Vous êtes sur la page 1sur 171

A FINITE ELEMENT METHOD PRIMER

FOR MECHANICAL DESIGN

Charles E. Knight, Jr.


Department of Mechanical Engineering
Virginia Polytechnic Institute
and State University
Copyright © 1994-2000 by Charles Knight

All rights reserved. No part of this book may be reproduced, stored in a retrieval
system, or transcribed in any form or by any means--electronic, mechanical,
photocopying, recording, or otherwise--without the prior written permission of
Charles Knight.

Library of Congress Cataloging-in-Publication Data

Knight, Charles E.
A finite element method primer for mechanical design / Charles E.
Knight, Jr.
p. cm.
Includes bibliographical references and index.
ISBN 0-534-93978-3
1. Finite element method. 2. Engineering design--Data processing.
3. Computer-aided design. I. Title.
TA347.F5K66 1993 93-27512
620r.0042r0151535--dc20 CIP

Printed and bound in the United States of America.


93 94 95 96 97 98 -- 10 9 8 7 6 5 4 3 2 1
This book is dedicated to my son and daughter, Michael and Laura
PREFACE

A Finite Element Method Primer for Mechanical Design was written to be


a supplementary text for the junior mechanical design course in a
traditional mechanical engineering curriculum. The intent is to introduce
the finite element method in the context of mechanical design. Emphasis
is on the practical aspects of proper modeling, checking, and interpretation
of results. Theoretical aspects are introduced as they are needed to help
understand the operation. The text covers truss, beam/frame, and two
dimensional solid engineering structures. This text is basically an excerpt
from The Finite Element Method in Mechanical Design.
The finite element method has rapidly become a vital tool for analysis of
mechanical designs. It has reached the point where practically every
design engineer has access to a finite element program through their
company's mainframe computer or on a micro-computer. This method is
extremely powerful in terms of the many different types of problems it can
solve. The method applies to many engineering fields, however this book
concentrates on the application to mechanical design.
Use of this tool does not guarantee correct results. It is a numerical
procedure involving approximations of theoretical behavior. In order to
produce correct results the model must be designed correctly, and be able
to reach numerical convergence (assuming the computer program is
without error). Correct results then depend primarily on the user's ability
to utilize the tool. It in no way supplants the engineer's responsibility to
do approximate engineering calculations, use good design practice, and
apply engineering judgment to the problem. Instead, it should supplement
these skills to ensure that the best design is obtained.
With the ready access of finite element analysis comes the need to
provide the understanding required to accurately and effectively use the
method. Most design engineers are not going to have the time nor the
inclination to study all the theoretical formulations and computer
algorithms much less write their own program. Therefore, most engineers
take the role of users of an in-house or commercial program.

iv
Preface v

Information About Accompanying Software


PC computer software is provided with this book that contains a finite
element program called FEPC. The program runs on an IBM PC or
compatible computer. It will solve two-dimensional problems using truss,
beam, plane stress, plane strain or axisymmetric solid elements. Dimen-
sional limits of the program are described in an abbreviated user's guide in
the Appendix. Also, you should review the README file on the disk for
start-up instructions. The complete user's guide is in a file with the
software. The program is provided as shareware.
Although technical support can not be provided, we will replace any
defective files. Also, you may contact the author regarding any program
errors or availability of program updates.

Acknowledgments

I would like to thank my former students in the course for their suggestions
in refining the material for this text. Thanks to my colleagues and
department head in the Department of Mechanical Engineering for their
encouragement and support. My appreciation goes to Jonathan Plant,
Monique Calello and the staff at PWS Publishing Company for their work,
understanding and patience throughout this project. Finally, thanks to the
following individuals who reviewed this text and provided many useful
comments and suggestions that helped to improve the final product: A.
Henry Hagerdoorn, University of Central Florida; H. Kazerooni, University
of California - Berkeley; Julius Wong, University of Louisville; and Larry
Stauffer, University of Idaho.

Charles E. Knight, Jr.


CONTENTS

CHAPTER 1
THE FINITE ELEMENT METHOD 1
1.1 General Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 One-Dimensional Spring System . . . . . . . . . . . . . . . . . . . . 3
1.3 Using a Computer Program . . . . . . . . . . . . . . . . . . . . . . . . 7
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

CHAPTER 2
TRUSSES 11
2.1 Direct Element Formulation . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 The Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3 The Analysis Step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.4 Output Processing and Evaluation . . . . . . . . . . . . . . . . . 20
2.5 Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.6 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

CHAPTER 3
BEAMS AND FRAMES 28
3.1 Element Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.2 The Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . 33
3.3 Output Processing and Evaluation . . . . . . . . . . . . . . . . . 34
3.4 Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.5 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

vi
Contents vii

CHAPTER 4
TWO-DIMENSIONAL SOLIDS 40
4.1 Element Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.2 The Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . 44
4.3 Computer Input Assistance . . . . . . . . . . . . . . . . . . . . . . . 47
4.4 The Analysis Step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.5 Output Processing and Evaluation . . . . . . . . . . . . . . . . . 50
4.6 Case Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.7 Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

APPENDIX 58
Using FEPC, FEPCIP, and FEPCOP . . . . . . . . . . . . . . . . . . . 59
Entering the Model in FEPCIP . . . . . . . . . . . . . . . . . . . . . . . . 59
The Analysis by FEPC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Graphic Results Using FEPCOP . . . . . . . . . . . . . . . . . . . . . . . 71
C H A P T E R 1

THE FINITE ELEMENT METHOD

1.1 General Overview

The finite element method is enjoying widespread use in many engineering


applications. Although first developed for structural analysis, it now solves
problems in heat transfer, fluid mechanics, acoustics, electromagnetics, and
other specialized disciplines. In conduction heat transfer, we solve for the
temperature distribution throughout the body with known boundary
conditions and material properties whether steady state or time dependent.
Application to fluid mechanics begins with steady inviscid incompressible
flow and progresses to very complex viscous compressible flow. The whole
area of computational fluid dynamics has made rapid progress in recent
years. Acoustics is another area where great strides are being made based
on finite element and boundary element numerical methods. Electromag-
netic solutions for magnetic field strength provide insight for design of
electromagnetic devices. Many of these capabilities are now being coupled
to yield solutions to fluid-structure interactions, convective heat transfer,
and other coupled problems.
The finite element method is a numerical method for solving a system of
governing equations over the domain of a continuous physical system. The
method applies to many fields of science and engineering, but this text
focuses on its application to structural analysis. The field of continuum
mechanics and theory of elasticity provide the governing equations.
The basis of the finite element method for analysis of solid structures is
summarized in the following steps. Small parts called elements subdivide
the domain of the solid structure illustrated in Figure 1-1. These elements
assemble through interconnection at a finite number of points on each
element called nodes. This assembly provides a model of the solid

1
2 Chapter 1 The Finite Element Method

structure. Within the domain of each element we assume a simple general


solution to the governing equations. The specific solution for each element
becomes a function of unknown solution values at the nodes. Application
of the general solution form to all the elements results in a finite set of
algebraic equations to be solved for the unknown nodal values. By
subdividing a structure in this manner, one can formulate equations for
each separate finite element which are then combined to obtain the solution
of the whole physical system. If the structure response is linear elastic, the
algebraic equations are linear and are solved with common numerical
procedures.

Figure 1-1. Two-Dimensional Continuum Domain

Since the continuum domain is divided into finite elements with nodal
values as solution unknowns, the structure loads and displacement
boundary conditions must translate to nodal quantities. Single forces like
F apply to nodes directly while distributed loads like P are converted to
equivalent nodal values. Supports like the grounding indicated by the
hatch in Figure 1-1 resolve into specified displacements for the supported
nodes.
At least two sources of error are now apparent. The assumed solution
within the element is rarely the exact solution. The error is the difference
between assumed and exact solutions. The magnitude of this error depends
on the size of the elements in the subdivision relative to the solution
variation. Fortunately, most element formulations converge to the correct
solution as the element size reduces. The second error source is the
precision of the algebraic equation solution. This is a function of the
computer accuracy, the computer algorithm, the number of equations, and
Section 1.2 One-Dimensional Spring System 3

the element subdivision. Both error sources are reduced with good
modeling practices.
In theory all solid structures could be modeled with three-dimensional
solid continuum elements. However, this is impractical since many
structures are simplified with correct assumptions without any loss of
accuracy, and to do so greatly reduces the effort required to reach a
solution. Different types of elements are formulated to address each class
of structure. Elements are broadly grouped into two categories, structural
elements and continuum elements.
Structural elements are trusses, beams, plates, and shells. Their
formulation uses the same general assumptions about behavior as in their
respective structural theories. Finite element solutions using structural
elements are then no more accurate than a valid solution using convention-
al beam or plate theory, for example. However, it is usually far easier to
get a finite element solution for a beam, plate, or shell problem than it is
using conventional theory.
Continuum elements are the two- and three-dimensional solid elements.
Their formulation basis comes from the theory of elasticity. The theory of
elasticity provides the governing equations for the deformation and stress
response of a linear elastic continuum subjected to external loads. Few
closed form or numerical solutions exist for two-dimensional continuum
problems, and almost none exist for three-dimensional problems; this
makes the finite element method invaluable.
An extensive literature has developed since the 1960s when the term
"finite element" originated. The first textbook appeared in 1967 [see
Reference 1.1]. The number of books and conference proceedings published
since then is near two hundred and the number of journal papers and other
publications is in the thousands. The engineer beginning study of the finite
element method may consult references [1.2], [1.3], [1.4], [1.5], [1.6], [1.7],
or [1.8] for formulation [1.9], [1.10], [1.11], [1.12], [1.13], or [1.14] for
structural and solid mechanics applications, and [1.15], [1.16], or [1.17] for
computer algorithms and implementation.

1.2 One-Dimensional Spring System

The fundamental operation of the finite element method is illustrated by


analysis of a one-dimensional spring system. A two-spring structure is
sketched in Figure 1-2. Each spring is an element identified by the number
in the box. The spring elements have a node at each end and they connect
at a common node. The number in the circle labels each node. The number
in the box labels each element. The subscripted u values are the node
displacements, i.e., degrees-of-freedom. There is an applied force F at node
3. We wish to solve for the node displacements and spring forces.
4 Chapter 1 The Finite Element Method

Figure 1-2. One-Dimensional Spring Structure

The first step is to formulate a general element. Figure 1-3 shows a


spring element. The element label is p with nodes i and j. For a displace-
ment formulation, assume positive displacement components of ui at node
i and uj at node j. The element has a spring constant k, so node forces
result when these displacements occur. Define fip as the force acting on
node i due to the node displacements of element p. Application of simple
equilibrium forms equations (1.1).

Figure 1-3. One-Dimensional Spring Element

fip  kp ui  kp uj
(1.1)
fjp  kp ui  kp uj

We write these in matrix form in equation (1.2) and then symbolically in


equation (1.3).

kp k p ui fip
 (1.2)
k p kp uj fjp
Section 1.2 One-Dimensional Spring System 5

[k ]{d }  {f } (1.3)

Here, [k] is the element stiffness matrix, {d} is the element node displace-
ment vector, and {f} is the element node internal force vector. These steps
complete the element formulation.
Now apply the general formulation to each element:

for element 1

k1 k 1 u1 f11
 (1.4)
k 1 k1 u2 f21

and for element 2

k2 k 2 u2 f22
 (1.5)
k 2 k2 u3 f32

The force components in the element equations are internal forces on the
nodes produced by the elements when the nodes displace. Equilibrium
requires that the sum of the internal forces equals the external force at
each node. Representing the external force by Fi, where i represents each
node, the equilibrium equations become:

at node 1 ˆ forces  0 v  f11  F1


at node 2 ˆ forces  0 v  f21  f22  F2 (1.6)
at node 3 ˆ forces  0 v  f32  F3

Substitute the element equations for the internal force terms in the
equilibrium equations (1.6), and that, in effect, performs the structure
assembly and yields the structure equations (1.7).

k1u1  k1u2  F1
 k1u1  k1u2  k2u2  k2u3  F2 (1.7)
 k2u2  k2u3  F3
6 Chapter 1 The Finite Element Method

These are written in matrix form in equation (1.8) and symbolically in


equation (1.9) where [K] is the structure stiffness matrix, {D} is the
structure node displacement vector, and {F} is the structure external force
vector.

k1 k 1 0 u1 F1

k 1 k1 k2 k2 u2  F2 (1.8)

0 k 2 k2 u3 F3

[K ]{D }  {F } (1.9)

The set of structure or system equations must now be solved. The spring
constants of the springs are known, so all terms in the structure stiffness
matrix are known. The applied forces are known and the node displace-
ments become the unknowns in this set of three simultaneous equations.
We get the solution by premultiplying both sides of equation (1.9) by the
inverse of [K]. However, in this case the inverse of [K] is singular, meaning
that we cannot get a unique solution. Physically, this means that the
structure can be in equilibrium at any location in the x space, and it is free
to occupy any of those positions. This allows rigid body motion. To have a
unique solution we must locate the structure; that is, apply boundary
conditions such as a fixed displacement on one of the nodes which is enough
to prevent rigid body motion.
If an external force F applies to node 3, and the spring attaches to the
wall at node 1, then it is natural to set the displacement of node 1 to zero.
This action zeroes the first column of terms in the structure stiffness
matrix, and that leaves three equations with two unknowns. If the value
of the reaction force at node 1 is unknown, then we may skip the first
equation and choose the second and third equations to solve for the
unknown displacements. If the external force on node 2 is zero then

k1 k2 k 2 u2 0
 (1.10)
k 2 k2 u3 F

Now we may get the solution of the resulting equations (1.10) by


premultiplying both sides of the equation by the inverse of this reduced
structure stiffness matrix. Using the solved displacements, we calculate
each element internal force by use of the individual element equations. In
Section 1.3 Using a Computer Program 7

this example the force calculation is trivial, but in more complex elements
this step determines stresses in elements of the structure.
Also, in this example the calculation of the reaction force at node 1 is
easily done from elementary equilibrium equations. However, in general
cases, the structure under analysis may be statically indeterminant and the
reaction forces at locations of support or fixed displacement may not be
known. In that situation, the equations involving these reaction forces are
stored and solved after the displacements are found to calculate any desired
reaction forces.
This detailed example illustrates most of the fundamental steps in the
finite element method. The finite element method obviously overpowers
the example case. However, a complex spring arrangement could use the
procedure for analysis, or if a computer program were written, solution for
a complex spring arrangement would come quickly with input of the spring
constants and connectivity. The major differences between this example
and actual practice are that (1) nodes usually have more than one
displacement component or degree-of-freedom, (2) the element formulation
is chosen to match the class of structure being analyzed, and (3) a large
number of equations must be solved.

1.3 Using a Computer Program

There are three stages that describe the use of any existing finite element
program. The preprocessing stage creates the model of the structure from
inputs provided by the analyst. A preprocessor then assembles the data
into a format suitable for execution by the processor in the next stage. The
processor is the computer code that generates and solves the system
equations. The third stage is postprocessing. The solution in numeric form
is very difficult to evaluate except in the most simple cases. The
postprocessor accepts the numeric solution, presents selected data, and
produces graphic displays of the data that are easier to understand and
evaluate.
Figure 1-4 draws a block diagram of a typical finite element computer
program. Before entering the program's preprocessor, the user should have
planned the model and gathered necessary data. In the pre-processor
block, the user defines the model through the commands available in the
preprocessor. The definition includes input and generation of all node point
coordinates, selection of the proper element from the program's element
library, input and generation of node connectivity to define all elements,
input of material properties, and specifying all displacement boundary
conditions, loads, and load cases. The completion of the preprocessing stage
results in creation of an input data file for the analysis processor.
8 Chapter 1 The Finite Element Method

PREPROCESSOR

INPUT DATA
Element File
Control Data, Materials, Node and Element
Load File
Definition, Boundary Conditions, Loads

FORM ELEMENT [k ]

Read Element Data, Calculate Element Element File


Stiffness Matrix, [k ]

FORM SYSTEM [K ]

Assemble Element [k ]s to Form the


System Stiffness Matrix, [K ]

APPLY DISPLACEMENT BOUNDARY CONDITIONS

COMPUTE DISPLACEMENTS

Solve the System Equations


[K ]{D } = {F } Load File
for the Displacements
{D } = [K ] -1{F }

COMPUTE STRESSES
Displacement,
Calculate Stresses and Output
Stress Files
Files for Postprocessor Plotting

POSTPROCESSOR

Figure 1-4. Finite Element Computer Program Block Diagram

The processor reads from the input data file each element definition,
calculates terms of the element stiffness matrix, and stores them in a data
array or on a disk file. The element type selection determines the form of
the element stiffness matrix. The next step is to assemble the structure
stiffness matrix by matrix addition of all element stiffness matrices. The
application of enough displacement boundary conditions to prevent rigid
References 9

body motion reduces the structure stiffness matrix to a nonsingular form.


Then the equation solution may be done with several different computer
algorithms, but most use some variation of Gaussian elimination. For a
large set of equations this is the most computationally intensive step in the
process. The solution here yields values for all node point displacement
components in the model. The node displacements associated with each
element combined with the element formulation matrix yield the element
strains. The element strains with the material properties yield the stresses
in each element. The processor then produces an output listing file with
data files for postprocessing.
Postprocessing takes the results files and allows the user to create
graphic displays of the structural deformation and stress components. The
node displacements are usually very small for most engineering structures
so they are magnified to show an exaggerated shape. Node displacements
are single-valued, but node values of stress are multi-valued if more than
one element is attached to a given node. Node stress values are usually
reached by extrapolation from internal element values and then averaged
for all elements attached to the node. Contour plots or other stress plots
desired by the user are created from the node values. In some post-
processing programs criterion plots of the factor of safety, stress ratio to
yield, or stress ratio to allowable values are also generated and displayed.
The engineer is then responsible for interpreting the results and taking
whatever action is proper. The user must estimate the validity of the
results first. This is very important because the tendency is to accept the
results without question. Experience, thorough checking of the modeling
assumptions and resulting predicted behavior, and correlation with other
engineering calculations or experimental results all contribute to estimat-
ing the validity of the results.

References

1.1 Zienkiewicz, O. C. and Cheung, Y. K., The Finite Element Method in


Structural and Continuum Mechanics, McGraw-Hill, London, 1967.
1.2 Akin, J. E., Finite Element Analysis for Undergraduates, Academic Press,
London, 1986.
1.3 Gallagher, R. H., Finite Element Analysis — Fundamentals, Prentice-Hall,
Englewood Cliffs, New Jersey, 1975.
1.4 Huebner, K. H. and Thornton, E. A., The Finite Element Method for
Engineers, John Wiley and Sons, New York, 1982.
1.5 Irons, B. and Ahmad, S., Techniques of Finite Elements, John Wiley and
Sons, New York, 1980.
1.6 Irons, B. and Shrive, N., Finite Element Primer, John Wiley and Sons, New
York, 1983.
1.7 Reddy, J. N., An Introduction to the Finite Element Method, McGraw-Hill,
New York, 1984.
10 Chapter 1 The Finite Element Method

1.8 Zienkiewicz, O. C. and Taylor, R. L., The Finite Element Method, Volume
1 — Basic Formulation and Linear Problems, Fourth Edition, McGraw-Hill,
New York, 1989.
1.9 Cook, R. D., Malkus, D. S., and Plesha, M. E., Concepts and Applications
of Finite Element Analysis, Third Edition, John Wiley and Sons, New York,
1989.
1.10 Fenner, D. N., Engineering Stress Analysis: A Finite Element Approach
with Fortran 77 Software, John Wiley and Sons, New York, 1987.
1.11 Potts, J. F. and Oler, J. W., Finite Element Applications with Microcomput-
ers, Prentice-Hall, Englewood Cliffs, New Jersey, 1989.
1.12 Ross, C. T. F., Finite Element Methods in Structural Mechanics, John Wiley
and Sons, New York, 1985.
1.13 Stasa, F. L., Applied Finite Element Analysis for Engineers, Holt, Rinehart
and Winston, New York, 1985.
1.14 Weaver, W. and Johnston, P. R., Finite Elements for Structural Analysis,
Prentice-Hall, Englewood Cliffs, New Jersey, 1984.
1.15 Akin, J. E., Application and Implementation of Finite Element Methods,
Academic Press, London, 1982.
1.16 Bathe, K. J. and Wilson, E. L., Numerical Methods in Finite Element
Analysis, Prentice-Hall, Englewood Cliffs, New Jersey, 1976.
1.17 Bathe, K. J., Finite Element Procedures in Engineering Analysis, Prentice-
Hall, Englewood Cliffs, New Jersey, 1981.
C H A P T E R 2

TRUSSES

The primary focus of this text is on the aspects of finite element analysis
that are more important to the user than the formulator or programmer.
However, for the user to employ the method effectively he or she must have
some understanding of the element formulations as well as some of the
computational aspects of the programming. Therefore, the next chapters
begin by looking at the element formulation for a given structural behavior
class before proceeding to model development and the proper modeling
approach. Most finite elements develop from use of an assumed displace-
ment approximation; therefore, the elements presented will all deal with
assumed displacement formulations.

2.1 Direct Element Formulation

This section presents the direct physical formulation of a truss element and
its spatial orientation to solve two-dimensional frameworks. A member of
a truss structure is like a one-dimensional spring. The member has a
length substantially larger than its transverse dimensions, and it has a
pinned connection to other members that eliminates all loads other than
axial load along the member length. It usually has a constant cross-section
area and modulus of elasticity along its length. The stiffness is then

AE
k  (2.1)
L

where A is the cross-section area, E is the modulus of elasticity, and L is


the member length.

11
12 Chapter 2 Trusses

A one-dimensional truss element then has an element formulation


identical to the one-dimensional spring given in equation (1.2). This is its
element stiffness matrix for one-dimensional displacement and loading
along the axis of the member. For member positioning in a two-dimen-
sional space as illustrated in Figure 2-1, each node has two components of
displacement u and v and two components of force p and q. This leads to
a set of element equations with an element stiffness matrix of size 4 by 4.

Figure 2-1. Two-Dimensional Truss Element

Derivation of the two-dimensional element stiffness matrix comes


through coordinate transformations [2.1], but first we expand the one-
dimensional stiffness matrix to two dimensions with the member lying
along the x axis. Assuming an order of components and equations of u and
v at node i followed by u and v at node j, the element equations are written
in equation (2.2).

k 0 k 0 u i p i

0 0 0 0 vi q i
 (2.2)
k 0 k 0 uj p j

0 0 0 0 vj q j

Notice that the terms relating displacement and force in the x direction
Section 2.1 Direct Element Formulation 13

are the spring constant of the member and the terms relating displacement
and force in the y direction are zero. A linear analysis always assumes that
the displacements are much smaller than the overall geometry of the
structure; therefore, the stiffness is based on the undeformed configuration.
In this case, because it is a motion perpendicular to the line of the member,
if we consider a vertical displacement component at one of the nodes, no
vertical force results because there is no axial stretch relative to the
undeformed configuration.
This formulation represents the element stiffness matrix in a local
element coordinate system that is aligned with the element axis. To
position the element at an arbitrary angle, O, from the x coordinate axis, we
perform a transformation of coordinate systems to derive the element
stiffness matrix in the x,y global coordinate system. In the system of
equations, the displacements and forces are both vectors, so they transform
through standard vector transformations. The displacement components
in global coordinates relate to local components through equation (2.3).

{d }  [T ]{d k } (2.3)

Here, {d'} are the global displacement components, [T] is the transforma-
tion matrix, and {d} are the local element coordinate displacement
components.
The transformation matrix is given by equation (2.4).

c s 0 0
s c 0 0
[T ]  (2.4)
0 0 c s
0 0 s c

Here, s is the sin O, and c is the cos O. Similarly, the force components in
the global coordinate system are given by

{f }  [T ]{f k } . (2.5)

The element stiffness matrix in the local coordinate system is defined in


matrix notation from equation (2.2) by
[k ]{d }  {f } . (2.6)
14 Chapter 2 Trusses

Making the substitutions for {d} and {f} given above yields

[k ][T ]{d k }  [T ]{f k } . (2.7)

The transformation matrix is an orthogonal matrix, meaning that



[T ]1  [T ] . (2.8)


Therefore, multiplying equation (2.7) by [T] produces

[T ] [k ][T ]{d k}  {f k } (2.9)

which makes

c2 cs c 2 cs

— cs s2 cs s 2
[k k ]  [T ] [k ][T ]  k (2.10)
c 2 cs c2 cs
cs s 2 cs s2

The use of this element formulation and equation assembly is shown


through the example truss structure pictured in (2.11). The elements and
nodes are numbered, and load and boundary conditions are shown. The
structure equations are

[K ]{D }  {F } (2.11)

where, [K] is the structure stiffness matrix, {D} is the node displacement
vector, and {F} is the applied load vector.
These equations come from applying the conditions of equilibrium to all
the nodes by setting the summation of internal forces equal to the applied
forces. The internal forces are given by the product of each element
stiffness matrix with its node displacements. This yields equation (2.12),
where the subscripts refer to the numbered elements. If the displacement
vector in each term of the equation above was identical, then we could

[k ]{d }|1  [k ]{d }|2  [k ]{d }|3  {F } (2.12)


Section 2.1 Direct Element Formulation 15

Figure 2-2. Example Truss Structure

factor it out and add the stiffness matrices term-by-term to produce the
structure stiffness matrix.
The displacement vector for each element must then expand to include
all the structure degrees-of-freedom, not just the ones associated with a
given element. In order for the matrix equation to be correct, a corre-
sponding expansion of the element stiffness matrix must accompany the
expansion of the displacement vector. It expands to the size of the
structure stiffness matrix which in this example becomes a 6-by-6 matrix.
The expansion simply adds rows and columns of zeroes to each element
stiffness matrix corresponding to the additional structure degrees-of-
freedom unused in the given element [2.1].
Applying this approach, the stiffness matrix for element 1 in the example
results from using equation (2.10) with a O value of 90 degrees. Rows and
columns of zeroes fill in equations and positions involving u1 and v1 as
shown in equation (2.13).

0 0 0 0 0
0 0 0 0 0 0 0 0 0
0 k1 0 k 1 0 0 0 0 0
  (2.13)
0 0 0 0 0 0 0 k1 0
0 k 1 0 k1 0 0 0 0 0
0 0 0 k 1 0

Similarly, the matrix for element 2 with O equal to 135 degrees and rows
16 Chapter 2 Trusses

and columns filled in equations and positions involving u2 and v2 results in


equation (2.14).

.5k2 .5k2 0 0 .5k2 .5k2


.5k2 .5k2 0 0 .5k2 .5k2
0 0 0 0 0 0
[k ] 2  (2.14)
0 0 0 0 0 0
.5k2 .5k2 0 0 .5k2 .5k2
.5k2 .5k2 0 0 .5k2 .5k2

Finally for element 3, O is 0 degrees and u3 and v3 are the additional


degrees-of-freedom in equation (2.15).

k3 0 k 3 0 0 0
0 0 0 0 0 0
k 3 0 k3 0 0 0
[k ] 3  (2.15)
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0

The summations of equation (2.12) are now carried out by adding the
expanded element stiffness matrices term-by-term. The resulting structure
stiffness matrix is in equation (2.16).

.5k2k3 .5k2 k 3 0 .5k2 .5


.5k2 .5k2 0 0 .5k2 .
k 3 0 k3 0 0
[K ]  (2.16)
0 0 0 k1 0 
.5k2 .5k2 0 0 .5k2 .
.5k2 .5k2 0 k 1 .5k2 .5k
Section 2.1 Direct Element Formulation 17

Before solving the equations, apply the displacement boundary


conditions. In this example the boundary conditions have components u2,
v2, and u3 equal to zero. Applying these to the system equations zeroes the
third, fourth, and fifth columns of the structure stiffness matrix. This
leaves six equations with three unknown displacements. In most problems,
the reaction forces in equations 3, 4, and 5 are also unknown, so choose
equations 1, 2, and 6 to solve for the displacement components, u1, v1, and
v3.
We get the solution by finding the inverse of the remaining 3-by-3
stiffness matrix. Multiplying the inverse with the load vector yields the
displacements. After finding the displacements, calculate the element
forces by use of the element equations.
This concludes the example and demonstration of two-dimensional truss
formulation. Of course, this extends easily to three dimensions. The
assembly of equations by expanding the element stiffness matrix to
structure size is useful for explanation of the process but impractical for a
large number of system equations. In computer programs the algorithm
only needs to place the terms of the element stiffness matrix in the correct
position in the structure stiffness matrix. This is easy to accomplish
because the structure equation sequence correlates to the node number.
Elements are defined by node numbers, thus providing the direct correla-
tion for positioning the terms.

2.2 The Finite Element Model

The practitioner of finite element analysis normally uses an existing


computer code. Because of the general complexity and sizable effort
required to create a finite element code, it is impractical to consider writing
a code for every specific problem that needs solving. So the job of the
practitioner is to use an existing code to solve the problem of interest. The
user in this case is responsible for creating the model of the structure, for
managing the execution of the program, and for interpreting results created
by the program.
In following the analysis procedure we reach the point of planning the
model. The arrangement of nodes and elements that describe the model is
known as the mesh. Using all the information that is known about the
problem, and knowing the capabilities of the program chosen to analyze the
problem, the mesh is planned to model the structure properly. In truss
structures, each member is modeled as one truss element with the
connections of truss members or elements at the node points. Based on the
formulation and general assumptions for truss members, these node
connections behave as pinned joints. This results in no flexural loading of
the member in that one member can swivel or hinge relative to the others
18 Chapter 2 Trusses

connected at any given node, yet will transmit axial load.


Since a truss element behaves exactly in agreement with the assump-
tions of a truss member, there is no need to divide a member into more
than one element. In fact, such a subdivision will cause the execution stage
of the program to fail. The failure is due to the zero stiffness against any
lateral force applied at a node connection where two members are in perfect
axial alignment. So just as a physical truss structure constructed in this
manner would collapse, the numerical solution of the problem defined in
this manner also should collapse.
A simple bridge structure is in Figure 2-3 and a corresponding finite
element mesh with elements and nodes numbered is in Figure 2-4.

Figure 2-3. Simple Bridge Structure

Figure 2-4. Simple Bridge Structure Finite Element Mesh

2.3 The Analysis Step

Operation of the analysis step is mostly transparent to the user. However,


there are some factors that the user must be familiar with to assure good
Section 2.3 The Analysis Step 19

results. Some of them are inherent in the computer hardware and


software, but some of them are under user control. For small truss
structure models, most programs have enough numerical accuracy and
performance to provide an accurate solution without much user concern.
Large models cause the most concern [2.1].
A large model is one in which there are many elements and nodes used
to represent the structure. (The structure itself is not necessarily large.)
With many system equations it becomes difficult to find a numerical
solution if the equation matrix is full. Even the inverse of a 10-by-10
matrix may be inaccurate if done by Gauss elimination with only a few
significant figures carried along in the mathematical operations. The
accuracy will improve, however, if the matrix has its nonzero terms
clustered near the diagonal. This reduces the number of operations and
reduces the roundoff error carried along in each operation. This kind of
matrix, with its nonzero terms near the diagonal, is a banded matrix.
If the truss structure model consists of thousands of nodes and elements,
then the bandwidth of the structure equations needs to be small. Keeping
it small reduces error and computing time. If the mesh plan does not have
a small bandwidth for the system of equations, then bandwidth minimizers
available in many programs should reduce it. There are several algorithms
available which will usually, but not always, find a better node or element
numbering pattern. Most programs keep the original numbering in the
model for documentation and presentation purposes by storing the
renumbered nodes and elements in translation tables. In some programs
the user has the option to keep the original numbering or change to the
new numbering plan.
The approximation error for the truss element is zero since the element
formulation is in exact agreement with the assumptions used to define a
truss member. During processor execution there are usually some prompts
of progress made displayed on the computer, and if errors occur, messages
appear. Sometimes these messages have meaning only to the computer
program developer, but some of them can be very helpful in determining
the model error.
The most common runtime errors involve incorrect definition of elements
or incorrect application of displacement boundary conditions. For example,
both conditions can produce an error message that the structure stiffness
matrix is not positive-definite or that a negative pivot or diagonal term in
the stiffness matrix appeared during equation reduction. For truss models
this can occur whenever there are not enough boundary conditions to
prevent rigid body motion. It can also occur when two elements connect in-
line resulting in zero lateral stiffness. It can also mean that the truss
structure itself is not kinematically stable associated with a kinematic
linkage of the members.
The error messages from the computer program should provide an
associated element number, node number, or equation number where the
20 Chapter 2 Trusses

error occurred. That makes it easier for the user to pinpoint the problem.
When execution completes without errors, then postprocessing may begin.

2.4 Output Processing and Evaluation

At this stage of the analysis, all programs have numerical results in the
form of a listing file of the problem. This file will include a summary of the
input data followed by numerical values of all node displacement compo-
nents and all element stress results. One of the important steps to take
here is to review the summary of the input printout, scanning it for errors
in input interpretation of the data entered or selection of default parame-
ters that are not appropriate for the problem at hand. This can usually be
done effectively if the program formats the data for easy viewing.
The results for displacements and stresses in the listing file for large
models are so lengthy that scanning is not practical. However, many
programs will print a summary of maximum values for displacement
components and stress magnitudes. Therefore, it is very desirable to
present the data graphically for more effective evaluation.
The first graphic of importance should be an exaggerated deformed
shape of the structure. All postprocessing programs will include this
graphic that uses the node displacements with a scale factor to exaggerate
the deformation and make it more apparent to the eye. The deflections in
most engineering structures are usually very small, and without an
exaggeration scale factor the deformed shape would look the same as the
undeformed shape. Program options usually exist either to provide both an
undeformed and deformed mesh simultaneously or an outline of the
undeformed object superimposed on the graphic of the deformed mesh.
The engineer must look at this plot critically and make sure that the
boundary conditions are correct and that the shape of the deformed
structure agrees qualitatively with the expected deformation. In truss
structures the deformed shape will obviously show each member or element
as a straight line connecting the nodes in an exaggerated deformed
position.
After thorough evaluation of the deformed shape, the graphics should
then turn to plots of the stress components. In continuum structures the
stress component plots relate to averaged quantities at the node points.
Truss structures have a stress in each member that is constant, and most
commercial postprocessing programs do not provide much in the way of
graphic presentation of these stresses. In this event the user must return
to the listing file and examine it for the highest stressed members.
The evaluation of the results determines whether we need to make
Section 2.5 Case Study 21

additional model refinements and whether the results have converged to


enough accuracy. For truss structures we know that the element formula-
tion is exact; therefore, in a correctly defined and processed model the
output results will be exact. So there is no need for refined modeling to
produce converged results in this case, but the modeling of loads and
boundary conditions may not be fully appropriate.
It is important to remember that this is a linear elastic analysis. One of
the potential failure modes is overstressing while another is elastic
buckling. The stresses compare to yield strength for the material to
determine if overstress failure occurs. To determine whether there is
potential for elastic buckling, the user must identify the members with a
significant compression load, and then use Euler buckling equations from
mechanics of materials to evaluate the potential for each member to buckle
[2.2]. This is not done in a linear elastic analysis computer program. If any
member has an inadequate safety factor against buckling, then the entire
structure should have a stability analysis conducted using a solution
algorithm available in some nonlinear computer codes.

2.5 Case Study

We show a typical truss structure in Figure 2-5. This structure has


members of two different cross-section areas and two different materials.
The mesh plan for this structure is illustrated in Figure 2-6 with a chosen
pattern for node and element numbers. The enforced boundary conditions

Figure 2-5. Two-Dimensional Truss Structure


22 Chapter 2 Trusses

symbolized by the triangular shapes have the triangle tip placed on the
node pointing in the direction of restraint. The loads symbolized by long
arrows apply to the indicated node points.

Figure 2-6. Truss Structure Finite Element Model

Table 2-1 lists the input data required to create this model. The title
line is first followed by the control data line. Node definition lines begin
with their number, with boundary condition restraints and coordinate
locations following. In this model all the z boundary conditions are
restrained and all the z coordinates are zero because this is a 2-D truss
element without any z degrees-of-freedom. Additional control data is next
followed by load data lines. The load data lines begin with the node
number of application with a direction and magnitude. The type of element
is a truss. The material data is shown with two table entries. Material 1
has a modulus of elasticity for steel with a cross-section area of 0.4 in2, and
material 2 is aluminum with a cross-section area of 0.7 in2. The element
definitions are given by the entry of two node numbers at the endpoints of
the element with a material table assignment.
This is such a small model that the equation bandwidth factor is
insignificant along with any question of numerical performance or
precision. Following program execution the deformed shape of the
structure is shown in Figure 2-7. Note particularly that the enforced
boundary conditions match, and the structure deforms in a manner that
agrees with its expected deformation. The displacement, load, and stress
results are in Table 2-2. Finally, the stress results are displayed in bar
graph form in Figure 2-8.
We may evaluate these results for the design by knowing the additional
property given by its yield strength. If we say the steel has a yield strength
of 50 kpsi and the aluminum has a yield strength of 30 kpsi, then the
lowest factor of safety will be in element number 9 which has a value of
Section 2.5 Case Study 23

Table 2-1. Input Data File for Truss Structure Model

***********|notes inside vertical bars are for data


explanation only|
Truss Case Study
|title line|
6 1 1 |6 nodes, 1 element group, 1
load case|
1 1 1 1 0.000 0.000 0.000 |node
number, |
2 0 0 1 10.000 0.000 0.000 |x,y,z
boundary|
3 0 0 1 20.000 0.000 0.000
|conditions, |
4 0 1 1 30.000 0.000 0.000
|0-free,1-fixed|
5 0 0 1 10.000 10.000 0.000 |x,y,z
|
6 0 0 1 20.000 10.000 0.000
|coordinates |
0 |number of inclined boundary
conditions|
1 3 |load case
1, 3 loads|
5 2 -5000.0 |node 5, y dir.,
-5000 value|
6 1 2000.0 |node 6, x dir.,

Figure 2-7. Truss Structure Deformed Shape

2.12. We see that members 4, 6, and 9 have compressive axial loads. If we


use the Euler buckling equation we have to know more than just its elastic
modulus and cross-section area. We also must know its cross-section shape
to find the area moment of inertia and the end conditions. If we assume
pinned end conditions and the cross-section shape is a solid circular rod
then the buckling loads for these elements would be 19250, 38500, and
19250 respectively. Thus, we also have a reasonable factor of safety
against buckling.
24 Chapter 2 Trusses

Table 2-2. Results Data File for Truss Structure Model

D I S P L A C E M E N T S
NODE X-DISP Y-DISP Z-DISP
1 0.000000 0.000000 0.000000
2 0.005833 -0.035436 0.000000
3 0.011667 -0.037694 0.000000
4 0.017500 0.000000 0.000000
5 0.015233 -0.035436 0.000000
6 0.008091 -0.037694 0.000000

S T R E S S E S I N T R U S S E L E M E N T
G R O U P 1
ELEM # FORCE STRESS

1 7000. 17500.
2 7000. 17500.
3 7000. 17500.
4 -7071. -10102.
5 0. 0.
6 -5000. -7143.
7 0. 0.
8 0. 0.

Figure 2-8. Truss Structure Member Stresses


Section 2.6 Closure 25

2.6 Closure

There are few situations in mechanical design where a truss element is the
right element for modeling the behavior. It is a simple element with which
to discuss and learn finite element concepts. It may make an important
contribution to the analysis by use as a boundary supporting spring or a
gap element that connects two or more separate parts of a machine that
must interact in the analysis. Therefore, the engineer must understand its
nature well to interpret its effect on the overall response of any analysis
that includes truss or truss-based elements.

Problems

2.1 A three-member truss structure is shown in Figure P2-1 with corre-


sponding node and element numbering for a finite element model.
Elements 1 and 2 are aluminum, and element 3 is steel. The cross-
section areas are 1.5 sq. in. for element 1 and 1.0 sq. in. for elements
2 and 3. Determine the displacement of node 2 and the stresses in
each member. Solve by use of a computer program and by hand
calculation. Report in a neat and concise informal engineering
communication. Please include the hand-calculated struc-ture
stiffness matrix in its full (8-by-8) and reduced form.

2.2 Design the derrick structure shown for a load capacity of 20 kips.
Choose a suitable steel and, using a factor of safety of 4.0, determine
the cross-section area for all the members. Recommend a cross-
section shape that will prevent any member from buckling.
26 Chapter 2 Trusses

2.3 Design a cantilevered boom to support the loads shown in Figure P2-
3. All members are steel with a cross-section area of 1 sq. in. The
material has an allowable stress of 20 kpsi. First determine if the
design is satisfactory as illustrated. Next redesign the structure
within the geometric boundaries shown and the same allowable
stress. A redesign may change the member arrangements, eliminate
members, or change cross-section areas. One of the redesign goals
should be to reduce the overall weight of the structure. Determine
a suitable cross-section shape to prevent buckling for each member
in compression.

2.4 Repeat Problem 2.3 for the boom in Figure P2-4.


References 27

References

2.1 Cook, R. D., Malkus, D. S., and Plesha, M. E., Concepts and Applications
of Finite Element Analysis, Third Edition, John Wiley and Sons, New York,
1989.
2.2 Popov, E. P., Introduction to Mechanics of Solids, Prentice-Hall, Inc.,
Englewood Cliffs, New Jersey, 1968.
C H A P T E R 3

BEAMS AND FRAMES

Application of straight beam theory readily solves simple beam problems


especially if the problem is statically determinant. If the beam is not
particularly simple, in that it may have cross-section changes, multiple
supports, or complex loading distributions, then we can use beam theory,
but it is very tedious to develop the solution by hand. Also, many 2-D or 3-
D framework structures may require solutions in which the truss member
assumption is inadequate and therefore needs the beam flexure formula-
tion. Further applications may include beam members as re-inforcement
members in combination beam, plate, and shell structures. These
applications are readily attacked with the finite element formulation.

3.1 Element Formulation

Here we follow the direct approach for formulating the element stiffness
matrix [3.1]. The element equations relating general displacement and
force components are given by

[k ]{d }  {f } (3.1)

where [k] is the element stiffness matrix, {d} is the node displacement
component column matrix, and {f} is the internal force component column
matrix. The stiffness matrix terms derive from superposition of simple
beam solutions. Apply a unit displacement of one component with the other
components held to zero and evaluate the magnitude of resulting force
components. For example, taking the element shown in Figure 3-1 and

28
Section 3.1 Element Formulation 29

applying a unit vertical displacement vi with ki = vj = kj = 0 results in the


deformation illustrated in Figure 3-2.

Figure 3-1. Beam Element

Figure 3-2. Deformed Beam Element

The forces shown, which are the superposition of the solutions for a
cantilever beam with an end load and an end moment, produce this
deformation. The superposition is done to yield a unit value of lateral
displacement with a zero slope at the end. The element equations written
in matrix form yield equation (3.2), which in turn yields the relations in
equation (3.3).

k11 k12 k13 k14 1 Fi

k21 k22 k23 k24 0 Mi


 (3.2)
k31 k32 k33 k34 0 Fj

k41 k42 k43 k44 0 Mj


30 Chapter 3 Beams and Frames

k11  Fi , k21  Mi , k31  Fj , and k41  M (3.3)

Using superposition of beam deflection equations available in any


mechanics of materials text, we write equations (3.4). Solve these
equations for the values of Fi and Mi in equations (3.5).

Fi L 3 Mi L 2
vi  1  
3EI 2EI
(3.4)
Fi L 2 Mi L
ki  0  
2EI EI

12EI
Fi 
L3
(3.5)
6EI
Mi 
L2

Use static equilibrium equations to get the values of Fj and Mj in equation


(3.6).

12EI 6EI
Fj   Mj  (3.6)
3
L L2

We now have all the terms of column 1 of the 4x4 element stiffness
matrix as shown in equation (3.7).

12EI
... ... ...
L3
6EI
... ... ...
L2
[k ]  (3.7)
12EI
 ... ... ...
L3
6EI
... ... ...
L2
Section 3.1 Element Formulation 31

Similarly applying a unit value of rotation for ki and fixing all other
components to zero, we derive the force and moment values in Figure 3-3.
These come from superposition of the same solutions for end load and
moment to satisfy the displacement conditions.

Figure 3-3. Deformed Beam Element

Notice that the sign convention employed here is common in the finite
element formulation such that the component's sign always agrees with the
positive direction of a right-handed coordinate system. This does not agree
with most beam sign conventions employed in mechanics of material texts.
Therefore, the user should be aware that the output components will
normally be expressed using this finite element sign convention. This
means, for example, that a positive value of moment at the first node of the
element will produce a tensile stress at the top surface of the beam. In
contrast, a positive moment on the second node will produce a compressive
stress at the top surface of the beam.
Obtain the remaining terms in the stiffness matrix by application of the
same procedures to the second node. The final element stiffness is then
given in equation (3.8).

12EI 6EI 12EI 6EI



L3 L2 L3 L2
6EI 4EI 6EI 2EI

L 2 L L 2 L
[k ]  (3.8)
12EI 6EI 12EI 6EI
  
L3 L2 L3 L2
6EI 2EI 6EI 4EI

L 2 L L 2 L
32 Chapter 3 Beams and Frames

This formulation provides an exact representation of a beam span within


the assumptions involved in straight beam theory, provided there are no
loads applied along the span. Therefore, in modeling considerations, place
a node at all locations where concentrated forces, or moments, act in
creating the element assembly. In spans where there is a distributed load,
the assumed displacement field does not completely satisfy the governing
differential equation; therefore, the solution is not exact but approximate.
One approach to modeling in this area is to make enough subdivisions of
the span with distributed load to lessen the error. If a work equivalent load
set acting on the nodes replaces the distributed load, then the influence of
any error in this element will not propagate to other elements. In other
words, the displacement components at the nodes will be correct if we use
the equivalent load set. The equivalent load components for a distributed
load on the element span are the negative of the end reaction force and
moment found in the solution of a fixed end beam with the same distrib-
uted load as shown by Logan [3.2].
This formulation provides the ability to analyze simple beams, but does
not account for the axial load that may exist in beam members connected
in a framework. By adding the truss element formulation by superposition
with the previous formulation, we have an element that can support both
lateral and axial loads. The axial stiffness terms at each node are added
to the element stiffness matrix formulation to create the frame element
stiffness matrix in equation (3.9).

AE AE
0 0  0
L L
12EI 6EI 12EI
0 0 
3 2
L L L3
6EI 4EI 6EI
0 0 
L 2 L L2
[k ]  (3.9)
AE AE
 0 0 0
L L
12EI 6EI 12EI
0   0
3 2
L L L3
6EI 2EI 6EI
0 0 
L 2 L L2

This assumes that superposition is valid for this case. If displacements


are small it will be accurate; however, there is an interaction that occurs
Section 3.2 The Finite Element Model 33

between axial and lateral loading on beams. If the axial load is tensile it
reduces the effect of lateral loads, and when the axial load is compressive
it amplifies the effect of lateral loads. To gain further information on this
interaction, consult an advanced mechanics of materials text [3.3] for the
equations that apply to members called beam-columns or struts. The
equations for these members are a nonlinear function of the size of lateral
displacement. Therefore, a linear analysis cannot account for the effect.
The user should be aware of this consideration. Remember that if the
axial load is tensile, the results from beam elements will be higher than
they actually are; thus results are conservative. Also, if the axial load is
compressive, the results will be less than actual and may be in serious
error. The size of error associated with the compressive loading is normally
quite small until the axial load exceeds roughly 25 percent of the Euler
column buckling load. In most cases a design should have a factor of safety
against buckling greater than four anyway.
Now the formulation includes the u and v displacement components and
the section rotation at the nodes in the element local coordinate system.
Using the coordinate transformations developed for truss members, we may
orient this two-dimensional beam element in 2-D space. Through this
transformation, then, the element formulation applies to any 2-D frame-
work.

3.2 The Finite Element Model

In planning the mesh for a structure to be modeled with beam elements,


the factors just revealed in element formulation provide guidance about the
proper element subdivision and connections. Since the element formulation
is exact for a beam span with no intermediate loads, then we need only one
element to model any member of the structure that has constant cross-
section properties and no intermediate loads. Where a span has a
distributed load, we may subdivide it with several elements to lessen the
error depending on the solution accuracy desired.
There should be a node placed at every location in the structure where
a point load is applied. Also, where frame members connect such that the
line element changes direction or cross-section properties change, we should
place a node and end an element at that point. Remember that the
connection of two or more elements at a node guarantees that each element
connecting at that node will have the same value of linear and rotation
displacement components at that node. Physically, think of this as a solid,
continuous, or welded configuration.
34 Chapter 3 Beams and Frames

3.3 Output Processing and Evaluation

A complete printout, or listing file, lists a reflection of model input data, the
displacement results including rotations, and output of stresses resulting
from moment, axial, and shear forces. The graphical presentation of the
deformed shape ideally would use the rotations at the nodes with the
assumed displacement shape function for the element to plot the actual
curved shape the elements take when loaded. However, most programs
only plot the deformed shape using the node translation displacements and
straight line connections to represent the elements. In this case it is
difficult to determine from the graphic if we applied the rotational
boundary conditions. In order to check boundary conditions and get a
smooth visualization of the deformation curvatures, the user may resort to
remodeling with several element subdivisions within each span.
The stresses in 2-D beam elements consist of a normal stress acting
normal to the beam cross section and a transverse shear stress acting on
the face of the cross section. The normal stress comes from superposition
of the axial stress that is uniform across the section with the bending stress
due to the moment on the section. This combination will result in the
maximum normal stress occurring either at the top or bottom surface. The
transverse shear stress is usually an average across the cross section
calculated by the transverse load divided by the area. This obviously does
not account for the shear stress variation that occurs across the section
from top to bottom [3.3]. The transverse shear stress must be zero at the
top and bottom surfaces and has some nonuniform distribution in between
that is a function of the cross-section geometry. This variation is usually
of minor importance, but the analyst may calculate it if desired.
Most of the available finite programs do not make graphical presentation
of the beam stress results. So it reverts to the engineer to evaluate the
stress output usually based on values from the printout listing. The
engineer also must check for Euler buckling in members that have an axial
compressive stress. If the factor of safety against buckling in these mem-
bers is less than about 4, then the stresses may need correction for the
interaction between the axial and flexural stress in that member.

3.4 Case Study

We show a simple beam structure in Figure 3-4 with two different cross
sections and two loads. It has simple supports, and we develop a mesh plan
in Figure 3-5 with five nodes and four elements. The model input data list
is in Table 3-1. The title line is first with the control data line next. Node
definition begins with its number, then boundary condition restraints and
Section 4.5 Output Processing and Evaluation 35

coordinate locations following. The z boundary condition for the 2-D beam
element applies to the rotation degree-of-freedom. All the z boundary
conditions are free in this model.

Figure 3-4. Simple Beam with a Cross-Section Change and Two


Loads

Figure 3-5. Finite Element Model of the Simple Beam

Load specification consists of the node number of application with the


load direction and magnitude. The type of element is a beam. The
material data is shown with two table entries. Entry 1 has a modulus of
elasticity for steel with a cross-section area of 0.4 in2, moment of inertia of
0.04 in4, and distance from neutral axis to the surface of 0.5 in. Entry 2 is
aluminum with a cross-section area of 0.7 in2, moment of inertia of 0.06 in4,
and distance from neutral axis of 0.75 in. The element definitions are given
by the entry of two node numbers at the endpoints of the element with a
material table assignment.
Following execution of the program, the deformed shape plot appears in
36 Chapter 3 Beams and Frames

Table 3-1. Input Data File for Beam Model

***********|notes inside vertical bars are for data explana-


tion only|
Beam Case Study
|title line|
5 1 1 |5 nodes, 1 element group, 1
load case|
1 1 1 0 0.000 0.000 0.000 |node
number, |
2 0 0 0 5.000 0.000 0.000 |x,y,z
boundary |
3 0 0 0 8.000 0.000 0.000 |condi-
tions |
4 0 0 0 10.000 0.000 0.000 |0-free,
1-fixed|
5 0 1 0 15.000 0.000 0.000 |x,y,z
coord. |
0 |number of inclined boundary
conditions|
1 2 |load case 1,
2 loads|

Figure 3-6, and the results printout is in Table 3-2. In this table the z
displacements are the angular rotations of the nodes. The axial stress is
the value from any axial load acting on the element. The flexure stress is
due to the resulting bending moment and is the value on the beam top
surface when the element definition has nodes I and J arranged left to
right. The average shear stress is simply the transverse shear load divided
by the cross-section area. The shape of the cross section determines the
actual shear stress distribution across the beam height. Finally, a bar
graph display of the beam element stresses is given in Figure 3-7.

Figure 3-6. Deformed Shape of the Simple Beam

These results show that while the finite element method provides
solutions as valid as straight beam theory will allow, there has been no
accounting for stress concentration effects where the cross-section change
Section 3.4 Case Study 37

Table 3-2. Results Data File for the Beam Model

D I S P L A C E M E N T S
NODE X-DISP Y-DISP Z-DISP
1 0.000000 0.000000 -0.006026
2 0.000000 -0.025789 -0.003422
3 0.000000 -0.031555 -0.000484
4 0.000000 -0.028968 0.003016
5 0.000000 0.000000 0.007182

S T R E S S E S I N B E A M E L E M E N T
G R O U P 1
ELEM AXIAL FLEXURE STRESS AVG SHEAR
# STRESS NODE I NODE J STRESS
1 0. 0. -15625. 625.
2 0. -15625. -13750. -125.
3 0. -13750. -12500. -83.

Figure 3-7. Beam Element Stress Display

occurred. The designer's job here is to take the results from these analyses
and then do more detailed modeling of the exact configuration where the
cross-section change occurred to evaluate the potential for failure at that
location.
38 Chapter 3 Beams and Frames

3.5 Closure

The use of beam elements in models provides the engineer with the
opportunity to solve rather complex beam structures or frameworks that
could not easily be done with conventional approaches. It also can include
the effects of the stiffness of supporting structures through connection with
truss elements or beam elements selected to approximate the support
stiffness. In the case of statically indeterminant structures where the
supports might have different stiffnesses, the finite element model will
provide much better solutions than we can get by conventional approaches.
It also can provide the loading that exists at localized areas where cross-
section changes or member connections occur. Then we can use the loading
in much more detailed models of those regions.

Problems

3.1 The structure shown in Figure P3-1 has a horizontal steel beam
welded to a rigid column on the left and simply supported on the
right end. There is also a steel rod with pinned attachments to the
column and the beam providing support for the beam. The beam
cross section is shown on the right, and the rod diameter is 25mm.
Evaluate the effectiveness of the steel rod for reducing stress in the
beam by analyzing models with and without the rod and comparing
results.

3.2 Curved beams can be approximated by a group of straight beam


elements. Opposed diametral forces load the thin circular ring in
Figure P3-2. Use symmetry and determine the number of elements
required to achieve 5 percent accuracy for the maximum moment and
the displacements along and perpendicular to the loaded diameter.
Assume that the thickness, t = r/10, and that the cross section of the
Problems 39

ring is square. The maximum moment and radial displacements are


given by

Pr Pr 3 8 Pr 3 4
Ma  Ga  _  Gb    1
_ 8EI _ 4EI _

where a is the diameter parallel to the loads, and b is perpendicular


to the loads.

3.3 Analyze the bicycle frame design sketched in Figure P3-3. Use a
vertical load of 150 lb. at the seat location and 25 lb. at the handle-
bar location and apply a load factor of 2.5 for inertial loading.
Assume for the first analysis that all the members are tubular steel
with a 1-in. outside diameter and 0.062-in. wall thickness. From the
first analysis, determine if any yield failures are likely if the
material is a high-carbon steel with a yield strength of 110 kpsi. If
yielding will occur, refine the design by replacement of highly
stressed members with a more substantial section or by altering the
design layout to eliminate yield failures. If the frame is
overdesigned, refine the design to reduce weight. Do the deflections
seem excessive? Is there a specific location that seems to be too
flexible?

References

3.1 Cook, R. D., Malkus, D. S., and Plesha, M. E., Concepts and Applications
of Finite Element Analysis, Third Edition, John Wiley and Sons, New York,
1989.
3.2 Logan, D. L., A First Course in the Finite Element Method, PWS-KENT
Publishing Co., Boston, Massachusetts, 1986
3.3 Cook, R. D. and Young, W. C., Advanced Mechanics of Materials,
Macmillan Publishing Co., New York, 1985.
C H A P T E R 4

TWO-DIMENSIONAL SOLIDS

While the finite element method is very helpful for the solution of truss,
beam, and frame problems, the real power of the method shows in
application to two- and three-dimensional solid analysis. There are very
few closed form solutions to two-dimensional problems, and they are only
available for simple geometries and loading conditions. The finite element
method, on the other hand, if correctly applied, can provide the solution to
most any two-dimensional problem. The correct application is of prime
importance, and the analyst makes decisions involving the layout and
planning of the model to represent the member under analysis. The correct
application must be done to limit solution errors.
Equations derived in theory of elasticity govern the solution to problems
in two dimensions. The finite element formulation must satisfy, at least
approximately, the relations among displacements, strains, and stresses to
find a solution for general two-dimensional problems.

4.1 Element Formulation

There are two shapes of elements used in two-dimensional analysis: the


triangle and the quadrilateral. The two basic element shapes may be
linear elements or quadratic elements, where linear and quadratic refer to
the order of the assumed polynomial displacement interpolation function
used within the element area. The linear triangle is the simplest and was
the first two-dimensional element developed. Analysts do not use it much
now because it requires many more elements to produce a converged and
accurate solution compared with the quadrilateral. However, we still
examine its formulation both for academic purposes and for its occasional

40
Section 4.1 Element Formulation 41

use in coarse to fine mesh transitions for refining models.


The triangular element illustrated in Figure 4-1 defines an area bounded
by the three sides connecting three node points. Within the element area
the displacement function is assumed to be of the form in equation (4.1),

u  a1  a2 x  a3 y
(4.1)
v  a4  a5 x  a6 y

where u and v are displacement components of a material point within the


element field, x and y are coordinates of the point, and ai, i=1,2,...6, are
constant coefficients to be determined. This is a linear distribution of the
two displacement components for any material point within the element
area. The linear function has three undetermined coefficients for each
component, and since we have three nodes we may evaluate the three
constants by use of the node point values of each component.

Figure 4-1. Triangular Two-Dimensional Finite Element

Application of the strain-displacement equations to the expressions for


u and v illustrates that all three strain components are constant within the
element for this assumed displacement field as derived in equation (4.2).

ju
Ix   a2
jx
jv
Iy   a6 (4.2)
jy
ju jv
Exy    a3  a5
jy jx
Also, for homogeneous material throughout the element, the stress-
strain relations are all constant; therefore, the stress components are also
constant.
42 Chapter 4 Two-Dimensional Solids

This displacement formulation also satisfies the compatibility require-


ments in the theory of elasticity [4.1] for the continuum. The compatibility
requirements are that no gaps or overlaps of material may occur during the
process of deformation under load. From these equations, we can see that
the continuous nature of the function enforces compatibility within the
element. On a triangular element, since the interpolation is linear, any
edge formed by connecting two nodes that is a straight line before
deformation will remain a straight line after deformation. Therefore, any
connecting element using the same two nodes for its shared edge satisfies
compatibility.
Some of the early finite element programs [4.2] used the triangle
element to create a quadrilateral element by subdividing a quadrilateral
shape into four triangles using the centroid of the quadrilateral as their
apex. After finding the stiffness matrix for each triangle element, assembly
of the triangles and condensation of the internal node resulted in the
stiffness matrix of the quadrilateral element. This was an effective way to
use the triangular element formulation and employ many more elements
without tedious input. However, the element of choice now is an
isoparametric quadrilateral formulation.
Next we examine the displacement basis for formulation of the isopara-
metric quadrilateral element. Taig [4.3] developed the element, and Irons
[4.4] published its formulation. The quadrilateral element formulation
derives from the formulation of a square element. It uses a co-ordinate
system transformation to convert the square to a quadrilateral. Begin with
the square element shown in Figure 4-2 with corner nodes.

Figure 4-2. Square Two-Dimensional Finite Element

Recognizing that four constants can be evaluated with four nodes, a


logical expression for the displacement function components becomes
Section 4.1 Element Formulation 43

u  a1  a2 x  a3 y  a4 xy
(4.3)
v  a5  a6 x  a7 y  a8 xy .

Use of the strain-displacement relations here shows that

I x  a2  a4 y
I y  a7  a8 x (4.4)
Exy  a3  a4 x  a6  a8 y .

These equations show that the strain approximation within the element
allows Ix to be a linear function of y and Iy to be a linear function of x, while
Exy is a linear function in x and y. So by the addition of one node we have
gained a much better approximate solution within the element field for the
general case where the strains vary with both x and y throughout the
structure domain. Since the stress-strain relations are constant, the stress
components may vary similarly within the element field.
Also, in this case the function satisfies compatibility within the element
because the function is continuous. Along the element edges for x =
constant or y = constant the displacement takes a linear form, and thus
remains a straight line between any two of the corner nodes. Therefore
element connections to other elements satisfy compatibility as long as
corner nodes of one element connect to the corner nodes of the adjacent
element. Connection of two adjacent elements to a third element such that
the edge of the third element spans two of the adjacent elements' edges, as
shown in Figure 4-3, violates compatibility.

Figure 4-3. Two-Dimensional Element Compatibility Violation


44 Chapter 4 Two-Dimensional Solids

Before continuing the discussion of the square element, make a change


of coordinate systems to transform the square into a quadrilateral shape.
Square elements can geometrically model very few structures. A co-ordinate
transformation from x,y to [,M produces the quadrilateral element sketched
in Figure 4-4. We call the element an isoparametric quadrilateral because
the same interpolation functions (parameters) used to define the displace-
ment field define the geometric transformation.

Figure 4-4. Two-Dimensional Quadrilateral Element

4.2 The Finite Element Model

The model plan should begin by choosing the type of element for use. The
linear triangle element can easily develop into a mesh inside almost any
arbitrary geometry. However, to produce accurate results there must be
many of these elements in the model.
In the line element models we have covered thus far, there was little
reason for element subdivision other than to define the geometry of the
structure. However, two-dimensional cases require element subdivision to
achieve an accurate solution. Since element subdivision is required and the
exact solution is unknown, a sequence of models with successive mesh
refinement is proper. Mesh refinement by further and further subdivision
using compatible elements converges to the exact solution. This procedure
is known as h-convergence because h is a common symbol for step size in
numerical operations, and its reduction leads to convergence.
Reaching a refined solution by increasing the order of polynomial
approximation within the element is another way to achieve convergence.
This has become known as p-convergence. A direct conversion of a linear
element mesh to quadratic elements will yield a more accurate solution.
Section 4.2 The Finite Element Model 45

This is the first step in the p-convergence method for numerical convergence
on the correct solution. The user may easily use h-convergence by successive
model building in all finite element programs. However, there are usually
only linear and quadratic order elements in the element library of most
programs that limit the pursuit of p-convergence. There are several new
commercial codes becoming available now, and a recent text by Szabo and
Babuska [4.5] provides good coverage of the p-convergence method theory
and application.
In planning the mesh, try to use symmetry whenever possible. The
advantages include a reduction of labor of model input, reduction of
computer time and cost, and a decrease in computer round-off error in the
equation solution because fewer equations exist in the model. There are
some drawbacks. Sometimes it becomes more difficult to picture the model.
Also, peak stresses may occur along symmetry lines and make it difficult to
locate elements properly to show the peak.
Recognize symmetry in two-dimensional objects by observation of
geometric patterns that may occur. These may develop by incrementing
plane sections, rotating sections about an axis, periodically rotating sections
about an axis, or by reflecting a section about a plane. For the symmetric
model to provide a solution, the load distribution must also be symmetric on
the object. In some cases, we can find solutions for anti-symmetric loading
conditions on symmetric objects by proper imposition of displacement
boundary conditions.
Displacement boundary conditions enforce symmetry by restricting node
points that lie on lines of symmetry to motion along the line of symmetry.
For example, look at the simply supported beam with central load in Figure
4-5. It has a vertical plane of symmetry at coordinate x = 0.

Figure 4-5. Simple Beam with Central Load

When the load applies, the beam will deflect downward and the displace-
ment of every material particle in the right half will be a mirror image of the
corresponding particle in the left half. So if the body is symmetric before
46 Chapter 4 Two-Dimensional Solids

loading, it is also symmetric after loading. Then we only need to model one
half of this beam. If we take the right half, then the outline of the model is
in Figure 4-6. The model load reduces by one half because each half of the
beam carries its share. The node points that lie on the plane where x = 0 are
restrained against x-direction motion, but left free to move in the y direction.

Figure 4-6. One-half Simple Beam Model with Central Load

In a 2-D model, rigid body motion could occur by x translation, y


translation, or rotation about the z axis. Examine the displacement
restraints applied for symmetry and support conditions, and determine if
these restraints will automatically prevent these three rigid body motions.
If not, then we need to apply additional boundary conditions to assure that
rigid body motion does not occur. If there is a possible rigid body motion and
a net external force acting on the body in that direction, then we have not
recognized all the support restraints because the body is not in static
equilibrium. If there is no net external force, then we may select any single
node location for restraint to prevent that motion.
For example, in the simply supported beam in Figure 4-5, assume we
model the whole beam and the only restraints are vertical fixed displace-
ment at the two support points. Then the body would be free to move in the
x direction with a rigid body translation. To prevent this motion, select any
node and apply one x-direction restraint. This one restraint is adequate to
prevent rigid body motion. In fact, applying more than one restraint
artificially prevents the structure from displacing normally and therefore
falsifies the solution. Application of displacement restraints to prevent rigid
body motion should not induce any stress conditions in the body.

4.3 Computer Input Assistance


Section 4.3 Computer Input Assistance 47

At this point, the analyst should have roughly defined the model. In two-
dimensional analyses, we want to develop an adequate subdivision within
the area defined by the geometrical boundary. For simple geometries and
loadings, typically a regular array of elements will be suitable and are not
very difficult to create with simple replication schemes. However, for more
complex geometry involving hundreds or thousands of elements, we need the
aid of an area or two-dimensional mesh generator. Most programs provide
a mesh generation capability in their preprocessor.
One approach for irregular areas is to perform a coordinate transforma-
tion mapping from an approximate fit of squares in an integer geometry to
the actual physical geometry. This mapping may be done by laying out a
rough equivalent of the actual geometry in an integer space where each
square in the integer space corresponds to an element. This procedure is
illustrated in Figure 4-7.

Figure 4-7. Two-Dimensional Mapping Mesh Generation

Sides in the integer geometry correspond to sides in the actual geometry.


Sides in the integer geometry must be horizontal or vertical, while sides in
the actual geometry can be either straight or curved line segments at any
orientation. Correspondence of the closed boundary of the actual geometry
with the closed boundary of the integer geometry sets up the mapping. Use
of the difference equations, that result from a finite difference approximation
of the Laplace equation, determines the interior node locations.
The interior locations result from an iterative solution beginning with a
linear interpolation between boundary nodes. Sometimes, in areas of
sharply concave boundaries of the actual geometry, linear interpolation of
48 Chapter 4 Two-Dimensional Solids

nodes may fall outside the boundary and may not pull back inside the
boundary with the iterations. Therefore, the user must examine the
generated mesh carefully before proceeding to make sure the mapping was
successful. Additional examples of this type of mesh generation are shown
in Figure 4-8. In this approach, the user usually has some control over the
resulting bandwidth and wavefront by selection of starting location and
direction for node and element generation.

Figure 4-8. Other Mapping Mesh Generation Examples

4.4 The Analysis Step

In most two-dimensional analyses, there are many more nodes and elements
used than with truss, beam, and frame models. Therefore, there is more
potential for error in both the analysis execution errors and overall
numerical precision errors. If we checked the model thoroughly in the
preprocessor, then we should have caught most execution errors. Execution
errors arise by not preventing rigid body motion in the model, improperly
defining any element, entering incorrect material and physical properties,
and many other factors. The error messages presented by the program
usually identify these errors rather easily when they occur.
Numerical precision errors may come about through element distortion,
Section 4.4 The Analysis Step 49

Figure 4-8. (Continued)

element compatibility violations, and stiffness matrix ill-conditioning caused


by large differences in stiffness values of elements [4.6]. Examples of severe
element distortion of quadrilateral elements are shown in Figure 4-9.
Ideally the elements would remain close to square shape. High aspect
ratios, large differences in side length, and very small or large inside angles
all contribute to numerical precision errors. In fact, inside angles greater
than 180 degrees may cause negative stiffnesses. Most commercial
programs will check the element distortion and issue warning messages or
cancel execution if the distortion is too high.
Even though most two-dimensional element formulations guarantee
satisfaction of the compatibility requirements in theory of elasticity,
modeling errors may still violate compatibility. Some of these are illustrated
in Figure 4-10.
To satisfy compatibility in corner-noded elements, each element side may
only join to a side of one other element. For elements with midside nodes,
each must connect to another element at the corners, and all three nodes on
a side must connect by use of common nodes.
The solution accuracy in two-dimensional analyses is very dependent on
the user's ability to evaluate the results and produce a numerically
converged solution. In the truss and beam elements, the element formula-
tion was exact, and therefore there was no concern about interpolation
accuracy. However, these two-dimensional elements require numerical
50 Chapter 4 Two-Dimensional Solids

Figure 4-9. Severe Element Distortion

Figure 4-10. Model Compatibility Errors

convergence. We accomplish this convergence through careful evaluation of


the output and refinement of the model.

4.5 Output Processing and Evaluation

Completion of the analysis run will produce a listing file and data files for
graphic postprocessing. As mentioned before, scan the printout file for
errors in interpretation of input data. The data that represent element
selection and options, analysis conditions, material and physical properties,
and these types of data are relatively easy to scan. Obviously, we cannot
Section 4.5 Output Processing and Evaluation 51

easily check the lists of node point and element definitions.


Graphic display of two-dimensional data results is necessary to have any
chance at making a complete evaluation. Display graphics will be useful for
overall checking of model response as well as location of critical areas. This
allows examination of the detailed listing file for specific and accurate values
in these critical areas. The deformed shape plots have an exaggeration
factor adequate to see the deformation. Check boundary condition
enforcement, and make a visual judgment as to whether the deformation
agrees with the expected response. In some cases, the exaggeration factor
may need to be very high to understand fully how the structure is respond-
ing. It may help to visualize the shape you would expect if the material of
the actual structure were very soft and easily deformed.
Postprocessors normally present stress results in a contour plot form with
a range of stress levels. In examining the plots we can make some simple
checks. The boundary conditions of the problem require that any stress
component perpendicular to a free surface must be zero. Any stress
component perpendicular to a pressure-loaded surface must equal the
pressure value. Where a plane of symmetry exists, the stress contour lines
should become normal to the symmetry boundary. There should not be any
abrupt changes in contour line direction and it should be continuous. The
plot should have expected or understandable shape and location of peak
values. These plots result from stress component values at the node points.
However, in general, the values at a node common to several elements will
not be equal. So there must be some manipulation of the computed stresses
to produce these contour plots.
Upon examination of the stress contours, the user must make some
judgment about the validity of the solution. With only results from one
model, we can never be sure that we have a converged or accurate solution.
The plan is then to produce multiple models using more refined meshes until
the solution has converged. The two-dimensional elements presented in this
chapter have mathematically guaranteed convergence to the exact solution
as the element size approaches zero. It is then appropriate to run multiple
solutions with refined meshes to estimate when the convergence has
occurred.
Although the individual stress components may be of interest, the failure
criterion should not be the maximum normal stress theory for ductile
materials. Its use can result in serious error when the minimum normal
stress is of the opposite sign. The maximum shear stress failure criterion is
accurate, but be sure to use the true maximum shear stress and not just the
maximum in-plane value. The Von Mises equivalent stress based on the
distortion energy theory is considered to be the most accurate for ductile
materials. If the material is brittle or a composite or some other class of
material, then the user must determine what failure criterion is proper to
use for the given material.
52 Chapter 4 Two-Dimensional Solids

4.6 Case Study

The case study analyzes a flat bar in tension with a central hole as a typical
stress concentration problem. We will analyze this case with the finite
element method and compare the results with the theoretical stress
concentration factor. The geometry is shown in Figure 4-11. By taking
advantage of symmetry, a one-quarter shaded section of the bar defines the
model geometry.

Figure 4-11. Flat Bar with a Central Hole in Tension

The plan for the first finite element model shown in Figure 4-12 has a
more refined mesh near the hole because the stress is naturally higher in
that area with steeper slopes of change. Displacement restraints apply to
the vertical symmetry edge to prevent displacement in the horizontal
direction and to the horizontal symmetry edge to prevent displacement in
the vertical direction. Node forces calculated and distributed on the right
edge provide a uniform stress there of 1 kpsi.

Figure 4-12. First Finite Element Model


Section 4.6 Case Study 53

After running the analysis, a contour plot of the x-direction stress


component appears as shown in Figure 4-13. The contour lines labels
correspond to the legend on the left. The element outlines are also shown.
The maximum level occurs at the edge of the hole as expected. A clearer
view of this region is shown in the enlargement of Figure 4-14. As
mentioned before, the finite element method gives an approximate, not an
exact, solution. An estimate of the error in the analysis is the range of stress
change relative to the average element value across an element. In this case
the corner element includes contour levels from 6 to 9 with a total range of
about 1500 psi, or about 750 psi from the average. The estimated error is
23 percent.

Figure 4-13. Contour Plot of the X Stress Component

Figure 4-14. Zoom View of the X Stress Contour Plot


54 Chapter 4 Two-Dimensional Solids

This error margin is large, so we should refine the model. A second model
produced was still not sufficiently accurate, so we created a third model.
This model is shown in Figure 4-15, and it is much more refined around the
hole. The zoom view of the x stress component is given in Figure 4-16. We
now have a stress range in the corner element of about 700 psi, or about 350
psi from the average, for a 9 percent estimated error. The maximum value
at the edge of the hole is 4300 psi. The nominal or average stress on the
reduced area section at the hole is 2000 psi which gives a stress concentra-
tion factor of 2.15. The theoretical stress concentration factor is 2.18, so the
actual error is only -1.4 percent.

Figure 4-15. Third Finite Element Model

Figure 4-16. Zoom View of the X Stress Contour Plot (Third Model)
Section 4.6 Case Study 55

4.7 Closure

The application of the finite element method to two- and three-dimensional


problems is where its power is really useful. There are very few closed form
solutions to these problems, especially for any but the simplest of geome-
tries. As shown by this case study, the engineer can reach a very accurate
solution by application of proper techniques and modeling procedures. The
accuracy is usually only limited by our willingness to model all the
significant features of the problem and pursue the analysis until we reach
convergence.

Problems

4.1 Show the convergence of finite element models of a simply supported


beam with a uniformly distributed load. Refine meshes to determine
the maximum displacement and maximum stress within about 5
percent accuracy. Use a beam length of eight times its height and a
unit thickness. Perform the study using each of the following element
types if available in the computer program:
(a) linear triangle,
(b) linear quadrilateral,
(c) parabolic triangle, and
(d) parabolic quadrilateral.
Referring to Figure P4-1, the flexural stress and the center
deflection including shear deformation are listed in the following
equations.

3qL 2
cx 
4bc 2
5qL 4 3qL 2(1  Y)
G  
16Ebc 3 5Ebc

4.2 A machine member illustrated in Figure P4-2 carries an end load.


Simple beam equations furnish the nominal flexure stresses in this
member but do not account for stress concentration effects. Deter-
mine the stress concentration factors for the two underside radius
locations. How do these compare with the values found for similar
geometries in many mechanics of materials or machine design books?
56 Chapter 4 Two-Dimensional Solids

4.3 Analyze the two tensile bar configurations shown in Figure P4-3.
Compare your results with published stress concentration factors. Is
there a significant interaction effect between the two geometrical
discontinuities for the bar in (b)?

4.4 Analyze the tensile loaded bar with an off-center hole shown in Figure
P4-4. Compare results with the closest published results available.
Make successive models with the hole moving closer to the side and
see if any pattern develops.
References 57

4.5 Determine the stress concentration factor(s) for the notched bar with
a center hole in Figure P4-5. Compare with published results for the
individual geometries and evaluate any interaction caused by their
proximity. You may wish to evaluate additional values of notch
radius or varying depths of a constant notch radius.

References

4.1 Timoshenko, S. and Goodier, J. N., Theory of Elasticity, McGraw-Hill, New


York, 1951.
4.2 Wilson, E. L. and Jones, R. M., Finite Element Stress Analysis of Axisym-
metric Solids, Aerospace Corporation, San Bernardino, CA, 1967, Air Force
Report No. BSD-TR-67-228, (AD-820991, N.T.I.S.).
4.3 Irons, B. and Ahmad, S., Techniques of Finite Elements, Ellis Horwood
Limited Publishers, West Sussex, England, 1980.
4.4 Irons, B. M., "Engineering Applications of Numerical Integration in
Stiffness Methods," AIAA J., Vol. 4, No. 11, 1966, pp. 2035-2037.
4.5 Szabo, B. and Babuska, I., Finite Element Analysis, John Wiley and Sons,
New York, 1991.
4.6 Cook, R. D., Malkus, D. S., and Plesha, M. E., Concepts and Applications
of Finite Element Analysis, Third Edition, John Wiley and Sons, New York,
1989.
APPENDIX

This user's guide is written to describe the operation of the FEPCIP,


FEPC, and FEPCOP programs. It assumes the user has been instructed
in use of the finite element method to solve stress analysis problems. The
programs will be described with their capabilities and general sequence of
use for solving a problem. The first-time user should refer to the README
file on the program disk for initial setup. In some parts of this guide the
user is referred to more complete documentation in the file on the program
disk.
These FEPC programs are provided as shareware for educational
purposes. They are copyrighted programs and you are not authorized to
sell or distribute them for COMMERCIAL purposes. You are free to use,
copy, and distribute them for NONCOMMERCIAL uses only if no fee is
charged for use, copying, or distribution. Specifically, the programs were
designed for use by students in university courses.
If you are an instructor and use the programs in a university course, I
would appreciate a simple registration (no fee) of your university, course
name or description, and approximate number of students taking the
course.
If you use the programs for COMMERCIAL purposes, i.e., employed
engineering, consulting, sponsored research, etc., a partial registration fee
to continue the software development would be appreciated. For a full
registration fee, you will receive the latest version, fully dimensioned, to
run on an IBM-PC or compatible with 640K RAM. Please state the current
version number of the software you are using. Send inquiries to:

Dr. C. E. Knight
Professor of Mechanical Engineering
914 Ballard Ct.
Blacksburg, VA 24060

The programs are continually under development and your comments


concerning present features or future enhancements would be appreciated.

58
Appendix 59

USING FEPC, FEPCIP, AND FEPCOP

FEPC is a program that performs finite element stress analysis of two-


dimensional truss, beam, plane solid, or axisymmetric solid structures.
There are two companion programs. FEPCIP is the FEPC INPUT
PROCESSOR that is used to input and check a model and prepare data
files for FEPC. FEPCOP is the FEPC OUTPUT PROCESSOR that reads
FEPC output data files and produces graphic displays.
The shareware version of the FEPC programs are currently dimensioned
to run in a PC with 640K memory. The dimension limits in the programs
are 600 nodes, 600 elements, 10 materials, 100 points, 30 lines, and 20
arcs. Also, the overall model size is limited in FEPC based on number of
nodes and average nodal bandwidth. For example, 600 nodes with an
average nodal bandwidth of 15, 500 with 18, or 400 with 24 are all
maximum model sizes that can be run in FEPC. Dimension limits of the
automatic mesh generation grid in FEPCIP are I=25 by J=60. Therefore,
a model can be built in FEPCIP even within the 600 node and element limit
that is too large to run in FEPC, so plan carefully.
The procedure for solving a problem is to run FEPCIP to create the
model, run FEPC to solve the equations, and run FEPCOP to display the
results. The FEPCIP program presents the user with menus for interactive
input, checking, and storing a model. This creates an analysis file used as
input for FEPC. Running FEPC produces a listing file of printout results
and files of results used as input for FEPCOP. Graphic displays of de-
formed shape and stress plots may then be produced by FEPCOP.

ENTERING THE MODEL IN FEPCIP

Before starting to enter the model, develop a node and element numbering
plan, boundary conditions, and the load placement for the model. With the
FEPCIP.EXE file in the current drive and directory, begin by typing

FEPCIP<CR>

where <CR> means to press the enter or return key. After the FEPCIP logo
appears, the program will continue after a short pause.
The screen will clear and the program will automatically detect the
proper graphics mode for the supported graphics cards.
After making the selection the main menu and graphics windows will
then appear along with a prompt to SELECT A FUNCTION KEY.
60 Appendix

INPUT PROCESSOR FINITE ELEMENT PERSONAL COMPUTER DATE TIME


TITLE:

F1 FILES MODEL
F2 MODEL DATA SUMMARY
F3 2D AUTOMSH WINDOW
F4 TITLE
MODEL
F6 CLEAR MEM GRAPHICS
F7 EXIT WINDOW
F8 VIEW OPTS
F9 DSPLY OPTS

SELECT A
FUNCTION KEY

Selection of a menu item by its function key brings up a branch menu for
many of the selections.
Key F1 branches to a menu for recalling a previously stored model,
storing a new model, or adding a title.
Key F2 branches to a menu for entering or editing all data required for
the model.
Key F3 branches to a menu for two-dimensional area mesh generation.
Key F4 prompts the user to input a title for the current model.
Key F6 will clear all the current model data from memory in order to
start entering a new model.
Key F7 exits the program.
Key F8 branches to a menu to change the current view of the model.
Key F9 branches to a menu to change the visibility of entities (nodes,
elements, loads, etc.) or labels (node numbers, element numbers)
on the next redraw of the model.
Every branch menu has a function key selection to return to the previous
menu. Many of the selections on the branch menus will branch to
additional menus. In each case, following completion of tasks on the
current menu, use the previous menu selection to step back through the
menus until the modeling is complete.
The general procedure for entering a model is to use the MODEL DATA
function key to access the menu for selecting the element type, defining the
material properties, defining nodes and elements, setting node displace-
ment restraints, and applying loads. For truss and beam element models,
all the model data are entered from this menu and its branch menus.
Two-dimensional solid models using plane stress, plane strain, or
axisymmetric elements may first use the 2D AUTOMSH selection to
generate the model mesh of nodes and elements. Once the nodes and
Appendix 61

elements are defined, return to the model data menu to complete the model
by material definition, setting node restraints and loads.
The model must be stored on a disk file before exiting the
program. The model data may be saved to disk at any time in the
progress of building the model. Two files are stored for all complete models
under a user-specified filename with file extensions of .MOD and .ANA. If
the model is incomplete only the .MOD file is stored and messages denoting
the data yet to be defined for the .ANA file are displayed. All the current
model data is saved in the .MOD file.
The program operates by using the function keys to select the operation
from the menu. When data is required, a prompt appears on the data entry
line just below the TITLE: header. The user types in the requested data
separated by commas or spaces, followed by the carriage return or enter
key.
When the prompt to detect an entity appears on the data entry
line, a cursor will appear. If a mouse exists and its driver is loaded,
then use the mouse to position the cursor and press the left button
to detect or the right button to abort.

FILES

F1 RCL FN.MOD Selecting FILES from the main menu branches to the
F2 STO FN.MOD submenu on the left. A previously formed and stored
& FN.ANA model may be recalled from disk by selecting F1 RCL
FN.MOD. The user is prompted to enter the filename,
FN, without its .MOD extension. The filename may
include the drive designation and path, but it may be a
F10 PREV MENU maximum of 20 characters long including the drive
designator characters. DO NOT enter any leading
SELECT A
FUNCTION KEY blank spaces in the input of the filename. If the file
cannot be found, an error message is displayed.
The current model may be stored on disk by selecting
F2 STO FN.MOD & FN.ANA. The model currently in memory will be
stored in FN.MOD assuming no errors. Also, if the model is complete and
ready for analysis, the input file for the FEPC program will be stored in
FN.ANA. If the model is incomplete, the user is given messages indicating
which data are missing. The store operation may be done at any time
during the progressive construction of the model in order to have a place to
restart in case of destruction of the current model data in memory. If the
files FN.MOD and FN.ANA already exist on the disk, they may be
overwritten with the user's consent by the current data in memory each
time the store function is executed.
After completing use of this branch menu, select F10 PREV MENU to
return to the main menu.
62 Appendix

MODEL DATA

Begin entering a new model by selecting F2 MODEL DATA on the FEPCIP


main menu. This produces the branch menu shown
F1 ELEM TYPE below.
F2 MATL PROP
If a truss or beam element model is to be entered, then
F3 NODE DEF
F4 ELEM DEF all the model data will be entered from this menu. If a
F5 RESTRAINTS two-dimensional solid is to be entered, then all the data
F6 LOADS may be entered from this menu, or the 2D AUTOMSH
selection may be used to generate the nodes and elements
F8 VIEW OPTS
F9 DSPLY OPTS for the mesh. Once these are generated they may be
F10 PREV MENU edited from this menu. If 2D AUTOMSH is to be used it
should be done first, or after the element type is selected
SELECT A and material property sets are defined; it will overwrite
FUNCTION KEY
any existing node and element definitions.

ELEMENT TYPE

The F1 ELEM TYPE selection displays the list of available elements.


Use the indicated function key to select the element for the model. Only
one element type may be used in a model. After selection the program
returns to the previous menu. See the documentation file on the program
disk for more detailed instructions.

MATERIAL PROPERTIES

The F2 MATL PROP selection branches to a menu for input and query
of material set definition. Up to 10 material property sets may be defined
and should be defined in numerical order. Material property sets may
include some physical properties depending on the element type. Each
element in the model has a material set number associated with it that
defines its material properties. If different material or physical properties
exist in different parts of the structure, then multiple material sets should
be defined before elements are defined so correct assignments may be made
at the time of element definition. See the documentation file on the
program disk for more detailed instructions.

NODE DEFINITION

The F3 NODE DEF selection branches to a menu to perform node


operations. Nodes for truss and beam element models all will be defined
in this section. If the 2D automesh option is used for the 2D plane and
axisymmetric models, then that should be done first and any additional
node operations will be done in this section. Node operations include
definition, generating a row of nodes between two defined nodes, moving,
Appendix 63

deleting, and querying nodes.


A node is defined by its number and coordinate position. A prompt will
appear on the data entry line to input a node. Simply enter the node
number and its X and Y coordinates. The data must be separated by
commas or spaces. All leading blanks are ignored, but do not enter any
trailing blanks. If the node is generated properly, it will be displayed on
the graphics screen if it lies inside the current window. The starting
window is 10 units by 10 units, but it will change automatically if any of
the view options are exercised. Autoscale will resize the window so that all
currently defined nodes fit inside. The prompt recycles so that the next
node may be input. Terminate input by tapping the return or enter key.
See the documentation file on the program disk for more detailed instruc-
tions.

ELEMENT DEFINITION

The F4 ELEM DEF selection branches to a menu to perform element


operations. Elements for truss and beam element models all will be defined
in this section. If the 2D automesh option is used for the 2D plane and
axisymmetric models, then that should be done first and any addi-tional
element operations will be done in this section. Element operations include
selecting material, definition, generating a row of elements from a starting
element, modifying, deleting, and querying elements.
Single elements are defined by user selection of nodes for each element.
The user is prompted to detect each node needed for the element definition.
The order of node selection on two node elements is of no consequence. The
nodes for four node elements must be picked in a counterclockwise
order surrounding the element area. Elements are numbered in
numerical order as they are defined. Their material set assigned is the
current material set. Each element is drawn on the graphics screen as it
is defined. The node prompts recycle to define the next element, and will
continue until the right mouse button is pressed or the return or enter key
is pressed at the node detect prompt to terminate element definition. See
the documentation file on the program disk for more detailed instructions.

RESTRAINTS

The F5 RESTRAINTS selection is for applying node displacement


boundary conditions. By default all nodes displacement components are
free to take on nonzero values appropriate to the structure response under
load. The components that must be zero for the model to behave properly
are specified to be fixed. The menus that appear allow the user to set
values for the restraints and then pick the nodes to which the set values
apply. See the documentation file on the program disk for more detailed
instructions.
64 Appendix

LOADS

The F6 LOADS selection is for applying loads to the model. Loads may
be node forces or element edge pressures (for 2-D solid elements). The
menus that appear for node forces allow the user to set component values
for the loads and then pick the nodes to which the set values apply. Edge
pressure is applied by input of the pressure value and then selection of the
elements to which the set pressure applies. Menu selections also allow
deletion or query of forces and pressures. See the documentation file on the
program disk for more detailed instructions.

2-D AUTOMESH GENERATION

This section of the program is used for area mesh generation of two-
dimensional plane stress, plane strain, or axisymmetric models. The
principle of the approach is a mapping of an integer area grid into the
geometrical area of the model. The geometrical area is defined using point
locations, lines, and arcs. The perimeter of the geometrical area is defined
by the complete set of lines and arcs that enclose the area.
The integer area grid will have lines that correspond to the lines and
arcs of the geometrical area. Plan the correspondence by imagining or
physically sketching on square grid paper the perimeter in the integer area.
Use integer coordinates I and J with a range of 1 to IMAX and 1 to JMAX,
respectively. IMAX and JMAX values are listed in the first section of this
guide. A 1-by-1 square in the integer area grid will map to an element in
the geometrical area. Grid points in the integer area grid will map to node
points in the geometrical area model.
Lines in the integer area can only be lines of constant I or lines of
constant J. The perimeter must be defined by a head-to-tail connection of
lines in a counterclockwise(ccw) direction around the area. The length of
line in the integer area is equal to the number of elements desired along
the corresponding line or arc in the geometrical area.
The process involves defining the geometrical points needed to describe
the model area, then defining lines or arcs using those points that complete
the model perimeter. Next, plan the corresponding integer area grid to be
mapped into the geometry of the model.
After all the geometric points, lines, and arcs have been entered, area
mesh generation can begin. The genmesh function presents a prompt to
pick the starting point of the area. This point on the geometry will
correspond to the 1,1 I,J coordinate location on the integer area. A series
of prompts then proceeds for the detection of a line or arc, the number of
elements on that line or arc, and the direction of the corresponding line in
the integer I,J area.
The first line or arc detected must have the selected starting point as one
of its endpoints. The next line or arc picked must have the other endpoint
Appendix 65

of the first line or arc as one of its endpoints. Each successive line or arc
picked must then connect to the other endpoint of the previous line or arc.
This continues until the perimeter of the geometry is closed and the
endpoint of the last line in the integer area must be back at the starting
point; i.e., the perimeter in the geometry area and the perimeter in the
integer area must close simultaneously. Both of these perimeters must
progress ccw around the area.
When a line or arc is picked, the entry of number of elements determines
the length of the line in the integer area. The direction entry chooses one
of four allowable line directions in the integer area. The directions are
labeled 1, 2, 3, and 4, which correspond to right(+I), up(+J), left(-I), and
down(-J), respectively, in the I,J coordinates.
The integer area of the model must lie in the positive quadrant of I,J
coordinates. Since the starting point in the integer area is at 1,1, and the
perimeter must be ccw, the direction for the first line or arc must be 1. The
direction for the second line or arc picked may be 1 or 2. Successive lines
may have any direction values as long as some lines with directions 1 and
2 are used before any with directions 3 or 4, so that the I and J coordinate
values always remain positive. The total number of elements on all lines
in the 1 direction must match the total number in the 3 direction, and the
total number in the 2 direction must match the number in the 4 direction.
The bandwidth of the structure stiffness matrix is minimized by making
the number of elements in the I direction smaller than in the J direction.
The limits are IMAX-1 elements in the I direction and JMAX-1 elements in
the J direction. However, no model may have more than the maximum
number of nodes or elements listed in the first section of this guide.
Mapping is an iterative process of distorting the integer area to fit in the
geometry area. After a few iterations a mesh will be drawn on the screen.
If it appears to be suitable then it can be accepted, or more iterations may
be requested to make it smoother. If it is unacceptable
F1 POINT then a different integer area may be tried.
F2 LINE The menu of functions for mesh generation is
F3 ARC
F4 GENMESH POINT

F8 VIEW OPTS Points are used to define lines and arcs that make up
F9 DSPLY OPTS the model's geometric perimeter. Two points are needed
F10 PREV MENU to define a line, and three points along the arc are needed
to define an arc. Points are input by their coordinate
SELECT A
FUNCTION KEY location. Selection of key F1 POINT brings up a submenu
for creating, modifying, or deleting points. See the
documentation file on the program disk for more detailed
instructions.

LINE
66 Appendix

A straight line may be used to represent all or part of any straight edge
on the model. More than one line on an edge might be used to produce
different element spacings along the edge. If two or more lines are used on
any single edge, they should be connected in series with no overlap.
Selection of key F2 LINE produces a submenu for creating, modifying, or
deleting lines. A line is defined by picking two points at the ends of the
line.

ARC

An arc may be used to represent all or part of any circular arc on the
model up to 180 degrees included angle. If more than one arc is used on a
circular arc of the model, then they should be connected in series. Three
points along the arc are needed for the definition. They are the two
endpoints and an intermediate point. Another point is created during
definition of the arc at the arc's center of curvature. This may cause the
autoscale function to reduce the model scale substantially if the arc radius
is very large, in order to fit all the points on the graphics screen. Selection
of key F3 ARC produces a submenu for creating, modifying, and deleting
arcs.

GENERATE MESH

Selection of the F4 GENMESH function begins a series of prompts and


inputs to define the meshing area. If the element type has not been
selected, the element menu will be presented for a choice. If more than one
material set has been defined, then the prompt to enter material set
number will appear. Enter the set number to be assigned to all the
elements defined using the mesh generator.
Following these conditional entries the prompt to detect the start point
appears. This is a geometric point on the model that corresponds to the 1,1
point in the I,J integer area. Next, the prompt to detect a line or arc begins
the sequence of perimeter definition. Detect the line or arc that connects
to the starting point and starts on the ccw path around the perimeter by
positioning the cursor on the line or arc center and pressing the spacebar
or the right mouse button. Following detection, enter the number of
elements along the line or arc. Then enter the direction number of the line
in the integer area (1, 2, 3, or 4). The set of prompts to detect line or arc,
enter number of elements, and enter direction all cycle until the user
terminates input by entry of the return key.
The user should be sure the geometry perimeter is closed before
terminating. The program checks that the integer area is closed, and if so
begins iterating on the mapping. This may take a few minutes. If the
integer area is not closed, a program message reports this condition and the
geometry is redrawn. Another trial to input the perimeter may begin with
Appendix 67

selection of the automesh function.


If the integer area did close, after a few iterations the mesh is drawn on
the screen with the prompt for more mesh iteration (Y or N). If it needs
additional smoothing, enter Y. More iterations will then be done and the
prompt will reappear. If it looks acceptable or is to be redone differently,
then enter N.
The next question asks if it's OK to keep (Y OR N). Enter Y to keep the
mesh, or enter N to discard this mesh and redo the GENMESH function
with another plan.
Once an acceptable mesh of nodes and elements is kept, return to the
MODEL DATA menu to apply the displacement boundary conditions and
loads, and perhaps define a material property set.
When the model data is complete, go to the files menu and store the
model and analysis files to disk. This also should be done periodically
during building of the model in case of unexpected termination of the
session or a different path chosen to complete the model.

VIEW OPTIONS

This selection appears on many of the branch menus to allow exercising


the view options without retracing the menus to reach them. The fol-
lowing menu appears on selection of view opts.
F1 AUTOSCALE
F2 ZOOM AUTOSCALE
F3 MAGNIFY
F4 CENTER
Selection of autoscale automatically scales the graphics
window to include all currently defined points and nodes.

ZOOM
F10 PREV MENU

SELECT A This function allows the user to select a portion of the


FUNCTION KEY graphics window which is then scaled to fit the full
window. The user is prompted to detect the two corners
of the zoom area.

MAGNIFY

This option changes the size of the model displayed. The user is
prompted to enter the magnification factor. A positive value must be
entered; values larger than one will increase the size of the drawing and
values smaller than one will decrease the size.
Subsequent use of the magnify command will enlarge (or decrease) the
model display with respect to its current size. For example, magnifying
your model by two and then by three produces an image six times larger
than the original.
68 Appendix

CENTER

The model may be moved by selecting a new center of the graphics


window. The user is prompted to locate the new center.

DISPLAY OPTIONS

Display options control which entities and labels are visible when a
graphics plot is done. A branch menu appears.
F1 ENTITY SW
F2 LABEL SW ENTITY SWITCH
F4 MONO/COLOR
An entity switch setting is off or on to control the
individual entity's visibility. Selection of this function
brings up a submenu listing all the entities and prompt-
F10 PREV MENU ing the user to select one.
Selection of a function key will produce a prompt to
SELECT A
FUNCTION KEY change its current setting by default. A Y or return key
entry will accept the prompt question.

LABEL SWITCH

This function controls the display of labels (numerals) for nodes and
elements on the graphic model. A submenu allows selection of a function
key producing a prompt to change its current setting by default. A Y or
return key entry will accept the default.

MONO/COLOR

This function switches the display between black and white or color
when the computer has a color graphics board. Switching the color to black
and white allows the screen graphics to be dumped to a black-and-white
printer without the loss of character intensity that sometimes happens in
such screen dumping. Since only the drawing color palette is changed with
this switch, the change occurs when the next drawing is done after the
switch. Execute the function key again to return to a color display.

THE ANALYSIS BY FEPC

When a model has been developed and saved, it is complete and ready to
Appendix 69

be processed by the finite element processor, FEPC. After exit from


FEPCIP, and before starting FEPC, be sure that the filename.ANA file can
be accessed by FEPC. Copy it to the FEPC diskette in the same directory
where FEPC.EXE resides, or use the drive and path designation in the
filename.

RUNNING FEPC

With the FEPC.EXE file in the current drive and directory, begin by
typing

FEPC<CR>

After the FEPC logo appears, a prompt will appear to enter the model
filename (20 characters max), with drive and/or path designation but
without the .ANA extension.
As the computations proceed, messages will appear on the screen
reporting the computation step in progress. If errors occur, error messages
will also appear on the screen. FEPC creates some other files as it runs.
There is a listing file of all the printed output labeled filename.LST. This
file should be studied by the user after an analysis to check the input data
interpreted by FEPC and all the numerical output. A file labeled file-
name.MSH stores the node and element data for FEPCOP. A file labeled
filename.NVL stores the node displacement and element stress data for
FEPCOP.
Some other files are also created during the FEPC run; these are deleted
upon normal termination of the program so the disk that stores the .ANA
file must have some excess space for these files during runtime. If the run
terminates abnormally some of these files may still be on the disk with
extensions of .ELM and .LOD. These and other output files will be
overwritten when running a model with the same filename.
If the FEPC run was successful, then FEPCOP may be used to display
the results in graphic form. If the run was not successful, then examine the
filename.LST file for data errors or error messages that may help to correct
the model.
The output from a FEPC run is stored in a listing file called file-
name.LST, where the filename is the same as the model file name entered
when beginning the FEPC analysis. This file includes a listing of all the
input data as well as all the numerical results.

FEPC ERROR MESSAGES


1 - OUT OF SPACE, MODEL IS TOO LARGE (I)
The model is too large to run in FEPC. Reduce the model size in FEPCIP
and try again. Consult the program limits given before.
70 Appendix

2 - NODE 'n' HAS BEEN PLACED ON AN INCLINED BOUNDARY, BUT IT IS


ALREADY CONSTRAINED AGAINST X OR Y DISP. OR BOTH
An inclined boundary angle is specified for node n, but an x or y restraint was
also specified which is incompatible. Edit the model in FEPCIP.
3 - FEPC.EXE file not found
The FEPC.EXE file must reside in the current drive and directory to execute.
4 - 5 No longer used.
6 - 'm' ELEM IS HIGHER THAN NO. OF ELEMENTS IN THE GROUP
The filename.ANA file has been corrupted because element number m is
higher than the total number of elements. The *.ANA file is an ASCII file so
it may be printed or edited. Examine its contents in comparison with the
data printout in the filename.LST file.
7 - ELEMENT NO 1 IS NOT DEFINED FIRST
The filename.ANA file has been corrupted because element number 1 is not
defined first in the list of element definitions. The *.ANA file is an ASCII file
so it may be printed or edited. Examine its contents in com-parison with the
data printout in the filename.LST file.
8 - YOU HAVE A ZERO LENGTH ELEMENT #'m'
A beam or truss element is defined using the same node for both ends or the
two nodes defining the element have coincident coordinate locations.
9 - BAD ELEMENT #'m'
A quadrilateral element is improperly defined or is too distorted. Check for
cw node order definition around the element ( it should be ccw), inside
angles between sides greater than 180 degrees, butterfly-shaped element,
or a triangle formed by using one node for two corners (this is legal if the last
two nodes in the element definition are the same).
10 - STIFFNESS MATRIX NOT POSITIVE DEFINITE, NEGATIVE STIFFNESS
DIAGONAL TERM FOR EQUATION 'n', VALUE = '#'
During solution of the system equations, a negative diagonal term is found
which means that the equations cannot be solved. The equation number
corresponds to the free node degree-of-freedom in the system ordered
consecutively with node numbers. These are listed in the filename.LST file
produced in the FEPC run. Find the node number from this list, then
examine the elements defined using this node number for errors. If the
equation number is 1 or the last equation number, then the error is probably
due to lack of sufficient displacement restraints to prevent rigid body motion.
11 - INCLINED BOUNDARY ANGLE MUST BE BETWEEN -89.99 AND +89.99 DEGR
The inclined boundary angle input is outside the allowable range.

GRAPHIC RESULTS USING FEPCOP

After a successful run by FEPC, results files, filename.MSH and file-


name.NVL, will have been created on the disk. These are the input files for
output processing by FEPCOP.
With the FEPCOP.EXE file in the current drive and directory, begin by
typing

FEPCOP<CR>
Appendix 71

where <CR> means to press the enter or return key. After the FEPCOP
logo appears, the program continues after a short pause.
The screen will clear and a prompt will appear to

ENTER MODEL FILE NAME (NO EXT) -

Enter the filename (20 characters max), with drive and/or path designation,
but without the .MSH or .NVL extension. Some messages will appear
noting the progress of calculations, and then the screen will clear and the
FEPCOP MAIN MENU is displayed along with a prompt to SELECT A
FUNCTION KEY.
F1 DEFORMED Selection of F1 DEFORMED brings up a branch menu.
F2 X-STRESS Selection of F1 PLOT in this branch produces a deformed
F3 Y-STRESS
F4 XY-STRESS
shape plot of the element mesh superimposed over the
F5 T-STRESS undeformed model. This plot shows the finite element
F6 VON MISES mesh when the node displacements are scaled and added
F7 TRUSS STRS to the node coordinates so that the deformed shape is
F8 BEAM STRS exaggerated. In truss and beam models the deformed
F9 OPTIONS
F10 EXIT
mesh is superimposed over the undeformed mesh plot.
In 2-D solid models the deformed mesh is superimposed
SELECT A over the outer boundary of the undeformed shape. The
FUNCTION KEY displacement scale factor may be changed in the OP-
TIONS menu to increase or decrease the plotted
deformation. An additional submenu appears for view
options of the plot.
Selection of F2 ANIMATE in the branch menu produces a sequential
mesh plot of truss and beam models or a boundary outline plot of 2-D
models, showing the progressive deformation as the load is applied cyclicly.
Press any key to terminate the animation.
The next five function key selections on the main menu show the stress
contour plots developed in 2-D solid models for the indicated components
of stress. Each function is accompanied by a legend of the contour values
and the view options menu. The T-stress component is nonzero only for the
axisymmetric element models and represents the hoop stress in the
axisymmetric structure. The Von Mises equivalent stress is calculated
based on the distortion energy failure theorem using all the stress
components calculated in the loaded model.

TRUSS STRS

This function is used to display the results in truss element models. The
user may select a plot of the axial force or stress in all truss elements. The
plot is in a bar chart format with the heights scaled to the maximum value
in any element. Plus or minus signs are drawn on the bar near the top to
indicate whether the member is in tension or compression.
72 Appendix

BEAM STRS

This function is used to display the results in beam element models. The
user may select to plot the axial, flexure, average transverse shear, or the
maximum combined axial plus flexure stress. These are also in bar chart
format with signs indicated near the top of each bar. The axial and
transverse shear stresses are constant along an element length, so one bar
per element is sufficient. The flexure stress component varies linearly
along the element length, so a bar is plotted for the value at each end. Two
bars are also plotted for the combined axial plus flexure stress. The sign
of the combined stress is the same as the sign of the axial stress which is
the combination producing the largest magnitude.

OPTIONS

F1 NODE SW Function key F9 OPTIONS produces a submenu.


F2 ELEM SW Selecting F1 adds node symbols to the 2-D stress plots,
F3 DISP SCALE and F2 adds element outlines inside the 2-D boundary
for stress plots. Both of these selections produce a
F10 PREV MENU prompt to switch the current setting. Answering Y or by
default a return key entry will make the change.
Selection F3 allows the scale factor for the deformed
SELECT A shape plots to be changed by prompting for a new scale
FUNCTION KEY
factor, with the current scale factor shown as the default
value. Enter a larger value to increase the exaggeration
or a smaller value to decrease it.
INDEX

Banded matrix 19 precision 48


Bandwidth 19, 48 reduction 19
Boundary condition roundoff 19
beam 34 runtime 19
Boundary conditions sources of 2
displacement 16 Euler buckling 21, 34
Compatibility 42, 43 Force
violation 43, 49 external 6
Computer program node 5
diagram 7 reaction 6
Coordinate transformations 12 Gauss elimination 19
Deformed shape 20, 22, 34, Gaussian elimination 9
35, 51 H-convergence 44
Elasticity Ill-conditioning 49
theory of 42 Interpolation functions 44
Element Mesh
distortion 48 refinement 44
frame 32 Mesh generation
isoparametric quadrilateral mapping 47
42 P-convergence 44
linear triangle 40 Postprocessor 7, 50
matrix 4 Preprocessor 7
truss 12, 14 Results
Elements evaluation 21
continuum 3 Rigid body motion 8, 46
spring 3 Sign convention
structural 3 beam 31
two-dimensional 40 Stiffness
Equations matrix 5, 13, 15, 31, 32
equilibrium 5, 14 Stress
system 5 contour 51
Error node 9
beam 33 transverse shear 34
beam element 32 Structure
estimate 53 beam 34
estimated 54 stiffness matrix 16
execution 48 truss 21
of approximation 19 Superposition 29, 32

73
74 Index

Symmetry 45
Vector transformation 13
Wavefront 48
shi20396_ch01.qxd 6/5/03 12:11 PM Page 1

Chapter 1

Problems 1-1 through 1-4 are for student research.

1-5 Impending motion to left


E

1 1

f f
A B
G
Fcr F ␪
D C ␪cr
Facc

Consider force F at G, reactions at B and D. Extend lines of action for fully-developed fric-
tion D E and B E to find the point of concurrency at E for impending motion to the left. The
critical angle is θcr . Resolve force F into components Facc and Fcr . Facc is related to mass and
acceleration. Pin accelerates to left for any angle 0 < θ < θcr . When θ > θcr , no magnitude
of F will move the pin.

Impending motion to right


E⬘ ⭈E

1 1

f f
A B
G
d

Fcr F⬘ ␪⬘
D C ⬘
␪cr
F⬘acc

Consider force F  at G, reactions at A and C. Extend lines of action for fully-developed fric-
tion AE  and C E  to find the point of concurrency at E  for impending motion to the left. The
critical angle is θcr . Resolve force F  into components Facc
 and F  . F  is related to mass
cr acc
and acceleration. Pin accelerates to right for any angle 0 < θ  < θcr . When θ  > θcr , no mag-
nitude of F  will move the pin.
The intent of the question is to get the student to draw and understand the free body in
order to recognize what it teaches. The graphic approach accomplishes this quickly. It is im-
portant to point out that this understanding enables a mathematical model to be constructed,
and that there are two of them.
This is the simplest problem in mechanical engineering. Using it is a good way to begin a
course.
What is the role of pin diameter d?
Yes, changing the sense of F changes the response.
shi20396_ch01.qxd 6/5/03 12:11 PM Page 2

2 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

1-6

(a) y Fy = −F − f N cos θ + N sin θ = 0 (1)
F
 T
Fx = f N sin θ + N cos θ − =0
r
F = N (sin θ − f cos θ)
T
␪ r x Ans.
T = Nr( f sin θ + cos θ)
N
fN
Combining
1 + f tan θ
T = Fr = KFr Ans. (2)
tan θ − f
(b) If T → ∞ detent self-locking tan θ − f = 0 ∴ θcr = tan−1 f Ans.
(Friction is fully developed.)

Check: If F = 10 lbf, f = 0.20, θ = 45◦ , r = 2 in


10
N= = 17.68 lbf
−0.20 cos 45◦ + sin 45◦
T
= 17.28(0.20 sin 45◦ + cos 45◦ ) = 15 lbf
r
f N = 0.20(17.28) = 3.54 lbf

θcr = tan−1 f = tan−1 (0.20) = 11.31◦

11.31° < θ < 90°

1-7
(a) F = F0 + k(0) = F0
T1 = F0r Ans.
(b) When teeth are about to clear
F = F0 + kx2
From Prob. 1-6
f tan θ + 1
T2 = Fr
tan θ − f

( F0 + kx2 )( f tan θ + 1)
T2 = r Ans.
tan θ − f

1-8
Given, F = 10 + 2.5x lbf, r = 2 in, h = 0.2 in, θ = 60◦ , f = 0.25, xi = 0, x f = 0.2
Fi = 10 lbf; Ff = 10 + 2.5(0.2) = 10.5 lbf Ans.
shi20396_ch01.qxd 6/5/03 12:11 PM Page 3

Chapter 1 3

From Eq. (1) of Prob. 1-6


F
N=
− f cos θ + sin θ
10
Ni = = 13.49 lbf Ans.
−0.25 cos 60◦ + sin 60◦
10.5
Nf = 13.49 = 14.17 lbf Ans.
10
From Eq. (2) of Prob. 1-6
1 + f tan θ 1 + 0.25 tan 60◦
K = = = 0.967 Ans.
tan θ − f tan 60◦ − 0.25
Ti = 0.967(10)(2) = 19.33 lbf · in
Tf = 0.967(10.5)(2) = 20.31 lbf · in

1-9
(a) Point vehicles
v

cars v 42.1v − v 2
Q= = =
hour x 0.324
Seek stationary point maximum
dQ 42.1 − 2v
=0= ∴ v* = 21.05 mph
dv 0.324
42.1(21.05) − 21.052
Q* = = 1367.6 cars/h Ans.
0.324
(b) v

l x l
2 2

 −1
v 0.324 l
Q= = +
x +l v(42.1) − v 2 v
Maximize Q with l = 10/5280 mi

v Q
22.18 1221.431
22.19 1221.433
22.20 1221.435 ←
22.21 1221.435
22.22 1221.434

1368 − 1221
% loss of throughput = 12% Ans.
1221
shi20396_ch01.qxd 6/5/03 12:11 PM Page 4

4 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

22.2 − 21.05
(c) % increase in speed = 5.5%
21.05
Modest change in optimal speed Ans.

1-10 This and the following problem may be the student’s first experience with a figure of merit.
• Formulate fom to reflect larger figure of merit for larger merit.
• Use a maximization optimization algorithm. When one gets into computer implementa-
tion and answers are not known, minimizing instead of maximizing is the largest error
one can make.

FV = F1 sin θ − W = 0

FH = −F1 cos θ − F2 = 0
From which
F1 = W/sin θ
F2 = −W cos θ/sin θ
fom = −S = −¢γ (volume)
.
= −¢γ(l1 A1 + l2 A2 )
F1 W l1
A1 = = , l2 =
S S sin θ cos θ
 
 F2  W cos θ
A2 =   =
S S sin θ
 
l2 W l2 W cos θ
fom = −¢γ +
cos θ S sin θ S sin θ
 
−¢γ W l2 1 + cos2 θ
=
S cos θ sin θ
Set leading constant to unity

θ◦ fom
θ* = 54.736◦ Ans.
0 −∞ fom* = −2.828
20 −5.86
30 −4.04 Alternative:
 
40 −3.22 d 1 + cos2 θ
45 −3.00 =0
dθ cos θ sin θ
50 −2.87
54.736 −2.828 And solve resulting tran-
60 −2.886 scendental for θ*.

Check second derivative to see if a maximum, minimum, or point of inflection has been
found. Or, evaluate fom on either side of θ*.
shi20396_ch01.qxd 6/5/03 12:11 PM Page 5

Chapter 1 5

1-11
(a) x1 + x2 = X 1 + e1 + X 2 + e2
error = e = (x1 + x2 ) − ( X 1 + X 2 )
= e1 + e2 Ans.
(b) x1 − x2 = X 1 + e1 − ( X 2 + e2 )
e = (x1 − x2 ) − ( X 1 − X 2 ) = e1 − e2 Ans.
(c) x1 x2 = ( X 1 + e1 )( X 2 + e2 )
e = x1 x2 − X 1 X 2 = X 1 e2 + X 2 e1 + e1 e2
 
. e1 e2
= X 1 e2 + X 2 e1 = X 1 X 2 + Ans.
X1 X2
 
x1 X 1 + e1 X 1 1 + e1 / X 1
(d) = =
x2 X 2 + e2 X 2 1 + e2 / X 2
 −1   
e2 . e2 e1 e2 . e1 e2
1+ =1− and 1+ 1− =1+ −
X2 X2 X1 X2 X1 X2
 
x1 X 1 . X 1 e1 e2
e= − = − Ans.
x2 X2 X2 X1 X2

1-12 √
(a) x1 = 5 = 2.236 067 977 5
X 1 = 2.23 3-correct digits

x2 = 6 = 2.449 487 742 78
X 2 = 2.44 3-correct digits
√ √
x1 + x2 = 5 + 6 = 4.685 557 720 28

e1 = x1 − X 1 = 5 − 2.23 = 0.006 067 977 5

e2 = x2 − X 2 = 6 − 2.44 = 0.009 489 742 78
√ √
e = e1 + e2 = 5 − 2.23 + 6 − 2.44 = 0.015 557 720 28
Sum = x1 + x2 = X 1 + X 2 + e
= 2.23 + 2.44 + 0.015 557 720 28
= 4.685 557 720 28 (Checks) Ans.

(b) X 1 = 2.24, X 2 = 2.45



e1 = 5 − 2.24 = −0.003 932 022 50

e2 = 6 − 2.45 = −0.000 510 257 22
e = e1 + e2 = −0.004 442 279 72
Sum = X 1 + X 2 + e
= 2.24 + 2.45 + (−0.004 442 279 72)
= 4.685 557 720 28 Ans.
shi20396_ch01.qxd 6/5/03 12:11 PM Page 6

6 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

1-13
(a) σ = 20(6.89) = 137.8 MPa
(b) F = 350(4.45) = 1558 N = 1.558 kN
(c) M = 1200 lbf · in (0.113) = 135.6 N · m
(d) A = 2.4(645) = 1548 mm2
(e) I = 17.4 in4 (2.54) 4 = 724.2 cm4
(f) A = 3.6(1.610) 2 = 9.332 km2
(g) E = 21(1000)(6.89) = 144.69(103 ) MPa = 144.7 GPa
(h) v = 45 mi/h (1.61) = 72.45 km/h
(i) V = 60 in3 (2.54) 3 = 983.2 cm3 = 0.983 liter

1-14
(a) l = 1.5/0.305 = 4.918 ft = 59.02 in
(b) σ = 600/6.89 = 86.96 kpsi
(c) p = 160/6.89 = 23.22 psi
(d) Z = 1.84(105 )/(25.4) 3 = 11.23 in3
(e) w = 38.1/175 = 0.218 lbf/in
(f) δ = 0.05/25.4 = 0.00197 in
(g) v = 6.12/0.0051 = 1200 ft/min
(h)  = 0.0021 in/in
(i) V = 30/(0.254) 3 = 1831 in3

1-15
200
(a) σ = = 13.1 MPa
15.3
42(103 )
(b) σ = = 70(106 ) N/m2 = 70 MPa
6(10−2 ) 2
1200(800) 3 (10−3 ) 3
(c) y = = 1.546(10−2 ) m = 15.5 mm
3(207)(6.4)(109 )(10−2 ) 4
1100(250)(10−3 )
(d) θ = 4 9 −3 4
= 9.043(10−2 ) rad = 5.18◦
79.3(π/32)(25) (10 )(10 )

1-16
600
(a) σ = = 5 MPa
20(6)
1
(b) I = 8(24) 3 = 9216 mm4
12
π
(c) I = 324 (10−1 ) 4 = 5.147 cm4
64
16(16)
(d) τ = = 5.215(106 ) N/m2 = 5.215 MPa
π(253 )(10−3 ) 3
shi20396_ch01.qxd 6/5/03 12:11 PM Page 7

Chapter 1 7

1-17
120(103 )
(a) τ = = 382 MPa
(π/4)(202 )
32(800)(800)(10−3 )
(b) σ = = 198.9(106 ) N/m2 = 198.9 MPa
π(32) 3 (10−3 ) 3
π
(c) Z = (364 − 264 ) = 3334 mm3
32(36)
(1.6) 4 (79.3)(10−3 ) 4 (109 )
(d) k = = 286.8 N/m
8(19.2) 3 (32)(10−3 ) 3
shi20396_ch02.qxd 7/21/03 3:28 PM Page 8

Chapter 2

2-1
(a) 12

10

0
60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210

(b) f/(Nx) = f/(69 · 10) = f/690

x f fx f x2 f/(Nx)
60 2 120 7200 0.0029
70 1 70 4900 0.0015
80 3 240 19 200 0.0043
90 5 450 40 500 0.0072
100 8 800 80 000 0.0116
110 12 1320 145 200 0.0174
120 6 720 86 400 0.0087
130 10 1300 169 000 0.0145
140 8 1120 156 800 0.0116
150 5 750 112 500 0.0174
160 2 320 51 200 0.0029
170 3 510 86 700 0.0043
180 2 360 64 800 0.0029
190 1 190 36 100 0.0015
200 0 0 0 0
210 1 210 44 100 0.0015
69 8480 1 104 600

8480
Eq. (2-9) x̄ = = 122.9 kcycles
69
 1/2
1 104 600 − 84802 /69
Eq. (2-10) sx =
69 − 1
= 30.3 kcycles Ans.
shi20396_ch02.qxd 7/21/03 3:28 PM Page 9

Chapter 2 9

2-2 Data represents a 7-class histogram with N = 197.

x f fx f x2
174 6 1044 181 656
182 9 1638 298 116
190 44 8360 1 588 400
198 67 13 266 2 626 688
206 53 10 918 2 249 108
214 12 2568 549 552
220 6 1320 290 400
197 39 114 7 789 900

39 114
x̄ = = 198.55 kpsi Ans.
197
 1/2
7 783 900 − 39 1142 /197
sx =
197 − 1
= 9.55 kpsi Ans.

2-3
Form a table:

x f fx f x2
64 2 128 8192
68 6 408 27 744
72 6 432 31 104
76 9 684 51 984
80 19 1520 121 600
84 10 840 70 560
88 4 352 30 976
92 2 184 16 928
58 4548 359 088

4548
x̄ = = 78.4 kpsi
58
 1/2
359 088 − 45482 /58
sx = = 6.57 kpsi
58 − 1
From Eq. (2-14)
  
1 1 x − 78.4 2
f (x) = √ exp −
6.57 2π 2 6.57
shi20396_ch02.qxd 7/21/03 3:28 PM Page 10

10 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

2-4 (a)

y f fy f y2 y f/(Nw) f (y) g(y)


5.625 1 5.625 31.640 63 5.625 0.072 727 0.001 262 0.000 295
5.875 0 0 0 5.875 0 0.008 586 0.004 088
6.125 0 0 0 6.125 0 0.042 038 0.031 194
6.375 3 19.125 121.9219 6.375 0.218 182 0.148 106 0.140 262
6.625 3 19.875 131.6719 6.625 0.218 182 0.375 493 0.393 667
6.875 6 41.25 283.5938 6.875 0.436 364 0.685 057 0.725 002
7.125 14 99.75 710.7188 7.125 1.018 182 0.899 389 0.915 128
7.375 15 110.625 815.8594 7.375 1.090 909 0.849 697 0.822 462
7.625 10 76.25 581.4063 7.625 0.727 273 0.577 665 0.544 251
7.875 2 15.75 124.0313 7.875 0.145 455 0.282 608 0.273 138
8.125 1 8.125 66.015 63 8.125 0.072 727 0.099 492 0.106 72
55 396.375 2866.859

For a normal distribution,


 1/2
2866.859 − (396.3752 /55)
ȳ = 396.375/55 = 7.207, sy = = 0.4358
55 − 1
  
1 1 x − 7.207 2
f ( y) = √ exp −
0.4358 2π 2 0.4358

For a lognormal distribution,


√ √
x̄ = ln 7.206 818 − ln 1 + 0.060 4742 = 1.9732, sx = ln 1 + 0.060 4742 = 0.0604
  
1 1 ln x − 1.9732 2
g( y) = √ exp −
x(0.0604)( 2π) 2 0.0604
(b) Histogram
f
1.2 Data
N
LN
1

0.8

0.6

0.4

0.2

0
5.63 5.88 6.13 6.38 6.63 6.88 7.13 7.38 7.63 7.88 8.13
log N
shi20396_ch02.qxd 7/21/03 3:28 PM Page 11

Chapter 2 11

2-5 Distribution is uniform in interval 0.5000 to 0.5008 in, range numbers are a = 0.5000,
b = 0.5008 in.
a+b 0.5000 + 0.5008
(a) Eq. (2-22) µx = = = 0.5004
2 2
b−a 0.5008 − 0.5000
Eq. (2-23) σx = √ = √ = 0.000 231
2 3 2 3
(b) PDF from Eq. (2-20)

1250 0.5000 ≤ x ≤ 0.5008 in
f (x) =
0 otherwise

(c) CDF from Eq. (2-21)



0 x < 0.5000
F(x) = (x − 0.5)/0.0008 0.5000 ≤ x ≤ 0.5008

1 x > 0.5008

If all smaller diameters are removed by inspection, a = 0.5002, b = 0.5008


0.5002 + 0.5008
µx = = 0.5005 in
2
0.5008 − 0.5002
σ̂x = √ = 0.000 173 in
2 3

1666.7 0.5002 ≤ x ≤ 0.5008
f (x) =
0 otherwise

0 x < 0.5002
F(x) = 1666.7(x − 0.5002) 0.5002 ≤ x ≤ 0.5008

1 x > 0.5008

2-6 Dimensions produced are due to tool dulling and wear. When parts are mixed, the distribution
is uniform. From Eqs. (2-22) and (2-23),
√ √
a = µx − 3s = 0.6241 − 3(0.000 581) = 0.6231 in
√ √
b = µx + 3s = 0.6241 + 3(0.000 581) = 0.6251 in

0.623
We suspect the dimension was in Ans.
0.625
shi20396_ch02.qxd 7/21/03 3:28 PM Page 12

12 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

2-7 F(x) = 0.555x − 33 mm


(a) Since F(x) is linear, the distribution is uniform at x = a
F(a) = 0 = 0.555(a) − 33

∴ a = 59.46 mm. Therefore, at x = b


F(b) = 1 = 0.555b − 33

∴ b = 61.26 mm. Therefore,



0 x < 59.46 mm
F(x) = 0.555x − 33 59.46 ≤ x ≤ 61.26 mm

1 x > 61.26 mm

The PDF is d F/dx , thus the range numbers are:



0.555 59.46 ≤ x ≤ 61.26 mm
f (x) = Ans.
0 otherwise

From the range numbers,

59.46 + 61.26
µx = = 60.36 mm Ans.
2
61.26 − 59.46
1

σ̂x = √ = 0.520 mm Ans.


2 3

(b) σ is an uncorrelated quotient F̄ = 3600 lbf, Ā = 0.112 in2


C F = 300/3600 = 0.083 33, C A = 0.001/0.112 = 0.008 929

From Table 2-6, for σ

µF 3600
σ̄ = = = 32 143 psi Ans.
µA 0.112
 1/2
(0.083332 + 0.0089292 )
σ̂σ = 32 143 = 2694 psi Ans.
(1 + 0.0089292 )
Cσ = 2694/32 143 = 0.0838 Ans.

Since F and A are lognormal, division is closed and σ is lognormal too.

σ = LN(32 143, 2694) psi Ans.


shi20396_ch02.qxd 7/21/03 3:28 PM Page 13

Chapter 2 13

2-8 Cramer’s rule



y x 2

x y x 3 yx 3 − x yx 2
a1 = = Ans.
x x 2 xx 3 − (x 2 ) 2

x 2 x 3

x y
2 xx y − yx 2
x x y
a2 = = Ans.
x x 2 xx 3 − (x 2 ) 2

x 2 x 3

x y x2 x3 xy
0 0.01 0 0 0
0.2 0.15 0.04 0.008 0.030
0.4 0.25 0.16 0.064 0.100
0.6 0.25 0.36 0.216 0.150
0.8 0.17 0.64 0.512 0.136
1.0 −0.01 1.00 1.000 −0.010
3.0 0.82 2.20 1.800 0.406

a1 = 1.040 714 a2 = −1.046 43 Ans.

Data Regression
x y y
0 0.01 0
0.2 0.15 0.166 286
0.4 0.25 0.248 857
0.6 0.25 0.247 714
0.8 0.17 0.162 857
1.0 −0.01 −0.005 71

y
0.3 Data
Regression
0.25

0.2

0.15

0.1

0.05

0.05 x
0 0.2 0.4 0.6 0.8 1
shi20396_ch02.qxd 7/21/03 3:28 PM Page 14

14 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

2-9
Data Regression
Su Se Se Su2 Su Se
0 20.356 75
60 30 39.080 78 3 600 1 800
64 48 40.329 05 4 096 3 072
65 29.5 40.641 12 4 225 1 917.5
82 45 45.946 26 6 724 3 690
101 51 51.875 54 10 201 5 151
119 50 57.492 75 14 161 5 950
120 48 57.804 81 14 400 5 760
130 67 60.925 48 16 900 8 710
134 60 62.173 75 17 956 8 040
145 64 65.606 49 21 025 9 280
180 84 76.528 84 32 400 15 120
195 78 81.209 85 38 025 15 210
205 96 84.330 52 42 025 19 680
207 87 84.954 66 42 849 18 009
210 87 85.890 86 44 100 18 270
213 75 86.827 06 45 369 15 975
225 99 90.571 87 50 625 22 275
225 87 90.571 87 50 625 19 575
227 116 91.196 51 529 26 332
230 105 92.132 2 52 900 24 150
238 109 94.628 74 56 644 25 942
242 106 95.877 01 58 564 25 652
265 105 103.054 6 70 225 27 825
280 96 107.735 6 78 400 26 880
295 99 112.416 6 87 025 29 205
325 114 121.778 6 105 625 37 050
325 117 121.778 6 105 625 38 025
355 122 131.140 6 126 025 43 310
5462 2274.5 1 251 868 501 855.5

m = 0.312 067 b = 20.356 75 Ans.

Se
140 Data
Regression
120

100

80

60

40

20

0 Su
0 100 200 300 400
shi20396_ch02.qxd 7/21/03 3:28 PM Page 15

Chapter 2 15

2
2-10 E = y − a0 − a2 x 2

∂E 
= −2 y − a0 − a2 x 2 = 0
∂a0

y − na0 − a2 x2 = 0 ⇒ y = na0 + a2 x2
Ans.
∂E 
=2 y − a0 − a2 x 2 (2x) = 0 ⇒ x y = a0 x + a2 x3
∂a2
Cramer’s rule

y x 2

x y x 3 x 3 y − x 2 x y
a0 = =
n x 2 nx 3 − xx 2

x x 3

n y
x x y nx y − xy
a2 = 2 =
n x nx 3 − xx 2

x x 3

Data Regression
x y y x2 x3 xy
20 19 19.2 400 8 000 380
40 17 16.8 1600 64 000 680
60 13 12.8 3600 216 000 780
80 7 7.2 6400 512 000 560
200 56 12 000 800 000 2400

800 000(56) − 12 000(2400)


a0 = = 20
4(800 000) − 200(12 000)
4(2400) − 200(56)
a2 = = −0.002
4(800 000) − 200(12 000)
y
25 Data
Regression

20

15

10

0 x
0 20 40 60 80 100
shi20396_ch02.qxd 7/21/03 3:28 PM Page 16

16 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

2-11

Data Regression
x y y x2 y2 xy x − x̄ (x − x̄) 2
0.2 7.1 7.931 803 0.04 50.41 1.42 −0.633 333 0.401 111 111
0.4 10.3 9.884 918 0.16 106.09 4.12 −0.433 333 0.187 777 778
0.6 12.1 11.838 032 0.36 146.41 7.26 −0.233 333 0.054 444 444
0.8 13.8 13.791 147 0.64 190.44 11.04 −0.033 333 0.001 111 111
1 16.2 15.744 262 1.00 262.44 16.20 0.166 666 0.027 777 778
2 25.2 25.509 836 4.00 635.04 50.40 1.166 666 1.361 111 111
5 84.7 6.2 1390.83 90.44 0 2.033 333 333

6(90.44) − 5(84.7)
m̂ = k = = 9.7656
6(6.2) − (5) 2
84.7 − 9.7656(5)
b̂ = Fi = = 5.9787
6
F
30 Data
Regression

25

20

15

10

0 x
0 0.5 1 1.5 2 2.5

5 84.7
(a) x̄ = ; ȳ = = 14.117
6 6
Eq. (2-37)

1390.83 − 5.9787(84.7) − 9.7656(90.44)
s yx =
6−2
= 0.556
Eq. (2-36)

1 (5/6) 2
sb̂ = 0.556 + = 0.3964 lbf
6 2.0333
Fi = (5.9787, 0.3964) lbf Ans.
shi20396_ch02.qxd 7/21/03 3:28 PM Page 17

Chapter 2 17

(b) Eq. (2-35)


0.556
sm̂ = √ = 0.3899 lbf/in
2.0333
k = (9.7656, 0.3899) lbf/in Ans.

2-12 The expression  = δ/l is of the form x/y. Now δ = (0.0015, 0.000 092) in, unspecified
distribution; l = (2.000, 0.0081) in, unspecified distribution;
C x = 0.000 092/0.0015 = 0.0613
C y = 0.0081/2.000 = 0.000 75
From Table 2-6, ¯ = 0.0015/2.000 = 0.000 75
 1/2
0.06132 + 0.004 052
σ̂ = 0.000 75
1 + 0.004 052
= 4.607(10−5 ) = 0.000 046
We can predict ¯ and σ̂ but not the distribution of .

2-13 σ = E
 = (0.0005, 0.000 034) distribution unspecified; E = (29.5, 0.885) Mpsi, distribution
unspecified;
C x = 0.000 034/0.0005 = 0.068,
C y = 0.0885/29.5 = 0.030
σ is of the form x, y
Table 2-6
σ̄ = ¯ Ē = 0.0005(29.5)106 = 14 750 psi
σ̂σ = 14 750(0.0682 + 0.0302 + 0.0682 + 0.0302 ) 1/2
= 1096.7 psi
Cσ = 1096.7/14 750 = 0.074 35

2-14
Fl
δ=
AE
F = (14.7, 1.3) kip, A = (0.226, 0.003) in2 , l = (1.5, 0.004) in, E = (29.5, 0.885) Mpsi dis-
tributions unspecified.
C F = 1.3/14.7 = 0.0884 ; C A = 0.003/0.226 = 0.0133 ; Cl = 0.004/1.5 = 0.00267 ;
C E = 0.885/29.5 = 0.03
Mean of δ:   
Fl 1 1
δ= = Fl
AE A E
shi20396_ch02.qxd 7/21/03 3:28 PM Page 18

18 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

From Table 2-6,


¯ Ā)(1/ Ē)
δ̄ = F̄ l(1/
1 1
δ̄ = 14 700(1.5)
0.226 29.5(106 )
= 0.003 31 in Ans.
For the standard deviation, using the first-order terms in Table 2-6,

. F̄ l¯ 2 1/2 1/2
σ̂δ = C F + Cl2 + C 2A + C E2 = δ̄ C F2 + Cl2 + C 2A + C 2E
Ā Ē
σ̂δ = 0.003 31(0.08842 + 0.002672 + 0.01332 + 0.032 ) 1/2
= 0.000 313 in Ans.
COV
Cδ = 0.000 313/0.003 31 = 0.0945 Ans.
Force COV dominates. There is no distributional information on δ.

2-15 M = (15 000, 1350) lbf · in, distribution unspecified; d = (2.00, 0.005) in distribution
unspecified.
32M 1350 0.005
σ= , CM = = 0.09 , Cd = = 0.0025
πd3 15 000 2.00
σ is of the form x/y, Table 2-6.
Mean:
32 M̄ . 32 M̄ 32(15 000)
σ̄ = = =
π d3 π d̄ 3 π(23 )
= 19 099 psi Ans.
Standard Deviation:
 2  1/2
σ̂σ = σ̄ C M + Cd23 1 + Cd23
.
From Table 2-6, Cd 3 = 3Cd = 3(0.0025) = 0.0075
 2  1/2
σ̂σ = σ̄ C M + (3Cd ) 2 (1 + (3Cd )) 2
= 19 099[(0.092 + 0.00752 )/(1 + 0.00752 )]1/2
= 1725 psi Ans.
COV:
1725
Cσ = = 0.0903 Ans.
19 099
Stress COV dominates. No information of distribution of σ.
shi20396_ch02.qxd 7/21/03 3:28 PM Page 19

Chapter 2 19

2-16
f (x)

␣ ␤
x
x1 x2

Fraction discarded is α + β. The area under the PDF was unity. Having discarded α + β
fraction, the ordinates to the truncated PDF are multiplied by a.
1
a=
1 − (α + β)
New PDF, g(x) , is given by

f (x)/[1 − (α + β)] x1 ≤ x ≤ x2
g(x) =
0 otherwise

More formal proof: g(x) has the property


 x2  x2
1= g(x) dx = a f (x) dx
x1 x1
 ∞  x1  ∞ 
1=a f (x) dx − f (x) dx − f (x) dx
−∞ 0 x2

1 = a {1 − F(x1 ) − [1 − F(x2 )]}


1 1 1
a= = =
F(x2 ) − F(x1 ) (1 − β) − α 1 − (α + β)

2-17
(a) d = U[0.748, 0.751]
0.751 + 0.748
µd = = 0.7495 in
2
0.751 − 0.748
σ̂d = √ = 0.000 866 in
2 3
1 1
f (x) = = = 333.3 in−1
b−a 0.751 − 0.748
x − 0.748
F(x) = = 333.3(x − 0.748)
0.751 − 0.748
shi20396_ch02.qxd 7/21/03 3:28 PM Page 20

20 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

(b) F(x1 ) = F(0.748) = 0


F(x2 ) = (0.750 − 0.748)333.3 = 0.6667
If g(x) is truncated, PDF becomes
f (x) 333.3
g(x)  500 g(x) = = = 500 in−1
f (x)  333.3 F(x2 ) − F(x1 ) 0.6667 − 0
a  + b 0.748 + 0.750
x µx = = = 0.749 in
0.748 0.749 0.750 0.751 2 2
b − a  0.750 − 0.748
σ̂x = √ = √ = 0.000 577 in
2 3 2 3

2-18 From Table A-10, 8.1% corresponds to z 1 = −1.4 and 5.5% corresponds to z 2 = +1.6.
k1 = µ + z 1 σ̂
k2 = µ + z 2 σ̂
From which
z 2 k1 − z 1 k2 1.6(9) − (−1.4)11
µ= =
z2 − z1 1.6 − (−1.4)
= 9.933
k2 − k1
σ̂ =
z2 − z1
11 − 9
= = 0.6667
1.6 − (−1.4)
The original density function is
  2 
1 1 k − 9.933
f (k) = √ exp − Ans.
0.6667 2π 2 0.6667

2-19 From Prob. 2-1, µ = 122.9 kcycles and σ̂ = 30.3 kcycles.


x10 − µ x10 − 122.9
z 10 = =
σ̂ 30.3
x10 = 122.9 + 30.3z 10
From Table A-10, for 10 percent failure, z 10 = −1.282
x10 = 122.9 + 30.3(−1.282)
= 84.1 kcycles Ans.
shi20396_ch02.qxd 7/21/03 3:28 PM Page 21

Chapter 2 21

2-20

x f fx f x2 x f/(Nw) f (x)
60 2 120 7200 60 0.002 899 0.000 399
70 1 70 4900 70 0.001 449 0.001 206
80 3 240 19 200 80 0.004 348 0.003 009
90 5 450 40 500 90 0.007 246 0.006 204
100 8 800 80 000 100 0.011 594 0.010 567
110 12 1320 145 200 110 0.017 391 0.014 871
120 6 720 86 400 120 0.008 696 0.017 292
130 10 1300 169 000 130 0.014 493 0.016 612
140 8 1120 156 800 140 0.011 594 0.013 185
150 5 750 112 500 150 0.007 246 0.008 647
160 2 320 51 200 160 0.002 899 0.004 685
170 3 510 86 700 170 0.004 348 0.002 097
180 2 360 64 800 180 0.002 899 0.000 776
190 1 190 36 100 190 0.001 449 0.000 237
200 0 0 0 200 0 5.98E-05
210 1 210 44 100 210 0.001 449 1.25E-05
69 8480

x̄ = 122.8986 sx = 22.887 19

x f/(Nw) f (x) x f/(Nw) f (x)


55 0 0.000 214 145 0.011 594 0.010 935
55 0.002 899 0.000 214 145 0.007 246 0.010 935
65 0.002 899 0.000 711 155 0.007 246 0.006 518
65 0.001 449 0.000 711 155 0.002 899 0.006 518
75 0.001 449 0.001 951 165 0.002 899 0.003 21
75 0.004 348 0.001 951 165 0.004 348 0.003 21
85 0.004 348 0.004 425 175 0.004 348 0.001 306
85 0.007 246 0.004 425 175 0.002 899 0.001 306
95 0.007 246 0.008 292 185 0.002 899 0.000 439
95 0.011 594 0.008 292 185 0.001 449 0.000 439
105 0.011 594 0.012 839 195 0.001 449 0.000 122
105 0.017 391 0.012 839 195 0 0.000 122
115 0.017 391 0.016 423 205 0 2.8E-05
115 0.008 696 0.016 423 205 0.001 499 2.8E-05
125 0.008 696 0.017 357 215 0.001 499 5.31E-06
125 0.014 493 0.017 357 215 0 5.31E-06
135 0.014 493 0.015 157
135 0.011 594 0.015 157
shi20396_ch02.qxd 7/21/03 3:28 PM Page 22

22 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

f
0.02 Histogram
PDF
0.018

0.016

0.014

0.012

0.01

0.008

0.006

0.004

0.002

0 x
0 50 100 150 200 250

2-21

x f fx f x2 f/(Nw) f (x)
174 6 1044 181 656 0.003 807 0.001 642
182 9 1638 298 116 0.005 711 0.009 485
190 44 8360 1 588 400 0.027 919 0.027 742
198 67 13 266 2 626 668 0.042 513 0.041 068
206 53 10 918 2 249 108 0.033 629 0.030 773
214 12 2568 549 552 0.007 614 0.011 671
222 6 1332 295 704 0.003 807 0.002 241
1386 197 39 126 7 789 204

x̄ = 198.6091 sx = 9.695 071

f
x f/(Nw) f (x) 0.045 Data
PDF
0.04
170 0 0.000 529
170 0.003 807 0.000 529 0.035

178 0.003 807 0.004 297 0.03


178 0.005 711 0.004 297 0.025
186 0.005 711 0.017 663 0.02
186 0.027 919 0.017 663
0.015
194 0.027 919 0.036 752
194 0.042 513 0.036 752 0.01

202 0.042 513 0.038 708 0.005


202 0.033 629 0.038 708 0 x
210 0.033 629 0.020 635 150 170 190 210 230

210 0.007 614 0.020 635


218 0.007 614 0.005 568
218 0.003 807 0.005 568
226 0.003 807 0.000 76
226 0 0.000 76
shi20396_ch02.qxd 7/21/03 3:28 PM Page 23

Chapter 2 23

2-22

x f fx f x2 f/(Nw) f (x)
64 2 128 8192 0.008 621 0.005 48
68 6 408 27 744 0.025 862 0.017 299
72 6 432 31 104 0.025 862 0.037 705
76 9 684 51 984 0.038 793 0.056 742
80 19 1520 121 600 0.081 897 0.058 959
84 10 840 70 560 0.043 103 0.042 298
88 4 352 30 976 0.017 241 0.020 952
92 2 184 16 928 0.008 621 0.007 165
624 58 4548 359 088

x̄ = 78.413 79 sx = 6.572 229

x f/(Nw) f(x) x f/(Nw) f(x)


62 0 0.002 684 82 0.081 897 0.052 305
62 0.008 621 0.002 684 82 0.043 103 0.052 305
66 0.008 621 0.010 197 86 0.043 103 0.031 18
66 0.025 862 0.010 197 86 0.017 241 0.031 18
70 0.025 862 0.026 749 90 0.017 241 0.012 833
70 0.025 862 0.026 749 90 0.008 621 0.012 833
74 0.025 862 0.048 446 94 0.008 621 0.003 647
74 0.038 793 0.048 446 94 0 0.003 647
78 0.038 793 0.060 581
78 0.081 897 0.060 581
f
0.09 Data
PDF
0.08

0.07

0.06

0.05

0.04

0.03

0.02

0.01

0 x
60 70 80 90 100

2-23
4 P̄ 4(40)
σ̄ = = = 50.93 kpsi
πd 2 π(12 )
4 σ̂ P 4(8.5)
σ̂σ = = = 10.82 kpsi
πd 2 π(12 )
σ̂s y = 5.9 kpsi
shi20396_ch02.qxd 7/21/03 3:28 PM Page 24

24 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

For no yield, m = S y − σ ≥ 0
m − µm 0 − µm µm
z= = =−
σ̂m σ̂m σ̂m
µm = S̄ y − σ̄ = 27.47 kpsi,
 1/2
σ̂m = σ̂σ + σ̂ Sy
2 2
= 12.32 kpsi
−27.47 m
z= = −2.230 0
12.32
From Table A-10, p f = 0.0129
R = 1 − p f = 1 − 0.0129 = 0.987 Ans.

2-24 For a lognormal distribution, 


Eq. (2-18) µ y = ln µx − ln 1 + C x2
 
Eq. (2-19) σ̂ y = ln 1 + C x2
From Prob. (2-23)
µm = S̄ y − σ̄ = µx
     
µ y = ln S̄ y − ln 1 + C Sy − ln σ̄ − ln 1 + Cσ
2 2

  
S̄ y 1 + Cσ2
= ln
σ̄ 1 + C S2y
   1/2
σ̂ y = ln 1 + C S2y + ln 1 + Cσ2
   
= ln 1 + C Sy 1 + Cσ
2 2

  
S̄ y 1 + Cσ2
ln
µ σ̄ 1 + C S2
z = − = −  
y

σ̂  
ln 1 + C S2y 1 + Cσ2

4 P̄ 4(30)
σ̄ = = = 38.197 kpsi
πd 2 π(12 )
4 σ̂ P 4(5.1)
σ̂σ = = = 6.494 kpsi
πd 2 π(12 )
6.494
Cσ = = 0.1700
38.197
3.81
C Sy = = 0.076 81
49.6
shi20396_ch02.qxd 7/21/03 3:28 PM Page 25

Chapter 2 25

  
49.6 1 + 0.170 
2
ln 
38.197 1 + 0.076 812
z = −   = −1.470
ln (1 + 0.076 812 )(1 + 0.1702 )
From Table A-10
p f = 0.0708
R = 1 − p f = 0.929 Ans.

2-25
(a) a = 1.000 ± 0.001 in
b = 2.000 ± 0.003 in
c = 3.000 ± 0.005 in
d = 6.020 ± 0.006 in
w̄ = d − a − b − c = 6.020 − 1 − 2 − 3 = 0.020 in
!
tw = tall = 0.001 + 0.003 + 0.005 + 0.006
= 0.015 in
w = 0.020 ± 0.015 in Ans.
(b) w̄ = 0.020
       

0.001 2 0.003 2 0.005 2 0.006 2
σ̂w = σ̂all =
2
√ + √ + √ + √
3 3 3 3
= 0.004 86 → 0.005 in (uniform)
w = 0.020 ± 0.005 in Ans.

2-26
V + V = (a + a)(b + b)(c + c)
V + V = abc + bca + acb + abc + small higher order terms
V . a b c
= + + Ans.
V̄ a b c
V̄ = ā b̄c̄ = 1.25(1.875)(2.75) = 6.4453 in3
V 0.001 0.002 0.003
= + + = 0.00296
V̄ 1.250 1.875 2.750
V
V = V̄ = 0.00296(6.4453) = 0.0191 in3

Lower range number:
V̄ − V = 6.4453 − 0.0191 = 6.4262 in3 Ans.
Upper range number:
V̄ + V = 6.4453 + 0.0191 = 6.4644 in3 Ans.
shi20396_ch02.qxd 7/21/03 3:28 PM Page 26

26 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

2-27
a
(a)
c b
w

wmax = 0.014 in, wmin = 0.004 in


w̄ = (0.014 + 0.004)/2 = 0.009 in
w = 0.009 ± 0.005 in

w̄ = x̄ − ȳ = ā − b̄ − c̄
0.009 = ā − 0.042 − 1.000
ā = 1.051 in

tw = tall
0.005 = ta + 0.002 + 0.002
ta = 0.005 − 0.002 − 0.002 = 0.001 in
a = 1.051 ± 0.001 in Ans.
 
(b) σ̂w = σ̂all
2
= σ̂a2 + σ̂b2 + σ̂c2

σ̂a2 = σ̂w2 − σ̂b2 − σ̂c2


     
0.005 2 0.002 2 0.002 2
= √ − √ − √
3 3 3
−6
σ̂a = 5.667(10 )
2
"
σ̂a = 5.667(10−6 ) = 0.00238 in
ā = 1.051 in, σ̂a = 0.00238 in Ans.

2-28 Choose 15 mm as basic size, D, d. Table 2-8: fit is designated as 15H7/h6. From
Table A-11, the tolerance grades are D = 0.018 mm and d = 0.011 mm.
Hole: Eq. (2-38)
Dmax = D + D = 15 + 0.018 = 15.018 mm Ans.
Dmin = D = 15.000 mm Ans.
Shaft: From Table A-12, fundamental deviation δ F = 0. From Eq. (2-39)
dmax = d + δ F = 15.000 + 0 = 15.000 mm Ans.
dmin = d + δ R − d = 15.000 + 0 − 0.011 = 14.989 mm Ans.

2-29 Choose 45 mm as basic size. Table 2-8 designates fit as 45H7/s6. From Table A-11, the
tolerance grades are D = 0.025 mm and d = 0.016 mm
Hole: Eq. (2-38)
Dmax = D + D = 45.000 + 0.025 = 45.025 mm Ans.
Dmin = D = 45.000 mm Ans.
shi20396_ch02.qxd 7/21/03 3:28 PM Page 27

Chapter 2 27

Shaft: From Table A-12, fundamental deviation δ F = +0.043 mm. From Eq. (2-40)
dmin = d + δ F = 45.000 + 0.043 = 45.043 mm Ans.
dmax = d + δ F + d = 45.000 + 0.043 + 0.016 = 45.059 mm Ans.

2-30 Choose 50 mm as basic size. From Table 2-8 fit is 50H7/g6. From Table A-11, the tolerance
grades are D = 0.025 mm and d = 0.016 mm.
Hole:
Dmax = D + D = 50 + 0.025 = 50.025 mm Ans.
Dmin = D = 50.000 mm Ans.
Shaft: From Table A-12 fundamental deviation = −0.009 mm
dmax = d + δ F = 50.000 + (−0.009) = 49.991 mm Ans.
dmin = d + δ F − d
= 50.000 + (−0.009) − 0.016
= 49.975 mm

2-31 Choose the basic size as 1.000 in. From Table 2-8, for 1.0 in, the fit is H8/f7. From
Table A-13, the tolerance grades are D = 0.0013 in and d = 0.0008 in.
Hole:
Dmax = D + (D) hole = 1.000 + 0.0013 = 1.0013 in Ans.
Dmin = D = 1.0000 in Ans.
Shaft: From Table A-14: Fundamental deviation = −0.0008 in
dmax = d + δ F = 1.0000 + (−0.0008) = 0.9992 in Ans.
dmin = d + δ F − d = 1.0000 + (−0.0008) − 0.0008 = 0.9984 in Ans.
Alternatively,
dmin = dmax − d = 0.9992 − 0.0008 = 0.9984 in. Ans.

2-32

W Di W

Do

Do = W + Di + W
D̄o = W̄ + D̄i + W̄
= 0.139 + 3.734 + 0.139 = 4.012 in

t Do = tall = 0.004 + 0.028 + 0.004
= 0.036 in
Do = 4.012 ± 0.036 in Ans.
shi20396_ch02.qxd 7/21/03 3:28 PM Page 28

28 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

2-33
Do = Di + 2W
D̄o = D̄i + 2W̄ = 208.92 + 2(5.33)
= 219.58 mm

t Do = t = t Di + 2tw
all
= 1.30 + 2(0.13) = 1.56 mm
Do = 219.58 ± 1.56 mm Ans.

2-34
Do = Di + 2W
D̄o = D̄i + 2W̄ = 3.734 + 2(0.139)
= 4.012 mm
 2 1/2
t Do = t 2 = tDo
+ (2 tw ) 2
all

= [0.0282 + (2) 2 (0.004) 2 ]1/2


= 0.029 in
Do = 4.012 ± 0.029 in Ans.

2-35

Do = Di + 2W
D̄o = D̄i + 2W̄ = 208.92 + 2(5.33)
= 219.58 mm

t Do = t 2 = [1.302 + (2) 2 (0.13) 2 ]1/2
all
= 1.33 mm
Do = 219.58 ± 1.33 mm Ans.

2-36
(a) F
w w= F−W
W
w̄ = F̄ − W̄ = 0.106 − 0.139
= −0.033 in

tw = t = 0.003 + 0.004
all
tw = 0.007 in
wmax = w̄ + tw = −0.033 + 0.007 = −0.026 in
wmin = w̄ − tw = −0.033 − 0.007 = −0.040 in

The minimum “squeeze” is 0.026 in. Ans.


shi20396_ch02.qxd 8/6/03 11:07 AM Page 29

Chapter 2 29

(b) Y
Ymax = D̄o = 4.012 in
Do w Ymin = max[0.99 D̄o , D̄o − 0.06]
= max[3.9719, 3.952] = 3.972 in
Y = 3.992 ± 0.020 in
Do + w − Y = 0
w = Y − D̄o
w̄ = Ȳ − D̄o = 3.992 − 4.012 = −0.020 in

tw = t = tY + t Do = 0.020 + 0.036 = 0.056 in
all
w = −0.020 ± 0.056 in

wmax = 0.036 in O-ring is more likely compressed than free prior to assembly of the
wmin = −0.076 in end plate.

2-37
(a) Figure defines w as gap.

F
w w=F−W
W
w̄ = F̄ − W̄
= 4.32 − 5.33 = −1.01 mm

tw = t = t F + tW = 0.13 + 0.13 = 0.26 mm
all
wmax = w̄ + tw = −1.01 + 0.26 = −0.75 mm
wmin = w̄ − tw = −1.01 − 0.26 = −1.27 mm

The O-ring is “squeezed” at least 0.75 mm.


(b) Y
Ymax = D̄o = 219.58 mm
Do w
Ymin = max[0.99 D̄o , D̄o − 1.52]
= max[0.99(219.58, 219.58 − 1.52)]
= 217.38 mm
Y = 218.48 ± 1.10 mm
From the figure, the stochastic equation is:
Do + w = Y
or, w = Y − Do
w̄ = Ȳ − D̄o = 218.48 − 219.58 = −1.10 mm

tw = t = tY + t Do = 1.10 + 0.34 = 1.44 mm
all
wmax = w̄ + tw = −1.10 + 1.44 = 0.34 mm
wmin = w̄ − tw = −1.10 − 1.44 = −2.54 mm
The O-ring is more likely to be circumferentially compressed than free prior to as-
sembly of the end plate.
shi20396_ch02.qxd 7/21/03 3:28 PM Page 30

30 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

2-38
wmax = −0.020 in, wmin = −0.040 in
w
b c d 1
w̄ = (−0.020 + (−0.040)) = −0.030 in
2
a 1
tw = (−0.020 − (−0.040)) = 0.010 in
2
b = 0.750 ± 0.001 in
c = 0.120 ± 0.005 in
d = 0.875 ± 0.001 in
w̄ = ā − b̄ − c̄ − d̄
−0.030 = ā − 0.875 − 0.120 − 0.750
ā = 0.875 + 0.120 + 0.750 − 0.030
ā = 1.715 in
Absolute:
tw = t = 0.010 = ta + 0.001 + 0.005 + 0.001
all
ta = 0.010 − 0.001 − 0.005 − 0.001
= 0.003 in
a = 1.715 ± 0.003 in Ans.
Statistical: For a normal distribution of dimensions

tw2 = t 2 = ta2 + tb2 + tc2 + td2
all
1/2
ta = tw2 − tb2 − tc2 − td2
= (0.0102 − 0.0012 − 0.0052 − 0.0012 ) 1/2 = 0.0085
a = 1.715 ± 0.0085 in Ans.

2-39
x n nx nx 2
93 19 1767 164 311
95 25 2375 225 625
97 38 3685 357 542
99 17 1683 166 617
101 12 1212 122 412
103 10 1030 106 090
105 5 525 55 125
107 4 428 45 796
109 4 436 47 524
111 2 222 24 624
136 13 364 1315 704

x̄ = 13 364/136 = 98.26 kpsi


 1/2
1 315 704 − 13 3642 /136
sx = = 4.30 kpsi
135
shi20396_ch02.qxd 7/21/03 3:28 PM Page 31

Chapter 2 31

Under normal hypothesis,


z 0.01 = (x0.01 − 98.26)/4.30
x0.01 = 98.26 + 4.30z 0.01
= 98.26 + 4.30(−2.3267)
.
= 88.26 = 88.3 kpsi Ans.

2-40 From Prob. 2-39, µx = 98.26 kpsi, and σ̂x = 4.30 kpsi.
C x = σ̂x /µx = 4.30/98.26 = 0.043 76
From Eqs. (2-18) and (2-19),
µ y = ln(98.26) − 0.043 762 /2 = 4.587
"
σ̂ y = ln(1 + 0.043 762 ) = 0.043 74

For a yield strength exceeded by 99% of the population,


z 0.01 = (ln x0.01 − µ y )/σ̂ y ⇒ ln x0.01 = µ y + σ̂ y z 0.01
From Table A-10, for 1% failure, z 0.01 = −2.326. Thus,
ln x0.01 = 4.587 + 0.043 74(−2.326) = 4.485
x0.01 = 88.7 kpsi Ans.
The normal PDF is given by Eq. (2-14) as
  
1 1 x − 98.26 2
f (x) = √ exp −
4.30 2π 2 4.30

For the lognormal distribution, from Eq. (2-17), defining g(x),


  
1 1 ln x − 4.587 2
g(x) = √ exp −
x(0.043 74) 2π 2 0.043 74

x (kpsi) f/(Nw) f (x) g (x) x (kpsi) f/(Nw) f (x) g (x)


92 0.000 00 0.032 15 0.032 63 102 0.036 76 0.063 56 0.061 34
92 0.069 85 0.032 15 0.032 63 104 0.036 76 0.038 06 0.037 08
94 0.069 85 0.056 80 0.058 90 104 0.018 38 0.038 06 0.037 08
94 0.091 91 0.056 80 0.058 90 106 0.018 38 0.018 36 0.018 69
96 0.091 91 0.080 81 0.083 08 106 0.014 71 0.018 36 0.018 69
96 0.139 71 0.080 81 0.083 08 108 0.014 71 0.007 13 0.007 93
98 0.139 71 0.092 61 0.092 97 108 0.014 71 0.007 13 0.007 93
98 0.062 50 0.092 61 0.092 97 110 0.014 71 0.002 23 0.002 86
100 0.062 50 0.085 48 0.083 67 110 0.007 35 0.002 23 0.002 86
100 0.044 12 0.085 48 0.083 67 112 0.007 35 0.000 56 0.000 89
102 0.044 12 0.063 56 0.061 34 112 0.000 00 0.000 56 0.000 89

Note: rows are repeated to draw histogram


shi20396_ch02.qxd 7/21/03 3:28 PM Page 32

32 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

0.16 Histogram
f (x)
0.14 g(x)

0.12

Probability density
0.1

0.08

0.06

0.04

0.02

0
90 92 94 96 98 100 102 104 106 108 110 112
x (kpsi)

The normal and lognormal are almost the same. However the data is quite skewed and
perhaps a Weibull distribution should be explored. For a method of establishing the
Weibull parameters see Shigley, J. E., and C. R. Mischke, Mechanical Engineering Design,
McGraw-Hill, 5th ed., 1989, Sec. 4-12.

2-41 Let x = (S fe ) 104


x0 = 79 kpsi, θ = 86.2 kpsi, b = 2.6
Eq. (2-28)
x̄ = x0 + (θ − x0 ) (1 + 1/b)
x̄ = 79 + (86.2 − 79) (1 + 1/2.6)
= 79 + 7.2 (1.38)
From Table A-34, (1.38) = 0.888 54
x̄ = 79 + 7.2(0.888 54) = 85.4 kpsi Ans.
Eq. (2-29)
σ̂x = (θ − x0 )[ (1 + 2/b) − 2 (1 + 1/b)]1/2
= (86.2 − 79)[ (1 + 2/2.6) − 2 (1 + 1/2.6)]1/2
= 7.2[0.923 76 − 0.888 542 ]1/2
= 2.64 kpsi Ans.
σ̂x 2.64
Cx = = = 0.031 Ans.
x̄ 85.4

2-42
x = Sut
x0 = 27.7, θ = 46.2, b = 4.38
µx = 27.7 + (46.2 − 27.7) (1 + 1/4.38)
= 27.7 + 18.5 (1.23)
= 27.7 + 18.5(0.910 75)
= 44.55 kpsi Ans.
shi20396_ch02.qxd 7/21/03 3:28 PM Page 33

Chapter 2 33

σ̂x = (46.2 − 27.7)[ (1 + 2/4.38) − 2 (1 + 1/4.38)]1/2


= 18.5[ (1.46) − 2 (1.23)]1/2
= 18.5[0.8856 − 0.910 752 ]1/2
= 4.38 kpsi Ans.
4.38
Cx = = 0.098 Ans.
44.55
From the Weibull survival equation
   
x − x0 b
R = exp − =1− p
θ − x0
   
x40 − x0 b
R40 = exp − = 1 − p40
θ − x0
   
40 − 27.7 4.38
= exp − = 0.846
46.2 − 27.7
p40 = 1 − R40 = 1 − 0.846 = 0.154 = 15.4% Ans.

2-43
x = Sut
x0 = 151.9, θ = 193.6, b = 8
µx = 151.9 + (193.6 − 151.9) (1 + 1/8)
= 151.9 + 41.7 (1.125)
= 151.9 + 41.7(0.941 76)
= 191.2 kpsi Ans.
σ̂x = (193.6 − 151.9)[ (1 + 2/8) − 2 (1 + 1/8)]1/2
= 41.7[ (1.25) − 2 (1.125)]1/2
= 41.7[0.906 40 − 0.941 762 ]1/2
= 5.82 kpsi Ans.
5.82
Cx = = 0.030
191.2

2-44
x = Sut
x0 = 47.6, θ = 125.6, b = 11.84
x̄ = 47.6 + (125.6 − 47.6) (1 + 1/11.84)
x̄ = 47.6 + 78 (1.08)
= 47.6 + 78(0.959 73) = 122.5 kpsi
σ̂x = (125.6 − 47.6)[ (1 + 2/11.84) − 2 (1 + 1/11.84)]1/2
= 78[ (1.08) − 2 (1.17)]1/2
= 78(0.959 73 − 0.936 702 ) 1/2
= 22.4 kpsi
shi20396_ch02.qxd 7/21/03 3:28 PM Page 34

34 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

From Prob. 2-42   


x − x0 b
p = 1 − exp −
θ − θ0
   
100 − 47.6 11.84
= 1 − exp −
125.6 − 47.6
= 0.0090 Ans.
y = Sy
y0 = 64.1, θ = 81.0, b = 3.77
ȳ = 64.1 + (81.0 − 64.1) (1 + 1/3.77)
= 64.1 + 16.9 (1.27)
= 64.1 + 16.9(0.902 50)
= 79.35 kpsi
σ y = (81 − 64.1)[ (1 + 2/3.77) − (1 + 1/3.77)]1/2
σ y = 16.9[(0.887 57) − 0.902 502 ]1/2
= 4.57 kpsi
   
y − y0 3.77
p = 1 − exp −
θ − y0
   
70 − 64.1 3.77
p = 1 − exp − = 0.019 Ans.
81 − 64.1

2-45 x = Sut = W[122.3, 134.6, 3.64] kpsi, p(x > 120) = 1 = 100% since x0 > 120 kpsi
   
133 − 122.3 3.64
p(x > 133) = exp −
134.6 − 122.3

= 0.548 = 54.8% Ans.

2-46 Using Eqs. (2-28) and (2-29) and Table A-34,


µn = n 0 + (θ − n 0 ) (1 + 1/b) = 36.9 + (133.6 − 36.9) (1 + 1/2.66) = 122.85 kcycles
σ̂n = (θ − n 0 )[ (1 + 2/b) − 2 (1 + 1/b)] = 34.79 kcycles
For the Weibull density function, Eq. (2-27),
 2.66−1   2.66 
2.66 n − 36.9 n − 36.9
f W (n) = exp −
133.6 − 36.9 133.6 − 36.9 133.6 − 36.9

For the lognormal distribution, Eqs. (2-18) and (2-19) give,


µ y = ln(122.85) − (34.79/122.85) 2 /2 = 4.771
"
σ̂ y = [1 + (34.79/122.85) 2 ] = 0.2778
shi20396_ch02.qxd 7/21/03 3:28 PM Page 35

Chapter 2 35

From Eq. (2-17), the lognormal PDF is


  2 
1 1 ln n − 4.771
f L N (n) = √ exp −
0.2778 n 2π 2 0.2778

We form a table of densities f W (n) and f L N (n) and plot.

n (kcycles) f W (n) f L N (n)


40 9.1E-05 1.82E-05
50 0.000 991 0.000 241
60 0.002 498 0.001 233
70 0.004 380 0.003 501
80 0.006 401 0.006 739
90 0.008 301 0.009 913
100 0.009 822 0.012 022
110 0.010 750 0.012 644
120 0.010 965 0.011 947
130 0.010 459 0.010 399
140 0.009 346 0.008 492
150 0.007 827 0.006 597
160 0.006 139 0.004 926
170 0.004 507 0.003 564
180 0.003 092 0.002 515
190 0.001 979 0.001 739
200 0.001 180 0.001 184
210 0.000 654 0.000 795
220 0.000 336 0.000 529

f (n)
0.014 LN
W
0.012

0.010

0.008

0.006

0.004

0.002

0
0 50 100 150 200 250
n, kcycles

The Weibull L10 life comes from Eq. (2-26) with a reliability of R = 0.90. Thus,
n 0.10 = 36.9 + (133 − 36.9)[ln(1/0.90)]1/2.66 = 78.1 kcycles Ans.
shi20396_ch02.qxd 7/21/03 3:28 PM Page 36

36 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

The lognormal L10 life comes from the definition of the z variable. That is,
ln n 0 = µ y + σ̂ y z or n 0 = exp(µ y + σ̂ y z)
From Table A-10, for R = 0.90, z = −1.282. Thus,
n 0 = exp[4.771 + 0.2778(−1.282)] = 82.7 kcycles Ans.

2-47 Form a table

x g(x)
i L(10−5) fi −5
f i x(10 ) 2 −10
f i x (10 ) (105 )
1 3.05 3 9.15 27.9075 0.0557
2 3.55 7 24.85 88.2175 0.1474
3 4.05 11 44.55 180.4275 0.2514
4 4.55 16 72.80 331.24 0.3168
5 5.05 21 106.05 535.5525 0.3216
6 5.55 13 72.15 400.4325 0.2789
7 6.05 13 78.65 475.8325 0.2151
8 6.55 6 39.30 257.415 0.1517
9 7.05 2 14.10 99.405 0.1000
10 7.55 0 0 0 0.0625
11 8.05 4 32.20 259.21 0.0375
12 8.55 3 25.65 219.3075 0.0218
13 9.05 0 0 0 0.0124
14 9.55 0 0 0 0.0069
15 10.05 1 10.05 101.0025 0.0038
100 529.50 2975.95

x̄ = 529.5(105 )/100 = 5.295(105 ) cycles Ans.


 1/2
2975.95(1010 ) − [529.5(105 )]2 /100
sx =
100 − 1
= 1.319(105 ) cycles Ans.
C x = s/x̄ = 1.319/5.295 = 0.249
µ y = ln 5.295(105 ) − 0.2492 /2 = 13.149
"
σ̂ y = ln(1 + 0.2492 ) = 0.245
  
1 1 ln x − µ y 2
g(x) = √ exp −
x σ̂ y 2π 2 σ̂ y
  
1.628 1 ln x − 13.149 2
g(x) = exp −
x 2 0.245
shi20396_ch02.qxd 7/21/03 3:28 PM Page 37

Chapter 2 37

105 g(x)
0.5

Superposed
0.4
histogram
and PDF

0.3

0.2

0.1

0
3.05(105) 10.05(105)
x, cycles

2-48
x = Su = W[70.3, 84.4, 2.01]
Eq. (2-28) µx = 70.3 + (84.4 − 70.3) (1 + 1/2.01)
= 70.3 + (84.4 − 70.3) (1.498)
= 70.3 + (84.4 − 70.3)0.886 17
= 82.8 kpsi Ans.
Eq. (2-29) σ̂x = (84.4 − 70.3)[ (1 + 2/2.01) − 2 (1 + 1/2.01)]1/2
σ̂x = 14.1[0.997 91 − 0.886 172 ]1/2
= 6.502 kpsi
6.502
Cx = = 0.079 Ans.
82.8

2-49 Take the Weibull equation for the standard deviation


σ̂x = (θ − x0 )[ (1 + 2/b) − 2 (1 + 1/b)]1/2

and the mean equation solved for x̄ − x0


x̄ − x0 = (θ − x0 ) (1 + 1/b)

Dividing the first by the second,

σ̂x [ (1 + 2/b) − 2 (1 + 1/b)]1/2


=
x̄ − x0 (1 + 1/b)

4.2 (1 + 2/b) √
= − 1 = R = 0.2763
49 − 33.8 2 (1 + 1/b)
shi20396_ch02.qxd 7/21/03 3:28 PM Page 38

38 Solutions Manual • Instructor’s Solution Manual to Accompany Mechanical Engineering Design

Make a table and solve for b iteratively

b 1 + 2/b 1 + 1/b (1 + 2/b) (1 + 1/b)


3 1.67 1.33 0.903 30 0.893 38 0.363
4 1.5 1.25 0.886 23 0.906 40 0.280
4.1 1.49 1.24 0.885 95 0.908 52 0.271
.
b = 4.068 Using MathCad Ans.
x̄ − x0 49 − 33.8
θ = x0 + = 33.8 +
(1 + 1/b) (1 + 1/4.068)
= 49.8 kpsi Ans.

2-50
x = S y = W[34.7, 39, 2.93] kpsi
x̄ = 34.7 + (39 − 34.7) (1 + 1/2.93)
= 34.7 + 4.3 (1.34)
= 34.7 + 4.3(0.892 22) = 38.5 kpsi
σ̂x = (39 − 34.7)[ (1 + 2/2.93) − 2 (1 + 1/2.93)]1/2
= 4.3[ (1.68) − 2 (1.34)]1/2
= 4.3[0.905 00 − 0.892 222 ]1/2
= 1.42 kpsi Ans.
C x = 1.42/38.5 = 0.037 Ans.

2-51

x (Mrev) f fx f x2
1 11 11 11
2 22 44 88
3 38 114 342
4 57 228 912
5 31 155 775
6 19 114 684
7 15 105 735
8 12 96 768
9 11 99 891
10 9 90 900
11 7 77 847
12 5 60 720
Sum 78 237 1193 7673

µx = 1193(106 )/237 = 5.034(106 ) cycles



7673(1012 ) − [1193(106 )]2 /237
σ̂x = = 2.658(106 ) cycles
237 − 1
C x = 2.658/5.034 = 0.528
shi20396_ch02.qxd 7/21/03 3:28 PM Page 39

Chapter 2 39

From Eqs. (2-18) and (2-19),


µ y = ln[5.034(106 )] − 0.5282 /2 = 15.292
"
σ̂ y = ln(1 + 0.5282 ) = 0.496
From Eq. (2-17), defining g(x),
  2 
1 1 ln x − 15.292
g (x) = √ exp −
x(0.496) 2π 2 0.496

x (Mrev) f/(Nw) g(x) · (106 )


0.5 0.000 00 0.000 11
0.5 0.046 41 0.000 11
1.5 0.046 41 0.052 04
1.5 0.092 83 0.052 04
2.5 0.092 83 0.169 92
2.5 0.160 34 0.169 92
3.5 0.160 34 0.207 54 0.25 Histogram
3.5 0.240 51 0.207 54 PDF

4.5 0.240 51 0.178 48 0.2


4.5 0.130 80 0.178 48
5.5 0.130 80 0.131 58
g(x)(106)

0.15
5.5 0.080 17 0.131 58
6.5 0.080 17 0.090 11
6.5 0.063 29 0.090 11 0.1

7.5 0.063 29 0.059 53


7.5 0.050 63 0.059 53 0.05
8.5 0.050 63 0.038 69
8.5 0.046 41 0.038 69 0
0 2 4 6 8 10 12
9.5 0.046 41 0.025 01 x, Mrev
9.5 0.037 97 0.025 01
10.5 0.037 97 0.016 18
10.5 0.029 54 0.016 18
11.5 0.029 54 0.010 51
11.5 0.021 10 0.010 51
12.5 0.021 10 0.006 87
12.5 0.000 00 0.006 87

ln x − µ y
z= ⇒ ln x = µ y + σ̂ y z = 15.292 + 0.496z
σ̂ y
L10 life, where 10% of bearings fail, from Table A-10, z = −1.282. Thus,
ln x = 15.292 + 0.496(−1.282) = 14.66
∴ x = 2.32 × 106 rev Ans.
MECHANICAL ENGINEERING DESIGN
TUTORIAL 4-14: STRESS CONCENTRATION

ORIGIN OF STRESS CONCENTRATIONS

Machine members often have regions in which the state of stress is significantly greater than
theoretical predictions as a result of:

1. Geometric discontinuities or stress raisers such as holes, notches, and fillets;


2. Internal microscopic irregularities (non-homogeneities) of the material created by such
manufacturing processes as casting and molding;
3. Surface irregularities such as cracks and marks created by machining operations.

These stress concentrations are highly localized effects which are functions of geometry and
loading. In this tutorial, we will examine the standard method of accounting for stress
concentrations caused by geometric features. Specifically, we will discuss the application of a
theoretical or geometric stress-concentration factor for determination of the true state of stress in
the vicinity of stress raisers.

THEORETICAL (GEOMETRIC) STRESS-CONCENTRATION FACTOR, K t AND K ts

In order to predict the “actual” stress resulting from a geometric stress raiser, a theoretical stress-
concentration factor is applied to the nominal stress. For a part subjected to a normal stress, the true
stress in the immediate neighborhood of the geometric discontinuity is calculated as:

σ max = Ktσ 0 (Text Eq. 4-48)

where,
Kt = Theoretical stress-concentration factor
σ 0 = Nominal normal stress

Similarly, we can also estimate the highly localized amplification of shear stress in the vicinity of a
geometric stress concentration,

τ max = Ktsτ 0
where,
K ts = Theoretical stress-concentration factor for shear
τ 0 = Nominal shear stress

The nominal stress of the above equations is typically derived from the elementary strength of
materials equations, using either a net or a gross cross section.


Text refers to Mechanical Engineering Design, 7th edition text by Joseph Edward Shigley, Charles R.
Mischke and Richard G. Budynas; equations and examples with the prefix T refer to the present tutorial.
Characteristics of Stress-Concentration Factors

1. Function of the geometry or shape of the part, but not its size or material;
2. Function of the type of loading applied to the part (axial, bending or torsional);
3. Function of the specific geometric stress raiser in the part (e.g. fillet radius,
notch, or hole)
4. Always defined with respect to a particular nominal stress;
5. Typically assumes a linear elastic, homogeneous, isotropic material.

Determination of Kt Value

The stress-concentration factor, associated with a specific geometry and loading condition of a part,
can be derived through experimentation, analysis or computational methods.

1. Experimental Methods. Optical methods, such as photoelasticity, are very


dependable and widely used for experimentally determining the stress
concentration at a point on a part. However, several alternative methods have
been used historically: the grid method, brittle-coating, brittle-model and strain
gauge.
2. Analytical Methods. The theory of elasticity can be used to analyze certain
geometrical shapes to calculate stress-concentration factors.
3. Computational Methods. Finite-element techniques provide a powerful and
inexpensive computational method of assessing stress-concentration factors.

Following are comparisons of stress-concentration factors derived using experimental, analytical


and computational methods for a rectangular filleted bar in tension and in pure bending. Text Figure
A-15-5 provides tensile test results for the bar in simple tension while Figure T6-2-1 shows

TEXT FIGURE A-15-5: Rectangular filleted bar in tension or simple


compression. σ 0 = F / A, where A = dt and t is the thickness.

Shigley, Mischke & Budynas Machine Design Tutorial 4-14: Stress Concentration 2/9
FIGURE T6-2-1: Stress distribution in a rectangular filleted bar in
simple tension obtained through photoelastic procedures. (S. P.
Timoshenko and J. N. Goodier, "Theory of Elasticity," Third Edition,
McGraw-Hill, Inc., 1969.)

the fringe pattern captured photographically from a photoelasticity experiment. Fringe


patterns are indicative of the stress intensity which is directly proportional to the maximum
shear stress and the principal stresses:

σ Intensity = 2τ max = σ 1 − σ 3 .

Stress-concentration factors can be developed from these contours.

Finally, Figure T6-2-2 contains the graphical results of a finite element analysis of the bar in
tension. Since the bar geometry and the loads applied to the bar are symmetrical with respect
to the longitudinal axis, the model only needs to incorporate the upper half of the bar; the
analytical results for the lower half of the bar will be a mirror image of those in the top half.

The finite element model plot contains contours of the σ x component of stress. However,
since the stress-concentration factor is applied to the dominant component of the stress, σ x
in this model, the finite element model can be queried for σ x to estimate the value of Kt
directly

σ x , FiniteElement
Kt =
σ0

where the nominal stress must be defined for the section geometry and applied loading.

Shigley, Mischke & Budynas Machine Design Tutorial 4-14: Stress Concentration 3/9
FIGURE T6-2-2: Stress contours of σ x generated by a
finite element model of one half of a rectangular filleted
bar in tension.

Similarly, Text Figure A-15-6, Figure T6-2-3 and Figure T6-2-4 respectfully provide results
obtained by applying bending and photoelastic testing and finite element analysis to a
rectangular filleted bar in pure bending.

TEXT FIGURE A-15-6: Rectangular filleted bar in pure bending.


σ 0 = Mc / I , where c = d / 2, I = td 3 /12 and t is the thickness.

Shigley, Mischke & Budynas Machine Design Tutorial 4-14: Stress Concentration 4/9
FIGURE T6-2-3: Stress distribution in a rectangular filleted bar in pure
bending obtained through photoelastic procedures. (By permission of S. P.
Timoshenko and J. N. Goodier; the figure was included in, "Theory of
Elasticity," Third Edition, McGraw-Hill, Inc., 1969.)

FIGURE T6-2-4: Stress contours of σ x generated by a


finite element model of one half of the rectangular filleted
bar in pure bending.

Stress-concentration factors, derived through many years of practice, have been catalogued
for numerous geometric features and loading configurations in two authoritative resources:

1. Pilkey, W. D., Peterson’s Stress Concentration Factors, 2nd ed., Wiley


Interscience, 1997.

Shigley, Mischke & Budynas Machine Design Tutorial 4-14: Stress Concentration 5/9
2. Young, W. C. and R. G. Budynas, Roark’s Formulas for Stress and Strain, 7th
ed., McGraw-Hill, 2001.

Application to Ductile and Brittle Materials for Static Loading

Ductile Materials. While stress concentration must be considered for fatigue and impact
loading of most materials, stress-concentration factors are seldom applied to ductile
materials under static loading. This design practice is justified by four points:

1. Areas of high stress caused by stress concentrations are highly localized and will
not dictate the performance of the part. Rather, it is assumed that the stress state
in the cross section as a whole is below the general yield condition;
2. If the magnitude of the loading is large enough to cause yielding due to the stress
concentration, the localized area will plastically deform immediately upon
loading;
3. Ductile materials typically work-harden (strain-strengthen) on yielding, resulting
in a localized increase in material strength;
4. The static load is never cycled.

It is important to note, that even though the stress-concentration factor is not usually applied
to estimate the stresses at a stress raiser in a ductile material, the higher state of stress does
in fact exist.

Ductile Material Practice: σ max = σ 0

Brittle Materials. Stress-concentration factors are always required for brittle materials,
regardless of the loading conditions, since brittle failure results in fracture. This type of
failure is characteristic of brittle materials which do not exhibit a yielding or plastic range.
As a consequence of brittle fracture, the part breaks into two or more pieces having no load
carrying capability. To avoid such catastrophic failure, the design practice is to always use a
stress-concentration factor for brittle materials to ensure that the state of stress is accurately
represented.

Brittle Material Practice: σ max = K tσ 0

Shigley, Mischke & Budynas Machine Design Tutorial 4-14: Stress Concentration 6/9
Example T6.2.1:

Problem Statement: A bar machined from an ASTM No. 20 cast iron, a brittle
material, is subjected to a static axial load.

Find: The critical section of the bar.

Solution Methodology:

1. Assume the stress concentrations do not interact and analyze the


localized effect of each stress concentration separately.
2. Compute the actual stress in the shoulder by taking into account the
stress concentration caused by a fillet radius in a rectangular bar in
tension.
3. Compute the actual stress in the region immediately adjacent to the
hole by applying the stress-concentration factor associated for a bar in
tension with a transverse hole.
4. Evaluate the critical section as the region having the highest actual
stress.

Schematic:

3/16” R. ¾ ” D.

1000 lb 1000 lb
D=2¼” d = 1½”

¼”

Solution:

1. Material Properties: Sut = 20 kpsi

2. Actual Stress in Shoulder


a. Stress-Concentration Factor from Text Figure A-15-5:
D 2.25 in. r 0.1875 in.
for = = 1.5 and = = 0.125 in., Kt = 1.95
d 1.5 in. d 1.5 in.

Shigley, Mischke & Budynas Machine Design Tutorial 4-14: Stress Concentration 7/9
b. Nominal stress, as defined in the caption of Text Figure A-15-5:
F F 1000 lb
σ0 = = = = 2666.7 psi = 2.67 kpsi
A0 dt (1.5 in.)(0.25 in.)
c. Actual stress at fillet:
σ max = Ktσ 0 = 1.95(2666.7 psi) = 5200 psi = 5.20 kpsi

Kt = 1.95

TEXT FIGURE A-15-5: Rectangular filleted bar in tension or simple


compression. σ 0 = F / A, where A = dt and t is the thickness.

3. Actual Stress at Hole Perimeter


a. From Text Figure A-15-1 shown on the next page:
d 0.75 in.
for = = 0.5, Kt = 2.19
w 1.5 in.
b. Nominal stress, as defined in the caption of Text Figure A-15-1:
F F 1000 lb
σ0 = = =
A0 ( w − d )t (1.5 in. − 0.75 in.)(0.25 in.)

= 5333.3 psi = 5.333 kpsi

Shigley, Mischke & Budynas Machine Design Tutorial 4-14: Stress Concentration 8/9
Kt = 2.19

TEXT FIGURE A-15-1: Bar in tension or simple compression with a


transverse hole. σ 0 = F / A, where A = ( w − d )t and where t is the thickness.

c. Actual stress at hole perimeter:


σ max = Ktσ 0 = 2.19(5.333 kpsi) = 11.68 kpsi

4. Since the actual stress at the hole is greater than the actual stress at the fillet, the
hole represents the critical section for this part.

Shigley, Mischke & Budynas Machine Design Tutorial 4-14: Stress Concentration 9/9
MECHANICAL ENGINEERING DESIGN
TUTORIAL 4 –15: PRESSURE VESSEL DESIGN

PRESSURE VESSEL DESIGN MODELS FOR CYLINDERS:

1. Thick-walled Cylinders
2. Thin-walled Cylinders

THICK-WALL THEORY

• Thick-wall theory is developed from the Theory of Elasticity which yields the state of
stress as a continuous function of radius over the pressure vessel wall. The state of
stress is defined relative to a convenient cylindrical coordinate system:

1. σ t — Tangential Stress
2. σ r — Radial Stress
3. σ l — Longitudinal Stress

• Stresses in a cylindrical pressure vessel depend upon the ratio of the inner radius to
the outer radius ( ro / ri ) rather than the size of the cylinder.
• Principal Stresses ( σ 1 , σ 2 , σ 3 )

1. Determined without computation of Mohr’s Circle;


2. Equivalent to cylindrical stresses ( σ t , σ r , σ l )

• Applicable for any wall thickness-to-radius ratio.

Cylinder under Pressure

Consider a cylinder, with capped ends, subjected to an internal pressure, pi, and an
external pressure, po,

ri
σr
σl
σt pi

σl σ
r σt
ro
po

FIGURE T4-15-1

Text Eq. refers to Mechanical Engineering Design, 7th edition text by Joseph Edward Shigley, Charles
R. Mischke and Richard G. Budynas; equations and figures with the prefix T refer to the present tutorial.
The cylinder geometry is defined by the inside radius, ri , the outside radius, ro , and the
cylinder length, l. In general, the stresses in the cylindrical pressure vessel ( σ t , σ r , σ l )
can be computed at any radial coordinate value, r, within the wall thickness bounded by
ri and ro , and will be characterized by the ratio of radii, ζ = ro / ri . These cylindrical
stresses represent the principal stresses and can be computed directly using Eq. 4-50 and
4-52. Thus we do not need to use Mohr’s circle to assess the principal stresses.

Tangential Stress:

pi ri2 − po ro2 − ri2 ro2 ( po − pi ) / r 2


σt = for ri ≤ r ≤ ro (Text Eq. 4-50)
ro2 − ri2

Radial Stress:

pi ri2 − po ro2 + ri2 ro2 ( po − pi ) / r 2


σr = for ri ≤ r ≤ ro (Text Eq. 4-50)
ro2 − ri2

Longitudinal Stress:

• Applicable to cases where the cylinder carries the longitudinal load, such as
capped ends.
• Only valid far away from end caps where bending, nonlinearities and stress
concentrations are not significant.

pi ri2 − po ro2
σl = for ri ≤ r ≤ ro (Modified Text Eq. 4-52)
ro2 − ri2

Two Mechanical Design Cases

1. Internal Pressure Only ( p o = 0 )


2. External Pressure Only ( p i = 0 )

Design Case 1: Internal Pressure Only

• Only one case to consider — the critical section which exists at r = ri .


• Substituting p o = 0 into Eqs. (4-50) and incorporating ζ = ro / ri , the
largest value of each stress component is found at the inner surface:

ro2 + ri2 ζ 2 +1
σ t (r = ri ) = σ t ,max = pi 2 2 = pi 2 = pi Cti (T-1)
ro − ri ζ −1

Shigley, Mischke & Budynas Machine Design Tutorial 4–15: Pressure Vessel Design 2/10
ζ 2 + 1 ro2 + ri 2
where Cti = 2 = is a function of cylinder geometry only.
ζ − 1 ro2 − ri 2
σ r (r = ri ) = σ r ,max = − pi Natural Boundary Condition (T-2)

• Longitudinal stress depends upon end conditions:

pi Cli Capped Ends (T-3a)


σl =
0 Uncapped Ends (T-3b)

1
where Cli = .
ζ −1
2

Design Case 2: External Pressure Only

• The critical section is identified by considering the state of stress at two


points on the cylinder: r = ri and r = ro. Substituting pi = 0 into Text
Eqs. (4-50) for each case:

r = ri σ r (r = ri ) = 0 Natural Boundary Condition (T-4a)

2ro2 2ζ 2
σ t (r = ri ) = σ t ,max = − po = − p o = − poCto (T-4b)
ro2 − ri2 ζ 2 −1

2ζ 2 2ro2
where, Cto = = .
ζ 2 − 1 ro2 − ri 2
r = ro σ r (r = ro ) = σ r ,max = − po Natural Boundary Condition (T-5a)

ro2 + ri2 ζ 2 +1
σ t (r = ro ) = − po = − po = − poCti (T-5b)
ro2 − ri2 ζ 2 −1

• Longitudinal stress for a closed cylinder now depends upon external


pressure and radius while that of an open-ended cylinder remains zero:

− poClo Capped Ends (T-6a)


σl =
0 Uncapped Ends (T-6b)

Shigley, Mischke & Budynas Machine Design Tutorial 4–15: Pressure Vessel Design 3/10
ζ2
where Clo = .
ζ 2 −1

Example T4.15.1: Thick-wall Cylinder Analysis

Problem Statement: Consider a cylinder subjected to an external pressure of


150 MPa and an internal pressure of zero. The cylinder has a 25 mm ID and a 50
mm OD, respectively. Assume the cylinder is capped.

Find:
1. the state of stress ( σ r , σ t , σ l ) at the inner and outer cylinder
surfaces;
2. the Mohr’s Circle plot for the inside and outside cylinder surfaces;
3. the critical section based upon the estimate of τ max .

Solution Methodology:
Since we have an external pressure case, we need to compute the state of
stress ( σ r , σ t , σ l ) at both the inside and outside radius in order to determine
the critical section.
1. As the cylinder is closed and exposed to external pressure only,
Eq. (T-6a) may be applied to calculate the longitudinal stress
developed. This result represents the average stress across the wall
of the pressure vessel and thus may be used for both the inner and
outer radii analyses.
2. Assess the radial and tangential stresses using Eqs. (T-4) and (T-5)
for the inner and outer radii, respectively.
3. Assess the principal stresses for the inner and outer radii based
upon the magnitudes of ( σ r , σ t , σ l ) at each radius.
4. Use the principal stresses to calculate the maximum shear stress at
each radius.
5. Draw Mohr’s Circle for both states of stress and determine which
provides the critical section.

Solution:

1. Longitudinal Stress Calculation:

OD 50 mm ID 25 mm
ro = = = 25 mm ; ri = = = 12.5 mm
2 2 2 2

Compute the radius ratio, ζ


ro 25 mm
ζ = = = 2.0
ri 12.5 mm

Shigley, Mischke & Budynas Machine Design Tutorial 4–15: Pressure Vessel Design 4/10
Then,
ζ2 (2) 2
Clo = = = 1.3333 mm 2
ζ − 1 (2) − 1
2 2

ζ2
σ l (r = ri ) = σ l (r = ro ) = − po 2 = − poClo = (−150MPa)(1.3333 mm 2 )
ζ −1
σ l = −200 MPa

2. Radial & Tangential Stress Calculations:

Inner Radius (r = ri)


2ζ 2 2(2) 2
Cto = 2 = = 2.6667
ζ − 1 (2) 2 − 1
2ro2
σ t (r = ri ) = σ t ,max = − po 2 2 = − poCto = (−150 MPa)(2.6667)
ro − ri
σt (r = ri ) = −400 MPa Compressive

σ r (r = ri ) = 0 Natural Boundary Condition for pi = 0

Outer Radius (r = ro)


ζ 2 + 1 (2)2 + 1
Cti = = = 1.6667
ζ 2 − 1 (2) 2 − 1
ro2 + ri 2
σ t (r = ro ) = σ t ,min = − po = − poCti = (−150 MPa)(1.6667)
ro2 − ri 2

σ t (r = ro ) = −250 MPa Compressive

σ r (r = ri ) = − po = −150 MPa Natural Boundary Condition

3. Define Principal Stresses:

Inner Radius (r = ri ) Outer Radius (r = ro )

σ 1 = σ r = 0 MPa σ 1 = σ r = −150 MPa


σ 2 = σ l = −200 MPa σ 2 = σ l = −200 MPa
σ 3 = σ t = −400 MPa σ 3 = σ t = −250 MPa

4. Maximum Shear Stress Calculations:

σ 1 − σ 3 0 − (−400)
Inner Radius (r = ri ) τ max (r = ri ) = = = 200 MPa
2 2

Shigley, Mischke & Budynas Machine Design Tutorial 4–15: Pressure Vessel Design 5/10
σ 1 − σ 3 (−150) − (−250)
Outer Radius (r = ro ) τ max (r = ro ) = = = 50 MPa
2 2

5. Mohr’s Circles:

Inner Radius (r = ri )
τ
FIGURE T4-15-2
τ max = 200 MPa

σ 1 = 0 MPa
σ
σ 3 = -400 MPa

σ 2 = -200 MPa

Outer Radius (r = ro )
τ
FIGURE T4-15-3

σ 3 = -250 MPa σ 1 = −150 MPa

• τmax = 50 MPa

σ 2 = -200 MPa

Critical Section

τ max ( r = ri ) = 200 MPa ⇐ Critical Section is at Inside Radius!

Shigley, Mischke & Budynas Machine Design Tutorial 4–15: Pressure Vessel Design 6/10
THIN-WALL THEORY

• Thin-wall theory is developed from a Strength of Materials solution which yields the
state of stress as an average over the pressure vessel wall.
• Use restricted by wall thickness-to-radius ratio:
t 1
 According to theory, Thin-wall Theory is justified for ≤
r 20
t 1
 In practice, typically use a less conservative rule, ≤
r 10

• State of Stress Definition:

1. Hoop Stress, σ t , assumed to be uniform across wall thickness.


2. Radial Stress is insignificant compared to tangential stress, thus, σ r  0.
3. Longitudinal Stress, σ l
S Exists for cylinders with capped ends;
S Assumed to be uniformly distributed across wall thickness;
S This approximation for the longitudinal stress is only valid far away
from the end-caps.

4. These cylindrical stresses (σ t , σ r , σ l ) are principal stresses (σ t , σ r , σ l ) which


can be determined without computation of Mohr’s circle plot.

• Analysis of Cylinder Section

FIGURE T4-15-4

di

t
FV

FHoop FHoop
Pressure Acting over
Projected Vertical Area

Shigley, Mischke & Budynas Machine Design Tutorial 4–15: Pressure Vessel Design 7/10
The internal pressure exerts a vertical force, FV, on the cylinder wall which is
balanced by the tangential hoop stress, FHoop.

FV = pA proj = p{( d i )(1)} = pd i


FHoop = σ t Astressed = σ t {(t )(1)} = σ t t

¦ Fy = 0 = FV − 2 FHoop = pd i − 2σ t t

Solving for the tangential stress,


pd i
σt = Hoop Stress (Text Eq. 4-53)
2t
• Comparison of state of stress for cylinder under internal pressure verses external
pressure:
Internal Pressure Only
pdi
σt = Hoop Stress
2t
σr = 0 By Definition
pdi σ t
σl = = Capped Case (Text Eq.4-55)
4t 2

External Pressure Only


pd o
σt = Hoop Stress
2t
σr = 0 By Definition
pd o σ t
σl = = Capped Case
4t 2

Example T4.15.2: Thin-wall Theory Applied to Cylinder Analysis

Problem Statement: Repeat Example T1.1 using the Thin-wall Theory


(po = 150 MPa, pi = 0, ID = 25 mm, OD = 50 mm).

Find: The percent difference of the maximum shear stress estimates found using
the Thick-wall and Thin-wall Theories.

Shigley, Mischke & Budynas Machine Design Tutorial 4–15: Pressure Vessel Design 8/10
Solution Methodology:

1. Check t/r ratio to determine if Thin-wall Theory is applicable.


2. Use the Thin-wall Theory to compute the state of stress
3. Identify the principal stresses based upon the stress magnitudes.
4. Use the principal stresses to assess the maximum shear stress.
5. Calculate the percent difference between the maximum shear stresses
derived using the Thick-wall and Thin-wall Theories.

Solution:
t 12.5 mm 1 1 1
1. Check t/r Ratio: = = ² or
r 25 mm 2 20 10

The application of Thin-wall Theory to estimate the stress state of this


cylinder is thus not justified.

2. Compute stresses using the Thin-wall Theory to compare with Thick-


wall theory estimates.

a. Hoop Stress (average stress, uniform across wall)


− po d o − (150 MPa )(50 mm)
σt = = = −300 MPa
2t 2(12.5 mm)

b. Radial Stress σ r = 0 by definition

c. Longitudinal Stress (average stress, uniform across wall)


−p d σ
σ l = o o = t = −150 MPa
4t 2

3. Identify Principal Stresses in terms of “Average” Stresses:

σ 1 = σ r = 0 MPa
σ 2 = σ l = −150 MPa
σ 3 = σ t = −300 MPa

4. Maximum Shear Stress Calculation:

σ 1 − σ 3 0 − (−300 MPa )
τ max = = = +150 MPa
2 2

5. Percent Difference between Thin- and Thick-wall Estimates for the


Critical Section:

Shigley, Mischke & Budynas Machine Design Tutorial 4–15: Pressure Vessel Design 9/10
τ max,Thin − τ max,Thick
% Difference = ∗100%
τ max,Thick
(+150) − (+200)
= ∗ (100%) = −25%
(+200)

Ÿ Thin -wall estimate is 25% low! ⇐

Shigley, Mischke & Budynas Machine Design Tutorial 4–15: Pressure Vessel Design 10/10
MECHANICAL ENGINEERING DESIGN
TUTORIAL 4-17: PRESS AND SHRINK FITS

APPLICATION OF THICK-WALL CYLINDRICAL PRESSURE VESSEL THEORY


RELATING TO STRESSES DEVELOPED FROM INTERFERENCE FITS:

1. Design application which uses the cylindrical pressure vessel Thick-Wall Theory.
2. Stresses develop between cylinders due to the contact pressure generated by an
interference fit. The interference fit is achieved by pressing a larger inside member
into the smaller opening of an outside member. In the specific case of a shaft press fit
into the hub of a gear, the outside diameter (OD) of the shaft is slightly larger than the
inside hole diameter (ID) of the hub. The diametral difference between the shaft OD
and the ID of the hub hole is referred to as the interference fit.

• The radial deformation required by the interference fit causes an interfacial


pressure, p, to develop at the nominal radius, at r = R. Consequently, radial
and tangential stresses, σ r and σ t , are produced.
• Assuming uncapped ends (σ l = 0), a biaxial state of stress exists for which
two non-zero principal stresses must be considered.
• From the cylindrical pressure vessel theory, the radial and tangential stresses
represent principal stresses.
• The length of the outer member is assumed to be equal to the length of the
inner member.

ro
δ
Outer
Member R

R ro
ri

Inner
Member ri

(a) End view of inner and outer (b) Cross-section of cylinders


members, press fit together. showing internal outside radius
larger than external inside radius
by a small amount of δ.

FIGURE T4-17-1 Interference fit of two


cylinders of finite length and equal lengths.


Text Eq. refers to Mechanical Engineering Design, 7th edition text by Joseph Edward Shigley, Charles
R. Mischke and Richard G. Budynas; equations with the prefix T refer to the present tutorial.
3. Referring to Fig. T4-17-1, the geometric features of the cylindrical parts are defined
as:

ri = the inside radius of the inner cylinder


R = nominal radius of internal outside radius and external inside radius after assembly
ro = outside radius of the outer cylinder
δ = radial interference

INSIDE CYLINDER

• Inner member experiences an external pressure, po = p, resulting in compressive


tangential and radial stresses.
• Thick-Wall Theory may be applied with ro = R:

§ R 2 + ri2 ·
(σ t )i r =R = − po¨ 2 2 ¸
= − pCit (Text Eq. 4-57)
© R − ri ¹
(σ r )i r = R = − po = − p

OUTSIDE CYLINDER

• Outer member only experiences internal pressure, pi = p, resulting in tensile


tangential stress and compressive radial stress.
• Thick-Wall Theory is, as always, applicable with ri = R:
§ ro2 + R 2 ·
(σ t )o r = R = pi ¨ 2 2 ¸
= pCot (Text Eq. 4-58)
© ro − R ¹
(σ r )i r = R = − pi = − p

DEFINITION OF INTERFACIAL PRESSURE

We presently have two equations and three unknowns for both the inside and outside
cylinder analyses. A third equation which relates the contact pressure and the interference
can be derived by examining the deformation of the members.

Deflection Equation

The total radial interference may be defined as:


δ total = δ i + δ o
where,

Shigley, Mischke & Budynas Machine Design Tutorial 4-17: Press and Shrink Fits 2/11
δ i = decrease in radius of inner cylinder
δ o = increase in radius of hole
The deformation may also be expressed as:
δ total = pRK i + pRK o (Modified Text Eq. 4-59)
where the outside member constant, Ko, is defined as,

1 ª§ ro2 + R 2 · º 1
Ko = «¨ 2 2 ¸ + ν o» = [Co +ν o ]
Eo ¬«© ro − R ¹ ¼» Eo

Using the radius ratio form defined for the cylindrical pressure vessel
formulation, ζ o = ro / R , we can define

ζ o2 + 1
Co = 2
ζ o −1
Note that the Co term is a function of geometry only, while the member constant
term Ko is a function of both geometry and material parameters. Similarly for the
inside member constant Ki,

1 ª§ R 2 + ri2 · º 1
Ki = «¨ 2 2 ¸ − ν i » = [Ci − ν i ]
Ei «¬© R − ri ¹ »¼ Ei

with ζ i = R / ri , Ci is defined as:

ζ i2 + 1
Ci =
ζ i2 − 1
For the case of a solid shaft, ri = 0, ζ i = R / ri = ∞ and Ci = 1.

We can now solve for the deformation for a given class of interference fits,

δ total = [ K o + K i ] pR
Or, rearranging, the contact pressure, p, may be expressed as a function of the
interference without assumptions regarding material property values:

ª 1 º δ total
p=« » (Modified Text Eq. 4-60)
¬ Ko + Ki ¼ R

Shigley, Mischke & Budynas Machine Design Tutorial 4-17: Press and Shrink Fits 3/11
Example T4.17.1: Shaft & Hub Shrink Fit

Problem Statement: A carbon-steel gear hub having a nominal hole diameter of 1 inch
is to be shrink-fitted to a carbon-steel shaft using a class FN4 fit. The hub has a nominal
thickness of ½ inch.

Find:
1. the maximum tangential and radial stresses in the hub and shaft when the
loosest fit is obtained;
2. the same as (a), except using the tightest fit as a condition.

Solution Methodology:
1. Using Table A-1-2 of this document, identify the diametral size ranges for the
shaft and the hole based on an ANSI US Customary Standard class FN4 fit
which is discussed below.
2. Calculate the interferences for the loosest and tightest fits.
3. Compute the interfacial pressure for the loosest and tightest fits.
4. For each fit, calculate the radial and tangential stresses for both the hub and
the shaft.

Schematic:

Shaft

Hub

Solution:

1. Shaft and Hub Size Ranges:


This problem specifies a class FN4 fit. The class FN4 fit is a specification
defined by the American National Standards Institute (ANSI), American
Standard Limits for Cylindrical Parts ANSI B4.1-1978. The FN4 fit is a
“Force fit suitable for parts which can be highly stressed or for shrink fits
where the heavy pressing forces required are impractical.” Excerpts from the
ANSI B4.1-1978 standard are provided in Table A-1 of this tutorial.

From Table A-1-2, for a nominal shaft/hole diameter of 1 in., the appropriate
size range is 0.95–1.19 in. The allowable tolerances, in thousandths of an
inch, are:


Text Eq. refers to Mechanical Engineering Design, 7th edition text by Joseph Edward Shigley, Charles
R. Mischke and Richard G. Budynas; equations with the prefix T refer to the present tutorial.
Largest Tolerance Smallest Tolerance
(10-3 in.) (10-3 in.)

Hub hole +0.8 –0.0


Shaft +2.3 +1.8

The shaft and hub size ranges for a nominal 1 in. diameter are:

Largest Diameter Smallest Diameter


(in.) (in.)

Hub hole 1.0008 1.0000


Shaft 1.0023 1.0018

2. Diametral and Radial Interference Calculations:


Loosest Fit Tightest Fit

Hub hole 1.0008 1.0000


Shaft 1.0018 1.0023
–0.0010 –0.0023

The radial interferences are therefore δ = –0.0005 in. (loosest fit) and
δ = –0.001 15 in. (tightest fit). The interference is taken as a positive number by
convention. Consequently, the radial interferences for the two cases are:

δ = 0.0005 in. (loosest fit)


δ = 0.001 15 in. (tightest fit)

3. Interfacial Pressure

The interface pressure can be computed from Modified Eq. (4-60):

ª 1 º δ total
p=« »
¬ Ko + Ki ¼ R

For this problem, the geometric features have been defined as:

d 1 in. d
ro = +t = + 0.5 in. = 1.0 in. ri = 0.0 R= = 0.5 in.
2 2 2

Thus, for the hub (outside member),

Shigley, Mischke & Budynas Machine Design Tutorial 4-17: Press and Shrink Fits 5/11
ro (1 in.)
ζo = = =2
R (0.5 in.)
ζ o2 + 1 (2) 2 + 1 5
Co = = = = 1.6667
ζ o2 − 1 (2) 2 − 1 3
1 1
Ko = [Co +ν o ] = [1.667 + 0.292] = 6.5289 ×10−8 (1/psi)
Eo ( 30 ×10 psi )
6

Similarly for the shaft (inside member)

R
ζi = =∞
ri
ζ i2 + 1
Ci = = 1.0
ζ i2 − 1
1 1
K i = [Ci − ν i ] = [1.0 − 0.292] = 2.360 ×10−8 (1/psi)
Ei ( 30 ×10 psi )
6

For the loosest fit case:


ª 1 º δ loosest fit ª 1 º § 0.0005 ·
ploosest fit = « » =« −8 −8 » ¨ ¸
¬ Ko + Ki ¼ R ¬ 6.5289 ×10 + 2.360 ×10 ¼ © 0.5 ¹
= 11 250 psi (loosest fit)

For the tightest fit case:

ª 1 º δ tightest fit ª 1 º § 0.00115 ·


ptightest fit = « » =« −8 −8 » ¨ ¸
¬ K o + Ki ¼ R ¬ 6.5289 × 10 + 2.360 × 10 ¼ © 0.5 ¹
= 25 875 psi (tightest fit)

4. Radial & Tangential Stress Calculations for Shaft and Hub:

Loosest Fit (p = 11 250 psi)

Hub: σ r (r = R) = − p = −11 250 psi

ro2 + R 2
σ t (r = R) = p = pCo = (11 250 psi)(1.6667) = 18 750 psi
ro2 − R 2
Shaft: σ r (r = R) = − p = −11 250 psi

R 2 + ri 2
σ t (r = R) = − p 2 2 = − pCi = (−11 250 psi)(1) = −11 250 psi
R − ri

Tightest Fit (p = 25 875 psi)

Shigley, Mischke & Budynas Machine Design Tutorial 4-17: Press and Shrink Fits 6/11
Hub: σ r (r = R) = − p = −25 875 psi

ro2 + R 2
σ t (r = R) = p = pCo = (25 875 psi)(1.6667) = 43 125 psi
ro2 − R 2
Shaft: σ r (r = R) = − p = −25 875 psi

R 2 + ri 2
σ t (r = R) = − p = − pCi = (−25 875 psi)(1) = −25 875 psi
R 2 − ri 2
Thus the shaft has equal radial and tangential stress for each tightness condition,
whereas, the hub’s tangential stress is consistently higher than its radial stress for
both the loosest and the tightest condition.

Note: Since the shaft length is greater than the hub length, which is typical in
practice, this design case violates one of the assumptions of the interfacial
pressure development. In this case, there would be an increase in the interfacial
pressure at each end of the hub. This condition of increased interfacial pressure
would typically be accounted for by applying a stress concentration factor, Kt, for
stresses calculated at points at the end of the hub such as,

σ t actual = K t , tangentialσ t
σ r actual = K t , radialσ r

Shigley, Mischke & Budynas Machine Design Tutorial 4-17: Press and Shrink Fits 7/11
Table A-1 LIMITS AND FITS FOR CYLINDRICAL PARTS†

The limits shown in the accompanying tabulations are in thousandths of an inch. The size ranges include all sizes
over the smallest size in the range, up to and including the largest size in the range. The letter symbols are defined
as follows:

RC Running and sliding fits are intended to provide a similar running performance, with suitable lubrication
allowance, throughout the range of sizes. The clearance for the first two classes, used chiefly as slide fits,
increases more slowly with diameter than the other classes, so that accurate location is maintained even at
the expense of free relative motion.
RC1 Close sliding fits are intended for the accurate location of parts which must assemble without perceptible
play.
RC2 Sliding fits are intended for accurate location but with greater maximum clearance than class RC1. Parts
made to this fit move and turn easily, but are not intended to run freely, and in the larger sizes may seize
with small temperature changes.
RC3 Precision running fits are about the closest fits which can be expected to run freely and are intended for
precision work at slow speeds and light journal pressures, but are not suitable where appreciable
temperature differences are likely to be encountered.
RC4 Close running fits are intended chiefly for running fits on accurate machinery with moderate surface speeds
and journal pressures, where accurate location and minimum play is desired.
RC5–RC6 Medium running fits are intended for higher running speeds, heavy journal pressures, or both.
RC7 Free running fits are intended for use where accuracy is not essential or where large temperature variations
are likely to be encountered, or under both of these conditions.
RC8–RC9 Loose running fits are intended for use where wide commercial tolerances may be necessary, together
with an allowance, on the external member.
L Locational fits are fits intended to determine only the location of the mating parts; they may provide rigid or
accurate location, as with interference fits, or provide some freedom of location, as with clearance fits.
Accordingly, they are divided into three groups: clearance fits, transition fits, and interference fits.
LC Locational clearance fits are intended for parts which are normally stationary but which can be freely
assembled or disassembled. They run from snug fits for parts requiring accuracy of location, through the
medium clearance fits for parts such as ball, race, and housing, to the looser fastener fits where freedom of
assembly is of prime importance.
LT Locational transition fits are a compromise between clearance and interference fits, for application where
accuracy of location is important, but either a small amount of clearance or interference is permissible.
LN Locational interference fits are used where accuracy of location is of prime importance and for parts
requiring rigidity and alignment with no special requirements for bore pressure. Such fits are not intended for
parts designed to transmit frictional loads form one part to another by virtue of the tightness of fit, since
these conditions are covered by force fits.
FN Force and shrink fits constitute a special type of interference fit, normally characterized by maintenance of
constant bore pressure throughout the range of sizes. The interference therefore varies almost directly with
diameter, and the difference between its minimum and maximum value is small so as to maintain the
resulting pressures within reasonable limits.
FN1 Light drive fits are those requiring light assembly pressures and producing more or less permanent
assemblies. They are suitable for thin sections or long fits or in cast-iron external members.
FN2 Medium drive fits are suitable for ordinary steel parts or for shrink fits on light sections. They are about the
tightest fits that can be used with high-grade cast-iron external members.
FN3 Heavy drive fits are suitable for heavier steel parts or for shrink fits in medium sections.
FN4–FN5 Force fits are suitable for parts which can be highly stressed or for shrink fits where the heavy pressing
forces required are impractical.

Extracted from American Standard Limits for Cylindrical Parts ANSI B4.1-1978, with the permission of the
th
publishers, The American Society of Mechanical Engineers, United Engineering Center, 345 East 47 Street, New
York 10017. Limit dimensions are tabulated in this standard for nominal sizes up to and including 200 in. An SI
version is also available.

Shigley, Mischke & Budynas Machine Design Tutorial 4-17: Press and Shrink Fits 8/11
Table A-1-1 RUNNING AND SLIDING FITS
Diameter Size Range (in.)
Class 0.00 - 0.12 0.12 - 0.24 0.24 - 0.40 0.40 - 0.71

RC1 Hole +0.20 -0.00 +0.20 -0.00 +0.25 -0.00 +0.30 -0.00
Shaft +0.10 -0.25 -0.15 -0.30 -0.20 -0.35 -0.25 -0.45
RC2 Hole +0.25 -0.00 +0.30 -0.00 +0.40 -0.00 +0.40 -0.00
Shaft -0.10 -0.30 -0.15 -0.35 -0.20 -0.45 -0.25 -0.55
RC3 Hole +0.40 -0.00 +0.50 -0.00 +0.60 -0.00 +0.70 -0.00
Shaft -0.30 -0.55 -0.40 -0.70 -0.50 -0.90 -0.60 -1.00
RC4 Hole +0.60 -0.00 +0.70 -0.00 +0.90 -0.00 +1.00 -0.00
Shaft -0.30 -0.70 -0.40 -0.90 -0.50 -1.10 -0.60 -1.30
RC5 Hole +0.60 -0.00 +0.70 -0.00 +0.90 -0.00 +1.00 -0.00
Shaft -0.60 -1.00 -0.80 -1.30 -1.00 -1.60 -1.20 -1.90
RC6 Hole +1.00 -0.00 +1.20 -0.00 +1.40 -0.00 +1.60 -0.00
Shaft -0.60 -1.20 -0.80 -1.50 -1.00 -1.90 -1.20 -2.20
RC7 Hole +1.00 -0.00 +1.20 -0.00 +1.40 -0.00 +1.60 -0.00
Shaft -1.00 -1.60 -1.20 -1.90 -1.60 -2.50 -2.00 -3.00
RC8 Hole +1.60 -0.00 +1.80 -0.00 +2.20 -0.00 +2.80 -0.00
Shaft -2.50 -3.50 -2.80 -4.00 -3.00 -4.40 -3.50 -5.10
RC9 Hole +2.50 -0.00 +3.00 -0.00 +3.50 -0.00 +4.00 -0.00
Shaft -4.00 -5.60 -4.50 -6.00 -5.00 -7.20 -6.00 -8.80

Diameter Size Range (in.)


Class 0.71 - 1.19 1.19 - 1.97 1.97 - 3 .15 3.15 - 4.73

RC1 Hole +0.40 -0.00 +0.40 -0.00 +0.50 -0.00 +0.60 -0.00
Shaft -0.30 -0.55 -0.40 -0.70 -0.40 -0.70 -0.50 -0.90
RC2 Hole +0.50 -0.00 +0.60 -0.00 +0.70 -0.00 +0.90 -0.00
Shaft -0.30 -0.70 -0.40 -0.80 -0.40 -0.90 -0.50 -1.10
RC3 Hole +0.80 -0.00 +1.00 -0.00 +1.20 -0.00 +1.40 -0.00
Shaft -0.80 -1.30 -1.00 -1.60 -1.20 -1.90 -1.40 -2.30
RC4 Hole +1.20 -0.00 +1.60 -0.00 +1.80 -0.00 +2.20 -0.00
Shaft -0.80 -1.60 -1.00 -2.00 -1.20 -2.40 -1.40 -2.80
RC5 Hole +1.20 -0.00 +1.60 -0.00 +1.80 -0.00 +2.20 -0.00
Shaft -1.60 -2.40 -2.00 -3.00 -2.50 -3.70 -3.00 -4.40
RC6 Hole +2.00 -0.00 +2.50 -0.00 +3.00 -0.00 +3.50 -0.00
Shaft -1.60 -2.80 -2.00 -3.60 -2.50 -4.30 -3.00 -5.20
RC7 Hole +2.00 -0.00 +2.50 -0.00 +3.00 -0.00 +3.50 -0.00
Shaft -2.50 -3.70 -3.00 -4.60 -4.00 -5.80 -5.00 -7.20
RC8 Hole +3.50 -0.00 +4.00 -0.00 +4.50 -0.00 +5.00 -0.00
Shaft -4.50 -6.50 -5.00 -7.50 -6.00 -9.00 -7.00 -10.50
RC9 Hole +5.00 -0.00 +6.00 -0.00 +7.00 -0.00 +9.00 -0.00
Shaft -7.00 -10.50 -8.00 -12.00 -9.00 -13.50 -10.00 -15.00

Shigley, Mischke & Budynas Machine Design Tutorial 4-17: Press and Shrink Fits 9/11
Table A-1-2 FORCE AND SHRINK FITS
Diameter Size Range (in.)
Class 0.00 - 0.12 0.12 - 0.24 0.24 - 0.40 0.40 - 0.56

FN1 Hole +0.25 -0.00 +0.30 -0.00 +0.40 -0.00 +0.40 -0.00
Shaft +0.50 +0.30 +0.60 +0.40 +0.75 +0.50 +0.80 +0.50
FN2 Hole +0.40 -0.00 +0.50 -0.00 +0.60 -0.00 +0.70 -0.00
Shaft +0.85 +0.60 +1.00 +0.70 +1.40 +1.00 +1.60 +1.20
FN3 Hole
Shaft
FN4 Hole +0.40 -0.00 +0.50 -0.00 +0.60 -0.00 +0.70 -0.00
Shaft +0.95 +0.70 +1.20 +0.90 +1.60 +1.20 +1.80 +1.40
FN5 Hole +0.60 -0.00 +0.70 -0.00 +0.90 -0.00 +1.00 -0.00
Shaft +1.30 +0.90 +1.70 +1.20 +2.00 +1.40 +2.30 +1.60
Diameter Size Range (in.)
Class 0.56 - 0.71 0.71 - 0.95 0.95 - 1.19 1.19 - 1.58

FN1 Hole +0.40 -0.00 +0.50 -0.00 +0.50 -0.00 +0.60 -0.00
Shaft +0.90 +0.60 +1.10 +0.70 +1.20 +0.80 +1.30 +0.90
FN2 Hole +0.70 -0.00 +0.80 -0.00 +0.80 -0.00 +1.00 -0.00
Shaft +1.60 +1.20 +1.90 +1.40 +1.90 +1.40 +2.40 +1.80
FN3 Hole +0.80 -0.00 +1.00 -0.00
Shaft +2.10 +1.60 +2.60 +2.00
FN4 Hole +0.70 -0.00 +0.80 -0.00 +0.80 -0.00 +1.00 -0.00
Shaft +1.80 +1.40 +2.10 +1.60 +2.30 +1.80 +3.10 +2.50
FN5 Hole +1.00 -0.00 +1.20 -0.00 +1.20 -0.00 +1.60 -0.00
Shaft +2.50 +1.80 +3.00 +2.20 +3.30 +2.50 +4.00 +3.00

Diameter Size Range (in.)


Class 1.58 - 1.97 1.97 - 2.56 2.56 - 3 .15 3.15 - 3.94

FN1 Hole +0.60 -0.00 +0.70 -0.00 +0.70 -0.00 +0.90 -0.00
Shaft +1.40 +1.00 +1.80 +1.30 +1.90 +1.40 +2.40 +1.80
FN2 Hole +1.00 -0.00 +1.20 -0.00 +1.20 -0.00 +1.40 -0.00
Shaft +2.40 +1.80 +2.70 +2.00 +2.90 +2.20 +3.70 +2.80
FN3 Hole +1.00 -0.00 +1.20 -0.00 +1.20 -0.00 +1.40 -0.00
Shaft +2.80 +2.20 +3.20 +2.50 +3.70 +3.00 +4.40 +3.50
FN4 Hole +1.00 -0.00 +1.20 -0.00 +1.20 -0.00 +1.40 -0.00
Shaft +3.40 +2.80 +4.20 +3.50 +4.70 +4.00 +5.90 +5.00
FN5 Hole +1.60 -0.00 +1.80 -0.00 +1.80 -0.00 +2.20 -0.00
Shaft +5.00 +4.00 +6.20 +5.00 +7.20 +6.00 +8.40 +7.00

Shigley, Mischke & Budynas Machine Design Tutorial 4-17: Press and Shrink Fits 10/11
Table A-1-2 FORCE AND SHRINK FITS (CONTINUED)

Diameter Size Range (in.)


Class 3.94 - 4.73 4.73 - 5.52 5.52 - 6.30 6.30 - 7.09

FN1 Hole +0.90 -0.00 +1.00 -0.00 +1.00 -0.00 +1.00 -0.00
Shaft +2.60 +2.00 +2.90 +2.20 +3.20 +2.50 +3.50 +2.80
FN2 Hole +1.40 -0.00 +1.60 -0.00 +1.60 -0.00 +1.60 -0.00
Shaft +3.90 +3.00 +4.50 +3.50 +5.00 +4.00 +5.50 +4.50
FN3 Hole +1.40 -0.00 +1.60 -0.00 +1.60 -0.00 +1.60 -0.00
Shaft +4.90 +4.00 +6.00 +5.00 +6.00 +5.00 +7.00 +6.00
FN4 Hole +1.40 -0.00 +1.60 -0.00 +1.60 -0.00 +1.60 -0.00
Shaft +6.90 +6.00 +8.00 +7.00 +8.00 +7.00 +9.00 +8.00
FN5 Hole +2.20 -0.00 +2.50 -0.00 +2.50 -0.00 +2.50 -0.00
Shaft +9.40 +8.00 +11.60 +10.00 +13.60 +12.00 +13.60 +12.00

Shigley, Mischke & Budynas Machine Design Tutorial 4-17: Press and Shrink Fits 11/11
MECHANICAL ENGINEERING DESIGN
TUTORIAL 4-20: HERTZ CONTACT STRESSES

CHARACTERISTICS OF CONTACT STRESSES

1. Represent compressive stresses developed from surface pressures between two curved
bodies pressed together;
2. Possess an area of contact. The initial point contact (spheres) or line contact
(cylinders) become area contacts, as a result of the force pressing the bodies against
each other;
3. Constitute the principal stresses of a triaxial (three dimensional) state of stress;
4. Cause the development of a critical section below the surface of the body;
5. Failure typically results in flaking or pitting on the bodies’ surfaces.

TWO DESIGN CASES

Two design cases will be considered,

1. Sphere – Sphere Contact (Point Contact Ÿ Circular Contact Area)


2. Cylinder – Cylinder Contact (Line Contact Ÿ Rectangular Contact Area)

SPHERE – SPHERE CONTACT

F F
x x
d1

y y
d2
2a

F F

z z

(a) Two spheres held in (b) Contact stress has an elliptical


contact by force F. distribution across contact over
zone of diameter 2a.

TEXT FIGURE 4-42 Two Spheres in Contact


Text. refers to Mechanical Engineering Design, 7th edition text by Joseph Edward Shigley, Charles R.
Mischke, and Richard G. Budynas; equations and figures with the prefix T refer to the present tutorial.
Consider two solid elastic spheres held in contact by a force F such that their point of contact
expands into a circular area of radius a, given as:

a = Ka 3 F (Modified Text Eq. 4-72)


1/ 3
ª 3 (1 − ν12 ) / E1 + (1 − ν 22 ) / E2 º
where K a = « »
¬8 (1/ d1 ) + (1/ d 2 ) ¼
F = applied force
ν1 ,ν 2 = Poisson's ratios for spheres 1 and 2
E1 , E2 = elastic modulii for spheres 1 and 2
d1 , d 2 = diameters of spheres 1 and 2

This general expression for the contact radius can be applied to two additional common cases:

1. Sphere in contact with a plane (d 2 = ∞);


2. Sphere in contact with an internal spherical surface or ‘cup’ (d 2 = − d ).

Returning to the sphere-sphere case, the maximum contact pressure, pmax , occurs at the center
point of the contact area.

3F
pmax = (Text Eq. 4-73)
2π a 2

State of Stress

• The state of stress is computed based on the following mechanics:

1. Two planes of symmetry in loading and geometry dictates that σ x = σ y ;


2. The dominant stress occurs on the axis of loading: σ max = σ z ;
3. The principal stresses are σ 1 = σ 2 = σ x = σ y and σ 3 = σ z given σ 1 , σ 2 ≥ σ 3 ;
4. Compressive loading leads to σ x , σ y , and σ z being compressive stresses.

• Calculation of Principal Stresses

ªª § 1 ·º º
1
σ x = − pmax «1 − ζ a tan −1 ¨¨
« ¸¸ » (1 + ν ) − » (Modified Text Eq. 4-74)
« ¬«
¬ © ζa ¹ ¼» (
2 1 + ζ a2 ) »
¼
= σ y = σ1 = σ 2

Shigley, Mischke & Budynas Machine Design Tutorial 4-20: Hertz Contact Stresses 2/10
− pmax
σ3 = σ z = (Modified Text Eq. 4-75)
1 + ζ a2

where ζ a = z / a = nondimensional depth below the surface


ν = Poisson's ratio for the sphere examined (1 or 2)

• Mohr’s Circle

Plotting the principal stresses on a Mohr’s circle plot results in: one circle, defined by
σ 1 = σ 2 , shrinking to a point; and two circles, defined by σ 1 , σ 3 and σ 2 , σ 3 , plotted
on top of each other. The maximum shear stress, τ max , for the plot is calculated as:

σ1 − σ 3 σ x − σ z σ y − σ z
τ max = = = (Modified Text Eq. 4-76)
2 2 2

If the maximum shear stress, τ max , and principal stresses, σ 1 , σ 2 , and σ 3 , are plotted
as a function of maximum pressure, p max , below the surface contact point, the plot of
Fig. 4-43 is generated. This plot, based on a Poisson’s ratio of ν = 0.3 , reveals that a
critical section exists on the load axis, approximately 0.48a below the sphere surface.
Many authorities theorize that this maximum shear stress is responsible for the
surface fatigue failure of such contacting elements; a crack, originating at the point of
maximum shear, progresses to the surface where lubricant pressure wedges a chip
loose and thus creates surface pitting.

TEXT FIGURE 4-43: Magnitude of the stress components below the


surface as a function of maximum pressure of contacting spheres.

Shigley, Mischke & Budynas Machine Design Tutorial 4-20: Hertz Contact Stresses 3/10
CYLINDER–CYLINDER CONTACT

Consider two solid elastic cylinders held in contact by forces F uniformly distributed along the
cylinder length l.

F x
F x
d1

l
y y
d2
2b

F F

z z

(a) Two right circular cylinders held (b) Contact stress has an elliptical
in contact by forces F uniformly distribution across contact zone
distributed along cylinder length l. of width 2b.

TEXT FIGURE 4-44 Two Cylinders in Contact

The resulting pressure causes the line of contact to become a rectangular contact zone of half-
width b given as:

b = Kb F (Modified Text Eq. 4-77)


1/ 2
ª 2 (1 − ν12 ) / E1 + (1 − ν 22 ) / E2 º
where Kb = « »
¬π l (1/ d1 ) + (1/ d 2 ) ¼
F = applied force
ν1 ,ν 2 = Poisson's ratios for cylinders 1 and 2
E1 , E2 = elastic modulii for cylinders 1 and 2
d1 , d 2 = diameters of spheres 1 and 2
l = length of cylinders 1 and 2 (l1 = l2 assumed)

Shigley, Mischke & Budynas Machine Design Tutorial 4-20: Hertz Contact Stresses 4/10
This expression for the contact half-width, b, is general and can be used for two additional cases
which are frequently encountered:

1. Cylinder in contact with a plane, e.g. a rail (d 2 = ∞);


2. Cylinder in contact with an internal cylindrical surface, for example the race of a
roller bearing (d 2 = − d ).

The maximum contact pressure between the cylinders acts along a longitudinal line at the center
of the rectangular contact area, and is computed as:

2F
pmax = (Text Eq. 4-78)
π bl

State of Stress

• The state of stress is computed based on the following mechanics:

1. One plane of symmetry in loading and geometry dictates that σ x ≠ σ y ;


2. The dominant stress occurs along the axis of loading: σ max = σ z ;
3. The principal stresses are equal to σ x , σ y , and σ z with σ 3 = σ z ;
4. Compressive loading leads to σ x , σ y , and σ z being compressive stresses.

• Calculation of Principal Stresses and Maximum Shear Stress

1
σ 3 = σ z = − pmax (Modified Text Eq. 4-81)
1 + ζ b2

­°σ x for 0 ≤ ζ b ≤ 0.436


σ1 = ®
°̄σ y for 0.436 ≤ ζ b

where,
σ x = −2ν pmax ª« 1 + ζ b2 − ζ b º» (Modified Text Eq. 4-79)
¬ ¼
ª§ 1 + 2ζ 2 · º
σ y = − pmax «¨ b ¸
− 2 ζb » (Modified Text Eq. 4-80)
«¨ 1 + ζ 2 ¸ »
¬© b ¹ ¼
ζb = z /b

Shigley, Mischke & Budynas Machine Design Tutorial 4-20: Hertz Contact Stresses 5/10
The maximum shear stress is thus given as:

­°τ1/ 3 = (σ z − σ x ) / 2 for 0 ≤ ζ b ≤ 0.436


τ max = ®
°̄τ1/ 3 = (σ z − σ y ) / 2 for 0.436 ≤ ζ b

When these equations are plotted as a function of maximum contact pressure up to a


distance 3b below the surface contact point, the plot of Fig. 4-45 is generated. Based
on a Poisson’s ratio of 0.3, this plot reveals that τ max attains a maxima for
ζ b = z / b = 0.786 and 0.3pmax.

TEXT FIGURE 4-45: Magnitude of stress components below


the surface as a function of maximum pressure for contacting
cylinders.

Example T4.20.1:

Problem Statement: A 6-in-diameter cast-iron wheel, 2 in wide, rolls on a flat


steel surface carrying a 800 lbf load.

Find:
1. The Hertzian stresses σ x , σ y , σ z and τ 1/3 in the cast iron wheel at
the critical section;
2. The comparative state of stress and maximum shear stress, arising
during a revolution, at point A located 0.015 inch below the wheel
rim surface.

Shigley, Mischke & Budynas Machine Design Tutorial 4-20: Hertz Contact Stresses 6/10
Solution Methodology:
1. Compute the value of the contact half-width, b.
2. Compute the maximum pressure generated by the normal force of
the wheel.
3. Use the results of steps (1) and (2) to calculate the contact stresses
in the cast iron wheel for the critical section, z/b = 0.786.
4. Evaluate the principal stresses based upon the contact stress
calculations.
5. Calculate the maximum shear stress.
6. Compare these results with those obtained by using Fig. 4-45.
7. During a single revolution of the wheel, point A will experience a
cycle of stress values varying from zero (when point A lies well
outside the contact zone) to a maximum state of stress (when A lies
within the contact zone and on the line of action of the 800 lbf
force.) We expect point A to “feel” the effects of a semi-elliptical
contact pressure distribution as point A moves into and through the
contact zone. Thus, we need to calculate the contact stresses for a
depth of z = 0.015 inch, which we expect to lie within the contact
zone.

Schematic:

Ecast iron = 14.5 × 106 psi


800 lbf 6 in
ν cast iron = 0.211

2 in

Esteel = 30 × 106 psi


ν steel = 0.292

Solution:

1. Compute contact half-width, b


Material Properties: E1 = Ecast iron = 14.5 × 106 psi; ν 1 = ν cast iron = 0.211
E2 = Esteel = 30.0 × 106 psi; ν 2 = ν steel = 0.292

Shigley, Mischke & Budynas Machine Design Tutorial 4-20: Hertz Contact Stresses 7/10
Dimensions: d1 = 6.0 in; d 2 = ∞; l = 2.0 in

b = Kb F (Modified Text Eq. 4-72)


1/ 2
ª 2 (1 − ν12 ) / E1 + (1 − ν 22 ) / E2 º
Kb = « »
¬π l (1/ d1 ) + (1/ d 2 ) ¼
1/ 2
­° 2 ª¬1 − (0.211) 2 º¼ /(14.5 ×106 ) + ª¬1 − (0.292) 2 º¼ /(30.0 × 106 ) ½°
=® ¾
°¯ π (2.0) (1/ 6.0) + (1/ ∞) °¿
= 4.291× 10−4 in/ lbf
b = Kb F = (4.291×10 −4 in/ lbf )(800 lbf )1/2 = 1.214 × 10-2 in

2. Maximum Pressure, pmax

2F 2(800 lbf)
pmax = = = 20 980 psi
π bl π (1.214 ×10−2 in)(2.0 in)

3. Hertz Contact Stresses in Cast Iron Wheel

At the critical section, ζ b = z / b = 0.786 ,

σ x = −2ν1 pmax ª« 1 + ζ b2 − ζ b º»
¬ ¼
= −2(0.211)(20 980 psi) ª 1+(0.786)2 − 0.786 º
«¬ »¼
= −4302 psi
ª§ 1 + 2ζ 2 · º
σ y = − pmax «¨ b ¸ − 2 ζb »
«¨ ¸ »
¬© 1 + ζ b
2
¹ ¼
­ ª 1 + 2(0.786) 2 º ½
° °
= (−20 980 psi) ® « » − 2(0.786) ¾
°¯ «¬ 1 + (0.786) »¼
2
°¿
= −3895 psi
1 −20 980 psi
σ z = − pmax =
1 + ζ b2 1+(0.786)2
= −16 490 psi

Shigley, Mischke & Budynas Machine Design Tutorial 4-20: Hertz Contact Stresses 8/10
Note that the small contact area involved in this type of problem gives rise to
very high pressure, relative to the applied force, and thus exceptionally high
stresses.

4. Since σ x , σ y , and σ z are all principal stresses, we can conclude:

σ 1 = σ y = − 3895 psi
σ 2 = σ x = − 4302 psi
σ 3 = σ z = −16 490 psi

5. Maximum Shear Stress

σ 1 − σ 3 σ y − σ z −3895 psi − (−16 490 psi)


τ max = τ1/ 3 = = =
2 2 2
= 6298 psi

6. Comparison with results based on Text Figure 4-45:

For z/b ≈ 0.75,

σ x ≈ −0.3 pmax = −6294 psi


σ y ≈ −0.2 pmax = −4196 psi
σ z ≈ −0.8 pmax = −16 780 psi
τ max ≈ 0.3 pmax = 6294 psi

Comparing these results with those calculated using a value of ν = 0.211, we


find that only σ x is a function of ν; σ y , σ z , and τ max are independent of ν

since the graphical estimates of their values are within 3 % of those obtained
from the plot which assumes a Poisson’s ratio of 0.3.

7. For a depth of 0.015 in below the cylinder surface,

0.015 in
ζb = = 1.236
1.214 × 10 −2 in

Substituting,

Shigley, Mischke & Budynas Machine Design Tutorial 4-20: Hertz Contact Stresses 9/10
σ x = −2ν pmax ª« 1 + ζ b2 − ζ b º»
¬ ¼
= −2(0.211)(20 980 psi) ª 1+(1.236)2 − 1.236 º
«¬ »¼
= −3133 psi

ª§ 1 + 2ζ 2 · º
σ y = − pmax «¨ b ¸ − 2 ζb »
«¨ 1 + ζ 2 ¸ »
¬© b ¹ ¼
­ ª 1 + 2(1.236) 2 º ½
° °
= (−20 980 psi) ® « » − 2 1.236 ¾
« 2 »
¯° ¬ 1 + (1.236) ¼ ¿°
= −1652 psi
1 −20 980 psi
σ z = − pmax =
1 + ζ b2 1+(1.236) 2
= −13 200 psi

σ 1 − σ 3 σ y − σ z −1652 − (−13 200)


τ max = = = = 5774 psi
2 2 2

As expected, at a depth corresponding greater than the critical section


( z/b = 1.236 > 0.786), the magnitudes of all three principal stresses are
smaller than those calculated for z/b = 0.786. The difference between the
principal stresses is also smaller and consequently, τmax also decreases.

Shigley, Mischke & Budynas Machine Design Tutorial 4-20: Hertz Contact Stresses 10/10
MECHANICAL ENGINEERING DESIGN
TUTORIAL 6, PART A: INTRODUCTION TO STATIC FAILURE THEORIES

APPROACH

Theories have been developed for the static failure of metals based upon the two classes of material
failure; ductile metals yield while brittle metals fracture. Thus separate failure theories exist for
ductile and brittle metals:

Failure Theories for Ductile Materials

1. Maximum Shear Stress (MSS)


2. Distortion Energy (DE)
3. Ductile Coulomb-Mohr (DCM)

Failure Theories for Brittle Materials

1. Maximum Normal Stress (MNS)


2. Brittle Coulomb-Mohr (BCM)

These theories have grown out of hypotheses and experimental data in the following manner.

1. Experimental failure data is first collected through tensile tests.


2. The state of stress is correlated to the experimental data using Mohr’s circle plots.
3. A failure theory is developed from a concept of the responsible failure mechanism.
4. A design envelope is established based upon the theoretical and empirical design
equations.

In light of the extensive dependence of failure theories on experimental data, we will first review
the acquisition and correlation of tensile test data to failure theory. Subsequently, the criteria and
application of specific failure theories will be discussed.

TENSILE TEST

d0

l0

TEXT FIGURE 3-1: A typical tension-test specimen. Some of the standard


dimensions used for d0 are 2.5, 6.25, and 12.5 mm and 0.505 in, but other sections and
sizes are in use. Common gauge lengths l0 used are 10, 25, and 50 mm and 1 and 2 in.


Text refers to Mechanical Engineering Design, 7th edition text by Joseph Edward Shigley, Charles R.
Mischke and Richard G. Budynas; equations and examples with the prefix T refer to the present tutorial.
The tensile test is a standardized test (ASTM Standard E8 or E8m) and thus allows for the sharing
of experimental data amongst researchers, typically in the form of stress-strain curves. Standard
dimensions for the test-specimen are provided in Text Figure 3-1 while a comparison of
characteristic stress-strain curves for ductile and brittle materials are shown in Text Figure 3-2.
These engineering stress-strain diagrams graphically demonstrate the difference in the failure
behavior of ductile and brittle metals, and the need for separate failure criteria. However, the curves
do not represent true values of stress and strain; rather, they are calculated based upon the original
specimen cross-sectional area, prior to loading.

P 4P
Engineering Stress σ= = (Text Eq. 3-1)
A0 π d 02

l − l0
Engineering Strain ε= (Text Eq. 3-2)
l0

u
Su f Sut u, f
Sf Sy y
Sy y
el
pl
Stress σ = P/A0

O a εy εu εf a
Strain ε Strain ε

(a) (b)

TEXT FIGURE 3-2: Stress-strain diagram obtained from the standard tensile test
(a) Ductile material; (b) brittle material. pl marks the proportional limit; el, the
elastic limit; y, the offset yield strength as defined by offset strain Oa; u, the
maximum or ultimate strength; and f, the fracture strength.

Referring to Text Fig. 3-2 (a), point el, the elastic limit, defines the onset of permanent set
while point a represents 0.2 percent permanent set with respect to the original gauge length
(ε = 0.002).

A measure of the “true” stress and strain can be obtained by taking simultaneous
measurements of the load and cross-sectional area during the tensile test experiment. Text

Shigley, Mischke & Budynas Machine Design Tutorial 6: Static Failure Theories 2/5
Figure 3-4 shows a typical true stress-strain diagram for a ductile material. The curve
between points u and f corresponds to a reduction in stress as the specimen necks down.

σf f
σu
u
True stress

εu εf
True strain

TEXT FIGURE 3-4: True stress- TEXT FIGURE 3-3: Tension specimen
strain diagram plotted in Cartesian after necking.
coordinates.

CORRELATION OF STATE OF STRESS WITH TEST DATA

For design, we need to relate the expected state of stress in a part to the actual state of stress
and thus, the material strength, as determined through the tensile test. We accomplish this
by applying principal stresses since they characterize a state of stress independent of the
original coordinate system.
y

σy =0
τ yx = τ xy = 0

σ x = P / A0 σ x = P / A0
x
Ÿ σ1 σ1
τ yx = τ xy = 0
σ2 =σ3 = 0
σy =0

(a) State of stress for simple tension. (b) Principal stresses for simple tension.

FIGURE 6A-1: Correlation of state of stress


with principal stresses for simple tension.

Shigley, Mischke & Budynas Machine Design Tutorial 6: Static Failure Theories 3/5
Since the tensile test generates a uniaxial state of stress, the principal stresses can be defined
as,

P
σ 1 = σ axial = and σ 2 = σ 3 = 0
A0

When plotted on a Mohr’s circle diagram, these stress values translate into what looks like a
single circle passing through the origin where σ 2 is coincident with σ 3 . Actually, there are
still three circles on the Mohr’s circle diagram. Two circles, defined by principal stresses
(σ 1,σ 2 ) and (σ 1,σ 3 ) , are drawn on top of each other. The third circle degenerates to a point
defined by principal stresses (σ 2,σ 3 ) .

σ3 σ2 σ1 σ

FIGURE 6A-2: Mohr’s circle for simple tension.

DEVELOPMENT OF STATIC FAILURE THEORIES

Design for static loading dictates that all loading variables remain constant:

1. Magnitude of load is constant;


2. Direction of load is constant;
3. Point of application of the load is fixed.

These conditions, in conjunction with criteria specific to ductile and brittle materials, have
been used in the development of the static failure theories outlined earlier.

Characteristics of Ductile Materials

1. The strain at failure is, ε f ≥ 0.05 , or percent elongation greater than five percent.
2. Ductile materials typically have a well defined yield point. The value of the
stress at the yield point defines the yield strength, Sy.
3. For typical ductile materials, the yield strength has approximately the same value
for tensile and compressive loading ( S yt ≈ S yc = S y ).

Shigley, Mischke & Budynas Machine Design Tutorial 6: Static Failure Theories 4/5
4. A single tensile test is sufficient to characterize the material behavior of a ductile
material, Sy and Sut.

Characteristics of Brittle Materials

1. The strain at failure is, ε f ≤ 0.05 or percent elongation less than five percent.
2. Brittle materials do not exhibit an identifiable yield point; rather, they fail by
brittle fracture. The value of the largest stress in tension and compression
defines the ultimate strength, Sut and Suc respectively.
3. The compressive strength of a typical brittle material is significantly higher than
its tensile strength, ( Suc  Sut ).
4. Two material tests, a tensile test and a compressive test, are required to
characterize the material behavior of a brittle material, Sut and Suc.

SUMMARY

This tutorial has attempted to provide a focused introduction to the development of static failure
theory by summarizing the theories associated with specific material classifications. In addition, the
experimental and analytical models, which have been employed historically to relate the
experimental data to strength quantities used for static design, are presented. Subsequent tutorials,
Static Failure of Ductile Materials and Static Failure of Brittle Materials, will respectively provide
detailed reviews and examples, respectively, of the failure theories associated with ductile and
brittle materials.

Shigley, Mischke & Budynas Machine Design Tutorial 6: Static Failure Theories 5/5
3-D STRESS IN MECHANICAL
DESIGN

August 2000

Copyright  2000 C. E. Knight


PURPOSE OF THE TUTORIAL

This tutorial is designed to introduce and place strong emphasis on the role of 3-D stress in the process of
mechanical design. Students in engineering are normally introduced to stress in its simplest one-
component form defined by load divided by area of cross section.. This is a valid definition of a pure 1-D
state of stress, but in many cases it seems to establish a baseline safe position for which many students
don’t want to venture forth. Carrying this attitude through the mechanical design process is a recipe for
failure.

Everything in the mechanical design realm has solid 3-D characteristics. The same is true for the state of
stress in the solid. In many simple cases the effective state of stress can be reduced to 2-D or 1-D, but only
after careful consideration. In the early stages of mechanical design, the locations of most likely stress
failure and the corresponding stress components acting at those locations must be identified. Once all the
stress components at a given location are determined, they may then be combined to find principal stresses,
maximum shear stress or other measures that are useful for predicting design success or failure. It is very
important to remember that stress components for one location in a machine part should never be
combined with stress components for a different location in the same part.

One of the interesting developments in visualizing the combining of 2-D stress components was the
creation of Mohr’s circle. This graphical representation of the 2-D stress transformation equations provides
a quick, accurate and visual protrayal of the 2-D state of stress. It finds the principal stresses and a
maximum shear stress (although this maximum shear stress may be quite misleading in 3-D stress).

Review of Two-Dimensional Mohr’s Circle - Graphical Approach

The beginning of a Mohr’s circle representation must be a stress element sketch of the 2-D state of stress as
shown in the figure.

This shows all potential non-zero 2-D stress components. The graphical Mohr’s circle uses coordinate
pairs of these data to make a plot. They are (σx, τxy ) and (σy , τyx). These two points establish the circle
diameter. By convention normal stresses, σ are positive in tension and negative in compression, however,
the shear stresses, τ, in the Mohr’s circle constructions are taken as positive if they make a cw moment
about the stress element. In the stress element above, τxy is ccw (-) while τyx is cw(+). This convention is
useful for determining the proper orientation of principal stresses and other components relative to the x,y
coordinates.

As an example, assume that σx is positive and τxy is positive (cw) with σy equal zero. First sketch the
normal stress axis along the horizontal and the shear stress axis along the vertical. Then plot the first
coordinate pair (σx, τxy ) at point A. Then plot the second pair (0, τyx ) at point B. These two points form
the diameter of the circle with its center at point C. Simple geometric triangles can then determine the
circle radius and all principal stress and peak shear stress values.
In a second example, assume that σx is smaller than σy , but both are positive, and that τxy is cw. Sketch the
normal stress, σ, and shear stress, τ axes and plot the coordinate pair (σx, τxy ) at point A and then (σy , τyx) at
point B. Connecting these points locates the circle center at point C. Geometrical calculations finish the
numerical values.

The Mohr’s circle gives a complete visual representation of the 2-D state of stress along with accurate
numerical values. However, there is a highly significant factor in mechanical design that has thus far been
neglected. That factor is the influence of the additional 3-D stress components on the design safety.

Three-Dimensional Stress in Mechanical Design

In the real world of applications all objects are 3-D. The general state of stress is pictured on the stress
element below. There are six independent stress components shown in a conventient Cartesian coordinate
system. It is readily seen that in the 2-D Mohr’s circle, the principal stresses are larger numerically than the
cartesion components unless they are already principal stresses. The same is true in 3-D stress. A qubic
equation can be solved for the three principal stress roots in the general stress case, however, in many
cases of mechanical design some of the principal stresses may be determined by inspection.
3-D Mohr’s Circles

Use of Mohr’s circles can again make visualization of the stress condition clearer to the designer. The
definition of the three circle diagram is sketched below. Note that the principal stress values are always
ordered by convention so the σ1 is the largest value in the tensile direction and σ3 is the largest value in the
compressive direction. Note also that there is one dominant peak shear stress in this diagram. Be
forewarned the principal stresses and this peak shear stress are going to play a strong role in
determining the factor of safety in mechanical design.

What about the two 2-D examples? How do they become 3-D representations? If the stress state is only
two-dimensional, then σz and all the shear stresses with z components are zero, therefore, σz = 0 is the third
principal stress. Only two principal stresses were found by the Mohr’s circle transformation. Since σz = 0,
it must not be important. Wrong!! Look at the modified examples below. In example 1, the second
principal stress, σ2 , becomes zero and the third principal stress, σ3 , is negative, but the overall range is the
same. In this case, there is no effect on the overall stress state.

However, the second example has the same first two principal stresses σ1 and σ2 , but σ3 is 0. This enlarges
the outermost circle which means that the overall state of stress has increased. In this case, there is an
appreciable contribution by the 3-D effect that must not be ignored in accounting for design safety.
Example 3

As a final example, use the stress conditions σx = 90 (T), τxy = 40 ccw, σy = 30 (T), and σz = - 25 (C).
First sketch the normal stress and shear stress axes and then plot the coordinate pair ( σx , τxy ) at A. Plot the
next coordinate pair ( σy , τyx ) at B. Connect points A and B to form the diameter of the 2-D Mohr’s circle
with center at C. Draw the circle and determine two of the principal stresses. The center C is located at a
stress value of 60. The triangle C, σx, and τxy form a 30, 40, 50 triangle, so the circle radius is 50. The two
principal stresses from the 2-D circle are 110 (T) and 10(T). Since there are no non-zero z component
shear stresses, σz is the third principal stress with a value of –25 (C). The maximum shear stress is at the
peak of the largest circle and is equal to half the difference between σ1 and σ3 .