h e stresses u s i n g t h e discarded s t r u c t u r a l e q u a t i o n s
( c o r r e s p o n d i n g t o p r e s c r i b e d d i s p l a c e m e n t boundary c o n d i t i o n s )
3. Averaging o r smoothing based on p r o j e c t i o n t e c h n i q u e s
4. Using i n f l u e n c e f u n c t i o n methods

157
 The f i r s t approach i s by f a r t h e most commonly used. Superconvergence
(increased accuracy and improved r a t e s of convergence) has been observed f o r
s t r e s s e s evaluated a t quadrature points. Note t h a t the s t r e s s e s ~-cannot have a
f a s t e r convergence than t h e s t r a i n energy. The second approach is p a r t i -
c u l a r l y useful f o r evaluating t h e s t r e s s e s a t the boundaries. The t h i r d and
fourth approaches improve the accuracy of s t r e s s predictions through the
f i l t e r i n g of spurious o s c i l l a t i o n s . A systematic assessment of the l a t t e r two
approaches is needed.

7. FUTURE D I R E C T I O N S FOR RESEARCH

Global-local a n a l y s i s s t r a t e g i e s have h i g h p o t e n t i a l f o r t h e r e l i a b l e and

e f f i c i e n t prediction of the nonlinear response of complex s t r u c t u r e s subjected
t o d i f f e r e n t loadings. To r e a l i z e t h i s p o t e n t i a l the global-local s t r a t e g y
must include the following seven key elements:
1 . Rational s e l e c t i o n of a hierarchy of mathematical models f o r
d i f f e r e n t p a r t s of the s t r u c t u r e and a s t r a t e g y f o r the adaptive refinement of
these models
2. - Use- of global (-- o r macro) elements f o r d i s c r e t i z a t i o n whenever approp-
r i a t e w i t h i n t e r f a c e conditions s a t i s f i e d v i a e x t e r i o r penalty method
3. Use of r e l i a b l e f a i l u r e c r i t e r i a and d i s c r e t e elements t h a t account
f o r the progressive f a i l u r e mechanisms
4. Application of operator s p l i t t i n- g i n- conjunction w i t h reduction
--
method f o r g e n e r a t i n g t h e response of the complex s t r u c t u r e by u s i n g large
perturbations from t h e response of a simpler s t r u c t u r e
5. Postprocessing t o increase the accuracy of s t r e s s c a l c u l a t i o n s
6. - Q u a l i t y control of numerical s o l u t i o n s
7. Exploiting the computational power of new m u l t E o c-
---I e-s s o r machines
through p a r a l l e l i z a t i o n of the problem formulation, computational s t r a t e g y a s
well as the numerical algorithms.
Each of the aforementioned key elements requires major development t o
reach t h e l e v e l of maturity needed f o r r o u t i n e inclusion i n the global-local
s t r a t e g y . To t h i s end, t h e r e a r e pacing items t h a t m u s t be addressed by t h e
research community. Among t h e items t h a t pace the progress of global-local
methodologies a r e t h e development of:
1 . C r i t e r i a and control parameters f o r s e l e c t i n g the mathematical model,
as well as adaptive s t r a t e g i e s f o r r e f i n i n g t h e model whenever needed. Also,
s t r a t e g i e s f o r blending regions of d i f f e r e n t s t r u c t u r a l behavior (e.g.,
boundary l a y e r , two/three dimensional models of t h e s t r u c t u r e ) .
2 . Reliable f a i l u r e c r i t e r i a and s h e l l elements t h a t account f o r t h e
composite delamination mechanisms.
3. Simple and accurate techniques f o r s t r e s s c a l c u l a t i o n s which provide
t h e same accuracy as t h a t of t h e s t r a i n energy.
4. Error estimation and adaptive improvement s t r a t e g i e s . T h i s is an
area which r e q u i r e s more a t t e n t i o n by researchers. In p a r t i c u l a r , e r r o r
estimators t h a t s a t i s f y t h e following four c r i t e r i a need t o be developed f o r
nonlinear analysis:
a ) provide r e l i a b l e l o c a l assessment of the e r r o r w i t h e x t r a p o l a t i o n t o
global estimation
b ) computationally inexpensive t o evaluate
c ) applicable t o a wide c l a s s of d i s c r e t e elements
d ) easy t o use i n conjunction w i t h adaptive improvement
5. P a r a l l e l computational s t r a t e g i e s f o r multiprocessor computers.

158
These s t r a t e g i e s i n c l u d e t h e u s e o f : a ) p r i m i t i v e v a r i a b l e s (e.g., three-
f i e l d mixed f o r m u l a t i o n ) ; b ) domain decomposition ( w i t h m i n i m i z a t i o n of
i n t e r f a c e s ) ; and c ) o p e r a t o r s p l i t t i n g t o uncouple t h e algebraic e q u a t i o n s .
I n a d d i t i o n , t h e i n t e n s e research e f f o r t c u r r e n t l y under way on p a r a l l e l
n u m e r i c a l a l g o r i t h m s (see, f o r example, NOOR C19831, and PADDON C19841) s h o u l d
be brought t o bear on g l o b a l - l o c a l methodologies. Due t o t h e wide v a r i e t y of
new p a r a l l e l computers, t h e idea of d e v e l o p i n g macro a l g o r i t h m s which a r e
e f f i c i e n t on d i f f e r e n t p a r a l l e l machines s h o u l d be i n v e s t i g a t e d . The
n u m e r i c a l tasks i n these a l g o r i t h m s a r e performed by d i f f e r e n t programs which
are o p t i m i z e d f o r each of t h e i n d i v i d u a l machines.

CONCLUDING REMARKS

A r e v i e w and an a s s e s s m e n t were made of g l o b a l - l o c a l s t r a t e g i e s f o r t h e

n o n l i n e a r a n a l y s i s of s t r u c t u r e s . To p r o v i d e a f o c u s f o r t h e d i s c u s s i o n two
benchmark problems of p o s t b u c k l i n g o f l a m i n a t e d composite c y l i n d r i c a l p a n e l s
were selected. The major characteristics of these problems were i d e n t i f i e d .
A number of g l o b a l - l o c a l a n a l y s i s s t r a t e g i e s were r e v i e w e d , t h e i r
p o t e n t i a l f o r s o l v i n g t h e benchmark problems d i s c u s s e d and t h e i r s h o r t c o m i n g s
d e l i n e a t e d . Also, e r r o r e s t i m a t i o n and p o s t p r o c e s s i n g t e c h n i q u e s were
reviewed.
The items t h a t pace t h e p r o g r e s s of g l o b a l - l o c a l methodologies a r e
i d e n t i f i e d and a r e , t h e r e f o r e , recommended as f u t u r e d i r e c t i o n s f o r
research. These i n c l u d e t h e c o u p l i n g o f d i f f e r e n t g l o b a l - l o c a l methodologies;
p o s t p r o c e s s i n g and stress c a l c u l a t i o n methods; q u a l i t y c o n t r o l and a d a p t i v e
improvement of numerical a l g o r i t h m s ; and e f f e c t i v e c o m p u t a t i o n a l s t r a t e g i e s
f o r new computing s y s t e m s .

REFERENCES

BABUSKA, I. & MILLER, A. (19841, "The P o s t p r o c e s s i n g Approach i n t h e F i n i t e

Element Method - Part 1 : C a l c u l a t i o n of Displacements, Stresses, and
Other Higher D e r i v a t i v e s of t h e D i s p l a c e m e n t s , " I n t e r n a t i o n a l J o u r n a l
f o r Numerical Methods i n E n g i n e e r i n g , Vol. 20, pp. 1085-1109.

BABUSKA, I. & G U I , W. ( 1 9 8 5 ) , "Basic P r i n c i p l e s o f Feedback and Adaptive

Approaches i n t h e F i n i t e Element Method," T e c h n i c a l Note BN-1042,
I n s t i t u t e f o r P h y s i c a l S c i e n c e and Technology, U n i v e r s i t y of
Maryland.

BELYTSCHKO, T. & HUGHES, T. J. R. ( e d s . ) ( 1 9 8 3 ) , Computational Methods f o r

T r a n s i e n t A n a l y s i s , North Holland P u b l i s h i n g Company.

BOHMER, K . & STETTER, H. J. ( e d s . ) (19841, - - C o r r e c t i o n Methods

Defect - Theory
and A p p l i c a t i o n s , Springer-Verlag.

CAREY, C. F. ( 1 9 8 2 ) , " D e r i v a t i v e C a l c u l a t i o n s From F i n i t e Element S o l u t i o n s , t f

Computer Methods i n Applied Mechanics and E n g i n e e r i n g , Vol. 35, pp. 1-14.

DELVES, L. M. & HALL, C . A. (19791, "An I m p l i c i t Matching P r i n c i p l e f o r G l o b a l

Element C a l c u l a t i o n s , " J o u r n a l of t h e I n s t i t u t e o f Mathematics and
A p p l i c a t i o n s , Vol. 23, pp. 223-234.

159
 DELVES, L. M. & PHILLIPS, C. (19801, "A Fast Implementation of the Global
Element Method," Journal of the Institute of Mathematics and Its
Applications, Vol. 25, pp. 177-197.

DONG, S. B. (1983), "Global-Local Finite Element Methods," in State-of-the-

Art Surveys
- -on Finite --- Element Technology, ed. by A. K. Noor and
W. D. Pilkey, American Society of Mechanical Engineers, pp. 451-474.

FELIPPA, C. A. (1978), "Iterative Procedures for Improving Penalty Function

Solutions of Algebraic Systems," International Journal for Numerical
Methods in Engineering, Vol. 12, pp. 821-836.

FRANK, R., HERTLING, J. & MONNET, J. P. (1983), !'The Application of Iterated

Defect Correction to Variational Methods for Elliptic Boundary Value
Problems,!! Computing, Vol. 30, pp. 121-135.

Theory of Elastic Thin --

GOLDENVEIZER, A. L. (1976), - Shells, Nauka.

HACKBUSCH, W. & TROTTENBERG, W. (eds.) (19821, - Multigrid Methods, _Lecture

Notes
-- in Mathematics
-- No.- 960. Proceedings of the conference held at
Koln-Portz, Germany, N O ~ .23-27, 1981, Springer-Verlag, Berlin.

HINTON, E. & CAMPBELL, J. (1974), "Local and Global Smoothing of Discontinuous

Finite Element Functions Using a Least Square Method," International
Journal for Numerical Methods in Engineering, Vol. 8, pp. 461-480.

HIRAI, I., UCHIYAMA, Y., MIZUTA, Y. & PILKEY, W. D. (19851, "An Exact Zooming
Method," Finite Elements in Analysis and Design, Vol. 1 , pp. 61-69.

HUGHES, T. J. R., RAEFSKY, A,, MULLER, A,, WINGET, J. & LEVIT, I. (1984),
"A Progress Report on EBE Solution Procedures in Solid Mechanics,"
Proceedings of the Second International Conference on Numerical Methods
for Nonlinear Problems, Barcelona, Spain, April 9-13, 1984.

IYENGAR, K. T. S. R., CHANDRASHEKHARAN, K. & SEBASTIAN, V. K. (1974), "On the

Analysis of Thick Rectangular Plates," Ingenieur-Archiv, Vol. 43,
PP* 317-3300
KAMIYA, N. & SAWAKI, Y. (19821, "An Integral Equation Approach to Finite De-
flection of Elastic Plates," International Journal of Nonlinear
Mechanics, Vol. 17, No. 3, pp. 187-194.

KAMIYA, N., SAWAKI, Y. & NAKAMURA, Y. (19841, "Postbuckling Analysis by the

Boundary Element Method," Engineering Analysis, Vol. 1, No. 1, pp. 40-44.

KARDESTUNCER, H. (ed.) (1985), Unification of--

Finite Element Software Systems,
North-Holland Publishing Company.

KELLY, D. W., GAGO, J. P. de S. R. & ZIENKIEWICZ, 0. C. (1983), "A Posteriori

Error Analysis and Adaptive Processes in the Finite Element Method: Part
I - Error Analysis; Part I1 - Adaptive Mesh Refinement," International
Journal for Numerical Methods in Engineering, Vol. 19, pp. 1593-1619
and 1621-1 656.

160
K N I G H T , N. F., GREENE, W. H. & STROUD, W. J. (19851, IINonlinear Response of
a B l a d e - S t i f f e n e d Graphite-Epoxy Panel w i t h a D i s c o n t i n u o u s S t i f f e n e r :
Work i n P r o g r e s s , " P r o c e e d i n g s of t h e Workshop on Computational Methods
of S t r u c t u r a l ' M e c h a n i c s and Dynamics, NASA Langley Research C e n t e r ,
Hampton, VA, June 19-21, 1985.

K N I G H T , N. F. & STARNES, J . H. ( 1 9 8 4 ) , IfPostbuckling Behavior of A x i a l l y Com-

pressed Graphite-Epoxy C y l i n d r i c a l P a n e l s w i t h C i r c u l a r H o l e s , f f i n
C o l l a p s e A n a l y s i s of S t r u c t u r e s , e d . by L. H. Sobel and K. Thomas,
PVP-Vol. 84, American S o c i e t y o f Mechanical E n g i n e e r s , pp. 153-167.

L I U , W . K . , BELYTSCHKO, T. & P A R K , K . C . ( 1 9 8 4 ) , I n n o v a t i v e Methods f o r

N o n l i n e a r Problems, P i n e r i d g e Press, Swansea, United Kingdom.

MIKLOSKO, J. & KOTOV, V. E. ( e d s . ) ( 1 9 8 4 ) , - ---

- S o f t w a r e and Hardware
Algorithms,
-
of P a r a l l e l Computers, Springer-Verlag.

MOTE, C . D. ( 1 9 7 1 ) , "Global-Local F i n i t e Element," I n t e r n a t i o n a l J o u r n a l f o r

Numerical Methods i n E n g i n e e r i n g , Vol. 3, pp. 565-574.

(19821, "On Making Large Nonlinear Problems S m a l l , " Computer

N O O R , A. K .
Methods i n Applied Mechanics and E n g i n e e r i n g , " Vol. 34, pp. 955-985.

( e d . ) (19831, ----
N O O R , A. K . ---
Impact of N e w Computing Systems on Computational
-- The American S o c i e t y o f Mechanical E n g i n e e r s .
Mechanics,

NOOR, A. K . (19851, "Reduction Method f o r t h e Nonlinear A n a l y s i s o f Symmetric

A n i s o t r o p i c P a n e l s , " P r o c e e d i n g s of t h e Europe-U.S. Symposium on F i n i t e
Element Methods f o r N o n l i n e a r Problems, Norwegian I n s t i t u t e o f
Technology, Trondheim, Norway, Aug. 12-15, 1985.

N O O R , A. K. & ANDERSEN, C. M. ( 1 9 8 2 ) , "Mixed Models and R e d u c e d / S e l e c t i v e

I n t e g r a t i o n Displacement Models f o r Nonlinear S h e l l A n a l y s i s , " I n t e r -
n a t i o n a l J o u r n a l f o r Numerical Methods i n E n g i n e e r i n g , Vol. 18,
pp. 1429-1 454.

NOOR, A. K. & PETERS, J. M . (19831, "Recent Advances i n Reduction Methods

f o r I n s t a b i l i t y A n a l y s i s of S t r u c t u r e s , " Computers and S t r u c t u r e s ,
Vol. 1 6 , pp. 67-80.

NOOR, A. K . & PETERS, J. M. (19851, "Model-Size Reduction Technique f o r t h e

A n a l y s i s of Symmetric A n i s o t r o p i c S t r u c t u r e s , " E n g i n e e r i n g Computations,
Vol. 2, NO. 4 , pp. 285-292.

NOOR, A. K . & P I L K E Y , W . D. ( e d s . ) ( 1 9 8 3 ) , S t a t e - o f - t h e - A r t Surveys on F i n i t e

---
Element Technology, The American S o c i e t y o f Mechanical E n g i n e e r s .

NOOR, A. K . , ANDERSEN, C . M. & T A N N E R , J . A. (19851, " E x p l o i t i n g Symmetries

i n t h e Modeling and A n a l y s i s of Tires,'I P r o c e e d i n g s of t h e Workshop on
Computational Methods of S t r u c t u r a l Mechanics and Dynamics, NASA Langley
Research C e n t e r , Hampton, V i r g i n i a , J u n e 19-21 , 1985.

161
 NOOR, A. K . , MATHERS, M. D. & ANDERSON, M. S. ( 1 9 7 7 ) , " E x p l o i t i n g Symmetries
f o r E f f i c i e n t P o s t b u c k l i n g A n a l y s i s o f Composite P l a t e s , " A I A A J o u r n a l ,
Vol. 15, NO. 1 , pp. 24-32.

NOOR, A. K . , STORASSLI, 0. 0. & FULTON, R. E. ( 1 9 8 4 ) , I1Impact of New Computing

Systems on C o m p u t a t i o n a l Mechanics and F l i g h t V e h i c l e S t r u c t u r e s
Technology,11 AGARD Report No. 706.

-
PADDON, D . J . ( e d . ) ( 1 9 8 4 ) , Supercomputers and P a r a l l e l Computation, C l a r e n d o n
~

Press, Oxford.

PATTIBIRAMAN, J . , RAMAMURTI, V. & REDDY, D. V. ( 1 9 7 4 ) , l l S t a t i c s and Dynamics

o f E l a s t i c S h e l l s w i t h C u t o u t s - A Review," J o u r n a l o f S h i p R e s e a r c h ,
Vol. 18, NO. 2 , pp. 113-126.

-~
SCHENDEL, U . (19841, I n t r o d u c t i o n t o Numerical Methods f o r P a r a l l e l ComFters, -
E l l i s Horwood, L t d .

SPECHT, B. ( 1 9 8 4 ) , "A General C o n s t r u c t i o n o f Local Error Estimates f o r

Conforming F i n i t e Elements," Computers and S t r u c t u r e s , Vol. 1 9 ,
pp. 81 5-822.

STANLEY, G . M. (Aug. 19851, "Continuum-Based S h e l l E l e m e n t s , " Ph.D.

Dissertation, Stanford University.

SZABO, B. A. ( 1 9 8 4 ) , llEstimation and C o n t r o l o f E r r o r Based on p-Convergence,"

P r o c e e d i n g s o f t h e I n t e r n a t i o n a l Conference on Accuracy Estimates and
Adaptive Refinements i n F i n i t e Element Computations, L i s b o n , P o r t u g a l ,
June 19-22, 1984.

UNDERWOOD, P. G. (1983), llDynamic Relaxation - A Review," i n C o m p u t a t i o n a l

Methods f o r T r a n s i e n t Dynamic A n a l y s i s , e d . by T. B e l y t s c h k o and
T. J. R. Hughes, North H o l l a n d P u b l i s h i n g Company.

UTKU, S. & MELOSH, R. J. (19841, " S o l u t i o n E r r o r s i n F i n i t e Element

A n a l y s i s , l l Computers and S t r u c t u r e s , Vol. 18, pp. 379-393.

VLASOV, V. Z. & LEONTEV, U . N. (19661, "Beams, P l a t e s and S h e l l s on E l a s t i c

F o u n d a t i o n s , It NASA TT-F-357.

WILKINS, D. J. (19831, "A P r e l i m i n a r y Damage Tolerance Methodology f o r Com-

p o s i t e S t r u c t u r e s , " i n F a i l u r e A n a l y s i s and Mechanisms o f F a i l u r e o f
F i b r o u s Composite S t r u c t u r e s , NASA CP-2278, pp. 67-93.

ZADUNAISKY, P. E. (19761, "On t h e Estimation of Errors P r o p a g a t e d i n t h e

Numerical I n t e g r a t i o n o f O r d i n a r y D i f f e r e n t i a l E q u a t i o n s , 1 1 Numer.
Math., Vol. 2 7 , pp. 21-39.

Z I E N K I E W I C Z , 0. C . , X I - K U I , L. & NAKAZAWA, S. ( 1 9 8 5 ) , " I t e r a t i v e S o l u t i o n

o f Mixed Problems and t h e S t r e s s Recovery P r o c e d u r e s ,11 Communications
i n Applied Numerical Methods, Vol. 1 , pp. 3-9.

162
TABLE 1. CHARACTERISTICS OF THE FINITE-ELEMENT MODELS USED FOR THE BENCHMARK
PROBLEMS

Blade-Stiffened Panel
(Knight, Greene 6 Stroud [1985])
Unstiffened Panel
Mesh 1 Mesh 2

Type of Element Mixed, 9-Noded + Hybrid 4-Noded

Number of 188 544

66
Elements (376) (1088)

Number of Displace-
1266 3504
ment Degrees of 1338
(2432) (7008)
Freedom'

TABLE 2. PARTIAL LIST OF THE COMMONLY USED TECHNIQUES FOR GLOBAL AND LOCAL
ANALYSES

o Global variational methods o Discrete element methods

o Discrete element methods o Conventional finite elements
o Conventional finite elements o Special elements
o Global element method o Superelements
o Boundary element method o Boundary element method
o Analytic solutions o Analytic solutions

163
 16-PLY UNSTIFRNEO PANEL BLADE STIFFENED PANEL

f1
I
I1

L = 1 = 0,3556 m
1 2
R = 0.381 m.
h = 2.276 x m.
E = 1.31 x 10l1 Pa
L
E = 1.303 x lolo Pa 25-PLY SKIN
T
9 Boundary conditions 1~45/02/~45/O~~45/Oj/~45/O~~45/02/~451,
G = 6 . 4 1 2 ~10 Pa
LT 9
C = 5.102 x 10 Pa A t x l = * L /2
n 1
u = &9/2, u =w= d = d = o
24-PLY BLADES
uLT= 0.392 1 2 1 2 I I 45/O2d; 45 I T
Fiber orientation: Atx,= t l /2
- 2
[ + 45/90/02/90/ 7 451 w= 0 3 0
1

F i g u r e 1. Benchmark problems c o n s i d e r e d i n p r e s e n t s t u d y .

- 128
LINEAR
SOLUTION '

-
L2
ETh3 - 64

- 32

1
0 17.5 35.0 52.5 70.0 0 0.15 0.30 0.45 0.60

F i g u r e 2. N o n l i n e a r r e s p o n s e of u n s t i f f e n e d p a n e l s u b j e c t e d t o end
s h o r t e n i n g ( s e e f i g . 1).

ORfGlNAL PAGE fS
OF POOR QUALITY
164
 =i)
fL.

Figure 3. Normalized contour pilots for displacements

NL2
and strains in untiffened panel at - 3
= 126.6
(see fig. 1). ETh

Figure 4 . Application of zooming technique to unstif€ened panel.

165
 GLOBAL GLOBAL ELEMENTS CONVENTIONAL FINITE ELEMENTS GLOBAL ELEMENTS
ANALY S I S

LOCAL CONVENTIONAL (OR SPECIAL) BOUNDARY ELEMENTS BOUNDARY ELEMENTS

ANALYSIS FINITE ELEMENTS

Figure 5. Simultaneous application of two discrete element methods to the

analysis of unstiffened panel.

&
ETh3
-64

-32
p Fi (4 VECTORS)

0 0.15 0.30 0.45 0.60 0 15 30 45 60

wJh 5
ETh4

Figure 6. Accuracy of normal displacement w and total strain energy U

obtained by reduction method. Unstiffened panel subjected to axial end
shortening (see fig. 1).

166
 I 1
aw
-I
I '1
-
aw
max.

Figure 7. Normalized contour plots for the global approximation

vectors - unstiffened panel subjected to axial end shortening
(see fig. 1).

-
C L A S S I CAL
B E A M THEORY - C L A S S I CA L THEORY
T H IN-IIVA LLED 7
B E A M THEORY - SHEAR
DEFORMAT I ON
PLATE THEORY -.
THEORY
BOUNDARY
7 LAYER THEORY

Figure 8. Heuristic approach for selecting a hierarchy of

mathematical models for blade stiffened panel.

167
 N89- 24648

APPLICATION OF THE P-VERSION OF THE FINITE-ELEMENT METHOD

TO GLOBAL-LOCAL PROBLEMS

Barna A. Szab6
Washington University
St. Louis, MO

1. INTRODUCTION

The following is a brief survey of some recent developments in finite-element

analysis technology which bear upon the three main research areas under consid-
eration in this workshop: (1) analysis methods; (2) software testing and quality
assurance; and (3) parallel processing.
The variational principle incorporated in a finite-element computer program,
together with a particular set of input data, determines the exact solution corre-
sponding to that input data. Most finite-element analysis computer programs are
based on the principle of virtual work. In the following we consider only programs
based on the principle of virtual work and denote the exact displacement vector
field corresponding to some specific set of input data by iiEx. The exact solution
i i E x is independent of the design of the mesh or the choice of elements. Except for
very simple problems, or specially constructed test problems, i i E x is not known.
We perform a finite-element analysis (or any other numerical analysis) because
we wish to make conclusions concerning the response of a physical system to
certain imposed conditions, as if iiEx were known. We know the finite-element
solution only which we denote by iiFE.The solution i i F E depends not only on the
variational principle and the input data but (also on the finite-element mesh and
the choice of elements. We will assume that the finite-elements are exactly and
minimally conforming and therefore the elements are completely characterized by
their polynomial degree. We therefore control iiFEby mesh design and the choice
of the polynomial degree of elements.
We wish to compute i i F E so that i i F E is close to i i E x in some sense. For example,
if we are interested in determining a stress intensity factor then we wish to have
the stress intensity factor computed from i i F E to be close to the stress intensity
factor computed from iiEx within some prespecified level of tolerance 7 . In general,
we wish to determine functionals Q , ( ~ F E ) (i = 1, 2 . . . , n ) so that:

The question naturally arises: how can we tell whether Q ; ( i i F E ) is close to Q ; ( i i E X )

if we do not know zEX? The answer is: by performing extensions. ,Both the
estimation and control of error are based on extensions.

PRECEDING PAGE BLANK NOT FILMED 169

 Extensions are systematic increases in the number of degrees of freedom either
by mesh refinement, increase in the polynomial degree of elements or a combination
of both. If the extension is by mesh refinement then the process is called h-
extension*. If the extension is by increase in the polynomial degree of elements
then the process is called p-extension**. If the extension is by a combination
of proper mesh refinement and concurrent increase in the polynomial degree of
elements then it is called h-p extension. Having performed an extension, we may
draw conclusions concerning the overall quality of the approximate solution and
the quality of any functional computed from C F E .

2. OVERALL QUALITY

The overall quality of approximation can be judged in terms of the estimated

error in energy norm and errors in equilibrium. Estimation of error in energy
norm is outlined in some detail and an example is presented. Procedures for
assessment of the quality of approximation in terms of errors in equilibrium are
briefly discussed.

2.1. Estimation of error in energy norm.

We know that the strain energy of the error U(ilEx - C F E ) must decrease mono-
tonically as we systematically refine the mesh or increase the polynomial degree
of elements and a well developed, elaborate theoretical basis exists for the estima-
tion of error in energy norm for the h-, p- and h-p extension processes. (See, for
example, [1,2,3,4,5].)
The error in energy norm is defined as:

where n represents the solution domain and U represents the strain energy. 11 U E X -
i i F E I I E ( n ) is closely related to the root- mean-square of error in stresses [6].

In the case of h- and p-extensions the estimate is of the form:

where k and p are positive constants, N is the number of degrees of freedom. In

the case of h-p extensions the estimate is of the form:

where k, 7 and e are positive constants. These estimators are 'sharp' for large N
values hence the 'less than or equal' (I) can be replaced by 'approximately equal'
( w ) in (3), (4) when N is large. Therefore from (3) for large N values we have:

log IldEX - ~ F IIE(n)

E log k - p log N (5)

If we plot log JJuEx

- uFEJJE(n)
versus log N we see a downward sloping straight line.
The absolute value of the slope is p , called the asymptotic rate of convergence.
When p is large then the error decreases rapidly as N is increased. When p is
small then the error decreases slowly. Of course, the error also depends on k which
* h represents the size of elements. h-Eztension involves letting h,, --t 0.

** p represents the polynomial degree of elements. p-Eztension involves letting pmin -+ 00.

170
is generally not known a priori, but can be estimated from data obtained from
properly performed extensions. This will be discussed later. When the estimate
is of the form (3)the rate of convergence is said to be algebraic.
When the estimate is of the form (4) and we plot 1ogI)iiEx - Z F E I I E ( n ) versus
logN then for large N values we see a downward curving line [2,3,4,5,7]. In this
case the rate of convergence is exponential:
log IliiEX - Z F E I I E ( n ) W l o g k - 7 (loge) N e (6)
where e is the base of the natural logarithm. If we plot l o g J J i i E x- iiFEIIE(n,versus
N e (not logN as before) then we see a downward sloping straight line. It is known
that under conditions which are generally satisfied in practice e 2 1/3 [5].
All error estimation techniques are based on extension. Because in general
the exact solution iiEx is not known, the only information available to us is how
the finite-element solution iiFE behaves when the number of degrees of freedom is
increased either through mesh refinement or increase in the polynomial degree of
elements. Such information, together with an estimate or hypothesis concerning
the magnitude of the error, or its rate of change with respect to N , is essential to
all error estimation. Of course, the estimate or hypothesis must be asymptotically
correct: as N + 00 the estimated error must approach zero at the same rate as the
true error does. Therefore the quality of error estimators should increase with N .
P-extension makes it convenient and inexpensive to obtain information con-
cerning the rate of change of U ( i i F E ) with respect to N . In the p-version hierarchic
basis functions are used. Therefore the stiffness matrices and load vectors corre-
sponding to polynomial degree p are embedded in the stiffness matrices and load
vectors of polynomial degree p + I. Once a solution is available for polynomial
degree pmas, all solutions corresponding to p = I, 2 , . .. ,pmaz - 1 can be readily and
inexpensively obtained. Specifically, we write:

Let us assume for the moment that U(CEX)> U ( i i F E ) . In that case:

We have three unknowns: U ( i i E x ) ,k and p. If we have three values of U ( i i F E )and

N corresponding to three different values of p, then we have three equations for
computing the unknowns. Let us denote these three values by Up, up-l,Up-2 and
Np, Np-l, Np--2 and U ( i i E x )by U . Then from (8) we have:
u - up Np-1
log -
log u - u,-1 NP
W (9)
Np-2
log -
Np-1
Denoting the right hand side of (9) by Q, we have:
u - up 8

u - up-l
To obtain an estimate of the exact strain energy U , we need to solve (10). The
solution is expected in the neighborhood of Up. Because convergence of the strain

171
 energy is monotonic, we know that U 2 Up when Up > Up-l. Conversely, U 5 Up
when Up < Up-l. Eq.(lO) would not be different if u(iiEx) < U ( i i F E ) ;therefore, the
restriction that U ( C E X ) > U ( C F E ) is not essential. Computational experience has
shown this estimate to be reliable and generally accurate, with the accuracy of the
estimate increasing with the accuracy of Up.

2.2. Example.
The following test problem is representative of plate and shell intersections
and reentrant corner problems in general. An L-shaped plane elastic body of
thickness t is loaded by tractions. The tractions are computed from a stress field
which satisfies the equilibrium and compatibility equations and the stress free
conditions along the reentrant edges. Specifically, the stress field corresponds to
the first (symmetric or 'Mode 1') term of the asymptotic expansion of iiEx about
the reentrant corner. (See, for example, [8].) Therefore the exact solution is
known. Specifically, the components of i i E x in the coordinate system shown in
Fig. 1 are:
A
u, = -r x [ ( K - Q(X
2G
+ 1)) COS X '6 - X COS(X - 2) 1'6 (1la)
A
uy = - r A
2G
[(K+ Q(X + 1))sinX 0 + X s i n ( X - 2) 6'1 (W
where A is a generalized stress intensity factor; X = 0.544483737;Q = 0.543075579; G is
the modulus of rigidity and K depends on Poisson's ratio Y only. For plane strain:
K. = 3 - 4v. We assume plane strain conditions and v = 0.3, therefore in this case
K = 1.8.

1 /

Fig. 1. L-shaped plane elastic body.

The stress tensor components are:

[T, =AX [p- Q(X + 1)) cos(^ - 1) e - (X - 1) cos(^ - 3) e]
cy = AX rx-l [ ( 2 + Q(X + 1)) cos(X - 1) 0 + (A - 1) cos(X - 3) 8)
T~~ = AX rx-' [(A - 1) sin(X - 3) B + Q(X + 1) s h ( X - 1) 6'1.

172
Because we know the exact displacement and stress fields we can compute the
strain energy of the exact solution:

A2azXt
~ ( C E X=) 4.15454423-
E
where E is the modulus of elasticity. The relative error in energy norm is defined
as follows:

Using the mesh shown in Fig. 2 finite-element solutions were obtained for
p = l to 8. The computations were performed by a new computer program, called
PROBE [9]. The number of degrees of freedom, the computed strain energy, the
estimated and true relative errors in energy norm, computed from eq. (14), are
shown in Table 1.

22

25a

Fig. 2. Mesh design.

The results presented in Table 1 are typical of the quality of the error estimate
we can obtain by means of the procedure described in Section 2.1. When the mesh
is strongly graded toward the point of singularity then the convergence path the
I
l o g ( e , ) E versus l o g N curve) looks like an inverted S [3,4,5,10]. For low N vaues
the rate of convergence is nearly exponential and the downward slope increases
with N . In this segment the estimated error is conservative. Near the inflection

173
 Table 1. Estimated and true relative error in energy norm.

_____ -- --
1 41 3.8860880 - - 25.42
2 119 4.1248326 - - 8.46
3 209 4.1481150 1.93 5.34 3.93
4 335 4.1526504 2.76 2.02 2.14
5 497 4.1536354 3.05 1.01 1.48
6 695 4.1539746 2.45 0.80 1.17
7 929 4.1541390 1.83 0.75 0.99
8 1199 4.1542378 1.39 0.75 0.86
-__-- __ __ --
00 00 4.1545442 1.09 - 0

point, (i.e. where the curvature of the convergence path changes from negative to
positive) the estimate is the least accurate and not conservative, nevertheless as
we see in this example, it remains close. The estimate then becomes progressively
more accurate as the asymptotic range of the p-extension is entered. In this case
the correct asymptotic rate of convergence is p = X = 0.5445. At p = 8 the computed
value of p is approximately 0.7 with B decreasing.

2.3. Equilibrium tests.

Smallness of error in energy norm is a necessary but not sufficient condition
for ensuring that the overall quality of the finite-element solution is good. It is
possible to produce examples where the estimated error in energy norm is small
(under 1 percent) yet the error in overall equilibrium is large (well over 10 percent).
Although we do not know Z E X we know that Z E x satisfies the equations of
equilibrium and the law of action and reaction. We can, therefore, assess the
quality of the finite-element solution by examining to what degree z F E satisfies
equilibrium and the law of action and reaction. Specifically, we can perform: (I)
overall equilibrium tests; (2) element by element equilibrium tests and (3) action-
reaction tests.
In the overall equilibrium test we 'cut' the structure from its supports and
integrate the tractions, computed from the Z F E , to obtain the reactions. In this
way a free body diagram is produced. The error in equilibrium must be small in
relation to the magnitude of applied forces. For example, we can define:

where r; are the (global) load vector components. The term F is a suitable measure
of the magnitude of the applied load.
In the element-by-element equilibrium test individual elements (or any group
of elements) are separated from the model and tested for equilibrium. Specifically,
we denote the element domain by Re and its boundary by dR,. We compute:

174
where u i j are the stress components computed from the finite element solution; Xi
represents the applied body force components and n, represents the unit normal
to an,. The summation convention is used. We should have 4-
well as the absolute values of each component of both expressions on the right
as

hand side of (16) small in relation to the magnitude of the applied loads. If lqje)l is
small but the absolute values of the integral expressions are not small then there
is a local error but, according to Saint-Venant's principle, the effect of the local
error will not be substantial at some distance from the element in question. If
Iqje)l is large, even after p-extension was performed, then element e, and possibly
its neighbors, should be subdivided. Thus the element by element equilibrium
test provides information about the quality of mesh design. In many cases minor
local refinement (for example, dividing one element into two elements) can have a
highly beneficial effect on the overall quality of approximation when p-extension
is used.
In the action-reaction test we compute the stress result ants along interelement
boundaries and external element boundaries where tractions are applied. Along
interelement boundaries the stress resultants computed for neighboring elements
should have nearly the same absolute value and opposite sense. Along external
boundaries the resultants of the applied tractions and the tractions computed from
the finite element solution should be nearly the same.
Examples of equilibrium tests are presented in [ll].

3. LOCAL QUALITY

Having ascertained that the overall solution quality is acceptable, we are ready
to compute the quantities which are of principal interest, i.e. g i ( c F E )Smallness
. of
error in energy and equilibrium does not guarantee that all functionals computed
from c F E are accurate. It is advisable to perform convergence tests on at least
the more difficult functionals. We demonstrate the procedure by computing the
direction and magnitude of the principal stresses at a point close to the reentrant
corner. We selected the point r = 0.025a; 6 = 30' in the coordinate system shown in
Fig. 1. The stress components computed from the exact solution are:

u2 = 3.67198AaX-l uy = 7.68696AaXd1 rzy = 0.698375AaX-l (17)

Therefore the principal stresses u1 and c2 and the direction of the first principal
stress u1 from the positive x-axis, denoted by d l , are:

In general functionals, other than the strain energy, do not converge mono-
tonically, nevertheless the fact that convergence has occurred should be obvious.
Here u1 and c2 happen to converge monotonically but el does not. We see that the
state of stress is known with sufficient accuracy for engineering purposes at p=4
(335 degrees of freedom, see Table 1). Extension beyond p=4 merely confirms
that convergence has occurred to within the range of precision normally expected
in engineering computations and thereby establishes reliability of the data.
This test problem demonstrates that accurate stress data can be obtained in
the very close proximity of stress singularities. Other examples and additional
discussion of this point are presented in [ 10,121.

175
 Table 2. Principal stresses at r = 0.025a; e = 30".

c1 61
P AaX-l AaX-l (degrees)
-- -__ ___ ___
1 7.096 2.854 92.1
2 7.441 3.047 84.1
3 7.532 3.294 81.6
4 7.731 3.483 80.9
5 7.754 3.525 80.6
6 7.773 3.545 80.5
7 7.786 3.551 80.5
8 7.791 3.553 80.4
__ ___ ___ ---
00 7.804 3.554 80.4

It is possible also to compute various functionals from ciFE using advanced

methods of extraction [13,14,15].For example,we may wishtodetermine the value
of the generalized stress intensity factor A [see eq.'s (lla,b), (12a,b,c)]. Such
procedures based on [ 13,14,15] have been implemented in PROBE [9].

4. CONCLUSIONS AND RECOMMENDATIONS

(1) Extensions are essential for both the estimation and control of error in finite-
element computations.
(2) We are in a much better position today than we were, even just one year
ago, from the point of view of understanding how an advanced finite element
software system should be designed so that (a) the solution is obtained at very
nearly the theoretically optimal efficiency and (b) the user is provided with the
capability to estimate and control the quality of engineering data computed
from the finite-element solution at a small marginal cost. This is because now
we understand the interplay between mesh design and the polynomial degree
of elements.
(3) P-extension, coupled with properly graded meshes, is the most efficient method
for controlling error in finite-element computations.
(4) The proper mesh design is such that points of singularity (and areas where the

d
solution changes rapidly over short distances are isolated by one or more layers
of small elements, with the elements grade in geometric progression toward
the points of singularity. In this way both the global and local behavior of the
solution can be represented without compromising the accuracy of either.
( 5 ) Implementation of advanced extraction methods for the the computation of
certain engineering data, such as stress intensity factors, will further increase
the efficiency and reliability of computations.
(6) The p-version is well suited for implementation on parallel processors because
the data are organized in relatively few, large units. This logical organization
reduces the overhead associated with parallel processing.
(7) The substantial increase of efficiency in finite-element computations through
the use of h-p extension and the availability of parallel and vector processing
technology make it possible and desirable to model plate and shell problems

176
 using hierarchic sequences of plate and shell theories in conjunction with fully
three-dimensional representation. The various plate and shell theories are
nothing more than specializations of the three-dimensional theory of elastic-
ity through restrictions imposed on the variation of the displacement field in
the direction of the normal. Such restrictions generally do not hold near sup-
ports, stiffeners, cut-outs, plate and shell intersections, etc. which are the
areas where cracking and delaminations originiate and therefore of the great-
est concern to analysts and designers. These areas can be properly modeled
by three-dimensional representation only. The use of hierarchic extensions to-
ward higher order plate and shell theories will permit us to assess and control
the quality of approximation in relation to three-dimensional theory.
Although linear theory is properly the first and most generally used approach
to structural modeling, it should be possible to ascertain by a posteriori anal-
ysis whether engineering conclusions drawn from a numerical model would be
different if geometric and material nonlinearities were considered. We can view
linear theory as the simplest of a hierarchic system of theories. Much the same
way as we estimate error by the use of extension processes within the frame-
work of linear theory, we should be able to estimate error by extension within
the hierarchic system of theories. This important area has not received much
attention in the past. Because it bears on the reliability of computed data, and
the engineering conclusions based on them, it deserves serious consideration.
(9) In some areas our ability to compute data is already greater than the material
scientists’ ability to tell us what data should be computed. For example, it
is not fully understood what parameters govern crack initiation. The reason,
at least in part, is that the conventional finite-element method tends to yield
’fuzzy’ data in areas where stresses change substantially over short distances.
Proper use of h-p extension, coupled with advanced extraction methods, per-
mits us to compute any stress field parameter with arbitrary precision. This
removes an uncertainty from the phenomenological characterization of mate-
rial response to various stress fields. Of course, such characterization can be
developed only through joint experimental-analytical investigations.

5. REFERENCES

[l] BabuSka, I., Szab6, B. and Katz, I. N., ”The p-Version of the Finite-element
Method”, SIAM J. Numer. Anal., Vol. 18, pp. 515-545,(1981).
[2]Babugka, I. and Szab6, B., ”On the Rates of Convergence of the Finite-element
Method”, Int. J. Num. Meth. Engng., Vol. 18,pp. 323-341 (1982).
[3]BabuSka, I., Gui, W. and Szab6, B. A., ”Performance of the h, p and h-p Ver-
sions of the Finite-element Method”, Institute for Physical Science and Tech-
nology, Laboratory for Numerical Analysis, Technical Note BN-1027 (Septem-
ber, 1984).
[4]Gui, W., ”The h-p Version of the Finite-element Method for One Dimensional
Problem”, Doctoral Dissertation, University of Maryland, (1985).
[5] Guo, B., ”The h-p Version of the Finite-element Method in Two Dimensions”, -
Mathematical Theory and Computational Experience”, Doctoral Dissertation,
University of Maryland (1985).
[6] Szab6, B. A. and BabuSka, I., ”Stress Approximations by the h- and p-Versions
of the Finite-element Method” Proc., 6th Invitational Symposium on the Uni-

177
 fication of Finite-elements, Finite Differences and Calculus of Variations”,
Edited by H. Kardestuncer, The University of Connecticut, May 1982.
[7] Szab6, B., ”Estimation and Control of Error Based on P-Convergence” Proc.
Int. Conf. on Accuracy Estimates and Adaptive Refinements in Finite-element
Computations (ARFEC), Lisbon, Portugal (1984).
[8] Williams, M. L., ” Stress Singularities Resulting from Various Boundary Con-
ditions in Angular Corners of Plates in Extension”, J. Appl. Mech. ASME,
pp. 526-528 (1952).
[9] Szab6, B. A., ”PROBE: Theoretical Manual”, Noetic Technologies Corp., St.
Louis, Missouri (1985).
[lo] Szab6, B. A., ”Mesh Design for the p-Version of the Finite-element Method”,
Report WU/CCM-85/2, Center for Computational Mechanics, Washington
University (1985). Presented at the Symposium on Recent Advances in Com-
putational Mechanics, Joint ASME,’ ASCE Mechanics Conference, Albuquer-
que, NM (June 24-26, 1985).
[ll] Szab6, B. A., ”Implementation of a Finite-element Software System with h-
and p-Extension Capabilities”, Proc., 8th Invitational UFEM Symposium:
Unification of Finite Element Software Systems, Edited by H. Kardestuncer,
The University of Connecticut, May 1985.
[12] Szab6, B. A., ”Computation of Stress Field Parameters in Areas of Steep
Stress Gradients”, Communications in Applied Numerical Methods, Vol. 2,
pp. 133-137 (1986).
[13] BabuSka, I. and Miller, A., ”The Post-Processing Approach in the Finite-
element Method” - Part 1: Calculation of Displacements, Stresses and Other
Higher Derivatives of the Displacements”, Int. J. Num. Meth. Engng., Vol.
20, pp. 1085-1109 (1984).
[14] BabuSka, I. and Miller, A., ”The Post-Processing Approach in the Finite-
element Method” - Part 2: The Calculation of Stress Intensity Factors”, Int.
J. Num. Meth. Engng., Vol. 20, pp. 1111-1129 (1984).
[15] BabuSka, I. and Miller, A., ”The Post-Processing Approach in the Finite-
element Method” - Part 3: A-Posteriori Error Estimates and Adaptive Mesh
Selection”, Int. J. Num. Meth. Engng., Vol. 20, pp. 2311-2324 (1984).

178
 Yednesday , June 19, 1985
LOCAL/GLOBAL NONLINEAR STRESS ANALYSIS

Questions and Answers following: "Some Issues i n Numerical Simulation of

Nonl i near Structural Response" by H. D. Hibbi tt.

Nelson Bauld, J r . , Clemson University: You used a l i n e a r constraint, a Riks

method, f o r computing the limit point on the buckling load. Have you t r i e d the
Crisfield method?

Hibbitt: We played around with a number of these methods, a n d Crisfield has

done quite a l o t of work on Riks' algorithms. We've t r i e d some of the things
he's mentioned. I think everybody has his own l i t t l e perturbation or variation
on the Riks concept. We're using one t h a t I found o u t a few months ago i s
exactly the' same as Issac Fried published. We're pretty happy with i t , i n most
cases. I t certainly has done as well as the alternates we've t r i e d . The linear
constraint makes i t very easy t o code a n d deal with i n a program l i k e ABAQUS.
I t ' s c r i t i c a l t o use some kind of automatic incrementation scheme along with
i t . By and large, i t seems t o work well. I n the problems where i t f a i l s , I've
certainly had fail ures with some of Cri s f i e l d ' s a1 ternate a1 g o r i thms as we1 1 .

Questions and Answers following: 'Canputerized Structural Mechanics for the

1990's" by B. F. Backman.

W . J . S t r o u d , NASA L a n g l e y Research Center: Bjorn, i n y o u r t a l k , you mentioned

computer program development and the usefulness, most often the lack of useful-
ness, of these computer programs t o industry people. T h a t i s really a problem.
I ' l l give you my personal o p i n i o n about some of these things. I t i s very d i f f i -
c u l t for NASA t o develop computer programs f o r industry. As a r e s u l t , except i n
sane very special cases throughout our history, we really have n o t had t h a t as
our goal. On the other hand, as a p a r t o f our research, we and our grantees and
contractors develop computer programs because t h a t is how we cal cul a t e numbers.
Ue t r y t o make those codes and the supporting technology available t o industry.
B u t i t ' s very hard, as you well know, t o develop computer programs for someone
e l s e ' s use. Would you l i k e t o comment on t h a t ?

179
Backman: I t h i n k NASA has a key role in t h i s area. When I referred t o second
generation programs I meant t h a t there i s a strong need f o r an organized inter-
face between the developers and the users, and I see NASA as having a key role
here. Maybe i t ' s a standing commission--maybe i t ' s an establ ished relation f o r
review--I d o n ' t know. B u t where the needs can be established a n d where the
f i r s t generation software could be used and evaluated by the industry and recom-
mendations coming back t o the developers, I think there exists an i n t e r e s t in
this. The only question i s how you stimulate t h i s t o the point where i t becomes
mutually beneficial.

Questions and Answers following: "Nonlinear Response of a Blade-Stiffened

Graphite-Epoxy Panel with a Discontinuous Stiffener: Work in Progress'' by N .
F . Knight.

K . C . Park, Lockheed Palo A l t o Research Laboratory: Norm, i s t h a t specimen

based on an optimized design f o r w h i c h the panel buckling load i s a b o u t the same
as the skin buckling l o a d ?

Knight: This panel was one t h a t ' s been around for a number o f years. I t was
f i r s t b u i l t as part of one of the early studies in our composite design program
a n d , I believe, i t was designed without the hole--as p a r t of a wing-type
structure near the t i p , so i t i s primarily a strength-type panel. Now w i t h the
hole, the characteristics o f the panel change.

Yung J . Chiang, Uniroyal , Inc.: By using the point-stress failure criterion,

You assume t h a t the failure occurs when the s t r e s s a t a distant do away from
the edge o f the hole reaches ultimate. What i s the s t r e s s t h a t you are
referring t o ?

Knight: We applied j u s t the average inplane s t r e s s resultant. I t was j u s t an

average longitudinal stress-type measure. We were using stress resul tants not
necessarily t o predict failure b u t t o guide us in defining the mesh refinement
needed near the hole in order t o predict the s t r e s s distribution. The elements
t h a t we were using do n o t have a transverse shear formulation, so we c a n ' t
really recover the shear s t r e s s distribution other t h a n from classical plate
theory by taking derivatives of the moments and things.

180
Chiang: Can the energy calculation be better t h a n s t r e s s for the prediction of
the f a i l u r e ?

Knight: W e d i d n ' t calcu a t e the energy so I d o n ' t really know. You can calcu-
l a t e a local energy norm and use t h a t t o predict f a i l u r e , b u t we did n o t make
t h a t calculation. I can t answer t h a t .

Questions and Answers fol lowing: "Computational Perspectives on Postbuckl i n g of

Composite Shells" by G. M. Stanley.

Gerald Goudreau, Lawrence Livermore National Laboratory: T h a t f i f t h buck1 ing

node t h a t you said was nonphysical, did i t vary with the different element types
t h a t you' ve studied?

Stanley: No, n o t really, the eigenvalues moved a r o u n d by a few percent here or

there. There was actually more of a discrepancy in the l i n e a r solution t h a n in
the buckling solutions.

Goudreau: Then t o w h a t do you a t t r i b u t e the nonphysical basis for t h a t mode?

Stanley: I'm n o t saying t h a t t h a t mode i s wrong. I'm saying t h a t t h a t mode

would n o t occur in practice, because any imperfection would prevent i t from
occurring.

J e r r o l d M. Housner, NASA Langley Research Center: Gary, I noticed t h a t i t ' s a

pretty rough problem once you get t o the buckling l o a d . I n the slide showing
experimental r e s u l t s , i t looks l i k e there was a d i s t i n c t softening of the panel
just prior t o i t s f i r s t f a i l ure--col lapse.

Stanley: Before you continue, t h a t distinction was a l i t t l e sharper in t h i s

picture than i t was in practice. T h a t was handdrawn.

Housner: B u t , in any case, the analysis doesn't seem t o pick t h a t up. The
softening i s in the l a s t 10 percent of the buckling load as you're coming u p the
load deflection curve. Could t h a t be--

Stanley: The analysis doesn't pick u p the peak; the analysis does roll over.

181
 Housner: I t does r o l l over f o r a s h o r t way I n o t i c e d , b u t what I ' m d r i v i n g a t
i s t h a t t h e s o f t e n i n g t h a t takes place--could t h a t be what t r i g g e r s t h e f a i l u r e
and i s n o t being d u p l i c a t e d i n t h e a n a l y s i s ?

Stanley: The s l i d e which shows s t r e s s contours i n c o l o r would h e l p us here.

As we move up t h e l o a d - s h o r t e n i n g curve--near, b u t before, t h e peak--the analy-
s i s p r e d i c t s very h i g h ( n e a r u l t i m a t e ) a x i a l s t r e s s e s i n t h e v i c i n i t y o f t h e
hole. So, i t ' s v e r y p o s s i b l e t h a t d e l a m i n a t i o n i s a l r e a d y o c c u r r i n g u p t h e r e .
T h i s s o f t e n i n g i s due t o m a t e r i a l i n s t a b i l i t y . (See f i g u r e , p. 207).* These
a r e t h e compressive a x i a l s t r e s s e s i n the, I t h i n k , f o u r t h l a y e r - - i n which t h e
f i b e r s a r e o r i e n t e d along t h e axis--and t h i s c o l o r s c a l e i s such t h a t t h e maxi-
mum i s w h i t e and t h e minimum i s b l a c k . E v e r y t h i n g i n w h i t e here, though, i s
a c t u a l l y beyond t h e nominal u l t i m a t e s t r e s s f o r t h i s p a r t i c u l a r composite. So
a t t h e b u c k l i n g l o a d we're a l r e a d y a t u l t i m a t e s t r e s s on both sides o f t h e hole.
T h i s shows t h a t as we r o l l around, t h e w h i t e r e g i o n begins t o r o t a t e a l i t t l e
b i t , t h e n i s u n l o a d i n g and drops o f f . When we're back on t h e secondary path, a l l
t h e l o a d i s p i c k e d up here and i t ' s a g a i n way beyond u l t i m a t e . You c a n ' t r e a d
i t f r o m t h e scale, b u t as we go up t h e curve, i t ' s maybe 50 p e r c e n t o r more
h i g h e r t h a n u l t i m a t e . So t h e r e a r e a l o t o f t h i n g s t h a t need t o be explained,
and I t h i n k t h a t we need t o t r y one t h i n g a t a time. I d o n ' t t h i n k we should
jump i n t o a 3-D a n a l y s i s u n t i l we have t r i e d some f a i l u r e c r i t e r i a - - j u s t degrad-
ing t h e m a t e r i a1 p r o p e r t i es.

Housner: T h a t ' s what y o u r n e x t s t e p i s g o i n g t o b e - - p u t t i n g i n f a i l u r e

c r i t e r i a?

Stanley: I t h i n k so. I p l a n t o use t h i s workshop, t h e feedback and o p i n i o n s

expressed here, t o h e l p f o r m u l a t e t h e n e x t phase o f t h e research, as a m a t t e r of
fact. But, t h a t ' s one p o s s i b l e d i r e c t i o n t h a t we m i g h t b e g i n next.

Questions and Answers following: "A Review o f Some Problems i n Global/Local

Stress Analysis'' by R. B. Nelson.

Stroud: I n t h e model t h a t Paolo R o b e r t i generated, what was t h e l o g i c behind

t h e model g e n e r a t i o n and r e f in m e n t ?

Nelson: I n words, t h e approach i s t o say: I'll t a k e a s p e c i f i c g r i d and then

I'll r e f i n e i t one t i n e t o see i f t h e s t r e s s e s change much. And I compare
182
s t r e s s e s across n e i g h b o r i n g elements. I f t h e s t r e s s change i s more than a s e t
t o l e r a n c e ( a n d t h e s t r e s s change was, by t h e way, a r o o t mean square t y p e Of

s t r e s s change), then I w i l l r e f i n e t h e model f u r t h e r i n these p a r t i c u l a r areas.

There a r e a number o f i s s u e s r e g a r d i n g t h e convergence t h a t should be used and
how one s h o u l d go about t h i s . There a r e a number o f a u t h o r s who have w r i t t e n on
t h a t topic. I ' m o n l y b r i n g i n g i t f o r w a r d t o show t h a t i t i s an area which I
t h i n k m e r i t s s e r i o u s research c o n s i d e r a t i o n .

Stroud: I agree w i t h you. I t h i n k t h a t many o f t h e t h i n g s t h a t you a l l u d e d t o

a r e n i c e t o have. Yet, today, i t i s d i f f i c u l t f o r us t o g e t support t o do some
t h i n g s t h a t m i g h t even suggest h a v i n g an e x p e r t system b u i l t i n t o i t because
t h a t ' s considered by many t o be a d i v e r s i o n . When you have a l i m i t e d amount o f
energy and funds t o p u t i n t o something, you c a n ' t go i n t o these d i v e r s i o n s .

Nelson: I t i s t r u e , b u t I t h i n k t h a t t h e development o f an automated technique

f o r node1 r e g e n e r a t i o n , o r adaption, o r refinement, something t h a t t h e engineer
r e a l l y s h o u l d n ' t have t o do, -
i s appropriate. The engineer should n o t have t o
t e n d t h e screens. So many people t e n d t h e screens, and I t h i n k they ought t o be
p u t t i n g t h e i r t i m e t o e x p l o r i n g whether o r n o t we need t o have a l a r g e deforma-
t i o n o p t i o n b e i n g t u r n e d on, o r whether o r n o t we a r e w o r r i e d about t h e r e b e i n g
a problem w i t h t h e m a t e r i a l model used t o model a composite. I s i t a mean
square o f s t r e s s o r a d e l a m i n a t i o n s t r e s s t h a t we should be l o o k i n g a t ? Those
k i n d s o f i s s u e s are e n g i n e e r i n g i s s u e s I t h i n k should be pursued.

Q u e s t i o n s and Answers f o l l o w i n g : "Some Comments on Global/Local A n a l y s i s " by

S. N. A t l u r i .

James C. Robinson, NASA Langley Research Center: I t h i n k when we s t a r t t a l k i n g

about e i t h e r g l o b a l l y o r l o c a l l y e x a c t s o l u t i o n s , as s t r u c t u a l engineers, we
s h o u l d show p r o p e r r e s p e c t f o r a n e g a t i v e f o r c e displacement curve. It looks
v e r y n i c e when i t J s p l o t t e d up i n a r i g i d t e s t machine, b u t when i t i s i n an
a c t u a l s t r u c t u r e t h a t i s surrounded by o t h e r e q u a l l y s e n s i t i v e elements and you
d O n J t do a dynamic s o l u t i o n , you have p r o b a b l y missed f a i l u r e a t t h e u l t i m a t e
load point. To l o o k a t i t from t h e r e on i s a f i n e academic e x e r c i s e , b u t s t r u c -
t u a l l y use1 ess.

Atluri: I agree w i t h you t o t a l l y . These a r e i l l u s t r a t i v e examples t o t e s t the

t h e o r y t h a t i s presented. In these examples , a d i spl acement-control 1ed 1o a d i ng
i s assumed.
183
 K . N. Shivakumar, AS&M, I n c . , Hampton. You mentioned t h a t i n a composite panel,
some t h e o r i e s would a l l o w you t o r e p l a c e a h o l e by a crack. What a r e t h e
t h e o r i e s ? To me, a crack has an e n t i r e l y d i f f e r e n t model from a hole, and you
may be m i s s i n g some p h y s i c s i n t h e problem.

Atluri: I f you c o n s i d e r an i s o t r o p i c p l a t e w i t h a h o l e i n i t , t h e s t r e s s
c o n c e n t r a t i o n i s always t h r e e , i f t h e s i z e o f t h e p l a t e i s r o u g h l y f i v e times
b i g g e r than t h e h o l e s i z e . But i n a composite l a m i n a t e , t h e r e i s a s i z e depend-
ence. Even if you c o n s i d e r a p l a t e f i v e times b i g g e r than t h e h o l e diameter,
t h e s t r e s s c o n c e n t a t i o n s t i l l depends upon t h e s i z e of t h e hole. And i n t h a t
sense, ifyou were t o p l o t t h e f a i l u r e s t r e s s o f t h e laminated p l a t e w i t h t h e
h o l e versus t h e h o l e r a d i u s , i t would vary as ra. And i n f r a c t u r e mechanics,
t h a t v a r i a t i o n i s - - i f you were t o c o n s i d e r r t o be t h e l e n g t h o f t h e crack--
t h e f a i l u r e s t r e s s f o r a cracked p l a t e would vary as r-.5, o r something l i k e
that. There i s no g r e a t t h e o r y behind i t - - j u s t an e m p i r i c a l r e l a t i o n . Mar a t
M I T has taken t h a t , and t h e r e i s a t h e o r y c a l l e d t h e Mar-Lin t h e o r y wherein t h e
f r a c t u r e s t r e s s i n a l a m i n a t e d h o l e v a r i e s as t h e diameter o f t h e h o l e t o some
power alpha. So t h e r e i s some evidence t h a t a h o l e i n a l a m i n a t e behaves l i k e a
crack. The p h y s i c s o f t h a t needs t o be e x p l o r e d much f u r t h e r .

Stanley: I agree w i t h y o u r advice t o i n c o r p o r a t e l o c a l l y e x a c t s o l u t i o n s i n

g l o b a l a n a l y s i s , b u t i n t h i s p a r t i c u l a r case o f t h e composite panel w i t h a hole,
what about t h e subsequent p r o p a g a t i o n o f f a i l u r e i n which t h e h o l e begins t o
distort--it's no l o n g e r a c i r c u l a r h o l e . Do you t h i n k i t ' s a c t u a l l y f e a s i b l e t o
c o n t i n u e t o f o l l o w t h e complete a n a l y s i s through w i t h l o c a l l y e x a c t s o l u t i o n s ?

Atluri: I t h i n k so. A t l e a s t these l o c a l l y e x a c t s o l u t i o n s c o u l d g i v e you a

b e t t e r handle on t h e three-dimensional s t r e s s s t a t e - - 1 i k e t h e t r a n s v e r s e shear
and so f o r t h . And when l a r g e geometric changes t a k e place, those s o l u t i o n s
would n o t be l o c a l l y exact, b u t they would be much b e t t e r t h a n t h e t r a d i t i o n a l
f i n i t e element b a s i s f u n c t i o n s you use.

Stanley: Perhaps they c o u l d be i n c o r p o r t e d as d e v i a t i o n a l b a s i s f u n c t i o n s on

t o p o f t h e normal shape f u n c t i o n s .

184
Questions and Answers following: "On Computational Schemes f o r Global/Local
S t r e s s Analysis" by J . N. Reddy.

Glenn Sahrmann, PRC Kentron: I n your analysis of the 2-D f r e e edge problem,
were the s t r e s s e s becoming s i ngul ar--unbounded?

Reddy: Without f r e e edge cap, yes, the analyses done by many people show t h a t
as you refine the mesh, oZ s t i l l keeps getting larger. So, i t i s unbounded.
B u t we know t h a t uZ must have some f i n i t e value. I d i d n o t address t h a t
problem here. My objective was i s to reduce the s t r e s s concentration by a d d i n g
a cap. We accounted f o r the cap in the f i n i t e element analysis. We also
modeled the 'cap as part of the region.

Questions and Answers f o l l owing: "G1 obal /Local F i n i t e E l ement Analysis of

Localized Stresses i n Prismatic Structures' by S. B. Dong.

Housner: After you compute what you feel i s the position of the interface
between the local and the f a r f i e l d solutions, have you gone back t o examine the
s e n s i t i v i t y of the solution t o the location of t h a t boundary--that interface?

Donz: Yes, i n f a c t , I had a student t h a t took the i n t e r f a c e a t different

regions. What happens i s t h a t when you have self-equil ibrated s t r e s s e s , those
which you subtract o u t from what the Bernoulli-Euler theory or Kirchoff theory
would give you i n terms of a plate analysis i n the f a r f i e l d , the s e l f -
equilibrated s t r e s s e s will decay. I f you take the f i n i t e element region very
close to the localized area, you would obviously need more global functions.
B u t t h a t ' s r e a l l y n o t a severe penalty because when you include 10 global func-
t i o n s , you can probably eliminate one or two layers of f i n i t e elements t h a t
would involve many more degress of freedom. A second advantage of t a k i n g the
f i n i t e element region very close to the localized area i s t h a t you will deter-
mine on a mathematical basis how the self-equilibrated s t e s s e s diffuse as they
go i n t o the i n t e r i o r rather t h a n by l e t t i n g the f i n i t e element solution--which
depends on the model ing--indicate how the sel f-equil i brated s t r e s s e s a r e
diffused. I f you have a complete s e t of global functions, t h e n t h e r e ' s no
d i f f i c u l ty i n going as close as you please t o the local ized region. By succes-
sively taking more and more global functions, you have a basis for evaluating
the e r r o r t h a t J s incurred i n the analysis.

185
 Housner: I guess you c o u l d go a l l t h e way back t o t h e i n c l u s i o n i t s e l f ifyou
r e a l l y wanted t o . Years ago, I dabbled a l i t t l e i n wave propagation and ifYOU
have t h e complete s e t o f waves, i n c l u d i n g those t h a t a t t e n u a t e , you can
a c t u a l l y , t h e o r e t i c a l l y , a t l e a s t , go a l l t h e way back t o t h e i n c l u s i o n .
Although, i n those days y e a r s ago, we d i d n ' t have t h e computers t o a c t u a l l y do
that.

Dong: I t h i n k ifyou go back t o t h e i n c l u s i o n , E l i Sternberg would be very

happy. Then you would c u t o u t a l l t h e f i n i t e elements, and, you would be where
he was b e f o r e a l l o f us messed up h i s conquest o f a l l t h e problems we have.

Q u e s t i o n s and Answers f o l l o w i n g : "Global /Local Methodologies and T h e i r Appl i c a -

t i o n s t o N o n l i n e a r Problems'' by A. K. Noor.

-
Park: What was t h e m o t i v a t i o n f o r choosing a symmetric boundary c o n d i t i o n ? We
have found t h a t a symmetric boundary c o n d i t i o n does n o t y i e l d t h e p h y s i c a l l y
correct buckling solution.

-
Noor: Well ,I t h i n k you have t o q u a l i f y t h e t y p e o f symmetry. I f you say
r e f l e c t i o n symmetry o r m i r r o r symmetry, t h e n I would agree w i t h you. What we
found i n these problems i s t h a t you have i n v e r s i o n symmetry which i s t y p i c a l of
I have looked a t t h e r e s u l t s t h a t were
t h e response o f many a n i s o t r o p i c panels.
presented by Norm K n i g h t and Jim Starnes and we a l s o looked a t our r e s u l t s . The
i n v e r s i o n symmetry was e x h i b i t e d by t h e response o f t h e panel up t o t h e maximum
l o a d g i v e n by t h e experiments.

-
Park: Do you f o r e s e e t h a t one may choose t h i s boundary c o n d i t i o n t o do c o l l a p s e
and a1 so postbuckl ing a n a l y s i s ?

-
Noor: L e t me p u t i t t h i s way. I f you s t a r t f r o m t h e b e g i n n i n g conducting a
n o n l i n e a r a n a l y s i s , t h e i n v e r s i o n symmetry would a1 ways be preserved u n t i l and
u n l e s s you have a branching p o i n t . And you can always d e t e c t t h e branching
p o i n t from y o u r equations. I f you g e t t h a t b r a n c h i n g p o i n t , then t h e r e i s a
I

p o s s i b i l i t y t h a t you m i g h t go on a branch where t h a t symmetry i s l o s t . However,

as you w i l l hear tomorrow, t h e r e a r e ways then t o s y n t h e s i z e t h e s o l u t i o n o f t h e
unsymmetric prob em from t h e symmetric and a n t symmetric s o l u t i o n s , even f o r t h e

186
n o n l i n e a r case. Summarizing what I s a i d , I t h i n k ifyou s t a r t from t h e begin-
n i n g w i t h a n o n l i n e a r s o l u t i o n , t h e i n v e r s i o n symmetry p r o p e r t i e s o f t h a t
n o n l i n e a r s o l u t i o n w i l l n o t change u n l e s s y o u have a branching p o i n t .

-
Park: How f a r have you c a r r i e d along y o u r s o l u t i o n i n y o u r example?

-
Noor: I n t h e u n s t i f f e n e d panel, up t o t h e maximum l o a d t h a t was g i v e n i n t h e
experiments.

-
Park: We happen t o d i s a g r e e w i t h t h a t because j u s t before t h e c o l l a p s e t h e r e i s
a b i f u r c a t i o n t h a t occurs about 10 o r 15 p e r c e n t below t h e c o l l a p s e l o a d , and
t h a t mode happens t o be antisymmetric.

-
Noor: We d i d n o t d e t e c t b i f u r c a t i o n up t o t h e l e v e l o f t h e l o a d i n g t h a t we had,
b u t i f you have b i f u r c a t i o n , you can l o o s e t h e symmetry. But even then, as I
s a i d , you can s y n t h e s i z e t h e unsymmetric s o l u t i o n from symmetric and
antisymmetric components. And y o u ' l l hear about t h i s tomorrow.

Stanley: I t h i n k i t would be good f o r us t o compare q u a n t i t a t i v e r e s u l t s , b u t

j u s t q u a l i t a t i v e l y , a r e t h e p a t h d e r i v a t i v e s more accurate than t h e modes i n
t h i s p a r t i c u l a r problem o r has t h i s been y o u r general experience?

-
Noor: Ne d i d n o t t r y t h e b u c k l i n g modes i n t h i s p a r t i c u l a r problem b u t I t r i e d
them i n several o t h e r p r o b l ems--pl a t e p r o b l ems , shear-1 oaded p l a t e s , a x i a l l y -
1oaded p l ates--and w i t h o u t any exception, you g e t c o n s i d e r a b l y h i g h e r accuracy
w i t h t h e p a t h d e r i v a t i v e s than w i t h equal number o f b u c k l i n g modes. Not t o
m e n t i o n a l s o t h a t t h e p a t h d e r i v a t i v e s a r e c o n s i d e r a b l y l e s s expensive
t o generate than the b u c k l i n g moies because i f you generate them a t zero
l o a d i n g o r z e r o end displacement, you need t o decompose o n l y t h e l i n e a r g l o b a l
s t i f f n e s s matrix.

Questions and Answers following: 'Application o f t h e p-version o f the F i n i t e

Element Method to Global/Local Problems' by 8. A. Szabo.

C a r l o s A . F e l i p p a , Lockheed P a l o A l t o Research L a b o r a t o r y : I think your theory

depends e s s e n t i a l l y on t h e assumption t h a t t h e elements a r e conforming.

-
Szabo: Yes.
187
Fel i p p a : Now, most production computer programs do not use conforming elements.
And i f you use nonconforming elements, a funny thing can happen. You can refine
the mesh and get a worse solution because the aspect r a t i o o f the elements may
deteriorate. How would you extend the theory t o programs t h a t use nonconforming
el ements?

Szabo: To me, a very important consideration i s t h a t there i s a n existing proof

t h a t the solution will converge in a certain norm. There are several methods,
such as methods involving nonconforming elements, which do n o t meet with the
Babuska-Brezzi criterion which guarantees the desired convergence properties.
This criterion i s d i f f i c u l t f o r engineers t o understand because i t i s based on
mathematical considerations. There are several methods in existence (methods
t h a t do n o t meet the Babuska-Brezzi c r i t e r i o n ) f o r which one could s e l e c t a s e t
of input d a t a a n d the methods would work well. I n other words, one could design
the mesh, select the i n p u t d a t a , e t c . , and the method would work. And somebody
else c o u l d select a d i f f e r e n t s e t of i n p u t d a t a , a n d because the rnethod does n o t
s a t i s f y the Babuska-Brezzi condition , i t may "blow u p " . Now i t frequently
happens t h a t a new method i s shown t o work on smooth problems, typically text-
book problems. And the researcher says, "Look, my method works because i t can
solve some specific problem l i k e the simply supported square plate." Any
numerical method should be able t o do t h i s . The real d i f f i c u l t y i s when these
methods are applied t o nonsmooth problems which are the important problems in
engineering. Then they may n o t work; they may give a completely nonsensical
r e s u l t . For t h i s reason, I choose t o stay away from those methods. Instead, I
use a method t h a t i s based on the principle of virtual work because I know t h i s
method s a t i s f i e s the Babuska-Brezzi condition. Nonconforming methods, which do
n o t s a t i s f y those conditions, may give you unpleasant surprises. For example,
you may be doing more work and getting worse results. On the other h a n d , i f you
use a method t h a t s a t i s f i e s the Babuska-Brezzi condition and i f you do more
work, ( f o r example, use f i n e r meshes) the analysis will give you better results.

Stroud: Because of Carlos' comments and some comments t h a t you j u s t made, do

you recommend t h a t conforming elements be used exclusively?

Szabo: Yes, until proofs are developed t h a t the other methods are robust--in
the sense t h a t no matter what (admissible) d a t a you select, the method should
work. And t h i s proof i s n o t available f o r the nonconforming elements, as f a r

188
as I know. Now i t c o u l d be t h a t somebody has j u s t worked i t o u t yesterday ( o r
something l i k e t h a t ) , b u t I have t a l k e d w i t h mathematicians about t h i s and they
have t h e view t h a t i t i s dangerous t o use methods which a r e n o t meeting t h a t
criterion.

Reddy: I t h i n k t h a t t h e LBB c o n d i t i o n i s n o t r e a l l y a necessary c o n d i t i o n , i t J s

a sufficient condition. And I t h i n k t h a t t h e r e a r e some elements which a r e
proven t o work even i f they do n o t s a t i s f y t h e LBB c o n d i t i o n . I t h i n k t h a t we
should n o t j u s t throw o u t e v e r y t h i n g t h a t w i l l n o t pass t h e LBB c o n d i t o n
because i t i s o n l y a s u f f i c i e n t c o n d i t i o n . Not necessary.

Szabo: I J v e heard t h a t argument. I J m n o t a mathematician; I ' v e o n l y p i c k e d up

i n f o m a t i o n on t h i s t o p i c by osmosis. But I b e l i e v e t h a t i t i s a necessary
c o n d i t i o n . B a s i c a l l y , t h e argument i s t h e f o l l o w i n g . Ify o u r method does n o t
meet t h e Babuska-Brezzi c o n d i t i o n , t h e n one can always s e l e c t i n p u t data (maybe
n o t t h e k i n d o f d a t a t h a t you a r e i n t e r e s t e d i n ) f o r which t h e method w i l l n o t
work, and f o r which you were n o t a b l e t o say ahead o f time, "Please do n o t make
my i n p u t d a t a l i k e t h i s o r t h a t , " because i t i s very d i f f i c u l t t o p r e d i c t which
s e t o f i n p u t d a t a w i l l cause t h e method t o f a i l . But please d o n J t p r e s s me on
i s s u e s t h a t r e l a t e t o f i n e r mathematical p o i n t s ; I ' m n o t a mathematician.

Reddy: I ' m o n l y t r y i n g t o l e t t h e o t h e r people h e r e know t h a t i f they check t h e

LBB c o n d i t i o n and i f i t f a i l s then t h e y should n o t g i v e up on t h e method. The
same t h ng about Lax-Milgram theorem f o r e x i s t e n c e . It i s a sufficiency
t h e o rem b u t n o t necessary. I t can be shown t h a t a problem has a s o l u t i o n b u t
may n o t meet t h e same c o n d i t i o n s a s s t a t e d i n t h e Lax-Milgram theorem o r Babuska
general zed theorem.

Szabo: I j u s t would l i k e t o i n v e r t y o u r question. I f you know you have a good

method , f o r example the method based on t h e p r i n c i p l e of v i r t u a l work, why would
YOU w i s h t o s e l e c t a d i f f e r e n t method which may g i v e you unpleasant s u r p r i s e s ?
I n o t h e r words, s i n c e we do have good methods t h a t we know a r e working, what
WOUl d be t h e reason f o r u s i n g methods--nonconfomi ng methods, f o r exampl e-- t h a t
have n o t g i v e n s a t i s f a c t o r y s o l u t i o n s a t v a r i o u s times?

Reddy: Well , very r a r e l y do we use methods o t h e r than those based on v i r t u a l

work p r i n c i p l e . F o r example, mixed-type f i n i t e elements a r e used i n c e r t a i n
problems where we t h i n k s t r e s s i s an i m p o r t a n t q u a n t i t y and t h a t should be known
189
on the boundary of the element, like hybrid methods and mixed methods. Those
are the cases where you need t o use mixed methods which are n o t based on the
virtual work principles--I should say traditional virtual work principles
because the mixed principles are also some kind of v i r t u a l work principles.

Szabo: I'm n o t against any particular method--or f o r it--all I am saying i s

t h a t I would like t o see a set of objective performance measures, convergence
characteristics in certain norms, proven a n d demonstrated, i n order for me t o be
convinced I should use t h a t method. Until t h a t i s available, I d o n ' t w a n t t o
use i t . And I have n o t seen really detailed performance studies on the methods
t h a t you have mentioned. I t could be t h a t they exist. All I'm saying i s t h a t I
haven't seen them a n d I ' m quite satisfied w i t h the performance of virtual work
fonnul ations.

Reddy: I think t h a t history shows i t works the other way a r o u n d . We, as

engineers, design methods which we think indeed do work. Then mathematicians,
or people interested i n mathematical aspects, explore the question o f whether i t
i s really a r i g h t method.

Szabo: I agree. I f we.waited f o r the mathematicians, I think we would s t i l l be

living i n caves. B u t , on the other h a n d , a large p a r t o f this meeting i s a b o u t
concerns of the performance of the various methods, and that's w h a t I wanted t o
address. I would love t o see similar performance studies presented for the
other methods as well. I haven't so f a r seen those.

Panel Discus s i on

Bjorn F. Backman, Boeing Military Airplane Company: I would l i k e t o s t a r t by

making a controversial statement. I think any p r o d u c t of engineering develop-
ment in the methods f i e l d i s an engineering tool. And I would 1 ike t o make the
distinction here, i n i t i a l l y , between s c i e n t i f i c development a n d engineering
development. I t h i n k science i s involved i n the pursuit of the t r u t h , while
engineering i s involved in the rational process o f making decisions i n an
environment of limited knowledge. To t h a t extent, I would have liked t o see one
of the two things. I would have liked the end product of computational
structual mechanics t o either be a tool t h a t directly applied t o the design
process or a tool t h a t comes w i t h a recipe t h a t says t h i s i s how you use i t in
order t o prepare the s e t of d a t a t o be used for the design process. That's from
190
one side. On t h e o t h e r s i d e I see i f we want t o - - i n an academic environment--to
pursue t h e s c i e n t i f i c s i d e o f e n g i n e e r i n g t o t h e e x c l u s i o n o f design
c o n s i d e r a t i o n s , t h e n I t h i n k NASA has an even more i m p o r t a n t r o l e i n making t h e
t r a n s l a t i o n i n t o t h e t o o l s t h a t can be a p p l i e d t o t h e design process. Thank
you.

W. Jefferson Stroud, NASA Langley Research Center: Yes, t h a t was a c o n t r o v e r -

s i a l statement. I want t o make sure I understand what you said, B j o r n . YOU

f e e l t h a t t h e CSM a c t i v i t y should produce products. I w i l l even go f u r t h e r and

say s o f t w a r e products.

Backman: Yes.

Stroud: Would you say t h a t merely d e v e l o p i n g t h e know-how t o produce t h e soft-

ware would be a s h o r t f a l l ?

Backman: We1 1 , i t reminds me o f t h e d i f f e r e n c e between s o l v i n g problerns i n

p r i n c i p l e and s o l v i n g problems. I f you l o o k a t any s o l u t i o n technique t h a t ' s
going t o r e q u i r e you 3 months t o come up w i t h a number. I f you want t o a p p l y
t h a t t e c h n i q u e t o t h e design process where t h e q u e s t i o n i s n o t what i s t h e
response, b u t t h e q u e s t i o n i s , what i s a good design--(For instance, f o r t h i s
c u r v e d panel t h a t we have been studying, where i s t h e optimum l o c a t i o n o f t h e
hole, o r what i s be b e s t pad-up around t h e hole, o r what i s t h e s e n s i t i v i t y o f
t h e design t o r e a l i s t i c boundary c o n d i t i o n s ? ) - - t h e n you see 3 months very
q u i c k l y t r a n s l a t e s i n t o a t l e a s t a hundred times t h a t l o n g i n o r d e r t o s o l v e t h e
d e s i g n problem. So I'm i m p l y i n g t h a t ify o u d o n ' t ask y o u r s e l f t h e q u e s t i o n ,
what am I g o i n g t o use t h i s technique f o r , o r y o u say t h i s i s t h e way you can
use i t i n t h e design process, you have o n l y s o l v e d t h e p r i n c i p l e n o t t h e
p r o b l em.

Gary M. S t a n l e y , Lockheed P a l o A l t o Research L a b o r a t o r y : I'd just like to

comment. I t h i n k t h a t what he's s a y i n g i s t h a t t h e implementational aspect Of

methods development i s n o n t r i v i a l and t h a t we s h o u l d be p u r s u i n g methods

imp1 ementation i s s u e s simultaneously w i t h methods development--rather than
w a i t i n g till l a t e r .

D a v i d H e r t i n g , MacNeal-Schwendler C o r p o r a t i o n : Wedr e s o r t o f coming i n from

l e f t f i e l d here. L e t me c u t i n , because U n i v e r s i t i e s and NASA have been

19 1
dancing t h i s nice l i t t l e dance, and we are involved in t h i s in several aspects.
We ( a t MSC) are involved in t h i s research ourselves, and we are also involved in
developing similar code. We are now o u t in the business o f paying f o r code
development ourselves, in e f f e c t , also doing w h a t NASAJs do n g . So t h i s i s a l l
very interesting t o us. However, I J d probably l i k e t o make a couple of
comments.

One thing I J m glad t o see i s NASA i s getting o u t of the role of trying t o

develop massive computer codes. I think t h a t ' s a good trend. I f you go into
the research mode i t J s going t o be much more valuable t o us, and much less
duplication. Also, I'm glad t o see t h a t you have practical applications. I'm
glad t o see people looking a t the r e a l i s t i c problems a n d developing techniques
t o solve them because we are also looking a t the same problems, particularly
t i r e s a n d cracks. W
e are also in the development mode.

However, I'm disturbed a b o u t a few things. I've kept quiet here during the
m o r n i n g , and maybe I should bring u p a couple o f points t h a t should stimulate
some discussion. One of the things i s the lack of knowledge a b o u t the
commercial codes. I speak f o r Swanson, I speak f o r Hibbit. A l o t of the
complaints t h a t were coming u p a b o u t the codes are n o t true. A l o t of these
capabil i t i e s e x i s t . The capability t o do the local ized nonlinear analysis i s
implicitly in NASTRAN, or MSC NASTRAN anyway. I'm sure ABAQUS has similar
capabil i t i e s .

This brings u p the previous point o f the ivory tower aspect. I t appears t h a t
you are in an ivory tower i f you're n o t aware of whatJs in the codes. I think
i t should be a p a r t o f the academic world t o look a t commercial codes. W e
certainly l o o k a t the academic world as f a r as looking a t w h a t the research i s
and what the current applications a n d h o t methods are. Let's do i t a l i t t l e
more equally.

A couple of other comments: Some o f these research projects are going t o

develop new methods, new codes. What's going t o happen t o these a f t e r they' re
done? There seems t o be a l o t of development a t NASA t h a t j u s t s o r t of goes
away and hides. We can't seem t o find i t . Sometimes we find i t a t COSMIC and
COSMIC says, "NO, that's our property," and we can't use i t . I f you're going t o
devel op a1 1 these nice el ement routi nes--sol vers, nonl inear methods--how do we

192
get those o u t t o the public? WeJ d certainly 1 ike t o use t h e m . There may be
some restrictions; we'd be glad t o work around them.

S t r o u d : You said a couple of things t h a t I d i d n ' t quite understand. A l t h o u g h

COSMIC charges for the software, I t h o u g h t you could get any software t h a t you
wanted.

Herting: Well , the question occurs because we'd 1 ike t o p u t i t i n NASTRAN and
distribute i t o u t t o everybody in the world.

S t r o u d : Worldwide distribution o f COSfIIC software i s different. I f software

i s developed w i t h government funds, a n d i f i t ' s considered t o the advantage of
the U n i t e d States t o have the software available o n l y t o United States compa-
nies, then there i s a restriction. B u t as f a r as COSMIC restricting software
from anybody in the United States, I don't believe t h a t ' s the case.

Herting: We've had some troubles j u s t f o r the United States. I think the prob-
lem i s , do we go o u t a n d require every customer t o pay f o r a COSMIC code t h a t i s
incorporated in MSC/NASTRAN? Take a program like CONMIEJ t h a t takes $200 or $300
t o purchase from COSMIC, our legal s t a f f t e l l s us t h a t we have to go collect
t h a t $200 or $300 from every delivery we make of CONMIN a n d ship i t off t o
COSMIC. And i t ' s really questions--details--like t h i s t h a t we haven't been able
t o work out.

S t r o u d : I c a n ' t address t h a t question. Your other question was something t h a t

I was r e a l l y c o n c e r n e d a b o u t when w e were c o n s i d e r i n g h a v i n g s o f t w a r e d e v e l o p e r s
a t t h i s workshop.Would the software developers say, "We can already do a l l
these things." I think t h a t a sharp person w i t h a good existing code can do a
l o t of things. I would not question t h a t one b i t . A sharp person with a good
existing commercial code could do a l l of the local/global stuff t h a t we've done.
B u t I think you agree t h a t i t would n o t be easy or routine. We want t o begin
now to look ahead a t w h a t we o u g h t t o be able to do 5 t o 10 years from now.

H . David Hibbitt, Hibbitt, Karlsson, and Sorenson, Inc.: I ' d l i k e t o address

the comment t h a t the academics should l o o k a t commercial codes. I d i d n ' t come
here t o be looked a t , I came here t o look. We commercial code vendors can learn
a l o t from academic work. O u r j o b i s to t r y and package i t t o make i t sensible
for the guys o u t there who have real problems. I don't really think i t ' s
193
t e r r i b l y important t h a t the academics look a t us. I came here, i n f a c t , t o look
a t an issue t h a t i s becoming important a n d t h a t I haven't heard discussed very
much except in very general terms. And t h i s i s the issue of parallel machines.
Langley, I think, i s the place where you have the f i n i t e element machine. O n
the other extreme, people a t Cray t e l l me t h a t the way t o do parallel processing
i s t o get the d a t a down i n t o a microlevel and do a l o t of local parallelism. 1
hope our code i s g o i n g t o have a long l i f e . So we really have t o l o o k a t these
parallel architectures and we have t o decide w h a t t o do. Now I haven't heard
anybody here discussing the f i n i t e element machine; I haven't heard anybody
addressing detailed issues of writing software on para1 1 el machi nes. I s there
somebody here who can help me with t h a t , a n d where i s he, and could he address
the question?

S t r o u d : You are right, Dave. W e are doing work in parallel processing. And
there i s no paper here today t h a t i s oriented specifically towards parallel
processing. Tomorrow, Joe Padovan will t a l k a b o u t i t . He's into t i r e model-
l i n g . We are h o p i n g t h a t w i t h i n the n e x t y e a r o r so we w i l l b e a b l e t o have a
conference t h a t will focus on parallel processing in structural analysis. We're
t r y i n g t o learn. We're buying a new multiple processer computer t h a t will
replace the f i n i t e element machine for parallel processing research. I t ' s
called a FLEX/32 Multicomputer. Each one o f the 20 processers i s a b o u t l i k e a
VAX 11/750. And I hope t h a t we will be able t o address some questions t h a t YOU
raised. Why d o n ' t you throw the question o f parallel processing t o the panel
and audience right now.

Hibbitt: I t h i n k t h a t the point t h a t I was getting t o , J e f f , i f I could j u s t

continue for a minute i s i t seems t h a t the philosophy you've take.n here a t
Langley i s t o go into parallelism a t the macro level , putting one f i n i t e element
on one processor, while i f you talk t o people, say a t Cray Research, the
impression I get i s t h a t they want t o do i t a l l on a micro level. There are
some very successful f i n i t e element applications t h a t have taken advantage of
t h i s approach. I t h i n k Jerry Goudreau i s here, and I think he knows a l o t a b o u t
t h i s . And I was hoping t h a t he would give us some good advice here. I s there a
right a n d a wrong way, or are b o t h ways successful?

S t r o u d : Let me comment on what our group has been doing in parallel processing.
F i r s t of a l l , the Finite Element Machine started o u t , in the mid-to-late 1 9 7 0 ' s ,
with the concept of a microprocessor for each f i n i t e element. By the beginning
i.9L
of t h e ~ O ' S ,we decided t h a t one processor f o r each f i n i t e element was n o t t h e
way t o go. We j u s t g o t 16 processors o p e r a t i o n a l on t h e F i n i t e Element Machine
about a n o n t h ago. Granted i t was n o t a h i g h p r i o r i t y research item, b u t i t was
c l e a r t h a t we here a t Langley c o u l d n o t develop a computer having several
hundred processors. So we and several u n i v e r s i t y grantees have been l o o k i n g a t
a l g o r i t h m s t h a t c o u l d e x p l o i t m u l t i p l e processors. F o r example, we a r e l o o k i n g
a t 1 i n e a r simultaneous e q u a t i o n s o l vers--both i t e r a t i v e and d i r e c t s o l u t i o n
techniques. Another example i s eigenvalue s o l u t i o n techniques i n which you have
eigenvalue s h i f t s . You c o u l d assign a value o f t h e s h i f t t o each o f t h e
processors and l e t each processor g e t t h e eigenvalues near i t s s h i f t parameter.
you c o u l d use some l o g i c t o i n s u r e t h a t y o u ' v e g o t t e n a l l t h e eigenvalues. A
t h i r d example i s s u b s t r u c t u r i n g , i n which you d i v i d e t h e s t r u c t u r e i n t o pieces,
a s s i g n those p i e c e s t o d i f f e r e n t processors, l e t each processor c a l c u l a t e t h e
response o f t h e s u b s t r u c t u r e assigned t o i t , and then work t h e i n t e r f a c e
problem. I n each case, computational speedups can occur i f several processors
a r e w o r k i n g on t h e i r assigned t a s k s simultaneously. But, t h e approach t h a t
we've t o t a l l y abandoned i s t h e i d e a of a microprocessor p e r f i n i t e element.

G e r a l d Goudreau, Lawrence Livermore N a t i o n a l L a b o r a t o r y : L e t me j u s t t a k e a

moment, n o t t o preempt t h e panel, b u t t o respond t o Dave a l i t t l e b i t about our
experience i n a Cray environment w i t h p r i m a r i l y t h e p i p e l i n e concept o f t h e
Cray. We a l s o have i m p l i c i t codes t h a t concern themselves w i t h o p t i m i z a t i o n Of
p r o f i l e e q u a t i o n s o l v e r s and eigen packages, b u t a l a r g e p o r t i o n o f our f i n i t e
element a p p l i c a t i o n s i s i n dynamic impact u s i n g e x p l i c i t f i n i t e element hydra
codes which d o n ' t have any m a t r i c e s t o deal w i t h whatsoever. And so w i t h o u t
l i n e a r a l g e b r a t o e x p l o i t , what a r e you g o i n g t o do t o t a k e advantage of t h e
Cray environment? And t h e t h i n g t h a t we h i t on, and t h i s i s John H a l l q u i s t , and
p r i m a r i l y i n t h e DYMA code a p p l i c a t i o n s , i s t h a t a b i g chunk o f t h e e f f o r t , i n
f a c t 95 p e r c e n t o f t h e e f f o r t , i s on what y o u would c a l l t h e r i g h t hand s i d e
e v a l u a t i o n , you know, f o f x and t, which i s r e a l l y f of the stress state-
I t ' s t h e element s t r e s s divergence i n s o l i d mechanics, and t h e r e a r e comparable
t h i n g s (B-transpose sigma) i n any of t h e s t r u c t u a l elements as w e l l . And r a t h e r
t h a n t r y t o v e c t o r i z e t h e element, meaning t h a t t h e l o o p o v e r whatever degrees
o f freedom a r e w i t h i n an element, we used element chunking. I n element
chunking, f i r s t you s c a l e r i z e every aspect o f t n e a l g o r i t h m w i t h i n t h e element
and then, i n t h e innermost do-loop, you t i g h t l y code a l o o p over a l a r g e number
o f elements. Idow some o f t h e v e c t o r machines r e q u i r e very l a r g e elements, and
i f you t u r n a l l t h e s c a l e r v a r i a b l e s t h a t a r e i n t h e element l o g i c i n t o v e c t o r s
195
Over a s e t of elements, you may have a horrendous amount o f storage r e q u i r e d t o
g e t l o n g enough vectors. But w i t h b o t h 64 and 128 v e c t o r l e n g t h s and w i t h t h e
memory t h a t t h e Crays have provided, we've been a b l e t o chunk a t t h a t l e v e l very
very e f f i c i e n t l y . And I t h i n k the B-transpose sigma p a r t that--and a l s o t h e
s t r a i n c o m p u t a t i o n - - t h e y ' r e j u s t about a t t h e l i m i t o f what we t h i n k we can g e t
o u t o f i t - - t h e Cray environment t h a t we l i v e i n . T h i s does p r e s e n t a problem
w i t h c o n s t i t u t i v e models, o r equations o f s t a t e because every t i m e a new one
comes a l o n g i t immediately slows down t h e code by a f a c t o r o f 2 t o 10. It's
o n l y a f t e r i t ' s proven i t s worth t h a t we then g e t i n and v e c t o r i z e o r o p t i m i z e
t h a t element
Now l e t me say a word about p a r a l l e l machines which i s s o r t o f concatenat-
i n g t h e p i p e i n e t o do more. The Cray XMP and f u t u r e Cray machines a r e promis-
i n g us more, and we have John H a l l q u i s t ' s p r o t e g e here, Dave Benson, who has
been i t c h i n g t o m u l t i t a s k t h e XMP. A problem we have a t Livermore i s t h a t when
t h e XMP i s n t i n t h e hands o f one research group, b u t i n t h e hands o f a l a b o r a -
t o r y w i t h 500 t o 1000 e q u a l l y i m p o r t a n t researchers, what you f i n d i s t h a t t h e
m u l t i t a s k i n g p r i o r i t y i s n o t f o r i n d i v i d u a l code, b u t f o r t h e o p e r a t i n g system
as a whole. And so a t Livermore the o p e r a t i n g system i s t r y i n g t o m u l t i t a s k a l l
these u s e r s t r y i n g t o simultaneously do t h e i r work. And so, a t t h e moment, a l l
we have access t o i s a s i n g l e processor and we h a v e n ' t g o t t e n advantage o f t h a t .
Now I was up a t White Oak, t h e Naval Weapons Center, y e s t e r d a y and found o u t
t h a t ABAQUS i s b e i n g e x p l o r e d by C S P I , a small a r r a y p r o c e s s i n g company up i n

-- and w e ' r e g e t t i n g one t o o I guess, not i n my group, b u t a n o t h e r

Massachusetts,
group a t t h e Laboratory. So I would be i n t e r e s t e d i n t a l k i n g f u r t h e r w i t h YOU

about--maybe a t a s m a l l e r l e v e l than a t a C r a y - l o o k i n g a t some o f these a r r a y

p r o c e s s i n g i d e a s and how we c o u l d proceed and then maybe nfluence the bigger
machines down t h e road.

I was a l s o hop ng t o hear from t h e

Moktar Salama, J e t P r o p u l s i o n Laboratory:
speakers more about how t h e i r problems f i t i n t o t h e broad o b j e c t i v e o f t h e CSM
a c t i v i t y , which as s t a t e d here, i s t o develop advanced s t r u c t u r a l a n a l y s i s tech-
n o l o g y t h a t w i l l e x p l o i t modern and emerging computers. I was hoping t h a t t h e
speakers would g i v e us an i d e a o f t h e amount o f computing t h a t t h e i r problems
take. Because i f t h e problems we're t a l k i n g about h e r e r e q u i r e l i t t l e e f f o r t ,
l i t t l e computing, t h e n t o d a y ' s computers a r e adequate. Unless t h e r e i s a g r e a t
deal o f computational i n t e n s i t y , a l a r g e amount o f data, o r u n l e s s t h e k i n d of
problem t h a t ' s b e i n g solved r e q u i r e s a r e a l t i m e s o l u t i o n t h e n these problems

196
d o n ' t f i t into the category of problems t h a t need modern computing beyond w h a t
we have today.

Barna A . Szabo, Washington University: I had occasion t o speak t o many aero-

space engineers in different companies. Their general opinion i s t h a t as the
1990's come along they foresee the use of 3-dimensional f i n i t e element tech-
nology in airframe design. The wing skins are getting thicker and more compli-
cated, and new materials are being used. There i s no d o u b t i n their minds t h a t
the aerospace industry will be directed toward very heavy computing, very large
computational loads, very complicated geometries, many el ements, new material S ,
singularities, damage tolerance requirements, a n d the l i k e . As we look t o the
1 9 9 0 ' ~I ~t h i n k i t i s clear t h a t parallel processing i s g o i n g t o be increasingly
important. And I believe they are very interested in the technoloqy I was
addressi ng--where we can o b t a i n high accuracy w i t h much fewer degrees of freedom
by properly designing the f i n i t e element spaces. There i s a real need o u t there
for us t o look a t how the airframes f o r the 1990's will be designed.
On the other hand, I cannot really speak a b o u t parallel systems because
such systems are not accessible t o us a t the university. We're very happy t o
f i n d time on the VAX 11/750. Even there I have t o twist the dean's arm. I
would certainly be very interested i n exploring different computer systems t o
find o u t exactly how we should design algorithms for those systems. Unless we
have access t o parallel systems t o experiment w i t h them--I d o n ' t t h i n k we can
really answer your questions a n d Dr. Hibbitt's questions, which are very impor-
t a n t questions. I t ' s n o t t h a t we're n o t interested in parallel processing.
I t ' s j u s t t h a t we f i r s t have t o get access t o t h e m .

Sal ama: We1 1 , I 'm real ly n o t saying t h a t the problems you discussed do n o t
require heavy computing. I ' m n o t saying t h a t . I ' m saying t h a t t h i s fact was n o t
brought out, that's all.

Stroud: The CSM g r o u p i s g o i n g t o be h a v i n g grants w i t h universities. We are

g o i n g t o make o u r Flex mu1 t i processor computer a v a i l ab1 e t o the universities
t h a t we are working with. We hope t h a t 1 2 t o 18 months from now we o u g h t t o be
able t o have a handle on several structural analysis methods t h a t look good on
mu1 t i p l e processor computers.

Richard B. Nelson, U. C . L . A . : I want t o second the comment t h a t Professor

Szabo made, and t h a t i s t h a t UCLA recently g o t some new computing equipment
197
which d i d involve an a r r a y processor. And now we are a l l looking around, saying
t o ourselves, now t h a t we have i t w h a t are we going t o do. Ploney i s the
mother's m i l k of research and for us t o go o u t a n d explore a r r a y processing i n a
big way we need t o have support for a relatively intensive program. With P h . 0 .
students, one of the things we do i s invest a great deal of time in h a v i n g a
student generally write, almost from scratch, a f i n i t e element code, or pick U P
SAP 4, which i s by now quite dog-eared, or possibly NONSAP, one of the codes
which we more or less have inherited from Berkeley more t h a n 10 years ago. I ' m
speaking as a researcher in need of a k i n d of t e s t bed t h a t we across the
country could pretty much rely on and plug o u r research e f f o r t s into so t h a t
people across the country could make use of our research product on a general
basis. I d o n ' t consider t h a t t o be particularly competitive with optimized
codes t h a t are specifical ly designed for certain industrial appl ications. B u t
rather t h a t i t gives people b o t h in industry a n d i n the academic cornunity the
opportunity t o communicate across technical l i n e s , and l e t ' s face i t , the
language o f comunication today i s t h r o u g h computers a n d software. And we have
t o develop some k i n d o f a liaison t h r o u g h channels o f software and t h a t ' s g o i n g
t o require significant support t o develop.

E. Thomas Moyer, J r . , The George Washington University: I ' d l i k e t o make two

comments, one on the hardware aspect of parallel processing. As the technology
gets cheap enough, rather than making time available a t one central location,
i t ' s becoming feasible for universities t o buy parallel processing computers.
F o r instance, with the new INTEL machines, you can pick u p a parallel system
with 32 processors in i t for under $100,000. As the prices f a l l , universities
should buy the computers either t h r o u g h a NASA program or an NSF program i n a
wide way so t h a t a l o t of researchers will have access t o them. A s f a r as the
comment t h a t was made with regard t o university professors not being aware Of
f i n i t e element codes capabilities and t h a t type of thing, I think t h a t the
universities should have a l o t stronger access t o the codes a t either the code
level or, a t l e a s t , a t the executable level a t a price, i f n o t nothing, then
extremely negligible. And t h a t I think would be productive b o t h t o the
commercial endeavor and t o the research environment as a whole. And I t h i n k
t h a t t h a t o u g h t t o be explored in more depth.

S t r o u d : My bet i s t h a t commercial software companies do provide their codes t o

universities a t quite reduced rates. (Agreement from audience. 1

198
Backman: I wanted t o address t h e q u e s t i o n about t h e needs o f t h e aerospace
i n d u s t r y . I t h i n k i t ' s q u i t e c l e a r t h a t f o r t h e 199O!s we have seen a number of
problems, many o f which perhaps have n o t been v e r y dominant before, t h a t r e q u i r e
b o t h hardware and s o f t w a r e improvements o f orders o f magnitude. You l o o k a t t h e
o p t i m i z a t i o n side, you l o o k a t t h e emerging technology i n t h e damage t o l e r a n c e
f i e 1 ds t h a t - - i n composites--requi r e del ami n a t i o n growth under compression , and
p o s t b u c k l e s t a t e s , and I see s u b s t a n t i a l development requirements. There i s
very l i t t l e doubt t h a t t h e r e a r e a number o f a p p l i c a t i o n s producing problem
s i z e s such t h a t p r e s e n t software and hardware i s n o t adequate. We s t i l l c a n ' t
g e t i n t o t h e l a r g e o p t i m i z a t i o n problems t h a t we need t o be doing. We r e a l l y
need t o g a i n a d d i t i o n a l f i r s t - h a n d hardware experience i n o r d e r t o t r u l y
understand and develop t h e requirements.

Stroud: We've touched on m u l t i p l e processor computers. What about t h e i d e a

t h a t if we a r e going t o have good s t r u c t u r a l a n a l y s i s i n t h e f u t u r e , we need t o
have b e t t e r s o f t w a r e a r c h i t e c t u r e i n o r d e r t o e x p l o i t t h e new computers.

Leonard A . Lopez, U n i v e r s i t y o f I l l i n o i s : I t h i n k we do need t o t a k e a very

s e r i o u s l o o k a t t h e a r c h i t e c t u r e o f computer systems f o r f i n i t e element
analysis. O r . Nelson t a l k e d about t e s t b e d systems. I am a s t r o n g b e l i e v e r i n
t h a t approach. We have been r u n n i n g a t e s t b e d system a t t h e U n i v e r s i t y o f
I l l i n o i s s i n c e 1975. The system uses a v i r t u a l machine concept and emphasizes
f l e x i b i l i t y , machine independence, easy a d d i t i o n o f elements, and easy a d d i t i o n
o f n o n l i n e a r m a t e r i a l models. B a s i c a l l y these were areas o f i n t e r e s t t o
i n d i v i d u a l s i n our department i n t h e e a r l y 1970!s, and t h i s c a p a b i l i t y has
p e r m i t t e d these researchers t o t r y new i d e a s w i t h o u t needing t o deal w i t h a l l o f
t h e o t h e r problems a s s o c i a t e d w i t h f i n i t e element techno1 ogy.
The design o f t h e POLO-FINITE t e s t b e d i s based on 1970 software and hard-
ware technology. The b i g g e s t problem we a r e f a c i n g today i s t h a t o u r expecta-
t i o n s have grown s i g n i f i c a n t l y , and t h a t t h e k i n d s o f problems b e i n g e n v i s i o n e d
by o u r f a c u l t y f a r exceed t h e c a p a c i t y o f t h e combined hardware/software t e s t -
bed. I t h i n k we now must proceed t o i n v e s t i g a t e and e x p l o i t t h e a r c h i t e c t u r e Of

p a r a l l e l (mu1t i - p r o c e s s o r ) computers i n o r d e r t o s o l v e t h e n e x t g e n e r a t i o n of
p r o b l ems.
There would be many o b j e c t i v e s f o r a m u l t i - p r o c e s s o r t e s t b e d system.
C l e a r l y one o f them must be t h a t t h e t e s t b e d be f l e x i b l e enough t o t a k e i n t o
account e x i s t i n g m u l t i - p r o c e s s i n g a r c h i t e c t u r e s , as w e l l as a r c h i t e c t u r e s which
have n o t y e t been conceived. I t w i l l t a k e y e a r s t o develop such a system and

199
p u t i t into production. The corresponding hardware technology i s changing very
rapidly, and i t will be necessary t o be able t o move the testbed t o various new
architectures as they are designed and implemented. I n order t o give you an
idea about how f a s t things are changing, I would l i k e t o again point t o o u r
experience a t the University of I l l i n o i s . Six months ago we were confined
primarily t o uti1 izing a Cyberll75 computer. Today we have two national centers
f o r supercomputers--one based on mu1 t i p l e Cray XMP!s and the other based on
the CEDAR machine. The CEDAR machine i s a research machine based on ALLIANT
e a1 so h a v e a network of Apol lo super-micro-computers which
computers ( t o d a y ) . W
can be used as a multi-processor network; i t i s also viable f o r f i n i t e element
calculations. All of these additions in hardware took place in six months.
Additional changes w i l l occur within the next one t o two years.
We, as individuals working i n f i n i t e element software technology, must be
careful n o t t o box ourselves i n , or exclude the kinds of things which will be
happening in the near future. I t h i n k we can accomplish t h i s objective by
seriously studying a n d redesigning the architecture o f o u r computer systems, and
developing new testbeds t h a t a r e flexible enough t o meet the challenges of the
future.

K . C. P a r k , Lockheed Palo Alto Research L a b o r a t o r y : You are opening a box of

worms. Dr. Bahram Nour-Omid and I have been exploring the Cal Tech Hypercube .
f o r the l a s t 6 months. I t took us 2-1/2 months t o successfully run a
one-dimensional wave propagation problem. And then the word I j u s t g o t from him
on the plane was, a f t e r struggling a b o u t 2 t o 2 - 1 / 2 months, we j u s t have been
able t o r u n a simple 2-D p l a t e bending problem. What t h i s means i s t h a t i t i s
very d i f f i c u l t t o implement software on parallel machines, a t the moment, mainly
because the supporting system u t i l i t i e s a r e n ' t there. And i t ' s g o i n g t o take
a b o u t 3 t o 5 years before we engineers can--in a routine manner--use a parallel
machine, e i t h e r a FLEX machine or a Hypercube, or t h e l i k e . Right now, f o r
example, the FLEX people i n s i s t t h a t they have a FORTRAN compiler. B u t when we
looked a t i t , i t really doesn't have a good solid FORTRAN cross compiler and
.
neither does the Cal Tech Hypercube. What t h i s means i s we have t o s t a r t a l l
over again using the C language. And imagine the conversion labor involved in
rewriting sophisticated f i n i t e element programs into the C language. So I think
we have 3 t o 5 years before we can s t a r t t o see the routine uses of the parallel
machines.

200
The o t h e r comment t h a t I would l i k e t o make i s t h a t p a r a l l e l machines a r e n o t
cheap. What t h a t means i s t h a t o n l y wealthy p r i v a t e companies, such as some of
t h e p r o f i t-maki ng f i n i t e element a n a l y s i s companies, 1 ike Marc , o r
MacNeal-Schwendler, may be a b l e t o a f f o r d p a r a l l e l machines. But i n t h e
meantime we want t o m i n i m i z e t h e t i m e i t takes t o move from u n i v e r s i t y research
products t o p r a c t i c a l applications--say on a p a r a l l e l machine. Then t h e c o s t of
t h e machine has g o t t o go down, o r somebody should p r o v i d e t h e funds. Otherwise
you w i l l n o t see a wide d i s s e m i n a t i o n o f p a r a l l e l machines.

Stroud: Today, i t ' s n o t so much t h e c o s t o f t h e m u l t i p l e processor machines,

i t ' s t h e l a c k o f software. T h a t ' s r i g h t , FLEX m i g h t n o t even have a FORTRAN
compiler--and t h a t ' s a problem. But, I c e r t a i n l y hope i t w o n ' t be 5 y e a r s
before we a r e r o u t i n e l y u s i n g t h e F l e x computer. I t h i n k t h a t we w i l l be a b l e
t o b e g i n t a l k i n g about s t r u c t u r a l a n a l y s i s on p a r a l l e l machines i n 18 months.

V i n c e n t Godino, B o l t , Beranek, and Newman: I would l i k e t o say a few words

about o u r experience i n p a r a l l e l processing. Under DARPA sponsorship, BBN has
developed a para1 1e l processor c a l l ed t h e B u t t e r f l y . We have a 128-node machi ne
c u r r e n t l y operational . The S t r u c t u r a l Mechanics department, which I head, has
been d e v e l o p i n g s t r u c t u r a l mechanics s o f t w a r e on t h i s machine i n c l u d i n g m a t r i x
o p e r a t i o n s , Gauss e l i m i n a t i o n , e t c . Based upon our experience on t h i s
p a r t i c u l a r machine, I must t a k e i s s u e w i t h some o f K. C. P a r k ' s comments. In
terms o f ease of implementation, our s c i e n t i f i c programmers, w i t h no p r e v i o u s
experience on any p a r a l l e l processors, were a b l e t o program several extremely
e f f e c t i v e a l g o r i t h m s i n a s h o r t t i m e and w i t h minimum h e l p from people
experienced on t h e machine. The B u t t e r f l y i s a v a i l a b l e t o DARPA c o n t r a c t o r s
t h r o u g h t h e ARPA net. Next g e n e r a t i o n p a r a l l e l processors c o n s i s t i n g of
thousands o f nodes a r e c u r r e n t l y under development. I guess I would l i k e t o
agree w i t h J e f f Stroud i n t h a t I t h i n k t h a t we!re a l o t c l o s e r t h a n f i v e y e a r s
b e f o r e these machines a r e g o i n g t o be r e a l t o o l s b e i n g used t o solve. s t r u c t u r a l
mechanics problems.

Jimmy Y. L. Ho, Lockheed, Sunnyvale: We need t o r e a l i z e t h a t t h e m a t e r i a l

p r o p e r t i e s and v a r i o u s o t h e r p r o p e r t i e s , such as j o i n t behavior, a r e s t a t i s t i c a l
quantities. They do n o t have a p r e c i s e value.

201
-
Szabo: The r e a l i t i e s a r e t h a t we do indeed deal w i t h s t o c h a s t i c d i f f e r e n t i a l
equations. The l o a d i n g , t h e m a t e r i a l p r o p e r t i e s , t h e thicknesses, and so on a r e
r e a l l y s t o c h a s t i c v a r i a b l e s . I ' m aware o f some r e s e a r c h i n t h i s area t h a t ' s
j u s t b e g i n n i n g t o address t h i s f a c t . Monte C a r l o methods a r e o u t o f t h e
q u e s t i o n because o f t h e s i z e o f p r a c t i c a l problems. We cannot perform so many
d i f f e r e n t analyses w i t h d i f f e r e n t i n p u t data. B u t i t i s p o s s i b l e t o study
c e r t a i n bounding c h a r a c t e r i s t i c s o f t h e d a t a and r e f 1 e c t those c h a r a c t e r i s t i c s
i n o u r answers i n an honest way. And t h e p o i n t r a i s e d here c o u l d r e a l l y be a
very i m p o r t a n t t o p i c f o r NASA t o t a k e up as a r e s e a r c h study--how t o be more
r e a l i s t i c i n t h e way we t r e a t computed data. The number i s n o t 10.75, but,
rather, i t i s a stochastic variable. Understanding t h e bounds w i t h i n which t h e
answer i s v a l i d i s very important. D e l i v e r i n g a number d o e s n ' t mean very much
u n l e s s I ' m a b l e t o s t a t e t h e confidence l e v e l I can a s s o c i a t e w i t h t h a t number.

R o b e r t Melosh, Duke U n i v e r s i t y : I see f o u r d i f f e r e n t groups o f people here, and

t h a t ' s p r o b a b l y t h e b a s i s f o r t h e c o n f l i c t . I see a group o f a n a l y s t s who want
t o be responsive t o t h e demands o f t h e i r employers and t u r n o u t good analyses,
and r i g h t now, as I see i t , t h e a n a l y s t has a v e r y h e r o i c j o b t o t u r n o u t
analyses t h a t he can depend on. There i s another group h e r e represented by
program developers who a r e responsive t o t h e needs o f t h e a n a l y s t s . They want
t o s u p p l y programs t o h e l p t h e a n a l y s t do h i s j o b , b u t they d o n ' t have t h e f i n a l
r e s p o n s i b i l i t y f o r the analysis. The engineer does. The t h i r d group i s t h e
group you r e p r e s e n t , J e f f , t h e Langley group who want t o do r e l e v a n t research t o
h e l p t h e f i r s t two groups. The f o u r t h group i n c l udes t h e u n i v e r s i t y people who
want t o do r e l e v a n t r e s e a r c h t o h e l p t h e f i r s t t h r e e groups. I t h i n k i f you
want t o c e n t e r on what you should do as a group, you have t o decide who you want
t o s a t i s f y i n t e n s o f s e t t i n g your p r i o r i t i e s . Do y o u want t o s a t i s f y t h e
a n a l y s t s a t t h a t end o f t h e extreme o r t h e r e s e a r c h e r a t t h e o t h e r end o f t h e
extreme? You have t o s e t t l e on t h a t i n o r d e r t o s e t t l e on what y o u r o b j e c t i v e s
w i l l be f o r research.

Bahram Nour-Omid, Lockheed P a l o A l t o L a b o r a t o r y : There a r e a couple o f i s s u e s

t h a t have come up i n my work on t h e p a r a l l e l p r o c e s s o r - - i n p a r t i c u l a r , t h e one
down i n Cal Tech. F i r s t o f a l l I ' d l i k e t o d i s t i n g u i s h between p a r a l l e l proces-
sors. The one based on s i n g l e i n s t r u c t i o n m u l t i - d a t a , which a r e Cray and Cyber
205 computers. And t h e n e x t g e n e r a t i o n computers which a r e g o i n g t o be m u l t i -

202
i n s t r u c t i o n , mu1 ti-data-stream computers. Now these computers come i n a v a r i e t y
of d i f f e r e n t a r c h i t e c t u r e s , some o f them a r e based on message passing between
processors, and o t h e r s a r e based on a shared memory environment. It i s quite
p o s s i b l e t o w r i t e programs--code--and g e t them t o work on these computers. But
whether those codes a r e g o i n g t o be e f f i c i e n t i s a completely d i f f e r e n t s t o r y .
One can w r i t e a Gauss e l i m i n a t i o n procedure, probably i n a m a t t e r o f months.
However, i t t u r n s o u t t h a t on c o n c u r r e n t processors, which a r e t h e m u l t i -
i n s t r u c t i o n , m u l t i - d a t a computers, t h e t i m e taken f o r a message t o go from one
processor t o another i s e q u i v a l e n t t o between 10 and 100 m u l t i p l i e s . Therefore,
one has t o l o o k a t a l g o r i t h m s t h a t u t i l i z e t h e g i v e n a r c h i t e c t u r e o f the compu-
ter. I t i s pointless t o look a t algorithms t h a t require a l o t o f i n t e r -
processor communication. One has t o l o o k a t a l g o r i t h m s which d o n ' t do t h a t . I n
P a r t i c u l a r , t h e most promising ones t h a t we have seen a r e i t e r a t i v e methods and
t h e t y p e which a r e based on p a r t i t i o n i n g o f a s t r u c t u r e i n t o substructures.

That b r i n g s up another problem which I t h i n k i s q u i t e unsolved and t h a t i s the

mapping o f a f i n i t e element mesh o n t o a s e t o f processors. People have done a
l o t o f work on p a r t i t i o n i n g meshes, r e g u l a r meshes--and mapping them on t h e
c o n c u r r e n t processors. B u t once t h e mesh g e t s q u i t e general, such as t h e ones
we have s e e n - - f i n i t e element mesh o f an a i r c r a f t - - t h e n I t h i n k i t ' s very h a r d t o
say how t o automate t h e process o f mapping those meshes on t h e computer.

Joop Nagtegaal, MARC A n a l y s i s Research Corp.: I would l i k e t o make two b r i e f

comments on two very d i f f e r e n t s u b j e c t s . One i s t h e p a r a l l e l processing. I
b e l i e v e t h e r e a r e a l o t o f a l g o r i t h m s a l r e a d y around which l e n d themselves t o
p a r a l l e l i z i n g computations. I know o f Gaussian e l i m i n a t i o n equation s o l v e r s
which can be para1 1 e l i z e d by s e l fcontinuous s u b s t r u c t u r i n g technique. Certainly
you can do a l l element c a l c u l a t i o n s i n p a r a l l e l . What I see i s m i s s i n g a t t h e
moment i s o p e r a t i n g s o f t w a r e on t h e s e m u l t i p r o c e s s o r machines which we can
program such t h a t they work. T h a t ' s what I see c u r r e n t l y as t h e b o t t l e n e c k .
B u t I see t h a t t h i n g s a r e changing.So, a l i t t l e b i t o f t i m e and I see a l o t o f
t h i n g s g o i n g on. The second comment on a whole d i f f e r e n t s u b j e c t i s on t h i s
u n c e r t a i n t y business, s t o c h a s t i c a n a l y s i s . I ' d l i k e t o add t h a t NASA i s spon-
s o r i n g such a p r o j e c t . I t ' s a d i f f e r e n t branch o f NASA, b u t I ' d l i k e t o t h i n k
o f NASA as one b i g o r g a n i z a t i o n . And t h e r e a r e c e r t i n l y some promising t h i n g s
g o i n g on. And I would be happy t o t a l k about t h a t w i t h anyone.

203
Stroud: NASA Lewis i s d o i n g t h a t work.

Nagtegaal: And t h e r e a r e several people from Lewis a t t h i s workshop.

Stroud: We're a1 1 one good team.

Nagtegaal: T h a t ' s what I l i k e t o t h i n k , yes, thank you.

Stanley: I ' m g e t t i n g t h e impression now from l i s t e n i n g t o a l l t h e comments--

from b o t h t h e panel and t h e audience--that people seem t o view t h i s i d e a of
g l o b a l / l o c a l a n a l y s i s as an i n t e r i m measure pending t h e f u l l - s c a l e u t i 1 i z a t i o n
o f p a r a l l e l p r o c e s s i n g - - t h a t as soon as computers become f a s t enough, t h e r e
w o n ' t be a need t o use g l o b a l f u n c t i o n s and e x a c t s o l u t i o n s . W e ' l l j u s t go i n
t h e r e w i t h b r u t e f o r c e w i t h 3-D f i n i t e element a n a l y s i s . I ' m wondering if
t h a t ' s t h e case, because i f t h a t i s t h e f e e l i n g t h e n perhaps we should r e t r a i n
i n midstream here. Maybe t h a t ' s a r h e t o r i c a l question. On t h e o t h e r hand, if
we do c o n t i n u e w i t h these g l o b a l / l c c a l methodologies, I would l i k e t o see them
become more problem o r i e n t e d , I ' d l i k e t o see them c o n t i n u e i n t h i s manner, I ' d
1 ike these benchmark p r o b l ems-- t h e NASA problems-- t o be d i s t r i b u t e d among people
t h a t a r e c o l l a b o r a t i n g i n t h i s area. And I would l i k e t o t h e n compare t h e
methods used t o s o l v e these problems based on a number o f c r i t e r i a , such as
(1) r e l i a b i l i t y ; ( 2 ) computational e f f o r t , which would i n c l u d e t h e a n a l y s t ' s
e f f o r t as w e l l ; (3) implementational e f f o r t , t h a t ' s i m p o r t a n t because r e t r e a d i n g
s o f t w a r e can be a huge c o s t and, f i n a l l y , ( 4 ) g e n e r a l i t y . I f worse comes t o
worst, we w i l l have t o b u i l d s p e c i a l purpose codes as t h e y do i n t h e f l u i d s
business. But, o p t i m a l l y , i t would be n i c e t o hang on t o t h e general purpose
b e n e f i t s we g e t from t h e f i n i t e element method by u s i n g such techniques as
reduced b a s i s g l o b a l f u n c t i o n s , based on f i n i t e element b a s i s f u n c t i o n s , i n t h e
f i r s t place, so t h a t we can automate t h i s g l o b a l / l o c a l i z a t i o n .

Ahmed K . Noor, George Washington U n i v e r s i t y : J u s t two comments. One w i t h

r e g a r d t o t h e q u e s t i o n about u n c e r t a i n t y and confidence i n t h e r e s u l t s . Short
of doing a s t o c h a s t i c a n a l y s i s I t h i n k we can g e t some c o n f i d e n c e i n t h e r e s u l t s
by c a r r y i n g o u t a s e n s i t i v i t y a n a l y s i s . S e n s i t i v i t y of t h e response w i t h
r e s p e c t t o t h e i n p u t parameters, f o r example. T h i s would g i v e us some c o n f i -
dence i n t h e numerical r e s u l t s . The second comment d e a l s w i t h p a r a l l e l proces-
sing. P a r a l l e l processing i s a system issue. U n f o r t u n a t e l y , much o f t h e work
t h a t i s b e i n g done has addressed c e r t a i n aspects o f t h e a n a l y s i s process. I n

204
p a r t i c u l a r , t h e numerical a1 gorithms f o r hand1 i n g t h e a1 g e b r a i c equations. This
i s very i m p o r t a n t , o f course, and computer t i m e spent i n a l g o r i t h m s i s dominant
i n present-day f i n i t e element a n a l y s i s , b u t i t m i g h t n o t be enough t o address
t h a t a l o n e i f we want a very e f f i c i e n t s o l u t i o n procedure. I t h i n k we have t o
address t h e q u e s t i o n from a h i g h e r l e v e l s t a r t i n g w i t h t h e f o r m u l a t i o n , l o o k i n g
a t t h e d i f f e r e n t phases o f t h e analyses, and, as I s a i d i n ' m y p r e s e n t a t i o n , t h a t
we m i g h t have t o move away from some o f t h e t r a d i t i o n a l f i n i t e element a n a l y s i s
processes 1 i k e , f o r example, bypassing t h e assembly process. T h i s i s t h e most
d i f f i c u l t p a r t o f t h e s o l u t i o n process f o r v e c t o r i z a t i o n o r p a r a l l e l i z a t i o n .
So, I t h i n k we have t o l o o k more i n t o t h e o v e r a l l computational s t r a t e g y r a t h e r
t h a n j u s t t h e numeri c a l a1 g o r i thm.

Nelson: T h i s comment has t o do w i t h something we h a v e n ' t t a l k e d a l o t about,

a l t h o u g h we saw a l o t o f p i c t u r e s . That was t h e idea o f t e s t i n g . I ' m reminded
o f an o l d T i n i u s Olsen t e s t i n g m o t t o which says t h a t one t e s t i s w o r t h a thou-
sand e x p e r t opinions. We want f i n i t e element a n a l y s i s t o be a t r u e e n g i n e e r i n g
tool. Which means t h a t t h e f i n i t e element s o f t w a r e has t o be c o n f i g u r e d t o be
flexible. I f , f o r example, we s t a r t l o o k i n g a t such t h i n g s as a NASA t e s t
panel, i n which we b e g i n t o see f a i l u r e such as d e l a m i n a t i o n and t h i n g s o f t h i s
nature, we want t o be a b l e t o i n c o r p o r a t e i n t h e s o f t w a r e new m a t e r i a l models o r
new f a i l u r e laws f o r composites. So t h e n we can have t i g h t feedback between
experiment and a n a l y s i s . We w i l l a l l b e n e f i t from h a v i n g codes t h a t a r e n o t
j u s t b i g g e r - - t h e codes a r e j u s t a l i t t l e b i t smarter, because we are.

J . N. Reddy, V i r g i n i a P o l y t e c h n i c I n s t i t u t e : I want t o make one comment i n

response t o t h e f i r s t comment made by Dr. Backman. Most o f my colleagues f r o m
u n i v e r s i t i e s , I hope, share a t l e a s t t h e s p i r i t o f what I ' m going t o say. I
t h i n k most o f t h e s t u d e n t s t h a t a r e coming o u t o f t h e u n i v e r s i t i e s , I am disap-
p o i n t e d t o say, a r e t a u g h t t h e r e c i p e - t y p e approach, f o r m u l a p l u g g i n g and
c h e c k i n g compared t o my o l d e r c o l l e a g u e s who have a l o t o f i n g e n u i t y and physics
background. My o l d e r c o l l e a g u e s a r e b e t t e r a t modeling than, I think, the
c u r r e n t students. And I d o n ' t want t o use t h e same k i n d o f approach as proposed
Or suggested by D r . Backman. I want t h e s t u d e n t s t o know more p h y s i c s t h a n
number c r u n c h i n g and formula plugging. I t h i n k i f they have an understanding Of

t h e b a s i c p h y s i c s and i f they have c e r t a i n t o o l s , when t h e y g e t o u t o f school I

I'll g i v e an example o f my own. A f t e r 1
t h i n k they can be t r a i n e d on t h e j o b .
graduated, I took a j o b w i t h Lockheed H u n t s v i l l e Space Company. Before t a k i n g
t h a t j o b , I o n l y worked w i t h H i l b e r t spaces and Sobolev spaces, and also, of
205
course, I l e a r n e d a g r e a t deal from P r o f e s s o r Oden and o t h e r people i n terms o f
p h y s i c s and mathematical modeling. A t my new j o b , I was asked t o do a 3-D
a n a l y s i s o f impact, hydrodynamic approximation. One week a f t e r I g o t i n t o t h e
j o b , I had a j o b o f f e r t o go t o the U n i v e r s i t y o f Oklahoma, b u t I decided t o
Stay and complete t h e p r o j e c t g i v e n t o me. And I d i d complete i t i n 6 months.
The p o i n t i s , i f you have t h e r i g h t background i n terms o f ' t h e mathematics and
physics, I t h i n k you can l e a r n number c r u n c h i n g very easily. But i f you t r a i n
y o u r s e l f w i t h o n l y number crunching, you may n o t g e t t h e p h y s i c a l understanding.
So I t h i n k we should be very c a r e f u l i n t r a i n i n g our students.

Stroud: My b e t i s t h a t B j o r n wasn't suggesting t h a t s t u d e n t s be t r a i n e d a

p a r t i c u a r way. Rather, he wanted t h e p r o d u c t o f our CSM e f f o r t s t o be s o f t -
ware.

Backman Yes. I ' m n o t sure what I s a i d t h a t caused t h i s comment. I t ' s possi-

b l e t h a t I h i n t e d t h a t t h e l e v e l of s o - c a l l e d blackbox e n g i n e e r i n g has t o be
increased. I t h i n k t h a t we a r e s l o w l y r e a c h i n g t h e p o i n t where p r o l i f e r a t i o n o f
e x p e r t s has reached a l e v e l such t h a t we simply c a n ' t r e p l a c e them f a s t enough
i n the production f i e l d . And what I ' m t r y i n g t o say i s , o f a l l these s t u d e n t s
t h a t a r e coming up w i t h t h e proper background i n science and s t r u c t u r a l
mechanics and so on, how many do you expect t o be p r o f i c i e n t i n t h e o p e r a t i n g
systems t h a t e x i s t on t h e machines they a r e u s i n g ? How many do you expect t o
know t h e d e t a i l s o f t h e c o m p i l e r s we use on t h e system? How many o f them do you
e x p e c t t o understand, i n d e t a i l , t h e e q u a t i o n s o l v e r s ? And how many do you
e x p e c t t o understand t h e i n n e r workings of t h e o p t i m i z e r s ? Somewhere i n h e r e we
a r e g o i n g t o have t o s t o p a s k i n g these people t o become renaissance
personalities. I ' m j u s t t r y i n g t o achieve some balance h e r e where maybe t h i n g s
l i k e f i n i t e element systems u l t i m a t e l y f i t i n t o t h e same category as o p e r a t i n g
systens. And maybe t h e e n g i n e e r i n g community w i l l be d e a l i n g w i t h t h e a c t u a l
design problems. I ' m , by a l l means, i n f a v o r o f a s o l i d background i n t h e
sciences. I ' d r a t h e r see t h e i n d u s t r y t a k i n g on some a d d i t i o n a l e d u c a t i o n a l
r e s p o n s i b i l i t y i n t h i s area. The p o i n t i s t h a t sooner o r l a t e r you e i t h e r have
t o r e q u i r e 10 y e a r s o r more o f h i g h e r e d u c a t i o n o r y o u have t o i d e n t i f y how
deeply i n t o t h e problem you want t o go. And, I t h i n k , a method l i k e t h i s one
t h a t D r . Szabo has been advocating i s probably a f i r s t s t e p towards t h e a b i l i t y
t o h e u r i s t i c a l l y understand t h e r e s u l t o f y o u r a n a l y s i s w i t h o u t knowing t h e g u t s
o f each step o f t h e s o l u t i o n . And I'll s t o p w i t h t h a t .

206
 ORIGINAL PAGE
BLACK AND WHITE PHOTOGRAPH
I
Stroud: Thank you. I t ' s c l e a r t h a t we have opened up a l o t o f t o p i c s and we
have n o t r e s o l v e d them. And opening them up i s c e r t a i n l y one of t h e o b j e c t i v e s
o f t h i s workshop. There w i l l be many more workshops and conferences t h a t w i l l
a t t e m p t t o r e s o l v e some o f these questions. I want t o thank you a l l f o r coming
and p a r t i c i p a t i n g . I n p a r t i c u l a r , I want t o thank t h e speakers.

* O r i g i n a l c o l o r s l i d e shown h e r e i n b l a c k and white.

207
 TIRE MODELING

PRECEDING PAGE BLANK NOT FILMED

209
 N89-24649
FEATURES AND CHARACTERIZATION NEEDS OF
RUBBER COMPOSITE STRUCTURES

Farhad Tabaddor
Goodrich Company
Akron, Ohio

Abstract

This paper outlines some of the major unique features of rubber

composite structures. The features covered are those related to
the material properties, but the analytical features are also
briefly discussed. It is essential to recognize these features
at the planning stage of any long-range analytical, experimental,
or application program.
The development of a general and comprehensive program which fully
accounts for all the important characteristics of tires, under
all the relevant modes of operation, may present a prohibitively
expensive and impractical task at the near future. There is therefore
a need to develop application methodologies which can utilize
the less general models, beyond their theoretical limitations and
yet with reasonable reliability, by proper mix of analytical,
experimental, and testing activities.

OUTLINE:

1. CORDS

2. RUBBER

3. RUBBER COMPOSITES

Single Ply
Laminate

4. ANALYTICAL AND
COMPUTATIONAL ASPECTS

PRECEDING PAGE BLANK NOT FILMED

211
INTRODUCTION

The textile-cord-reinforced-rubber composites used in various

industrial products differ in many respects from the classical
rigid composites. There are a great number of papers, books, and
professional journals devoted to rigid composites. The literature
on rubber composites, however, is very limited as compared to
classical composites. To understand the mechanics of such composites,
it is essential to develop an in-depth understanding of the way in
which the internal variables of the constituents participate and
interact in responding to external agents, i.e., mechanical,
thermal or other environmental forces. It is, however, important
to first study the properties of each constituent before dealing
with the material properties of rubber composites.

CORDS

The textile cord reinforcements are structural members which may be

viewed as one dimensional in a microscopic sense. The cords
generally consist of several yarns twisted together. The yarns
also consist of numerous filament components organized in a
geometrical array with the view towards enhancing certain target
properties. The cord properties therefore depend on the properties
of the filaments and the geometrical organization as well as the
interfacial characteristics. The filament itself has also been known
to possess an internal structure at the microscopic level.

In view of the above consideration, the cord itself is a complex

structure which may be studied at the microscopic level (figs. 1
and 2 ) . The way in which the cord properties are related to
filament properties, geometries (ref. 1) and other variables is
the subject of textile mechanics (fig. 3 ) . Properties such as
strength, stiffness, and fatigue characteristics can be con-
trolled by internal variables when their relationships are well
understood.

From a higher scale of continuum mechanics we are, however,

concerned only with the phenomenological properties of the cords as
experimentally obtained without any attempt for relating such

212
measured values to the micro-structure in the sense of textile
mechanics. The properties of interest at this continuum level are
therefore the effective cord properties.
Such an approach enables us to bypass the complexity of the textile
mechanics in our formulation of composite properties. The
limitation is, however, that such measured properties serve as
llaveragesll
only and therefore the continuum elements should be at
least in the order of cord diameters.

Figure 1. Typical tire cord.

213
 ORlC I N AL
BLACK AND WHITE

Figure 2 . Cross s e c t i o n of t i r e cord (2).

0 5 IO 15 20 25
STRAIN

(1) RAYON (5) NYLON-66

(2) NOMEX (6) STEEL
(3) POLYESTER (7) GLASS
(4) NYLON-6

TESTED AT 70°F.,65% R.H.

INSTRON - 100% STRAIN RATE

Figure 3. Stress-strain characteristics of typical tire cords (2).

2 14
The cord is the major load-bearing member of rubber composite
structures and as such should provide strength and many other
characteristics of interest. Some of the expected performance
characteristics of,thetire, and for that matter any other
structure, can be directly related to cord properties. Let us
consider some of the tire performance characteristics which are
affected by cord properties. A partial list is given in Table 1.

Table 1
TIRE PERFORMANCE AFFECTED BY CORDS
o Burst Strength
o Bruise Resistance
o Tire Endurance (Separations)
o Power Loss
o Tread Wear
o High-speed Endurance
o Tire Size and Shape
o Groove Cracking
o Flat Spotting and Non-Uniformity
o Tire Cornering Force
o Tire Spring Rate
o Noise

215
Tables 2 to 4 of reference ( 3 ) provide a list of tire
requirements and the related cord requirements. Such relations
should be viewed with caution and qualifications.

Table 2
(REF. 3 )
RADIAL PASSENGER

Vehicle Trends Tire Requirements Cord Requirements

High Performance Low Aspect Ratio Modulus
(Increased Cornering' Lateral Stiffness
Forces 1
Dimensional Stability
Downsizing Downsizing - Monoply Fatigue
Modulus
Tenacity
Front Wheel Drive Improved Tread Wear Dimensional Stability
Lateral Stiffness
Retreading Aged Adhesion
Thermal Stability
Fatigue
Fuel Economy Rolling Resistance Tenacity
Modulus
Hysteresis

216
 Table 3
(REF. 3)
RADIAL TRUCK

Vehicle Trends Tire Requirements Cord Requirements

Increased Drive Increased Durability Tenacity
Position Loads Strength/Area Ratio
Improved Fuel Lighter Weight Tenacity
Economy/New Low Profile Fatigue
Transportation Retreading Aged Adhesion
Act Fatigue
Thermal Stability
Rolling Resistance Tenacity
Modulus
Hysteresis

Table 4
(REF. 3)
RADIAL AIRCRAFT

Tire Requirements Cord Requirements

High Deflection Tenacity
Modulus
Thermal Stability
Reduced Weight Tenacity
Strength/Area Ratio
Retread Aged Adhesion
Dimensional/Thermal Stability
Fatigue

217
 CORD VISCOELASTICITY
The textile cords used in rubber-reinforced composites are often
nonlinear viscoelastic. The viscoelastic deformations are associated
with the loss of energy. The dissipated energy appears as heat and
leads to temperature rise which in turn affects t h e material properties.
The cord is therefore an important contributor to the energy loss in
rubber-composite structures such as tires ( 4 ) . Figures 4 and 5 show
some of the viscoelastic properties of polyester cords.

Figure 4 . Carpet plot of 1 0 0 0 / 2 polyester real modulus.

I 218
 .!e

.I 6 7-1 1000/2 POLYESTER I bo

” I
.I4

.I 2
TAN 6

.IO

.O 8

.06

.04 I

.9
STRAIN x 15 MAXIMUM

F i g u r e 5. Carpet p l o t of 1 0 0 0 / 2 p o l y e s t e r loss m o d u l u s .

219
SUMMARY ON CORDS DISCUSSION

o Cords are structures and as such their effective properties are

geometry and boundary condition dependent.

o The effective cord properties are different when cords are

embedded in the rubber matrix.

o The calculation of the three-dimensional effective properties

by analytical homogenization is very complex and impractical.

o The properties of some cords such as nylon are highly temper-

ature sensitive, particularly around the glass transition
temperature.

o The properties are different in tension and compression.

o Cords are often nonlinear and viscoelastic.

220
RUBBER

The physical properties of rubber compounds depend on various

processing parameters and components. Here again, we are not
concerned as how these properties are related to the molecular
structure of the rubber and the physics of rubber vulcanization.
The focus is on determination of the relevant properties by proper
experiment and within the framework of continuum physics. The most
important property is the elasticity of the rubber, which is
distinctly different from other conventional materials. The most
distinctive features of rubber elasticity are the deformability and
the rapid recovery of the deformations when loads are removed.
Rubber remains elastic at extension ratios of several hundred
percent. Such elastic characteristics make the rubber unique in
this respect. In fact, the major developments in the theory of
large elastic deformations evolved around application to rubber
elasticity. The rubber elasticity is an important subject in
understanding finite-element analysis of such products.

RUBBER E L A S T I C I T Y

To understand rubber elasticity, we may first examine the

thermodynamics of reversible processes. The first and the second
laws of thermodynamics s t a t e t h a t ( 5 )
dE = Tds + dw
where E is the internal energy, T is the absolute temperature, S
is the entropy and w is the work done on the system. Experimental
work has shown that the rubber elasticity resides basically in the
entropy term. The rubber elasticity therefore has an entirely
different molecular origin than other elastic materials whose
elasticities are primarily associated with the increase in internal
energy through changes in molecular or atomic spaces.

221
 Much work has been carried out to formulate rubber elasticity.
One example of the molecular approach is one which considers the
molecular chain length having Gaussian distribution. The elasticity
parameters are calculated from such quantities as finite molecular
length and molecular weight between crosslinks. The entropy change
resulting from Gaussian theory leads to

in which N is t e number of network chains per unit volume

andX1 12 and '3 are the three prjncipal extension ratios
along the three mutually perpendicular axes of strain for pure
homogeneous strains.
The above equation provides a first order approximation to rubber
elasticity but is not adequate over a large range of deformations.

i 222
PHENOMENOLOGICAL APPROACH
An entirely different approach (6) is the phenomenological approach
of continuum mechanics. In this approach, the existence of a strain
energy function is postulated. It has been shown that such a strain
energy function should depend on deformation gradients. The equa-
tions for isotropic incompressible elastic media are as follows:

w = w (I1, 12, 13)

I1 = x,2 + x22 + 13
2

I2 = x 12x 22 + AllQ
2 2 + x2x3
2 2

For the incompressible case

13 = 1

In series form

223
An alternative form, s u g g e s t e d in ( 7 ) , is as follows

w = t (A1) + t (A2> + t (A3>

The stress-strain relations in the principal directions (the

stresses are force per deformed area) are

2 24
The one-dimensional stress-strain relation can be obtained from
the three-dimensional relations as shown in figure 6.

0.4

0.2

Figure 6. One-dimensional stress-strain relation.

225
It can be seen that the tangent compressive moduli increase
significantly as the compressive strain increases. This feature
realistically describes the rubber response in compression. In the
finite-element treatment of rubber materials, the rubber is often
modeled as linear, on the ground that the strains are small and
that the linear constitutive law should then provide a reasonable
approximation. This view is very disputable. It suffices to say
that the compressive strains remain small due to stiffening of
rubber in compression. This feature can only be handled properly
by nonlinear constitutive laws. In the linear model of rubber, the
material stiffness remains unchanged and therefore moderate
compressive forces produce high compressive strains that are not
seen in the actual structure. Such exceedingly large strains result
in distortion of elements which quickly leads to severe numerical
problems.

A difficult problem, in dealing with rubber elasticity, arises from

the incompressibility constraint condition. The incompressibility
condition leads to certain simplifications in the exact analysis of
the problem, basically because of reduction in the number of
unknown parameters. Such is not, however, the case with the finite-
element approach. In the variational formulation, for example, the
Lagrange multiplier introduces an additional unknown scalar
function into the finite-element formulation. This unknown should
be accommodated at the element level. The procedure results in a
significant increase in the number of total unknowns a n d a l s o may
result in an ill-conditioned stiffness matrix. A great deal of
research has been carried out to find the most suitable element for
handling incompressibility. The subject is still open but the fact
remains that the incompressibility imposes an additional burden in
numerical analysis. The incompressibility condition, aside from
being inconvenient in the finite-element analysis, is an
approximation for rubber-like materials. Such an approximation
becomes increasingly less accurate as the percentage of carbon
black increases in the rubber compound. The exact enforcement
of incompressibility is therefore not actually needed.

226
NEARLY INCOMPRESSIBLE MATERIALS

The strain energy function for nearly incompressible materials can

be obtained by a series expansion of the strain energy function
about (13-1), and retaining the leading terms of up to second order
in (13-1) as follows

w = cc cij (I1 - 3) i ( I 2 - 3 9 + w2 - (I3 - 1) + wg ( I 3 - 1)* + ...

ij

One special case of the above equations is when the function W2

and W 3 are considered constants. The constants are described as
follows:

227
 The near incompressibility can now be enforced by assigning large
values to H. In fact, as H approaches infinity, (13-L) approaches
zero so that the strain energy remains finite. The higher the
value of H, the closer the incompressibility would be satisfied.
H is referred to as the penalty number. In finite-element analysis,
however, the large values of H can lead to overriding stiffness
which results in numerical problems. The penalty method nevertheless
permits the satisfaction of near incompressibility without increasing
the number of unknowns in finite-element formulation. Figure 7 shows
a set of typical properties for the rubber.

.. -.2

C,,= 0472 (kg Imrn’)

C0l = 0211

I
C = - 00414
/ C,,= - 000262
8
I c,o= 00112
I H I =- ( c , o + z c o l )
I
I H = 11.138
I
I
I

Figure 7. Stress-strain relation, (8).

228
FRACTURE AND FATIGUE PROPERTIES

The fatigue of the rubber has been the subject of many investigations
in the past. The'rubber fatigue is intimately related to the nature
of rubber fracture and cut growth. The fracture mechanics approach
for rubber was first adopted by Rivlin and Thomas (9) and Thomas (10)
who promoted the concept of the tear energy in describing the cut
growth mechanism. The tear energy approach has been applied to the
study of the crack growth problem and to the description of fatigue
behavior of rubber. We only consider the mechanically induced fatigue
and this therefore excludes the fatigue caused by or resulting from
non-mechanical sources such as aggressive environment. Some of the
non-mechanical sources, such as ozone cracking in elastomers, may,
however, be more damaging than mechanical sources.
Bussels early results (11) on NR compounds, shown in Table 5, clearly
demonstrate the unique feature of rubber fatigue.
Table 5
EFFECT OF STRAIN CYCLE

STRAIN CYCLE LIFE, MINUTE

5 - 60% Extension 10
10 - 60% II
25
15 - 60% II 90

17 - 60% II
150+

20 - 60% 00 00

Some experimental fatigue data on various compounds are shown in

Figures 8 to 11 taken from References (12-14).

229
 LMIN - L o
%MINIMUM STRAIN = LO ( 100)70

F i g u r e 8. F a t i g u e l i f e o f NR as a f u n c t i o n of m i n i m u m s t r a i n
r a n g e ( f r o m ref. 12).

0 9
I o3 lo4 lo5 IO6 '01 IOD IO
NUMBER OF CYCLES TO BREAK

Figure 9. Imposed extension versus number o f c y c l e s to break

(from r e f . 1 2 ) .

230
 I

DUMEELLS OF REVOLUTION
1
75% TOTAL STRAIN CYCLE

IO4 t 7. MINIUUN STRAIN

F i g u r e 10. C y c l e of f a i l u r e v e r s u s minimum s t r a i n
f o r NR and SBR (from refs. 1 3 and 1 4 ) .

1 LIFE V 5 % MINIMUM STRAIN

-
- NR - SER

7 5 H R 25EE

IC0 NR

25 h?-75SE

100 SBR

t I
IO5' -:5 -3: 1/2
%
6
MINMUM STRAIN
+ 50

F i g u r e 11. C y c l e of f a i l u r e v e r s u s minimum s t r a i n for

v a r i o u s c o m p o s i t i o n s (from r e f s . 1 3 and 14).

231
 SUMMARY ON RUBBER

o Rubber remains elastic at extension ratios of several hundred.

o Rubber is almost incompressible.

o The rubber elasticity can be most conveniently expressed

through the strain energy function.

o Rubber has unique fatigue properties. Extensive experimental

work needs to be carried out for fatigue characterization of
various rubber compounds.

o Appropriate failure theories are needed for interpretation of

stress data from finite-element analysis.

o Properties are temperature dependent.

232
CORD REINFORCED RUBBER COMPOSITE - SINGLE PLY

In the macro-mechanics approach to composites, the actual

heterogeneous medium, Figure 12, is replaced by an "equivalent"
homogeneous medium. The equivalent medium has the same geometry as
the heterogeneous medium but different material properties. The
vveffectivevf
properties of the latter may belong to a different
class of symmetry than those of the constituent materials. For
instance, for isotropic constituents, the effective properties may
be isotropic, orthotropic or anisotropic depending on the internal
geometry of the composite.
Let us consider a two-dimensional composite made of an isotropic
matrix and reinforced by doubly periodic arrays of isotropic
fibers as shown in Figure 13. A compo-site representation of the
problem is a homogeneous medium with the same external geometry and
boundary conditions as those of the heterogeneous case, but with
orthotropic properties. The effective orthotropic properties are
then defined for a vvrepresentativevv
element. It is very important
to first understand what is meant by that. Hashin (15) defined and
discussed this question and has drawn some analogies to the case of
homogeneous continua. In such continua, the hypothesis is that the
continua retain their properties even for infinitesimal elements.
Due to the discrete nature and microstructure of homogeneous media,
the infinitesimal element of continuum mechanics should be large
compared to the scales of the microstructure. The infinitesimal
element therefore exhibits properties which are some statistical
averages of the microstructure properties. On the other hand, the
infinitesimal elements should be very small compared to the
dimensions of the continua. It then follows that the physical
quantities at a point, such as moduli, stress and strain components,
are in reality associated with averages over infinitesimal elements
and not with a geometrical point. Due to the complex nature of the
microstructure, the properties of the continuum infinitesimal
elements are determined experimentally as opposed to calculating
the averages from a microstructural theory. The same comments also
apply to the lvrepresentativev'
element of composite theories, that
is to say that a Ivrepresentativevv
element is the infinitesimal
element of composite materials. It therefore should be large
compared to the dimensions of its material phases. A
Ivrepresentativevv
element, as defined, would retain and represent
the properties of the composite continua and furthermore these
properties would be insensitive to boundary conditions.

233
 Figure 12. Representative element.

2 34
A distinction must be made between a "representative" element and
unit elements or unit cells. The latter is defined as building
blocks, so that the continua can be constructed by repeated use of
such units. For example, consider the composite of Figure 13.
There are a number of unit cell configurations which can equally
serve as building blocks. Some of these possible choices are shown
in Figure 14. Figure 15 shows the boundary effects on shear
deformations. The average properties of these units, unlike
those of a llrepresentativell
element, are highly boundary condition
.
dependent (16) A llrepresentativell element, however, consists of a
large number of unit cells.
From the definition of a Ilrepresentativell element, it is apparent
that the properties calculated or experimentally measured would be
rather insensitive to boundary conditions. The problem, however,
is that such a calculation would be a formidable task. Most
published works dealing with a calculation of effective properties
use a unit cell as the basis of their computations. Even for unit
cells, the exact solutions can be obtained for only some simple
geometries and simple boundary conditions. Another approach
adopted is to determine the upper and the lower bounds for these
properties through approximate solutions of energy formulations.
These bounds are, however, far apart for composites with high
cord-to-matrix stiffness ratios, such as rubber composites, and are
therefore of little practical use.
The immediate question is, therefore, how sensitive these properties
are to boundary conditions specified over the surface of the
unit cell, and, furthermore, how the element boundary conditions are
influenced by global boundary conditions. These questions have not
been fully investigated in the literature. It can, however, be
stated that such sensitivities should depend on the relative
stiffnesses of the cord and the matrix. As the stiffness of the
matrix approaches that of the reinforcing materials, the
effective properties should become less sensitive to boundary
conditions and obviously independent of the boundary conditions in
the limiting case of identical constituents1 properties. These
sensitivities are thus of greater concern in rubber composites t h a n
conventional rigid composites. The ratio of cord to rubber
stiffness can exceed 30,000 in some composites which is far greater
than those of rigid composites.

235
 Figure 13. Componsite with doubly periodic f i b e r arrays.

(11) (111)

Figure 14. Several u n i t c e l l s .

236
 ( a ) Undeformed (b) Deformed
F i g u r e 15. m i f o r m s h e a r s t r a i n boundary c o n d i t i o n .

237
SUMMARY ON BASICS OF RUBBER COMPOSITES

o The calculation of effective properties of rubber composites

by homogenization is subject to more limitations than rigid
composites. These limitations need to be established on a
sound theoretical basis.

o In experimental determination of the effective properties,

the effects of size and the boundary conditions should be
investigated.

o The homogenized properties of a single ply cannot be easily

calculated since all the 3-D homogenized cord properties are
not generally available, with sufficient accuracy. Some of
the effective composite properties, however, may prove to be
somewhat insensitive to inaccuracies in cord properties.

o A necessary characteristic of a composite material is

statistical homogeneity. A representative element, used for
calculation of the e f f e c t i v e properties, should be large
compared to typical phase regions. A representative element
of a single ply does not have such a characteristic in the
thickness direction. The out of plane properties of a single
ply, therefore, are subject to question.

o The effective properties are different in tension and

compression.

238
CHARACTERIZATION OF LAMINATED COMPOSITES
We now consider a laminated structure, composed of N layers of
cord-reinforced composite materials, as shown in F i g u r e 16.
Each layer of heterogeneous composite may then be modeled as
homogeneous but orthotropic with respect to the proper local
coordinate of each layer. The constitutive equations of the
laminated composite, however, must be obtained with respect to
a global coordinate system XYZ, as shown in Figure 16. The
transformation relations may then be used for the appropriate
layers to carry out the required transformation. The layers
are numbered from top to bottom and no symmetry is assumed with
respect to any axis. The stress and moment resultants for the
laminated structure in terms of stresses are defined as follows:

Figure 16. Geometry of typical laminated composites.

239
 ~~ ~

The displacement components and the resulting constitutive relations

are

240
SUMMARY ON LAMINATED COMPOSITES

o Each ply may be homogenized, subject to the same limitations

as those of a single ply.

o Complete homogenization is not possible since the coupling

between the forces and the moments can only be accounted for
by preserving nonhomogeneity in the thickness direction.

o Classical kinematics constraints, such as the Kirchhoff

hypothesis, do not apply.

o Plate/shell type stiffnesses are extremely hard to measure.

o Constitutive relations are nonlinear due to angle change

between the cords of adjacent plies.

241
ANALYTICAL AND COMPUTATIONAL FEATURES

KINEMATICS
The rubber composite structures undergo large strains as well as
large rotations. The major kinematic features are listed in table 6.

Table 6

KINEMATIC FEATURES

LARGE RUBBER STRAINS

LARGE ROTATIONS
CORD ANGLE CHANGES
NEAR INCOMPRESSIBILITY CONDITION

CONSTITUTIVE RELATIONS FOR COMPOSITES

Single ply
The single-ply composites can be considered as linear orthotropic
when referred to axes of symmetry. For large displacements but
moderately small strains, the rubber and the cords can be considered
linear, but the strain-displacement gradient relations are nonlinear.
The most appropriate form of constitutive equations, in such c a s e , is
tS = tC t
o ij o ijkl o"k1
where tCijkl are orthotropic properties referred to coordinates
of initial material symmetry. The term kSij are the components of
the second Piolla-Kirchhoff stress tensor and t
o ~ k l are the compo-
nents of the Lagrangian strain tensor. This form is invariant under
a rigid body motion and therefore needs to be updated due to the cord
angle change. This does not hold for constitutive equations

242
which utilize other stress and strain measures. The major drawback
of modeling each ply separately is the increased size of the
finite-element problem.

Several plies
It is often more convenient to lump several plies together in one
element and therefore reduce the size of the problem. The preceding
equation can be utilized but not longer remains unchanged due to the
change in the angle between the minus and the plus ply. In such
models the orthotropic properties of the combined plies continually
change as functions of the cord angle.

Composites with nonlinear constituents

It is often necessary to account for the rubber nonlinearities in
the finite-element analysis, even for small strains. The reason
for such a need is that the typical compressive forces can produce
large rubber strains if the rubber stiffening in not properly
accounted for in the material modeling. These unrealistic large
strains may lead to element distortion and eventual loss of
numerical stability in the finite-element model. No rigorous
nonlinear composite constitutive equations have appeared in the
literature for rubber composites. This awaits further research
in this field.

Various modelinglevelsmay be required depending on the nature

of the problem. These models and their respective features are
listed in Table 7.

243
 Table 7

MODE L I N G FEATURES

ELEMENT -
TYPE-
FOR COMPOSITES FEATURES

MEMBRANE ELEMENTS REQUIRES 2-D MATERIAL PROPERTIES,

INTERPLIES MODELED INDEPENDENTLY
ONE ELEMENT PER PLY AND ONE ELEMENT
PER INTERPLY IN THE THICKNESS
DIRECTION, NO NEED FOR MATERIAL
UPDATE DUE TO THE CORD ANGLE CHANGE

TWO- AND THREE-DIMENSIONAL

ELEMENTS REPRESENTING:

SINGLE PLY REQUIRES 3-D MATERIAL PROPERTIES,

ONE ELEMENT PER PLY IN THICKNESS
DIRECTION, NO NEED FOR MATERIAL
UPDATE DUE TO THE CORD ANGLE
CHANGE

SEVERAL PLIES REQUIRES 3-D MATERIAL PROPERTIES,

LESS ELEMENTS, MATERIAL PROPERTIES
MUST BE UPDATED DUE TO THE CORD
ANGLE CHANGE, CANNOT MODEL THE
COUPLING BETWEEN THE BENDING
AND IN-PLANE FORCES

SHELL-TYPE ELEMENTS INCREASED MATERIAL PROPERTY INPUT,

LEAST NUMBER OF ELEMENTS,
CAN MODEL THE COUPLING BETWEEN THE
BENDING AND IN-PLANE FORCES,
KINEMATICS CONSTRAINTS OF THE
CORRESPONDING SHELL THEORY, MATERIAL
PROPERTIES MUST BE UPDATED DUE TO
THE CORD ANGLE CHANGE

I
~

I
244
SUMMARY OF MAJOR ANALYTICAL AND COMPUTATIONAL NEEDS AND FEATURES
o Large elastic strains.
o Large rotations.
o Incompressibility condition, natural rubber can be considered
as incompressible, but filled rubber exhibits compressibility.

o Nonconservative loading.
o Element type and aspect ratio.
o Self-adoptive schemes for load increments and step size.
o Contact algorithms for frictional loadings.
o Finite-element formulation in rotating coordinate system.
o Substructuring for localized analysis.
o Stresses are very erratic at regions of sudden change in
stiffness, such as cord-rubber interface. When calculated
from the finite-element displacement method, proper smoothing
algorithms must be developed for nonlinear problems.
o Sensitivity analysis for uncertain input, material properties
or other variables.

245
 I REFERENCES

I 1. Riding G. "Study of the Geometrical Structure of Multi-Ply

Yarns," Journal of Textile Institute, vol. 52, pp. 366-381,
1961.
2. Clark, S.K., "Mechanics of Pneumatic Tires," National Bureau of
Standards Monograph 122, First Edition, 1971.
3. Oblath, R.M., "Toward 21st Century - The Challenges Ahead of
Tire Reinforcements," Akron Rubber Group, Akron, Ohio,
Jan. 26, 1984.
4. Tabaddor, F. and Clark, S.K., "Linear and Nonlinear Viscoelastic
Constitutive Relations of Cord Reinforced Elastomers," Proceed-
ings of the International Symposium on Constitutive Equations
for Non-Linear Materials, University of Arizona, Tuscon,
Jan. 1983.

I
I
5. Treloar, L.R.G., "The Elasticity of a Network of Long Chain
Molecules (I and II)," Trans. Faraday Society, vol. 40, p. 59,
1944.
I
6. Rivlin, R . S . and Saunders, D . W . , "Large Elastic Deformations of
Isotropic Materials - VII. Experiments on the Deformation of
Rubber," Phil. Trans. Royal Society, vol. A243, p. 251, 1948.
I
7. Valanis, K.C. and Landel, R.F., "The Strain Energy Function of a
Hyper-Elastic Material in Terms of the Extension Ratios,"
Journal of Applied Physics, vol. 38, no. 7, pp. 2997-3002, 1967.
8. Tavazza, G. and Rimondi, G., "Stress and Strain Calculations of
Rubber Goods," Proceedings of the International Rubber Confer-
ence, Kiev USSR, 1978.
9. Rivlin, R.S. and Thomas, A.G., J Polymer Science, vol. 10,
p. 291, 1953.

I 10. Thomas, A.G., J. Polymer Science, vol. 31, p. 467, 1958.

11. Busses, W . F . , Ind. Eng. Chem., vol. 26, p. 1194, 1934.
12. Cadwell, S.M., Merril, R.A., Sloman, C.M. and Yost, F.L., Ind.
Eng. Chem. Anal., vol. 12(1), p. 19, 1940.
13. Beatty, J.R., Rubber Chemistry and Technology, no. 5,
pp. 1341-1346, Dec. 1964.
14. Beatty, J.R., J. of Elastomers and Plastics, vol. 11, p . 147,
April 1974.
I 15. Hashin, Z., "Analysis of Composite Materials--A Survey," Journal
of Applied Mechanics, vol. 50, September 1983.

246
16. Tabaddor, F., "Mechanical Properties of Cord-Rubber Composites,"
Composite Structures, vol. 3 , pp. 3 3 - 5 3 , 1 9 8 5 .

247
 N89=24650
NONLINEAR HIERARCHICAL SUBSTIIUCTURAL PARALLELISM

AND COMPUTER ARCHITECTURE

Joe Padovan
The University of Akron
Akron, Ohio
Abstract

This paper investigates computer architecture in conjunction with the

algorithmic structures of nonlinear finite-element analysis. To help set the stage
for this goal, the development is undertaken by considering the wide-ranging needs
associated with the analysis of rolling tires which possess the full range of
kinematic, material and boundary condition induced nonlinearity in addition to gross
and local cord-matrix material properties.

1. Introduction
-
With the advent of the finite-element method (FEM), the analysis of large-scale
structure is finally possible. While large-scale linear finite-element simulations
are relatively economical, such is not the case for nonlinear situations involving
geometric, material and boundary induced nonl inearit~l-~. There are numerous
aerospace and commercial structures which require full-scale nonlinear analysis to
enable their improved design. This includes such structural systems as gas turbines,
space structures, aircraft structure, autos, etc. Perhaps the most commonplace of
such structures is the tire, which serves as a component to a wide variety of
aerospace and auto systems.

To bypass the difficulties associated with nonlinear FE analysis, significant

work has been channeled into two main areas, namely:

i) The development of algorithmic improvements, element-element, 5 constrained

7
Newton/Raphson (NR) , and hierarchical least squares,

ii) The design of new computer architecture enabling hardware speedup, i.e., as in
vector processors (Cray, Cyber 205 and true parallel machines8s9)

In the context of such thrusts, not enough e€fort has been undertaken to
consider how algorithmic structures might effect machine architecture or vice versa.

Based on the foregoing comments, this paper will investigate machine architec-
ture in conjunction with algorithmic structure. To achieve this goal, the develop-
ment will be undertaken by considering the wide-ranging needs associated with the
analysis of tires. This approach was taken since, as will be seen in later sections,
the needs of tire modeling embody essentially all the requirements of nonlinear
continuum mechanics, namely10

i) Material nonlinearity

ii) Inelastic behavior

249
PRECEDING PAGE BLANK NOT FILMED
 iii) Large deformation/strain kinematics

iv) Complex inertial fields

v) Nonlinear boundary conditions

vi) Microstructure

I vii) Thermomechanical response

1 viii) Solid fluid interaction

All this leads to the development of what is called hierarchical substructural

parallelism which enables bottom-up/top-down modeling. l 1 Overall a nonlinear
multilevel substructuring scheme is overviewed which enables the simplification of
the data based management (DBM) of parallel-type operators while still yielding
enhanced computational speeds as well as reducing core requirements.

In the sections that follow, detailed tire modeling discussions embody the
diversity of needs of nonlinear simulations, various types of current machine
architectures, and potentials of hierarchical substructural parallelism. Examples
that define enhanced properties will also be given.

2. - - ----of FEM Vis-;-vis

Shortcominp ---I-- -- Analysis
Tire Structural

Noting Figure 1, t h e tire possesses a very regionalized/substructural form of

construction. Overall it consists o f : 12

I i) Carcass plies, steel/glass/Kevlar cord-rubber composites

ii) Belt plies (same as above)

I
iii) Bead, bundled steel cords

iv) Thread conFiguration

v) Regionalized rubber types

vi) Belt edges, turnup plies

I 250
The operating environment consists of:

i) The tire-road interface which involves varying pavement textures,

flexibilities and resulting frictional characteristics13,14

ii) The tire-rim interface

iii) The tire-rim-suspension behavior

iv) Cornering, braking and accelerating maneuvers

v) Standing, steady/transient rolling10,13-15

vi) Obstacle/hole envelopment roll over events13,14

vii) Pressurization16,17

A s seen from Figures 2 and 3, the pressurization and subsequent loading into
standing contact can lead to large deformations and associated rotations. For
instance Table 1 illustrates comparisons of the deflection fields generated from
linear and nonlinear FE simulations.

In this context, it follows that there are several sources of response

nonlinearity, namely

i) Large deformation kinematics

ii) The road-tire-rim interfaces

iii) Bimodular behavior of cord-rubber composites in transitions from

tension to compression

iv) Thermomechanical interactions

v) Material nonlinearity

vi) Local large strain levels in various regions of the tire;

belt edges, bead region, and tread

vii) Dynamic impact interactions

Each of the foregoing sources of nonlinearity initiates different forms of response

behavior.

For instance, from a kinematics point of view, the pressurization process causes
rotations and deflections which lead to an overall stiffening of the tire.
Similarly, as with Hertizian contact problems, the tire-road interface also exhibits
hardening-type properties, namely, the hub force-deflection response is stiffening in
character as noted in Figure 4.

251
 In a d d i t i o n t o t h e f o r e g o i n g modeling d i f f i c u l t i e s , i n g e n e r a l t h e t i r e r e s p o n s e
needs t o be h a n d l e d i n s e v e r a l l e v e l s , namely

i ) Cord-matrix and r e g i o n a l i z e d r u b b e r i n t e r f a c e s

i i) Who l e co r d - r ubbe r p 1i e s / 1ami n a e

i i i ) F u l l l a m i n a t e s t r u c t u r e s , s e v e r a l p l i e s a s i n b e l t and
carcass laminates

i v ) Full (global) structure

A s one p r e c e e d s from ( i ) - ( i v ) , a "bottom-up" m o d e l i n g l l a p p r o a c h i s r e q u i r e d w h e r e i n

f i n e d e t a i l i s handled a t t h e l o w e s t l e v e l w h i l e t h e upper l e v e l models are i n -
c r e a s i n g l y c o a r s e r s o a s t o r e d u c e o v e r a l l d e g r e e s of freedom i n a g l o b a l model.
Once t h e g l o b a l - l e v e l model i s s o l v e d what i s needed i s a "top-down" scheme'' t o
p r o v i d e p r o p e r mechanics i n f o r m a t i o n a t t h e c o n s t i t u e n t l e v e l . Such a n a p p r o a c h i s
n e c e s s a r y i f p r o p e r stress and s t r a i n € i e l d s a r e t o be c a p t u r e d hence e n a b l i n g p r o p e r
d e s c r i p t i o n of i n t e r n a l . f i e l d s .

C u r r e n t FE models of t i r e s s t a r t from l e v e l ( i i i ) and p r o c e e d t o ( i v ) . In this

way, a t r u e l o c a l - l e v e l d e s c r i p t i o n of mechanical f i e l d s i s n o t p o s s i b l e .

3. Types of P-----
arallelism

M u l t i p r o c e s s o r computers f a l l b a s i c a l l y i n t o two main c a t e g o r i e s , namely

i ) V e c t o r p r o c e s s o r s ( C r a y , Cyber 205)

i i ) True p a r a l l e l p r o c e s s o r s ( F l e x , Goodyear)

Compared w i t h s i n g l e p r o c e s s o r u n i t s (IBM 3 0 8 4 , CDC7600), v e c t o r p r o c e s s o r s

e n a b l e q u i c k e r more e E f i c i e n t h a n d l i n g of m a t r i x m a n i p u l a t i o n s . T h i s i s a c h i e v e d
t h r o u g h t h e u s e of m u l t i p l e p r o c e s s o r s which o p e r a t e s i m u l t a n e o u s l y on a s u c c e s s i o n
oE m a t r i x e l e m e n t s . Data t r a n s f e r f o r s u c h o p e r a t i o n s i s t y p i c a l l y from a s i n g l e
common c o r e s t o r a g e .

I n t r u e p a r a l l e l p r o c e s s o r s , d i f f e r e n t f u n c t i o n s / o p e r a t i o n s a r e performed i n
s e p a r a t e p r o c e s s o r s . I n s u c h machines d a t a t r a n s f e r u s u a l l y i n v o l v e s b o t h a common
c o r e as w e l l as i n d i v i d u a l . l o c a l p r o c e s s o r c o r e s . F o r s u c h machines v e r y h i g h s p e e d s
c a n be r e a l i z e d .

I n t h e c o n t e x t of programming l a n g u a g e s , v e c t o r p r o c e s s o r s t y p i c a l l y can be
programmed i n enhanced v e r s i o n s of FORTRAN o r t h e l i k e . For t r u e p a r a l l e l p r o c e s -
s o r s , o v e r a l l programming is g e n e r a l l y a c h i e v e d a t two l e v e l s . A t t h e l o c a l pro-
c e s s o r l e v e l , l a n g u a g e s s u c h as FORTRAN can be employed. A t t h e t o t a l s y s t e m l e v e l ,
machine c o n t r o l l a n g u a g e MCL i s u s u a l l y employed.

252
4. Classical
- -Solution Algorithm
The solution of large-scale FE simulations typically involves either some
variant of the NewtonjRaphson scheme NR, or an explicitlimplicit time integration
procedure. For the current demonstration purposes, the presentation will concentrate
on static equation solvers. The most recent improvements € o r such problems fall into
several categories, namely

i) Element-by-element preconditioners (Hughes et a ~ ~ )

ii) Constrained NR procedures of Padovan and Arechaga6

iii) Constrained hierarchical least-squares algorithms of

Padovan and Lackney7

Assuming large deformation kinematics along with potential material

nonlinearity, the governing FE formulation takes the form697

where S_ is the second Piola Kirchoff stress tensor, F, is the nodal force vector
and G is the vector of body forces. Typically (1) is nonlinear and must be solved
via N< schemes. After expansion into truncated Taylor series, ( 1 ) yields the
following NR algorithm

;k T
AG- + - - - .f[Bi] Sidv
[Ki]AYi+l = F+AF -
R

where [K] defines the tangent stiffness matrix, that is697

such that [Si] i s the prestress matrix and [DTi] is the tangent material stiEEness.
A s noted earlier, the solution involves either the use of constrained procedures6 for
1
appropriate load increment control or a direct Gaussian-type inversion scheme.

To date such methodologies have been employed either in single processor or

vector processor machines. The shortcomings of the FEM outlined in the previous
section are essentially a direct outgrowth of the limitations of the architecture of
single and vector-processor-type machines. In the next section, the intrinsic
structure of the INR algorithm will be explored to define new computer architectures
to bypass such difficulties.

253
 5. Hierarchical Substructurint?

From a conceptual point of view, the I N R scheme defined by ( 2 ) does not confine
the FEM scheme to a particular type of computer configuration. Rather the problems
of speed and storage are essentially hardware based. Specifically the main questions
and problems evolve out of the need to define architectures which enable the use of
multiple processors so as to enhance overall machine speed as well as memory size.
While the CRAY and CYBER systems are certainly a step in the right direction, they
fall short of the ultimate requirements. Currently very large-scale FE models can
easily outstrip the available core storage and machine CPU speeds.

In seeking to develop new computer architectures one is faced with the fact that

i) Vector processors require extensive cores as well as complex logic flows

ii) True parallel processors still await the fruition of properly organized DBM

Based on the foregoing, this paper seeks to develop what is called a hierarch€cal
form of substructural parallelism. Following the pioneering efforts of the NASA
Langley groupgy9 specifically, a nonlinear FE simulation, say of the tire, can be
logically divided into a hierarchy of substructural groups defined by a variety of
attributes, namely

i) Material group

ii) Geometric conEiguration

iii) Kinematic behavior

I iv) Boundary conditions

At the lowest rung of the hierarchy, items (i)-(iv) are employed to define the
specific local level substructural groups. The choice of the number of Eirst-order
groups is contingent on:

i) Minimizing core requirements of local level processors

ii) Minimizing number of perimeter nodes so that higher order substructural

groups also have reduced core requirements for associated processors.

A s can be seen, the main thrust is to maintain in core solutions for each local
substructural processor.

Noting Figure 5, a given FE simulation can be broken up into a number of sub-

structural levels. At each level internal nodes are eliminated to enable assembly
through perimeter nodes. In terms of (21, the NR algorithm and its constrained
counterpart can be substructured to yield the following first-level algorithms, that
is :

254
 k = 1,2,....Number of first-level substructure such that

where ( )ks'() d notes the first level kth substructure, ( li+l the (i+llth
iteration, ('
@-i+l 'ke the nodal load increment, [Ki] the substructural tangent

stiffness, A:i+l (lSk) the nodal deflection increment and E!:ik) the body force
increment .
To enable assembly into second-order substructural groups, ( 4 ) is partitioned
into internal and perimeter nodes yielding

Employing (7)-(10) we obtain the following relationships for the inner and perimeter I
nodes
, where

I Assembling (11) yields the second-level substructural relationships, namely

k = 1,2,.. . Number of second level substructure

By partitioning (16) into inner and perimeter degrees of freedom we yield the third-
order substructural relations after the appropriate assembly process. Continuing the
partitioning and assembly process yields the various higher order substructural
relations specifically

wherein the associated inner and perimeter partitions take the form

256
such that

Based on ( 1 1 ) - ( 2 4 ) , we see that the overall nonlinear hierarchical substructuring

requires a forward calculation phase as well as a backward stage. The forward phase
involves the use of (111, (131, ( 1 4 1 , ( 1 9 ) , ( 2 1 ) and (22). In contrast the backward
phase, which involves the definition of inner nodes, incorporates the use of ( 1 2 ) ,
(15), (16), (20), ( 2 3 ) and ( 2 4 ) . In terms of the forward iterative algorithms, the
overall required machine architecture takes the form defined in Figure 6 . Note the
common data buses linking successive substructural levels need only provide access to
perimeter data. In this way, significantly less data need to be accessed by the
global-level DBM. This applies throughout the forward phase of the iteration
process. Overall the steps handled by each of the succeeding levels involve
assembly, inner/perimeter partitioning, and setting up effective stiffnesses €or the
forward and backward phases. In terms of ( 2 1 ) - ( 2 4 ) , the stiffnesses associated wi h
the perimeter and inner nodes involve an inverse of the inner partition of the ktF;
substructural stiffness. All such manipulations must be performed by processors
dedicated to each of the k individual substructures associated with the various
hierarchical levels.

Once the forward loop of calculations is complete, the perimeter data must be
back tracked to the inner nodes o f each of the various substructures at the different
substructural levels. The overall flow of control/calculation is depicted in
Figure 7. A s can be seen, the perimeter data are used to determine the inner nodal
incremental excursions. This is achieved through the use of the family of expres-
sions defined by ( 2 0 ) . Once the back substitutions to the succeeding levels up to
and including the first are completed, the standard norm type convergence checks must
be implemented to ascertain the quality of convergence. Contingent on the conver-
gence check, the iteration process can be cycled through the forward and backward
phases of the substructural hierarchy.

257
6. Discussion

To illustrate the hierarchical substructural scheme, consider the three-level

simulation defined in Figure 8. The number of nodes and substructure associated with
the example are given in Figure 9. Based on the number of inner and perimeter
variables depicted, the expressions defining the number of respective nodes are given
by :

i) Level 1

Perimeter Nodes = 2(R1 + $2 + 23)

Inner Nodes = (21 - 2)(R2 - 2)

ii) Level 2

iii) Level 3

Perimeter Nodes = 2mlnl(Rl - 1) + 2m2n2(R2 - 1)

+ m 1n 1 + m 2n 2 - m1m2 + 1 (30)

Employing (25-30) we see that the storage effectiveness of each of the various levels
is expressed by the relations

&') = - ~
Per
. i- me-ter
~ - -
Perimeter + Inner

where k denotes the level number. In the context oE (31), it follows that

p = 2 ( t 1n1 + t2n2 - n1 - n,) (33)

n1II 1(n2 + 1) + E2n2(nl + 1) - 3n1n2 - 4(nl + n2) + 1

2mlnl(Ql - 1)' t 2m2n2(I12 - 1)

&3) = (34)
L m n,(m2 + 1) + II m n (m + 1) -mlm2(n1 + n2 + 1) - mlnl - m2n2 + 1
1 1 , 2 2 2 1

258
 Consider the case wherein

El = 100, L2 = 50

n =5, n2=4
1

m1 = 3 , m2 = 4

In terms of the foregoing, Table 2 gives the total number of

0 degrees of freedom

0 processors required at each level

perimeter/inner nodes

as well as the storage effectiveness of each of the substructural leve s . Noting

that a straight solution of the given problem would require a 1.2 x 10!t order
stiffness matrix, it follows from Table 2 that very significant storage savings as
well as speed enhancements can be achieved.

In the context of the foregong development, it follows that hierarchical sub-

structural parallelism has decided advantages over vector-type processors, namely:

i) Global common core is reduced in size

ii) Substructures are handled in smaller local cores which could employ
vector processors and which are controlled by local DBM

iii) Data transfer between succeeding levels of substructural hierarchy

are reduced thereby reducing load on DBM

iv) Various substructures are updated, inverted, and assembled

simultaneously hence emhancing the overall speed

v) The overall addressing requirements are reduced since the size of

individual substructural zones is much smaller

iv) Extensive use of cash memory (Ram Disk) can be made at the local
level thereby reducing disk 1/0

vii) Backward and forward steps follow natural formulational lines

viii) Element-to-element or hierarchical least-squares algorithms can

be employed at the local substructural level

ix) Linear/nonlinear problem partitioning can be more logically handled

x) Overall control of the machine is more logical and less difficult

since local processors are essentially autonomous within updating
and inverting phases of the operation

259
 xi) The MCL can be patterned about well-defined substructuring methodology; the
transfer of control from level to level is contingent on the
monitoring status of stiffness/inversion calculations

xii) The data base manager needs only to deal with data residing on perimeters
of the substructure; as noted earlier, this significantly reduces the amount
of data transferred between levels.

A s discussed earlier, the modeling of tires in their use environment represents

perhaps one of the most comprehensive single component nonlinear structural response
problems currently available. This follows from the fact that geometric-, material-,
and boundary-induced nonlinearity all simultaneously act to define the global
response behavior. Due to their regionalized/ substructural form of construction,
tires represent a good modeling problem to help define the architecture of high-level
multiprocessor machines. In this context, a hierarchical form of substructural
parallelism has decided advantages over other forms o f multiprocessors. A s has been
seen such a procedure has several theoretical advantages for nonlinear problems.
These evolve about the simplified DBM structure, reduced data flow, smaller global
core, and reduced addressing requirements.

Overall Euture work in this area should

0 Place qreater emphasis on algorithmic architecture and its possible effects on

machine structure

0 Establish proper control configuration f o r hierarchical DBM

Extend scheme to constrained incremental Newton/Raphson (INR) least-squares

algorithms as well as transient schemes

0 Apply concept to available parallel processors

0 Structure procedure so as to enable either direct or iterative solutions at

substructural level

0 Establish criteria to enable determination of quality of convergence at

local substructural level

260
 6. References
~

I 1. Zienkiewicz, O.C., The Finite Element Method, McGraw Hill Company (1983).

2. Bathe, K.J., Finite Element Procedures in Engineering Analysis, Prentice Hall

(1983).

3. Oden, J.T., Finite Elements of Nonlinear Continuum,-McGraw Hill Co. (1972).

4. Cook, R.D., Concepts and Applications of Finite Element Analysis, Wiley (1983).

5. Hughes, T.J.R., Winget, J. and Levit, I., Comp. Meth. Appl. Mech. Engineering,
36, 241 (1983).

6. Padovan, J. and Arechaga, T., Int. Jr. Engineering Science, 20, 1077 (1982).

7. Padovan, J. and Lackney, .J., Computers and Structures, 19, 535 (1984).

8. Noor, A.K., Storaasli, 0.0. and Fulton, R.E., ASME Special Publ., H00275, 1
(1983).

9. Bostic, S.W. and Fulton, R.D., AIAA Paper No. 85-0783-CP (1985).

10. Padovan, J. and Tovichakchaikul, S . , Computers and Structures, 18, 191 (1984).

11. Padovan, J., Global-Local Finite Element Modeling, 126th Meeting of the ACS
Rubber Division (1984).

12. Clark, S., Mechanics of Pneumatic Tires, U.S. Department o f Transportation

(1980).

13. Padovan, J. and Nakajima, Y., Numerical Simulation of Tire Skidding Events
Involving Impacts with Holes and Bumps, 4th Annual Tire Society Meeting (1985).

14. Padovan, J. and Nakajima, Y., Numerical Simulation of Tire Transiently

Sliding/Rolling Over Holes and Bumps, ACS Meeting (April 1985).
15. Padovan, J. and Paramadilok, O., Transient and Steady State Viscoelastic Rolling
Contact, Computers and Structures, 20, 545-553 (1985).

16. Trinko, J.J., Ply and Rubber Stresses and Contact Forces for a Loaded Radial
Tire, First Annual Meeting of Tire Society, Akron (1982).

17. Tabaddor, F. and Stafford, J., Computers and Structures, 13, 737 (1981).

261
 TABLE 1 COMPARISON OF LINEAR AND NONLINEAR FE SIMULATION OF PRESSURIZED TIRE

PRESSURE LINEAR ( I N ) VONLINEAR ( I N )

(PSI 1 MAX I Y U Y DEFLECT I ON MAXIMUM DEFLECT I O N

CROWN SIDEWALL CROWN SIDEWALL

5. .003 ,036 003

10. ,006 ,073 006 ,047

15. ,009 .i n 9 009 064

30. . 01 8 21 8 01 8 ,107

TABLE 2 COMPARISON OF HIERARCHICAL SUBSTRUCTURAL PARALLEL AND SINGLE

PROCESSOR SYSTEMS

LEVEL

GLOBAL

*D o f F/P - deqrees o f freedom per processor

**Total nuniher o f p e r i m e t e r a n d i n n e r 0 o f F

262
 T R E A D STOCK
/

I B E ~ . D*BE40 FILLER

F i g u r e 1 T i r e geometry and c r o s s s e c t i o n .

INITIAL SHAPE
- - - EX PER1MEN TAL
-.-. FINITE ELEMENT

Figure 2 P r e s s u r e t i r e profile.

263
 Figure 3 Loading into standing contact.

n spring rate - 2 O p s i
500 --
rl
v Predicted

Experimental

-4 -8
Radial Deflection (in-)

Figure 4 Force deflection characteristics into standing contact.

2 64
 @ Steel belt
@ Textile carcass
@ Inner liner
@ Belt edge cushion
@ Side wall
@ Bead filler
@) Textile finishing
@ Bead

Figure 5 Substructural zones of tire.

COMMON DATA BUS

1 MAIN 1 / 0 BUS
-----------
LOCAL STIFFNESS UPDATE

LEVEL 1 1 k ... SUBSTRUCTURE ASSEMBLED

PERIMETER-INNER PARTI-
TIONING

-- ---_-___ -
COMMON DATA BUS
-_---_-----
PERIMETER DATA 1/0

SUBSTRUCTURE ASSEMBLED
PERIMETER-INNER PARTI-
TIONING

COMMON
..
DATA BUS
I
_----------

1
COMMON DATA BUS PERIMETER DATA 1 / 0
__---------
GLOBAL SUBSTRUCTURE
ASSEMBLED A N D SOLVED
LEVEL j

Figure 6 Flow of control: forward loop.

265
 --------____
COMMON DATA BlJS
- M- A-I -N -1-/ 0- _B U_S _ _
INNER VARIABLES CALC.

ALL F I E L D VARIABLES

UPDATED

-- -------_ -

LEVEL
L COMMON DATA BUS PERIMETER
__-_-------
INNER VARIABLES
DATA 1/0

CALC.

COMMON

.
DATA BUS
J
e

I
----_------
I COMMON DATA BUS
i PERIMETER DATA
---_---___-
I / O

INNER VARIABLES CALC.

LEVEL j
I 3
Figure 7 Flow of control: backward loop.

-
-
LEVEL

LEVEL 2
1 (LOCAL)

<INTERMEDIATE)

’/
Figure 8 Example oE three-level hierarchical substructural system.

266
 LEVEL 1

LEVEL 2

LEVEL

Figure 9 Number of perimeter substructures at each level.

267
 ADAPTIVE METHODS, ROLLING CONTACT,
AND NONCLASSICAL F R I C T I O N LAWS

J. T. Oden
The U n i v e r s i t y of Texas a t A u s t i n
A u s t i n , Texas

ABSTRACT

R e s u l t s and methods on t h r e e d i f € e r e n t areas oE contemporary r e s e a r c h a r e

o u t l i n e d . These i n c l u d e a d a p t i v e methods, t h e r o l l i n g c o n t a c t problem f o r
f i n i t e d e f o r m a t i o n of a h y p e r e l a s t i c o r v i s c o e l a s t i c c y l i n d e r , and non-
c l a s s i c a l f r i c t i o n laws f o r modeling dynamic f r i c t i o n phenomena.

PRECEDING PAGE RANK NOT FILMED 269

 1. INTRODUCTION

T h i s p a p e r a d d r e s s e s t h r e e s u b j e c t s t h a t impact on t h e computer s i m u l a -
t i o n of n o n l i n e a r t i r e b e h a v i o r : a d a p t i v e methods, which r e p r e s e n t schemes
f o r a s s e s s i n g n u m e r i c a l e r r o r and a u t o m a t i c a l l y a d a p t i n g t h e mesh s o as t o
improve a c c u r a c y ; t h e r o l l i n g c o n t a c t problem, which i s a t t h e h e a r t of t i r e
a n a l y s i s ; and new f r i c t i o n laws, which a r e e s s e n t i a l i n d e v e l o p i n g r e a l i s t i c
models of f r i c t i o n a l c o n t a c t . Space l i m i t a t i o n s p r e c l u d e a d e t a i l e d d i s c u s -
s i o n of t h e s e i s s u e s ; b u t f u r t h e r d e t a i l s c a n be found i n r e c e n t p a p e r s and
r e p o r t s by t h e a u t h o r and h i s c o l l e a g u e s [ l - 1 7 1 .

2. ROLLING CONTACT

The g e n e r a l r o l l i n g c o n t a c t problem a s a b a s i s f o r n o n l i n e a r t i r e a n a l -
y s i s i n v o l v e s some of t h e most c h a l l e n g i n g and d i f f i c u l t problems i n s t r u c -
t u r a l mechanics. Among t h e c o m p l i c a t i n g f e a t u r e s are t h e p r e s e n c e of
u n i l a t e r a l c o n t a c t , f r i c t i o n , i n e r t i a e f f e c t s , multi-parameter b i f u r c a t i o n s ,
t h e emergence of s t a n d i n g waves, v i s c o e l a s t i c and t h e r m a l e f f e c t s , large
d e f o r m a t i o n s , t h e n e c e s s i t y of modeling of complex materials s u c h as f i b e r -
r e i n f o r c e d r u b b e r s , and t h e p r e s e n c e of n o n - c o n s e r v a t i v e p r e s s u r i z a t i o n
l o a d i n g s . A f i r s t s t e p toward r e s o l v i n g s u c h i s s u e s i s t h e f o r m u l a t i o n of
c o r r e c t k i n e m a t i c s and v a r i a t i o n a l p r i n c i p l e s f o r a s p e c i a l case: t h e s t e a d y -
s t a t e r o l l i n g of a h y p e r e l a s t i c o r v i s c o e l a s t i c c y l i n d e r i n c o n t a c t w i t h a
r i g i d f o u n d a t i o n and i n a s t a t e of p l a n e s t r a i n .

The k i n e m a t i c a l s i t u a t i o n is shown i n F i g . 1 where t h e geometry of t h e

r e f e r e n c e c o n f i g u r a t i o n ( a r i g i d s p i n n i n g c y l i n d e r w i t h no c o n t a c t ) is com-
p a r e d t o t h e geometry of a deformed c y l i n d e r i n s t e a d y - s t a t e r o l l i n g c o n t a c t
w i t h a rough ( f r i c t i o n a l ) roadbed ( f o u n d a t i o n ) .

Key f e a t u r e s of t h e k i n e m a t i c d e s c r i p t i o n a r e l i s t e d as f o l l o w s :

1) T i m e a p p e a r s only i m p l i c i t l y i n t h e f o r m u l a t i o n ; i f (R,@, Z,) are

material c o o r d i n a t e s , t h e r e f e r e n t i a l c o o r d i n a t e s a r e

r = R, 8 = @ + u t , z = Z

where w i s t h e a n g u l a r v e l o c i t y of t h e r i g i d , r e f e r e n c e c y l i n d e r .

2) I f t h e motion i s d e f i n e d by t h e map

xi = xi (r, e, z) i = 1, 2 , 3

where xi = s p a t i a l C a r t e s i a n c o o r d i n a t e s of p a r t i c l e s i n t h e c u r r e n t c o n f i -
g u r a t i o n , t h e n v e l o c i t y and a c c e l e r a t i o n components are

2
v =w-
axi a = u 2- a- X i
i ae i 2
ae

270
 3) The u n i l a t e r a l c o n t a c t c o n s t r a i n t can be expressed i n t h e form

where
rC i s t h e e x t e r i o r c o n t a c t s u r f a c e and @ i s t h e d i s t a n c e Erom t h e
hub c e n t e r t o t h e foundation. This c o n d i t i o n can a l s o be w r i t t e n

where (.I+ d e n o t e s t h e p o s i t i v e p a r t OF (.).

4 ) The time h i s t o r y of deformation can be expressed i n terms of s t r a i n s

of p a r t i c l e s l o c a t e d on t h e same c i r c u l a r a r c i n t h e r e f e r e n c e c o n f i g u r a t i o n .
For example, i f i s t h e Green-Saint Venant s t r a i n t e n s o r , i t s time h i s t o r y
over an i n t e r n a l 0 - T - t s a t i s f i e s :

This p r o p e r t y makes i t p o s s i b l e t o i n c o r p o r a t e v i s c o e l a s t i c e f f e c t s i n t o t h e
r o l l i n g c o n t a c t problem i n a s t r a i g h t f o r w a r d way.

For i l l u s t r a t i o n purposes, we c o n s i d e r a c l a s s of r o l l i n g c o n t a c t prob-

lems i n which t h e f o l l o w i n g c o n s t i t u t i v e p r o p e r t i e s hold:

a ) The material i s e i t h e r an i s o t r o p i c h y p e r e l a s t i c material char-

a c t e r i z e d by a s t r a i n energy f u n c t i o n

(or W = W ( 1 1 ] . 1 2 ) if I3 = 1 - an i n c o m p r e s s i b l e material) in which I l , 1 2 , T 3

a r e t h e p r i n c i p a l i n v a r i a n t s of t h e deformation t e n s o r C, = zTF,, o r t h e
material i s a v i s c o e l a s t i c m a t e r i a l c h a r a c t e r i z e d by a l i n e a r v i s c o e l a s t i c
p e r t u r b a t i o n of a h y p e r e l a s t i c m a t e r i a l ; e.g.,

w i t h y a v i s c o s i t y parameter, k a m a t e r i a l k e r n e l , and S, t h e second

Piola-Kirchhoff stress t e n s o r .

271
 ___-

S ) The normal s t i f f n e s s of t h e c o n t a c t i n t e r f a c e obeys a c o n s t i t u t i v e

l a w of t h e t y p e

m
n
tn = cn(x2 - I{)+ on rc

where t, i s t h e normal stress and cn and m, a r e material c o n s t a n t s .

I f m, = 1 and cn = 1 / ~ where , E is a positive constant, t h i s r e l a t i o n
c o i n c i d e s w i t h t h e normal c o n t a c t s t r e s s a s s o c i a t e d w i t h a n e x t e r i o r p e n a l t y
a p p r o x i m a t i o n of t h e u n i l a t e r a l c o n s t r a i n t c o n d i t i o n .

c ) If the cylinder is given a prescribed velocity vo relative to

t h e roadway, t h e s l i p v e l o c i t y on t h e c o n t a c t s u r f a c e i s

wT - vo - i, = v
0
- wagX

5) The f r i c t i o n l a w i s ( w i t h the f r i c t i o n a l s t r e s s )
LT

t I <
I -T llltnl = > WT = 0

I -Tt I = pJtnJ = > WT = -XI

V a r i a t i o n a l p r i n c i p l e s f o r var ous r o l l i n g c o n t a c t problems are summa-

r i z e d i n F i g s . 2-8, b e g i n n i n g w i t h t h e p u r e s p i n n i n g of a c y l i n d e r w i t h o u t
c o n t a c t and p r o g r e s s i n g t o t h e g e n e r a l v a r i a t i o n a l i n e q u a l i t y f o r f i n i t e de-
f o r m a t i o n r o l l i n g c o n t a c t w i t h f r i c t i o n . V a r i o u s s p a c e s of a d m i s s i b l e func-
t i o n s a r e d e f i n e d i n t h e s e f i g u r e s as w e l l a s s e v e r a l n o n l i n e a r forms. I n
particular,

A(x, n) = t h e i n t e r n a l v i r t u a l work produced by t h e P i o l a - K i r c h h o f f

N N
stresses To

B(x, n) = t h e v i r t u a l work produced by i n e r t i a ( r a d i a l a c c e l e r a t i o n )

N N
e f f e c t s p e r u n i t of a n g u l a r v e l o c i t y v e l o c i t y w

C(x, 0) = work term due t o normal compliance of t h e i n t e r f a c e

N N

I(x, 5,
N N
n)
N
= a v i r t u a l work term r e p r e s e n t i n g t h e work done by t h e
hydrostatic pressure p ( p r e s e n t when t h e material i s
incompressible)

j ( x , rl) = t h e v i r t u a l work term due t o f r i c t i o n a l forces

N N

f(n> N
= t h e v i r t u a l work due t o e x t e r n a l f o r c e s

272
 A f i n i t e - e l e m e n t code h a s been d e v e l o p e d based on t h i s g e n e r a l v a r i a -
t i o n a l p r i n c i p l e which has t h e f o l l o w i n g € e a t u r e s :

1. B i q u a d r a t i c (Q2) e l e m e n t s a r e used t o a p p r o x i m a t e t h e d i s p l a c e m e n t
f i e l d a n d , f o r i n c o m p r e s s i b l e materials, e l , d i s c o n t i n u o u s l i n e a r e l e m e n t s are
used t o approximate h y d r o s t a t i c p r e s s u r e s

I 2. The f r i c t i o n a l f u n c t i o n s are r e g u l a r i z e d i n a s t a n d a r d way

3. A R i k s - C r i s f i e l d method w i t h Newton-Raphson c o r r e c t i o n s i s used t o

~

s o l v e t h e n o n l i n e a r s y s t e m s of e q u a t i o n s c h a r a c t e r i z i n g t h e d i s c r e t e problem

To d a t e , a n e x t e n s i v e s e t of n u m e r i c a l s o l u t i o n s has been o b t a i n e d u s i n g
t h e s e c o n c e p t s and methods. Here o n l y one example i s c i t e d , which i s i n t e r -
e s t i n g b e c a u s e of t h e s l o w emergence of s t a n d i n g waves as t h e a n g u l a r v e l o c i t y
i s i n c r e a s e d f o r a f i x e d p e n e t r a t i o n H of a hollow r u b b e r c y l i n d e r i n t o a
rough roadway w i t h c o e f f i c i e n t of f r i c t L o n u = 0.03. Computed deformed
s h a p e s and stress c o n t o u r s are shown i n F i g s . 9-13 f o r v a r i o u s v a l u e s of w.
We n o t i c e t h e s t e a d y development of more-or-less p e r i o d i c wavelets on t h e ex-
t e r i o r s u r f a c e which meet a t p o i n t s a t which s i n g u l a r i t i e s a p p e a r t o be
f o r m i n g . The p r e s e n c e of E r i c t i o n on t h e c o n t a c t s u r f a c e d e s t r o y s t h e symme-
t r y of t h i s w a v e l e t p a t t e r n . Mild v i s c o e l a s t i c e f f e c t s , s u c h as t h o s e i n
r u b b e r s a t moderate t e m p e r a t u r e s , do n o t a p p r e c i a b l y a l t e r t h e s t r u c t u r e of
t h e s e deformed g e o m e t r i e s .

The g e n e r a l i t y of t h e Formulation and o f t h e methods employed h e r e makes

i t p o s s i b l e t o s t u d y n u m e r i c a l l y a wide range of r o l l i n g c o n t a c t problems.
F u r t h e r work s h a l l i n v o l v s g e n e r a l i z a t i o n s o f t h e s e r e s u l t s t o t h r e e -
d i m e n s i o n a l r o l l i n g c o n t a c t problems which i n c l u d e t h e e f f e c t s o € t u r n i n g ,
s t e e r i n g f o r c e s , and t i l t i n g r e l a t i v e t o t h e roadway p l a n e .

3. ADAPTIVE METHODS

We s h a l l now t u r n t o t h e i m p o r t a n t s u b j e c t of a d a p t i v e f i n i t e - e l e m e n t
methods. Adaptive methods s h o u l d have a s i g n i f i c a n t impact on n o t o n l y t i r e
analysis but also on general computational structural mechanics in the rela-
t i v e l y near future.

I n g e n e r a l , a d a p t i v e methods s e e k t o change t h e s t r u c t u r e of a n a p p r o x i -
mate method t o improve t h e q u a l i t y of t h e s o l u t i o n . By s t r u c t u r e , w e mean t h e
mesh d e n s i t y , l o c a t i o n s of nodes, and o r d e r of t h e l o c a l p o l y n o m i a l s . By
q u a l i t y of a n a p p r o x i m a t i o n , we g e n e r a l l y mean t h e e r r o r i n a p p r o x i m a t i o n i n
some a p p r o p r i a t e norm. There a r e , t h u s , two p r i m a r y a s p e c t s of any a d a p t i v e
E i n i t e - e lement method :

1) The e s t i m a t i o n of t h e e r r o r

2) The a d a p t i v e s t r a t e g y

I n t h e f i r s t o € t h e s e , i t i s assumed t h a t a n i n i t i a l a p p r o x i m a t i o n of t h e
s o l u t i o n i s known, p e r h a p s from a c o m p u t a t i o n on a c o a r s e mesh, and t h a t t h i s
r o u g h s o l u t i o n c a n be u s e d t o o b t a i n a n a p o s t e r i o r i e s t i m a t e of t h e l o c a l

273
 e r r o r o v e r e a c h f i n i t e element. Once a n estimate oE t h e l o c a l e r r o r i s known,
one must c a l l upon some t e c h n i q u e t o r e d u c e t h e l o c a l e r r o r and t h e r e b y i m -
p r o v e t h e q u a l i t y of t h e s o l u t i o n .

T h e r e a r e two g e n e r a l t y p e s of methods we have s t u d i e d f o r a p o s t e r i o r i

e r r o r e s t i m a t i o n of t h e l o c a l e r r o r o v e r e a c h f i n i t e e l e m e n t . Once a n e s t i -
mate of t h e l o c a l e r r o r i s known, one must c a l l upon some t e c h n i q u e t o r e d u c e
t h e l o c a l e r r o r and t h e r e b y improve t h e q u a l i t y of t h e s o l u t i o n .

T h e r e are two g e n e r a l t y p e s of methods we have s t u d i e d f o r a p o s t e r i o r i

e r r o r estimation:

1. R e s i d u a l methods

2. I n t e r p o l a t i o n ( o r a p r i o r i ) methods

A s t h e name i m p l i e s , r e s i d u a l methods make u s e of element r e s i d u a l s - t h e

r e s i d u a l o r u n b a l a n c e l e f t o v e r when t h e a p p r o x i m a t e s o l u t i o n i s s u b s t i t u t e d
i n t o t h e g o v e r n i n g e q u a t i o n s and edge c o n d i t i o n s on e a c h e l e m e n t .

The r e s i d u a l i t s e l f ( e . g . , t h e e q u i l i b r i u m u n b a l a n c e i n element f o r c e s )
i s n o t n e c e s s a r i l y a good i n d i c a t i o n of l o c a l e r r o r . I n d e e d , t h e l o c a l
r e s i d u a l c a n be n e a r l y z e r o w h i l e t h e e r r o r can be q u i t e l a r g e . For t h i s
reason, i t i s g e n e r a l l y necessary t o c a l c u l a t e c e r t a i n l o c a l e r r o r i n d i c a t o r s
which bound t h e e r r o r above and below i n a p p r o p r i a t e norms. The c a l c u l a -
9,
t i o n of e r r o r i n d i c a t o r s g e n e r a l l y r e q u i r e s t h e s o l u t i o n of s p e c i a l l o c a l
problems o v e r e a c h e l e m e n t i n which t h e e l e m e n t r e s i d u a l s e n t e r as d a t a . F o r
example, i n t h e model e l l i p t i c problem,

2
-Au = f i n Q CIR

u = on an

( w i t h A t h e L a p l a c i a n and f given), the finite-element solution Uh

satisfies

I Vuh Vvh dxdy = fvhdxdy

n n

f o r a r b i t r a r y test €unction Vh, and o v e r e a c h e l e m e n t K of a mesh, the

residual is

r h = - Au,, - f

274
 Over element K, a n e r r o r i n d i c a t o r #Ik i s computed which s a t i s E i e s

f, V#k Vv dxdy = I r h vdxdy + #I -an

auh
vd s
K K aK

for v i n H1(K). One c a n show t h a t t h e e r r o r eh = u - Uh i n t h e e n e r g y norm

(1 lehl Il,Q = {I lVehl2 dxdy 11'2) s a t i s € i e s t h e bound
n

V a r i o u s r e s i d u a l methods d i f f e r i n t h e way t h e s e e r r o r i n d i c a t o r s a r e
d e f i n e d and c a l c u l a t e d . There a r e some r e s i d u a l t e c h n i q u e s which c a n produce
s h a r p l o c a l e r r o r estimates i n v i r t u a l l y any norm f o r c e r t a i n classes of prob-
lems. (See Demkowicz and Oden 1 4 , 51). These methods are n o t r e s t r i c t e d t o
l i n e a r problems and have been used t o produce e r r o r estimates i n h i g h l y non-
l i n e a r problems ( s e e [ 7 , 161).

The i n t e r p o l a t i o n methods make u s e of t h e f a c t t h a t t h e i n t e r p o l a t i o n

e r r o r o v e r an e l e m e n t K of d i a m e t e r hK o v e r a n e l e m e n t on which poly-
n o m i a l s of d e g r e e p a r e used i s

where u i s a g i v e n smooth f u n c t i o n , I$,u is its interpolant,

+ u2 + u 2 )dxdy

8
a n d C i s a c o n s t a n t independen of hK and u. The i d e a behind i n t e r p o l a -
t i o n methods is t o estimate lu12,K u s i n g r e s u l t s of a coarse-mesh approxima-
t i o n ( e . g . , u s i n g f i n i t e - d i f f e r e n c e methods o r e x t r a c t i o n methods [ 6 ] ) . The
c o n s t a n t C c a n n o t , i n g e n e r a l , be d e t e r m i n e d , s o s u c h i n t e r p o l a t i o n methods
c a n o n l y be used t o assess r e l a t i v e e r r o r i n v a r i o u s f i n i t e e l e m e n t s .

Once a n estimate of t h e e r r o r i s o b t a i n e d , t h e l o c a l e r r o r is reduced

a d a p t i v e l y u s i n g one of t h e f o l l o w i n g t e c h n i q u e s :
 1. h-methods: t h e mesh s i z e h i s reduced and t h e number of e l e m e n t s
i s i n c r e a s e d i n r e g i o n s of h i g h e r r o r .

2. p-methods: t h e mesh i s f i x e d , b u t t h e l o c a l o r d e r p of t h e
polynomial s h a p e f u n c t i o n s i s i n c r e a s e d .

3. Moving mesh methods: t h e nodes i n a f i n i t e - e l e m e n t mesh are moved

and c o n c e n t r a t e d i n areas of h i g h e r r o r .

4 . Combined methods: t h e s e i n v o l v e c o m b i n a t i o n s of t h e above t h r e e

techniques.

We have developed t e s t codes which employ a l l of t h e s e methods. The

r e s u l t s of some tests a r e g i v e n i n F i g s . 14-20, and s p e c i f i c comments f o l l o w .

1. I n F i g . 14 we see a d i s t o r t e d mesh o b t a i n e d u s i n g a moving mesh

s t r a t e g y on t h e d r i v e n c a v i t y problem f o r a n i n c o m p r e s s i b l e v i s c o u s f l u i d (see
[71).

2. F i g u r e s 15, 16, and 17 c o n t a i n computed r e s u l t s i n a d a p t i v e schemes

based on r e s i d u a l methods d e v i s e d by Demkowicz and Oden [ 4 , 51. The r e s u l t s
shown h e r e a r e f o r t r a n s i e n t h e a t c o n d u c t i o n problems w i t h dominate c o n v e c t i o n
e f f e c t s and f o r n o n l i n e a r B u r g e r s ' e q u a t i o n v e c t o r - v a l u e d s o l u t i o n s which
s i m u l a t e the Navier-Stokes e q u a t i o n s . A s p e c i a l h-method i s u s e d here which
employs a f i n e g r i d and a c o a r s e - g r i d a p p r o x i m a t i o n t o estimate e r r o r .

3. R e s u l t s oE a new p-method f o r Navier-Stokes e q u a t i o n s i n two-

d i m e n s i o n s a r e i l l u s t r a t e d i n F i g s . 18, 19 ( s e e [ l , 161). Here a r a t h e r
c o a r s e mesh i s used and e r r o r s are reduced by i n c r e a s i n g l o c a l polynomial
d e g r e e s from 1 t o 2 t o 3 Different shading i n these figures indicates d i f -
f e r e n t l e v e l s of l o c a l L i e r r o r , w i t h b l a c k c e l l s i n d i c a t i n g a n e r r o r of less
t h a n 5 p e r c e n t , g r e y a n e r r o r of less t h a n 10 p e r c e n t , and w h i t e a n e r r o r of
o v e r 20 p e r c e n t . Such l a r g e l o c a l e r r o r s a r e reduced b e f o r e t h e s o l u t i o n i s
a l l o w e d t o advance i n t i m e . Remarkably, t h e s o - c a l l e d e f f e c t i v i t y i n d e x 0
f o r t h i s proljlem (which r e p r e s e n t t h e r a t i o of t h e e s t i m a t e d l o c a l e r r o r t o
t h e a c t u a l l o c a l e r r o r ) f o r a n L' - norm v a r i e d from around 1.001 t o 0.860
f o r t h e time r a n g e s c o n s i d e r e d i n a t e s t example. T h i s s u g g e s t s t h a t
r e s i d u a l - t y p e e r r o r estimates based on p-type s t r a t e g i e s can be v e r y a c c u r a t e ,
e v e n f o r t r a n s i e n t n o n l i n e a r problems on c o a r s e meshes.

F i g u r e 20 shows r e f i n e d mesh p a t t e r n s f o r a c l a s s of l i n e a r e l l i p t i c
problems i n which a v e r y f a s t v e c t o r i z a b l e h-method i s used i n c o n j u n c t i o n
w i t h a n i n t e r p o l a t i o n - t y p e e r r o r e s t i m a t o r ( s e e [ 6 1 ) . One i n t e r e s t i n g a s p e c t
of t h i s work, i n d i c a t e d by d i f f e r e n t s h a d i n g s of e l e m e n t s i n t h e f i g u r e , i s
t h a t t h e d i s t i n c t i o n between "optimal" meshes d e t e r m i n e d u s i n g a v e r y s o p h i s -
t i c a t e d e r r o r e s t i m a t o r ( s e e [171) and v e r y c r u d e estimates (C61) i s n e g l i g i -
b l e whenever s t r o n g s i n g u l a r i t i e s a r e p r e s e n t .

4. NON-CLASSICAL FRICTION LAWS

I n o u r most r e c e n t c a l c u l a t i o n s of r o l l i n g c o n t a c t , we have employed

s p e c i a l i n t e r f a c e c o n s t i t u t i v e e q u a t i o n s f o r t h e normal compliance OF t h e

276
i n t e r f a c e and t h e t a n g e n t i a l . f r i c t i o n a l f o r c e s . Some of t h e s e laws a r e men-
t i o n e d i n S e c t i o n 2 of t h i s paper ( s e e a l s o Fig. 6 ) . The p h y s i c a l i n t e r p r e t a -
t i o n and t h e m o t i v a t i o n of such models are d i s c u s s e d i n r e f e r e n c e s [14, 15, and
181.

ACKNOWLEDGMENT

The work r e p o r t e d h e r e was supported by s e v e r a l a g e n c i e s : t h e work

on a d a p t i v e methods by The O f f i c e of Naval Research under Contract
N00014-84-K-0409; t h e work on r o l l i n g c o n t a c t by Contract NAS1-17359 from t h e
NASA Langley Research Center, and work on n o n - c l a s s i c a l f r i c t i o n by t h e A i r
Force O f f i c e of S c i e n t i f i c Research under Contract F49620-84-0024.

277
 REFERENCES

1. DEMKOIJICZ, L . , ODEN, J . T . , STROUBOULIS, T. and CAMPOS L., V a r i a t i o n a l

P r i n c i p l e s and A d a p t i v e Schemes f o r a Class of Flow Problems w i t h Moving
B o u n d a r i e s , F i n i t e Elements
----I--- -------
i n Flow Problems, Vol. V I , John Wiley and
S o n s , L t d . , London, 1985.

2. ODEN, J . T . , and DEMKOWICZ, L., A P o s t e r i o r i E r r o r Estimates and

A d a p t i v i t y i n F i n i t e Element Methods, I n t e g r a l Methods i n S c i e n c e and
_____l-_l

E n g i n e e r i n g , Hemisphere P u b l i s h e r s , W a s h i n g t i o n , 1986.

3. ODEN, J.T., and DEMKOWICZ, L . , A d a p t i v e F i n i t e Element Methods € o r Com-

p l e x Problems i n S o l i d and F l u i d Mechanics, I n t e r n a t i o n a l Conference on
F i n i t e Elements i n Computational Mechanics, 2-6 December, 1985, Bombay
India.

4. DEMKOWICZ, J . , and ODEN, J . T . , An A d a p t i v e C h a r a c t e r i s t i c P e t r o v - G a l e r k i n

F i n i t e Element Method € o r Convection-Dominated L i n e a r and N o n l i n e a r
P a r a b o l i c Problems i n One Space V a r i a b l e s , Texas I n s t i t u t e For Computa-
t i o n a l Mechnics, TICOM R e p- o r t 85-3, A u s t i n , TX, J u n e , 1985.

5. DEMKOWICZ, L., and ODEN, J . T . , An A d a p t i v e C h a r a c t e r i s t i c P e t r o v - G a l e r k i n

F i n i t e Element Method f o r Convection-Dominated L i n e a r and N o n l i n e a r
P a r a b o l i c Problems i n Two Space V a r i a b l e s , TICOM R e p o r t 85-7, 1985.

6. DEMKOWICZ, L . , DEVLOO, P.Y., and ODEN, J . T . , On a n H-Type Mesh Refinement

S t r a t e g y Based on M i n i m i z a t i o n of I n t e r p o l a t i o n E r r o r s , Computer Methods
i n Applied Mechanics and E n g i n e e r i n g . Vol. 53, pp. 67-89, 1985.

7. DEMKOWICZ, L., and ODEN J . T . , On a Mesh O p t i m i z a t i o n Method Based on

M i n i m i z a t i o n of I n t e r p o l a t i o n E r r o r , L n t . J. Eng. S c i . , Vol. 24, No. 1,
pp. 55-68, 1986s.

8. ODEN J . T . , DEMKOWICZ, L . , and STROUBOULIS, T., A d a p t i v e F i n i t e Element

For Flow Problems With Moving B o u n d a r i e s . I. V a r i a t i o n a l P r i n c i p l e s and
A P o s t e r i o r i Estimates, Computer Methods i n Applied Mechanics and
E n g i n e e r i x , Vol. 46, pp. 217-251, 1984.

9. DEMKOWICZ, L., and ODEN J . T . , An A d a p t i v e C h a r a c t e r i s t i c P e t r o v - G a l e r k i n

F i n i t e Element Method F o r Convection-Dominated L i n e a r and N o n l i n e a r
P a r a b o l i c Problems i n Two Space V a r i a b l e , C o m p u t a t i o n a l Methods - in
A p p l i e d--Mechanics, ( t o a p p e a r ) .

10. ODEN J . T . , DEMKOWICZ, L . , and STROUBOULIS, T., A d a p t i v e Methods € o r

I n c o m p r e s s i b l e Viscous Flow Problems w i t h Moving B o u n d a r i e s , - Accuracy
E s t i m a t e s and Adaptive Refinements i n F i n i t e Element Computations J o h n
Wiley and Sons L t d . , London, ( t o a p p e a r ) , ( P r o c e e d i n g s , L i s b o n P r o t u g a l ,
J u n e 19-22, 1984).

11. STROUBOULIS, T . , ODEN J.T., and DEMKOWICZ, L . , A P o s t e r i o r i E r r o r

Estimates f o r Some G a l e r k i n Space-Time Approximations of t h e Unsteady
Navier-Stokes E q u a t i o n s , =OM REPORT 85-5, J U N E 1985.

278
12. ODEN J . T . , RECKER, E . B . , LIN, T.L., and DEMKOWICZ, L., F o r m u l a t i o n and
F i n i t e Element A n a l y s i s of a G e n e r a l Class of R o l l i n g C o n t a c t Problems
w i t h F i n i t e E l a s t i c Deformation, Mathematics of F i n i t e Elements w i t h
A p p l i c a t i o n s , Academic P r e s s L t d . , London, ( t o a p p e a r ) .

13. ODEN J . T . , BECKER, E.R., L I N , T.L., and HSIEH, K.T., Numerical A n a l y s i s

of Some Problems R e l a t e d t o t h e Mechanics of Pneumatic Tires: F i n i t e
Deformation of Cord-Reinforced Rubber Composites, Symposium on Advances
and
- ---
--Trends i n S t r u c t u r e s and Dynamics, NASA CN, NASA Langley R e s e a r c h
C e n t e r , ( t o a p p e a r ) . NASA CP 2335, Oct. 1984, pp 297-307.

14. ODEN J.T., and LIN, T.L., On t h e G e n e r a l R o l l i n g C o n t a c t Problem f o r

F i n i t e Deformations of a V i s c o e l a s t i c C y l i n d e r , TICOM -
REPORT 85-,
-- A u s t i n ,
TX, 1985.

15. ODEN J . T . , and MARTINS, J . A . C . , Models and C o m p u t a t i o n a l Methods f o r

Dynamic F r i c t i o n Phenomena, Computer Methods i n Applied Mechanics ---------
and
E n g i n e e r i n g , Vol. 52, pp. 527-634, 1985.

16. ODEN J . T . , and DEMKOWICZ, L., E x t r a c t i o n Methods f o r S e c o n d - D e r i v a t i v e s

i n F i n i t e Element Approximations of L i n e a r E l a s t i c i t y Problems,
Communictions i n Applied Numerical
.__-____---- ---Methods,
- Vol. 1, pp. 137-139, 1985.

17. MARTINS, J.A.C., and ODEN J . T . , I n t e r f a c e Models, V a r i a t i o n a l P r i n c i p l e s

and Numerical S o l u t i o n s f o r Dynamic F r i c t i o n Problems, Mechanics O E
--
Material I n t e r f a c e s , ( P r o c e e d i n g s S e s s i o n on Mechanics of Material
I n t e r f a c e s a i d Mechanics of C o n t a c t Problems, a t t h e ASCE/ASME Mechanics
C o n f e r e n c e , Albuquerque, NM, J u n e 24-26, 1985)

279
 Figure 1

M:W = POLYNOMIAL OF IC, IIc, Illc

V = SOBOLEV SPACE OF ORDER ( 1 $1

= (w'.P(Q~))~ for some p. 2 i p < m

Figure 2

280
 EM =B U

ROL I ING CONTACT WITHOUT

FRICTION

3( = UNILATERAL CONSTRAINT SET

= ( q f v l q,iH onr,l

INCOMPRESSIBLF MATERIAL

Iq E v I q 2 I H on Tc , detVq = 1 a.e. in no)

Figure 4

281
 ROI I ING CONTACT WITH FRICTION

'u = SPACE OF VELOCITY-MOTIONS

5 [ q 6 v I (waeq1,q2) 6 v I

YARlATlONALITY B.V.P.

Figure 6

I 282
- Q = SPACE OF MULTIPLIER
detVq E Lp(Qo) ---> 9 = LP’(Q0)

(l/p) + (l+p’) = 1

i(p,)(,~) = TRILINEAR FORM

= p adjV)( : V q dvo

Figure 7

Figure 8

283
 C1=80 p s i C2=20 p s i VIS=O FRIC=.3 DISP=R, - H = .1 i n .

300 rpm

Figure 9

C1=80 pl s i C2=20 p s i VIS=O FRIC =.3 DISP=R, - H = .1 i n .

300 rpm

Figure 10

2 84
C1=80 p s i C2 = 20 p s i VIS=O FRIC=.3 DISP=R, - H = .1 i n .

'00 rpm

Figure 1 1

C1=80 p s i C2=20 p s i VIS=O FRIC=.3 DISP=R, - H = .1 i n .

300 rpm

Figure 12

285
 C1=80 p s i C2=20 p s i VIS=O FRIC=.3 OISP=R, - H = .1 i n .

.6 rpm

Figure 13

D r i v e n c a v i t y problem. Optimal mesh a f t e r 8 FE r e c a l c u l a -

tions. a = 4 , ~ ~ Y = 4O ,
Figure 14

286 c- 4
 ORIGlNAL PAGE
AND WHITE PHOTOGRAPH

Heat equation w i t h a dominating c o n v e c t i o n -

s o l u t i o n a f t e r 1 t i m e step, t = 0 . 0 2

F i g u r e 15

Heat e q u a t i o n w i t h a dominating c o n v e c t i o n -
s o l u t i o n a f t e r 25 t i m e steps, t = 0.5

F i g u r e 16

287
 ORlG INAL PAGE
BLACK AND WHITE PH07-0GRAPH

.,c,
Burger's equations. First component o f the
s o l u t i o n a f t e r 1 time s t e p , t = 0.02

Figure 1 7

I TIME=@.1888E-i-88

Problem w i t h exponential solution. Deviations o f the

local e f f e c t i v i t y r a t i o s e from unity. Parameters: h = 2 ,
A t = 0.1, 6 = 0.25, M = 4 K,1
Figure 18

288
 ORIGINAL PAGE
BLACK AND WHITE PHOT'OG'R~H

T"

Problem w i t h exDonentia1 s o l u t i o n . Deviations o f the local

effectivity ratios e from u n i t y . Parameters h = 2,
KY5
A t = 0.1, 6 = 0.5, M = 4
F i g u r e 1.9

I l l
m %
H
m R
m I
W
W
3
I
I1
-c

-E
m
m
z
w
W
w
w
3
J
W m

W
E :
Z
H Z
H
I
I-
-I
W
z
\
I-
cn
>
m W E
r- W
-I
R
t- 0 a
z W 3

z
H
LL
W
$
W
a

LL
\
I-
cn
R
2
W
U
\
I-
Z
w W H

1. P r e c o n d i t i o n e d J a c o b i Conjugate G r a d i e n t Scheme
2. 4-El ements/Ref inement
3 . Rates = Uniform/Const. = (CI1i2
F i g u r e 20

289
 CONTACT SOLUTION ALGORITHMS

J o h n T. T i e l k i n g
Texas A and M U n i v e r s i t y
C o l l e g e S t a t i o n , Texas

EXPANDED ABSTRACT

Two a l g o r i t h m s f o r o b t a i n i n g s t a t i c c o n t a c t s o l u t i o n s are d e s c r i b e d i n t h i s
p r e s e n t a t i o n . Although t h e y w e r e d e r i v e d f o r c o n t a c t problems i n v o l v i n g s p e c i f i c
s t r u c t u r e s ( a t i r e and a s o l i d r u b b e r c y l i n d e r ) , t h e y a r e s u f f i c i e n t l y g e n e r a l t o
b e a p p l i e d t o o t h e r s h e l l - o f - r e v o l u t i o n and solid-body c o n t a c t problems.

The s h e l l - o f - r e v o l u t i o n c o n t a c t a l g o r i t h m i s a method of o b t a i n i n g a p o i n t
l o a d i n f l u e n c e c o e f f i c i e n t m a t r i x f o r t h e p o r t i o n of s h e l l s u r f a c e t h a t i s ex-
pected t o c a r r y a c o n t a c t load. I f t h e s h e l l i s s u f f i c i e n t l y l i n e a r with r e s p e c t t o
c o n t a c t l o a d i n g , a s i n g l e i n f l u e n c e c o e f f i c i e n t m a t r i x c a n b e u s e d t o o b t a i n a good
a p p r o x i m a t i o n of t h e c o n t a c t p r e s s u r e d i s t r i b u t i o n . O t h e r w i s e , t h e m a t r i x w i l l b e
updated t o r e f l e c t n o n l i n e a r l o a d - d e f l e c t i o n behavior.

The solid-body c o n t a c t a l g o r i t h m u t i l i z e s a Lagrange m u l t i p l i e r t o i n c l u d e t h e

c o n t a c t c o n s t r a i n t i n a p o t e n t i a l e n e r g y f u n c t i o n a l . The s o l u t i o n i s found by
a p p l y i n g t h e p r i n c i p l e of minimum p o t e n t i a l e n e r g y . The Lagrange m u l t i p l i e r i s
i d e n t i f i e d as t h e c o n t a c t l o a d r e s u l t a n t f o r a s p e c i f i c d e f l e c t i o n .

A t p r e s e n t , o n l y f r i c t i o n l e s s c o n t a c t s o l u t i o n s h a v e been o b t a i n e d w i t h t h e s e
a l g o r i t h m s . A s l i d i n g t r e a d element h a s b e e n developed t o c a l c u l a t e f r i c t i o n s h e a r
f o r c e i n t h e c o n t a c t r e g i o n of t h e r o l l i n g s h e l l - o f - r e v o l u t i o n t i r e model. T h i s
e l e m e n t a l l o w s a r e l a t i v e l y g e n e r a l , non-Coulomb, f r i c t i o n l a w t o b e s p e c i f i e d f o r
t h e c o n t a c t i n t e r f a c e . I t h a s t h e added a d v a n t a g e of a l l o w i n g f r i c t i o n t o b e cal-
c u l a t e d i n t h e c o n t i n u o u s i n t e r f a c e and, when coupled w i t h t h e solid-body c o n t a c t
a l g o r i t h m , w i l l p e r m i t a n a l y t i c i n v e s t i g a t i o n of v a r i o u s continuum f r i c t i o n t h e o r i e s
t h a t have been proposed.

T h e o u t l i n e of f u t u r e d i r e c t i o n s f o r t h e development of c o n t a c t s o l u t i o n
algorithms is:

I. SHELL-OF-REVOLUTJON CONTACT ALGORITHM

11. SOLID-BODY CONTACT ALGORJTHM

111. STATTC AND ROLLJNG CONTACT F R I C T T O N

IV. FUTURE D I R E C T I O N S FOR CONTACT S O L U T I O N ALGORTTHMS

291
 I. SHELL-OF-REVOLUTION CONTACT ALGORITHM

A s h e l l whose geometry and material p r o p e r t i e s are axisymmetric c a n b e eco-

n o m i c a l l y modeled by s h e l l - o f - r e v o l u t i o n f i n i t e e l e m e n t s . The SAMMSOR/SNASOR pro-
grams ( r e f s . 1 , 2 ) , f o r example, p e r m i t n o n l i n e a r b e h a v i o r of o r t h o t r o p i c s h e l l s of
r e v o l u t i o n t o b e c a l c u l a t e d , i n c l u d i n g r e s p o n s e t o nonaxisymmetric l o a d s . The
a l g o r i t h m d e s c r i b e d h e r e w a s developed t o c a l c u l a t e t h e s h e l l d e f l e c t i o n i n r e s p o n s e
t o a n o d a l p o i n t l o a d , u t i l i z i n g t h e c a l c u l a t e d r e s p o n s e t o a sequence of harmoni-
c a l l y v a r y i n g r i n g l o a d s on t h e node. The p o i n t l o a d s o l u t i o n i s t h e n u s e d t o con-
s t r u c t a n i n f l u e n c e c o e f f i c i e n t m a t r i x , from which t h e s h e l l c o n t a c t s o l u t i o n i s
obtained.

292
 FINITE-ELEMENT TIRE MODEL

The tire is modeled here by an assembly of axisymmetric shell elements con-

nected to form a meridian of arbitrary curvature and following the carcass mid-
surface. The elements are homogeneous orthotropic, with moduli determined by the
ply structure of a particular tire. Details of this model are given in reference 3.
If the deformation is symmetric about the wheel plane only one-half of the meridian
is modeled, as shown in figure 1. The finite elements are joined at nodal circles,
referred to here as nodes. Node 12 in figure 1 is located at the tire bead and is
given in built-in end condition.

Figure 1

293
 S I N G L E HARMONIC R I N G LOADS

The f i n i t e - e l e m e n t t i r e model w i l l respond t o s i n g l e harmonic r i n g l o a d s on

t h e n o d a l c i r c l e s i n a d d i t i o n t o a u n i f o r m i n f l a t i o n p r e s s u r e l o a d . An a p p r o x i -
m a t e l y l i n e a r r i n g l o a d - d e f l e c t i o n r e s p o n s e i s o b t a i n e d when a n i n d i v i d u a l r i n g l o a d
i s a p p l i e d t o any node of t h e p r e s s u r i z e d t i r e model. An example r i n g load-
d e f l e c t i o n c a l c u l a t i o n f o r a p a s s e n g e r t i r e model i s shown i n f i g u r e 2. A harmonic
sequence of s t i f f n e s s matrices i s o b t a i n e d by a p p l y i n g a sequence of s i n g l e harmonic
r i n g l o a d s t o e a c h of t h e nodes t h a t may b e i n t h e tire-pavement c o n t a c t r e g i o n .

TOTAL
RADIAL
3000 -
RING
LOAD (Ib)
2000 -
PRESSURE

I
0 0.5 I .o 1.5
RADIAL DEFLECTION (in)

CROWN LOAD-DEFLECTION DATA CALCULATED WITH A UNIFORM R I N G LOAD APPLIED TO THE CROWN NODE

SINGLE H A W N I C RING LOADS APPLIED TO A FINITE-ELEMENT NODE

Figure 2
I
2 94
 TRANSFER FUNCTION DEFINITION

A s a consequence of the l i n e a r i t y of t h e r i n g l o a d - d e f l e c t i o n r e s p o n s e , t h e
a p p l i c a t i o n of a s i n g l e harmonic r i n g l o a d produces a displacement f i e l d t h a t
varies c i r c u m f e r e n t i a l l y i n t h e same harmonic as t h e a p p l i e d r i n g l o a d . The d e f i n i -
t i o n of t h e t r a n s f e r f u n c t i o n Tn as t h e r a t i o o f t h e o u t p u t and i n p u t a m p l i t u d e s
i s g i v e n below ( r e f . 4 ) . S i n c e each node responds d i f f e r e n t l y , a t r a n s f e r f u n c t i o n
m a t r i x T i k l n i s used t o s t o r e t h e s t i f f n e s s i n f o r m a t i o n g e n e r a t e d by t h e r i n g l o a d s .
The p a r t i t i o n s of t h i s matrix are determined by t h e d i r e c t i o n of t h e r i n g l o a d . ( F i g . 3 ) .

S i n g l e Harmonic R i n g Load An cos ne ( i n p u t )

S i n g l e Harmonic Displacement Bn cos ne ( o u t p u t )

Bn
TRANSFER FIJNCTTON T, = A
n

= nth harmonic t r a n s f e r f u n c t i o n r e l a t i n g displacement o f node i t o an

Tikln
n t h harmonic r i n g l o a d on node k

LOAD TYPE

Circum-
Radial fe rential Axial
r- I 1

z
W

I I I
a

n=0,1,. .. .,N/2
Figure 3

295
 POINT LOAD VECTOR { p ) AND THE DISCRETE FOURIER TRANSFORM (DFT)

T h i s a p p l i c a t i o n of t h e d i s c r e t e F o u r i e r t r a n s f o r m u s e s a n even number of
p o i n t s ( N ) , e q u a l l y spaced around t h e c i r c u m f e r e n c e . The example shown i n f i g u r e
4 u s e s N = 8 p o i n t s . A u n i t l o a d i s a p p l i e d a t any p o i n t , s a y p o i n t 0. The DFT
of t h e l o a d v e c t o r y i e l d s a s e t of N c o e f f i c i e n t s , which a r e a p p r o x i m a t e v a l u e s
Gj9
of t h e c o e f f i c i e n t s of t h e c o n v e n t i o n a l F o u r i e r s e r i e s d e f t n e d on t h e c o n t i n u o u s
i n t e r v a l 0 < 8 < 27~and r e p r e s e n t i n g t h e u n i t p o i n t l o a d . The p o i n t l o a d i s
a p p l i e d , s e q u e n t i a l l y , i n t h e r a d i a l , a x i a l , and c i r c u m f e r e n t i a l d i r e c t i o n s .

INFLUENCE COEFFICIENT GENERAT I O N

I1

{ P I = ( 1 , 0, 0, 0, 0, 0, 0, 0 ) load Lector
1 N-1
-iPn/N
DFT G . = - gk$k We = e
k=O

9k = {PI ’ G.J = -N1 j = 0, 1, . . ., N-1

Figure 4

296
 INVERSE DISCRETE FOURIER TRANSFORM (IDFT) AND THE INFLUENCE COEFFICIENTS

Having t h e u n i t p o i n t l o a d r e p r e s e n t e d by a c o n v e n t i o n a l F o u r i e r series, whose

c o e f f i c i e n t s an are approximately g i v e n by t h e DFT c o e f f i c i e n t s , t h e t r a n s f e r
f u n c t i o n s T i k l n a r e a p p l i e d , on each harmonic, t o o b t a i n t h e c o e f f i c i e n t s b of t h e
F o u r i e r series r e p r e s e n t i n g t h e r e s p o n s e of t h e n o d a l c i r c l e t o t h e u n i t p oPi n t
l o a d . The i n v e r s e d i s c r e t e F o u r i e r t r a n s f o r m i s t h e n used t o e v a l u a t e t h e d i s -
placements, u , a t t h e N p o i n t s . These d i s p l a c e m e n t s a r e t h e elements of t h e
i n f l u e n c e c o e y f i c i e n f matrix [AijkL]. (Fig. 5 ) .

INPUT SERIES COEFFICIENTS a 2 Gn -- 1

n
OUTPUT S E R I E S COEFFICIENTS b = anTikln = -1 T i k l n
n

DFT OF DISPLACEMENT VECTOR Gn b

n

I!-1
IDFT u ik = G WWmn
n e
m = 0, 1, . . . ) N-1
n=O

ik
INFLUENCE COEFFICIENTS
jkl
Ai
= u j -1 j = 1, 2, . . .) N

ik
SHIFT: Aijke = u j -1 j = l, &+1, . . ., N

-
SYMMETRY: Akeij - Aijkl

Figure 5

297
 INFLUENCE COEFFICIENT MATRIX

The influence coefficient matrix relates the radial, axial, and circumferen-
tial components of the displacement of points on the tire surface to the radial,
axial, and circumferential components of load at these points. The radial response
partition, shown in figure 6, is used to obtain a solution for frictionless con-
tact, in which the axial and circumferential force components are known to be zero.
The matrix here covers 3 points on each of 5 nodes. The point separation with this
matrix is 11.25 degrees.

Pke = load a t p o i n t e on node k

d.. = d e f l e c t i o n of p o i n t j on node i
- . - 1J --
dl 1 All11 p11

d21 A2111 A2121 p21

d31 A3111 A3121 A3131 '31

d4 1 A4111 A4121 h131 4141 4' 1

d51 A5111 A5121 k131 A5141 A5151 '51

dl2 A1211 A122 1 h231 A1241 A1251 A1212 '1 2

d22 12211 A2221 A2231 A2241 A2251 A2212 A2222 p22

d32 \32 11 A322 1 A3231 A3241 A3251 A3212 A3222 A3232 '32
d42 4211 A422 1 4231 4241 4251 4212 4222 4232 h242 '42

d52
. d -
5211 A5221 A5231 4241 k251 A5252 A5222 k232 4242
- '52- -
h252

Figure 6

298
 TOROIDAL SHELL CONTACT SCHEMATIC

A c y l i n d r i c a l c o o r d i n a t e system i s used t o l o c a t e p o i n t s on t h e t o r o i d a l
s u r f a c e . The c o o r d i n a t e s r , 0, and z i n d i c a t e t h e r a d i a l , c i r c u m f e r e n t i a l , and
a x i a l d i r e c t i o n s , r e s p e c t i v e l y . The t i r e e q u a t o r l i e s i n t h e r-0 p l a n e (wheel
p l a n e ) and a t i r e m e r i d i a n is i n an r-z p l a n e .

A f t e r t h e i n f l a t i o n s o l u t i o n h a s been o b t a i n e d , t h e t i r e model i s d e f l e c t e d
a g a i n s t a f r i c t i o n l e s s , f l a t s u r f a c e . The c o n t a c t i n g s u r f a c e i s p e r p e n d i c u l a r
t o t h e wheel p l a n e and p o s i t i o n e d a t t h e s p e c i f i e d loaded r a d i u s R l , as shown
i n f i g u r e 7 . The v e r t i c a l l o a d and t h e c o n t a c t p r e s s u r e d i s t r i b u t i o n a r e unknown,
a priori.

Figure 7

299
 DEFLECTED MERIDIAN

The d e f l e c t e d s h a p e of t h e m e r i d i a n p a s s i n g t h r o u g h t h e c e n t e r of c o n t a c t i s
shown i n f i g u r e 8. T h i s s h a p e i s c a l c u l a t e d by t h e f i n i t e - e l e m e n t t i r e model f o r
t h e s p e c i f i e d t i r e d e f l e c t i o n of one i n c h . The t i r e l o a d t h a t w i l l produce a one
i n c h t i r e d e f l e c t i o n i s c a l c u l a t e d t o b e 10,590 l b . F i g u r e 8 a l s o shows t h e
m e r i d i a n p r i o r t o i n f l a t i o n and t h e c a l c u l a t e d s h a p e of t h e m e r i d i a n of t h e i n f l a t e d
t i r e , p r i o r t o c o n t a c t l o a d i n g . These f i n i t e - e l e m e n t m e r i d i a n s f o l l o w t h e carcass
m i d s u r f a c e , a s i n d i c a t e d i n Eigure 1. Geometric and m a t e r i a l p r o p e r t y d a t a on t h e
Space S h u t t l e n o s e g e a r t i r e w e r e u s e d f o r t h e c a l c u l a t e d r e s u l t s shown i n f i g u r e s
8 , 9 , and 10.

I -- --_ \ 6 - 1.0 in ( s p e c i f i e d )

.,I --__Uninflated Shape

1 I I I I 1
0 1 2 3 4 5

a (in)

Figure 8

300
 CONTACT PRESSURE DISTRIBUTIONS

The s t a t i c c o n t a c t p r e s s u r e v a l u e s ( p s i ) c a l c u l a t e d f o r two d i f f e r e n t l o a d s
on t h e S h u t t l e nose g e a r t i r e are shown i n f i g u r e 9. The number of f i n i t e - e l e m e n t
p o i n t s i n t h e c o n t a c t r e g i o n i n c r e a s e s as t h e t i r e l o a d i n c r e a s e s . A rough e s t i -
m a t e of t h e c o n t a c t boundary i s o b t a i n e d by e x t r a p o l a t i o n of t h e p r e s s u r e d i s t r i -
b u t i o n , I n t e g r a t i o n of t h e p r e s s u r e d i s t r i b u t i o n g i v e s t h e t i r e l o a d .

NOSE GEAR T I R E CONTACT PRESSURE DISTRIBUTIONS

3 2 ~ 8 . 8 20 PR TYPE V I 1
I n f l a t i o n Pressure = 300 psi
_ _ _ _ _ _ _ _ - ---------- - -
/
/
A- - 703.1
-- -- . \

I
I

i 191.2 327.2 191.2

\

' \
I 117.5 \
I I
I 364.2 482.6 psi 364.2 I
I I
I 117.5 I
\ I

'.--- _--
\ 191.2 327.2 191.2 I
/
/
703.1 /

- - - -- _ _ _ _ _ _ _ _ - - -------
/

(a) 6 = 0.75 i n . F, = 5.708 l b

--- - - - - - - - - - -- ---_
I
/ _---
c _ _ _ _ _ _

527.9
--.-
/ \
I 388.6 822.3 388.6 I
I 318.5 285.3
I
285.3 I
I
I 48.9 173.0 48.9 I
I
I 401.3 529.7 psi 401.3 1
I I
I 48.9 173.0 48.9
I
I 285.3 318.5 285.3 I
\ I
\ 388.6 822.3 388.6 /
\ I

'----- ~

-- ---- - _ _ _ _ _ _ _ _ - ----- _---

527.9 #MM

(b) 6 = 1.00 i n , F, = 10,590 l b

Figure 9

301
 T I R E LOAD VERSUS T I R E DEFLECTION

An i m p o r t a n t t e s t of a t i r e model i s i t s a b i l i t y t o c a l c u l a t e a s t a t i c
l o a d - d e f l e c t i o n c u r v e . F i g u r e 1 0 compares t h e l o a d - d e f l e c t i o n c u r v e c a l c u l a t e d
f o r t h e S h u t t l e n o s e g e a r t i r e w i t h measured d a t a f o r a s i m i l i a r a i r c r a f t t i r e .
Although t h e s e a r e b o t h 3 2 x 8.8 Type V I 1 t i r e s , c o n s t r u c t i o n a l d e t a i l s c a n
a l t e r t h e l o a d - d e f l e c t i o n c u r v e (and many o t h e r a s p e c t s of t i r e b e h a v i o r ) . The
c o r d used i n t h e t e s t t i r e i s unknown and may be q u i t e d i f f e r e n t from t h e nylon
c o r d i n t h e Space S h u t t l e t i r e .

Load-Deflection Curves For 32 x 8 . 8 Type VI1 Aircraft Tires

30000

Tire Calculated
Load
Fz (lb)
25000
/
/*
/

/
/ / 24 P r

'/
/
/ 285 Psi
/ Measured Data
20000
(Similiar Tire)

15000

10000

-
5000

0 1 1 I I I I I I I
0 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 2.25

Tire Deflect ion 6 (in)

Deflection

Figure 10

302
I
 11. SOLID-BODY CONTACT ALGORITHM

A s o l i d body w i t h a c y l i n d r i c a l s u r f a c e i s o f t e n loaded i n c o n t a c t a g a i n s t a
r i g i d s u r f a c e . The c o n t a c t l o a d may r e v o l v e around t h e body, as i n t h e case of
a r o l l e r o r a s o l i d t i r e , o r may remain s t a t i o n a r y i f t h e c y l i n d e r i s u s e d as a
s u p p o r t cushion.

I n t e r f a c i a l f r i c t i o n , p r e s e n t i n a l l c o n t a c t problems, i s c u r r e n t l y an a c t i v e
f i e l d of r e s e a r c h . The s t u d y of f r i c t i o n a l behavior i s f a c i l i t a t e d i f t h e c o n t a c t
r e g i o n i s r e l a t i v e l y l a r g e , as i s produced when t h e body i s h i g h l y deformable.
T h i s makes i t easier t o c a l c u l a t e d i s t r i b u t i o n s of normal p r e s s u r e and t a n g e n t i a l
motion ( s l i p ) i n t h e i n t e r f a c e . I n t h e case of rubber c o n t a c t , t h e behavior
d e v i a t e s s u f f i c i e n t l y from t h e Coulomb f r i c t i o n l a w t h a t o t h e r , more p h y s i c a l l y
r e a l i s t i c l a w s , can be e a s i l y t e s t e d . S i n c e f r i c t i o n i s a microscopic phenomenon,
a c o n t a c t s o l u t i o n g i v i n g continuous d i s t r i b u t i o n s of i n t e r f a c i a l p r e s s u r e and
s l i p i s d e s i r a b l e f o r a n a l y t i c purposes. The c o n t a c t a l g o r i t h m d e s c r i b e d h e r e
p r o v i d e s a continuum s o l u t i o n f o r f r i c t i o n l e s s c o n t a c t , t h e f i r s t ' s t e p toward
a n a l y s i s of f r i c t i o n i n t h e continuous c o n t a c t i n t e r f a c e .

303
 A PLANE STRAIN CONTACT PROBLEM

An e l a s t i c s e m i c y l i n d e r of r a d i u s R i s bonded t o a f i x e d s u r f a c e , f i g u r e
l l ( a ) . A c o n t a c t l o a d i s a p p l i e d by a r i g i d p l a t e t h a t d e f l e c t s t h e s e m i c y l i n d e r
as shown i n f i g u r e l l ( b ) .

The problem i s f o r m u l a t e d i n terms of c y l i n d r i c a l material c o o r d i n a t e s

( r , 0 , z ) which i d e n t i f y p o i n t s i n t h e undeformed body, Bo. A p o i n t Po i n Bo i s
l o c a t e d by C a r t e s i a n c o o r d i n a t e s x 1 and xg axes shown i n f i g u r e l l ( a ) and

x = r cos 0 x2 = r sin 0 x 3= z
1

The c o n t a c t l o a d i s assumed t o produce a p l a n e s t r a i n d e f o r m a t i o n . P o i n t Po

moves t o p o s i t i o n P i n t h e deformed body, B. P o i n t P i s l o c a t e d by t h e C a r t e s i a n
c o o r d i n a t e s y . . With p l a n e s t r a i n , y, = y l ( r , O ) , y, = y 2 ( r , 0 ) , and y, = X,z
where X 3 = 1 2s a s p e c i f i e d c o n s t a n t e x t e n s i o n r a t i o .

t x2

t y2

F i g u r e 11

304
 GEOMETRIC DESCRIPTION

The m e t r i c t e n s o r s g
i B and G i j , g i v e n below, c o m p l e t e l y d e s c r i b e t h e e l a s t i c
s e m i c y l i n d e r b e f o r e and a f t e r d e f o r m a t i o n . S i n c e y3 i s known a priori, t h e
problem i s s o l v e d by f i n d i n g t h e f u n c t i o n s y , ( r , 8 ) and y 2 ( r , e ) which d e t e r m i n e
Gij. The d i s p l a c e m e n t f i e l d i s n o t u t i l i z e d i n t h i s f o r m u l a t i o n b u t i t c a n , of
c o u r s e , b e found when x and y a r e known. ( F i g . 1 2 ) .
i i

( i m p l i e d sum
with a = 1,2)

The Green/Saint-Venant s t r a i n t e n s o r components are d e f i n e d a s

y i j = 3i(Gij -

Und e f ormed Deformed

Metric [ g . . I Metric [ G . . ]
1J 1J

Figure 12

305
 MATERIAL DESCRIPTION

The material i s assumed t o be h y p e r e l a s t i c s o t h a t i t s c o n s t i t u t i v e p r o p e r t i e s

are contained i n a s t r a i n energy d e n s i t y , W. I s o t r o p y i s a l s o assumed. For
p l a n e s t r a i n of an i s o t r o p i c m a t e r i a l , t h e s t r a i n energy i s known t o b e a f u n c t i o n
of only t h e f i r s t and t h i r d s t r a i n i n v a r i e n t s , I, and I 3 ( r e f . 5)

For g e n e r a l deformation, t h e s t r a i n i n v a r i e n t s are g i v e n by

ij
11=g G.. I,= G / g
1J
2
where g = d e t [g. .] = r f o r t h e semicylinder and G = d e t [ G . .].
1J 1J

When t h e m a t e r i a l i s a l s o assumed t o b e i n c o m p r e s s i b l e ( I 3 = 1 ) t h e c o n s t i t u t i v e
behavior i s n o t completely determined by t h e s t r a i n energy d e n s i t y . H y d r o s t a t i c
p r e s s u r e becomes an a d d i t i o n a l unknown, which can be determined as a Lagrange
m u l t i p l i e r ( r e f . 6 ) . T h i s d i f f i c u l t y i s avoided i f a compressible material model
i s used.

The m a t e r i a l d e s c r i p t i o n s e l e c t e d f o r t h e c o n t a c t problem solved h e r e i s t h e

compressible neo-Hookean model developed f o r continuum rubber by B l a t z and KO
( r e f . 7 ) . The Blatz-Ko model may b e expressed as

where p i s t h e c l a s s i c a l s h e a r modulus, K i s t h e b u l k modulus, and k i s a parameter

r e l a t e d t o atomic r e p u l s i o n . When I, = 1, t h e Blatz-Ko model r e d u c e s t o t h e
neo-Hookean model f o r i n c o m p r e s s i b l e m a t e r i a l . For s m a l l s t r a i n s i t r e d u c e s t o
t h e energy d e n s i t y g i v i n g Hooke's l a w f o r compressible i s o t r o p i c material.

306
 CONTACT CONSTRAINT

The assumption of f r i c t i o n l e s s c o n t a c t w i t h a r i g i d s u r f a c e , p o s i t i o n e d per-

p e n d i c u l a r t o t h e y, axis ( s e e f i g . l l ( b ) ) , i m p l i e s a geometric c o n s t r a i n t on1.y on
t h e s o l u t i o n f u n c t i o n y 2 ( r , 0 ) . Within t h e c o n t a c t r e g i o n , whose e x t e n t i s n o t
known a p r i o r i , t h e deformed s u r f a c e i s f l a t and i t i s known t h a t y 2 = Rl
where Rp. i s t h e s p e c i f i e d l o c a t i o n of t h e c o n t a c t s u r f a c e . Outside t h e c o n t a c t
r e g i o n , and i n t h e i n t e r i o r of B , t h e s o l u t i o n must s a t i s f y y, < RL. This i n -
e q u a l i t y c o n s t r a i n t o n y i s converted t o an e q u a l i t y c o n s t r a i n t by i n t r o d u c i n g a
new f u n c t i o n s(r,f3), d e f i n e d by t h e f o l l o w i n g e q u a t i o n
2
y, + s = Rl ( c o n s t r a i n t equation)

which i s v a l i d everywhere on t h e boundary and i n t h e i n t e r i o r of B. The f u n c t i o n

s(r,0), called a slack variable, h a s been used p r e v i o u s l y i n o p t i m i z a t i o n problems
w i t h a n i n e q u a l i t y c o n s t r a i n t . Reference 8 g i v e s several examples of t h e u s e of
slack variables.

The c o n t a c t problem i s solved by minimizing t h e s t r a i n energy i n B , s u b j e c t

t o t h e c o n s t r a i n t e q u a t i o n g i v e n above. S i n c e p l a n e s t r a i n i s assumed, t h e energy
i s uniform a l o n g t h e a x i s of t h e s e m i c y l i n d e r . Using symmetry, i n t e g r a t i o n of t h e
energy d e n s i t y i s t a k e n over one-half of t h e r-8 p l a n e contained i n Bo.

The c o n s t r a i n t e q u a t i o n is brought i n t o t h e energy d e n s i t y f u n c t i o n a l by

means of a Lagrange m u l t i p l i e r f u n c t i o n A(r,O). The c o n t a c t problem i s t h e n
governed by t h e f o l l o w i n g f u n c t i o n a l

where

Although r W i s p o s i t i v e d e f i n i t e , F i s n o t p o s i t i v e d e f i n i t e due t o t h e addi-

t i o n of t h e c o n s t r a i n t . T h e r e f o r e , I may o n l y be regarded as being made sta-
t i o n a r y i n s t e a d of minimized by equilibrium s o l u t i o n functions. The i n t e g r a l
of W, however, i s minimized by t h e e q u i l i b r i u m s o l u t i o n and t h i s i s used as a
check d u r i n g t h e s o l u t i o n f i n d i n g p r o c e s s .

Through a d d i t i o n a l a n a l y s i s (J. T. T i e l k i n g , Texas A and M U n i v e r s i t y ,

unpublished d a t a ) t h e Lagrange m u l t i p l i e r f u n c t i o n is shown t o be a n unknown
c o n s t a n t , i d e n t i f i e d as t h e r e s u l t a n t load i n t h e c o n t a c t region. The c o n s t r a i n t
c o n d i t i o n may then be removed from t h e i n t e g r a l , t h e s l a c k v a r i a b l e is no longer
needed, and t h e c o n t a c t problem i s now governed by

The s o l u t i o n i s obtained by f i n d i n g y a ( r , 8 ) and t h e c o n s t a n t X which make

I ( y a , A ) s t a t i o n a r y . (Fig. 1 3 ) .

307
 I
Exact
-

/ I \

I I I
/ / / / / / / - / R /

Example: M= N = 2

Figure 13

308
 SOLUTION FUNCTIONS

The numerical s o l u t i o n i s o b t a i n e d by a p p l i c a t i o n of t h e p r i n c i p l e of
s t a t i o n a r y p o t e n t i a l energy ( r e f . 6), u s i n g t h e f u n c t i o n a l I ( y a , A ) i n which X is
an unknown c o n s t a n t . The s o l u t i o n f u n c t i o n s y a r e taken a s two-dimensional
f i n i t e series
a
M N

i-1 j=l

M N

where a and bi. a r e unknown c o e f f i c i e n t s . The symmetry and geometric boundary

c o n d i t i g j i s e v i d e a t i n f i g u r e 11a r e m e t by

ij i
(r,e) = r s i n (2je)
yl

ij
i
y, (rye) = r s i n [ ( 2 j -1.)0]

The above f u n c t i o n s a l l o w t h e energy d e n s i t y t o be i n t e g r a t e d , t h e r e b y re-

ducing t h e f u n c t i o n a l I t o an a l g e b r a i c f u n c t i o n of t h e 2 x M X N + 1 unknown
c o n s t a n t s a i j y bij. and A

The f u n c t i o n a l I i s made s t a t i o n a r y by t h e c o n s t a n t s obtained from t h e following

s e t of simultaneous n o n l i n e a r e q u a t i o n s :

aaij
31
= O and----
81
abij
- 0 f o r i = 1,2,...,M and j = 1 , 2 , ..., N

T h i s system i s solved i n an i t e r a t i v e manner by t h e Newton-Raphson method. Using

the s t a r t i n g values a = b = X = 0 , f i v e o r six i t e r a t i o n s (which g i v e s u c c e s s i v e
i'
c o r r e c t i o n s t o t h e s e c a n s t a k i s ) are u s u a l l y s u f f i c i e n t . The i t e r a t i o n s are con-
t i n u e d u n t i l t h e c o r r e c t i o n s appear t o have n e g l i g i b l e e f f e c t on t h e s o l u t i o n
f u n c t i o n s y9 ( r y e ) . The energy d e n s i t y i s e v a l u a t e d a f t e r each i t e r a t i o n t o
check f o r minimization.

309
 DEFORMATION SOLUTION

Numerical r e s u l t s h a v e been o b t a i n e d u s i n g material c o n s t a n t s 1-1 = 75 p s i

( s h e a r modulus), K = 475,000 p s i ( b u l k m o d u l u s ) , and k = 13.3 i n t h e Blatz-Ko
model. The v a l u e s of K and k a r e t a k e n from r e f e r e n c e 7 where t h e y a r e shown t o
g i v e a good f i t t o h y d r o s t a t i c compression d a t a on Butyl t r e a d r u b b e r ( p o l y i s o -
b u t y l e n e ) . The s h e a r modulus i s b e l i e v e d t o b e a r e a l i s t i c estimate, b a s e d on
T r e l o a r ' s s t a t e m e n t ( r e f . 9) t h a t t h e s h e a r modulus of r u b b e r i s lower t h a n t h e
b u l k modulus by a f a c t o r of a b o u t 104 .
The computer-generated drawing below shows c o o r d i n a t e c i r c l e s and r a d i i
b e f o r e d e f o r m a t i o n (xl and x2, dashed l i n e s ) and t h e deformed c o n f i g u r a t i o n
(yl and y 2 , s o l i d l i n e s ) of t h e s e c i r c l e s and r a d i i . The d e f o r m a t i o n i s pro-
duced by a 10-percent d e € l e c t i o n of t h e c o n t a c t i n g s u r f a c e (shown d a s h e d ) . The
d e f o r m a t i o n s o l u t i o n , y1 ( r y e ) and y2 ( r y e ) , i s o b t a i n e d i n a 16-term series
f o r each f u n c t i o n ; t h e s t r a i n e n e r g y i s minimized by t h e c o e f f i c i e n t s a and
b i j f o r i = 1 , 2 , 3 , 4 and j = 1 , 2 , 3 , 4 , found a f t e r s i x Newton-Raphson i t e r a t i % .
T h i s c o m p u t a t i o n t o o k 3 0 s e c o n d s of CPU t i m e on a mainframe computer (Amdahl
470/V8). The Lagrange m u l t i p l i e r o b t a i n e d i n t h i s s o l u t i o n i s X=93.1, i n t e r -
p r e t e d a s a 93.1 l b l o a d needed f o r a 10-percent d e f l e c t i o n if t h e s e m i c y l i n d e r
extends 1 inch i n t h e z-direction. (Fig. 14).

--- Undeflected

10% Deflection

Figure 14

310
 111. STATIC AND ROLLING CONTACT FRICTION

S t a t i c Contact Friction. A body i s brought i n t o s t a t i c c o n t a c t by motion

p e r p e n d i c u l a r t o t h e c o n t a c t p l a n e . During t h i s motion, t h e c o n t a c t boundary
expands u n t i l t h e r e s u l t a n t of t h e normal c o n t a c t p r e s s u r e r e a c h e s e q u i l i b r i u m
w i t h t h e e x t e r n a l l o a d a p p l i e d t o t h e body. A s t h e c o n t a c t r e g i o n i s formed, s h e a r
f o r c e s are generated by t a n g e n t i a l motion of c o n t a c t i n g s u r f a c e p o i n t s . These
s h e a r f o r c e s a r e f r i c t i o n a l and t r a n s i e n t , r e a c h i n g e q u i l i b r i u m l e v e l s when t h e
body i t s e l f comes i n t o e q u i l i b r i u m . Although t h e body may be assumed e l a s t i c , and
thus conservative, t h e f r i c t i o n a l shear f o r c e s are not conservative.

The formidable problem of c a l c u l a t i n g a s t a t i c c o n t a c t s o l u t i o n i n c l u d i n g t h e

e f f e c t of f r i c t i o n i s a l l e v i a t e d somewhat by assuming Coulomb's l a w of f r i c t i o n
i s v a l i d i n t h e c o n t a c t r e g i o n . An a l g o r i t h m f o r i n c l u d i n g Coulomb f r i c t i o n i n
a s t a t i c c o n t a c t problem h a s been developed by R o t h e r t e t a l . ( r e f . l o ) . Although
Coulomb f r i c t i o n may b e t a k e n f o r a n approximate a n a l y s i s of f r i c t i o n a l c o n t a c t ,
mathematical and p h y s i c a l u n c e r t a i n t i e s a r i s e when i t i s assumed. N o n c l a s s i c a l
f r i c t i o n l a w s have been proposed by Oden and h i s coworkers (e.g., r e f . 11). A l -
though t h e s e appear t o have been developed mainly f o r metals, t h e y may b e a p p l i c -
a b l e t o more deformable material such as rubber.

Rolling Contact Friction. T h i s d i s c u s s i o n i s l i m i t e d t o s t e a d y r o l l i n g

under a c o n s t a n t l o a d . Neglecting h y s t e r e t i c e f f e c t s i n t h e body, t h e power i n -
p u t t o m a i n t a i n s t e a d y r o l l i n g i s balanced by t h e work r a t e of t h e f r i c t i o n f o r c e s
i n t h e c o n t a c t r e g i o n . I n s t e a d y r o l l i n g , a c o n t a c t r e g i o n , whose boundary is
f i x e d by t h e l o a d , is c o n t i n u o u s l y generated. The normal p r e s s u r e and s l i d i n g
v e l o c i t y a t a g i v e n l o c a t i o n w i t h i n t h e c o n t a c t boundary do n o t change w i t h time
s o s t e a d y - s t a t e f r i c t i o n a l behavior i s maintained. The r o l l i n g c o n t a c t problem
w i t h f r i c t i o n i s t h e r e f o r e much easier t o a n a l y z e and p r o v i d e s a mechanism f o r
t h e s t u d y of n o n c l a s s i c a l f r i c t i o n t h e o r i e s . An a l g o r i t h m f o r c a l c u l a t i n g f r i c t i o n
i n t h e c o n t a c t r e g i o n of a t i r e r o l l i n g a t c o n s t a n t v e l o c i t y w i l l b e d e s c r i b e d
next .

311
 SLIDING TREAD MODEL

T h i s i s a t r e a d element model developed t o c o n v e r t a f r i c t i o n l e s s r o l l i n g

c o n t a c t s o l u t i o n i n t o a s o l u t i o n f o r r o l l i n g c o n t a c t w i t h f r i c t i o n . The element
Passes through t h e c o n t a c t r e g i o n w i t h t h e v e l o c i t y found f o r f r i c t i o n l e s s c o n t a c t .
This i s termed t h e carcass v e l o c i t y , V,, whose d i s t r i b u t i o n i s symmetric about t h e
c e n t e r of t h e f o o t p r i n t a s sketched below

e n t r y 4 V exit
C

/ / / / 1 1 1

For a f r e e - r o l l i n g t i r e , t h e amplitude of t h e f o o t p r i n t s l i d i n g v e l o c i t y i s
v e r y s m a l l . A t 60 mph (1056 i p s ) , t h e peak V, i s c a l c u l a t e d (by t h e a u t h o r ) t o
be about 50 i p s i n a f r i c t i o n l e s s f o o t p r i n t .

I n f r e e - r o l l i n g , t h e normal c o n t a c t p r e s s u r e d i s t r i b u t i o n , p , i s e s s e n t i a l l y
unchanged by f r i c t i o n . The s l i d i n g v e l o c i t y d i s t r i b u t i o n , however, i s s i g n i f i -
c a l l y a l t e r e d i n a n i n t e r a c t i v e manner. A h y s t e r e t i c t h e o r y of tire-pavement
I f r i c t i o n proposed by Schapery ( r e f . 1 2 ) g i v e s t h e dependence of t h e f r i c t i o n
f o r c e , F,, on t h e a c t u a l s l i d i n g v e l o c i t y , Vs, and normal p r e s s u r e , p, a t a
I p o i n t i n t h e c o n t a c t r e g i o n . T h i s i s expressed as

( h y s t e r e t i c theory)

where a , b, and B are material f r i c t i o n p r o p e r t i e s . The s l i d i n g tread element

model, shown i n f i g u r e 1 5 ( a ) , i s v i s c o e l a s t i c w i t h s t i f f n e s s and damping parameters
K and c. S l i d i n g f r i c t i o n , F,, c a u s e s t h e element t o deform, t h e r e b y i n f l u e n c i n g
t h e s l i d i n g v e l o c i t y Vs. The f o l l o w i n g n o n l i n e a r d i f f e r e n t i a l e q u a t i o n i s
i n t e g r a t e d t o c a l c u l a t e Vs.

dVS
K(Vc-Vs) + CV C

I
I n t h i s e q u a t i o n , dF,/dVs i s t h e r a t e of change of s l i d i n g f r i c t i o n w i t h s l i d i n g
v e l o c i t y . T h i s c a n b e o b t a i n e d by d i f f e r e n t i a t i n g t h e h y s t e r e t i c t h e o r y g i v e n
above o r measured e x p e r i m e n t a l l y . F o o t p r i n t t r a n s i t t i m e , t, i s t a k e n as t h e
independent v a r i a b l e . The t i m e i s e q u i v a l e n t t o l o c a t i o n i n t h e f o o t p r i n t f o r
steady r o l l i n g .

312
 Figure 1 5 ( b ) shows a schematic diagram of t h e s l i d i n g t r e a d model and i t s
function i n converting a f r i c t i o n l e s s s l i d i n g v e l o c i t y d i s t r i b u t i o n i n t o s l i d i n g
v e l o c i t y i n f l u e n c e d by f r i c t i o n .

SL I DING
TREAD
MODEL

+I
vs BDa6_T-L”y””
I

I I
I
I
I
I I
I I

(b)
F i g u r e 15

313
 IV. FUTURE DIRECTIONS FOR CONTACT. SOLUTION ALGORITHMS

A s i n o t h e r areas of s o l i d mechanics, f u t u r e r e s e a r c h on c o n t a c t problems

w i l l be d i r e c t e d towards o b t a i n i n g s o l u t i o n s v a l i d f o r l a r g e d e f o r m a t i o n s . The
f i n i t e - e l e m e n t method seems p a r t i c u l a r l y w e l l s u i t e d f o r a p p l i c a t i o n t o c o n t a c t
problems and s p e c i a l elements have a l r e a d y been developed f o r t h i s purpose. Con-
tinuum mechanics r e s e a r c h on c o n t a c t problems should n o t be n e g l e c t e d , however.
A large-deformation c o n t a c t s o l u t i o n i n terms of c o n t i n u o u s f u n c t i o n s w i l l prove
v a l u a b l e i n t h e a n a l y s i s of c o n t a c t w i t h f r i c t i o n and t h e assessment of f r i c t i o n
l a w s now being proposed f o r deformable b o d i e s .

A t r u e c o n t a c t problem i s one i n which t h e c o n t a c t boundary and i n t e r f a c i a l

p r e s s u r e d i s t r i b u t i o n s a r e unknown a priori. A t p r e s e n t , i t a p p e a r s t h a t such
problems w i l l be d i s p l a c e m e n t p r e s c r i b e d : D e f l e c t i o n of t h e body toward t h e
c o n t a c t s u r f a c e i s s p e c i f i e d and i n t e g r a t i o n of t h e c a l c u l a t e d normal component
of t h e c o n t a c t p r e s s u r e g i v e s t h e r e s u l t a n t l o a d . Some e f f o r t should b e d i r e c t e d
towards a l o a d - s p e c i f i e d c o n t a c t problem, perhaps u t i l i z i n g t h e p r i n c i p l e of
s t a t i o n a r y complementary energy t o c a l c u l a t e t h e i n t e r f a c i a l p r e s s u r e d i s t r i b u -
t i o n subject t o t h e prescribed load c o n s t r a i n t . Validated s o l u t i o n s f o r f r i c -
t i o n l e s s c o n t a c t a r e e s s e n t i a l p r i o r t o i n c l u d i n g t h e e f f e c t of f r i c t i o n on t h e
contact solution.

I n the a n a l y s i s of f r i c t i o n , i t seems t h a t t h e s t u d y of r o l l i n g c o n t a c t a s
a s t e a d y - s t a t e problem h a s much t o o f f e r . A s friction i s a n i n t e r a c t i v e phe-
nomenon, a t l e a s t i n r e g a r d t o s l i d i n g v e l o c i t y , s o p h i s t i c a t e d a l g o r i t h m s a r e
needed t o g e n e r a t e t h e f r i c t i o n a l c o n t a c t s o l u t i o n from t h e s o l u t i o n f o r f r i c -
t i o n l e s s c o n t a c t (which w i l l undoubtably be t h e s t a r t i n g p o i n t ) .

The f o l l o w i n g schematic, f i g u r e 16, o u t l i n e s a p r o g r e s s i o n of r e s e a r c h on

c o n t a c t problems. L i n e a r s l i d i n g c o n t a c t i s excluded from t h e o u t l i n e as t h i s
i s u s u a l l y a t r a n s i e n t s i t u a t i o n l e a d i n g t o a c c e l e r a t e d wear a n d a b r a s i o n .
Much more w i l l be gained by r e s e a r c h focused on r o l l i n g c o n t a c t which w i l l , i n
any c a s e , i n c l u d e s l i d i n g .

31 4
 PROGRESSION OF RESEARCH ON FRICTIONAL CONTACT PROBLEMS

Large Deformation
Contact Problems

Influence Coefficient Lagrange Multiplier

Algorithms Algorithms
Finite-Element Methods Continuum Methods

+Frictionless Contact

Friction Laws
I
I
I
Surface
Characterization
J

I
r’
I

Static Contact with Friction

Frictional Contact

Rolling Contact with Friction

S t e a d y - S t a t e ProbZem

I
I I I
Assessment of Nonclassical
Friction Laws

Analysis of Wear

Figure 16

315
 REFERENCES

1. T i l l e r s o n , J . R . : and Haisler, W.E.: SAMMSOR I1 - A F i n i t e Element Program t o

Determine S t i f f n e s s and Mass Matrices of S h e l l s of R e v o l u t i o n . TEES-RPT-70-18,
Texas A&M U n i v e r s i t y , O c t . 1970.

2. Haisler, W.E.; and S t r i c k l i n , J . A . : SNASOR I1 - A F i n i t e Element Program f o r

t h e S t a t i s N o n l i n e a r A n a l y s i s of S h e l l s of R e v o l u t i o n . TEES-RPT-70-20, Texas
A&M U n i v e r s i t y , O c t . 1970.

3. Schapery, R.A.; and T i e l k i n g , J . T . : I n v e s t i g a t i o n of Tire-Pavement I n t e r -

a c t i o n During Maneuvering: Theory and R e s u l t s . Report No. FHWA-RD-78-72,
F e d e r a l Highway A d m i n i s t r a t i o n , J u n e 1977.

4. T i e l k i n g , J . T . ; and Schapery, R.A.: A Method f o r S h e l l C o n t a c t A n a l y s i s .

Computer Methods i n Applied Mechanics and E n g i n e e r i n g , Vol. 26, No. 2,
pp. 181-195, May 1981.

5. Green, A.E.; and Adkins, J . E . : L a r g e E l a s t i c Deformations. Oxford Univer-

s i t y P r e s s , London, 1960.

6. Levinson, M.: The A p p l i c a t i o n of t h e P r i n c i p l e of S t a t i o n a r y P o t e n t i a l

Energy t o Some Problems i n F i n i t e E l a s t i c i t y . J o u r n a l of Applied Mechanics,
Vol. 32, NO. 3, 1965, pp. 656-660.

7. B l a t z , P . J . ; and KO, W.L.: A p p l i c a t i o n of F i n i t e E l a s t i c Theory t o t h e

Deformation of Rubbery Materials. T r a n s a c t i o n s of t h e S o c i e t y of Rheology,
Vol. 6, 1962, pp. 223-261.

8. B e v e r i d g e , G.S-G.; and S c h e c t e r , R.S.: Optimization: Theory and P r a c t i c e .

M c G r a w - H i l l , 1970, p. 264.

9. T r e l o a r , L.R.G.: The P h y s i c s of Rubber E l a s t i c i t y , 2nd e d i t i o n . Oxford

U n i v e r s i t y P r e s s , London, 1958, p. 74.

10. R o t h e r t , H.; I d e l b e r g e r , H . ; J a c o b i , W . ; and Niemann, L.: On G e o m e t r i c a l l y

N o n l i n e a r C o n t a c t Problems w i t h F r i c t i o n . Computer Yethods i n Applied
Mechanics and E n g i n e e r i n g , Vol. 5 1 , 1985, pp. 139-155.

11. Oden, J . T . ; and P i r e s , E.B.: N o n l o c a l and N o n l i n e a r F r i c t i o n Laws and Varia-

t i o n a l P r i n c i p l e s f o r C o n t a c t Problems i n E l a s t i c i t y . J o u r n a l of A p p l i e d
Mechanics, Vol. 50, No. 1, March 1983, pp. 67-76.

12. Schapery, R.A.: A n a l y t i c a l Models f o r t h e Deformation and Adhesion Com-

p o n e n t s of Rubber F r i c t i o n . T i r e S c i e n c e and Technology, Vol. 6 , No. 1,
F e b r u a r y 1978, pp. 3-47.

316
 N89=24653
EXPLOITING SYMMETRIES IN THE MODELING
AND ANALYSIS OF TIRES

Ahmed K. Noor, C. M. Andersen, and John A. Tanner

NASA Langley Research Center
Hampton, Virginia 23665

INTRODUCTION

In recent years nonlinear analysis of static and dynamic problems has become
the focus of intense research efforts. This endeavor has prompted the development
of versatile and powerful finite-element discretization methods as well as of im-
proved numerical methods and software systems for nonlinear static and dynamic
analysis of structures and solids. One of the most challenging applications of
computational structural mechanics is the numerical simulation of the response of
aircraft tires during taxi, takeoff and landing operations. The commonly used
models for predicting the tire response are reviewed in Refs. 1 to 3 . Figure 1
lists some of the difficulties encountered in the modeling and analysis of tires and
their implications.

f h t , the tire is a composite structure composed of rubber and textile con-

stituents which exhibit anisotropic nonhomogeneous material properties. The lami-
nated carcass of the aircraft tire is thick enough to allow significant transverse
shear deformation. Second, the tire geometry is complicated and due to the presence
of unavoidable imperfections, the cross section is unsymmetric; and t k i r t d , the tire
is subjected to inflation pressure and to a variety of unsymmetric mechanical and
thermal loads which can result in large structural rotations and deformation, as
well as to a variation in the characteristics of the tire constituents. Moreover,
the detailed stress and temperature distribution in the tire may require the use of
three-dimensional finite elements in certain regions of the tire.

The aforementioned difficulties make the computational expense of the numerical

simulation of the tire response prohibitive. Hence, the need for the development of
modeling techniques and analysis methods to reduce this expense. Among the inodeling
strategies which show promise in reducing the cost is the exploitation of symmetries
and q u ~ i - ~ y m m ~ ~ by the tire response.
exhibited

i D IFF I CULTY IMPLICATION

o TIRE SHAPE
- COMPLEX GEOMETRY
- IMPERFECTIONS
o TIRE MATERIAL AND CONSTRUCTION o LARGE MODEL S I Z E
- ANISOTROPY
- NONHOMOGENEITY o COMPUTATIONALLY EXPEN IVE

o TIRE LOADS
- UNSYMMETRY
Figure 1
317
 OBJECTIVES AND SCOPE

The objectives of this paper are listed in Figure 2. They are

1) To review the different types of symmetry exhibited by the tire response

2) To present simple and efficient computational procedures for reducing the

size of the analysis model of tires

3 ) To discuss the potential of the proposed techniques and their application

to practical, quasi-symmetric tire problems

To sharpen the focus of the study, discussion is limited to two-dimensional

shell models of the tire, with elliptic cross-section and linear material response.
The analytical formulation is based on a Sanders-Budiansky-type shell theory with
transverse shear deformation, anisotropic material behavior, and geometric nonli-
nearities (moderate rotations) included (Refs. 4 and 5 ) . Displacement finite-
element models are used for the discretization. However, the procedure presented
herein is expected to be particularly useful for the analysis of three-dimensional
tire models.

0 B JECTl VES

0 REVIEW SYMMETRIES PRESENT I N TIRES

PRESENT TECHNIQUES FOR MODEL-S I ZE REDUCTIONS I N Q U A S I -SYMMETRI C

PRO B LEM S

0 D I S C U S S POTENTIAL APPLICATIONS OF TECHNIQUES

SCOPE

TWO-DIMENSIONAL SHELL MODELS

0 LINEAR MATERIAL RESPONSE

0 ELLIPTIC CROSS SECTION

0 DISPLACEMENT F. E. MODELS

Figure 2

318
I SYMMETRIES EXHIBITED BY TIRE RESPONSE

The three types of symmetry commonly exhibited by the tire response are shown
in Figs. 3 and 4. Also, the difference between the symmetries of orthotropic and
anisotropic tires are illustrated.

The first type of symmetry is the aXh! h y t ? ? m w exhibited by tires whose

geometric, material characteristics, and loading are independent of the circumferen-
tial coordinate, i.e., axisymmetric. The response of these tires will also be axi-
symmetric. An example of this situation is that of a tire subjected to uniform
inflation pressure. For an orthotropic tire the generalized displacements u, w and
$ 1 are axisymmetric and v and $2 are zero. In contradistinction, the five gener-
alized displacements are nonzero for an anisotropic tire.

Re&tection ( O / t m&tOhl h y m m ~with respect to coordinate planes is exhibited

by the response of orthotropic tires when subjected to loadings that exhibit the
same type of symmetry. Anisotropic tires, on the other hand, exhibit /tatationat ( O I L
h V U A i 0 n ) hynnntfky with respect to the center of symmetry shown in Fig. 4. If
these symmetries are exploited in the finite-element analysis, the size of analysis
model for an anisotropic tire is twice that of the corresponding orthotropic tire
(see, for example, Refs. 6 and 7).

When the external loading exhibits p h o d i c (on t A a n A ~ a t i o n c d )aymmd&.j, the

tire response also exhibits periodic symmetry. In orthotropic tires this is demon-
strated by the presence of more than two planes of reflection symmetry. In aniso-
tropic tires periodic symmetry is demonstrated by the presence of more than two
centers of rotational (inversion) symmetry. Again, the size of the analysis model
for an anisotropic tire with periodic symmetry is twice that of the corresponding
orthotropic tire (see Fiq. 4 ) .

A X I A L SYMMETRY - e. g. INFLATION PRESSURE

ANI SOTROP IC
U u, v, w, A*
V

ORTHOTROPIC
u, W # + I
V v,+2 = o

ANALY S IS MODEL

TYP CAL MER D l A N

Figure 3

319
 REFLECT1ON AND
ROTAT IONAL SYMMETRIES --_ ..
w t 0 RTHOTROP IC
(REFLECT1ON AND
RCITAT13NAL SYMMETRIES)

s +

ANISOTROPIC
(ROTAT IO NAL SYMMETRY
ONLY)

I I I
PER I OD I C SYMMETRY 1

Note: Shading shows analysis region.

Figure 4

320
 QUASI-SYMMETRIC PROBLEMS

Figure 5 lists a number o f quasi-symmetric problems. The three basic quasi-

symmetric problems are the ones for which either the tire material, loading, or geo-
metry are not symmetric, but the other two are symmetric. Here material anisotropy
is considered to be a source of reflection unsymmetry. The unsynmetry in geometry
can be caused by the presence of unsymmetric imperfections in the tire. In a prac-
tical situation combination of the three basic types of unsymmetry can exist. In the
present study, a computational procedure is presented for reducing the size of the
analysis models for quasi-symmetric problems of tires to those of the corresponding
symmetric problems.

CASE MATER I A L LOAD ING GEOMETRY

- -

SYMMETR IC A X I A L L Y SYMMETRIC

I1

111

Figure 5

321
 BASIC IDEA OF MODEL-REDUCTION TECHNIQUE
FOR QUASI-SYMMETRIC PROBLEMS

Figure 6 lists the two key elements of the model reduction technique when ap-
plied to the finite-element analysis of anisotropic tires with symmetric geometry
subjected to symmetric loading. The two elements are: a) decomposition of the
stiffness matrix into the sum of an orthotropic and nonorthotropic (anisotropic)
parts; and b) successive application of the finite-element method and the classical
Rayleigh-Ritz technique. The finite-element method is first used to generate few
global approximation vectors (or modes). Then the amplitudes of these modes are
computed by using the Rayleigh-Ritz technique.

A N I S O T R O P I C MATERIALS

0 DECOMPOSITION OF MATRICES I N GOVERNING FIKITE-ELEMENT EQUATIONS

INTO ORTHOTROPIC AND NONORTHOTROPIC PARTS

0 SUCCESSIVE A P P L l CATIO\! OF
- FINITE-ELEMENT A N A L Y S I S TO GENERATE A FEW GLOBAL A P P R O X I M A T I O N
VECTORS U S I N G SAME S I Z E MODEL A S FOR THE ORTHOTROPIC CASE

- C L A S S I C A L RAYLEIGH-RITZ TECHNIQUE TO COMPUTE AMPLITUDES OF

A P PROX I M A T ION VECTORS

Figure 6

322
 MATHEMATICAL FORMULATION

I Figure 7 outlines the mathematical formulation for the proposed model-reduction

technique when applied to linear problems of anisotropic tires. The global stiff-
ness matrix [K] is decomposed into orthotropic and nonorthotropic matrices [KIo
and [K],, respectively. The nonorthotropic matrix [Kla is multiplied by a tracing
parameter X which identifies & the nonorthotropic material coefficients. The
original finite-element equations correspond to the case A = l .

The global approximation vectors are selected to be the solution corresponding

to X=O (zero nonorthotropic matrix) and its various-order derivatives with respect
to X (path derivatives). The path derivatives are obtained by successive differen-
tiation of the governing finite-element equations with respect to A . Note that the
coefficient matrix appearing on the left-hand sides of the recursion formulas is
[KIo, and that the size of the analysis region used in evaluating each of the global
approximation vectors is the same as that for the orthotropic case (X=O).

The vector {XIo and its path derivatives are now chosen as approximation vec-
tors, and the vector of nodal displacements for the anisotropic tire, {XI, is ex-
pressed as a linear combination of these vectors. A Rayleigh-Ritz technique is used
to replace the original finite-element equations by a reduced system of equations in
the unknown parameters, { $ I , which represent the amplitudes of the global approxi-
mation vectors.

GOVERNING FINITE-ELEMENT EQUATIONS

[K] I X ; = j P ;
LET [K] = [K], + h[K],
h = TRACING PARAMETER

BAS1 S REDUCTION

WHERE

I1 V jt = VECTOR OF AMPLITUDES OF MODES

GLOBAL APPROXIMATION VECTORS

[K], ;xio = IPI

I OBTAINED FROM SAME SIZE MODEL
A S FOR THE ORTHOTROPIC CASE

REDUCED EQUATIONS ( E R A Y L E I G H - R ITZ TECHNIQUE)

Figure 7

323
 APPLICATION TO ANISOTROPIC TIRES

As a first application of the proposed model-reduction technique consider the

anisotropic tire subjected to the symmetric localized loading shown in Figure 8. As
shown in Figure 4, the tire response exhibits rotational (inversion) symmetry. It
does not exhibit reflection symmetry, and therefore, the analysis model consists of
half the tire. As can be seen from the contour plots of Figure 8, all the global
approximation vectors exhibit reflection symmetry (and antisymmetry) and therefore,
they can be obtained by analyzing only one quadrant of the tire (same size model as
that used for analyzing the corresponding orthotropic tire).

GLOBAL A P P R O X I M A T I O N VECTORS ARE ALL

DOUBLY SYMMETR I C AND THE SYMMETRY
PATTERNS ARE KNOWN

Figure 8

324
 ACCURACY AND CONVERGENCE O F SOLUTIONS OBTAINED
BY PROPOSED TECHNIQUE

A n i n d i c a t i o n of t h e a c c u r a c y and convergence of t h e s o l u t i o n s o b t a i n e d by t h e
proposed model r e d u c t i o n t e c h n i q u e i s given i n Fig. 9. The s t a n d a r d of comparison
i s t a k e n t o be t h e d i r e c t f i n i t e - , e l e m e n t s o l u t i o n of t h e a n i s o t r o p i c t i r e . As can
be seen from Fig. 9, t h e s o l u t i o n s o b t a i n e d by t h e proposed t e c h n i q u e are h i g h l y
a c c u r a t e even when a small number of approximation v e c t o r s are used. Numerical ex-
periments have shown t h a t for h i g h l y a n i s o t r o p i c t i r e s no more than f i v e a p p r o x i -
mation v e c t o r s are needed.

-}
1.0

W Full 1
l o

1 I 1 I 1 I I

U = TOTAL STRAIN ENERGY

-
U

FULL
Oa5 t
1 2 3 4 5 6 7
NUMBER OF APPROX. VECTORS

Figure 9

325
 APPLICATION OF MODEL-REDUCTION TECHNIQUE TO PREDICTING NONLINEAR
RESPONSE OF ORTHOTROPIC TIRES SUBJECTED TO ASYMMETRIC LOADING

As another application of the proposed model-reduction technique, consider the

orthotropic tire subjected to the localized loading shown in Fig. 10. The analysis
model consists of one quadrant of the tire. For linear problems, the decomposition
of the loading into symmetric and antisymmetric components and the consequent reduc-
tion of the size of the analysis model to one-octant of the tire are well known.
However, it is generally assumed that such a decomposition is not useful for non-
linear problems in which the principle of superposition is not applicable.

The foregoing model-reduction technique can be used to reduce the size of the
analysis model to one octant of the tire. This is accomplished by: 1) decomposi-
tion of the given loading into symmetric and antisymmetric components with load
parameters p1 and p2, respectively; and 2) use of the multiple-parameter reduction
technique described in Refs. 8 and 9. The vector of nodal displacements of the tire
is approximated by a linear combination of few global vectors or modes. These vec-
tors are selected to be the various-order derivatives of the displacment vector with
respect to load parameters p1 and p2. Each of these vectors is evaluated at
p1=p2=0. Figure 10 shows that the global approximation vectors exhibit periodic
symmetries (and/or antisymmetries), and therefore, each vector can be evaluated using
only one octant of the tire.

p1 p2

A S A S A

A S A S A
W

ASYMMETRIC LOAD I NG U
S S S S S
S and A refer to lines
of reflection symmetry
and antisymmetry, - '0
r e spect ive Iy.

Figure 10

326
 POTENTIAL OF PROPOSED MODEL-REDUCTION TECHNIQUE

The proposed model reduction technique appears to have high potential for anal-
ysis of practical tire problems. In particular, in the presence of combinations of
unsymmetries in the material (viz. anisotropy), the geometry or loading on the tire,
the same size model can be used as for symmetric material (viz., orthotropic), geo-
metry and loading (Fig. 1 1 ) . This is accomplished by the introduction of a tracing
parameter for each of these unsymmetric effects; and the successive application of a
reduction method with each of these parameters. The global approximation vectors
are selected to be the various-order derivatives with respect to the tracing para-
meters.

0 ASYMMETRIC LOAD1 NG AND A N I SOTROPY

0 ANISOTROPY AND I N I T I A L UNSYMMETRIC IMPERFECTIONS

0 ASYMMETRIC LOADING AND I N I T I A L UNSYMMETRIC IMPERFECTIONS

THREE-DIMENSIONAL MODELS

MIXED FI NITE-ELEMENT MODELS

F i g u r e 11

327
 SUMMARY

In summary, a computational procedure is presented for reducing the size of the

analysis models of tires having unsymmetric material, geometry and/or loading. The
two key elements of the procedure when applied to anisotropic tires are: a) decom-
position of the stiffness matrix into the sum of an orthotropic and nonorthotropic
parts; and b) successive application of the finite-element method and the classical
Rayleigh-Ritz technique. The finite-element method is first used to generate ~ Q L U
global approximation vectors (or modes). Then the amplitudes of these modes are
computed by using the Rayleigh-Ritz technique.

The proposed technique has high potential for handling practical tire problems
with anisotropic materials, unsymmetric imperfections and asymmetric loading. It is
also particularly useful for use with three-dimensional finite-element models of
tires.

MODEL-SIZE REDUCTION PROCEDURE PRESENTED FOR A N A L Y S I S OF TIRES

BASED O N

- DECOMPOSITION OF MATRICES

- SUCCESSIVE A P P L I C A T I O N S OF F I N I T E ELEMENTS AND

RAYLEIGH-RITZ TECHNIQUES

0 PROPOSED TECHNIQUE H A S H I G H POTENTIAL FOR HANDLING

- ASYMMETR I C LOAD I NG

- ANISOTROPIC MATERIALS

- UNSYMMETRIC IMPERFECTIONS

- THREE D I M E N S I O N A L MODELS
Figure 1 2

328
 REFERENCES

1. C l a r k , S. K. (ea.), "Mechanics of Pneumatic T i r e s , " U . S . D e p t . o f T r a n s p o r t a -

t i o n , N a t i o n a l Highway T r a f f i c S a f e t y A d m i n i s t r a t i o n , Washington, D.C., 1981.

2. Tanner, J. A. ( c o m p i l e r ) , " T i r e Modeling," NASA CP-2264, 1983.

3. Noor, A. K. and Tanner, J. A., "Advances and Trends i n t h e Development of Compu-

t a t i o n a l Models f o r T i r e s , " Computers and S t r u c t u r e s , 1985.

4. S a n d e r s , J. L., " N o n l i n e a r T h e o r i e s f o r Thin S h e l l s , " Q u a r t e r l y of Applied

Mathematics, V o l . 21, A p r i l 1963, pp. 21-36.

5. Budiansky, B., "Notes on N o n l i n e a r S h e l l Theory," J o u r n a l of Applied Mechanics,

Vol. 35, June 1968, pp. 393-401.

6. Noor, A. K. and Camin, R. A., "Symmetry C o n s i d e r a t i o n s f o r A n i s o t r o p i c S h e l l s , "

Computer Methods i n Applied Mechanics and E n g i n e e r i n g , V o l . 9, 1976, pp. 317-
335.

7. Noor, A. K., Mathers, M. D. and Anderson, M. S., " E x p l o i t i n g Symmetries f o r

E f f i c i e n t P o s t b u c k l i n g A n a l y s i s of Composite P l a t e s , " A I A A J o u r n a l , V o l . 15,
J a n u a r y 1977, pp. 24-32.

8. Noor, A. K. and P e t e r s , J. M., "Recent Advances i n Reduction Methods f o r I n s t a -

b i l i t y A n a l y s i s of S t r u c t u r e s , " Computers and S t r u c t u r e s , V o l . 16, 1983, pp. 67-
80.

9. N o o r , A. K., Andersen, C. M. and Tanner, J. A., "Mixed Models and Reduction

Techniques f o r Large R o t a t i o n , N o n l i n e a r A n a l y s i s of S h e l l s of R e v o l u t i o n w i t h
A p p l i c a t i o n t o T i r e s , " NASA TP-2343, October 1984.

329
 PANEL DISCUSSION, Wednesday, June 19 - TIRE HODEL-ING
J . A. Tanner, NASA Langley Research Center: You've heard f i v e papers t h i s morn-
i n g i n t h e general area o f t i r e modeling. However, I b e l i e v e t h a t t h e c o n t e n t
o f t h e papers i s a c t u a l l y broader than t i r e modeling and t h a t we a r e r e a l l y
d e a l i n g w i t h t h e more general i s s u e s a s s o c i a t e d w i t h computational s t r u c t u r a l
mechanics. I was r u s h i n g some o f t h e a u t h o r s t h i s morning, t r y i n g t o r u s h them
t h r o u g h t h e i r p r e s e n t a t i o n s , and I am concerned t h a t t h e y may n o t have made a l l
t h e p o i n t s they wanted t o make. I n an e f f o r t t o make amends, I would l i k e t o
ask each a u t h o r t o t a k e about 2 o r 3 minutes t o summarize any p o i n t s t h a t he
would l i k e t o emphasize and t o address p o i n t s n o t made i n h i s p r e s e n t a t i o n .
A f t e r t h e summary statements, t h e f l o o r w i l l be opened f o r a d d i t i o n a l questions.
I am g o i n g t o s t a r t immediately t o my r i g h t w i t h D r . Tabaddor who gave t h e f i r s t
paper t h i s morni ng.

Farhad Tabbador, 9. F. Goodrich Company: I would j u s t l i k e t o r e i t e r a t e t h e

point on t h e m a t e r i a l p r o p e r t i e s o f composites. There i s a g r e a t need f o r
e f f o r t s i n c h a r a c t e r i z a t i o n o f composites, and I t h i n k t h a t t h e o r i e s developed
o v e r t h e p a s t decade o r so f o r r i g i d composites do n o t q u i t e apply t o s o f t
composites. I would recommend t h a t maybe some s o r t o f g r a n t e f f o r t m i g h t be
i n i t i a t e d by NASA t o do t h i s b a s i c work.

Tanner: D r . Padovan, i n y o u r p r e s e n t a t i o n you d e a l t w i t h t h e a r c h i t e c t u r a l

c h a r a c t e r i s t i c s o f t h e new computers t h a t w i l l be coming on l i n e . I thought
maybe you would have some more t h i n g s you wanted t o say on t h a t subject.

J o e Padovan, U n i v e r s i t y o f Akron: I t h i n k we would l i k e t o see t h e a v a i l a b i l i t y

of some r e a l number c r u n c h i n g power so t h a t we c o u l d r e a l l y push models t o t h e
l i m i t . Secondly, I t h i n k t h a t we a r e a l l t o o i n t e r e s t e d i n t h e l o c a l d e t a i l
i n f o r m a t i o n i n t h e model and I t h i n k we a r e n o t r e a l l y t r y i n g t o c a p t u r e t h e
t r u e essence o f what a t i r e does, and t h a t i s t o p r o v i d e good r i d e c h a r a c t e r i s t i c s .
I t h i n k more e f f o r t s h o u l d be p u t towards c h a r a c t e r i z i n g t h e o v e r a l l f e a t u r e s of
t h e t r u e s t r u c t u r e and l e a v e most o f t h e s t r e s s a n a l y s i s t o t h e t i r e i n d u s t r y
where i t belongs. Now I am speaking f o r m y s e l f . I t h i n k t h e dynamic c h a r a c t e r -
i s t i c s o f t h e t i r e a r e most i m p o r t a n t . More e f f o r t has t o be p u t i n t o t h a t k i n d
o f modeling. T h i r d l y , more money has t o be p u t i n t o t h e e f f o r t i n general. If
you r e a l l y want m a j o r changes i n t h e t e c h n o l o g i e s t h a t have been i n p l a c e f o r 40
y e a r s then y o u have t o spend a g r e a t deal o f money.
331
PRECEDING PAGE U N M ROT FlLMED
Tanner: D r . Oden, i n y o u r p r e s e n t a t i o n t h i s m o r n i n g y o u s p e n t some t i m e t a l k i n g
a b o u t e r r o r a n a l y s i s . The t i r e i n d u s t r y w i l l n o t u s e any a n a l y s i s t o o l s u n l e s s
t h e i r s o l u t i o n accuracies a r e v e r i f i e d . I imagine t h a t e r r o r a n a l y s i s i s c r i t i -
c a l t o t h e v e r i f i c a t i o n o f any m o d e l i n g s t r a t e g y . Would y o u c a r e t o comment
f u r t h e r on t h i s s u b j e c t ?

J . T i n s l e y Oden, The C o m p u t a t i o n a l Mechanics Company, I n c . : It i s g r a t i f y i n g t o

see t h e word " e r r o r " used more and more i n m e e t i n g s such as t h i s because I
b e l i e v e t h e u s e r s o f f i n i t e e l e m e n t methods have f i n a l l y r e a c h e d a p o i n t where
t h e y a r e b e g i n n i n g t o demand some a s s u r a n c e as t o t h e r e 1 i a b i l i t y and q u a l i t y of
solutions. We have s p e n t 10 y e a r s i n d e v e l o p i n g e l e m e n t s , t h e n a l g o r i t h m s , and
now machines t o h a n d l e t h e a l g o r i t h m s , b u t how good a r e t h e answers? Methods
a r e e v o l v i n g t h a t w i l l someday a l l o w us t o answer t h a t q u e s t i o n w i t h some degree
o f confidence. However, t h e r e a r e methods a t o u r d i s p o s a l now t h a t a r e based on
e s t i m a t i o n s o f e r r o r s t h a t w i l l a l l o w us t o i n c o r p o r a t e a d a p t i v e s t r a t e g i e s i n t o
very general f i n i t e element s t r u c t u r a l c a l c u l a t i o n s . I h e a r t i l y support the
a c t i v i t y i n t h i s a r e a . I t h i n k we a r e g o i n g t o see some e r r o r e s t i m a t i o n and
a d a p t i v i t y become an i n t e g r a l p a r t of f i n i t e e l e m e n t c a l c u l a t i o n s i n t h e n e a r
future.

Tanner: D r . T i e l k i n g , t h e c e n t r a l t h a n e o f y o u r p a p e r was c o n t a c t a l g o r i t h m s
and y o u p r e s e n t e d r e s u l t s f r o m s t a t i c c o n t a c t , b o t h w i t h and w i t h o u t f r i c t i o n .
I was p a r t i c u l a r l y i n t e r e s t e d i n y o u r l a s t comment, however, when y o u i n d i c a t e d
t h a t f u t u r e s t u d i e s o u g h t t o be concerned w i t h r o l l i n g c o n t a c t . I thought YOU

m i g h t w a n t t o expand a l i t t l e on t h a t t h o u g h t .

J . T. T i e l k i n g , Texas A&M U n i v e r s i t y : Yes, I w o u l d l i k e t o say a l i t t l e b i t

more a b o u t t h e s t u d y o f c o n t a c t w i t h f r i c t i o n . O f course, a l l c o n t a c t i n v o l v e s
friction. The a s s u m p t i o n o f f r i c t i o n l e s s c o n t a c t i s a n a l o g o u s t o t h e a s s u m p t i o n
o f a r i g i d body--we r e a l l y d o n ' t have such s t r u c t u r e s . I feel that the f i n i t e
e l a n e n t models c e r t a i n l y do have a p l a c e i n t h e s t u d y o f f r i c t i o n , b u t f r i c t i o n
f o r c e s v a r y c o n t i n u o u s l y t h r o u g h t h e i n t e r f a c e , t h e y depend, o f course, on t h e
v a r i a t i o n s o f s e v e r a l v a r i a b l e s , t y p i c a l l y i n t e r f a c i a l p r e s s u r e , normal p r e s -
sure, and s l i d i n g v e l o c i t y . I t i s now r e c o g n i z e d t h a t t h e r e i s s l i d i n g v e l o c i t y
everywhere i n t h e t i r e c o n t a c t r e g i o n . The a s s u m p t i o n o f r e g i o n s o f adhesion
and s l i d i n g a r e good f o r a p p r o x i m a t i o n s , b u t n o t f o r more a c c u r a t e r e s u l t s . I
f e e l t h a t t h e c o n t i n u o u s s l i d i n g models s h o u l d n o t be n e g l e c t e d . One o f my
m o t i v a t i o n s f o r d o i n g t h a t p l a i n s t r a i n c y l i n d e r p r o b l e m was t o g e t a l a r g e

332
deformation c o n t a c t s o l u t i o n which c o u l d be used i n t h e s t u d y o f f r i c t i o n .
Therefore, I am very pleased t o see t h a t you a r e n o t r e s t r i c t i n g t h e CSM
a c t i v i t y t o f i n i t e element methods and t h a t you w i l l e n t e r t a i n a l t e r n a t i v e
methods o f s o l u t i o n . F i n a l l y , I a p p r e c i a t e y o u r concern f o r u n i v e r s i t y funding.
F o r e i g n governments have been sending t h e i r b e s t s t u d e n t s t o our u n i v e r s i t i e s
f o r a l o n g t i m e and we a r e very happy t o have them, b u t we would l i k e t o be a b l e
t o fund our own students. I t h i n k o r g a n i z a t i o n s such as NASA c o u l d c e r t a i n l y do
a g r e a t deal t o see t h a t s t u d e n t s from our own c o u n t r y can be funded as w e l l as
the top students f r a n foreign countries.

Tanner: Now I w i l l l e t t h e f i r s t two a u t h o r s on t h e l a s t paper have t h e i r say.

A . K . Noor, George Washington U n i v e r s i t y : I would l i k e t o make a few comments

about t h e paper. I t h i n k f i r s t of a l l t h i s was r e a l l y a study t h a t aims a t
developing a general computational s t r a t e g y . I n very general terms, when we a r e
c o n f r o n t e d w i t h a complex problem, we generate t h e s o l u t i o n t o t h a t problem
s t a r t i n g from a s i m p l e r problem o r perhaps even from a h e i r a c h y o f s i m p l e r prob-
lems. I t h i n k t h i s i s v e r y much an e n g i n e e r i n g - t y p e approach. These s i m p l e r
problems w i l l each be a s s o c i a t e d w i t h a c o n t r o l parameter. What we a r e d o i n g i s
e s s e n t i a l l y u s i n g a p e r t u r b a t i o n technique. What we have heard i n t h e l a s t
p r e s e n t a t i o n was an e x t e n s i o n o f t h e c l a s s i c a l p e r t u r b a t i o n technique which i s
l i m i t e d t o small v a l u e s o f t h e p e r t u r b a t i o n parameter. The t e c h n i q u e t h a t we
have developed i s n o t l i m i t e d t o small values o f t h e p e r t u r b a t i o n parameter; i t
works w e l l even i f you have very l a r g e values f o r t h e p e r t u r b a t i o n parameter.
Our example was f o r a h i g h l y a n i s o t r o p i c t i r e . The p e r t u r b a t i o n parameter was
i n d i c a t i v e o f t h e degree of a n i s o t r o p y . The technique, I t h i n k , can be essen-
t i a l l y summed up as a combination o f o p e r a t o r s p l i t t i n g , where we d i d t h e s p l i t -
t i n g on t h e m a t r i c e s , and a m u l t i p l e parameter extended p e r t u r b a t i o n . The o t h e r
comment which I would l i k e t o make i s t h a t y o u g e t t h e s e n s i t i v i t y i n f o r m a t i o n
f r e e because a l l t h a t i n f o r m a t i o n i s needed i n t h e d a i l y use o f t h e equations.
I n o t h e r words, t h e s e n s i t i v i t y i n f o r m a t i o n i s a by p r o d u c t o f t h e s o l u t i o n
algorithm. The f i n a l comment i s t h a t t h e CPLJ t i m e was reduced by a f a c t o r Of

e i g h t o r more f o r t h e problems t h a t you have seen b u t t h a t r e d u c t i o n f a c t o r can

be much h i g h e r f o r dynamic problems.

333
C. M. Anderson, C o l l e g e o f W i l l i a m and Mary: Well , t o paraphrase o u r paper j u s t
one o t h e r way, i f the geometry, m a t e r i a l p r o p e r t i e s , and l o a d i n g i n a s t r u c t u r e
e x h i b i t a k i n d o f symmetry, t h e n you may w e l l e x p e c t t h e s o l u t i o n w i l l e x h i b i t
t h a t same symmetry. I t i s very easy t o t a k e t h a t symmetry i n t o account and
reduce y o u r model s i z e . Nhat we a r e saying i s t h a t one can now i n t r o d u c e a
h i g h e r symmetry group i n t o t h e problem. F o r n o n l i n e a r problems which do n o t
have t h a t f u l l symmetry, one can f i n d b a s i s f u n c t i o n s which r e p r e s e n t t h e h i g h e r
symmetry group. The s o l u t i o n t h a t we a r e l o o k i n g f o r w i l l be a l i n e a r combina-
t i o n o f those b a s i s f u n c t i o n s . Therefore, we can s t i l l achieve t h e reduced
model s ze even though t h e f u l l symmetry i s n o t p r e s e n t i n t h e f i n a l s o l u t i o n
t h a t we a r e 1 ooki ng f o r .

Tanner: I have a few c l o s i n g comments. F o r t h e p a s t 2 y e a r s , NASA and t h e U.S.

t i r e i n d u s t r y have been i n v o l v e d i n a j o i n t program c a l l e d t h e N a t i o n a l T i r e
Modeling Program. The program p r o v i d e s a forum f o r t e c h n i c a l d i s c u s s i o n s w i t h -
o u t i n f r i n g i n g on t h e p r o p r i e t a r y r i g h t s o f t h e i n d i v i d u a l t i r e companies. The
o b j e c t i v e o f t h e program i s t o develop t i r e a n a l y s i s t o o l s t o h e l p s t r e a m l i n e
t h e t i r e design process. The number o f t i r e i n d u s t r y r e p r e s e n t a t i v e s i n v o l v e d
i n t h i s workshop i s an i n d i c a t i o n o f t h e s t r o n g i n d u s t r y s u p p o r t f o r t h e
p r og ram.

The l a s t p o i n t t h a t I want t o make i n v o l v e s t h e r e l a t i o n s h i p between t i r e model-

i n g s t u d i e s and computational s t r u c t u r a l mechanics i n general. The paper t h a t
D r . Tabaddor presented t h i s morning g i v e s an i n d i c a t i o n o f t h e m a t e r i a l p r o p e r t y
concerns t h a t e x i s t i n t h e t i r e i n d u s t r y . A1 though some o f t h e concerns expres-
sed by D r . Tabaddor a r e unique t o t h e r u b b e r - c o r d composites, I b e l i e v e t h a t
many of h i s m a t e r i a l p r o p e r t y concerns a r e common t o b o t h s o f t and r i g i d compos-
ites. I t h i n k t h a t many of t h e t e c h n i c a l c h a l l e n g e s t h a t we f a c e i n our t i r e
modeling e f f o r t s a r e common t o each o f t h e d i s c i p l i n e s r e p r e s e n t e d h e r e today.
T h i s CSM workshop i s a g r e a t way o f g e t t i n g a l o t o f people t o g e t h e r t o f i n d o u t
what we have i n common i n our t e c h n i c a l p u r s u i t s . We should do t h i s more o f t e n .

Any q u e s t i o n s from t h e audience?

Barna A. Szabo, Washinaton U n i v e r s i t y : I ' d l i k e t o address my q u e s t i o n t o P r o f .

Oden. You mentioned t h e e r r o r e s t i m a t i o n which uses t h e e r r o r i n d i c a t o r s t h a t
YOU presented. I understand t h a t you a r e computing t h e c o n t r i b u t i o n o f e r r o r i n
t h e energy norm o f each element t o t h e t o t a l e r r o r o f approximation. You would

334
l i k e t o see t h a t each element c o n t r i b u t e s about t h e same e r r o r i f t h e mesh i s
p r o p e r l y designed. We found t h a t i n some cases we can have very small e r r o r s i n
t h e energy norm, something of l e s s t h a n 1 p e r c e n t and, y e t , have more t h a n 20
percent e r r o r i n equilibrium. F o r example, i f you c u t t h e s t r u c t u r e o f f t h e
supports and recompute t h e r e a c t i o n s from t h e f i n i t e element s o l u t i o n u s i n g t h e
d i r e c t method, we s t i l l can have a f a i r l y l a r g e e r r o r i n e q u i l i b r i u m . I t i s
i m p o r t a n t t o l o o k a t t h e e r r o r i n energy norm b u t i t i s a l s o i m p o r t a n t t o check
other quantities o f interest. Have you observed t h e same t h i n g ? I am p a r t i c u -
l a r l y fond of t h e e q u i l i b r i u m check because i t can be appl i e d t o n o n l i n e a r prob-
lems as w e l l as t o l i n e a r problems and a l s o i t i s meaningful t o engineers. One
l a s t comment, i f an element o r group o f elements i s found t o be o u t o f e q u i l i b -
r i u m t h e n t h e p o s s i b i l i t y o f p o l l u t i o n o f t h a t e r r o r a r i s e s , whereas i f t h e group
O f elements i s i n e q u i l i b r i u m t h e n a c c o r d i n g t o S t . Venant's p r i n c i p l e you would

expect t h e e r r o r t o be l o c a l i z e d . T h i s i s a q u a l i t a t i v e view, b u t , neverthe-

l e s s , i t has a c e r t a i n c o n v i n c i n g f l a v o r t o i t t h a t engineers w i l l a p p r e c i a t e .

-
Oden: The presence o r absence o f e q u i l i b r i u m g l o b a l l y has t o do w i t h t h e r e s i d -
ual i n a s o l u t i o n when you a r e t a l k i n g about l i n e a r o r n o n l i n e a r e l l i p t i c prob-
lems. The c a l c u a t i o n o f t h e e r r o r i n d i c a t o r does compute a l o c a l r e s i d u a l . It
computes a l o c a l r e s i d u a l t h a t e n t e r s t h e r i g h t - h a n d s i d e o f t h e equation. The
problem i s t h a t t h e r e s i d u a l by i t s e l f i s n o t an i n d i c a t i o n o f e r r o r . I f y o u r
r e s i d u a l i s l a r g e , chances a r e t h e e r r o r i s l a r g e , b u t t h e converse i s n o t t r u e .
You can have a system o f f o r c e s i n e q u i l i b r i u m . and have an enormous e r r o r . That
i s t h e c o m p l i c a t i n g f e a t u r e o f these k i n d s o f e r r o r e s t i m a t i o n . One must s e l e c t
an i n d i c a t i o n o f t h e e r r o r t h a t has t h e r i g h t a s y m t o t i c b e h a v i o r such t h a t as
t h e i n d i c a t o r goes t o zero, so a l s o does t h e e r r o r go t o zero. The energy norm
exhibits t h i s characteristic. The p o i n t s you r a i s e a r e s u b t l e b u t v e r y impor-
t a n t issues i n c a l c u l a t i n g e r r o r estimators. These p o i n t s a r e t h e c o m p l i c a t i n g
f e a t u r e i n t h i s p a r t i c u l a r k i n d o f s t r a t e g y because one has t o design an e r r o r
i n d i c a t o r w i t h the proper asymtotic properties. There a r e many methods f o r
e s t i m a t i n g l o c a l e r r o r , b u t t h e one I am speaking o f now, wh l e expensive is
t h e b e s t i n t h a t i t overcomes t h e very problems t h a t you a r e r e f e r r i n g t o It
produces a r o b u s t and a c c u r a t e e s t i m a t i o n of t h e l o c a l e r r o r which does, n
f a c t , v a n i s h a t a r a t e equal t o t h e v a n i s h i n g o f t h e r e s i d u a as t h e mesh i s
r e f i ned.

335
G e r a l d Goudreau, Lawrence Livermore N a t i o n a l L a b o r a t o r y : I'm new t o t i r e model-
i n g and t h e one t h i n g I d i d n ' t g e t o u t o f t h i s morning was a f e e l i n g f o r how YOU

g e t y o u r composite m a t e r i a l models o f a t i r e . I g e t a f e e l i n g t h a t you go o u t

and measure r u b b e r p r o p e r t i e s , even g e t f i n i t e d e f o r m a t i o n o r f i n i t e s t r a i n
models of rubber, you do l o a d d e f l e c t i o n t e s t s on cords and, y e t , somehow t h i s
i s a l l p u t t o g e t h e r i n t o some k i n d o f a model which may be o r t h o t r o p i c , maybe
a n i s o t r o p i c , b u t who's doing what? Anyone can respond.

Joe Padovan: Measurements o f t h e m a t e r i a l p r o p e r t i e s o f t i r e c o n s t i t u e n t s have

been g o i n g on f o r many years. The i n d u s t r y does n o t measure rubber p r o p e r t i e s
and c o r d p r o p e r t i e s s e p a r a t e l y b u t r a t h e r t h e aggregate c o r d / r u b b e r composite.
T y p i c a l l y , t h e r u l e t h a t has been used f o r m i x t u r e s i s H a l p i n - T s a i , although
t h i s does n o t f u l l y apply i n a n o n l i n e a r s e t t i n g l i k e t h e t i r e .

Goudreau: Are these small s t r a i n models then?

Padovan: There a r e n o t r e a l l y any 1 arge d e f o r m a t i o n c o n s t i t u t i v e models except

f o r some v e r y i s o l a t e d m a t e r i a l s . I n t h e case o f r u b b e r / c o r d composites, t h e r e
c e r t a i n l y a r e none y e t . B a s i c a l l y what you do i s assume H a l g i n - T s i a o r whatever
l i n e a r model you a r e u s i n g works and then assume l a r g e s t r a i n s and repeal
K i r c h h o f f ' s assumptions on t h e o t h e r s i d e o f t h e equation.

Goudreau: How a r e y o u c h a r a c t e r i z i ng t h i s a n i s o t r o p y ?

Padovan: The a n i s o t r o p y is hand1 ed t h r o u g h s t a n d a r d 1aminate theory.

Goudreau: You're t e s t i n g t h i n g s lamina b y lamina then?

Padovan: Yes, lamina by lamina.

Marion G. P o t t i n g e r , B. F. Goodrich Company: I would l i k e t o make some

comments. S t r u c t u r a l mechanics l o v e t o c a l c u l a t e t h e t i r e s t r e s s e s and s t r a i n s ,
b u t t i r e u s e r s a r e concerned w i t h performance c h a r a c t e r i s t i c s which a r e v e r y
d i f f e r e n t things.
Since t h e cords a r e p r e t t y b i g and you a r e t r e a t i n g t h e t i r e
as a continuum, t h e r e a r e some d e f i n i t i v e l i m i t a t i o n s on what you can do. We
a r e w o r r y i n g about b e i n g exact, b u t how does t h i s exactness f i t i n w i t h t h e
t r i e d and proven design r u l e s o f thumb. The o l d design r u l e s o f thumb a r e very

336
i m p o r t a n t t o design people and so a r e t h e methods o f q u a l i t a t i v e c h a r a c t e r i z a -
t i o n . F i n a l l y , I would l i k e t o mention t h e q u e s t i o n o f combining t h e a n a l y t i c a l
and t h e experimental and t h e q u e s t i o n o f experimental v e r i f i c a t i o n . N o t o n l y do
you have t o s a t i s f y y o u r s e l f t h a t t h i n g s r e a l l y work, you a l s o have t o s a t i s f y
1awyers.

Richard B. Nelson, U.C.L.A.: I am a u s e r o f f i n i t e elements, b u t I a l s o use

c a r s and t i r e s . My e x p e r i e n c e i s t h a t heat, abuse, and, d e l a m i n a t i o n a r e what
kill tires. I wonder i f t h e r e has been a t t e n t i o n g i v e n i n t h e i n d u s t r y t o t h e
need t o develop models which a r e capable o f p r e d i c t i n g h e a t b u i l d - u p and what
t h a t does i n t h e way o f g e n e r a t i n g premature t i r e delaminations.

Tanner: O f course, we a r e a l l extremely i n t e r e s t e d i n t h e thermal c h a r a c t e r i s -

t i c s o f t h e t i r e c o n s t i t u e n t s . As D r . Tabaddor p o i n t e d o u t t h i s morning, t h e
c o n s t i t u e n t s a r e h i g h l y s e n s i t i v e t o temperature. I n t h e area o f a i r c r a f t
ti res, we a r e v e r y concerned about t h e t r a n s i e n t temperatures t h a t a r e generated
d u r i n g normal t a x i o p e r a t i o n s . Truck and automobile t i r e people have s i m i l a r
concerns except t h a t they deal w i t h r o l l i n g f o r many many hundreds o f m i l e s a t a
t i m e a t s t a b i 1 ized temperatures.

D r . Tabaddor: With r e s p e c t t o D r . Ne1 son's question, e x p e r i m e n t a l l y , o f course,

t h e r e i s much concern about t h e e f f e c t o f t h e temperature on m a t e r i a l proper-
t i e s , e s p e c i a l l y f a t i g u e and f r a c t u r e p r o p e r t i e s . There i s a s i g n i f i c a n t degra-
d a t i o n o f t h e s e q u a l i t i e s w i t h i n c r e a s i n g temperature. The a n a l y t i c a l model
w i l l a l s o be e x t r e m e l y complex because y o u have a mechanical model and r o l l i n g
c o n t a c t w h i l e v i s c o e l a s t i c i t y generates h e a t and t h a t h e a t i n t u r n a f f e c t s t h e
t i r e material properties.

Dr. Tielking: One o b j e c t i v e o f t h e r o l l i n g c o n t a c t a n a l y s i s which i n c l u d e s

f r i c t i o n and v i s o e l a s t i c i t y i s t o g e t t h e energy d i s s i p a t i o n due t o t i r e f l e x i n g
i n t o a heat conduction analysis. There i s some p r e l i m i n a r y work a l r e a d y under
way i n t h e N a t i o n a l T i r e Modeling Program t h a t ' s i n v o l v e d i n t r a n s i e n t h e a t con-
duction analysis f o r r o l l i n g t i r e s . Sam C l a r k has been i n v o l v e d i n t h a t t o a
g r e a t e x t e n t b o t h e x p e r i m e n t a l l y and c o m p u t a t i o n a l l y over t h e 1a s t couple of
years.

Joop Nagtegaal, MARC A n a l y s i s Research Corp.: I have had q u i t e a b i t o f e x p e r i -

ence w i t h t h e t i r e i n d u s t r y . I would l i k e t o make a comment on t h e remarks from

337
 t h e gentleman o f Goodrich. I n t h e l a t e 7 0 ' s when I was d e a l i n g w i t h several
t i r e i n d u s t r i e s on a n o t h e r c o n t i n e n t , they would pose a simple problem t h a t
c o u l d be solved, b u t t h e n say t h a t t h a t i s n o t t h e whole problem. I would have
t o account f i r s t f o r one e f f e c t and then a n o t h e r u n t i l , f i n a l l y , t h e problem
would become unsolvable. Now t i r e problems a r e e x t r e m e l y complicated, I under-
s t a n d t h a t , b u t I t h i n k i t would be e x t r e m e l y u s e f u l i f t h i s N a t i o n a l T i r e
Modeling Program c o u l d come up w i t h a c l e a r s e t o f problems t o solve. Maybe
t h i s i s a l r e a d y b e i n g done because I see a l o t o f p o s i t i v e c o n t r i b u t i o n s here
which a r e very d i f f e r e n t from many years ago.

338
l- N a l w l k m u l r s and
Snace Mrn~ruslralm

1. Report No.
Report Documentation Page
2. Government Accession No. 3. Recipient’s Catalog No.

NASA CP- 3034, Part 1

5. Report Date

May 1989
6. Performing Organization Code

7. Authods) 8. Performing Organization Report No.

W. Jefferson Stroud, Jerrold M. Housner, L-16560

John A. Tanner, and Robert J. Hayduk, Editors 10. Work Unit No.

9. Performing Organization Name and Address 505-63-01-10

NASA Langley Research Center 11. Contract or Grant No.
Hampton, VA 23665-5225
13. Type of Report and Period Covered
12. Sponsoring Agency Name and Address
National Aeronautics and Space Administration Conference Publication
Washington, DC 20546-0001 14. Sponsoring Agency Code

I 15. Supplementary Notes

16. Abstract
The conference publication contains the proceedings of the Workshop on Computatione
Methods for Structural Mechanics and Dynamics held at NASA Langley Research Center,
June 19-21, 1985. The Workshop was organized into the following four sessions:

(1) Local/Global Nonlinear Stress Analysis

(2) Tire Modeling
(3) Transient Dynamics
(4) Multibody Dynamics

Transcriptions of discussions are also included.

17. Key Words (Suggested by Author(s)) 18. Distribution Statement

Local/Global Parallel Processing Unclassified - Unlimited
Global/Local
Tire Modeling

1 Transient Dynamics
Multibody Dynamics
19. Security Classif. (of this report)
I Subject Category
20. Security Classif. (of this pagel
-
21. No. of pages
39
22. Price

342 A1 5
Unclassified unclassified