Lecture Notes On FEM
Uploaded by
Abdeldjalil DebbousLecture Notes On FEM
Uploaded by
Abdeldjalil DebbousLecture Notes: An Introduction to the Finite Element Method
Lecture Notes:
An Introduction to the
Finite Element Method
Yijun Liu
CAE Research Laboratory
Mechanical Engineering
University of Cincinnati
Cincinnati, OH 45221-0072, U.S.A.
E-mail: [email protected]
Web: www.yijunliu.com
(Last Updated: October 21, 2015)
© 1997-2013 by Yijun Liu, University of Cincinnati.
© 1997-2013 Yijun Liu, University of Cincinnati i
Lecture Notes: An Introduction to the Finite Element Method
Copyright Notice
© 1997-2013 by Yijun Liu, University of Cincinnati.
All rights reserved. Permissions are granted for personal and educational uses only at the
University of Cincinnati. Any other uses of these lecture notes (such as for classroom
lectures outside the University of Cincinnati, trainings elsewhere, and those of a
commercial nature) are prohibited, unless such uses have been granted in writing by the
author.
© 1997-2013 Yijun Liu, University of Cincinnati ii
Lecture Notes: An Introduction to the Finite Element Method
Table of Contents
Copyright Notice ............................................................................................................................. ii
Table of Contents ........................................................................................................................... iii
Preface .............................................................................................................................................v
Chapter 1. Introduction ................................................................................................................. 1
I. Some Basic Concepts .............................................................................................................. 1
II. Review of Matrix Algebra..................................................................................................... 6
III. Spring Element................................................................................................................... 11
IV. Summary ............................................................................................................................. 20
V. Problems............................................................................................................................... 20
Chapter 2. Bar and Beam Elements............................................................................................ 21
I. Linear Static Analysis............................................................................................................ 21
II. Bar Element .......................................................................................................................... 21
III. Beam Element ..................................................................................................................... 39
IV. Summary ............................................................................................................................. 54
V. Problems............................................................................................................................... 54
Chapter 3. Two-Dimensional Elasticity Problems ..................................................................... 57
I. Stress State in Structures ....................................................................................................... 57
II. 2-D (Plane) Elasticity Problems ........................................................................................... 57
III. Finite Elements for 2-D Problems ..................................................................................... 62
IV. Summary ............................................................................................................................. 77
V. Problems............................................................................................................................... 78
Chapter 4. Modeling and Solution Techniques ........................................................................... 80
I. Symmetry............................................................................................................................... 80
II. Substructures (Superelements) ............................................................................................. 81
III. Equation Solving................................................................................................................. 83
IV. Nature of Finite Element Solutions .................................................................................... 85
V. Convergence of FEA Solutions............................................................................................ 86
VI. Adaptivity (h-, p-, and hp-Methods) ................................................................................... 86
VII. Summary ........................................................................................................................... 87
VIII. Problems........................................................................................................................... 87
© 1997-2013 Yijun Liu, University of Cincinnati iii
Lecture Notes: An Introduction to the Finite Element Method
Chapter 5. Plate and Shell Elements ........................................................................................... 89
I. Plate Theory........................................................................................................................... 89
II. Plate Elements ..................................................................................................................... 97
III. Shells and Shell Elements ................................................................................................ 101
IV. Summary ........................................................................................................................... 106
V. Problems............................................................................................................................. 106
Chapter 6. Three-Dimensional Elasticity Problems .................................................................. 108
I. 3-D Elasticity Theory .......................................................................................................... 108
II. Finite Element Formulation ............................................................................................... 111
III. Typical 3-D Solid Elements .............................................................................................. 113
IV. Solids of Revolution (Axisymmetric Analysis)................................................................ 120
V. Summary ............................................................................................................................ 125
VI. Problems .......................................................................................................................... 125
Chapter 7. Structural Vibration and Dynamics ......................................................................... 127
I. Basic Equations ................................................................................................................... 127
II. Free Vibration of Multiple DOF Systems .......................................................................... 132
III. Damping ............................................................................................................................ 135
IV. Modal Equations ............................................................................................................... 136
V. Frequency Response Analysis ........................................................................................... 137
VI. Transient Response Analysis ............................................................................................ 138
VII. Summary ......................................................................................................................... 144
VIII. Problems........................................................................................................................ 144
Chapter 8. Thermal Analysis ..................................................................................................... 146
I. Temperature Field................................................................................................................ 146
II. Thermal Stress Analysis ..................................................................................................... 148
III. Summary ........................................................................................................................... 153
IV. Problems .......................................................................................................................... 153
References for Further Studies ................................................................................................... 155
© 1997-2013 Yijun Liu, University of Cincinnati iv
Lecture Notes: An Introduction to the Finite Element Method
Preface
This is a basic introduction to the finite element method (FEM) for undergraduate
students in engineering or other readers who have no previous experience with the FEM. The
lecture notes cover the basic concepts in the FEM using the simplest mechanics problems as
examples, and lead to the discussions and applications of the one-dimensional (1-D) bar and
beam, 2-D plane stress and plane strain, plate and shell, and 3-D solid elements in the analyses of
structural stresses, vibrations and dynamics. The understanding of the FEM principles and
procedures, and correct usage of the FEM software are emphasized throughout the notes.
These lecture notes have been developed by the author for the undergraduate courses on
the FEM in the Mechanical Engineering Department at the University of Cincinnati since 1997.
The materials in these notes are aimed for students in mechanical, civil and aerospace
engineering, who need a general background in the FEM so that they can apply the FEM in their
design and analysis of components, structures or systems using available commercial FEM
software. For students who will conduct research on the FEM, these lecture notes should only
serve as an introduction and they should consult to the references listed at the end of the notes for
more rigorous treatment of the subject in order to have the necessary theoretical background and
programming skills in developing new capabilities in the FEM.
The lecture notes include eight chapters and can be used for an undergraduate FEM
course in one semester (15 weeks with three one-hour lectures each week) or in two quarters (20
weeks with three one-hour lectures each week, with four or five accompanying computer
laboratory sessions each quarter). Chapter 1 gives a basic introduction to the concept of the FEM
using the spring system as examples. It also reviews the matrix algebra that is essential for the
FEM. Chapter 2 introduces the bar and beam elements and outlines the general procedures in the
formulations and application of the FEM. Chapter 3 covers 2-D problems in elasticity, that is,
plane stress and plane strain elements. Chapter 4 discusses various modeling techniques in the
FEM and related topics, such as error indicators and how to evaluate the FEM results. Chapter 5
gives an introduction to the plate and shell elements, emphasizing the correct use of these types
of elements. Chapter 6 provides the formulations and applications of the FEM in general 3-D
elasticity problems. Chapter 7 is an introduction of the FEM in structural vibration and dynamics
analysis, covering normal modes, harmonic and transient responses of structures using the FEM.
Chapter 8 covers the basics in thermal analysis of structures using the FEM. Exercise problems
and/or projects using FEM software packages are provided at the end of each chapter. Further
readings are provided in the Reference section to conclude the lecture notes.
The author thanks many of his former undergraduate and graduate students at the
University of Cincinnati for their suggestions on the earlier versions of these lecture notes and
for their contributions to many of the examples used in the lecture notes.
Yijun Liu
Cincinnati, Ohio, USA
Winter 2007
© 1997-2013 Yijun Liu, University of Cincinnati v
Lecture Notes: Introduction to the Finite Element Method
Chapter 1. Introduction
I. Some Basic Concepts
A Simple Idea
The finite element method (FEM), or finite element analysis (FEA), is based on the idea
of building a complicated object with simple blocks, or, dividing a complicated object into small
and manageable pieces. Application of this simple idea can be found everywhere in everyday life
(see, e.g., Figure 1.1), as well as in engineering. For example, children play the toy Lego by
using many small pieces of simple geometries to build various objects such as trains, ships or
buildings. With more and smaller pieces, these objects often look more realistic. As another
example, a digital image, which looks smooth and colorful, is in fact composed of millions of
dots that just have one simple color.
(a) (b)
Figure 1.1. Objects built with simple and small pieces: (a) a fire engine built with Lego®; and
(b) a house built with many elements – bricks, beams, columns, panels and so on.
In mathematical terms, this is simply the use of the limit concept, that is, to approach or
represent a smooth objects with a finite number of simple pieces and increasing the number of
such pieces will increase the accuracy of this representation. For example, ancient people used
this concept to estimate the area of a circle well before the formula A = πR 2 was established
(where R is the radius of the circle). In doing so, a circle is approximated by a polygon or divided
into many triangles (Figure 1.2). The area of one triangle is given by:
1
Si = Ri Li ,
2
where Ri is the height and Li the base length of the triangle. The area A of the circle can be
therefore obtained in the following manner:
N
1 1
S N = ∑ S i = N Ri Li → RLtotal = π R 2 = A, as N → ∞ ,
i =1 2 2
where N is the total number of triangles (or elements) and Ltotal = NLi is, in the limit, the
circumference of the circle, which is 2πR as it is known now.
© 1997-2013 Yijun Liu, University of Cincinnati 1
Lecture Notes: Introduction to the Finite Element Method
“Element” Si
R
Li
Ri
Figure 1.2. Approximation of the area of a circle using a finite number of triangles.
From the above mentioned examples, one can conclude that objects with complicated
geometries can be represented by many small pieces (or elements) with simple geometries. As
the number of such pieces increase, the representation becomes more accurate. This is exactly
the same concept used in the development of the FEM as one will see in later chapters.
Why the Finite Element Method?
Computers have revolutionized the practice of engineering. Design of a product that used
to be done by tedious hand drawings has been replaced by computer-aided design (CAD) using
computer graphics. Analysis of a design used to be done by hand calculations and many of the
testing have been replaced by computer simulations using computer-aided engineering (CAE)
software. Together, CAD, CAE and CAM (computer-aided manufacturing) have dramatically
changed the landscape of engineering. For example, a car, that used to take five to six years from
design to product, can now be produced starting from the concept design to the manufacturing
within about 18 months using the CAD/CAE/CAM technologies. A company without adopting
the CAD/CAE technologies is deemed to lose ground in the competitive market place. FEM is
the most widely applied simulation tool in CAE or one of the most powerful calculators available
for engineering students.
Applications of the FEM in Engineering
There are numerous applications of the FEM in industries today and below is only a very
short list:
• Mechanical/Aerospace/Automobile/Civil/Electrical Engineering
• Structure stress and dynamic analysis
• Thermal/fluid flows
• Electrostatics/Electromagnetics
• Geomechanics
• Biomechanics
© 1997-2013 Yijun Liu, University of Cincinnati 2
Lecture Notes: Introduction to the Finite Element Method
Figure 1.3. Modeling of gear coupling using the FEM.
A Brief History of the FEM
An account of the historical development of finite element method and the computational
mechanics in general was given by O. C. Zienkiewicz recently, which can be found in Ref. [1]. A
few major milestones are as follows:
1943 ----- Courant (Variational methods which laid the foundation for FEM)
1956 ----- Turner, Clough, Martin and Topp (Stiffness method)
1960 ----- Clough (Coined “Finite Element”, solved plane problems)
1970s ----- Applications on “mainframe” computers
1980s ----- Microcomputers, development of pre- and postprocessors (GUI)
1990s ----- Analysis of large structural systems, nonlinear and dynamic problems
Figure 1.4. Simulating a can drop with dynamic FEM.
© 1997-2013 Yijun Liu, University of Cincinnati 3
Lecture Notes: Introduction to the Finite Element Method
FEM in Structural Analysis (The Procedure in Using the FEM)
• Divide structure into pieces (elements with nodes, Figure 1.5)
• Describe the behavior of the physical quantities on each element
• Connect (assemble) the elements at the nodes to form an approximate system of
equations for the whole structure
• Solve the system of equations involving unknown quantities at the nodes (for example,
the displacements)
• Calculate desired quantities (for example, strains and stresses) at selected elements
(a) (b)
Figure 1.5. (a) A plate with a hole; and (b) A FEM discretization (mesh).
Computer Implementations
A typical FEM software has the following three key components:
• Preprocessor (used to build FE models, apply loads and constraints)
• FEA solver (assemble and solve the FEM system of equations)
• Postprocessor (sort and display the results)
The computer graphical-user interface (GUI) of the ANSYS software is shown in
Figure 1.6. Other FEM packages have similar interfaces.
© 1997-2013 Yijun Liu, University of Cincinnati 4
Lecture Notes: Introduction to the Finite Element Method
File
menu
Quick
tool
menu
Main
menu
Graphics
window
Figure 1.6. GUI of the ANSYS software.
Available Commercial FEM Software Packages
There are many commercial FEM packages, or CAD/CAE software in general, available
for conducting FEA in nearly all fields of engineering. The following is only a short list:
• ANSYS (General purpose, PCs and workstations)
• UG/NX (Complete CAD/CAM/CAE package)
• NASTRAN (General purpose FEA on mainframes and PCs)
• ABAQUS (Nonlinear and dynamic analyses)
• COSMOS (General purpose FEA)
• ALGOR (PCs and workstations)
• PATRAN (Pre/Post Processor)
• HyperWorks/HyperMesh (Pre/Post Processor)
• Dyna-3D (Crash/impact analysis)
• Others
© 1997-2013 Yijun Liu, University of Cincinnati 5
Lecture Notes: Introduction to the Finite Element Method
Objectives of This Course
• Understand the fundamental ideas of the FEM
• Know the behavior and usage of each type of elements covered in this course
• Be able to prepare a suitable FE model for a given problem
• Can interpret and evaluate the quality of the results (know the physics of the problems)
• Be aware of the limitations of the FEM (do not misuse the FEM)
II. Review of Matrix Algebra
Linear System of Algebraic Equations
a 11 x1 + a 12 x 2 + ... + a 1n x n = b1
a 21 x1 + a 22 x 2 + ... + a 2 n x n = b2
(1.1)
.......
a n1 x1 + a n 2 x 2 + ... + a nn x n = bn
where x1, x2, ..., xn are the unknowns.
In matrix form:
Ax = b (1.2)
where
a11 a12 ... a1n x1 b1
a a 22 ... a 2 n x b
A = [aij ] = 21 , x = {xi } = 2 , b = {bi } = 2 (1.3)
... ... ... ... : :
xn bn
a n1 an 2 ... a nn
A is called a n×n (square) matrix, and x and b are (column) vectors with dimension n.
Row and Column Vectors
w1
v = [v1 v2 v3 ] w = w2 (1.4)
w3
Matrix Addition and Subtraction
For two matrices A and B, both of the same size (m×n), the addition and subtraction are
defined by
© 1997-2013 Yijun Liu, University of Cincinnati 6
Lecture Notes: Introduction to the Finite Element Method
C = A + B with cij = aij + bij
(1.5)
D = A − B with d ij = aij − bij
Scalar Multiplication
λA = [λaij ] (1.6)
Matrix Multiplication
For two matrices A (of size l×m) and B (of size m×n), the product of AB is defined by
m
C = AB with cij = ∑ aik bkj (1.7)
k =1
where i = 1, 2, ..., l; j = 1, 2, ..., n.
Note that, in general, AB ≠ BA , but ( AB)C = A ( BC) (associative).
Transpose of a Matrix
If A = [aij], then the transpose of A is
A T = [a ji ] (1.8)
Notice that
( AB)T = BT A T (1.9)
Symmetric Matrix
A square (n×n) matrix A is called symmetric, if
A = AT or a ij = a ji (1.10)
Unit (Identity) Matrix
1 0 ... 0
0 1 ... 0
I= (1.11)
... ... ... ...
0 0 ... 1
Note that AI = A, Ix = x.
Determinant of a Matrix
The determinant of square matrix A is a scalar number denoted by det A or |A|. For
2×2 and 3×3 matrices, their determinants are given by
a b
det = ad − bc (1.12)
c d
and
© 1997-2013 Yijun Liu, University of Cincinnati 7
Lecture Notes: Introduction to the Finite Element Method
a11 a12 a13
det a 21 a 22 a 23 = a11a 22 a 33 + a12 a 23a 31 + a 21a 32 a13
(1.13)
a 31 a 32 a 33
− a13a 22 a 31 − a12 a 21a 33 − a 23a 32 a11
Singular Matrix
A square matrix A is singular if det A = 0, which indicates problems in the system
(nonunique solutions, degeneracy, etc.)
Matrix Inversion
For a square and nonsingular matrix A ( det A ≠ 0 ), its inverse A-1 is constructed in such
a way that
AA −1 = A −1 A = I (1.14)
The cofactor matrix C of matrix A is defined by
Cij = ( −1)i + j M ij (1.15)
where Mij is the determinant of the smaller matrix obtained by eliminating the ith row and jth
column of A.
Thus, the inverse of A can be determined by
1
A −1 = CT (1.16)
det A
We can show that ( AB) −1 = B −1 A −1 .
Examples:
−1
a b 1 d − b
(A) c d = ( ad − bc ) − c a .
Checking:
−1
a b a b 1 d − b a b 1 0
c d c d = ( ad − bc ) − c a c d = 0 1 .
−1 T
1 −1 0 3 2 1 3 2 1
− 1 2 − 1 1 2 2 1 = 2 2 1 .
(B) =
( 4 − 2 − 1)
0 − 1 2 1 1 1 1 1 1
Checking:
© 1997-2013 Yijun Liu, University of Cincinnati 8
Lecture Notes: Introduction to the Finite Element Method
1 − 1 0 3 2 1 1 0 0
− 1 2 − 1 2 2 1 = 0 1 0 .
0 − 1 2 1 1 1 0 0 1
If det A = 0 (i.e., A is singular), A-1 does not exist.
The solution of the linear system of equations (Eq. (1.1)) can be expressed as (assuming
the coefficient matrix A is nonsingular)
x = A −1b
Thus, the main task in solving a linear system is to find the inverse of the coefficient matrix.
Solution Techniques for Linear Systems of Equations
• Gauss elimination methods
• Iterative methods
We will briefly review the two methods in Chapter 4.
Positive Definite Matrix
A square (n×n) matrix A is said to be positive definite, if for all nonzero vector x of
dimension n,
x T Ax > 0
Note that positive definite matrices are nonsingular. Later on we will see that all stiffness
matrices are positive definite and the above condition is a statement that the strain energy in a
structure should be positive if the structure is constrained and the stiffness matrix is nonsingular.
Differentiation and Integration of a Matrix
Let A(t ) = aij (t ) , then the differentiation is defined by
d daij (t )
A (t ) = (1.17)
dt dt
and the integration by
∫ A(t )dt = ∫ a (t )dt
ij (1.18)
© 1997-2013 Yijun Liu, University of Cincinnati 9
Lecture Notes: Introduction to the Finite Element Method
Types of Finite Elements
We are now ready to study the various finite elements. All the elements developed in the
FEM can be categorized into the following three types according to their geometries.
1-D (Line) Elements:
Figure 1.6. 1-D elements: Springs, trusses, beams, pipes, etc.
2-D (Plane) Elements:
Figure 1.7. 2-D elements: Membranes, plates, shells, etc.
3-D (Solid) Elements:
Figure 1.8. 3-D elements for 3-D fields (temperature, displacement, stress, velocity, etc.).
We will start with the 1-D spring element as an example to study the basic concept and
ingredients in the FEM.
© 1997-2013 Yijun Liu, University of Cincinnati 10
Lecture Notes: Introduction to the Finite Element Method
III. Spring Element
“Everything important is simple!”
One Spring Element
We first study a single spring element (Figure 1.9) and then a system of spring elements.
x
i j
fi ui uj fj
k
Figure 1.9. One spring element.
Two nodes: i, j
Nodal displacements: ui, uj (m, mm)
Nodal forces: fi, fj (Newton)
Spring constant (stiffness): k (N/m, N/mm)
Relationship between spring force F and elongation Δ is shown in Figure 1.10.
Linear
F
Nonlinear
Figure 1.10. Force-displacement relation in a spring.
We only consider linear problems in this introductory course. In the linear portion of the
curve shown in Figure 1.10, we have
F = k∆ , with ∆ = u j − ui . (1.19)
© 1997-2013 Yijun Liu, University of Cincinnati 11
Lecture Notes: Introduction to the Finite Element Method
where k = F / ∆ (> 0) is the stiffness of the spring (the force needed to produce a unit stretch).
Consider the equilibrium of forces for the spring. At node i, we have
f i = − F = − k (u j − ui ) = kui − ku j fi i F
and at node j,
F j fj
f j = F = k (u j − ui ) = −kui + ku j
In matrix form,
k − k ui f i
− k = (1.20)
k u j f j
or,
ku = f (1.21)
where
k = (element) stiffness matrix
u = (element) nodal displacement vector
f = (element) nodal force vector
From the derivation, we see that the first equation in (1.20) represents the equilibrium of
forces at node i, while the second equation in (1.20) that of forces at node j. Note also that k is
symmetric. Is k singular or nonsingular? That is, can we solve the equation in (1.20)? If not,
why?
Spring System
x
k1 k2
1 2 3
u1, F1 u2, F2 u3, F3
Figure 1.11. A system of two spring elements.
For a system of multiple spring elements, we first write down the stiffness equation for
each spring and then “assemble” them together to form the stiffness equation for the whole
system. For example, for the two-spring system shown in Figure 1.11, we proceed as follows:
For element 1, we have
© 1997-2013 Yijun Liu, University of Cincinnati 12
Lecture Notes: Introduction to the Finite Element Method
k1 − k1 u1 f11
− k = (1.22)
1 k1 u2 f 21
and for element 2,
k2 − k 2 u2 f12
− k = (1.23)
2 k 2 u3 f 22
where f i m is the (internal) force acting on local node i of element m (i = 1, 2).
Assemble the stiffness matrix for the whole system:
Consider the equilibrium of forces at node 1,
F1 1 f11
F1 = f 1
1
at node 2,
f21 2 f12
F2 = f + f
2
1
1
2
F2
and at node 3,
f22 3 F3
F3 = f 2
2
Using (1.22) and (1.23), we obtain
F1 = k1u1 − k1u2
F2 = −k1u1 + ( k1 + k 2 )u2 − k 2 u3
F3 = −k 2 u2 + k 2 u3
In matrix form, we have
k1 − k1 0 u1 F1
− k
k1 + k 2 − k 2 u2 = F2 (1.24)
1
0 − k2 k 2 u3 F3
or
Ku = F (1.25)
in which, K is the stiffness matrix (structure matrix) for the entire spring system.
An alternative way of assembling the whole stiffness matrix:
© 1997-2013 Yijun Liu, University of Cincinnati 13
Lecture Notes: Introduction to the Finite Element Method
“Enlarging” the stiffness matrices for elements 1 and 2, we have
k1 − k1 0 u1 f11
− k
k1 0 u2 = f 21 ,
1
0 0 0 u3 0
and
0 0 0 u1 0
0 k
− k 2 u2 = f12
2
0 − k 2 k 2 u3 f 22
Adding the two matrix equations (i.e., using superposition), we have
k1 − k1 0 u1 f11
− k
k1 + k 2 − k 2 u2 = f 21 + f12
1
0 − k2 k 2 u3 f 22
This is the same equation we derived by using the force equilibrium concept.
Energy approach:
We can also obtain the result using an energy method, for example, the principle of
minimum potential energy. In fact, the energy approach is more general and considered the
foundation of the FEM. To proceed, we consider the strain energy U stored in the spring system
shown in Figure 1.11.
1 2 1 2 1 T 1 T
U= k1∆1 + k 2 ∆ 2 = ∆1 k1∆1 + ∆ 2 k 2 ∆ 2
2 2 2 2
However,
u u
∆1 = u2 − u1 = [− 1 1] 1 , ∆ 2 = u3 − u2 = [− 1 1] 2
u2 u3
We have
1 k − k1 u1 1 k − k 2 u2
U= [u1 u2 ] 1 + [u2 u3 ] 2 = (enlarging…)
2 − k1 k1 u2 2 − k 2 k 2 u3
© 1997-2013 Yijun Liu, University of Cincinnati 14
Lecture Notes: Introduction to the Finite Element Method
k1 − k1 0 u1
1
= [u1 u2 u3 ] − k1 k1 + k 2 − k 2 u2 (1.26)
2
0 − k2 k 2 u3
The potential of the external forces is
F1
Ω = − F1u1 − F2 u2 − F3u3 = −[u1 u2 u3 ] F2 (1.27)
F
3
Thus, the total potential energy of the system is
k1 − k1 0 u1 F1
1
Π = U + Ω = [u1 u2 u3 ] − k1 k1 + k 2 − k 2 u2 − [u1 u2 u3 ] F2 (1.28)
2
0 − k2 k 2 u3 F
3
which is a function of the three nodal displacements ( u1 , u2 , u3 ). According to the principle of
minimum potential energy, for a system to be in equilibrium, the total potential energy must be
minimum, that is, dΠ = 0 , or equivalently,
∂Π ∂Π ∂Π
= 0, = 0, = 0, (1.29)
∂u1 ∂u2 ∂u3
which yield the same three equations as in (1.24).
Boundary and load conditions:
Assuming that node 1 is fixed, and same force P is applied at node 2 and node 3, that is
u1 = 0 and F2 = F3 = P
we have from Eq. (1.24)
k1 − k1 0 0 F1
− k
k1 + k 2 − k 2 u2 = P
1
0 − k2 k 2 u3 P
which reduces to
k1 + k 2 − k 2 u2 P
−k =
2 k 2 u3 P
© 1997-2013 Yijun Liu, University of Cincinnati 15
Lecture Notes: Introduction to the Finite Element Method
and
F1 = −k1u2
Unknowns are
u
U = 2 and the reaction force F1 (if desired).
u3
Solving the equations, we obtain the displacements
u2 2 P / k1
=
u3 2 P / k1 + P / k 2
and the reaction force
F1 = −2 P
Checking the Results
• Deformed shape of the structure
• Equilibrium of the external forces
• Order of magnitudes of the obtained values
Notes About the Spring Elements
• Spring elements are only suitable for stiffness analysis
• They are not suitable for stress analysis of the spring itself
• There are spring elements with stiffness in the lateral direction, spring elements for
torsion, etc.
Example 1.1
k1 k2 P k3
x
1 2 3 4
Given: For the spring system shown above,
© 1997-2013 Yijun Liu, University of Cincinnati 16
Lecture Notes: Introduction to the Finite Element Method
k1 = 100 N/mm, k 2 = 200 N/mm, k 3 = 100 N/mm
P = 500 N, u1 = u4 = 0
Find: (a) the global stiffness matrix
(b) displacements of nodes 2 and 3
(c) the reaction forces at nodes 1 and 4
(d) the force in the spring 2
Solution:
(a) The element stiffness matrices are (make sure to put proper unit after each number)
100 − 100
k1 = (N/mm)
− 100 100
200 − 200
k2 = (N/mm)
− 200 200
100 − 100
k3 = (N/mm)
− 100 100
Applying the superposition concept, we obtain the global stiffness matrix for the spring system
u1 u2 u3 u4
100 − 100 0 0
− 100 100 + 200 − 200 0
K=
0 − 200 200 + 100 − 100
0 0 − 100 100
or
100 − 100 0 0
− 100 300 − 200 0
K=
0 − 200 300 − 100
0 0 − 100 100
which is symmetric and banded.
Equilibrium (FE) equation for the whole system is
© 1997-2013 Yijun Liu, University of Cincinnati 17
Lecture Notes: Introduction to the Finite Element Method
100 − 100 0 0 u1 F1
− 100 300 − 200 0 u2 F2
=
0 − 200 300 − 100 u3 F3
0 0 − 100 100 u4 F4
(b) Applying the BCs u1 = u4 = 0 , F2 = 0 and F3 = P , and “deleting” the 1st and 4th
rows and columns, we have
300 − 200 u2 0
− 200 300 u = P
3
Solving this equation, we obtain
u2 P / 250 2
= = ( mm)
u3 3P / 500 3
(c) From the 1st and 4th equations system FE equation, we obtain the reaction forces
F1 = −100u2 = −200( N)
F4 = −100u3 = −300( N )
(d) The FE equation for spring (element) 2 is
200 − 200 ui f i
− 200 200 u = f
j j
Here i = 2, j = 3 for element 2. Thus we can calculate the spring force as
u 2
F = f j = − f i = [− 200 200] 2 = [− 200 200] = 200 (N)
u 3 3
Check the results:
Draw the free-body diagram (FBD) of the system and consider the equilibrium of the
forces.
F1 = -200 N F4 = -300 N
P = 500 N
Equilibrium of the forces is satisfied!
© 1997-2013 Yijun Liu, University of Cincinnati 18
Lecture Notes: Introduction to the Finite Element Method
Example 1.2
4 k4 F1
k1 1
2
4 k2 F2 k3
1
2 3 3 5
x
Problem: For the spring system with arbitrarily numbered nodes and elements, as shown
above, find the global stiffness matrix.
Solution: First we construct the following element connectivity table
Element Connectivity Table
Element Node i (1) Node j (2)
1 4 2
2 2 3
3 3 5
4 2 1
which specifies the global node numbers corresponding to the local node numbers for each
element.
Then we write the element stiffness matrix for each element
u4 u2 u2 u3 u3 u5 u2 u1
k − k1 , k − k2 , k − k3 , k − k4
k1 = 1 k2 = 2 k3 = 3 k4 = 4
− k1 k1 − k 2 k 2 − k 3 k 3 − k 4 k 4
Finally, applying the superposition method, we obtain the global stiffness matrix as follows
u1 u2 u3 u4 u5
k4 − k4 0 0 0
− k k1 + k 2 + k 4 − k2 − k1 0
4
K= 0 − k2 k2 + k3 0 − k3
0 − k1 0 k1 0
0 0 − k3 0 k 3
The matrix is symmetric, banded, but singular.
© 1997-2013 Yijun Liu, University of Cincinnati 19
Lecture Notes: Introduction to the Finite Element Method
IV. Summary
In this chapter, the basic concepts in the finite element method are introduced. The spring
element is used as an example to show how to establish the element stiffness matrices, to
assemble the finite element equations for a system from element stiffness matrices, and to solve
the FE equations. Verifying the FE results is emphasized. The concepts and procedures
introduced in this chapter are very simple and yet very important for studying the finite element
analyses of other problems.
V. Problems
Problem 1. Answer the following questions briefly:
(a) What is the physical meaning of the FE equations (for either an element
or the whole structure)?
(b) What is the procedure in using the FEM?
Problem 2. For the following given matrix and vector
1 − 2 6 0
A=2 2 3, b = 3
2
− 1 3 0
Find:
(a) Determinant det A;
(b) Inverse A-1;
(c) Solution of the equation Ax = b;
(d) Value of the quadratic form xTAx.
Problem 3. A spring system is shown below
1 k1
F3 k3 F4
k2 3 4
2
x
Given: k1 = 80 N/mm, k2 = 100 N/mm, k3 = 160 N/mm, F3 = 200 N, F4 = 100 N ,
and nodes 1 and 2 are fixed;
Find:
(a) Global stiffness matrix;
(b) Displacements of nodes 3 and 4;
(c) Reaction forces at nodes 1 and 2;
(d) Forces in springs 1 and 2.
© 1997-2013 Yijun Liu, University of Cincinnati 20
Lecture Notes: Introduction to the Finite Element Method
Chapter 2. Bar and Beam Elements
I. Linear Static Analysis
Most structural analysis problems can be treated as linear static problems, based on the
following assumptions
1. Small deformations (loading pattern is not changed due to the deformed shape)
2. Elastic materials (no plasticity or failures)
3. Static loads (the load is applied to the structure in a slow or steady fashion)
Linear analysis can provide most of the information about the behavior of a structure, and
can be a good approximation for many analyses. It is also the bases of nonlinear analysis in
most of the cases.
II. Bar Element
Consider a uniform prismatic bar:
ui uj
fi i j fj
x A, E
L
Figure 2.1. Notations for a bar element.
L, A, E length, cross-sectional area, and elastic modulus of the beam, respectively
u, ε , σ displacement, strain, and stress, respectively (all functions of x)
Strain-displacement relation:
du
ε= (2.1)
dx
Stress-strain relation:
σ = Eε (2.2)
Equilibrium equation:
dσ
+ f =0 (2.3)
dx
© 1997-2013 Yijun Liu, University of Cincinnati 21
Lecture Notes: Introduction to the Finite Element Method
where f is the body force (force per volume, such as gravitational and magnetic forces) inside the
bar.
Stiffness Matrix --- Direct Method
Assuming that the displacement u is varying linearly along the axis of the bar, that is, in
terms of the two nodal values ui and u j , we can write (derive this)
x x
u ( x ) = 1 − u i + u j (2.4)
L L
we have
u j − ui ∆
ε= = ( ∆ = elongation) (2.5)
L L
E∆
σ = Eε = (2.6)
L
We also have
F
σ= (F = force in bar) (2.7)
A
Thus, (2.6) and (2.7) lead to
EA
F= ∆ = k∆
L
EA
where k = is the stiffness of the bar. That is, the bar behaves like a spring in this case and
L
we conclude that the element stiffness matrix is
EA EA
− k L −
k
k= = L
− k k − EA EA
L L
or
EA 1 − 1
k= (2.8)
L − 1 1
This can be verified by considering the equilibrium of the forces at the two nodes.
Element equilibrium equation is
EA 1 − 1 ui f i
= (2.9)
L − 1 1 u j f j
Degree of Freedom (DOF): Number of components of the displacement vector at a node. For 1-
D bar element: one DOF at each node.
© 1997-2013 Yijun Liu, University of Cincinnati 22
Lecture Notes: Introduction to the Finite Element Method
Physical Meaning of the Coefficients in k: The jth column of k (here j = 1 or 2) represents the
forces applied to the bar to maintain a deformed shape with unit displacement at node j and zero
displacement at the other node.
Stiffness Matrix --- A Formal Approach
We derive the same stiffness matrix for the bar using a formal approach which can be
applied to many other more complicated situations.
Define two linear shape functions as follows
N i (ξ ) = 1 − ξ , N j (ξ ) = ξ (2.10)
where
x
ξ= , 0 ≤ξ ≤1 (2.11)
L
Ni(ξ) Nj(ξ)
1 1
i j ξ i j ξ
ξ=0 ξ=1 ξ=0 ξ=1
Figure 2.2. The shape of the shape functions.
From (2.4) we can write the displacement as
u( x ) = u(ξ ) = N i (ξ )ui + N j (ξ )u j
or
ui
u = [N i N j ] = Nu (2.12)
u j
Strain is given by (2.1) and (2.12) as
du d
ε= = N u = Bu (2.13)
dx dx
where B is the element strain-displacement matrix, which is
dξ
B=
d
dx
[N i (ξ ) N j (ξ )] =
d
dξ
[N i (ξ ) N j (ξ )]•
dx
that is,
B = [− 1 / L 1 / L] (2.14)
Stress can be written as
© 1997-2013 Yijun Liu, University of Cincinnati 23
Lecture Notes: Introduction to the Finite Element Method
σ = Eε = EBu (2.15)
Consider the strain energy stored in the bar
σ TεdV = ∫ (u T B T EBu )dV
1 1
U=
2V∫ 2V
(2.16)
1
= u T ∫ (B T EB )dV u
2 V
where (2.13) and (2.15) have been used.
The potential of the external forces is written as (this is by definition, and remember the
negative sign)
Ω = − f i ui − f j u j = −u T f (2.17)
The total potential of the system is
Π =U +Ω
which yields by using (2.16) and (2.17)
1 T
Π= u ∫ (B T EB )dV u − u T f (2.18)
2 V
Setting dΠ = 0 by the principle of minimum potential energy, we obtain (verify this)
∫ (B EB )dV u = f
T
V
or
ku = f (2.19)
where
k = ∫ (B T EB )dV (2.20)
V
is the element stiffness matrix.
Expression (2.20) is a general result which can be used for the construction of other types
of elements.
Now, we evaluate (2.20) for the bar element by using (2.14)
− 1 / L
L
EA 1 − 1
k = ∫ E [− 1 / L 1 / L]Adx=
0
1/ L L − 1 1
which is the same as we derived earlier using the direct method.
Note that from (2.16) and (2.20), the strain energy in the element can be written as
© 1997-2013 Yijun Liu, University of Cincinnati 24
Lecture Notes: Introduction to the Finite Element Method
1
U = u T ku (2.21)
2
In the future, once we obtain an expression like (2.16), we can immediately recognize that the
matrix within the square bracket is the stiffness matrix. Recall that for a spring, the strain energy
can be written as
1 2 1 T
U= k∆ = ∆ k∆
2 2
Thus result (2.21) goes back to the simple spring case again.
Example 2.1
1 2A,E 2 A,E
1 2 P 3 x
L L
Problem: Find the stresses in the two-bar assembly which is loaded with force P, and
constrained at the two ends, as shown in the above figure.
Solution: Use two 1-D bar elements.
For element 1,
u1 u2
2 EA 1 − 1
k1 =
L − 1 1
For element 2,
u2 u3
EA 1 − 1
k2 =
L − 1 1
Imagine a frictionless pin at node 2, which connects the two elements. We can assemble the
global FE equation as follows
2 − 2 0 u1 F1
EA
− 2 3 − 1 u2 = F2
L
0 − 1 1 u3 F3
Load and boundary conditions (BCs) are
u1 = u3 = 0, F2 = P
FE equation becomes
© 1997-2013 Yijun Liu, University of Cincinnati 25
Lecture Notes: Introduction to the Finite Element Method
2 − 2 0 0 F1
EA
− 2 3 − 1 u2 = P
L
0 − 1 1 0 F3
“Deleting” the 1st row and column, and the 3rd row and column, we obtain
EA
[3]{u2 } = {P}
L
Thus,
PL
u2 =
3EA
and
u1 0
PL
u2 = 1
u 3 EA 0
3
Stress in element 1 is
u
σ 1 = Eε 1 = EB1u1 = E [− 1 / L 1 / L] 1
u2
u2 − u1 E PL P
=E = − 0 =
L L 3EA 3A
Similarly, stress in element 2 is
u
σ 2 = Eε 2 = EB 2 u 2 = E [− 1 / L 1 / L] 2
u3
u3 − u 2 E PL P
=E = 0 − =−
L L 3EA 3A
which indicates that bar 2 is in compression.
Check the results: Draw the FBD and check the equilibrium of the structures.
-2P/3 P -P/3
Notes:
• In this case, the calculated stresses in elements 1 and 2 are exact. It will not help if
we further divide element 1 or 2 into smaller elements.
© 1997-2013 Yijun Liu, University of Cincinnati 26
Lecture Notes: Introduction to the Finite Element Method
• For tapered bars, averaged values of the cross-sectional areas should be used for the
elements.
• We need to find the displacements first in order to find the stresses, and thus this
approach is called the displacement based FEM.
Example 2.2
∆
1 A,E 2
1 2 P 3 x
L L
Problem: Determine the support reaction forces at the two ends of the bar shown above,
given the following
P = 6.0 ×104 N, E = 2.0 ×104 N/mm 2 ,
A = 250 mm 2 , L = 150 mm, ∆ = 1.2 mm
Solution:
We first check to see if contact of the bar with the wall on the right will occur or not. To
do this, we imagine the wall on the right is removed and calculate the displacement at the right
end
PL (6.0 × 10 4 )(150)
∆0 = = = 1.8mm > ∆ = 1.2mm
EA ( 2.0 × 10 4 )( 250)
Thus, contact occurs and the wall on the right should be accounted for in the analysis.
The global FE equation is found to be
1 − 1 0 u1 F1
EA
− 1 2 − 1 u2 = F2
L
0 − 1 1 u3 F3
The load and boundary conditions are
F2 = P = 6.0 × 10 4 N
u1 = 0, u3 = ∆ = 1.2mm
© 1997-2013 Yijun Liu, University of Cincinnati 27
Lecture Notes: Introduction to the Finite Element Method
FE equation becomes
1 − 1 0 0 F1
EA
− 1 2 − 1 u2 = P
L
0 − 1 1 ∆ F3
The 2nd equation gives
[2 − 1] 2 = {P}
EA u
L ∆
that is,
EA
[2]{u2 } = P + EA ∆
L L
Solving this, we obtain
1 PL
u2 = + ∆ = 1.5mm
2 EA
and
u1 0
u 2 = 1.5( mm)
u 1.2
3
To calculate the support reaction forces, we apply the 1st and 3rd equations in the global
FE equation.
The 1st equation gives
u1
F1 =
EA
[1 − 1 0]u2 = EA (− u2 ) = −5.0 × 104 N
L u L
3
and the 3rd equation gives,
u1
F3 =
EA
[0 − 1 1]u2 = EA (− u2 + u3 ) = −1.0 × 10 4 N
L u L
3
Check the results.!
© 1997-2013 Yijun Liu, University of Cincinnati 28
Lecture Notes: Introduction to the Finite Element Method
Distributed Load
q
i j
x
fiq fjq
i j
Figure 2.3. Conversion of a distributed load on one element.
Distributed axial load q (N/mm, N/m, lb/in) can be converted to two equivalent nodal
forces using the shape functions. Consider the work done by the distributed load q,
L L L
N i ( x)
u( x ) q( x )dx = ∫ (Nu ) q( x )dx = [ui u j ]∫
1 1 1
∫
T
Wq = q( x )dx
20 20 2 0
N j ( x )
L
(2.22)
1
= uT ∫ N T q( x )dx
2 0
The work done by the equivalent nodal forces are
1 q 1 1
W fq = f i ui + f jq u j = uT f q (2.23)
2 2 2
Setting Wq = W f q and using (2.22) and (2.23), we obtain the equivalent nodal force vector
f i q L T L
N i ( x)
f q = q = ∫ N q( x )dx = ∫ q( x )dx (2.24)
f j 0 0
N j ( x )
which is valid for any distributions of q. If q is a constant, we have
1 − x / L
L
qL / 2
fq = q ∫ dx = (2.25)
0
x/L qL / 2
In an assembly of bar elements, equivalent forces are added at each node as shown below.
q
1 2 3
qL/2 qL qL/2
1 2 3
Figure 2.4. Conversion of a distributed load on two elements.
© 1997-2013 Yijun Liu, University of Cincinnati 29
Lecture Notes: Introduction to the Finite Element Method
Bar Elements in 2-D and 3-D Spaces
2-D Case
x
y j
Y
ui’ θ
i vi
ui
X
Figure 2.5. Local and global coordinates for a bar in 2-D space.
Local Global
x, y X, Y
ui' ,vi' ui ,vi
1 DOF at each node 2 DOFs at each node
Note that lateral displacement vi’ does not contribute to the stretch of the bar within the
linear theory.
Transformation
u
ui' = ui cosθ + vi sin θ = [l m] i
vi
u
vi' = −ui sin θ + vi cosθ = [− m l ] i
vi
where l = cosθ ,m = sin θ .
In matrix form,
ui' l m ui
'= (2.26)
vi − m l vi
or,
~
u i' = Tu i
where the transformation matrix
© 1997-2013 Yijun Liu, University of Cincinnati 30
Lecture Notes: Introduction to the Finite Element Method
~ l m
T= (2.27)
− m l
~ ~
is orthogonal, that is, T −1 = T T .
For the two nodes of the bar element, we have
ui' l m 0 0 ui
'
vi − m l 0 0 vi
'= (2.28)
u j 0 0 l m u j
v 'j 0
0 − m l v j
or,
~
T 0
u ' = Tu with T= ~ (2.29)
0 T
The nodal forces are transformed in the same way,
f ' = Tf (2.30)
Stiffness Matrix in the 2-D Space
In the local coordinate system, we have
EA 1 − 1 ui f i
' '
=
L − 1 1 u 'j f j'
Augmenting this equation, we write
1 0 −1 0 ui' f i '
EA 0 0 0 0 vi' 0
=
L − 1 0 1 0 u 'j f j'
0 0 0 0 v 'j 0
or,
k 'u' = f '
Using transformations given in (29) and (30), we obtain
k ' Tu = Tf
Multiplying both sides by TT and noticing that TTT = I, we obtain
T T k ' Tu = f (2.31)
© 1997-2013 Yijun Liu, University of Cincinnati 31
Lecture Notes: Introduction to the Finite Element Method
Thus, the element stiffness matrix k in the global coordinate system is
k = TT k ' T (2.32)
which is a 4×4 symmetric matrix.
Explicit form,
ui vi uj vj
l2 lm − l 2 − lm
(2.33)
EA lm m2 − lm − m 2
k=
L − l 2 − lm l2 lm
− lm − m
2
lm m2
Calculation of the directional cosines l and m:
X j − Xi Y j − Yi
l = cosθ = , m = sin θ =
L L
The structure stiffness matrix is assembled by using the element stiffness matrices in the usual
way as in the 1-D case.
Element Stress
ui
u '
1 1 l m 0 0 vi
σ = Eε = EB = E −
i
'
u j L L 0 0 l m u j
v j
That is,
ui
v
E i
σ= [− l − m l m] (2.34)
L u j
v j
which can be used to evaluate the element stress once the nodal displacements are known.
© 1997-2013 Yijun Liu, University of Cincinnati 32
Lecture Notes: Introduction to the Finite Element Method
Example 2.3
3
45o
2 P2
Y 2 P1
1
45o
X
1
A simple plane truss is made of two identical bars (with E, A, and L), and loaded as
shown in the above figure.
Find:
(a) displacement of node 2;
(b) stress in each bar.
Solution:
This simple structure is used here to demonstrate the FEA procedure using the bar
element in 2-D space.
In local coordinate systems, we have
EA 1 − 1
k 1' = = k '2
L − 1 1
These two matrices cannot be assembled together, because they are in different coordinate
systems. We need to convert them to global coordinate system OXY.
2
Element 1: θ = 45o , l=m=
2
Using formula (2.32) or (2.33), we obtain the stiffness matrix in the global system
u1 v1 u2 v2
1 1 − 1 − 1
1 1 − 1 − 1
EA
k 1 = T1 k 1T1 =
T '
2 L − 1 − 1 1 1
− 1 − 1 1 1
© 1997-2013 Yijun Liu, University of Cincinnati 33
Lecture Notes: Introduction to the Finite Element Method
2 2
Element 2: θ = 135o , l=− ,m =
2 2
u2 v2 u3 v3
1 −1 −1 1
1 − 1
EA − 1 1
k 2 = T2 k 2 T2 =
T '
2 L − 1 1 1 − 1
1 −1 −1 1
Assemble the structure FE equation,
u1 v1 u2 v2 u3 v3
1 1 −1 −1 0 0 u1 F1 X
1 1 −1 −1 0 0 v1 F1Y
EA − 1 − 1 2 0 − 1 1 u 2 F2X
=
2 L − 1 − 1 0 2 1 − 1 v2 F2Y
0 0 −1 1 1 − 1 u3 F3 X
0 0 1 − 1 − 1 1 v3 F3Y
Load and boundary conditions (BC):
u1 = v1 = u3 = v3 = 0, F2 X = P1 , F2Y = P2
Condensed FE equation,
EA 2 0 u2 P1
=
2 L 0 2 v 2 P2
Solving this, we obtain the displacement of node 2,
u2 L P1
=
v 2 EA P2
Using formula (2.34), we calculate the stresses in the two bars,
0
E 2 L 0
σ1 = [− 1 − 1 1 1] = 2 (P1 + P2 )
L 2 EA P1 2 A
P2
© 1997-2013 Yijun Liu, University of Cincinnati 34
Lecture Notes: Introduction to the Finite Element Method
P1
P
σ2 =
E 2
[1 − 1 − 1 1] L 2 = 2 (P1 − P2 )
L 2 EA 0 2 A
0
Check the results:
Check the equilibrium conditions, symmetry, antisymmetry, etc.
Example 2.4 (Multipoint Constraint)
y’
x’
P 3
2 2
1 Y
L
3
1 45o
X
For the plane truss shown above,
P = 1000 kN, L = 1m, E = 210GPa ,
A = 6.0 × 10 − 4 m 2 for elements 1 and 2,
A = 6 2 × 10 − 4 m 2 for element 3.
Determine the displacements and reaction forces.
Solution:
We have an inclined roller at node 3, which needs special attention in the FE solution.
We first assemble the global FE equation for the truss.
Element 1: θ = 90o , l = 0, m = 1
u1 v1 u2 v2
0 0 0 0
0 − 1
( 210 × 10 9 )(6.0 × 10 −4 ) 0 1 ( N/m)
k1 =
1 0 0 0 0
0 − 1 0 1
Element 2: θ = 0o , l = 1, m = 0
© 1997-2013 Yijun Liu, University of Cincinnati 35
Lecture Notes: Introduction to the Finite Element Method
u 2 v 2 u3 v3
1 0 −1 0
0
( 210 × 10 9 )(6.0 × 10 −4 ) 0 0 0
( N/m)
k2 =
1 − 1 0 1 0
0 0 0 0
1 1
Element 3: θ = 45o , l= ,m =
2 2
u1 v1 u3 v3
0.5 0.5 − 0.5 − 0.5
0.5
−4
0.5 − 0.5 − 0.5
( 210 × 10 )(6 2 × 10 )
9
( N/m)
k3 =
2 − 0.5 − 0.5 0.5 0.5
− 0.5 − 0.5 0.5 0.5
The global FE equation is,
0.5 0.5 0 0 − 0.5 − 0.5 u1 F1 X
1.5 0 − 1 − 0.5 − 0.5 v1 F1Y
5 1 0 −1 0 u2 F2 X
1260 × 10 =
1 0 0 v 2 F2Y
1.5 0.5 u3 F3 X
Sym. 0.5 v3 F3Y
Load and boundary conditions (BCs):
u1 = v1 = v2 = 0, and v3' = 0,
F2 X = P, F3 x ' = 0.
From the transformation relation and the BC, we have
2 2 u3 2
v3' = − = ( −u3 + v3 ) = 0,
2 2 v3 2
that is,
u3 − v 3 = 0
This is a multipoint constraint (MPC).
Similarly, we have a relation for the force at node 3,
2 2 F3 X 2
F3 x ' = = ( F3 X + F3Y ) = 0,
2 2 F3Y 2
that is,
© 1997-2013 Yijun Liu, University of Cincinnati 36
Lecture Notes: Introduction to the Finite Element Method
F3 X + F3Y = 0
Applying the load and BC’s in the structure FE equation by “deleting” the 1st, 2nd and
4th rows and columns, we have
1 − 1 0 u2 P
1260 × 10 − 1 1.5 0.5 u3 = F3 X
5
0 0.5 0.5 v3 F3Y
Further, from the MPC and the force relation at node 3, the equation becomes,
1 − 1 0 u2 P
1260 × 10 − 1 1.5 0.5 u3 = F3 X
5
0 0.5 0.5 u3 − F3 X
which is
1 − 1 P
u2
1260 × 10 − 1 2 = F3 X
5
u3
0 1 − F
3X
The 3rd equation yields,
F3 X = −1260 × 105 u3
Substituting this into the 2nd equation and rearranging, we have
1 − 1 u2 P
1260 × 105 =
− 1 3 u3 0
Solving this, we obtain the displacements,
u2 1 3P 0.01191
= 5 = ( m)
u3 2520 × 10 P 0.003968
From the global FE equation, we can calculate the reaction forces,
F1 X 0 − 0.5 − 0.5 − 500
F 0 − 0.5 − 0.5 u − 500
1Y 2
F2Y = 1260 × 10 0 0 u3 = 0.0 ( kN )
5
0
F
− 1 1.5 0.5 v3 − 500
3X
F3Y 0 0.5 0.5 500
Check the results!
A general multipoint constraint (MPC) can be described as,
∑Au
j
j j =0
© 1997-2013 Yijun Liu, University of Cincinnati 37
Lecture Notes: Introduction to the Finite Element Method
where Aj’s are constants and uj’s are nodal displacement components. In FE software, users only
need to specify this relation to the software. The software will take care of the solution process.
3-D Case
y
x j
Y
i
z
X
Z
Figure 2.6. Local and global coordinates for a bar in 3-D space.
Local Global
x, y, z X, Y, Z
ui' ,vi' ,wi' ui ,vi ,wi
1 DOF at each node 3 DOFs at each node
Element stiffness matrices are calculated in the local coordinate systems and then
transformed into the global coordinate system (X, Y, Z) where they are assembled. The
transformation relation is
ui' l X lY l Z ui
'
vi = m X mY m Z vi (2.35)
w' n nY n Z wi
i X
where (l X , lY , l Z ), ( m X , mY , mZ ) and ( n X , nY , n Z ) are the direction cosines of the local x, y and z
coordinate axis in the global coordinate system, respectively. FEM software packages will do
this transformation automatically.
Input data for bar elements:
• (X, Y, Z) for each node
• E and A for each element (Length L can be computed from the coordinates of the two
nodes)
© 1997-2013 Yijun Liu, University of Cincinnati 38
Lecture Notes: Introduction to the Finite Element Method
III. Beam Element
Simple Plane Beam Element
y
vi, Fi vj, Fj
i j
E,I x
θi, Mi θj, Mj
L
Figure 2.7. Notion for a simple beam element in 2-D.
L, I, E length, moment of inertia of the cross-sectional area and elastic modulus
v = v (x ) deflection (lateral displacement) of the neutral axis of the beam
dv
θ= rotation of the beam about the z-axis
dx
Q = Q (x ) (internal) shear force
M = M (x ) (internal) bending moment about z-axis
Fi , M i , F j , M j applied (external) lateral forces and moments at node i and j, respectively
Elementary Beam Theory:
We have the following results from the simple beam theory
d 2v
EI = M ( x) (2.36)
dx 2
dM dQ
= Q ( x ), = q( x ) (2.37)
dx dx
where q(x ) is the distributed load in the lateral direction. Combining (2.36) and (2.37), we have
d 4v
EI 4 = q( x ) (2.38)
dx
Bending stress in the beam is given by
My
σ =− (2.39)
I
Simple beam theory and thus the simple beam elements are valid for “long” slender beams, for
example, for beams with aspect ratios (length/height) larger than 10.
© 1997-2013 Yijun Liu, University of Cincinnati 39
Lecture Notes: Introduction to the Finite Element Method
Direct Method
We first apply the direct method to establish the beam stiffness matrix using the results
from elementary beam theory. The FE equation for a beam takes the form
k11 k12 k13 k14 vi Fi
k k 22 k 23 k 24 θ i M i
21 =
k 31 k 32 k 33 k 34 v j F j
k 41 k 42 k 43 k 44 θ j M j
Recall that each column in the stiffness matrix represents the forces needed to keep the structure
in a special deformed shape. For example, the first column represents the forces/moments to
keep the shape with vi = 1, θ i = v j = θ j = 0 as shown in Figure 2.8 (a). Thus, using the results
from strength of materials for a cantilever beam with a force k11 and moment k 21 applied at the
free end, we have
k11 L3 k 21 L2 k11 L2 k 21 L
vi = − =1 and θi = − + =0
3EI 2 EI 2 EI EI
Solving this system of equations, we obtain k11 and k 21 . Using the equilibrium conditions of the
beam, we obtain k 31 and k 41 , and thus the first column of the stiffness matrix.
0.15 1 0
1
0 0.5 1
0.1
-0.05
0.5 0.5
0.05
-0.1
0 0 0
0 0.5 1 0 0.5 1 0 0.5 1 -0.15
(a) N1 ( x) (b) N 2 ( x) (c) N 3 ( x) (d) N 4 ( x)
Figure 2.8. The shape of the shape function for the simple beam element.
Element stiffness equation (local node: i, j or 1, 2):
12 6 L − 12 6 L vi Fi
2
EI 6 L 4 L − 6 L 2 L θ i M i
2
= (2.40)
L3 − 12 − 6 L 12 − 6 L v j F j
2
6 L 2 L − 6 L 4 L θ j M j
2
Formal Approach
To derive the stiffness matrix in (2.40), we introduce four shape functions (Figure 2.8),
© 1997-2013 Yijun Liu, University of Cincinnati 40
Lecture Notes: Introduction to the Finite Element Method
N 1 ( x ) = 1 − 3x 2 / L2 + 2 x 3 / L3
N 2 ( x ) = x − 2 x 2 / L + x 3 / L2
(2.41)
N 3 ( x ) = 3x 2 / L2 − 2 x 3 / L3
N 4 ( x ) = − x 2 / L + x 3 / L2
Then, we can represent the deflection as,
vi
θ
i
v ( x ) = Nu = [N 1 ( x ) N 2 ( x ) N 3 ( x ) N 4 ( x )] (2.42)
v j
θ j
which is a cubic function. Notice that,
N1 + N 3 = 1
N 2 + N 3L + N 4 = x
which implies that the rigid-body motion is represented correctly by the assumed deformed
shape of the beam.
Curvature of the beam is,
d 2v d 2
= Nu = Bu (2.43)
dx 2 dx 2
where the strain-displacement matrix B is given by,
B=
d2
dx 2
[ ]
N = N 1" ( x ) N 2" ( x ) N 3" ( x ) N 4" ( x )
(2.44)
6 12 x 4 6 x 6 12 x 2 6x
= − 2 + 3 − + 2 2
− 3 − + 2
L L L L L L L L
Strain energy stored in the beam element is
L T L
1 1 My 1 My 1 T 1
U = ∫ σ T εdV = ∫ ∫ − − dAdx = ∫ M Mdx
2V 2 0 A I E I 20 EI
L T L
1 d 2v d 2v 1
= ∫ 2 EI 2 dx = ∫ (Bu ) EI (Bu )dx
T
2 0 dx dx 20
1 T T
L
= u ∫ B EIBdx u
2 0
We conclude that the stiffness matrix for the simple beam element is
L
k = ∫ B T EIBdx (2.45)
0
© 1997-2013 Yijun Liu, University of Cincinnati 41
Lecture Notes: Introduction to the Finite Element Method
Applying the result in (2.44) and carrying out the integration, we arrive at the same stiffness
matrix as given in (2.40).
Combining the axial stiffness (from the bar element), we obtain the stiffness matrix of a
general 2-D beam element,
ui vi θi uj vj θj
EA EA
L 0 0 − 0 0
L
12 EI 6 EI 12 EI 6 EI
0 0 −
L3 L2 L3 L2
0 6 EI 4 EI 6 EI 2 EI
0 − 2 (2.46)
k= L2 L L L
EA EA
− 0 0 0 0
L L
12 EI 6 EI 12 EI 6 EI
0 − 3 − 0 − 2
L L2 L3 L
0 6 EI 2 EI 6 EI 4 EI
0 − 2
L2 L L L
3-D Beam Element
The element stiffness matrix is formed in the local (2-D) coordinate system first and then
transformed into the global (3-D) coordinate system to be assembled. Details for 3-D beam
elements can be found in the references listed at the end of the lecture notes.
Example 2.5
Y
P
M
1 2
1 E,I 2 3 X
L L
Given: The beam shown above is clamped at the two ends and acted upon by the force P
and moment M in the mid-span.
Find: The deflection and rotation at the center node and the reaction forces and
moments at the two ends.
Solution: Element stiffness matrices are
© 1997-2013 Yijun Liu, University of Cincinnati 42
Lecture Notes: Introduction to the Finite Element Method
v1 θ1 v2 θ2
12 6 L − 12 6 L
2
EI 6 L 4 L − 6 L 2 L
2
k1 = 3
L − 12 − 6 L 12 − 6 L
2
6L 2 L − 6L 4 L
2
v2 θ2 v3 θ3
12 6 L − 12 6 L
6 L 4 L2 − 6 L 2 L2
EI
k2 = 3
L − 12 − 6 L 12 − 6 L
2
6L 2 L − 6L 4 L
2
Global FE equation is
v1 θ1 v2 θ2 v3 θ3
12 6 L − 12 6 L 0 0 v1 F1Y
6 L 4 L2 − 6 L 2 L2 0 0 θ1 M 1
EI − 12 − 6 L 24 0 − 12 6 L v 2 F2Y
3 2 =
L 6L 2 L 2
0 8L 2
− 6 L 2 L θ 2 M 2
0 0 − 12 − 6 L 12 − 6 L v3 F3Y
0 0 6 L 2 L2 − 6 L 4 L2 θ 3 M 3
Loads and constraints (BC’s) are
F2Y = − P, M2 = M, v1 = v3 = θ1 = θ 3 = 0
Reduced FE equation
EI 24 0 v2 − P
=
L3 0 8L2 θ 2 M
Solving this, we obtain
v2 L − PL2
=
θ 2 24 EI 3M
From the global FE equation, we obtain the reaction forces and moments
F1Y − 12 6 L 2 P + 3M / L
M − 6 L 2 L2 v
1 EI 2 1 PL + M
= 3 =
F3Y L − 12 − 6 L θ 2 4 2 P − 3M / L
M 3 2
− PL + M
6L 2 L
Stresses in the beam at the two ends can be calculated using the formula
© 1997-2013 Yijun Liu, University of Cincinnati 43
Lecture Notes: Introduction to the Finite Element Method
My
σ =σx = −
I
Note that the FE solution is exact for this problem according to the simple beam theory, since no
distributed load is present between the nodes. Recall that (Eq. (2.38))
d 4v
EI = q( x )
dx 4
If q(x)=0, then exact solution for the deflection v is a cubic function of x, which is exactly what
described by the shape functions given in (2.42).
Equivalent Nodal Loads of Distributed Transverse Load
q(x)
i x L j
Fiq Fjq
Mi q
Mj q
i j
Figure 2.9. Conversion of the distributed lateral load to nodal forces and moments.
To convert a distributed load to nodal forces and moments, we consider again the work
done by the distributed load q
L L L
1 1 1
∫ v ( x ) q( x )dx = ∫ (Nu ) q( x )dx = u T ∫ N T q( x )dx
T
Wq =
20 20 2 0
The work done by the equivalent nodal forces (and moments) is
Fi q
q
M 1
W fq =
1
[vi θi vj θ j ] qi = u T f q
2 Fj 2
M qj
© 1997-2013 Yijun Liu, University of Cincinnati 44
Lecture Notes: Introduction to the Finite Element Method
By equating Wq = W f q , we obtain the equivalent nodal force vector as
L
f q = ∫ N T q( x )dx (2.47)
0
which is valid for arbitrary distributions of q(x ) . For constant q, we have the results shown in
Figure 2.10 (verify this). An example of this result is given in Figure 2.11.
i x L j
qL/2 qL/2
qL2/12
qL2/12
i j
Figure 2.10. Conversion of a constant distributed lateral load to nodal forces and moments.
L L
qL qL/2
qL2/12
L L
Figure 2.11. Conversion of a constant distributed lateral load on two beam elements.
© 1997-2013 Yijun Liu, University of Cincinnati 45
Lecture Notes: Introduction to the Finite Element Method
Example 2.6
y
p
1 E,I 2 x
L
Given: A cantilever beam with distributed lateral load p as shown above.
Find: The deflection and rotation at the right end, the reaction force and moment at the
left end.
Solution: The work-equivalent nodal loads are shown below,
y
f
m
1 E,I 2 x
L
where
f = pL / 2, m = pL2 / 12
Applying the FE equation, we have
12 6 L − 12 6 L v1 F1Y
6 L 4 L2 − 6 L 2 L2 θ M
EI 1 = 1
L − 12 − 6 L 12 − 6 L v2 F2Y
3
2
6 L 2 L − 6 L 4 L θ 2 M 2
2
Load and constraints (BCs) are
F2Y = − f , M2 = m
v1 = θ1 = 0
Reduced equation is
EI 12 − 6 L v2 − f
L3 − 6 L 4 L2 θ = m
2
Solving this, we obtain
© 1997-2013 Yijun Liu, University of Cincinnati 46
Lecture Notes: Introduction to the Finite Element Method
v2 L − 2 L2 f + 3Lm − pL4 / 8EI
= = (A)
θ 2 6 EI − 3Lf + 6m − pL / 6 EI
3
These nodal values are the same as the exact solution. Note that the deflection v(x) (for 0
< x< L) in the beam by the FEM is, however, different from that by the exact solution. The exact
solution by the simple beam theory is a 4th order polynomial of x, while the FE solution of v is
only a 3rd order polynomial of x.
If the equivalent moment m is ignored, we have,
v2 L − 2 L2 f − pL4 / 6 EI
= = (B)
θ 2 6 EI − 3Lf − pL / 4 EI
3
The errors in (B) will decrease if more elements are used. The equivalent moment m is often
ignored in the FEM applications. The FE solutions still converge as more elements are applied.
From the FE equation, we can calculate the reaction force and moment as,
F1Y EI − 12 6 L v 2 pL / 2
= 3 2 =
M 1 L − 6 L 2 L θ 2 5 pL / 12
2
where the result in (A) has been used. This force vector gives the total effective nodal forces
which include the equivalent nodal forces for the distributed lateral load p given by,
− pL / 2
− pL / 12
2
The correct reaction forces can be obtained as follows,
F1Y pL / 2 − pL / 2 pL
= − = 2
M 1 5 pL / 12 − pL / 12 pL / 2
2 2
Check the results:
Draw the FBD for the FE model (with the equivalent nodal force vector) and the check
the equilibrium condition.
pL/2 pL/2
5pL2/12 pL2/12
© 1997-2013 Yijun Liu, University of Cincinnati 47
Lecture Notes: Introduction to the Finite Element Method
Example 2.7
Y
P
1 E,I 2
2 X
1 3 k
L L 4
Given: P = 50 kN, k = 200 kN/m, L = 3 m, E = 210 GPa, I = 2×10-4 m4.
Find: Deflections, rotations and reaction forces.
Solution:
The beam has a roller (or hinge) support at node 2 and a spring support at node 3. We
use two beam elements and one spring element to solve this problem.
The spring stiffness matrix is given by
v3 v4
k − k
ks =
− k k
Adding this stiffness matrix to the global FE equation (see Example 2.5), we have
v1 θ1 v2 θ2 v3 θ3 v4
12 6 L − 12 6 L 0 0 0 v1 F1Y
4L − 6L 2L
2 2
0 0 0 θ1 M 1
24 0 − 12 6L 0 v2 F2Y
EI =
− θ 2 M 2
2 2
8 L 6 L 2 L 0
L
3
12 + k ' − 6 L − k ' v3 F3Y
4 L2 0 θ 3 M 3
Symmetry k ' v4 F4Y
in which
L3
k' = k
EI
is used to simplify the notation.
We now apply the boundary conditions
v1 = θ1 = v2 = v4 = 0,
M 2 = M 3 = 0, F3Y = − P
© 1997-2013 Yijun Liu, University of Cincinnati 48
Lecture Notes: Introduction to the Finite Element Method
‘Deleting’ the first three and seventh equations (rows and columns), we have the following
reduced equation
8L2 − 6L 2 L2 θ 2 0
EI
3
− 6 L 12 + k ' − 6 L v3 = − P
L
2 L2
− 6L 4 L2 θ 3 0
Solving this equation, we obtain the deflection and rotations at node 2 and node 3,
θ 2 3
PL2
v3 = − 7 L
θ EI (12 + 7k ' )
3 9
The influence of the spring k is easily seen from this result. Plugging in the given numbers, we
can calculate
θ 2 − 0.002492 rad
v 3 = − 0.01744 m
θ − 0.007475 rad
3
From the global FE equation, we obtain the nodal reaction forces as,
F1Y − 69.78 kN
M − 69.78 kN ⋅ m
1
=
F2Y 116.2 kN
F4Y 3.488 kN
Checking the results: Draw free body diagram of the beam
69.78 kN 50 kN
1 2 3
69.78 kN⋅m
116.2 kN 3.488 kN
Sum the forces and moments to verify that equilibrium of the beam is satisfied.
© 1997-2013 Yijun Liu, University of Cincinnati 49
Lecture Notes: Introduction to the Finite Element Method
FE Analysis of Frame Structures
Members in a frame are considered to be rigidly connected (for example, welded
together). Both forces and moments can be transmitted through their joints. We need the general
beam element (combinations of bar and simple beam elements) to model frames.
Example 2.8
Y 500 lb/ft
3000 lb 1 2
1
8 ft
E, I, A 2 3
3 4 X
12 ft
Given: E = 30 × 10 6 psi, I = 65 in. 4 , A = 6.8 in. 2
Find: Displacements and rotations of the two joints 1 and 2.
Solution:
For this example, we first convert the distributed load to its equivalent nodal loads to
obtain the following FE mode.
3000 lb 3000 lb
72000 lb-in.
3000 lb 1 1 2 72000 lb-in.
2 3
3 4
© 1997-2013 Yijun Liu, University of Cincinnati 50
Lecture Notes: Introduction to the Finite Element Method
In local coordinate system, the stiffness matrix for a general 2-D beam element is
ui vi θi uj vj θj
EA EA
L 0 0 − 0 0
L
12 EI 6 EI 12 EI 6 EI
0 0 −
L3 L2 L3 L2
0 6 EI 4 EI 6 EI 2 EI
0 − 2
k= L2 L L L
EA EA
− L 0 0
L
0 0
12 EI 6 EI 12 EI 6 EI
0 − 3 − 0 − 2
L L2 L3 L
0 6 EI 2 EI 6 EI 4 EI
0 − 2
L2 L L L
Element Connectivity Table
Element Node i (1) Node j (2)
1 1 2
2 3 1
3 4 2
For element 1, we have
u1 v1 θ1 u2 v2 θ2
141.7 0 0 − 141.7 0 0
0 0.784 56.4 0 − 0.784 56.4
0 56.4 5417 0 − 56.4 2708
k 1 = k 1 ' = 10 ×
4
− 141.7 0 0 141.7 0 0
0 − 0.784 − 56.4 0 0.784 − 56.4
0 56.4 2708 0 − 56.4 5417
For elements 2 and 3, the stiffness matrix in local system is
ui ' vi ' θi ' uj' vj' θj'
212.5 0 0 − 212.5 0 0
0 2.65 127 0 − 2.65 127
0 127 8125 0 − 127 4063
k 2 ' = k 3 ' = 10 4 ×
− 212.5 0 0 212.5 0 0
0 − 2.65 − 127 0 2.65 − 127
0 127 4063 0 − 127 8125
© 1997-2013 Yijun Liu, University of Cincinnati 51
Lecture Notes: Introduction to the Finite Element Method
where i = 3, j = 1 for element 2, and I = 4, j = 2 for element 3.
The transformation matrix T is
l m 0 0 0 0
− m l 0 0 0 0
0 0 1 0 0 0
T=
0 0 0 l m 0
0 0 0 − m l 0
0 0 0 0 0 1
We have l = 0, m = 1 for both elements 2 and 3. Thus,
0 1 0 0 0 0
− 1 0 0 0 0 0
0 0 1 0 0 0
T=
0 0 0 0 1 0
0 0 0 − 1 0 0
0 0 0 0 0 1
Using the transformation relation
k = T T k' T
we obtain the stiffness matrices in the global coordinate system for elements 2 and 3
u3 v3 θ3 u1 v1 θ1
2.65 0 − 127 − 2.65 0 − 127
0 212.5 0 0 − 212.5 0
− 127 0 8125 127 0 4063
k 2 = 10 4 ×
− 2.65 0 127 2.65 0 127
0 − 212.5 0 0 212.5 0
− 127 0 4063 127 0 8125
and
u4 v4 θ4 u2 v2 θ2
2.65 0 − 127 − 2.65 0 − 127
0 212.5 0 0 − 212.5 0
− 127 0 8125 127 0 4063
k 3 = 10 4 ×
− 2.65 0 127 2.65 0 127
0 − 212.5 0 0 212.5 0
− 127 0 4063 127 0 8125
© 1997-2013 Yijun Liu, University of Cincinnati 52
Lecture Notes: Introduction to the Finite Element Method
Assembling the global FE equation and noticing the following boundary conditions
u3 = v3 = θ3 = u4 = v4 = θ 4 = 0
F1 X = 3000 lb, F2 X = 0, F1Y = F2Y = −3000 lb,
M 1 = −72000 lb ⋅ in., M 2 = 72000 lb ⋅ in.
we obtain the condensed FE equation
144.3 0 127 − 141.7 0 0 u1 3000
0 213.3 56.4 0 − 0.784 56.4 v1 − 3000
127 56 . 4 13542 0 − 56 . 4 2708 θ1 − 72000
10 4 × =
− 141 . 7 0 0 144 . 3 0 127 u 2 0
0 − 0.784 − 56.4 0 213.3 − 56.4 v 2 − 3000
0 56.4 2708 127 − 56.4 13542 θ 2 72000
Solving this, we obtain
u1 0.092in.
v − 0.00104in.
1
θ1 − 0.00139rad
=
u2 0.0901in.
v2 − 0.0018in.
θ 2 − 3.88 × 10 rad
−5
To calculate the reaction forces and moments at the two ends, we employ the element FE
equations for element 2 and element 3 with known nodal displacement vectors. We obtain
F3 X − 672.7lb
F3Y = 2210 lb
M 60364 lb ⋅ in.
3
and
F4 X − 2338lb
F4Y = 3825 lb
M 112641lb ⋅ in.
4
Check the results:
Draw the free-body diagram of the frame as shown below. Equilibrium is maintained
with the calculated forces and moments. Recall that the problem we solved is the one with the
equivalent loads, not the one with the distributed load. Thus the corresponding FBD for the FE
model should be applied for verifying the results.
© 1997-2013 Yijun Liu, University of Cincinnati 53
Lecture Notes: Introduction to the Finite Element Method
3000 lb 3000 lb
72000 lb-in.
3000 lb
72000 lb-in.
112641 lb-in.
60364 lb-in.
672.7 lb 2338 lb
2210 lb 3825 lb
IV. Summary
In this chapter, we studied the bar elements which can be used in truss analysis and the
beam elements which are used in frame analysis. The concept of the shape functions is further
explored and the derivations of the stiffness matrices using the energy approach are emphasized.
Treatment of distributed loads is discussed and several examples are studied.
V. Problems
Problem 1. Using (2.47) derive the results of the equivalent nodal forces and moments for a
beam element with uniformly distributed lateral load.
Problem 2. The plane truss is loaded with force P as shown below. Constants E and A for
each bar are as shown in the diagram. Determine:
(a) the nodal displacement;
(b) the reaction forces;
(c) the stresses in bar elements.
Y L
4 3
E, A
E ,2 2 A
L
o
45
45o
1 2
E, A X
P
© 1997-2013 Yijun Liu, University of Cincinnati 54
Lecture Notes: Introduction to the Finite Element Method
Problem 3. The plane truss is loaded with force P as shown below. Constants E and A for
each bar are as shown in the diagram. Determine:
(a) the nodal displacements,
(b) the stresses in each bar elements.
Y 3
P
45o E ,2 2 A
L
E, A
1 45o 2
E, A X
Problem 4. The plane truss is supported as shown below. The Young’s modulus E is the
same for all the bars. The cross-sectional areas are shown in the figure. Suppose
that the node 2 settles by an amount of δ as shown. Determine the stresses in each
bar element using the FEM.
3
Y
2 2A
A L
1 A 2
δ X
L
Problem 5. The cantilever beam is supported by a spring at the end as shown in the figure.
Using FEM, determine the deflection and rotation at the node 2.
Y
E,I P
1 2 X
k
L
3
Problem 6. Determine the nodal displacement, rotations and reaction forces for the propped
cantilever beam shown below. The beam is assumed to have constant EI and
length 2L. It is supported by a roller at midlength and is built in at the right end.
© 1997-2013 Yijun Liu, University of Cincinnati 55
Lecture Notes: Introduction to the Finite Element Method
Y
P
1 E,I 2
1 2 X
3
L L
Problem 7. The 2-D frame is supported as shown in the figure. Constants E, A, I of the beam
and the length L are given. Determine the displacement and rotation at node 2.
Y
1 L 2 X
Problem 8. The plane frame is subjected to the uniformly distributed load and is fixed at the
ends as shown in the figure. Assume E = 30×106 psi, A= 100 in.2 and
I = 1000 in.4 for both elements of the frame. Find:
(a) the displacement and rotation of node 2;
(b) the reaction forces and moments at both ends.
1000 lb/ft
2 3
Y
40 ft
45o X
1
30 ft
Problem 9. Using an FEM software package (ANSYS, NASTRAN, or ABAQUS), solve the
frame problem in Example 2.8.
© 1997-2013 Yijun Liu, University of Cincinnati 56
Lecture Notes: Introduction to the Finite Element Method
Chapter 3. Two-Dimensional Elasticity Problems
The finite element method for deformation and stress analyses of two-dimensional (2-D)
structural models will be discussed in this chapter. First, the basic equations in elasticity theory
(see Refs. [9, 10] for more details) for stress analysis are reviewed. Then several types of 2-D
finite elements are introduced. Applications of these elements are demonstrated and their
accuracies and efficiencies are discussed.
I. Stress State in Structures
In general, the stresses and strains in a structure consist of six components, that is
(Figure 3.1),
σ x ,σ y ,σ z ,τ xy ,τ yz ,τ zx for stresses,
and
ε x ,ε y ,ε z ,γ xy ,γ yz ,γ zx for strains.
σy
τ xy
τ yz
σx
τ zx
y
σz
x
z
Figure 3.1. Stress components at a point in a structure.
Under certain conditions, the state of stresses and strains can be simplified. A general 3-D
structure analysis can, therefore, be reduced to a 2-D analysis.
II. 2-D (Plane) Elasticity Problems
Plane stress
In the plane stress case, any stress component related to the z direction is zero, that is,
σ z =τ yz =τ zx = 0 (ε z ≠ 0) (3.1)
© 1997-2013 Yijun Liu, University of Cincinnati 57
Lecture Notes: Introduction to the Finite Element Method
A thin planar structure with constant thickness and loading within the plane of the structure (xy-
plane) can be regarded as a plane stress case (Figure 3.2).
y y
p
x z
Figure 3.2. An example of a plane stress case.
Plane strain
In the plane strain case, any strain component related to the z direction is zero, that is,
ε z =γ yz =γ zx = 0 (σ z ≠ 0) (3.2)
A long structure with a uniform cross section and transverse loading along its length (z-
direction), such as a tunnel, can be regarded as a plane strain case (Figure 3.3).
y y
p
x z
Figure 3.3. An example of a plane strain case.
Stress-Strain-Temperature (Constitutive) Relations
For elastic and isotropic materials, we have the following stress-strain relation for 2-D
cases
ε x 1/ E −ν / E 0 σ x ε x 0
ε y = − ν / E 1/ E 0 σ y + ε y 0 (3.3)
γ 0 0 1 / G τ xy γ xy 0
xy
or,
ε = E−1σ + ε 0
© 1997-2013 Yijun Liu, University of Cincinnati 58
Lecture Notes: Introduction to the Finite Element Method
where ε 0 is the initial strain (for example, due to a temperature change), E the Young’s
modulus, ν the Poisson’s ratio and G the shear modulus. Note that
E
G= (3.4)
2(1 + ν )
which means that there are only two independent materials constants for homogeneous and
isotropic materials.
We can also express stresses in terms of strains by solving the above equation
σ x 1 ν 0 ε x ε x 0
E ν 1 ε − ε
σ y =
0
y y 0
(3.5)
τ 1 − ν
2
xy 0 0 (1 − ν ) / 2 γ xy γ xy 0
or,
σ = Eε + σ 0
where σ 0 = −Eε 0 is the initial stress.
The above relations are valid for plane stress case. For plane strain case, we need to
replace the material constants in the above equations in the following fashion
E ν
E→ ; ν→ ; G→G (3.6)
1 −ν 2 1 −ν
For example, the stress is related to strain by
σ x 1 − ν ν 0 ε x ε x 0
E ν ε − ε
σ y =
1 −ν 0
y y 0
τ (1 + ν )(1 − 2ν )
xy 0 0 (1 − 2ν ) / 2 γ xy γ xy 0
in the plane strain case.
Initial strain due to a temperature change (thermal loading) is given by the following for
the plane stress case
ε x 0 α∆T
ε y 0 = α∆T (3.7)
γ 0
xy 0
© 1997-2013 Yijun Liu, University of Cincinnati 59
Lecture Notes: Introduction to the Finite Element Method
where α is the coefficient of thermal expansion, ∆T the change of temperature. For the plane
strain case, α should be replaced by (1 + ν )α in (3.7). Note that if the structure is free to deform
under thermal loading, there will be no (elastic) stresses in the structure.
Strain and Displacement Relations
For small strains and small rotations, we have,
∂u ∂v ∂u ∂v
εx = , εy = , γ xy = +
∂x ∂y ∂y ∂x
In matrix form, we write
ε x ∂ / ∂x 0
u
ε y = 0 ∂ / ∂y , or ε = Du (3.8)
γ ∂ / ∂y ∂ / ∂x v
xy
From this relation, we know that the strains (and thus stresses) are one order lower than
the displacements, if the displacements are represented by polynomials.
Equilibrium Equations
In elasticity theory, the stresses in the structure must satisfy the following equilibrium
equations
∂σ x ∂τ xy
+ + fx = 0
∂x ∂y
(3.9)
∂τ xy ∂σ y
+ + fy = 0
∂x ∂y
where fx and fy are body forces per unit volume (such as gravity forces). In the FEM, these
equilibrium conditions are satisfied in an approximate sense.
Boundary Conditions
ty
p
y tx
St
Su
x
Figure 3.4. Boundary conditions for a structure.
© 1997-2013 Yijun Liu, University of Cincinnati 60
Lecture Notes: Introduction to the Finite Element Method
The boundary S of the body can be divided into two parts, Su and St (Figure 3.4). The
boundary conditions (BCs) can be described as
u = u, v = v, on S u
(3.10)
tx = tx , t y = t y , on S t
in which tx and ty are tractions (stresses on the boundary) and the barred quantities are those with
known values. In the FEM, all types of loads (distributed surface loads, body forces,
concentrated forces and moments, etc.) are converted to point forces acting at the nodes.
Exact Elasticity Solution
The exact solution (displacements, strains and stresses) of a given problem must satisfy
the equilibrium equations (9), the given boundary conditions (10) and compatibility conditions
(structures should deform in a continuous manner, no cracks or overlaps in the obtained
displacement fields).
Example 3.1
y
A plate is supported and loaded with distributed force p as shown in the figure. The
material constants are E and ν.
The exact solution for this simple problem can be found easily as follows.
p p
Displacement: u= x, v = −ν y
E E
p p
Strain: εx = , ε y = −ν , γ xy = 0
E E
Stress: σ x = p, σ y = 0, τ xy = 0
Exact (or analytical) solutions for simple problems are numbered (suppose there is a hole
in the plate or the roller support are replaced by clamped ones!). That is why we need the FEM!
© 1997-2013 Yijun Liu, University of Cincinnati 61
Lecture Notes: Introduction to the Finite Element Method
III. Finite Elements for 2-D Problems
A General Formula for the Stiffness Matrix
Displacements (u, v) in a plane element are interpolated from nodal displacements (ui, vi)
using shape functions Ni as follows,
u1
v
u N 1 0 N2 0 1
= u or u = Nd (3.11)
v 0 N1 0 N2 2
v
2
where N is the shape function matrix, u the displacement vector and d the nodal displacement
vector. Here we have assumed that u depends on the nodal values of u only, and v on nodal
values of v only.
From strain-displacement relation (Eq.(3.8)), the strain vector is,
ε = Du = DNd, or ε = Bd (3.12)
where B = DN is the strain-displacement matrix.
Consider the strain energy stored in an element,
σ T ε dV = ∫ (σ x ε x + σ y ε y + τ xy γ xy )dV
1 1
U=
2V∫ 2V
1
= ∫ (Eε )T ε dV = 1 ∫ ε T Eε dV = 1 d T ∫ BT EBdV d
2V 2V 2 V
1 T
= d kd
2
From this, we obtain the general formula for the element stiffness matrix
k = ∫ BT EBdV (3.13)
V
Note that unlike the 1-D cases, E here is a matrix which is given by the stress-strain relation
(e.g., Eq.(3.5) for plane stress).
The stiffness matrix k defined by (3.13) is symmetric since E is symmetric. Also note
that given the material property, the behavior of k depends on the B matrix only, which in turn
on the shape functions. Thus, the quality of finite elements in representing the behavior of a
structure is mainly determined by the choice of shape functions. Most commonly employed 2-D
elements are linear or quadratic triangles and quadrilaterals.
© 1997-2013 Yijun Liu, University of Cincinnati 62
Lecture Notes: Introduction to the Finite Element Method
Constant Strain Triangle (CST or T3)
This is the simplest 2-D element (Figure 3.5), which is also called linear triangular
element.
v3
3
(x3, y3) u3
v
v2
v1 u 2
(x, y) u2
y 1 (x2, y2)
u1
(x1, y1)
x
Figure 3.5. Linear triangular element (T3).
For this element, we have three nodes at the vertices of the triangle, which are numbered
around the element in the counterclockwise direction. Each node has two degrees of freedom
(can move in the x and y directions). The displacements u and v are assumed to be linear
functions within the element, that is,
u = b1 + b2 x + b3 y , v = b4 + b5 x + b6 y (3.14)
where bi (i = 1, 2, ..., 6) are constants. From these, the strains are found to be,
ε x = b2 , ε y = b6 , γ xy = b3 + b5 (3.15)
which are constant throughout the element. Thus, we have the name “constant strain triangle”
(CST).
Displacements given by (3.14) should satisfy the following six equations
u1 = b1 + b2 x1 + b3 y1
u2 = b1 + b2 x2 + b3 y 2
v3 = b4 + b5 x3 + b6 y 3
Solving these equations, we can find the coefficients b1, b2, ..., and b6 in terms of nodal
displacements and coordinates. Substituting these coefficients into (3.14) and rearranging the
terms, we obtain
© 1997-2013 Yijun Liu, University of Cincinnati 63
Lecture Notes: Introduction to the Finite Element Method
u1
v
1
u N 1 0 N2 0 N3 0 u2
= (3.16)
v 0 N1 0 N2 0 N 3 v 2
u3
v3
where the shape functions (linear functions in x and y) are
1
N1 = {( x2 y3 − x3 y 2 ) + ( y 2 − y3 ) x + ( x3 − x2 ) y}
2A
1
N2 = {( x3 y1 − x1 y3 ) + ( y3 − y1 ) x + ( x1 − x3 ) y} (3.17)
2A
1
N3 = {( x1 y 2 − x2 y1 ) + ( y1 − y 2 ) x + ( x2 − x1 ) y}
2A
and
1 x1 y1
1
A = det 1 x 2 y2 (3.18)
2
1 x3 y 3
is the area of the triangle (Prove this!).
Using the strain-displacement relation (3.8), results (3.16) and (3.17), we have
u1
v
εx y 23 0 y 31 0 y12 0 1
1 u
ε y = Bd = 0 x32 0 x13 0 x21 2 (3.19)
γ 2A v
xy x32 y 23 x13 y 31 x21 y12 2
u3
v3
where xij = xi - xj and yij = yi - yj (i, j = 1, 2, 3). Again, we see constant strains within the element.
From stress-strain relation (Eq.( 3.5), for example), we see that stresses obtained using the CST
element are also constant.
Applying formula (3.13), we obtain the element stiffness matrix for the CST element
k = ∫ B T EBdV = tA( B T EB) (3.20)
V
© 1997-2013 Yijun Liu, University of Cincinnati 64
Lecture Notes: Introduction to the Finite Element Method
in which t is the thickness of the element. Notice that k for CST is a 6 by 6 symmetric matrix.
Both the expressions of the shape functions in (3.17) and their derivations are lengthy and
offer little insight into the behavior of the element.
η=0
ξ=0 η=b
3
ξ=a η=1
ξ=1 (a, b)
2
1
Figure 3.6. The natural coordinate system defined on the triangle.
We introduce the natural coordinates (ξ ,η ) on the triangle (Figure 3.6). Then the shape
functions can be represented simply by
N1 = ξ , N 2 = η , N 3 = 1 − ξ − η (3.21)
Notice that,
N1 + N 2 + N 3 = 1 (3.22)
which ensures that the rigid-body translation is represented by the chosen shape functions. Also,
as in the 1-D case,
1, at node i;
Ni = (3.23)
0, at the other nodes
and varies linearly within the element. The plot for shape function N1 is shown in Figure 3.7. N2
and N3 have similar features.
ξ=0
3
N1
ξ=1
1
2
1
Figure 3.7. Plot of the shape function N 1 for T3 element.
© 1997-2013 Yijun Liu, University of Cincinnati 65
Lecture Notes: Introduction to the Finite Element Method
We have two coordinate systems for the element: the global coordinates (x, y) and the
natural (local) coordinates (ξ ,η ) . The relation between the two is given by
x = N 1 x1 + N 2 x 2 + N 3 x3
(3.24)
y = N 1 y1 + N 2 y 2 + N 3 y 3
or,
x = x13ξ + x23η + x3
(3.25)
y = y13ξ + y 23η + y 3
where xij = xi - xj and yij = yi - yj (i, j = 1, 2, 3) as defined earlier.
Displacement u or v on the element can be viewed as functions of (x, y) or (ξ ,η ) . Using
the chain rule for derivatives, we have,
∂ u ∂ x ∂ y ∂ u ∂ u
∂ ξ ∂ ξ ∂ ξ ∂ x ∂ x
∂ u = ∂ x = J
∂ y ∂u ∂u
(3.26)
∂ η ∂ η ∂ η ∂ y ∂ y
where J is called the Jacobian matrix of the transformation.
From (3.25), we calculate
x y13 1 y 23 − y13
J = 13 , J −1 = (3.27)
x 23 y 23 2 A − x 23 x13
where det J = x13 y 23 − x 23 y13 = 2 A has been used (A is the area of the triangle. Prove this!).
From (3.26), (3.27), (3.16) and (3.21) we have
∂ u ∂ u
∂ x 1 y 23 − y13 ∂ ξ 1 y 23 − y13 u1 − u3
∂ u = = (3.28)
2 A − x23 x13 ∂ u 2 A − x23 x13 u2 − u3
∂ y ∂ η
Similarly,
∂ v
∂ x 1 y 23 − y13 v1 − v3
∂ v = (3.29)
2 A − x23 x13 v2 − v3
∂ y
© 1997-2013 Yijun Liu, University of Cincinnati 66
Lecture Notes: Introduction to the Finite Element Method
Using the results in (3.28) and (3.29), and the relations ε = Du = DNd = Bd , we obtain the
strain-displacement matrix,
y 23 0 y 31 0 y12 0
1
B= 0 x32 0 x13 0 x21 (3.30)
2A
x32 y 23 x13 y 31 x21 y12
which is the same as we derived earlier in (3.19).
Applications of the CST Element:
• Use in areas where the strain gradient is small.
• Use in mesh transition areas (fine mesh to coarse mesh).
• Avoid using CST in stress concentration or other crucial areas in the structure, such as
edges of holes and corners.
• Recommended for quick and preliminary FE analysis of 2-D problems.
Linear Strain Triangle (LST or T6)
v3
u3
3
v6 v5
u6 6 5 u5
v2
v1
y
u4 2 u2
1 4
u1 v4
x
Figure 3.8. Quadratic triangular element (T6).
This type of elements (Figure 3.8) is also called quadratic triangular element. There are
six nodes on this element: three corner nodes and three mid-side nodes. Each node has two
degrees of freedom (DOFs) as before. The displacements (u, v) are assumed to be quadratic
functions of (x, y),
u = b1 + b2 x + b3 y + b4 x 2 + b5 xy + b6 y 2
(3.31)
v = b7 + b8 x + b9 y + b10 x 2 + b11 xy + b12 y 2
where bi (i = 1, 2, ..., 12) are constants. From these, the strains are found to be
© 1997-2013 Yijun Liu, University of Cincinnati 67
Lecture Notes: Introduction to the Finite Element Method
ε x = b2 + 2b4 x + b5 y
ε y = b9 + b11 x + 2b12 y (3.32)
γ xy = (b3 + b8 ) + (b5 + 2b10 ) x + (2b6 + b11 ) y
which are linear functions. Thus, we have the “linear strain triangle” (LST), which provides
better results than the CST.
In the natural coordinate system we defined earlier, the six shape functions for the LST
element are
N1 = ξ ( 2ξ − 1)
N 2 = η ( 2η − 1)
N 3 = ζ ( 2ζ − 1)
(3.33)
N 4 = 4ξη
N 5 = 4ηζ
N 6 = 4ζ ξ
in which ζ = 1 − ξ − η . Each of these six shape functions represents a quadratic form on the
element as shown in Figure 3.9.
ξ=0
3
ξ=1/2
6 5
ξ=1 N1
1
4 2
1
Figure 3.9. Plot of the shape function N 1 for T6 element.
Displacements can be written as,
6 6
u = ∑ N i ui , v = ∑ N i vi (3.34)
i =1 i =1
The element stiffness matrix is still given by k = ∫ B T EBdV , but here BTEB is quadratic
V
in x and y. In general, the integral has to be computed numerically.
© 1997-2013 Yijun Liu, University of Cincinnati 68
Lecture Notes: Introduction to the Finite Element Method
Linear Quadrilateral Element (Q4)
η v3
v4 u3
η =1 u4
3 ξ
4
v2
v1
2 u2
1
y u1
η = −1
ξ = −1 ξ =1
x
Figure 3.10. Linear quadrilateral element (Q4).
There are four nodes at the corners of the quadrilateral element (Figure 3.10). In the
natural coordinate system (ξ ,η ) , the four shape functions are,
1
N1 = (1 − ξ )(1 − η )
4
1
N 2 = (1 + ξ )(1 − η )
4 (3.35)
1
N 3 = (1 + ξ )(1 + η )
4
1
N 4 = (1 − ξ )(1 + η )
4
4
Note that ∑N
i =1
i = 1 at any point inside the element, as expected.
The displacement field is given by
4 4
u = ∑ N i ui , v = ∑ N i vi (3.36)
i =1 i =1
which are bilinear functions over the element. The stress and strain fields are constant on this
type of elements.
Quadratic Quadrilateral Element (Q8)
This is the most widely used element for 2-D problems due to its high accuracy in
analysis and flexibility in modeling.
© 1997-2013 Yijun Liu, University of Cincinnati 69
Lecture Notes: Introduction to the Finite Element Method
η =1 7
3 ξ
4
6
8
5 2
1
y
η = −1
ξ = −1 ξ =1
x
Figure 3.11. Quadratic quadrilateral element (Q8).
There are eight nodes for this element (Figure 3.11), four corners nodes and four midside
nodes. In the natural coordinate system (ξ ,η ) , the eight shape functions are,
1
N1 = (1 − ξ )(η − 1)(ξ + η + 1)
4
1
N2 = (1 + ξ )(η − 1)(η − ξ + 1)
4
1
N3 = (1 + ξ )(1 + η )(ξ + η − 1)
4
1
N4 = (ξ − 1)(η + 1)(ξ − η + 1)
4 (3.37)
1
N5 = (1 − η )(1 − ξ 2 )
2
1
N6 = (1 + ξ )(1 − η 2 )
2
1
N7 = (1 + η )(1 − ξ 2 )
2
1
N8 = (1 − ξ )(1 − η 2 )
2
8
Again, we have ∑N
i =1
i = 1 at any point inside the element.
The displacement field is given by
8 8
u = ∑ N i ui , v = ∑ N i vi (3.38)
i =1 i =1
© 1997-2013 Yijun Liu, University of Cincinnati 70
Lecture Notes: Introduction to the Finite Element Method
which are quadratic functions over the element. Strains and stresses over a quadratic
quadrilateral element are linear functions, which are better representations. A model of fiber-
reinforced composite materials using the Q8 elements is shown in Figure 3.12.
Notes:
• Q4 and T3 are usually used together in a mesh with linear elements.
• Q8 and T6 are usually applied in a mesh composed of quadratic elements.
• Quadratic elements are preferred for stress analysis, because of their high accuracy and
the flexibility in modeling complex geometry, such as curved boundaries.
Figure 3.12. Analysis of composite materials (mesh and contour stress plots).
Transformation of Loads
Concentrated load (point forces), surface traction (pressure loads) and body force
(weight) are the main types of loads applied to a structure. Both traction and body forces need to
be converted to nodal forces in the FE model, since they cannot be applied to the FE model
directly. The conversions of these loads are based on the same idea (the equivalent-work
concept) which we have used for the cases of bar and beam elements.
© 1997-2013 Yijun Liu, University of Cincinnati 71
Lecture Notes: Introduction to the Finite Element Method
qB
q fB
qA
fA
s
B B
A L A
Figure 3.13. Traction applied on the edge of a Q4 element.
Suppose, for example, we have a linearly varying traction q on a Q4 element edge, as
shown in the Figure 3.13. The traction is normal to the boundary. Using the local (tangential)
coordinate s, we can write the work done by the traction q as
L
1
2 ∫0
Wq = t un ( s ) q( s )ds
where t is the thickness, L the side length and un the component of displacement normal to the
edge AB.
For the Q4 element (linear displacement field), we have
un ( s ) = (1 − s / L)unA + ( s / L)unB
The traction q(s), which is also linear, is given in a similar way
q( s ) = (1 − s / L) q A + ( s / L) qB
Thus, we have,
1 1 − s / L q A
L
Wq = t ∫ [u nA u nB ]
[1 − s / L s / L ] q ds
2 0 s / L B
1
L
(1 − s / L) 2 ( s / L)(1 − s / L) q A
= [u nA u nB ]t ∫ ds q
0 ( s / L )(1 − s / L ) ( s / L) 2
2 B
1 tL 2 1 q A
= [unA u nB ]
2 6 1 2 q B
1 f
= [unA u nB ] A
2 fB
and hence the equivalent nodal force vector is
f A tL 2 1 q A
=
f B 6 1 2 q B
© 1997-2013 Yijun Liu, University of Cincinnati 72
Lecture Notes: Introduction to the Finite Element Method
Note, for constant q, we have
f A qtL 1
=
fB 2 1
For quadratic elements (either triangular or quadrilateral), the traction is converted to
forces at three nodes along the edge, instead of two nodes. Traction tangent to the boundary, as
well as body forces, are converted to nodal forces in a similar way.
Stress Calculation
The stress in an element is determined by the following relation,
σ x εx
σ y = E ε y = EBd (3.39)
τ γ
xy xy
where B is the strain-nodal displacement matrix and d is the nodal displacement vector which is
known for each element once the global FE equation has been solved.
Stresses can be evaluated at any point inside the element (such as the center) or at the
nodes. Contour plots are usually used in FEA software packages (during post-process) for users
to visually inspect the stress results.
The von Mises Stress:
The von Mises stress is the effective or equivalent stress for 2-D and 3-D stress analysis.
For a ductile material, the stress level is considered to be safe, if
σ e ≤ σY
where σ e is the von Mises stress and σ Y the yield stress of the material. This is a generalization
of the 1-D (experimental) result to 2-D and 3-D situations.
The von Mises stress is defined by
1
σe = (σ 1 − σ 2 ) 2 + (σ 2 − σ 3 ) 2 + (σ 3 − σ 1 ) 2 (3.40)
2
in which σ 1 , σ 2 and σ 3 are the three principle stresses at the considered point in a structure.
For 2-D problems, the two principle stresses in the plane are determined by
© 1997-2013 Yijun Liu, University of Cincinnati 73
Lecture Notes: Introduction to the Finite Element Method
2
σx +σ y σx −σ y
σ1 =
P
+ + τ xy2
2 2
(3.41)
2
σx +σ y σx −σ y
σ 2P = − + τ xy2
2 2
Thus, we can also express the von Mises stress in terms of the stress components in the xy
coordinate system. For plane stress conditions, we have,
σ e = (σ x + σ y ) 2 − 3(σ xσ y − τ xy2 ) (3.42)
Averaged Stresses:
Stresses are usually averaged at nodes in FEA software packages to provide more
accurate stress values. This option should be turned off at nodes between two materials or other
geometry discontinuity locations where stress discontinuity does exist.
Example 3.2
A square plate with a hole at the center is under a tension load p in x direction as shown
in the figure.
A
x
B
The dimension of the plate is 10 in. x 10 in., thickness is 0.1 in. and radius of the hole is 1
in. Assume E = 10x106 psi, v = 0.3 and p = 100 psi. Find the maximum stress in the plate.
FE Analysis:
This is a plane stress case. From the knowledge of stress concentrations, we should
expect the maximum stresses occur at points A and B on the edge of the hole. Value of this
© 1997-2013 Yijun Liu, University of Cincinnati 74
Lecture Notes: Introduction to the Finite Element Method
stress should be around 3p (= 300 psi) which is the exact solution for an infinitely large plate
with a circular hole.
We use the ANSYS software to do the modeling (meshing) and analysis, using quadratic
triangular (T6), linear quadrilateral (Q4) and quadratic quadrilateral (Q8) elements. The FEM
results by using the three different elements are compared and their accuracies and efficiencies
are discussed. One mesh plot and one stress contour plot are shown below.
An FE mesh (T6, 1518 elements)
FE stress plot and deformed shape (T6, 1518 elements)
© 1997-2013 Yijun Liu, University of Cincinnati 75
Lecture Notes: Introduction to the Finite Element Method
The stress calculations with several meshes are listed in the following table, along with
the number of elements and DOFs used.
Table. FEA stress results
Elem. Type No. of Elem. Total DOFs Max. σ (psi)
Q4 506 1102 312.42
Q4 3352 7014 322.64
Q4 31349 64106 322.38
... ... ... ...
T6 1518 6254 320.18
T6 2562 10494 321.23
T6 24516 100702 322.24
... ... ... ...
Q8 501 3188 320.58
Q8 2167 13376 321.70
Q8 14333 88636 322.24
The converged results are obtained with all three types of elements with the differences in
the maximum stress values less than 0.05%. However, Q8 and T6 elements are more efficient
and converge much faster than the Q4 elements which is a linear representation and cannot
model curved boundaries accurately. If the required accuracy is set at 1%, then the mesh with
501 Q8 elements should be sufficient. Note also that we need to check the deformed shape of the
plate for each model to make sure the BCs are applied correctly. Less elements should be enough
to achieve the same accuracy with a better or “smarter” mesh (mapped mesh). We will redo this
example in the next chapter employing the symmetry features of the problem.
Further Discussions
(a) Know the behaviors of each type of elements:
T3 and Q4: linear displacement, constant strain and stress;
T6 and Q8: quadratic displacement, linear strain and stress.
(b) Choose the right type of elements for a given problem:
When in doubt, use higher order elements (T6 or Q8) or a finer mesh.
(c) Avoid elements with large aspect ratios and corner angles (Figure 3.14):
Aspect ratio = Lmax / Lmin
© 1997-2013 Yijun Liu, University of Cincinnati 76
Lecture Notes: Introduction to the Finite Element Method
where Lmax and Lmin are the largest and smallest characteristic lengths of an element,
respectively.
Elements with distorted shapes
Elements with normal shapes
Figure 3.14. Elements with distorted (irregular) and normal (regular) shapes.
(d) Make sure the elements are connected properly:
Don not leave unintended gaps or free elements in FE models (Figure 3.15).
A C
B D
Improper connections (gaps along AB and CD)
Figure 3.15. Unintended gaps in the FE mesh.
IV. Summary
2-D elements for analyzing plane stress and plane strain problems are discussed in this
chapter. Linear triangular (T3) and linear quadrilateral (Q4) elements are good for deformation
analysis and not accurate for stress analysis. Quadratic triangular (T6) and quadratic quadrilateral
(Q8) elements are good for stress concentration problems and for models with curved
boundaries. Whenever possible (as allowed by the computing resources), higher-order elements
(T6 or Q8) elements should be applied in FE stress analysis of 2-D structures.
© 1997-2013 Yijun Liu, University of Cincinnati 77
Lecture Notes: Introduction to the Finite Element Method
V. Problems
Problem 1. List the boundary conditions in Example 3.1.
Problem 2. The plate shown below is constrained at the left end and loaded with a linearly
varying pressure load at the right end. Constants E, v and thickness t are given.
y
po
h
x
Suppose we have found the displacement field as follows:
p0
u= xy ,
Eh
p
v = − 0 ( x 2 + νy 2 )
2 Eh
Find:
(a) strains in the plate;
(b) stresses in the plate;
(c) check if or not the equilibrium equations are satisfied by the stresses;
(d) check if or not the boundary conditions are satisfied by the solution.
Optional:
Assume E = 10x106 psi, v = 0.3, po = 100 psi, L = 12 in., h = 4 in. and thickness
t = 0.1 in. Use an FEM software to check your results.
Problem 3. Derive the shape functions in (3.17) for T3 elements and prove (3.18).
Problem 4. From (3.27), prove det J = x13 y 23 − x 23 y13 = 2 A and discuss why “bad shaped”
elements can cause numerical errors in the FEM.
Problem 5. Using a FEM software, design a steel shelf bracket. Some dimensions of the
bracket are fixed as shown in the figure, while others can be changed. The shape
and topology near the lower part of the bracket can also be changed, including
adding additional openings. The goal is to use as less material as possible for the
bracket, while to support the given distributed load p.
© 1997-2013 Yijun Liu, University of Cincinnati 78
Lecture Notes: Introduction to the Finite Element Method
(a) For steel, use E = 200 GPa, v = 0.32, and yield stress σ Y = 250 MPa.
(b) Use a factor of safety = 2.0 for the design.
(c) Report the configuration, dimensions, and total volume of the bracket of your
final design.
p = 250 N/cm2
R = 1 cm
H = 12 cm
Bracket thickness:
t = 0.2 cm
L = 20 cm
Problem 6. Similar to the previous problem, design a steel bracket. Some dimensions of the
bracket are fixed as shown in the figure, while others can be changed. The shape
and topology of the bracket can also be changed. The goal of this design is to use
least material for the bracket, while to support the given loads.
(a) For steel, use E = 200 GPa, v = 0.32, and yield stress σ Y = 250 MPa.
(b) Use a safety factor of 1.5 for the design.
(c) Report the configuration, all dimensions, and the total volume of the bracket
of your final design.
Radius of all three holes: R = 10 mm;
Thickness: max. 3 mm.
60 mm
Bolted to a frame
300 N
60 mm
300 N
200 mm
© 1997-2013 Yijun Liu, University of Cincinnati 79
Lecture Notes: Introduction to the Finite Element Method
Chapter 4. Modeling and Solution Techniques
In this chapter, we discuss several techniques in the modeling and solution process of
using the FEM. Applying these techniques can greatly improve the efficiencies and accuracies of
the finite element analysis.
I. Symmetry
Symmetry features of a structure are the first thing one should look into and explore in
the FE modeling and analysis. The model size can be cut almost in half and the solution
efficiencies can be improved by several times. A structure possesses symmetry if its components
are arranged in a periodic or reflective manner. Types of symmetries are (Figure 4.1):
• Reflective (mirror, bilateral) symmetry
• Axisymmetry
• Rotational (cyclic) symmetry
• Translational symmetry
• Others (or combinations of the above)
(a) (b) (c) (d)
Figure 4.1. Some examples of symmetry: (a) reflective symmetry; (b) axisymmetry;
(c) rotational symmetry; and (d) translational symmetry.
In the FEM, symmetry properties can be applied to
• Reduce the size of the problems and thus save CPU time, disk space, post-processing
efforts, and so on
• Simplify the modeling task
• Check the FEM results (make sure the results are symmetrical if the geometry and
loading of the structure are symmetrical)
Symmetry properties of a structure should be fully exploited and retained in the FE model
to ensure the efficiency and quality of FE solutions.
© 1997-2013 Yijun Liu, University of Cincinnati 80
Lecture Notes: Introduction to the Finite Element Method
An Example
For the problem of a plate with a center hole as discussed in Example 3.2 of the previous
chapter, we redo the FEA mesh using the symmetry features of the plate. To do this, we first
model just one quarter of the plate using mapped mesh, and then reflect the model (with the
mesh) twice to obtain the model and mesh for the entire plate, as shown in Figure 4.2. Only 896
Q8 elements are used in this symmetrical model and the results are comparable to those in
Chapter 3 using more elements with the free mesh. The quarter model can also be applied in the
analysis, if the boundary conditions are also symmetrical about the xz and yz planes.
Figure 4.2. Results using symmetry features for Example 3.2 (mesh and stress contour plots).
In vibration or buckling analysis, however, the symmetry concept should not be used in
the FEA solutions (it is still applicable in the modeling stage), since symmetric structures often
have antisymmetric vibration or buckling modes.
II. Substructures (Superelements)
Another very useful technique for analyzing very large FEA models of mechanical
systems is to apply the concept of substructures or superelements. Substructuring is a process of
analyzing a large structure as a collection of (natural) components. The FEA models for these
components are called substructures or superelements (SE). The physical meaning of a
substructure is simply a finite element model of a portion of the structure. Mathematically, it
presents a boundary matrix which is condensed by eliminating the interior points and keeping
only the exterior or boundary points of the portion of the structure. In other words, instead of
solving the FEA system of equations once, one can use partitions of the matrix so that larger
models can be solved on relatively smaller computers. More details of the theory and
implementations of the substructures or superelements can be found in the documentation of the
FEA software packages (such as ANSYS or Nastran).
Figure 4.3 shows an FEA model of a truck used to conduct the full vehicle static or
dynamic analysis. The entire model can have several millions of DOFs that can be beyond the
© 1997-2013 Yijun Liu, University of Cincinnati 81
Lecture Notes: Introduction to the Finite Element Method
capabilities of some computers. Using the substructuring technique, one can build the FEA
model for each subsystem first (such as the cab, chassis, steering system, suspension system,
payload, and so on) and then condense the FEA equations to smaller ones relating only DOFs on
the interfaces between the subsystems and residing on a residual structure (e.g., the chassis). The
condensed system is much smaller than the original system and can be solved readily.
Figure 4.3. An FEA model of a truck analyzed using substructures.
The advantages of using the substructuring technique are:
• Good for large problems (which will otherwise exceed your computer capabilities)
• Less CPU time per run once the superelements have been processed (i.e., matrices have
been condensed and saved)
• Components may be modeled by different groups
• Partial redesign requires only partial reanalysis (reduced cost)
• Efficient for problems with local nonlinearities (such as confined plastic deformations)
which can be placed in one superelement (residual structure)
• Exact for static deformation and stress analysis
The disadvantages of using the substructuring technique are:
• Increased overhead for file management
• Increased initial time for setting up the system
• Matrix condensations for dynamic problems introduce new approximations
© 1997-2013 Yijun Liu, University of Cincinnati 82
Lecture Notes: Introduction to the Finite Element Method
III. Equation Solving
There are two types of solvers used in the FEA for solving the linear systems of algebraic
equations, mainly, the direct methods and iterative methods.
Direct Methods (Gauss Elimination):
• Solution time proportional to NB2 (with N being the dimension of the matrix, B the
bandwidth of the FEA systems)
• Suitable for small to medium problems (with DOFs in the 100,000 range), or slender
structures (small bandwidth)
• Easy to handle multiple load cases
Iterative Methods:
• Solution time is unknown beforehand
• Reduced storage requirement
• Suitable for large problems, or bulky structures (large bandwidth, converge faster)
• Need to solve the system again for different load cases
An Example - Gauss Elimination:
Solve the following given system of equations:
8 − 2 0 x1 2
− 2 4 − 3 x = − 1 or Ax = b . (4.1)
2
0 − 3 3 x3 3
Forward Elimination:
(1) 8 − 2 0 2
Form (2) − 2 4 − 3 − 1 ; (4.2)
(3) 0 − 3 3 3
(1) + 4 x (2) ⇒ (2):
(1) 8 − 2 0 2
(2) 0 14 − 12 − 2 ; (4.3)
(3) 0 − 3 3 3
14
(2) + (3) ⇒ (3):
3
© 1997-2013 Yijun Liu, University of Cincinnati 83
Lecture Notes: Introduction to the Finite Element Method
(1) 8 − 2 0 2
(2) 0 14 − 12 − 2 ; (4.4)
(3) 0 0 2 12
Back Substitutions (to obtain the solution):
x3 = 12 / 2 = 6 1.5
x2 = (−2 + 12 x3 ) / 14 = 5 or x = 5 . (4.5)
x1 = (2 + 2 x2 ) / 8 = 1.5 6
An Example - Iterative Method:
The Gauss-Seidel Method (as an example):
Ax = b (A is symmetric) (4.6)
N
or ∑a
j =1
ij x j = bi , i = 1, 2, ..., N .
Start with an estimate x ( 0 ) of the solution vector and then iterate using the following:
1 i −1 N
bi − ∑ aij x j ∑a
( k +1) ( k +1) (k )
xi = − ij xj ,
aii j =1 j =i +1 (4.7)
for i = 1, 2, ..., N .
In vector form,
−1
[
x ( k +1) = A D b − A L x ( k +1) − A L x ( k ) ,
T
] (4.8)
where
A D = 〈 aii 〉 is the diagonal matrix of A,
A L is the lower triangular matrix of A,
T
such that A = AD + AL + AL . (4.9)
Iterations continue until solution x converges, i.e.
x ( k +1) − x ( k )
≤ε, (4.10)
x(k )
where ε is the tolerance for convergence control.
Iterative solvers with moderate selections of the tolerance are usually much faster than
direct solvers in solving large-scale models. However, for ill-conditioned systems, direct solvers
should be applied to ensure the accuracy of the solutions.
© 1997-2013 Yijun Liu, University of Cincinnati 84
Lecture Notes: Introduction to the Finite Element Method
IV. Nature of Finite Element Solutions
Some observation of the FEA models and solutions:
• FEA model – A mathematical model of the real structure, based on many approximations
• Real structure -- Infinite number of nodes (physical points or particles), thus infinite
number of DOFs
• FEA model – finite number of nodes, thus finite number of DOFs
In particular, one can argue that the displacement field is controlled (or constrained) by
the values at a limited number of nodes (Figure 4.4).
4
Recall that on an element: u = ∑ Nα uα
α =1
Figure 4.4. Elements in an FEA model
Therefore, we have the so called stiffening effect:
• FEA Model is stiffer than the real structure
• In general, displacement results are smaller in magnitudes than the exact values
Hence, the FEM solution of displacement is a lower bound of the exact solution.
∆ (Displacement)
Exact Solution
FEA Solutions
No. of DOFs
Figure 4.5. Convergence of FEM solutions with exact solution
That is, FEA displacement solutions approach the exact solution from below, which can
be used to monitor the FEA solutions. However, this is true for the displacement based FEA.
© 1997-2013 Yijun Liu, University of Cincinnati 85
Lecture Notes: Introduction to the Finite Element Method
V. Convergence of FEA Solutions
As the mesh in an FEA model is “refined” (with smaller and smaller elements)
repeatedly, the FEA solution will converge to the exact solution of the mathematical model of
the problem (the model based on bar, beam, plane stress/strain, plate, shell, or 3-D elasticity
theories or assumptions). Several types of refinements have been devised in the FEA, which
include:
h-refinement: Reduce the size of the element (“h” refers to the typical size of the
elements)
p-refinement: Increase the order of the polynomials on an element (linear to
quadratic, etc.; “p” refers to the highest order in a polynomial)
r-refinement: Re-arrange the nodes in the mesh
hp-refinement: Combination of the h- and p-refinements (to achieve better results)
With any of the above type of refinements, the FEA solutions will converge to the
analytical solutions of the mathematical models. Some FEA software can automate the process
of refinements in the FEA solutions to achieve the so called adaptive solutions.
VI. Adaptivity (h-, p-, and hp-Methods)
Adaptive FEA represents the future of the FEA applications. With proper error control,
automatic refinements of an FEA mesh can be generated by the program until the converged
FEA solutions are obtained. With the adaptive FEA capability, users’ interactions are reduced, in
the sense that a user only need to provide a good initial mesh for the model (even this step can be
done by the software automatically).
Error estimates are crucial in the adaptive FEA. Interested readers can refer to Ref. [2] for
more details. In the following, we introduce one type of the error estimates.
We first define two stress fields:
σ --- element by element stress field (discontinuous across elements)
σ*--- averaged or smoothed stress field (continuous across elements)
Then, the error stress field can be defined as:
σ E = σ - σ* (4.11)
Compute strain energies,
M
1
U = ∑U i , Ui = ∫ 2σ
T
E −1σdV ; (4.12)
i =1 Vi
M
1
U * = ∑ U i* , U i* = ∫ σ *T E −1σ *dV ; (4.12)
i =1 Vi
2
© 1997-2013 Yijun Liu, University of Cincinnati 86
Lecture Notes: Introduction to the Finite Element Method
M
1
U E = ∑U E i , U E i = ∫ σ TE E −1σ E dV ; (4.14)
i =1 Vi
2
where M is the total number of elements, Vi is the volume of the element i.
One error indicator - the relative energy error, is defined as:
1/ 2
UE
η= . (0 ≤ η ≤ 1) (4.15)
U + U E
The indicator η is computed after each FEA solution. Refinement of the FEA model continues
until, say
η ≤ 0.05.
When this condition is satisfied, we conclude that the converged FE solution is obtained.
Some examples of using different error estimates in the FEA solutions can be found in
Ref. [2].
VII. Summary
In this chapter, we briefly discussed a few modeling techniques and concepts related to
FEA solutions. For symmetrical structures, the symmetry features should be explored in both
modeling/meshing stage and solution stage (if the BCs are symmetrical as well). Substructuring
or using superelements is a useful technique for solving large-scale problems with constrained
computing resources. Convergence of the FEA solutions is the important goal in FEA and should
be monitored by using the error estimates and employing the adaptive FEA capabilities in the
software.
VIII. Problems
Problem 1. Suppose that we need to find out the in-plane effective modulus of a composite
reinforced with long fibers aligned in the z-direction and distributed uniformly. A
2-D elasticity model shown below can be used for this study with the FEA. The
effective modulus can be estimated by using the formula E eff = σ x ( ave ) / ε x ( ave ) ,
where the averaged stress and strain are evaluated along the vertical edge on the
right side of the model. Assume for the matrix E = 10 GPa, v = 0.35, and for the
fibers E = 100 GPa, v = 0.3. The unit cell has a dimension of 1x1 μm2, and the
radius of the fibers is 0.2 μm.
Start with 1x1 cell, 2x2 cells, 3x3 cells, … and keep increasing the number of the
cells as you can. Report the value of the effective Young’s modulus of the
© 1997-2013 Yijun Liu, University of Cincinnati 87
Lecture Notes: Introduction to the Finite Element Method
composite in the x and y direction. Employ symmetry features of the model in
generating the meshes for your analysis.
One unit cell
Problem 2. Suppose that a “meshed panel” will be used in a design in order to reduce the
weight. For this purpose, we need to find out the in-plane effective modulus of
this panel in the x- or y-direction. A sample piece of the panel similar to the one
shown below can be used for this study. Employ symmetry and study the effects
of the numbers of cells used in the model on the computed moduli.
Assume the panel is made of aluminum with E = 70 GPa, v = 0.35, a = 10 mm,
b = 6 mm, c = 1.5 mm, and thickness t = 1 mm.
y a b c
One cell
© 1997-2013 Yijun Liu, University of Cincinnati 88
Lecture Notes: Introduction to the Finite Element Method
Chapter 5. Plate and Shell Elements
Many structure members can be categorized as plates and shells [11], which are
extensions of the 1-D straight beams and curved beams to 2-D cases, respectively. Some of the
structures that can be modeled as plates are:
• Shear walls
• Floor panels
• Shelves
While those that can be modeled as shells include:
• Sea shells and egg shells (the wonder of the nature)
• Various containers, pipes, and tanks
• Bodies of cars, boats, aircraft, etc.
• Roofs of buildings (the Superdome), etc.
Figure 5.1 shows two recent engineering wonders that are constructed mainly using plate
and shell structure members.
(a) The new Boeing 787 aircraft (b) The National Grand Theatre in Beijing
Figure 5.1. Examples of plate and shell structures.
The advantages of using plate and shell structures are their light weight, superior load-
carrying capabilities, and sometimes, simply their artistic appeals.
I. Plate Theory
A plate has the following characteristics:
• A flat surface
• Applied with lateral loading
• Bending behavior dominates
© 1997-2013 Yijun Liu, University of Cincinnati 89
Lecture Notes: Introduction to the Finite Element Method
Forces and Moments Acting on the Plate
∆y
y
∆x My
q(x,y)
Qy Mxy
t Mx
Qx
x Mid surface
Mxy
Figure 5.2. Forces and moments acting on an infinitesimally small element in a plate.
Stresses in the Plate
τyz y
σy
τxy
τxz τxy
σx
x
Figure 5.3. Stresses acting on the infinitesimally small element in the plate.
© 1997-2013 Yijun Liu, University of Cincinnati 90
Lecture Notes: Introduction to the Finite Element Method
Relations Between the Forces and Stresses:
Bending moments (per unit length):
t/2
Mx = ∫ σ x zdz , ( N ⋅ m / m) (5.1)
−t / 2
t/2
My = ∫ σ y zdz , ( N ⋅ m / m) (5.2)
−t / 2
Twisting moment (per unit length):
t/2
M xy = ∫ τ xy zdz , ( N ⋅ m / m) (5.3)
−t / 2
Shear Forces (per unit length):
t/2
Qx = ∫ τ xz dz , ( N / m) (5.4)
−t / 2
t/2
Qy = ∫ τ yz dz , ( N / m) (5.5)
−t / 2
Maximum bending stresses:
6M x 6M y
(σ x ) max = ± , (σ y ) max = ± . (5.6)
t2 t2
Note that:
• Maximum stress is always at z = ±t / 2
• No bending stresses at midsurface (similar to the beam model)
© 1997-2013 Yijun Liu, University of Cincinnati 91
Lecture Notes: Introduction to the Finite Element Method
Thin Plate Theory (Kirchhoff Plate Theory)
Assumptions (similar to those for the simple beam theory):
A straight line along the normal to the mid surface remains straight and normal to the
deflected mid surface after loading, that is, there is no transverse shear deformation (Figure 5.4):
γ xz = γ yz = 0 .
∂w
z ∂x
w
x
Figure 5.4. Deflection and rotation after loading of a plate according to Kirchhoff plate theory.
Displacement:
w = w( x, y ), (deflection)
∂w
u = −z , (5.7)
∂x
∂w
v = −z .
∂y
Strains:
∂ 2w
ε x = −z 2 ,
∂x
∂ 2w
ε y = −z 2 , (5.8)
∂y
∂ 2w
γ xy = −2 z .
∂x∂y
Note that there is no stretch of the mid surface due to the deflection of the plate.
© 1997-2013 Yijun Liu, University of Cincinnati 92
Lecture Notes: Introduction to the Finite Element Method
Stresses (plane stress state):
σ x 1 ν 0 ε x
E ε ,
σ y = ν 1 0 y (5.9)
τ 1 − ν
2
xy 0 0 (1 − ν ) / 2 γ xy
or,
∂2w
2
σ x 1 ν 0 ∂2x
E ∂ w
σ y = − z 2
ν 1 0 2 . (5.10)
τ 1 −ν ∂2y
0 0 (1 − ν )
xy
∂ w
∂x∂y
Note the main variable: deflection w = w( x, y ) .
Governing Equation:
D∇ 4 w = q( x, y ) , (5.11)
where
∂4 ∂4 ∂4
∇ ≡ ( 4 + 2 2 2 + 4 ),
4
∂x ∂x ∂y ∂y
Et 3
D= (the bending rigidity of the plate), (5.12)
12(1 − ν 2 )
q = lateral distributed load (force/area).
Compare the 1-D equation for straight beam:
d 4w
EI = q( x) . (5.13)
dx 4
Note: Equation (5.11) represents the equilibrium condition in the z-direction. To see
this, refer to the Figure 5.2. showing all the forces on a plate element. Summing the forces in the
z-direction, we have,
Qx ∆y + Qy ∆x + q∆x∆y = 0, (5.14)
© 1997-2013 Yijun Liu, University of Cincinnati 93
Lecture Notes: Introduction to the Finite Element Method
which yields,
∂Qx ∂Qy
+ + q ( x, y ) = 0 . (5.15)
∂x ∂y
Substituting the following relations into the above equation, we obtain Eq. (5.11):
Shear forces and bending moments:
∂M x ∂M xy ∂M xy ∂M y
Qx = + , Qy = + , (5.16)
∂x ∂y ∂x ∂y
∂2w ∂2w ∂2w ∂2w
M x = D 2 + ν 2 , M y = D 2 + ν 2 . (5.17)
∂x ∂y ∂y ∂x
The fourth-order partial differential equation, given in (5.11) and in terms of the
deflection w(x,y), needs to be solved under certain given boundary conditions.
Boundary Conditions:
∂w
Clamped: w = 0, = 0; (5.18)
∂n
Simply supported: w = 0, M n = 0 ; (5.19)
Free: Qn = 0, Mn = 0; (5.20)
where n is the normal direction of the boundary (Figure 5.5). Note that the given values in the
boundary conditions shown above can be non-zero values as well.
s n
boundary
Figure 5.5. The boundary of a plate.
© 1997-2013 Yijun Liu, University of Cincinnati 94
Lecture Notes: Introduction to the Finite Element Method
Examples:
A square plate (Figure 5.6) with four edges clamped or hinged, and under a uniform load
q or a concentrated force P at the center C.
y
C
L
L
x Given: E, t, and ν = 0.3
Figure 5.6. A square plate.
For this simple geometry, Eq. (5.11) with boundary condition (5.18) or (5.19) can be
solved analytically. The maximum deflections are given in the Table 5.1 for the four different
cases.
Table 5.1. Deflection at the Center (wc)
Clamped Simply supported
Under uniform load q 0.00126 qL4/D 0.00406 qL4/D
Under concentrated force P 0.00560 PL2/D 0.0116 PL2/D
in which: D= Et3/(12(1-v2)).
These values can be used to verify the FEA solutions.
Thick Plate Theory (Mindlin Plate Theory)
If the thickness t of a plate is not “thin”, for example, when t / L ≥ 1 / 10 (L = a
characteristic dimension of the plate main surface), then the thick plate theory by Mindlin should
be applied. This theory accounts for the angle changes within a cross section, that is,
γ xz ≠ 0, γ yz ≠ 0 .
© 1997-2013 Yijun Liu, University of Cincinnati 95
Lecture Notes: Introduction to the Finite Element Method
This means that a line which is normal to the mid surface before the deformation will not be so
after the deformation (Figure 5.7).
∂w
θy ≠ −
z ∂x
w
∂w x
∂x
Figure 5.7. Displacement and rotation based on the Mindlin thick plate theory.
New independent variables:
θ x and θ y : rotation angles of a line, which is normal to the mid surface before the
deformation, about x- and y-axis, respectively.
New relations:
u = zθ y , v = − zθ x ; (5.21)
and
∂θ y ∂θ ∂θ y ∂θ x
εx = z , ε y = − z x , γ xy = z ( − ),
∂x ∂y ∂y ∂x
(5.22)
∂w ∂w
γ xz = + θ y , γ yz = −θ .
∂x ∂y x
Note that if we imposed the conditions (or assumptions) that
∂w ∂w
γ xz = + θ y = 0, γ yz = − θ x = 0, (5.23)
∂x ∂y
then we can recover the relations applied in the thin plate theory.
Main variables are: w( x, y ), θ x ( x, y ) and θ y ( x, y ) .
The governing equations and boundary conditions can be established for thick plates
based on the above assumptions.
© 1997-2013 Yijun Liu, University of Cincinnati 96
Lecture Notes: Introduction to the Finite Element Method
II. Plate Elements
Kirchhoff Plate Elements:
A 4-Node Quadrilateral Element:
z y
Mid surface 4
3
1 2
t
∂w ∂w ∂w ∂w
w1 , , w2 , ,
∂x 1 ∂y 1 ∂x 2 ∂y 2
Figure 5.8. A 4-node quadrilateral element with 3 DOFs at each node.
∂w ∂w
DOFs at each node: w, , .
∂x ∂y
On each element, the deflection w(x,y) is represented by
4
∂w ∂w
w( x, y ) = ∑ N i wi + N xi ( )i + N yi ( )i , (5.24)
i =1 ∂x ∂y
where Ni, Nxi and Nyi are shape functions. This is an incompatible element [4]. The stiffness
matrix is still of the form
k = ∫ BT EBdV , (5.25)
V
where B is the strain-displacement matrix, and E the Young’s modulus (stress-strain) matrix.
© 1997-2013 Yijun Liu, University of Cincinnati 97
Lecture Notes: Introduction to the Finite Element Method
Mindlin Plate Elements:
Two Quadrilateral Elements:
z y z y
4 4 7 3
3
8 6
x x
1 2 1
t 5 2
t
(a) 4-node quadrilateral (b) 8-node quadrilateral
Figure 5.9. 4-node and 8-node quadrilateral plate elements.
DOFs at each node: w, θx and θy.
On each element, the displacement and rotations are represented by:
n
w( x, y ) = ∑ N i wi ,
i =1
n
θ x ( x, y ) = ∑ N iθ xi , (5.26)
i =1
n
θ y ( x, y ) = ∑ N iθ yi .
i =1
For these elements:
• There are three independent fields within each element.
• Deflection w(x, y) is linear for Q4, and quadratic for Q8.
© 1997-2013 Yijun Liu, University of Cincinnati 98
Lecture Notes: Introduction to the Finite Element Method
Discrete Kirchhoff Element:
This is a triangular element. First, start with a 6-node triangular element (Figure 5.10),
z y 3
4 6
1 2
t 5 x
Figure 5.10. A 6-node triangular element with 5 DOFs at each corner node and 2 DOFs at each
mid node.
∂w ∂w
DOFs at corner nodes: w, , ,θ ,θ ;
∂x ∂y x y
DOFs at mid side nodes: θx , θ y .
Total DOFs = 21.
Then, impose conditions γ xz = γ yz = 0 , etc., at selected nodes to reduce the DOFs (using
relations in Eq. (5.22)), to obtain the discrete Kirchhoff triangular (DKT) element (Figure 5.11):
z y 3
1 2
x
Figure 5.11. Discrete Kirchhoff triangular element with 3 DOFs at each node.
∂w ∂w
DOFs at each node: w,θ x = , θ y = .
∂x ∂y
Total DOFs = 9 (DKT element).
Note that w(x, y) is incompatible for DKT elements [4]; however, its convergence is
faster (w is cubic along each edge) and it is efficient.
© 1997-2013 Yijun Liu, University of Cincinnati 99
Lecture Notes: Introduction to the Finite Element Method
Test Problem:
We consider a square plate with its four edges clamped and a concentrated force P
applied at the center (Figure 5.12). Using 4-node plate elements, we obtain results in Table 5.2.
y
C
L
L
x L/t = 10, ν = 0.3
Figure 5.12. A clamped square plate with a concentrated load P.
Table 5.2. ANSYS result for deflection wc.
Number of Elements wc (× PL2/D)
2×2 0.00593
4×4 0.00598
8×8 0.00574
16×16 0.00565
: :
Exact Solution 0.00560
Questions: Why results converge from “above”? Contradiction to what we learnt
about the nature of the FEA solution?
Reason: This is an incompatible element (See comments on page 177 of Cook’s
textbook [4]).
© 1997-2013 Yijun Liu, University of Cincinnati 100
Lecture Notes: Introduction to the Finite Element Method
III. Shells and Shell Elements
Shells are thin structure members which span over curved surfaces. The thickness t of a
shell is usually much smaller than the other dimensions of the shell and thus it can be represented
mathematically by a 2-D surface in space, with the thickness as a parameter (Figure 5.13).
Figure 5.13. A shell structure member and its mathematical representation.
Forces in Shells:
Unlike the plate models, there are two types of forces in shells, that is:
Membrane forces (in plane forces) + Bending forces (out of plane forces)
(cf. plates: bending forces only)
Figure 5.14. Forces and moments in a shell structure member.
© 1997-2013 Yijun Liu, University of Cincinnati 101
Lecture Notes: Introduction to the Finite Element Method
Example: A Cylindrical Container
p p
internal forces:
p
p
membrane stresses
dominate
Figure 5.15. Internal forces acting in a cylinder under internal pressure load.
Shell Theories:
Similar to the plate theories, there are two types of theories for modeling shells,
according to the thickness of the shells to be studied:
• Thin shell theory
• Thick shell theory
Shell theories are the most complicated ones to formulate and analyze in mechanics.
Many of the contributions were made by Russian scientists in the 1940s and 1950s, due to the
need to develop new aircraft and other light-weight structures. Interested readers can refer to
Ref. [11] for in-depth studies on this subject. These theoretical work have laid the foundations
for the development of various finite elements for analyzing shell structures.
© 1997-2013 Yijun Liu, University of Cincinnati 102
Lecture Notes: Introduction to the Finite Element Method
Flat Shell Elements:
plane stress element plate bending element
flat shell element
Figure 5.16. Combination of plane stress element and plate bending element yields a flat shell
element.
cf.: bar + simple beam element => general beam element (for modeling curved beams).
DOFs at each node:
w
v
u θx
θy
Figure 5.17. Q4 or Q8 shell elements.
© 1997-2013 Yijun Liu, University of Cincinnati 103
Lecture Notes: Introduction to the Finite Element Method
Curved Shell Elements:
θz
i
w
v
i u θx
θy
Figure 5.18. A 8-node curved shell element and the DOFs at a typical node i.
• Curved shell elements are based on the various shell theories;
• They are the most general shell elements (flat shell and plate elements are subsets);
• Complicated in formulation.
Test Cases:
L L/2
L/2
q F
A
R A
80o R
F
(a) Roof (b) Pinched Cylinder
F2
F R F
b
A
A F L
F1
F
(c) Pinched Hemisphere (d) Twisted Strip (90o)
Figure 5.19. Four test cases where analytical solutions are available.
© 1997-2013 Yijun Liu, University of Cincinnati 104
Lecture Notes: Introduction to the Finite Element Method
For the four cases shown in Figure 5.19, check the table on page 188 of Cook’s textbook
[4] for the exact values of displacement ∆A under various loading conditions, which can be used
to verify the FEA results in shell analysis.
Figure 5.20 is a stamping part analyzed using shell elements. The bracket has a uniform
thickness and is fixed at the four bolt hole positions. A load is applied through a pin passing
through the two holes in the lower part of the bracket. Note the one layer of elements on the edge
of each hole (Figure 5.20 (a)), which is a common practice to models holes. Note also that this
layer of elements on the edge of each hole has been masked in the stress contour plot
(Figure 5.20 (b)), due to inaccurate stress results near the constraint locations. To reduce the true
stress levels in the bracket, the thickness can be changed, the shape of the bracket can be
modified, and the model is re-meshed and re-analyzed, all of which are very easy to carry out
with the shell elements.
(a) (b)
Figure 5.20. Stress analysis of a bracket using shell elements: (a) The FEA model; (b) Stress
contour plot.
Cautions in Applying Shell Elements:
In many cases, however, the plate and shell models may not be adequate for analyzing a
structure member, even if it is considered thin. For example, the structure component has a
nonuniform thickness (turbine blades, vessels with stiffeners, thin layered structures, etc.), see
Figure 5.21, or has a crack for which detailed stress analysis is needed. In such cases, one should
turn to 3-D elasticity theory and apply solid elements which will be discussed in the next chapter.
© 1997-2013 Yijun Liu, University of Cincinnati 105
Lecture Notes: Introduction to the Finite Element Method
(a) (b)
Figure 5.21. Cases in which shell elements are not adequate: (a) Casting parts; (b) Parts with
nonuniform thickness. 3-D solid elements should be applied in such cases.
IV. Summary
In this chapter, we discussed the main aspects of the plate and shell theories and the plate
and shell elements used for analyzing plate and shell structures. Plates and shells can be regarded
as the extensions of the beam elements from 1-D line elements to 2-D surface elements. Plates
are usually applied in modeling flat thin structure members, while the shells in modeling curved
thin structure members. In applying the plate and shell elements, one should keep in mind the
assumptions used in the development of these types of elements. In cases where these
assumptions are no longer valid, one should turn to general 3-D theories and solid elements.
V. Problems
Problem 1. The roof structure shown in Figure 5.19 (a) is loaded by its own weight with q =
90 lbf/in2. The dimensions and material constants are: R = 25 in., L = 50 in., t =
0.25 in., E = 432×106 psi and v = 0.0. The two straight edges are free, while the
two curved edges have a “diaphragm” support (meaning that y and z DOFs are
constrained, but x (along the axis) and all rotational DOFs are unconstrained).
Use shell elements to find the maximum displacement and von Mises stress in the
structure. Verify your results (Note that the value of the analytical solution for the
displacement at the mid point A of the straight edge is 0.3024 in).
© 1997-2013 Yijun Liu, University of Cincinnati 106
Lecture Notes: Introduction to the Finite Element Method
Problem 2. Consider a glass cup placed on a table, as shown in the figure. Using an FEA
software package and 8-node shell elements, find the maximum displacement and
von Mises stress in the cup when the cup is applied with a pressure load of 10
N/mm2 on the inner wall. Assume that the cup has a uniform thickness, E =
70 GPa and v = 0.17.
40 mm
150 mm
Thickness = 4 mm
30 mm
Problem 3. A fuel tank, with a total length = 5 m, diameter = 1 m, and thickness = 0.01 m, is
shown below. Using the FEA, find the deformation and stresses when the tank is
applied with an internal pressure p = 100 MPa and placed on the ground.
Assume the Young’s modulus E = 200 GPa and Poisson’s ratio υ = 0.3.
© 1997-2013 Yijun Liu, University of Cincinnati 107
Lecture Notes: Introduction to the Finite Element Method
Chapter 6. Three-Dimensional Elasticity Problems
Solid elements based on 3-D elasticity theory [9, 10] are the most general elements for
stress analysis when the simplified bar, beam, plane stress/strain, plate/shell elements are no
longer valid or accurate. In this chapter, we first review the elasticity equations in 3-D and then
discuss a few types of 3-D finite elements commonly used for 3-D stress analysis.
I. 3-D Elasticity Theory
Stress State:
There are six stress components at each point in a 3-D elastic body (Figure 6.1).
y
z
y,v
σy
τ yx
τ yz
τ xy
τ zy
σx
τ zx
τ xz x, u
σz
z, w
Figure 6.1. State of stress in a 3-D element.
© 1997-2013 Yijun Liu, University of Cincinnati 108
Lecture Notes: Introduction to the Finite Element Method
The six stress components shown in Figure 6.1 can be wriiten as a vector:
σx
σ
y
σ
σ ={ σ } = z , or [ σ ]. (6.1)
τ xy
ij
τ yz
τ zx
Strains:
Similarly, the six strain components in 3-D can be expressed as:
ε x
ε y
ε ,
ε = { ε }=
γ
z
or [ ε ]. ij (6.2)
γ
xy
yz
γ zx
Stress-Strain Relations:
The stress-strain relation in 3-D is given by:
1 − v v v 0 0 0
v 1− v v 0 0 0 ε
σx x
σ v v 1− v 0 0 0 ε
y y
σ z 1 − 2v (6.3)
E 0 0 0 0 0 ε z .
= 2 γ
τ xy (1 + v)(1 − 2v) xy
τ yz 1 − 2v
0 0 0 0 0 γ yz
2
τ zx γ
1 − 2v zx
0 0 0 0 0
2
Or in a matrix form:
σ = Eε .
Displacements:
The displacement field can be described as:
u ( x, y, z ) u1
u = v ( x, y , z ) = u 2 . (6.4)
w( x, y, z ) u
3
© 1997-2013 Yijun Liu, University of Cincinnati 109
Lecture Notes: Introduction to the Finite Element Method
Strain-Displacement Relations:
Strain field is related to the displacement field as given below:
∂u ∂v ∂w
εx = , εy = , εz = ,
∂x ∂y ∂z
(6.5)
∂v ∂u ∂w ∂v ∂u ∂w
γ xy = + , γ yz = + , γ xz = + .
∂x ∂y ∂y ∂z ∂z ∂x
These six equations can be written in the following index or tensor form:
ε ij =
(
1 ∂ui ∂u j
+
2 ∂x j ∂xi ) , i, j = 1, 2,3 .
Or simply,
1
ε ij =
2
( ui, j + u j ,i ). (tensor notation)
Equilibrium Equations:
The stresses and body force vector f at each point satisfy the following three equilibrium
equations for elastostatic problems:
∂σ x ∂τ xy ∂τ xz
+ + + f x = 0,
∂x ∂y ∂z
∂τ yx ∂σ y ∂τ yz
+ + + f y = 0, (6.6)
∂x ∂y ∂z
∂τ zx ∂τ zy ∂σ z
+ + + f z = 0.
∂x ∂y ∂z
Or in index or tensor notation:
σ ij , j + fi = 0 .
Boundary Conditions (BCs):
At each point on the boundary Γ and in each direction, either displacement or traction
(stress on the boundary) should be given, that is:
ui = ui , on Γu ( specified displacement );
(6.7)
ti = ti , on Γσ ( specified traction );
in which the barred quantities denote given values, and the traction (stress on a surface) is
defined by ti = σ ij n j , or in a matrix form:
© 1997-2013 Yijun Liu, University of Cincinnati 110
Lecture Notes: Introduction to the Finite Element Method
t x σ x τ xy τ xz nx
t y = τ xy σ y τ yz n y .
t τ
z xz τ yz σ z nz
with n being the normal (Figure 6.2).
p
Γσ
n
Γ ( = Γ u + Γσ )
Γu
Figure 6.2. The boundary of an elastic domain.
Stress Analysis:
For 3-D stress analysis, one needs to solve equations in (6.3), (6.5) and (6.6) under the
BCs in (6.7) in order to obtain the stress, strain and displacement fields (15 equations for 15
unknowns for 3-D problems). Analytical solutions are often difficult to find and thus numerical
methods such as the FEA is applied in 3-D stress analysis.
II. Finite Element Formulation
We first summarize the FEA formulation for 3-D elasticity problems, which are straight
forward extensions of the FEA formulations for 1-D bar and 2-D elasticity problems.
Displacement Field:
As in the FEA formulations for 1-D and 2-D problems, we first interpolate the
displacement fields within a 3-D element using shape functions Ni:
N
u= ∑N u ,
i =1
i i
N
v= ∑N
i =1
i vi , (6.8)
N
w= ∑N
i =1
i wi ,
in which ui , vi , and wi are nodal values of the displacement on the element, and N is the number
of nodes on that element. In matrix form, we have:
© 1997-2013 Yijun Liu, University of Cincinnati 111
Lecture Notes: Introduction to the Finite Element Method
u
1
v
1
u
N1 0 0 N2 0 0 w
1
v = 0 N1 0 0 N2 0 u
2
(6.9)
w
(3×1) 0 0 N1 0 0 N2 (3×3 N ) v
2
w
2
(3N ×1)
Or in a matrix form:
u=Nd .
Using relations (6.5) and (6.8), we can derive the strain vector to obtain:
ε =B d
in which B is the matrix relating the nodal displacement vector d to the strain vector ε. Note that
the dimensions of the B matrix are 6×3N.
Stiffness Matrix:
Once the B matrix is found, one can apply the following familiar expression to determine
the stiffness matrix for the element:
k = ∫ BT E B dv. (6.10)
v
The dimensions of the stiffness matrix k are 3N×3N. A numerical quadrature is often needed to
evaluate the above integration, which can be expensive if the number of nodes is large, such as
for higher-order elements.
A Note of the Rigid-Body Motions:
Note that there are six rigid-body motions for 3-D bodies:
3 translations and 3 rotations.
These rigid-body motions (causes of singularity of the system of equations) must be
removed from the FEA model for stress analysis to ensure the accuracy of the analysis. On the
other hand, over constrains can also cause inaccurate or unwarranted results.
© 1997-2013 Yijun Liu, University of Cincinnati 112
Lecture Notes: Introduction to the Finite Element Method
III. Typical 3-D Solid Elements
We can classify the type of elements for 3-D problems as follows (Figure 6.3)according
to their shapes and the orders of the shape functions constructed on the elements:
Tetrahedron:
linear (4 nodes) quadratic (10 nodes)
Hexahedron (brick):
linear (8 nodes) quadratic (20 nodes)
Penta:
linear (6 nodes) quadratic (15 nodes)
Figure 6.3. Different types of 3-D solid elements.
Whenever possible, try to apply higher-order (quadratic) elements, such as 10-node
tetrahedron and 20-node brick elements for 3-D stress analysis. Avoid using the linear, especially
the 4-node tetrahedron elements in 3-D stress analysis, because they are inaccurate for such
purposes. However, it is fine to use them for deformation analysis or in vibration analysis (see
next chapter).
In the following section, we will examine the element formulation for the 8-node brick
element.
© 1997-2013 Yijun Liu, University of Cincinnati 113
Lecture Notes: Introduction to the Finite Element Method
Element Formulation:
Linear Hexahedron Element
3
4
y 2
8 1
7
5 6
x
z
mapping (xyz↔ξηζ)
(-1≤ ξ,η,ζ ≤ 1)
η
(-1,1,-1) 4 3 (1,1,-1)
(-1,1,1) 8 7 (1,1,1)
o
ξ
(-1,-1,-1) 1 2 (1,-1,-1)
(-1,-1,1) 5 6 (1,-1,1)
ζ
Figure 6.4. Mapping an element to the natural coordinate system.
Displacement Field in the Element:
8 8 8
u = ∑ N i ui , v = ∑ N i vi , w = ∑ N i wi . (6.11)
i =1 i =1 i =1
Shape Functions:
© 1997-2013 Yijun Liu, University of Cincinnati 114
Lecture Notes: Introduction to the Finite Element Method
1
N1 (ξ ,η , ζ ) = (1 − ξ ) (1 − η ) (1 − ζ ) ,
8
1
N 2 (ξ ,η , ζ ) = (1 + ξ ) (1 − η ) (1 − ζ ) ,
8
1
N 3 (ξ ,η , ζ ) = (1 + ξ ) (1 + η ) (1 − ζ ) , (6.12)
8
1
N 8 (ξ ,η , ζ ) = (1 − ξ ) (1 + η ) (1 + ζ ) .
8
Note that we have the following relations for the shape functions:
N i ( ξ j ,η j , ζ j ) = δ ij , i, j =1,2,, 8.
8
∑ N ( ξ ,η ,ζ ) =1.
i =1
i
Coordinate Transformation (Mapping):
8 8 8
x = ∑ N i xi , y = ∑ N i yi , z = ∑ N i zi . (6.13)
i =1 i =1 i =1
That is, the same shape functions are used as for the displacement field. This is again called an
isoparametric element.
Jacobian Matrix:
∂u ∂x ∂y ∂z ∂u
∂ξ ∂ξ ∂ξ ∂ξ ∂x
∂u ∂x ∂y ∂z ∂u
= ∂η .
∂η ∂η ∂η ∂y (6.14)
∂u ∂x ∂y ∂z ∂u
∂ζ ∂ζ ∂ζ ∂ζ ∂z
≡ J Jacobian matrix
Inverting this relation, we have:
∂u ∂u
∂x ∂ξ
∂u −1 ∂u ∂u ∂N i
8
= J , with = ∑ ∂ξ ui and so on. (6.15)
∂y ∂η ∂ξ i =1
∂u ∂u
∂z ∂ζ
and similarly for v and w. These relations lead to the following expression for the strain:
© 1997-2013 Yijun Liu, University of Cincinnati 115
Lecture Notes: Introduction to the Finite Element Method
∂u
∂x
∂v
ε x ∂y
ε
y ∂w
ε ∂z
ε= z = = use (6.15) = B d .
γ
+
xy
∂v ∂u
γ yz ∂x ∂y
∂w ∂v
γ zx +
∂y ∂z
∂u ∂w
+
∂z ∂x
where d is the nodal displacement vector, that is:
ε = B d. (6.16)
Strain energy is evaluated as:
1 T 1
U= ∫
2V
σ ε dV = ∫ (Eε)T ε dV
2V
1 T
2 V∫
= ε Eε dV (6.17)
1 T T
= d ∫ B E B dV d.
2 V
That is, the element stiffness matrix is
k = ∫ BT E B dV . (6.18)
V
In ξηζ coordinates:
dV = (det J ) d ξ dη d ζ (6.19)
Therefore,
1 1 1
k=∫ ∫ ∫ B E B (det J ) dξ dη dζ .
T
(6.20)
−1 −1 −1
It is easy to verify that the dimensions of this stiffness matrix is 24x24.
Note that in general, 3-D elements do not use rotational DOFs.
© 1997-2013 Yijun Liu, University of Cincinnati 116
Lecture Notes: Introduction to the Finite Element Method
Treatment of Distributed Loads:
Distributed loads need to be converted into nodal forces using the equivalent energy
concept as discussed in earlier chapters. Figure 6.5 shows the result of a pressure load converted
to nodal forces for a 20-node hexahedron element. Note the direction of the forces at the four
corner nodes, which is not intuitive at all.
pA/3 pA/12
p
Surface Area =A Nodal forces for 20-node brick element
Figure 6.5. Equivalent nodal forces on a 20-node brick element for a constant distributed load p.
Stresses:
To compute the the stresses within an element, one uses the following relation once the
nodal displacement vector is known for that element:
σ =Eε = EBd .
Stresses are evaluated at selected points (Gaussian points or nodes) on each element.
Stress values at the nodes are often discontinuous and less accurate. Averaging of the stresses
from surrounding elements around a node are often employed to smooth the stress field results.
The von Mises stress for 3-D problems is given by:
1
σ e = σ VM = (σ 1 − σ 2 ) 2 + (σ 2 − σ 3 ) 2 + (σ 3 − σ 1 ) 2 . (6.21)
2
where σ 1 , σ 2 and σ 3 are the three principal stresses.
3-D stress analysis using solid elements is one of the most difficult tasks in FEA.
Meshing structures with complicated geometries can be very tedious and time consuming. Great
care need to be taken to make sure that the FEA mesh is in good quality (for example, with no
distorted elements). Computing cost is another factor. For structures with stress concentrations,
large FEA models are often needed, which can run hours or days to solve even on today’s best
computers. A good CAE engineer should be able to decided where to apply a fine mesh and
where not to, in order to trike a balance between the cost and accuracy for an FEA task.
© 1997-2013 Yijun Liu, University of Cincinnati 117
Lecture Notes: Introduction to the Finite Element Method
Examples:
Figure 6.6 show a drag link FEA model using solid elements. Although the structure has
a slender shape, it has a bended angle and holes. 3-D solid elements are needed for the stress
analysis in this case. Great care is taken in meshing this part, where 20-node brick elements are
used for better accuracy in the stress analysis. Buckling analysis may also be conducted for
slender structures when they are under compressions. More information about buckling analysis
using FEA can be found in the references or in documents of FEA software packages.
(a) (b)
(c)
Figure 6.6. FEA for a drag link: (a) The model; (b) Mesh for the right end; (c) Stress distribution
due to tension loads applied at the two ends.
© 1997-2013 Yijun Liu, University of Cincinnati 118
Lecture Notes: Introduction to the Finite Element Method
Figure 6.7 show a 3-D FEA of a gear coupling which is applied to transmit powers
through two aligned rotating shafts. Contact stresses and failure modes are to be determined
based on detailed 3-D FE models. This analysis requires the use of nonlinear FEA options, which
are readily available now in almost all FEA software packages.
(a) (b)
(c)
Figure 6.7. Analysis of a gear coupling: (a) Ring gear; (b) Hub gear; (c) High contact stresses in
the gear teeth obtained using nonlinear FEA.
© 1997-2013 Yijun Liu, University of Cincinnati 119
Lecture Notes: Introduction to the Finite Element Method
IV. Solids of Revolution (Axisymmetric Analysis)
Many mechanical parts can be considered as solids of revolution or axisymmetric solids
(Figure 6.8). If the applied loads on these axisymmetric parts are also axisymmetric, the 3-D
analysis can be simplified by using the axisymmetric models.
(a) A baseball bat (b) A shaft
Figure 6.8. Examples of axisymmetric solids.
Cylindrical Coordinates:
Axisymmetric models are based on the cylindrical coordinate (r, θ, z) (Figure 6.9).
z, w
θ
r, u
(a) (b)
z, w
σz
θ
r, u
τ rz
r σθ σr
(c) (d)
Figure 6.9. (a) An axisymmetric body; (b) The cylindrical coordinates; (c) An axisymmetric
model; (d) Stress components in the cylindrical coordinates.
© 1997-2013 Yijun Liu, University of Cincinnati 120
Lecture Notes: Introduction to the Finite Element Method
In this cylindrical coordinates, one can establish the strain-displacement and stress-strain
relations, and the equilibrium equations.
Displacement Field:
u = u (r , z ), w = w(r , z ), v=0 ( No circumferential component ) .
Strain-Displacement Relation (Figure 6.10):
∂u u ∂w
εr = , εθ = , εz = ,
∂r r ∂z
(6.22)
∂w ∂u
γ rz = + , γ rθ = γ zθ = 0.
∂r ∂z
u
(r+u)dθ
dθ rdθ
Figure 6.10. The geometric relations used in deriving strain-displacement relations.
Stresses-Strain Relation:
σ r 1 − v v v 0 ε r
σ v 1− v v 0 ε
θ E θ
= v v 1− v 0 (6.23)
σ z (1 + v) (1 − 2v) 1 − 2v ε z
τ rz 0 0 0 γ rz
2
Axisymmetric Elements (Figure 6.11):
z z
η
2 2 2
r, u 3
r, u
ξ
3 3
4
1
1 1
A 3-node element (ring) A 4-node element (ring)
Figure 6.11. Axisymmetric 3-node and 4-node elements.
© 1997-2013 Yijun Liu, University of Cincinnati 121
Lecture Notes: Introduction to the Finite Element Method
Formulation of the axisymmetric elements are similar to other 2-D plane stress/strain
elements. The stiffness matrix, for example, is given by:
k = ∫ BT E B rdr dθ dz , (6.24)
V
or
2π 1 1
k= ∫ ∫ ∫ B E B r (det J ) dξ dη dθ
T
0 −1 −1
1 1
(6.25)
= 2π ∫ ∫ B E B r (det J ) dξ dη ,
T
−1 −1
in which B is the matrix relating the nodal displacement vector to strain vector in the cylindrical
coordinate system.
These axisymmetric elements have planar shapes, but actually represent rings in the
circumferential directions of the axisymmetric solids.
Applications:
Many rotating parts, such as a flywheel (Figure 6.12), can be modeled using
axisymmetric elements.
z
ω angular velocity (rad/s)
Figure 6.12. Cross-section of a flywheel that is rotating with an angular velocity ω.
Body forces in rotating parts are given by the following formulas:
fr = ρ r ω 2 , equivalent radial centrifugal/ inertial force;
(6.26)
fz = − ρ g, gravitational force;
where ρ is the mass density and g is gravitational acceleration.
© 1997-2013 Yijun Liu, University of Cincinnati 122
Lecture Notes: Introduction to the Finite Element Method
Other examples include cylinders subject to internal pressure (Figure 6.13) and press fit
(Figure 6.14).
p
r0
q = ( p ) 2π r0
Figure 6.13. A cylinder with internal radius r0 is subjected to internal pressure p.
ri
r0
ri + δ
(a) ring (sleeve) (b) shaft
At r = ri :
uo − ui = δ
⇒ MPC
“i” “o”
(c) Interface condition (Multi-Point Constraint (MPC))
Figure 6.14. An example of press fit.
© 1997-2013 Yijun Liu, University of Cincinnati 123
Lecture Notes: Introduction to the Finite Element Method
A more advanced example is the Belleville (conical) spring shown in Figure 6.15:
z
p
(a) (b)
δ
(c)
Figure 6.15. A Belleville (conical) springs.
This is a geometrically nonlinear (large deformation) problem (because of the nonlinear
behavior shown in the force-displacement curve as in Figure 6.15(c). Iteration approaches
(incremental methods) need to be employed to solve this type of problems.
Examples of axisymmetric solids shown in Figures 6.12-15 can be used as test problems.
One can build both 3-D solid models and corresponding 2-D axisymmetric models to compare
the modeling and computational efficiencies. The accuracy of the results for the two type of
models should be within certain tolerance if correct boundary conditions and same mesh
patterns/densities are employed.
© 1997-2013 Yijun Liu, University of Cincinnati 124
Lecture Notes: Introduction to the Finite Element Method
V. Summary
In this chapter, we discussed the 3-D (solid) elements for elasticity problems, that is,
general 3-D deformation and stress analyses. Solid elements are the most accurate elements and
should be applied when the bar, beam, plane stress/strain, plate/shell elements are no longer valid
or accurate. Especially for stress concentration problems, higher-order solid elements, such as
10-node tetrahedron or 20-node hexahedron (brick) elements should be employed in the FEA.
For rotating parts or solids of revolution and under axisymmetric loading, the axisymmetric
elements are most effective and efficient.
VI. Problems
Problem 1. For a tapered bar shown below, study the deformation and stresses in the bar with
a 3-D model using solid elements and a 1-D model using 1-D bar elements.
Assume R1 = 1 m, R2 = 0.5 m, L = 5 m, force F = 3000 N, the Young’s modulus
E = 200 GPa and Poisson’s ratio υ = 0.3. The bar is fixed at the left end. Compare
the results from the 3-D model and 1-D model.
F
2R1
R2
Problem 2. An open cylinder shown below has an inner radius a = 1 m, outer radius
b = 1.1 m, length L = 10 m and is applied with a pressure load p = 10 GPa on the
inner surface.
(a) Compute the stresses in the cylinder using 2-D plane stress, 2-D axisymmetric
and 3-D solid models. Compare the FEA results based on these models.
(b) Consider the same cylinder, added with a hole of radius r = 0.2 m on the
cylinder wall, centered at the mid-length, and along the radial direction.
Assume the Young’s modulus E = 200 GPa and Poisson’s ratio υ = 0.3.
© 1997-2013 Yijun Liu, University of Cincinnati 125
Lecture Notes: Introduction to the Finite Element Method
a
b L
Problem 3. For the rotating part sketched below, assume that it is made of steel with the
Young’s modulus E = 200 GPa, Poisson’s ratio υ = 0.3, and mass density
ρ = 7850 kg/m3. Assume that the part is rotating at a speed of 1000 RPM about
the z axis. Ignore the gravitational force. Compute the stresses in the part using
the FEA with a full 3-D model and an axisymmetric model. Compare the results
with the two models.
z
ω
12 mm
30 mm
50 mm
5 mm 8 mm
30 mm 16 mm
© 1997-2013 Yijun Liu, University of Cincinnati 126
Lecture Notes: Introduction to the Finite Element Method
Chapter 7. Structural Vibration and Dynamics
In this chapter, we first review the basic equations and their solutions for structural
vibration and dynamic analysis. Then, we discuss the FEA formulations for solving vibration and
dynamic responses. Guidelines in modeling and solving such problems are provided.
There are three main types of problems for structural vibration and dynamic analyses:
• Natural frequencies and modes (f(t) = 0);
• Frequency response (f(t) = fo sinωt);
• Transient response (f(t) is arbitrary);
where f(t) is the dynamic force applied on the structure, t the time, and ω the circular frequency
(Figure 7.1)
f(t)
Figure 7.1. A dynamic force applied to the structure.
I. Basic Equations
A. Single DOF System
First, let us consider a single degree of freedom (DOF) system shown in Figure 7.2.
k
f=f(t) m - mass
m k - stiffness
c
c - damping
f (t ) - force
FBD: ku
m f(t)
cu
x, u
Figure 7.2. A single DOF system with damping.
© 1997-2013 Yijun Liu, University of Cincinnati 127
Lecture Notes: Introduction to the Finite Element Method
From the free-body diagram (FBD) and Newton’s law of motion (ma = f), we have:
m u = f(t) − k u − c u ,
that is:
m u + c u + k u = f(t) , (7.1)
where u is the displacement, u = du / dt the velocity, and u = d u / dt the acceleration.
2 2
Free Vibration (Normal Mode Analysis):
If the applied force f(t) = 0 and no damping (c = 0), Eq. (7.1) becomes:
m u + k u = 0 . (7.2)
The physical meaning of this equation is: inertia force + elastic/stiffness force = 0.
Although there is no applied force, the structural can have nonzero displacement or
experience vibrations under contain initial conditions (ICs). To solve for such nontrivial
solutions, we assume:
u(t) = U sin ωt ,
where ω is the circular frequency of oscillation, U the amplitude. Substituting this into Eq. (7.2)
yields:
−U ω2 m sin ωt + kU sin ωt = 0
that is:
−ω 2 m + k U = 0 .
For nontrivial solutions for U, we must have:
−ω 2 m + k = 0 ,
which yields
k
ω= . (7.3)
m
This is the circular natural frequency of the single DOF system (rad/s). The cyclic frequency
(1/s = Hz) is ω/2π.
Equation (7.3) is a very important result in free vibration analysis, which says that the
natural frequency of a structural is proportional to the square-root of the stiffness of the structure
and inversely proportional to the square-root of the total mass of the structure.
The typical response of the system in undamped free vibration is sketched in Figure 7.3.
© 1997-2013 Yijun Liu, University of Cincinnati 128
Lecture Notes: Introduction to the Finite Element Method
u
u=Usinwt
U
t
U
T=1/f
Undamped Free Vibration
Figure 7.3. Typical response in an undamped free vibration.
For non-zero damping c, where
0 < c < cc = 2mω = 2 k m (cc = critical damping) (7.4)
we have the damped natural frequency:
ωd = ω 1 − ξ 2 , (7.5)
where
ξ = c / cc (7.6)
is called the damping ratio.
For structural damping: 0 ≤ ξ < 0.15 (usually 1~5%)
ωd ≈ ω . (7.7)
That is, we can ignore damping in normal mode analysis.
The typical response of a system in damped free vibration is sketched in Figure 7.4. We
can see that damping has the effect of reducing the vibration of the system.
Figure 7.4. Typical response of a free vibration with a nonzero damping c < cc .
© 1997-2013 Yijun Liu, University of Cincinnati 129
Lecture Notes: Introduction to the Finite Element Method
B. Multiple DOF Systems
Equation of Motion:
For multiple DOF systems, the equation of motion can be written as:
+ Cu + Ku = f (t ) ,
Mu (7.8)
in which: u nodal displacement vector;
M mass matrix of the structure;
C damping matrix;
K stiffness matrix;
f forcing vector.
The physical meaning of Eq. (7.8) is :
Inertia forces + Damping forces + Elastic forces = Applied forces
We already know how to determine the stiffness matrix K for a structure, as discussed in
previous chapters. The main tasks in vibration analysis is to determine the mass matrix and
damping matrix for the structure.
Mass Matrices:
There are two types of mass matrices: lumped mass matrices and consistent mass
matrices. The former is empirical and easier to determine, and the latter is analytical and more
involved in their computing.
We use a bar element to illustrate the lumped mass matrix (Figure 7.5).
1 ρ, A, L 2
ρAL ρAL
m1 = m2 =
2 2
u1 u2
Figure 7.5. The lumped mass for a 1-D bar element.
For this bar element, the lumped mass matrix for the element is found to be:
ρ AL
2 0
m= ,
0 ρ AL
2
which is a diagonal matrix and thus is easier to compute.
In general, we apply the following element consistent mass matrix:
© 1997-2013 Yijun Liu, University of Cincinnati 130
Lecture Notes: Introduction to the Finite Element Method
m=
∫ ρN
V
T
NdV , (7.9)
where N is the same shape function matrix as used for the displacement field, and V is the
volume of the element.
Equation (7.9) is obtained by considering the kinetic energy within an element:
1 T 1
Κ= u mu (cf. mv 2 )
2 2
1 1
= ∫ ρ u 2 dV = ∫ ρ ( u ) udV
T
2 V 2 V
1 (7.10)
= ∫ ρ ( Nu ) ( Nu ) dV
T
2 V
1
= u T ∫ ρ NT NdV u .
2
V
m
For the bar element (linear shape function), the consistent mass matrix is:
1 − ξ
m=
∫V
ρ
ξ
[1 − ξ ξ ]ALdξ
1 / 3 1 / 6 u1 (7.11)
= ρAL
1 / 6 1 / 3 u2
which is a non-diagonal matrix.
As in the case for stiffness matrices, element mass matrices are established in local
coordinates first, then transformed to global coordinates, and finally assembled together to form
the global structure mass matrix M.
v1
v2
θ1
ρ, A, L θ2
Figure 7.6. The lumped mass for a 1-D simple beam element.
For a simple beam element (Figure 7.6), the consistent mass matrix can be found readily
by applying the four shape functions listed in Eq. (2.41). We have:
© 1997-2013 Yijun Liu, University of Cincinnati 131
Lecture Notes: Introduction to the Finite Element Method
m = ∫ ρ NT NdV
V
156 22 L 54 −13L v1
2
−3L2 θ (7.12)
ρ AL 22 L 4 L 13L
= 1
420 54 13L 156 −22 L v2
−13L −3L
2
−22 L 4 L2 θ2
Selecting a proper unit system is very important in vibration or dynamic analysis. Two
choices of the units are listed in Table 7.1. Make sure they are consistent in the FEA models.
Table 7.1. Units in dynamic analysis
Choice I Choice II
t (time) s s
L (length) m mm
m (mass) kg Mg
a (accel.) m/s2 mm/s2
f (force) N N
ρ (density) kg/m3 Mg/mm3
II. Free Vibration of Multiple DOF Systems
Free vibration or normal mode analysis aims to study the dynamic characteristics of a
structure, which include:
• Natural frequencies;
• Normal modes (shapes).
Let f(t) = 0 and C = 0 (ignore damping) in the dynamic equation (7.8) and obtain:
+ Ku = 0
Mu (7.13)
Assume that displacements vary harmonically with time, that is:
u(t ) = u sin(ωt ),
(t ) = ω u cos(ωt ),
u
(t ) = −ω 2 u sin(ωt ),
u
where u is the vector of the amplitudes of the nodal displacements.
© 1997-2013 Yijun Liu, University of Cincinnati 132
Lecture Notes: Introduction to the Finite Element Method
Substitutes these into Eq. (7.13) yields:
K − ω 2 M u = 0 . (7.14)
This is a generalized eigenvalue problem (EVP). The trivial solution is u = 0 for any values
of ω (not interesting). Nontrivial solutions ( u ≠ 0 ) exist if and only if:
K − ω 2M = 0 (7.15)
This is an n-th order polynomial of ω2, from which we can find n solutions (roots) or eigenvalues
ωi (i = 1, 2, …, n). These are the natural frequencies (or characteristic frequencies) of the
structure.
The smallest non-zero eigenvalue ω1 is called the fundamental frequency.
For each ωi , Eq. (7.14) gives one solution or eigen vector:
[K − ω i
2
]
M ui = 0 .
ui (i=1, 2, …, n) are the normal modes (or natural modes, mode shapes, etc.).
Properties of the Normal Modes:
Normal modes satisfy the following properties:
uiT K u j = 0 , uiT M u j = 0 , for i ≠ j, (7.16)
if ωi ≠ ω j . That is, modes are orthogonal (thus independent) to each other with respect to K and
M matrices.
Normal modes are usually normalized such that:
uTi Mui = 1, uTi Kui = ωi2 . (7.17)
Notes:
• Magnitudes of displacements (modes) or stresses in normal mode analysis have no
physical meanings.
• For normal mode analysis, no support of the structure is necessary.
• ωi = 0 means there are rigid-body motions of the whole or a part of the structure. This
can be applied to check the FEA model (to see if there are rigid-body motions,
mechanisms or free elements in the FEA models).
• Lower modes are more accurate than higher modes in the FEA calculations (because of
less spatial variations in the lower modes, leading to that fewer elements/wave length are
needed).
© 1997-2013 Yijun Liu, University of Cincinnati 133
Lecture Notes: Introduction to the Finite Element Method
Example 7.1:
Consider the free vibration of a cantilever beam with one element as shown below.
y
v2
ρ, A, EI θ2
1 2 x
L
We have the following equation for the free vibration (EVP):
v2 0
K − ω M θ = 0 ,
2
2
where
EI 12 −6 L ρ AL 156 −22 L
K= , M= .
L3 −6 L 4 L2 420 −22 L 4 L2
The equation for determine the natural frequencies is:
12 − 156λ −6 L + 22 Lλ
= 0,
−6 L + 22 Lλ 4 L2 − 4 L2 λ
2 4
in which λ = ω ρ AL / 420 EI .
Solving the EVP, we obtain:
1
EI 2 v2 1
ω1 = 3.533 4
, = 1.38 ,
ρ AL θ 2 1
L
1
EI 2 v2 1
ω2 = 34.81 4
, = 7.62 .
ρ AL θ 2 2
L
The exact solutions of the first two natural frequencies for this problem are:
1 1
EI 2 EI 2
ω1 = 3.516 4
, ω2 = 22.03 4
.
ρ AL ρ AL
We can see that for the FEA solution with one beam element, mode 1 is calculated much more
accurately than mode 2. More elements are needed in order to compute mode 2 more accurately.
The first three mode shapes of the cantilever beam is shown in the insert above.
© 1997-2013 Yijun Liu, University of Cincinnati 134
Lecture Notes: Introduction to the Finite Element Method
III. Damping
There are two commonly used models for viscous damping.
A. Proportional Damping (Rayleigh Damping)
In this damping model, the damping matrix C is assumed to be proportional to the
stiffness and mass matrices in the following fashion:
C = αK + βM , (7.18)
where the constants α and β are found from the following two equations:
αω1 β αω 2 β
ξ1 = + , ξ2 = + , (7.19)
2 2ω1 2 2ω 2
with ω1 , ω2 , ξ1 and ξ 2 (damping ratios) being specified by the user. The plots of the above
two equations are shown in Figure 7.7.
Figure 7.7. Two equations for determining the proportional damping coefficients.
B. Modal Damping
In this damping model, the viscous damping is incorporated in the modal equations to be
discussed in the next section.
© 1997-2013 Yijun Liu, University of Cincinnati 135
Lecture Notes: Introduction to the Finite Element Method
IV. Modal Equations
Use the normal modes (modal matrices), we can transform the coupled system of
dynamic equations to uncoupled system of equations or modal equations.
We have:
K − ωi 2 M ui = 0, i = 1, 2, ..., n , (7.20)
where the normal modes u i satisfy:
T
ui K u j = 0,
T for i ≠ j,
ui M u j = 0,
and
uTi Mui = 1,
T for i = 1, 2, …, n.
ui Kui = ωi ,
2
Form the modal matrix:
Φ ( n×n ) = [u1 u 2 u n ] (7.21)
We can verify that:
ω12 0 0
T 0 ω 22
Φ KΦ = Ω = (Spectral matrix),
0
(7.22)
0 0 ω n2
ΦT MΦ = I.
Transformation for the displacement vector:
u = z1u1 + z2 u 2 + + zn u n = Φz , (7.23)
where
z1 (t )
z2 (t )
z=
zn (t )
are called the principal coordinates.
Substitute (7.23) into the dynamic equation (7.8) and obtain:
+ CΦz
MΦz + KΦz = f (t ).
Pre-multiply this result by ΦT, and apply (7.22):
© 1997-2013 Yijun Liu, University of Cincinnati 136
Lecture Notes: Introduction to the Finite Element Method
z + Cφ z
+ Ωz = p(t ), (7.24)
where Cφ = α I + β Ω if proportional damping is applied, and p = ΦT f (t ) .
If we introduce modal damping:
2ξ1ω1 0 0
0 2ξ 2ω 2
Cφ =
, (7.25)
0
2ξ nω n
where ξi is the damping ratio at mode i, Eq. (7.24) becomes:
zi + 2ξ iωi zi + ωi2 zi = pi (t ), i = 1, 2, …, n. (7.26)
Equations in (7.24) with modal damping, or in (7.26), are called modal equations. These
equations are uncoupled, second-order differential equations, which are much easier to solve
than the original dynamic equation which is a coupled system.
To recover u from z, apply transformation (7.23) again, once z is obtained from (7.26).
Notes:
• Only the first few modes may be needed in constructing the modal matrix Φ (that is, Φ
could be an n×m rectangular matrix with m<n). Thus, significant reduction in the size of
the system can be achieved.
• Modal equations are best suited for structural vibration problems in which higher modes
are not important (that is, for structural vibrations, but not for structures under impact or
shock loadings).
V. Frequency Response Analysis
Frequency response analysis is also called harmonic response analysis, when the applied
dynamic load is a sine or cosine functions. In this case, the equation of motion is:
+ Cu + Ku = F
Mu sin
ω
t (7.27)
Harmonic loading
A. Modal Method
In this approach, we apply the modal equations, that is:
zi + 2ξiωi zi + ωi2 zi = pi sin ωt , i = 1, 2, …, m.
(7.28)
These are uncoupled equations. The solutions for z are in the form:
© 1997-2013 Yijun Liu, University of Cincinnati 137
Lecture Notes: Introduction to the Finite Element Method
pi ωi2
zi (t ) = sin(ωt − θ i ),
(7.29)
(1 − ηi2 ) 2 + (2ξiηi ) 2
where
2ξiηi
θi = arctan 1 − η 2 , phase angle;
i
ηi = ω ωi ;
c ci
ξi = i = , damping ratio.
cc 2mωi
The response of each mode zi is similar to that of a single DOF system. Once the natural
coordinate vector z is known, we can recover the real displacement vector u from z using
Eq. (7.23).
B. Direct Method
In this approach, we solve Eq. (7.27) directly, that is, compute the inverse of the
coefficient matrix, which is in general much more expensive than the modal method.
iω t
Using complex notation to represent the harmonic response, we have u = ue and
Eq. (7.27) becomes:
K + iωC − ω M
u = F
2
(7.30)
Inverting the matrix K + iωC − ω 2 M , we can obtain the displacement amplitude vector u .
However, this equation is expensive to solve for large systems and the matrix K + iωC − ω 2 M
can become ill-conditioned if ω is close to any natural frequency ωi of the structure. Therefore,
the direct method is only applied when the system of equations is small and the frequency is
away from any natural frequency of the structure.
VI. Transient Response Analysis
In the transient response analysis, also called dynamic response/time-history analysis, we
are interested in computing the responses of the structures under arbitrary time-dependent
loading (Figure 7.8).
© 1997-2013 Yijun Liu, University of Cincinnati 138
Lecture Notes: Introduction to the Finite Element Method
f(t)
(a)
u(t)
t
(b)
Figure 7.8. (a) A step type of loading; (b) Structural response to the step loading.
To compute the transient responses, integration through time is employed (Figure 7.9).
u(t)
u1
u n u n+1
u2
t0 t1 t2 t n t n+1 t
Figure 7.9. Computing the responses by integration through time.
We write the equation of motion at instance t n , n = 0, 1, 2, 3, ⋅⋅⋅, as:
© 1997-2013 Yijun Liu, University of Cincinnati 139
Lecture Notes: Introduction to the Finite Element Method
n + Cu n + Ku n = f n .
Mu (7.31)
Then, we introduce time increments: ∆t=tn+1-tn, n=0, 1, 2, 3, ⋅⋅⋅, and integrate through the time.
There are two categories of methods for transient analysis as described in the following
sections.
A. Direct Methods (Direct Integration Methods)
Central Difference Method:
Approximate the velocity and acceleration vectors by using finite difference:
1
n =
u (u n +1 − u n −1 ),
2∆t
1
n
u = (u n +1 − 2u n + u n −1 ) (7.32)
( ∆t ) 2
Dynamic equation becomes,
1 1
M (u n + 1 − 2u n + u n −1 ) + C (u n +1 − u n −1 ) + Ku n = f n ,
2∆t
2
(∆t )
which yields
Au n +1 = F(t ) (7.33)
where
1 1
A = M+ C,
( ) 2∆t
2
∆t
F (t ) = f − K − 2 M u − 1 M − 1 C u .
n
( ∆t )
2 n
( ∆t )
2
2∆t
n −1
We compute un+1 from un and un-1, which are known from the previous time step. The solution
procedure is repeated or marching from t0,t1, t n ,t n +1, , until reach the specified maximum
time. This method is unstable if ∆t is too large.
Newmark Method:
We use the following approximations:
(∆t ) 2
u n +1 ≈ u n + ∆tu n + [(1 − 2β )un + 2βun +1 ], → (un +1 = )
2 (7.34)
u n +1 ≈ u n + ∆t [(1 − γ )u
n + γu n +1 ],
© 1997-2013 Yijun Liu, University of Cincinnati 140
Lecture Notes: Introduction to the Finite Element Method
where β and γ are chosen constants. These lead to the following equation:
Au n +1 = F (t ) (7.35)
where
γ 1
A=K+ C+ M,
β ∆t β ( ∆t ) 2
F (t ) = f (f n +1 , γ , β , ∆t , C, M , u n , u
n ,u
n ).
This method is unconditionally stable if
1
2β ≥ γ ≥ .
2
1 1
For example, we can use γ = , β = , which gives the constant average acceleration method.
2 4
Direct methods can be expensive, because of the need to compute A-1, repeatedly for each
time step if nonuniform time steps are used.
B. Modal Method
In this method, we first do the transformation of the dynamic equations using the modal
matrix before the time marching:
m
u= ∑ u z (t ) =Φz,
i =1
i i
i = 1,2,⋅⋅⋅, m. (7.36)
zi + 2ξ i ωi zi + ωi z i = pi (t ),
Then, solve the uncoupled equations using an integration method. We can use, for example,
10% of the total modes (m = n/10). The advantages of the modal method are as follows:
• Uncoupled system;
• Fewer equations;
• No inverse of matrices;
• More efficient for large problems.
However, the modal method is less accurate if higher modes are important, which is the
case for structures under impact or shock loading. Table 7.2 summarizes the advantages and
disadvantages of the direct and modal methods for transient response analysis.
© 1997-2013 Yijun Liu, University of Cincinnati 141
Lecture Notes: Introduction to the Finite Element Method
Table 7.2. Comparisons of the methods
Direct Methods Modal Method
• Small models • Large models
• More accurate (with small ∆t) • Higher modes ignored
• Single loading • Multiple loading
• Shock loading • Periodic loading
• … • …
Cautions in Dynamic Analysis
• Symmetry model should not be used in the dynamic analysis (normal modes, etc.)
because symmetric structures can have non-symmetric modes. However, symmetry can
still be applied in creating the FEA model of a symmetric structure.
• Mechanism or rigid body motion means ω = 0. Can use this to check FEA models to see
if they are properly connected and/or supported.
• Input for FEA: Loading F(t) or F(ω) can be very complex and data can be enormous in
real engineering applications (for example, the load data for a car) and thus they often
need to be filtered first before being used as input for FEA.
Example 7.2:
Figure 7.10. FEA model of a front bumper and supporting brackets.
© 1997-2013 Yijun Liu, University of Cincinnati 142
Lecture Notes: Introduction to the Finite Element Method
Figure 7.10 shows a front bumper and the supporting brackets in a car. The model was
applied to study the dynamic responses of the bumper. Shell elements were used for this study
and the natural frequencies and vibration modes were obtained first. Figure 7.11 shows the first
mode of the bumper when it is constrained at the bracket locations. The frequency response of
the bumper was also analyzed using the same FE model, as shown by the red curve in
Figure 7.12 and with the acceleration of the two brackets as the input. Several modifications of
the bumper design were also studied with the goal to increase the base natural frequency (for
example, from below 30 Hz to above 35 Hz) and to reduce the magnitudes of the frequency
responses. The improved responses are shown by the other three curves in Figure 7.12.
Figure 7.11. The first vibration mode of the bumper.
Reference Model
Figure 7.12. Frequency response of the bumper from 0 to 50 Hz.
© 1997-2013 Yijun Liu, University of Cincinnati 143
Lecture Notes: Introduction to the Finite Element Method
Example 7.3:
One of the most interesting applications of the dynamic analysis with FEA is to conduct
crash analysis and virtual drop tests of various products. The most popular FEA software for
such analyses is the LS-DYNA package from Livermore Software Technology Corporation
(LSTC). Figure 7.13 is an example of crash analysis of a car using LS-DYNA. Figure 1.4 is an
example of drop test simulation of a soda can, also using LS-DYNA.
Figure 7.13. Car crash analysis using LS-DYNA (from LSTC website www.lstc.com).
More information about how to perform the impact analysis and drop test using the
dynamic FEA, especially the software LS-DYNA, can be found from the LSTC website and the
documents with the LS-DYNA software package.
VII. Summary
In this chapter, we first reviewed the equation of motion for both single DOF and
multiple DOF systems and discussed how to compute the mass and damping matrices in the FEA
formulations. Then, we discussed the methods for solving normal modes, harmonic responses,
and transient responses for structural vibration and dynamic problems. The advantages and
disadvantages of the direct method and modal method are discussed. Several examples of
vibration and dynamic analyses are also discussed to show the applications of the FEA in
vibration and dynamic analyses.
VIII. Problems
Problem 1. For the cantilever beam studied in Example 7.1, apply more beam elements and
investigate the convergence of the FEA solutions for the first ten natural
frequencies and normal modes of the beam.
© 1997-2013 Yijun Liu, University of Cincinnati 144
Lecture Notes: Introduction to the Finite Element Method
Problem 2. For a square plate with edge length = 1 m, thickness = 0.005 m, Young’s modulus
E = 70 GPa, Poisson’s ratio υ = 0.3, and mass density ρ = 2800 kg/m3, find the
first five natural frequencies and normal modes using the FEA when the plate is:
(a) Clamped at the four edges;
(b) Simply supported at the four edges;
(c) Free at the four edges (not supported).
Study the convergence of your FEA results and discuss the effects of the support
on the natural frequencies of the plate.
Problem 3. A fuel tank, with a total length = 5 m, diameter = 1 m, and thickness = 0.01 m, is
shown below. Using the FEA, find the first ten natural frequencies and
corresponding normal modes, when:
(a) The tank is not supported at all;
(b) The tank is constrained along the circumferences in the radial directions at the
two locations 1 m away from the two ends.
Carefully build your FE mesh (using shell elements) so that the symmetry of the
tank is reserved and the boundary conditions in part (b) can be applied readily.
Assume the tank is made of steel with the Young’s modulus E = 200 GPa,
Poisson’s ratio υ = 0.3, and mass density ρ = 7850 kg/m3.
© 1997-2013 Yijun Liu, University of Cincinnati 145
Lecture Notes: Introduction to the Finite Element Method
Chapter 8. Thermal Analysis
In this chapter, we will discuss briefly the thermal analysis using the FEA. Thermal
stresses due to changes of the temperatures are common in most engineering systems, such as
cars, airplanes, bridges, electronic devices, and many consumer products.
The two main objectives in thermal analysis are:
• Determine the temperature field (steady state or unsteady state);
• Determine the thermal stresses in structures due to the temperature changes.
I. Temperature Field
For the temperature field in a 1-D space, such as a bar (Figure 8.1), we have the
following Fourier heat conduction equation:
∂T
f x = −k , (8.1)
∂x
where,
fx = heat flux per unit area,
k = thermal conductivity,
T = T(x, t) = temperature field.
T(x, t)
Figure 8.1. The temperature field T(x, t) in a 1-D bar model.
For 3-D case, we have:
fx ∂T ∂x
f y = −Κ ∂T ∂y , (8.2)
f ∂T ∂z
z
where, fx, fy, fz = heat flux in the x, y and z direction, respectively. In the case of isotropic
materials, the conductivity matrix is:
k 0 0
Κ = 0 k 0 . (8.3)
0 0 k
The equation of heat flow is given by:
© 1997-2013 Yijun Liu, University of Cincinnati 146
Lecture Notes: Introduction to the Finite Element Method
∂f ∂f ∂f ∂T
− x + y + z + qv = cρ (8.4)
∂x ∂y ∂z ∂t
in which,
qv = rate of internal heat generation per unit volume,
c = specific heat,
ρ = mass density.
For steady state case ( ∂T ∂t = 0 ) and isotropic materials, we can obtain:
k∇ 2T = −qv . (8.5)
This is a Poisson equation, which needs to be solved under given boundary conditions.
Boundary conditions for steady state heat conduction problems are (Figure 8.2):
T = T, on ST ; (8.6)
∂T
Q ≡ −k = Q, on S q . (8.7)
∂n
Note that at any point on the boundary S = ST S q , only one type of BCs can be specified.
y Sq
ST
x
Figure 8.2. Boundary conditions for heat conduction problems.
Finite Element Formulation for Heat Conduction:
For heat conduction problems, we can establish the following FE equation:
KT T = q (8.8)
where,
KT = conductivity matrix,
T = vector of nodal temperature,
q = vector of thermal loads.
© 1997-2013 Yijun Liu, University of Cincinnati 147
Lecture Notes: Introduction to the Finite Element Method
The element conductivity matrix is given by:
k T = ∫ BT ΚBdV . (8.9)
V
This is obtained in a similar way as for the structural analysis, that is, by starting with the
interpolation T = NTe for the temperature field (with N being the shape function matrix and Te
the nodal temperature). Note that there is only one DOF at each node for the thermal problems.
For transient (unsteady state) heat conduction problems, we have:
∂T
≠ 0.
∂t
In this case, we need to apply finite difference schemes (use time steps and integrate in time), as
in the transient structural analysis, to obtain the transient temperature fields.
II. Thermal Stress Analysis
To determine the thermal stresses due to temperature changes in structures, we can
proceed to:
• Solve Eq. (8.8) first to obtain the temperature (change) fields.
• Apply the temperature change ΔT as initial strains (or initial stresses) to the structure to
compute the thermal stresses due to the temperature change.
At temperature T1
At temperature T2
εo
Figure 8.3. Expansion of a bar due to increase in temperature.
1-D Case:
To understand the stress-strain relations in cases of solids undergo temperature changes,
we first examine the 1-D case (Figure 8.3). We have for the thermal strain (or initial strain):
ε o = α∆T , (8.10)
in which,
α = the coefficient of thermal expansion,
© 1997-2013 Yijun Liu, University of Cincinnati 148
Lecture Notes: Introduction to the Finite Element Method
∆T = T2 − T1 = change of temperature.
Total strain is given by:
ε = εe + εo (8.11)
with ε e being the elastic strain due to mechanical load.
That is, the total strain can be written as:
ε = E −1σ + α∆T , (8.12)
Or, inversely, the stress is given by:
σ = E (ε − ε o ) . (8.13)
Example 8.1:
Consider the bar under thermal load ∆T as shown in Figure 8.3.
(a) If no constraint on the right-hand side, that is, the bar is free to expand to the
right, then we have:
ε = ε o , ε e = 0, σ = 0 ,
from Eq. (8.13), that is, there is no thermal stress in this case!
(b) If there is a constraint on the right-hand side, that is, the bar can not expand to the
right, then we have:
ε = 0, ε e = −ε o = −α∆T , σ = − Eα∆T ,
from Eqs. (8.11) and (8.13). Thus, thermal stress exists!
From this simple example, we see that the way in which the structure is constrained has a
critical role in inducing the thermal stresses.
2-D Cases:
For plane stress, we have:
ε x α∆T
ε o = ε y = α∆T . (8.14)
γ 0
xy o
For plane strain, we have:
ε x (1 + ν )α∆T
ε o = ε y = (1 + ν )α∆T , (8.15)
γ 0
xy o
© 1997-2013 Yijun Liu, University of Cincinnati 149
Lecture Notes: Introduction to the Finite Element Method
in which, ν is the Poisson’s ratio.
3-D Case:
ε x α∆T
ε α∆T
y
ε z α∆T
εo = = . (8.16)
γ
xy 0
γ yz 0
γ zx o 0
Observation: Temperature changes do not yield shear strains.
In both 2-D and 3-D cases, the total strain can be given by the following vector equation:
ε = εe + εo . (8.17)
And the stress-strain relation is given by:
σ = Eε e = E(ε − ε o ) . (8.18)
Notes on FEA for Thermal Stress Analysis:
• Need to specify α for the structure and ∆T on the related elements (which experience
the temperature change).
• Note that for linear thermoelasticity, same temperature change will yield same stresses,
even if the structure is at two different temperature levels.
• Differences in the temperatures during the manufacturing and working environment are
the main cause of thermal (residual) stresses.
Example 8.2:
First, we study a heat sink model taken from Ref. [8] for thermal analysis. A heat sink is
a device commonly used to dissipate heat from a CPU in a computer. In this heat sink model, a
given temperature field ( T = 120 ) is specified on the bottom surface and a heat flux condition
∂T
( Q ≡ −k = −0.2 ) is specified on all the other surfaces. The 20-node brick elements are used
∂n
and the FE mesh shown in Figure 8.4 has 127,149 nodes. This mesh for the volume was obtained
by extruding the cross section meshed with quadrilateral area elements so that a mapped mesh
was obtained. The computed temperature distribution on the heat sink using ANSYS is shown in
Figure 8.5. The cooling effect of the heat sink is most evident.
© 1997-2013 Yijun Liu, University of Cincinnati 150
Lecture Notes: Introduction to the Finite Element Method
Figure 8.4. A heat-sink model used for heat conduction analysis.
Figure 8.5.Computed temperature distribution in the heat-sink.
Example 8.3:
Next, we study the thermal stresses in structures due to temperature changes. For this
purpose, we employ the same model of a plate with a center hole (Figure 8.6) as used in Chapter
3 and Chapter 4 to show the relation between the thermal stresses and constraints. We assume
that the plate is made of steel with the Young’s modulus E = 200 GPa, Poisson’s ratio ν = 0.3
and thermal expansion coefficient α = 12x10-6 1/°C. The plate is applied with a uniform
temperature increase of 100 °C.
© 1997-2013 Yijun Liu, University of Cincinnati 151
Lecture Notes: Introduction to the Finite Element Method
Figure 8.6. A square plate with a center hole and under a uniform temperature load.
Figure 8.7 shows the computed thermal stresses in the plate under two different types of
constraints. When the plate is constrained (roller support) at the left side only, the plate expands
uniformly in both the x and y directions, which causes no thermal stresses (Figure 8.7 (a), note
that the numbers, ranging from 10-6 to 10-3, are actually machine zeros). However, when the
plate is constrained at both the left and right sides, the plate can expand only in the y direction
and significant thermal stresses are induced (Figure 8.7 (b)), especially near the edge of the hole.
(a)
© 1997-2013 Yijun Liu, University of Cincinnati 152
Lecture Notes: Introduction to the Finite Element Method
(b)
Figure 8.7. Thermal (von Mises) stresses in the plate: (a) When the plate is constrained at left
side only (thermal stresses = 0); (b) When the plate is constrained at both left and right sides.
In many cases, the changes of the material properties of a structure should be considered
as well when the temperature changes are significant, especially when the structure is exposed to
high temperatures such as in an aircraft engine. Cyclic temperature fields can also cause thermal
fatigue of structures and lead to failures. All these phenomena can be modeled with the FEA and
interested readers can consult with the documents of the FEA software at hand.
III. Summary
In this chapter, we briefly discussed the governing equations for heat conduction
problems and the FEA formulation. Thermal stresses due to changes of temperatures in
structures are also discussed and the effects of constraints of the structures on the thermal
stresses are emphasized.
IV. Problems
Problem 1. Study the heat conduction problem in a simple annular region shown below, using
∂T
the FEA. Assume a = 1, b = 2, Ta = 100, and Qb = −k = −200 . Determine the
∂n
temperature field and heat flux in this region and compare your FEA results with
the analytical solution.
© 1997-2013 Yijun Liu, University of Cincinnati 153
Lecture Notes: Introduction to the Finite Element Method
a b
O
Sa
V
Sb
Problem 2. For the same glass cup model studied in Chapter 5, as shown in the figure below,
determine the thermal stresses when the inner surfaces of the cup experience a
temperature change from a room temperature of 20 °C to 100 °C, while all other
surfaces are kept at the same room temperature of 20 °C. Assume that the cup has
a uniform thickness, E = 70 GPa, v = 0.17 and the coefficient of thermal
expansion α = 8.0×10-6 ⁄ °C.
40 mm
150 mm
Thickness = 4 mm
30 mm
© 1997-2013 Yijun Liu, University of Cincinnati 154
Lecture Notes: Introduction to the Finite Element Method
References for Further Studies
1. O. C. Zienkiewicz, “The birth of the finite element method and of computational
mechanics,” International Journal for Numerical Methods in Engineering, 60, 3-10 (2004).
2. O. C. Zienkiewicz and R. L. Taylor, The Finite Element Method, 5th ed (McGraw-Hill,
New York, 2000).
3. J. N. Reddy, An Introduction to the Finite Element Method, 3rd ed (McGraw-Hill, New
York, 2006).
4. R. D. Cook, Finite Element Modeling for Stress Analysis (John Wiley & Sons, Inc., New
York, 1995).
5. R. D. Cook, D. S. Malkus, M. E. Plesha, and R. J. Witt, Concepts and Applications of
Finite Element Analysis, 4th ed (John Wiley & Sons, Inc., New York, 2002).
6. K. J. Bathe, Finite Element Procedures (Prentice Hall, Englewood Cliffs, NJ, 1996).
7. T. R. Chandrupatla and A. D. Belegundu, Introduction to Finite Elements in Engineering,
3rd ed (Prentice Hall, Upper Saddle River, NJ, 2002).
8. S. Moaveni, Finite Element Analysis - Theory and Application with ANSYS, 3rd ed
(Prentice-Hall, Upper Saddle River, NJ, 2007).
9. Y. C. Fung, A First Course in Continuum Mechanics, 3rd ed (Prentice Hall, Englewood
Cliffs, 1994).
10. P. Timoshenko and J. N. Goodier, Theory of Elasticity, 3rd ed (McGraw-Hill, New York,
1987).
11. S. P. Timoshenko and S. Woinowsky-Krieger, Theory of Plates and Shells, 2nd ed
(McGraw-Hill, New York, 1987).
© 1997-2013 Yijun Liu, University of Cincinnati 155
You might also like
- Concepts and Applications of Finite Element Analysis 4ed PDF100% (3)Concepts and Applications of Finite Element Analysis 4ed PDF733 pages
- Applied Finite Element Analysis - Larry J. Segerlind75% (4)Applied Finite Element Analysis - Larry J. Segerlind222 pages
- FEM - Cont. & Chapt-1 - Slide PresentationNo ratings yetFEM - Cont. & Chapt-1 - Slide Presentation10 pages
- (G. Lakshmi Narasaiah) Finite Element Analysis PDFNo ratings yet(G. Lakshmi Narasaiah) Finite Element Analysis PDF349 pages
- CE6502 Finite Element Method in Structures - Course Outline 2019.20 IINo ratings yetCE6502 Finite Element Method in Structures - Course Outline 2019.20 II4 pages
- Rose School: European School of Advanced Studies in Reduction of Seismic RiskNo ratings yetRose School: European School of Advanced Studies in Reduction of Seismic Risk49 pages
- Desai and Abel - Intro To Finite Element MethodNo ratings yetDesai and Abel - Intro To Finite Element Method323 pages
- Finite Element Methods: Prof. D.Neeraja Hced/HfdtNo ratings yetFinite Element Methods: Prof. D.Neeraja Hced/Hfdt91 pages
- C I V 5 0 3 Finite Element Analysis: DR Zubair Syed, MieaustNo ratings yetC I V 5 0 3 Finite Element Analysis: DR Zubair Syed, Mieaust57 pages
- Finite Element Analysis Theory and Programming by C. S. Krishnamoorthy82% (17)Finite Element Analysis Theory and Programming by C. S. Krishnamoorthy137 pages
- Finite Element Analysis G. Lakshmi NarasaiahNo ratings yetFinite Element Analysis G. Lakshmi Narasaiah349 pages
- Y. M. Desai, T. I Eldho, A. H. Shah - Finite Element Method With Applications in Engineering-Pearson Education (2011)100% (1)Y. M. Desai, T. I Eldho, A. H. Shah - Finite Element Method With Applications in Engineering-Pearson Education (2011)489 pages
- Lecture Notes - Finite Element Method: January 2020No ratings yetLecture Notes - Finite Element Method: January 20206 pages
- Chapter I Review of Related Studies and Literature89% (18)Chapter I Review of Related Studies and Literature5 pages
- Rbi Script - Plaza, Marvin - G11 Math - Q1-W2100% (2)Rbi Script - Plaza, Marvin - G11 Math - Q1-W26 pages
- Rheotherm: Circulating Water Flow and Fouling SensorNo ratings yetRheotherm: Circulating Water Flow and Fouling Sensor2 pages
- Assignment 1 - Linear Programming I - With AnswersNo ratings yetAssignment 1 - Linear Programming I - With Answers2 pages
- Draft - Master Direction On Outsourcing of Information Technology (IT) ServicesNo ratings yetDraft - Master Direction On Outsourcing of Information Technology (IT) Services23 pages
- Cambridge 1 Syllabus Planer Nov - Dec 2023No ratings yetCambridge 1 Syllabus Planer Nov - Dec 20233 pages
- Cross-Border Interbank Payment System (CIPS)No ratings yetCross-Border Interbank Payment System (CIPS)40 pages
- Statistical Approach To Quality ManagementNo ratings yetStatistical Approach To Quality Management57 pages
- CSE 101 - Introduction To Computers I: TopicNo ratings yetCSE 101 - Introduction To Computers I: Topic39 pages
- Black Box Fairness Testing of Machine Learning ModelsNo ratings yetBlack Box Fairness Testing of Machine Learning Models11 pages
- Identify Your Imac Model - Apple SupportNo ratings yetIdentify Your Imac Model - Apple Support1 page
- Properties and Classifications of Bamboo For ConstNo ratings yetProperties and Classifications of Bamboo For Const11 pages
- Dfs-tp-Ah-001-Us 012h10 Lgtechpaper Errorcodes Ls 091-121 HSV 3No ratings yetDfs-tp-Ah-001-Us 012h10 Lgtechpaper Errorcodes Ls 091-121 HSV 32 pages