Finit Element Simulation Using ANSYS

ENGINEERING – GENERAL
“A must read for those interested in exploring the possibilities of using the finite element
method (FEM) as a practical analysis tool for structural mechanics, stress analysis,
Finite Element
Simulations
vibration, heat transfer, and fluid dynamic problems. … Alawadhi systematically introduces
the theory of FEA and demonstrates a step-by-step procedure for practical analysis using
ANSYS. This book is a must-have reference for students, academics, and practitioners
in the field of mechanical, civil, environmental, and design engineering.”
Using ANSYS
—Dr. Arun Arjunan, University of Wolverhampton, UK
Uses a Step-By-Step Technique Directed with Guided Problems

and Relevant Screen Shots
Simulation use is on the rise, and more practicing professionals are depending on the
Second Edition
reliability of software to help them tackle real-world mechanical engineering problems.
Finite Element Simulations Using ANSYS, Second Edition offers a basic understanding
of the principles of simulation in conjunction with the application of ANSYS. Employing
a step-by-step process, the book presents practical end-of-chapter problems that are
solved using ANSYS and explains the physics behind them. The book examines structure,
solid mechanics, vibration, heat transfer, and fluid dynamics. Each topic is treated in a
way that allows for the independent study of a single subject or related chapter.
What’s New in the Second Edition:
• Introduces the newest methods in modeling and meshing for

finite element analysis
• Modifies ANSYS examples to comply with the newest version of ANSYS
• Replaces many ANSYS examples used in the first edition with more
general, comprehensive, and easy-to-follow examples
• Adds more details to the theoretical material on the finite element
• Provides increased coverage of finite element analysis for
heat transfer topics
• Presents open-ended, end-of-chapter problems tailored to serve as
class projects
Finite Element Simulations Using ANSYS, Second Edition functions as a fundamental

reference for finite element analysis with ANSYS methods and procedures, as well as a
guide for project and product analysis and design.
K24312
ISBN: 978-1-4822-6197-4
Esam M. Alawadhi
90000
9 781482 261974
K24312_Cover_PubGr.indd All Pages 8/10/15 11:35 AM

Finite Element
Simulations
Using ANSYS
Second Edition
This page intentionally left blank
Finite Element
Simulations
Using ANSYS
Second Edition
Esam M. Alawadhi
Kuwait University
Safat, Kuwait
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2016 by Taylor & Francis Group, LLC
CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works

Version Date: 20150728
International Standard Book Number-13: 978-1-4822-6198-1 (eBook - PDF)
This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been
made to publish reliable data and information, but the author and publisher cannot assume responsibility for the valid-
ity of all materials or the consequences of their use. The authors and publishers have attempted to trace the copyright
holders of all material reproduced in this publication and apologize to copyright holders if permission to publish in this
form has not been obtained. If any copyright material has not been acknowledged please write and let us know so we may
rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or uti-
lized in any form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopy-
ing, microfilming, and recording, or in any information storage or retrieval system, without written permission from the
publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://
www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923,
978-750-8400. CCC is a not-for-profit organization that provides licenses and registration for a variety of users. For
organizations that have been granted a photocopy license by the CCC, a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for
identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
Contents
Preface ix
Acknowledgments xi
About the author xiii
1 Introduction to finite element method and ANSYS 1
1.1 The finite element method and structural analysis 1

1.2 Stress analysis of pin-jointed bar 5
1.3 The finite element method and thermal analysis 6
1.4 Heat transfer through a composite wall 8
1.5 Introduction to ANSYS software 10
2 Bar and beam structures 19
2.1 Finite element method for a horizontal bar element 19

2.2 Finite element method for an arbitrary-oriented
bar element 23
2.3 Analyzing a plane bar structure 26
2.4 Analyzing a plane bar structure using ANSYS 29
2.5 Finite element method for a horizontal beam element 42
2.6 Analyzing a horizontal beam structure 49
2.7 Analyzing a horizontal beam structure using ANSYS 51
2.8 Development of an arbitrary-oriented beam element 67
2.9 Distributed load on a beam element 70
2.10 Analyzing beam structure under a transient loading
using ANSYS 72
v
vi Contents
3 Solid mechanics 95
3.1 Stress–strain relations 95

3.2 Development of triangular plane stress
and plane strain element 100
3.3 Analyzing rectangular plate subjected to forces 105
3.4 Development of rectangular plane stress element 109
3.5 Analyzing a plate with a hole subjected to tensile
pressure using ANSYS 112
3.6 Axisymmetric elements 126
3.7 Displacement analysis of a vessel under transient
loading with ANSYS 130
3.8 Contact element analysis 145
3.9 Two horizontal cylinders in contact analysis
using ANSYS 147
4 Vibration 171
4.1 Vibration analysis 171

4.2 Development of stiffness equations for axial
vibration of a bar element 172
4.3 Natural frequencies of axial vibration of a bar element 173
4.4 Development of stiffness equations for flexural
vibration of a beam element 175
4.5 Natural frequencies of the flexural vibration of
a beam element 177
4.6 Natural frequencies of the flexural vibration of
beam element using ANSYS 179
4.7 Development of stiffness equations for vibration of
an oriented beam element 190
4.8 Harmonic vibration of a plate with holes using ANSYS 192
4.9 Three-dimensional vibration of shaft with disks
using ANSYS 207
5 Heat transfer 225
5.1 Introduction to heat conduction 225

5.2 Finite element formulation for conductive heat transfer 228
5.3 Finite element method for one-dimensional
heat conduction 231
5.4 Heat transfer through a composite wall 233
5.5 Finite element method for two-dimensional
heat conduction 235
5.6 Heat conduction in a solid plane 238
5.7 Thermal analysis of fin and chip using ANSYS 243
5.8 Finite element method for transient heat transfer 264
5.9 Unsteady thermal analyses of a masonry brick
using ANSYS 265
Contents vii
6 Fluid mechanics 293
6.1 Governing equations for fluid mechanics 293

6.2 Finite element method for fluid mechanics 296
6.3 Entrance length in developing flow in a channel
using ANSYS 298
6.4 Studying flow around a half cylinder in a channel
using ANSYS 318
7 Multiphysics 343
7.1 Introduction 343

7.2 Thermal and structural analysis of a thermocouple
using ANSYS 345
7.3 Chips cooling in a channel using ANSYS 361
8 Meshing guide 393
8.1 Mesh refinement 393

8.2 Element distortion 395
8.3 Mapped mesh 396
8.4 Mapped mesh with ANSYS 398
Bibliography 411
Index 413
Preface
Due to the complexity of modern-day problems in mechanical e ngineering,

engineers, in practice, seldom rely only on theory or experiments. The use
of engineering software is becoming prevalent among academics as well as
practicing engineers. Commercial finite element software such as ANSYS
(ANSYS, Canonsburg, Pennsylvania) and Abaqus (Dassault Systèmes,
Vélizy Villacoublay, France) for analysis and design is commonly found
in use in universities, industry, and research centers. This software has
become more reliable, reputable, easy to use, and trustworthy, and it allows
much time to be saved.
This book focuses on the use of ANSYS in solving practical
mechanical engineering problems. ANSYS is extensively used in the
design cycle by industry leaders in the United States and around the
world. Additionally, ANSYS is available in computer labs for students
in most universities around the world. Courses such as computer-aided
design, modeling, and simulation, and major design courses all uti-
lize ANSYS as a tool for analyzing various mechanical components.
Graduate students also use ANSYS in their finite element courses as
a complement for the theoretical study of the finite element method.
This book provides mechanical engineering students and engi-
neers with the fundamental knowledge of numerical simulation using
ANSYS. The book serves most of the disciplines of mechanical engi-
neering: structure, solid mechanics, vibration, heat transfer, and fluid
dynamics, with adequate background material to explain the physics
behind the computations. Each physical phenomenon is treated inde-
pendently in a way that enables readers to pick out a single subject or a
related chapter and study it. Instructors can cover appropriate chapters
depending on the objectives of the course. The required basic knowl-
edge of the finite element method relevant to each physical phenomenon
is illustrated at the beginning of the respective chapter. The general
theory of the finite e lement, however, is presented briefly and concisely
because the theory is well documented by other finite element books.
ix
x Preface
For example, in the heat transfer chapter, the theory is first explained,
the governing equations are derived, the modeling techniques are pre-
sented, and finally practical problems are solved using ANSYS in a
step-by-step technique. Each chapter independently discusses a single
physical phenomenon, while the last chapter is devoted to multiphysics
analyses and problems. The finite element solution is greatly affected
by the quality of the mesh, and therefore, a separate chapter on meshing
is included as a guide that emphasizes the basics of the meshing tech-
niques. Practical end-of-chapter problems are provided in each chapter
to challenge the reader’s understanding.
Undergraduate and graduate engineers will use this book as a part of
their courses, either when studying the basics of applied finite elements,
or in mastering practical tools of engineering modeling. Engineers in
industry can use this book as a guide for better design and analysis
of their products. In all mechanical engineering curricula, junior- and
senior-level courses use some type of engineering modeling software,
and, oftentimes, this software is ANSYS. Senior students also use
ANSYS in their senior design projects. Graduate-level finite element
courses frequently use ANSYS to complement the theoretical analysis
of finite elements. The course that uses this book should be taken after
the introduction to design courses and the basic thermal–fluid courses.
Courses such as that in senior design can be taken after this course.
In this second edition of the book, new sections are added, and
ANSYS examples are modified to be in compliance with the new version
of ANSYS. Most ANSYS examples in the first edition are replaced by
more general, comprehensive, and easy-to-follow examples. In the finite
element theoretical part, more details are added, especially for the heat
transfer chapter. Additionally, open-ended problems are added at the end
of each chapter, which can serve as class projects.
Acknowledgments
I profited greatly from discussion with faculty members and engineers at

Kuwait University. I want to mention particularly Professor Ahmed
Yigit.
xi
About the author
Esam M. Alawadhi is a professor of mechanical engineering at Kuwait

University, Kuwait City, Kuwait. He earned his doctor of philosophy in
mechanical engineering from Carnegie Mellon University, Pittsburgh,
Pennsylvania in May 2001. His research focuses on renewable energy,
thermal management of electronics devices, energy conservation for
buildings, fluid flow stability, and phase-change heat transfer.
xiii
Cha p t e r ON E
Introduction to finite element

method and ANSYS
1.1 The finite element method and structural analysis
The finite element method was developed in the 1950s and has been
continuously enhanced since then. Rapid advances in computing power
and the drastic drop in cost make the finite element method affordable.
Now, it is a commonly used method for solving a wide range of problems,
and the potential of the finite element method is enormous. The finite
element method is typically found in the aerospace, automotive, elec-
trical, hydraulic, biomedical, nuclear, and structural engineering fields,
among many others. The first step in the finite element solution proce-
dure is to divide the domain into elements, and this process is called
the domain discretization. The variable distribution across each element
can be defined by linear, quadratic, or trigonometric function. The ele-
ments distribution in the domain is called the finite element mesh. The
elements are connected at points called nodes. For example, consider a
section of a bridge, as shown in Figure 1.1a. The bridge is divided into
linear elements and connected by nodes, as shown in Figure 1.1b.
After the region is discretized, the governing equations for the ele-
ment must be established for the required physics. Material properties,
such as modulus of elasticity for structural analysis, should be avail-
able. The equations are assembled to obtain the global equation for the
mesh, which describes the behavior of the body as a whole. Generally,
the global governing equation has the following form:
[K]{U} = {F} (1.1)
where [K] is called the stiffness matrix; {U} is the nodal degree of
freedom, such as the displacements for structural analysis; and {F}
is the nodal external force, such as forces for structural analysis. The
[K] matrix is a singular matrix, and therefore it cannot be inverted.
1
2 Finite Element Simul ations Using ANSYS
Node
Element
(a) (b)
FIGURE 1.1 (a) A section of a bridge and (b) finite element mesh.
Consider a one-dimensional bar with initial length L that is subjected to

a tensile force at its ends, as shown in Figure 1.2a. The cross-sectional
area of the bar is A. The element can be modeled using a single element
with two nodes, i and j, as shown in Figure 1.2b.
Assume that the displacement function of the bar u(x) varies linearly
along the length of the bar. The expression of the displacement is repre-
sented as
u(x) = a + bx (1.2)
The displacement at nodes i and j are ui and uj, respectively. Then
ui(x) = a + bxi (1.3)
uj(x) = a + bxj (1.4)
where xi is the x-coordinate for node i, and xj is the x-coordinate

for node j. Solving for constants a and b, it is found that
a = (uixj − ujxi)/L (1.5)
b = (uj − ui) / L (1.6)
where L is the initial length of the element and is equal to (xi − xj).
Substituting constants a and b into the displacement equation (Equation 1.2)
and rearranging, the displacement function becomes
xj − x x − xi
u (x) = ui + u j (1.7)
L L
F F Element
L
x Node i Node j
(a) (b)
FIGURE 1.2 A one-dimensional bar element. (a) A single one-dimensional bar element and (b) an
element with two nodes.
Introduc tion to finite element me thod and ANSYS 3
or
u(x) = Niui − Njuj (1.8)
where Ni and Nj are called the shape functions of the element. When
the element is loaded, it will be in an equilibrium position. The sum
of the strain energy γ and work w done by external forces is the
potential energy π of the element. The potential energy at the equi-
librium position has the minimum value, and the potential energy is
defined as
π = γ − w (1.9)
For a single bar element, the strain energy in the bar is given by
xj
1
γ=
2 ∫ σ ε A dx (1.10)
xi
Since the strain is related to the stress by the modulus of elasticity

(σ = Eε), the strain energy is expressed as
xj
AE 2
γ=
2 ∫
ε dx (1.11)
xi
The strain is equal to the elongation of the element in the x-direction
ε = (uj − ui) / L (1.12)
Then, the strain energy in Equation 1.11 becomes
AE
γ=
2L
( u j − u i ) (1.13)
2
The strain energy can be expressed in the matrix form as follows:
   u i  1
u i u j   1 −1  
AE 
γ= = {U}T [K ]{U} (1.14)
2L    −1 1  u j  2
 
where [K] is the stiffness matrix for the element. The external forces are
forces acting at the nodes. The work done by the applied forces at the
nodes is expressed as
w = uiFi + ujFj = {U}T{F} (1.15)

Finally, the work done and strain energy are substituted into the
potential energy for a single element. The potential energy is expressed as
1
π= {U}T [ K ]{U} − {U}T {F } (1.16)
2
For the minimum potential energy, the displacement must be equal

to zero:
∂π
= 0 (1.17)
∂{U}
or
∂π ∂π
= 0 and = 0 (1.18)
∂ui ∂u j
Equation 1.16 is differentiated using the expressions (1.18), and the

results are
AE

L
( ui − u j ) − Fi = 0 (1.19)
AE

L
( − ui + u j ) − Fj = 0 (1.20)
Equations 1.19 and 1.20 are expressed in the matrix form as follows:
AE  1 −1   u i   Fi 
 −   = 0 (1.21)
L  −1 1   u j   Fj 
   
and symbolically,
[K]{U} − {F} = 0 (1.22)
where
AE  1 −1 
[K ] = (1.23)
L  −1 1 
The derivation of the stiffness matrix is valid only for one bar ele-
ment. In practice, a model consists of many elements of different proper-
ties. The total potential energy of a number of elements is
π= ∑(γ
e =1
e
− w ) (1.24)
Minimizing Equation 1.24, the result is
∑ K  {U} − {F} = 0 (1.25)

e =1
e
1.2 Stress analysis of pin-jointed bar
Use the finite element method to calculate the maximum displacement

of the bar assembly shown in Figure 1.3a. Given that A1 = 90 mm2,
A2 = 75 mm2, A3 = 50 mm2, E1 = 100 GPa, E2 = 110 GPa, and E3 = 180 GPa.
Before starting to solve the problem, all bar elements and nodes of
the assembly should be numbered, as suggested in Figure 1.3b.
There are three main steps for solving this problem using the finite
element method: (1) the stiffness matrix for all bar elements is deter-
mined, and then stiffness matrices are assembled to obtain the global
stiffness matrix; (2) the boundary conditions are applied to the global
stiffness matrix; and (3) rows and a column of the global stiffness matrix
are eliminated to remove the singularity of the stiffness matrix, and
finally we solve for the displacements.
Step 1: Construct the stiffness matrix for all bar elements

The stiffness matrix for element 1, which has nodes 1 and 2, is calcu-
lated using matrix (1.23):
K (1)  =
( )
90 × 10 −6 100 × 10 9  1 −1   3 −3 
= 108 
−3   

30 × 10  −1 1   −3 3 
The stiffness matrix for element 2, which has nodes 2 and 3, uses
K ( 2)  =
( )
75 × 10 −6 110 × 10 9  1 −1   1.5 −1.5 
= 108 
−3   

55 × 10  −1 1   −1.5 1.5 
(1)
(2)
(3)
50 kN
30.00 mm 55.00 mm 45.00 mm
(a)
Element (1) Element (2) Element (3)
Node (1) Node (2) Node (3) Node (4)

(b)
FIGURE 1.3 (a) Bar assembly. (b) Elements and nodes for the bar assembly.
K (3)  =
(
50 × 10 −6 180 × 10 9  1 ) −1   2
= 108 
−2 
45 × 10 −3  −1  2 
 1   −2
Assembling the elements’ stiffness matrices to form the global stiff-

ness matrix,
e=3
[K ] = ∑ K
e =1
(e)


 3 −3 0 0 
 −3 4.5 −1.5 0 
[ K ] = 108  
 0 −1.5 3.5 −2 
 0 0 −2 2 

Step 2: Apply the boundary conditions to the global stiffness matrix

using Equation 1.22
 3 −3 0 0   u1 = 0   F1 
   u 2   
−3 4.5 −1.5 0   F2 = 0 
108   =  
 0 −1.5 3.5 −2   u3   F3 = 0 
 0 0 −2 2   u 4   F4 = 50 
  
Step 3: Eliminate the first row and column from the global stiffness
matrix to remove the singularity
 4.5 −1.5 0   u2   0 
 u   
108  −1.5 3.5 −2 =  0 
  3
 0 −2 2   u 4   50 

Solve for displacements:
u2 = 1.667 × 10 −7 m
u3 = 5.0 × 10 −7 m
u4 = 7.5 × 10 −7 m (Maximum)
1.3 The finite element method and thermal analysis
Heat transfer in a one-dimensional space can be easily analyzed using

the finite element method. Consider a wall that is made by common
bricks, as shown in Figure 1.4. The outdoor surface temperature is Ti,
while the indoor surface temperature is Tj.
Ti Tj
Wall
Outdoor Indoor
FIGURE 1.4 One-dimensional heat con-

L
duction in a wall.
The Fourier law indicates that the heat flow across the wall Q is pro-
portional to temperature gradient dT/dx in the direction of heat transfer.
That is,
dT
Q = − kA (1.26)
dx
where k is the thermal conductivity of the wall and A is the area per-
pendicular to the heat flow. For steady-state conditions, the temperature
varies linearly along the length of the wall as follows:
dT Tj − Ti (1.27)
=
dx L
Hence, the heat flow across the indoor and outdoor surfaces of the
wall can be expressed as, respectively,
Tj − Ti
Q i = − kA (1.28)
L
Ti − Tj
Q j = − kA (1.29)
L
Equations 1.28 and 1.29 are expressed in the following matrix

form:
kA  1 −1   Ti   Q i 
 −   = 0 (1.30)
L  −1 1   Tj   Q j 
   
This is analogous to the bar element equation (Equation 1.21). The

nodal forces are substituted by nodal heat flow, nodal displacements by
nodal temperatures, and constant AE/L by kA/L; therefore,
[K]{T} − {Q} = 0 (1.31)

where
kA  1 −1 
[K ] = (1.32)
L  −1 1 
1.4 Heat transfer through a composite wall
Consider a wall that is constructed by stone bricks at the outdoor surface

(k = 1.25 W/m · °C), common brick (k = 0.75 W/m · °C), and plaster at the
indoor surface (k = 0.9 W/m · °C), as shown in Figure 1.5. The outdoor
surface temperature is To = 45°C, while the indoor surface temperature
is Ti = 25°C. Determine the temperature at the interfaces, and the results
should be per unit area of the wall.
Step 1: Construct the stiffness matrix for all elements

The stiffness matrix for element 1, which has nodes 1 and 2, is calcu-
lated using expression (1.32):
K (1)  =
1.25 × 1  1 −1   25 −25 
=
5 × 10 −2  −1 1   −25 25 

K ( 2)  =
0.75 × 1  1 −1   5 −5 
=
15 × 10 −2  −1 1   −5 5 

Outdoor Indoor
(To) (Ti)
Stone Common
brick brick
Plaster
5 cm 15 cm 3 cm
(a)
Element (1) Element (2) Element (3)
Node (1) Node (2) Node (3) Node (4)

(b)
FIGURE 1.5 The wall composition and finite element mesh. (a) The wall composition and (b) the
finite element presentation.
K (3)  =
0.9 × 1  1 −1   30 −30 
=
3 × 10 −2  −1 1   −30 30 

Assembling the elements’ stiffness matrices to form a global stiffness

matrix,
e=3
[K ] = ∑ K
e =1
(e)


 25 −25 0 0 
 −25 30 −5 0 
[ K ] = 108  
 0 −5 35 −30 
 0 0 −30 30 

Step 2: Apply the boundary conditions to the global stiffness matrix

using Equation 1.31
 25 −25 0 0   T1 = 45   Q1 
 −25 30 −5     
0 T2   Q2 = 0 
  =  
 0 −5 35 −30  T3   Q3 = 0 
 0 0 −30 30   T4 = 25   Q 4 
   
Then Q1 and Q2 are required to maintain the constant temperature at

the external surfaces.
Step 3: Eliminate the rows and columns, numbers 1 and 4, from

the global stiffness matrix to remove the singularity from the stiff-
ness matrix
The temperatures at nodes 1 and 4 are known. Hence, only Equations 2
and 3 are needed to solve the problem:
 25 −5   T2   25 × 45 
 −5  =  
 35   T3   30 × 25 

Solving for temperatures,
T2 = 42.07°C
T3 = 27.44°C
1.5 Introduction to ANSYS software
The ANSYS software is the most advanced package for single- and
multiphysics simulations, offering enhanced tools and capabilities that
enable engineers to complete their jobs in an efficient manner. ANSYS
includes significant capabilities, expanding functionality, and integra-
tion with almost all computer-aided design (CAD) drawing software,
such as pro-engineers, AutoCAD, and Solid Edge. In addition, ANSYS
has the best-in-class solver technologies, a coupled physics for complex
simulations, meshing technologies customizable for physics, and com-
putational fluid dynamics.
ANSYS can solve problems in structural, thermal, fluid, acoustics,
and multiphysics:
Structural:
• Linear
• Geometric and material nonlinearities
• Contact
• Static
• Dynamic
• Transient, natural frequency, harmonic response, response spec-
trum, random vibration
• Buckling
• Topological optimization
Thermal:
• Steady state or transient

• Conduction
• Convection
• Radiation
• Phase change
Fluid mechanics:
• Steady state or transient

• Incompressible or compressible
• Laminar or turbulent
• Newtonian or non-Newtonian
• Free, forced, or mixed convection heat transfer
• Conjugate solid/fluid heat transfer
• Surface-to-surface radiation heat transfer
• Multiple species transport

• Free surface boundaries
• Fan models and distributed resistances
• Stationary or rotating reference frames
Acoustics:
• Fully coupled fluid/structural

• Near and far field
• Harmonic, transient, and modal
Multiphysics:
• Thermal–mechanical
• Thermal–electric
• Thermal–electric–structural
• Piezoelectric
• Piezoresistive
• Peltier effect
• Thermocouple effect
• Electromechanical circuit simulator
ANSYS is not a single program, but it is a family of CAD programs,

as shown in Figure 1.6. The ANSYS/Multiphysics package consists of
ANSYS/Emag, which is for magnetic field analysis, ANSYS/Flotran
for fluid mechanics, ANSYS/LS-DYNA for dynamics analysis, and
ANSYS/Mechanical for structural and thermal analyses.
ANSYS/ ANSYS/
Mechanical Emag
ANSYS/
Multiphysics
ANSYS/ ANSYS/
Flotran LS-DYNA
FIGURE 1.6 The ANSYS
family.
To start ANSYS:
Double click on the Mechanical APDL Product Launcher icon in

the desktop
or, go to
Start > All Programs > ANSYS > Mechanical APDL Product
Launcher
The ANSYS Product Launcher window will appear as shown. First,

select the ANSYS Multiphysics in License. Then, type the working direc-
tory path for ANSYS files in the Working Directory. All ANSYS files will
be stored in this directory, including images and AVI movies. The session
name is specified in the Job Name. Finally, click on Run to start the ANSYS.
A select ANSYS Multiphysics in License

B change Working Directory to C:\, or any directory
C change the initial Job Name to Example, or any name
Run
The ANSYS interface has the following components:

• Utility menu
• Toolbar
• Main menu
• View tools
• Graphics
• Comment prompt
In addition, the ANSYS Output window will show up. The output
window is dynamically listing important information during the pre-
processor, solution, and postprocessor. Additionally, the calculated
results in the postprocessor are shown in this window. Warnings in
the ANSYS Output should be carefully considered to avoid unex-
pected errors.
Utility menu
The Utility menu is used for the file operations, listing and plotting,
and changing display options. In the pull-down File menu, the following
tasks can be performed:
A Clearing current job and starting a new job

B Resuming a job, and the database file can be located
C Saving ANSYS job
D Reading ANSYS input file
E Listing and file operations
F Importing and exporting files
G Generating a summary file
H Exiting ANSYS job
The pull-down List menu is for listing the model components, such as
keypoints, areas, and nodes. In addition, the properties of the material can
be listed. The boundary conditions imposed on the model can also be listed.
The pull-down Plot menu is similar to the List menu. Plotting

g eometry’s components, such as keypoints and areas, can be performed
in this menu. In the PlotCtrls menu, printing the model in the ANSYS
graphics, changing the style of the ANSYS graphics, or changing the
quality of the graphics can be done. The Workplane is for the grids setup.
ANSYS Main Menu

Most ANSYS jobs are done using the Main Menu, from building the
model to viewing the results. In the preprocessor, the material properties
and real constants are specified, element type is selected, and modeling
operations and meshing tools are available. In the solution, the boundary
conditions are imposed, and the solution setup parameters are specified.
In the postprocessor, the ANSYS results are presented. List, plot results,
and path operations can be performed.
The three tasks are summarized as follows:

Preprocessor:
1. Element type
2. Material properties
3. Real constants
4. Modeling
5. Meshing
Solution:
1. Boundary condition
2. Solution setup parameters
Postprocessor:
1. Plot results
2. List results
3. Path operation
PROBLEM 1.1
Use the finite element method to calculate the maximum displacement
of the bar assemblies shown in Figures 1.7 to 1.9.
E = 155 GPa
A = 100 mm2 E = 180 GPa
A = 50 mm2 50 kN
FIGURE 1.7 Two bars

assembly with a tensile
35.00 mm 55.00 mm
force for at one end.
E = 150 GPa E = 200 GPa E = 150 GPa

A = 75 mm2 A = 125 mm2 A = 75 mm2
150 kN 150 kN
55.00 mm 35.00 mm 55.00 mm
FIGURE 1.8 Three bars assembly with a tensile force for at both ends.
100 kN
E = 150 GPa
D = 0.45 cm
E = 150 GPa 2 cm
D = 1.1 cm
E = 150 GPa
D = 1.75 cm
E = 150 GPa 1 cm
D = 2.25 cm
0.5 cm
0.25 cm
FIGURE 1.9 A structure with a vertical tensile force.
PROBLEM 1.2
Consider a wall shown in Figure 1.10. The outdoor surface
temperature is To = 42°C, while the indoor surface temperature is
Ti = 23.5°C. Determine the temperature at the interfaces, and the
results should be per unit area of the wall. The thermal conductivity
of stone brick, thermal insulation, common brick, and indoor finish-
ing is 1.25 W/m · °C, 0.85 W/m · °C, 0.2 W/m · °C, and 1.75 W/m · °C,
respectively.
Insulation
Outdoor Indoor
(To) (Ti)
Stone Common
brick brick
Indoor
finishing
10.5 cm 22.5 cm 1.25 cm

5.2 cm
FIGURE 1.10 Heat transfer in a wall composed of four different materials.
Al Fe Cu
205 W/m . ºC 73 W/m .ºC 386 W/m .ºC
T1 = 100˚C T2 = 25˚C
2.5 cm 5.5 cm 4.5 cm
FIGURE 1.11 Heat transfer in a rod composed of three different materials.
PROBLEM 1.3
Consider a thermal rod that is composed of three different materials, as
shown in Figure 1.11. The left circular area is maintained at a tempera-
ture of T1 = 100°C, while the right one is maintained at a temperature
of T2 = 25°C. The entire lateral surface is well insulated. Determine the
heat flow through the rod, if its diameter is 1.5 cm.
Cha p t e r T WO
Bar and beam structures
2.1 Finite element method for a horizontal bar element
The bar structures are composed of straight bars and welded at the
interactions. The bar structures can be found in bridges and roofs.
The derivation of the stiffness matrix for a bar element is applicable to
the pin-connected structures. The bar element is assumed to have a con-
stant cross-sectional area A, uniform modulus of elasticity E, and initial
length L. The bar is subjected to tensile forces along the local axis and
applied at its ends. There are two coordinate systems: local (x, y) and
global (x,y) coordinates. Figure 2.1 shows the local and global coordi-
nate systems.
For each element, there are two local forces acting on the nodes: F1x
and F1y are acting on node 1, and F2x and F2y are acting on node 2. The
nodal degrees of freedom are the four local displacements, two at each
node: d1x and d1y are at node 1, and d2x and d2y are at node 2. Figure 2.2
shows a bar element lying along the local x-coordinate.
The strain–displacement relationship is obtained from Hooke’s law
as follows:
σ x = Eε x (2.1)
where
du
εx = (2.2)
dx
and therefore,
Aσ x = Fx (2.3)
Then u is the local axial displacement in the x-direction and Fx is the

tensile force in the local x-direction. Note that the bar element cannot
sustain shear forces. Substituting σx and εx into Hooke’s law (2.1) yields:
d  du 
 AE  = 0 (2.4)
dx  dx 
19
FIGURE 2.1 The local and global θ

x
coordinate systems.
Node 1 A, E Node 2
Fx1 Fx2
FIGURE 2.2 A bar ele-
ment lying along the local x
L
x-coordinate.
Assuming a linear displacement along the local x-axis of the bar, the
displacement function is expressed as
u ( x ) = a1 + a 2 x (2.5)
The displacement values at nodes are used to determine the values of

the constants a1 and a2. Now, the displacement function is expressed as
a function of the nodal displacement d1x and d2x, which can be achieved
by evaluating u at the nodes, and solving for a1 and a2 as follows:
u ( 0 ) = d1x = a1 (2.6)
u ( L ) = d2x = a 2 L + d1x (2.7)
and solving for a2,
d2x − d1x
a2 = (2.8)
L
Hence, the displacement function (2.5) becomes
d −d 
u ( x ) =  2x 1x  x + d1x (2.9)
 L 
The displacement function (2.9) is expressed in the following form:
u ( x ) = N1 d1 + N 2 d2 (2.10)
where N1 and N2 are called the shape functions of the displacement

function, which is associated with degrees of freedom d1 and d2. The
shape functions are defined as
Bar and be a m struc tures 21
x
N1 = 1 − (2.11)
L
and
x
N2 = (2.12)
L
The displacement function can be written in the following matrix
form:
 d1x 
u = [ N1 N 2 ]   (2.13)
 d2x 
The strain is the elongation in the x-direction, which is equal to the

derivative of the displacement function as follows:
u ( x + ∆x ) − u ( x ) du
εx = = (2.14)
∆x dx
In terms of the nodal displacement, the strain is written as
du d2x − d1x
εx = = (2.15)
dx L
The strain (2.15) is substituted into Hooke’s law (2.1) to obtain the
force:
 d − d1x 
Fx = AE  2x  (2.16)
 L
The nodal force at node 1 is acting in the negative x-direction and

should have a negative sign as follows:
F1x = − Fx (2.17)
Hence, the nodal force at node 1 becomes
F1x =
AE
L
(
d1x − d2x (2.18))
The nodal force at node 2 is acting in the positive x-direction and
should have a positive sign as follows:
F2x = Fx (2.19)
F2x =
AE
L
(
d2x − d1x (2.20))
The nodal force in the x-direction is expressed in a matrix form as

follows:
 F1x  AE  1 −1   d1x 
 =   (2.21)
L  −1 1   d2x
 F2x   
and similarly for the nodal forces in the y-direction:
 F1y   
  AE  1 −1   d1y 
 =     (2.22)
 F2y  L  −1 1   d2y 
{}
Since {F } = K  d , the stiffness matrix for a bar element in local
coordinates can be written as
AE  1 −1 
K  = (2.23)
L  −1 1 
The stiffness matrix can be determined using the strain energy

expression along with Castigliano’s theorem for a bar. The strain energy
of a bar in tension is expressed as follows:
L
Fx2
U=
∫
0
2EA
dx (2.24)
In the theorem, the change of strain energy is equal to the virtual

displacement times the external force at a point as follows:
∂U = Pi ∂ di (2.25)
Equation 2.25 in terms of external force is
∂U
Pi = (2.26)
∂ di
The axial force can be written in terms of displacement function as
∂u
Fx = σ x A = E ε x A = EA (2.27)
∂x
Using Equation 2.10 in Equation 2.27, the axial force is expressed as
( )
Fx = EA N1′ d1x + N ′2 d2x (2.28)
The strain energy of a bar (2.24) in tension is expressed as
∫ ( N′ d )
EA 2
U= 1 1x + N ′2 d2x dx (2.29)
2
0
Using Castigliano’s first theorem, the nodal force in the x-direction at

node 1 is written as
L
δ ∂U
F1x =
∂ d1x
= EA
∫ ( N′ d
0
1 1x )
+ N ′2 d2x N1′ dx
(2.30)
L  L 
0
∫  
0
∫
= EA  N1′ N1′ dx  d1x + EA  N1′ N ′2 dx  d2x
 
and at node 2,
L
δ ∂U
F2x =
∂ d2x
= EA
∫ ( N′ d
0
1 1x )
+ N ′2 d2x N ′2 dx
(2.31)
L  L 
0
∫  
0
∫
= EA  N1′ N ′2 dx  d1x + EA  N ′2 N ′2 dx  d2x
 
Nodal forces in the x-direction in the matrix form are expressed as

 F1x   k11 k12   d1x 
 =    (2.32)
k 21 k 22
 F2x     d2x 
Or, symbolically,
{F } = [K ]{d } (2.33)
x x
where [K] is the stiffness matrix, in which the matrix coefficients are
defined as
L

∫
k ij = EA N ′i N ′j dx (2.34)
0
Substituting the shape functions into Equation 2.32, the stiffness

equation is expressed as
 F1x  EA  1 −1   d1x 
 =    (2.35)
 F2x  L  −1 1   d2x 
2.2 Finite element method for an arbitrary-oriented

bar element
The local coordinate system is always chosen to represent an individual

element, whereas the global coordinate system is chosen for the whole
structure. In order to relate the global displacement components to a local
one, the transformation matrix is used. Figure 2.3 shows a bar element
lying along the local x-axis. The local x-axis is oriented at an angle θ
Fy2, dy2
, d x2
F x2
2
de Fx2, dx2
No
L
E, y
A,
1 θ
de x
No
Fy1, dy1
, d x1
F x1
Fx1, dx1
FIGURE 2.3 Relationship between the local and global displacements.
measured counterclockwise from the global x-axis. In the local coor-

dinate, each node has a nodal force Fx and nodal displacement dx along
its axis. In the global coordinate, each node has a horizontal force Fx,
vertical force Fy, horizontal displacement dx, and vertical displacement
dy. Hence, each element has four degrees of freedom.
From Figure 2.3, the displacement in the x-direction for nodes 1 and
2 in a local coordinate system can be obtained from its global displace-
ments as follows:
d1x = d1x cos θ + d1y sin θ (2.36)
d2x = d 2x cos θ + d 2y sin θ (2.37)
and in the y-direction,
d1y = − d1x sin θ + d1y cos θ (2.38)
d2y = − d 2x sin θ + d 2y cos θ (2.39)
The transformation for nodal displacements for each element can be

written in matrix form as follows:
 d1x  d 
   cos θ sin θ 0 0   1x 
 d1y   − sin θ cos θ 0 0   d1y 
 =     (2.40)
 d2x   0 0 cos θ sin θ   d2x 
d   0 0 − sin θ cos θ   
 2y   d2y 
and symbolically,
{d} = [ T ]{d} (2.41)

The nodal force is transformed from the global to the local coordinate
system in a similar manner:
 F1x   
   cos θ sin θ 0 0   F1x 
 F1y   − sin θ cos θ 0 0   F1y 
 =    (2.42)
 F2x   0 0 cos θ sin θ   F2x 
 F2y   0 0 − sin θ cos θ   F2y 
   
and symbolically,
{F} = [ T ]{F} (2.43)

The global element nodal force vector is related to the global
displacement vector using the global stiffness matrix as follows:
 F1x   d1x 
   
 F1y   d1y 
  = [K ]   (2.44)
 F2x   d2x 
 F2y  d 
   2y 
where
 1 0 −1 0 
AE  0 0 0 0 
[K ] =   (2.45)
L  −1 0 1 0 
 0 0 0 0 
{} {}
Substituting d = [ T ]{d} and {F } = [ T ]{F } into {F } = K  d , the
following equation is obtained:
[ T ]{F } = K  [ T ]{d} (2.46)
By multiplying both sides of the above equation by the inverse of the

transformation matrix, the nodal force vector becomes
{F } = [ T ]−1 K  [ T ]{d} (2.47)

Since the transformation matrix is orthogonal, its transport and

inverse are equal. The nodal force vector is expressed as
{F } = [ T ]T K  [ T ]{d} (2.48)

Since the global coordinate force equation {F} = [K]{d} is equal to
nodal force in Equation 2.48, the global stiffness matrix for an element
is written as
[ K ] = [ T ]T K  [ T ] (2.49)
Expanding the stiffness matrix yields
 cos θ cos θ cos θ sin θ − cos θ cos θ − cos θ sin θ 

 
AE  cos θ sin θ sin θ sin θ − sin θ cos θ − sin θ sin θ 
[K] =
L  − cos θ cos θ − sin θ cos θ cos θ cos θ sin θ cos θ 
 
 − cos θ sin θ − sin θ sin θ sin θ cos θ sin θ sin θ 
 
(2.50)
Assemble the global stiffness and force matrices using the direct
stiffness method to obtain matrices for the entire domain:
[K ] = ∑ K
e =1
(e)
 (2.51)
{F } = ∑ {F } (2.52)
e =1
(e)
{d} = ∑ {d } (2.53)
e =1
(e)
The global stiffness matrix is related to the global nodal force

matrix and global displacement matrix for the whole structure using the
following expression:
{F}=[K]{d} (2.54)
2.3 Analyzing a plane bar structure
The bar structure shown in Figure 2.4a is composed of three bars.

All bars have the same length, cross-sectional area, and modulus of
elasticity. A horizontal force is applied at Point 2 with F = 100 kN.
F3y, d3y
3 Node 3 F3x, d3x
Element 2 Element 3
F1y, d1y F2y, d2y
1 2
F F1x, d1x F2x, d2x
Node 1 Element 1 Node 2
(a) (b)
FIGURE 2.4 (a) A bar structure and (b) nodes and elements.
Determine the horizontal and vertical displacements at all connections,

given that E = 100 GPa, A = 1.0 × 10 −4 m2, and L = 1 m.
Before starting to solve the problem, all elements and nodes of the
structure should be numbered as suggested in Figure 2.4b.
Step 1: Construct the stiffness matrix for each element

The stiffness matrix for Element 1, which has nodes 1 and 2 with θ = 0o,
is calculated using the following matrix (2.50):
1 1 2 2
 4 0 −4 0  1
AE  0 0 0 0  1
K (1)  =  
4L  −4 0 4 0  2
 0 0 0 0  2
The stiffness matrix for Element 2, which has nodes 1 and 3 with
θ = 60o, is calculated as follows:
1 1 3 3
 1 −1 − 3  1
3
 
AE  3 3 − 3 −3  1
K ( 2)  =  
4L  −1 − 3 1 3  3
 − 3 −3 3 3  3

The stiffness matrix for Element 3, which has nodes 2 and 3 with
θ = 120°, is calculated as follows:
2 2 3 3
 1 − 3 −1 3  2
 
AE  − 3 3 − 3 −3  2
K (3)  =  
4L  −1 3 1 − 3  3
 3 −3 − 3 3  3

Step 2: Assemble the elements’ stiffness m

atrices to form a global
stiffness matrix using Equation 2.51
1 1 2 2 3 3
 5 3 −4 0 −1 − 3  1
 
 3 3 0 0 − 3 −3  1
 
[K ] =
AE  −4 0 5 − 3 −1 3  2
4L  0 0 − 3 3 3 −3  2
 
 −1 − 3 −1 3 2 0  3
 
 − 3 −3 3 −3 0 6  3
Step 3: Apply the boundary conditions to the global stiffness

matrix using Equation 2.54
 F1x   d1x = 0 
   
 F1y = 0   d1y 
 F2x = 10 5   d 2x 
  = [K ]  d = 0 
 F2y = 0   2y 
 F3x   d 3x = 0 
   
 F3y d =0
  3y 
Step 4: Eliminate rows and columns from the global

stiffness matrix to remove the singularity
Since the stiffness matrix is singular, at a minimum, one equation must
be eliminated to remove the singularity. The first, fourth, fifth, and sixth
columns are multiplied by zero displacements, and therefore, they can be
deleted with the corresponding rows, and the result is
=
(
 F1y = 0  1 × 10 −4 100 × 10 9  3 0   d1y ) 
 5   0 5  
 F2x = 10  1× 4    d 2x 
Step 5: Solve for d1y and d2x

d1y = 0 m
d2x = 0.008 m
2.4 Analyzing a plane bar structure using ANSYS
The bar structure shown in Figure 2.4a is composed of three bars. All

bars have the same length, cross-sectional area, and modulus of elastic-
ity. A horizontal force is applied at Point 2, F = 100 kN. Determine using
ANSYS the horizontal and vertical displacements at all connections and
the reaction at the supports, given that E = 100 GPa, A = 1.0 × 10 –4 m2,
and L = 1 m.
Double click on the Mechanical APDL Product Launcher icon
A select ANSYS Multiphysics in License

B change Working Directory to C:\ or any directory
C change the initial Job Name to bar, or any name
Run
This example is limited to structural analysis. Hence, select Structural

in Preferences and leave other physics unselected. This will reduce selec-
tion options in the preprocessor and solution tasks.
Main Menu > Preferences
A select Structural in Preferences for GUI Filtering
OK
Next, the element type is selected. For the present problem, the
element type is Link, which is equivalent to a bar.
Main Menu > Preprocessor > Element Type > Add/Edit/Delete
Add...
The selected link element will not support pressure on the elements
or moment on the nodes. However, this element can be used for three-
dimensional analyses.
A select Link
B select 3D finit stn 180
OK
Close
The cross-sectional area of the bars is required for the analysis. The
initial strain is zero by default.
Main Menu > Preprocessor > Real Constants > Add/Edit/Delete
Add...
Notice that the selected element type is shown in the Element Type
for Real Constants. Here, the cross-sectional area of the bars is specified.
OK
A type 1e-4 in Cross-sectional area AREA
OK
Close
Material properties of the bars are specified in the following steps. The
bars are elastic and independent of the direction isotropic. Only the modu-
lus of elasticity is required and the zero Poisson ratio, or any value, is just
to avoid an error message from ANSYS.
Main Menu > Preprocessor > Material Props > Material Models
A click on Structural > Linear > Elastic > Isotropic
A type 100e9 in EX
B type 0 in PRXY
OK
Close the Material Model Behavior window
The process of modeling the bar structure is started here. First, three
nodes are created, followed by creation of the elements. The x- and
y-coordinates of each node are specified in ANSYS. The coordinate for
node 1 is (0,0), for node 2 is (1,0), and for node 3 is (0.5,0.866). The
Apply button will not close the window, allowing for additional inputs,
while the OK will close the window.
Main Menu > Preprocessor > Modeling > Create > Nodes >
In Active CS
A type 1 in Node number

B type 0 and 0 in X,Y,Z Location in active CS
Apply

Apply

B type 0.5 and 0.866 in X,Y,Z Location in active CS
OK
ANSYS graphics show the created nodes
Main Menu > Modeling > Create > Elements > Auto Numbered >
Thru Nodes
Click on node 1 then 2. Then in Elements from Nodes window, click on
Apply
Apply
OK
ANSYS graphics show the three elements
The preprocessor task is ended at this point. The solution task starts
here. Nodal forces and displacements are applied. Starting with forces or
displacements will not affect the solution. A zero nodal displacement at

a certain direction means that the node is fixed at that direction.
Main Menu > Solution > Define Loads > Apply > Structural >
Displacement > On Nodes
Click on node 1. Then in Elements from Nodes window, click on
OK
A select UX in Lab2 DOFs to be constrained

B type 0 in Displacement value
Apply
OK
A select UY in Lab2 DOFs to be constrained

Apply
OK
A select UX in Lab2 DOFs to be constrained

B select UY in Lab2 DOFs to be constrained
C type 0 in Displacement value
OK
Force/Moment > On Nodes
Click on node 2. Then in Apply F/M on Nodes window, click on
OK
A select FX in Direction of force/mom

B type 100e3 in Force/moment value
OK
The positive FX means that the force at node 2 is in the positive

x-direction. The ANSYS graphics will show the applied force and its
direction with a red arrow. Reapplying the force at node 2 will automati-
cally delete the force and apply the new value.
ANSYS graphics show the force with d irection and

the displacement constraints on nodes
The final step is to initiate the solution. ANSYS will assemble
the stiffness matrices, apply the boundary conditions, and solve the
equations.
Main Menu > Solution > Solve > Current LS
OK
Close
Results can be viewed in the general postprocessor task. Inspecting

the deformation of the bar structure will help to identify if the problem
is solved correctly. Node 2 should move to the right as a result of the
applied force, while node 3 should be stationary.
Main Menu > General Postproc > Plot Results > Deformed Shape
A select Def + unreformed
OK
ANSYS graphics show the bar before and after deformation

From the above figure, the result is as expected. Nodes 1 and 3 are
fixed, while node 2 is moved in the direction of the applied force. In
the following steps, the nodal results are presented. First, the nodal
displacements are listed, and second, the reactions at the supports are
listed.
Main Menu > General Postproc > List Results > Nodal Solution
A click on Nodal Solution > DOF Solution > X-Component of

displacement
OK
A list of nodal displacement in the x-direction is shown. In addition, the

maximum nodal displacement is shown at the end. The nodal displace-
ment in the y-direction can be determined by selecting Y-Component in
the Nodal Solution. The results are identical to the previous example, and
the reactions at the supports are listed as follows:
Main Menu > General Postproc > List Results > Reaction Solution
A select All items

OK
As shown, the forces are balanced in the x- and y-directions. The sum
of the forces in the x-direction is equal to −100 × 103, which is equal to
the applied force, and the sum of the forces in the y-direction is equal
to zero.
2.5 Finite element method for a horizontal beam element
A beam is defined as a long and slender structural member that can be

subjected to transverse loadings. Therefore, the beam can be twisted
and bended. The bar elements can sustain the transverse loadings only.
Consider a horizontal beam element, as shown in Figure 2.5. The beam
has an initial length L, modulus of elasticity E, and moment of inertia I.
The local displacement and rotation at nodes 1 and 2 are (dx1, θ1) and
(dx2, θ2), respectively. The local nodal force and bending moment at
nodes 1 and 2 are (F1y, M1) and (F2y, M2), respectively.
All moments and rotations are positive if their direction is counter-
clockwise and negative if their direction is in a clockwise direction. The
horizontal forces are positive if the direction is positive in the x-direction
and negative if the direction is negative in the x-direction. The vertical
forces are positive if the direction is positive in the y-direction and nega-
tive if the direction is negative in the y-direction. Consider a differential
beam element, as shown in Figure 2.6.
1 I, E 2 x
M1, θ1 M2, θ2
L
Fy1, dy1 Fy2, dy2
FIGURE 2.5 Beam element subjected to forces and moments.

M M + dM
V a b V + dV
FIGURE 2.6 A differential

dx
beam element.
Applying the moments balance at Point b yield,
− Vdx + dM = 0 (2.55)
or
dM
V = (2.56)
dx
The curvature of the beam is related to the applied moment. The θ is
the rotation and the function v(x) is the transverse displacement in the
y-direction. Figure 2.7 shows a section of a horizontal deflected beam.
The rotation of the deflection is expressed as
dv ( x )
θ = (2.57)
dx
The axial strain is related to the axial displacement using the following
relationship:
du
εx = (2.58)
dx
On the other hand, the axial displacement is related to the transverse
displacement by
dv
u = −y (2.59)
dx
Substituting the relationship (2.59) into (2.58) yields
d2v
ε x = −y (2.60)
dx 2
y
θ
v(x)
x FIGURE 2.7 Horizontal deflected
dx
beam.
For a cross section of the beam shown in Figure 2.8, the differen-
tial force and moment along the center of the beam are expressed as,
respectively,
dF = σx dA (2.61)
dM = σx ydA (2.62)
Using Hooke’s equation, the differential moment becomes

d2v
dM = Eε x ydA = Ey 2 dA (2.63)
dx
Integrating Equation 2.63 for the beam section yields
d 2v
M = E
∫
dx 2
y 2 dA (2.64)
Since the first moment of inertia is equal to ∫ y2 dA, Equation 2.64 is
expressed as
d2v
M = EI (2.65)
dx 2
Also, the shearing force (2.56) is expressed as
d3 v
V = EI (2.66)
dx 3
The variations of beam curvature are approximated by a third-order
polynomial, and the constants in the polynomial are determined using
the boundary conditions as follows:
v(x) = a1x3 + a2x2 + a3x + a4 (2.67)
Using expression (2.57), the rotation of the deflection can be deter-

mined as
θ(x) = 3a1x2 + 2a2x + a3 (2.68)
Applying the boundary conditions,
v(0) = dy1 : a4 = dy1 (2.69)
v(L) = dy2 : a1L3 + a2L2 + a3L + a4 = dy2 (2.70)
dF
y
FIGURE 2.8 Beam cross section.

θ(0) = θ1 : a3 = θ1 (2.71)
θ(L) = θ2 : 3a1L2 + 2a2L + a3 = θ2 (2.72)
Solving Equations 2.69–6.72 for a1, a2, a3, and a4 and substituting
the solution into Equation 2.67 yields
2 1
v ( x ) =  3 ( d1y − d 2y ) + 2 ( θ1 + θ2 )  x 3
L L 
(2.73)
3 1
+  − 2 ( d1y − d 2y ) − ( 2θ1 − θ2 )  x 2 + θ1x + d1y
 L L 
or
3x 2 2x 2 3x 2 x2
v ( x ) = d1y + θ1x − 2
d1y − θ1 + 2 d 2y − θ2
L L L L
(2.74)
2x 2 x3 2x 2 x3
+ 3 d1y + 2 θ1 − 3 d 2y + 2 θ2
L L L L
The displacement function v(x) can be expressed in terms of shape

functions as
v(x) = N1d1y + N2θ1 + N3d2y + N1θ2 (2.75)
and the shape functions are
2 3
x x
N1 = 1 − 3   + 2   (2.76)
 L  L
 x2   x3 
N 2 = x − 2   + 2  2  (2.77)
 L L 
2 3
x x
N 3 = 3   − 2   (2.78)
 L  L
 x 2   x3 
N 4 = −   +  2  (2.79)
 L L 
Finally, the element stiffness matrix is derived using the equilibrium

approach. The nodal shear force and bending moment are calculated
using Equations 2.65 and 2.66, respectively:
d 3 v ( 0 ) EI
F1y = EI = 3 (12d1y + 6Lθ1 − 12d 2y + 6Lθ2 ) (2.80)
dx 3 L
d 2 v ( 0 ) EI
M1 = − EI
dx 2
( )
= 3 6d1y + 4L2 θ1 − 6Ld 2y + 2L2 θ2 (2.81)
L
d 3 v ( L ) EI
F2y = − EI = 3 ( −12 d1y − 6Lθ1 + 12d 2y − 6Lθ2 ) (2.82)
dx 3 L
d 2 v ( L ) EI
M2 = EI
dx 2
( )
= 3 6 Ld1y + 2L2 θ1 − 6Ld 2y + 4L2 θ2 (2.83)
L
The nodal forces and moments can be expressed in matrix form,
{F} = [K]{d}, as follows:
 F1y   12 6L −12 6L   d1y 

    
 M1  EI  6L 4L2 −6L 2L2   θ1 
 =   (2.84)
 F2y  L3  −12 −6L 12 −6L   d 2y 
 M 2   6L 2L2 −6L 4L2   θ2 
   
where
 12 6L −12 6L 
 
EI 6L 4L2 −6L 2L2 
[K ] = 3  (2.85)
L  −12 −6L 12 −6L 
 6L 2L2 −6L 4L2 

Assemble the stiffness matrices and force and displacement vectors:
[K ] = ∑ K
e =1
(e)
 (2.86)
N
{F } = ∑ {F } (2.87)
e =1
(e)
{d} = ∑ {d } (2.88)
e =1
(e)
The global stiffness matrix is used to relate the nodal force vector to
nodal displacement vector for the whole structure:
{F} = [K]{d} (2.89)
The stiffness equation for the horizontal beam element can also be
obtained by using Castigliano’s theorem. The strain energy for a beam
with uniform material properties and cross-sectional area is
L L 2
1 EI  ∂2 v 
U=
2EI ∫
( M )2 dx =
0
2  ∂x 2  ∫
dx (2.90)
0
The second derivative of displacement function is
v ′′ ( x ) = N1′′ d1y + N ′′θ

2 1 + N 3′′ d 2y + N ′′θ
4 2 (2.91)
and the second derivative of the shape functions is
6 x
N1′′= − + 12 3 (2.92)
L2 L
4 x
N ′′2 = − + 6 2 (2.93)
L L
6 x
N 3′′ = − 12 3 (2.94)
L2 L
6 x
N ′′4 = − + 6 2 (2.95)
L L
In Castigliano’s first theorem, the change of strain energy is equal

to the virtual displacement times external force at a point. Applying
Castigliano’s theorem at node 1 gives
∂U ∂  EI L  ∂2 v  2 
F1y = =
∂d1y ∂d1y
 ∫
 2  ∂x 2 
0
dx  (2.96)

Substituting the second derivative of the displacement function into

Equation 2.80 gives
∂  EI 
L
∫ ( N′′d 4 2 ) dx  (2.97)
2
F1y =  1 1y + N ′′θ
2 1 + N 3′′ d 2y + N ′′θ
∂d1y  2 
0
Simplifying expression (2.97) yields
L
EI
2 ( N1′′d1y + N ′′θ 4 2 ) N1′′ dx (2.98)
F1y =
2 ∫
0
2 1 + N 3′′ d 2y + N ′′θ
and finally,
F1y = k11v1 + k12θ1 + k13v2 + k14θ2 (2.99)
where
L

∫
k ij = EI N ′′i N ′′j dx (2.100)
0
The nodal force in the y-direction at node 1 is

EI
F1y =
L3
(12 d1y + 6Lθ1 − 12d2y + 6Lθ2 ) (2.101)
If instead of virtual displacement, the virtual rotation ∂θ at a point
where moment M is applied, the moment in terms of strain energy is
∂U
M1 = (2.102)
∂θi
Applying Castigliano’s theorem at node 1 gives
∂  EI  ∂2 v  
L 2
∂U
M1 = = 
∂θ1 ∂θ1  2  ∂x 
0
2 ∫dx  (2.103)

Substituting the second derivative of the displacement function into

Equation 2.103 gives
∂  EI 
L
∫ ( N′′ d 4 2 ) dx  (2.104)
2
M1 =  1 1y + N ′′θ
2 1 + N 3′′ d 2y + N ′′θ
∂θ1  2 
0
Since the modulus of elasticity and moment are independent of x,

Equation 2.104 becomes
L
EI
2 ( N1′′d1y + N ′′θ 4 2 ) N ′′
M1 =
2 ∫
0
2 1 + N 3′′ d 2y + N ′′θ 2 dx (2.105)
or
M1 = k21v1 + k22θ1 + k23v2 + k24θ2 (2.106)
where
L

∫
k ij = EI N ′′i N ′′j dx (2.107)
0
Finally, the nodal moment at node 1 is expressed as

EI
M1 =
L3
( )
6 d1y + 4L2 θ1 − 6Ld 2y + 2L2 θ2 (2.108)
By using the same procedure for nodal force and moment at node 2,
the stiffness equation for a horizontal beam can be expressed as
 F1y   12 6L −12 6L   d1y 

    
 M1  EI  6L 4L2 −6L 2L2   θ1 
 = 3   (2.109)
 F2y  L  −12 −6L 12 −6L   d 2y 
 M 2   6L 2L2 −6L 4L2   θ2 
   
2.6 Analyzing a horizontal beam structure
A horizontal beam structure, shown in Figure 2.9, is made of two solid

cylinders with different materials and radii. Determine the displacement
and slope at the points where force or moment is applied. For Beam 1:
E = 210 GPa, D = 5 cm; and for Beam 2: E = 180 GPa, D = 4 cm, where
D is the diameter of the cylinder.
There are unlimited options for elements and nodes distribution, and
some of these options are shown in Figure 2.10. Increasing the number of
elements will definitely enhance the accuracy of the results, but only up
to a certain number of elements. After this number, the results become
independent of the number of elements. The first mesh contains just two
elements, which is the minimum to solve this problem. The second mesh
contains 3 elements, the third mesh contains 6 elements, the fourth mesh
contains 12 elements, and the fifth mesh contains 24 elements. For an
illustration purpose, the first mesh is selected because it has the mini-
mum number of elements.
The first moments of inertia of the first and second beams are required
to solve the problem, and they are
π 4 π
( )
4
I1 = R = 2.5 × 10 −2 = 3.067 × 10 −7 m 4
4 4
500 N
Beam 1 Beam 2
250 N . m
2m 1m
FIGURE 2.9 Beam structure.
F2y
Element 1 Element 2
Mesh 1
Node 1 Node 2 Node 3 M3
Mesh 2
Mesh 3
Mesh 4
Mesh 5
FIGURE 2.10 Meshes for the beam structure.

π 4 π
( )
4
I2 = R = 2.0 × 10 −2 = 1.256 × 10 −7 m 4
4 4
First, the stiffness matrix for each element is obtained using
Equation 2.85. For the first element, which has nodes 1 and 2, the stiffness
matrix is
 12 12 −12 12 
[K (1) ] =
210 × 10 9
( 3.067 × 10 )
−7 
12 16 −12 8 

23  −12 −12 12 −12 
 12 8 −12 16 
1 1 2 2
 9.66 9.66 −9.66 9.66  1
 9.66 12.88 −9.66 6.44  1
[K (1) ] = 10 4  
 −9.66 −9.66 9.66 −9.66  2
 9.66 6.44 −9.66 12.88  2
For the second element, which has nodes 2 and 3, the stiffness
matrix is
 12 6 −12 6 
[K ( 2) ] =
180 × 10 9
(1.256 × 10 )
−7 

6 4 −6 2 

13  −12 −6 12 −6 
 6 2 −6 4 
2 2 3 3
 27.13 13.56 −27.13 13.56  2

 13.56 9.04 −13.56 4.52  2
[K ( 2) ] = 10 4  
 −27.13 −13.56 27.13 −13.56  3
 13.56 4.52 −13.56 9.04  3
Assembling [K(1)] and [K(2)] using Equation 2.86 yields:
 F1y  d =0 
   9.66 9.66 −9.66 9.66 0 0   1y 
 M 1   9.66 12.88 −9.66 6.44 0 0   θ1 = 0 
 
 F2y = −500 
4  −9.66 −9.66 36.79 3.9 −27.13 13.56   d 2y 
  = 10  
 9.66 21.88 −13.56 4.56   θ2 
 M2 = 0   0
6.44 3.9

 F3y = 0  0 −27.13 −13.56 27.13 −13.56  d 3y 
 
   0 0 13.56 4.52 −13.56 9.04   θ 
 M 3 = − 250   3 
The first and second columns and rows are deleted to remove the
singularity from the stiffness matrix, and it becomes
 F2y = −500   
 
 36.79 3.9 −27.13 13.56   d 2y 
 M2 = 0   3.9 21.92 −13.56 
4.56  θ2 
 = 10  
4
 
 F3y = 0   −27.13 −13.56 27.13 −13.56   d 3y 
 M3 = −250   13.56 4.52 −13.56 9.04   θ3 
   
There are four equations and four unknowns, and solving for
displacements and rotations, the results are
d2y = –0.0279 m
d3y = –0.0562 m
θ2 = –0.0227 rad
θ3 = –0.0337 rad
2.7 Analyzing a horizontal beam structure using ANSYS
The horizontal beam structure shown in Figure 2.9 is made of two solid

cylinders with different materials and radii. Determine the maximum
displacement and reactions at the support using ANSYS. Divide each
meter of the beam by 50 elements. For Beam 1: E = 210 GPa, D = 5 cm,
and for Beam 2: E = 180 GPa, D = 4 cm.
Double Click on the Mechanical APDL Product Launcher icon

A select Structural
OK
Add...
The beam element can support the moment and the selected element
has two nodes.
A select Beam
B select 2 node 188
OK
Options…
The illustrated theory for the beam in this chapter is based on the
third-order polynomial for deflection function. Therefore, in options,
the element behavior should be changed to a cubic form.
A select Cubic Form. in Element behavior K3

OK
Close
The beam cross-sectional area is specified in the beam section. In this

example, two beams with different radii and material properties will be
modeled.
A Click on Structural > Linear > Elastic > Isotropic
The following window will appear, the modulus of elasticity for

the first beam will be specified, and any value for the Poisson ratio is
required to avoid an error message.
A type 210e9 in EX
B type 0 in PRXY
OK
In the following steps, the material of the second beam will be speci-
fied. The ID for the first beam is one, while the ID for the second beam
is two.
In the Define Material Models Behavior
A click on Material, then New Model
A type 2 in Define Material ID
OK
A click Material Model Number 2

B click on Structural > Linear > Elastic > Isotropic
The following window will appear, the modulus of elasticity for the
second beam will be specified, and any value for the Poisson ratio is
required to avoid an error message.
A type 180e9 in EX
B type 0 in PRXY
OK

Main Menu > Preprocessor > Sections > Beam > Common Sections
A select the solid cylinder in Sub-Type

B type 0.025 in R
Apply
Meshview
ANSYS graphics show the properties of the solid cylinder

A type 2 in ID
B select the geometry of a solid cylinder in Sub-Type
C type 0.02 in R
OK
The process of modeling the beam structure is started here. First,

three keypoints are created followed by creating the lines. The x- and
y-coordinates for all keypoints are specified: (0,0) is the coordinate for
Keypoint 1, (0,2) for Keypoint 2, and (0,3) for Keypoint 3.
Main Menu > Preprocessor > Modeling > Create > Keypoints >
In Active CS
A type 1 in Keypoint number

Apply

Apply

OK
Main Menu > Modeling > Create > Lines > Lines > Straight Line
Click on Keypoint 1 then 2, and then click on Keypoint 2 then 3. In
Create Straight Line window, click on
OK
ANSYS graphics show the three keypoints and two lines

The modeling task is ended at this point. Before meshing the lines,
the radius and material properties should be specified for the two lines.
All lines by default have ID number 1. Hence, only properties of Beam 2
should be changed to ID number 2. Each meter of the beam is divided
into 50 elements. Therefore, the first beam will be 100 elements, while
the second beam will be 50 elements.
Main Menu > Preprocessor > Meshing > Mesh Tool
A select Lines
B click on Set
Select Line 2 only. In Line Attributes window, click on
OK
The following Line Attributes window will show up. By selecting
number 2 for material number and element section, the properties of
number 2 are assigned to the line number 2.
A select 2 in Material number

B select 2 in Element section
OK
A click on Set in Lines

In Element Size on Picked Lines window, click on
Pick All
In Element Size on Picked Lines window, there are two options:
specify the length of elements or the number of element divisions for
a line. For the present example, each meter of the beam is divided into
50 elements. Or, the length of the elements is 0.02.
A type 0.02 in Element edge length
OK
A click on Mesh
In Mesh Lines window, click on
Pick All
Utility Menu > Plot > Nodes
ANSYS graphics show the nodes

Modeling and meshing are completed at this point. Next, the bound-
ary conditions are applied starting with the support and then the force
and moment. This order is not important for the solution. Loads are
applied at the keypoints, which will automatically be transferred to the

nodes.
Main Menu > Solution > Define Loads > Apply >
Structural > Displacement > On Keypoints
Click on Keypoint number 1. In Apply U,ROT on KPs window,

click on
OK
A select All DOF

OK
Force/Moment > On Keypoints
Click on Keypoint 2. Then, in Apply F/M on KPs window, click on
OK
A select FY in Direction of force/mom

B type –500 in Force/moment value
Apply
Click on Keypoint 3. Then, in Apply F/M on KPs window, click on
OK
A select MZ in the Direction of force/mom

B type –250 in the Force/moment value
OK
ANSYS graphics show the nodal force and moment with direction
The final step is to run the ANSYS solution. ANSYS will assemble
problem.
OK
Close
A select Def + undeformed
OK
ANSYS graphics show the beams before and after applying the loads
Main Menu > General Postproc > List Results > Nodal Solution
A click on Nodal Solution > DOF Solution > Y-Component of

displacement.
OK
A list of nodal displacements in the y-direction is shown. The maxi-

mum nodal displacement is shown at the end of the file, and node 102 has
the highest deflection of –0.057388. Comparing with only two elements
mesh solution in the previous example, maximum nodal displacement

was −0.0562 m, which is close to the ANSYS result. The nodal reactions
are determined in the following step:
Main Menu > General Postproc > List Results > Reaction Solution
A select All items
OK
The results are for node 1 only, the fixed node boundary, and the
moment reaction is listed. Notice that the reaction force and moment
at the fixed boundary are equal to in negative value the applied force
and moment, which ensures the balance of force and moment in the
structure.
2.8 Development of an arbitrary-oriented beam element
Practically, the beam elements are not horizontal, but they can be
arbitrary oriented in two- or three-dimensional space to form a com-
plex structure. For two-dimensional space, each node possesses three
degrees of freedom: x-displacement, y-displacement, and rotation.

Additionally, there are forces in the x- and y-directions and a bend-
ing moment at each node. Figure 2.11 shows an arbitrary-oriented
beam element at an angle φ with the global x-axis. At node 1, the local
x-direction and y-direction forces are F1x and F1y, respectively, and the
bending moment is M1. At node 2, the local x-direction and y-direction
forces are F2x and F2y, respectively, and the bending moment is M2.
These forces and bending moments cause the beam to bend. Node 1 is
displaced in the local x- and y-directions d1x and d1y, and rotated at an
angle θ1, while node 2 is displaced in the local x- and y-directions d2x
and d2y, and rotated at an angle θ2.
The beam has modulus of elasticity E, uniform cross-sectional
area A, moment of inertia I, and initial length L. Since the solu-
tion for axial displacements, the transverse deflection, and rotation
is independent, the bar and horizontal beam are combined to form
an axial flexural beam element. The bar element has a stiffness
equation of
 F1x  AE  1 −1   d1x 
 =    (2.110)
 F2x  L  −1 1   d2x 
and a horizontal beam element has a stiffness equation of
 F1y   12 6L −12 6L  d1y 

 
M1  EI  6L 4L2 −6L 2L2  θ1 



 = 3    (2.111)
 F2y  L  −12 −6L 12 −6L  d2y 
  6L 2L2 −6L 4L2 
 M 2    θ2 
Fy2, dy2
Fx2, dx2
2
de
No M2, θ2
I
L,
E, y
A,
Fy1, dy1
1 ϕ
de x
No
M1, θ1
Fx1, dx1
FIGURE 2.11 An arbitrary-oriented beam element at an angle φ.

Hence, the stiffness equations of bar and beam are combined as
 EA EA 
 0 0 − 0 0 
 L L 
 12 EI 6 EI 12 EI 6 EI 
 F1x   0 0 −  d 
   L3 L2 L3 L2   1x 

 F1y   6 EI 4 EI 6 EI 2 EI   d1y 
 M   0 0 −  
 1  L2 L L2 L   θ1 
 =
   
 F2x   − EI 0 0
EI
0 0   d2x 
 F2y   L L   d2y 
   12 EI 6 EI 12 EI 6 EI   
 M 2   0 − 3 − 2 0 3
− 2   θ2 
 L L L L 
 6 EI 2 EI 6 EI 4 EI 
 0 0 − 2
 L2
L L L 

(2.112)
The axial flexural beam element is lying along the local x-axis and
the local x-axis is oriented at an angle φ measured counterclockwise
from the global x-axis. The transformation matrix [T] is used to transfer
the local nodal forces and moments, and nodal displacements and rota-
tion from local coordinate to the global one. The nodal displacements
and rotation vector in a symbolic form are as follows:
{d} = [ T ]{d} (2.113)

and the nodal forces and bending moment vector in a symbolic form are
{F} = [ T ]{F} (2.114)

where
 cos θ sin θ 0 0 0 0 
 − sin θ cos θ 0 0 0 0 
 
0 0 1 0 0 0
[ T ] =   (2.115)

0 0 0 cos θ sin θ 0
 0 0 0 − sin θ cos θ 0 
 
 0 0 0 0 0 1 
{}
Since {F } = K  d , and using the transformation matrix, the
following equation is obtained:
[ T ]{F } = K  [ T ]{d} (2.116)

By multiplying both sides of Equation 2.116 by the inverse of the

transformation matrix, the global force vector is expressed as
{F } = [ T ]−1 K  [ T ]{d} (2.117)
Since the transformation matrix is orthogonal, its transport and

inverse are equal, and therefore,
{F } = [ T ]T K  [ T ]{d} (2.118)
Since the global force vector, {F} = [K]{d}, is equal to the global
force vector in Equation 2.118, the global stiffness matrix can be
written as
[ K ] = [ T ]T K  [ T ] (2.119)
The stiffness matrices, force vectors, and displacement vectors are

assembled as follows:
[K ] = ∑ K
e =1
(e)
 (2.120)
{F } = ∑ {F } (2.121)
e =1
(e)
{d} = ∑ {d } (2.122)
e =1
(e)
Finally, the global stiffness matrix is used to relate the global nodal
force vector to global displacement vector for the entire structure by
{F} = [K]{d} (2.123)
2.9 Distributed load on a beam element
Unlike the bar elements, beam elements can support a distributed load.
The distributed load can be uniform, varying linearly, or randomly dis-
tributed. The work equivalence method is used to replace the distributed
load by concentrated nodal forces and moments. Consider the beam sub-
jected to a distributed load, as shown in Figure 2.12a.
The work due to the distributed load is expressed as

∫
Wdist = P ( x ) v ( x ) dx (2.124)
0
y
P(x) F1y F2y
x M1 M2
L L
(a) (b)
FIGURE 2.12 (a) Beam that is subjected to a distributed load and (b) its equivalent nodal forces and
moments.
where P(x) is the distributed load on beam with length L and v(x) is the
transverse displacement of a beam. The work due to the discrete nodal
forces and moments is given by
Wdisc = M1θ1 + M2θ2 + F1yd1y + F2yd2y (2.125)
Hence, the distributed load can be replaced by equivalent nodal

forces and moments, as shown in Figure 2.12b. M1, M2, F1y, and F2y are
determined by setting Wdist = Wdisc, as follows:
L
 2 1 3 1
θ + θ2 ) x 3 + − 2 (d1y − d 2y ) − (2θ1 − θ2 ) x 2
∫P(x)  L (d
0
3 1y − d 2y ) +
L2( 1
  L L 
+ θ1x + d1y  dx = M1θ1 + M 2 θ2 + F1y d1y + F2y d 2y

(2.126)
For example, consider the case where the applied distributed pressure
is uniform on a horizontal beam, P(x) = −Po. Integrating Equation 2.126
with respect to x between 0 and L gives the following:
2 1
−  3 ( d1y − d 2y ) + 2 ( θ1 + θ2 ) 
L L 
L4  3 1
− − ( d1y − d 2y ) − ( 2θ1 − θ2 )  (2.127)
4  L2 L 
L3 L2
− θ1 − d1y L = M1θ1 + M2 θ2 + F1y d1y + F2y d 2y
3 2
then,
Po L
F1y = − (2.128)
2
Po L2
M1 = − (2.129)
12
Po L
F2y = − (2.130)
2
Po L2
M2 = (2.131)
12
Or, in matrix form,
 Po L 
 − 
 2 
 F1y   Po L2 
  −
 M   12 
{Fdist } =  1  =   (2.132)
 F2y   −
Po L 
 M2   2 
   
Po L2
 
 12 
In general, the concentrated nodal forces and moments, which are

equivalent to the distributed load, are added to the applied nodal forces
and moments to obtain the actual global nodal forces and moments as
follows:
[K]{d} = {F} + {Fdist} (2.133)
For a linearly increasing pressure along a beam, P(x) = −Pox, the dis-
tributed load can be replaced by equivalent nodal forces and moments,
M1, M2, F1y, and F2y. Equation 2.126 is integrated, and the equivalent
nodal forces and moments are determined by setting Wdist = Wdisc, and
the final result is
 3 PL 
− 
 20 
 F1y   PL2 
 
 M1   30 
{Fdist } =  +   (2.134)
 F2y   − 7PL 
 M2   20 
  
5PL2 
− 
 96 
2.10 nalyzing beam structure under a transient loading

A
using ANSYS
For the three-dimensional beam structure shown in Figure 2.13, a tran-

sient harmonic force, with a frequency of f = 1/500 s−1 and amplitude of
A = 100 kN, is applied at the shown location. Use ANSYS to determine
9 Fx(t)
4m
8 7
5 6
4m
4 3
y
1 x 2
m
z FIGURE 2.13 A structure
4
4m
under transient loading.
the displacement at the point where the force is applied as a function of

time. Also, create an animation file. The total time for the loading process
is 2500 seconds. The beams are made of circular cross-sectional pipe,
E = 200 GPa, and with inner and outer radii of Ri = 0.12 m, Ro = 0.15 m,
respectively. Mesh each beam by 10 elements. Use the following formula
to simulate the applied transient force:
Fx (t) = A sin (2πft)
Double clicks on the Mechanical APDL Product Launcher icon

A select Structural
OK
Add...
The beam elements can support the moment and have three-
dimensional capability.
A select Beam
B select 2 node 188
OK
Options…
OK
Close
For material properties, only the modulus of elasticity is required.

In this example, all beams have the same geometry and material
properties.

A type 200e9 in EX
B type 0 in PRXY
OK
The beam cross-sectional properties are specified in sections. The
geometry of the cross-sectional area of the pipe will be selected and
radii are specified. The number of divisions along the circumference is
20 to ensure a smooth circular cross section.
A select the cross-sectional geometry of a pipe

B type 0.12 in Ri
C type 0.15 in Ro
D type 20 in N
Apply
Meshview
OK
ANSYS graphics will show geometrical properties of the pipe
The process of modeling the beam structure is started here. First,

keypoints are created, followed by creating the line connecting the key-
points. The x- and y-coordinates of each keypoint are specified. The
coordinate for Keypoint 1 is (0,0,0).
In Active CS

B type 0, 0, and 0 in X,Y,Z Location in active CS
Apply
Similarly for other keypoints, (4,0,0) is the coordinate for Keypoint 2;

(4,0,–4) is for Keypoint 3; (0,0,–4) is for Keypoint 4; (0,4,0) is for
Keypoint 5; (4,4,0) is for Keypoint 6; (4,4,–4) is for Keypoint 7; (0,4,–4)
is for Keypoint 8; and (2,8,–2) is for Keypoint 9. To view the keypoints at
different directions, the Pan-Zoom-Rotate is used:
Utility Menu > PlotCtrls > Pan-Zoom-Rotate
A this group is for viewing a geometry at different directions

B this group is for zooming at a specific region in ANSYS graphics
C this group is for zooming and panning
D this group is for rotating the geometry
E this group is for fitting the geometry, resetting, closing, and helping
Close
click on Keypoint 1 then 5
OK
ANSYS graphics show the keypoints and lines

The modeling task is ended at this point. Before meshing the lines,
each meter of the beam is divided into 10 elements. Therefore, the
beams’ element length is 0.1 m.

In Element Sizes on Picked Lines window, click on
Pick All
The following window will show up. In Element Sizes on Picked
Lines window, there are two options, specifying either the length of
e lements or the number of element divisions. For the present example,

each meter of the beam should be divided into 10 elements. Alternatively,
the length of the element is 0.1.
A type 0.1 in Element edge length

OK
A click on Mesh
Pick All
Utility Menu > Plot > nodes

Modeling and meshing is completed at this point. Next, the solution
is switched to transient, and unsteady parameters will be specified. The
boundary conditions are applied starting with the supports, and then
the transient force. This order is not important for solution.
Main Menu > Solution > Analysis Type > New Analysis
A select Transient
OK
OK
The total time duration for the process is 2500 seconds. To ensure
that the obtained results are accurate, the total time duration is divided
into 100 time steps. Hence, the time step for this problem is 25 seconds.
Results for all time steps are stored by selecting Write every substep
option in Frequency. Otherwise, only the result at time step 2500 seconds
will be stored.
Main Menu > Solution > Analysis Type > Sol’n Control
A type 2500 in Time end of loadstep

B select Time increment
C type 25 in Time step size
D select Write every substep
OK
The function editor is used to apply transient force formula as a nodal

force. This technique is simple and convenient for this problem since an
equation for the force is given. The function is plotted to ensure correct
typing.
Displacement > On Keypoints
Click on Keypoint numbers 1, 2, 3, and 4. In Apply U,ROT on KPs
window, click on
OK
A select All DOF

OK
Main Menu > Solution > Define Loads > Apply > Functions >
Define/Edit
A type the equation: 100e3*sin(2*3.1415/500*{TIME})

B click on GRAPH
A type 0 and 2500 in X-Axis Range

B type 100 in Number Of Points
Graph
Close
Then 0 and 2500 are the range of the data on the x-axis, while 100 is
the number of points to be plotted. Number Of Points has nothing to do
with the accuracy of the results, and a higher number will just create a
smooth plot. The equation should be saved.
ANSYS graphics show the oscillating function of the force

In Equation Editor window, click on File then Save
Save the file as ForceX, and this file name is optional. After saving
the function, it is required to load it to the ANSYS solution using the
read file.
Save
Close the Function Editor window
Main Menu > Solution > Define Load > Apply > Functions >
Read File
Select the file ForceX from Open window, and then click on
Open
A type ForceX in the Table parameter name. This name is optional,

and should not be the same as the file name of the function.
OK
Utility Menu > Plot > Lines

Click on Keypoint number 9. In Apply F/M on KPs window, click on
OK
A select FX in the Direction of force/mom

B select Existing table
OK
The following window will show up to select the function.
A select FORCEX
OK
The final step is to run the ANSYS solution. ANSYS will assemble the
stiffness matrices, apply the boundary conditions, and solve the problem.
OK
Close
An animation of the deformed structure from time = 0 to 500 seconds

can be easily accomplished using animate in the PlotCtrls option in Utility
Menu. The Number of the frames in the animation over time is the number
of pictures in the avi file, while the animation time delay is the display
period between two pictures. Fifty frames produce a good resolution file.
With 0.5 s, the file duration is 250 s (500 frames × 0.5 delay).
Main Menu > General Postproc
Utility Menu > PlotCtrls > Animate > Over time …
A type 50 in Number of animation frames

B select Time Range
C type 0 and 500 in Range Minimum, Maximum

D select DOF solution
E select Deformed Shape
OK
ANSYS creates an animation for the deformation.

The displacement at any point in the structure can be displayed as a
function of time using the time history postprocessor. For this problem,
the displacement at the point where the force is applied as a function of
time is required.
Main Menu > TimeHist Postpro
A click on the green + button
A click on Nodal Solution > DOF Solution > X-Component of

displacement.
OK
Click on the Keypoint number 9, where the force is applied, and in

Node for Data window, click on
OK
A click on the graph button
OK
ANSYS graphics show the displacement history of the selected location

As shown, the maximum displacement in the x-direction is

0.00420747 m.
PROBLEM 2.1
For the bar structures shown in Figure 2.14a through 2.14d, determine
the horizontal and vertical nodal displacements, using the finite element
method, given that E = 210 GPa, A = 2.5 × 10 –4 m2, F = 0.5 kN, and
L = 0.5 m.
PROBLEM 2.2
For the bar structures shown in Figure 2.14a through 2.14d, determine
the horizontal and vertical nodal displacements, using ANSYS, given
that E = 210 GPa, A = 2.5 × 10 –4 m2, F = 0.5 kN, and L = 0.5 m.
PROBLEM 2.3
The horizontal beam structures, shown in Figure 2.15a through 2.15d,
are made of different materials and cross-sectional areas. Determine
the nodal displacements and slopes, using the finite element method,
given E1 = 270 GPa, E2 = 215 GPa, E3 = 170 GPa, I1 = 2.5 × 10 –6 m4,
I2 = 5 × 10 –6 m4, and I3 = 7.5 × 10 –6 m4.
L/2
L F 45˚
2×F
(a) (b)
2×F
L 3×F
L/2
L L
F
(c) (d)
FIGURE 2.14 Bar structures.

E1, I1 E1, I1
1.25 kN . m
2.75 m 3.25 m
(a)
0.5 kN 0.1 kN
E1, I1 E2, I2
2.5 m 2.5 m
(b)
1.25 kN 0.75 kN
E1, I1 E2, I2 E1, I1
1.0 m 1.0 m 1.0 m
(c)
1.5 kN
E1, I1 E2, I2 E3, I3
1.25 kN . m
1.5 m 3.0 m 1.0 m
(d)
FIGURE 2.15 Horizontal beam structures.
PROBLEM 2.4
The beam structures, shown in Figure 2.16a through 2.16d, are made
of solid cylinders with different materials and radii. Determine the
maximum nodal displacement and reaction at the support(s) using
ANSYS. Divide each meter of the beam by 25 elements, given
E1 = 170 GPa, E2 = 210 GPa, E3 = 270 GPa, D1 = 2 cm, D2 = 5 cm, and
D3 = 7 cm.
0.5 kN 0.75 kN
E1, D1 E2, D2
120 cm 120 m
(a)
0.75 kN
E1, D1 E2, D2 E3, D3
1.25 kN . m
125 cm 250 cm 100 cm
(b)
2.5 kN
0.1 kN
100 cm All beams 0.2 kN . m

E1, D1
75 cm 75 cm 150 cm
(c)
1.5 kN . m
D2 D2
75 cm All beams
D2 D2
E1
D1 D1
0.1 kN
75 cm 75 cm
(d)
FIGURE 2.16 Two-dimensional beam structures.
PROBLEM 2.5
For the three-dimensional beam structure, shown in Figure 2.17, tran-
sient harmonic forces with a frequency of f = 1/250 s−1 and amplitude of
A = 0.85 kN are applied at the shown locations. Use ANSYS to deter-
mine the displacement at the points where the forces are applied as a
function of time. Also, create an animation for the loading process. The
total time duration for the loading process is 1250 seconds. The beams
are made of rectangular cross-sectional pipe, as shown in Figure 2.17,
and E = 180 GPa. Mesh each beam by 10 elements. Use the following
formula to simulate the applied transient forces:
Fx (t) = A sin (2 π f t)
Fx(t)
Fx(t)
5m
1.0 cm 10 cm
10 cm
Beam cross section
5m
m
5
5m
FIGURE 2.17 Three-dimensional beam structure with transient forces.
PROBLEM 2.6
For the beam structure shown in Figure 2.18, two transient harmonic
forces are applied at the shown locations. Use ANSYS to determine the
displacement at the points where the forces are applied as a function
of time. Also, create an animation for the loading process. The total
time duration for the loading process is 5000 seconds. The beams are
made of circular cross-sectional pipe, E = 200 GPa, with inner and
outer radii of R i = 0.1 m, Ro = 0.11 m, respectively. Mesh each beam
by 10 elements. Use the following formulas to simulate the applied
transient forces:
2 πt 
F1 ( t ) = 100 sin 
 500 
2 πt 
F2 ( t ) = 200 sin 
 1000 
F1(x) F2(x)
1.25 m 2.5 m 1.25 m
FIGURE 2.18 Horizontal beam structure with transient forces.

Cha p t e r TH REE
Solid mechanics
3.1 Stress–strain relations
In this chapter, finite element development of two-dimensional solid ele-

ments is described. The development is limited to plane stress–strain
elements. The plane stress is defined as the state of stress in which the
normal and shear stresses are perpendicular to the plane. Alternatively,
the loads on the body are in the xy-plane only. The plane strain is defined
as the state of strain in which the normal and shear strain are perpen-
dicular to the plane. Stress is defined as the magnitude of force ΔF on a
unit area ΔA, as the unit area approaches zero,
∆F
σ = lim (3.1)
∆A → 0 ∆A
Figure 3.1 shows a two-dimensional stress acting on an element

with width dx and height dy. The element is treated as two-dimensional
with a unit depth. The normal stresses σx and σy are acting in the x- and
y-directions, respectively. Shear stress τxy is acting in the y-direction and
normal to the x-plane. Shear stress τyx is acting in the x-direction and normal
to the y-plane. From the moment of equilibrium, τxy must be equal to τyx.
The equilibrium equation for two-dimensional stresses can be
obtained, and the net force in the x-direction should be zero:
 ∂σ x   ∂τ yx 
 σ x +  dy − σ x dy +  τ yx + dx − τ yx dx (3.2)
∂x  ∂ y 
Simplifying Equation 3.2 yields
∂σ x ∂τ yx
+ = 0 (3.3)
∂x ∂y
Similarly, the equilibrium equation for two-dimensional stresses in
the y-direction is
∂σ y ∂τ xy
+ = 0 (3.4)
∂x ∂y
95
σy + dσy
τyx + dτyx
τxy + dτxy
σx σx + dσx
τxy
y τyx
FIGURE 3.1 Two-dimensional σy

x
state of stress.
Strain is defined as the magnitude of elongation Δδ due to stresses

divided by its original length ΔL, as ΔL approaches zero,
∆δ
ε = lim (3.5)
∆L → 0 ∆L
Figure 3.2 shows a state of two-dimensional strain. There are two

types of strain: longitudinal strain due to stress, as shown in Figure 3.2a,
and shearing strain due to shear stress, as shown in Figure 3.2b. The
shearing strain is defined as the change in value of the originally right
angle in an unstrained state. In Figure 3.2a, u is the displacement in the
x-direction and v is the displacement in the y-direction.
From Figure 3.2a, the strain in the x-direction can be expressed as
 u + du dx  − u
 dx  du
ε x = lim = (3.6)
dx → 0 dx dx
dv dy
dy
du dy
dy
dy dy
v
u dv dx
y y dx
du dx
dx dx
dx
x x
(a) (b)
FIGURE 3.2 (a) Longitudinal and (b) shear strains.

Solid mechanics 97
Similarly in the y- and z-directions, the strains are, respectively,
 dv 
 v + dy dy  − v dv
ε y = lim = (3.7)
dy → 0 dy dy
 w + dw dz  − w
 dz  dw
εz = lim = (3.8)
dz → 0 dz dz
As shown in Figure 3.2b, the shearing strain in the xy, xz-,

and yz- planes is equal to the sum of changes in angle, respectively,
∂v ∂u
dx dy
∂y ∂v ∂ u
γ xy = γ yx = ∂x + = + (3.9)
dx dy ∂x ∂ y
∂w ∂u
dx dz
∂ ∂w ∂ u
γ xz = γ zx = x + ∂z = + (3.10)
dx dz ∂x ∂z
∂v ∂w
dz dy
∂y ∂v ∂w
γ yz = γ zy = ∂z + = + (3.11)
dz dy ∂z ∂ y
The stresses and strains can be presented in a vector form as follows:
 σx 
 
 σy 
 σ z 
{σ } =  (3.12)
τ xy 
 
 τ xz 
 τ yz 
 
 ∂u 
 
 ∂x 
 ∂v 
 εx   ∂x 
   
 εy   ∂w 
 εz   ∂z 
{ε} =  =   (3.13)
γ xy ∂ u ∂v
   + 
 γ xz   ∂ y ∂x 
 γ yz   ∂ u ∂w 
   + 
 ∂z ∂x 
 ∂w ∂v 
 + 
 ∂ y ∂z 
According to Hooke’s law, the stress and strain are related by the
modulus of elasticity E. The stress–strain relation in the x-direction is
expressed as
σx = Eεx (3.14)
Elongations in the y- and z-directions are caused by stress in the

x-direction and stress lateral direction. Using Poisson’s ratio v, which is
defined as the lateral strain over the axial strain, the elongation in the
y- and z-directions due to x-direction stresses can be determined as follows:
σx
ε y = ε z = − vε x = − v (3.15)
E
The shearing stress and the shearing strain are related by modulus of
shear G. For the xy-plane,
τxy = Gγxy (3.16)
where
E
G= (3.17)
2 (1 + v )
Hence, the generalized Hooke’s laws are written as
εx =
1
E
( )
σ x − v ( σ y + σ z ) (3.18)
1
εy =
E
(σ y − v (σ x + σ z )) (3.19)
εz =
1
E
( )
σ z − v ( σ x + σ y ) (3.20)
2 (1 + v )
γ xy = τ xy (3.21)
E
2 (1 + v )
γ yz = τ yz (3.22)
E
2 (1 + v )
γ zx = τzx (3.23)
E
Expressing Equations 3.18–3.20 and 3.21–3.23 in terms of strain com-
ponents as
vE E
σx =
(1 + v )(1 − 2v )
( ε x + ε y + εz ) +
1 + v
ε x (3.24)
vE E
σy =
(1 + v )(1 − 2v )
( ε x + ε y + εz ) +
1+ v
ε y (3.25)
vE
σz =
(1 + v )(1 − 2v )
( εx + ε y + εz ) + 1 +E v εz (3.26)
Solid mechanics 99
E
τ xy = γ xy (3.27)
2 (1 + v )
E
τ yz = γ yz (3.28)
2 (1 + v )
E
τzx = γ zx (3.29)
2 (1 + v )
For plane stress cases, all stresses in the z-direction are assumed zero,
σz = τyz = τzx, and strain–stress relations are expressed as
1
εx =
E
(σ x − vσ y ) (3.30)
1
εy =
E
(σ y − vσ x ) (3.31)
−v
εz =
E
(σ x + σ y ) (3.32)
2 (1 + v )
γ xy = τ xy (3.33)
E
Stress–strain relations are expressed as
E
σx =
1 − v2
( εx + vε y ) (3.34)
E
σy =
1 − v2
( ε y + vεx ) (3.35)
E
τ xy = γ xy (3.36)
2 (1 + v )
For the plane strain case, all strains in the z-direction are assumed to
be zero, εz = γyz = γzx, and strain–stress relations are expressed as
vE E
σx =
(1 + v )(1 − 2v )
( εx + εy ) +
1+ v
ε x (3.37)
vE
σy =
(1 + v )(1 − 2v )
( εx + ε y ) + 1 +E v ε y (3.38)
vE
σz = = v ( σ x + σ y ) (3.39)
(1 + v )(1 − 2v )
E
τ xy = γ xy (3.40)
2 (1 + v )
Stress–strain relations are expressed as
1+ v
εx =
E
((1 − v ) σ x − vσ y ) (3.41)
1+ v
εy =
E
((1 − v ) σ y − vσ x ) (3.42)
2 (1 + v )
γ xy = τ xy (3.43)
E
In matrix form, the stress–strain relations in the state of plane stress
and plane strain are expressed as, respectively,
 σx   1 v 0  
  E  v 1 0   εx 
 σy =    εy  (3.44)
 τ xy  (
1 − v2 )  0 0 1 − v  
   2   γ xy 
 
 σx   1− v v 0  
  E  v 1− v 0   εx 
 σy =    εy  (3.45)
 τ xy  (1 + v )(1 − 2v )  1 − 2v  
  γ xy
0 0
   2 
 
The stress–strain relationships are symbolically expressed as
{σ} = [D]{ε} (3.46)
where [D] is called the stress–strain matrix, or simply the D matrix.
3.2 evelopment of triangular plane stress

D
and plane strain element
Figure 3.3a shows a plate subjected to tensile stress. The plate is divided

into linear triangular elements, as shown in Figure 3.3b. The linear trian-
gular element has three nodes and nodes can be displaced, but the element’s
sides remain straight. As shown in Figure 3.3c, the nodes are named i, j,
and m. At each node, there are two degrees of freedom, displacement in the
x- and y-directions. The coordinates for the three nodes i, j, and k are (xi,yi),
(xj,yj), and (xk,yk), respectively. The nodal displacement vector is given by
 ui 
 
 vi 
 uj 
{d} =   (3.47)
 vj 
 um 
 
 vm 
Therefore, each element has six degrees of freedom. Since it is a
l inear element, linear displacement functions are selected for the x- and
y-directions as follows:
u(x,y) = a1 + a2x + a3y (3.48)

Solid mechanics 101
(a)
(b)
vm
(xm, ym) um
vi
(xi, yi)
ui vj
y
uj
(xj, yj)
x
(c)
FIGURE 3.3 (a) Plate subjected to tensile stress, (b) dividing the plate into linear triangular elements,
and (c) description of a linear triangular element.
v(x,y) = a4 + a5x + a6y (3.49)
The displacement equations can be solved because there are six con-
stants, a1 to a6, and six equations. The nodal x- and y-displacements for
all nodes are expressed in the following matrix form:
 ui   1 xi y i   a1 
    
 uj  =  1 xj y j   a2  (3.50)
 um   1 xm y m   a 3 
    
 vi   1 xi yi   a 4 
    
 vj  =  1 xj y j   a5  (3.51)
 vm   1 xm y m   a 6 
    
and solving for a’s,
 a1   αi α j α m   ui 
  1   
 a2 =  βi β j βm   u j  (3.52)
 a3  2A  γ γ j γ m   u m 
   i  
and
 a4   αi α j α m   vi 
  1   
 a5 =  βi β j βm   v j  (3.53)
 a6  2A  γ γ j γ m   v m 
   i  
where A is the area of the element, which is equal to
1
A=  x i ( y j − y m ) + x j ( y m − y i ) + x m ( y i − y j )  (3.54)
2
The α, β, and γ are defined as
αi = xj ym − yj xm αj = xm yi − ym xi αm = xi yj − yi xj
βi = yj − ym βj = ym − yi βm = yi − yj (3.55)
γi = xm − xj γj = xi − xm γm = xj − xi
Substituting the values of α’s, β’s, and γ’s into Equations 3.48 and
3.49 yields
1  ( α i + β i x + γ i y ) u i + ( α j + β jx + γ j y ) u j 
u ( x, y ) =   (3.56)
2A  + ( α m + β m x + γ m y ) u m 
1  ( α i + β i x + γ i y ) v i + ( α j + β jx + γ j y ) v j 
v ( x, y ) =   (3.57)
2A  + ( α m + β m x + γ m y ) v m 
The strain vector for a two-dimensional element is given in
Equation 3.13. Using the displacement in the v- and the u-directions, the
strain vector is expressed as
 εx   β i u i + β ju j + β m u m 
  1  
{ε} =  ε y =  γ i v i + γ jv j + γ m v m  (3.58)
 γ xy  2A  γ u + β v + γ u + β v + γ u + β v 
   i i i i j j j j m m m m 
The strain vector can be expressed in a matrix form as follows:
 ui 
 β 0 β j 0 βm 0  vi 
1  i  uj 
{ε} = 0 γi 0 γj 0 γm   (3.59)
2A  γ  vj
 i βi γ j βj γ m βm   um 
 vm 
 
Solid mechanics 103
The matrix (3.59) is symbolically expressed as
{ε} = [B]{d} (3.60)
The stress–strain relationship is symbolically expressed as
{σ} = [D][B]{d} (3.61)
The principle of the minimum potential energy theory is used to

obtain the stiffness matrix [K]. The total potential energy is the sum of
the strain energy U and the potential energy of the total applied nodal
forces Fp. The strain energy is given as

∫
U = {ε}T {σ } dV (3.62)
V
Using Equations 3.60 and 3.61 in Equation 3.62, the strain energy
becomes
∫ [ B] {d} [ D][ B]{d} dV (3.63)

T T
U=
V
The total applied nodal forces can be expressed as
{Fp} = −{d}T{F} (3.64)
Applying Castigliano’s first theorem,

∂U
Fi = (3.65)
∂d i
Using expressions (3.63) and (3.64) in the definition of Castigliano’s
first theorem yields
∂U
∫ [ B] {d} [ D][ B]{d} dV = {d} {F} (3.66)
T T T

∂{d}
V
Taking the first variation of the potential energy with respect to the
nodal displacement vector yields
∫ [ B] [ D][ B]{d} dV = {F} (3.67)

T

V
Since the nodal displacements vector is related to the nodal forces

vector by [K]{d} = {F}, the stiffness matrix is expressed as

∫
[ K ] = [ B]T [ D ][ B] dV (3.68)
V
For a constant thickness element t, Equation 3.68 is rewritten as

∫
[ K ] = [ B]T [ D ][ B] t dx dy (3.69)
A
The matrices [D] and [B] are not functions of x or y, and thus the
integration of dx dy is simply the area of the element A as follows:
[K] = tA[B]T[D][B] (3.70)
The global stiffness matrix is obtained by assembling the stiffness

matrices for all elements as follows:
N
[K ] = ∑ K
e =1
(e)
 (3.71)
and the nodal forces are also assembled to form a global force vector as
follows:
N
[F ] = ∑ F
e =1
(e)
 (3.72)
Also, the global displacements vector can be obtained by
[d] = ∑ d
e =1
(e)
 (3.73)
Finally,
{F} = [K]{d} (3.74)
To determine the nodal displacements, the global stiffness matrix

must be formulated. The following steps are for solving solid mechanics
problems:
a. Divide the domain into elements and create a finite element mesh.
b. Formulate the [B] matrix for all elements.
c. Formulate the [D] matrix for all elements.
d. Formulate the [K] matrix for all elements using calculated [B] and
[D] matrices.
e. Assemble the [K] matrices to create the global [K] matrix.
f. Formulate the stiffness equation {F} = [K]{d}.
g. Apply the boundary conditions to the stiffness equation to remove
the singularity.
h. Solve the stiffness equations to determine the unknowns.
Solid mechanics 105
3.3 Analyzing rectangular plate subjected to forces
The thin plate, as shown in Figure 3.4a, is subjected to two forces at its

right edges. Determine the nodal displacements at points where forces
are applied, given E = 200 GPa, v = 0.3, and t = 0.01 m. Consider the
problem as a plane stress.
50 kN
1m
100 kN
2m
(a)
(b)
(0,1) (2,1)
3 4
Element 2
y Element 1
1 2
(0,0) x (2,0)
(c)
FIGURE 3.4 (a) Thin plate subjected to two forces, (b) element distribution for example, and (c) finite
element mesh.
There are unlimited options for elements and nodes distributions, and
some of these options are shown in Figure 3.4b. An increasing num-
ber of elements will definitely enhance the accuracy of the results, but
up to a certain number of elements. After which, the results become
independent of the number of the elements. The first mesh contains just
two elements, which is the minimum number of elements to solve this
problem. The second mesh contains four elements, the third mesh also
contains four elements, but with a different elements distribution. The
last mesh has random elements distribution. For illustration purposes,
the first mesh is selected for this example, because it has the least num-
ber of elements. Figure 3.4c shows the finite element mesh for the pres-
ent problem.
First, the [B], [D], and [K] matrices for Element 1 are formulated.
Element 1 has the coordinate shown in Figure 3.4c. Node 1 is named (i)
and its coordinate is (0,0), Node 2 is named (j) and its coordinate is (2,0),
and Node 3 is named (m) and its coordinate is (0,1).
The β’s and γ’s are required for the [B] matrix, and they are calcu-
lated using Equations 3.55; A is calculated using (3.54). We have
βi = yj − ym = 0 − 1 = −1
βj = ym − yi = 1 − 0 = 1
βm = yi − yj = 0 − 0 = 0
γi = xm − xj = 0 − 2 = −2
γj = xi − xm = 0 − 0 = 0
γm = xj − xi = 2 − 0 = 2
2A = 1 m2
Then, the [B] matrix is formulated using (3.59):
 −1 0 1 0 0 0 
1 
[ B] = 0 −2 0 0 0 2
2 
 −2 −1 0 1 2 0 
and [D] matrix is formulated using (3.44):
 1 0.3 0 
200 × 10 9  0.3 1 0 
[D] =  
1 − 0.3 
2
1 − 0.3 
0 0
 2 
 
Solid mechanics 107
Finally, the [K] matrix for Element 1 is obtained using expression (3.70):
1 1 2 2 3 3
  1
1.318 0.714 –0.549 –0.385 –0.769 –0.330
  1
 0.714 2.390 –0.330 –0.192 –0.385 –2.198 
 –0.549 –0.330 0.549 0 0 0.330  2
(1)
K  = 10 
11

 –0.385 –0.192 0 0.192 0.385 0  2
 –0.769 –0.385 0 0.385 0.770 0 
  3
 –0.330 –2.198 0.330 0 0 2.198  3

Element 2 has the coordinate shown in Figure 3.4c. Node 2 is named

(i) and its coordinate is (2,0), Node 4 is named (j) and its coordinate is
(2,1), and Node 3 is named (m) and its coordinate is (0,1). The β’s and
γ’s are
βi = yj − ym = 1 − 1 = 0
βj = ym − yi = 1 − 0 = 1
βm = yi − yj = 0 − 1 = −1
γi = xm − xj = 0 − 2 = −2
γj = xi − xm = 2 − 0 = 2
γm = xj − xi = 2 − 2 = 0
2A = 1 m2
Then the [B] matrix is formulated using (3.59):
 0 0 1 0 −1 0 
1 
[ B] = 0 −2 0 2 0 0
2 
 −2 0 2 1 0 −1 
and [D] matrix is formulated using (3.44):
 1 0.3 0 
200 × 10 9  0.3 1 0 
[D] =  
1 − 0.3 
2
1 − 0.3 
0 0
 2 
 
Finally, the [K] matrix for Element 2 is obtained using expression (3.70):
2 2 4 4 3 3
 2
0.769 0 –0.769 –0.385 0 0.385 
 0 2.198 –0.329 –2.198 0.330 0  2
 
–0.769 –0.330 1.318 0.714 –0.549 –0.385  4
K ( 2)  = 1011 
 –0.385 –2.198 0.714 2.390 –0.330 –0.192  4
 0 0.330 –0.549 –0.330 0.549 0 
  3
 0.385 0 –0.385 –0.192 0 0.192  3
Using Equation 3.71, the global matrix is formulated as shown:
1 1 2 2 3 3 4 4
1
 1.318 0.714 –0.549 –0.385 –0.769 –0.330 0 0 
 0.714 2.390 –0.330 –0.192 –0.385 –2.198 0 0  1
 
 –0.549 –0.330 1.319 0 0 0.715 –0.769 –0.385  2
11  –0.385 –0.192 0 2.390 0.715 0 –0.329 –2.198  2
[ K ] = 10 
–0.769 –0.385 0 0.714 1.319 0 0.549 –0.330  3
 
 0.330 –2.198 0.715 0 0 2.390 –0.385 –0.192  3
 0 0 –0.769 –0.330 –0.549 –0.385 1.318 0.714  4
 0 0 –0.385 –2.198 –0.330 –0.192 0.714 2.39 
 4
The force–displacement equation is assembled using expression (3.74):
 F1x   d1x = 0 
   
 F  d1y = 0 
1y

 F2x = 100 × 10 3   d 2x 
   
 F2y = 0   d 2y 
  = [K ]  
 F3x   d 3x = 0 
 F3y   d 3y = 0 
   
 F4x = 0   d 4x 
 F4y = 50 × 10 3   d 4y 
   
Eliminating the first, second, fifth, and sixth columns and rows,

 F2x = 100 × 10 3 
   1.319 0 −0.769 −0.385  d 2x 
 F2y   0 2.39 −0.329 −2.198   
d 2y 
 = 10  
11
 
 F4x   −0.769 −0.330 1.318 0.714  d 4x 
 F4y = 50 × 10 3  
 −0.385 −2.198 0.714 2.39   d 4y 
   
Finally, solving the above equations, the nodal displacements are
d2x = 1.665 × 10 –6 m
Solid mechanics 109
d2y = 2.836 × 10 –6 m
d4x = 1.156 × 10 –8 m
d4y = 3.082 × 10 –6 m
3.4 Development of rectangular plane stress element
The rectangular element has four nodes, and nodes can be displaced
but sides remain straight. It is assumed that the stress components are
constant within each element. At each node, there are two degrees of
freedom, displacement in the x- and y-directions, as shown in Figure 3.5.
Each node is subjected to two forces: vertical and horizontal. The width
and height of the element are a and b, respectively. Node 1 has coordi-
nate (0,0) and displacement (u1,v1), Node 2 has coordinate (a,0) and dis-
placement (u2,v2), Node 3 has coordinate (a,b) and displacement (u3,v3),
and Node 4 has coordinate (0,b) and displacement (u4,v4). The nodal
displacement vector is given by
 u1 
 
 v1 
 u2 
 v2 
{d} =  
 (3.75)
 u3 
 v3 
 u4 
 
 v4 
Fy4, v4 Fy3, v3
Fx4, u4 Fx3, u3
b
y
Fx1, u1 Fx2, u2
x
a
Fy1, v1 Fy2, v2
FIGURE 3.5 Rectangular element with displacements and forces at the nodes.

Therefore, there are eight degrees of freedom for each element, and
the displacement functions for the x- and the y-directions are
u(x,y) = a1 + a2x + a3y + a4xy (3.76)
v(x,y) = a5 + a6x + a7y + a8xy (3.77)
The values of constants a1 to a8 are obtained by substituting the eight

nodal coordinates in the displacement functions in terms of element
dimensions, a and b. Rearranging the displacement functions as
u(x,y) = f1u1 + f2u2 + f3u3 + f4u4 (3.78)
v(x,y) = f1v1 + f2v2 + f3v3 + f4v4 (3.79)
where functions f1, f2, f3, and f4, are defined as follows:
x y
f1 =  1 −   1 −  (3.80)
 a b
x y
f2 = 1 −  (3.81)
a b
xy
f3 = (3.82)
ab
y x
f4 = 1 −  (3.83)
b a
The strain vector for a two-dimensional element is given in Equation 3.13.

Using the displacements v and u, the strain vector is expressed as
 u1 
 v1 
 
 εx   ρ1 0 ρ2 0 ρ3 0 ρ4 0   u2 
    v2 
{ε} =  ε y  =  0 µ1 0 µ 2 0 µ 3 0 µ 4    (3.84)
u3
 γ xy   µ1 ρ1 µ 2 ρ2 µ 3 ρ3 µ 4 ρ4   
    v3 
 u4 
 v4 
 
where
1 y
ρ1 = −  1 −  (3.85)
a b
1 y
ρ2 = 1 −  (3.86)
a b
y
ρ3 = (3.87)
ab
Solid mechanics 111
y
ρ4 = − (3.88)
ab
1 x
µ1 = −  1 −  (3.89)
b a
x
µ2 = − (3.90)
ab
x
µ3 = (3.91)
ab
1 x
µ4 = 1− (3.92)
b a
The strain vector (3.84) is symbolically expressed as
{ε} = [B]{d} (3.93)
and the stress–strain relationship is expressed as
{σ} = [D][B]{d} (3.94)
The principle of the minimum potential energy theory is used to

obtain the stiffness matrix [K]. The total potential energy πp is the sum
of the strain energy U and the potential energy of the total applied nodal
forces Fp. For a constant thickness element t, the stiffness matrix is
rewritten as

∫
[ K ] = [ B]T [ D ][ B] t dx dy (3.95)
A
Using Equation 3.95, the stiffness equation for the rectangular ele-
ment with four nodes in the state of plane stress element is calculated as
 C1 C2 C4 C5 − C1 /2 − C2 C7 − C5 
 
 C2 C 3 − C 5 C 6 − C 2 − C3 /2 C 5 C8 
 C − C5 C1 − C2 C7 C5 − C1 /2 C2 
 4

 C5 C6 − C2 C3 − C5 C8 C2 − C3 /2 
[K ] =  
 − C1 /2 − C2 C7 − C5 C1 C2 C4 C5 
 − C2 − C3 /2 C5 C8 C2 C3 − C5 C6 
 
 C7 C5 − C1 /2 C2 C4 − C5 C1 − C2 
 −C C8 C2 − C3 /2 C5 C6 − C2 C3 
 5

(3.96)
where
b 1 − v a  Et
C1 =  + (3.97)
 3a 6 b  1 − v2
v 1 − v  Et
C2 =  + (3.98)
4 8  1 − v2
b 1 − v b  Et
C3 =  + (3.99)
 3a 6 a  1 − v2
b 1− v a  Et
C4 =  − + (3.100)
 3a 12 b  1 − v2
v 1 − v  Et
C5 =  + (3.101)
4 8  1 − v2
a 1 − v b  Et
C6 =  + (3.102)
 6b 6 a  1 − v2
b 1 − v a  Et
C7 =  + (3.103)
 6a 6 b  1 − v2
a 1− v b  Et
C8 =  − + (3.104)
 3b 12 a  1 − v2
The global stiffness matrix is obtained by assembling the stiffness

matrices for all elements as follows:
N
[K ] = ∑ K
e =1
(e)
 (3.105)
and the nodal forces are assembled to form a global force vector as
follows:
N
[F ] = ∑ F
e =1
(e)
 (3.106)
Also, the global displacement vector is obtained by

N
[ d ] = ∑  d(e)  (3.107)
e =1
Finally,
{F} = [K]{d} (3.108)
3.5 nalyzing a plate with a hole subjected

A
to tensile pressure using ANSYS
The square plate with a hole shown in Figure 3.6a is subjected to tensile

pressure at both vertical sides. Use ANSYS to determine the maximum
stress in the x-direction. Also, compare the ANSYS result with maximum
stress using the stress concentration factor chart. The applied tensile
Solid mechanics 113
ϕ 1.25 cm
P P
σ(ave)
P
5 cm σ(max)
(a) (b)
K
3.0
2.75
0.025 m P
2.5
2.25
2.0
R 0.00625 m
d/H
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.025 cm
(c) (d)
FIGURE 3.6 (a) A plate with a hole subjected to tensile pressure. (b) A plate with hole under
stress, and stress distribution near the hole. (c) Stress concentration factor for a plate with a hole.
(d) Symmetry boundary conditions for the problem.
pressure is 100 kN/m2, and let E = 270 GPa, v = 0.3, and consider the
plate as a plane stress with thickness of t = 0.005 m.
For the plate with a hole shown in Figure 3.6a, the stress along the
vertical symmetry line is typically assumed uniform. However, from the
experimental observations, the stress is not uniform, but it has a maxi-
mum value near the hole, and it is greater than the average stress, as
illustrated in Figure 3.6b.
The complexity of maximum stress can be conveniently treated using
the stress concentration factor K. The maximum stress is equal to the
average stress multiplied by the stress concentration factor. The defini-
tion of the stress concentration factor is
σ ( max )
K= (3.109)
σ ( ave )
The average stress is calculated as
P×H
σ ( ave ) = (3.110)
(H − d)
where H is the height of the plate, 5 cm for the present problem; P is
plate thickness; and d is the hole diameter. The stress concentration for
a solid plate with a hole can be obtained using Figure 3.6c. The x-axis
represents the ratio of the diameter of the hole to the height of the plate,
while the y-axis is the corresponding stress concentration factor. Note

that decreasing the diameter of the hole increases the stress concentra-
tion up to three times the average stress.
ANSYS is employed to determine the maximum stress. This exam-
ple is limited to structural analysis. Hence, select Structural only in the
Preferences. Solid element is used, and its shape is rectangular with four
nodes.
A select Structural
OK
Add…
Solid mechanics 115
A select Solid
B select Quad 4 node 182
OK
The Quad 4 node 182 is a rectangular-shaped element with four
nodes. Triangular elements are not available in the ANSYS elements
library. A quadratic element with eight nodes is available. The plate has
a thickness. In the option, select plane stress with thickness. In Real
Constant window, the thickness is specified.
Options…
A select Plane strs w/thk
OK
Close the Element Type window

Close
Add…
OK
Solid mechanics 117
A type 0.005 in Thickness
OK
Close the Element Type for Real Constants window
Close
For the material properties, the modulus of elasticity and Poisson’s
ratio are required to solve the problem.
A type 270e9 in EX
B type 0.3 in PRXY
OK

The geometry is modeled by creating a square and a circle. Boolean
operation is utilized to remove the circle from the square using sub-
traction. The advantage of symmetry in the problem is considered.
Only the upper right quarter is considered. Figure 3.6d shows the con-
sidered geometry for the problem. Notice that the imposed boundary
condition at the vertical symmetry line is zero displacement in the
x-direction and zero displacement in the y-direction at the horizontal
symmetry line.
Main Menu > Preprocessor > Modeling > Create >

Areas > Rectangle > By 2 Corners
A type 0 in WP X
B type 0 in WP Y
C type 0.025 in Width
D type 0.025 in Height
OK
Solid mechanics 119
Main Menu > Preprocessor > Modeling > Create > Areas > Circle >
Solid Circle
A type 0 in WP X
B type 0 in WP Y
C type 0.00625 in Radius
OK
ANSYS graphics show the square and the circle

Now, the circular area should be subtracted from the square area to
form a plate with a hole. In Boolean operations, there are a number of
operations such as adding, subtracting, dividing, and many more.
Main Menu > Preprocessor > Modeling > Operate >

Booleans > Subtract > Areas
Click on square area, and then in Subtract Areas window, click on
Apply
Click on circular area, and then in Subtract Areas window, click on
OK
A free mesh is generated using the smart mesh option, and the mesh
refinement is 1. The high stresses are expected to be at a region close to
the hole, and therefore, more elements will be added at that region using
refinement at the line.
A select Smart Size

B set the level to 1
Mesh
Solid mechanics 121
In Mesh Areas window, click on
Pick All
A select Lines
B click on Refine
Click on the curved line at the hole, and then in Refine Mesh at Line
window, click on
OK
A turn on the Yes to activate Advanced options

OK
A type 3
OK
ANSYS graphics show the mesh

Structural > Displacement > On Lines
In the ANSYS graphics, click on the left vertical line, where zero
x-direction displacement is applied. Then, in Apply U,ROT on Lines
window, click on
OK
OK
A select UX
Apply
Solid mechanics 123
In the ANSYS graphics, click on the horizontal bottom line, where

zero y-direction displacement is applied. Then, in Apply U,ROT on
Lines window, click on
OK
A select UY
OK
Pressure is applied at the right vertical side of the plate. The negative
pressure means that the pressure is tensile.
Structural > Pressure > On Lines
In the ANSYS graphics, click on the right vertical line. Then, in

Apply PRES on lines window, click on
OK
A type –100e3 in Load PRES value

OK
The final step is to run the solution. ANSYS will assemble the stiff-
ness matrices, apply the boundary conditions, and solve the problem.
Results can be plotted and listed in the General Preprocessor task.
OK
Close
The above windows indicate that the solution task is completed suc-
cessfully. The next step is for getting the results. The first figure shows the
plate before and after deformation. Inspecting this figure is very impor-
tant to determine if the problem is solved correctly. The second figure is
the stress contours in the x-direction. This figure is used to determine the
maximum stress in the x-direction, and the stress concentration factor.
OK
Solid mechanics 125
ANSYS graphics show the plate before and after deformation

The above figure indicates that the left vertical and bottom horizontal
lines are fixed, while the right vertical line is moved to the right in the
direction of the applied pressure. The result is as expected.
Main Menu > General Postproc > Plot Results > Contour Plot >
Nodal Solution
A click on Nodal Solution > Stress > X-Component of stress
OK
ANSYS graphics show the x-direction stress contours

The stress contours show the location of the maximum stress in

the x-direction. As expected, it is at the top of the hole. The accuracy
of the result can additionally be improved if the number of elements in
the model is increased. As shown in the stress contours, the maximum
stress is 360,874 N/m2. The maximum stress can also be obtained using
the stress concentration in Figure 3.6c. The average stress is
100 × 10 3 × 0.025
σ ave = = 133333.33 N/m 2
0.025 − 0.00625
Using Figure 3.6c, the stress concentration should be
K = 2.34
and the maximum stress is calculated as follows:
σmax = Kσave = 2.34 × 133333.33 = 312,000 N/m2
Comparing the ANSYS maximum stress with maximum stress

obtained from Figure 3.6c, the error is calculated as
312000 − 360874
Error = (100 ) = 15.6%
312400
3.6 Axisymmetric elements
An axisymmetric solid that is subjected to axisymmetric loading is com-

monly found, and it is considered as a special case. Hence, a complex
three-dimensional solid model is simplified to a two-dimensional model.
Moreover, there are no additional degrees of freedom in the simplifica-
tion. The axisymmetric solid model to be considered in this illustration is
a cylinder with rectangular cross-sectional area, as shown in Figure 3.7a.
Because of symmetry in the geometry and loads, no tangential degrees
of freedom are assumed and, therefore, all derivations are independent
of the circumferential angular variables. The plane strain finite element
that is developed in this chapter can be extended to an axisymmetric
element, by replacing the integration along the thickness to along the cir-
cumference. The cross-sectional area of the axisymmetric solid model is
shown in Figure 3.7b.
A linear triangular element is used to mesh the cross-sectional area,
as shown in Figure 3.7c. The nodal displacement functions in the r- and
the z-directions are
u(r,z) = a1 + a2r + a3z (3.111)
w(r,z) = a4 + a5r + a6z (3.112)

Solid mechanics 127
Z
Ro
Ri Z
Ro
Ri
wm
H um
H wi
ui wj
r z
r uj
r
(a) (b) (c)
FIGURE 3.7 (a) Axisymmetric solid model, (b) cross-sectional area of the axisymmetric solid model,
and (c) an axisymmetric element.
The cylindrical coordinates for the three nodes are (r1, z1), (r2, z2),
and (r3, z3). The displacement functions can be expressed in matrix
form for all nodes as
 u1   1 r1 z1 0 0 0  a1 
    
 u2   1 r2 z2 0 0 0  a2 
 u3   1 r3 z3 0 0 0  a3 
 =    (3.113)
 z1   0 0 0 1 r1 z1  a4 
 z2   0 0 0 1 r2 z2  a5 
 z3   0 0 0 1 r3 z3  a6 
    
and symbolically:
{d} = [B]{a} (3.114)
The constants a’s can be obtained by inverting the matrix [B] as

follows:
{a} = [B]−1{d} (3.115)

The inverse of the matrix [B] is calculated as
 ω1 ω2 ω3 0 0 0 
 
 ω4 ω5 ω6 0 0 0 
1 ω7 ω8 ω9 0 0 0 
[ B] = 
−1
 (3.116)
γ 0 0 0 ω1 ω2 ω3 
 0 0 0 ω4 ω5 ω6 
 0 0 0 ω7 ω8 ω9 
 
where
ω1 = r2z3 − r3z2 ω2 = r3z1 − r1z3 ω3 = r1z2 − r2z1

ω 4 = z 2 − z 3 ω5 = z3 − z1 ω6 = z1 − z2 (3.117)
ω7 = r3 − r2 ω8 = r1 − r3 ω9 = r2 − r1
γ = r1 (z2 − z3) + r2 (z3 − z1) + r3 (z1 − z2)
The unit elongation of the element in radial direction is expressed as

∂u
εr = (3.118)
∂r
and the total tangential strain is given by
u ∂v
εθ = + (3.119)
r r ∂θ
where the second term is the tangential strain due to the displacement in
the y-direction. The shearing strain is expressed as
v ∂ u ∂v
γ rθ = − + + (3.120)
r r ∂θ ∂r
For an axisymmetric deformation, the displacement in the θ-direction
is equal to zero, and hence, the strain and shearing strain are expressed as
∂u
εr = = a 2 (3.121)
∂r
u a1 az
εθ = = + a 2 + 3 (3.122)
r r r
∂w
εz = = a 6 (3.123)
∂z
∂ u ∂w
γ rz = + = a 3 + a 5 (3.124)
∂z ∂r
γrθ = 0 (3.125)
γθz = 0 (3.126)
Strain and shearing strain are expressed in matrix form as
 a1 
    
εr 0 1 0 0 0 0  a2 
   
 εθ   1/r 1 z /r 0 0 0 a3  (3.127)
 =  
 εz   0 0 0 0 0 1  a4 
 γ rz   0 0 1 0 1 0 
 a5 
   
a6
 
Solid mechanics 129
Equation 3.127 is symbolically expressed as
{ε} = [G]{a} (3.128)
Using Equation 3.115 in Equation 3.128 yields
{ε} = [A]{d} (3.129)
[A] = [G][B]−1 (3.130)
For an isotropic material, and the case of thermal strain due to tem-
perature rise ΔT, the strain–stress relationship is expressed as
   1− ν ν ν 0     α ∆T 
σr    εr
  ν 1− ν ν 0   
 σθ  E    εθ   α ∆T 
  =  ν ν 1 − ν 0   −  
 σz  (1 + ν )(1 − 2 ν )  1  
εz   α ∆T 
 τ rz   0 0 0 − ν   γ rz   0 
 2     
(3.131)
where α is the thermal coefficient of expansion. The symbolic expres-

sion for Equation 3.131 is
{σ} = [C][A]{d} − [C]{εi} (3.132)
Castigliano’s first theorem, which is the same as Equation 3.74 with

an additional term for thermal strain, is expressed as follows:
{F} = [K]{d} − {Fi} (3.133)
where the stiffness matrix is obtained as

∫
[ K ] = [ A ]T [ C ][ A ] dV (3.134)
Putting Equation 3.130 into Equations 3.133 and 3.134 yields
( ) ∫ [G] [ C ]{ε i } dV (3.135)

T
 F i  = [ B]−1
T

Finally, the stiffness matrix is
[ K ] = ([ B]−1 )  [ G ]T [ C ][ G ] dV  [ B]−1 (3.136)

 
∫
T

 
3.7 isplacement analysis of a vessel under

D
transient loading with ANSYS
The vessel shown in Figure 3.8a is subjected to internal transient pres-

sure loading. Determine the maximum displacement in the radial direc-
tion of the vessel as a function of time. Take the advantage of symmetry
in the problem to reduce the computational size. The inner and outer
radii are 0.25 and 0.265 m, respectively, and the height of the vessel
is 0.75 m. Let E = 180 GPa and v = 0.33. The upper and lower hori-
zontal surfaces are fixed. The total time duration for the process is
1000 seconds. The internal pressure is a function of time according to
the following equation:
2π
P ( t ) = 100 × 10 3 + 100 × 10 3 sin  ( t + 75)
 100
This example is limited to structural analysis. Hence, select

Structural only. Solid element is used, and its shape is rectangular with
four nodes. The following steps are for selecting the element type and
its behavior:
Ro
Ri
Ro
dx = 0
Ri dy = 0
H
P(t) H/2
y
x
dy = 0
(a) (b)
FIGURE 3.8 (a) A vessel subjected to time-dependent pressure at the inner surface. (b) The axisym-
metric model for the problem.
Solid mechanics 131
A select Structural
OK
Add…
A select Solid
OK
In the element type, Quad means that a quadratic element is selected.

The first digit of number 42 is the number of nodes in the element, and
the second digit is the number of degrees of freedom in each node,
which are the x- and y-displacements. In the following steps, the element
behavior is changed from plane stress to axisymmetric. The results from
the axisymmetric analysis should be more accurate than the results from
three-dimensional elements, and the computational time will be less.
Options…
A select Axisymmetric in Element behavior
OK
Close the Element Types window
Close
Solid mechanics 133
For the material properties, the elasticity and Poisson ratio are
required to solve the problem.
The following windows will show up.
A type 180e9 in EX
B type 0.33 in PRXY
OK
The geometry is simply a rectangle. The advantage of symmetry in

the problem is considered, and only the cross-sectional area of the ves-
sel is considered. Figure 3.8b shows the considered geometry used to
solve the problem. Notice that the axis of rotation is the y-axis, and no
geometry is allowed in the negative x-axis region. The geometry rotates
about the positive y-axis. Additionally, only the upper half of the vessel
will be modeled.

A type 0.25 in WP X
B type 0 in WP Y
C type .015 in Width
D type .375 in Height
OK
Solid mechanics 135
A select Smart Size

Mesh
Pick All
ANSYS graphics show the mesh of the cross-sectional area of the vessel
Modeling and meshing are completed at this point. Next, the solution
is switched to transient, and unsteady parameters are specified. The
boundary conditions are applied starting with the displacements, then
the transient pressure. This order is not important for solution.
A select Transient
OK
OK
The total time duration for the process is 1000 seconds and to ensure
obtaining accurate results, the process duration is divided into 200 time
steps. Hence, the time step for this problem is 5 seconds. Results for
all time steps are stored by selecting the Write every substep option in
Frequency.
Solid mechanics 137
Main Menu > Solution > Analysis Type > Sol’n Control
A type 1000 in Time at end of loadstep

D select Write every substep
OK
The function editor is used to apply the transient pressure equation at

the inner surface. This technique is simple and convenient for this problem
because an equation for the pressure is given. The upper horizontal line is
fixed, while the lower horizontal line is fixed in the y-direction only due to
the symmetry. Transient pressure is applied at the left vertical line.
In the ANSYS graphics, click on the top lines that are fixed, and in
Apply U,ROT on Lines window, click on
OK
A select All DOF

Apply
In the ANSYS graphics, click on the bottom horizontal line, and in

Apply U,ROT on Lines window, click on
OK
A select UY
OK
Functions > Define/Edit
A Type the equation: 100e3+100e3*sin(2*3.1415/100*({time}+75))

B File → Save
Solid mechanics 139
Save the file as PreTran, and this file name is optional. After saving
the function, it is required to load it to the ANSYS solution using the
read file, and then close the Function Editor window.
Functions > Read File
Select the file PreTran, then open, and the following window will
show up.
A type PreTran in Table parameter name and this name is optional,

and it should not be the same as the file name of the function.
OK
Structural > Pressure > On Lines
In the ANSYS graphics, click on the left vertical line where pressure
is applied, and then in Apply PRES on lines window, click on
OK
A select Existing table
OK
A select PRETRAN
OK
The final step in the solution task is to run the solution. ANSYS
will assemble the stiffness matrices, apply the boundary conditions,
and solve the problem. Results can be plotted and listed in the General
Preprocessor task.
Solid mechanics 141
OK
Close
The above window indicates that the solution task is successfully

accomplished. The next step is for getting the results. First, the results
at time of 75 seconds is uploaded, and the deformed shape and displace-
ment contours in the x-direction are displayed.
Main Menu > General Postproc > Read Results > By Pick
A select set 15
Read
Close
OK
ANSYS graphics show the vessel wall before and after deformation
The above figure indicates that the upper line is fixed, while the
b ottom line is moved to the outside. The maximum displacement is
exactly at the middle of the vessel, as expected.
Nodal Solution
Solid mechanics 143
A click on Nodal Solution → DOF Solution → X-Component of

displacement
OK
ANSYS graphics show the contours of the displacement in the x-direction

The maximum displacement is equal to 0.231 × 10 –5 m. To visual-
ize the results in three-dimensional space, the vessel wall is partially
expanded in 180° as follows:
Utility Menu > PlotCtrls > Style > Symmetry Expansion >

2D Axi-Symmetric
A select 1/2 expansion
OK
Utility Menu > PlotCtrls > Pan-Zoom-Rotate

Click on the isometric view.
ANSYS graphics show the x-direction displacement

contours in three dimensions
Animation of the deformation from time = 0–100 seconds, one cycle,
can be easily done using animate in the PlotCtrol options. The Number
of the frames in the animate over time is the number of pictures in the
avi file, while the animation time delay is the display period between
two pictures. Fifty animation frames produce a good resolution anima-
tion file and with 0.5 second, the animation file duration is 25 seconds
(50 frames × 0.5 delay).

B select Time Range
C type 0 and 100 for Range Minimum, Maximum
Solid mechanics 145
D select DOF solution

E select Translation UX
OK
ANSYS creates an animation file for the deformation.
3.8 Contact element analysis
In this section, the contact pressure distribution between two bodies

held in contact is determined. Figure 3.9 shows two parallel cylinders of
length L in contact. The upper cylinder has a radius of r1, while the lower
cylinder has a radius of r2. Force F is applied to the upper and lower
cylinders. At the contact region, a semielliptical contact pressure distri-
bution is developed within each cylinder. The width of the contact line
is 2a, and the contact pressure varies from zero at the end of the contact
line to the largest value Po at the center of the contact line.
The contact width and maximum contact pressure can be expressed,
respectively, as
1/2
 4Fr1r2  1 − v12 1 − v 22  
a=  +  (3.137)
 πL ( r1 + r2 )  E1 E 2  
2 F
Po = (3.138)
π aL
where E is the modulus of elasticity and v is Poisson’s ratio. If the two
cylinders are made of the same material and Poisson’s ratio is equal to
0.3, the expressions (3.137) and (3.138) are simplified as follows:
1/ 2
 F  r1 r2  
a = 1.52    (3.139)
 EL  r1 + r2  
r1
Po
2a
r2
F
FIGURE 3.9 Two cylinders in contact.
1/2
 FE  r1 + r2  
Po = 0.418     (3.140)
 L  r1r2  
Expressions (3.139) and (3.140) can be used to determine the contact
width and maximum pressure of a contact problem of a circular cylinder
and a flat surface, as shown in Figure 3.10. If the radius of the second
cylinder is very large, r2 = ∞, the contact width and maximum contact
pressure are expressed as
1/2
Fr
a = 1.52  1  (3.141)
 EL 
1/2
 FE 
Po = 0.418   (3.142)
 Lr1 
When two spheres are in contact with different materials and loaded
with force F, the contact area will have a circular shape of diameter a.
The contact area and maximum stress can be expressed as
( ) ( )
1/3
 3F 1 − v12 /E1 + 1 − v 22 /E 2 
a=  (3.143)
 4 1/r1 + 1/r2 
F
Po = 1.5 (3.144)
πa 2
If the sphere of radius r1 contacting a large flat body of the same mate-
rial, the expressions of the contact area, and the maximum stress for the
contact can be obtained by substituting r2 = ∞ in Equation 3.143, and then
the contact diameter and maximum pressure become
1/3
3Fr
a = 0.88  1  (3.145)
 E 
1/3
 FE 2 
Po = 0.62  2  (3.146)
 4r 
r1
FIGURE 3.10 A cylinder and a flat plate in contact.
Solid mechanics 147
3.9 Two horizontal cylinders in contact

analysis using ANSYS
Two horizontal cylinders are placed close to each other, as shown in

Figure 3.11. The two cylinders have the same radius, r = 0.01 m, and force
of 10 kN is applied to upper and lower cylinders, as shown in the figure.
As a result, the cylinders will move and meet. Determine the maximum
pressure at the contact region, given that E = 200 GPa and v = 0.3 for
both cylinders.
Contact problems are highly nonlinear and require large computer
resources to solve. Additionally, the physics behind the contact problem
is relatively complex, and it is important to understand the physics of the
problem to solve it as accurately as possible. Too many parameters are
required and just using the ANSYS default settings will give a reason-
able result. Contact problems present two main difficulties. First, the
regions of contact cannot be accurately predicted. The contact region
depends on the loads, materials, boundary conditions, and other fac-
tors. Second, most contact problems need to account for friction. There
are several friction models to choose from, and all are nonlinear, which
makes solution convergence even more difficult. The basic steps for per-
forming a contact analysis are as follows:
1. Create the model geometry and mesh.
2. Identify the contact pairs. The contact pairs are the region in which
contact might occur during the deformation of the model. Once
potential contact surfaces are identified, they must be defined via
target and contact elements.
3. Select contact and target surfaces. Contact elements are con-
strained against penetrating the target surface, but target elements
can penetrate through the contact surface. If a convex surface is
F
FIGURE 3.11 A contact element problem.
expected to meet a flat surface, the convex surface should be the

contact surface. If one surface is smaller than the other, the smaller
surface should be the contact surface.
4. Apply the boundary conditions. Any type of boundary conditions
can be applied in the contact problems.
5. Define solution options and load steps. Convergence behavior for
contact problems depends on the geometry and boundaries of the
problem. The time step size must be small enough to ensure the
convergence. The time step size is specified by a number of steps
or the time step size itself. However, the best way to set an accurate
time step size is to turn on the automatic time stepping in ANSYS
solution options.
6. Solve the contact problem. Ensure that the problem is fully
converged.
7. Finally, get the contact results. Results from a contact analysis
consist mainly of displacements, stresses, strains, reaction forces,
and the contact information, such as contact pressure.

in Preferences. Solid element is used, and its shape is rectangular with
four nodes. The contact simulation requires another two element types:
the target and contact elements. These two elements will be introduced
into the model when the contact wizard tool is activated.
A select Structural
OK
Solid mechanics 149
Add...
A select Solid
OK
Close
The following window will show up. For the material properties,
the modulus of elasticity and Poisson’s ratio are required to solve the
problem.
A type 200e9 in EX
B type 0.3 in PRXY
OK
Next, the geometry of the problem is created, which is two circles.

The expected contact surfaces should be defined in the contact wizard.
Additionally, the contact region should have a fine mesh for accurate
numerical predication. Hence, a small circle of a radius of 0.002 m is
created at the contact region and will have a fine mesh.
Solid mechanics 151
Main Menu > Preprocessor > Modeling > Create > Areas >
Circle > Solid Circle
A type 0 in WP X
B type 0.01 in WP Y
Apply
A type 0 in WP X
B type –0.01 in WP Y
Apply
A type 0 in WP X
B type 0 in WP Y
OK
ANSYS graphics show the created three circles

Solid mechanics 153
First, the three areas are overlapped. Second, since there is s ymmetry
in the problem, a vertical line is created to split the entire mode by half.
Only the right symmetry side is considered. Finally, extra areas are
deleted.
Main Menu > Preprocessor > Modeling > Operate > Booleans >
Overlap > Areas
In Overlap Areas window, click on
Pick All
ANSYS Main Menu > Preprocessor > Modeling > Create >
Lines > Lines > Straight Line
Create a vertical line by clicking on two keypoints. The first at the top
of the upper circle, and the second at the bottom of the lower circle, and
then in create Straight Line window, click on
OK
ANSYS graphics show the created three overlapped

circles and a vertical line
ANSYS Main Menu > Preprocessor > Modeling >
Operate > Booleans > Divide > Area by Line
To select all three circles, in Divide Area by Line window, click on
Pick All
Click on the created vertical line. In Divide Area by Line window,
click on
OK
Main Menu > Preprocessor > Modeling > Delete > Area and below
Select the left areas and the small area on the right side that is between
the two circles. Then, in Delete Area and Below window, click on
OK
ANSYS graphics show the final geometry

A Click on Set in Lines

Solid mechanics 155
Select the lines that are far away from the contact area, as illustrated
in the following figure.
Apply
A type 100 in No. of element divisions
Apply
Select the lines that are close to the contact area, as illustrated in the
following figure.
OK
OK
Solid mechanics 157
A Mesh
In Mesh Areas, click on
Pick All
ANSYS graphics shows the mesh

In this problem, the top surface of the lower circle is the target surface
because it is stationary, while the bottom surface of the upper circle is
the contact surface because it is moving. Both contact and target sur-
faces are associated with the deformable bodies. These two surfaces
together comprise the contact pair. The Contact Manager is very effec-
tive in defining, viewing, and editing the contact pairs. In addition, all
contact pairs for the entire model can be managed. The Contact Wizard,
which is accessed from the Contact Manager, makes the process of cre-
ating contact pairs very efficient.
Main Menu > Preprocessor > Modeling > Create > Contact Pair
A click on the contact wizard
The contact wizard is used to select the target and contact surfaces.
First, the target surface should be selected and then the contact surface.
A click on Lines
B click on Pick Target
Click on the top surfaces of the lower circle, as shown in the following
figure.
Solid mechanics 159
In Select Lines for Target window, click on
OK
Then in the Contact Wizard window, click on
Next >
A click on Lines
B click on Pick Contact
Click on the bottom surfaces of the upper circle, as shown in the

following figure.
In Select Lines for Contact window, click on
OK
Then in the Contact Wizard, click on
Next >
Create >
Solid mechanics 161
Finish
ANSYS graphics show the contact region

Close the Contact Manager window
In the ANSYS graphics, click on all vertical left lines, where zero
x-displacement is applied, and then in Apply U,ROT on Lines, click on
OK
A select UX
OK
Forces are applied at the top keypoint of the upper circle and bottom
keypoint of the lower circle.
In the ANSYS graphics, click on the bottom keypoint at the lower

circle, where the force is applied, and then in Apply F/M on KPs
window, click on
OK

B type 10e3 in Force/moment value
Apply
In the ANSYS graphics, click on the top keypoint at the upper circle,
where the force is applied, and then in Apply F/M on KPs window,
click on
OK

B type −10e3 in Force/moment value
OK
Solid mechanics 163
OK
Yes
Close
The above window indicates that the solution task is accomplished
successfully. The next step is getting the results. The scale is changed to
true scale for actual dimensional display in ANSYS graphics.
Utility Menu > PlotCtrls > Style > Displacement Scaling
A select 1.0 (true scale)
OK
Nodal Solution
A click on Nodal Solution > Stress > Y-Component of stress
OK
ANSYS graphics show the contours of stress in the y-direction
The above figure indicates that the maximum stress in the y-direction
is equal to 0.274 × 109 Pa. The maximum stress in the y-direction can be
theoretically obtained using expression (3.140) as follows:
Po = 0.418
FE r1 + r2
= 0.418
( )
10 × 10 3 200 × 10 9 0.01 + 0.01
L r1r2 1 0.01 × 0.01
= 0.264 × 10 9 Pa
The theoretical maximum stress is very close to the result obtained by

ANSYS with marginal error of 3.78%.
Solid mechanics 165
PROBLEM 3.1
The thin plates, as shown in Figure 3.12a and 3.12b, are subjected to force.
Determine the nodal displacements at the point where the force is applied,
using the finite element method, given E = 180 GPa, v = 0.33, and t =
0.02 m. Consider the plates as plane stress problems, and use only one
triangular element.
PROBLEM 3.2
The thin plates, as shown in Figure 3.13a and 3.13b, are subjected to
force(s). Determine the nodal displacements, using the finite element
method, given E = 240 GPa, v = 0.3, and t = 0.01 m. Use linear triangu-
lar elements and the suggested two-element mesh. Consider the plates as
plane stress problems.
PROBLEM 3.3
The thin plates, as shown in Figure 3.14a and 3.14b, are subjected to
force(s). Determine the nodal displacements, using the finite element
method, given E = 195 GPa, v = 0.3, and t = 0.015 m. Use linear trian-
gular elements and the suggested finite element mesh. Consider them as
plane stress problems.
PROBLEM 3.4
The square plate with a hole, as shown in Figure 3.15, is subjected to ten-
sile pressure at both vertical sides. Use ANSYS to determine the maxi-
mum stress in the x-direction. Also, compare the ANSYS result with
maximum stress using the stress concentration factor figure. The applied
pressure is 75 kN/m2, and let E = 230 GPa, v = 0.30, and consider the
plate as a plane stress with thickness of t = 0.01 m. Take the advantage
of symmetry in the problem to reduce the computational size.
1m 1.5 m 100 kN
100 kN
2m 1.5 m
(a) (b)
FIGURE 3.12 Thin plate subjected to force.

50 kN
0.25 m
100 kN
0.75 m
(a)
0.1 m
15 kN
0.1 m
(b)
0.25 m
85 kN
0.25 m 0.25 m
(a)
75 kN
0.5 m
75 kN
1.0 m 1.0 m
(b)

Solid mechanics 167
ϕ 1.0 cm
P P
5 cm FIGURE 3.15 A plate with a hole

subjected to tensile pressure.
PROBLEM 3.5
A notched rectangular plate is in tension, as shown in Figure 3.16.
Pressure (P) is applied at the left and right vertical sides of the plate,
given applied pressure P = 200 kPa, E = 180 GPa, v = 0.3, and t = 0.05 m,
where t is the thickness of the plate. Determine maximum displacement
in the x-direction and calculate the stress concentration factor (K) using
the ANSYS result. Take advantage of symmetry in the problem to reduce
the computational size.
PROBLEM 3.6
A flat plate is axially loaded, as shown in Figure 3.17. Pressure P = 150 kPa
is applied at the left and right vertical sides of the plate. E = 210 GPa,
v = 0.33, and t = 0.01 m, where t is the thickness of the plate. Determine
using ANSYS the maximum displacement in the x-direction and the stress
concentration factor (K). Take the advantage of symmetry in the problem
to reduce the computational size.
0.06 m
0.04 m 0.02 m 0.04 m
FIGURE 3.16 A notched rectangular plate subjected to tensile pressure.

R 1.0 cm
3.5 cm
3.5 cm 4.0 cm 3.5 cm
FIGURE 3.17 A plate subjected to tensile pressure.
PROBLEM 3.7
Determine the maximum displacement in the radial direction of the
vessel shown in Figure 3.18. Take the advantage of symmetry in the
problem to reduce the computational size. The inner and outer radii
are 0.15 and 0.1625 m, respectively, and the height of the vessel is
0.95 m. Let E = 230 GPa and v = 0.3. Pressure of P = 1.5 MPa is
applied at the inner surface of the vessel. The upper and lower hori-
zontal surfaces are fixed.
PROBLEM 3.8
The square plate with a hole shown in Figure 3.15 is subjected to tran-
sient tensile pressure at both vertical sides. Use ANSYS to determine
the maximum stress in the x-direction as a function of time, and let
E = 230 GPa, v = 0.30, and consider the plate as a plane stress with
Ro
Ri
FIGURE 3.18 A vessel subjected

to pressure at the inner surface.
Solid mechanics 169
thickness of t = 0.01 m. Take the advantage of symmetry in the problem

to reduce the computational size. The total time duration for the process
is 500 seconds. The applied pressure (in N/m2) is a function of time
according to the following expression:
2π 
P ( t ) = 75 × 10 3 + 75 × 10 3 sin  t
 50 

PROBLEM 3.9
The square plate geometry with a hole shown in Figure 3.16 is subjected
to transient tensile pressure at both vertical sides. Use ANSYS to deter-
mine the maximum stress in the x-direction as a function of time, and
let E = 180 GPa, v = 0.33 and consider the plate as a plane stress with
thickness of t = 0.02 m. Take the advantage of symmetry in the problem
to reduce the computational size. The total time duration for the process
is 2500 seconds. The applied pressure (in N/m2) is a function of time
according to the following expression:
2π 
P ( t ) = 125 × 10 3 + 125 × 10 3 sin  t
 500 

PROBLEM 3.10
A horizontal cylinder is placed on a flat surface, as shown in Figure 3.19.
The cylinder has a radius of r = 0.025 m, and a force of 25 kN is applied
to the cylinder. The cylinder will compress and penetrate the flat sur-
face. Determine the maximum stress at the contact region, given that
E = 180 GPa and v = 0.3 for the cylinder and flat plate.
PROBLEM 3.11
Two horizontal cylinders are placed close to each other, as shown in
Figure 3.20. The two cylinders have different radii, r1 = 0.01 m and
FIGURE 3.19 A horizontal cylinder on a flat plate.

r1
r2
FIGURE 3.20 A contact element problem.
r2 = 0.02 m, and force of 50 kN is applied only to the upper cylinder,

while the lower cylinder is placed on a flat surface. As a result, the upper
cylinder will move downward and meet the lower cylinder. Determine
the maximum stress at the contact region. Given that E = 180 GPa and
v = 0.33 for both cylinders.
Cha p t e r F OUR
Vibration
4.1 Vibration analysis
In this chapter, the basic formulation of the motion of a single degree

of freedom is presented. The aim of this introduction is to discuss some
important concepts necessary for solving and understating vibration
problems. Figure 4.1 shows a mass–spring system that is subjected to a
time-dependent force F(t), where m is the mass of the system and k is the
stiffness of the spring.
Applying Newton’s second law, the equation of motion of the spring–
mass element in the x-direction is expressed as
+ kx = F ( t ) (4.1)
mx
Equation 4.1 can also be symbolically expressed as
[ m ]{x} + [ k ]{x} = {F } (4.2)

where [k] is the stiffness matrix, [m] is mass matrix, {x} is the nodal
displacement vector, and {x} is the acceleration vector. If the applied
force F(t) is equal to zero, then Equation 4.1 becomes a homogenous
ordinary differential equation. The solution of this equation gives the
natural frequency of the oscillating mass:
k
ω2 = , (4.3)
m
where ω is called the natural frequency of the free vibration of the mass.
The natural frequency depends on the stiffness of the spring and the
mass of the system. Substituting (4.3) into the equation of motion of free
vibration (4.1) yields
x + ω 2 x = 0 (4.4)

The solution of Equation 4.4 is
x(t) = A cos(ωt) + B sin(ωt) (4.5)
where the constants A and B can be determined using the initial
conditions.
171
k
m F(t)
FIGURE 4.1 Spring–mass system subjected to a time-dependent force.
4.2 evelopment of stiffness equations for

D
axial vibration of a bar element
Lumped mass formulation is one of the common ways to formulate the

stiffness matrix for dynamics analysis. Figure 4.2 shows a bar with ini-
tial length L, modulus of elasticity E, cross-sectional area A, and den-
sity ρ. To account for the mass inertia of the bar, the mass is distributed
and concentrated in the nodes, as shown in Figure 4.2.
The inertia forces are added to the nodal forces for the bar element
equation, which is used in static analyses, as follows:
 ρAL 
 F1 −
u1  EA 
 2  1 −1   d1  (4.6)
 =   
 F2 − ρAL
u2  L  −1 1   d 2 
 2 
The negative values of the inertial forces indicate that they are resist-
ing the elastic forces F1 and F2. Equation 4.6 is rearranged as follows:
 d 
 F1  EA  1 −1   d1  ρAL  1 0   1 
 =   +     (4.7)
 F2  L  −1 1   d 2  2  0 1  
 d2 
or symbolically,
{F } = [ k ]{d} + [ m ]{d} (4.8)

For free vibration, the nodal force vector is equal to zero. For a har-
monic motion, it is assumed that the displacements vary harmonically
with respect to time with frequency ω:
{d} = {U} sin(ωt) (4.9)
The second derivative of the displacement vector is the acceleration:
{d} = −ω 2
{U} sin ( ωt ) (4.10)
Expression (4.9) into (4.10) yields
{d} = −ω {d} (4.11)

2
Vibr ation 173
Node 1 A, ρ Node 2
F1 F2
L
ρAL ρAL FIGURE 4.2 Bar element and

m1 = m2 =
2 2 the concentrated mass at the
nodes.
Substituting the acceleration vector in Equation 4.7 yields
 F1  EA  1 −1   d1  2 ρAL  1 0   d1 
 =  −ω    (4.12)

 F2  L  −1 1   d 2  2  0 1   d 2 
Equation 4.12 can be rearranged as
 F1   EA  1 −1  ρAL  1 0   d1 
 =   − ω2     (4.13)
 F2   L  −1 1  2  0 1   d 2 
The global stiffness matrix can be obtained by assembling the stiff-

ness matrix for each element as follows:
N
[K ] = ∑ K
e =1
(e)
 (4.14)
The nodal forces are also assembled to form a global force vector:
[F ] = ∑ F
e =1
(e)
 (4.15)
Similarly for the displacement vector,
[ d ] = ∑  d(e)  (4.16)
e =1
Finally, the nodal forces and the displacement relationship is

expressed as
{F} = [K]{d} (4.17)
4.3 Natural frequencies of axial vibration of a bar element
Determine the first two natural frequencies of a cantilever bar, as shown

in Figure 4.3. The total length of the bar is 2 m. Model the bar with two
elements only, and let E = 200 GPa, A = 2 × 10 –3 m2, and ρ = 5000 kg/m3.
E, A, ρ
2m
Element 1 Element 2
FIGURE 4.3 Cantilever bar and Node 1 Node 2 Node 3
finite element mesh.
Element 1 has nodes 1 and 2, and its length is L = 1.0 m. Using

Equation 4.13 for nodal forces and displacement relationship,
 F1   EA  1 −1  ρAL  1 0    d1 
 =    − ω2    
 F2   L  −1 1  2  0 1    d 2 
Nodal displacement at node 1 and nodal force at node 2 are equal

to zero. Element 2 has nodes 2 and 3, and its length is L = 1 m. Using
Equation 4.13,
 F2   EA  1 −1  ρAL  1 0   d 2 
 =   − ω2    
 F3   L  −1 1  2  0 1   d 3 

Nodal forces at node 3 are equal to zero. Assembling the nodal and
displacement vectors and stiffness matrix for both elements yields
 F1    1 −1 0   1 0 0   d1 = 0 
   EA  2 ρAL   
 F2 = 0 =

−1 2 −1 − ω 0 2 0   d 2 
   
 F3 = 0   L  0 −1 1  2
 0 0 1   d 3 
    
Since the first column is multiplied by zero displacement, it can be

deleted. The first row can also be deleted, and then
 0   EA  2 −1  ρAL  2 0   d 2 
 =   − ω2    
 0   L  −1 1  2  0 1   d 3 

The above equations are simplified as
 0   2 −1   2 0 
 =  − λ
 0   −1 1 
 0 1 
Vibr ation 175
where λ is the eigenvalue,
ρL2
λ = ω2
2E
then,
 0   2 − 2λ −1 
 =  
 0   −1 1− λ 
The determinant of the above matrix yields the following equation:
2λ2 − 4λ + 1 = 0
And the solution for the above equation is λ1 = 0.29289 and λ2 = 1.70711.
The natural frequencies are calculated as
2λ1E
ω1 = = 4840.58 Hz
ρL2

2λ 2 E
ω2 = = 11686.23 Hz
ρL2

The exact solution for the present problem is
nπ E
ω=
2L ρ

where n is the mode number. The exact solutions for the first and the
second frequencies are ω1 = 4967.15 Hz and ω2 = 9934.3 Hz, respec-
tively. The exact results are close to the finite element solution with error
in the first and the second frequencies of 2.55% and 17.64%, respec-
tively. The error can be reduced if the number of elements is increased.

D
flexural vibration of a beam element
Figure 4.4 shows a horizontal beam element. The variation of curvature

of the beam is assumed to be a polynomial of the third order, and the con-
stants in the polynomial are determined using the boundary conditions.
There are four degrees of freedom for the element, two vertical displace-
ments and two rotations. The displacement function is expressed as
dy(x) = a1x3 + a2x2 + a3x + a4 (4.18)
Using expression (2.68), the rotation is expressed as
θ(x) = 3a1x2 + 2a2x + a3 (4.19)

E, I, A, ρ
x
1 2
θ1 L θ2
FIGURE 4.4 Horizontal d1y d2y
beam element.
The constant a’s are obtained by using the boundary conditions at the
nodes as follows:
dy(x = 0) = d1y (4.20)
dy(x = L) = d2y (4.21)
∂v
θ(x = 0) = = θ1 (4.22)
∂x
∂v
θ(x = L) = = θ2 (4.23)
∂x
Therefore, the displacement function (4.18) is expressed as
3x 2 2x 2 3x 2 x2
d y ( x ) = d1y + θ1x − d1y − θ1 + d 2y − θ2
L2 L L2 L
(4.24)
2x 2 x3 2x 2 x3
+ 3 d1y + 2 θ1 − 3 d 2y + 2 θ2
L L L L
Rearranging the displacement function (4.24) gives
d(x) = f1(x)d1 + f2(x)θ2 + f3(x)d2 + f4(x)θ2 (4.25)
where f1(x), f2(x), f3(x), and f4(x) are shape functions, and they are
expressed as
2 3
x x
f1 ( x ) = 1 − 3   + 2   (4.26)
 L  L
 x2   x3 
f2 ( x ) = 1 − 3   + 2  2  (4.27)
 L L 
2 3
x x
f3 ( x ) = 3   − 2   (4.28)
 L  L
 x 2   x3 
f4 ( x ) = −   +  2  (4.29)
 L L 
Vibr ation 177
The kinetic and strain energies are, respectively,
L 2
ρA  ∂v 
T=
∫   dx (4.30)
2  ∂x 
0
L 2
EI  ∂2 v 
U=
∫
2  ∂x 2 
0
dx (4.31)
Using the Lagrange equation,
d  ∂T  ∂T
 + = F (4.32)
dt  ∂θ  ∂θ
Substituting kinetic and strain energies into the Lagrange equation
gives
[ m ]{d} + [ k ]{d} = {F ( t )} (4.33)
In the case of free vibration, and with the expression {}

d = −ω 2 {d} ,
the Lagrange equation (4.33) becomes
−ω2[m]{d} + [k]{d} = {0} (4.34)
Expanding Equation 4.34 yields
 156 22L 54 −13L   d1y 

  
ρAL  22L 4L2 13L −3L2   θ1 
−ω 2  
420  54 13L 156 −22L   d 2y 
 −13L −3L2 −22L 4L2   
   θ2 
(4.35)
 12 / L   
2
6 / L −12 / L 2
6 / L  d1y
  0 
  
EI  6 / L 4 −6 / L 2  θ1   0
+  =  
L  −12 / L2 −6 / L −12 / L2 −6 / L   d 2y   0 
 6/L −6 / L 4     0 
 2
 θ2 
4.5 atural frequencies of the flexural

N
vibration of a beam element
Determine the first natural frequency of the beam shown in Figure 4.5a.

The beam is 2 m in length and fixed at both ends. The beam has a square
cross section with height and width of 0.05 m, ρ = 5000 kg/m3, and
E = 200 GPa.
The minimum number of nodes required to determine the natural
frequency is 3, because two of the nodes are fixed. In this example,
E, A, I, ρ
Element 1 Element 2
2m
Node 1 Node 2 Node 3
(a) (b)
FIGURE 4.5 (a) Horizontal beam with fixed ends and (b) finite element mesh.
the beam is divided into two elements only. The finite element mesh
for the horizontal beam is shown in Figure 4.5b. In the next ANSYS
example, the beam is meshed with 50 elements. Element 1 has a length
of L = 1 m, nodes 1 and 2, and its stiffness and mass matrices are,
respectively,
ρAL  156 −22L 

[ m1 ] =  
420  −22L 4L2 

EI  12 / L2 −6 / L 
K (1)  =  
L  −6 / L 4 

Element 2 has a length of L = 1 m, nodes 2 and 3, and its stiffness and

mass matrices are, respectively,
ρAL  156 22L 

[m2 ] =  
420  22L 4L2 

EI  12 / L2 6/L 
K ( 2)  =  
L  6 / L 4 

Applying Equation 4.34:
ρAL  312 0   d 2y  EI  24 / L2 0   d 2y   0 
−ω 2   +   =  
420  0 8L2   θ2  L  0 8   θ2   0 
Simplifying the above equation yields
ρAL  312 0  EI  24 / L2 0 = 0 

−ω 2  +    
420  0 8L2  L  0 8   0 
Vibr ation 179
Solving for the first natural frequency,
1/ 2
5.68  EI 
ω1 =

( L )2  ρA 
The exact solution for the first natural frequency is given by
1/ 2
5.59  EI 
ω1 =

( L )2  ρA 
which is very close to the finite element result.
4.6 atural frequencies of the flexural vibration

N
of beam element using ANSYS
For a horizontal beam shown in Figure 4.5a, determine the first five

natural vibration frequencies. Also, create animation for the third
mode. Mesh the beam with 50 elements. The beam has a square
cross section with height and width of 0.05 m, ρ = 5000 kg/m 3, and
E = 200 GPa.
A select Structural
OK
Add...
A select Beam
B select 2 node 188
OK
Options…
Vibr ation 181
The illustrated theory for the beam in this chapter is based on the
third-order polynomial for displacement function. Therefore, in the
option, the element behavior should be changed to cubic form.
OK
Close
The beam cross section is specified in sections. For material proper-

ties, the modulus of elasticity and density are required.

The following window will appear, the Young modulus for the beam
is specified, and any value for the Poisson ratio is required to avoid an
error message. Also, the density of the beam is required.
A type 200e9 in EX
B type 0.3 in PRXY
OK
A click on Structural > Density
A type 5000 in DENS
OK
Close the Define Material Model Behavior window

The cross section of the beam is a square. In the cross section, the
square section type is the default section. The width and height of the
beam’s cross section should be specified.
Vibr ation 183
A type 0.05 in B
B type 0.05 in H
Apply
Review
OK
ANSYS graphics will show the geometrical

properties of the beam cross section
Modeling the beam structure takes place in the following steps. First, two
keypoints are created. Then, a line that connects the keypoints is created. The
x- and y-coordinates of each keypoint are specified for ANSYS. Then, (0,0)
is the coordinate for Keypoint 1, and (0,2) is the coordinate for Keypoint 2.
In Active CS

Apply

OK
Click on Keypoints 1 and 2. In Create Straight Line window, click on
OK
Vibr ation 185
A click Set in Lines

Pick All
The following window will show up. In Element Sizes on Picked
Lines window, there are two options, either specifying the length of the
elements or the number of element divisions. For the present example,
the beam is divided into 50 elements.
OK
A click on Mesh

Pick All
ANSYS graphics show the nodes

The analysis type is changed from Static to Modal in New Analysis.
The required number of free vibration modes is five, and this is requested
in analysis options.
A select Modal
OK
Main Menu > Solution > Analysis Type > Analysis Options
Vibr ation 187
A type 5 in No. of modes to extract

B type 5 in NIMODE No. of modes to expand
OK
OK
Boundary conditions are specified in solution. Both ends of the beam
are fixed and no forces are applied because it is a free vibration problem.
The displacement is applied on the keypoints.
Displacement > On Keypoints
Click on Keypoints 1 and 2. Then, in Apply U,ROT on KPs window,
click on
OK
A select UX, UY, UZ, ROTX, ROTY, and ROTZ

OK
problem.
OK
Close
In the following steps, the natural frequencies of the beam are listed.
Notice that the frequencies 1 and 2 are the same, because the beam is
vibrating similarly in the y- and the z-directions.
Main Menu > General Postproc > Results Summary
Close
The exact solution for the first natural frequency for this geometry is
given by
1/ 2
5.59  EI 
ω1 =
( L / 2 )2  ρA 
where L is the beam length. The exact solution for the present geometry
is ω1 = 510.377 rad/s or 81.23 Hz, which is very close to the ANSYS
results. Next, the deformation shape of the third mode is displayed.
Vibr ation 189
A select Set 3
Read
Close
OK
ANSYS graphics show the beam before and after deformation

The animation of the third vibration mode is displayed in the follow-
ing steps. An AVI file is created and stored in the working directory.
The number of frames is the number of screen shoots for the animation
file. The 50-frame file is specified for this example and a higher value
will enhance the resolution of the animation, but the size of the file will
be increased. The time delay is the time between each screen shot, and
increasing the time delay will increase the duration of the animation.
Utility Menu > PlotCtrls > Animate > Mode Shape
A type 50 in No. of frames to create

B select Def + undeformed
OK
Animation will be shown on the screen for the third mode.

D
vibration of an oriented beam element
Figure 4.6 shows an axial flexural frame element oriented at an angle

θ in the xy-plane. There are three degrees of freedom at each node,
displacement in x-direction, displacement in y-direction, and rotation.
For a static beam, the stiffness equation is expressed as
{F} = [T]T[K][T]{d} (4.36)
The stiffness equation is extended for vibration by adding the ω 2 [ m ]

matrix to the stiffness matrix as follows:
{F } = [ T ]T ([ K ] − ω 2 [ m ])[ T ]{d} (4.37)

where
 F1x 
 
 F1y 
 M1  (4.38)
{F } =  
 F2x 
 F2y 
 
 M 2 
Vibr ation 191
Fy2, dy2
M2, θ2
de2
No Fx2, dx2
,ρ
, I, A y
L ,E
1 θ
Fx1, dx1 de x
No
M1, θ1
Fy1, dy1
FIGURE 4.6 An axial flexural frame element.
 cosθ sinθ 0 0 0 0 
 −sinθ cosθ 0 0 0 0 
 
0 0 1 0 0 0
[ ] 
T =  (4.39)

0 0 0 cosθ sinθ 0
 0 0 0 −sinθ cosθ 0 
 
 0 0 0 0 0 1 
 0 0 0 0 0 0 
 
 0 12 / L2 6/L 0 −12 / L2 6/L 
EI  0 6/L 4 0 −6 / L 2 
[ ]
K =  (4.40)
L  0 0 0 0 0 0
 
 0 −12 / L2 −6 / L 0 12 / L2 −6 / L 
 0 6/L 2 0 −6 / L 4 
and
 0 0 0 0 0 0 
 0 156 22L 0 54 −13L 
 
ρAL  0 22L 4L2 0 13L −3L2 
[m] =  (4.41)
420  0 0 0 0 0 0
 
 0 54 13L 0 156 −22L 
 0 −13L −3L2 0 −22L 4L2 
The equation of motion for the two-dimensional flexural beam

element is
{F } = ([ K ] − ω 2 [ m ]) {d} (4.42)
4.8 Harmonic vibration of a plate with holes using ANSYS
The rectangular plate with six holes, as shown in Figure 4.7a, is sub-
jected to forcing function. The forcing function is shown in Figure 4.7b.
Create a graph showing the relationship between the displacement at the
point where the force is applied and the load’s frequency. Let t = 0.001 m,
E = 210 GPa, v = 0.25, and ρ = 5000 kg/m3. The frequency of the load is
varied between 1 and 2000 Hz, and the number of subsets is 100.

This example is limited to vibration analysis. Hence, select Structural.
The geometry is meshed with quadratic four-node elements.
A select Structural
OK
F(t)
R 0.025 m
0.1 m
0.075 m 0.1 m 0.075 m
0.65 m
(a)
F, kN
0.1
1.0 t, sec
(b)
FIGURE 4.7 (a) A plate with six holes and (b) forcing function.
Vibr ation 193
Material properties of the structure are defined in the following steps.

The modulus of elasticity and density are required to solve this problem.
In addition, the plate has a thickness, and the thickness is specified in
real constants.
Add...
A select Solid
OK
Options…
A select Plane strs w/thk in Element behavior K3
OK
Close
Add...
OK
Vibr ation 195
A type 0.001 in Thickness THK
OK
Close the Real Constants window
Close
A type 210e9 in EX
B type 0.25 in PRXY
OK
Main Menu > Material Props > Material Models
A type 5000 in DENS
OK

The geometry is modeled by creating a rectangle and circles.
Boolean operation is utilized to subtract the circles from the rectangle.
Alternatively, the circles can be overlapped and then deleted.
Rectangle > By 2 Corners
Vibr ation 197
A type 0 in WP X
B type 0 in WP Y
C type .65 in Width
D type .1 in Height
OK
Solid Circle
A type .075 in WP X
B type .05 in WP Y
C type .025 in Radius
OK
The other circles are created using the copy area in the modeling. The
number of circles including the original circle is six, and the distance
between the circles is 0.1 m.
Main Menu > Preprocessor > Modeling > Copy > Areas
In the ANSYS graphics, select the solid circular area, and then in
Copy Areas window, click on
OK
A type 6 in Number of copies

B type .1 DX X-offset in active CS
OK
Subtract > Areas
Click on the center of the rectangular area to select it. Then, in Subtract
Areas window, click on
OK
Click on all circular areas to select them. Then, in Subtract Areas
window, click on
OK
ANSYS graphics showing the final geometry

Vibr ation 199
The geometry is meshed with quadratic four-node elements. A free mesh

is generated using the smart mesh option. The mesh refinement is 1.
A select Smart Size

Then, in Mesh Tool, click on
Mesh
In Mesh Areas, click on
Pick All
The analysis type is changed from Static to Harmonic. The frequency of

the load is varied between 1 and 2000 Hz, and the number of subsets is 100.
A select Harmonic
OK
Boundary conditions are applied as follows. The left vertical line of
the beam is fixed, while the other lines are free. Force in the negative
y-direction is applied at the keypoint at the upper right corner.
Displacement > On Lines
In the ANSYS graphics, click on the left line where displacement is
applied. Then, in Apply U,ROT on Lines window, click on
OK
A select All DOF

B set VALUE to 0
OK
Vibr ation 201
The Function Editor is used to apply the transient force formula as

a nodal force. This technique is simple and convenient for this problem
because an equation for the force is given. Notice that the force unit
given in Figure 4.7b is in kN.
Define/Edit
A type the equation: −100−100*{TIME}

In the Equation Editors click on File then Save
Save the file as FY_Time, and the file name is optional. After saving
the function, it has to be loaded to the ANSYS solution using the read
file.
Save
Close the Function Editor
Read File
Select the file FY_Time.func then

Open
A type FY_Time in Table parameter name, and this name is optional

OK
Force/Moment > On Keypoint
Click on keypoint where the force is applied, and then in Apply F/M
on KPs window, click on
OK

OK
Vibr ation 203
A select FY_TIME
OK
Main Menu > Solution > Load Step Opts > Time/Frequency >
Freq and Substeps
A type 0 and 2000 in Harmonic freq range

B type 100 in Number of substeps
C select Stepped
OK
problem.
OK
Close
A graph that shows the displacement at a specific location in the plate
as a function of frequency can be done with the Time history. In the fol-
lowing steps, the displacement at the node where the force is applied is
plotted as a function of frequency.
A click on the green + button

Vibr ation 205

displacement
OK
In the ANSYS graphics, click on the node at the upper right corner of
the plate and then in Node for Data window, click on
OK
A click on the graph button

ANSYS graphics show the y-displacement as a function of frequency

The jumps in the displacement shown in the above figure occurred
at the natural frequency of the first and the second modes of the plate.
Next, an animation is created for the frequency of 1000 Hz.
Main Menu > General Postproc > Read Results > By Time/Freq
A type 1000 in TIME Value of time or freq
OK
Utility menu > Plot > Areas


OK
Vibr ation 207
ANSYS graphics show the plate before and after deformation

OK
Animation will be shown in ANSYS graphics.
4.9 Three-dimensional vibration of shaft

with disks using ANSYS
For the geometry shown in Figure 4.8, the ends of the shaft are fixed,
while the two disks and shaft are vibrating freely. The total shaft length
is 0.655 m. Determine the first five natural vibration frequencies.
R 0.3
R 0.03
Fixed Fixed
0.125 m 0.25 m
0.015 m
FIGURE 4.8 Shaft and two disks.
Also, create animation for the third mode. Let m, E = 180 GPa, v = 0.3,
and ρ = 7000 kg/m3. Use free mesh with Tet 4 node element and set
smart size to 5.

only. The three-dimensional solid element must be used and the type of
the element is tetrahedral with four nodes.
Vibr ation 209
A select Structural
OK
Add...
A select Solid
B select Tet 4 node 285
OK
Close
For the material properties, the modulus of elasticity and density are

The following window will show up:
Vibr ation 211
A type 180e9 in EX
B type 0.3 in PRXY
OK
Main Menu > Material Props > Material Models
A type 7000 in DENS
OK

The used technique for modeling the geometry is quite simple. The
following steps are performed:
1. The shaft is modeled as a long, solid cylinder.

2. The first disk is modeled as a thin, solid cylinder and then it is
moved to its position in the shaft.
3. The second disk is created by copying the first disk.
4. The shaft and two disks are added to create one volume.
Volumes > Cylinder > Solid Cylinder
A type 0 in WP X
B type 0 in WP Y
D type 0.655 in Depth
OK
ANSYS Utility Menu > PlotCtrls > Pan-Zoom-Rotate …
Vibr ation 213
Click Iso button to show the isometric view of the shaft and zoom
in and out.
ANSYS graphics show the isometric view of the shaft
Volumes > Cylinder > Solid Cylinder
A type 0 in WP X
B type 0 in WP Y
D type 0.015 in Depth
OK
The disk is not in its correct position, and it has to be moved along
the z-axis to its position at a distance of 0.25 m from its current position.
ANSYS Main Menu > Preprocessor > Modeling > Move/Modify >
Volume
Click on the disk and then in Move Volumes window, click on
OK
A type 0 in DX X-offset in active CS

B type 0 in DY Y-offset in active CS
C type 0.25 in DZ Z-offset in active CS
OK
The second disk is created by copying the first disk. The offset dis-
tance between the two disks is the spacing plus the thickness of the disk,
which is 0.14 m.
ANSYS Main Menu > Preprocessor > Modeling > Copy > Volumes
Click on the disk, and then in Copy Volumes window, click on
OK
A type 0 in DX X-offset in active CS

B type 0 in DY Y-offset in active CS
C type 0.14 in DZ Z-offset in active CS
OK
Vibr ation 215
Next, the three volumes are added to create one volume.
ANSYS Main Menu > Preprocessor > Modeling > Operate >
Booleans > Add > Volumes
In Add Volumes windows, click on
Pick All
A select Smart Size

Mesh
In Meshed Areas window, click on
Pick All

The analysis type is changed from Static to Modal. The required
number of free modes to be calculated is five.
A select Modal
OK
Main Menu > Solution > Analysis Type > Analysis Options
A type 5 in No. of modes to extract

B type 5 in NIMODE No. of modes to expand
OK
Vibr ation 217
OK
The left and right surfaces of the shaft are fixed, while the other
surfaces are free. No forces are applied.
Displacement > On Areas
In the ANSYS graphics, click on the left and right surfaces of the
shaft, which are fixed. Then in Apply U,ROT on Areas window, click on
OK
A select All DOF

OK
Close
Main Menu > General Postproc > Results Summary
ANSYS lists the natural frequencies of the shaft with two disks. Next,
the deformed shape of the shaft for the mode number 4 is displaced,
followed by an animation.
A select Set 4
Read
Close
OK
Vibr ation 219
ANSYS graphics show the vibrating shaft with

two disks before and after deformation

OK
Animation will be shown for the third mode.
PROBLEM 4.1
Determine the first natural vibration frequency of a horizontal canti-
lever bar using finite element method, as shown in Figure 4.9. The
total bar length is 3 m. Model the bar with three elements only and let
E = 180 GPa, A = 4 × 10 –3 m2, and ρ = 7000 kg/m3.
E, A, ρ
3m
FIGURE 4.9 A simple cantilever bar.
PROBLEM 4.2
A horizontal cantilever bar is composed of two different bars of dif-
ferent cross sections and lengths, as shown in Figure 4.10. Determine
the first natural vibration frequency of the bar using the finite element
method. Model the bar with three elements only, as suggested in the
figure, and let E = 210 GPa, ρ = 4500 kg/m 3, A1 = 4 × 10 –3 m 2, and
A 2 = 2 × 10 –3 m 2.
PROBLEM 4.3
A horizontal beam structure is composed of three different beams
of different material properties, as shown in Figure 4.11. Determine
the first natural vibration frequency of the structure using the finite
element method. Model the structure with three elements only and
let E1 = 205 GPa, ρ1 = 4250 kg/m3, E2 = 270 GPa, ρ2 = 3000 kg/m3,
E3 = 180 GPa, ρ3 = 4000 kg/m3, and the beams have a square cross-
sectional area with height and width of 0.1 m.
PROBLEM 4.4
A horizontal cantilever bar structure is composed of two different bars
of different cross-sectional areas and lengths, as shown in Figure 4.10.
Determine the first five natural vibration frequencies of bar structure and
let E = 210 GPa, ρ = 4500 kg/m3; A1 = 4 × 10 –3 m2, and A2 = 2 × 10 –3 m2.
PROBLEM 4.5
A horizontal beam structure is composed of three different beams
of different material properties, as shown in Figure 4.11. Determine
the first five natural vibration frequencies of beam structure using
ANSYS. Also, create an animation for the second mode. Mesh each
E, A1, ρ E, A2, ρ
1m 2m
Element 1 Element 2 Element 3
Node 1 Node 2 Node 3 Node 4
FIGURE 4.10 A cantilever bar made of two different bars.
E1, ρ1 E2, ρ2 E3, ρ3
0.5 m 0.5 m 0.5 m
FIGURE 4.11 A horizontal beam structure made of three different beams.

Vibr ation 221
beam with 50 elements and let E1 = 205 GPa, ρ1 = 4250 kg/m 3,

E2 = 270 GPa, ρ2 = 3000 kg/m3, E3 = 180 GPa, ρ3 = 4000 kg/m3, and
the beam has a square cross-sectional area with height and width
of 0.01 m.
PROBLEM 4.6
For the plate with circular holes, as shown in Figure 4.12, the left verti-
cal line is fixed. Determine the first five natural vibration frequencies of
structure using ANSYS. Also, create an animation for the third mode.
Let t = 0.002 m, E = 200 GPa, and ρ = 7500 kg/m3.
PROBLEM 4.7
For the rectangular cross-sectional plate with square holes, as shown
in Figure 4.13, both ends are fixed. Determine the first five natural
vibration frequencies of the structure using ANSYS. Also, create an
animation for the third mode. Let t = 0.001 m, E = 270 GPa, and
ρ = 6000 kg/m 3.
PROBLEM 4.8
The geometry shown in Figure 4.14a is subjected to the transient force.
The relationship between force and time is shown in Figure 4.14b.
Create a graph showing the relationship between the displacement
R 0.025 m
0.1 m
0.05 m 0.1 m 0.05 m
FIGURE 4.12 A plate with circular holes.
0.04 m
0.01 m 0.02 m 0.02 m
FIGURE 4.13 A rectangular plate with square holes.

F(t)
0.08 m
0.02 m 0.44 m 0.02 m

(a)
F, N
250
t, sec.
1.0
(b)
FIGURE 4.14 (a) A rectangular plate with a rectangular hole and (b) forcing function.
at the point where the force is applied and the load’s frequency.
Let t = 0.002 m, E = 230 GPa, and ρ = 4000 kg/m3. The frequency
of the load is varied between 1 and 5000 Hz, and the number of
subset is 50.
PROBLEM 4.9
For the geometry shown in Figure 4.15, the ends of the shaft are fixed,
while the disk and the shaft are vibrating freely. Determine the first five
natural vibration frequencies for the shaft. Also, create an animation for
the third mode. Let m, E = 180 GPa and ρ = 7000 kg/m3. Use free mesh
with Tet 4 node element, and set smart size to 5.
PROBLEM 4.10
For the airplane wing shown in Figure 4.16, determine the first five natu-
ral vibration frequencies. Let E = 180 GPa, v = 0.25, and ρ = 3500 kg/m3.
The wingspan is 5 m, and the left surface of the wing is fixed. Solve the
problem as a three-dimensional problem. Use Tet 10-node elements with
smart size set to 1.
Vibr ation 223
R 0.2
R 0.05 R 0.025
Fixed Fixed
0.025 m
0.2 m 0.15 m
FIGURE 4.15 A shaft and disk.
R 0.4 m
2.5 m
FIGURE 4.16 Airplane wing.

Cha p t e r FIV E
Heat transfer
5.1 Introduction to heat conduction
Whenever a temperature gradient exists in a solid, heat will flow from

the high-temperature region to the low-temperature region. The basic
governing heat conduction equation is obtained by considering a plate
with a surface area A and thickness Δx, as shown in Figure 5.1. One side
is maintained at temperature T1 and the other side is at temperature T2.
Experimental observation indicates that the rate of heat flow is directly
proportional to the area and temperature difference, but inversely pro-
portional to the plate thickness. The proportionality sign is replaced by
an equal sign by introducing the constant k as follows:
T1 − T2
Q = kA (5.1)
∆x
The constant k is the thermal conductivity of the plate. This property
depends on the type of the plate’s material. Equation 5.1 is also called
Fourier’s law. Fourier’s law can be expressed in differential form in the
direction of the normal coordinate:
dT
Q = − kA (5.2)
dn
Also, Fourier’s law can be expressed for multidimensional heat flux
flow as follows:
 ∂T ∂T ∂T 
q = −k  î + ĵ + k̂ (5.3)
 ∂x ∂y ∂z 
An energy balance can be applied to a differential volume dx dy dz,

for conduction analysis in a Cartesian coordinate, as shown in Figure 5.2.
The objective of this energy balance is to obtain the temperature distribu-
tion within the solid. The temperature distribution is used to determine
the heat flow at a certain surface, or to study the thermal stress. The
heat flux perpendicular to the surface of the control volume is indicated
225
Δx
T1
T2
FIGURE 5.1 Heat transfer

through a plate.
qy
qz+dz
qx qx+dx
qz
FIGURE 5.2 Differential control volume qy+dy

for the energy balance.
by the terms, qx, qy, and qz. The heat flux at the opposite surface can
then be expressed using the Taylor series expansion of the first order as
follows:
∂q x
q x + dx = q x + dx (5.4)
∂x
∂q y
q y + dy = q y + dy (5.5)
∂y
∂q z
q z + dz = q z + dz (5.6)
∂z
Energy can be generated in the medium, and the expression of the

heat generation is
E gen = q dx dy dz (5.7)
where q is the generated heat per unit volume, W/m3. If the heating pro-
cess is unsteady, the total energy of the control volume can be increased
or decreased. The energy storage term is expressed as
∂T
E st = ρCP dx dy dz (5.8)
∂t
He at tr ansfer 227
The sum of the energy generation in the control volume and net heat
flow should be equal to the energy stored in the control volume. The
energy conservation can be expressed in the following mathematical
form:
( )
E gen + E in − E out = E st (5.9)
Substituting expressions (5.4–5.8) into (5.9), the energy conservation

equation becomes
 ∂q ∂q y ∂q  ∂T

q dx dy dz +  x dy dz + dx dz + z dx dy  = ρCP dx dy dz
 ∂x ∂y ∂z  ∂t
(5.10)
x Q ′′,
The Q ′′, y and Q z′′ are obtained from Fourier’s law (5.2) as follows:
dT
Q ′′x = − k dx (5.11)
dx
dT
Q ′′y = − k dy (5.12)
dy
dT
Q z′′ = − k dz (5.13)
dx
Finally, the conduction energy equation per unit volume, in a

Cartesian coordinate, is expressed as
∂  ∂T  ∂  ∂T  ∂  ∂T  ∂T
k + k + k  + q = ρCp (5.14)
∂x  ∂x  ∂ y  ∂ y  ∂z  ∂z  ∂t
When the system reaches the steady-state condition, the term ∂T/∂t is
equal to zero. If the thermal conductivity is independent of the direction,
the conduction energy equation can be written in a simpler form as
∂2 T ∂2 T ∂2 T q ρCP ∂T
+ + + = (5.15)
∂x 2 ∂ y 2 ∂z 2 k k ∂t
The energy equation is a partial differential equation with second
order in space and first order in time. The boundary conditions along its
surface as well as the initial condition must be specified. For the initial
condition, the temperature distribution in the system must be provided.
In heat transfer problems, there are three types of boundary conditions:
temperature, heat flux, and convection.
The constant temperature, also called the Dirichlet condition, corre-
sponds to a situation for which the surface is maintained at a fixed tem-
perature. The mathematical expression for this boundary condition is
T (x, t) = Ts (5.16)
The second type of boundary condition, which is called the Neumann

condition, corresponds to a constant heat flux applied to a surface. The
heat flux is related to the temperature gradient at the surface by Fourier’s
law as follows:
∂T
−k = q (5.17)
∂x
A special case of this boundary condition is an insulated boundary
condition, and the heat flux should be zero:
∂T
−k = 0 (5.18)
∂x
The third boundary condition corresponds to convection at a surface.
The conduction–convection heat balance at the wall surface must be sat-
isfied. The heat transfer coefficient (h) should be known, as well as the
fluid bulk temperature (T∞):
∂T
−k = h [ T∞ − T ( x, t )] (5.19)
∂x
5.2 Finite element formulation for conductive heat transfer
The finite element method is an efficient way to solve conduction prob-

lems. The heat transfer solution can be used to estimate the heat flow at
the boundaries or to determine the temperature distribution for thermal-
stress analysis. In this chapter, the variation formulation is used to obtain
the conductive equation. It is accomplished by minimizing the following
potential function:
1   ∂T  2  ∂T  
2
I=
2 ∫  ∂x 
V
 ∂ y  
V S
∫ ∫
 k x   + k y    dV − Q T dV − q T dS
(5.20)
1
∫
h ( T − T∞ ) dS
2
+
2
S
where the first term is heat conduction in the solid, the second term is
volumetric heat generation, the third term is surface heat flux, and the
fourth term is surface heat convection. For any surface, either surface
heat flux or convection is applied. The temperature function T within
each element can be expressed in terms of shape functions as
 T1 
 
 T2 
{T} =  N1 N 2 N 3 ... N i   T3  (5.21)
 
 
 Ti 
He at tr ansfer 229
and the heat flux in the x- and y-directions is given by
 ∂T 
 q x   kx   
0 ∂x 
 q = − 0   (5.22)
 y   ky   ∂T 
 ∂y 
 
The temperature gradient vector can be expressed in terms of shape
functions as
 T1 
 ∂T   ∂ N1 ∂N 2 ∂N3 ∂N i  
   ...  T2 
∂x   ∂x ∂x ∂x ∂x
{g} =  ∂T
= 


T3  (5.23)
   ∂ N1 ∂N 2 ∂N3
...
∂N i
 
 ∂y   ∂ y ∂y ∂y ∂y   
   Ti 
or symbolically,
{g} = [B] {T} (5.24)
where [B] is a derivative of the shape functions. The thermal conductiv-
ity matrix is
 kx 0 
[D] =  0  (5.25)
ky 
 
Hence, the heat flux vector can be written in terms of the temperature
gradient vector and thermal conductivity matrix as follows:
 q x 
 q  = [ D ]{g} (5.26)
 y 
Potential function (5.20) can be expressed as
1  T
I=
2  ∫V
∫
{g} [D]{g} dV − {T}T [ N ]T Q dV − {T}T [ N ]T qdS
V
∫
S
(5.27)
∫ ( )
1
h  {T}T [ N ] − T∞  dS
T 2
+
2  
S
Using the expression of the temperature gradient (5.24) and expand-

ing the surface convection term, the potential function (5.27) can be
written as
1
{ T }T ∫ [ B] [ D][ B]{T} dV − {T} ∫ [ N ] Q dV − {T} ∫ [ N ] q dS
T T T T T
I=
2
V V S
+
1
2∫ 
T T
(
h {T}T [ N ] [ N ]{T} − {T}T [ N ] + [ N ]{T} T∞ − T∞2  dS
 )
S
(5.28)
Minimizing the potential equation (5.28) with respect to temperature

vector, it gives
∂I
∫ [ B] [ D][ B]{T} dV − ∫ [ N ] Q dV − ∫ [ N ] q dS
T T T
=0=
∂{T}
V V S
(5.29)
∫
+ h [ N ] [ N ]{T} − [ N ] T∞  dS
T T
Rearranging Equation 5.29 as
∫ [ B] [ D][ B]{T} dV + ∫ h [ N ] [ N ]{T} dS

T T
V S
(5.30)
∫ [ N ] Q dV + ∫ [ N ] q dS + ∫ h [ N ] T dS
T T T
= ∞
V S S
Equation 5.30 can be expressed in terms of conductive matrix, nodal

temperature, and thermal forces as
[K] {T} = {FQ} + {Fq} + {Fc} (5.31)
where

∫ V
∫
[ K ] = [ B]T [ D ][ B] dV + h [ N ]T [ N ] dS (5.32)
S
The first term for the conductive matrix (5.32) is for axial conduction
through the element, while the second term is convection at the external
surface of the element, and the thermal forces are
{FQ } = ∫ [ N ]T Q dV (5.33)
V
{Fq } = ∫ [ N ]T q dS (5.34)
S

∫
{Fc } = h [ N ]T T∞ dS (5.35)
S
The term {FQ} is volumetric heat generation in the element, {Fq}

is applied heat flux on the external surface of the element, and {Fc} is
applied convection on the external surface of the element. The inte-
grations are over the surface where convection or heat flux is applied.
A symbolic expression for element conduction equation is
[K] {T} = {F} (5.36)

He at tr ansfer 231
where
{F} = {FQ} + {Fq} + {Fc} (5.37)
The global conductive matrix is obtained by assembling the conduc-

tive matrix for each element as follows:
N
[K ] = ∑ K
e =1
(e)
 (5.38)
The thermal force vectors are assembled to form a global one as

follows:
N
{F } = ∑ {F } (5.39)
e =1
(e)
Also, the temperature vectors are assembled to form a global one as

follows:
N
{ T} = ∑ {T } (5.40)
e =1
(e)
5.3 F inite element method for one-dimensional

heat conduction
Consider one-dimensional heat conduction, as shown in Figure 5.3,

where the length of the element is L. The temperature function is similar
to the displacement function, which is
T(x) = N1T1 + N2T2 (5.41)
where T1 and T2 are nodal temperatures, and N1 and N2 are shape func-
tions, which are defined as
x
N1 = 1 − (5.42)
L
x
N2 = (5.43)
L
The shape functions can be presented in matrix form as
 x x 
[ N ] =  N1 N 2  =  1 −  (5.44)
 L L 
Node 1 A, k Node 2
x FIGURE 5.3 One-dimensional heat

L
conduction.
The thermal conductivity of the material is assumed to be isotropic.

The [B] and [D] matrices are calculated as
 ∂ N1 ∂N 2   1 1 
[ B] =  =  (5.45)
 ∂x ∂x   − L L 
 kx 0   k 0 
[D] =  0 =
k y   0
(5.46)
 k 

In Equation 5.32, and since the element has a uniform cross-sectional

area, the integral over differential volume dV is replaced by A dx, where
A is the cross-sectional area of the element, and the integral over differ-
ential surface dS is replaced by P dx, where P is the circumference of the
element. The element conductive matrix (5.32) is expressed as
T
 x x   k 0  x x 

∫
[K ] =  1 −
L
 
L   0 k  
1−
L
 A dx
L 
(5.47)
T
 x x   x x 
∫
+ h 1−
 L
  1−
L   L
 dA
L 
After performing matrix operations and integration, the element con-

ductive matrix (5.47) is expressed as
Ak  1 −1  hPL  2 1   1 0 
[K ] = + + hA  (5.48)
L  −1 1  6  1 
2   0 0 
The first term of the conductive matrix is axial conduction, the sec-
ond term is perimeter convection, and the third term is convection at the
surface of node 1. If the convection is applied at node 2, the conductive
matrix (5.48) becomes
Ak  1 −1  hPL  2 1   0 0 
[K ] = + + hA  (5.49)
L  −1 1  6  1 
2   0 1 
The thermal nodal forces can be calculated using expressions

(5.33–5.35) as follows:
T
 x x  QAL  1 
{FQ } = ∫  1−
L
 QA dx =
L 

2  1 
 (5.50)

T
 x x  qPL  1   1 
{Fq } = ∫  1−  qP dx =   + qA   (5.51)
 L L  2  1   0 
He at tr ansfer 233
T
 x x  hT∞ PL  1   1 

∫
{Fc } = h  1 −
L
 T∞ P dx =
L 

2  1 
 + hT∞ A  
 0 
(5.52)
In Equations 5.51 and 5.52, the first term after integration is for heat
flux or convection applied at the perimeter and the second term is for
heat flux or convection applied at node 1. If the convection is applied at
node 2, Equations 5.51 and 5.52 become, respectively,
T
 x x  qPL  1   0 
{Fq } = ∫  1−  qP dx = 
2  1 
 + qA   (5.53)
 L L   1 
T
 x x  hT∞ PL  1   0 
{Fc } = ∫ h  1−
 L
 T∞ P dx =
L 

2  1 
 + h T∞ A  
 1 
(5.54)
5.4 Heat transfer through a composite wall
Consider a composite wall shown in Figure 5.4. The wall is composed

of three layers with different thermal conductivities: k A = 0.1 W/m · °C,
k B = 0.2 W/m · °C, and kC = 0.25 W/m · °C. A convection boundary con-
dition is applied at the left surface, h = 15 W/m2 · °C and Ti = 25°C,
while the right surface is at a fixed temperature, To = 50°C. Calculate the
temperature at the interfaces and heat flow through the wall. Given that
LA = 5 cm, LB = 15 cm, and LC = 2.5 cm.
The heat conduction in the wall is a one-dimensional heat transfer
problem and can be solved using the finite element method using three
elements only, as shown in Figure 5.4. Since the temperature varia-
tion is linear in each layer, having more than one element in each layer
would not affect the results. The convection thermal load is applied at
node 1 and temperature load at node 4. The results are per unit area A.
To
Ti
kA kB kC
h
LA LB LC
Element 1 Element 2 Element 3

Convection
Node 1 Node 2 Node 3 Node 4
FIGURE 5.4 Heat conduction in a composite wall and finite element mesh.

For element 1, convection load is applied at node 1, and no nodal thermal

load is applied at node 2, there is no convection or heat flux applied at
the surface of the element, and no heat is generated in the element. The
conductive matrix and nodal thermal load vector, respectively, are
k  −1   1 0 
K (1)  = A  1 + 0 + h 
L A  −1 1 
 0 0 
0.1  1 −1   1 0   17 −2 
= + 15  =
0.05  −1 1   0 0   −2 2 

1   375 
{F( ) } = {F( ) } = hT 
1
c
1
∞ = 
0   0 


For elements 2 and 3, no nodal thermal loads are applied on nodes,

there is no convection or heat flux applied at the element, and no heat is
generated in the element. The conductive matrix and nodal thermal load
vector are for elements 2 and 3, respectively,
k 
K ( 2)  = B  1 −1  0.2  1 −1   1.33 −1.33 
= =
L B  −1 1  0.15  −1 1   −1.33 1.33 
k 
K (3)  = C  1 −1  0.25  1 −1   10 −10 
= =
L C  −1 1  0.025  −1 1   −10 10 
{F( ) } = 
2 0 

0 
{F( ) } = 
3 0 

0 
Assembling the conductive matrix and nodal thermal forces and

temperatures yields, it gives
 17 −2 0 0  T1   375 
 −2     
3.33 −1.33 0 T2   0 
  =  0 
 0 −1.33 11.33 −10  T3   
 0 0 −10 10   T4   Q 4 

where Q4 is heat flow at node 4, which is required to maintain the tem-

perature at node 4 at 50°C. Since the temperature at node 4 is known,
T4 = 50°C, the last equation is modified, so that T4 has a value of 50°C.
He at tr ansfer 235
Consequently, the third equation number 3 is also modified. The

conductive matrix and nodal thermal forces and temperatures are
expressed as
 17 −2 0 0  T1   375 
 −2    
3.33 −1.33 0 T2   0 
  =  
 0 −1.33 11.33 0  T3   500 
 0 0 0 1   T4   50 

Then, there are three equations and three unknowns, the solution for
the equations is
T1 = 26.175°C
T2 = 34.981°C
T3 = 48.237°C
5.5 F inite element method for two-dimensional

heat conduction
The simplest type of element used for heat conduction is a three-node

triangular element, as shown in Figure 5.5. The nodes are named i, j, and
m, and the temperature is linearly varying across the element.
The temperature function is the same as the displacement function
used in structural analysis:
 Ti 
{T} =  N i N j N m   Tj  (5.55)

 Tm 
 
and the shape functions are given as
1
Ni =
2A
(α i + βi x + γ i y ) (5.56)
Node m
(xm, ym)
(xi, yi)
Node i
y
Node j
(xj, yj)
FIGURE 5.5 Three-node trian-
x
gular element.
1
Nj =
2A
(α j + β jx + γ jy ) (5.57)
1
Nm =
2A
(α m + βm x + γ m y ) (5.58)
where α, β, and γ are defined as
α i = x j y m − y jx m α j = xm yi − ym xi α m = xi y j − yi x j
βi = y j − ym β j = ym − yi βm = yi − y j (5.59)
γ i = xm − x j γ j = xi − xm γ m = x j − xi
The [B] matrix is calculated using Equation 5.23 as
 ∂N i ∂N j ∂N m 
  
∂x ∂x ∂x  = 1  βi βj βm 
[ B] =   2A  γ i
 (5.60)
∂N i ∂N j ∂N m γj γm 
   
 ∂ y ∂y ∂y 
where A is the area of the element. The thermal conductivity [D]

matrix is
 kx 0 
[D] =  0  (5.61)
ky 
 
The thermal conductive matrix (5.32)
∫ ∫
[ K ] = [ B]T [ D ][ B] dV + h [ N ]T [ N ] dS
V S
Assuming the element has a constant thickness t, the thermal conduc-

tive matrix can be written for a triangular element as
 2 1 0 
hLt   (5.62)
[ K ] = tA [ B] [ D ][ B] +
T
1 2 0
6  
 0 0 0 
where A is the area of the element and L is the element side length where
the convection is applied. The second term in Equation 5.62 represents
convection on the side of the element, and the volumetric heat generation
can be considered as a nodal thermal force as follows:
 1 
QAt  
{FQ } = ∫ [ N ]T Q dV =
3 
 1  (5.63)

V
 1 
He at tr ansfer 237
where the heat generation is equally distributed on the nodes. The heat
flux on a surface is considered as nodal thermal load as
{Fq } = ∫ [ N ]T q dS = ∫  N i
T
N j N m  q dS (5.64)

S S
If the heat flux is applied on a specific side of the element, the heat
flux vector is
 
q L ij  1 
{Fq } = 2 
 1  if heat flux applied between nodes i and j (5.65)

 0 
 0 
{Fq } = qL2jm  1

 if heat flux applied between nodes j and m (5.66)
 1 
 
 1 
qL mi  
{Fq } = 2 
 0  if heat flux applied between nodes m and i (5.67)

 1 
where Lij is the side length between nodes i and j, Ljm is the side length
between nodes j and m, and Lmi is the side length between nodes m
and i. The convection on a surface is considered as nodal thermal load
as follows:
∫ ∫
T
{Fc } = [ N ]T h T∞ dS =  N i N j N m  h T∞ dS (5.68)
S S
If the convection is applied on a specific side of the element, the con-

vection vector is
 1 
hT∞ L ij  
{Fc } =  1  if convection applied between nodes i and j
2  
(5.69)  0 
 0 
hT∞ L jm  
{Fc } =  1  if convection applied between nodes j and m
2  
(5.70)  1 
 1 
hT∞ L mi  
{Fc } =  0  if convection applied between nodes m and i
2  
(5.71)  1 
5.6 Heat conduction in a solid plane
The two-dimensional body, shown in Figure 5.6a, is subjected to

convection at the vertical left side, with T∞ = 50°C and h = 20 W/m2 · °C.
At the right side, a fixed temperature boundary is imposed, with
To = 100°C, while the two horizontal sides are well insulated. The
sides’ length of the body is 2 m, and it has a thermal conductivity of
25 W/m · °C. Determine the temperature at the center of the body.
There are unlimited options for elements distributions, and some of
these options are shown in Figure 5.6b. Increasing the number of elements
will definitely enhance the accuracy of the results, but up to a certain
number of elements. After which, the results become independent of the
number of the elements. The first mesh contains two elements, which is the
minimum number of elements required to solve this problem. However,
this mesh cannot predict the temperature at the center of the body because
there is no node at the center of the body. The second mesh contains 4 ele-
ments, the third mesh contains 8 elements, and the fourth mesh contains
16 elements.
For an illustration purpose, the second mesh is selected because it has
the least number of elements. Figure 5.6c shows the nodes and elements
distribution for the geometry, and it consists of four elements and five nodes.
First, the [B], [D], and [K] matrices for element 1 are formulated. The name
of node 1 is (i) and its coordinate is (0,0), the name of node 2 is (j) and its
coordinate is (2,0), and the name of node 5 is (m) and its coordinate is (1,1).
The values of β’s and γ’s are required for the [B] matrix and are calculated
using Equation 5.59, and A is the area of the element, so we have
βi = yj − ym = 0 − 1 = −1
βj = ym − yi = 1 − 0 = 1
βm = yi − yj = 0 − 0 = 0
γi = xm − xj = 1 − 2 = −1
γj = xi − xm = 0 − 1 = −1
γm = xj − xi = 2 − 0 = 2
A = 0.5(2)1 = 1 m2
Then, the [B] matrix is formulated using Equation 5.60:
1 0 
[ B] =  −1 1
2  −1 −1 2 

and the [D] matrix is formulated using Equation 5.61:
 25 0 
[D] = 
 0 25 
He at tr ansfer 239
Insulated
T∞ = 50˚C
To = 100˚C
h = 20 W/m2 ·˚C
2m
(a)
(b)
Node 4 Node 3
Elem. 3
Elem. 2 Node 5
Elem. 4
y
Elem. 1
Node 1 x Node 2
(c)
FIGURE 5.6 (a) The two-dimensional body subjected to convection and fixed temperature bound-
ary conditions, (b) suggested finite element meshes, and (c) the elements distribution.
No convection is applied, and therefore, the convection term in

Equation 5.62 is equal to zero. Finally, the [K] matrix for element 1 is
calculated using Equation 5.62:
1 2 5
 12.5 0 −12.5  1
K  = 
(1) 
0 12.5 −12.5  2
 
 −12.5 −12.5 25  5

For element 2, the [B], [D], and [K] matrices are formulated. The name
of node 1 is (i) and its coordinate is (0,0), the name of node 5 is (j) and its
coordinate is (1,1), and the name of node 4 is (m) and its coordinate is (0,2).
The values of β’s and γ’s are required for the [B] matrix, and calculated
using Equation 5.59, and A is the area of the element. We then have
βi = yj − ym = 1 − 2 = −1
βj = ym − yi = 2 − 0 = 2
βm = yi − yj = 0 − 1 = −1
γi = xm − xj = 0 − 1 = −1
γj = xi − xm = 0 − 0 = 0
γm = xj − xi = 1 − 0 = 1
A = 0.5(2)1 = 1 m2
1 
[ B] =  −1 2 −1 
2  −1 0 1 

and the [D] matrix is the same as for element 1. No convection is

applied, and therefore, the convection term in Equation 5.62 is equal
to zero. Finally, the [K] matrix for element 2 is calculated using
Equation 5.62:
1 5 4
 12.5 −12.5 0  1
K  =  −12.5
( 2) 
25 −12.5  5
 
 0 −12.5 12.5  4
For element 3, the [B], [D], and [K] matrices are formulated. The
name of node 4 is (i) and its coordinate is (0,2), the name of node 5
is (j) and its coordinate is (1,1), and the name of node 3 is (m) and
its coordinate is (2,0). The values of β’s and γ’s are required for
He at tr ansfer 241
the [B] matrix, and are calculated using Equation 5.59, and A is the

area of the element. We have
βi = yj − ym = 1 − 2 = −1
βj = ym − yi = 2 − 2 = 0
βm = yi − yj = 2 − 1 = 1
γi = xm − xj = 2 − 1 = 1
γj = xi − xm = 0 − 2 = −2
γm = xj − xi = 1 − 0 = 1
A = 0.5(2)1 = 1 m2
1 
[ B] =  −1 0 1 
2  1 −2 1 

and the [D] matrix is the same as for element 1. No convection is

applied, and therefore, the convection term in Equation 5.62 is equal
to zero. Finally, the [K] matrix for element 3 is calculated using
Equation 5.62:
4 5 3
 12.5 −12.5 0  4
K (3)  =  −12.5 25 −12.5  5
 
 0 −12.5 12.5  3

For element 4, the [B], [D], and [K] matrices are formulated. The
name of node 2 is (i) and its coordinate is (2,0), the name of node 3 is (j)
and its coordinate is (0,2), and the name of node 5 is (m) and its coordi-
nate is (2,0). The values of β’s and γ’s are required for the [B] matrix, and
calculated using Equation 5.59, and A is area of the element. We have
βi = yj − ym = 2 − 1 = 1
βj = ym − yi = 1 − 0 = 1
βm = yi − yj = 0 − 2 = −2
γi = xm − xj = 1 − 2 = −1
γj = xi − xm = 2 − 1 = 1
γm = xj − xi = 2 − 0 = 2
A = 0.5(2)1 = 1 m2
1 
[ B] =  1 1 −2 
2 −1 1 2 

and the [D] matrix is the same as for element 1. Convection is applied
on the side i–j of the element, and therefore, the convection term in
Equation 5.62 is not equal to zero. Finally, the [K] matrix for element 4
is calculated using Equation 5.62:
2 3 5 2 3 5
 12.5 −12.5 0  2  13.3 66.7 0  2
 −12.5 25 −12.5  3 +  6.67 13.3 0  3
   
 0 −12.5 12.5  5  0 0 0  5

Finally, the [K] matrix for the entire body is obtained by adding [K]
matrices for elements 1, 2, 3, and 4 using the expression (5.38), and the
result is
1 2 3 4 5
 25 0 0 0 −25  1
5
 0 38.33 6.67 0 −25  2
[K] = ∑ [K ] =  0
(i)
6.67 38.33 0 −25

 3
i =1  0 0 0 25 −25  4
 −25 −25 −25 −25 100 
  5

Convection is applied on the side i–j of the element 4, the convection

force vector is
 1   1000 
hT∞ L ij    
{Fc } =  1  =  1000 
2    
 0   0 
Using total conductive matrix, [K], and nodal force in expression

(5.36) yields
[K]{T} = {F}
 25 0 0 0 −25  T1   Q1 
 0 38.33 6.67    
0 −25 T2   1000
   
 0 6.67 38.33 0 −25  T3 =  1000 
 0 0 0 25 −25  T4   Q4 
 −25 −25 −25 −25 100    
   T5   0 
He at tr ansfer 243
where Q1 and Q4 are heat flow at nodes 1 and 4, respectively, which are
required to maintain the temperature at the nodes. The temperature of
nodes 1 and 4 are known, and therefore, the above equations should be
modified. In nodal force vector, Q1 and Q4 are replaced by 50°C, and
consequently, equations 1 and 4 are modified. The conductive matrix
and nodal thermal forces and temperatures are expressed as
 1 0 0 0 0  T1   100 
    
0 38.33 6.67 0 −25 T2   1000
   
 0 6.67 38.33 0 −25  T3 =  1000 
 0 0 0 1 0  T4   100 
 0 −25 −25 0 100   
5000

   T5   

There are three equations and three unknowns. The T5 is the tem-
perature at the center of the body, which is 84.62°C. The temperature at
nodes 2 and 3 are equal to 69.33°C.
5.7 Thermal analysis of fin and chip using ANSYS
The fin shown in Figure 5.7 is used to manage the temperature of an

electronic chip that generates heat. Heat is generated in the heat source
within the chip, and its value is 25 Watts. The heat transfer process
is steady. Heat convection is applied at the entire external surface,
h = 15 W/m2 · °C and To = 20°C, while the bottom surface of the chip
is well insulated. Determine the maximum and average temperatures at
the bottom surface of the chip. The thermal conductivities of the used
materials: kfin = 110 W/m · °C, kchip = 1.25 W/m · °C, and khs = 2.5 W/m · °C.
This example is limited to thermal analysis. Hence, select Thermal

in the preferences. The Solid element is used, and its shape is Triangle
with six nodes.
Convection Fin
Convection
6.0 cm
Convection
Chip
3.0 cm
1.0 cm Heat
source
5.0 cm Insulation
9.0 cm
FIGURE 5.7 Heated chip with fin, and the boundary conditions.

A select Thermal
OK
Main Menu > Preprocessor > Element Types > Add/Edit/Delete
Add...
He at tr ansfer 245
A select Solid
B select Triangl 6node 35
OK
Close
Only the thermal conductivity is required to solve the problem. Note

that the thermal conductivities of the fin, chip, and heater are differ-
ent. By default, all areas will be assigned to material number 1. In this
problem, 1 is the material number of the fin, 2 for the chip, and 3 for the
heater.
A click on Thermal > Conductivity > Isotropic
A type 110 in KXX
OK
In the Define Material Models Behavior: Material > New Model
Make sure that the number 2 is in Define Material ID
OK
He at tr ansfer 247
A select Material Model Number 2

B click on Thermal > Conductivity > Isotropic
A type 1.25 in KXX
OK
OK

A type 2.5 in KXX
OK
The thermal conductivities are listed using list properties in utility

menu.
Utility Menu > List > Properties > All Materials

He at tr ansfer 249
The geometry is created using the ANSYS graphics. Setting up the

workspace is done using the WP Settings. Snap is enabled to allow
the mouse-click on the ANSYS graphics with an increment. Spacing
is the distance between the vertical or horizontal grids. The size of
the grids is specified in the Minimum and Maximum. The space is
divided into squares with side length of 0.01 m. Both the total width
and the height of the grids are 0.09 m. This setup makes the modeling
easy by creating keypoints, lines, and areas in the ANSYS graphics.
The grids should be first activated by selecting Display Working Plane
in the Utility Menu.
ANSYS Utility Menu > WorkPlane > Display Working Plane
Utility Menu > WorkPlane > WP Settings …
A select Grid Only

B type 0.01 in Snap Incr
C type 0.01 in Spacing
D type 0 in Minimum
E type 0.09 in Maximum
OK
Click on zoom in and out, until the ANSYS graphics show all the grids.
ANSYS graphics show the grids

The keypoints are created first, and then lines. Finally, fin, heater,
and chip areas are created. The thermal conductivities of the chip,
heater, and fin are assigned using element attribute in the meshing tools
window.
Keypoints > On Working Plane
Click on the ANSYS graphics window at the location of the key-

points, as shown in the following figure.
OK
Now, the grids are deactivated by selecting Display Working Plane
in Utility Menu.
ANSYS graphics show the keypoints

He at tr ansfer 251
ANSYS Main Menu > Preprocessor > Modeling > Create > Lines >
Lines > Straight Line
Click on two keypoints to create a line, and continue to create all

lines. The created lines are shown in the following figure.
OK
ANSYS graphics show the lines

ANSYS Main Menu > Preprocessor > Modeling > Create > Areas >
Arbitrary > By Lines
Click on chip lines to create a large rectangle area, and then in Create
Area by Lines window, click on
Apply
Click on heater lines to create a small rectangle area, and then in
Create Area by Lines window, click on
Apply
Click on fin lines to create the fin area, and then in Create Area by
OK
ANSYS graphics show the areas

The area between the heater and the chip is separated, and the area
of the heater is just over the area of the chip. The two areas should be
connected by common lines. Overlap merges the two areas, and the
boundary lines of the heater are sheared with the chip.
Overlap > Areas
Pick All
A select Area
B click on Set
He at tr ansfer 253
Click the chip’s area only
The above window will show up. Because the chip and heater
have the same centroid, ANSYS is inquiring about which area should
be selected. The chip or heater selection can be switched by clicking on
Prev and Next buttons in the Multiple_Entities window. Make sure
that the chip’s area is highlighted. In Multiple_Entities window, click on
OK
In Area Attributes window, click on
OK
By selecting number 2, the properties of number 2 in the material

model are assigned to the chip. The fin by default has the properties of
number 1 in the material model.
Apply
Click the heater area only
The chip or heater selection can be switched by clicking on Prev

and Next buttons in the Multiple_Entities window. Make sure that the
heater’s area is highlighted.
OK
In Area Attributes window, click on
OK
By selecting number 3, the properties of number 3 in the material
model are assigned to the heater.
OK
To ensure that thermal conductivity of the fin and chip are assigned
correctly, the components are colored according to their material number.
He at tr ansfer 255
The material coloring has no effect on the solution. The geometry is meshed

with triangular six-node elements. A free mesh is generated using the smart
mesh option. The mesh refinement is 1.
Utility Menu > PlotCtrls > Numbering …
A select Material numbers

B select Colors only
OK
Utility Menu > Plot > Areas
ANSYS graphics show the heater, fin, and chip with different colors
A select Smart Size

Mesh
Pick All
Close the Mesh Tool window

Close
Boundary conditions are applied in the solution task. A volumetric

heat generation is applied to the heater only. The convection bound-
ary condition is applied to the fin surface and the vertical sides of
the chip. The bottom surface of the chip is well insulated, and the
He at tr ansfer 257
zero heat flux simulates the insulation boundary condition. However,

if no boundary condition is specified at any external surface, ANSYS
will consider it as an insulated boundary condition. No boundary con-
ditions are applied at a common line between the chip and fin, and
the chip and heater. The heat generation must be per unit volume.
The applied heat generation is divided by the area of the heater
because the problem is two dimensional. The heater volumetric heat
generation is calculated as
25
Q= = 50,000 W /m 3
0.05 × 0.01 × 1
Main Menu > Solution > Define Loads > Apply > Thermal > Heat
Generat > On Areas
Click on the heater area, where heat generation is applied.
Because the chip and heater are overlapped, and having the same
centroid, ANSYS is inquiring about which area should be selected. The
chip or heater selection can be switched by clicking on Prev and Next
buttons in the Multiple_Entities window. Make sure that the heater’s area
is highlighted.
OK
In Apply HGEN on areas window, click on
OK
A type 50000 in Load HGEN value

OK
Main Menu > Solution > Define Load > Apply > Thermal >
Convection > On Lines
Click on fin external surfaces and two vertical surfaces of the chip
where the convection boundary condition is applied. Then in Apply
CONV on lines window, click on
OK
A type 15 in Film coefficient

B type 20 in Bulk temperature
OK
The preprocessor and solution tasks are completed.
OK
He at tr ansfer 259
Close
The solution is successfully completed, and no error messages are

posted. In the postprocessor, the temperature contours should be carefully
inspected to ensure that the boundary conditions are applied correctly.
Nodal Solution
A click on Nodal Solution > DOF Solution > Nodal Temperature
OK
ANSYS graphics show the temperature contours

The temperature contours indicate that the maximum temperature is

located at the bottom surface of the chip, and is equal to 26.431°C. The
default number of contours is 9, and this number can be increased to a
higher value for better data analysis. First, the graphics device must be
changed to win32C, and then the number of contours can be increased up
to 114 contours. Notice that increasing the number of contours does not
mean that the accuracy of the result is improved. The vector plot showing
the heat flow from the chip to the fin can be presented. The red arrow is
for high value of the heat flux. The average temperature at the base of the
chip is calculated using the path operation. To create a path, there are two
options: Arbitrary and Circular paths. The Arbitrary path can be made
from straight-line segments by clicking on the ANSYS graphics, and the
grids should be enabled. For this example, the Arbitrary path is utilized.
Main Menu > General Postproc > Path Operations > Define Path >
On Working Plane
OK
Click on the ANSYS graphics window at the right and left bottom
corners of the chip, as shown in the following figure.
He at tr ansfer 261
A click on left bottom corner of the chip

B click on right bottom corner of the chip
Then in On Working Plane window, click on
OK
A name the path as base

B type 100 in Number of divisions
OK
The name of the path is optional. The number of the data set is the
maximum number of field variables. The number of division is 20 by
default, and increasing this number to 100 will produce a smoother plot.
Next, the field variable is assigned to the path for plotting. This can be
accomplished by using the Map onto Path in the path operation. Only one
variable can be selected. For this example, the temperature is selected.
Main Menu > General Postproc > Path Operations >

Map onto Path
A select Temperature TEMP

OK
Now, the temperature is ready to be plotted. In the Plot path item, there
are two options. The temperature can either be plotted or listed. The list
results can be exported to another software, such as EXCEL.
Main Menu > General Postproc > Path Operations > Plot path
Items > On Graph
A select TEMP
OK
ANSYS graphics show the temperature distribution at the base

The temperature at the exit is perfectly parabolic due to the sym-
metry in the problem. The above graph indicates that the maximum
temperature at the bottom surface of the chip is 26.079°C. The aver-
age temperature at the base can be determined using the integration in
the path operation. The value of the integration must be divided by the
path length to get the average value of the variable, and the path length
He at tr ansfer 263
is 0.09 m. Hence, number 11.111 is entered in the FACT and will be
multiplied by the integration result. Selecting S in the Lab2 means that
the integration is performed along the path.
Main Menu > General Postproc > Path Operation > Integrate
A type 11.1111 in FACT Factor

B select TEMP in Lab1 1st Path item
C select S in Lab2 2nd Path item
OK
The ANSYS output window shows the value of the average tempera-
ture at the base which is 25.076°C.
5.8 Finite element method for transient heat transfer
Finite element formulation for transient conduction heat transfer is pre-

sented in this section. The conduction energy equation in per unit vol-
ume and in Cartesian coordinate can be expressed as
∂  ∂T  ∂  ∂T  ∂  ∂T  ∂T
 kx + ky +  kz  + q = ρCp (5.72)
∂x  ∂x  ∂ y  ∂ y  ∂z  ∂z  ∂t
where Cp is the specific heat of the material. The finite element method is
an efficient way to solve transient conduction problems. The heat transfer
solution can be used to estimate the heat flow at the system’s boundary,
or to determine the temperature distribution for thermal-stress analysis.
In this section, the variation formulation is used to obtain the conductive
equation, and it is accomplished by minimizing the following potential
function:
1   ∂T  
2
 ∂T  1
∫ ∫ ∫
h(T − T∞ ) dS
∫
2
I= k x    dV −  Q −ρCp  T dV − q T dS +
2  ∂x  ∂t 2
v v s s
(5.73)
The second term in Equation 5.73 adds an additional contribution for

transient as follows:
ΩQ = −
∫ ( Q − ρC T ) T dV (5.74)
V
p
where T = ∂T / ∂ t. The temperature at nodes can be expressed in terms

of the shape functions as follows:
{T} = [ N i ]{T i } (5.75)
Substituting temperature function (5.75) into the transient term in the

heat conduction equation (5.74) yields
ΩQ = −
∫ ([ N ]{T } Q − ρ C [ N ]{T }[ N ]{T }) dV (5.76)
V
i i p i i i i
Minimizing the transient term with respect to the nodal temperature

is as follows:
∫ ([ N ] Q − ρ C [ N ][ N ]{T }) dV (5.77)
∂Ω Q
=−
∂{Ti }
i p i i i
V
He at tr ansfer 265
The second term in Equation 5.77 is an additional term that should be

added to the conductive matrix, and it is expressed as

∫ (ρC [ N ][ N ]) dV {T } = [ m ]{T } (5.78)
V
p i i i i
where [m] is the element mass matrix. Equation 5.78 is added to the
conductive matrix to account for transient heat transfer as
[ K ]{T} + [ m ]{T i } = {F } (5.79)
where

∫
V
∫
[ K ] = [ B]T [ D ][ B] dV + h [ N ]T [ N ] dS (5.80)
S
and
{F} = {FQ} + {Fq} + {Fc} (5.81)
{FQ} is volumetric heat generation in the element, {Fq} is applied heat

flux on the external surface of the element, and {Fc} is applied convec-
tion on the external surface of the element. The integration is over the
surface in which convection or heat flux is applied. The global conduc-
tive matrix can be obtained by assembling the conductive matrixes as
follows:
[K ] = ∑ [Ke =1
(e)
] (5.82)
and the nodal heat force are also assembled to form a global one as
follows:
{F } = ∑ {Fe =1
(e)
} (5.83)
Also, the global temperature vector can be obtained as
{ T} = ∑ {Te =1
(e)
} (5.84)
5.9 Unsteady thermal analyses of a masonry brick

using ANSYS
The cross-sectional area of a masonry brick, as shown in Figure 5.8a,

is made of cement and hollow cylinders. The purpose of the numerical
8 × D 0.04 m To and ho
0.04 m
Air R 0.02 m
0.04 m
0.08 m Cement
0.08 m Cement
Air
0.04 m
0.04 m 0.08 m 0.04 m

Ti and hi
(a) (b)
FIGURE 5.8 (a) Masonry brick with boundary conditions. (b) The considered geometry for numeri-
cal simulation.
simulation is to study the heat flow from an outdoor to the indoor space.
The initial temperature of the brick is 25°C.
Property Air Cement

Density (kg/m3) 1.125 2500
Conductivity (W/m · K) 0.025 0.8
Specific heat (J/kg · K) 1005 750
At the indoor surface of the brick, the convection boundary condition

is applied with time-independent temperature and heat transfer coeffi-
cient, Ti = 20°C and hi = 10 W/m2 · °C. At the outdoor surface, the heat
transfer coefficient is time independent, ho = 25 W/m2 · °C, but the tem-
perature is time dependent. The following expression is used to simulate
the outdoor temperature variation:
2π 
T ( t ) = 35 + 5 Sin  t
 86400 
To ensure having a periodic condition, the simulation is kept run-

ning for 3 days, and the results for the last day are presented. Since
there is symmetry in the geometry and boundary conditions, only
one-eighth of the geometry is considered. Figure 5.8b shows the con-
sidered geometry for numerical simulation. Outdoor convection is
applied at the upper lateral line, and indoor convection is applied at
the lower lateral line. The two vertical lines are well insulated, which
show temperature history at the indoor and outdoor surfaces of the
brick for 1 day.
He at tr ansfer 267

A select Thermal
OK
Add…
A select Solid
B select Triangl 6node 35
OK
Close
For the material properties, thermal conductivity, specific heat,

and density are required to solve the problem because the problem is
unsteady. Note that the material properties of the air and cement are
different. By default, all areas will be assigned to material number 1.
In this problem, the material number of the air is 1, and 2 for the cement.
A click on Thermal > Conductivity > Isotropic
A type 0.025 in KXX
OK
He at tr ansfer 269
Click on Thermal > Specific Heat
A type 1005 in C
OK
Click on Thermal > Density
A type 1.125 in DENS
OK
OK

A type 0.8 in KXX

OK
Click on Thermal > Specific Heat
A type 750 in C
OK
He at tr ansfer 271
Click on Thermal > Density
A type 2500 in DENS

OK
The geometry is modeled by creating a rectangle and circles. The
Boolean operation is utilized to merge the areas.
A type 0 in WP X
B type 0 in WP Y
OK
Solid Circle
A type 0.04 in WP X
B type 0.04 in WP Y
Apply
A type 0.04 in WP X
B type 0.12 in WP Y
OK
He at tr ansfer 273

The cement and the air areas are separated, and the area of the air is
just over the area of the cement. The two areas should be connected by
common lines. Overlap operation merges the areas, and the boundary
lines of the air will be sheared with the cement. Then, the extra areas of
the air are deleted.
Overlap > Areas
Pick All
Main Menu > Preprocessor > Modeling > Delete > Area and Below
Click on extra air areas to select them, and in Delete Area and Below
window, click on
OK
ANSYS graphics show the final geometry

A select Areas
B click on Set
Select the brick area only. In Area Attributes window, click on
OK
The following windows will show up. By selecting number 2,

the properties of number 2 in the material model are assigned to the
cement. The air by default has the properties of number 1 in the mate-
rial model.
OK
He at tr ansfer 275
To ensure that thermal conductivities of the air and cement are

assigned correctly, the components are colored according to their mate-
rial number. This has no effect on the solution. The geometry is meshed
with triangular six-node elements. A free mesh is generated using the
smart mesh option. The mesh refinement is 1.
A select Material numbers

B select Colors only
OK

A select Smart Size

Mesh
Pick All
Close the Mesh Areas window
Close
ANSYS graphics show the brick and air with

different colors, and finite element mesh
The type of the analysis is changed to Transient, and a Full solution
method is selected. This solution method fully imposes the boundary con-
ditions at time 0. The solution control window is used to specify the total
time duration, time step, and number of output results to be stored for the
postprocessor. The total time duration is 3 days, or 259,200 seconds, and
the total time duration is solved by approximately 1000 steps, or the s olution
time step size is 250. If less time step size is used, such as 125 seconds, the
results will be more accurate, but the computational time will be doubled.
Data will be stored at every two subset. Hence, there will be 518 data sets
available for the postprocessor task.
A select Transient
OK
He at tr ansfer 277
OK
Main Menu > Solution > Analysis Type > Sol’n Controls
A type 259200 in the Time at end of loadstep

D select Write every Nth substep
E type 2 in where N =
F click on Transient tab
In the Transient tab, selection between stepped loading or ramped

loading is specified. If stepped loading is selected, loads are fully
applied at the time = 0. If ramped loading is selected, loads are linearly
increasing over the entire process. In this example, the stepped loading
is selected.
A select Stepped loading.
OK
Boundary conditions and initial conditions are applied in the solution
task. The function editor is used to apply a transient temperature formula
for a convective boundary condition on the upper horizontal surface.
This technique is convenient for this problem since an equation for the
temperature is given.
ain Menu > Solution > Define Loads > Apply > Functions >
M
Define/Edit
A type the equation: 35+5*sin(2*3.1415/86400*{TIME})
In Equation Editor, click on File then Save
Save the file as Tout, and this file name is optional. After saving the
function, it is required to load it to the ANSYS solution using the read file.
Save
He at tr ansfer 279
Close the Equation Editor window

Read File
Select the file Tout.func, then
Open
A type Tout in the Table parameter name, and this name is optional,
and should not be the same as the file name of the function.
Main Menu > Solution > Define Loads > Apply > Thermal >
Click on the bottom line where the indoor convective boundary con-
dition is applied, and then in Apply CONV on lines window, click on
OK
A type 10 in the VALI Film coefficient

B type 20 in the VAL2I Bulk temperature
OK
Click on the upper line where the outdoor convective boundary con-
dition is applied, and then in Apply CONV on lines window, click on
OK

OK
A select TOUT
OK
He at tr ansfer 281
Main Menu > Solution > Define Loads > Apply > Initial Condit’n >
Define
To select all nodes in the domain, in Define Initial Conditions

window, click on
Pick All
A select TEMP in DOF to be specified

B type 25 in the VALUE Initial value of DOF
OK
The solution task is now completed, and the model is now ready to be
solved. During the solution task, ANSYS output windows will show the
progress of the solution, and carefully monitor the run.
OK
Close
The temperature contours are presented at 12 PM of the third day, or

at time = 216,000 seconds. First, the time step is loaded, and then the
temperature contours are plotted.
Main Menu > General Postproc > Read Results > By Time/Freq
A type 216000 in TIME Value of time or freq
OK
Nodal Solution
OK
He at tr ansfer 283
ANSYS graphics show the temperature contours at time = 216,000 seconds

Determining the temperature history at a specific location in the
domain is required. Here, the temperature history at the upper and bot-
tom surfaces is presented in graphical form. Results are presented for
the third day only. Hence, the time range is from 172,800 to 259,200
seconds as follows:
A click on the data properties button

A select X-Axis
B select Specified
C type 172800 in Min
D type 259200 in Max
OK
A click on green + button
Apply
Click at right bottom corner, and in Node for Date window, click on
Apply
He at tr ansfer 285
OK
Click at right upper corner, and in Node for Date window, click on
OK
A click on Ctrl key, and then click on TEMP_3 and TEMP_4 to

select both temperatures
B click on the graph button
OK
ANSYS graphics show the temperature history at the selected locations

Animation of the temperature contours from time = 172,800 to
259,200 seconds can be easily accomplished using animate in the
PlotCtrls. The number of the frames in the animate over time is the num-
ber of pictures in the avi file, while the animation time delay is the dis-
play period between two pictures.

B select Time Range
C type 172800 and 259200 in Range Minimum, Maximum
D select Temperature TEMP
OK
The ANSYS will show an animation of the heating process of the

brick. The animation file will be stored in the working directory, and its
format is avi.
He at tr ansfer 287
PROBLEM 5.1
Consider a composite wall shown in Figure 5.9. The wall is composed
of four layers with different thermal conductivities: k A = 0.22 W/m · °C,
k B = 0.16 W/m · °C, kC = 0.25 W/m · °C, and k D = 0.35 W/m · °C. A convec-
tive boundary condition is applied at the left surface, hi = 7.5 W/m2 · °C
and Ti = 22°C, and at the right surface, h = 15 W/m2 · °C and Ti = 45°C.
Calculate the temperature at the interfaces, and heat flow through the
wall using the finite element method. Given that LA = 2.5 cm, LB = 15 cm,
LC = 20 cm, and LD = 2.5 cm.
PROBLEM 5.2
Consider a composite pipe shown in Figure 5.10. The pipe is composed
of three pipes with different thermal conductivities (k A = 0.22 W/m · °C,
k B = 0.16 W/m · °C, and kC = 0.25 W/m · °C), and radii (rA = 0.03 m,
rB = 0.02 m, and rC = 0.01 m). A convective boundary condition is
applied at the right surface: h1 = 10 W/m2 · °C and T1 = 25°C, and fixed
temperature at the left surface, T2 = 45°C. The external surface of the
pipe is well insulated. Calculate the temperature at the interfaces, and
heat flow through the pipe using finite element method. Given that
LA = 0.05 m, LB = 0.075 m, and LC = 0.025 m.
PROBLEM 5.3
The two-dimensional body shown in Figure 5.11 is subjected to a fixed
temperature boundary at the left and right vertical sides, T L = 100°C
and T R = 50°C, respectively. The two horizontal sides are well insu-
lated. The side length of the body is 1 m, and it has thermal conduc-
tivity of 10 W/m · °C. Determine the temperature at the center of the
body using the finite element method. Consider the suggested four
elements mesh.
Ti
To
kA kB kC kD
hi
ho
LA LB LC LD FIGURE 5.9 Heat conduction in a

composite wall.
kA kB kC
T1 T2
h1
LA LB LC FIGURE 5.10 Heat conduction in a

composite pipe.
Insulated
TL = 100˚C TR = 50˚C
1m
FIGURE 5.11 Two-dimensional body subjected to fixed temperature boundary conditions and sug-
gested finite element mesh.
PROBLEM 5.4
The two-dimensional body shown in Figure 5.12 is subjected convection
at the inclined side, with TR = 50°C and h = 20 W/m2 · °C. At the left ver-
tical side, a fixed temperature boundary is imposed, TL = 100°C, while
the two horizontal sides are well insulated. It has thermal conductivity
of 10 W/m · °C. Determine the temperature at the inclined surface of the
body using the finite element method. Consider the suggested three ele-
ments mesh.
PROBLEM 5.5
The fin shown in Figure 5.13 is used to manage the temperature
of an electronic chip that generates heat. Heat is only generated
in the chip, and its value is 15 Watts. The heat transfer process is
steady. Heat convection is applied along the entire external surfaces,
h = 10 W/m 2 · °C and To = 25°C, while the bottom surface of the chip
is well insulated. Determine the maximum and average tempera-
tures at the bottom surface of the chip, given k fin = 75 W/m · °C, and
kchip = 0.95 W/m · °C.
PROBLEM 5.6
Four cylindrical heaters are embedded in a conductive epoxy for heat-
ing purposes. An aluminum alloy fin is used to enhance heat flow out of
the heaters. Figure 5.14 shows the geometry and the thermal conductivity
of the components. Each heater generates 12.5 Watts, and free convec-
tion, h = 10 W/m2 · °C and T = 20°C, is applied at all external surfaces.
Determine:
1. Maximum operating temperature in the device.
2. Temperature distribution a long path A–A.
3. Average temperature a long path A–A.
He at tr ansfer 289
0.5 m
Insulated
TR = 50°C
h = 20 W/m2  °C
0.5 m
TL = 100°C 1m
FIGURE 5.12 Two-dimensional body subjected convection and fixed temperature boundary condi-
tions and the suggested finite element mesh.
Convection
6.0 cm
Convection
Convection
Fin
3.0 cm
Chip
Insulation
9.0 cm
FIGURE 5.13 Heated chip with fin, and the boundary conditions.
Epoxy
(15 W/m  K) Heater
2 cm 1.5 cm Fin (25 W/m  K)
4 × R 0.5 cm
(85 W/m  K)
A
1.5 cm
2 cm
1 cm
1 cm
A
3 cm 3 cm
FIGURE 5.14 Cylindrical heaters with fin.

PROBLEM 5.7
A masonry brick, as shown in Figure 5.15, is made of cement and
air holes. At the indoor surface, convective boundary conditions are
applied with time-independent temperature and heat transfer coeffi-
cient, Ti = 22.5°C and h i = 7.5 W/m 2 · °C. At the outdoor surface, heat
transfer coefficient is time independent, ho = 27.5 W/m 2 · °C, but the
temperature is time dependent. The initial temperature of the brick
is 20°C.
Property Air Cement

Density (kg/m3) 1.125 2400
Conductivity (W/m · K) 0.025 0.82
Specific heat (J/kg · K) 1005 740
The following expression simulates the outdoor temperature:
2π 
T ( t ) = 30 + 10 Sin  t
 86400 

To ensure having a periodic condition, the simulation should be kept

running for 3 days, and the results for the last day should be presented.
Show temperature histories at the indoor and outdoor surfaces of the
brick for the last day.
PROBLEM 5.8
For the electronic board shown in Figure 5.16, the initial temperature of
the entire system is 20°C. At time = 0, 24 Watts are generated in the pro-
cessor, and convection is applied to the boundaries with h = 5 W/m2 · ˚C
and To = 20°C, while the board bottom surface is well insulated. The total
To and ho
0.05 m
Air
0.10 m
Cement
0.05 m 0.05 m
0.10 m Ti and hi
FIGURE 5.15 Masonry brick and the boundary conditions.

He at tr ansfer 291
0.005 m 0.005 m
Fin
0.045 m
Processor
Chip
Board
0.01 m
0.01 m 0.015 m
0.005 m
0.015 m 0.065 m 0.035 m 0.04 m 0.015 m
FIGURE 5.16 Electronic board.
time duration of the device is 10 hours. Solve this transient process using
a time step of 25 seconds. Determine the following:
a. Does the system reach the steady-state condition?
b. What is the maximum temperature of the processor and chip at
time = 5 hours?
c. Show a temperature history at the center of the processor and chip.
d. Show the temperature distribution along the external surface of
the chip at time = 5 and 7.5 hours.
Property Chip Fin Processor Board

Density (kg/m )
3 900 2050 920 910
Conductivity (W/m · K) 2.1 45 2.1 3.2
Specific heat (J/kg · K) 2200 4500 1300 1150
Cha p t e r SIX
Fluid mechanics
6.1 Governing equations for fluid mechanics
The mass conservation equation in a differential form can be obtained

by applying the mass conservation principle on a differential control
volume as shown in Figure 6.1. Considering the control volume, the net
mass flow rate in the x-, y-, and z-directions can be expressed as
∂
x-direction:
∂x
(ρu ) dx dy dz (6.1)
∂
y-direction:
∂y
(ρv ) dx dy dz (6.2)
∂
z-direction:
∂z
(ρw ) dx dy dz (6.3)
The rate of change of mass inside the control volume can be obtained
from the Reynolds transport theory as follows:
∂ρ ∂ρ

∫ ∂t dV = ∂t dx dy dz (6.4)
CV
The net mass flux into the control volume should be equal to the rate
of change of mass inside the control volume. The mass conservation in
differential form can be expressed as
∂ρ ∂
+
∂ t ∂x
(ρu ) + ∂∂y (ρv ) + ∂∂z (ρw ) = 0 (6.5)
Newton’s second law on a differential control volume, as shown in
Figure 6.1, can be used to obtain the conservation of momentum equa-
tion. The net forces on the control volume should be balanced with the
acceleration of the control volume times its mass as follows:

a dm = ∑ dF (6.6)
293
my
mz+dz
mx mx+dx
mz
my+dy
FIGURE 6.1 Fixed control volume.

The acceleration vector field a is obtained from the total time deriva-
tive of the velocity vector
dV du dv dw
a= = i+ j+ k (6.7)
dt dt dt dt
Each component of the velocity fields is a function of the space and

time. Using the chain rule, the scalar time derivative can be obtained
du ( x, y,z, t ) ∂ u ∂ u dx ∂ u dy ∂ u dz
= + + + (6.8)
dt ∂ t ∂x dt ∂ y dt ∂z dt
where u = dx/dt is the local velocity component in the x-direction, v = dy/dt

is the local velocity component in the y-direction, and w = dw/dt is the
local velocity component in the z-direction. The total derivative of u is
the acceleration in the x-direction
du ( x, y,z, t ) ∂ u du du du
ax = = +u +v +w (6.9)
dt ∂t dx dy dz
The acceleration in the y- and z-directions can be expressed as,

respectively,
dv ( x, y,z, t ) ∂v dv dv dv
ay = = +u +v +w (6.10)
dt ∂t dx dy dz
du ( x, y,z, t ) ∂w dw dw dw
az = = +u +v +w (6.11)
dt ∂t dx dy dz
The mass of the control volume must be equal to volume times the
density as follows:
dm = ρ dx dy dz (6.12)
Fluid mechanics 295
The forces on the control volume are of two types: body and surface.
The body force is due to the gravity:

dFb = ρg dx dy dz (6.13)
while the surface forces are due to the surface stresses, including normal
and parallel stresses. The surface stresses in the x-, y-, and z-directions
are as follows:
 ∂σ ∂τ yx ∂τ zx 
dFsx =  xx + + dx dy dz (6.14)
 ∂x ∂y ∂z 
 ∂σ yy ∂τ xy ∂τzy 
dFsy =  + + dx dy dz (6.15)
 ∂y ∂y ∂z 
 ∂σ ∂τ ∂τ yz 
dFsz =  zz + xz + dx dy dz (6.16)
 ∂x ∂y ∂z 
The equations of motion in the x-, y-, and z-directions can be

expressed as, respectively,
 ∂u ∂u ∂u ∂u 
x-direction: ρ  +u +v +w 
 ∂ t ∂ x ∂ y ∂z 
(6.17)
 ∂σ ∂τ yx ∂τ zx 
=  ρgx + xx + +
 ∂x ∂y ∂z 
 ∂v ∂v ∂v ∂v 
y-direction: ρ  +u +v +w 
 ∂ t ∂ x ∂ y ∂z 
(6.18)
 ∂σ yy ∂τ xy ∂τ zy 
=  ρg y + + +
 ∂x ∂y ∂z 
 ∂w ∂w ∂w ∂w 
z-direction: ρ  +u +v +w
 ∂ t ∂ x ∂ y ∂z 
(6.19)
 ∂σ ∂τ ∂τ yz 
=  ρgz + zz + xz +
 ∂x ∂y ∂z 
For Newtonian fluid, the stress components are obtained from the
theory of elasticity, and they are
∂u
σ xx = − P + 2µ (6.20)
∂x
∂v
σ yy = − P + 2µ (6.21)
∂y
∂w
σ zz = − P + 2µ (6.22)
∂z
 ∂ u ∂v 
τ xy = τ yx = µ  + (6.23)
 ∂ y ∂x 
 ∂v ∂w 
τ yz = τ zy = µ  + (6.24)
 ∂z ∂ y 
 ∂w ∂v 
τzy = τ yz = µ  + (6.25)
 ∂ y ∂z 
Substituting the stress equations into the equations of motion, we have
 ∂u ∂u ∂u ∂u  ∂P  ∂2 u ∂2 u ∂2 u 
ρ
+u +v +w  =− + µ  2 + 2 + 2  + ρgx
 ∂t ∂x ∂y ∂z  ∂x  ∂x ∂y ∂z 
(6.26)
 ∂v ∂v ∂v ∂v  ∂P  ∂2 v ∂2 v ∂2 v 
ρ +u +v +w  =− + µ  2 + 2 + 2  + ρg y
 ∂t ∂x ∂y ∂z  ∂y  ∂x ∂y ∂z 
(6.27)
 ∂w ∂w ∂w ∂w  ∂P  ∂2 w ∂2 w ∂2 w 
ρ +u +v +w  =− + µ  2 + 2 + 2  + ρgz
 ∂t ∂x ∂y ∂z  ∂z  ∂x ∂y ∂z 

(6.28)
Equations 6.26–6.28 are called the Navier–Stokes equations. They

are nonlinear and nonhomogenous partial differential equations.
6.2 Finite element method for fluid mechanics
The finite element method is utilized to solve the governing equations

and to discretize the computational domain for fluid flow problems.
Four nodes of quadrilateral elements were used for the numerical dis-
cretization. ANSYS has only this type of element in its library for fluid
dynamics. Weighted integral statements of the mass, momentum, and
energy conservations over a typical element Ω are given by
∫ w f dΩ (6.29)
Ω
1 1
∫ w f dΩ (6.30)
Ω
2 2
Fluid mechanics 297

∫ w f dΩ (6.31)
Ω
3 3
where f1, f2, and f3 are mass, momentum, and energy conservations,
respectively. Then w1, w2, and w3 are weight functions, which are equal
to the interpolation functions. The choice of the weight function is
restricted to the spaces of approximation functions used for pressure,
velocity fields, and temperature. The pressure, velocity fields, and tem-
perature are approximated as follows:
P ( x, t ) = ∑Φ ( x) P ( t ) = Φ
l =1
l l
T
{P} (6.32)
u i ( x, t ) = ∑Ψ (x) u ( t ) = Ψ
n =1
n
n
i
T
{u i } (6.33)
T ( x, t ) = ∑Θ
m =1
m ( x ) Tm ( t ) = Θ T {T} (6.34)
where Φ, ψ, and θ are vectors of the shape functions, and P, ui, and
T are vectors of nodal value of the pressure, velocity components, and
temperature, respectively. The weight functions have the following
correspondence:
w1 ≅ Φ, w2 ≅ Ψ, w3 ≅ Θ (6.35)
The mass and momentum conservations can be symbolically written

in the following matrix form:
[AT] {u} = 0 (6.36)
[C] {u} + [K] {u} – [A] {P} = {F} (6.37)
[D] {T} + [L] {T} = {G} (6.38)
The coefficients are defined by
∂ψ
Ai =
∫ ∂x Φ dΩ (6.39)
Ω
i
T
∂ψ T

∫
Ci ( u j ) = ρψ u i
Ω
∂x i
dΩ (6.40)
∂ψ ∂ψ T
K ij = µ
∫ Ω
∂x j ∂x i
dΩ (6.41)

∫
Fi = ψτ i dΓ (6.42)
Γ
∂Θ T

∫
D i ( u j ) = ρ CP ψu i
Ω
∂x i
dΩ (6.43)
∂Θ ∂Θ T

∫
L ij = k
Ω
∂x i ∂x j
dΩ (6.44)

∫
Ω
∫
G = Θ Q ′′′dΩ + Θ q ′′dΓ (6.45)
Γ
where k is the thermal conductivity, Cp is the specific heat, and τi is the

component of the total boundary stress, which is the sum of the viscous
boundary stress and the hydrostatic boundary stress. The q″ is the heat
flux applied at the boundary of the elements.
6.3 Entrance length in developing flow in

a channel using ANSYS
Water flow in a two-dimensional channel, as shown in Figure 6.2,

develops in the axial direction. At the inlet, the flow is uniform and
the flow velocity is 0.005 m/s. The exit condition is a reference zero
pressure. The density and viscosity of water are ρ = 998.3 kg/m3 and
μ = 1.002 × 10 –3 (1.002e-3) Pa · s, respectively. Determine the entrance
length and show that the mass conservation principle is satisfied.

This example is limited to fluid analysis. Hence, select Fluid FLOTRAN.
The 2D FLOTRAN 141 element is used, and its shape is rectangle with
Wall
Uniform Flow 0.05 m

flow in out
Wall
1.0 m
FIGURE 6.2 Channel with developing flow.

Fluid mechanics 299
four nodes. The 3D FLOTRAN 142 element is for three-dimensional

analysis. The density and viscosity are the only needed properties to solve
this problem.
A select FLOTRAN CFD
OK
Add…
A select 2D FLOTRAN 141
OK
Close
A click on CFD > Density

Fluid mechanics 301
A type 998.3 in DENS
OK
Click on CFD > Viscosity
A type 1.002e-3 in VISC
OK
A type 0 in WP X
B type 0 in WP Y
C type 1 in Width
OK
The smart mesh number of one will produce insufficient mesh density
to have accurate results. The elements in the domain can be additionally
increased by using the lines size control in the Mesh Tool. Lines are divided
into segments, which will be elements in these lines. The lines are divided
by either specifying the number of divisions or the lengths of the segments.
In this example, the number of divisions is specified. The vertical and lat-
eral lines will be divided by 25 and 500 segments, respectively.
Fluid mechanics 303
Click on two horizontal lines. Then, in Element Sizes on Picked

OK
A type 500 in NDIV No. of element divisions
Apply
Click on two vertical lines. Then, in Element Sizes on Picked Lines

window, click on
OK
A type 25 in NDIV No. of element divisions
OK
Fluid mechanics 305
A click on Mesh
Pick All

Zero velocity is imposed at the upper and lower walls to simulate wall
boundary condition. At the inlet, the velocity is uniform. At the exit,
zero pressure is imposed to simulate the exit boundary condition.
Main Menu > Solution > Define Loads > Apply > Fluid/CFD >
Velocity > On Lines
Click on the two horizontal lines. Then in Apply VELO load on lines
window, click on
OK
A type 0 in VX Load value

B type 0 in VY a Load value
OK
Fluid mechanics 307
Velocity > On Lines
Click at the inlet line. Then, in Apply VELO load on lines window, click on
OK
A type 0.005 in VX Load value

OK
Main Menu > Solution > Define Load > Apply > Fluid/CFD >
Pressure DOF> On Lines
Click on the exit line, and then in Apply PRES on lines window, click on
OK
A type 0 in PRES Pressure value
OK
The present problem is steady state and adiabatic. Hence, keep the
default setting in execution control. Notice that ANSYS is capable of
simulating turbulent and compressible flow. The maximum number of
iterations is 1000, and an additional 1000 iterations are required if the
termination criterion is not satisfied. The termination criterion for the
velocity components and pressure is 1 × 10 –6. The iterations will stop
if the maximum number of iterations is reached or the termination cri-
terion is satisfied. The material properties are MP, which means that
the ANSYS will use properties stored in the Material Properties in the
preprocessor task to solve the problem.
Main Menu > Solution > FLOTRAN Set Up > Execution Ctrl
A type 1000 in EXEC Global iteration

B type 1e-6 in the termination criterion for VX Velocity component
C type 1e-6 in the termination criterion for VY Velocity component
D type 1e-6 in the termination criterion for PRES Pressure
OK
Fluid mechanics 309
Main Menu > Solution > FLOTRAN Set Up > Fluid Properties
A select MP table in Density

B select MP table in Viscosity
OK
A confirmation window will show up. Read it carefully to avoid

an unexpected error. The value –1 indicates that the property is not
available. The conductivity and specific heat are not required to solve
the problem.
OK
The following step will initiate the numerical iterations. Carefully

examine the normalized rate of change for all field variables. The nor-
malized rate of change should reach termination criterion for all field
variables to declare the convergence. Otherwise, additional iterations are
required. When the normalized rate of change is decreasing, the solution
process is approaching the convergence. If the solution is diverged, there
are either incorrect boundary conditions or fluid properties.
Main Menu > Solution > FLOTRAN Set Up > Run FLOTRAN
Close
ANSYS graphics show the solution convergence

The normalized rate of change reaches the termination criterion.
Hence, the convergence is reached. The Last Set is selected in the Read
Results to ensure that the data from the last iteration set are loaded for
the postprocessor. Otherwise, no results will be shown in the postpro-
cessor. A plot of velocity vectors is presented in the postprocessor task,
followed by velocity profile along the centerline of the channel and at
the exit.
Fluid mechanics 311
Main Menu > General Postproc > Read Results > Last Set
Nodal Solution
A select X-Component of fluid velocity

OK
ANSYS graphics show the vector for the velocity

The red contours are for the maximum velocity in the channel, while
the blue contours are for minimum velocity. The velocity is maximum
at the centerline and zero at the wall. The developing region is clearly
visible at the entrance region of the channel. Notice that the velocity
has a parabolic velocity profile at the exit of the channel. To determine
the entrance length, a plot of the x-velocity component along the channel’s

centerline is created using the path operation in the postprocessor. The
x-velocity component should be increased from its initial value at the
inlet until it becomes unchanged. The path is created by specifying two
points, one at the inlet and the other at the exit along the centerline.
The number of division is the number of points used to create the plot.
A higher number of divisions will create a smoother plot.
Utility Menu > WorkPlane > WP Setting
A select Grid Only

OK
Fluid mechanics 313
On Working Plane
The path is arbitrary. In On Working Plane window, click on
OK
Click on the ANSYS graphics window at the right and left ends of the
centerline of the channel as shown below.
A click on left end of the centerline

B click on right end of the centerline
In On Working Plane window, click on
OK
A type center in Define Path Name; the name of the path is optional
OK
Main Menu > General Postproc > Path Operation > Map onto Path
A select Velocity VX
OK
The velocity in the x-direction is ready to be plotted. In the Plot Path

Item, there are two options. The velocity can be either plotted or listed.
The list results can be exported to another graphical software such as
EXCEL. First, the grids are removed from ANSYS graphics.
Main Menu > General Postproc > Path Operation > Plot Path Item >
On Graph
A select VX
OK
Fluid mechanics 315
ANSYS graphics show the x-velocity profile

along the centerline of the channel
The figure indicates that the flow is fully developed at the exit of the
channel. The x-velocity component becomes invariant at a distance of
0.75 m from the entrance. Experimentally, the entrance length can be
determined using the following equation:
L = 0.056 Re H
where Re is the Reynolds number and H is the channel’s height. The

Reynolds number is calculated using the following expression:
ρVH 998.3 × 0.005 × 0.05

Re = = = 249.07
µ 1.002 × 10 −3
Then, the entrance length is 0.697 m, which is close to the ANSYS

results. The error between the two methods can be additionally reduced
if a finer mesh is used. To determine the average velocity at the exit, the
path operation is used to plot the velocity profile with integration.
Main Menu > General Postproc > Path Operation > Define Path >
On Working Plane
The path is arbitrary. In On Working Plane window, click on
OK
Click on the ANSYS graphics window at the top and bottom

c orners at the exit of the channel, and then in On Working Plane win-
dow, click on
OK
A type Exit in Define Path Name; the name of the path is optional
OK
OK
Main Menu > General Postproc > Path Operation > Plot Path Item >
On Graph
A select VX
OK
Fluid mechanics 317
ANSYS graphics show the x-velocity profile at the exit of the channel
The velocity profile at the exit is perfectly parabolic. This graph can
be compared to the analytical velocity profile of a fully developed flow
to ensure that the obtained solution is accurate. The average velocity
at the exit can be determined using the integration in the path opera-
tion. The value of the integration must be divided by the path length to
get the average value. The number 20 in the Factor is the inverse of the
path length. Selecting S in the Lab2 means that the integration is per-
formed along the path.
A type 20 in FACT Factor

B select VX in 1st Path item
C select S in 2nd Path item
OK
The ANSYS Output window shows the average velocity at the exit,
which is 0.48 × 10 –2 m/s. The mass flow rate at the exit and inlet should
be the same as follows:
i =m
m e
or
(ρVA)i = (ρVA)e
Since the density and cross-sectional area of the inlet and exit are the
same, then
Vi = Ve
Comparing the inlet velocity, which is 0.005 m/s, to the exit velocity,
which is 0.48 × 10 –2 m/s, the two velocities are close to each other with
an error of 4%.
6.4 S tudying flow around a half cylinder in a channel

using ANSYS
Air at low velocity enters a channel as shown in Figure 6.3. A half

cylinder with a circular cross section is placed at the bottom wall of
the channel. The inlet velocity has fully developed velocity profile with
an average velocity of Um = 0.005 m/s. The exit condition is a refer-
ence zero pressure. Air’s density and viscosity are ρ = 1.25 kg/m3 and
y
Flow
x H = 0.2 m
Half cylinder out
0.15 m 0.1 m 0.75 m
FIGURE 6.3 Flow over a half cylinder in a channel.

Fluid mechanics 319
μ = 17.7 × 10 –6 Pa · s, respectively. Determine the pressure drop in the

channel and drag force on the half cylinder. Use the following equation
for the velocity profile at the inlet:
3  2y 
2
u ( y) = U m 1 −   
2   H 
OK
Add…

OK
Close
Since this problem is fluid dynamics, density and viscosity are
required to solve this problem. The geometry is modeled by creating a
rectangle and a circle. The Boolean operation is utilized to remove the
circle from the rectangle using subtraction.
A click on CFD > Density

Fluid mechanics 321
A type 1.25 in DENS
OK
OK

A type 0 in WP X
C type 1 in Width
OK
Solid Circle
A type 0.2 in WP X
OK
Fluid mechanics 323
Subtract > Areas
Click on rectangle area, and then in Subtract Area window, click on
Apply
Click on circular area, and then in Subtract Area window, click on
OK
A free mesh is generated using the smart mesh option, and the mesh
refinement is 1. More elements will be added to the computational
domain using refinement at elements.
A select Smart Size

C Click on Mesh
Pick All
D select Elements
E click on Refine
In Refine Mesh at Area window, click on
Pick All
OK
ANSYS graphics show the final mesh

The modeling and meshing tasks are now completed. The hydrody-
namics boundary conditions are applied. At the inlet, Vx is a function of
the y-direction and the Vy is 0. At the exit, zero pressure is imposed to
simulate a free exit boundary condition. Zero velocity components are
applied at the surface of the cylinder and lateral surfaces of the channel
to simulate a wall boundary condition.
The inlet velocity profile can be applied easily with the function edi-
tor of ANSYS. ANSYS will create a data table from the velocity func-
tion and apply it to the selected lines.
Fluid mechanics 325
Define/Edit
A type the equation 1.5*0.005*(1−(2*{Y}/0.2)^2)

In Equation Editor window, click on File then Save
Save the file as VinY. The file name is optional.
Save
Close the Function Editor window
After saving the function, it is required to load the function to the
ANSYS solution using the read file in function.
Main Menu > Solution > Define Loads > Apply > Functions > Read File
Select VinY.func
Open
A type VinY in the Table parameter name. The name of the table
is optional. The function “VinY” will be shown later when the
boundary conditions are applied.
OK
Velocity > On Lines
Click on the channel entrance line, and then in Apply VELO load on
lines window, click on
OK
A select Existing table in Apply VX load as a

OK
Fluid mechanics 327
A select VINY
OK
Velocity > On Lines
Click on the upper and lower lines of the channel, and the surface of
the cylinder, and then in Apply VELO load on lines window, click on
OK

OK
Pressure DOF > On Lines
Click on the channel’s exit line, and then on Apply PRES on lines
window, click on
OK

OK
The maximum number of iterations is 1000, and an additional
1000 iterations is required if the termination criterion was not satisfied.
The termination criterion for the velocity components and pressure is
1 × 10 –6. The iterations will stop if the maximum number of iterations
is reached or the termination criterion is satisfied for all field variables.
A type 1000 in EXEC Global iterations

B type 1e-6 in VX Velocity component
C type 1e-6 in VY Velocity component
D type 1e-6 in PRES Pressure
OK
Fluid mechanics 329
By default, the shear stress will not be calculated by ANSYS, which is

required to calculate the lift and drag forces. To calculate the shear stress,
it must be selected in the additional output in the FLOTRAN setup.
Main Menu > Solution > FLOTRAN Set Up > Additional Out >
RFL Output Derived
A select TAUW Output wall shear stress?

OK

OK
OK
OK
ANSYS graphics show the solution convergence

Fluid mechanics 331
As shown in the normalized rate of change, the solution did not reach
the maximum number of iterations, which is 1000, and all field variables
reach the termination criterion, which is 1 × 10 –6. Therefore, the solution
is converged. A plot of velocity vectors is presented in the postprocessor
task, followed by pressure drop, and lift and drag calculations.
Main Menu > General Postproc > Plot Results > Vector Plot >
Predefined
OK
ANSYS graphics show the vector for the velocity

The red arrows are for the maximum velocity in the channel, while
the blue arrows are for minimum velocity. The wake flow behind the
cylinder is visible in the figure, and the flow is very slow at this region.
The flow velocity is maximum above the cylinder. Notice that the veloc-
ity has a parabolic profile at the inlet and exit of the channel. To calculate
the pressure drop, the average pressure at the inlet is determined using
path operation in the postprocessor. The path at the inlet is created by
specifying two points. The grids should be enabled.
Utility Menu > WorkPlane > Display Working Plane
A select Grid Only

OK
Fluid mechanics 333
Main Menu > General Postproc > Path Operation > Define Path >
On Working Plane
OK

Click on zoom in and out until the ANSYS graphics show all grids.
Click on the ANSYS graphics window at the top and bottom corners at
the inlet of the channel, and then in On Working Plane window, click on
OK
A type inlet in Define Path Name. The name of the path is optional
OK
A select Pressure PRES
OK
Utility Menu > WorkPlane > Display Working Plane

Main Menu > General Postproc > Path Operation > Plot path Item >
On Graph
A select PRES
OK
ANSYS graphics show the pressure distribution at the inlet of the channel
Fluid mechanics 335
The average pressure at the inlet can be determined using the

integration in the path operation. The value of the integration must
be divided by the path length to get the average value of the vari-
able. The number 5 in the Factor is the inverse of the path length.
Selecting S in the Lab2 means that the integration is performed along
the path.
A type 5 in FACT Factor

B select PRES in Lab1 1st Path item
C select S in Lab2 2nd Path item
OK
The ANSYS Output window shows the average pressure at the

inlet, which is 0.34299 × 10 –4 N/m 2. At the exit, the pressure is speci-
fied as a boundary condition and equal to zero. Hence, the pressure
drop in the channel is equal to 0.34299 × 10 –4 N/m 2. To determine
the lift and drag forces on the surface of the cylinder, nodes along
the surface of the cylinder should be selected first. Then, ANSYS

automatically integrates the pressure and shear stress along the

surface of the cylinder. The nodal selection is performed in the fol-
lowing steps:
Utility Menu > PlotCtrls > Pan-Zoom-Rotate …
Zoom the area around the cylinder.
Utility Menu > Select > Entities
A select Lines
OK
Carefully select the two lines at the surface of the cylinder, and then
in Select Lines window, click on
OK
Fluid mechanics 337
Utility Menu > Select > Entities
A select Nodes
B select Attached to
C select Lines, all
OK
ANSYS graphics show the nodes at the cylinder

Main Menu > General Postproc > Nodal Calcs > Surface Integral
A select Both
OK
ANSYS shows the net pressure in the x- and y-directions, which are
Px = 1.534646078 × 10 –6 N/m2
Py = –6.556592423 × 10 –7 N/m2
Notice that Py has a negative value because the pressure in the y-direction
is downward. The net shear stresses in the x- and y-directions are
τx = 6.823577481 × 10 –7 N/m2
τy = 2.67938876 × 10 –7 N/m2
Fluid mechanics 339
Hence, the net drag force is (Px + τx) × S, and the net lift force is
(Py + τy) × S, where S is the circumference of the half cylinder, and it
is equal to 0.05π. To visualize the flow streamlines, the flow is traced
with particles. First, the trace points will be defined at the entrance and
behind the cylinder to capture the wake flow. Then, the trace particles
are plotted. Finally, an animation is created.
Utility Menu > Select > Everything
Main Menu > General Postproc > Plot Results > Defi Trace Pt
Click on all grid points at the entrance and just one grid point behind
the cylinder, and then in Define Trace Point window, click on
OK
Main Menu > General Postproc > Plot Results > Plot Flow Tra
A select VSUM
OK
ANSYS graphics show the streamlines

Utility Menu > PlotCtrls > Animate > Particle Flow

B type 0.25 in Particle Size factor
C select VSUM in Plot Flow Trace
OK
ANSYS graphics show the animation of the particles flow in the channel.
PROBLEM 6.1
Air at low velocity enters a channel as shown in Figure 6.4. An object
is placed in the channel, at the location shown. The inlet velocity is uni-
form, 0.005 m/s. The exit condition is a reference zero pressure. Let
ρ = 1.25 kg/m3 and μ = 17.7 × 10 –6 Pa · s. Determine the pressure drop
in the channel and net drag force on the object. Also show that the mass
conservation principle is satisfied.
PROBLEM 6.2
Water at low velocity enters a channel as shown in Figure 6.5. A trian-
gular cross-sectional area cylinder is placed in the channel, at the loca-
tion shown. The inlet velocity has a fully developed profile. The exit
condition is a reference zero pressure. The properties of water are
Flow Flow 0.12 m

in out
R0.03 R0.03
0.15 m 0.09 m 0.5 m
FIGURE 6.4 Flow over an object in a channel.

Fluid mechanics 341
y
Flow Flow
x 1.0 cm 4 cm
in out
1.0 cm
4 cm
25 cm
FIGURE 6.5 Channel with a triangular cylinder.
ρ = 998.3 kg/m3 and μ = 1.002 × 10 –3 Pa · s. Use the following equation

for the velocity profile at the inlet:
3  2y 
2
u ( y ) = 0.0001 1 −   
2   H 
where H = 0.04 m. Determine the pressure drop in the channel, and
drag and lift forces on the cylinder, and show that the mass conservation
principle is satisfied.
PROBLEM 6.3
Two cylinders in a wind tunnel, as shown in Figure 6.6, are used to simu-
late heat exchanger tubes. The inlet velocity is 0.0025 m/s, and air is the
working fluid (μ = 20 × 10 –6 Pa · s, ρ = 1.25 kg/m3). The exit condition is
a reference zero pressure. Determine the following:
a. Is mass balance satisfied?
b. What is the maximum shear stress on the first and second
cylinders?
c. What is the average shear stress on the first and second cylinders?
d. What is the pressure drop in the channel?
e. Is the exit flow fully developed?
PROBLEM 6.4
Water at low velocity enters a backward-facing step channel as shown
in Figure 6.7. The height in the channel is increased at a distance 0.5 m
from the entrance as shown in the figure. The inlet velocity has a fully
2 × R 0.05
Flow Flow
0.4 m
in out
0.2 m
0.4 m 0.3 m
2.5 m
FIGURE 6.6 Two cylinders in a wind tunnel.

y
Flow x Flow
in 0.4 m
out
0.1 m
0.5 m
2.0 m
FIGURE 6.7 Flow over a backward-facing step.
developed profile. The exit condition is a reference zero pressure. Water

is the working fluid, and its properties are ρ = 997.1 kg/m3 and μ =
1.003 × 10 –3 Pa · s. Use the following equation for the velocity profile at
the inlet:
3  2y 
2
u ( y ) = 0.001 1 −   
2   H 
where H = 0.15 m. Flow over a backward-facing step generates a recircu-

lation zone due to the separation flow obtained from the adverse pressure
gradients in the fluid flow, and the flow reattached again. Determine the
reattachment length.
Cha p t e r S E V EN
Multiphysics
7.1 Introduction
In practice, the engineering problems to be simulated are m ultiphysics,

and there has not been sufficient capability in the past to address the
full physics of the problems. Therefore, there has been a tendency
to simplify the problems by either focusing on the primary physics
or decoupling the physics. For example, for heat exchanger analysis,
fluid flow is solved first, followed by the thermal solution. Recently,
significant improvements have been achieved in both hardware and
software capabilities that made the multiphysics simulations inexpen-
sive and accurate. As computing power increases, the ability to model
the full physics is becoming a practical possibility. Hence, there is an
increasing demand for multiphysics simulation software. ANSYS has
the capability to model such complicated simulations. ANSYS can
effectively simulate the thermal–structural, thermal–fluid, and fluid–
structural problems.
In this chapter, the setting up of problems involving multiphysics
is introduced. The level of coupling for multiphysics simulations is
illustrated. Examples of thermal–structural and thermal–fluid sys-
tems are presented. The coupling between the physical phenomena
can be classified into three levels: low, medium, and high. With a
low level, it may be sufficient to use a simple one-way coupling with
file transfer between two physics, as analysis is thermal–structural
and thermal–fluid with temperature-dependent properties. In the
t hermal–structural system, the thermal analysis is done first, and then
the temperature distribution is transferred to the structural analysis as
a boundary condition. This analysis is fast, and the thermal and struc-
tural analysis is completely separate, because the structural defor-
mation has a negligible effect on the temperature distribution. The
process is shown in Figure 7.1.
The problem with a medium level of coupling is that the solution
of each physics depends on another but requires no mesh movement,
such as thermal–fluid analysis with temperature-dependent properties.
343
In the thermal–fluid analysis, the properties of the fluid, such as the
density and viscosity, are a function of temperature. In this analysis, fluid
is solved first, followed by thermal solution. The properties of the fluid
are updated, and then the fluid is solved again. This process is continued
until the solution is fully converged. The process is shown in Figure 7.2.
If all properties of the fluid are independent of the temperature, the heat
transfer and fluid flow can be solved separately, and the simulation is
considered as a low-level analysis.
The fluid–structural analysis is considered as a high-level analysis.
It requires a degree of compatibility in the solver technologies and
often involves a mesh movement, which may be accounted for using
the Arbitrary Lagrangian Eulerian method. The computation time for
this analysis is high, and it requires high-performance computers. The
process is shown in Figure 7.3. There are two types of fluid–structural
analyses. In the first type, the solid is moving in a specified way and
the fluid pressure has no effect on the movement of the solid. In the
second type, which is much more complex, the fluid pressure affects
the movement of the structure, and possible structural deformation
occurs.
Boundary
conditions
Thermal Structural
Geometry Thermal Structural
results results
Properties
FIGURE 7.1 Low-level analysis.
Boundary
conditions
Fluid Thermal
Geometry Fluid results Thermal results
Properties
Properties
update
FIGURE 7.2 Medium-level analysis.
Boundary
conditions
Fluid Structural
Geometry Fluid Structural
results results
Properties
Geometry
update
FIGURE 7.3 High-level analysis.

Multiphysics 345
7.2 Thermal and structural analysis of a thermocouple

using ANSYS
Furnaces use thermocouples to control their temperature, and they

come in various designs. The common design of thermocouples is
the two-plate design. It consists of aluminum and carbon steel plates
attached to each other as shown in Figure 7.4. The left end is fixed
and maintained at 100°C, and other surfaces are exposed to a free
convection boundary condition with h = 5 W/m 2 · °C and T∞ = 15°C.
Determine the maximum displacement in the y-direction. The ther-
mophysical properties of the aluminum and carbon steel are shown
in Table 7.1.
This problem is considered as multiphysics, and physics are not
coupled because the structure and heat transfer are solved sepa-
rately. The deformation of the thermocouple has no effect on the
heat transfer. The heat transfer must be solved first, and then the
structure.
A solid thermal element is used, and its shape is quadratic with four
nodes. This element will be replaced by a structural element when the
thermal part is completely solved.
T = 15°C, h = 5 W/m2 . °C
Aluminum
0.005 m
T = 100°C
0.005 m
Carbon steel
T = 15°C, h = 5 W/m2 . °C
0.1 m
FIGURE 7.4 A thermocouple consists of aluminum and carbon steel plates.
Table 7.1 Thermophysical properties of aluminum and

carbon steel
Property Aluminum Carbon steel

Thermal conductivity (W/m · K) 83 111
Young modulus (Pa) 70 × 109 210 × 109
Poisson’s ratio 0.33 0.29
Thermal expansion (1/K) 23 × 10–6 12 × 10–6
A select Structural
B select Thermal
OK
Add...
Multiphysics 347
A select Solid in Thermal Mass

OK
Close
The thermal conductivity is required for the thermal part, and the
modulus of elasticity and Poisson’s ratio are required for the struc-
tural part. Additionally, the thermal expansion of both m aterials is
important to account for the deflection of the thermocouple. There
will be two different sets of properties for aluminum and carbon
steel. The material number 1 in the Define Material Model Behavior
is for aluminum, while material number 2 is for carbon steel.
A type 70e9 in EX
B type 0.33 in PRXY
OK
Click on Structural > Thermal Expansion > Secant Coefficient >
Isotropic
A type 23e-6 in ALPX

OK
Multiphysics 349
Click on Thermal > Conductivity > Isotropic
A type 83 in KXX
OK
In the Define Material Models Behavior menu: Material >

New Model
OK
Select material number 2 in the Material Models Defined, and then
Click on Structural > Linear > Elastic > Isotropic
A type 210e9 in EX
B type 0.29 in PRXY
OK
Click on Structural > Thermal Expansion > Secant Coefficient >

Isotropic
A type 12e-6 in ALPX
OK
Click on Thermal > Conductivity > Isotropic
A type 111 in KXX
OK
Close the Material Models Behavior window

The geometry of the problem is relatively simple. Two rectangles are
created, and then they are glued.
Multiphysics 351
Rectangle by 2 Corners
A type 0 in WP X
B type 0 in WP Y
C type 0.1 in Width
Apply
A type 0 in WP X
B type 0.005 in WP Y
C type 0.1 in Width

OK
Glue > Areas
In Glue Areas window, click on

Pick All
A select Areas in Element Attributes

B click on Set in Element Attributes
Multiphysics 353
select the carbon steel area only, and then in Area Attributes window,
click on
OK

OK
A select Material numbers in Elem/Attrib numbering

B select Colors only in Numbering shown with
OK
A select Smart Size

Mesh
In Mesh Area window, click on
Pick All
Close the Mesh Tool window
Close
Multiphysics 355
Temperature > On Lines
Click on the two left vertical lines where the temperature bound-
ary condition is applied, and then in Apply TEMP on Lines window,
click on
OK
A select TEMP
B type 100 in VALUE Load TEMP value
C select Yes
OK
Click on lines where the convection boundary condition is applied:
the two external horizontal lines and the right two vertical lines. Do not
click on the line between the two materials. Then in Apply CONV on
lines window, click on
OK

B type 15 in Bulk temperature
OK
Now, the problem is ready to be solved as a heat transfer problem.
Close
Multiphysics 357
The thermal solution is now completed. The thermal elements must

be replaced by structural elements with the same element type. This can
be done using the Switch element type in the preprocessor. The tempera-
ture contours are plotted to ensure that the thermal solution is correct
before solving it as a structural problem.
Nodal Solution
A click Nodal Solution > DOF Solution > Nodal Temperature

OK

The temperature contours show the left side is maintained at 100°C,
and the temperature is decreasing in the positive x-direction due to
the convective cooling. In general, the temperature contours are as
expected. Next, the element type is switched to structural type, and
the nodal temperature solution from thermal analysis is loaded to the
structural nodes.
Main Menu > Preprocessor > Element Type > Switch Elem Type
A select Thermal to Struc in Change element type

OK
Close
Main Menu > Solution > Define Loads > Apply > Structure >
Temperature > from Thermal Analysis
A click on Browse …
Select from the thermal results file that has an extension of rth, and
then click on
Open
OK
Multiphysics 359
The structural boundary condition is the zero displacement at the left

side of the thermocouple.
Displacement > On Lines
In the ANSYS graphics, click on the two vertical left lines where
zero displacement is applied. Then, in Apply U,ROT on Lines window,
click on
OK
A select All DOF

B type 0 in VALUE Displacement value
OK
OK
Close

OK
ANSYS graphics show the thermocouple before and after deformation
The plot shows downward deflection of the thermocouple because the

thermal expansion of the aluminum is higher than that of the carbon steel.
Nodal Solution
Multiphysics 361

displacement
OK
ANSYS graphics show the displacement in the y-direction

The ANSYS results indicate that the maximum displacement in the
y-direction is equal to 0.755 × 10 –3 m.
7.3 Chips cooling in a channel using ANSYS
Study the thermal characteristics of two electronic chips mounted on

a channel’s bottom wall using ANSYS. The configuration is shown in
Figure 7.5. The working fluid is air, and thermophysical properties of
the chips are listed in Table 7.2. The inlet velocity is 0.01 m/s, while the
exit condition is a reference zero pressure. The inlet temperature is 25°C,
and 5 W are generated in each chip. Consider the problem as a steady
fluid flow and heat transfer. Determine the following:
1. Maximum velocity in the x-direction, and temperature in the domain
2. Satisfaction in mass and energy balances
3. Heat transfer coefficient distribution at the surface of the first chip
Electronic manufacturers are continuously providing the market with
high-performance devices, but with high heat dissipation. Heat highly
affects the performance and durability of electronic devices more than
any other factor, and operating electronic devices at a temperature higher
Flow Chips Flow

in out 0.075 m
0.025 m
0.1 m 0.05 m 0.05 m 0.35 m
FIGURE 7.5 Channel with two chips.

Table 7.2 Thermophysical properties of

the air and electronic chips
Property Air Chips

Density (kg/m3) 1.25 2500
Conductivity (W/m · °C) 0.025 0.25
Specific heat (J/kg · °C) 1006.2 850
Viscosity (N · s/m2) 18.5 × 10–6 —
than its recommended operating value will significantly affect their

reliability and functionality. Thermal analysis of a channel containing
multiple heated blocks subjected to forced convection flow is extensively
addressed in the literature because it simulates integrated circuit chips
placed on a horizontal board. In this analysis, researchers are focusing
on the temperature distribution within the chips, the maximum tempera-
ture of the chips, and distribution of the local Nusselt number along the
chips’ surface. Temperature distribution within the chips is typically
used to predict the reliability of some of its components, such as the
solder joints, and to establish a guide for safe operating conditions. The
Nusselt number is used to estimate the heat flow out of the chips and to
determine the required cooling load. In addition, researchers are inter-
ested in the friction factors on the surface of the chips, and the pressure
drop in the channel, which are used to measure the required fan power.
This problem is considered as multiphysics and physics are not
coupled because the fluid flow and heat transfer are solved separately.
ANSYS will solve the fluid flow first, and then it uses the velocity com-
ponents to solve for the temperature field. ANSYS Flotran CFD can solve
fluid dynamics problems, as well as the thermal problems. The selected
FLOTRAN 141 element has temperature as a degree of freedom.

OK
Multiphysics 363
Add...

OK
Close
There will be different sets of properties for air and chips. The
aterial number 1 in the Material Model Behavior is for the fluid only,
m
and material numbers 2 to 10 are for the solids.
A click on CFD > Conductivity > Density
A type 0.025 in KXX

OK
Click on CFD > Specific Heat
A type 1006.2 in C
OK
Multiphysics 365
Click on CFD > Density
A type 1.25 in DENS

OK

OK
In the Define Material Model Behavior menu: Material > New Model
OK

B click on CFD > Conductivity > Viscosity
A type 0.25 in KXX

OK
Click on CFD > Specific Heat
A type 850 in C
OK
Multiphysics 367
Click on CFD > Density
A type 2500 in DENS

OK

The geometry is created by first creating a large rectangular area for
fluid flow. Then, the two chips are created. Overlap is utilized to insert
the chips into the fluid rectangular area.
Rectangle by 2 Corners
A type 0 in WP X
B type 0 in WP Y
C type 0.6 in Width
Apply
A type 0.1 in WP X
B type 0 in WP Y
Apply
A type 0.2 in WP X
B type 0 in WP Y
OK
Multiphysics 369
Overlap > Areas
Pick All
The following step is for changing the material properties of the chips
from number 1 to 2. By selecting number 2, the properties of number 2
in the material model are assigned to the chips. Air by default has the
properties of number 1 in the material model.
A select Areas
B click on Set
Select both chips, and then in Area Attributes window, click on
OK
OK
To ensure that the properties of air and chips are assigned correctly,
the air and chips are colored according to their material number in the
material model. This step has no effect on the solution.
A select Material numbers in Elem/Attrib numbering

B select Colors only in Numbering shown with
OK
Multiphysics 371
The smart mesh number 1 has insufficient mesh density to have

a ccurate results for fluid dynamics problems. The number of elements in
the domain is increased by using the area size control in the Mesh Tool.
A select Smart Size

C Mesh
In Mesh Area window, click on
Pick All
D select Areas in Refine at

E click on Refine
In Refine Mesh at Area window, click on
Pick All
OK

Using refine at elements instead of refine at areas will generate a
similar mesh. Notice that the ANSYS properties of the material model
behavior will be used for air and chips.
Close the Mesh Tool Window

Temperature> On Lines
Click at the inlet line of the channel to specify the inlet temperature,
and in Apply TEMP on Lines window, click on
OK
Multiphysics 373
A select TEMP
B type 25 in VALUE Load TEMP value
C select Yes
OK
The heat generation must be per unit volume. The applied heat genera-
tion is divided by the area of the chip because the problem is two dimen-
sional. The chip’s volumetric heat generation is calculated as follows:
5
Q= = 4000 W/m 2
0.05 × 0.025
Main Menu > Solution > Define Loads > Apply > Thermal > Heat
Generat > On Areas
Click on both chips. Then, in Apply HGEN on ARs window, click on
OK
A type 4000 in Load HGEN value
OK
The top and bottom boundaries are wall and insulated. Zero x- and
y-velocities should be imposed. There is no need to impose a velocity bound-
ary at the bottom surface of the chips. By default, any unassigned thermal
boundary conditions will be considered as insulated. Do not impose any
boundary conditions on the surface of the chips that are exposed to the flow.
Velocity > On Lines
Click on the inlet of the channel. Then in Apply VELO load on lines
window, click on
OK
A type 0.01 in VX Load value

OK
Velocity > On Lines
Click on the top and bottom walls. Then, in Apply VELO load on lines
window, click on
OK

OK
Multiphysics 375
Pressure DOF > On Lines
Click on channel’s exit. Then, in Apply PRES on lines window, click on
OK
OK
The fluid flow and heat transfer are solved at the same time. In
s olution options, the problem will be solved as steady state, and the sys-
tem is thermal. Flow can be changed from laminar to turbulent and from
incompressible to compressible in this window.
Main Menu > Solution > FLOTRAN Set Up > Solution Options
A change to thermal in TEMP Adiabatic or thermal?

OK
In execution control, it is required to specify the number of iterations

for fluid flow. Then 1000 iterations will be sufficient to reach the conver-
gence criteria for all field variables, which is 1 × 10 –6 (1e-6).
A type 1000 in EXEC Global iterations

B change the termination criteria for VX Velocity component to 1e-6
C change the termination criteria for VY Velocity component to 1e-6
D change the termination criteria for Pressure to 1e-6
E change the termination criteria for Temperature to 1e-6
OK
Multiphysics 377

C select MP table in Conductivity
D select MP table in Specific heat
OK
OK
For the heat transfer coefficient calculation, the flow bulk temperature
is required. In the reference condition, the bulk temperature is specified,
which is equal to the flow inlet temperature.
Main Menu > Solution > FLOTRAN Set Up > Flow Environment >
Ref Conditions
A type 25 in Reference (bulk) temp

OK
ANSYS graphics show a normalized rate of change for field variables
Close
ANSYS graphics show the normalized rate of change of the field
variables. Note the solution is fully converged. All field variables reach
1 × 10 –6 at the iteration number 185, which is less than the specified 1000
iterations. In read results, the last set must be read to have the results
from the last iteration.
Nodal Solution
A click on Nodal Solution > DOF Solution > X-Component of fluid

velocity
OK
Multiphysics 379
ANSYS graphics show the x-component velocity contours
For the obtained results, the maximum and average velocity in the
x-direction occurred above the chips, and the maximum velocity is
0.021396 m/s. This is expected because this region has the smallest
cross-sectional area that forces the flow to accelerate. At the walls, the
velocity is 0. Notice that the velocity has a parabolic profile at the exit
of the channel.
Nodal Solution

OK

As shown in the temperature contours, the maximum temperature is
73.2043°C, and this occurred at the second chip. To ensure that mass and
energy balances are satisfied, the average velocity in the x-direction and
temperature at the exit are determined using the path operation in the
postprocessor. The path at the exit is created by specifying two points.
A select Grid Only

OK
Multiphysics 381

Utility Menu > PlotCtrls > Pan-Zoom-Rotate …
Click on zoom in and out until the ANSYS graphics show all grids.
On Working Plane
OK
Click on the ANSYS Graphics window at the exit’s top and bottom
corners of the channel, and then in On Working Plane window, click on
OK
A type Exit in Define Path Name; the name of the path is optional
OK
Main Menu > General Postproc > Path Operations > Map onto Path
Apply
A select Temperature TEMP
OK
Plot Path Item > On Graph
A select VX
OK
Multiphysics 383
ANSYS graphics show the velocity profile at the exit of the channel
Main Menu > General Postproc > Path Operation >

Plot Path Items > On Graph
A select TEMP
OK
ANSYS graphics show the temperature profile at the exit of the channel

The average velocity and temperature at the exit are determined

using the integration in the path operation. The value of the integration
must be divided by the path length to get the average value of the vari-
able. The number 13.333 in the Factor is the inverse of the path length.
Selecting S in the Lab2 means that the integration is performed along
the path.
Main Menu > General Postproc > Path Operations > Integrate
A type 13.333 in Factor

B select VX in 1st Path item
Apply
A type 13.333 in Factor

B select TEMP in 1st Path item
OK
Multiphysics 385
The ANSYS Output window shows the average velocity and tempera-
ture at the exit, which are 0.95993 × 10 –2 m/s and 36.052°C, respectively.
The mass flow rate at the exit and inlet should be the same as follows:
i =m
m e
or
ρ V A)i = ( ρ V A)e
(
Since the density and cross-sectional area of the inlet and exit are the
same, then
Vi = Ve
Comparing the inlet velocity, which is 0.01 m/s, to the exit velocity,
which is 0.0095993 m/s, the two velocities are close to each other with
an error of 4%. The heat gained by air in the channel can be determined
using the first law of thermodynamics:

Q = m Cp ( Te − Ti )
or
Q = ( ρ V A) Cp( Te – Ti )
Q = (1.25 × 0.01 × 0.075) × 1006.2 × (36.052 − 25) = 10.425 Watts
The heat gained Q should be equal to the generated heat in the two
chips. Comparing the heat gain to the generated heat in the two chips,
which is 10 Watts, the energy balance is satisfied with an error of 4.25%.
The error found in mass and energy balances can be reduced if the
computational domain is meshed with a finer mesh. The path opera-
tion is performed in the following steps to obtain the distribution of the
average heat transfer coefficient around the first chip. Here, the film
coefficient is selected.

On Working Plane
OK
Click on the ANSYS Graphics window at locations A, B, C, D, and E,
as shown in the following figure, and then in On Working Plane window,
click on
OK
A type Chip1 in Define Path Name; the name of the path is optional
OK
Multiphysics 387
Main Menu > General Postproc > Path Operations > Map onto Path
A select Other quantities

B select Film coeff HFLM
OK
Now, the stored variable, HFLM, is ready to be plotted. In the Plot path
item, there are two options. The stored data can be either plotted or listed.
The list results can be exported into other software such as EXCEL.

Plot of Path Item > On Graph
A select HFLM
OK
ANSYS graphics show the heat transfer coefficient around the first chip

The average heat transfer coefficient at the chip surface can be

d etermined using the integration in the path operation. The value of
the integration must be divided by the path length to get the average
value of the variable. Selecting S in the Lab2 means that the integration
is performed along the path.
A type 10 in Factor
B select HFLM in 1st Path item
OK
The ANSYS Output window shows the value of the average heat
transfer coefficient around the first chip, which is 1.642 W/m2 · °C.
PROBLEM 7.1
The thermocouple shown in Figure 7.6 consists of copper and carbon
steel plates attached to each other. The left end is fixed and maintained
at 200°C, and the thermocouple is exposed to a free convection
boundary condition with h = 15 W/m2 · °C and T∞ = 20°C. Display the
y-displacement contours and determine the maximum displacement in
Multiphysics 389
T = 20°C, h = 15 W/m2 . °C
Copper
0.0005 m
T = 200°C 0.001 m
Carbon steel
T = 20°C, h = 15 W/m2 . °C
0.05 m
FIGURE 7.6 A thermocouple consists of copper and carbon steel plates.
the y-direction. The thermophysical properties of the used materials are

shown Table 7.3.
PROBLEM 7.2
A pipeline used for transporting highly corrosive liquid is made of con-
centric carbon steel and aluminum pipes as shown in Figure 7.7. The inner
pipe is made of aluminum to prevent corrosion, while the outer pipe is
made of carbon steel to add strength to the pipeline in Table 7.4. At the
inner surface of the pipe, the temperature is 100°C, while on the outer sur-
face the temperature is 25°C. Due to thermal expansion variation between
the carbon steel and aluminum, a thermal stress is developed at the contact
surface that could damage the pipeline. Determine the maximum stress in
the pipe. The displacement at inner surface of the pipe is fixed.
Table 7.3 Thermophysical properties of the used materials
Property Copper Carbon steel

Thermal conductivity (W/m · K) 401 111
Young’s modulus (GPa) 140 210
Thermal expansion (1/K) 16.5 × 10–6 12 × 10–6
Aluminum
Carbon steel
8.0 cm
10.0 cm FIGURE 7.7 A composite
11.0 cm
pipeline.
Table 7.4 Thermophysical properties of aluminium and

carbon steel
Property Aluminum Carbon steel

Thermal conductivity (W/m · °C) 83 111
Young’s modulus (Pa) 70 × 109 210 × 109
Thermal expansion (1/K) 23 × 10–6 12 × 10–6
PROBLEM 7.3
Study the thermal characteristics of an electronic chip mounted on a
channel using ANSYS. The configuration is shown in Figure 7.8. The
working fluid is air, and thermophysical properties of the chip and air
are listed in Table 7.5. The inlet velocity is 0.0125 m/s, while the exit
condition is a reference zero pressure. The inlet temperature is 20°C, and
5 Watts are generated in a heat source. Consider the problem as a steady-
state heat transfer. Determine the following:
a. Maximum velocity in the x-direction and temperature in the
domain
b. Satisfaction of the mass and energy balances
c. Heat transfer coefficient distribution at the surface of the chip.
PROBLEM 7.4
The two cylinders in a wind tunnel, as shown in Figure 7.9, are used
to simulate heat exchanger tubes. The inlet velocity and tempera-
ture are 0.002 m/s and 25°C, respectively. Air is the working fluid
(μ = 20 × 10 –6 Pa · s, Cp = 1000 J/kg · °C, k = 0.025 W/m · °C, and
ρ = 1.25 kg/m3). The surfaces of the cylinders are maintained at 50°C.
The exit condition is a reference zero pressure.
0.0175 m
Flow Heat source Flow
Chips 0.045 m
in out
0.005 m 0.015 m
0.05 m 0.025 m 0.25 m
FIGURE 7.8 A channel with a chip.
Table 7.5 Thermophysical properties of

the chip and air
Property Air Chip

Density (kg/m3) 1.25 2250
Conductivity (W/m · °C) 0.025 0.4
Specific heat (J/kg · °C) 1005 950
Viscosity (N s/m2) 20 × 10–6 —
Multiphysics 391
2 × R 0.05
Flow Flow
0.4 m
in out
0.2 m
0.4 m 0.3 m
2.5 m
FIGURE 7.9 Two cylinders in a wind tunnel.
a. Show that the mass balance is satisfied.

b. Ensure maximum shear stress on the first and second cylinders.
c. Find the average heat transfer coefficient on the first and second
cylinders.
d. Determine pressure drop in the channel.
PROBLEM 7.5
Water at low velocity enters a channel as shown in Figure 7.10. A trian-
gular cross-sectional area cylinder is placed in the channel, at the shown
location. The inlet velocity has a fully developed profile, and its tem-
perature is uniform at 25°C. The exit condition is a reference zero pres-
sure. The properties of water are ρ = 997.1 kg/m3, Cp = 4180 J/kg °C,
k = 0.6076 W/m · °C, and μ = 0.891 × 10 –3 Pa · s. Use the following
equation for the velocity profile at the inlet:
3  2y 
2
u ( y ) = 0.0001 1 −   
2   H 
where H = 0.04. Ten Watts are generated in the triangular cylinder, which
has properties of ρ = 4600 kg/m3, Cp = 1055 J/kg · °C, and k = 76.5 W/m · °C.
a. Show that the mass balance is satisfied.
b. Show that the energy balance is satisfied.
c. Determine the maximum shear stress at the surface of the triangular
cylinder.
d. Determine the average heat transfer coefficient in a triangular
cylinder.
e. Determine pressure drop in the channel.
Flow x 1.0 cm
Flow 4 cm
in out
1.0 cm
4 cm
25 cm
FIGURE 7.10 Channel with a triangular object.

Cha p t e r E IGHT
Meshing guide
8.1 Mesh refinement
The accuracy of the finite element model can be enhanced by increasing

either the number or the order of the elements. An increasing number of
elements is called h-refinement, while increasing the order of the elements
is called p-refinement. Increasing the number of elements or their order
will lead to a significant increase in the computational time and required
memory to solve the problem. Consider the stress analysis of a plate shown
in Figure 8.1a. A linear quadratic element is used to mesh the domain.
The stress concentration factor is required, and the exact solution is avail-
able from theory of elasticity. Figure 8.1b shows the finite element mesh.
The mesh consists of 55 elements. To study the effect of the number of
elements on the solution, the number of elements is increased and the
stress concentration is obtained for the corresponding elements number.
Figure 8.2 shows the error of the stress concentration for different num-
bers of elements using linear quadratic elements. The domain is meshed
with three different meshes, A, B, and C. The number of elements of mesh C
is higher than that of mesh B, and the number of elements of mesh B is
higher than that of mesh A. The figure indicates that as the number of ele-
ments increases, the relative error is decreased to approach a fixed value.
Adding more elements to the mesh C will insignificantly reduce the error.
When the number of elements has no effect on solution, the solu-
tion is called a mesh-independent solution. On the other hand, increas-
ing the order of the elements reduces the error for the same mesh size.
Figure 8.3 indicates that replacing the linear quadratic elements with
cubical ones reduces the percentage of the error by about 10%.
With high-order elements, the mesh-independent process is faster
than with low-order elements, as shown in Figure 8.4. In addition, a
mesh with the linear quadratic elements is becoming mesh independent
at a very slow rate.
The higher order elements lead to more accurate results than lower
order elements. However, the time required to complete the solution
is important, especially for very large meshes. In addition to time,
the higher order elements need a large amount of memory that could not
393
(a) (b)
FIGURE 8.1 (a) Geometry of a plate and (b) finite element mesh.

Error (%)
A B C
Number of elements
FIGURE 8.2 Error of the stress concentration for different mesh sizes using linear quadratic elements.
20
Error (%)
15
10
FIGURE 8.3 Percentage of error with different element orders.
be afforded. Figure 8.5 indicates that in spite of a low mesh convergence

rate of linear quadratic elements, the time required to solve the problem
is nearly independent of the number of elements in the mesh. In addi-
tion, increasing the order of the elements significantly increases the time
Meshing guide 395
Error (%)
A B C
Number of elements
FIGURE 8.4 Error of the stress concentration for different numbers of e lements and orders.
Normalized computational time
A B C
Number of elements
FIGURE 8.5 Normalized computational time for different quadrilateral elements.
8.2 Element distortion
When a complex geometry is meshed using an automatic mesh gen-

erator, elements at some critical regions in the geometry are distorted
and the accuracy of the results is significantly decreased at these
regions. Hence, this section concerns the element distortions and
identifies these elements. In addition, guidelines about good element
shapes are also introduced. In general, there are four types of ele-
ments distortion:
1. Mid-side node off-center

2. Taper
3. Skew
4. Aspect ratio
The mid-side node off-center type is when side nodes are shifted
from the center of the side. As the node moves away from the center,
the errors increase. In practice, the mid-side node should not be shifted

more than one-eighth of the side’s length. Rectangular elements can be
skewed or tapered. The angles of the rectangular elements should be 90°.
For a rectangular element, and when the height-to-width ratio exceeds 2,
the element has an aspect-ratio problem. It turns out that the aspect ratio
for a perfectly shaped rectangular element is 1 and 1.15 for an equilateral
triangle element. Aspect ratios higher than the recommended value can
only occur in noncritical regions of the model, such as in uniform temper-
ature or stress regions. Figure 8.6 shows the major types of element dis-
tortions. Element distortions are not limited to two-dimensional spaces,
they can also occur in three-dimensional spaces. The most common one
is the warping. Warping occurs when the face of a solid element does not
lie in the same plane, and the warping angle should not be more than 10°.
8.3 Mapped mesh
Meshes can be classified as free or mapped meshes. The elements of the

free meshes are randomly distributed in the domain with different sizes
and shapes, and element distortions are most likely to appear in these
meshes. Figure 8.7 shows a free mesh of a corrugated channel used for
fluid flow analysis. Most of the finite element software generates this
type of mesh. Poor results are always associated with free meshes. The
free meshes should be only limited to very complex geometries.
Mapped meshes have organized distributed elements, and the meshes
are almost free of the element distortions. Figure 8.8 shows a mapped
mesh of a corrugated channel. The mesh designers can fully control the
(a)
(b)
(c)
FIGURE 8.6 Major types of element distortions: (a) mid-side node off-center, (b) taper and skew,
and (c) aspect ratio.
Meshing guide 397
FIGURE 8.7 Free mesh of a corrugated channel.
FIGURE 8.8 Mapped mesh of a corrugated channel.
size of the elements and their distribution in the computational domain.

These meshes are typically used in finite element simulations because of
their flexibility and ability to mesh complex geometries. Mesh designers
should have a strong understanding of the physics of the problem before
generating the mapped meshes. For example, for solid mechanics prob-
lems, the areas with high and uniform stress gradients should be identi-
fied, and the mesh should be designed accordingly.
For fluid problems, the regions close to the walls have very high veloc-
ity gradients due to shear forces, and elements should be dense at the
region close to the walls. The geometry shown in Figure 8.9 is fin, and
a thermal analysis is required. High temperature gradients at the upper
region are expected. Therefore, elements must be concentrated at this
region. The geometry is meshed with three different mapped meshes.
The mesh in Figure 8.9a is unacceptable because the elements are not
concentrated in the upper region, and poor results are expected. In the
second mesh (Figure 8.9b), the upper region of the geometry has more
(a) (b) (c)
FIGURE 8.9 A fin (a) without elements concentration, (b) with only elements concentration at the
upper region, and (c) with elements concentration at the upper region with gradual changes.
elements than the first mesh, but an abrupt change in element size should
be avoided, because errors in the large elements will be transferred to
the small elements. In addition, performing matrix operations of these
meshes is difficult. The third mesh, Figure 8.9c, is the ideal mesh.
Elements are concentrated at the upper region, and the change from the
large elements in the lower region to small elements in the upper region
is gradual.
8.4 Mapped mesh with ANSYS
The square plate with a hole shown in Figure 8.10 is subjected to tensile

pressure at both vertical sides. Use ANSYS to create a mapped mesh of
the plate and concentrate elements at the region close to the hole where
high stress gradients are expected. Use quadratic elements with four
nodes to mesh the plate.
A select Structural
OK
ϕ 1.25 cm
P P
FIGURE 8.10 Square plate with

5 cm
a hole.
Meshing guide 399
Add...
A select Solid
OK
Close
The geometry is modeled by creating a square and a circle. A Boolean
operation is utilized to remove the circle from the square using subtrac-
tion. The advantage of symmetry in the problem is considered. Only
the upper right quarter is considered. To create a mapped mesh with
ANSYS, the computational domain must be divided into areas, and each
area must have four sides only.
A type 0 in WP X
B type 0 in WP Y
OK
Circle > Solid Circle
Meshing guide 401
A type 0.0 in WP X
B type 0.0 in WP Y
OK
Main Menu > Preprocessor > Modeling > Operate >

Booleans > Subtract > Areas
Click on square area, and then in Subtract Areas window, click on
Apply
Click on circular area, and then in Subtract Areas window, click on
OK
ANSYS graphics show the lines of the geometry

As shown in the figure, the geometry is one area which is composed
of five lines. Therefore, the mapped mesh cannot be created. The area
should be divided into areas that each should be composed of four sides,
and there are many options. As a suggestion for this example, the area is
divided into two areas along the symmetry line of the geometry. First,
a keypoint is created at the origin. Second, a line is created between the
keypoints at the origin and upper right corner of the plate. Third, the
area is divided by a line, which has just been created. The new Keypoint
will have number 10, and this number is selected to avoid a conflict with
existing keypoints.

Keypoints > In Active CS
A type 10 in NPT Keypoint number

OK
Lines > Lines > Straight Line
Click on keypoints at the origin, and at the origin and upper right
corner of the geometry. Then, in Create Straight Line window, click on
OK
ANSYS graphics show the created line

ANSYS Main Menu > Preprocessor > Modeling >
Operate > Booleans > Divide > Area by Line
Click on area to select it. Then, in Divide Area by Line window,
click on
Apply
Click on the just created line to select it. Then, in Divide Area by Line
window, click on
OK
Meshing guide 403

Now the geometry is ready to be meshed. It consists of two areas,
and each area is composed of four lines. Elements should be concen-
trated at the region close to the hole since high stress gradients are
expected at this region. The lines should be divided into segments, and
the two opposite lines of an area must have the same number of division.

The lines indicated in the following figure should have the same line
division and spacing ratio.
Select the lines as shown in the figure. Then in Element Sizes on

Picked Lines window, click on
Apply
A type 25 in the NDIV
The number 25 means that the selected lines will be divided into
25 segments, and each segment will be one element.
Apply
Meshing guide 405
Select the lines, as shown in the following figure. Then in Element

Sizes on Pick Lines window, click on
OK
A type 50 in NDIV
B type 10 in SPACE
The number 50 means that the selected lines will be divided into
50 segments, and the segments are gradually increasing by a spacing
ratio of 10.
OK
ANSYS graphics show the lines division

Notice that the division of the horizontal and vertical lines is wrongly
implemented. The segment lengths are increasing toward the hole, and
they should be reversed. The reversion can be easily done using flip in
Mesh Tool as follows:
A click on Flip in Lines

Select the horizontal and vertical lines, which have a wrong division.
Then in Flip Line Bias window, click on
OK
Meshing guide 407
ANSYS graphics show the lines division
A select Mapped
Mesh
Pick All
ANSYS graphics show a mapped mesh
PROBLEM 8.1
Create a mapped mesh for the geometries shown in Figure 8.11, and the
geometries are used for solid mechanics.
F R 1.0 cm
1m P P 3.5 cm
2m 3.5 cm 4.0 cm 3.5 cm

(a) (b)
R 0.01 m
R 0.01 m
(c)
FIGURE 8.11 Geometries for solid mechanics: (a) a rectangular plate subjected to vertical and hori-
zontal forces, (b) a notched rectangular plate subjected to a tensile pressure, and (c) two horizontal
cylinders on a flat plate with a force.
Meshing guide 409
PROBLEM 8.2
geometries are used for heat transfer.
PROBLEM 8.3
geometries are used for fluid mechanics.
Insulated
T∞
To
h
2m
(a)
Convection
Fin
Convection
6.0 cm
Convection
Chip
3.0 cm
1.0 cm
Heat
source
5.0 cm Insulation
9.0 cm
(b)
0.005 m 0.005 m
Fin
0.045 m
Processor
Chip
Board
0.01 m
0.015 m
0.01 m
0.005 m
0.015 m 0.065 m 0.035 m 0.04 m 0.015 m
(c)
FIGURE 8.12 Geometries for heat transfer: (a) two-dimensional body subjected to fixed t emperature
and convection boundary conditions, (b) a chip and fin with heat generation in the heat source, and
(c) electronic board assembly.
Flow
in Flow
0.4 m
out
0.1 m
0.5 m
2.0 m
(a)
y
Flow Flow
x 0.2 m
in out
0.15 m 0.1 m 0.75 m
(b)
y
Flow x Flow
1.0 cm 4 cm
in out
1.0 cm
4 cm
25 cm
(c)
FIGURE 8.13 Geometries for fluid mechanics: (a) flow over a backward-facing step, (b) flow over
a half cylinder in a channel, and (c) flow over a triangular cylinder in a channel.
Bibliography
V. A da m s and A . A s k e na z i. Building Better Product with Finite Element

Analysis. ONWORD Press, Santa Fe, NM.
A nsys Th e ory. ANSYS online manual. ANSYS, Canonsburg, PA.
T. C h a n d ru pat l a and A . Be l e gu n du. Introduction to Finite Elements
in Engineering. 3rd edition. Prentice Hall, Upper Saddle River, NJ.
M. Faga n. Finite Element Analysis: Theory and Practice. Pearson Education,
Harlow, Essex, England.
S. M oav e n i. Finite Element Analysis: Theory and Application with ANSYS.
Prentice Hall, Upper Saddle River, NJ.
J. R e ddy. An Introduction to the Finite Element Method. 2nd edition.
McGraw-Hill, New York, NY.
E . Thom p s on. Introduction to the Finite Element Method: Theory,
Programming, and Applications. Wiley, Hoboken, NJ.
T. Ya ng. Finite Element Structural Analysis. Prentice Hall, Englewood
Cliffs, NJ.
411
ENGINEERING – GENERAL
“A must read for those interested in exploring the possibilities of using the finite element
method (FEM) as a practical analysis tool for structural mechanics, stress analysis,
Finite Element
Simulations
vibration, heat transfer, and fluid dynamic problems. … Alawadhi systematically introduces
the theory of FEA and demonstrates a step-by-step procedure for practical analysis using
ANSYS. This book is a must-have reference for students, academics, and practitioners
in the field of mechanical, civil, environmental, and design engineering.”
Using ANSYS
—Dr. Arun Arjunan, University of Wolverhampton, UK
Uses a Step-By-Step Technique Directed with Guided Problems

and Relevant Screen Shots
Simulation use is on the rise, and more practicing professionals are depending on the
Second Edition
reliability of software to help them tackle real-world mechanical engineering problems.
Finite Element Simulations Using ANSYS, Second Edition offers a basic understanding
of the principles of simulation in conjunction with the application of ANSYS. Employing
a step-by-step process, the book presents practical end-of-chapter problems that are
solved using ANSYS and explains the physics behind them. The book examines structure,
solid mechanics, vibration, heat transfer, and fluid dynamics. Each topic is treated in a
way that allows for the independent study of a single subject or related chapter.
What’s New in the Second Edition:
• Introduces the newest methods in modeling and meshing for

finite element analysis
• Modifies ANSYS examples to comply with the newest version of ANSYS
• Replaces many ANSYS examples used in the first edition with more
general, comprehensive, and easy-to-follow examples
• Adds more details to the theoretical material on the finite element
• Provides increased coverage of finite element analysis for
heat transfer topics
• Presents open-ended, end-of-chapter problems tailored to serve as
class projects
Finite Element Simulations Using ANSYS, Second Edition functions as a fundamental

reference for finite element analysis with ANSYS methods and procedures, as well as a
guide for project and product analysis and design.
K24312
ISBN: 978-1-4822-6197-4
Esam M. Alawadhi
90000
9 781482 261974
K24312_Cover_PubGr.indd All Pages 8/10/15 11:35 AM

Finit Element Simulation Using ANSYS

Uploaded by

Copyright:

Available Formats

Finit Element Simulation Using ANSYS

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Finit Element Simulation Using ANSYS

Uploaded by

Copyright:

Available Formats

What is the book about?

What is the book about?

What are the main topics covered in the book?

What are the main topics covered in the book?

ENGINEERING – GENERAL

Uses a Step-By-Step Technique Directed with Guided Problems

What’s New in the Second Edition:

• Introduces the newest methods in modeling and meshing for

Finite Element Simulations Using ANSYS, Second Edition functions as a fundamental

K24312_Cover_PubGr.indd All Pages 8/10/15 11:35 AM

No claim to original U.S. Government works

International Standard Book Number-13: 978-1-4822-6198-1 (eBook - PDF)

1 Introduction to finite element method and ANSYS 1

1.1 The finite element method and structural analysis 1

2 Bar and beam structures 19

2.1 Finite element method for a horizontal bar element 19

3.1 Stress–strain relations 95

4.1 Vibration analysis 171

5 Heat transfer 225

5.1 Introduction to heat conduction 225

6 Fluid mechanics 293

6.1 Governing equations for fluid mechanics 293

7.1 Introduction 343

8 Meshing guide 393

8.1 Mesh refinement 393

Due to the complexity of modern-day problems in mechanical e­ ngineering,

I profited greatly from discussion with faculty members and engineers at

Esam M. Alawadhi is a professor of mechanical engineering at Kuwait

Introduction to finite element

1.1 The finite element method and structural analysis

[K]{U} = {F} (1.1)

FIGURE 1.1 (a) A section of a bridge and (b) finite element mesh.

Consider a one-dimensional bar with initial length L that is subjected to

The displacement at nodes i and j are ui and uj, respectively. Then

ui(x) = a + bxi (1.3)

uj(x) = a + bxj (1.4)

where xi is the x-coordinate for node i, and xj is the x-coordinate

a = (uixj − ujxi)/L (1.5)

b = (uj − ui) / L (1.6)

u(x) = Niui − Njuj (1.8)

Since the strain is related to the stress by the modulus of elasticity

The strain is equal to the elongation of the element in the x-direction

ε = (uj − ui) / L (1.12)

Then, the strain energy in Equation 1.11 becomes

The strain energy can be expressed in the matrix form as follows:

w = uiFi + ujFj = {U}T{F} (1.15)

For the minimum potential energy, the displacement must be equal

Equation 1.16 is differentiated using the expressions (1.18), and the

[K]{U} − {F} = 0 (1.22)

Minimizing Equation 1.24, the result is

∑ K  {U} − {F} = 0 (1.25)

1.2 Stress analysis of pin-jointed bar

Use the finite element method to calculate the maximum displacement

Step 1: Construct the stiffness matrix for all bar elements

30.00 mm 55.00 mm 45.00 mm

Element (1) Element (2) Element (3)

Node (1) Node (2) Node (3) Node (4)

Assembling the elements’ stiffness matrices to form the global stiff-

Step 2: Apply the boundary conditions to the global stiffness matrix

Solve for displacements:

Due to the complexity of modern-day problems in mechanical e ngineering,

Double click on the Mechanical APDL Product Launcher icon in

2.2 Finite element method for an arbitrary-oriented