A First Course in The Finite Element Method - Lecture Slides PDF

CIVL 7/8117 Chapter 1 - Introduction to FEM 1/26
A First Course in Finite Elements

Introduction
• The finite element method has become a powerful tool

for the numerical solution of a wide range of engineering
problems.
• Applications range from deformation and stress analysis of

automotive, aircraft, building, and bridge structures to field
analysis of heat flux, fluid flow, magnetic flux, seepage,
and other flow problems.

Introduction
• With the advances in computer technology and CAD

systems, complex problems can be modeled with relative
ease.
• Several alternative configurations can be tried out on a

computer before the first prototype is built.
• All of this suggests that we need to keep pace with these

developments by understanding the basic theory, modeling
techniques, and computational aspects of the finite
element method.

Introduction
• In this method of analysis, a complex region defining a
continuum is discretized into simple geometric shapes
called finite elements.
• The material properties and the governing relationships

are considered over these elements and expressed in
terms of unknown values at element corners.
• An assembly process, considering the loading and

constraints, results in a set of equations.
• Solution of these equations gives an approximate behavior

of the continuum.






Historical Background
• Basic ideas of the finite element method originated

from advances in aircraft structural analysis.
• In 1941, Hrenikoff presented a solution of elasticity
problems using the “frame work method.”
• Courant’s paper, which used piecewise polynomial
interpolation over triangular subregions to model
torsion problems, appeared in 1943.
• Turner et al. derived stiffness matrices for truss,
beam, and other elements and presented their
findings in 1956.
• The term “finite element” was first coined and used
by Clough in 1960.

• In the early 1960s, engineers used the method for

approximate solution of problems in stress analysis,
fluid flow, heat transfer, and other areas.
• A book by Argyris in 1955 on energy theorems and
matrix methods laid a foundation for further
developments in finite element studies.
• The first book on finite elements by Zienkiewicz and
Chung was published in 1967.
• In the late 1960s and early 1970s, finite element
analysis was applied to nonlinear problems and
large deformations.

• Until the early 1950s, matrix methods and the associated
finite element method were not readily adaptable for solving
complicated problems because of the large number of
algebraic equations that resulted.
• Hence, even though the finite element method was being

used to describe complicated structures, the resulting large
number of equations associated with the finite element
method of structural analysis made the method extremely
difficult and impractical to use.
• With the advent of the computer, the solution of thousands of

equations in a matter of minutes became possible.

• Mathematical foundations were laid in the 1970s.
• New element development, convergence studies, and

other related areas fall in this category.
• Today, developments in distributed or multi-node

computers and availability of powerful microcomputers
have brought this method within reach of students and
engineers working in small industries.

General Steps of the Finite Element Method
• Typically, for the structural stress-analysis problem, the

engineer seeks to determine displacements and
stresses throughout the structure, which is in equilibrium
and is subjected to applied loads.
• For many structures, it is difficult to determine the

distribution of deformation using conventional methods,
and thus the finite element method is necessarily used.

• There are two general approaches associated with the

finite element method.
• One approach, called the force, or flexibility method,

uses internal forces as the unknowns of the problem.
• To obtain the governing equations, first the equilibrium

equations are used. Then necessary additional equations
are found by introducing compatibility equations.
• The result is a set of algebraic equations for determining

the redundant or unknown forces.

• The second approach, called the displacement, or

stiffness method, assumes the displacements of the
nodes as the unknowns of the problem.
• The governing equations are expressed in terms of nodal

displacements using the equations of equilibrium and an
applicable law relating forces to displacements.

• These two approaches result in different unknowns (forces

or displacements) in the analysis and different matrices
associated with their formulations (flexibilities or
stiffnesses).
• It has been shown that, for computational purposes, the

displacement (or stiffness) method is more desirable
because its formulation is simpler for most structural
analysis problems.

• The finite element method involves modeling the
structure using small interconnected elements called finite
elements.
• A displacement function is associated with each finite

element.

• Every interconnected element is linked, directly or
indirectly, to every other element through common (or
shared) interfaces, including nodes and/or boundary lines
and/or surfaces.
• The total set of equations describing the behavior of each

node results in a series of algebraic equations best
expressed in matrix notation.

Basic Ingredients - Discrete Problems
The basic steps or building blocks of any application of FEM

to a mathematical or physical problem are:
1. Discretization
2. Interpolation
3. Elemental Description or Formulation
4. Assembly
5. Constraints
6. Solution
7. Computation of Derived Variables

Basic Ingredients - Discrete Problems
Whether the FEM is used for
(1) setting up and solving a problem by hand,

(2) using an existing FEM program to solve a problem,
or
(3) generating or writing a program to solve a class of
problems;
These basic steps are important in setting-up and solving

the FEM model.
Step 1 - Discretize and Select Element Types
• Step 1 involves dividing the body into an equivalent system

of finite elements with associated nodes and choosing the
most appropriate element type.
• The total number of elements used and their variation in size

and type within a given body are primarily matters of
engineering judgment.
• The elements must be made small enough to give usable

results and yet large enough to reduce computational effort.
• Small elements (and possibly higher-order elements) are

generally desirable where the results are changing rapidly,
such as where changes in geometry occur, whereas large
elements can be used where results are relatively constant.
• Primary line elements consist of bar (or truss) and beam

elements.
• They have a cross-sectional area but are usually represented

by line segments.
• The simplest line element (called a linear element) has two

nodes, one at each end, although higher-order elements
having three nodes or more (called quadratic, cubic, etc.
elements) also exist.
• The basic two-dimensional (or plane) elements are loaded by

forces in their own plane (plane stress or plane strain
conditions). They are triangular or quadrilateral elements.
• The simplest two-dimensional elements have corner nodes

only (linear elements) with straight sides or boundaries
although there are also higher-order elements, typically with
mid-side nodes (called quadratic elements) and curved
sides.


• The most common three-dimensional elements are

tetrahedral and hexahedral (or brick) elements; they are
used when it becomes necessary to perform a three-
dimensional stress analysis.
• The basic three dimensional elements have corner nodes

only and straight sides, whereas higher-order elements with
mid-edge nodes (and possible mid-face nodes) have curved
surfaces for their sides

• The axisymmetric element is developed by rotating a

triangle or quadrilateral about a fixed axis located in the plane
of the element through 360°.
• This element can be used when the geometry and loading of

the problem are axisymmetric.

Step 2 - Select a Displacement Function
Step 2 involves choosing a displacement function within each

element.
The function is defined within the element using the nodal

values of the element.
Linear, quadratic, and cubic polynomials are frequently used

functions because they are simple to work with in finite
element formulation.
The functions are expressed in terms of the nodal unknowns

(in the two-dimensional problem, in terms of an x and a y
component).
Hence, the finite element method is one in which a

continuous quantity, such as the displacement throughout the
body, is approximated by a discrete model composed of a set
of piecewise-continuous functions defined within each finite
domain or finite element.
Step 3 - Define the Strain/Displacement

and Stress/Strain Relationships
Strain/displacement and stress/strain relationships are

necessary for deriving the equations for each finite element.
For one-dimensional small strain deformation, say, in the x

direction, we have strain x, related to displacement u by
du
x 
dx
Stresses must be related to the strains through the
stress/strain law (generally called the constitutive law). The
simplest of stress/strain laws, Hooke’s law, often used in stress
analysis, is given by:
 x  E x
Step 4 - Derive the Element Stiffness Matrix and Equations
Direct Equilibrium Method - According to this method, the

stiffness matrix and element equations relating nodal forces
to nodal displacements are obtained using force equilibrium
conditions for a basic element, along with force/deformation
relationships.
This method is most easily adaptable to line or one-

dimensional elements (spring, bar, and beam elements)
Work or Energy Methods - To develop the stiffness matrix

and equations for two- and three-dimensional elements, it is
much easier to apply a work or energy method.
The principle of virtual work (using virtual displacements),

the principle of minimum potential energy, and Castigliano’s
theorem are methods frequently used for the purpose of
derivation of element equations.
We will present the principle of minimum potential energy

(probably the most well known of the three energy methods
mentioned here)
Methods of Weighted Residuals - The methods of weighted

residuals are useful for developing the element equations
(particularly popular is Galerkin’s method).
These methods yield the same results as the energy

methods, wherever the energy methods are applicable.
They are particularly useful when a functional such as

potential energy is not readily available.
The weighted residual methods allow the finite element

method to be applied directly to any differential equation
Step 5 - Assemble the Element Equations

and Introduce Boundary Conditions
The individual element equations generated in Step 4 can
now be added together using a method of superposition
(called the direct stiffness method) whose basis is nodal
force equilibrium (to obtain the global equations for the whole
structure).
Implicit in the direct stiffness method is the concept of

continuity, or compatibility, which requires that the structure
remain together and that no tears occur anywhere in the
structure.
The final assembled or global equation written in matrix form
F   K d 
is:
Step 6 - Solve for the Unknown Degrees of Freedom

(or Generalized Displacements)
Once the element equations are assembled and modified to
account for the boundary conditions, a set of simultaneous
algebraic equations that can be written in expanded matrix
form as:
 F1   K11 K12  K1n   d1 
F   K  
 2   21 K 22  K 2 n  d 2 
   
        
Fn  K n1 K n 2  K nn  d n 
where n is the structure total number of unknown nodal

degrees of freedom.
These equations can be solved for the d’s by using an

elimination method (such as Gauss’s method) or an iterative
method (such as Gauss Seidel’s method)
Step 7 - Solve for the Element Strains and Stresses
For the structural stress-analysis problem, important

secondary quantities of strain and stress (or moment and
shear force) can be obtained in terms of the displacements
determined in Step 6.
Step 8 - Interpret the Results
The final goal is to interpret and analyze the results for use
in the design/analysis process.
Determination of locations in the structure where large

deformations and large stresses occur is generally important
in making design/analysis decisions.
Advantages of the Finite Element Method
The finite element method has been applied to numerous

problems, both structural and non-structural. This method has
a number of advantages that have made it very popular.
1. Model irregularly shaped bodies quite easily
2. Handle general load conditions without difficulty
3. Model bodies composed of several different materials

because the element equations are evaluated individually
4. Handle unlimited numbers and kinds of boundary

conditions
Advantages of the Finite Element Method
5. Vary the size of the elements to make it possible to use

small elements where necessary
6. Alter the finite element model relatively easily and cheaply
7. Include dynamic effects
8. Handle nonlinear behavior existing with large

deformations and nonlinear materials
The finite element method of structural analysis enables the

designer to detect stress, vibration, and thermal problems
during the design process and to evaluate design changes
before the construction of a possible prototype.
End of
Introduction
CIVL 7/8117 Chapter 2 - The Stiffness Method 1/32
Chapter 2 – Introduction to the Stiffness

(Displacement) Method
Learning Objectives
• To define the stiffness matrix
• To derive the stiffness matrix for a spring element
• To demonstrate how to assemble stiffness matrices into
a global stiffness matrix
• To illustrate the concept of direct stiffness method to
obtain the global stiffness matrix and solve a spring
assemblage problem
• To describe and apply the different kinds of boundary
conditions relevant for spring assemblages
• To show how the potential energy approach can be used
to both derive the stiffness matrix for a spring and solve
a spring assemblage problem
The Stiffness (Displacement) Method
This section introduces some of the basic concepts on which

the direct stiffness method is based.
The linear spring is simple and an instructive tool to illustrate
the basic concepts.
The steps to develop a finite element model for a linear spring
follow our general 8 step procedure.
1. Discretize and Select Element Types - Linear spring

elements
2. Select a Displacement Function - Assume a variation
of the displacements over each element.
3. Define the Strain/Displacement and Stress/Strain
Relationships - use elementary concepts of equilibrium
and compatibility.
4. Derive the Element Stiffness Matrix and Equations -

Define the stiffness matrix for an element and then
consider the derivation of the stiffness matrix for a linear-
elastic spring element.
5. Assemble the Element Equations to Obtain the Global

or Total Equations and Introduce Boundary
Conditions - We then show how the total stiffness matrix
for the problem can be obtained by superimposing the
stiffness matrices of the individual elements in a direct
manner.
The term direct stiffness method evolved in reference to

this method.
6. Solve for the Unknown Degrees of Freedom (or

Generalized Displacements) - Solve for the nodal
displacements.
7. Solve for the Element Strains and Stresses - The

reactions and internal forces association with the bar
element.
8. Interpret the Results

1. Select Element Type - Consider the linear spring shown

below. The spring is of length L and is subjected to a
nodal tensile force, T directed along the x-axis.
f1x f2x
Note: Assumed sign conventions
2. Select a Displacement Function - A displacement

function u(x) is assumed.
u  a1  a2 x
In general, the number of coefficients in the displacement

function is equal to the total number of degrees of freedom
associated with the element. We can write the
displacement function in matrix forms as:
a 
u  1 x 1 x 2  1 
a2 2 x 1

We can express u as a function of the nodal displacements
ui by evaluating u at each node and solving for a1 and a2.
u( x  0)  u1  a1 

 Boundary Conditions
u( x  L )  u2  a2L  a1 

Solving for a2:
u2  u1
a2 
L
Substituting a1 and a2 into u gives:
 u  u1   x x
u  2  x  u1   1   u1    u2
 L   L L
In matrix form:  x x   u1 
u   1    
 L  L  u2 
Or in another form:
u 
u  N1 N2   1 
u2 
Where N1 and N2 are defined as:

x x
N1  1  N2 
L L
The functions Ni are called interpolation functions
because they describe how the assumed displacement
function varies over the domain of the element. In this case
the interpolation functions are linear.
u  a1  a2 x
 x x   u1 
u   1    
 L  L  u2 
u 
u  N1 N2   1 
u2 
x
N1  1 
L
N1  N2  1
x
N2 
L
3. Define the Strain/Displacement and Stress/Strain

Relationships - Tensile forces produce a total elongation
(deformation)  of the spring. For linear springs, the force
T and the displacement u are related by Hooke’s law:
f1x f2 x
T T
T  k
where deformation of the spring  is given as:

f1x  T
  u(L )  u(0)   u2  u1
f2 x  T
4. Step 4 - Derive the Element Stiffness Matrix and

Equations - We can now derive the spring element
stiffness matrix as follows:
Rewrite the forces in terms of the nodal displacements:
T  f1x  k  u2  u1   f1x  k  u1  u2 
T  f2 x  k  u2  u1   f2 x  k  u1  u2 
We can write the last two force-displacement relationships

in matrix form as:
 f1x   k k   u1 
   
f2 x   k k  u2 

This formulation is valid as long as the spring deforms
along the x axis. The coefficient matrix of the above
equation is called the local stiffness matrix k:
 k k 
k 
 k k 
5. Step 4 - Assemble the Element Equations

The global stiffness matrix and the global force vector
are assembled using the nodal force equilibrium equations,
and force/deformation and compatibility equations.
N N
K  K    k ( e ) F  F    f ( e )
e 1 e 1
where k and f are the element stiffness and force matrices

expressed in global coordinates.
6. Step 6 - Solve for the Nodal Displacements
Solve the displacements by imposing the boundary

conditions and solving the following set of equations:
F   K d  F  Kd
7. Step 7 - Solve for the Element Forces
Once the displacements are found, the forces in each

element may be calculated from:
T  k  k  u2  u1 
The Stiffness Method – Spring Example 1
Consider the following two-spring system shown below:
where the element axis x coincides with the global axis x.

For element 1:  f1x   k1 k1   u1 
   
f3 x   k1 k1  u3 
For element 2: f3 x   k 2 k 2  u3 

   
f2 x   k 2 k 2  u2 
Both continuity and compatibility require that both elements

remain connected at node 3. Element number
u3(1)  u3(2)
We can write the nodal equilibrium equation at each node as:
F1x  f1x (1) F3 x  f3 x (1)  f3 x (2) F2 x  f2 x (2)
Therefore the force-displacement equations for this spring

system are:
F1x  k1u1  k1u3 F2 x  k 2u3  k2u2
Element 1 Element 2
F3 x   k1u1  k1u3    k2u3  k 2u2 
In matrix form the above equations are:

 F1x   k1 0 k1   u1 
    
F2 x    0 k2 k 2  u2  F  Kd

F   k k k1  k 2  u3 
 3x   1 2
where F is the global nodal force vector, d is called the

global nodal displacement vector, and K is called the
global stiffness matrix.
Assembling the Total Stiffness Matrix by Superposition
Consider the spring system defined in the last example:
The elemental stiffness matrices may be written for each

element.
For element 1: For element 2:
u1 u3 u3 u2
 k k1  u1  k  k 2  u3
k (1)   1  k (2)   2 
 k1 k1  u3  k 2 k 2  u2
Write the stiffness matrix in global format for element 1 as

follows:
 1 0 1  u1  f1x 
(1)
   
k1  0 0 0  u2   f2(1)x 
 
 1 0 1  u3  f3(1)x 
For element 2:
0 0 0   u1  f1x 
(1)
   
k 2 0 1 1 u2   f2(1)x 
 
0 1 1  u3  f3(1)x 
Apply the force equilibrium equations at each node.
f1(1)
x
  0   F1x 
   (2)   
 0   f2 x   F2 x 
f (1)  f (2)  F 
 3x   3x   3x 
The above equations give:
 k1 0 k1   u1   F1x 
 0    
k2 k 2  u2   F2 x 
 
 k1 k 2 k1  k2  u3  F3 x 
To avoid the expansion of the each elemental stiffness

matrix, we can use a more direct, shortcut form of the
stiffness matrix.
u1 u3 u3 u 2
 k k1  u1  k  k 2  u3
k (1)   1  k (2)   2 
 k1 k1  u3   k 2 k 2  u2
The global stiffness matrix may be constructed by directly
adding terms associated with the degrees of freedom in k(1)
and k(2) into their corresponding locations in the K as
follows: u u u
1 2 3
 k1 0 k1  u1
K   0 k2  k 2  u2
 
 k1 k2 k1  k2  u3
Boundary Conditions
In order to solve the equations defined by the global

stiffness matrix, we must apply some form of constraints or
supports or the structure will be free to move as a rigid body.
Boundary conditions are of two general types:

1. homogeneous boundary conditions (the most common)
occur at locations that are completely prevented from
movement;
2. nonhomogeneous boundary conditions occur where finite
non-zero values of displacement are specified, such as the
settlement of a support.
Consider the equations we developed for the two-spring

system. We will consider node 1 to be fixed u1 = 0. The
equations describing the elongation of the spring system
become:
 k1 0 k1   0   F1x 
 0    
k2 k 2  u2   F2 x 
 
 k1 k 2 k1  k 2  u3  F3 x 
Expanding the matrix equations gives:
F1x  k1u3
F2 x  k2u3  k2u2 

 Solve for u2 and u3
F3 x  k 2u2   k1  k 2  u3 
The second and third equation may be written in matrix form

as:
 k2 k 2  u2  F2 x 
 k k  k  u   F 
 2 1 2 3  3x 
Once we have solved the above equations for the unknown

nodal displacements, we can use the first equation in the
original matrix to find the support reaction.
F1x  k1u3
For homogeneous boundary conditions, we can delete the

row and column corresponding to the zero-displacement
degrees-of-freedom.
Let’s again look at the equations we developed for the two-

spring system.
However, this time we will consider a nonhomogeneous
boundary condition at node 1: u1 = .
The equations describing the elongation of the spring
system become:
 k1 0 k1      F1x 
 0    
k2 k 2  u2   F2 x 
 
   
 k1 k 2 k1  k 2  u3  F3 x 
Expanding the matrix equations gives:
F1x  k1  k1u3 F2 x  k2u3  k2u2
F3 x  k1  k1u3  k2u3  k2u2

By considering the second and third equations because

they have known nodal forces we get:
F2 x  k2u3  k2u2 F3 x  k1  k1u3  k2u3  k2u2
 k2 k 2  u2   F2 x 
 k k  k  u   F  k  
 2 1 2 3  3x 1 
For nonhomogeneous boundary conditions, we must transfer

the terms from the stiffness matrix to the right-hand-side force
vector before solving for the unknown displacements.
Once we have solved the above equations for the unknown

nodal displacements, we can use the first equation in the
original matrix to find the support reaction.
F1x  k1  k1u3

Consider the following three-spring system:
The elemental stiffness matrices for each element are:

1 3 3 4
 1 1 1  1 1
k (1)  1000 
3
 k (2)  2000  
 1 1 3
 1 1 4
4 2
 1 1 4
k (3)  3000  
 1 1 2
Using the concept of superposition (the direct stiffness

method), the global stiffness matrix is:
Element 1
 1000 0 1000 0 
 0 3000 0 3000  Element 3
K 
 1000 0 3000 2000 
 
 0 3000 2000 5000 
Element 2
The global force-displacement equations are:
 1000 0 1000 0   u1   F1x 

 0 3000 0 3000  u2  F2 x 
     
 1000 0 3000 2000  u3  F3 x 
 
 0 3000 2000 5000  u4  F4 x 
We have homogeneous boundary conditions at

nodes 1 and 2 (u1 = 0 and u2 = 0).
The global force-displacement equations reduce to:
 1000 0 1000 0   u1   F1x 

 0 3000 0 3000  u2  F2 x 
     
 1000 0 3000 2000  u3  F3 x 
 
 0 3000 2000 5000  u4  F4 x 
Substituting for the known force at node 4 (F4x = 5,000 lb)

gives:
 
 3000 2000  u3    0 
 2000 5000  u  5,000 
   4 
Solving for u3 and u4 gives:
10 15
u3  in u4  in
11 11
To obtain the global forces, substitute the displacement in

the force-displacement equations.
 F1x   1000 0 1000 0  0 
F   0 3000   0 
 2x   3000 0
 
 
F
 3x   1000 0 3000 2000  10 11
F4 x   0 
3000 2000 5000  15 11
Solving for the forces gives:

10,000 45,000
F1x   lb F2 x   lb
11 11
55,000
F3 x  0 F4 x  lb  5,000lb
11
Next, use the local element equations to obtain the force in

each spring.
For element 1:  f   1000 1000   0 
 
1x
 10 
f3 x   1000 1000   11
The local forces are: 10,000 10,000

f1x   lb f3 x  lb
11 11
A free-body diagram of the spring element 1 is shown below.


each spring.
For element 2: f3 x   2000 2000  10 11
   
f4 x   2000 2000  15 11

f3 x   lb f4 x  lb
11 11

each spring.
For element 3: f   3000 3000  15 
 
4x 11
  
f2 x   3000 3000   0 

f4 x  lb f2 x   lb
11 11
Consider the following four-spring system:
The spring constant k = 200 kN/m and the displacement

 = 20 mm.
Therefore, the elemental stiffness matrices are:
 1 1
k (1)  k (2)  k (3)  k ( 4)  200   kN / m
 1 1
Using superposition (the direct stiffness method), the global

stiffness matrix is:
Element 1 Element 2 Element 3 Element 4
 200 200 0 0 0
 200 400 200 0 0
 
K 0 200 400 200 0
 
 0 0 200 400 200 
 0 0 0 200 200 

 200 200 0 0 0   u1   F1x 
 200 400 200 0 0  u2  F2 x 
    
 0 200 400 200 0  u3   F3 x 
 
 0 0 200 400 200  u4  F4 x 
   
 0 0 0 200 200  u5  F5 x 
Applying the boundary conditions (u1 = 0 and u5 = 20 mm)

and the known forces (F2x, F3x, and F4x equal to zero) gives:
 400 200 0 0   u2   0 
 200 400 200 0   u3   0 
   
 0 200 400 200   u4   0 
 
 0 0 200 200  0.02  F5 x 

Rearranging the first three equations gives:
 400 200 0  u2  0 

 200 400 200  u   0 
  3  
 0 200 400  u4  4 
Solving for u2, u3, and u4 gives:

u2  0.005 m u3  0.01m u4  0.015 m
Solving for the forces F1x and F5x gives:
F1x  200(0.005)  1.0kN
F5 x  200(0.015)  200(0.02)  1.0kN


each spring.
For element 1:  f1x   200 200   0 
   
f2 x   200 200  0.005 
f1x  1.0kN f2 x  1.0kN
For element 2: f2 x   200 200  0.005 

   
f3 x   200 200   0.01 
f2 x  1.0kN f3 x  1.0kN

each spring.
For element 3: f3 x   200 200   0.01 
   
f4 x   200 200  0.015 
f3 x  1.0kN f4 x  1.0kN
For element 4: f4 x   200 200  0.015 

   
f5 x   200 200   0.02 
f4 x  1.0kN f5 x  1.0kN
Consider the following spring system:
The boundary conditions are: u1  u3  u4  0
The compatibility condition at node 2 is:
u2(1)  u2(2)  u2(3)  u2
Using the direct stiffness method: the elemental stiffness

matrices for each element are:
1 2 2 3 2 4
 k k1   k k 2  2  k k3 
k (2)   2
1 2
k (1)   1   k (3)   3 
 k1 k1  2  k 2 k 2  3
 k 3 k3  4
Using the concept of superposition (the direct stiffness

method), the global stiffness matrix is:
1 2 3 4
 
 k1 k1 0 0  1
 k k  k 2  k 3 k k 3  2
K 1 1 2

 0 k 2 k2 0  3
 0 k3 0 k3  4
 
Applying the boundary conditions (u1 = u3 = u4 = 0) the

stiffness matrix becomes:
1 2 3 4
 
 k1 k1 0 0  1
 k k  k 2  k 3 k 2 k 3  2
K 1 1 
 0 k 2 k2 0  3
 0 k 3 0 k3  4
 
Applying the known forces (F2x = P) gives:
P   k1  k 2  k3  u2
P
Solving the equation gives: u2 
k1  k 2  k3
Solving for the forces gives:
F1x  k1u2 F3 x  k2u2 F4 x  k3u2

Potential Energy Approach to Derive Spring

Element Equations
One of the alternative methods often used to derive the
element equations and the stiffness matrix for an element is
based on the principle of minimum potential energy.
This method has the advantage of being more general than

the methods involving nodal and element equilibrium
equations, along with the stress/strain law for the element.
The principle of minimum potential energy is more adaptable

for the determination of element equations for complicated
elements (those with large numbers of degrees of freedom)
such as the plane stress/strain element, the axisymmetric
stress element, the plate bending element, and the three-
dimensional solid stress element.
Total Potential Energy
The total potential energy p is defined as the sum of the

internal strain energy U and the potential energy of the
external forces :
p  U  
Strain energy is the capacity of the internal forces (or
stresses) to do work through deformations (strains) in the
structure.
The potential energy of the external forces  is the capacity

of forces such as body forces, surface traction forces, and
applied nodal forces to do work through the deformation of
the structure.
Recall the force-displacement relationship for a linear

spring:
F  kx
The differential internal work (or strain energy) dU in the
spring is the internal force multiplied by the change in
displacement which the force moves through:
dU  Fdx   kx  dx

x
1 2
The total strain energy is: U   dU    kx  dx  kx
L 0
2
The strain energy is the area under the force-displacement

curve. The potential energy of the external forces is the
work done by the external forces:
  Fx

1 2
Therefore, the total potential energy is:  p  kx  Fx
2
The concept of a stationary value of a function G is shown
below:
dG
0
dx
The function G is expressed in terms of x.

To find a value of x yielding a stationary value of G(x), we
use differential calculus to differentiate G with respect to x
and set the expression equal to zero.
We can replace G with the total potential energy p and the

coordinate x with a discrete value di. To minimize p we first
take the variation of p (we will not cover the details of
variational calculus):
 p  p  p
 p   d1   d 2  ...   dn
d1 d 2 d n
The principle states that equilibrium exist when the di define

a structure state such that p = 0 for arbitrary admissible
variations di from the equilibrium state.
To satisfy p = 0, all coefficients associated with di must be

zero independently, therefore:
 p  p
0 i  1, 2,, n or 0
d i  d 
Total Potential Energy – Example 5
Consider the following linear-elastic spring system subjected

to a force of 1,000 lb.
Evaluate the potential energy for various displacement

values and show that the minimum potential energy also
corresponds to the equilibrium position of the spring.
The total potential energy is defined as the sum of the

p  U   U
1 2
kx   Fx
2
 p
The variation of p with respect to x is:  p  x  0
x
 p
Since x is arbitrary and might not be zero, then: 0
x
Using our express for p, we get:

1 2 1
p  kx  Fx  500( lb in ) x 2  1,000lb  x
2 2
 p
 0  500 x  1,000 x  2.0 in
x
If we had plotted the total
potential energy function p
for various values of
deformation we would get:
Let’s derive the spring element equations and stiffness

matrix using the principal of minimum potential energy.
Consider the linear spring subjected to nodal forces shown
below:
The total potential energy p

1
p  k  u2  u1   f1x u1  f2 x u2
2
2
Expanding the above express gives:
p 
1
2
 
k u22  2u1u2  u12  f1x u1  f2 x u2

below:
Recall:  p  p
0 i  1, 2,, n or 0
d i  d 
Therefore:
 p k
  2u2  2u1   f1x  0
u1 2
 p k
  2u2  2u1   f2 x  0
u2 2

below:
Therefore: k  u1  u2   f1x
k  u1  u2   f2x
In matrix form the equations are:
 f1x   k k   u1 
   
f2 x   k k  u2 
Obtain the total potential energy of the spring system shown

below and find its minimum value.
The potential energy p for element 1 is:
1
 p(1)  k1  u3  u1   f1x u1  f3 x u3
2

1
 p(2)  k2  u4  u3   f3 x u3  f4 x u4
2
Obtain the total potential energy of the spring system shown

below and find its minimum value.

1
 p(3)  k3  u2  u4   f2 x u2  f4 x u4
2
The total potential energy p for the spring system is:

3
 p    p( e )
e 1
Minimizing the total potential energy p:

 p
 0  k1u3  k1u1  f1x (1)
u1
 p
 0  k3u2  k3u4  f2 x (3)
u2
 p
 0  k1u3  k1u1  k 2u4  k 2u3  f3 x (1)  f3 x (2)
u3
 p
 0  k 2u4  k 2u3  k3u2  k3u4  f4 x (2)  f4 x (3)
u4
In matrix form:  k1 0 k1 0   u1   f1x 

 0 k  
k3 u2     
 3 0
   
f2 x 
(2) 
 k1 0 k1  k 2 k 2  u3   f3 x  f3 x 
(1)
 
 0 k3 k 2 k 2  k3  u4  f4 x  f4 x 
(2) (3)
Using the following force equilibrium equations:

f1x (1)  F1x
f2 x (3)  F2 x
f3 x (1)  f3 x (2)  F3 x
f4 x (2)  f4 x (3)  F4 x

 1000 0 1000 0   u1   F1x 
 0 3000 0 3000  u2  F2 x 
     
 1000 0 3000 2000  u3  F3 x 
 
 0 3000 2000 5000  u4  F4 x 
The above equations are identical to those we obtained

through the direct stiffness method.
Homework Problems
1. Do problems 2.4, 2.10, and 2.22 on pages 66 - 71 in

your textbook “A First Course in the Finite Element
Method” by D. Logan.
End of Chapter 2
CIVL 7/8117 Chapter 3a - Development of Truss Equations 1/80
Chapter 3a – Development of Truss Equations
Learning Objectives
• To derive the stiffness matrix for a bar element.
• To illustrate how to solve a bar assemblage by the direct
stiffness method.
• To introduce guidelines for selecting displacement
functions.
• To describe the concept of transformation of vectors in
two different coordinate systems in the plane.
• To derive the stiffness matrix for a bar arbitrarily oriented
in the plane.
• To demonstrate how to compute stress for a bar in the
plane.
• To show how to solve a plane truss problem.
• To develop the transformation matrix in three-
dimensional space and show how to use it to derive the
stiffness matrix for a bar arbitrarily oriented in space.
• To demonstrate the solution of space trusses.
Development of Truss Equations

Having set forth the foundation on which the direct stiffness
method is based, we will now derive the stiffness matrix for
a linear-elastic bar (or truss) element using the general
steps outlined in Chapter 2.
We will include the introduction of both a local coordinate

system, chosen with the element in mind, and a global or
reference coordinate system, chosen to be convenient (for
numerical purposes) with respect to the overall structure.
We will also discuss the transformation of a vector from the

local coordinate system to the global coordinate system,
using the concept of transformation matrices to express
the stiffness matrix of an arbitrarily oriented bar element in
terms of the global system.

Next we will describe how to handle inclined, or skewed,
supports.
We will then extend the stiffness method to include space

trusses.
We will develop the transformation matrix in three-dimensional

space and analyze a space truss.
We will then use the principle of minimum potential energy

and apply it to the bar element equations.
Finally, we will apply Galerkin’s residual method to derive the

bar element equations.



Stiffness Matrix for a Bar Element

Consider the derivation of the stiffness matrix for the linear-
elastic, constant cross-sectional area (prismatic) bar
element show below.
This application is directly applicable to the solution of pin-

connected truss problems.

element show below.
where T is the tensile force directed along the axis at

nodes 1 and 2, x is the local coordinate system directed
along the length of the bar.

element show below.
The bar element has a constant cross-section A, an initial

length L, and modulus of elasticity E.
The nodal degrees of freedom are the local axial

displacements u1 and u2 at the ends of the bar.

du
The strain-displacement relationship is:   E 
dx
From equilibrium of forces, assuming no distributed loads
acting on the bar, we get:
A x  T  constant
Combining the above equations gives:
du
AE  T  constant
dx
Taking the derivative of the above equation with respect to the
local coordinate x gives:
d  du 
 0
dx 
AE
dx 

The following assumptions are considered in deriving the bar
element stiffness matrix:
1. The bar cannot sustain shear force: f1y  f2 y  0
2. Any effect of transverse displacement is ignored.
3. Hooke’s law applies; stress is related to strain:  x  E x


Step 1 - Select Element Type
We will consider the linear bar element shown below.

A linear displacement function u is assumed: u  a1  a2 x
The number of coefficients in the displacement function, ai, is

equal to the total number of degrees of freedom associated
with the element.
Applying the boundary conditions and solving for the unknown
coefficients gives:
 u  u1   x x   u1 
u  2  x  u1 u   1    
 L   L  L  u2 

u 
Or in another form: u  N1 N2   1 
u2 
where N1 and N2 are the interpolation functions gives as:

x x
N1  1  N2 
L L
The linear displacement
function u plotted over the
length of the bar element is
shown here.

du u2  u1
The stress-displacement relationship is: x  
dx L
We can now derive the element stiffness matrix as follows:

T  A x
Substituting the stress-displacement relationship into the
above equation gives:
 u  u1 
T  AE  2 
 L 

The nodal force sign convention, defined in element figure, is:

f1x  T f2 x  T
therefore, u u   u  u1 
f1x  AE  1 2  f2 x  AE  2 
 L   L 
Writing the above equations in matrix form gives:
 f1x  AE  1 1  u1 
   
f2 x  L  1 1 u2 
Notice that AE/L for a bar element is analogous to the spring

constant k for a spring element.

The global stiffness matrix and the global force vector are
assembled using the nodal force equilibrium equations, and
force/deformation and compatibility equations.
n n
K  K    k ( e ) F  F    f ( e )
e 1 e 1
Where k and f are the element stiffness and force matrices

expressed in global coordinates.

Step 6 - Solve for the Nodal Displacements
Solve the displacements by imposing the boundary conditions
and solving the following set of equations:
F  Ku
Step 7 - Solve for the Element Forces

Once the displacements are found, the stress and strain in
each element may be calculated from:
du u2  u1
x    x  E x
dx L

Example 1 - Bar Problem
Consider the following three-bar system shown below. Assume
for elements 1 and 2: A = 1 in2 and E = 30 (106) psi and for
element 3: A = 2 in2 and E = 15 (106) psi.
Determine: (a) the global stiffness matrix, (b) the displacement

of nodes 2 and 3, and (c) the reactions at nodes 1 and 4.

For elements 1 and 2:
1 2 node numbers for element 1
1  30  10  6
1 1 lb 6  1 1 lb
k (1)  k (2)   1 1 in  10  1 1 in
30    
For element 3:
 2  15  106   1 1 lb 6  1 1 lb

k (3)
  1 1 in  10  1 1 in
30    
As before, the numbers above the matrices indicate the
displacements associated with the matrix.

Assembling the global stiffness matrix by the direct stiffness
methods gives: E1 E2 E3
 1 1 0 0 
 1 2 1 0 
K  106  
 0 1 2 1
 
 0 0 1 1
Relating global nodal forces related to global nodal
displacements gives:
 F1x   1 1 0 0   u1 
F   2 1 0  u2 
 2x  6  1  
   10
F3 x   0 1 2 1 u3 
F4 x   
 0 0 1 1 u4 

Example 1 – Bar Problem
The boundary conditions are: u1  u4  0
 F1x   1 1 0 0   0 
F   2 1 0  u2 
 2x  6  1  
   10
F3 x   0 1 2 1 u3 
F4 x   
 0 0 1 1  0 
Applying the boundary conditions and the known forces

(F2x = 3,000 lb) gives:
3,000  6  2 1  2 
u
   10    
 0   1 2  u3 

Example 1 – Bar Problem
Solving for u2 and u3 gives: u2  0.002 in

u3  0.001 in
The global nodal forces are calculated as:

 F1x   1 1 0 0   0  2,000 
F   2 1 0  0.002  3,000 
 2x  6 1
   10     lb
F3 x   0 1 2 1  0.001  0 
 
F4 x   0 0 1 1  0   1,000 

Selecting Approximation Functions for Displacements
Consider the following guidelines, as they relate to the one-
dimensional bar element, when selecting a displacement
function.
1. Common approximation functions are usually

polynomials.
2. The approximation function should be continuous within
the bar element.

function.
3. The approximating function should provide interelement

continuity for all degrees of freedom at each node for
discrete line elements, and along common boundary lines
and surfaces for two- and three-dimensional elements.

function.
For the bar element, we must ensure that nodes common to
two or more elements remain common to these elements
upon deformation and thus prevent overlaps or voids
between elements.
The linear function is then called a conforming (or

compatible) function for the bar element because it
ensures both the satisfaction of continuity between adjacent
elements and of continuity within the element.

function.
4. The approximation function should allow for rigid-body
displacement and for a state of constant strain within the
element.
Completeness of a function is necessary for

convergence to the exact answer, for instance, for
displacements and stresses.

The interpolation function must allow for a rigid-body
displacement, that means the function must be capable of
yielding a constant value.
Consider the follow situation: u  a1 a1  u1  u2
Therefore: u  N1u1  N2u2   N1  N2  a1
Since u = a1 then: u  a1   N1  N2  a1
This means that: N1  N2  1
The displacement interpolation function must add to unity at

every point within the element so the it will yield a constant
value when a rigid-body displacement occurs.

Transformation of Vectors in Two Dimensions
In many problems it is convenient to introduce both local and
global (or reference) coordinates.
Local coordinates are always chosen to conveniently

represent the individual element.
Global coordinates are chosen to be convenient for the whole

structure.

Given the nodal displacement of an element, represented by
the vector d in the figure below, we want to relate the
components of this vector in one coordinate system to
components in another.

Let’s consider that d does not coincide with either the local or
global axes. In this case, we want to relate global
displacement components to local ones. In so doing, we will
develop a transformation matrix that will subsequently be
used to develop the global stiffness matrix for a bar element.

We define the angle to be positive when measured
counterclockwise from x to x’. We can express vector
displacement d in both global and local coordinates by:
d  u1i  v1 j  u1i  v1 j

Consider the following diagram:
Using vector addition: a  b  i
Using the law of cosines, we get: | a || i | cos  | a | cos 
Similarly: | b || i | sin | b | sin


The vector a is in the i direction and b is in the j direction,

therefore:
a | a | i   cos  i b | b |   j    sin   j 

The vector i can be rewritten as: i  cos  i  sin j

The vector j can be rewritten as: j  sin i  cos  j
Therefore, the displacement vector is:

u1i  v1 j  u1  cos  i  sin j   v1  sin i  cos  j   u1i  v1 j

Combining like coefficients of the local unit vectors gives:

u1 cos  v1 sin  u1
u1   C S  u1  C  cos 
   
u1 sin  v1 cos  v1 v1   S C  v1  S  sin

The previous equation relates the global displacement d to
the d local displacements
 C S
The matrix is called the transformation matrix.  
 S C 
The figure below shows u expressed in terms of the global

coordinates x and y.
u   Cu  Sv

Example 2 - Bar Element Problem
The global nodal displacement at node 2 is u2 = 0.1 in and
v2 = 0.2 in for the bar element shown below. Determine the
local displacement.
Using the following expression we just derived, we get:
u   Cu  Sv
u2  cos 60o (0.1)  sin 60o (0.2)  0.223 in

Global Stiffness Matrix
We will now use the transformation relationship developed
above to obtain the global stiffness matrix for a bar element.

We known that for a bar element in local coordinates we have:
 f1x  AE  1 1  u1 

    f   kd
f2x  L  1 1 u2 
We want to relate the global element forces f to the global

displacements d for a bar element with an arbitrary
orientation.
 f1x   u1 
f   
 1y   v1 
 k  f = kd
f2 x  u2 
f2 y  v 2 

Using the relationship between local and global components,
we can develop the global stiffness matrix.
We already know the transformation relationships:
u1  u1 cos   v1 sin u2  u2 cos   v 2 sin
Combining both expressions for the two local degrees-of-

freedom, in matrix form, we get:
 u1  d = T *d
 u1  C S 0 0   v1 
   
u2   0 0 C S  u2  C S 0 0 
T*   
v 2   0 0 C S

A similar expression for the force transformation can be
developed.
 f1x 
 
 f1x  C S 0 0   f1y 
     f  T*f
f2x   0 0 C S  f2 x 
f2 y 
Substituting the global force expression into element force

equation gives:
f  = kd  T * f  kd
Substituting the transformation between local and global
d = T * d  T * f  kT *d

The matrix T* is not a square matrix so we cannot invert it.
Let’s expand the relationship between local and global
displacement.
 u1   C S 00   u1 
 v    S 0   v1 
 1  C 0
 
  d = Td
u2   0 0 C S  u2 
v 2   0 
0 S C  v 2 
where T is:
C S 0 0
 S C 0 0
T 
0 0 C S
 
0 0 S C 

We can write a similar expression for the relationship between
local and global forces.
 f1x   C S 00   f1x 
f     
 1y  S C 0 0   f1y 
     f  = Tf
f
 2x   0 0 C S  f2 x 
 
f2y   0 0 S C  f2 y 
Therefore our original local coordinate force-displacement

expression
 f1x  AE  1 1  u1 

    f  = kd
f2x  L  1 1 u2 

May be expanded:  f1x  1 0 1 0   u1 
f  
 1y  AE
0 0 0 0  v1 
     
f2 x  L  1 0 1 0  u2 
f2y   
0 0 0 0  v 2 

f   kd  Tf  kTd
Multiply both side by T -1 we get: f  T -1kTd
where T-1 is the inverse of T. It can be shown that: T 1  T T

The global force-displacement equations become: f = T TkTd
Where the global stiffness matrix k is: k  T TkT
Expanding the above transformation gives:

 C 2 CS C 2 CS 
 2
CS S 
2
AE  CS S
k
L  C 2 CS C 2 CS 
 
 CS S
2
CS S2 
We can assemble the total stiffness matrix by using the above

element stiffness matrix and the direct stiffness method.
n n
K  K    k ( e ) F  F    f ( e ) F  Kd
e 1 e 1

Local forces can be computed as:
 f1x  1 0 1 0   u1  1 0 1 0   C S 00   u1 
f  
 1y  AE
0 0 0 0   v1  AE  0 0 0 0   S C 0 0   v1 
          
f2 x  L  1 0 1 0  u2  L  1 0 1 0  0 0 C S  u2 
    
f2y  0 0 0 0  v 2  0 0 0 0  0 0 S C  v 2 
 f1x   Cu1  Sv1  Cu2  Sv 2 

f    
 1y  AE  0

  
f
 2 x L  Cu1  Sv 1  Cu 2  Sv 2 
 
f2y   0 

For the bar element shown below, evaluate the global stiffness
matrix. Assume the cross-sectional area is 2 in2, the length
is 60 in, and the E is 30 x 106 psi.
 C 2 CS C 2 CS 
 2
CS S 
2
AE  CS S
k
L  C 2 CS C 2 CS 
 
 CS S
2
CS S2 
3 1
Therefore: C  cos30o  S  sin30o 
2 2

The global elemental stiffness matrix is:
 3 3 
 3  3
4 4 4 4
 
k
 
(2 in 2 ) 30  106 psi 

3
4
1
4
 3
4
1 
4 
60 in  3  3 3 3 
 4 4 4 4 
 3 3 
  4
1
4 4
1 
4 
Simplifying the global elemental stiffness matrix is:

 0.750 0.433 0.750 0.433 
 0.433 0.250 0.433 0.250  lb
k  106   in
 0.750 0.433 0.750 0.433 
 
 0.433 0.250 0.433 0.250 

Computation of Stress for a Bar in the x-y Plane
For a bar element the local forces are related to the local
displacements by:
 f1x  AE  1 1  u1 
   
f2x  L  1 1 u2 
The force-displacement equation for f2x is:

AE u 
f2x   1 1  1 
L u2 
The stress in terms of global displacement is:
 u1 
 
C S 0 0  v1  E
     Cu1  Sv1  Cu2  Sv 2 
E
   1 1 
L  0 0 C S  u2  L
v 2 

For the bar element shown below, determine the axial stress.
Assume the cross-sectional area is 4 x 10-4 m2, the length is
2 m, and the E is 210 GPa.
The global displacements are known as

u1 = 0.25 mm, v1 = 0, u2 = 0.5 mm,
and v2 = 0.75 mm.
E

L
 Cu1  Sv1  Cu2  Sv 2 
210  106  1 3 1 3 
   (0.25)  (0)  (0.5)  (0.75) KN m
2  2 4 2 4 
  81.32  103 kN m  81.32 MPa
2

Solution of a Plane Truss
We will now illustrate the use of equations developed above
along with the direct stiffness method to solve the following
plane truss example problems.
A plane truss is a structure composed of bar elements all lying

in a common plane that are connected together by
frictionless pins.
The plane truss also must have loads acting only in the
common plane.

Example 5 - Plane Truss Problem
The plane truss shown below is composed of three bars
subjected to a downward force of 10 kips at node 1. Assume
the cross-sectional area A = 2 in2 and E is 30 x 106 psi for all
elements.
Determine the x and y displacement at node 1 and stresses in
each element.

Element Node 1 Node 2  C S

1 1 2 90o 0 1
2 1 3 45o 0.707 0.707
3 1 4 0o 1 0

 C 2 CS C 2 CS 
Example 5 - Plane Truss Problem 
AE  CS S
2 2
CS S 
k
L  C 2 CS C 2 CS 
 
The global elemental stiffness matrix are:  CS S
2
CS S2 
u1 v1 u2 v2
0 0 0 0
 1 lb
(2 in )(30  10 psi ) 0
2 6
1 0
 in
element 1: C 0 S 1  k (1) 
120 in 0 0 0 0
 
0 1 0 1
u1 v1 u3 v3
1 1 1 1
 1 1 lb
(2 in 2 )(30  106 psi )  1 1
 in
element 2: C 2
S 2
 k(2) 
2 2
240 2 in  1 1 1 1
 
 1 1 1 1
u1 v1 u4 v4
1 0 1 0
 0  lb
(2 in 2 )(30  106 psi )  0 0 0
 in
element 3: C 1 S  0  k(3) 
120 in  1 0 1 0
 
0 0 0 0

The total global stiffness matrix is:
 1.354 0.354 0 0 0.354 0.354 1 0  element 1
 0.354 1.354 0 1 0.354 0.354 0 0 
 0 0 0 0 0 0 0 0 element 2
K  5  10  0.354
5 0 1 0 1 0 0 0 0  lb
in
element 3
 0.354 0 0 0.354 0.354 0 0

 0.354
1
0.354 0 0 0.354 0.354 0 0

 0 
0 0 0 0 0 1 0
 0 0 0 0 0 0 0
The total global force-displacement equations are:

 0   1.354 0.354 0 0 0.354 0.354 1 0  u1 
10, 000  0.354 1.354 0 1 0.354 0.354 0 0  v 1 
 F2 x   0 0 0 0 0 0 0 0  0 
 F2 y    0 
 F3 x   5  10  0.354
5 0 1 0 1 0 0 0 0
 F3 y   0.354 0 0 0.354 0.354 0 0

 0 
 F   0.354
1
0.354 0 0 0.354 0.354 0 0
  00 
 0  0 
0 0 0 0 0 1 0
 F 
4x
  
 4y 
0 0 0 0 0 0 0

Applying the boundary conditions for the truss, the above
equations reduce to:
 0   1.354 0.354 0 0 0.354 0.354 1 0  u1 

10, 000  0.354 1.354 0 1 0.354 0.354 0 0  v 1 
 F2 x   0 0 0 0 0 0 0 0  0 
 F2 y  0  0 
 F3 x   5  10  0.354
5 0 1 0 1 0 0 0
 
 F3 y   0.354 0 0 0.354 0.354 0 0
 0 
 F   0.354
1
0.354 0 0 0.354 0.354 0 0
 0
0 0 
 0  0
0 0 0 0 0 1
 F4 x   0  
 4y 
0 0 0 0 0 0

Applying the boundary conditions for the truss, the above
  5 1.354 0.354   1 
0 u
   5  10    
10,000  0.354 1.354  v1 
Solving the equations gives: u1  0.414  10 2 in

v1  1.59  10 2 in
E
The stress in an element is:   Cu 1  Sv 1  Cu 2  Sv 2 
L 
where i is the local node number


1 1 2 90o 0 1
2 1 3 45o 0.707 0.707
3 1 4 0o 1 0
E
 Cu 1  Sv 1  Cu 2  Sv 2 
L 
30  106
element 1  (1) 
120
 v1   3,965 psi
30  106
element 2  (2)  
120
(0.707)u1  (0.707)v1   1,471 psi


1 1 2 90o 0 1
2 1 3 45o 0.707 0.707
3 1 4 0o 1 0
E
 Cu 1  Sv 1  Cu 2  Sv 2 
L 
30  106
element 3  (3)

120
 u1   1,035 psi

Let’s check equilibrium at node 1:
F x  f1(2)
x cos(45) f1x
(3)
F y x sin(45) f1x 10,000 lb

 f1(2) (1)

Let’s check equilibrium at node 1:
F x  (1,471 psi )(2 in 2 )(0.707)  (1,035 psi )(2 in 2 )  0
F y  (3,965 psi )(2 in 2 )  (1,471 psi )(2 in 2 )(0.707)  10,000  0

Develop the element stiffness matrices and system equations
for the plane truss below.
Assume the stiffness of each element is constant. Use the

numbering scheme indicated. Solve the equations for the
displacements and compute the member forces. All elements
have a constant value of AE/L

Develop the element stiffness matrices and system equations
for the plane truss below.
Member Node 1 Node 2 Elemental 

Stiffness
1 1 2 k 0
2 2 3 k 3/4
3 1 3 k /2

Compute the elemental stiffness matrix for each element. The
general form of the matrix is:
 C 2 CS C 2 CS 
 2
CS S 
2
AE  CS S
k
L  C 2 CS C 2 CS 
 
 CS S
2
CS S2 

Stiffness
1 1 2 k 0
2 2 3 k 3/4
3 1 3 k /2

For element 1: u
1
v
1
u
2
v
2
 1 0 1 0  u1
 0 0 0 0 v1
k (1) k 
 1 0 1 0 u2
 
 0 0 0 0 v2

Stiffness
1 1 2 k 0
2 2 3 k 3/4
3 1 3 k /2

For element 2: u2 v2 u3 v3
 1 1 1 1 u2
 
k 1 1 1 1 v2
k (2)  
2  1 1 1 1 u3
 
 1 1 1 1 v3

Stiffness
1 1 2 k 0
2 2 3 k 3/4
3 1 3 k /2

For element 3: u1 v1 u3 v3
0 0 0 0 u1
0 1 0 1 v1
k (3) k 
0 0 0 0 u3
 
0 1 0 1 v3

Stiffness
1 1 2 k 0
2 2 3 k 3/4
3 1 3 k /2

Assemble the global stiffness matrix by superimposing the
elemental global matrices.
u1 v1 u2 v2 u3 v3
 2 0 2 0 0 0 u1 element 1
 0 2 0 0 0 2  v1 element 2
  element 3
k  2 0 3 1 1 1 u2
K  
2 0 0 1 1 1  1 v2
 0 0 1 1 1  1 u3
 
 0 2 1 1 1 3 v3

The unconstrained (no boundary conditions satisfied)
equations are:  2 0 2 0 0 0   u1   F1x 
 0    
 2 0 0 0 2  v1   F1y 
k  2 0 3 1 1 1  u2   P1 
     
2 0 0 1 1 1 1  v 2   P2 
 0 0 1 1 1 1  u3   F3 x 
    
 0 2 1 1 1 3   v 3   F3 y 
The displacement at nodes 1 and 3 are zero in both directions.

Applying these conditions to the system equations gives:
 2 0 2 0 0 0   0   F1x 
 0  
 2 0 0 0 2  0   F1y 
k  2 0 3 1 1 1  u2   P1 
     
2 0 0 1 1 1 1  v 2   P2 
 0 0 1 1 1 1  0   F3 x 
    
 0 2 1 1 1 3   0   F3 y 

Applying the boundary conditions to the system equations
gives:
k  3 1 u2   P1 
  
2  1 1  v 2   P2 
Solving this set of equations is fairly easy. The solution is:

P1  P2 P1  3P2
u2  v2 
k k

Using the force-displacement relationship the force in each
member may be computed.
Member (element) 1:  f1x   Cu1  Sv1  Cu2  Sv 2 
f   
 1y  0
 k 
C 1 S0 f
 2 x  Cu1  Sv 1  Cu 2  Sv 2
f2 y   
 0 
 P  P2 
f1x  k  Cu2   k   1    P1  P2  fy 1  0
 k 
 P  P2 
f2 x  k Cu2   k  1   P1  P2 fy 2  0
 k 

Member (element) 2: f2 x   Cu2  Sv 2  Cu3  Sv 3 
f   
 2y  
0

1 1    k
C S f3 x   Cu2  Sv 2  Cu3  Sv 3 
2 2
f3 y   
 0 
f2 x  k Cu2  Sv 2 
 P  P2   1   P1  3P2   1  
 k  1         2P2
 k   2  k   2 
f3 x  k  Cu2  Sv 2 
 P  P2   1   P1  3P2   1  
 k  1        2P2
 k   2   k  2 

Member (element) 3: f1x  0 f1y  0
f3 x  0 f3 y  0
The solution to this simple problem can be readily checked by

using simple static equilibrium equations.

Consider the two bar truss shown below.
Determine the displacement in the y direction of node 1 and

the axial force in each element.
Assume E = 210 GPa and A = 6 x 10-4 m2

The global elemental stiffness matrix for element 1 is:
3 4
cos  (1)   0 .6 sin  (1)   0. 8
5 5
 0.36 0.48 0.36 0.48 

 0.64 0.48 0.64 
210  106 (6  10 4 )  0.48
k (1) 
5  0.36 0.48 0.36 0.48 
 
 0.48 0.64 0.48 0.64 
Simplifying the above expression gives:

u1 v1 u2 v2
 0.36 0.48 0.36 0.48 

 0.48 0.64 0.48 0.64 
k (1)  25,200 
 0.36 0.48 0.36 0.48 
 
 0.48 0.64 0.48 0.64 

cos  ( 2 )  0 sin  ( 2 )  1
0 0 0 0 
 0 1
(210  106 )(6  10 4 ) 0 1
k (2) 
4 0 0 0 0
 
0 1 0 1

u1 v1 u3 v3
0 0 0 0 
0 1.25 0 1.25 
k (2)  25,200  
0 0 0 0 
 
0 1.25 0 1.25 

The total global equations are:
 F1x   0.36 0.48 0.36 0.48 0 0   u1 
F   0.48   element 1
 1y   1.89 0.48 0.64 0 1.25   v1 
F2 x   0.36 0  u2  element 2
0.48 0.36 0.48 0
   25,200   
F
 2y   0.48 0.64 0.48 0.64 0 0  v 2 
F3 x   0 0 0 0 0 0  u3 
    
 3 y 
F  0 1.25 0 0 0 1.25  v 3 
The displacement boundary conditions are:

u1   u2  v 2  u3  v 3  0

The total global equations are:
 F1x   0.36 0.48 0.36 0.48 0 0   u1 
P  
 F1y
 0.48 1.89 0.48 0.64 0 1.25  v1 
F2 x   0.36 0.48 0.36 0.48 0 0  0 
   25,200   
 0.48 0.64
F
 2y  0.48 0.64 0 0  0 
F3 x   0 0 0 0 0 0  0 
    
 3 y 
F  0 1.25 0 0 0 1.25   0 
By applying the boundary conditions the force-displacement

P  25,200(0.48  1.89v1 )

Solving the equation gives: v1  (2.1 105 )P  0.25
By substituting P = 1,000 kN and  = -0.05 m in the above

equation gives:
v1  0.0337m
The local element forces for element 1 are:

 u1  0.05 
 f1x   1 1 0.6 0.8 0 0  v1  0.0337 
   25,200    
f2 x   1 1  0 0 0.6 0.8   u2 
 v2 
The element forces are: f1x  76.6 kN f2 x  76.7 kN

Tension

The local element forces for element 2 are:
 u1  0.05 
 
 f1x   1 1 0 1 0 0  v1  0.0337 
   31,500     
f3 x   1 1 0 0 0 1  u3 
 v3 
The element forces are: f1x  1,061 kN f3 x  1,061 kN
Compression

Transformation Matrix and Stiffness Matrix for
a Bar in Three-Dimensional Space
Let’s derive the transformation matrix for the stiffness matrix for
a bar element in three-dimensional space as shown below:

The coordinates at node 1 are x1, y1, and z1, and the coor-
dinates of node 2 are x2, y2, and z2. Also, let x, y, and z be
the angles measured from the global x, y, and z axes,
respectively, to the local axis.

The three-dimensional vector representing the bar element is
gives as:
d  ui  vj  wk  u  i  v  j  w  k

Taking the dot product of the above equation with i gives:
u(i  i)  v ( j  i)  w (k  i)  u 
By the definition of the dot product we get:

x2  x1 y 2  y1 z2  z1
i  i   Cx j  i   Cy k  i   Cz
L L L
where L  ( x2  x1 )2  ( y 2  y1 )2  ( z2  z1 )2
Cx  cos x Cy  cos y Cz  cos  z

where Cx, Cy, and Cz are projections of i on to i, j, and k,
respectively.

Therefore: u   Cx u  Cy v  Czw
The transformation between local and global displacements is:
 u1 
v  d  T *d
 1

 u1  Cx Cy Cz 0 0 0   w1 
   
u 
 2  0 0 0 C x
C y C z   u2  C C C 0 0 0 
v2  T*   x y z 
   0 0 0 C x C y Cz 
w 2 

The transformation from the local to the global stiffness matrix
is: C 0  x
C 0 
 y 
C 0  AE  1 1 Cx Cy Cz 0 0 0 
k  T TkT k z   1 1  0 0 0 C C C 
 0 Cx  L   x y z
 0 Cy 
 
 0 Cz 
 Cx2 C xC y C xC z Cx2 CxCy CxCz 

 Cy2 
 C xCy Cy2 Cy Cz CxCy Cy Cz 
AE  CxCz Cz2 CxCz Cy Cz Cz 
2
C y Cz
k  
L  Cx2 CxCy CxCz Cx2 C xCy C xC z 
 C C Cy2 Cy Cz C xC y Cy2 C y Cz 
 x y 
 CxCz Cy Cz Cz2 C xC z C y Cz Cz2 

The global stiffness matrix can be written in a more convenient

form as:
 Cx 2 C xCy C x Cz 
  AE    
  CxCy Cy 2 C y Cz  k  
L   
CxCz C y Cz Cz 2 


Example 8 – Space Truss Problem
Consider the space truss shown below. The modulus of
elasticity, E = 1.2 x 106 psi for all elements. Node 1 is
constrained from movement in the y direction.
To simplify the stiffness matrices

for the three elements, we will
express each element in the
following form:
AE    
k  
L   

Consider element 1: L(1)  ( x2  x1 )2  ( y 2  y1 )2  ( z2  z1 )2
L(1)  ( 72)2  (36)2  80.5 in
72
Cx   0.89
80.5
36
Cy   0.45
80.5
Cz  0
 0.79 0.40 0
   0.40 0.20 0

 0 0 0 

Consider element 1:
u v w u v w
1 1 1 2 2 2
(0.302 in )(1.2  10 psi )  

2 6
  lb
k    
in
80.5 in 

Consider element 2: L(2)  ( x3  x1 )2  ( y 3  y1 )2  ( z3  z1 )2
L(2)  ( 72)2  (36)2  (72)2  108 in

72
Cx   0.667
108
36
Cy   0.33
108
72
Cz   0.667
108
 0.45 0.22 0.45 

   0.22 0.11 0.45 

 0.45 0.45 0.45 

Consider element 2:
u v w u v w
1 1 1 3 3 3
(0.729 in )(1.2  10 psi )  

2 6
  lb
k    
in
108 in 

Consider element 3: L(3)  ( x 4  x1 )2  ( y 4  y1 )2  ( z4  z1 )2
L(3)  ( 72)2  ( 48)2  86.5 in
72
Cx   0.833
86.5
Cy  0
48
Cz   0.550
86.5
0.69 0 0.46 
   0 0 0 

0.46 0 0.30 

Consider element 3:
u v w u v w
1 1 1 4 4 4
(0.187 in )(1.2  10 psi )  

2 6
  lb
k    
in
86.5 in 
The boundary conditions are:
u2  v 2  w 2  0
u3  v 3  w 3  0
u4  v 4  w 4  0
v1  0

Canceling the rows and the columns associated with the
boundary conditions reduces the global stiffness matrix to:
u1 w1
 9,000 2,450 
K 
 2,450 4,450 

 9,000 2,450   u1   0 
 2,450   
 4,450  w1  1,000 
Solving the equation gives:
u1  0.072 in w1  0.264 in

It can be shown, that the local forces in an element are:
 ui 
v 
 i
 fix  AE  Cx Cy Cz Cx Cy Cz  w i 
    
f jx  L  Cx Cy Cz Cx Cy Cz   u j 
v j 
 
w j 
The stress in an element is:
 ui 
v 
 i
E w i 
  Cx  Cy  Cz Cx Cy Cz   
L uj 
v j 
 
w j 

The stress in element 1 is:  0.072 
 0 
 
1.2  10 6  0.264 
 (1)  0.89 0.45 0  0.89 0.45 0 
80.5  0 
 0 
 
 (1)  955 psi  0 
The stress in element 2 is:  0.072 

 0 
 
1.2  106  0.264 
 (2)  0.667  0.33  0.667  0.667 0.33 0.667   
108  0 
 0 
 
 0 
 (2)  1,423 psi

The stress in element 3 is:  0.072 
 0 
 
1.2  106  0.264 
 (3)  0.83 0 0.55  0.83 0  0.55   
86.5  0 
 0 
 
 (3)  2,843 psi  0 

Inclined, or Skewed Supports
If a support is inclined, or skewed, at some angle  for the
global x axis, as shown below, the boundary conditions on
the displacements are not in the global x-y directions but in
the x’-y’ directions.

Inclined, or Skewed, Supports
We must transform the local boundary condition of v’3 = 0
(in local coordinates) into the global x-y system.

Therefore, the relationship between of the components of the
displacement in the local and the global coordinate systems
at node 3 is:
u '3   cos  sin   u3 
   
v '3    sin  cos   v 3 
We can rewrite the above expression as:
 cos  sin  
d '3   [t3 ]d3  t3    sin cos  

We can apply this sort of transformation to the entire
displacement vector as:
d '  [T1]d or d  [T1]T d '

Where the matrix [T1]T is:  [I ] [0] [0] 
[T1 ]  [0]
T
[I ] [0] 
 
[0] [0] [t3 ]
Both the identity matrix [I] and the matrix [t3] are 2 x 2 matrices.
The force vector can be transformed by using the same

transformation.
f '  [T1]f 
In global coordinates, the force-displacement equations are:
f   [K ]d 

Applying the skewed support transformation to both sides of
the equation gives:
[T1 ] f   [T1 ][K ] d 
By using the relationship between the local and the global

displacements, the force-displacement equations become:
f '  [T1][K ][T1]T d '
Therefore the global equations become:
 F1x   u1 
F   v 
 1y   1
 F2 x   u2 
  [T1 ][K ][T1 ]
T
  
 F2 y   v2 
F ' 3 x  u '3 
   
F '3 y  v '3 

Consider the plane truss shown below. Assume E = 210 GPa,
A = 6 x 10-4 m2 for element 1 and 2, and A = 2 (6 x 10-4)m2
for element 3.
Determine the stiffness matrix
for each element.
 C 2 CS C 2 CS 
 2
CS S 
2
AE  CS S
k
L  C 2 CS C 2 CS 
 
 CS S
2
CS S2 

 C 2 CS C 2 CS 
 2
CS S 
2
AE  CS S
k
L  C CS C
2 2
CS 
 
 CS S
2
CS S2 
cos  (1)  0 sin (1)  1

u1 v1 u2 v2
0 0 0 0
 0 1
(210  106 kN / m 2 )(6  104 m 2 ) 0 1 
k (1) 
1m 0 0 0 0
 
0 1 0 1

 C 2 CS C 2 CS 
 2
CS S 
2
AE  CS S
k
L  C CS C
2 2
CS 
 
 CS S
2
CS S2 
cos  (2)  1 sin (2)  0
u2 v2 u3 v3
1 0 1 0 
 0 0 
(210  106 kN / m 2 )(6  104 m 2 )  0 0
k (2)

1m  1 0 1 0
 
0 0 0 0

 C 2 CS C 2 CS 
 2
CS S 
2
AE  CS S
k
L  C CS C
2 2
CS 
 
 CS S
2
CS S2 
2 2
cos (3)  sin (3) 
2 2
u1 v1 u3 v3
1 1 1 1
2 
6 2 4
(210  10 kN / m )(6 2  10 m )  1 1 1 1
k (3)  
2 2m  1 1 1 1
 
 1 1 1 1

Using the direct stiffness method, the global stiffness matrix is:
 0.5 0.5 0 0  0.5  0.5
 0.5 1.5 0  1  0.5  0.5
 
 0 0 1 0 1 0
K  1,260  10 5 N
m 0 1 0 1 0

0

 0.5  0.5  1 0 1.5 0.5
 
 0.5  0.5 0 0 0.5 0.5
We must transform the global displacements into local

coordinates. Therefore the transformation [T1] is:
1 0 0 0 0 0 
0 1 0 0 0 0 
 [I ] [0] [0]   
0 
[T1 ]  [0] [0] 
0 1 0 0 0

[I ]
  
[0] [0] [t3 ] 0 0 0 1 0 0 
0 0 0 0 2
2
2 
2
  2

0 0 0 0 2
2
2

The first step in the matrix transformation to find the product of
[T1][K]. T  1K 
1 0 0 0 0 0 
 0.5 0.5 0 0 0.5 0.5 
0 0 

1 0 0 0
  0.5 1.5 0 1 0.5 0.5 

T1 K  0
 1, 260  105 N 
m 0
0
0
1
0
0
1
0
0
0 
0

 0
0
0
1
1
0
0
1
1
0
0
0
  
0 0 0 0 2 2
  0.707 0.707 0.707 0 1.414 0.707 
 
2 2
0 0 0.707 0 0.707 0
0 0 0 0  2
2
2

2
 0.5 0.5 0 0 0.5 0.5 

 0.5 1.5 0 1 0.5 0.5 
T1 K   1,260  10 m  00
5 N 0
1
1
0
0
1
1
0
0
0
 0.707 0.707 0.707 0 1.414 0.707


 0 0 0.707 0 0.707 0

The next step in the matrix transformation to find the product of
[T1][K][T1]T.
T1 K  T1 
T
0.5 0.5  
1 0 0 0 0 0 
 0.5 0.5 0 0
0
 0.5 0 1 0.5 0.5  1 0 0 0 0 

1.5
 
1 0  0 0 1 0 0 0 
T1 K T1   1,260  105 N m 
T 0 0 1 0
 0 1 0 1 0 0  0 0 0 1 0 0

 0.707 0.707 0.707 0 1.414 0.707   


  0 0 0 0 2 2
0  
2 2
 0 0 0.707 0 0.707 
0 0 0 0 2
2
2
2 
 0.5 0.5 0 0 0.707 0

 0.5 1.5 0 1 0.707 0
T1 K T1   1,260  10  0.707 

T 5N 0 0 1 0 0.707
m 0 1 0 1 0 0
 0.707 0.707 0.707 0 1.5 0.5


 0 0 0.707 0 0.5 0.5 

The displacement boundary conditions are: u1  v1  v 2  v '3  0
 F1x   0.5 0.5 0 0 0.707 0   u1

    
 F1y   0.5 1.5 0 1 0.707 0   v1
    
F
 2x   0 0 1 0 0.707 0.707   u2 
  1,260  10 N m 
5
  
F
 2y   0 1 0 1 0 0   v 2 
   0.707  
F '
 3x  0.707 0.707 0 1.5 0.5  u '3 
 
F '   0 0 0.707 0 0.5 0.5  v '3 
 3y  

By applying the boundary conditions the global force-

displacement equations are:
 1 0.707   u2  F2 x  1,000 kN 

1,260  105 N 
m  0.707 1.5  u '3  F '3 x  0 

Solving the equation gives: u2  11.91 mm u '3  5.61 mm


Example 9 – Space Truss Problem  F1x 
F 
 u1 
 v 
 1y   1
 F2 x   u2 
  [T1 ][K ][T1 ]
T
  
 F2 y 
The global nodal forces are calculated as: F '3 x 
 v2 
u '3 
   
F '3 y  v '3 
 F1x 
F   0.5 0.5 0 0 0.707 0  0

 1y   0.5 1.5 0 1 0.707 0  0

 F2 x   0 0 1 0 0.707 0.707  11.91
  1,260  10 mm 
2N
 1  0 
 F 2 y  
0 0 1 0 0
 
F ' 3 x   0.707 0.707 0.707 0 1.5 0.5
  5.61 
   0 0 0.707 0 0.5 0.5   0 
F '3 y 
Therefore:
F1x  500 kN F1y  500 kN
F2 y  0 F '3 y  707 kN

Potential Energy Approach to Derive Bar Element Equations
Let’s derive the equations for a bar element using the principle
of minimum potential energy.
The total potential energy, p, is defined as the sum of the
p  U  
The differential internal work (strain energy) dU in a one-
dimensional bar element is:
dU   x ( y )( z )( x )d  x

If we let the volume of the element approach zero, then:

dU   x d  x dV
Summing the differential energy over the whole bar gives:

 x   x  1
U      x d  x  dV     E x d  x  dV   E x dV
2
V   V   2
0 0 V
For a linear-elastic material (Hooke’s law) as shown below:
1
 x  E x U    x  x dV
V
2

The internal strain energy statement becomes

1
U    x  x dV
2V
The potential energy of the external forces is:

M
    X bu dV   Txus dS   fixui
V S i 1
where Xb is the body force (force per unit volume), Tx is the

traction (force per unit area), and fix is the nodal concentrated
force. All of these forces are considered to act in the local x
direction.

Apply the following steps when using the principle of minimum

potential energy to derive the finite element equations.
1. Formulate an expression for the total potential energy.
2. Assume a displacement pattern.
3. Obtain a set of simultaneous equations minimizing the

total potential energy with respect to the displacement
parameters.

Consider the following bar element, as shown below:
L
A
 p    x  x dx  f1x u1  f2 x u2
20
  X bu dV   Txus dS
V S
We can approximate the axial displacement as:

u  x x
u  N1 N2   1  N1  1  N2 
u2  L L

Using the stress-strain relationships, the axial strain is:

du  dN1 dN2   u1 
x    
dx  dx dx  u2 
where N1 and N2 are the interpolation functions gives as:

 1 1   u1 
 x      x   [B]{d }
 L L  u2 
 1 1
B  
 L L 
The axial stress-strain relationship is:  x   [D ]  x 

Where [D] = [E] for the one-dimensional stress-strain

relationship and E is the modulus of elasticity.
Therefore, stress can be related to nodal displacements as:

 x   [D][B]d 
The total potential energy expressed in matrix form is:
L
A
 p    x   x  dx  d  P    u  X b  dV   u Tx  dS
T T T T
20 V S
where {P} represented the concentrated nodal loads.


If we substitute the relationship between û and d̂ into the

energy equations we get:
x x
L

 
A
p   d  B  D  B d  dx  d  P
T T T T
20
  d  N   X b  dV   d  Ns  Tx  dS
T T T T
V S
In the above expression for potential energy p is a function of

the d, that is: p = p(u1 , u2 ).
However, [B] and [D] and the nodal displacements u are not a
function of x.

Integration the energy expression with respect to x gives:

AL
p  d  [B ]T [D]T [B]d  d  f 
T T
2
where
f   P   [N ]T  Xb  dV   [N ]T  X b  dS
V S
We can define the surface tractions and body-force matrices

as:
fs    [N ]T Tx  dS fb    [N ]T  Xb  dV
S V

Minimization of p with respect to each nodal displacement

requires that:
 p  p
0 0
u1 u2
For convenience, let’s define the following
U   d 
* T
[B ]T [D ]T [B ] d 
 1
 L   1 1   u1 
U   u
*
1 u 2    [E ]  
 L
 
L  u2 
 1
 L 

U* 
E 2
L2

u1  2u1u2  u22 
The loading on a bar element is given as:
d f   u1f1x  u2f2 x

T
Therefore, the minimum potential energy is:

 p AE
  2u1  2u2   f1x  0
u1 2L
 p AE
  2u1  2u2   f2 x  0
u2 2L

The above equations can be written in matrix form as:

 p AE  1 1  u1   f1x 
  1 1  u    f   0
 d  L    2   2x 
AE  1 1
The stiffness matrix for a bar element is: k   L  1 1
This form of the stiffness matrix obtained from the principle of

minimum potential energy is identical to the stiffness matrix
derived from the equilibrium equations.

Consider the bar shown below:
The energy equivalent nodal forces due to the distributed load

are:
 x
L 1  
 
f0    [N ]T Tx  dS f0    1x     L Cx dx
f
S f2x  0  x 
 L 

 Cx 2 Cx 3  L
 x  CL 
2
1  
f1x  L  L   2  6 
0
3L
     Cx dx   
  2
f2x  0  x 
L
 Cx 3  CL 
 L   3L 0   3 
 
The total load is the area under the distributed load curve, or:
1 CL2
F  (L )(CL ) 
2 2
The equivalent nodal forces for a linearly varying load are:

F 1 2F 2
f1x   of the total load f2 x   of the total load
3 3 3 3

Consider the axially loaded bar shown below. Determine the
axial displacement and axial stress. Let E = 30 x 106 psi,
A = 2 in2, and L = 60 in. Use (a) one and (b) two elements in
the finite element solutions.

The one-element solution:
The distributed load can be converted into equivalent nodal

forces using:
 x
L 1 
F1x   L 
 0   x 
F  [ N ] T
T dS F0        10x dx
S F2x  0  x 
 L 

 x  10L2 10L2   10L2 

 - + -
 2 3   6 
1
F1x  L  L 
     10x dx    2
F2x  0  x 
2
 - 10 L  - 10L 
 L   3   3 
F1x   6,000 lb 
  
F2x  12,000 lb 

 1 1
k (1)  106  
 1 1
The element equations are:
 1 1 u1   6,000 

106       u1  0.006 in
 1 1  0  R2 x  12,000 
The second equation gives:
106 (u1 )  R2 x  12,000  R2 x  18,000 lb

The axial stress-strain relationship is: { x }  [D]{ x }

{ x }  E [B ] d 
 1 1   u1   u2  u1 
 E      E 
 L L  u2   L 
 0  0.006 
 30  106    3,000 psi (T )
 60 

The two-element solution:
The distributed load can be converted into equivalent nodal

forces.
For element 1, the total force of the triangular-shaped
distributed load is:
1
(30 in.)(300 lb in )  4,500 lb
2

Based on equations developed for the equivalent nodal force of

a triangular distributed load, develop in the one-element
problem, the nodal forces are:
 1 
 (4,500)
 f1x (1)   3   1,500 lb 
 (1)     
f2 x   2 (4,500) 3,000 lb 
 3 

For element 2, the applied force is in two parts: a triangular-

shaped distributed load and a uniform load. The uniform load
is:
(30 in )(300 lb / in )  9,000 lb
The nodal forces for element 2 are:

 1 1 
   (9,000)  (4,500) 
f2 x    2
(2)
3  6,000 lb 
 (2)     
f3 x    1 (9,000)  2 (4,500)  7,500 lb 
  2 3  


The final nodal force vector is:

 F1x  f1x   1,500 
(1)
   (1) (2)   
F2 x   f2 x  f2 x    9,000 
F   f3(2)  R  7,500 
 3x   x   3x 
The element stiffness matrices are:
1 2
2 3
2AE  1 1
k(1)  k(2) 
L  1 1

The assembled global stiffness matrix is:

 1 1 0  element 1
6  
K  2  10 1 2 1
  element 2
 0 1 1
The assembled global force-displacement equations are:
 1 1 0   u1   1,500 
   
2  106  1 2 1 u2    9,000 
 
 0 1 1  0  R3 x  7,500 

After the eliminating the row and column associated with u3x,
we get:
 1 1  u1   1,500 
2  106    
 1 2 u2  9,000 
Solving the equation gives: u1  0.006 in

u2  0.00525 in
Solving the last equation gives:
2  106 u2  R3 x  7,500  R3 x  18,000

The axial stress-strain relationship is:

 1 1   d1x 
 x (1)  E    
 L L  d 2 x 
 1 1   0.006 
 E     750 psi (T )
 30 30  0.00525 

The axial stress-strain relationship is:

 1 1  d 2 x 
 x (2)  E    
 L L  d 3 x 
 1 1  0.00525 
 E     5,250 psi (T )
 30 30   0 

Comparison of Finite Element Solution to Exact Solution
In order to be able to judge the accuracy of our finite element
models, we will develop an exact solution for the bar element
problem.
The exact solution for the displacement may be obtained by:
L
1
AE 0
u P ( x )dx
where the force P is shown on the following free-body diagram.
1
P( x ) 
2
 x (10 x )  5 x 2

Therefore:
L x
1 1 5x3
AE 0 AE o
u P ( x )dx u 5 x dx 
2
 C1
3 AE
Applying the boundary conditions:

5x3 5L3
u (L )  0   C1  C1  
3 AE 3 AE
The exact solution for axial displacement is:
u (L ) 
5
3 AE

x 3  L3   (x) 
P( x ) 5x 2
A

A

A plot of the exact solution for displacement as compared to
several different finite element solutions is shown below.

A plot of the exact solution for axial stress as compared to
several different finite element solutions is shown below.

A plot of the exact solution for axial stress at the fixed end
(x = L) as compared to several different finite element
solutions is shown below.

Galerkin’s Residual Method and Its Application
to a One-Dimensional Bar
There are a number of weighted residual methods.
However, the Galerkin’s method is more well-known and will be
the only weighted residual method discussed in this course.
In weighted residual methods, a trial or approximate function is
chosen to approximate the independent variable (in our
case, displacement) in a problem defined by a differential
equation.
The trial function will not, in general, satisfy the governing
differential equation.
Therefore, the substitution of the trial function in the differential
equation will create a residual over the entire domain of the
problem.

Therefore, the substitution of the trial function in the differential
equation will create a residual over the entire domain of the
problem.
 RdV  minimum
V
In the residual methods, we require that a weighted value of

the residual be a minimum over the entire domain of the
problem.
The weighting function W allows the weighted integral of the
residuals to go to zero.
 RW dV  0
V

Using Galerkin’s weighted residual method, we require the
weighting functions to be the interpolation functions Ni.
Therefore:
 RN
V
i dV  0 i  1, 2, , n

Example 12 - Bar Element Formulation
Let’s derive the bar element formulation using Galerkin’s
method. The governing differential equation is:
d  du 
 0
dx 
AE
dx 
Applying Galerkin’s method we get:

d  du 
L
 dx  AE dx  N
0
i dx  0 i  1, 2, , n
We now apply integration by parts using the following general

formula:
 udv  uv   vdu

If we assume the following:
dNi
u  Ni du  dx
dx
d  du  du
dv   v  AE
dx 
AE dx
dx  dx
then integration by parts gives:
 udv  uv   vdu
L
d  du 
L L
du du dN
0 dx  AE dx  Ni dx  Ni AE dx 0  0 AE dx dxi dx  0

Recall that:
du dN1 dN2 du  1 1   u1 
 u1  u2    
dx dx dx dx  L L  u2 
Our original weighted residual expression, with the

approximation for u becomes:
 u1 
L
 1 1
L
dNi du
AE   L dx    Ni AE
0
dx  L  u2  dx 0

Substituting N1 for the weighting function Ni gives:
 u1 
L
L
dN  1 1 du
AE  1   dx    N1AE
0
dx  L L  u2  dx 0
 1  1
L
1  u1  AEL
AE       dx    2  u1  u2 
0
L L L  u2  L
L
du du
N1AE  AE  AE x x 0
 A x x 0
 f1x
dx 0 dx x 0
AE
 u1  u2   f1x
L

Substituting N2 for the weighting function Ni gives:
 u1 
L
 1 1
L
dN2 du
AE   L dx    N AE
L 
2
0
dx  u2  dx 0
 1  1
L
1  u1  AEL
AE      dx    2  u1  u2 
0 
L L L  u2  L
L
du du
N2 AE  AE  AE x x L
 A x x L
 f2 x
dx 0 dx x L
AE
  u1  u2   f2 x
L

Writing the last two equations in matrix form gives:
AE  1 1  u1   f1x 
 1 1 u   f 
L    2   2x 
This element formulation is identical to that developed from

equilibrium and the minimum potential energy approach.

Problems:
2. Verify the global stiffness matrix for a three-dimensional
bar. Hint: First, expand T* to a 6 x 6 square matrix, then
expand k to 6 x 6 square matrix by adding the appropriate
rows and columns of zeros, and finally, perform the matrix
triple product k = TTk’T.
3. Do problems 3.4, 3.10, 3.12, 3.15a,b, 3.18, 3.23, 3.37,

3.43, 3.48, 3.50, and 3.55 on pages 146 - 165 in your
textbook “A First Course in the Finite Element Method” by
D. Logan.
4. Use SAP2000 and solve problems 3.63 and 3.64.
End of Chapter 3a
CIVL 7/8117 Chapter 3b - Symmetry and Bandwidth 1/16
Chapter 3b – Development of Truss Equations
Learning Objectives
• To derive the stiffness matrix for a bar element.
• To illustrate how to solve a bar assemblage by the direct
stiffness method.
• To introduce guidelines for selecting displacement
functions.
• To describe the concept of transformation of vectors in
two different coordinate systems in the plane.
• To derive the stiffness matrix for a bar arbitrarily oriented
in the plane.
• To demonstrate how to compute stress for a bar in the
plane.
• To show how to solve a plane truss problem.
• To develop the transformation matrix in three-
dimensional space and show how to use it to derive the
stiffness matrix for a bar arbitrarily oriented in space.
• To demonstrate the solution of space trusses.
Symmetry and Bandwidth

In this section, we will introduce the concepts of symmetry to
reduce the size of a problem and of banded-symmetric
matrices and bandwidth.
In many instances, we can use symmetry to facilitate the

solution of a problem.
Symmetry means correspondence in size, shape, and

position of loads; material properties; and boundary
conditions that are mirrored about a dividing line or plane.
Use of symmetry allows us to consider a reduced problem

instead of the actual problem. Thus, the order of the total
stiffness matrix and total set of stiffness equations can be
reduced.
Symmetry and Bandwidth - Example 1

Solve the plane truss problem shown below. The truss is
composed of eight elements and five nodes.
A vertical load of 2P is applied at node 4. Nodes 1 and 5 are

pin supports. Bar elements 1, 2, 7, and 8 have an axial
stiffness of AE and bars 3, 4, 5, and 6 have an axial stiffness
of AE.

In this problem, we will use a plane of symmetry.
The vertical plane perpendicular to the plane truss passing
through nodes 2, 4, and 3 is the plane of symmetry because
identical geometry, material, loading, and boundary
conditions occur at the corresponding locations on opposite
sides of this plane.

For loads such as 2P, occurring in the plane of symmetry,
one-half of the total load must be applied to the reduced
structure.
For elements occurring in the plane of symmetry, one-half of
the cross-sectional area must be used in the reduced
structure.
Element  C S C2 S2 CS
1 45° 0.707 0.707 0.5 0.5 0.5
2 315° 0.707 -0.707 0.5 0.5 -0.5
3 0° 1 0 1 0 0
4 270° 0 -1 0 1 0
5 90° 0 1 0 1 0

u1 v1 u2 v2 u1 v1 u3 v3
 1 1 1 1  1 1 1 1
   
AE  1 1 1 1 AE  1 1 1 1
k (1)  k (2) 
2L  1 1 1 1 2L  1 1 1 1
   
 1 1 1 1  1 1 1 1
1 45° 0.707 0.707 0.5 0.5 0.5
2 315° 0.707 -0.707 0.5 0.5 -0.5
3 0° 1 0 1 0 0
4 270° 0 -1 0 1 0
5 90° 0 1 0 1 0

u1 v1 u4 v4 u2 v2 u4 v4
1 0 1 0  0 0 0 0
 0 0  1
AE 0 0
 AE 0 1 0

k (3)  k ( 4) 
L 1 0 1 0 2L 0 0 0 0
   
0 0 0 0 0 1 0 1
1 45° 0.707 0.707 0.5 0.5 0.5
2 315° 0.707 -0.707 0.5 0.5 -0.5
3 0° 1 0 1 0 0
4 270° 0 -1 0 1 0
5 90° 0 1 0 1 0

u3 v3 u4 v4
0 0 0 0
 1
AE 0 1 0

k (5) 
2L 0 0 0 0
 
0 1 0 1
1 45° 0.707 0.707 0.5 0.5 0.5
2 315° 0.707 -0.707 0.5 0.5 -0.5
3 0° 1 0 1 0 0
4 270° 0 -1 0 1 0
5 90° 0 1 0 1 0

Since elements 4 and 5 lie in the plane of symmetry, one half
of their original areas have been used in developing the
stiffness matrices.
The displacement boundary conditions are:
u1  v1  u2  u3  u4  0
By applying the boundary conditions the force-displacement

 2 0 1 v 2   0 
AE     
0 2 1 v 3    0 
2L  
1 1 2  v 4  P 

We can solve the above equations by separating the matrices
in submatrices (indicated by the dashed lines). Consider a
general set of equations shown below:
K11 K12   d1   0  K11d1  K12d 2  0

K K  d   F 
 21 22   2    K 21d1  K 22d 2  F
Solving the first equation for d1 gives: d1  K111K12d 2
Substituting the above equation in the second matrix equation

K 21  K111K12d 2   K 22d 2  F
gives:
Simplifying this expression gives:
K 22 
 K 21K111K12 d 2  F

The previous equations can be written as: kc d 2  F
where: kc  K 22  K 21K111K12
d 2   kc  F
1
Therefore, the displacements d2 are:
If we apply this solution technique to our example global

stiffness equations we get:
  1 

AE  0   2 
1
 1 1  1
kc  [1]       
L   2 2  0 1  1 

  2 
Simplifying:
AE   1   AE  1  2L
 kc 
1
kc  [1]     
L   2  L  2  AE
2PL
Therefore, the displacements d2 are: d2  v 4  
AE
The remaining displacements can be found by substituting the
result for v4 in the global force-displacement equations.
 1

v 2   1 0   2   2PL 
        
v 3  0 1  1   AE 
 2 
Expanding the above equations gives the values for the
displacements.
 PL 

 2   AE 
v
  
v 3   PL 
 AE 
The coefficient matrix (stiffness matrix) for the linear equations

that occur in structural analysis is always symmetric and
banded.
Because a meaningful analysis generally requires the use of a

large number of variables, the implementation of
compressed storage of the stiffness matrix is desirable both
from the viewpoint of fitting into memory (immediate access
portion of the computer) and computational efficiency.
Another method, based on the concept of the skyline of the

stiffness matrix, is often used to improve the efficiency in
solving the equations.
The skyline is an envelope that begins with the first nonzero

coefficient in each column of the stiffness matrix (see the
following figure).
In skyline, only the coefficients between the main diagonal

and the skyline are stored.
In general, this procedure takes even less storage space in

the computer and is more efficient in terms of equation
solving than the conventional banded format.
A matrix is banded if the nonzero terms of the matrix are

gathered about the main diagonal.
To illustrate this concept, consider the plane truss shown on

below.
We can see that element 2 connects
nodes 1 and 4.
Therefore, the 2 x 2 submatrices at

positions 1-1, 1-4, 4-1, and 4-4 will
have nonzero coefficients.
The total stiffness matrix of the plane truss, shown in the

figure below, denotes nonzero coefficients with X’s.
The nonzero terms are within the some band. Using a banded
storage format, only the main diagonal and the nonzero
upper codiagonals need be stored.
We now define the semibandwidth: nb as nb = nd(m + 1)
where nd is the number of degrees of freedom per node and

m is the maximum difference in node numbers determined
by calculating the difference in node numbers for each
element of a finite element model.
In the example for the plane truss shown above,

m = 4 - 1 = 3 and nd = 2;
nb = 2(3 + 1) = 8
Execution time (primarily, equation-solving time) is a function

of the number of equations to be solved.
Without using banded storage of global stiffness matrix K, the

execution time is proportional to (1/3)n3, where n is the
number of equations to be solved.
Using banded storage of K, the execution time is proportional

to n(nb)2
The ratio of time of execution without banded storage to that

using banded storage is then (1/3)(n/nb)2
Execution time (primarily, equation-solving time) is a function

of the number of equations to be solved.
square matrix upper triangular matrix
banded matrix skyline matrix

For the plane truss example, this ratio is (1/3)(24/8)2 = 3
Therefore, it takes about three times as long to execute the

solution of the example truss if banded storage is not used.
Several automatic node renumbering schemes have been

computerized.
This option is available in most general-purpose computer

programs. Alternatively, the wavefront or frontal method are
popular for optimizing equation solution time.
In the wavefront method, elements, instead of nodes, are

automatically renumbered.
In the wavefront method the assembly of the equations

alternates with their solution by Gauss elimination.
The sequence in which the equations are processed is

determined by element numbering rather than by node
numbering.
The first equations eliminated are those associated with

element 1 only.
Next the contributions to stiffness coefficients from the

adjacent element, element 2, are eliminated.
If any additional degrees of freedom are contributed by

elements 1 and 2 only these equations are eliminated
(condensed) from the system of equations.
As one or more additional elements make their contributions

to the system of equations and additional degrees of freedom
are contributed only by these elements, those degrees of
freedom are eliminated from the solution.
This repetitive alternation between assembly and solution was

initially seen as a wavefront that sweeps over the structure in
a pattern determined by the element numbering.
The wavefront method, although somewhat more difficult to

understand and to program than the banded-symmetric
method, is computationally more efficient.
A banded solver stores and processes any blocks of zeros

created in assembling the stiffness matrix.
These blocks of zero coefficients are not stored or processed

using the wavefront method.
Many large-scale computer programs are now using the

wavefront method to solve the system of equations.
Homework Problems
5. Do problem B.9 on pages 864 in your textbook “A First

Course in the Finite Element Method” by D. Logan.
Determine the bandwidths of the plane trusses shown in
the figure below. What conclusions can you draw
regarding labeling of nodes?
Homework Problems
6. Solve the following truss problems. You may use

SAP2000 to do truss analysis.
5k 7 8
10 ft
a) For the plane truss shown below,

10 ft
determine the nodal displacements and
element stresses. 10 k
5 6
Nodes 1 and 2 are pin joints.

10 ft
Let E = 107 psi and the A = 2.0 in2 for
all elements. 10 k 3 4
10 ft
1 2
Homework Problems
b) For the 25-bar truss shown below, determine the displacements and
elemental stresses. Nodes 7, 8, 9, and 10 are pin connections.
Let E = 107 psi and the A = 2.0 in2 for the first story and A = 1.0 in2 for
the top story. Table 1 lists the coordinates for each node. Table 2 lists
the values and directions of the two loads cases applied to the 25-bar
space truss.
2 1 Node x (in) y (in) z (in)
1 -37.5 0.0 200.0
2 37.5 0.0 200.0
6
5 3 -37.5 37.5 100.0
3
4 4 37.5 37.5 100.0
5 37.5 -37.5 100.0
9 10
7 6 -37.5 -37.5 100.0
7 -100.0 100.0 0.0
8 8 100.0 100.0 0.0
9 100.0 -100.0 0.0
10 -100.0 -100.0 0.0
Note: 1 in = 2.54 cm
Homework Problems
b) For the 25-bar truss shown below, determine the displacements and
Let E = 107 psi and the A = 2.0 in2 for the first story and A = 1.0 in2 for
the top story. Table 1 lists the coordinates for each node. Table 2 lists
the values and directions of the two loads cases applied to the 25-bar
space truss.
2 1
Case Node Fx (kip) Fy (kip) Fz (kip)
6 1 1.0 10.0 -5.0

5
3 2 0.0 10.0 -5.0
4 1
3 0.5 0.0 0.0
10 6 0.5 0.0 0.0

9
7 1 0.0 20.0 -5.0
2
2 0.0 -20.0 -5.0
8
Note: 1 kip = 4.45 kN
Homework Problems
c) For the 72-bar truss shown below, determine the displacements and
Let E = 107 psi and the A = 1.0 in2 for the first two stories and A = 0.5
in2 for the top two stories. Table 3 lists the values and directions of the
two loads cases applied to the 72-bar space truss.
Case Node Fx (kip) Fy (kip) Fz (kip)
17 18
17 0.0 0.0 -5.0
60 in 18 0.0 0.0 -5.0
1
19 0.0 0.0 -5.0
13 14
20 0.0 0.0 -5.0
60 in 2 17 5.0 5.0 -5.0
Note: 1 kip = 4.45 kN.
9 10
15 14
18 3
60 in 16
4 9 8
5 10 17 2
6 13
7
11
1 6
60 in 5
12
1 2
120 in (b)
(a)
End of Chapter 3b
CIVL 7/8117 Chapter 4 - Development of Beam Equations - Part 1 1/36
Chapter 4a – Development of Beam Equations
Learning Objectives
• To review the basic concepts of beam bending
• To derive the stiffness matrix for a beam element
• To demonstrate beam analysis using the direct stiffness
method
• To illustrate the effects of shear deformation in shorter
beams
• To introduce the work-equivalence method for replacing
distributed loading by a set of discrete loads
• To introduce the general formulation for solving beam
problems with distributed loading acting on them
• To analyze beams with distributed loading acting on
them
Chapter 4a – Development of Beam Equations
Learning Objectives
• To compare the finite element solution to an exact
solution for a beam
• To derive the stiffness matrix for the beam element with
nodal hinge
• To show how the potential energy method can be used
to derive the beam element equations
• To apply Galerkin’s residual method for deriving the
beam element equations
Development of Beam Equations

In this section, we will develop the stiffness matrix for a beam
element, the most common of all structural elements.
The beam element is considered to be straight and to have

constant cross-sectional area.

We will derive the beam element stiffness matrix by using the
principles of simple beam theory.
The degrees of freedom associated with a node of a beam

element are a transverse displacement and a rotation.

We will discuss procedures for handling distributed loading
and concentrated nodal loading.
We will include the nodal shear forces and bending moments

and the resulting shear force and bending moment diagrams
as part of the total solution.

We will develop the beam bending element equations using
the potential energy approach.
Finally, the Galerkin residual method is applied to derive the

beam element equations
Beam Stiffness
Consider the beam element shown below.
The beam is of length L with axial local coordinate x and

transverse local coordinate y.
The local transverse nodal displacements are given by vi and
the rotations by ϕi. The local nodal forces are given by fiy and
the bending moments by mi.
Beam Stiffness
At all nodes, the following sign conventions are used on the
element level:
1. Moments are positive in the counterclockwise direction.
2. Rotations are positive in the counterclockwise direction.
3. Forces are positive in the positive y direction.
4. Displacements are positive in the positive y direction.
Beam Stiffness
At all nodes, the following sign conventions are used on the
global level:
1. Bending moments m are positive if they cause the beam
to bend concave up.
2. Shear forces V are positive is the cause the beam to
rotate clockwise.
Beam Stiffness
(+) Bending Moment
(-) Bending Moment

Beam Stiffness
(+) Shear Force
(-) Shear Force
Beam Stiffness
The differential equation governing simple linear-elastic beam
behavior can be derived as follows. Consider the beam
shown below.
Beam Stiffness
The differential equation governing simple linear-elastic beam
behavior can be derived as follows. Consider the beam
shown below.
w ( x )dx  dx 
 2 
 
Write the equations of equilibrium for the differential element:

 dx 
  Mright side  0  M  Vdx  w ( x )dx   dx 2  0
 2 

  Fy  0  V  (V  dV )  w ( x )dx
Beam Stiffness
From force and moment equilibrium of a differential beam
element, we get:
dM
M right side 0   Vdx  dM  0 or V 
dx
dV
F y 0   wdx  dV  0 or w 
dx
d  dM 
w 
dx  dx 
Beam Stiffness
The curvature  of the beam is related to the moment by:
1 M
 
 EI
where  is the radius of the deflected curve, v is the
transverse displacement function in the y direction, E is the
modulus of elasticity, and I is the principle moment of inertia
about y direction, as shown below.
Beam Stiffness
dv
The curvature,  for small slopes   is given as:
dx
d 2v
 2
dx
Therefore: d 2v M d 2v
  M  EI
dx 2 EI dx 2
Substituting the moment expression into the moment-load

equations gives:
d 2  d 2v 
 EI   w  x 
dx 2  dx 2 
For constant values of EI, the above equation reduces to:
 d 4v 
EI  4   w  x 
 dx 
Beam Stiffness
Step 1 - Select Element Type
We will consider the linear-elastic beam element shown below.
Beam Stiffness
Assume the transverse displacement function v is:
v  a1x 3  a2 x 2  a3 x  a4
The number of coefficients in the displacement function ai is
equal to the total number of degrees of freedom associated
with the element (displacement and rotation at each node).
The boundary conditions are:
v ( x  0)  v1 v ( x  L)  v 2
dv dv
 1  2
dx x 0 dx x L
Beam Stiffness

coefficients gives:
v (0)  v1  a4 v (L )  v 2  a1L3  a2L2  a3L  a4
dv (0) dv (L )
 1  a3  2  3a1L2  2a2L  a3
dx dx
Solving these equations for a1, a2, a3, and a4 gives:
2 1 
v   3 v1  v 2   2 1  2   x 3
L L 
 3 1 
   2 v1  v 2    21  2   x 2  1x  v1
 L L 
Beam Stiffness
In matrix form the above equations are: v  [N ] d 
v1 
 
d    1  [N ]  N1 N2 N3 N4 
v 2 
2 
where
N1 
1
L3

2 x 3  3 x 2L  L3  N2 
L3

1 3
x L  2 x 2L2  xL3 
N3 
1
L3

2 x 3  3 x 2 L  N4 
L3

1 3
x L  x 2L2 
Beam Stiffness
N1, N2, N3, and N4 are called the interpolation functions for a
beam element.
1.000 N1 1.000 N2
0.800 0.800
0.600 0.600
0.400 0.400
0.200 0.200
0.000 0.000
0.00 1.00 0.00 1.00
-0.200 -0.200
1.000 N3 1.000 N4
0.800 0.800
0.600 0.600
0.400 0.400
0.200 0.200
0.000 0.000
0.00 1.00 0.00 1.00
-0.200 -0.200
Beam Stiffness
du
The stress-displacement relationship is:  x  x, y  
dx
where u is the axial displacement function.
We can relate the axial displacement to the transverse

displacement by considering the beam element shown
below:
Beam Stiffness
dv
u  y
dx
Beam Stiffness
One of the basic assumptions in simple beam theory is that
planes remain planar after deformation, therefore:
du  d 2v 
 x  x, y    y  2 
dx  dx 
Moments and shears are related to the transverse
displacement as:
 d 2v   d 3v 
m  x   EI  2  V  x   EI  3 
 dx   dx 
Beam Stiffness
Step 4 - Derive the Element Stiffness Matrix
and Equations
Use beam theory sign convention for shear force and bending
moment.
m+
m+
V+ V+
Beam Stiffness
and Equations
Using beam theory sign convention for shear force and
bending moment we obtain the following equations:
d 3v EI
f1y  V  EI  12v1  6L1  12v 2  6L2 
dx 3 x 0
L3
d 3v EI
f2 y  V  EI 3   12v1  6L1  12v 2  6L2 
dx x L
L3
d v
2
EI
m1  m  EI
dx 2

L3 
6Lv1  4L21  6Lv 2  2L22 
x 0
d 2v EI
m2  m  EI 2
dx

L3 
6Lv1  2L21  6Lv 2  4L22 
x L
Beam Stiffness
and Equations
 f1y   12 6L 12 6L  v1   f1y   v1 

m   6L 2L2   1  m   
 1  EI  6L 4L2
   1  1
      k  
 f2 y  L  12 6L 6L  v 2   f2 y  v 2 
3
12
 
m2   6L 2L2 6L 4L2  2  m2  2 
where the stiffness matrix is:

 12 6L 12 6L 
 6L 2L2 
EI 6L 4L2
k 3  
L  12 6L 12 6L 
 
 6L 2L2 6L 4L2 
Beam Stiffness
and Equations
Beam stiffness based on Timoshenko Beam Theory
The total deflection of the beam at a point x consists of two

parts, one caused by bending and one by shear force. The
slope of the deflected curve at a point x is:
dv
  x   x
dx
Beam Stiffness
and Equations
The relationship between bending moment and bending

deformation is:
d  x 
M  x   EI
dx
Beam Stiffness
and Equations
The relationship between shear force and shear deformation is:
V  x   ks AG   x 
where ksA is the shear area.

Beam Stiffness
and Equations
You can review the details in your book, but by including the
effects of shear deformations into the relationship between
forces and nodal displacements a modified elemental
stiffness can be developed.
Beam Stiffness
and Equations
 12 6L 12 6L 

EI  6L  4    L2 6L  2    L2  
12EI
k 3
L 1     12 6L 12 6L  ks AGL2

 6L  2    L2 6L  4    L2 
Beam Stiffness
Consider a beam modeled by two beam elements (do not
include shear deformations):
Assume the EI to be constant throughout the beam. A force of

1,000 lb and moment of 1,000 lb-ft are applied to the mid-
point of the beam.
Beam Stiffness
The beam element stiffness matrices are:
v1 1 v2 2
 12 6L 12 6L 
 6L 2L2 
EI  6L 4L2

k (1)  3
L  12 6L 12 6L 
 
 6L 2L2 6L 4L2 
v  v 
2 2 3 3
 12 6L 12 6L 
 6L 2L2 
EI 6L 4L2
k (2)  3  
L  12 6L 12 6L 
 
 6L 2L2 6L 4L2 
Beam Stiffness
In this example, the local coordinates coincide with the global
coordinates of the whole beam (therefore there is no
transformation required for this problem).
The total stiffness matrix can be assembled as:
 F1y   12 6L 12 6L 0 0   v1 
M   6L 4L2  
 1 6L 2L 2
0 0   1 
 
F2 y  EI  12 6L 12  12 6L  6L 12 6L  v 2 
M   3  2  
 2  L  6L 2L 6L  6L 4L  4L 6L 2L  2 
2 2 2
F3 y   0 0 12 6L 12 6L  v 3 

    
M3   0 0 6L 2L 2
6L 4L2  3 
Element 1 Element 2
Beam Stiffness
The boundary conditions are: v1  1  v 3  0
 F1y   12 6L 12 6L 0 0   v01 

M   6L 4L2  
 1 6L 2L2 0 0   01 
 
F2 y  EI  12 6L 12  12 6L  6L 12 6L  v 2 
M   3   
 2  L  6L 2L 6L  6L 4L  4L
2 2 2
6L 2L2  2 
F3 y   0 0 12 6L 12 6L  v03 
    
M3   0 0 6L 2L2 6L 4L2  3 
Beam Stiffness
By applying the boundary conditions the beam equations

reduce to:
 1,000 lb  24 0 6L  v 2 
  EI  2  
1,000 lb ft   3  0 8L 2L  2 
2
  L 6L 2L2 4L2   

 0    3
Beam Stiffness
Solving the above equations gives:
875L3  375L2 125L2  625L 125L2  125L

v2   in 2   rad 3   rad
12EI 4EI EI

v1 
 
d v 
2
 d N   
2
m  x   EI  2   EI  2   1 
 dx   dx  v 2 
2 
The second derivative of N is linear; therefore m(x) is linear.
Beam Stiffness
875L3  375L2 125L2  625L 125L2  125L

v2   in 2   rad 3   rad
12EI 4EI EI

v1 
 
d v 
3
 d 3N   1 
V  x   EI  3   EI  2   
 dx   dx  v 2 
2 
The third derivative of N is a constant; therefore V(x) is
constant.
Beam Stiffness
Example 1 - Beam Problem
Consider the beam shown below. Assume that EI is constant
and the length is 2L (no shear deformation).

v
1
1 v
2
2 v
2
2 v
3
3
 12 6L 12 6L   12 6L 12 6L 
 6L 2L2   6L 2L2 
EI 6L 4L2 EI 6L 4L 2
k (1)  3   k (2)  3  
L  12 6L 12 6L  L  12 6L 12 6L 
   
 6L 2L2 6L 4L2   6L 2L2 6L 4L2 
Beam Stiffness
The local coordinates coincide with the global coordinates of
the whole beam (therefore there is no transformation required
for this problem).
The total stiffness matrix can be assembled as:
 F1y   12 6L 12 6L 0 0   v1 
M   6L 4L2 6L 2L2  
 1 0 0   1 
 
F2 y  EI  12 6L 24 0 12 6L  v 2 
M   3   
 2  L  6L 2L
2
0 8L 6L 2L2  2 
2
F3 y   0 0 12 6L 12 6L  v 3 

    
M3   0 0 6L 2L2 6L 4L2  3 
Element 1 Element 2
Beam Stiffness
The boundary conditions are: v 2  v 3  3  0
 F1y   12 6L 12 6L 0 0  v1 

M   6L 4L2 6L 2L2  
 1 0 0   1 
 
F2 y  EI  12 6L 24 0 12 6L  v02 
M   3   
 2  L  6L 2L
2
0 8L 6L 2L2  2 
2
F3 y   0 0 12 6L 12 6L  v03 

    
M3   0 0 6L 2L2 6L 4L2  03 
Beam Stiffness
By applying the boundary conditions the beam equations

reduce to:
 P   12 6L 6L  v1 
  EI  2 
 0   3  6L 4L 2L  1 
2
 0  L  6L 2L2 8L2   
    2
 7L 
Solving the above equations gives:  3 
v1   
  PL2  

 1   3 
  4EI  
 2  
 1 
Beam Stiffness
The positive signs for the rotations indicate that both are in the
counterclockwise direction.
The negative sign on the displacement indicates a deformation
in the -y direction.
 F1y   12 6L 12 6L 0 0   7 L 3 
M   6L 4L2 6L 2L2
 1 0 0  3 
  
F2 y  P  12 6L 24 0 12 6L   0 
M     
 2  4L  6L 2L
2
0 8L2 6L 2L2   1 
F3 y   0 0 12 6L 12 6L   0 
    
M3   0 0 6L 2L2 6L 4L2   0 
Beam Stiffness
The local nodal forces for element 1:
 f1y   12 6L 12 6L    7L 3   P 
   2    
 m1  P  6L 4L 6L 2L   3   0 
2
    
 f2 y  4L  12 6L 12 6L   0   P 
m2   2 
 6L 2L 6L 4L   1  PL 
2
 f2 y   12 6L 12 6L  0   1.5P 
m   6L 4L2 6L 2L2   1  PL 
 2 P       
  
 f3 y  4L  12 6L 12 6L  0   1.5P 
m3   2 
 6L 2L 6L 4L  0   0.5PL 
2
Beam Stiffness
The free-body diagrams for the each element are shown
below.
Combining the elements gives the forces and moments for the
original beam.
Beam Stiffness
Therefore, the shear force and bending moment diagrams are:
Beam Stiffness
Consider the beam shown below. Assume E = 30 x 106 psi and
I = 500 in4 are constant throughout the beam. Use four
elements of equal length to model the beam.

vi i v ( i 1) ( i 1)
 12 6L 12 6L 
 6L 2L2 
EI 6L 4L2
k( i )  3  
L  12 6L 12 6L 
 
 6L 2L2 6L 4L2 
Beam Stiffness
Using the direct stiffness method, the four beam element
stiffness matrices are superimposed to produce the global
stiffness matrix.
Element 1 Element 2
Element 3
Element 4
Beam Stiffness
The boundary conditions for this problem are:
v1  1  v 3  v 5  5  0
Beam Stiffness
v1  1  v 3  v 5  5  0
After applying the boundary conditions the global beam

24 0 6L 00  v 2  10,000 lb 
 0 8L2 2L2 00  2   0 

EI    
6L 2L2 8L2 6L 2L2  3    0 
L 
3
 v  10,000 lb 
0 0 6L 24 0  4
   
 0 0 2L2
0 8L2  4   0 
Beam Stiffness
Substituting L = 120 in, E = 30 x 106 psi, and I = 500 in4 into
the above equations and solving for the unknowns gives:
v 2  v 4  0.048 in 2  3  4  0
The global forces and moments can be determined as:
F1y  5 kips M1  25 kips·ft

F2 y  10 kips M2  0
F3 y  10 kips M3  0
F4 y  10 kips M4  0
F5 y  5 kips M5  25 kips·ft
Beam Stiffness
 f1y   12 6L 12 6L   0   5 kips 
m   6L 2L2   0   25 k·ft 
 1  EI  6L 4L2
  3   
 f2 y  L  12 6L 12 6L  0.048  5 kips 
 
m2   6L 2L2 6L 4L2   0   25 k·ft 
 f2 y   12 6L 12 6L  0.048   5 kips 

m   6L 2L2   0  25 k·ft 
 2  EI  6L 4L2

  3  
 f3 y  L  12 6L 12 6L   0   5 kips 
 
m3   6L 2L2 6L 4L2   0  25 k·ft 
Beam Stiffness
 f3 y   12 6L 12 6L   0   5 kips 
m   6L 2L2   0   25 k·ft 
 3  EI  6L 4L2

    
 f4 y  L  12 6L 6L  0.048  5 kips 
3
12
 
m4   6L 2L2 6L 4L2   0   25 k·ft 
 f4 y   12 6L 12 6L  0.048   5 kips 

m   6L 2L2   0  25 k·ft 
 4  EI  6L 4L2
     
 f5 y  L  12 6L 6L   0   5 kips 
3
12
 
m5   6L 2L2 6L 4L2   0  25 k·ft 
Beam Stiffness
Note: Due to symmetry about the vertical plane at node 3, we

could have worked just half the beam, as shown below.
Line of symmetry
Beam Stiffness
Consider the beam shown below. Assume E = 210 GPa and
I = 2 x 10-4 m4 are constant throughout the beam and the
spring constant k = 200 kN/m. Use two beam elements of
equal length and one spring element to model the structure.
Beam Stiffness
v1 1 v2 2 v2 2 v3 3
 12 6L 12 6L   12 6L 12 6L 
 6L 2L2   6L 2L2 
EI 6L 4L EI 6L 4L
2 2
k (1)  3   k (2)  3  
L  12 6L 12 6L  L  12 6L 12 6L 
   
 6L 2L2 6L 4L2   6L 2L2 6L 4L2 
The spring element stiffness matrix is:

v3 3 v4
v3 v4
 k 0 k 
 k k 
k (3)
  k (3)
0 0 0
 k k   
 k 0 k 
Beam Stiffness
Using the direct stiffness method and superposition gives the
global beam equations.
Element 2
Element 1
 F1y   12 6L 12 6L 0 0 0   v1 
M 
 1
 6L 4L2 6L 2L2 0 0 0   1 
 
F2 y   12 6L 24 0 12 6L 0  v 2 
  EI    kL3
 M2   3  6L 2L
2
0 8L2 6L 2L2 0  2  k' 
F  L EI
 0 0 12 6L 12  k ' 6L k '  v 3 
 3y    
 M3   0 0 6L 2L2 6L 4L2 0  3 
F   0 0 0 0 k ' 0 k '  v 4 
 4y 
Element 3
Beam Stiffness
The boundary conditions for this problem are: v1  1  v 2  v 4  0
 F1y   12 6L 12 6L 0 0 0   v01 

M 
 1
 6L 4L2 6L 2L2 0 0 0   01 
 
F2 y   12 6L 24 0 12 6L 0  v02 
  EI    kL3
 M2   3  6L 2L
2
0 8L2 6L 2L2 0  2  k' 
F  L  0 0 12 6L 12  k ' 6L k '  v 3 
EI
 3y    
 M3   0 0 6L 2L2 6L 4L2 0  3 
 
F4 y   0 0 0 0 k ' 0 k '  v04 
Beam Stiffness

M2   8L2 6L 2L2  2   0 
  EI     
F3 y   3  6L 12  k ' 6L  v 3   P 
 M  L  2L2 6L 4L2  3   0 
 3 
Solving the above  3PL2  1 
 
equations gives: EI  12  7k '  

2  
   7PL    kL3
3
1
v 3       k' 
   EI  12  7k '   EI
 3 
9PL2  1 
  
 EI  12  7k '  
Beam Stiffness
Substituting L = 3 m, E = 210 GPa, I = 2 x 10-4 m4, and
k = 200 kN/m in the above equations gives:
v 3  0.0174 m
2  0.00249 rad
3  0.00747 rad
Substituting the solution back into the global equations gives:
 F1y   12 6L 12 6L 0 0 0  0 
M   6L 4L2 6L 2L2  
  0 0 0 0
1
  
F2 y   12 6L 24 0 12 6L 0  0 
  EI   
M
 2  3 
6L 2L 2
0 8L 2
 6L 2L 2
0  0.00249 rad 
F  L
 0 0 12 6L 12  k ' 6L k '   0.0174 m 
 3y
  
 M3   0 0 6L 2L2 6L 4L2 0   0.00747 rad 
F   0 k ' k '   
 4y  0 0 0 0 0 
Beam Stiffness
Substituting L = 3 m, E = 210 GPa, I = 2 x 10-4 m4, and
k = 200 kN/m in the above equations gives:
v 3  0.0174 m
2  0.00249 rad
3  0.00747 rad
Substituting the solution back into the global equations gives:
 F1y   69.9 kN 
M   
 1  69.7 kN  m 
F2 y   116.4 kN 
   
 M2    0 
F   50 kN 
 3y   
 M3   0 
F   3.5 kN 
 4y   
Beam Stiffness
Distributed Loadings
Beam members can support distributed loading as well as
concentrated nodal loading.
Therefore, we must be able to account for distributed loading.
Consider the fixed-fixed beam subjected to a uniformly
distributed loading w shown the figure below.
The reactions, determined from structural analysis theory, are

called fixed-end reactions.
Beam Stiffness
In general, fixed-end reactions are those reactions at the ends
of an element if the ends of the element are assumed to be
fixed (displacements and rotations are zero).
Therefore, guided by the results from structural analysis for the

case of a uniformly distributed load, we replace the load by
concentrated nodal forces and moments tending to have the
same effect on the beam as the actual distributed load.
Beam Stiffness
The figures below illustrates the idea of equivalent nodal loads
for a general beam. We can replace the effects of a uniform
load by a set of nodal forces and moments.
Beam Stiffness
Work Equivalence Method
This method is based on the concept that the work done by
the distributed load is equal to the work done by the discrete
nodal loads. The work done by the distributed load is:
L
Wdistributed   w  x  v  x  dx
0
where v(x) is the transverse displacement. The work done by

the discrete nodal forces is:
Wnodes  m11  m22  f1y v1  f2 y v 2
The nodal forces can be determined by setting

Wdistributed = Wnodes for arbitrary displacements and rotations.
Beam Stiffness
Example 4 - Load Replacement
Consider the beam, shown below, and determine the
equivalent nodal forces for the given distributed load.
Using the work equivalence method or: Wdistributed  Wnodes

L
 w  x  v  x  dx  m 
0
1 1  m22  f1y v1  f2 y v 2
Beam Stiffness
Evaluating the left-hand-side of the above expression with:
w  x   w
2 1 
v ( x )   3 v1  v 2   2 1  2   x 3
L L 
 3 1 
   2 v1  v 2    21  2   x 2  1x  v1
 L L 
gives:
L
Lw L2w
 w v  x  dx 
0
2
v1  v 2  
4
1  2   Lw v 2  v1 
L2w L2w
  21  2   1  wLv1
3 2
Beam Stiffness
Using a set of arbitrary nodal displacements, such as:
v1  v 2  2  0 1  1
The resulting nodal equivalent force or moment is:

L
m11  m22  f1y v1  f2 y v 2   w  x  v  x  dx

0
 wL2 2 2 L2  wL2
m1     Lw  w
 4 3 2  12
Beam Stiffness
Using a set of arbitrary nodal displacements, such as:
v1  v 2  1  0 2  1
The resulting nodal equivalent force or moment is:

L
m11  m22  f1y v1  f2 y v 2   w  x  v  x  dx

0
 wL2 wL2  wL2

m2     
 4 3  12
Beam Stiffness
Setting the nodal rotations equal zero except for the nodal
LW Lw
f1y    Lw  Lw  
2 2
LW Lw
f2 y   Lw  
2 2
Summarizing, the equivalent nodal forces and moments are:
End of Chapter 4a
Chapter 4b – Development of Beam Equations
Learning Objectives
•To review the basic concepts of beam bending
•To derive the stiffness matrix for a beam element
•To demonstrate beam analysis using the direct stiffness
method
•To illustrate the effects of shear deformation in shorter
beams
•To introduce the work-equivalence method for replacing
distributed loading by a set of discrete loads
•To introduce the general formulation for solving beam
problems with distributed loading acting on them
•To analyze beams with distributed loading acting on them
Chapter 4b – Development of Beam Equations
Learning Objectives
•To compare the finite element solution to an exact solution
for a beam
•To derive the stiffness matrix for the beam element with
nodal hinge
•To show how the potential energy method can be used to
derive the beam element equations
•To apply Galerkin’s residual method for deriving the beam
element equations
Beam Stiffness
General Formulation
We can account for the distributed loads or concentrated loads
acting on beam elements by considering the following
formulation for a general structure:
F = Kd - F0
where F0 are the equivalent nodal forces, expressed in

terms of the global-coordinate components.
These forces would yield the same displacements as the
original distributed load.
If we assume that the global nodal forces are not present
(F = 0) then:
F0 = Kd
Beam Stiffness
General Formulation
We now solve for the displacements, d, given the nodal
forces F0.
Next, substitute the displacements and the equivalent nodal
forces F0 back into the original expression and solve for the
global nodal forces.
F = Kd - F0
This concept can be applied on a local basis to obtain the local
nodal forces in individual elements of structures as:
f = kd - f0
Beam Stiffness
Consider the beam shown below; determine the equivalent
nodal forces for the given distributed load.
The work equivalent nodal forces are shown above.

Using the beam stiffness equations:  wL 
  2 
 f1y   12 6L 12 6L  v1   wL2 
m   4L2 6L 2L2   1   12 

 1  EI  6L    
  3 
 f2 y  L  12 6L 12 6L  v 2   wL 

m2   
 6L 2L2
6L 4L2  2   2 
 wL2 
 
 12 
Beam Stiffness
Apply the boundary conditions: v1  1  0
 12 6L 12 6L  v1 

EI  6L 4L2 6L 2L2   1 
 
L3  12 6L 12 6L  v 2 
 
 6L 2L2 6L 4L2  2 
We can solve for the displacements

 wL   wL4 

 2  EI  12 6L  v 2  v 2   8EI 
 2  3        3 
 wL  L  6L 4L2  2  2   wL 

 12   6EI 
Beam Stiffness
In this case, the method of equivalent nodal forces gives the
exact solution for the displacements and rotations.
To obtain the global nodal forces, we will first define the
product of Kd to be Fe, where Fe is called the effective
global nodal forces. Therefore:
 wL 
 2 
 F 1y 
e
 12 6L 12 6L   0   5wL2 
 e   
 2  
 M 1  EI  6L 4L 6L 2L   0   12 
2
 e  3  wL4    
F 2 y  L  12 6L 12 6L   8EI    wL 
 M e 2   2 
 6L 2L 6L 4L   wL 6EI   2 
2 3
 wL2 
 
 12 
Beam Stiffness
Using the above expression and the fix-end moments in:
 wL   wL   wL 
 2   2   
     2
 F1y   5wL2   wL2   wL 
M   
 1 12   12    2 
F = Kd - F0     
    
F2 y    wL    wL   0 
 M2   2   2   
 wL2   wL2   
     0 
 12   12   
 F  0  F  wL
y 1y
2
wL L
 M  0 

1  wL 
2 2
Beam Stiffness
Example 6 - Cantilever Beam
Consider the beam, shown below, determine the vertical
displacement and rotation at the free-end and the nodal
forces, including reactions. Assume EI is constant
throughout the beam.
We will use one element and replace the concentrated load

with the appropriate nodal forces.
Beam Stiffness
The beam stiffness equations are:  P
2 
 f1y   12 6L 12 6L  v1   PL 
m   
 1 EI
 6L 4L2 6L 2L2   1   8 
  3      
 f2 y  L  12 6L 12 6L  v 2   P 

m2   
 6L 2L2 6L 4L2  2   2 
 PL 
 
 8 
Apply the boundary conditions: v1  1  0
 12 6L 12 6L  v1 

EI  6L 4L2 6L 2L2   1 
 
L3  12 6L 12 6L  v 2 
 
 6L 2L2 6L 4L2  2 
Beam Stiffness
The beam stiffness equations become:
 P  5PL3 
 2  EI  12 
6L  v 2  v 2   48EI 
  3      2 
 PL  L  6L 4L2  2  2   PL 

 8   8EI 
To obtain the global nodal forces, we begin by evaluating the

effective nodal forces.  P 
 2 
 F e1y   12 6L 12 6L   0   
 e   2    3PL 
 M 1  EI  6L 4L 6L 2L   0 
2
 8 
 e    5 PL3   
F 2 y  L  12 6L 12 6L  
3
48 EI  P 
 2 
 2
 M 2 
e
 6L 2L 6L 4L    PL 8 EI 
2 2
 PL 
 
 8 
Beam Stiffness
Using the above expression in the following equation, gives:
 P   P   
 2  2  P 
 F1y       
M   3PL   PL   PL 
 1  8   8   2 
F = Kd - F0        
F2 y  P P
      0 
 M2   2  2   
 PL   PL   
F      0 
 8   8   
Kd = Fe F0 F
Beam Stiffness
In general, for any structure in which an equivalent nodal
force replacement is made, the actual nodal forces acting on
the structure are determined by first evaluating the effective
nodal forces Fe for the structure and then subtracting off the
equivalent nodal forces F0 for the structure.
Similarly, for any element of a structure in which equivalent

nodal force replacement is made, the actual local nodal
forces acting on the element are determined by first
evaluating the effective local nodal forces for the element
and then subtracting off the equivalent local nodal forces
associated only with the element.
f = kd - f0
Beam Stiffness
Comparison of FE Solution to Exact Solution
We will now compare the finite element solution to the exact
classical beam theory solution for the cantilever beam
shown below.
Both one- and two-element finite element solutions will be

presented and compared to the exact solution obtained by
the direct double-integration method.
Let E = 30 x 106 psi, I = 100 in4, L = 100 in, and uniform load
w = 20 Ib/in.
Beam Stiffness
To obtain the solution from classical beam theory, we use the
double-integration method:
 M( x ) 
y    
 EI 
where the double prime superscript indicates differentiation
twice with respect to x and M is expressed as a function of x
by using a section of the beam as shown:

  Fy  0  V ( x )  wL  wx
wx 2 wL2
  M  0  M(x )  

 wLx 
2 2
Beam Stiffness
To obtain the solution from classical beam theory, we use the
double-integration method:
 M( x ) 
y    
 EI 
w  x2 L2 
EI   2
y    Lx   dx dx
2
w  x 3 Lx 2 xL2 
EI   6
y      C1  dx Boundary Conditions
2 2  y (0)  0 y (0)  0
w x 4
Lx3
x L 
2 2
y      C1x  C2
EI  24 6 4 
w  x 4 Lx 3 x 2L2 
y    
EI  24 6 4 
Beam Stiffness
Recall the one-element solution to the cantilever beam is:
 wL4 

v 2   8EI 
  3 
2   wL 

 6EI 
Using the numerical values for this problem we get:
 20 lb in 100 in  
4
 
 2 
v  8 
30  10 6

psi 100 in 4
  0.0833 in 
    
2   20 lb in 100 in    0.00111 rad 
3
 4 
 
 6 30  10 psi 100 in 
6
Beam Stiffness
The slope and displacement from the one-element FE
solution identically match the beam theory values evaluated
at x = L.
The reason why these nodal values from the FE solution are
correct is that the element nodal forces were calculated on
the basis of being energy or work equivalent to the
distributed load based on the assumed cubic displacement
field within each beam element.
Beam Stiffness
Values of displacement and slope at other locations along the
beam for the FE are obtained by using the assumed cubic
displacement function.

1
  1

v ( x )  3 2 x 3  3 x 2L v 2  3 x 3L  x 2L2 2
L L
The value of the displacement at the midlength v(x = 50 in) is:
v ( x  50 in )  0.0278 in
Using beam theory, the displacement at v(x = 50 in) is:
v ( x  50 in )  0.0295 in
Beam Stiffness
In general, the displacements evaluated by the FE method
using the cubic function for v are lower than by those of
beam theory except at the nodes.
This is always true for beams subjected to some form of

distributed load that are modeled using the cubic
displacement function.
The exception to this result is at the nodes, where the beam

theory and FE results are identical because of the work-
equivalence concept used to replace the distributed load by
work-equivalent discrete loads at the nodes.
Beam Stiffness
The beam theory solution predicts a quartic (fourth-order)
polynomial expression for a beam subjected to uniformly
distributed loading, while the FE solution v(x) assumes a
cubic (third-order) displacement behavior in each beam all
load conditions.
The FE solution predicts a stiffer structure than the actual one.
This is expected, as the FE model forces the beam into

specific modes of displacement and effectively yields a
stiffer model than the actual structure.
However, as more elements are used in the model, the FE

solution converges to the beam theory solution.
Beam Stiffness
For the special case of a beam subjected to only nodal
concentrated loads, the beam theory predicts a cubic
displacement behavior.
The FE solution for displacement matches the beam theory

solution for all locations along the beam length, as both v(x)
and y(x) are cubic functions.
Beam Stiffness
Under uniformly distributed loading, the beam theory solution
predicts a quadratic moment and a linear shear force in the
beam.
However, the FE solution using the cubic displacement

function predicts a linear bending moment and a constant
shear force within each beam element used in the model.
Beam Stiffness
We will now determine the bending moment and shear force
in the present problem based on the FE method.
M ( x )  EIy   EI
d 2 Nd 
 EI
 d N d
2
dx 2 dx 2
M ( x )  EI Bd
 12 x   4 6 x   6 12 x   2 6 x  
B    L2 
6
   3     2 
 L3   L L2   L2 L   L L 
 6 12 x   4 6x 
M  EI   2  3  v1     2  1
 L L   L L 
 6 12 x   2 6x  
  2  3  v 2     2  2 
L L   L L  
Beam Stiffness
We will now determine the bending moment and shear force
in the present problem based on the FE method.
Position MFE Mexact
X=0 -83,333 lb-in -100,000 lb-in
X = 50 in -33,333 lb-in -25,000 lb-in
X = 100 in 16,667 lb-in 0
Beam Stiffness
The plots below show the displacement, bending moment ,
and shear force over the beam using beam theory and the
one-element FE solutions.
Beam Stiffness
The FE solution for displacement matches the beam theory
solution at the nodes but predicts smaller displacements
(less deflection) at other locations along the beam length.
Beam Stiffness
The bending moment is derived by taking two derivatives on
the displacement function. It then takes more elements to
model the second derivative of the displacement function.
Beam Stiffness
The shear force is derived by taking three derivatives on the
displacement function. For the uniformly loaded beam, the
shear force is a constant throughout the singIe-element
model.
Beam Stiffness
To improve the FE solution we need to use more elements in
the model (refine the mesh) or use a higher-order element,
such as a fifth-order approximation for the displacement
function.
Beam Stiffness
Beam Element with Nodal Hinge
Consider the beam, shown below, with an internal hinge. An
internal hinge causes a discontinuity in the slope of the
deflection curve at the hinge and the bending moment is
zero at the hinge.
Beam Stiffness
For a beam with a hinge on the right end:
[K ] [K12 ]
11
 f1y   12 6L 12 6L   v 1 
m    6L 2L2   1 
 1  EI  6L 4L2
 
  3
 f2 y  L  12 6 L 12 6L  v 2 
m2   
 6L 2L 2
 6L 4L2  2 
[K 21 ] [K 22 ]
The moment m2 is zero and we can partition the matrix to
eliminate the degree of freedom associated with 2 .
f1   K11 K12   d1  f1  K11d1  K12d 2
    
f2  K21 K22  d2  f2  K 21d1  K 22d 2
Beam Stiffness
v1   f1y 
 f1y   12 6L 12 6L   v 1  d1   1  f1  m1 
m    6L 2L2   1  v  f 
 1  EI  6L 4L2
   2  2y 
  3
 f2 y  L  12 6 L 12 6L  v 2 
m2   
 6L 2L 2
 6L 4L2  2  d2   2  f2   m2 
d 2  K 22 1  f2  K 21d1 
f1  K11d1  K12d 2  K11d1  K12 K 22 1  f2  K 21d1  

m2  0
fc  kc d1 fc  f1  K12K 22 1f2 kc  K11  K12K 22 1K 21
Beam Stiffness
We can condense out the degree of freedom by using the
partitioning method discussed earlier.
kc  K11  K12K 22 1K 21
 12 6L 12  6L 
EI   EI  2  1
kc  3 6L 4L2 6L  3  2L  2 6L 2L2  6L 
L   L
6L  4L
 12 6L 12  
Therefore, the condensed stiffness matrix is:
 1 L 1
kc  3 L L2 L 
3EI 
L  
 1 L 1 
Beam Stiffness
The element force-displacement equations are:
 f1y   1 L 1 v1 
  3EI  L L2 L    
m1   3   1
f  L  1 L 1  v 2 
 2y 
Expanding the element force-displacement equations and
maintaining m2 = 0 gives:
 f1y   1 L 1 0   v1 
   L L2 L 0   1 
 m1  3EI   
  3
 f2 y  L  1 L 1 0  v 2 
m2   
0 0 0 0  2 
Beam Stiffness
Once the displacements d1 are found, d2 may be computed:
 m2  0 v1  
  
d 2  K 22 1
 f2  K 21d1  2   K 22 1
 m2   K 21  1  
 v  
  2 
In this case, d2 is 2 (the rotation at right side of the element

at the hinge.
Beam Stiffness
For a beam with a hinge on the left end:
 f1y   12 6L 12 6L   v 1 
m   6L  6L 2L2   1 
 1  EI 4L2
  3   
 f2 y  L  12  6L 12 6L  v 2 
 
m2  6L 4L2  2 
2
 6L 2L
Rewriting the equations to move 1 and m1 to the first row:
 m1   4L2 6L 6L 2L2   1 

f    
 1y  EI  6L 12 12 6L   v 1 
  3  
 f2 y  L  6L 12 12 6L  v 2 
m2   2 
 2L 6L 6L 4L2  2 
Beam Stiffness
For a beam with a hinge on the left end:
[K ] [K12 ]
11
 m1   4L2 6L 6L 2L2   1 

f    
 1y  EI  6L 12 12 6L   v 1 
    
 f2 y  L  6L 12 6L  v 2 
3
12
m2   2 
 2L 6L 6L 4L2  2 
[K 21 ] [K 22 ]
The moment m1 is zero and we can partition the matrix to
eliminate the degree of freedom associated with 1 .
f1   K11 K12   d1  f1  K11d1  K12d 2
    
f2  K21 K22  d2  f2  K 21d1  K 22d 2
Beam Stiffness
 m1   4L2 6L 6L 2L2   1  d1  1 f1  m1

f    
 1  EI  6L 12 12 6L   v 1  v1 
  3    f1y 
 f2  L  6L 12 12 6L  2   2  v 2 
 v d  f2    f2 y 
m2   2    m 
 2L 6L 6L 4L2  2   2  2
d1  K111  f1  K12d 2 
f2  K 21d1  K 22d 2  K 22d 2  K 21 K111  f1  K12d 2  

m1  0
fc  kc d 2 fc  f2  K 21K111f1 kc  K 22  K 21K111K12
Beam Stiffness
We can condense out the degree of freedom by using the
partitioning method discussed earlier.
kc  K 22  K 21K111K12
 12 12 6L   6L 
EI   EI   1
kc  3 12 12 6L  3 6L  2 6L 6L 2L2 
L   L
 2L2  4L
 6L 6L 4L2   
Therefore, the condensed stiffness matrix is:
 1 1 L 
kc  3 1 1 L 
3EI 
L  
 L L L2 
Beam Stiffness
The element force-displacement equations are:
 f1y   1 1 L   v 1 
  3EI  1 1 L  v 
 f2 y   3   2
m  L  L L L   1 
2 
 2

maintaining m1 = 0 gives:
 f1y  1 0 1 v1 

L
m  0 0   1 
 1  3EI 
0 0
 
  3
 f2 y  L L 0 L L  v 2 
m2   
 1 0 L 1  2 
Beam Stiffness
Once the displacements d2 are found, d1 may be computed:
 m1  0  v 1  
  
d1  K 11
1
 f1  K12d 2  1  K 11
1
 m1  K12 v 2  
   
  1 
In this case, d1 is 1 (the rotation at right side of the element

at the hinge.
Beam Stiffness
For a beam element with a hinge at its left end, the element
force-displacement equations are:
 f1y   1 1 L   v 1 
  3EI  1 1 L  v 
 f2 y   3   2
m  L  L L L  2 
2 
 2
maintaining m1= 0 gives:
 f1y  1 0 1 L   v 1 
m  0 0   1 
 1  3EI 
0 0
 
  3
 f2 y  L  1 0 1 L  v 2 
m2   
L 0 L L2  2 
Beam Stiffness
Example 7 - Beam With Hinge
In the following beam, shown below, determine the vertical
displacement and rotation at node 2 and the element forces
for the uniform beam with an internal hinge at node 2.
Assume EI is constant throughout the beam.
Beam Stiffness
The stiffness matrix for element 1 (with hinge on right) is:
 f1y   1 L 1 0   v1 
m   L L2 L 0   1 
 1  3EI   
  3
 f2 y  L  1 L 1 0  v 2 
m2   
0 0 0 0  2 
v1 1 v2 2
 f1y   1 a 1 0   v1 
m   0   1 
 1  3EI  a a a
2
  3  
 f2 y  a  1 a 1 0  v 2 
 
m2  0 0 0 0  2 
Beam Stiffness
The stiffness matrix for element 2 (with no hinge) is:
 f2   12 6L 12 6L  v 2 
m   6L 2L2  2 
 2  EI  6L 4L2
 
  
 f3  L  12 6 L 6L  v 3 
3
12
m3   
 6L 2L2  6L 4L2  3 
v2 2 v3 3
 f2   12 6b 12 6b  v 2 
m   2
6 b 2b 2  2 
 2  EI  6b 4b
 
  3
 f3  b  12 6 b 12 6b  v 3 
m3   
 6b 2b 2 6 b 4b 2  3 
Beam Stiffness
The assembled equations are:
 a33 3
 a33 0 0 0
 f1y   3
a2
 v1 
m   a2
3
a  a32 0 0 0  
  1
 3  3 6 
 1
 f2 y  12
 3 6
 12
b  v 2 
 a3 a2 b3 a3 2 3 2
   EI 
b b
2 
m2  0 0 6 4
 b2
6

b  2
 b2 b

 f3 y   0 0  12  b62 12
 b62  v 3 
  
b3 b3
 
m3  4 3 
 0 0 6
b2
2
b  b62 b
Beam Stiffness
The boundary conditions are: v1  v 3  1  3  0
 a33 3
 a33 0 0 0
 f1y   3
a2
  v01 
m   a2
3
 3
0 0 0   0 
a a2
 1  3  3 6 
 1
 f2 y 
3
 12 6
 12
b 2  v 2 
 a3 a2 a3 b3 b2 b3
   EI  2 
m
 2 0 0 6 4
 6

b  2
 b2 b b2

 f3 y   0 0  12  b62 12
 b62  v03 
  
b3 b3
  0 
 
m 4 
3
 0 0 6
b2
2
b  b
6
2 b
3
Beam Stiffness
 a 3 b 3P 
 3 12 6   
 a3  b3 b2  v 2  P 
3

v 2   3 b  a EI 
3

EI         
 6 4  2   0  2   a 3 b 2P 
 b 2 b   
 2 b3  a3 EI 
 
Beam Stiffness
The element force-displacement equations for element 1 are:
   b3P 
 0   b3  a3 
   f1y   
 f1y   1 a 1  
   ab P 
3
  3EI   
  m1    3 3 
2
 1
m 3 
a a a

0 
f  a  1 a 1  
 2y   
 f   b  a 
3 3
abP
  2 y   b3P 
 
 3  b3  a3  EI   3 3 
   b a 
Beam Stiffness
The slope 2 on element 1 may be found using the
condensed matrix:  m2  0 v 1  
1   
d 2  K 22  f2  K 21d1  2 element1  K 22  m2   K 21  1  
1
 v  
  2 
  
  
  0 
a  EI  
2 1  0  3  6a 2a 2 6a   0 
4EI  a  
 
3 3
abP

  3  b  a  EI  
3 3

a 2b3P
2 1  
 
2 b 3  a 3 EI
Beam Stiffness
The element force-displacement equations for element 2 are:
 a3P 
 a 3 b 3P   b3  a3 
 
 f2 y   12 6b 12 6b   3  
b  a EI 
3 3 
 f2 y  


m   2    0
 2  EI 6b 4b 6b 2b   m2   
2
a 3 b 2P 
  3      
 f3 y  b  12 6b 12 6b   2
 2
 
b3  a3 EI   f3 y   a P 
3
m3   6b 2b 6b 4b    m3   b  a 

2 3 3
 0   3 
 0   ba P 
 b 3  a 3 
Beam Stiffness
Displacements and rotations on each element.
a3 b3P v 3  3  0
v1  1  0 v2  
1
 
3 b3  a3 EI
3
2 2
2 3
a 3 b 2P
2 1  
abP
2 2 

2 b3  a3 EI   
2 b3  a 3 EI
Beam Stiffness
Potential Energy Approach to Derive
Beam Element Equations
Let’s derive the equations for a beam element using the
principle of minimum potential energy.
The procedure for applying the principle of minimum potential

energy is similar to that used for the bar element.
The total potential energy p is defined as the sum of the

p  U  
Beam Stiffness
The differential internal work (strain energy) dU in a one-
dimensional beam element is:
U   21  x  x dV
V
For a single beam element, shown below, subjected to both

distributed and concentrated nodal forces, the potential
energy due to forces (or the work done these forces) is:
2 2
    Ty v dS   Piy v i   mi i
S i 1 i 1
Beam Stiffness
If the beam element has a constant cross-sectional area A,
then the differential volume of the beam is given as:
dV  dA dx
The differential element where the surface loading acts is

given as: dS = b dx (where b is the width of the beam
element).
Therefore the total potential energy is:
 p    21  x  x dA dx   Ty vb dx    Piy v i  mi i 
2
x A x i 1
Beam Stiffness
d 2v
The strain-displacement relationship is:  x   y
dx 2
We can express the strain in terms of nodal displacements
and rotations as:
 12 x  6L 6 xL  4L2 12 x  6L 6 xL  2L2 
 x  y 
L3 L3 L3 L3  d 
 
 x   y [B]d
12 x  6L 6 xL  4L2 12 x  6L 6 xL  2L2 
[B ]   
 L3 L3 L3 L3 
Beam Stiffness
The stress-strain relationship in one-dimension is:
 x   [E ] x 
where E is the modulus of elasticity. Therefore:
 x    y [E ][B]d   x    y [B]d
The total potential energy can be written in matrix form as:
 p    21  x   x  dA dx   bTy v  dx  d  P
T T T
x A x
Beam Stiffness
If we define, w  bTy as a line load (load per unit length) in the
y direction and the substitute the definitions of x and x the
total potential energy can be written in matrix form as:
L L
E 2
p    y d  [B ]T [B ] d  dA dx   w d  [N ]T dx  d  P 
T T T
0 A
2 0
Use the following definition for moment of inertia: I   y 2dA

A
Then the total potential energy expression becomes:

L L
EI
p   d  [B ]T [B ]d  dx   w d  [N ]T dx  d  P
T T T
0
2 0
Beam Stiffness
Differentiating the total potential energy with respect to the
displacement and rotations (v1, v2, 1 and 2) and equating
each term to zero gives:
L L
EI  [B ]T [B ] dx d    w [N ]T dx  P   0
0 0
L
The nodal forces vector is: f    w [N ]T dx  P

0
L
The elemental stiffness matrix is: k   EI  [B]T [B] dx
0
Beam Stiffness
Integrating the previous matrix expression gives:
L
k   EI  [B]T [B] dx
0
 12 6L 12 6L 
 4L2 6L 2L2 
EI  6L
k   L3  12 6L 12

6L 
 
 6L 2L2 6L 4L2 
Beam Stiffness
Galerkin’s Method to Derive Beam Element Equations
The governing differential equation for a one-dimensional

beam is:
 d 4v 
EI  4   w  0
 dx 
We can define the residual R as:
L
  d 4v  
0  dx 4   Ni dx  0
 EI    w i  1, 2, 3, and 4
 
If we apply integration by parts twice to the first term we get:
L L
 EI v xxxx  Ni dx   EI v xx   Ni ,xx  dx  EI Ni v xxx    Ni ,x  v xx 

L
0
0 0
where the subscript x indicates a derivative with respect to x

Beam Stiffness
Since v  [N ] d  , then the second derivative of v with respect

to x is:
12 x  6L 6 xL  4L2 12 x  6L 6 xL  2L2 
v xx   3
L3 L3 L3  d 
 L 
or
v x x  [B ] d 
Therefore the integration by parts becomes:

L L
  Ni ,xx  EI [B] dx d    Ni w dx  NiV   Ni ,x  m  d 

L
0
0
0 0
i  1, 2, 3, and 4
Beam Stiffness
The above expression is really four equations (one for each

Ni) and can be written in matrix form as:
L L
 [B] EI [B] dx d     [N ] w dx  [N ]x m  [N ] V
T T T T L
0
0 0
The interpolation function in the last two terms can be

evaluated:
[N ] x  x  0   [0 1 0 0] [N ]x  x  L   [0 0 0 1]
[N ]  x  0   [1 0 0 0] [N ]  x  L   [0 0 1 0]
Beam Stiffness
Therefore, the last two terms of the matrix form of the Galerkin
formulation become (see the figure below):
i  1  V (0) i 2  m(0)
i 3  V (L ) i 4  m( L )
Beam Stiffness
When beam elements are assembled, as shown below:
Two shear forces and two moments form adjoining elements

contribute to the concentrated force and the concentrated
moment at the node common to both elements.
Beam Stiffness
Problems:
7. Verify the four beam element equations are contained in

the following matrix expression.
L L
EI  [B ] [B ] dx d    w [N ]T dx  P   0
T
0 0
8. Do problems 4.10, 4.12, 4.18, 4.22, 4.40 and 4.47 on

pages 229 - 237 in your textbook “A First Course in the
Finite Element Method” by D. Logan.
Beam Stiffness
Problems:
9. Work problem 4.36 (on page 231 in your textbook “A

First Course in the Finite Element Method” by D. Logan)
using the SAP2000.
Attempt to select the lightest standard W section to

support the loads for the beam.
The bending stress must not exceed 160 MPa and the
allowable deflection must not exceed (L = 6 m)/360
End of Chapter 4b
CIVL 7/8117 Chapter 5 - Plane Frame and Grid Equations - Part 1 1/52
Chapter 5a – Plane Frame and Grid Equations
Learning Objectives
• To derive the two-dimensional arbitrarily oriented
beam element stiffness matrix
• To demonstrate solutions of rigid plane frames by
the direct stiffness method
• To describe how to handle inclined or skewed
supports
Plane Frame and Grid Equations

Many structures, such as buildings and bridges, are
composed of frames and/or grids.
This chapter develops the equations and methods for solution

of plane frames and grids.
First, we will develop the stiffness matrix for a beam element

arbitrarily oriented in a plane.

We will then include the axial nodal displacement degree of

freedom in the local beam element stiffness matrix.

Then we will combine these results to develop the stiffness

matrix, including axial deformation effects, for an arbitrarily
oriented beam element.
We will also consider frames with inclined or skewed supports.


Two-Dimensional Arbitrarily Oriented Beam Element
We can derive the stiffness matrix for an arbitrarily oriented

beam element, in a manner similar to that used for the bar
element.
The local axes and are located along the beam element and
transverse to the beam element, respectively, and the global
axes x’ and y’ are located to be convenient for the total
structure.

The transformation from local displacements to global
displacements is given in matrix form as:
u   C S  u  C  cos 
    
v   S C  v  S  sin
Using the second equation for the beam element, we can
relate local nodal degrees of freedom to global degree of
freedom:
u1 
v1  Su1  Cv1
v1   S C 0 0 0 0  v1 
    0 0 0  1 
 1  0 1 0
 
  
v
  
2  0 0 0 S C 0  u2  d  Td

2  0 
0 0 0 0 1  v 2 
 
2 

For a beam, we will define the following as the
transformation matrix:
 S C 0 0 0
0
0 0 1 0 0 0
T 
0 0 0 S C 0 
 
0 0 0 0 0 1
Notice that the rotations are not affected by the orientation of

the beam.

Substituting the above transformation into the general form of
the stiffness matrix TTk’T gives:
u1 v1 1 u2 v2 2
 12S 2 12SC 6LS 12S 2 12SC 6LS 
 
 12SC 12C 2 6LC 12SC 12C 2 6LC 
EI  6LS 6LC 4L2 6LS 6LC 2L2 
k 3  
L  12S 2 12SC 6LS 12S 2 12SC 6LS 
 12SC 12C 2 6LC 12SC 12C 2 6LC 
 
 6LS 6LC 2L2 6LS 6LC 4L2 

Let’s now consider the effects of an axial force in the general
beam transformation.
Recall the simple axial deformation, define in the spring

element:  f   AE  1 1  u  
 
1x 1
   
f2x  L  1 1 u2 

Combining the axial effects with the shear force and bending
moment effects, in local coordinates gives:
 f1x   C 1 0 0 C 1 0 0   u 1 
f   0 12C 2 6 LC 2 0  12C 2 6 LC 2   v 1 
 1y    
 m 1   0 6 LC 2 4C 2 L2 0  6 LC 2 2C 2 L2   1 
   
 f2x   C 1 0 0 C1 0 0   u 2 
 f2y   0  12C 2  6 LC 2 0 12C 2  6 LC 2  v 2 
    
 m 2   0 6 LC 2 2C 2 L2 0  6 LC 2 4C 2 L2    2 
AE EI
C1  C2 
L L3

Therefore:
 C1 0 0 C 1 0 0 
 0 12C 2 6 LC 2 0  12C 2 6 LC 2 
 
 0 6 LC 2 4C 2 L2 0  6 LC 2 2C 2 L2 
k   
 C 1 0 0 C1 0 0 
 0  12C 2  6 LC 2 0 12C 2  6 LC 2 
 
 0 6 LC 2 2C 2 L2 0  6 LC 2 4C 2 L2 
The above stiffness matrix include the effects of axial force in

the x’ direction, shear force in the y’ direction, and bending
moment about the z’ axis.

The local degrees of freedom may be related to the global
degrees of freedom by:
 u1   C S 0 0 0 0   u1 
 v    S C 0 0 0 0   v 
 1   1

 1   0 0 1 0 0 0   1  d  Td
   
u2   0 0 0 C S 0  u2 
v 2   0 0 0 S C 0  v 2 
    
2   0 0 0 0 0 1 2 
where T has been expanded to include axial effects


Substituting the above transformation T into the general form
of the stiffness matrix gives:
 12I 2  12I  6I  12I   12I  6I 

 AC  2 S  A  2  CS  S   AC 2  2 S 2    A  2  CS  S
2
 L  L  L  L   L  L 
 12I 6I  12I   12I 2  6I 
 AS  2 C 2
2
C   A  2  CS   AS  2 C 
2
C 
 L L  L   L  L 
 6I 6I 
 C
k   L  2I 
E 4I S
L L 
 12I  12I  6I 
 AC 2  2 S 2  A   CS S 
 L  L2  L 
 12I 6I 
 symmetric AS 2  2 C 2  C
 L L 
 4I 

The analysis of a rigid plane frame can be accomplished by
applying stiffness matrix.
A rigid plane frame is: a series of beam elements rigidly

connected to each other; that is, the original angles made
between elements at their joints remain unchanged after the
deformation.

Furthermore, moments are transmitted from one element to
another at the joints.
Hence, moment continuity exists at the rigid joints.

In addition, the element centroids, as well as the applied
loads, lie in a common plane.
We observe that the element stiffnesses of a frame are

functions of E, A, L, I, and the angle of orientation  of the
element with respect to the global-coordinate axes.

Rigid Plane Frame Example 1
Consider the frame shown in the figure below.
The frame is fixed at nodes 1 and 4 and subjected to a

positive horizontal force of 10,000 lb applied at node 2 and to
a positive moment of 5,000 lb-in applied at node 3.

Let E = 30 x 106 psi and A = 10 in2 for all elements, and let I =
200 in4 for elements 1 and 3, and I = 100 in4 for element 2.

Element 1: The angle between x and x’ is 90°
C 0 S 1
12I 12(200) 6I 6(200)

  0.167 in 2   10.0 in 3
120 
2
L2 L 120
E 30  106
  250,000 lb 3
L 120 in

u1 v1 1 u2 v2 2
 0.167 0 10 0.167 0 10 
 0 10 0 0 10 0 
 
 10 0 800 10 0 400  lb
k (1)  250,000  
 0.167 0 10 0.167 0 10  in
 0 10 0 0 10 0 
 
 10 0 400 10 0 800 

C 1 S0
12I 12(100) 6I 6(100)

  0.0835 in 2   5.0 in 3
120 
2 2
L L 120
E 30  106
  250,000 lb 3
L 120 in

u2 v2 2 u3 v3 3
 10 0 0 10 0 0 
 0 0.0835 5 0 0.0835 5 
 
 0 5 400 0 5 200  lb
k (2)  250,000  
 10 0 0 10 0 0  in
 0 0.0835 5 0 0.0835 5 
 
 0 5 200 0 5 400 

C 0 S  1
12I 12(200) 6I 6(200)

  0.167 in 2   10.0 in 3
120 
2 2
L L 120
E 30  106
  250,000 lb 3
L 120 in

u3 v3 3 u4 v4 4
 0.167 0 10 0.167 0 10 
 0 10 0 0 10 0 
 
 10 0 800 10 0 400  lb
k (3)  250,000  
 0.167 0 10 0.167 0 10  in
 0 10 0 0 10 0 
 
 10 0 400 10 0 800 

u1  v1  1  u4  v 4  4  0

u2 v2 2 u3 v3 3
10,000  10.167 0 10 10 0 0  u2 
 0   0 10.0835 5 0 0.0835 5  v 2 
   
 0  5  10 5 1200 0 5 200  2 
   2.5  10   
 0   10 0 0 10.167 0 10  u3 
 0   0 0.0835 5 0 10.0835 5  v 3 
    
 5,000   0 5 200 10 5 1200  3 

Solving the above equations gives: u2   0.211 in 
v   0.00148 in 
 2  
2  0.00153 rad 
  
u3   0.209 in 
v 3   0.00148 in 
   
3  0.00149 rad 
x


Element 1: The element force-displacement equations can
be obtained using f’ = k’Td. Therefore, Td is:
 C S 0 0 0 0
 S C 0 0 0 0 

 0 0 1 0 0 0
 C 0 S  1
T  
 0 0 0 C S 0
 0 0 0 S C 0 
 
 0 0 0 0 0 1
0 1 0 0 0 0  u1  0   0
 1    

0 0 0 0 0 v1  0
   0

0 0 1 0 0 0  1  0   0
Td     
0 0 0 0 1 0  u2  0.211 in   0.00148 in 
0 0 0 1 0 0  v 2  0.00148 in   0.211 in 
    
0 0 0 0 0 1 2  0.00153 rad  0.00153 rad 

Element 1: Recall the elemental stiffness matrix is:
 C1 0 0 C1 0 0  C1  
2

AE 10 in 30  10 psi
6

 2.5  106 lb in
 0 12C2 6LC2 

12C2 6LC2 0
 L 120 in
 0 6LC2 4C2L2 0 6LC2 2C2L2 
k   
 C1
 
0 0 C1 0 0 
EI 200 in 30  10 psi
4 6
 0 12C2 6LC2 0 12C2 6LC2 

2C2L2 6LC2

4C2L2  C2    3,472.22 lb in
 0
120 in 
6LC2 0
L3 3
The local force-displacement equations are:

 10 0 0 10 0 0  0
 0 0.167 10 0 0.167 10   0
  
5  0 10 800 0 10 400   0
f (1)  kTd  2.5  10   
 10 0 0 10 0 10   0.00148 in 
 0 0.167 10 0 0.167 10   0.211 in 
  
 0 10 400 0 10 800   0.00153 rad 

Element 1: Simplifying the above equations gives:
 f1x  3,700 lb 
 f    4,990 lb 
 1y   
 m1   376 k  in 
  
 f2x   3,700 lb 
 f2y  4,990 lb 
   
m2   223 k  in 

 C S 0 0 0 0
 S C 0 0 0 0 

 0 0 1 0 0 0
 C 1 S0
T  
 0 0 0 C S 0
 0 0 0 S C 0 
 
 0 0 0 0 0 1
1 0 0 0 0 0  u2  0.211 in   0.211 in 

0  
1 0 0 0 0 v 2  0.00148 in    0.00148 in 
   
0 0 1 0 0 0  2  0.00153 rad  0.00153 rad 
Td      
0 0 0 1 0 0  u3  0.209 in   0.209 in 
0 0 0 0 1 0  v 3  0.00148 in   0.00148 in 
    
0 0 0 0 0 1 3  0.00149 rad  0.00149 rad 

Element 2: The local force-displacement equations are:
 C1 0 0 C1 0 0  C1  
2

AE 10 in 30  10 psi
6

 2.5  106 lb in
 0 12C2 6LC2 

12C2 6LC2 0
 L 120 in
 0 6LC2 4C2L2 0 6LC2 2C2L2 
k   
 C1
 
0 0 C1 0 0 
EI 100 in 30  10 psi
4 6
 0 12C2 6LC2 0 12C2 6LC2 
 2
6LC2

4C2L2  C2    1,736.11 lb in
 0
120 in 
6LC2 2C2L 0
L3 3

 10 0 0 10 0 0  0.211 in 
 0 0.0833 5 0 0.0833 5   0.00148 in 
  
5  0 5 400 0 5 200  0.00153 rad 
f (2)  kTd  2.5  10   
 10 0 0 10 0 0  0.209 in 
 0 0.0833 5 0 0.0833 5   0.00148 in 
  
 0 5 200 0 5 400  0.00149 rad 

 f2x   5,010 lb 
 f    3,700 lb 
 2y   
m2  223 k  in 
  
 f3x   5,010 lb 
 f3y   3,700 lb 
   
m3   221 k  in 

 C S 0 0 0 0
 S C 0 0 0 0 

 0 0 1 0 0 0
 C 0 S  1
T  
 0 0 0 C S 0
 0 0 0 S C 0 
 
 0 0 0 0 0 1
0 1 0 0 0  u3  0.209 in
0   0.00148 in 
1 0 0 0 0 v 3  0.00148 in   0.209 in 
0 
   
0 0 1 0 0  3  0.00149 rad  0.00149 rad 
0
Td     
0 0 0 0 1 0  u4  0   0 
0 0 0 1 0 0  v 4  0   0 
    
0 0 0 0 0 1 4  0   0 

 C1 0 0 C1 0 0  C1  
2

AE 10 in 30  10 psi
6

 2.5  106 lb in
 0 12C2 6LC2 

12C2 6LC2 0
 L 120 in
 0 6LC2 4C2L2 0 6LC2 2C2L2 
k   
 C1
 
0 0 C1 0 0 
EI 200 in 30  10 psi
4 6
 0 12C2 6LC2 0 12C2 6LC2 

2C2L2 6LC2

4C2L2  C2    3,472.22 lb in
 0
120 in 
6LC2 0
L3 3

 10 0 0 10 0 0   0.00148 in 
 0 0.167 10 0 0.167 10   0.209 in 
  
 0 10 800 0 10 400   0.00149 rad 
f (3)  kTd  2.5  105   
 10 0 0 10 0 10   0 
 0 0.167 10 0 0.167 10   0 
  
 0 10 400 0 10 800   0 

 f3x   3,700 lb 
 f    5,010 lb 
 3y   
m3   226 k  in 
  
 f4x   3,700 lb 
 f4y   5,010 lb 
   
m4   375 k  in 

Check the equilibrium of all the elements:

The frame is fixed at nodes 1 and 3 and subjected to a

positive distributed load of 1,000 lb/ft applied along element 2.
Let E = 30 x 106 psi and A = 100 in2 for all elements, and
let I = 1,000 in4 for all elements.

First we need to replace the distributed load with a set of
equivalent nodal forces and moments acting at nodes 2 and 3.
For a beam with both end fixed, subjected to a uniform

distributed load, w, the nodal forces and moments are:
wL (1,000 lb / ft )40 ft
f2 y  f3 y     20k
2 2
wL2 (1,000 lb / ft )(40 ft )2
m2  m3     133,333 lb  ft
12 12
 1,600 k  in

If we consider only the parts of the stiffness matrix associated
with the three degrees of freedom at node 2, we get:
Element 1: The angle between x and x’ is 45º
C  0.707 S  0.707
12I 12(1,000) 6I 6(1,000)

  0.0463 in 2   11.785 in 3
  L 12  30 2
2 2
L 12  30 2
E 30  106
  58.93 k 3
L 12  30 2 in

u2 v2 2
 50.02 49.98 8.33 
k (1)
 58.93  49.98 50.02 8.33  k
  in
 8.33 8.33 4000 
u2 v2 2
 2,948 2,945 491 
k (1) 
 2,945 2,948 491  k
  in
 491 491 235,700 

C 1 S 0
12I 12(1,000) 6I 6(1,000)

  0.0521 in 2   12.5 in 3
12  40  12  40
2 2
L L
E 30  106
  52.5 k 3
L 12  40 in

u2 v2 2
100 0 0 
k (2)
 62.50 0 0.052 12.5  k

  in
 0 12.5 4,000 
u2 v2 2
6,250 0 0 
k (2)
  0 3.25 781.25  k
  in
 0 781.25 250,000 

The global beam equations reduce to:
 0   9,198 2,945 491  u2 
    
 20 k   2,945 2,951 290  v 2 

  290 485,700  2 
1,600 k  in   491
Nodal equivalent forces
u2   0.0033 in 
   
v 2    0.0097 in 
  0.0033 rad 
 2  

u2   0.0033 in 
   
v 2    0.0097 in 
  0.0033 rad 
 2  

x

Element 1: The element force-displacement equations can be
obtained using f’ = k’Td. Therefore, Td is:
 0.707 0.707 0 0 0 0  0   0 
 0.707 0.707 0 0 0 0   0   0 
    
 0 0 1 0 0 0  0   0 
Td      
 0 0 0 0.707 0.707 0  0.0033 in   0.00452 in 
 0 0 0 0.707 0.707 0   0.0097 in   0.0092 in 
    
 0 0 0 0 0 1 0.0033 rad  0.0033 rad 
Recall the elemental stiffness matrix is a function of

values C1, C2, and L
AE (100)30  106 EI 30  106 (1,000)
C1    5,893 k in C2    0.2273 k in
 
3
L 12  30 2 L3 12  30 2

 5,893 0 10 5,893 0 0  0 
 0 2.730 694.8 0 2.730 694.8   0 
  
 10 694.8 117,900 0 694.8 117,000   0 
  kTd  
f(1)  
 5,893 0 0 5,983 0 0   0.00452 in 
 0 2.730 694.8 0 2.730 694.8   0.0092 in 
  
 0 694.8 117,000 0 694.8 235,800  0.0033 rad 
Simplifying the above equations gives:

 f1x   26.64 k 
f   
 1y   2.268 k 
 m1   389.1 k  in 
  
 f2x   26.64 k 
 f2y   2.268 k 
   
 2 
m  778.2 k  in 

obtained using f’ = k’Td. Therefore, Td is:
1 0 0 0 0 0   0.0033 in   0.0033 in 
0 1 0 0 0 0   0.0097 in   0.0097 in 
    
0 0 1 0 0 0   0.0033 rad  0.0033 rad 
Td      
0 0 0 1 0 0  0   0 
0 0 0 0 1 0  0   0 
    
0 0 0 0 0 1  0   0 
Recall the elemental stiffness matrix is a function of

values C1, C2, and L
AE (100)30  106 EI 30  106 (1,000)
C1    6,250 k in C2    0.2713 k in
12  40 12  40 
3
L L3

 6,250 0 0 6,250 0 0   0.0033 in 
 0 3.25 781.1 0 3.25 781.1   0.0097 in 
  
 0 781.1 250,000 0 781.1 125,000   0.0033 rad 
  kTd  
f(2)  
 6,250 0 0 6,250 0 0  0 
 0 3.25 781.1 0 3.25 781.1   0 
  
 0 781.1 125,000 0 781.1 250,00  0 
Simplifying the above equations gives:

 20.63 k 
 2.58 k 

832.57 k  in 
kd   
 20.63 k 
 2.58 k 
 
 412.50 k  in 

Element 2: To obtain the actual element local forces, we must
subtract the equivalent nodal forces.
f  kd  f0
 f2x   20.63 k   0   20.63 k 

 f    2.58 k   20 k   17.42 k 
 2y       
m2  832.57 k  in  1600 k  in   767.4 k  in 
     
 f3x   20.63 k   0   20.63 k 
 f3y   2.58 k   20 k   22.58 k 
       
m3  412.50 k  in   1600 k  in  2,013 k  in 

The local forces in both elements are:
Element 1 Element 2

In this example, we will illustrate the equivalent joint force
replacement method for a frame subjected to a load acting on
an element instead of at one of the joints of the structure.
Since no distributed loads are present, the point of application

of the concentrated load could be treated as an extra joint in
the analysis.

This approach has the disadvantage of increasing the total

number of joints, as well as the size of the total structure
stiffness matrix K.

For small structures solved by computer, this does not pose a

problem.

However, for very large structures, this might reduce the

maximum size of the structure that could be analyzed.

The frame is fixed at nodes 1, 2, and 3 and subjected to a

concentrated load of 15 k applied at mid-length of element 1.
Let E = 30 x 106 psi, A = 8 in2, and let I = 800 in4 for all
elements.

1. Express the applied load in the
element 1 local coordinate
system (here x’ is directed from
node 1 to node 4).
2. Next, determine the equivalent

joint forces at each end of
element 1, using the Table D-1
in Appendix D.


3. Then transform the equivalent joint forces from the local
coordinate system forces into the global coordinate system
forces, using the equation: f = TTf
These global joint forces are:
4. Then we analyze the structure, using the equivalent joint

forces (plus actual joint forces, if any) in the usual manner.
5. The final internal forces developed at the ends of each

element may be obtained by subtracting Step 2 joint forces
from Step 4 joint forces.

Element 1: The angle between x and x’ is 63.43º
C  0.447 S  0.895
12I 12(800)
  0.0334 in 2
12  44.7 
2 2
L
u v 
6I 6(800) 4 4 4
  8.95 in 3  90.0 178 448 
L 12  44.7
k (1)
  178 359 244  k in
 
E 30  10 6
 448 244 179,000 
  55.9 k 3
L 12  44.7 in

Element 2: The angle between x and x’ is 116.57º
C  0.447 S  0.895
12I 12(800)
  0.0334 in 2
12  44.7 
2 2
L
u v 
6I 6(800) 4 4 4
  8.95 in 3  90.0 178 448 
L 12  44.7
k (2)   178 359 244  k in
 
E 30  10 6
 448 244 179,000 
  55.9 k 3
L 12  44.7 in

Element 3: The angle between x and x’ is 0º (from nodes 4 to 3)
C 1 S0
12I 12(800)
  0.0267 in 2
12  50 
2 2
L
u v 
6I 6(800) 4 4 4
  8.0 in 3  400 0 0
L 12  50
k (3)
 0 1.334 400  k in
 
E 30  10 6
 0 400 160,000 
  50.0 k 3
L 12  50 in

The global beam equations reduce to:
 7.5 k  582 0 896  u4 

    
 0  0 719 400  v 4 

900 k  in  896 518,000  4 
   400
u4  0.0103 in 
   
v 4    0.000956 in 
  0.00172 rad 
 4  


obtained using f’ = k’Td  C S 0 0 0 0
 S C 0 0 0 0 
 
C  0.447 S  0.895  0 0 1 0 0 0
T 
 0 0 0 C S 0
 0 0 0 S C 0 
 
 0 0 0 0 0 1
 0.447 0.895 0 0 0 0   u1  0   0 
 0.895 0.447 0 0 0  
0  v1  0   0 
    
 0 0 1 0 0 0   1  0   0 
Td       
 0 0 0 0.447 0.895 0  u40.0103 in   0.00374 in 
 0 0 0 0.895 0.447 0  v0.000956
 in   0.00963 in 
     
4

 0 0 0 0 0 1  0.00172
4 rad  0.00172 rad 


 C1 0 0 C1 0 0 
 0 12C2 6LC2 0 12C2 6LC2 
 
 0 6LC2 4C2L2 0 6LC2 2C2L2 
k   
 C1 0 0 C1 0 0 
 0 12C2 6LC2 0 12C2 6LC2 
 
 0 6LC2 2C2L2 0 6LC2 4C2L2 
AE (8)30  106 EI 30  106 (800)

C1    447.2 k in C2    0.155 k in
L 12  44.72 L3 12  44.72 3


 447 0 0 447 0 0  0 
 0 1.868 500.5 0 1.868 500.5   0 
  
 0 500.5 179,000 0 500.5 89,490   0 
  kTd  
f(1)  
 447 0 0 447 0 0   0.00374 in 
 0 1.868 500.5 0 1.868 500.5   0.00963 in 
  
 0 500.5 89,490 0 500.5 179,000  0.00172 rad 
 1.67 k  k(1) Td
 0.88 k 
 
158 k  in 
  kd  
f(1) 
 1.67 k 
 0.88 k 
 
311 k  in 

Element 1: To obtain the actual element local forces, we must

subtract the equivalent nodal forces.
f  kd  f0
 f1x   1.67 k   3.36 k   5.03 k 

 f    0.88 k    
 1y    
6.71 k
  7.59 k 

 m1   158 k  in   900 k  in  1,058 k  in 
    
 f4x   1.67 k   3.36 k   1.68 k 
 f4y   0.88 k   6.71 k   5.83 k 
       
m4   311 k  in   900 k  in   589 k  in 


 S C 0 0 0 0 
 
C  0.447 S  0.895  0 0 1 0 0 0
T 
 0 0 0 C S 0
 0 0 0 S C 0 
 
 0 0 0 0 0 1
 0.447 0.895 0 0 0 0  u2  0   0 

 0.895 0.447 0 0 0   
0 v 2  0   0 
    
 0 0 1 0 0 0  2  0   0 
Td    u   
 0 0 0 0.447 0.895 0   40.0103 in   0.00546 in 
 0 0 0 0.895 0.447 0  v0.000956
 in   0.00879 in 
    
4

 0 0 0 0 0 1 40.00172
 rad  0.00172 rad 


 C1 0 0 C1 0 0 
 0 12C2 6LC2 0 12C2 6LC2 
 
 0 6LC2 4C2L2 0 6LC2 2C2L2 
k   
 C1 0 0 C1 0 0 
 0 12C2 6LC2 0 12C2 6LC2 
 
 0 6LC2 2C2L2 0 6LC2 4C2L2 
AE (8)30  106 EI 30  106 (800)

C1    447.2 k in C2    0.155 k in
L 12  44.72 L3 12  44.72 3


 447 0 0 447 0 0  0 
 0 1.868 500.5 0 1.868 500.5   0 
  
 0 500.5 179,000 0 500.5 89,490   0 
  kTd  
f(2)  
 447 0 0 447 0 0  0.00546 in 
 0 1.868 500.5 0 1.868 500.5   0.00879 in 
  
 0 500.5 89,490 0 500.5 179,000  0.00172 rad 
 2.44 k  k(2) Td
 0.877 k 
 
158 k  in 
  kd  
f(2) 
 2.44 k 
 0.877 k 
 
312 k  in 

Element 2: Since there are no applied loads on element 2,

there are no equivalent nodal forces to account for.
Therefore, the above equations are the final local nodal forces
 2.44 k 
 0.877 k 
 
158 k  in 
  kd  
f(2) 
 2.44 k 
 0.877 k 
 
312 k  in 


 S C 0 0 0 0 
 
C 1 S 0  0 0 1 0 0 0
T 
 0 0 0 C S 0
 0 0 0 S C 0 
 
 0 0 0 0 0 1
1 0 0 0 0 0  u40.0103 in   0.0103 in 
0 
0  v0.000956

1 0 0 0
 4 in   0.000956 in 
  
0 0 1 0 0 0  40.00172
 rad  0.00172 rad 
Td       
0 0 0 1 0 0  u3  0   0 
0 0 0 0 1 0  v 3  0   0 
     
0 0 0 0 0 1 3  0   0 


 C1 0 0 C1 0 0 
 0 12C2 6LC2 0 12C2 6LC2 
 
 0 6LC2 4C2L2 0 6LC2 2C2L2 
k   
 C1 0 0 C1 0 0 
 0 12C2 6LC2 0 12C2 6LC2 
 
 0 6LC2 2C2L2 0 6LC2 4C2L2 
AE (8)30  106 EI 30  106 (800)

C1    400 k in C2    0.111 k in
12  50 12  50 
3
L L3


 400 0 0 400 0 0   0.0103 in 
 0 1.335 400 0 1.335 400   0.000956 in 
  
 0 400 160,000 0 400 80,000  0.00172 rad 
  kTd  
f(3)  
 400 0 0 400 0 0  0 
 0 1.335 400 0 1.335 400   0 
  
 0 400 80,000 0 400 160,000   0 
 4.12 k  k(3) Td
 0.687 k 
 
275 k  in 
  kd  
f(3) 
 4.12 k 
 0.687 k 
 
137 k  in 

Element 3: Since there are no applied loads on element 3,

there are no equivalent nodal forces to account for.
Therefore, the above equations are the final local nodal forces

The frame shown below is fixed at nodes 2 and 3 and subjected

to a concentrated load of 500 kN applied at node 1.
For the bar, A = 1 x 10-3 m2, for the beam, A = 2 x 10-3 m2,
I = 5 x 10-5 m4, and L = 3 m. Let E = 210 GPa for both
elements.
Bar
Beam

C 1 S 0
12I 12(5  10 5 )
  6.67  105 m 2
L2 (3)2
6I 6(5  10 5 ) u1 v1 1
  104 m3
L 3
2 0 0 
k (1)
 70  10 0 0.067 0.10  kN
3 
E 210  106   m
  70  106 kN / m 3
L 3 0 0.10 0.20 

C  0.707 S  0.707
u1 v1
k (2)

 
103 m 2 210  106 kN m2 0.5 0.5 
0.5 0.5  m
kN
4.24 m  
u1 v1
0.354 0.354  kN
k (2)  70  103   m
0.354 0.354 

Assembling the elemental stiffness matrices we obtain the
global stiffness matrix:
 2.354 0.354 0 
K  70  10 0.354 0.421 0.10  kN m
 3
 
 0 0.10 0.20 
The global equations are:
 0  2.354 0.354 0  u1 

    
500 kN   70  10 kN m 0.354 0.421 0.10  v1 
3
  0.10 0.20  1 

 0   0

u1   0.00388 m 
   
v1    0.0225 m 
  0.0113 rad 
 1  

Bar Element: The bar element force-displacement equations
can be obtained using f’ = k’Td
C  0.707 S  0.707
 u1 
 
 f1x  AE  1 1 C S 0 0   v1 
   
f3x  L  1 1   0 0 C S  u3 
v 3 
AE
f1x  Cu1  Sv1   670 kN
L
AE
f3x   Cu1  Sv1   670 kN
L

Beam Element: The bar element force-displacement equations
 C S 0 0 0 0
 S C 0 0 0 0 
  C 1
 0 0 1 0 0 0
T  S 0
 0 0 0 C S 0
 0 0 0 S C 0 
 
 0 0 0 0 0 1
1 0 0 0 0 0   u0.00388
1 m   0.00388 m 
0   

1 0 0 0 0  v10.0225
  m   0.0225 m 
  
0 0 1 0 0 0   0.0113
1 rad  0.0113 rad 
d  Td       
0 0 0 1 0 0  u2  0   0 
0 0 0 0 1 0  v 2  0   0 
     
0 0 0 0 0 1  2  0   0 

 C1 0 0 C1 0 0 
 0 12C2 6LC2 0 12C2 6LC2 
 
 0 6LC2 4C2L2 0 6LC2 2C2L2 
k   

 1C 0 0 C1 0 0 
 0 12C2 6LC2 0 12C2 6LC2 
 
 0 6LC2 2C2L2 0 6LC2 4C2L2 
AE (0.002)210  106
C1    140  103 kN
m
L 3
EI 210  106 (5  10 5 )
C2    388.89 kN m
3 
3
L3

2 0 0 2 0 0   0.00388 m 
 0 0.067 0.10 0 0.067 0.10   0.0225 m 
  
3  0 0.10 0.20 0 0.10 0.10  0.0113 kN  m 
  kd  70  10 
f(1)  
 2 0 0 2 0 0  0 
 0 0.067 0.10 0 0.067 0.10   0 
  
0 0.10 0.10 0 0.10 0.20   0 
 f1x   473 kN  k(1) d
 f    26.5 kN 
 1y   
 m1   0.0 
  
f(1) 
 f2x   473 kN 
 f2y   26.5 kN 
   
m2  78.3 kN  m 


The frame is fixed at nodes 1 and 3 and subjected to a moment

of 20 kN-m applied at node 2
Assume A = 2 x 10-2 m2, I = 2 x 10-4 m4, and E = 210 GPa for all
elements.


C 0 S 1
12I 12(2  10 4 )
  1.5  10 4 m 2
L2 (4)2
6I 6(2  104 )
  3  10 4 m 3
L 4
E 210  106
  5.25  107 kN m3

L 4 u
2
v
2 2
0.015 0 0.03 
 5.25  10  0 0  kN m
The parts of k associated (1) 5
k 2
with node 2 are:  
 0.03 0 0.08 


C 1 S0
12I 12(2  10 4 )
2
 2
 9.6  10 4 m 2
L (5)
6I 6(2  104 )
  2.4  10 4 m 3
L 5
E 210  106
  4.2  107 kN m3
L 5 u v 
2 2 2
2 0 0 
k  4.2  10 0 0.0096 0.024  kN m
5 
The parts of k associated (2)
with node 2 are:  
0 0.024 0.08 

Assembling the elemental stiffness matrices we obtain the
global stiffness matrix:
0.8480 0 0.0158 
K  10 6
0 1.0500 0.0101 kN m
 
0.0158 0.0101 0.0756 
 0  0.8480 0 0.0158  u2 

    
 0   10  0
6
1.0500 0.0101 v 2 

20 kN  m   
  0.0158 0.0101 0.0756  2 

u2  4.95  10 m 
6
   6 
v 2   2.56  10 m 
   2.66  10 4 rad 
 2  

obtained using f’ = k’Td
 C S 0 0 0 0
 S C 0 0 0 0 
  C 0
 0 0 1 0 0 0
T  S 1
 0 0 0 C S 0
 0 0 0 S C 0 
 
 0 0 0 0 0 1
0 1 0 0 0 0   u1  0   0 
 1 0 0 0 0 0   v1  0   0 
    
0 0 1 0 0 0   1  0   0 
d  Td     6   6 
0 0 0 0 1 0  u24.95
  10 m    2.56  10 m 
0 0 0 1 0 0  v22.56
  106 m   4.95  10 6 m 
     
0 0 0 0 0 1  2.66 4
2   10 rad 
4
2.66  10 rad 

 C1 0 0 C1 0 0 
 0 12C2 6LC2 0 12C2 6LC2 
 
 0 6LC2 4C2L2 0 6LC2 2C2L2 
k   

 1C 0 0 C1 0 0 
 0 12C2 6LC2 0 12C2 6LC2 
 
 0 6LC2 2C2L2 0 6LC2 4C2L2 
AE (2  102 )210  106

C1    1.05  106 kN
m
L 4
EI 210  106 (5  10 5 )
C2    388.89 kN m
3
3
L3

 200 0 0 200 0 0  0 
 0 1.5 3 0 1.5 3   0 
  
3  0 3 8 0 3 4  0 
  kd  5.25  10 
f(1)  6 
 200 0 0 200 0 0   2.56  10 m 
 0 1.5 3 0 1.5 3   4.95  10 6 m 
  
 0 3 4 0 3 8  2.66  10 4 rad 
 f1x   2.69 kN  k(1) d

f   4.2 kN 
 1y   
 m1   5.59 kN  m 
  
f(1) 
 f2x   2.69 kN 
 f2y   4.2 kN 
   
m2  11.17 kN  m 

 C S 0 0 0 0
 S C 0 0 0 0 
  C 1
 0 0 1 0 0 0
T  S 0
 0 0 0 C S 0
 0 0 0 S C 0 
 
 0 0 0 0 0 1
1 0 0 0 0 0  u 24.95  106 m   4.95  10 6 m 

0 1 0 0 0 0  v  6   
 2 2.56  10 m   2.56  10 m 
6

0 0 1 0 0 0  2.66
2  104 rad  2.66  10 4 rad 
d  Td    u    
0 0 0 1 0 0   3  0   0 
0 0 0 0 1 0  v 3  0   0 
      
0 0 0 0 0 1 3  0   0 

 C1 0 0 C1 0 0 
 0 12C2 6LC2 0 12C2 6LC2 
 
 0 6LC2 4C2L2 0 6LC2 2C2L2 
k   

 1C 0 0 C1 0 0 
 0 12C2 6LC2 0 12C2 6LC2 
 
 0 6LC2 2C2L2 0 6LC2 4C2L2 
AE (2  10 2 )210  106
C1    0.84  106 kN
m
L 5
EI 210  106 (2  10 4 )
C2    336 kN m
5
3
L3

 200 0 0 200 0 0   4.95  106 m 
 0  
0.96 2.40 0 0.96 2.40   2.56  106 m 
 
 0 2.40 8 0 2.40 4  2.66  10 4 rad 
  kd  4.2  103 
f(2)  
 200 0 0 200 0 0  0 
 0 0.96 2.40 0 0.96 2.40   0 
  
 0 2.40 4 0 2.40 8  0 
 f2x   4.16 kN  k(2) d
 f    2.69 kN 
 2y   
m2   8.92 kN  m 
  
f(2) 
 f3x   4.16 kN 
 f3y   2.69 kN 
   
m3   4.47 kN  m 

Inclined or Skewed Supports
If a support is inclined, or skewed, at some angle  for the
global x axis, as shown below.
The boundary conditions on the displacements are not in the

global x-y directions but in the x’-y’ directions.

We must transform the local boundary condition of v’3 = 0 (in
local coordinates) into the global x-y system.

Therefore, the relationship between of the components of the
displacement in the local and the global coordinate systems at
node 3 is:
u '3   cos  sin  0  u3 

    
v '3     sin  cos  0  v 3 

 '   0 1 3 
 3  0
We can rewrite the above expression as:
 cos  sin  0
d '3   [t3 ]d3  t3     sin cos  0

 0 0 1

We can apply this sort of transformation to the entire
displacement vector as:
d '  [Ti ]d or d  [Ti ]T d '

where the matrix [Ti] is:  [I ] [0] [0] 
[Ti ]  [0] [I ] [0] 
 
[0] [0] [t3 ]
Both the identity matrix [I] and the matrix [t3] are 3 x 3 matrices.

The force vector can be transformed by using the same
transformation.
f '  [Ti ]f 
In global coordinates, the force-displacement equations are:
f   [K ]d
Applying the skewed support transformation to both sides of the
force-displacement equation gives:
[Ti ] f   [Ti ][K ] d 
By using the relationship between the local and the global

displacements, the force-displacement equations become:
[Ti ] f   [Ti ][K ][Ti ]T d '  f '  [Ti ][K ][Ti ]T d '

Therefore the global equations become:
 F1x   u1 
F  v 
 1y   1
 M1   1 
   
 F2 x   u2 
 F2 y   [Ti ][K ][Ti ]
T
 v2 
M   
 2   2
F '3 x  u '3  Elemental
F '  v '  coordinates
 3y   3
 M3   3 
End of Chapter 5a

Grid Equations
A grid is: a structure on which the loads are applied
perpendicular to the plane of the structure, as opposed to a
plane frame where loads are applied in the plane of the
structure.
Both torsional and bending moment continuity are maintained at

each node in a grid element.

Grid Equations
Examples of a grid structure are floors and bridge deck
systems.

Grid Equations
Examples of a grid structure are floors and bridge deck
systems.

Grid Equations
A representation of the grid element is shown below:
The degrees of freedom for a grid element are: a vertical

displacement v’i (normal to the grid), a torsional rotation ’ix
about the x’ axis, and a bending rotation ’iz about the z’ axis.
The nodal forces are: a transverse force f’iy, a torsional m’ix

moment about the x’ axis, and a bending moment m’iz about
the z’ axis.

Grid Equations
Let’s derive the torsional rotation components of the element
stiffness matrix.
Consider the sign convention for nodal torque and angle of twist
shown the figure below.
Nodal Elemental
A linear displacement function is assumed.   a1  a2 x 

coefficients gives:     
   2 x 1x  x   1x
 L 

Grid Equations
Let’s derive the torsional rotation components of the element
stiffness matrix.
Consider the sign convention for nodal torque and angle of twist
shown the figure below.
Nodal Elemental
1x 
Or in matrix form:    N1 N2    
2 x 
x x
where: N1  1  N2 
L L

Grid Equations
To obtain the relationship between the shear strain  and the
angle of twist ’ consider the torsional deformation of the bar
as shown below.
If we assume that all radial lines, such as OA, remain straight

during twisting or torsional deformation, then the arc length
AB is: d 
AB   max dx   R d    max  R
dx 

Grid Equations
as shown below.
At any radial position, r, we have, from similar triangles OAB

and OCD:
d  r
 r  2 x  1x 
dx  L

Grid Equations
as shown below.
The relationship between shear stress and shear strain is:

  G
where G is the shear modulus of the material.

Grid Equations
as shown below.
 J
From elementary mechanics of materials, we get: mx 
R
Where J is the polar moment of inertia for a circular cross
section or the torsional constant for non-circular cross
sections.

Grid Equations

Grid Equations
GJ
Rewriting the above equation we get: mx   2 x  1x 
L
m1x  mx
The nodal torque sign convention gives:
m2 x  mx
GJ GJ
Therefore: m1x   1x  2 x  m2 x   2 x  1x 
L L

 m1x  GJ  1 1 1x 
    
m2 x  L  1 1  2 x 

Grid Equations
Combining the torsional effects with shear and bending effects,
we obtain the local stiffness matrix equations for a grid
element.
 f1y   L3
12 EI
0 6 EI
L2
 12L3EI 0 6 EI
L2
  v 
 m   0 GJ
 GJL
 1
0  1x 
 1x   6EI
0 0
L
 
 m1z   L2
2 EI 
0 4 EI
 6LEI2 0   
   12EI   1z 
L L

f 
 2 y    L3 0  6LEI2 12EI
0  6LEI2   v 2 
L3
m2 x   0  GJL GJ

0  2 x 
 
0 0
 L
 
 4 EI   
 2 z   6LEI2
m 0 2 EI
L  6LEI2 0 L   2z 

Grid Equations
The transformation matrix relating local to global degrees of
freedom for a grid is:
1 0 0 0 0 0 
0 C S 0 0 0 
 
0 S C 0 0 0 
TG   
0 0 0 1 0 0 
0 0 0 0 C S 
 
0 0 0 0 S C 
where  is now positive taken counterclockwise from x to x’ in
the x-z plane: therefore:
x j  xi z j  zi
C  cos   S  sin 
L L

Grid Equations
The transformation matrix relating local to global degrees of
freedom for a grid is:
1 0 0 0 0 0 
0 C S 0 0 0 
 
0 S C 0 0 0 
TG   
0 0 0 1 0 0 
0 0 0 0 C S 
 
0 0 0 0 S C 
The global stiffness matrix for a grid element arbitrary oriented
in the x-z plane is given by:
kG  TG TkG TG

Grid Example 1
The frame is fixed at nodes 2, 3, and 4, and is subjected to a

load of 100 kips applied at node 1.
Assume I = 400 in4, J = 110 in4, G = 12 x 10 3 ksi, and

E = 30 x 10 3 ksi for all elements.

Grid Example 1
To facilitate a timely solution, the boundary conditions at nodes

2, 3, and 4 are applied to the local stiffness matrices at the
beginning of the solution. v 2  2 x  2 z  0
v 3  3 x  3 z  0
v 4  4 x  4 z  0

Grid Example 1
Recall the general elemental stiffness matrix:
 f1y   L3
12 EI
0 6 EI
L2
 12L3EI 0 6 EI
L2
  v 
 m   0 GJ
 GJ
 1 
0  1x
 1x   6EI
0 0
L L
 
 m1z   L2
2 EI 
0 4 EI
 6LEI2 0   
   12EI   1z 
L L

f 
 2 y    L3 0  6LEI2 12 EI
0  6LEI2   v 2 
L3
m2 x   0  GJ GJ

0  2 x 
 
0 0
 L L
 
 4 EI   
 2 z   6LEI2
m 0 2 EI
L
 6LEI2 0 L   2z 

Grid Example 1
Recall the general transformation matrix:
1 0 0 0 0 0 
0 C S 0 0 0 
  kG  TG TkG TG
0 S C 0 0 0 
TG   
0 0 0 1 0 0 
0 0 0 0 C S 
 
0 0 0 0 S C 

Grid Example 1
Beam Element 1:
x2  x1 0  20
C  cos    0.894
L(1) 22.36
z2  z1 20  10
S  sin    0.447
L(1) 22.36
12EI 12(30  103 )(400) 6EI 6(30  103 )(400)

  7.45 k   1,000 k
L3 (22.36  12)3 in L2 (22.36  12)2
4EI 4(30  103 )(400) GJ (12  103 )(110)

  179,000 k  in   4,920 k  in
L (22.36  12) L (22.36  12)

Grid Example 1
Beam Element 1:
The global stiffness matrix for element 1, considering only the
parts associated with node 1, and the following relationship:
kG  TG TkG TG
1 0 0  1 0 0 
TG  0 0.894 0.447 
 TG T
 0 0.894 0.447 

   
0 0.447 0.894  0 0.447 0.894 
v1 1x 1z
 7.45 0 1,000 
k(1)  0 4,920 0 k
  in
1,000 0 179,000 

Grid Example 1
Beam Element 1:
kG  TGT kG TG
1 0 0   7.45 0 1,000   1 0 0 
kG (1)  0 0.894 0.447   0 4,920 0  0 0.894 0.447 
   
0 0.447 0.894  1,000 0 179,000  0 0.447 0.894 
v1 1x 1z
 7.45 447 894 
kG (1)   447 39,700 69,600  k
  in
 894 69,600 144,000 

Grid Example 1
Beam Element 2:
x3  x1 0  20
C  cos    0.894
L(2) 22.36
z3  z1 0  10
S  sin    0.447
L(2) 22.36
12EI 12(30  103 )(400) 6EI 6(30  103 )(400)

  7.45 k   1,000 k
L3 (22.36  12)3 in L2 (22.36  12)2
4EI 4(30  103 )(400) GJ (12  103 )(110)

  179,000 k  in   4,920 k  in
L (22.36  12) L (22.36  12)

Grid Example 1
Beam Element 2:
kG  TGT kG TG
1 0 0  1 0 0 
TG  0 0.894 0.447 
 TG T
 0 0.894 0.447 

   
0 0.447 0.894  0 0.447 0.894 
v1 1x 1z
 7.45 0 1,000 
k(2)
  0 4,920 0 k
  in
1,000 0 179,000 

Grid Example 1
Beam Element 2:
kG  TGT kG TG
1 0 0   7.45 0 1,000   1 0 0 
kG (2)  0 0.894 0.447   0 4,920 0  0 0.894 0.447 
   
0 0.447 0.894  1,000 0 179,000  0 0.447 0.894 
v1 1x 1z
 7.45 447 894 
kG (2)
 447 39,700 69,600  k

  in
 894 69,600 144,000 

Grid Example 1
Beam Element 3:
x4  x1 20  20
C  cos   0
L(3) 10
z4  z1 0  10
S  sin    1
L(3) 10
12EI 12(30  103 )(400) 6(30  103 )(400)

  83.3 k / in  5,000 k
L3 (10  12)3 (10  12)2
4EI 4(30  103 )(400) GJ (12  103 )(110)

  400,000 k  in   11,000 k  in
L (10  12) L (10  12)

Grid Example 1
Beam Element 3:
kG  TGT kG TG
1 0 0  1 0 0
TG  0 0 1 TG T
  0 0 1
   
0 1 0  0 1 0 
v1 1x 1z
 83.3 0 5,000 
k(3)  0 11,000 0 k
  in
5,000 0 400,000 

Grid Example 1
Beam Element 3:
kG  TGT kG TG
 1 0 0   83.3 0 5,000   1 0 0 
kG (3)  0 0 1  0 11,000 0   0 0 1
   
0 1 0  5,000 0 400,000  0 1 0 
v1 1x 1z
 83.3 5,000 0 
kG (3)  5,000 400,000 0 k
  in
 0 0 11,000 

Grid Example 1
Superimposing the three elemental stiffness matrices gives:
v1 1x 1z
 98.2 5,000 1,790 

KG  5,000 479,000 0 
 
 1,790 0 299,000 
F1y  100 k   98.2 5,000 1,790   v1 

     
 M1x  0    5,000 479,000 0
  1x 
 M  0   1,790 299,000  1z 
 1z   0

Grid Example 1
 v1   2.83 in 
Solving the above equations gives:    
1x    0.0295 rad 
   
1z  0.0169 rad 

Grid Example 1
 v1   2.83 in 
1x    0.0295 rad 
   
1z  0.0169 rad 

Grid Example 1
 v1   2.83 in 
1x    0.0295 rad 
   
1z  0.0169 rad 
The results indicate that the y displacement at node 1 is
downward as indicated by the minus sign.
The rotation about the x-axis is positive.
The rotation about the z-axis is negative.
Based on the downward loading location with respect to the

supports, these results are expected.

Grid Example 1
Beam Element 1: The grid element force-displacement
equations can be obtained using f’ = k’GTGd
x2  x1 0  20
C  cos    0.894
L(1) 22.36
z2  z1 20  10
S  sin    0.447
L(1) 22.36
1 0 0 0 0 0  1 0 0 0 0 0 
0 C S 0 0 0  0 0.894 0.447 0 0 0 
   
0 S C 0 0 0  0 0.447 0.894 0 0 0 
TG     
0 0 0 1 0 0  0 0 0 1 0 0 
0 0 0 0 C S  0 0 0 0 0.894 0.447 
   
 0 0 0 0 S C   0 0 0 0 0.447 0.894 

Grid Example 1
1 0 0 0 0 0   2.83 in   2.83 in 
0 0.894 0.447  
0 0 0   0.0295 rad  0.0339 rad 
  
0 0.447 0.894 0 0 0  0.0169 rad  0.00192 rad 
TG d      
0
0 0 0 1 0 0  0   
0 0 0 0 0.894 0.447   0   0 
    
 0 0 0 0 0.447 0.894   0   0 

Grid Example 1
 7.45 0 1,000 7.45 0 1,000   2.83 in 
 0 4,920 0 0 4,920 0  0.0339 rad 
  
 1,000 0 179,000 1,000 0 89,500  0.00192 rad 
f(1)  kTd    
 7.45 0 1,000 7.45 0 1,000   0 
 0 4,920 0 0 4,920 0  0 
  
 1,000 0 89,500 1,000 0 179,000  0 
k(1) d'
 f1y   19.2 k 
 m   167 k  in 
 1x   
 m1z  2,480 k  in 
  
 f2y   19.2 k 
m2 x   167 k  in 
   
 m2 z   2,660 k  in 

Grid Example 1
x3  x1 0  20
C  cos    0.894
L(2) 22.36
z3  z1 0  10
S  sin    0.447
L(2) 22.36
1 0 0 0 0 0  1 0 0 0 0 0 
0 C S 0 0 0  0 0.894 0.447 0 0 0 
   
0 S C 0 0 0  0 0.447 0.894 0 0 0 
TG     
0 0 0 1 0 0  0 0 0 1 0 0 
0 0 0 0 C S  0 0 0 0 0.894 0.447 
   
0 0 0 0 S C  0 0 0 0 0.447 0.894 

Grid Example 1
1 0 0 0 0   2.83 in   2.83 in 
0
0 0.894 0.447  
0 0 0   0.0295 rad  0.0188 rad 
  
0 0.447 0.894 0 0 0  0.0169 rad   0.0283 rad 
TG d      
0
0 0 0 1 0 0  0   
0 0 0 0 0.894 0.447   0   0 
    
 0 0 0 0 0.447 0.894   0   0 

Grid Example 1
 7.45 0 1,000 7.45 0 1,000   2.83 in 
 0 4,920 0 0 4,920 0  0.0188 rad 
  
 1,000 0 179,000 1,000 0 89,500   0.0283 rad 
f(2 )  kTd    
 7.45 0 1,000 7.45 0 1,000   0 
 0 4,920 0 0 4,920 0  0 
  
 1,000 0 89,500 1,000 0 179,000   0 
k(2) d'
 f1y   7.23 k 
 m   92.5 k  in 
 1x   
 m1z  2,240 k  in 
  
 f3y   7.23 k 
m3 x   92.5 k  in 
   
 m3 z   295 k  in 

Grid Example 1
x4  x1 20  20
C  cos   0
L(3) 10
z4  z1 0  10
S  sin    1
L(3) 10
1 0 0 0 0 0  1 0 0 0 0 0
 0 C S 0 0 0   0 0 1 0 0 0
   
0 S C 0 0 0  0 1 0 0 0 0
TG     
0 0 0 1 0 0  0 0 0 1 0 0
0 0 0 0 C S  0 0 0 0 0 1
   
0 0 0 0 S C  0 0 0 0 1 0

Grid Example 1
1 0 0 0 0 0   2.83 in   2.83 in 
0  
0 1 0 0 0   0.0295 rad  0.0169 rad 
  
0 1 0 0 0 0  0.0169 rad  0.0295 rad 
TG d      
0
0 0 0 1 0 0  0   
0 0 0 0 0 1  0   0 
    
0 0 0 0 1 0  0   0 

Grid Example 1
 83.3 0 5,000 83.3 0 5,000   2.83 in 
 0 11,000 0 0 11,000 0  0.0169 rad 
  
5,000 0 400,000 5,000 0 200,000  0.0295 rad 
f(3 )  kTd    
 83.3 0 5,000 83.3 0 5,000   0 
 0 11,000 0 0 11,000 0  0 
  
 5,000 0 200,000 5,000 0 400,000  0 
k(3) d'
 f1y   88.1 k 
 m   186 k  in 
 1x   
 m1z   2,340 k  in 
  
 f4y   88.1 k 
m4 x   186 k  in 
   
 m4 z   8,240 k  in 

Grid Example 1

Grid Example 1 – Forces in the y-direction

Grid Example 1 - Moment about the y’ axis

Grid Example 1 - Torsional Moment about the x’ axis

Grid Example 1

Grid Example 1
To check the equilibrium of node 1, the local forces and
moments for each element need to be transformed to global
coordinates. Recall, that:
f   Tf  f  T T f  T T  T -1
Since we are only checking the forces and moments at node 1,

we need only the upper-left-hand portion of the transformation
matrix TG
Element 1:
 f1y   1 0 0   19.2 k   19.2 k 

      
m1x   0 0.894 0.447   167 k  in    1,260 k  in 
     
m1z  0 0.447 0.894   2,480 k  in  2,150 k  in 

Grid Example 1
To check the equilibrium of node 1, the local forces and
moments for each element need to be transformed to global
coordinates. Recall, that:
f   Tf  f  T T f  T T  T -1
Element 2:
 f1y   1 0 0  7.23 k   7.23 k 
       
m1x   0 0.894 0.447   92.5 k  in    1,080 k  in 
m  0 0.447 0.894  2,240 k  in  1,960 k  in 
 1z      
Element 3:
 f1y   1 0 0   88.1 k   88.1 k 
      
m1x   0 0 1 2,340 k  in   2,340 k  in 
m  0 1 0   186 k  in   186 k  in 
 1z      

Grid Example 1
Check the equilibrium of node 1. Element 1:
 f1y   19.2 k 
   
Remember that forces and m1x    1,260 k  in 
   
moments from each element are  m1z  2,150 k  in 
equal in magnitude but opposite
in sign. Element 2:
 f1y   7.23 k 
   

 1x  
m 1,080 k  in 
 m   1,960 k  in 
 1z   
Element 3:
 f1y   88.1 k 
   
m1x    2,340 k  in 
 m   186 k  in 
 1z   

Grid Example 1
Check the equilibrium of node 1.
M 1x  1,260  1,080  2,340  0.0 k  in
M 1z  2,150  1,060  186  4.0 k  in
F 1y  100  7.23  19.2  88.1  0.07 k


Grid Example 2
Element 1: nodes 1 to 2
The frame is fixed at nodes 1 and 3, and is subjected to a load

of 22 kN applied at node 2.
Assume I = 16.6 x 10-5 m4, J = 4.6 x 10-5 m4, G = 84 GPa,

and E = 210 GPa for all elements.

Grid Example 2
To facilitate a timely solution, the boundary conditions at nodes

1 and 3 are applied to the local stiffness matrices at the
beginning of the solution.
v1  1x  1z  0
v 3  3 x  3 z  0

Grid Example 2
Recall the general elemental stiffness matrix:
 f1y   L3
12 EI
0 6 EI
L2
 12L3EI 0 6 EI
L2
  v 
 m   0 GJ
 GJL
 1
0  1x 
 1x   6EI
0 0
L
 
 m1z   L2
2 EI 
0 4 EI
 6LEI2 0   
   12EI   1z 
L L

f 
 2 y    L3 0  6LEI2 12EI
0  6LEI2   v 2 
L3
m2 x   0  GJL GJ

0  2 x 
 
0 0
 L
 
 4 EI   
 2 z   6LEI2
m 0 2 EI
L
 6LEI2 0 L   2z 

Grid Example 2
Recall the general transformation matrix:
1 0 0 0 0 0 
0 C S 0 0 0 
  kG  TG TkG TG
0 S C 0 0 0 
TG   
0 0 0 1 0 0 
0 0 0 0 C S 
 
0 0 0 0 S C 

Grid Example 2
Beam Element 1: from nodes 1 to 2
x x 3
C  cos  2 (1) 1   1
L 3
z2  z1 0
S  sin   0
L(1) 3
12EI 12(210  106 )(16.6  10 5 )
  1.55  10 4 kN / m
L3 (3)3
6EI 6(210  106 )(16.6  10 5 )
  2.32  104 kN
L2 (3)2
4EI 4(210  106 )(16.6  10 5 )
  4.65  10 4 kN·m
L 3
GJ (84  106 )(4.6  10 5 )
  0.128  104 kN·m
L 3

Grid Example 2
Beam Element 1:
kG  TGT kG TG
1 0 0 1 0 0
TG  0 1 0  TGT
 0 1 0 
   
0 0 1 0 0 1
v2 2 x 2 z
 1.55 0 2.32
k(1)
 10
4  0 0.128 0  KN m
 
 2.32 0 4.65 

Grid Example 2
Beam Element 1:
kG  TGT kG TG
 1 0 0   1.55 0 2.32  1 0 0 
kG (1)  10 4 0 1 0   0 0.128 0  0 1 0  kN
    m
0 0 1  2.32 0 4.65  0 0 1
v2 2 x 2 z
 1.55 0 2.32
kG (1)
 10
4  0 0.128 0  kN
  m
 2.32 0 4.65 

Grid Example 2
Beam Element 2: from nodes 3 to 2
x x 0
C  cos  3 (2) 2   0
L 3
z3  z2 3
S  sin    1
L(1) 3
12EI 12(210  106 )(16.6  10 5 )
  1.55  10 4 kN / m
L3 (3)3
6EI 6(210  106 )(16.6  10 5 )
  2.32  104 kN
L2 (3)2
4EI 4(210  106 )(16.6  10 5 )
  4.65  10 4 kN·m
L 3
GJ (84  106 )(4.6  10 5 )
  0.128  104 kN·m
L 3

Grid Example 2
Beam Element 2:
kG  TGT kG TG
1 0 0  1 0 0
TG  0 0 1 TGT
  0 0 1
   
0 1 0  0 1 0 
v2 2 x 2 z
 1.55 0 2.32
k(2)  104  0 0.128 0  KN m
 
 2.32 0 4.65 

Grid Example 2
Beam Element 2:
kG  TGT kG TG
 1 0 0   1.55 0 2.32  1 0 0 
kG (2)  10 4 0 0 1  0 0.128 0  0 0 1 kN
    m
0 1 0   2.32 0 4.65  0 1 0 
v2 2 x 2 z
1.55 2.32 0 
kG (2)  10 4  2.32 4.65 0  kN
  m
 0 0 0.128 

Grid Example 2
Superimposing the two elemental stiffness matrices gives:
v2 2 x 2 z
 3.10 2.32 2.32
KG  10  2.32 4.78
4
0  kN
  m
 2.32 0 4.78 
F2 y  22 kN   3.10 2.32 2.32  v 2 

  4   
 M2 x  0   10  2.32 4.78 0  2 x 

 M 0   
 2z  
 2.32 0 4.78  2 z 

Grid Example 2
 v  0.00259 m 
Solving the above equations gives:  2   
2 x    0.00126 rad 
   
2 z  0.00126 rad 

Grid Example 2
 v  0.00259 m 
2 x    0.00126 rad 
   
2 z  0.00126 rad 

Grid Example 2
 v  0.00259 m 
2 x    0.00126 rad 
   
2 z  0.00126 rad 
The results indicate that the y displacement at node 1 is
downward as indicated by the minus sign.
The rotation about the x-axis is positive.
The rotation about the z-axis is negative.
Based on the downward loading location with respect to the

supports, these results are expected.

Grid Example 2
x2  x1 3
C  cos   1
L(1) 3
z2  z1 0
S  sin   0
L(1) 3
1 0 0 0 0 0  1 0 0 0 0 0
0 C S 0 0 0  0 1 0 0 0 0
   
0 S C 0 0 0  0 0 1 0 0 0
TG     
0 0 0 1 0 0  0 0 0 1 0 0
0 0 0 0 C S  0 0 0 0 1 0
   
0 0 0 0 S C  0 0 0 0 0 1

Grid Example 2
1 0 0 0 0 0  0   0 
0 1 0 0 0 0  0   0 
    
0 0 1 0 0 0  0   0 
TG d      
0 0 0 1 0 0   0.00259 m   0.00259 m 
0 0 0 0 1 0   0.00126 rad   0.00126 rad 
    
0 0 0 0 0 1 0.00126 rad  0.00126 rad 

Grid Example 2
 1.55 0 2.32 1.55 0 2.32   0 
 0 0.128 0 0 0.128 0  0 
  
4  2.32 0 4.65 2.32 0 2.33   0 
f(1)  kTd  10   
 1.55 0 2.32 1.55 0 2.32  0.00259 m 
 0 0.128 0 0 0.128 0   0.00126 rad 
  
 2.32 0 2.33 2.32 0 4.65  0.00126 rad 
k(1) d'
 f1y   11.0 kN 
 m   1.50 kN  m 
 1x   
 m1z   31.0 kN  m 
  
 f2y  11.0 kN 
m2 x   1.50 kN  m 
   
 m2 z   1.50 kN  m 

Grid Example 2
x3  x 2 0
C  cos   0
L(2) 3
z3  z2 3
S  sin    1
L(1) 3
1 0 0 0 0 0  1 0 0 0 0 0
0 C S 0 0 0  0 0 1 0 0 0
   
0 S C 0 0 0  0 1 0 0 0 0
TG     
0 0 0 1 0 0  0 0 0 1 0 0
0 0 0 0 C S  0 0 0 0 0 1
   
0 0 0 0 S C  0 0 0 0 1 0

Grid Example 2
1 0 0 0 0 0  0   0 
0     
0 1 0 0 0 0 0
    
0 1 0 0 0 0  0   0 
TG d      
0   0.00259 m   0.00259 m
0 0 0 1 0 
0 0 0 0 0 1  0.00126 rad  0.00126 rad 
    
0 0 0 0 1 0  0.00126 rad  0.00126 rad 

Grid Example 2
 1.55 0 2.32 1.55 0 2.32   0 
 0 0.128 0 0 0.128 0  0 
  
4  2.32 0 4.65 2.32 0 2.33   0 
f(2 )  kTd  10   
 1.55 0 2.32 1.55 0 2.32  0.00259 m 
 0 0.128 0 0 0.128 0  0.00126 rad 
  
 2.32 0 2.33 2.32 0 4.65  0.00126 rad 
k(2) d'
 f2y  11.0 kN 
m   1.50 kN  m 
 2x   
 m2 z   1.50 kN  m 
  
 f3y   11.0 kN 
m3 x   1.50 kN  m 
   
 m3 z  31.0 kN  m 

Grid Example 2
The resulting free-body diagrams:

Grid Example 2 – Forces in the y-direction

Grid Example 2 – Torsional Moment about the x’ axis

Grid Example 2 – Moment about the z’ axis

Beam Element Arbitrarily Oriented in Space
In this section, we will develop a beam element that is arbitrarily
oriented in three-dimensions.
This element can be used to analyze three-dimensional frames.
Let consider bending about axes, as shown below.

The y’ axis is the principle axis for which the moment of inertia
is minimum, Iy
The right-hand rule is used to establish the z’ axis and the

maximum moment of inertia, Iz

Bending in the x’-z’ plane: The bending in the x’-z’ plane is
defined by m’y
The stiffness matrix for bending the in the x’-z’ plane is:
 12L 6L2 12L 6L2 

EI y  6L2 4L3 6L2

2L3 
ky  4 

L  12L 6L2 12L 6L2 
 
 6L
2
2L3 6L2 4L3 
where Iy is the moment of inertia about the y’ axis (the weak

axis), therefore: Iy < Iz

Bending in the x’-y’ plane: The bending in the x’-y’ plane is
defined by m’z
The stiffness matrix for bending the in the x’-z’ plane is:
 12L 6L2 12L 6L2 

 
EIz  6L
2
4L3 6L2 2L3 
kz  4
L  12L 6L2 12L 6L2 
 
 6L
2
2L3 6L2 4L3 
where Iz is the moment of inertia about the z’ axis (the strong

axis).

Direct superposition of the bending stiffness matrices with the
effects of axial forces and torsional rotation give:
 AE AE 
 L 0 0 0 0 0  0 0 0 0 0 
L
 
 0 12 E I z
0 0 0
6 E Iz
0 
12 E I z
0 0 0
6 E Iz 
 L3 L2 L3 L2 
 
 0 12 E I y 6 E Iy 12 E I y 6 E Iy
0 0  0 0 0  0  0 
 L3 L2 L3 L2 
 
 0 GJ GJ
0 0 0 0 0 0 0  0 0 
 L L 
 6 E Iy 4 E Iy 6 E Iy 2 E Iy 
 0 0  0 0 0 0 0 0 
 L2 L L2 L 
 6 E Iz 4 E Iz 6 E I z 2 E Iz 
 0 0 0 0 0 0 0 0 
k   L2 L L2 L 

A E AE
 0 0 0 0 0 0 0 0 0 0 
 L L 
 12 E I z 6 E I z 12 E I z 6 E Iz 
 0 0 0 0 0 0 0 0 
 L3 L2 L3 L2 
 12 E I y 6 E Iy 12 E I y 6 E Iy 
 0 0  0 0 0 0 0 0 
 L3 L2 L3 L2 
 GJ GJ 
 0 0 0  0 0 0 0 0 0 0 
 L L 
 6 E Iy 2 E Iy 6 E Iy 4 E Iy 
 0 0  0 0 0 0 0 0 
 L2 L L2 L 
 6 E Iz 2 E Iz 6 E I z 4 E Iz 
 0 L2
0 0 0
L
0
L2
0 0 0
L 

The global stiffness matrix may be obtained using: k  T TkT
 3 x 3  Cxx  Cyx  Czx  

where:  
3 x 3  
T   3 x 3  Cxy  Cyy  Czy  
 3 x 3  Cxz Cyz Czz 

 3 x 3 
the direction cosines, Cij’, are defined as shown below:

The direction cosines of the x’ axis are:
x   cos  xx i  cos  yx  j  cos  zx k
x2  x1 y 2  y1 z2  z1
cos  xx   l cos  yx   m cos  zx   n
L L L
The y’ axis is selected to be perpendicular

to the x’ and the z axes is such a way
that the cross product of global z with x’
results in the y’ axis as shown in the
figure.
i j k
1 m l
z  x  y   0 0 1   i  j D  l 2  m2
D D D
l m n

The z’ axis is determined by the condition that z’ = x’  y’
i j k
1 ln mn
z  x   y   l m n   i j  Dk
D D D
m l 0
Therefore, the transformation matrix becomes:

 
 l m n 
Cxx  Cyx  Czx    
 
 0 
m l
3 x 3  Cxy  Cyy  Czy  
 D D 
Cxz Cyz Czz   ln 
mn
  D 
 D D 

There are two exceptions that arise when using the above
expressions for mapping the local coordinates to the global
system:
(1) when the positive x’ coincides with z
For the this case, it is assumed that y’ is y.
 0 0 1
   0 0 0 
 1 0 0 

There are two exceptions that arise when using the above
expressions for mapping the local coordinates to the global
system:
(2) when the positive x’ is in the opposite direction as z
For the second case, it is assumed that y’ is y.
0 0 1
  0 0 0 
 1 0 0 

If the effects of axial force, both shear forces, twisting moment,
and both bending moments are considered, the stiffness
matrix for a frame element is:
AE AE 
 L 0 0 0 0 0  0 0 0 0 0 
L
 
 0 12 E Iz 6 E Iz 12 E Iz 6 E Iz 
0 0 0 0 0 0 0
 L 1   y 
3
L2 1   y  L3 1   y  L 1   y  
2
 
 12 E I y 6 E I y 12 E I y 6 E Iy 
 0 0 0 0 0 0 0 0 
 L3 1   z  L2 1   z  L3 1   z  L2 1   z  
 
 0 GJ GJ 
0 0 0 0 0 0 0  0 0
 L L 
 
 0 6 E I y  4   z  E Iy 6 E Iy  2   z  E Iy 
0 0 0 0 0 0 0
 L 1   z 
2
L 1   z  L 1   z 
2
L 1   z  
 

 0
6 E Iz
0 0 0
4    E I
y z
0
6 E I z
0 0 0
 2   y  E Iz 
 L 1   y 
2
L 1    L 1   y 
2
L 1   y  
k   A E y

 0 0 0 0 0
AE
0 0 0 0 0 
 L L 
 
 0 12 E Iz 6 E I z 12 E Iz 6 E Iz 
0 0 0 0 0 0 0
 L3 1   y  L2 1   y  L3 1   y  L2 1   y  
 
 12 E I y 6 E Iy 12 E I y 6 E Iy 
 0 0 0 0 0 0 0 0 
 L3 1   z  L2 1   z  L3 1   z  L2 1   z  
 GJ GJ 
 0 0 0  0 0 0 0 0 0 0 
 L L 

 0
6 E I y  2   z  E Iy 6 E Iy  4   z  E Iy 

0 0 0 0 0 0 0
 L2 1   z  L 1   z  L2 1   z  L 1   z  
 
 6 E Iz 2    E I
y z 6 E Iz  4   y  E Iz 
 0 0 0 0 0 0 0 0 
 L2 1   y  L 1    y L2 1   y  L 1   y  

12EI y 12EI z
In this case the symbol  are:  y  2
z 
GAs L GAs L2
Where As is the effective beam cross-section in shear.
AE AE 
 L 0 0 0 0 0  0 0 0 0 0 
L
 
 0 12 E Iz 6 E Iz 12 E Iz 6 E Iz 
0 0 0 0 0 0 0
 L 1   y 
3
L2 1   y  L3 1   y  L 1   y  
2
 
 12 E I y 6 E I y 12 E I y 6 E Iy 
 0 0 0 0 0 0 0 0 
 L3 1   z  L2 1   z  L3 1   z  L2 1   z  
 
 0 GJ GJ 
0 0 0 0 0 0 0  0 0
 L L 
 
 0 6 E I y  4   z  E Iy 6 E Iy  2   z  E Iy 
0 0 0 0 0 0 0
 L2 1   z  L 1   z  L2 1   z  L 1   z  
 

 0
6 E Iz
0 0 0
4    E I
y z
0
6 E I z
0 0 0
 2   y  E Iz 
 L 1   y 
2
L 1    L 1   y 
2
L 1   y  
k   A E y

 0 0 0 0 0
AE
0 0 0 0 0 
 L L 
 
 0 12 E Iz 6 E I z 12 E Iz 6 E Iz 
0 0 0 0 0 0 0
 L 1   y 
3
L 1   y 
2
L 1   y 
3
L 1   y  
2
 
 12 E I y 6 E Iy 12 E I y 6 E Iy 
 0 0 0 0 0 0 0 0 
 L3 1   z  L2 1   z  L3 1   z  L2 1   z  
 GJ GJ 
 0 0 0  0 0 0 0 0 0 0 
 L L 

 0
6 E I y  2   z  E Iy 6 E Iy  4   z  E Iy 

0 0 0 0 0 0 0
 L 1   z 
2
L 1   z  L 1   z 
2
L 1   z  
 
 6 E Iz 2    E I
y z 6 E Iz  4   y  E Iz 
 0 0 0 0 0 0 0 0 
 L 1   y 
2
L 1    y L 1   y 
2
L 1   y  

Recall the shear modulus of elasticity or the modulus of rigidity,
G, is related to the modulus of elasticity and the Poisson’s
ratio,  as:
E
G
2 1   
This is the form of the stiffness matrix used by SAP2000 for its
frame element.

If y and z are set to zero, the stiffness matrix reduces to:
 AE AE 
 L 0 0 0 0 0  0 0 0 0 0 
L
 
 0 12 E I z
0 0 0
6 E Iz
0 
12 E Iz
0 0 0
6 E Iz 
 L3 L2 L3 L2 
 
 0 12 E I y 6 E Iy 12 E I y 6 E Iy
0 0  0 0 0  0  0 
 L3 L2 L3 L2 
 
 0 GJ GJ
0 0 0 0 0 0 0  0 0 
 L L 
 6 E Iy 4 E Iy 6 E Iy 2 E Iy 
 0 0  0 0 0 0 0 0 
 L2 L L2 L 
 6 E Iz 4 E Iz 6 E I z 2 E Iz 
 0 0 0 0 0 0 0 0 
k   L2 L L2 L 

A E AE
 0 0 0 0 0 0 0 0 0 0 
 L L 
 12 E I z 6 E I z 12 E I z 6 E Iz 
 0 0 0 0 0 0 0 0 
 L3 L2 L3 L 2

 12 E I y 6 E Iy 12 E Iy 6 E Iy 
 0 0  0 0 0 0 0 0 
 L3 L2 L3 L2 
 GJ GJ 
 0 0 0  0 0 0 0 0 0 0 
 L L 
 6 E Iy 2 E Iy 6 E Iy 4 E Iy 
 0 0  0 0 0 0 0 0 
 L2 L L2 L 
 6 E Iz 2 E Iz 6 E Iz 4 E Iz 
 0 L2
0 0 0
L
0
L2
0 0 0
L 

Example Frame Application
A bus subjected to a static roof-crush analysis. In this model
599 frame elements and 357 nodes are used.

Example Frame Application
A bus subjected to a static roof-crush analysis. In this model
599 frame elements and 357 nodes are used.

Concept of Substructure Analysis
Sometimes structures are too large to be analyzed as a single
system or treated as a whole; that is, the final stiffness matrix
and equations for solution exceed the memory capacity of the
computer.
A procedure to overcome this problem is to separate the whole

structure into smaller units called substructures.

Concept of Substructure Analysis
For example, the space frame of an airplane, as shown below,
may require thousands of nodes and elements to completely
model and describe the response of the whole structure.
If we separate the aircraft into substructures, such as parts of

the fuselage or body, wing sections, etc., as shown below,
then we can solve the problem more readily and on computers
with limited memory.
Beam Stiffness
Problems:
11. Do problems 5.1, 5.8, 5.13, 5.28, and 5.51 on pages 299
- 327 in your textbook “A First Course in the Finite
Element Method” by D. Logan.
12. Do problems 5.20, 5.23, 5.25, 5.35, and 5.53 on pages

299 - 327 in your textbook “A First Course in the Finite
Element Method” by D. Logan. You may use the
SAP2000 to do frame analysis.
End of Chapter 5b
CIVL 7/8117 Chapter 6 - Plane Stress/Plane Strain Stiffness Equations - Part 1 1/78
Chapter 6a – Plane Stress/Strain Equations
Learning Objectives
• To review basic concepts of plane stress and plane
strain.
• To derive the constant-strain triangle (CST)

element stiffness matrix and equations.
• To demonstrate how to determine the stiffness

matrix and stresses for a constant strain element.
• To describe how to treat body and surface forces

for two-dimensional elements.
Chapter 6a – Plane Stress/Strain Equations
Learning Objectives
• To evaluate the explicit stiffness matrix for the
constant-strain triangle element.
• To perform a detailed finite element solution of a

plane stress problem.
Plane Stress and Plane Strain Equations

In Chapters 2 through 5, we considered only line elements.
Line elements are connected only at common nodes, forming
framed or articulated structures such as trusses, frames, and
grids.
Line elements have geometric properties such as cross-
sectional area and moment of inertia associated with their
cross sections.

However, only one local coordinate along the length of the
element is required to describe a position along the element
(hence, they are called line elements).
Nodal compatibility is then enforced during the formulation of
the nodal equilibrium equations for a line element.
This chapter considers the two-dimensional finite element.

Two-dimensional (planar) elements are thin-plate elements
such that two coordinates define a position on the element
surface.
The elements are connected at common nodes and/or along
common edges to form continuous structures.

Nodal compatibility is then enforced during the formulation of
the nodal equilibrium equations for two-dimensional
elements.
If proper displacement functions are chosen, compatibility
along common edges is also obtained.

The two-dimensional element is extremely important for:
(1) Plane stress analysis, which includes problems such

as plates with holes, fillets, or other changes in
geometry that are loaded in their plane resulting in local
stress concentrations.
Plane Stress Problems

(1) Plane stress analysis, which includes problems such

as plates with holes, fillets, or other changes in
geometry that are loaded in their plane resulting in local
stress concentrations.

(2) Plane strain analysis, which includes problems such

as a long underground box culvert subjected to a
uniform load acting constantly over its length or a long
cylindrical control rod subjected to a load that remains
constant over the rod length (or depth).
Plane Strain Problems

(2) Plane strain analysis, which includes problems such

as a long underground box culvert subjected to a
uniform load acting constantly over its length or a long
cylindrical control rod subjected to a load that remains
constant over the rod length (or depth).

We begin this chapter with the development of the stiffness
matrix for a basic two-dimensional or plane finite element,
called the constant-strain triangular element.
The constant-strain triangle (CST) stiffness matrix derivation is

the simplest among the available two-dimensional elements.
We will derive the CST stiffness matrix by using the principle

of minimum potential energy because the energy formulation
is the most feasible for the development of the equations for
both two- and three-dimensional finite elements.

Formulation of the Plane Triangular Element Equations
We will now follow the steps described in Chapter 1 to formulate

the governing equations for a plane stress/plane strain
triangular element.
First, we will describe the concepts of plane stress and plane
strain.
Then we will provide a brief description of the steps and basic
equations pertaining to a plane triangular element.

Plane Stress
Plane stress is defined to be a state of stress in which the

normal stress and the shear stresses directed
perpendicular to the plane are assumed to be zero.
That is, the normal stress z and the shear stresses xz and yz
are assumed to be zero.
Generally, members that are thin (those with a small z
dimension compared to the in-plane x and y dimensions) and
whose loads act only in the x-y plane can be considered to be
under plane stress.

Plane Strain
Plane strain is defined to be a state of strain in which the

strain normal to the x-y plane z and the shear strains xz
and yz are assumed to be zero.
The assumptions of plane strain are realistic for long bodies

(say, in the z direction) with constant cross-sectional area
subjected to loads that act only in the x and/or y directions and
do not vary in the z direction.

Two-Dimensional State of Stress and Strain
The concept of two-dimensional state of stress and strain and

the stress/strain relationships for plane stress and plane strain
are necessary to understand fully the development and
applicability of the stiffness matrix for the plane stress/plane
strain triangular element.

A two-dimensional state of stress is shown in the figure below.


The infinitesimal element with sides dx and dy has normal

stresses x and y acting in the x and y directions (here on the
vertical and horizontal faces), respectively.

The shear stress xy acts on the x edge (vertical face) in the y
direction. The shear stress yx acts on the y edge (horizontal
face) in the x direction.


Since xy equals yx, three independent stress exist:
     x  y  xy 
T
Recall, the relationships for principal stresses in two-

dimensions are:
x y x y 
2
1       xy   max
2
2  2 
x y  y 
2
2    x    xy   min
2
2  2 


Also, p is the principal angle which defines the normal whose
direction is perpendicular to the plane on which the maximum
or minimum principle stress acts.
2 xy
tan 2 p 
x y


The general two-dimensional state of strain at a point is show
below.


u v u v
x  y   xy  
x x y x


u v u v
x  y   xy  
x x y x
The strain may be written in matrix form as:
     x y  xy 
T


For plane stress, the stresses z, xz, and yz are assumed to
be zero. The stress-strain relationship is:
 x  1  0   x 
  E   
 y    1 0   y 
  1  
2
 xy  0 0 0.5 1      xy 
 x   x  1  0 
    E  
 y   D    y  [D ]  2 
 1 0 
    1 
 xy   xy  0 0 0.5 1    
is called the stress-strain matrix (or the constitutive matrix),

E is the modulus of elasticity, and  is Poisson’s ratio.


For plane strain, the strains z, xz, and yz are assumed to be
zero. The stress-strain relationship is:
 x  1    0   x 
     
0   y 
E
 y   1 
  1   1  2   0
 
 xy   0 0.5     xy 
 x   x  1    0 
      0 
E
 y   D    y  [D ]  1 
    1   1  2   0 
0.5   
 xy   xy   0
is called the stress-strain matrix (or the constitutive matrix),

E is the modulus of elasticity, and  is Poisson’s ratio.


The partial differential equations for plane stress are:
 2u  2u 1     2u  2v 
    
x 2 y 2 2  y 2 xy 
 2v  2v 1     2v  2u 
    
x 2 y 2 2  y 2 xy 

Consider the problem of a thin plate subjected to a tensile load

as shown in the figure below:

Discretize the thin plate into a set of triangular elements. Each

element is define by nodes i, j, and m.

We use triangular elements because boundaries of irregularly

shaped bodies can be closely approximated, and because the
expressions related to the triangular element are
comparatively simple.

This discretization is called a coarse-mesh generation if few

large elements are used.
Each node has two degrees of freedom: displacements in the x
and y directions.

We will let ui and vi represent the node i displacement

components in the x and y directions, respectively.

The nodal displacements for an element with nodes i, j, and m

are:
 di 
d    d j 
d 
 m
where the nodes are ordered counterclockwise around the

element, and
u 
d i    i 
v i 

The nodal displacements for an element with nodes i, j, and m

are:
 ui 
v 
 i
u 
d    j 
v j 
um 
 
v m 

Step 2 - Select Displacement Functions

u( x, y )
The general displacement function is:  i    
v ( x, y )
The functions u(x, y) and v(x, y) must be compatible with the
element type.
Step 3 - Define the Strain-Displacement and

Stress-Strain Relationships
The general definitions of normal and shear strains are:
u v u v
x  y   xy  
x x y x


For plane stress, the stresses z, xz, and yz are assumed to
be zero. The stress-strain relationship is:
 x  1  0   x 
  E   
 y    1 0   y 
  1  
2
 xy  0 0 0.5 1      xy 


For plane strain, the strains z, xz, and yz are assumed to be
zero. The stress-strain relationship is:
 x  1    0   x 
  E   
 y    1 0
 y 
  1   1  2   0

 xy   0 0.5     xy 

Using the principle of minimum potential energy, we can derive

the element stiffness matrix.
f   [k ]d 
This approach is better than the direct methods used for one-
dimensional elements.

Step 5 - Assemble the Element Equations and

Introduce Boundary Conditions
The final assembled or global equation written in matrix form is:
F   [K ]d 
where {F} is the equivalent global nodal loads obtained by

lumping distributed edge loads and element body forces at the
nodes and [K] is the global structure stiffness matrix.


Once the element equations are assembled and modified to
account for the boundary conditions, a set of simultaneous
algebraic equations that can be written in expanded matrix
form as:
F   [K ]d 
Step 7 - Solve for the Element Forces (Stresses)
For the structural stress-analysis problem, important secondary
quantities of strain and stress (or moment and shear force)
can be obtained in terms of the displacements determined in
Step 6.

Derivation of the Constant-Strain Triangular Element
Stiffness Matrix and Equations

Consider the problem of a thin plate subjected to a tensile load
as shown in the figure below:
 ui 
v 
 i
u 
d    j 
v j 
um 
 
v m 

We will select a linear displacement function for each triangular

element, defined as:
u
u( x, y )
Linear representation of u(x, y)
u
 i    
y m
v ( x, y )
ui
x
 a  a2 x  a3 y 
uj
(xi, yi) (xm, ym)  1 
(xj, yj) a4  a5 x  a6 y 
A linear function ensures that the displacements along each

edge of the element and the nodes shared by adjacent
elements are equal.

We will select a linear displacement function for each triangular

element, defined as:
 a1 
a 
 2
 a  a x  a3 y   1 x y 0 0 0  a3 
 i    1 2   
a4  a5 x  a6 y  0 0 0 1 x y  a4 
a5 
 
a6 

To obtain the values for the a’s substitute the coordinated of the
nodal points into the above equations:
ui  a1  a2 xi  a3 y i v i  a4  a5 xi  a6 y i
u j  a1  a2 x j  a3 y j v j  a4  a5 x j  a6 y j
um  a1  a2 xm  a3 y m v m  a4  a5 xm  a6 y m

Solving for the a’s and writing the results in matrix forms gives:
 ui  1 xi y i   a1 
    
y j  a2  a   x  u
1
 u j   1 x j 

u  1 x y m  a3 
 m  m


The inverse of the [x] matrix is:
i
 i  j m 
1  
1
[x]  i  j m  m
2A  j
  i  j  m 
 i  x j y m  y j xm i  y j  y m  i  xm  x j
 j  xi y m  y i xm  j  ym  yi  j  xi  xm
 m  xi y j  y i x j m  y i  y j  m  x j  xi


The inverse of the [x] matrix is:
 i  j m  1 xi yi
1  
1
[x]  i  j m  2A  1 xj yj
2A 
  i  j  m  1 xm ym
2 A  xi  y j  y m   x j  y m  y i   xm  y i  y j 
where A is the area of the triangle



The values of a may be written matrix form as:
 a1   i  j  m   ui 
  1   
a2    i  j m   u j 
a  2 A    j  m  um 
 3  i
a4   i  j m   vi 
  1   
a5  
2 A  i  j m   v j 
a    i  j  m  v m 
 6


Expanding the above equations
 a1 
 
u  1 x y  a2 
 
a3 
Substituting the values for a into the above equation gives:
 i  j  m   ui 
u  1 x y    i  j  m   u j 
1
2A  
  i  j  m  um 


We will now derive the u displacement function in terms of the
 i ui   j u j   mum 
1  
u  1 x y    i u i   j u j   m um 
2A
  i ui   j u j   mum 
Multiplying the matrices in the above equations gives:
u ( x, y ) 
1
2A

 i   i x   i y  u i    j   j x   j y  u j
  m   m x   m y um 


We will now derive the v displacement function in terms of the
 i v i   j v j   mv m 
1  
v   1 x y    i v i   j v j   mv m 
2A
  i v i   j v j   mv m 
v ( x, y ) 
1
2A

 i   i x   i y  v i    j   j x   j y  v j
  m   m x   m y v m 


The displacements can be written in a more convenience form
as:
u ( x, y )  N i u i  N j u j  N m u m
v ( x, y )  N i v i  N j v j  N m v m
where:
1
Ni   i   i x   i y 
2A
Nj 
1
2A
 j   j x   j y 
1
Nm   m   m x   m y 
2A


The elemental displacements can be summarized as:
u( x, y ) Ni ui  N j u j  Nmum 
 i     
v ( x, y ) Ni v i  N j v j  Nmv m 
In another form the above equations are:  ui 
v 
 i
N i 0 Nj 0 Nm 0  uj 
{ }    
0 Ni 0 Nj 0 Nm   v j 
um 
 
{ }  [N ]{d } v m 


In another form the equations are: { }  [N ]{d }
N 0 Nj 0 Nm 0 
N    0i Ni 0 Nj 0 Nm 

The linear triangular shape functions are illustrated below:

Ni Nj Nm
1 y y y
1
x m x m x m
i i i
j j j


So that u and v will yield a constant value for rigid-body
displacement, Ni + Nj + Nm = 1 for all x and y locations on the
element.

Ni Nj Nm
1 y y y
1
x m x m x m
i i i
j j j


Ni Nj Nm
1 y y y
1
x m x m x m
i i i
j j j


element.
For example, assume all the triangle displaces as a rigid body in
the x direction: u = u0
u0  u0  u0  N i  N j  N m 
0
 
Ni 0 N j 0 Nm 0  u0   Ni  N j  Nm  1
{ }     
 0 Ni 0 N j 0 Nm   0 
u0 
 
0


element.
the y direction: v = v0
0 v 0  v 0  Ni  N j  Nm 
v 
 0
N i 0 N j 0 N m 0   0   Ni  N j  Nm  1
{ }     
 0 Ni 0 N j 0 Nm  v 0 
0
 
v 0 


The requirement of completeness for the constant-strain triangle
element used in a two-dimensional plane stress element is
illustrated in figure below.
The element must be able to translate uniformly in either the x
or y direction in the plane and to rotate without straining as
shown


The reason that the element must be able to translate as a rigid
body and to rotate stress-free is illustrated in the example of a
cantilever beam modeled with plane stress elements.
By simple statics, the beam elements

beyond the loading are stress free.
Hence these elements must be free
to translate and rotate without
stretching or changing shape.


Elemental Strains: The strains over a two-dimensional

element are:
 u 
 
  x   x 
   v 
{ }    y    
   y 
 xy   u v 
  
 y x 


Substituting our approximation for the displacement gives:

u 
x
 u,x 
x
 Ni ui  N j u j  Nm um 
u,x  Ni ,x ui  N j ,x u j  Nm,x um
where the comma indicates differentiation with respect to that

variable.


The derivatives of the interpolation functions are:

1  
Ni ,x   i   i x   i y   i
2 A x 2A
j m
N j ,x  Nm, x 
2A 2A
Therefore:
u

1
x 2 A
  i ui   j u j   m um 


In a similar manner, the remaining strain terms are
approximated as:
v

1
y 2 A
  i v i   j v j   mv m 
u v
 
1
y x 2 A
  i u i   i v i   j u j   j v j   m u m   mv m 


We can write the strains in matrix form as:
 u   ui 
  v 
  x   x   i 0  j 0 m 0   i 
   v  1   u 
{ }    y       0 i 0  j 0 m   j 
   y  2 A     j
v
 m  m   j 
 xy   u v   i i j
um 
    
 y x  v m 
 di 
 
{ }  [B ]{d } { }  Bi Bj Bm   d j 
 
d m 


Stress-Strain Relationship: The in-plane stress-strain
relationship is:
 x   x 
    { }  [D ][B ]{d }
 y   [D ]   y 
   
 xy   xy 
For plane stress [D] is: For plane strain [D] is:
1  0  1    0 
E   [D ] 
E   1 
[D ]   1 0 0
1 2   1   1  2   0 
0 0 0.5 1      0 0.5   


The total potential energy is defined as the sum of the internal
strain energy U and the potential energy of the external forces
:  U   
p b p s
1 1
Where the strain energy is: U 
2V { }T { }dV   { }T [D]{ }dV
2V
The potential energy of the body force term is:
 b    { }T { X }dV
V
where {} is the general displacement function and {X} is the

body weight per unit volume.


:  U   
p b p s
1 1
2V { }T { }dV   { }T [D]{ }dV
2V
The potential energy of the concentrated forces is:
 p   {d }T {P }
where {P} are the concentrated forces and {d} are the nodal
displacements.


:  U   
p b p s
1 1
2V { }T { }dV   { }T [D]{ }dV
2V
The potential energy of the distributed loads is:
s    { }T {T }dS
S
where {} is the general displacement function and {T} are

the surface tractions.


Then the total potential energy expression becomes:
1
 p   d  [B ]T [D][B ]d  dV   d  [N ]T { X }dV
T T
2V V
 d  P   d  [N ]T {T }dS
T T
The nodal displacements {d} are independent of the general x-

y coordinates, therefore
1
 p  d   [B ]T [D][B ]dV d   d   [N ]T { X }dV
T T
2 V V
 d  P  d   [N ]
T T T
{T }dS
S


We can define the last three terms as:
f    [N ]T { X }dV  P   [N ]T {T }dS
V S
Therefore:
1
p  d   [B]T [D][B]dV d   d  f 
T T
2 V
Minimization of p with respect to each nodal displacement

requires that:
 p
  [B ]T [D][B ]dV d   f   0
 d  V


The above relationship requires:
 [B ]
T
[D][B ]dV d   f 
V
The stiffness matrix can be defined as:
[k ]   [B ]T [D][B ]dV
V
For an element of constant thickness, t, the above integral

becomes:
[k ]  t  [B ]T [D][B ] dx dy
A


The integrand in the above equation is not a function of x or y
(global coordinates); therefore, the integration reduces to:
[k ]  t [B ]T [D][B ] dx dy
A
[k ]  tA [B ]T [D][B ]
where A is the area of the triangular element.



Expanding the stiffness relationship gives:
 [k ii ] [k ij ] [kim ] 
 
[k ]   [k ji ] [k jj ] [k jm ] 
[k mi ] [kmj ] [kmm ]
where each [kii] is a 2 x 2 matrix define as:

[kii ]  [Bi ]T [D ][Bi ] tA [kij ]  [Bi ]T [D][B j ] tA
[kim ]  [Bi ]T [D][Bm ] tA


Recall:
 i 0  j 0
1 
Bi   2A  0  i  B j  
1 
0 j

2A 
  i  i    j  j 
 m 0
1 
Bm   2A  0  m 
  m  m 

Step 5 - Assemble the Element Equations to Obtain the

Global Equations and Introduce the Boundary
Conditions
The global stiffness matrix can be found by the direct stiffness
method. N
[K ]   [ k ( e ) ]
e 1
The global equivalent nodal load vector is obtained by lumping

body forces and distributed loads at the appropriate nodes as
well as including any concentrated loads.
N
{F }   {f ( e ) }
e 1


Conditions
The resulting global equations are: F   K d 
where {d} is the total structural displacement vector.
In the above formulation of the element stiffness matrix, the

matrix has been derived for a general orientation in global
coordinates.
Therefore, no transformation form local to global coordinates is
necessary.


Conditions
However, for completeness, we will now describe the method to
use if the local axes for the constant-strain triangular element
are not parallel to the global axes for the whole structure.


Conditions
To relate the local to global displacements, force, and stiffness
matrices we will use:
d   Td f   Tf k  T T k T


Conditions
The transformation matrix T for the triangular element is:
 C S 0 0 0 0
 S C 0 0 0 0 
  C  cos 
 0 0 C S 0 0
T   S  sin
 0 0 S C 0 0 
 0 0 0 0 C S
 
 0 0 0 0 S C 

Step 7 - Solve for Element Forces and Stress
Having solved for the nodal displacements, we can obtain

strains and stresses in x and y directions in the elements by
using:
{ }  [B ]{d } { }  [D ][B ]{d }

Plane Stress Example 1
Consider the structure shown in the figure below.
Let E = 30 x 106 psi,  = 0.25, and t = 1 in.
Assume the element nodal displacements have been

determined to be u1 = 0.0, v1 = 0.0025 in, u2 = 0.0012 in,
v2 = 0.0, u3 = 0.0, and v3 = 0.0025 in.

First, we calculate the element  ’s and  ’s as:
 i  y j  y m  0  1  1  i  xm  x j  0  2  2
 j  y m  y i  0  ( 1)  2  j  xi  xm  0  0  0
 m  y i  y j  1  0  1  m  x j  xi  2  0  2

Therefore, the [B] matrix is:
 i 0 j 0 m 0  1 0 2 0 1 0 
1    1  0 2 0 0 0 2 
B   2 A  0 i 0 j 0 m 
2(2)  
  i i j j m 
m  
 2 1 0 2 2  1
 i  y j  y m  0  1  1  i  xm  x j  0  2  2
 j  y m  y i  0  ( 1)  2  j  xi  xm  0  0  0
 m  y i  y j  1  0  1  m  x j  xi  2  0  2

For plane stress conditions, the [D] matrix is:
 1 0.25 0 
30  106 
[D ]  2 
0.25 1 0 
1  (0.25) 
 0 0 0.375 
Substitute the above expressions for [D] and [B] into the general
equations for the stiffness matrix:
[k ]  tA [B ]T [D ][B ]

[k ]  tA [B ]T [D][B ]
 1 0 2
 0 2 1
  1 0.25 0   1 0 2 0 1 0 
(2)30  106  2 0 1   1 
k   0.25 1 0 0 2 0 0 0 2 
4(0.9375)  2 0 2    2(2)  
 0 0 0.375   2 1 0 2 2 1
 1 0 2 
 
 0 2 1

Performing the matrix triple product gives:
 2.5 1.25 2 1.5 0.5 0.25 

1.25 4.375 1 0.75 0.25 3.625 
 
  2  1 4 0 2 1  lb
k  4  106  
 1.5 0.75 0 1.5 1.5 0.75  in
 0.5 0.25 2 1.5 2.5 1.25 
 
0.25 3.625 1 0.75 1.25 4.375 

The in-plane stress can be related to displacements by:

{ }  [D ][B ]{d }
0.0 
0.0025 in 
 x   1 0.25 0   1 0 2 0 1 0   
  30  10 
6
 1   0.0012 in 
 y   
0.25 1 0
 2(2) 
0 2 0 0 0 2 
 0.0 
  0.9375  0        
 xy   0 0.375   2 1 0 2 2 1 0.0 
 
0.0025 in 

The stresses are:  x   19,200 psi 

   
 y    4,800 psi 
   15,000 psi 
 xy   
Recall, the relationships for principal stresses and principal

angle in two-dimensions are:
x y  y 
2
1  2 xy 
1    x    xy   max
2
p  tan1  
2  2  2   x   y 
x y  y 
2
2    x    xy   min
2
2  2 

Therefore:
2
19,200  4,800  19,200  4,800 
1     15,000   28,639 psi
2
 
2  2 
2
19,200  4,800  19,200  4,800 
2     15,000   4,639 psi
2
 
2  2 
1  2( 15,000) 
p  tan1    32.3
o
2 19,200  4,800 

Treatment of Body and Surface Forces
The general force vector is defined as:
f    [N ]T { X }dV  P   [N ]T {T }dS

V S
Let’s consider the first term of the above equation.
fb    [N ]T { X }dV  Xb 
X   
V  Yb 
where Xb and Yb are the weight densities in the x and y
directions, respectively.
The force may reflect the effects of gravity, angular velocities, or

dynamic inertial forces.

For a given thickness, t, the body force term becomes:
fb    [N ]T { X }dV  t  [N ]T { X }dA

V A
 Ni 0 
 0 Ni 
 
 Xb 
N
[N ]T   j
0  X   
  Yb 
 0 Nj 
Nm 0 
 
 0 Nm 

The integration of the {fb} is simplified if the origin of the

coordinate system is chosen at the centroid of the element, as
shown in the figure below.
With the origin placed at the centroid, we can use the definition
of a centroid.
 y dA  0
A
 x dA  0
A

Recall the interpolation functions for a plane stress/strain

triangle:
Ni 
1
2A
 i   i x   i y  Nj 
1
2A
 j   j x   j y 
1
Nm   m   m x   m y 
2A
With the origin placed at the centroid, we can use the definition
of a centroid.
 y dA  0
A
 x dA  0
A

bh
Therefore the terms in the integrand are: A
2
  x dA  0
i   y dA  0
A
i
A
 i  x j y m  y j xm           0  
b 2h h bh
 2  3   3  3
 j  xm y i  y m xi   0           
h b 2h bh
 3   2  3  3
 m  xi y j  y i x j              
b h b h bh
 2  3   2  3  3
1 1 tA
i   j  m 
2A t  i dA  t  dA 
3 A
2A A
3 3

Therefore the terms in the integrand are:
fb    [N ]T { X }dV  t  [N ]T { X }dA

V A
The body force at node i is given as:

tA  X b 
fbi    
3  Yb   fbix   Xb 
f  Y 
The general body force vector is:  biy 
 b
 f  tA  X 
fb    fbjx    Y b 
 bjy  3  b 
fbmx   Xb 
   
fbmy   Yb 

The third term in the general force vector is defined as:
fs    [N ]T {T }dS
S
Let’s consider the example of a uniform stress p acting between

nodes 1 and 3 on the edge of element 1 as shown in figure
below.
p  p
T    px    
 y  0 

The third term in the general force vector is defined as:
fs    [N ]T {T }dS
S
 N1 0
0 N1 
 
N 0 p  p
[N ]T   2 
T    px    
  0 
0 N2  y
 N3 0
 
0 N3 
evaluated at x = a

Therefore, the traction force vector is:
fs    [N ]T {T }dS
S
 N1 0  N1p 
0 N1   0 
   
t L
N 0  p L
N2 p 
fs      2    dy dz  t    dy
0 0
0 N2   0  0
0 
 N3 0  N3 p 
   
0 N3  x = a  0 x=a

The interpolation function for i = 1 is:

1
Ni   i   i x   i y 
2A
For convenience, let’s choose the coordinate system shown in

the figure below.
 i  x j y m  y j xm
with i = 1, j = 2, and m = 3
1  x 2 y 3  y 2 x3
  0  0    0  a   0

The interpolation function for i = 1 is:

1
Ni   i   i x   i y 
2A
For convenience, let’s choose the coordinate system shown in

the figure below.
Similarly, we can find:

1  0 1  a
ay
N1 
2A

The remaining interpolation function, N2 and N3 are:

ay L(a  x ) Lx  ay
N1  N2  N3 
2A 2A 2A
Evaluating Ni along the 1-3 edge of the element (x = a) gives:
ay
N1 
2A
N2  0
a L  y 
N3 
2A

Substituting the interpolation function in the traction force vector
expression gives:
 N1p   y 
 0   0 
   
L
N p  atp
L
 0 
fs    [N ]T {T }dS  t   02  dy  2 A   0  dy
S 0  0 
 N3 p  L  y 
   
 0   0 

Therefore, the traction force vector is:
 fs1x  L2   1
f    0 
 s1y   0  
 f  atp  0  pLt 0 
fs    [N ]T {T }dS  fs 2 x       
S  s2y  4 A  0 2 0 
fs 3 x  L2   1
     
fs 3 y   0 0 
aL
A
2

The figure below shows the results of the surface load

equivalent nodal for both elements 1 and 2:
From Element 1
From Element 2

For the CS triangle, a distributed load on the element edge can

be treated as concentrated loads acting at the nodes
associated with the loaded edge.
However, for higher-order elements, like the linear strain triangle

(discussed in Chapter 8), load replacement should be made
by using the principle of minimum potential energy.
For higher-order elements, load replacement by potential

energy is not equivalent to the apparent statically equivalent
one.

Explicit Expression for the Constant-Strain Triangle
Stiffness Matrix
Usually the stiffness matrix is computed internally by computer
programs, but since we are not computers, we need to
explicitly evaluate the stiffness matrix.
For a constant-strain triangular element, considering the plane

strain case, recall that: [k ]  tA[B ]T [D ][B ]
where [D] for plane strain is:

1    0 
[D ] 
E   1 0 
1   1  2   0 
0.5   
 0

Stiffness Matrix
Substituting the appropriate definition into the above triple
product gives:
[k ]  tA[B ]T [D ][B ]
 i 0 i 
0 i  i 
 1    0   i 0 j 0 m 0
j 0 j   
0 0
tE
[k ]    1 i j m 
 j  
0 0
4 A(1   )(1  2 ) 0 j 
  0 0.5      i i j j m  m 
m  
0
 m 0
 
0 m m 

Stiffness Matrix
Substituting the appropriate definition into the above triple
product gives:
The stiffness matrix is a function of the global coordinates x and

y, the material properties, and the thickness and area of the
element.

Plane Stress Problem 2
Consider the thin plate subjected to the surface traction shown

in the figure below.
Assume plane stress conditions. Let E = 30 x 106 psi,  = 0.30,

and t = 1 in.
Determine the nodal displacements and the element stresses.

Discretization
Let’s discretize the plate into two elements as shown below:

20 in.
10 in.
This level of discretization will probably not yield practical results

for displacement and stresses; however, it is useful example
for a longhand solution.

Discretization
For element 2, The tensile traction forces can be converted into
nodal forces as follows:
fs 3 x   1  1 5,000 lb 
f  0  0   0 
 s3y       
 fs1x  pLt 0  1,000 psi (1 in )10 in 0   0 
fs 2         
 fs1y  2 0  2 0
   0 
fs 4 x   1  1 5,000 lb 
       
fs 4 y  0  0   0 

Discretization
For element 2, The tensile traction forces can be converted into
nodal forces as follows:
20 in. 5 kips
F   A  1,000 psi 10 in 1in 

10 in.
 10,000 lb
5 kips

The governing global matrix equations are: {F }  [K ]{d }
Expanding the above matrices gives:
 F1x   R1x   d1x   0 

F    d   0 
 1y   R1y   1y   
F2 x   R2 x  d 2 x   0 
       
F2 y   R2 y  d 2 y   0 
    [ K ]    [K ]  
F3 x  5,000 lb  d 3 x  d 3 x 
F3 y   0  d 3 y  d 3 y 
       
F4 x  5,000 lb  d 4 x  d 4 x 
F4 y   0     
  d 4 y  d 4 y 

Assemblage of the Stiffness Matrix
The global stiffness matrix is assembled by superposition of the
individual element stiffness matrices.
The element stiffness matrix is: [k ]  tA[B ]T [D ][B ]
 i 0 j 0 m 0
1  
B   2 A  0 i 0 j 0 m 
  i i j j m  m 
1    0 
[D ] 
E   1 0 
1   1  2   0 
0.5   
 0

For element 1: the coordinates are xi = 0, yi = 0, xj = 20, yj = 10,

xm = 0, and ym = 10. The area of the triangle is:
bh (20)(10)
A   100 in.2
2 2
 i  y j  y m  10  10  0  i  xm  x j  0  20  20
 j  y m  y1  10  0  10  j  xi  xm  0  0  0
 m  y i  y j  0  10  10  m  xi  x j  20  0  20

 0 0 10 0 10 0 
1 
[B ]  0 20 0 0 0 20  1
200   in
 20 0 0 10 20 10 
 i  y j  y m  10  10  0  i  xm  x j  0  20  20
 j  y m  y1  10  0  10  j  xi  xm  0  0  0
 m  y i  y j  0  10  10  m  xi  x j  20  0  20


1  0   1 0.3 0 
E   30  10
6
[D ]   1 0  0.3 1 0  psi
1  2   0.91  
0 0 0.5 1      0 0 0.35 
[k ]  tA [B ]T [D ][B ]

 0 0 20 
Therefore:  0 20 0 
   1 0.3 0 
6
 
0  lb 3
30(10 ) 10 0 0
[B ] [D ] 
T
  0.3 1
200(0.91)  0 0 10    in
 0
 0 0.35 
 10 0 20 
 
 0 20 10 
 0 0 7 
 6 20 0 
 
30(10 6
)  10 3 0  lb
[B ]T [D]    3
200(0.91)  0 0 3.5  in
 10 3 7 
 
 6 20 3.5 

[k ]  tA [B ]T [D][B ]
 0 0 7 
 6 20 0 
   0 0 10 0 10 0 
(0.15)(106 )  10 0  1 
20 
3
1(100)    0 20 0 0 0
0.91  0 0 3.5  200  
 10 3 7   20 0 0 10 20 10 
 
 6 20 3.5 


u1 v1 u3 v3 u2 v2
 140 0 0 70 140 70 
 0 400 60 0 60 400 
 
75,000  0 60 100 0 100 60  lb
[k (1) ]    in
0.91  70 0 0 35 70 35 
 140 60 100 70 240 130 
 
 70 400 60 35 130 435 

Rearranging the rows and columns gives:

u1 v1 u2 v2 u3 v3
 140 0 140 70 0 70 
 0 400 60 400 60 0 
 
75,000  140 60 240 130 100 70  lb
[k (1) ]    in
0.91  70 400 130 435 60 35 
 0 60 100 60 100 0
 
 70 0 70 35 0 35 

For element 2: the coordinates are xi = 0, yi = 0, xj = 20, yj = 0,

xm = 20, and ym = 10. The area of the triangle is:
bh (20)(10)
A   100 in.2
2 2
 i  y j  y m  0  10  10  i  xm  x j  20  20  0
 j  y m  y1  10  0  10  j  xi  xm  0  20  20
m  y i  y j  0  0  0  m  xi  x j  20  0  20

 10 0 10 0 0 0 
1  1
[B ]  0 0 0 20 0 20
200   in
 0 10 20 10 20 0 
 i  y j  y m  0  10  10  i  xm  x j  20  20  0
 j  y m  y1  10  0  10  j  xi  xm  0  20  20
m  y i  y j  0  0  0  m  xi  x j  20  0  20


1  0   1 0.3 0 
E   30  10
6
[D ]   1 0  0.3 1 0  psi
1  2   0.91  
0 0 0.5 1      0 0 0.35 
[k ]  tA [B ]T [D][B ]

 10 0 0 
Therefore:  0 0 10 
   1 0.3 0 
30(106 )  10 20  
0 
0
[B ] [D ] 
T
  0.3 1
200(0.91)  0 20 10  
 0 0 0.35 
 0 0 20 
 
 0 20 0 
 10 3 0 
 0 0 3.5 
 
30(10 6
)  10 3 7 
[B ]T [D]   
200(0.91)  0 20 3.5 
 6 0 7 
 
 6 20 0 

[k ]  tA [B ]T [D][B ]
 10 3 0 
 0 0 3.5 
   10 0 10 0 0 0 
(0.15)(106 )  10 3 7  1  
1(100)    0 0 0 20 0 20
0.91  0 20 3.5  200  
 0 10 20 10 20 0 
 6 0 7 
 
 6 20 0 


u1 v1 u4 v4 u3 v3
 100 0 100 60 0 60 
 0 35 70 35 70 0 
 
75,000  100 70 240 130 140 60 
[k (2) ]   
0.91  60 35 130 435 70 400 
 0 70 140 70 140 0 
 
 60 0 60 400 0 400 

Rearranging the rows and columns gives:

u1 v1 u3 v3 u4 v4
 100 0 0 60 100 60 
 0 35 70 0 70 35 
 
75,000  0 70 140 0 140 70 
[k (2) ]   
0.91  60 0 0 400 60 400 
 100 70 140 60 240 130 
 
 60 35 70 400 130 435 

In expanded form, element 1 is:
u1 v1 u2 v2 u3 v3 u4 v4
 28 0 28 14 0 14 0 0 
 0 80 12 80 12 0 0 0 
 
 28 12 48 26 20 14 0 0 
 
375,000  14 80 26 87 12 7 0 0

[k (1) ] 
0.91  0 12 20 12 20 0 0 0 
 
 14 0 14 7 0 7 0 0 
 0 0 0 0 0 0 0 0 
 
 0 0 0 0 0 0 0 0 

In expanded form, element 2 is:
u1 v1 u2 v2 u3 v3 u4 v4
 20 0 0 0 0 12 20 12 
 0 7 0 0 14 0 14 7 
 
 0 0 0 0 0 0 0 0 
 0 0 0 0 0 0 0
 0
375,000  
[k (2) ] 
0.91  0 14 0 0 28 0 28 14 
 
 12 0 0 0 0 80 12 80 
 20 14 0 0 28 12 48 26 
 
 12 7 0 0 14 80 26 87 

Using the superposition, the total global stiffness matrix is:
u1 v1 u2 v2 u3 v3 u4 v4
 48 0 28 14 0 26 20 12 
 0 87 12 80 26 0 14 7 
 
 28 12 48 26 20 14 0 0 
 80 26 87 7

375,000  14 12 0 0

[k ] 
0.91  0 26 20 12 48 0 28 14 
 
 26 0 14 7 0 87 12 80 
 20 14 0 0 28 12 48 26 
 
 12 7 0 0 14 80 26 87 

Applying the boundary conditions: d1x  d1y  d 2 x  d 2 y  0
 R1x   48 0 28 14 0 26 20 12   d01x 

 R   0  
87 12 80 26 0 14 7   d01y 
 1y   
 R2 x   28 12 48 26 20 14 0 0   d02 x 
    
 R2 y  375,000  14 80 26 87 12 7 0 0  d02 y 
 
  
5,000 lb  0.91  0 26 20 12 48 0 28 14  d 3 x 
 0   
 26 0 14 7 0 87 12 80  d 3 y 
   
 500 lb   20 14 0 0 28 12 48 26  d 4 xx 
 0   
 12 7 0 0 14 80 26 87   
d 4 yy 

The governing equations are:
5,000 lb   48 0 28 14  d3 x 
 0   0  
  375,000  87 12 80  d 3 y 
    
5,000 lb  0.91  28 12 48 26  d 4 x 
 0   
 14 80 26 87  d 4 y 
Solving the equations gives:
d 3 x  609.6 
d   
 3y   4.2
   10
d
 6
 
663.7
 in
 4x   
d 4 y   104.1

The exact solution for the displacement at the free end of the
one-dimensional bar subjected to a tensile force is:
PL (10,000)20
   670  10 6 in
AE 10(30  10 )
6
The two-element FEM solution is:
d 3 x  609.6 
d   
 3y   4.2
   10 6
   in
d 4 x  663.7 
d 4 y   104.1

The in-plane stress can be related to displacements by:
{ }  [D ][B ]{d }
 d ix 
d 
0 
iy 
1  0  i 0 j 0 m
E     d jx 
{ }  2 
 1 0  0 i 0 j 0 m   
2 A(1   ) d jy
0 0 0.5 1       i
 i j j m  m   
d mx 
 
d my 

Element 1: { }  [D][B ]{d }

 0.0 
 0.0 
 x   1 0.3 0  0 0 10 0 10 0  
  6
30(10 )(10 )  609.6 
0  0 
6
 y   
0.3 1

20 0 0 0 20 
 
  0.96(200)  4.2 
 xy   0 0   20
0.35  0 0 10 20 10 

0.0 
 
 0.0 
 x  1,005 psi 
   
 y    301 psi 
   2.4 psi 
 xy   

Element 2: { }  [D][B ]{d }

 0.0 
 0.0 
 x   1 0.3 0   10 0 10 0 0 0  
  30(10 )(10 ) 
6
663.7 
0  0 
6
 y   
0.3 1

0 0 20 0 20 
 104.1
  0.96(200)
 0.35   
 xy  0 0   0 10 20 10 20 0 

609.6 
 
 4.2 
 x   995 psi 
   
 y    1.2 psi 
   2.4 psi 
 xy   

The principal stresses and principal angle in element 1 are:
2
1005  301  1005  301 
1      (2.4)  1005 psi
2
2  2 
2
1005  301  1005  301 
2      (2.4)  301 psi
2
2  2 
1  2(2.4) 
 p  tan 1    0o
2 1005  301

The principal stresses and principal angle in element 2 are:
2
995  1.2  995  1.2 
1      ( 2.4)2  995 psi
2  2 
2
995  1.2  995  1.2 
2      ( 2.4)  1.1 psi
2
2  2 
1  2( 2.4) 
 p  tan 1  0
o
2  995  1.2 

Problems
12. Do problems 6.6a, 6.6c, 6.7, 6.10a-c, 6.11, and 6.13 on
13. Rework the plane stress problem given on page 356 in your
D. Logan using Matlab code FEM_2Dor3D_linelast_standard to
do analysis.
Start with the simple two element model. Continuously

refine your discretization by a factor of two each time until
your FEM solution is in agreement with the exact solution.
How many elements did you need?


FEM_2Dor3D_linelast_standard
function FEM_2Dor3D_linelast_standard
%
% Example 2D and 3D Linear elastic FEM code
% Currently coded to run either plane strain or plane stress (2DOF) or general 3D but
% could easily be modified for axisymmetry too.
%
% Variables read from input file;
% nprops No. material parameters
% materialprops(i) List of material parameters
% ncoord No. spatial coords (2 for 2D, 3 for 3D)
% ndof No. degrees of freedom per node (2 for 2D, 3 for 3D)
% (here ndof=ncoord, but the program allows them to be different
% to allow extension to plate & beam elements with C^1 continuity)
% nnode No. nodes
% coords(i,j) ith coord of jth node, for i=1..ncoord; j=1..nnode
% nelem No. elements
% maxnodes Max no. nodes on any one element (used for array dimensioning)
% nelnodes(i) No. nodes on the ith element
% elident(i) An integer identifier for the ith element. Not used
% in this code but could be used to switch on reduced integration,
% etc.
% connect(i,j) List of nodes on the jth element
% nfix Total no. prescribed displacements
% fixnodes(i,j) List of prescribed displacements at nodes
% fixnodes(1,j) Node number
% fixnodes(2,j) Displacement component number (1, 2 or 3)
% fixnodes(3,j) Value of the displacement
% ndload Total no. element faces subjected to tractions
% dloads(i,j) List of element tractions
% dloads(1,j) Element number
% dloads(2,j) face number
% dloads(3,j), dloads(4,j), dloads(5,j) Components of traction
% (assumed uniform)
%

% To run the program you first need to set up an input file, as described in
% the lecture notes. Then change the fopen command below to point to the file.
% Also change the fopen command in the post-processing step (near the bottom of the
% program) to point to a suitable output file. Then execute the file in
% the usual fashion (e.g. hit the green arrow at the top of the MATLAB
% editor window)
%
%
% ==================== Read data from the input file ===========================
%
%
% YOU NEED TO CHANGE THE PATH & FILE NAME TO POINT TO YOUR INPUT FILE
%
infile=fopen ('Logan_p364_2_element.txt','r');
outfile=fopen('Logan_p364_2_element_results.txt','w');

Logan_p364_2_element.txt
E
G
2 1   
No._material_props: 3
Shear_modulus: 11.5385
Poissons_ratio: 0.3
Plane_strain/stress: 0
No._coords_per_node: 2
No._DOF_per_node: 2
No._nodes: 4
Nodal_coords:
0.0 0.0
0.0 10.0
20.0 10.0
20.0 0.0
No._elements: 2
Max_no._nodes_on_any_one_element: 3
element_identifier; no._nodes_on_element; connectivity:
20 in. 1 3 1 3 2
2 3 1 4 3
No._nodes_with_prescribed_DOFs: 4
Node_#, DOF#, Value:
1 1 0.0
1 2 0.0
10 in. 2 1 0.0
2 2 0.0
No._elements_with_prescribed_loads: 1
Element_#, Face_#, Traction_components
2 2 1.0 0.0

12
10
-2
0 2 4 6 8 10 12 14 16 18 20
Nodal Displacements:
Node Coords u1 u2
1 0.0000 0.0000 -0.0000 -0.0000
2 0.0000 10.0000 -0.0000 0.0000
3 20.0000 10.0000 0.6096 0.0042
4 20.0000 0.0000 0.6637 0.1041
Strains and Stresses

Element; 1
int pt Coords e_11 e_22 e_12 s_11 s_22 s_12
1 6.6667 6.6667 0.0305 0.0000 0.0001 1.0048 0.3014 0.0024
Element; 2
1 13.3333 3.3333 0.0332 -0.0100 -0.0001 0.9952 -0.0012 -0.0024

Node Coords u1 u2
1 0.0000 0.0000 -0.0000 -0.0000
2 0.0000 10.0000 -0.0000 0.0000
3 20.0000 10.0000 0.6096 0.0042
4 20.0000 0.0000 0.6637 0.1041

Element; 1
1 6.6667 6.6667 0.0305 0.0000 0.0001 1.0048 0.3014 0.0024
Element; 2
1 13.3333 3.3333 0.0332 -0.0100 -0.0001 0.9952 -0.0012 -0.0024

Node Coords u1 u2
1 0.0000 0.0000 -0.0000 -0.0000
2 0.0000 10.0000 -0.0000 0.0000
3 20.0000 10.0000 0.6096 0.0042
4 20.0000 0.0000 0.6637 0.1041

Element; 1
1 6.6667 6.6667 0.0305 0.0000 0.0001 1.0048 0.3014 0.0024
Element; 2
1 13.3333 3.3333 0.0332 -0.0100 -0.0001 0.9952 -0.0012 -0.0024

FEM_2Dor3D_linelast_standard – 8 elements
Logan_p364_8_element.txt
No._material_props: 3
Shear_modulus: 11.5385
Poissons_ratio: 0.3
Plane_strain/stress: 1
No._coords_per_node: 2
No._DOF_per_node: 2
No._nodes: 9
Nodal_coords:
0.0 0.0
10.0 0.0
20.0 0.0
0.0 5.0
10.0 5.0
20.0 5.0
0.0 10.0
10.0 10.0
20.0 10.0
12
7 8 9 No._elements: 8
Max_no._nodes_on_any_one_element: 3
10 element_identifier; no._nodes_on_element; connectivity:
1 3 1 2 4
2 3 2 5 4
8 3 3 2 3 5
4 3 3 6 5
6
4 5 6 5 3 4 5 7
6 3 5 8 7
7 3 5 6 8
4 8 3 6 9 8
No._nodes_with_prescribed_DOFs: 6
Node_#, DOF#, Value:
2
1 1 0.0
1 2 0.0
0 4 1 0.0
1 2 3 4 2 0.0
7 1 0.0
-2 7 2 0.0
No._elements_with_prescribed_loads: 2
0 2 4 6 8 10 12 14 16 18 20 Element_#, Face_#, Traction_components
4 1 1.0 0.0
8 1 1.0 0.0

12
10
-2
0 2 4 6 8 10 12 14 16 18 20
Node Coords u1 u2
1 0.0000 0.0000 0.0000 -0.0000
2 10.0000 0.0000 0.2489 0.0317
3 20.0000 0.0000 0.5461 -0.0231
4 0.0000 5.0000 -0.0000 0.0000
5 10.0000 5.0000 0.2706 -0.0276
6 20.0000 5.0000 0.5758 -0.0893
7 0.0000 10.0000 0.0000 -0.0000
8 10.0000 10.0000 0.3068 -0.0923
9 20.0000 10.0000 0.6082 -0.1541

12
10
-2
0 2 4 6 8 10 12 14 16 18 20

Element; 1
1 3.3333 1.6667 0.0249 0.0000 0.0016 1.0051 0.4308 0.0366
Element; 2
1 6.6667 3.3333 0.0271 -0.0119 0.0008 0.8875 -0.0106 0.0183
Element; 3
1 13.3333 1.6667 0.0297 -0.0119 -0.0006 0.9949 0.0355 -0.0132
Element; 4
1 16.6667 3.3333 0.0305 -0.0132 -0.0001 1.0035 -0.0066 -0.0026

12
10
-2
0 2 4 6 8 10 12 14 16 18 20

Element; 5
1 3.3333 6.6667 0.0271 -0.0000 -0.0014 1.0927 0.4683 -0.0318
Element; 6
1 6.6667 8.3333 0.0307 -0.0129 -0.0010 1.0147 0.0079 -0.0230
Element; 7
1 13.3333 6.6667 0.0305 -0.0129 0.0005 1.0086 0.0053 0.0122
Element; 8
1 16.6667 8.3333 0.0301 -0.0130 0.0001 0.9931 -0.0017 0.0035




End of Chapter 6a
Chapter 6b – Plane Stress/Strain Equations
Learning Objectives
• To derive the bilinear four-noded rectangular (Q4)
element stiffness matrix.
• To compare the CST and Q4 model results for a

beam bending problem and describe some of the
CST and Q4 elements

We will now develop the four-noded rectangular plane
This element is an isoparametric formulation of a general

quadrilateral element (see Chapter 10).
This element is also called the bilinear rectangle because of

the linear terms in x and y for the x and y displacement
functions.

The "Q4" symbol represents the element as a quadrilateral
with four corner nodes.
Two advantages of the rectangular element over the triangular

element are:
1. ease of data input and
2. simpler interpretation of output stresses.

The "Q4" symbol represents the element as a quadrilateral
with four corner nodes.
A disadvantage of the rectangular element is that the simple

linear-displacement rectangle with its associated straight
sides poorly approximates the real boundary condition
edges.

The usual steps outlined in Chapter 1 will be followed to
obtain the element stiffness matrix and related equations.
1. Discretize and Select Element Type
2. Select a Displacement Function - Assume a variation of the
displacements over each element.
3. Define the Strain/Displacement and Stress/Strain Relationships -
use elementary concepts of equilibrium and compatibility.

4. Derive the Element Stiffness Matrix and Equations - Define the
stiffness matrix for an element and then consider the derivation of the
stiffness matrix for a linear-elastic spring element.
5. Assemble the Element Equations to Obtain the Global or Total

Equations and Introduce Boundary Conditions

6. Solve for the Unknown Degrees of Freedom (or Generalized
Displacements) - Solve for the nodal displacements.
7. Solve for the Element Strains and Stresses - The reactions and
internal forces association with the bar element.
8. Interpret the Results

1. Discretize and Select Element Type
Consider the rectangular element shown below (all interior
angles are 90°) with corner nodes 1-4 (again labeled
counterclockwise) and base and height dimensions of 2b
and 2h, respectively
 u1 
v 
 1
u2 
 
   2 
v
d 
u3 
v 3 
 
u4 
v 4 

2. Select a Displacement Function
For a compatible displacement field, the element dis-
placement functions u and v must be linear along each
edge because only two points (the corner nodes) exist
along each edge.
u( x, y )
 i    
v ( x, y )
 a  a x  a3 y  a4 xy 
 1 2 
a5  a6 x  a7 y  a8 xy 

There are a total of eight generalized degrees of freedom
(a's) and a total of eight specific degrees of freedom (u1, v1 at
node 1 through u4, v4 at node 4) for the element.
u( x, y )
 i    
v ( x, y )
 a  a x  a3 y  a4 xy 
 1 2 
a5  a6 x  a7 y  a8 xy 

We can proceed in the usual manner to solve for the a's and
obtain:
1
u ( x, y )   b  x  h  y  u1   b  x  h  y  u2
4bh 
  b  x  h  y  u3   b  x  h  y  u4 
1
v ( x, y )   b  x  h  y  v1   b  x  h  y  v 2
4bh 
  b  x  h  y  v 3   b  x  h  y  v 4 

These displacement expressions, can be expressed
equivalently in terms of the interpolation functions and
unknown nodal displacements as:
u( x, y )  N1u1  N2u2  N3u3  N4u4

v ( x, y )  N1v1  N2v 2  N3v 3  N4v 4
where:
N1 
 b  x  h  y  N2 
 b  x  h  y 
4bh 4bh
N3 
 b  x  h  y  N4 
 b  x  h  y 
4bh 4bh

The shape functions are visually deceiving. There is no
curvature in directions parallel to any side; however, there is a
twist due to the xy term in the element representation.
4 3 4 3
N1 N2
1 2 1 2
4 3 N4 4 3
N3
1 2 1 2

The shape functions are visually deceiving. There is no
curvature in directions parallel to any side; however, there is a
twist due to the xy term in the element representation.

In expanded form, the equations become:
 u1 
v 
 1
u2 
u  x, y   N1 0 N2  
0 N3 0 N4 0  v 2 
   
v  x, y    0 N1 0 N2 0 N3 0 N4  u3 
v 3 
 
u4 
{ }  [N ]{d } v 4 


displacement, N1 + N2 + N3 + N4 = 1 for all x and y locations
on the element.
the y direction: v = v0 0
v 
 0
0
 
u  x, y   N1 0 N2 0 N3 0 N4 0  v 0 
   
v  x, y    0 N1 0 N2 0 N3 0 N4   0 
v 0 
 
v ( x, y )  v 0  v 0  N1  N2  N3  N4  0
v 0 
 N1  N2  N3  N4  1

3. Define the Strain-Displacement andStress-Strain
Relationships
The general definitions of normal and shear strains are:
u v u v
x  y   xy  
x y y x
The strains over a two-dimensional element are:
 u 
 
  x   x 
   v 
{ }    y    
   y 
 xy   u v 
  
 y x 

Relationships
u 
 u,x   N1u1  N2u2  N3u3  N4u4 
x x
u,x  N1,x u1  N2,x u2  N3,x u3  N4,x u4

variable.

Relationships
The derivatives of the interpolation functions with respect to x
are:
1  h  y 
N1,x   b  x  h  y   
4bh x 4bh
1  h  y 
N2,x   b  x  h  y  
4bh x 4bh
1  h  y 
N3,x   b  x  h  y  
4bh x 4bh
1  h  y 
N4,x   b  x  h  y   
4bh x 4bh

Relationships
The derivatives of the interpolation functions with respect to y
are:
1  b  x 
N1,y   b  x  h  y   
4bh y 4bh
1  b  x 
N2,y   b  x  h  y   
4bh y 4bh
1  b  x 
N3,y   b  x  h  y  
4bh y 4bh
1  b  x 
N4,y   b  x  h  y  
4bh y 4bh

Relationships
We can write the strains in matrix form as: { }  [B ]{d }
  h  y  0 h  y  0
1 
B   4bh  0  b  x  0  b  x 
   b  x    h  y    b  x   h  y 
h  y  0  h  y  0 
0 b  x  0  b  x  
b  x  h  y  b  x    h  y  

Relationships
From equations, we observe that x is a function of y, y is a
function of x, and xy is a function of both x and y.
The stresses are again given as:
 x   x 
    { }  [D ][B ]{d }
 y   [D ]   y 
   
 xy   xy 
For plane stress [D] is: For plane strain [D] is:
1  0  1    0 
E   [D ] 
E   1 
[D ]   1 0 0
1 2   1   1  2   0 
0 0 0.5 1      0 0.5   

4. Derive the Element Stiffness Matrix and Equations
The above relationship requires:
h b
  [B ]
T
[D ][B ] t dx dy d   f 
h b
h b
[k ]    [B ]
T
[D][B ] t dx dy
h b
For an element of constant thickness, t, the above integral

becomes:
h b
[k ]  t   [B ]
T
[D][B ] dx dy
h b

Because the [B] matrix is a function of x and y, integration must
be performed.
The [k] matrix for the rectangular element is now of order 8 x 8.
A numerical evaluation for [k] using b = 4 in., h = 2 in., t = 1 in.,

E = 30 x 106 psi, and  = 0.3.
This double integral was solved using Mathcad.


1.35E+10 5.49E+09 -1.69E+09 -4.22E+08 -6.75E+09 -5.49E+09 -5.06E+09 4.22E+08
5.49E+09 2.45E+10 4.22E+08 9.28E+09 -5.49E+09 -1.22E+10 -4.22E+08 -2.15E+10
-1.69E+09 4.22E+08 1.35E+10 -5.49E+09 -5.06E+09 -4.22E+08 -6.75E+09 5.49E+09
-4.22E+08 9.28E+09 -5.49E+09 2.45E+10 4.22E+08 -2.15E+10 5.49E+09 -1.22E+10

[k] =
-6.75E+09 -5.49E+09 -5.06E+09 4.22E+08 1.35E+10 5.49E+09 -1.69E+09 -4.22E+08
-5.49E+09 -1.22E+10 -4.22E+08 -2.15E+10 5.49E+09 2.45E+10 4.22E+08 9.28E+09
-5.06E+09 -4.22E+08 -6.75E+09 5.49E+09 -1.69E+09 4.22E+08 1.35E+10 -5.49E+05
4.22E+08 -2.15E+10 5.49E+09 -1.22E+10 -4.22E+08 9.28E+09 -5.49E+09 2.45E+10

Steps 5 - 7
Steps 5 through 7, which involve assembling the global stiffness
matrix and equations, determining the unknown nodal
displacements, and calculating the stress, are identical to
those in Section 6.2 for the CST.
However, the stresses within each element now vary in both the
x and y directions.

Numerical Comparison of CST to Q4 Element Models and
Element Defects.
Table 6-1 compares the free end deflection and maximum
principal stress for a cantilevered beam modeled with various all
triangular CST elements or all rectangular Q4 elements.

Element Defects.
Table 6-1 compares the free end deflection and maximum
principal stress for a cantilevered beam modeled with various all
triangular CST elements or all rectangular Q4 elements.

Element Defects.
We observe from the displacement results that the CST element
models produce stiffer models than the actual beam behavior, as the
deflections are predicted to be smaller than classical beam theory predicts.
We also observe that the CST model converges very slowly to the classical
beam theory solution.
This is partly due to the element predicting only constant stress within each
element when for a bending problem; the stress actually varies linearly
through the depth of the beam.
This problem is rectified by using the linear-strain triangle (LST) element as

described in Chapter 8.

Element Defects.
The results indicate that the Q4 element model predicts more accurate
deflection behavior than the CST element model.
The two-row model of Q4 elements yields deflections very close to that

predicted by the classical beam deflection equation, whereas the two-row
model of CST elements is quite inaccurate in predicting the deflection.
As the number of rows is increased to four and then eight, the deflections are
predicted increasingly more accurately for the CST and Q4 element models.
The two-noded beam element model gives the identical deflection as the
classical equation ( = PL3/3EI) as expected (see discussion in Section 4.5)
and is the most appropriate model for this problem when you are not
concerned, for instance, with stress concentrations.

Element Defects.
It has been shown for a beam subjected to pure bending, the CST has a
spurious or false shear stress and hence a spurious shear strain in parts of
the model that should not have any shear stress or shear strain.
This spurious shear strain absorbs energy; therefore, some of the energy that
should go into bending is lost.
The CST is then too stiff in bending, and the resulting deformation is smaller
than actually should be.
This phenomenon of excessive stiffness developing in one more modes of

deformation is sometimes described as shear locking or parasitic shear.

Element Defects.
It should be noted that using a single row of Q4 elements with their linear
edge displacement is not recommended to accurately predict the stress
gradient through the depth of the beam.

Element Defects.
As mentioned previously, the CST element has constant strain and stress
within it, while the Q4 element normal strain x and hence the normal stress
x is linear in the y direction.
Therefore, the CST is not able simulate the bending behavior nearly as well
as the Q4 element.
The classical beam theory/bending stress equation predicts a linear stress

variation through the depth the beam given by x = -My/I
As shown when comparing the principal stresses for each model, as more
rows are used, the stresses approach the classical bending stress of 20 MPa
with the Q4 approaching the classical solution much faster as indicated by
comparing the two-row solutions for Q4 and CST models.

Element Defects.
This brief description of some of the limitations in using the CST and Q4
elements does not prevent us from using them to model plane stress and
plane strain problems.
It just requires us to use a fine mesh as opposed to a coarse one, particularly

where bending occurs and where in general large stress gradients will
results.
Also, we must make sure our computer program can handle Poisson's ratios
that approach 0.5 (if that is desired, such as in rubber-like materials).
For common materials, such as metals, Poisson's ratio is around 0.3, so

locking should not be of concern.
End of Chapter 6b
CIVL 7/8117 Practical Considerations in Modeling 1/43
Chapter 7 – Practical Considerations in Modeling
Learning Objectives
•To present concepts that should be considered when
modeling for a situation by the finite element method,
such as aspect ratio, symmetry, natural subdivisions,
sizing of elements and the h, p, and r methods of
refinement, concentrated loads and infinite stress,
infinite medium, and connecting different kinds of
elements
•To describe some of the approximations inherent in

finite element solutions
•To illustrate convergence of solution and introduce

the patch test for convergence of solution
Chapter 7 – Practical Considerations in Modeling
Learning Objectives
•To discuss the interpretation of stresses in an
element, including a common method of averaging
the nodal values (also called smoothing)
•To present a flowchart of typical finite element

processes used for the analysis of plane stress and
plane strain
•To describe a computer assisted step-by-step

solution of a bicycle wrench
•To demonstrate various real-world applications

where plane stress/strain element models are
applicable
Practical Considerations in Modeling

Introduction
In this section we will discuss some modeling considerations

and guidelines, including mesh size, natural subdivisions,
and the use of symmetry and associated boundary
conditions.
We will also introduce the concept of static condensation,

which enables us to apply the basis of the CST stiffness
matrix to a quadrilateral element.

Finite Element Modeling
Finite element modeling is partly an art guided by visualizing
physical interactions taking place within a body.
In modeling the user is confronted with the difficult task of

understanding physical behavior taking place and
understanding the physical behavior of various elements
available for use.
Matching the appropriate finite element to the physical

behavior being modeling is one of many decisions that must
be made by the modeler.
Understanding the boundary conditions can be one of the

most difficult tasks a modeler must face in construction a
useable finite element model.

Aspect Ratio and Element Shape
The aspect ratio is define as the ratio of the longest
dimension to the shortest dimension of a quadrilateral
element.
In general, as the aspect ratio increases, the inaccuracy of the

finite element solution increases.
Good Bad

Consider the five different finite element model shown in the
figure below.

A plot of the resulting error in the displacement at point A of
the beam verse aspect ratio is given.

In addition, the numerical answers are given in the following
table.
Case Aspect Number of Number of Point A Point B % Error

Ratio Nodes Elements
1 1.1 84 60 -1.093 -0.346 5.2
2 1.5 85 64 -1.078 -0.339 6.4
3 3.6 77 60 -1.014 -0.238 11.9
4 6.0 81 64 -0.886 -0.280 23.0
5 24.0 85 64 -0.500 -0.158 56.0
Exact Solution -1.152 -0.360

In general, elements that yield the best results are compact
and regular in shape will: (1) aspect ratios near one; and (2)
corner angles of quadrilaterals near 90°.

Minimum Support Conditions to
Suppress Rigid Body Motions in 2D
stable stable unstable


Use of Symmetry
The use of symmetry will often expedite the modeling of a
problem.
Symmetry allows us to consider a reduced problem instead of

the actual problem.
This will allow us to use a finer discretization of element with

less computational cost.

Use of Symmetry
problem.

Use of Symmetry
problem.

Use of Symmetry
problem.

Use of Symmetry
problem.

Use of Symmetry
problem.

Natural Subdivisions at Discontinuities
There are a variety of natural subdivisions for finite element
discretizations.
For example, natural locations of nodes occur at concentrated

loads or discontinuities in loading, other types of boundary
conditions, and abrupt changes in geometry of materials.

Sizing of Elements and Mesh Refinement
A discretization depends on the geometry of the structure, the
loading, and the boundary conditions.
For example, areas of high, rapidity changing stresses require

a finer mesh than regions where the stress is constant.

Where Finer Meshes Should Be Used

Use of Symmetry
Here the use of symmetry is applied to a soil mass subjected
to a foundation loading (66 nodes and 50 elements).

Use of Symmetry
Note that at the place of symmetry the displacements in the
direction perpendicular to the plane must be zero.

Use of Symmetry
This is modeled by rollers at nodes 2 - 6.

Use of Symmetry
The figure below illustrates the use of triangular elements for
transitions from smaller quadrilaterals to larger
quadrilaterals.

Use of Symmetry
The transitions are required since CST elements do not have
immediate nodes along their edges.

Use of Symmetry
If an element had an intermediate node, the resulting
equations would be inconsistent with the energy formulation
for the CST equations.

Use of Symmetry


Elements Must Not Cross Interfaces


Mesh Revision (Refinement)
Mesh revision or refinement is used to increase the accuracy
of the results produced by a finite element model.
First, start with a basic mesh using the fewest, reasonable
number of elements, and obtain a benchmark result.
Then refine the model by doing actions like increasing the
mesh density and compare the results.
This process continues until the results converge to some
user-defined criteria.
Consider this original mesh for the discussion of types of
methods:

Mesh Revision (Refinement)
Consider this original mesh for the discussion of types of
methods:

h Method of Refinement
In the h method, we use the particular element based on the
shape functions for that element.
Elements of the same kind are then added or made smaller.
Uniform or non-uniform refinements are possible as shown
below:

p Method of Refinement
In the p method, the polynomial p is increased to higher-order
polynomials based on the degree of accuracy specified.
These iterations are repeated until a convergence is reached.

r Method of Refinement
In the r method, the nodes are rearranged or relocated without
changing the number of elements of the polynomial degree
of the field quantities.

Infinite Medium
A typical example of infinite medium is a soil foundation
problem.
The guideline for the finite element model is that enough

material must be included such that the displacements at
nodes and stresses within the elements become negligibly
small at locations far from the foundation load.
The level of discretization can be determined by a trail-and-

error procedure in which the horizontal and vertical
distances from the load are varied and the resulting effects
on the displacements and stresses are observed.

Infinite Medium
For a homogeneous soil mass, experience has shown the
influence of a footing becomes insignificant if the horizontal
distance of the model is taken as approximately four and six
times the width of the footing and the vertical distance is
taken as approximately four to ten times the width of the
footing.

Infinite Medium
footing.

Infinite Medium
footing.

Checking the Model
The discretized finite element model should be checked
carefully before results are computed.
Ideally, a model should be checked by an analyst not involved

in the preparation of the model, who is then more likely to be
objective.
Preprocessors with their detailed graphical display capabilities

now make it comparatively easy to find errors, particularly
with a misplaced node or missing element or a misplaced
load or boundary condition.
Preprocessors include the ability to color, shrink, rotate, and

section a model mesh.

Checking the Results and Typical Postprocessor Results
An analyst should probability spend as much time processing,
checking, and analyzing results as spent in data preparation.

An analyst should probability spend as much time processing,
checking, and analyzing results as spent in data preparation.

The wrench in this example is modeled by 307 constraint
strain triangular elements (plane stress assumption). Below
is a plot of the deformed shape of the wrench over the
original mesh.


Equilibrium and Compatibility of Finite Element Results
An approximate solution for a stress analysis problem using
the finite element method based on assumed displacement
fields does not generally satisfy all the requirements for
equilibrium and compatibility that an exact theory-of-
elasticity solution satisfies.
However, remember that relatively few exact solutions exist.
Hence, the finite element method is a very practical one for

obtaining reasonable, but approximate, numerical solutions.

We now describe some of the approximations generally
inherent with finite element solutions.
1. Equilibrium of nodal forces and moments is satisfied.
This is true because the global equation F = Kd is a nodal

equilibrium equation whose solution for d is such that the
sums of all forces and moments applied to each node are
zero.
Equilibrium of the whole structure is also satisfied because

the structure reactions are included in the global forces,
and hence, in the nodal equilibrium equations.

2. Equilibrium within an element is not always satisfied.
However, for the constant-strain bar and the constant-

strain triangle, element equilibrium is satisfied.
Also the cubic displacement function is shown to satisfy

the basic beam equilibrium differential equation, and
hence, to satisfy element force and moment equilibrium.

3. Equilibrium is not usually satisfied between elements.
A differential element including parts of two adjacent finite

elements is usually not in equilibrium (see the figure
below).

A differential element including parts of two adjacent finite

elements is usually not in equilibrium (see the figure
below).

For line elements, such as used for truss and frame

analysis, interelement equilibrium is satisfied.
However, for two- and three-dimensional elements,

interelement equilibrium is not usually satisfied.
Also, the coarseness of the mesh causes this lack of

interelement equilibrium to be even more pronounced.

4. Compatibility is satisfied within an element as long as the

element displacement field is continuous; hence, individual
elements do not tear apart.
5. In the formulation of the element equations, compatibility is

invoked at the nodes.
Hence, elements remain connected at their common

nodes. Similarly, the structure remains connected to its
support nodes because boundary conditions are invoked
at these nodes.

6. Compatibility may or may not be satisfied along

interelement boundaries.
For line elements such as bars and beams, interelement

boundaries are merely nodes.
The constant-strain triangle remain straight sided when

deformed and therefore, interelement compatibility exists
for these elements. Incompatible elements, those that
allow gaps or overlaps between elements, can be
acceptable and even desirable.

Convergence of Solution
When the mesh size is reduced - that is the number of
elements is increased - we are ensured of monotonic
convergence of the solution when compatible and complete
displacement functions are used.


Case Number of Number of Nodes Aspect

Elements Ratio Point A
1 12 21 2 -0.740
2 24 39 1 -0.980
3 32 45 3 -0.875
4 64 85 1.5 -1.078
5 80 105 1.2 -1.100
Exact Solution -1.152

Patch Test
The Patch test is used to test the convergence of a solution of
an element being used in the model.
The test requires that the element must be able to

accommodate both rigid-body motion and a constant state of
strain, as both are possible within a structure.
The test can also be applied to determine if sufficient Gauss

points have been used in the numerical integration process
to evaluate the stiffness matrix for isoparametric
formulations.



Interpretation of Stresses
In the stiffness or displacement formulation of the finite
element method, used in this course, the primary quantities
determined are the interelement nodal displacements of the
assemblage.
Secondary quantities, such as stress and strain, are computed

based on these nodal displacements.
In the case of the bar and constant-strain triangles, stresses

are constant over the element.
For these elements, it is common practice to assign the stress

to the centroid of the element with acceptable results.

An alternative procedure sometimes is to use an average
(possibly weighted) value of the stresses evaluated at each
node of the element.
This averaging method is often based interpolating the

element nodal values using the element shape functions.
The averaging method is called smoothing.
While the results from smoothing may be pleasing to the eye,

they may not indicate potential problems with the model and
the results.

You should always view the unsmoothed contour plots as well.
Highly discontinuous contours between elements in a region

of an unsmoothed plot indicate modeling problems and
typically require additional refinement of the element mesh
in the suspect region.

Static Condensation
Let’s consider the concept of static condensation and used it
to develop the stiffness matrix of a quadrilateral element.
Consider a general quadrilateral element as shown below.

Static Condensation
An imaginary node 5 is temporary introduced at the
intersection of the diagonals of the quadrilateral to create
four triangles.

Static Condensation
We can superimpose the stiffness matrices of the four
triangles to create the stiffness matrix of the quadrilateral
element, where the internal imaginary node 5 degrees of
freedom are said to be condensed out so that they never
enter into the final equations.

Static Condensation
Let’s start by partitioning the equilibrium equations:
K11 K12  de  Fe 

K     
 21 K22   di   Fi 
where di is the vector of displacements corresponding to the
imaginary internal node, Fi is the vector of loads at the
internal node, and de and Fe are the actual displacements
and loads, respectively.
Rewriting the above equations we gives:
K11de  K12d i  Fe
K 21de  K 22d i  Fi

Static Condensation
Solving the second equations for di gives:
di  K221K21de  K221Fi
Substituting the above equation, we obtain the condensed

equilibrium equation: k d  F
c c c
where k  K  K K 1K
c 11 12 22 21
Fc  Fe  K12K 22 1Fi
where kc and Fc are called the condensed stiffness matrix

and the condensed load vector, respectively.

Static Condensation
An advantage of the four-CST quadrilaterals is that the
solution becomes less dependent on the skew of the
subdivision mesh.
The skew means a directional stiffness bias that is built into a

model through certain discretization patterns.

Static Condensation
The stiffness matrix of a typical triangular element, call it
element 1, labeled with nodes 1, 2, and 5 is given as:
k11
(1) (1)
k12 (1)
k15 
 (1) (1) 
k (1)   k 21 (1)
k 22 k 25 
k51
(1) (1)
k52 (1)
k55 
 
where kij (1) is a 2 x 2 matrix.

Static Condensation
The assembled stiffness matrix
for the quadrilateral is:
(u1,v1 ) (u2 ,v 2 ) (u3 ,v 3 ) (u4 ,v 4 ) (u5 ,v 5 )

 k11
(1)
 k11
( 4) (1)
k12 0 ( 4)
k14 (1)
k15  k15
(4)

 (2) 
 k 21
(1) (1)
k 22  k 22
(2) (2)
k23 0 (1)
k 25  k 25 
 0 (2)
k32 (2)
k33  k33(3) (3)
k34 (2)
k35  k35
(3)

k    k ( 4) 0 (3)
k 43 (3)
k 44  k 44
(4) (3)
k 45 (4) 
 k 45
 41 
 k55  k55
(1) (2) 
 k51
(1)
 k51
( 4) (1)
k52  k52
(2) (2)
k53  k53
(3) (3)
k54  k54
(4)

  k55
(3)
 k55( 4)


Example Problem
Consider the quadrilateral with internal node 5 and
dimensions as shown below. Apply the static condensation
technique.
Using the CST stiffness matrix for plain strain, we get:

1 2 5
3 4 5
 1.5 1.0 0.1 0.2 1.6 1.2 

 1.0 3.0 0.2 2.6 0.8 5.6 
 
E  0.1 0.2 1.5 1.0 1.6 1.2 
[k (1) ]  [k (3) ]   
4.16  0.2 2.6 1.0 3.0 0.8 5.6 
 1.6 0.8 1.6 0.8 3.2 0.0 
 
 1.2 5.6 1.2 5.6 0.0 11.2 

Example Problem
Consider the quadrilateral with internal node 5 and
dimensions as shown below. Apply the static condensation
technique.
Using the CST stiffness matrix for plain strain, we get:

4 1 5
2 3 5
 1.5 1.0 0.1 0.2 1.6 1.2 
 1.0 3.0 0.2 2.6 0.8 5.6 
 
E  0.1 0.2 1.5 1.0 1.6 1.2 
[k ]  [k ] 
(2) ( 4)
 
4.16  0.2 2.6 1.0 3.0 0.8 5.6 
 1.6 0.8 1.6 0.8 3.2 0.0 
 
 1.2 5.6 1.2 5.6 0.0 11.2 

Example Problem
The resulting assembled matrix before static condensation is:

Example Problem
The resulting assembled matrix before static condensation is:

Flowchart for the Solution of Place Stress/Strain Problems
The following flowchart is
typical for a finite element
process used for the
analysis of plane stress and
plane strain problems.


A bicycle company is disappointed with the negative feedback
they have received on their latest model, and they have
pinpointed the problem to an outdated bicycle crank design.
They have outsourced the task of analyzing the crank to you,

providing you with the geometry of the bicycle crank and
attached pedal shaft shown below.





Coarser Mesh Finer Mesh

DMX 0.026148 in 0.026651 in
SMX 25,308 psi 27,942 psi
The maximum displacement at the tip of shaft is 1.9% greater

and the maximum stress is 10% greater


Below is a finite element model and the von Mises stress plot
for a beam welded to a column by top and bottom fillet
welds. The material is steel with E = 205 GPa and  = 0.25

After mesh refinement around the top weld to double the
number of elements in the weld, the maximum Von Mises
stress was determined to be 87.3 Mpa compares reasonably
well with that obtained by the classical method where a
value of 94 MPa was obtained.

Problems
14. Do problems 7.2, 7.3, and 7.4 on pages 421 - 436 in your
D. Logan.
15. Work problems 7.13, 7.16, and 7.22 on pages 421 - 436 in
your textbook “A First Course in the Finite Element Method”
by D. Logan.
End of Chapter 7
CIVL 7/8117 Chapter 8 - Linear-Strain Triangle Equations 1/27
Chapter 8 – Linear-Strain Triangle Equations
Learning Objectives
• To develop the linear-strain triangular (LST)
• To describe how the LST stiffness matrix can be

determined.
• To compare the differences in results using the

CST and LST elements.
Development of the Linear-Strain Triangle Equations

Introduction
In this section we will develop a higher-order triangular element,
called the linear-strain triangle (LST).
This element has many advantages over the constant-strain
triangle (CST).
The LST element has six nodes and twelve displacement
degrees of freedom.
The displacement function for the triangle is quadratic.

Derivation of the Linear-Strain Triangular Elemental
The procedure to derive the LST element stiffness matrix and

element equations is identical to that used for the CST
element.
Consider the triangular element shown in the figure below:

Each node has two degrees of freedom: displacements in

the x and y directions.
We will let ui and vi represent the node i displacement
components in the x and y directions, respectively.

The nodal displacements for an LST element are:

u1 
v 
 1
u2 
 
 d1  v 2 
d  u 
 2  3
d  v
d  d3   u3 
 4  4
d5  v 4 
   
d 6  u5  The nodes are ordered
v 5 
  counterclockwise around the
u6  element
v 
 6

Consider a straight-sided triangular element shown below:
4 2
3
6
y 5
1
x
The variation of the displacements over the element may be

expressed as:
u( x, y )  a1  a2 x  a3 y  a4 x 2  a5 xy  a6 y 2
v ( x, y )  a7  a8 x  a9 y  a10 x 2  a11xy  a12 y 2


The displacement compatibility among adjoining elements is
satisfied because the three nodes defining adjacent sides
define a unique a parabola.
The CST and LST triangles are variations of the Pascal
triangles as show below.

The general element displacement functions are:
 a1 
a 
 2
1 x y x2 xy y2 0 0 0 0 0 0  · 
   2  
0 0 0 0 0 0 1 x y x xy y2  · 
 a11 
 
a12 
  M * a

To obtain the values for the a’s substitute the coordinates of the
nodal points into the above equations:
 u1   1 x1 y1 x12 x1y 1 y12 0 0 0 0 0 0   a1 

u   2 2  
 2  1 x2 y2 x 2 x2 y 2 y 2 0 0 0 0 0 0   a2 
                
    
u6   1 x6 y6 x62 x6 y 6 y 62 0 0 0 0 0 0   a6 
  2  
v1   0 0 0 0 0 0 1 x1 y1 x12 x1y1 y1  a7 

                
    
v 5   0 0 0 0 0 0 1 x5 y5 x52 x5 y 5 y 52   a11 

v 6   x62 y 6  a12 
2
 0 0 0 0 0 0 1 x6 y6 x6 y 6

Solving for the a’s and writing the results in matrix form gives
1
 a1   1 x1 y1 x12 x1y 1 y12 0 0 0 0 0 0   u1 
a     
 2  1 x2 y2 x22 x2 y 2 y 22 0 0 0 0 0 0  u2 
                 
     
 a6   1 x6 y6 x62 x6 y 6 y 62 0 0 0 0 0 0  u6 
  2   
 a7   0 0 0 0 0 0 1 x1 y1 x12 x1y 1 y1 v1
  
                 
    
 a11   0 0 0 0 0 0 1 x5 y5 x52 x5 y 5 y 52  v 5 

a12  
 0 0 0 0 0 0 1 x6 y6 x62 x6 y 6 y 6  v 6 
2
a   x  u
1
where [x] is the 12 x 12 matrix on the right-hand-side of the

above equation.

The “best” way to invert [x] is to use a computer.

Note that only the 6 x 6 part of [x] really need be inverted.
  M * a a   x  u   M *  x  u

1 1

The general displacement expressions in terms of interpolation

functions and the nodal degrees of freedom are:
  N u where N   M *  x 

1

Step 3 - Define the Strain-Displacement and Stress-Strain
Relationships
element are:
 u 
 
  x   x 
   v 
{ }    y    
   y 
 xy   u v 
  
 y x 

Step 3 - Define the Strain-Displacement and Stress-Strain
Relationships
element are:
u v  a1 
a 
0 1 0 2x y 0 0 0 0 0 0 0  2 
 · 
    0 0 0 0 0 0 0 1 0 0 x 2y   
 ·
 0 0 1 0 x 2y 0 1 0 2x y 0   
 a11 
 
a12 

Observe that the strains are linear over the triangular element;
therefore, the element is called a linear-strain triangle (LST).
The above equation may be written in matrix form as
   M 'a
where [M ’] is based on derivatives of [M*].
0 1 0 2x y 0 0 0 0 0 0 0
M '   0 0 0 0 0 0 0 1 0 0 x 2y 

 0 0 1 0 x 2y 0 1 0 2x y 0 

If we substitute the values of a’s into the above equation gives:
   M 'a a   x  u

1
    B d
where [B] is a function of the nodal coordinates (x1, y1) through

(x6, y6).
B   M ' x 
1

The stresses are given as:  x   x 

   
 y   [D ]   y 
   
 xy   xy 
For plane stress, [D] is:
1  0 
E  
[D ]  2 
 1 0 
1 
0 0 0.5 1    
For plane strain, [D] is: 1    0 

[D ] 
E   1  0 
1   1  2   0 
0.5   
 0


[k ]   [B ]T [D][B ]dV
V
However, [B] is now a function of x and y; therefore, we must

integrate the above expression to develop the element
stiffness matrix.

The [B] matrix is:
 1 0 2 0 3 0 4 0 5 0 6 0
1 
B   2 A  0 1 0 2 0 3 0 4 0 5 0  6 
  1 1  2 2 3 3 4 4 5 5 6  6 
The β’s and the γ’s are functions of x and y as well as the nodal
coordinates.

The [B] matrix is:
 1 0 2 0 3 0 4 0 5 0 6 0
1 
B   2 A  0 1 0 2 0 3 0 4 0 5 0  6 
  1 1  2 2 3 3 4 4 5 5 6  6 
The stiffness matrix is a 12 x 12 matrix and is very cumbersome

to compute in explicit form.
However, if the origin of the coordinates is the centroid of the

element, the integrations become more amenable.
Typically, the integrations are computed numerically.

The element body forces and surface forces should not be
automatically lumped at the nodes.
The following integration should be computed:
fb    [N ]T { X }dV
V
fs    [N ]T {T }dS
S

The element equations are:
 f1x   k11 k12 k13  k1,12   u1 

f   k  k2,12  v1 
 1y   21 k22 k23
  
f2 x    k31 k32 k33  k3,12  u2 
      

   
    
f6 y  k12,1 k12,2 k12,3  k12,12  v 6 
12  1 12  12 12  1

Steps 5, 6, and 7
Assembling the global stiffness matrix, determining the global

displacements, and calculating the stresses, are identical to
the procedures used for CST elements.

Example LST Stiffness Determination
Consider a straight-sided triangular element shown below:

The triangle has a base dimension of b and a height h, with mid-

side nodes.
We can calculate the coefficients a1 through a6 by evaluating
the displacement u at each node.
u( x, y )  a1  a2 x  a3 y  a4 x 2  a5 xy  a6 y 2
u1  u (0,0)  a1
u2  u(b,0)  a1  a2b  a4 b 2
u3  u(0, h )  a1  a3 h  a6 h 2

The triangle has a base dimension of b and a height h, with mid-

side nodes.
We can calculate the coefficients a1 through a6 by evaluating
the displacement u at each node.
2 2
b h b h b bh h
u4  u  ,   a1  a2  a3  a4    a5  a6  
2 2 2 2 2 4 2
2
 h h h
u5  u  0,   a1  a3  a6  
 2 2 2
2
b  b b
u6  u  ,0   a1  a2  a4  
2  2 2

Solving the above equations simultaneously for the a’s gives:

4u6  3u1  u2
a1  u1 a2 
b
4u5  3u1  u3 2  u2  2u6  u1 

a3  a4 
h b2
4  u1  u 4  u5  u6  2  u3  2u5  u1 
a5  a6 
bh h2

The u displacement equation is:
 4u  3u1  u2   4u  3u1  u3 
u( x, y )  u1   6  x 5 y
 b   h 
 2  u2  2u6  u1   2  4  u1  u4  u5  u6  
 x   xy
 b2   bh 
 2  u3  2u5  u1   2
 y
 h2 

The v displacement equation can be determined in a manner
identical to that used for the u displacement:
 4v  3v1  v 2   4v  3v1  v 3 
v ( x, y )  v1   6  x 5 y
 b   h 
 2 v 2  2v 6  v1   2  4 v1  v 4  v 5  v 6  
 x   xy
 b2   bh 
 2 v 3  2v 5  v1   2
 y
 h2 

The general form of the displacement expressions in terms of
the interpolation functions is given as  u1 
v 
u  N1 0 N2 0 N3 0 N 4 0 N5 0 N6 0   
1
    
v   0 N1 0 N2 0 N3 0 N4 0 N5 0 N6   
u
 6
v 6 
where the interpolation functions are:
3x 3y 2x2 4xy 2y 2 x 2x2

N1  1     N2    2
b h b2 bh h2 b b
y 2y 2
N3    2
h h

The general form of the displacement expressions in terms of
the interpolation functions is given as  u1 
v 
u  N1 0 N2 0 N3 0 N 4 0 N5 0 N6 0   
1
    
v   0 N1 0 N2 0 N3 0 N4 0 N5 0 N6   
u
 6
u6 
where the interpolation functions are:
4xy 4y 4 xy 4y 2
N4  N5    2
bh h bh h
4 x 4 xy 4 x 2
N6    2
b bh b

The element interpolation functions N have two basic shapes.
The behavior of the functions N1, N2, and N3 is similar except

referenced at different nodes.
The shape function N1 is shown below:

1 3
5
N1
1
4
6
y
2
x

The second type of interpolation function is valid for functions

N4, N5, and N6. The function N5 is shown below:
5
N5
1
4
6
y
x

The element strain is given as:
   B d
where the [B] matrix is:
 1 0  2 0 3 0 4 0 5 0 6 0
1 
B   2 A  0  1 0 2 0 3 0 4 0 5 0  6 
  1 1  2 2 3 3 4 4 5 5 6  6 
 N i   Ni 
i  2A    i  2A  
 x   y 

 N  bh
Therefore, since  i  2 A  i  A
 x  2
3x 3y 2x2 4xy 2y 2 4hx

N1  1     1  3h   4y
b h b2 bh h2 b
x 2x2 4hx
N2    2  2  h 
b b b
y 2y 2
N3    2 3  0
h h

 N  bh
Therefore, since  i  2 A  i  A
 x  2
4xy
N4   4  4y
bh
4y 4 xy 4y 2
N5    2 5  4y
h bh h
4 x 4 xy 4 x 2 8hx
N6    2  6  4h   4y
b bh b b

 N  bh
Therefore, since  i  2 A  i  A
 y  2
3x 3y 2x2 4xy 2y 2 4by

N1  1      1  3 b   4x
b h b2 bh h2 h
x 2x2
N2    2 2  0
b b
y 2y 2 4by
N3    2  3  b 
h h h

 N  bh
Therefore, since  i  2 A  i  A
 y  2
4xy
N4   4  4x
bh
4y 4 xy 4y 2  5  4b  4 x 
8by
N5    2
h bh h h
4 x 4 xy 4 x 2
N6    2  6  4x
b bh b

The stiffness matrix for a constant thickness element can be
obtained by substituting the β’s and the γ’s into the [B] and
then substituting [B] into the following expression and
evaluating the integral numerically.
[k ]   [B ]T [D][B ]dV
V

Comparison of Elements
For a given number of nodes, a better representation of true
stress and displacement is generally obtained using LST
elements than is obtained using the same number of nodes
a finer subdivision of CST elements.
For example, a single LST element gives better results than four
CST elements.

Consider the following cantilever beam with E = 30 x 106 psi,
 = 0.25, and t = 1 in.

Table 1 lists the series of tests run to compare results using the
CST and LST elements.
Table 1. Comparison of CST and LST results

Series of Test Number of Degrees of Number of
Runs Nodes Freedom, nd Elements
A-1 4 x 16 85 160 128 CST
A-2 8 x 32 297 576 512 CST
B-1 2 x 8 85 160 32 LST
B-2 4 x 16 297 576 128 LST

Table 2 shows comparisons of free-end (tip) deflection and
stress for each element type used to model the cantilever
beam.
Tip Deflection x Location

Runs nd Bandwidth, nd
(in) (ksi) (x, y)
A-1 160 14 -0.29555 67.236 (2.250,11.250)
A-2 576 22 -0.33850 81.302 (1.125,11.630)
B-1 160 18 -0.33470 58.885 (4.500,10.500)
B-2 576 22 -0.35159 69.956 (2.250,11.250)

Exact Solution -0.36133 80.000 (0,12)

The larger the number of degrees of freedom for a given type
of triangular element, the closer the solution converges to the
exact one (compare run A-1 to run A-2, and B-1 to B-2).

(in) (ksi) (x, y)
A-1 160 14 -0.29555 67.236 (2.250,11.250)
A-2 576 22 -0.33850 81.302 (1.125,11.630)
B-1 160 18 -0.33470 58.885 (4.500,10.500)
B-2 576 22 -0.35159 69.956 (2.250,11.250)
Exact Solution -0.36133 80.000 (0,12)

For a given number of nodes, the LST analysis yields
somewhat better results than the CST analysis (compare run
A-1 to run B-1).

(in) (ksi) (x, y)
A-1 160 14 -0.29555 67.236 (2.250,11.250)
A-2 576 22 -0.33850 81.302 (1.125,11.630)
B-1 160 18 -0.33470 58.885 (4.500,10.500)
B-2 576 22 -0.35159 69.956 (2.250,11.250)

Exact Solution -0.36133 80.000 (0,12)

Although the CST element is rather poor in modeling bending,
we observe that the element can be used to model a beam in
bending if sufficient number of elements is used.

(in) (ksi) (x, y)
A-1 160 14 -0.29555 67.236 (2.250,11.250)
A-2 576 22 -0.33850 81.302 (1.125,11.630)
B-1 160 18 -0.33470 58.885 (4.500,10.500)
B-2 576 22 -0.35159 69.956 (2.250,11.250)
Exact Solution -0.36133 80.000 (0,12)

In general, both the LST and CST analyses yield sufficient
results for most plane stress/strain problems provided a
sufficient number of elements are used.

(in) (ksi) (x, y)
A-1 160 14 -0.29555 67.236 (2.250,11.250)
A-2 576 22 -0.33850 81.302 (1.125,11.630)
B-1 160 18 -0.33470 58.885 (4.500,10.500)
B-2 576 22 -0.35159 69.956 (2.250,11.250)

Exact Solution -0.36133 80.000 (0,12)

The results of Table 2 indicate:
That the LST model might be preferred over the CST model
for plane stress applications when a relatively small number
of nodes is used.
The use of triangular elements of higher order, such as the

LST, is not visibly more advantageous when large numbers
of nodes are used, particularly when the cost of the
formation of element stiffnesses, equation bandwidth, and
overall complexities involved in the computer modeling are
considered.

Most commercial programs incorporate the use of CST and/or

LST elements for plane stress/strain problems although these
elements are used primarily as transition elements (usually
during mesh generation).
Also, recall that finite element displacements will always be less
than the exact ones, because finite element models are
always predicted to be stiffer than the actual structures when
using the displacement formulation of the finite element
method.

A comparison of CST and LST models of a plate subjected to

parabolically distributed edge loads is shown below.

The LST model converges to the exact solution for horizontal

displacement at point A faster than does the CST model.

However, the CST model is quite acceptable even for modest

numbers of degrees of freedom.

For example, a CST model with 100 nodes (200 degrees of
freedom) often yields nearly as accurate a solution as does
an LST model with the same number of degrees of freedom.

Problems
16. Do problems 8.3, 8.6, and 8.7 on pages 448 - 450 in your
D. Logan.
End of Chapter 8
CIVL 7/8117 Chapter 9 - Axisymmetric Elements 1/63
Chapter 9 – Axisymmetric Elements
Learning Objectives
• To review the basic concepts and theory of
elasticity equations for axisymmetric behavior.
• To derive the axisymmetric element stiffness

matrix, body force, and surface traction equations.
• To demonstrate the solution of an axisymmetric

pressure vessel using the stiffness method.
• To compare the finite element solution to an exact

solution for a cylindrical pressure vessel.
• To illustrate some practical applications of

axisymmetric elements.
Axisymmetric Elements
Introduction
In previous chapters, we have been concerned with line or one-
dimensional elements (Chapters 2 through 5) and two-
dimensional elements (Chapters 6 through 8).
In this chapter, we consider a special two-dimensional element

called the axisymmetric element.
This element is quite useful when symmetry with respect to

geometry and loading exists about an axis of the body being
analyzed.
Problems that involve soil masses subjected to circular footing

loads or thick-walled pressure vessels can often be analyzed
using the element developed in this chapter.
Introduction
We begin with the development of the stiffness matrix for the
simplest axisymmetric element, the triangular torus, whose
vertical cross section is a plane triangle.
We then present the longhand solution of a thick-walled

pressure vessel to illustrate the use of the axisymmetric
element equations.
This is followed by a description of some typical large-scale

problems that have been modeled using the axisymmetric
element.
Derivation of the Stiffness Matrix
In this section, we will derive the stiffness matrix and the body
and surface force matrices for the axisymmetric element.
However, before the development, we will first present some

fundamental concepts prerequisite to the understanding of the
derivation.
Axisymmetric elements are triangular tori such that each
element is symmetric with respect to geometry and loading
about an axis such as the z axis.
Hence, the z axis is called the axis

of symmetry or the axis of
revolution.
Each vertical cross section of the

element is a plane triangle.
The nodal points of an axisymmetric

triangular element describe
circumferential lines.
In plane stress problems, stresses exist only in the x-y plane.
In axisymmetric problems, the radial displacements develop

circumferential strains that induce stresses r ,  , z and rz
where r, , and z indicate the radial, circumferential, and
longitudinal directions, respectively.
Triangular torus elements are often used to idealize the

axisymmetric system because they can be used to simulate
complex surfaces and are simple to work with.
For instance, the axisymmetric problem of a semi-infinite half-
space loaded by a circular area (circular footing) can be
solved using the axisymmetric element developed in this
chapter.
For instance, the axisymmetric problem of a domed pressure
vessel can be solved using the axisymmetric element
developed in this chapter.
For instance, the axisymmetric problem of stresses acting on
the barrel under an internal pressure loading.
For instance, the axisymmetric problem of an engine valve stem
can be solved using the axisymmetric element developed in
this chapter.
For instance, an axisymmetric specimen loaded under tension-
compression.
Because of symmetry about the z axis, the stresses are
independent of the  coordinate.
Therefore, all derivatives with respect to  vanish, and the

displacement component v (tangent to the  direction), the
shear strains r and z and the shear stresses r and z are
all zero.
Consider an axisymmetric ring element and its cross section to

represent the general state of strain for an axisymmetric
problem.
The displacements can be expressed for element ABCD in the
plane of a cross-section in cylindrical coordinates.
We then let u and w denote the displacements in the radial and

longitudinal directions, respectively.
The side AB of the element is displaced an amount u, and side
CD is then displaced an amount u + (u/ r) in the radial
direction.
u
The normal strain in the radial direction is then given by:  r 
r
The strain in the tangential direction depends on the tangential
displacement v and on the radial displacement u.
However, for axisymmetric deformation behavior, recall that the

tangential displacement v is equal to zero.
The tangential strain is due only to the radial displacement.
Having only radial displacement u, the new length of the arc AB

is (r + u)d, and the tangential strain is then given by:
 
 r  u  d  rd 
u
rd r
Consider the longitudinal element BDEF to obtain the
longitudinal strain and the shear strain.
The element displaces by

amounts u and w in the
radial and longitudinal
directions at point E.
The element displaces

additional amounts:
(w/z)dz along line BE and
(u/r)dr along line EF.
Furthermore, observing lines EF and BE, we see that point F
moves upward an amount (w/r)dr with respect to point E
and point B moves to the right an amount (u/z)dz with
respect to point E.
The longitudinal normal

strain is given by:
w
z 
z
The shear strain in the r-z
plane is:
u w
 rz  
z r
Summarizing the strain-displacement relationships gives:
u u w u w
r    z   rz  
r r z z r
The isotropic stress-strain relationship, obtained by simplifying

the general stress-strain relationships, is:
 r  1    0 0  r 
    1 0    z 
 z E 
0
 
 

   1   1  2   0 0 1 0    
 
 rz   0 0 0 0.5     rz 
The procedure to derive the element stiffness matrix and
element equations is identical to that used for the plane-stress
in Chapter 6.
An axisymmetric solid is shown discretized below, along with a
typical triangular element.
The procedure to derive the element stiffness matrix and
element equations is identical to that used for the plane-stress
in Chapter 6.
The stresses in the axisymmetric problem are:
The element displacement functions are taken to be:
u(r , z )  a1  a2r  a3 z
w (r , z )  a4  a5 r  a6 z
The nodal displacements are:  ui 

w 
 di   
i
u 
d    d j    j 
d   w j 
 m  u 
 
m
 m
w
The function u evaluated at node i is: u(ri , zi )  a1  a2ri  a3 zi
The general displacement function is then expressed in matrix

form as:
 a1 
a 
 2
 a1  a2 r  a3 z   1 r z 0 0 0  a3 
 i      
 4
a  a5 r  a6 
z 0 0 0 1 r z  a4 
a5 
 
a6 
By substituting the coordinates of the nodal points into the
equation we can solve for the a's:
 ui  1 ri zi   a1 
    
z j  a2  a   x  u
1
 u j   1 r j  
u  1 r zm  a3 
 m  m
 w i  1 ri zi  a4 
    
z j  a5  a   x  w 
1
 w j   1 r j  
w  1 r zm  a6 
 m  m
Performing the inversion operations we have:
 i  j m  1 ri zi
1  
1
[x]  i  j m  2A  1 rj zj
2A 
  i  j  m  1 rm zm
2 A  ri  z j  zm   r j  zm  zi   rm  zi  z j 
where A is the area of the triangle

i
 i  j m 
1   j m
1
[x]  i  j m 
2A 
  i  j  m 
 i  r j zm  z j rm  i  z j  zm  i  rm  r j
 j  ri zm  zi rm  j  zm  zi  j  ri  rm
 m  ri z j  zi r j  m  zi  z j  m  r j  ri
The values of a may be written matrix form as:
 a1   i  j  m   ui 
  1   
a2  
2 A  i  j m   u j 
a    i  j  m  um 
 3
a4   i  j m   wi 
  1   
a5  
2 A  i  j m   w j 
a    i  j  m  w m 
 6
Expanding the above equations:
 a1 
 
u  1 r z a2 
a 
 3
Substituting the values for a into the above equation gives:

 i  j  m   ui 
u  1 r z    i  j  m   u j 
1
2A  
  i  j  m  um 
We will now derive the u displacement function in terms of the
coordinates r and z.
 i ui   j u j   mum 
1  
u  1 r z    i u i   j u j   m um 
2A
  i ui   j u j   mum 
u(r , z ) 
1
2A

 i   i r   i z  u i   j   j r   j z  u j
  m   m r   m z um 
We will now derive the w displacement function in terms of the
coordinates r and z.
 i w i   j w j   mw m 
1  
w   1 r z    i w i   j w j   mw m 
2A
  i w i   j w j   mw m 
w (r , z ) 
1
2A

 i   i r   i z  w i    j   j r   j z  w j
  m   m r   m z w m 
The displacements can be written in a more convenience form
as:
u ( r , z )  Ni u i  N j u j  N m u m
w ( r , z )  N i w i  N j w j  N mw m
where:
1
Ni   i   i r   i z 
2A
Nj 
1
2A
 j   j r   j z 
1
Nm   m   m r   m z 
2A
The elemental displacements can be summarized as:
 u ( r , z )   Ni ui  N j u j  Nmum 
 i     
w (r , z ) Ni w i  N j w j  Nmw m 
 ui 
w 
 i
N i 0 Nj 0 Nm 0   uj 
{ }    
0 Ni 0 Nj 0 Nm   w j 
 um 
 
{ }  [N ]{d } w m 
In another form the equations are:
N 0 Nj 0 Nm 0 
N    0i Ni 0 Nj 0 Nm 


Ni Nj Nm
1 z z y
1
r m r m r m
i i i
j j j

Ni Nj Nm
1 z z y
1
r m r m r m
i i i
j j j
So that u and w will yield a constant value for rigid-body
displacement, Ni + Nj + Nm = 1 for all r and z locations on the
element.

Ni Nj Nm
1 z z y
1
r m r m r m
i i i
j j j
element.
the x direction: u = u0
u0  u0  u0  N i  N j  N m 
0
 
Ni 0 N j 0 Nm 0  u0   Ni  N j  Nm  1
{ }     
 0 Ni 0 N j 0 Nm   0 
u0 
 
0
element.
the z direction: w = w0
 0  w w N N N
w  0 0 i j m
 0
N i 0 N j 0 N m 0   0   Ni  N j  Nm  1
{ }     
 0 Ni 0 N j 0 Nm  w 0 
0
 
w 0 

element are:
 u 
 r 
r     a2 
 w   
  
a6

   z  
{ }   z       a1 a3 z 
    u    a2 
 rz    r r 
r 
 u w   a3  a5 
  
 z r 

element are:
 u 
 r   a1 
r     0 1 0 0 0 0  a 
w   
 0 0 0 0 0 1  
2
   
   z  a3 
{ }   z      1 z   
    u   1 0 0 0  a4 
 rz   r  r r
 a 
 u w   0 0 1 0 1 0  5 
   a6 
 z r 

u 
 u,r   Ni ui  N j u j  Nmum 
r r
u,r  Ni ,r ui  N j ,r u j  Nm,r um

variable.
The derivatives of the interpolation functions are:

1  
Ni ,r   i   i r   i z   i
2 A r 2A
j m
N j ,r  Nm,r 
2A 2A
Therefore:
u

1
r 2 A
  i ui   j u j   m um 
In a similar manner, the remaining strain terms are
approximated as:
w

z 2 A
1
  i w i   j w j   mw m 
u 1   i iz    z
    i   ui   j   j  j  u j
r 2 A  r r   r r 
  z 
  m   m  m  um 
 r r  
u w

z r

2A
1
  i ui   i w i   j u j   j w j   m u m   mw m 
 ui 
r   i 0 j 0 m 0  
 v
  
0 i 0 j 0  m   i 
 z 1 uj 
   2A  i     i z 0  j     j z 0  m     m z 0   v 
   
r
i
r r
j
r r
m
r
 j 
 u 
 rz   i i j j m  m   m 
v m 
 di 
 
{ }  Bi Bj Bm   d j 
{ }  [B ]{d }  
d m 
 ui 
w 
r   i
  u 
 
{ }   z   Bi  B j  Bm   j 
 w
    j
 rz   um 
 
w m 
Stress-Strain Relationship: The in-plane stress-strain
relationship is:
 r   r 
   
 z z  { }  [D ][B ]{d }
   [ D ]  

    
 xy   rz 
1    0 0 
  1 0 0 
E  
[D ] 
1   1  2   0 0 1 0 
 
 0 0 0 0.5   
[k ]   [B ]T [D][B ]dV
V
For a circumferential differential element the integral becomes:
[k ]  2  [B ]T [D][B ] r dr dz
A
After integrating along the circumferential boundary, the [B]

matrix is a function of r and z.
Therefore, [k] is a function of r and z and is of order 6 x 6.
We can evaluate [k] by one of three methods:
1. Numerical integration (Gaussian quadrature) as

discussed in Chapter 10.
2. Explicit multiplication and term-by-term integration.

3. Evaluate [B] for a centroidal point  r , z  of the element
B  r , z    B 
ri  r j  rm zi  z j  zm
r r  zz 
3 3
3. Evaluate [B] for a centroidal point  r , z  of the element
As a first approximation: [k ]  2 rA B  D  B 

T
If the triangular subdivisions are consistent with the final

stress distribution (that is, small elements in regions of
high stress gradients), then acceptable results can be
obtained by Method 3.
Distributed Body Forces
Loads such as gravity (in the direction of the z axis) or
centrifugal forces in rotating machine parts (in the direction of
the r axis) are considered to be body forces.
The body forces can be found by:
Rb 
fb   2  N 
T
  r dr dz
A  Zb 
Where Rb = 2r for a machine part moving with a constant
angular velocity  about the z axis, with material mass density
 and radial coordinate r, and Zb is the body force per unit
volume due to the force of gravity.
The body forces can be found by:
Rb 
fb   2  N 
T
  r dr dz
A  Zb 
Considering the body force at node i, we have
T R  N 0
fbi   2  Ni   b  r dr dz Ni    0i
T
A  Zb   Ni 
Multiplying and integrating yields

2 Rb 
fbi     Ar
3  Zb 
The origin of the coordinates is the centroid of the element, and
Rb is the radially directed body force per unit volume evaluated
at the centroid of the element.
The body forces at nodes j and m are identical to those given
for node i. Hence, for an element, we have
Rb 
 
 Zb 
2 Ar Rb 
fb     Rb   2  r
3  Zb 
Rb 
 
 Zb 
Surface Forces
fs    Ns  T  dS
T
Surface forces can be found by
S
Where again [Ns] denotes the shape function matrix evaluated

along the surface where the surface traction acts.
For example, along the vertical face jm

of an element, let uniform loads pr, and
pz, be applied along surface r = rj.
Surface Forces
For instance, for node j, substituting Nj gives
zm
 j   j r   j z 0   pr 
f    21A 
sj
0  j   j r   j z   pz 
2 r j dz
zj 
Evaluated at r = rj and z
Surface Forces
Integrating the equations explicitly along with similar
evaluations for fsi and fsm the total distribution of surface force
to nodes i, j, and m is
0
0
 
2 r j  zm  z j   pr 
fs    
2  pz 
 pr 
 
 pz 
Steps 5 - 7
Steps 5 through 7, which involve assembling the total stiffness

matrix, total force matrix, and total set of equations; solving for
the nodal degrees of freedom; and calculating the element
stresses, are analogous to those of Chapter 6 for the CST
element.
Steps 5 - 7
The stresses are not constant in each element.
They are usually determined by one of two methods that we use

to determine the LST element stresses.
1. Either we determine the centroidal element stresses, or

2. We determine the nodal stresses for the element and
then average them.
The latter method has been shown to be more accurate in

some cases.
Example 1
For the element of an axisymmetric body rotating with a

constant angular velocity  = 100 rev/min, evaluate the
approximate body force matrix.
Include the weight of the material, where the weight density

w = 0.283 lb./in.3. Dimensions are inches.
Zb  0.283 lb./in.3
Example 1
Let evaluate the approximate body force matrix.
The body forces per unit volume evaluated at the centroid of the
element are:
Rb   2  r
2
  1min   0.283 lb./in.3
 100rpm   2 rad
   2.333in.
 60 sec    32.2 ft.2  12in. 
rev

 ft. 

s
 0.187lb./in.3
2 Ar

 
2 0.5in.2  2.333in.
 2.44in.2
3 3
Example 1
Let evaluate the approximate body force matrix.
The body forces per unit volume evaluated at the centroid of the
element are:
Rb   0.187   0.457 

    0.691
 Zb  0.283
   
2 Ar Rb   0.187  lb.  0.457 
fb      2.44in. 
3
 3   lb.
3  Zb  0.283  in. 0.691
Rb   0.187   0.457 
     
 Zb  0.283  0.691
Example 2
For the long, thick-walled cylinder under internal pressure p
equal to 1 psi, determine the displacements and stresses.
Example 2
First discretize the cylinder into four triangular elements.
A horizontal slice of the cylinder represents the total cylinder

behavior.
A coarse mesh of elements

is used for simplicity's sake
Example 2
The governing global matrix equation is:
 F1r   u1 
F  w 
 1z   1
 F2r   u2 
   
F2 z  w 2  [K] is a matrix of order 10 x 10
 F3 r   u3 
   K   
F3 z  w 3 
 F4 r   u4 
   
F4 z  w 4 
 F5 r   u5 
   
F5 z  w 5 
Example 2
Assemblage of the Stiffness Matrix
The [K] matrix is assembled in the usual manner by

superposition of the individual element stiffness matrices.
For simplicity's sake, we will evaluate [B] for a centroidal point
 r , z  of the element.
[k ]  2 rA B  D  B 
T
Example 2
Assemblage of the Stiffness Matrix: Element 1
r r zz
 ri  0.50   zi  0.00 
       
 r j   1.00  in.  z j   0.00  in.
    z  0.25 
rm  0.75   m  
 i 0 j 0 m 0
 0 i 0 j 0  m 
1 
B    i z j  jz m  mz 
2A   i  i 0  j  0  m  0
 r r r r r r 
 i i j j m  m 
Example 2
r r zz
 ri  0.50   zi  0.00 
       
 r j   1.00  in.  z j   0.00  in.
    z  0.25 
rm  0.75   m  
 i  r j zm  z j rm  (1.0)(0.25)  (0.0)(0.75)  0.25in.2
 j  ri zm  zi rm  (0.75)(0.0)  (0.25)(0.5)  0.125in.2
 m  ri z j  zi r j  (0.5)(0.0)  (0.0)(1.0)  0
Example 2
r r zz
 ri  0.50   zi  0.00 
       
 r j   1.00  in.  z j   0.00  in.
    z  0.25 
rm  0.75   m  
 i  z j  zm  0.25in2  i  rm  r j  0.25in.2
 j  zm  zi  0.25in2  j  ri  rm  0.25in.2
 m  zi  z j  0  m  r j  ri  0.5in.2
Example 2
r r zz
 ri  0.50   zi  0.00 
       
 r j   1.00  in.  z j   0.00  in.
    z  0.25 
rm  0.75   m  
3 3
ri zi
r   0.75in. z   0.0833in.
i 1 3 i 1 3
1
A (0.5)(0.25)  0.0625in.2
2
Example 2
r r zz
 ri  0.50   zi  0.00 
       
 r j   1.00  in.  z j   0.00  in.
    z  0.25 
rm  0.75   m  
 0.25 0 0.25 0 0 0 
 0 0.25 0 0.25 0 0.5  1
1  
B  
0.125 0.0556 0 0.0556 0 0.0556 0  in.
 
 .025 0.25 0.25 0.25 0.5 0 
Example 2
r r zz
 ri  0.50   zi  0.00 
       
 r j   1.00  in.  z j   0.00  in.
    z  0.25 
rm  0.75   m  
1    0 0 
  1 0 0 
E
  1   1  2  0
D  
   0 1 0 
 
 0 0 0 0.5   
Assume that E = 30 x 106 psi and  = 0.3

Example 2
r r zz
 ri  0.50   zi  0.00 
       
 r j   1.00  in.  z j   0.00  in.
    z  0.25 
rm  0.75   m  
0.7 0.3 0 0 
0.3 0.7 0 0 
 
D   57.7 106  0 0 0.7 0  psi
 
 0 0 0 0.2
Example 2
r r zz
 ri  0.50   zi  0.00 
       
 r j   1.00  in.  z j   0.00  in.
    z  0.25 
rm  0.75   m  
 0.158 0.0583 0.0361 0.05 
 0.075 0.175 0.075 0.05 
 
B  D 4 x 4 
T 
57.7 106   0.192 0.0917 0.114 0.05 
 
0.125  0.075 0.175 0.075 0.05 
6x 4
 0.0167 0.0166 0.0388 0.1

 
 0.15 0.35 0.15 0
Example 2
r r zz
 ri  0.50   zi  0.00 
       
 r j   1.00  in.  z j   0.00  in.
    z  0.25 
rm  0.75   m  
i=1 j=2 m=5
 54.46 29.45 31.63 2.26 29.37 31.71
 29.45 61.17 11.33 33.98 31.72 95.15 
 
6  31.63 11.33 72.59 38.52 20.31 49.84  lb.
(1)

k   10  33.98 38.52 61.17

22.66 95.15  in.
 2.26
 29.37 31.72 20.31 22.66 56.72 9.06 
 
 31.71 95.15 49.84 95.15 9.06 190.31
Example 2
r r zz
 ri  1.00   zi  0.00 
       
 r j   1.00  in.  z j   0.50  in.
    z  0.25 
rm  0.75   m  
 i 0 j 0 m 0
 0 i 0 j 0  m 
1 
B    z j  jz m  mz 
2A  i  i  i 0  j  0  m  0
r r r r r r 
 i i j j m  m 
Example 2
r r zz
 ri  1.00   zi  0.00 
       
 r j   1.00  in.  z j   0.50  in.
    z  0.25 
rm  0.75   m  
 i  r j zm  z j rm  (1.0)(0.25)  (0.5)(0.75)  0.125in.2
 j  ri zm  zi rm  (0.75)(0.0)  (0.25)(1.0)  0.25in.2
 m  ri z j  zi r j  (1.0)(0.5)  (0.0)(1.0)  0.5in.2
Example 2
r r zz
 ri  1.00   zi  0.00 
       
 r j   1.00  in.  z j   0.50  in.
    z  0.25 
rm  0.75   m  
 i  z j  zm  0.25in2  i  rm  r j  0.25in.2
 j  zm  zi  0.25in2  j  ri  rm  0.25in.2
 m  zi  z j  0.5in2  m  r j  ri  0
Example 2
r r zz
 ri  1.00   zi  0.00 
       
 r j   1.00  in.  z j   0.50  in.
    z  0.25 
rm  0.75   m  
3 3
ri zi
r   0.9167in. z   0.25in.
i 1 3 i 1 3
1
A (0.5)(0.25)  0.0625in.2
2
Example 2
r r zz
 ri  1.00   zi  0.00 
       
 r j   1.00  in.  z j   0.50  in.
    z  0.25 
rm  0.75   m  
i=2 j=3 m=5
 85.75 46.07 52.52 12.84 118.92 33.23 
 46.07 74.77 12.84 41.54 45.32 33.23 
 
6  12.84 46.07 118.92 33.23  lb.
 
52.52 85.74
k   10 
(2)

 12.84 41.54 46.07 74.77 45.21 33.23  in.
 118.92 45.32 118.92 45.21 216.41 0
 
 33.23 33.23 33.23 33.23 0 66.46 
Example 2
r r zz
 ri  1.00   zi  0.50 
       
 r j   0.50  in.  z j   0.50  in.
    z  0.25 
rm  0.75   m  
 i 0 j 0 m 0
 0 i 0 j 0  m 
1 
B    i z j  jz m  mz 
2A   i  i 0  j  0  m  0
 r r r r r r 
 i i j j m  m 
Example 2
r r zz
 ri  1.00   zi  0.50 
       
 r j   0.50  in.  z j   0.50  in.
    z  0.25 
rm  0.75   m  
 i  r j zm  z j rm  (0.5)(0.25)  (0.5)(0.75)  0.25in.2
 j  ri zm  zi rm  (1.0)(0.25)  (0.5)(0.75)  0.125in.2
 m  ri z j  zi r j  (1.0)(0.5)  (0.5)(0.5)  0.25in.2

Example 2
r r zz
 ri  1.00   zi  0.50 
       
 r j   0.50  in.  z j   0.50  in.
    z  0.25 
rm  0.75   m  
 i  z j  zm  0.25in2  i  rm  r j  0.25in.2
 j  zm  zi  0.25in2  j  ri  rm  0.25in.2
 m  zi  z j  0  m  r j  ri  0.5in.2
Example 2
r r zz
 ri  1.00   zi  0.50 
       
 r j   0.50  in.  z j   0.50  in.
    z  0.25 
rm  0.75   m  
3 3
ri zi
r   0.75in. z   0.417in.
i 1 3 i 1 3
1
A (0.5)(0.25)  0.0625in.2
2
Example 2
r r zz
 ri  1.00   zi  0.50 
       
 r j   0.50  in.  z j   0.50  in.
    z  0.25 
rm  0.75   m  
i=3 j=4 m=5
 72.58 38.52 31.63 11.33 20.31 49.84 
 38.52 61.17 2.26 33.98 22.66 95.15 
 
    29.37 31.72  lb.
 
31.63 2.26 54.46 29.45
k (3)   106  
 11.33 33.98 29.45 61.17 31.72 95.15  in.
 20.31 22.66 29.37 31.72 56.72 9.06 
 
 49.84 95.15 31.72 95.15 9.06 190.31
Example 2
r r zz
 ri  0.50   zi  0.00 
       
 r j   0.75  in.  z j   0.25  in.
    z  0.50 
rm  0.50   m  
 i 0 j 0 m 0
 0 i 0 j 0  m 
1 
B    z j  jz m  mz 
2A  i  i  i 0  j  0  m  0
r r r r r r 
 i i j j m  m 
Example 2
r r zz
 ri  0.50   zi  0.00 
       
 r j   0.75  in.  z j   0.25  in.
    z  0.50 
rm  0.50   m  
 i  r j zm  z j rm  (0.75)(0.5)  (0.25)(0.5)  0.25in.2
 j  ri zm  zi rm  (0.5)(0.5)  (0.0)(0.5)  0.25in.2
 m  ri z j  zi r j  (0.5)(0.25)  (0.0)(0.75)  0.125in.2
Example 2
r r zz
 ri  0.50   zi  0.00 
       
 r j   0.75  in.  z j   0.25  in.
    z  0.50 
rm  0.50   m  
 i  z j  zm  0.25in2  i  rm  r j  0.25in.2
 j  zm  zi  0.5in2  j  ri  rm  0
 m  zi  z j  0.25in.2  m  r j  ri  0.25in.2
Example 2
r r zz
 ri  0.50   zi  0.00 
       
 r j   0.75  in.  z j   0.25  in.
    z  0.50 
rm  0.50   m  
3 3
ri zi
r   0.5833in. z   0.25in.
i 1 3 i 1 3
1
A (0.5)(0.25)  0.0625in.2
2
Example 2
r r zz
 ri  0.50   zi  0.00 
       
 r j   0.75  in.  z j   0.25  in.
    z  0.50 
rm  0.50   m  
i=1 j=5 m=4
 41.53 21.90 66.45 21.14 20.39 0.75 
 21.92 47.57 36.24 21.14 0.75 26.43 
 
6 66.45 36.24 169.14 0 66.45 36.24  lb.
 
k   10 
( 4)

 21.14 21.14 0 42.28 21.14 21.14  in.
 20.39 0.75 66.45 21.14 41.53 21.90 
 
 0.75 26.43 36.24 21.14 21.90 47.57 
Example 2
Using superposition of the element stiffness matrices, where we
rearrange the elements of each stiffness matrix in order of
increasing nodal degrees of freedom, we obtain the global
stiffness matrix as:
Example 2
1 2 3 4 5
 95.99 51.35 36.63 2.26 0 0 20.39 0.75 95.82 52.86 

 51.35 108.74 11.33 33.98 0 0 33.98 26.43 67.96 116.3 
 
36.63 11.33 158.34 84.59 52.52 12.84 0 0 139.2 83.07 
 2.26 33.98 84.59 135.94 12.84 41.54 0 0 67.98 128.4

 
 12.84 31.63 139.2 83.07  lb.
K   10 
6
0 0 52.52 158.33 84.59 11.33
0 0 12.84 41.54 84.59 135.94 2.26 33.98 67.98


128.4 in.
 
 20.39 33.98 0 0 31.63 2.26 95.99 51.35 95.82 52.86 
 0.75 26.43 51.35 116.3

 
0 0 11.33 33.98 108.74 67.96
95.82 67.96 139.2 67.98 139.2 67.98 95.82 67.96 498.99 0

 
52.86 116.3 83.07 128.4 83.07 128.4 52.86 116.3 0 489.36 
Example 2
The applied nodal forces are given as:
 pr  0.785 
p   0 
 z  
2 ri  z j  zm   0   0 
fs       lb.
2 0  0 
 pr  0.785 
   
 pz   0 
2  0.5in. 0.5in.
F1r  F4 r  1psi   0.785 lb.
2
Example 2
The resulting equations are:
0.785
 95.99 36.63 0.75 95.82 52.86   u1 
   0
51.35 2.26 0 0 20.39
 51.35 11.33 26.43 67.96 116.3  w1  
  
108.74 33.98 0 0 33.98
 0
36.63 11.33 158.34 84.59 52.52 12.84 0 0 139.2 83.07   u2 
 2.26    
0
  
33.98 84.59 135.94 12.84 41.54 0 0 67.98 128.4 w2

lb.  12.84 31.63 139.2  u3  
83.07   0
10  
0 0 52.52 158.33 84.59 11.33
  lb.
6
41.54 2.26 67.98 128.4

 w 0
 3 
in. 0 0 12.84 84.59 135.94 33.98

 20.39 33.98 0 0 31.63 2.26 95.99 51.35 95.82 52.86   u4  0.785
 0.75     
26.43 51.35 w
116.3  4 
 
0 0 11.33 33.98 108.74 67.96 0

95.82 0   u5  

67.96 139.2 67.98 139.2 67.98 95.82 67.96 498.99
   0
52.86 116.3 83.07 128.4 83.07 128.4 52.86 116.3 0 489.36  
w5  
 0
Example 2
The nodal displacements are:
 u1   0.0322
 w   0.00115 
 1  
 u2   0.0219 
   
w 2   0.00206 
 u3   0.0219 
  
w  0.00206
 
6
  10 in.
 3  
 u4   0.0322
   
w 4  0.00115 
 u5   0.0244 
   
w 5   0.0 
Example 2
The results for nodal displacements are as expected because
radial displacements at the inner edge are equal (u1 = u4) and
those at the outer edge are equal (u2 = u3).
Example 2
In addition, the axial displacements at the outer nodes and
inner nodes are equal but opposite in sign (w1 = -w4 and w2 =
-w3) as a result of the Poisson effect and symmetry.
Example 2
Finally, the axial displacement at the center node is zero
(w5 = 0), as it should be because of symmetry.
Example 2
Determine the stresses in each element as:    D  B  d 
For element 1:
0.7 0.3 0 0 
0.3 0.7 0 0 
 
D   57.7 106  0 0 0.7 0  psi
 
 0 0 0 0.2
 0.25 0 0.25 0 0 0 
 0 0.25 0 0.25 0 0.5  1
1  
B  
0.125 0.0556 0 0.0556 0 0.0556 0  in.
 
 .025 0.25 0.25 0.25 0.5 0 
Example 2
For element 1:
 u1 
 r  w  0.338 
   1 
 z  u2   0.0126 

   [ D ]  B 
  w    0.942 psi

    2  
 xy   u5   0.1037 
 
w 5 
Example 2
For element 2:
 u2 
 r  w  0.105 
   2 
 z  u3   0.0747 

   [ D ]  B 
  w    0.690  psi

    3  
 xy   u5   0.0 
 
w 5 
Example 2
For element 3:
 u3 
 r  w  0.337 
   3 
 z  u4   0.0125 

   [ D ]  B 
  w    0.942  psi

    4  
 xy   u5   0.1037 
 
w 5 
Example 2
For element 4:
 u1 
 r  w 
 1 0.470 
  0.1493 
 z  u5   
   [ D ]  B 
  w      psi

    5  1.426 
 xy   u4   0.0 
 
w 4 
Example 2
For element 1:
  r   0.338 
   
 z  0.0126 
   psi
    0.942
 xy  0.1037 
For element 3:
  r   0.337 
   
 z  0.0125 
     psi

   0.942 
 xy   0.1037 
Example 2
For element 4:
  r  0.470 
   
 z  0.1493 
   psi
    1.426 
 xy   0.0 
For element 2:
  r   0.105 
   
 z  0.0747 
     psi

   0.690 
 xy   0.0 
Example 2
The figure below shows the exact solution along with the results
determined here and the other results.
Example 2
Observe that agreement with the exact solution is quite good
except for the limited results due to the very coarse mesh
used in the longhand example.
Example 2
Stresses have been plotted at the center of the quadrilaterals
and were obtained by averaging the stresses in the four
connecting triangles.
Applications
Consider the finite element model of a steel-reinforced concrete
pressure vessel.
Applications
The vessel is a thick-walled cylinder with flat heads.
Applications
An axis of symmetry (the z axis) exists such that only one-half
of the r-z plane passing through the middle of the structure
need be modeled.
Applications
The concrete was modeled by using the axisymmetric triangular
element developed in this chapter.
Applications
The steel elements were laid out along the boundaries of the
concrete elements so as to maintain continuity (or perfect
bond assumption) between the concrete and the steel.
Applications
The vessel was then subjected to an internal pressure as
shown in the figure.
Applications
Note that the nodes along the axis of symmetry should be
supported by rollers preventing motion perpendicular to the
axis of symmetry.
Applications
The figure below shows a finite element model of a high-
strength steel die used in a thin-plastic-film-making process
Applications
The die is an irregularly shaped disk. An axis of symmetry with
respect to geometry and loading exists as shown.
Applications
The die was modeled by using simple quadrilateral axisym-
metric elements. The locations of high stress were of primary
concern.
Applications
The figure shows a plot of the von Mises stress contours for the
die. The von Mises (or equivalent, or effective) stress is often
used as a failure criterion in design.
Applications
The figure shows a stepped 4130 steel shaft with a fillet radius
subjected to an axial pressure of 1,000 psi in tension.
Applications
Fatigue analysis for reversed axial loading required an accurate
stress concentration factor to be applied to the average axial
stress of 1,000 psi.
Applications
Fatigue analysis for reversed axial loading required an accurate
stress concentration factor to be applied to the average axial
stress of 1,000 psi.
Applications
The figure below shows the resulting maximum principal stress
plot using a computer program.
Problems
17. Work problems 9.2, 9.3, and 9.5 on pages 475 - 485 in your
D. Logan.
18. Work problem 9.15 on pages 478 - 479 in your textbook “A

First Course in the Finite Element Method” by D. Logan.
Use the FEATools software to model the problem.

End of Chapter 9
CIVL 7/8117 Chapter 10 - Isoparametric Elements 1/85
Chapter 10 – Isoparametric Elements
Learning Objectives
• To formulate the isoparametric formulation of the
bar element stiffness matrix
• To present the isoparametric formulation of the
plane four-noded quadrilateral (Q4) element
stiffness matrix
• To describe two methods for numerical
integration— Newton-Cotes and Gaussian
Quadrature —used for evaluation of definite
integrals
• To solve an explicit example showing the
evaluation of the stiffness matrix for the plane
quadrilateral element by the four-point Gaussian
quadrature rule
Chapter 10 – Isoparametric Elements
Learning Objectives
• To illustrate by example how to evaluate the
stresses at a given point in a plane quadrilateral
element using Gaussian quadrature
• To evaluate the stiffness matrix of the three-noded
bar using Gaussian quadrature and compare the
result to that found by explicit evaluation of the
stiffness matrix for the bar
• To describe some higher-order shape functions for
the three-noded linear strain bar, the improved
bilinear quadratic (Q6), the eight- and nine-noded
quadratic quadrilateral (Q8 and Q9) elements, and
the twelve-noded cubic quadrilateral (Q12)
element
• To compare the performance of the CST, Q4, Q6,
Q8, and Q9 elements to beam elements
Isoparametric Elements
Introduction
In this chapter, we introduce the isoparametric formulation of
the element stiffness matrices.
After considering the linear-strain triangular element (LST) in

Chapter 8, we can see that the development of element
matrices and equations expressed in terms of a global
coordinate system becomes an enormously difficult task (if
even possible) except for the simplest of elements such as
the constant-strain triangle of Chapter 6.
Hence, the isoparametric formulation was developed.
Introduction
The isoparametric method may appear somewhat tedious
(and confusing initially), but it will lead to a simple computer
program formulation, and it is generally applicable for two-
and three-dimensional stress analysis and for nonstructural
problems.
The isoparametric formulation allows elements to he created

that are nonrectangular and have curved sides.
Numerous commercial computer programs (as described in

Chapter 1) have adapted this formulation for their various
libraries of elements.
Introduction
First, we will illustrate the isoparametric formulation to develop
the simple bar element stiffness matrix.
Use of the bar element makes it relatively easy to understand

the method because simple expressions result.
Then, we will consider the development of the isoparametric

formulation of the simple quadrilateral element stiffness
matrix.
Introduction
Next, we will introduce numerical integration methods for
evaluating the quadrilateral element stiffness matrix.
Then, we will illustrate the adaptability of the isoparametric

formulation to common numerical integration methods.
Finally, we will consider some higher-order elements and their

associated shape functions.
Isoparametric Formulation of the Bar Element
The term isoparametric is derived from the use of the same
shape functions (or interpolation functions) [N] to define the
element's geometric shape as are used to define the
displacements within the element.
Thus, when the interpolation function is u = a1 + a2s for the

displacement, we use x = a1 + a2s for the description of the
nodal coordinate of a point on the bar element and, hence, the
physical shape of the element.
Isoparametric element equations are formulated using a natural
(or intrinsic) coordinate system s that is defined by element
geometry and not by the element orientation in the global-
coordinate system.
In other words, axial coordinate s is attached to the bar and

remains directed along the axial length of the bar, regardless
of how the bar is oriented in space.
There is a relationship (called a transformation mapping)

between the natural coordinate systems and the global
coordinate system x for each element of a specific structure.
First, the natural coordinate s is attached to the element, with
the origin located at the center of the element.
The s axis need not be parallel to the x axis-this is only for

convenience.
Consider the bar element to have two degrees of freedom-axial

displacements u1 and u2 at each node associated with the
global x axis.
For the special case when the s and x axes are parallel to each
other, the s and x coordinates can be related by:
L
x  xc  s
2
Using the global coordinates x1 and x2 with xc =(x1 + x2)/2, we
can express the natural coordinate s in terms of the global
coordinates as:
  x  x2    2 
s  x  1  
 2    x2  x1  
The shape functions used to define a position within the bar are
found in a manner similar to that used in Chapter 3 to define
displacement within a bar (Section 3.1).
We begin by relating the natural coordinate to the global

coordinate by:
x  a1  a2s
Note that -1 ≤ s ≤ 1.
Solving for the a's in terms of x1 and x2, we obtain:
 1
x    1  s  x1  1  s  x2 
2
In matrix form:
x  1 s 1 s
x  N1 N2   1  N1  N2 
 x2  2 2
The linear shape functions map the s coordinate of any point in
the element to the x coordinate.
For instance, when s = -1, then x = x1 and

when s = 1, then x = x2
x  1 s 1 s
x  N1 N2   1  N1  N2 
 x2  2 2
x  1 s 1 s
x  N1 N2   1  N1  N2 
 x2  2 2
x  1 s 1 s
x  N1 N2   1  N1  N2 
 x2  2 2
When a particular coordinate s is substituted into [N] yields the
displacement of a point on the bar in terms of the nodal
degrees of freedom u1 and u2.
Since u and x are defined by the same shape functions at the

same nodes, the element is called isoparametric.
x  1 s 1 s
x  N1 N2   1  N1  N2 
 x2  2 2
Step 3 - Strain-Displacement and Stress-Strain Relationships
We now want to formulate element matrix [B] to evaluate [k].
We use the isoparametric formulation to illustrate its

manipulations.
For a simple bar element, no real advantage may appear

evident.
However, for higher-order elements, the advantage will become

clear because relatively simple computer program
formulations will result.
To construct the element stiffness matrix, determine the strain,

which is defined in terms of the derivative of the displacement
with respect to x.
The displacement u, however, is now a function of s so we must

apply the chain rule of differentiation to the function u as
follows:
du du dx du du du dx
 x   x  
ds dx ds dx dx ds ds
du u2  u1
The derivative of u with respect to s is: 
ds 2
dx x2  x1 L
The derivative of x with respect to s is:  
ds 2 2
1   u1 
 x   
1
Therefore the strain is:  
 L L  u2 
Since {} = [B]{d}, the strain-displacement matrix [B] is:
B    L
1 1
 L 
Recall that use of linear shape functions results in a constant [B]

matrix, and hence, in a constant strain within the element.
For higher-order elements, such as the quadratic bar with three

nodes, [B] becomes a function of natural coordinates s.
The stress matrix is again given by Hooke's law as:
   E    E B d 
L
The stiffness matrix is:  k    B  E B  A dx
T
However, in general, we must transform the coordinate x to s

because [B] is, in general, a function of s.
L 1
 f ( x ) dx   f (s ) J  ds
0 1
where [J] is called the Jacobian matrix.
In the one-dimensional case, we have |[J]| = J.
dx L
For the simple bar element: J   ds  2
The Jacobian determinant relates an element length (dx) in the

global-coordinate system to an element length (ds) in the
natural-coordinate system.
In general, |[J]| is a function of s and depends on the numerical

values of the nodal coordinates.
This can be seen by looking at for the equations for a

quadrilateral element.
The stiffness matrix in natural coordinates is:

1
L
k   2  B  E B  A ds
T
1
For the one-dimensional case, we have used the modulus of

elasticity E = [D].
Performing the simple integration, we obtain:
AE  1 1
k    1 1
L  
For higher-order one-dimensional elements, the integration in

closed form becomes difficult if not impossible.
Even the simple rectangular element stiffness matrix is difficult

to evaluate in closed form.
However, the use of numerical integration, as described in

Section 10.3, illustrates the distinct advantage of the
isoparametric formulation of the equations.
Determine the body-force matrix using the natural coordinate

system s. The body-force matrix is:
fb    [N ]T  X b  dV fb    [N ]T  X b  Adx

V V
Substituting for N1 and N2 and using dx = (L/2)ds
1  s 
1   ALX b 1
fb   A   2  X b  ds 
L

1 
1 s  2 2 1
 2 
The physical interpretation of the results for {fb} is that since AL

represents the volume of the element and Xb the body force
per unit volume, then ALXb is the total body force acting on the
element.
The factor ½ indicates that this body force is equally distributed

to the two nodes of the element.
1  s 
1   ALX b 1
fb   A   2  X b  ds 
L

1 
1 s  2 2 1
 2 
Determine the surface-force matrix using the natural coordinate

system s. The surface-force matrix is:
fs    [Ns ]T Tx  dS

S
Assuming the cross section is constant and the traction is

uniform over the perimeter and along the length of the
element, we obtain: L
fs    [Ns ]T Tx  dx

0
where we now assume {Tx} is in units of force per unit length.
1  s 
1   L 1
fs     2 Tx  ds  Tx   
L
1 
1 s  2 2 1
 2 
Since {Tx} is in force-per-unit-length {Tx}L is now the total force.
The ½ indicates that the uniform surface traction is equally

distributed to the two nodes of the element.
1  s 
1   L 1
fs     2 Tx  ds  Tx   
L
1 
1 s  2 2 1
 2 
Note that if {Tx} were a function of x (or s), then the amounts of
force allocated to each node would generally not be equal and
would be found through integration.
Isoparametric Formulation of the Quadrilateral Element
Recall that the term isoparametric is derived from the use of the
same interpolation functions to define the element shape as
are used to define the displacements within the element.
The approximation for displacement is:

u  a1  a2s  a3t  a4st
The description of a coordinate point in the plane element is:

x  a1  a2s  a3t  a4st
The natural-coordinate systems s-t defined by element

geometry and not by the element orientation in the global-
coordinate system x-y.
Much as in the bar element example, there is a transformation

mapping between the two coordinate systems for each
element of a specific structure, and this relationship must be
used in the element formulation.
We will now formulate the isoparametric formulation of the
simple linear plane quadrilateral element stiffness matrix.
This formulation is general enough to be applied to more

complicated (higher-order) elements such as a quadratic
plane element with three nodes along an edge, which can
have straight or quadratic curved sides.
Higher-order elements have additional nodes and use different

shape functions as compared to the linear element, but the
steps in the development of the stiffness matrices are the
same.
We will briefly discuss these elements after examining the linear

plane element formulation.
Step 1 Select Element Type

The natural s-t coordinates are attached to the element, with
the origin at the center of the element.
The s and t axes need not be

orthogonal, and neither has to be
parallel to the x or y axis.
The orientation of s-t coordinates is

such that the four corner nodes and
the edges of the quadrilateral are
bounded by +1 or -1

The natural s-t coordinates are attached to the element, with
the origin at the center of the element.
This orientation will later allow us to

take advantage more fully of
common numerical integration
schemes.

Consider the quadrilateral to have eight degrees of freedom u1,
v1, ... , u4, and v4 associated with the global x and y directions.
The element then has straight sides but is otherwise of
arbitrary shape.

For the special case when the distorted element becomes a
rectangular element with sides parallel to the global x-y
coordinates, the s-t coordinates can be related to the global
element coordinates x and y by
x  xc  bs y  y c  ht
where xc and yc are the global coordinates of the element

centroid.

Assuming global coordinates x and y are related to the natural
coordinates s and t as follows:
x  a1  a2s  a3t  a4st y  a5  a6s  a7t  a8st
Solving for the a's in terms of x1, x2, x3, x4, y1, y2, y3, y4, we
obtain
1
x 1  s 1  t  x1  1  s 1  t  x2  1  s 1  t  x3  1  s 1  t  x4 
4
1
y 1  s 1  t  y1  1  s 1  t  y 2  1  s 1  t  y 3  1  s 1  t  y 4 
4
Step 1 Select Element Type  x1 

In matrix form: y 
 1
 x2 
 
 x  N1 0 N2 0 N3 0 N4 0  y 2 
   
 y   0 N1 0 N2 0 N3 0 N4   x3 
y3 
where:  
 x4 
N1 
1  s 1  t  N2 
1  s 1  t   y 4 
4 4
N3 
1  s 1  t  N4 
1  s 1  t 
4 4
N1 t N2 t
4 3 4 3
s s
1 2 1 2
N3
t N4
t
4 3 4 3
s s
1 2 1 2


These shape functions are seen to map the s and t coordinates
of any point in the square element to those x and y
coordinates in the quadrilateral element.
P (s, t )

Consider square element node 1 coordinates, where s = -1 and
t = -1 then x = x1 and y = y1.

Other local nodal coordinates at nodes 2, 3, and 4 on the
square element in s-t isoparametric coordinates are mapped
into a quadrilateral element in global coordinates x2, y2 through
x4, y4 .

Also observe the property that N1 + N2 + N3 + N4 = 1 for all
values of s and t.

We have always developed the element interpolation functions
either by assuming some relationship between the natural and
global coordinates in terms of the generalized coordinates a's
or, similarly, by assuming a displacement function in terms of
the a's.
However, physical intuition can often guide us in directly
expressing shape functions based on the following two criteria
set forth in Section 3.2 and used on numerous occasions:
n
N
i 1
i 1 i  1, 2, ,...,n
Step 2 Select of Displacement Functions

The displacement functions within an element are now similarly
defined by the same shape functions as are used to define the
element geometric shape:
 u1 
v 
 1
u2 
 
u  N1 0 N2 0 N3 0 N4 0  v 2 
   
v   0 N1 0 N2 0 N3 0 N4  u3 
v 3 
 
u4 
v 4 
Step 3 Strain-Displacement and Stress-Strain Relationships

We now want to formulate element matrix [B] to evaluate [k].
However, because it becomes tedious and difficult (if not

impossible) to write the shape functions in terms of the x and
y coordinates, as seen in Chapter 8, we will carry out the
formulation in terms of the isoparametric coordinates s and t.
This may appear tedious, but it is easier to use the s- and t-

coordinate expressions.
This approach also leads to a simple computer program

formulation.

To construct an element stiffness matrix, we must determine the
strains, which are defined in terms of the derivatives of the
displacements with respect to the x and y coordinates.
The displacements, however, are now functions of the s and t

coordinates.
The derivatives u/x and v/y are now expressed in terms of

s and t.
Therefore, we need to apply the chain rule of differentiation.


The chain rule yields:
f f x f y f f x f y
   
s x s y s t x t y t
The strains can then be found; for example, x = u/x

Consider Cramer’s rule for small systems:
 a1 b1   x   c1 
a   
 2 b2   y  c2 
c1 b1 a1 c1
c2 b2 a2 c2
x y
a1 b1 a1 b1
a2 b2 a2 b2

Using Cramer’s rule, which involves the determinants of
matrices, we can obtain:
f y x f
s s s s
f y x f
f f
 t t  t t
x x y y x y
s s s s
x y x y
t t t t

The determinant in the denominator is the determinant of the
Jacobian matrix [J].
x y
s s
J 
x y
t t
We now want to express the element strains as:    B d 
Where [B] must now be expressed as a function of s and t.


The usual relationship between strains and displacements given
in matrix form as:
 u     
   0 
 x  x 
 x   
  
{ }    y   
v       u 
  0  
   y   y  v 
 xy   u v       
    
 y x   y x 
Where the rectangular matrix on the right side is an operator
matrix; that is, ( )/x and ( )/y represent the partial
derivatives of any variable we put inside the parentheses.

Evaluating the determinant in the numerators, we have
  1  y    y    
  
x J   t  s s t 
  1  x    x    
  
y J   s t t s 
Where |[J]| is the determinant of [J].


We can obtain the strains expressed in terms of the natural
coordinates (s-t) as:
 y    y    
 0 
 x   t s s t 
  1  x    x     u 
 y    0  
   J   s t t s  v 
 xy   x    x    y    y  
  
 s t t s t s s t 

We can express the previous equation in terms of the shape
functions and global coordinates in compact matrix form as:
   D 'N d 
 y    y    
 0 
 t s s t 
1  x    x  
D '   0 
J   s t t s 
 x    x    y    y  
  
 s t t s t s s t 

The shape function matrix [N] is the 2 x 8 {d} is the column
matrix.
B   D ' N 
38 32 28
The matrix multiplications yield
1
B  s, t    B  B2  B3  B4 
J   1
 a  Ni ,s   b  Ni ,t  0 
 
Bi    0 c  Ni ,t   d  Ni ,s  
 
c  Ni ,t   d  Ni ,s  a  Ni ,s   b  Ni ,t  

Here i is a dummy variable equal to 1, 2, 3, and 4, and
1
a  y1  s  1  y 2  s  1  y 3 1  s   y 4 1  s  
4
1
b  y1  t  1  y 2 1  t   y 3 1  t   y 4  1  t  
4
1
c  x1  t  1  x2 1  t   x3 1  t   x 4  1  t  
4
1
d  x1  s  1  x2  s  1  x3 1  s   x 4 1  s  
4

Using the shape functions, we have
1 1
N1,s   t  1 N1,t   s  1
4 4
where the comma followed by the variable s or t indicates

differentiation with respect to that variable; that is,
N1,s = N1/s and so on.

The determinant |[J]| is a polynomial in s and t and is tedious to
evaluate even for the simplest case of the linear plane
quadrilateral element.
However, we can evaluate |[J]| as
 0 1 t t  s s 1 
t 1 0 s  1 s  t 
1
J   8  X c    Yc 
T
 s  t s  1 0 t 1 
 
1  s s  t t  1 0 
 Xc    x1 x2 x4  Yc    y1 y 2 y4 
T T
x3 y3

We observe that |[J]| is a function of s and t and the known
global coordinates x1, x2, ... , y4.
Hence, [B] is a function of s and t in both the numerator and the

denominator and of the known global coordinates x1 through
y4.
The stress-strain relationship is a function of s and t.
   D B d 
Step 4 Derive the Element Stiffness Matrix and Equations

We now want to express the stiffness matrix in terms of s-t
coordinates.
For an element with a constant thickness h, we have
[k ]    [B ]T [D ][B ] h dx dy
A
However, [B] is now a function of s and t, we must integrate with

respect to s and t.

Once again, to transform the variables and the region from x
and y to s and t, we must have a standard procedure that
involves the determinant of [J].
1 1
  f ( x, y ) dx dy
A
   f (s, t ) J  ds dt
1 1
where the inclusion of |[J]| in the integrand on the right side of

equation results from a theorem of integral calculus.

We also observe that the Jacobian (the determinant of the
Jacobian matrix) relates an element area (dx dy) in the global
coordinate system to an elemental area (ds dt) in the natural
coordinate system.
For rectangles and parallelograms, J is the constant value J =

A/4, where A represents the physical surface area of the
element.
1 1
[k ]    [B ]
T
[D][B ] h  J  ds dt
1 1

The |[J]| and [B] are complicated expressions within the integral.
Integration to determine the element stiffness matrix is usually

done numerically.
The stiffness matrix is of the order 8 x 8.
1 1
[k ]    [B ]
T
[D][B ] h  J  ds dt
1 1

Body Forces - The element body-force matrix will now be
determined from
1 1
fb     [N ]T  X b  h  J  ds dt
1 1
 8  1  8  2   2  1
Like the stiffness matrix, the body-force matrix has to be
evaluated by numerical integration.

Surface Forces - The surface-force matrix, say, along edge t =
1 with overall length L, is
1
L
fs    [Ns ]T T  h ds
2 1
 4  1  4  2   2  1
fs 3s 
  T
 fs 3t  hL N3 0  ps 
1
0 N4
   N4 
  ds
fs 4s  2 1  0 N3 0
along t 1
 pt 
 fs 4t 

Surface Forces - For the case of uniform (constant) ps, and pt,
along edge t = 1, the total surface-force matrix is
1
L
fs    [Ns ]T T  h ds
2 1
 4  1  4  2   2  1
fs 3s   ps 
f   
 s 3t  hL  pt 
   
fs 4s  2  ps 
 fs 4t   pt 
Steps 5 - 7
Steps 5 through 7, which involve assembling the global stiffness

matrix and equations, determining the unknown nodal
displacements, and calculating the stress, are identical to
those in presented in previous chapters.
Example 1
For the four-noded linear plane quadrilateral element shown

below with a uniform surface traction along side 2-3, evaluate
the force matrix by using the energy equivalent nodal forces.
Let the thickness of the element be h = 0.1 in.
1
hL
fs    [Ns ]T T  ds
2 1
Example 1
5  8   4  0  5
2 2
With length of side 2-3 given by: L 
fs 2s 
  T
 fs 2t  hL N2 0  ps 
1
0 N3
    N3 
  dt
fs 3s  2 1  0 N2 0
along s 1
 pt 
 fs 3t 
Example 1
Substituting for L, the surface traction matrix, and the thickness
h = 0.1 we obtain
fs 2s 
 
 fs 2t   0.1in. 5in. N2
T
1
0 N3 0 2,000 
 
2  0
1 
0 N3 

 0 
 dt
fs 3s  N2
along s 1
 fs 3t 
Example 1
Simplifying gives:
fs 2s  2,000 N2  N2 
f  1 
0  0 1
 s 2t 
  0.25in.2 
  dt  500lb.    dt
f
 s 3s  1
 2,000 N 3 1
 N3 
 fs 3t     
 0  0
evaluated along s  1 evaluated along s  1
Example 1
Substituting the shape functions, we have
 s  t  st  1
fs 2s   4  1  t 
f    1 
 s 2t 
1
 0  0 
   500lb.   dt  250lb.    dt
f s  t  st  1  t  1
 s 3s  1
  1
 fs 3t  4  
   0 
 0 
evaluated along s  1
Example 1
Performing the integration gives:
fs 2s  1  t   1 500 
f  1   0   0 
 s 2t  0
   250lb.    dt  500lb.     lb.
f
 s 3s  1
 t  1  1 500 
 fs 3t       
 0  0   0 
Newton-Cotes and Gaussian Quadrature
In this section, we will describe two methods for numerical
evaluation of definite integrals, because it has proven most
useful for finite element work.
The Newton-Cotes methods for one and two intervals of

integration are the well known trapezoid and Simpson's one-
third rule, respectively.
We will then describe Gauss' method for numerical evaluation of

definite integrals.
After describing both methods, we will then understand why the

Gaussian quadrature method is used in finite element work.
The Newton-Cotes method is a common technique for
evaluation of definite integrals.
1
To evaluate the integral    y dx
1
we assume the sampling points of y(x) are spaced at equal

intervals.
Since the limits of integration are from -1 to 1 using the

isoparametric formulation, the Newton-Cotes formula is given
by
1 n
   y dx  h Ci y i  h C0 y 0  C1y1  C2 y 2    Cn y n 
1 i 0
The constants Ci are the Newton-Cotes constants for numerical
integration with i intervals.
The number of intervals will be one less than the number of

sampling points, n.
The term h is the interval between the limits of integration (for

limits of integration between -1 and 1 this makes h = 2).
1 n
1 i 0
The Newton-Cotes constants have been published and are
summarized in the table below for i = 1 to 6.
Intervals, No. of
i Points, n C0 C1 C2 C3 C4 C5 C6
1 2 1/2 1/2 (trapezoid rule)
2 3 1/6 4/6 1/6 (Simpson's 1/3 rule)
3 4 1/8 3/8 3/8 1/8 (Simpson's 3/8 rule)
4 5 7/90 32/90 12/90 32/90 7/90
5 6 19/288 75/288 50/288 50/288 75/288 19/288
6 7 41/840 216/840 27/840 272/840 27/840 216/840 41/840
1 n
1 i 0
The case i = 1 corresponds to the well known trapezoid rule
illustrated below.
1 n
y y 
   y dx  h Ci y i  2hC00y 0 1C1y1yC
0  y    Cn y n 
2 y 21
1 i  0  2 
The case i = 2 corresponds to the well-known Simpson one-
third rule.
It has been shown that the formulas for i = 3 and i = 5 have the
same accuracy as the formulas for i = 2 and i = 4, respectively.
Therefore, it is recommended that the even formulas with i = 2

and i = 4 be used in practice.
1 n
1 i 0
To obtain greater accuracy one can then use a smaller interval
(include more evaluations of the function to be integrated).
This can be accomplished by using a higher-order Newton-

Cotes formula, thus increasing the number of intervals i.
It has been shown that we need to use n equally spaced

sampling points to integrate exactly a polynomial of order at
most n - 1.
1 n
1 i 0
On the other hand, using Gaussian quadrature we will show that
we use unequally spaced sampling points n and integrate
exactly a polynomial of order at most 2n - 1.
For instance, using the Newton-Cotes formula with n = 2

sampling points, the highest order polynomial we can integrate
exactly is a linear one.
However, using Gaussian quadrature, we can integrate a cubic

polynomial exactly.
Gaussian quadrature is then more accurate with fewer sampling
points than Newton-Cotes quadrature
This is because Gaussian quadrature is based on optimizing the

position of the sampling points (not making them equally
spaced as in the Newton-Cotes method) and also optimizing
the weights Wi (see the table below).
Order N Points ui Weights wi

1 n
   y dx   Wi y ( xi ) 1
2
0.000000000
±0.577350269
2.00000000
1.00000000
1 i 0 3 0.000000000 0.88888889
±0.774596669 0.55555556
4 ±0.339981044 0.65214515
±0.861136312 0.34785485
Newton-Cotes Example
Using the Newton-Cotes method with i = 2 intervals (n = 3
sampling points), evaluate the integrals:
1 1
  x 
    x 2  cos    dx    3 x  x  dx
1   2  1
Using three sampling points means we evaluate the function

inside the integrand at x = -1, x = 0, and x = 1, and multiply
each evaluated function by the respective Newton-Cotes
numbers.
1 n
   y dx  h Ci y i  2  61 y 0  64 y1  61 y 2 
1 i 0
1 1
  x 
    x 2  cos    dx    3 x  x  dx
1   2  1
1
  x 
    x 2  cos    dx  2  61 1.8775826   64 1  61 1.8775826  
1   2 
 2.5850550 0.027% error
1
  x 
    x 2  cos    dx  2.5843688
1   2 
1 1
  x 
    x 2  cos    dx    3 x  x  dx
1   2  1
1
   3 x  x  dx  2  61 1.3333333   4
6 1  61  2 
1
 2.4444444 0.706% error
1
   3 x  x  dx  2.427305
1
Gaussian Quadrature
1

To evaluate the integral:   y dx
1
where y = y(x), we might choose (sample or evaluate) y at the

midpoint y(0) = y1 and multiply by the length of the interval, as
shown below to arrive at I = 2y1, a result that is exact if the
curve happens to be a straight line.
Gaussian Quadrature
Generalization of the formula leads to:
1 n
   y dx   Wi y  xi 
1 i 1
That is, to approximate the integral, we evaluate the function at

several sampling points n, multiply each value yi by the
appropriate weight Wi , and add the terms.
Gauss's method chooses the sampling points so that for a given

number of points, the best possible accuracy is obtained.
Sampling points are located symmetrically with respect to the

center of the interval.
Gaussian Quadrature
1 n
   y dx   Wi y  xi 
1 i 1
In general, Gaussian quadrature using n points (Gauss points)

is exact if the integrand is a polynomial of degree 2n - 1 or
less.
In using n points, we effectively replace the given function y =

f(x) by a polynomial of degree 2n- 1.
The accuracy of the numerical integration depends on how well

the polynomial fits the given curve.
Gaussian Quadrature
1 n
   y dx   Wi y  xi 
1 i 1
If the function f(x) is not a polynomial, Gaussian quadrature is

inexact, but it becomes more accurate as more Gauss points
are used.
Also, it is important to understand that the ratio of two

polynomials is, in general, not a polynomial; therefore,
Gaussian quadrature will not yield exact integration of the
ratio.
Gaussian Quadrature - Two-Point Formula
To illustrate the derivation of a two-point (n = 2) consider:
1
   y dx  W1y  x1   W2 y  x2 
1
There are four unknown parameters to determine: W1, W2, x1,

and x2.
Therefore, we assume a cubic function for y as follows:
y  C0  C1x  C2 x 2  C3 x 3
In general, with four parameters in the two-point formula, we
would expect the Gauss formula to exactly predict the area
under the curve.
1
2
A  C
1
0 
 C1x  C2 x 2  C3 x 3 dx  2C0  C2
3
However, we will assume, based on Gauss's method, that
W1 = W2 and that x1 = x2 as we use two symmetrically located
Gauss points at x = ±a with equal weights.
The area predicted by Gauss's formula is

AG  W y ( a )  W y (a )  2W C0  C2a 2 
If the error, e = A - AG, is to vanish for any C0 and C2, we must
have, in the error expression:
 
e  A  AG   2C0  32 C2   C0  C2a 2 2W
e
 0  2  2W  W 1
C0
e 2 1  0.5773....
 0   2a 2W  a
C2 3 3
Now W = 1 and a = 0.5773 ... are the Wi’s and ai’s (xi’s) for the
two-point Gaussian quadrature as given in the table.
Gaussian Quadrature Example
Use three-point Gaussian Quadrature evaluate the integrals:
1 1
  x 
    x 2  cos    dx    3 x  x  dx
1   2  1
3
 x  Order N Points ui Weights wi
   Wi  xi 2  cos  i  1 0.000000000 2.00000000
i 1  2  2 ±0.577350269 1.00000000
3 0.000000000 0.88888889
5
 1.5259328  ±0.774596669 0.55555556
9 4 ±0.339981044 0.65214515
0.34785485
8 ±0.861136312
 1.0 
9
5
 1.5259328   2.5843698 0.00004% error
9
Use three-point Gaussian Quadrature evaluate the integrals:
1 1
  x 
    x 2  cos    dx    3 x  x  dx
1   2  1
3 Order N Points ui Weights wi

   Wi 3 xi  xi  1 0.000000000 2.00000000
i 1 2 ±0.577350269 1.00000000
3 0.000000000 0.88888889
5
 1.2015923  ±0.774596669 0.55555556
9 4 ±0.339981044 0.65214515
0.34785485
8 ±0.861136312
 1.0 
9
5
 1.5673475   2.4271888 0.00477% error
9
In two dimensions, we obtain the quadrature formula by
integrating first with respect to one coordinate and then with
respect to the other as
1 1 1
n 
 
1
1 f ( s, t ) ds dt  1 
i 1
Wi f  si , t   dt

n 
  W j   Wi f  si , t j  
n
j 1  i 1 
  Wi W j f  si , t j 
n n
i 1 j 1
For example, a four-point Gauss rule (often described as a 2 x 2
rule) is shown below with i = 1, 2 and j = 1, 2 yields
2 2
   Wi W j f  si , t j   W1 W1 f  s1, t1   W1 W2 f  s1, t 2 
i 1 j 1
 W2 W1 f  s2 , t1   W2 W2 f  s2 , t 2 
The four sampling points are

at si and ti = ±0.5773... and
Wi = 1.0
In three dimensions, we obtain the quadrature formula by
integrating first with respect to one coordinate and then with
respect to the other two as
1 1 1
 f (s, t, z ) ds dt dz  Wi W j Wk f  si , t j , zk 
n n n
  
1 1 1 i 1 j 1 k 1
Evaluation of the Stiffness Matrix by Gaussian Quadrature
For the two-dimensional element, we have shown in previous
chapters that
[k ]    [B ]T [D][B ] h dx dy
A
where, in general, the integrand is a function of x and y and

nodal coordinate values.
We have shown that [k] for a quadrilateral element can be
evaluated in terms of a local set of coordinates s-t, with limits
from -1 to 1within the element.
1 1
[k ]    [B ]
T
[D][B ] h  J  ds dt
1 1
Each coefficient of the integrand [B]T [D] [B] |[J]| evaluated by

numerical integration in the same manner as f(s, t) was
integrated.
A flowchart to evaluate [k] for an element using four-point
Gaussian quadrature is shown here.
The explicit form for four-point Gaussian quadrature (now using
the single summation notation with i = 1, 2, 3, 4), we have
1 1
[k ]    [B ]
T
[D][B ] h  J  ds dt
1 1
 B  s1, t1   D  B  s1, t1   J  s1, t1   W1 W1

T
 B  s2 , t 2   D  B  s2 , t 2   J  s2 , t 2   W2 W2

T
 B  s3 , t3   D  B  s3 , t3   J  s3 , t3   W3 W3

T
 B  s4 , t 4   D  B  s4 , t 4   J  s4 , t 4   W4 W4

T
where s1=t1= -0.5773, s2=-0.5773, t2=0.5773, s3=0.5773,

t3=-0.5773, and s4=t4=0.5773 and W1=W2=W3=W4=1.0
Evaluate the stiffness matrix for the quadrilateral element shown
below using the four-point Gaussian quadrature rule.
Let E = 30 x 106 psi and = 0.25. The global coordinates are

shown in inches. Assume h = 1 in.
Using the four-point rule, the four points are:
 s1, t1    0.5773,  0.5773 

 s2 , t2    0.5773, 0.5773 
 s3 , t3    0.5773,  0.5773 
 s4 , t 4    0.5773, 0.5773 
With W1 = W2 = W3 = W4 = 1.0
k   B  s1, t1  D  B  s1, t1  J  s1, t1 

T
 B  s2 , t2   D  B  s2 , t 2   J  s2 , t 2  

T
 B  s3 , t3   D  B  s3 , t3   J  s3 , t3  

T
 B  s4 , t 4   D  B  s4 , t 4   J  s4 , t 4  

T
First evaluate |[J]| at each Gauss, for example:
J  0.5773,  0.5773  

Recall:
 0 1 t t  s s 1 
t 1 0 s  1 s  t 
1
J   8  X c    Yc 
T
 s  t s  1 0 t 1 
 
1  s s  t t  1 0 
 Xc    x1 x2 x4  Yc    y1 y 2 y4 
T T
x3 y3
For this example:
 Xc   3 5 5 3  Yc    2 2 4 4
T T
Recall:
J  0.5773,  0.5773  

 0 1   0.5773   0.5773    0.5773   0.5773   1  2 
 
1  0.5773   1 0  0.5773   1   0.5773    0.5773   2 
 3 3 
 4 
5 5
8  0.5773    0.5773    0.5773   1 0  0.5773   1
  
 1   0.5773   0.5773    0.5773    0.5773   1 0  
 4 
 1.000
Similarly: J  0.5773,0.5773    1.000
J  0.5773,  0.5773    1.000
J  0.5773,0.5773    1.000

To evaluate [B] consider:
B  0.5773,  0.5773  
1
 B1  B2  B3  B4 
 J  0.5773,  0.5773  
where
 a  Ni ,s   b  Ni ,t  0 
 
Bi    0 c  Ni ,t   d  Ni ,s  
 
c  Ni ,t   d  Ni ,s  a  Ni ,s   b  Ni ,t  
For this example:
1
a  y1  s  1  y 2  s  1  y 3 1  s   y 4 1  s  
4
1
 2   0.5773   1  2    0.5773   1
4
 4 1   0.5773    4 1   0.5773   
 1.000
Similar computations are used to obtain b, c, and d.

The shape functions are computed as:
1 1
N1,s   t  1    0.5773   1  0.3943
4 4
1 1
N1,t   s  1    0.5773   1  0.3943
4 4
Similarly, [B2], [B3], and [B4] must be evaluated like [B1] at

(-0.5773, -0.5773).
We then repeat the calculations to evaluate [B] at the other

Gauss points.
Using a computer program written specifically to evaluate [B], at
each Gauss point and then [k], we obtain the final form of
[B(-0.5773, -0.5773)], as
B  0.5773,  0.5773  
 0.1057 0 0.1057 0 0 0.1057 0 0.3943 

  0.1057 0.1057 0.3943 0.1057 0.3943 0 0.3943 0 
 
 0 0.3943 0 0.1057 0.3943 0.3943 0.1057 0.3943 
With similar expressions for [B(-0.5773, 0.5773)], and so on.

The matrix [D] is:
1  0  32 8 0 
E    8 32 0   106 psi
[D ]   1 0 
1 2    
0 0 0.5 1      0 0 12
Finally, [k] is:

1466 500 866 99 733 500 133 99 
 500 1466 99 133 500 733 99 866 
 
 866 99 1466 500 133 99 733 500 
 
99 133 500 1466 99 866 500 733 
[k ]  10 4 
 733 500 133 99 1466 500 866 99 
 
 500 733 99 866 500 1466 99 133 
 133 99 733 500 866 99 1466 500 
 
 99 866 500 733 99 133 500 1466 
Evaluation of Element Stresses
The stresses are not constant within the quadrilateral element.

   D B d 
In practice, the stresses are evaluated at the same Gauss
points used to evaluate the stiffness matrix [k].
The common method used in computer programs is to evaluate

the stresses in all elements at a shared node and then use an
average of these element nodal stresses to represent the
stress at the node.
The stresses are not constant within the quadrilateral element.

   D B d 
Stress plots obtained in these programs are based on this
average nodal method.
The following example illustrates the use of Gaussian

quadrature to evaluate the stress matrix at the s = 0, t = 0
locations of the element.
For the rectangular element shown below, assume plane stress

conditions with E = 30 X 106 psi ,  = 0.3, and displacements
u1 = 0, v1 = 0, u2 = 0.001 in., v2 = 0.0015 in., u3 = 0.003 in.,
v3 = 0.0016 in., u4 = 0, and v4 = 0.
Evaluate the stresses, x, x, and xy at s = 0, t = 0.
First, evaluate [B] at s = 0, t = 0.

1
B  B  B2  B3  B4 
J   1
1
  B1  0,0   B2  0,0   B3  0,0   B4  0,0   
 J  0,0    
For the rectangular element shown below, assume plane stress

conditions with E = 30 X 106 psi ,  = 0.3, and displacements
u1 = 0, v1 = 0, u2 = 0.001 in., v2 = 0.0015 in., u3 = 0.003 in.,
v3 = 0.0016 in., u4 = 0, and v4 = 0.
 0 1 0 1 2 
 1 0 1 0   2 
1
J  0,0    3 5 5 3    
8  0 1 0 1 4 
 
 1 0 1 0  4 
J  0,0    1
Recall, [Bi] is:

 a  Ni ,s   b  Ni ,t  0 
 
Bi    0 c  Ni ,t   d  Ni ,s  
 
c  Ni ,t   d  Ni ,s  a  Ni ,s   b  Ni ,t  
with: a  1 b  0 c 1 d  0
Differentiating the shape functions with respect to s and t and

then evaluating at s = 0, t = 0, we obtain:
1 1 1 1
N1,s   N2,s  N3,s  N4,s  
4 4 4 4
1 1 1 1
N1,t   N2,t   N3,t  N4,t 
4 4 4 4
Therefore [B] is:

  41 0  41 0  41 0    41 0
B1    0  41 

B2    0  41 
 B3   0 41  B4    0 1 
4 
  41  41    41 4 
1
  41 41   41  41 
The element stress matrix {} is then obtained by substituting

[B] and the plane stress [D] matrix into the definition as:
   D B d 
   D B d 
 1 0.3 0 
30  106 
 0.3 1 0 
1  0.09    0 
 0 0 0.35   0 
 
 0.001 
 0.25 0 0.25 0 0.25 0 0.25 0  
0.0015 
 0 0.25 0 0.25 0 0.25 0 0.25  
  0.003 
 
 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25  0.0016 
 
 0 
 3.321  0 
   1.071104 psi
1.417 
 
Higher-Order Shape Functions – Linear Strain Bar
In general, higher-order element shape functions can be

developed by adding additional nodes to the sides of the linear
element.
This results in higher-order strain variations and convergence
occurs at a faster rate using fewer elements.
The trade-off is that there is a substantial increase in required
computational power.
Another advantage of higher-order elements is that curved
boundaries of irregularly-shaped bodies can be approximated
more closely than simple straight-sided linear elements.
We have been working with the linear strain bar element

throughout the text.
The linear strain bar (also called a quadratic isoparametric bar

element) shown below has three coordinates of nodes in the
global coordinates.
For the three-noded linear strain bar isoparametric element we

will determine the shape functions, N1, N2, and N3, and the
strain-displacement matrix [B].
Assume the general axial displacement function to be a

quadratic:
x  a1  a2s  a3s 2
Evaluating the a’s in terms of the nodal coordinates, we obtain

x  1  x1  a1  a2  a3
x  0   x3  a1
x 1  x2  a1  a2  a3
Substituting the values for a1, a2, and a3 into the general
equation for x, we obtain
 x  x1   x  x2  2 x3  2
x  a1  a2s  a3s 2  x3   2  s 1 s
 2   2 
Combining like terms gives:

 s  s  1   s  s  1 
x 
2
 x1  
2
 
 x2  1  s x3
2
   
 x1  x 
   s  s  1 s  s  1  1
x  N1 N2 N3   x 2    1  s   x2 
2
   2 2  x 
 x3   3
Therefore the shape functions
s  s  1 s  s  1
N1  N2  N3  1  s 2
2 2
Now determine the strain-displacement matrix [B] as:

 u1 
 
 B  u2 
du du ds
 
dx ds dx  
u3 
Using an isoparametric formulation the displacement function is:
u2 u u u 2u
u  u3  s  1 s  1 s2  2 s2  3 s2
2 2 2 2 2
du u2 u1  1  1
   u1s  u2s  2u3s   s   u1   s   u2   2s  u3
ds 2 2  2  2
dx L
We have previously showed that:   J 
ds 2
This relationship holds for the higher-order one-dimensional
elements as well as for the two-noded constant strain bar
element as long as node 3 is at the geometry center of the
bar.
Using this relationship gives:

du du ds 2  1  1 
   s   u1   s   u2   2s  u3 
dx ds dx L  2  2 
 2s  1   2s  1   4s 
  u1    u2    u3
 L   L   L 
In matrix form:
 u1 
du  2s  1 2s  1 4s   
   u2 
dx  L L L  
u3 
The axial strain becomes:
 u1   u1 
du  2s  1 2s  1 4s     
x     u2   B  u2 
dx  L L L   u 
u3   3
Where the gradient matrix [B] is:
2s  1 2s  1
B   
4s 
 
 L L L 
Substituting the expression for [B] in the stiffness matrix, we

obtain
1
L
k   2  B  E B  A ds
T
1
  2s  1  2s  1 2s  1  4s  2s  1 
2
 
 L2 L2 L2 
AEL   2s  1 2s  1  2s  1  4s  2s  1  ds
1 2
2 1 
 
L2 L2 L2 

  4s  2s  1  4s  2s  1  2s  1 
2
 L2 L2 L2 
 
Substituting the expression for [B] in the stiffness matrix, we

obtain
1
L
k   2  B  E B  A ds
T
1
 4s 2  4s  1
1
4s 2  1 8s 2  4s 
AE  
 
2L 1  4s 2  1 4s 2  4s  1 8s 2  4s  ds
 8s 2  4s 8s 2  4s 16s 2 

1
4 3 4 3 8 
 3 s  2s  s s s  s 3  2s 2 
2
3 3
   4.67 0.667 5.33 
AE  4 3 4 3 8
 s s s  2s 2  s  s 3  2s 2   AE 0.667 4.67 5.33 
2L  3 3 3  2L  
 8 8 16 3   5.33 5.33 10.67 
  s 3  2s 2  s 3  2s 2 s
 3 3 3  1
Now let’s illustrate how to evaluate the stiffness matrix for the
three-noded bar element using two-point Gaussian
quadrature.
We will compare the results to that obtained by the explicit

integration performed.
Recall the stiffness matrix for this element is:

1
L
k     B  E B  A ds
T
2 1
 4s 2  4s  1
1
4s 2  1 8s 2  4s 
AE  
 
2L 1  4s 2  1 4s 2  4s  1 8s 2  4s  ds
 8s 2  4s 8s 2  4 s 16s 2 

Using two-point Gaussian quadrature, we evaluate the stiffness

matrix at two points:
1 1
s1    0.57735 s2   0.57735
3 3
w1  1 w2  1
We then evaluate each term in the integrand at each Gauss

point and multiply each term by its weight (here weights are1).
We then add those Gauss point evaluations together to obtain

the final term for each element of the stiffness matrix.
For two-point evaluation, there will be two terms added together

to obtain each element of the stiffness matrix. We proceed to
evaluate the stiffness matrix term by term as follows:
We will evaluate the stiffness matrix term by term as follows:

2
k11    w i  2si  1   2  0.57735   1  2  0.57735   1
2 2 2
i 1
 4.6667
2
k12    w i  2si  1 2si  1
i 1
  2  0.57735   1  2  0.57735   1

2
  2  0.57735   1  2  0.57735   1

2
 0.6667

2
k13    w i  4si  2si  1   4  0.57735  2  0.57735   1
i 1
  4  0.57735    2  0.57735   1
 5.3333
2
k22    w i  2si  1   2  0.57735   1   2  0.57735   1
2 2 2
i 1
 4.6667

2
k23    w i  4si  2si  1   4  0.57735  2  0.57735   1
i 1
  4  0.57735    2  0.57735   1
 5.3333
2
k33    w i 16si   16  0.57735   16  0.57735 
2 2 2
i 1
 10.6667
By symmetry, the [k21] equals the [k21], etc. Therefore, from the
evaluations of the terms, the final stiffness matrix is
 4.67 0.667 5.33 

k   2L 0.667 4.67 5.33 
AE 
 5.33 5.33 10.67 
The results obtained from Gaussian quadrature are identical to

those obtained analytically by direct explicit integration of each
term in the stiffness matrix.
To further illustrate elements with improved physical behavior,

we start with the Q6 element, and then to further illustrate the
concept of higher-order elements, we will consider the
quadratic (Q8 and Q9) elements and cubic (Q12) element
shape functions.
We then compare results for a cantilever beam model meshed

with the numerous element types described in this and
previous chapters, such as the CST, Q4, Q6, Q8, and Q9
elements.
Higher-Order Shape Functions – Bilinear Quadratic (Q6)
An improved element to remove the shear locking inherent in

the Q4 element is to add two internal degrees of freedom per
displacement function (g1 –g4) to the Q4 element displacement
functions.
This element is then called a Q6 element.

4
  
u  s, t    Ni ui  g1 1  s 2  g 2 1  t 2 
i 1
4
  
v  s, t    Ni v i  g 3 1  s 2  g 4 1  t 2 
i 1
These are the shape functions derived for the

isoparametric Q4 element
The displacement field is enhanced by modes that describe the

state of constant curvature (also called bubble modes) that are
represented by g1 through g4.
These corrections allow the elements to curve between the

nodes and can then model bending with either s or t axis as
the neutral axis
The magnitude of these modes is determined by minimizing the

internal strain energy in the element.
The additional degrees of freedom are condensed out before

the element stiffness matrix is developed.
Hence, only the degrees of freedom associated with the four

corner nodes appear.
The element can model pure bending exactly if it is a

rectangular shape.
Because the g1 –g4 degrees of freedom are internal and not

nodal degrees of freedom, they are not connected to other
elements.
There is a possibility that the edges of two adjacent elements

may have different curvatures and thus the displacement field
along this common edge may be incompatible.
This incompatibility will occur under certain loading conditions,

such as shown:
Higher-Order Shape Functions – Quadratic Rectangle (Q8)
A quadratic isoparametric element with four corner nodes and

four additional mid-side nodes. This eight-noded element is
often called a Q8 element.
The shape functions of the quadratic element are based on the

incomplete cubic polynomial such that coordinates x and y
are:
x  a1  a2s  a3t  a4st  a5s 2  a6t 2  a7s 2t  a8st 2
y  a9  a10s  a11t  a12st  a13s 2  a14t 2  a15s 2t  a16st 2
These functions have been chosen so that the number of

generalized degrees of freedom (2 per node times 8 nodes
equals 16) are identical to the total number of a's.

are:
The literature also refers to this eight-noded element as a

"serendipity" element as it is based on an incomplete cubic,
but it yields good results in such cases as beam bending.

are:
We are also reminded that because we are considering an

isoparametric formulation, displacements u and v are of
identical form as x and y, respectively.
To describe the shape functions, two forms are required: one for
corner nodes and one for mid-side nodes. The four corner
nodes are:
1
N1  1  s 1  t  s  t  1
4
1
N2  1  s 1  t  s  t  1
4
1
N3  1  s 1  t  s  t  1
4
1
N4  1  s 1  t  s  t  1
4
The four mid-side nodes are:
1
N5  1  s 1  t 1  s 
2
1
N6  1  s 1  t 1  t 
2
1
N7  1  s 1  t 1  s 
2
1
N8  1  s 1  t 1  t 
2
The displacement functions are given by:

U 1 
V 
1

U
2 
V 
2

u  N1 0 N2 0 N3 0 N4 0 N5 0 N6 0 N7 0 N8 0  U 
   
3
v   0 N1 0 N2 0 N3 0 N4 0 N5 0 N6 0 N7 0 N8  V
3 
U 
 
4
 
U 
 
8
V
8 
x 
du
 DN d  B   DN 
dx
Recall the [D’] operator is:
 y    y    
 0 
 t s s t 
1  x    x  
D '   0 
J   s t t s 
 x    x    y    y  
  
 s t t s t s s t 
x 
du
 DN d  B   DN 
dx
To evaluate the matrix [B] and the matrix [k] for the eight-noded
quadratic isoparametric element, we now use the nine-point
Gauss rule (often described as a 3 X 3 rule).
Results using 2 X 2 and 3 X 3 rules have shown significant

differences, and the 3 X 3 rule is recommended.
By adding a ninth node at s = 0, t = 0, we can create an element

called a Q9.
This is an internal node that is not connected to any other

nodes. We then add the a17s2t2 and a18s2t2 terms to x and y
equations, respectively, and to u and v.
The element is then called a Lagrange element as the shape

functions can be derived using Lagrange interpolation
formulas.
We now present a comparison of results for a cantilever beam

meshed with the various plane elements as described in this
and previous Chapters 6 and 8.
Below, the CST, Q4, Q6, Q8, and Q9 element mesh solutions
are compared to the classical beam element.
Note that the Q6 element (or Q4 incompatible) removes the

shear locking that occurs with the Q4 element and yields
excellent results for the displacement even with a single row of
rectangular elements.
However, small angles of trapezoidal distortion (say 15° from

the vertical) make the elements much too stiff.
Also parallel distortion reduces accuracy of the elements but to

a smaller amount than the trapezoidal distortion.
The Q8 and Q9 elements perform very well considering only
one row and two elements or fewer total degrees of freedom
(d.o.f) are used compared to the Q6 mesh.
The Q9 element with the additional internal node yields slightly
better single row results than the Q8
Higher-Order Shape Functions – Cubic Rectangle (Q12)
The cubic (Q12) element has four corner nodes and additional
nodes taken to be at one-third and two-thirds of the length
along each side.
The shape functions of the cubic element are based on the
incomplete quartic polynomial:
x  a1  a2s  a3t  a4s 2  a5st  a6t 2  a7 s 2t  a8st 2
 a9s 3  a10t 3  a11s 3t  a12st 3
For the corner nodes (i = 1, 2, 3, 4),
1 si   1, 1, 1,  1
Ni  1  ssi 1  tti  9  s 2  t 2   10 
32 t i   1,  1, 1, 1
For nodes on sides s = ± 1 (i = 7, 8, 11, 12),

9
Ni  1  ssi 1  9tti  1  t 2  si  1 t i   31
32
For nodes on sides t = ± 1 (i = 5, 6, 9, 10),
9
Ni 
32

1  9ssi 1  tti  1  s 2  si   31 t i  1
Having the shape functions for the Q9 quadratic element or for

the Q12 cubic element, we can obtain [B] and then set up [k]
for numerical integration for plane element.
The cubic element requires a 3 X 3 rule (nine points) to evaluate

the matrix exactly.
We then conclude that what is really desired is a library of

shape functions that c used in the general equations
developed for stiffness matrices, distributed load, and body
and can be applied not only to stress analysis but to
nonstructural problems as well.
Problems
19. Work problems 10.1, 10.6a, 10.8, 10.15dg, and 10.17b on

20. Write a computer program to evaluation of the [k] stiffness

matrix for the Q4 element by Gaussian quadrature. Check
your stiffness matrix values with the Example 10.4 in the
textbook. In addition, develop your code in such a way that
it could be easily extended to the Q8, Q9, and Q12
elements.
End of Chapter 10
CIVL 7/8117 Chapter 12 - Plate Bending Elements 1/34
Chapter 12 – Plate Bending Elements
Learning Objectives
• To introduce basic concepts of plate bending.
• To derive a common plate bending element
stiffness matrix.
• To present some plate element numerical
comparisons.
• To demonstrate some computer solutions for plate
bending problems.
Learning Objectives
stiffness matrix.
comparisons.
bending problems.
Learning Objectives
stiffness matrix.
comparisons.
bending problems.
Development of the Plate Bending Element

Introduction
In this section we will begin by describing elementary concepts
of plate bending behavior and theory.
The plate element is one of the more important structural

elements and is used to model and analyze such structures as
pressure vessels, chimney stacks, and automobile parts.
A large number of plate bending element formulations exist that

would require lengthy chapter to cover.

Introduction
The purpose in this chapter is to present the derivation of the
stiffness matrix for one of the most common plate bending
finite elements and then to compare solutions to some
classical problems for a variety of bending elements in the
literature.

Basic Concepts of Plate Bending
A plate can be considered the two-dimensional extension of a
beam in simple bending.
Both plates and beams support loads transverse or

perpendicular to their plane and through bending action.
A plate is flat (if it were curved, it would be a shell).
A beam has a single bending moment resistance, while a plate

resists bending about two axes and has a twisting moment.
We will consider the classical thin-plate theory or Kirchhoff

plate theory.

Basic Behavior of Geometry and Deformation
Consider the thin plate in the x-y plane of thickness t measured
in the z direction shown in the figure below:
The plate surfaces are at z = ±t/2, and its midsurface is at z = 0.
1.The plate thickness is much smaller than its inplane

dimensions b and c (that is, t << b or c)

If t is more than about one-tenth the span of the plate, then

transverse shear deformation must be accounted for and the
plate is then said to be thick.

2. The deflection w is much less than the thickness t

(than is, w/t << 1).

Kirchhoff Assumptions
Consider the differential slice cut from the plate by planes
perpendicular to the x axis as show in the figure below:
Loading q causes the plate to deform laterally or upward in the

z direction and, the defection w of point P is assumed to be
a function of x and y only; that is w = w(x, y) and the plate
does not stretch in the z direction.

The line a-b drawn perpendicular to the plate surface before

loading remains perpendicular to the surface after loading.

1. Normals remain normal. This implies that transverse shears

strains yz = 0 and xz = 0. However xy does not equal to zero.
Right angles in the plane of the plate may not remain right
angles after loading. The plate may twist in the plane.

2. Thickness changes can be neglected and normals undergo

no extension. This means that z = 0.

3. Normal stress z has no effect on in-plane strains x and y in

the stress-strain equations and is considered negligible.

4. Membrane or in-plane forces are neglected here, and the

plane stress resistance can be superimposed later (that is,
the constant-strain triangle behavior of Chapter 6 can be
superimposed with the basic plate bending element
resistance).

4. Therefore, the in-plane deflections in the x and y directions at

the midsurface, z = 0, are assumed to be zero; u(x, y, 0) = 0
and v(x, y, 0) = 0.

Based on Kirchhoff assumptions, at any point P the
displacement in the x direction due to a small rotation  is:
 w 
u  z  z  
 x 
At the same point, the displacement in the y direction is:
 w 
v  z  z  
 y 
The curvatures of the plate are then given as the rate of change
of the angular displacements of the normals and defined as:
 2w  2w 2 2w
x   y    xy  
x 2 y 2 xy

Using the definitions for in-plane strains, along with the
curvature relationships, the in-plane strain/displacement
equations are:
 2w  2w  2w
 x  z 2  y  z 2  xy  2z
x y xy
The first of the above equations is used in beam theory.
The remaining two equations are new to plate theory.


Stress/Strain Relationship
Based on the third Kirchhoff assumption, the plane stress
equations that relate in-plane stresses to in-plane strains for
an isotropic material are:
x 
E
1 2
 x   y 
y 
E
1 2
 y   x 
 xy  G xy

The in-plane normal stresses and shear stress are shown acting
on the edges of the plate shown in figure below:
Similar to the stress variation in a beam, the stresses vary

linearly in the z direction from the midsurface of the plate.

The in-plane normal stresses and shear stress are shown acting
on the edges of the plate shown in figure below:
The transverse shear stresses yz and xz are also present, even
though transverse shear deformation is neglected.
These stresses vary quadratically through the plate thickness.

The bending moments acting along the edge of the plate can be
related to the stresses by:
t /2 t /2 t /2
Mx  
 t /2
z x dz My  
 t /2
z y dz M xy  
 t /2
z xy dz

Substituting strains for stresses gives:
 E 
t /2
2  x
Mx   z    y   dz
 t /2 
1 
 E 
t /2
2  y
My   z    x   dz
 t /2 
1  
t /2
M xy  
 t /2
zG xy dz

Using the strain/curvature relationships, the moment expression
become:
D(1   )
M x  D  x   y  M y  D  y   x  M xy   xy
2
where D = Et3/[12(1 -  2)] is called the bending rigidity of the
plate.
The maximum magnitude of the normal stress on each edge of
the plate are located at the top or bottom at z = t/2.
6M x
For example, it can be shown that:  x 
t2

The equilibrium equations for plate bending are important in
selecting the element displacement fields.
The governing differential equations are:

Qx Qy
 q 0
x y
M x M xy
  Qx  0
x y
M y M xy
  Qy  0
y x
where q is the transverse distributed loading and Qx and Qy
are the transverse shear line loads.

The transverse distributed loading q and the transverse shear
line loads Qx and Qy are the shown below:
Qx Qy
 q 0
x y
M x M xy
  Qx  0
x y
M y M xy
  Qy  0
y x

Substituting the moment/curvature expressions in the last two
differential equations list above, solving for Qx and Qy, and
substituting the results into the first equation listed above, the
governing partial differential equation for isotropic, thin-plate
bending may be derived as:
  4w 2 4w  4w 
D 4  2 2  4   q
 x x y y 
where the solution to the thin-plate bending is a function of the

transverse displacement w.

Substituting the moment/curvature expressions in the last two
differential equations list above, solving for Qx and Qy, and
substituting the results into the first equation listed above, the
governing partial differential equation for isotropic, thin-plate
bending may be derived as:
  4w 2 4w  4w 
D 4  2 2  4   q
 x x y y 
If we neglect the differentiation with respect to the y direction,

the above equation simplifies to the equation for a beam and
the flexural rigidity D of the plate reduces to the EI of the beam
when the Poisson effect is set to zero.

Potential Energy of a Plate
The total potential energy of a plate is given as:
U
1
2 V
 x x   y  y   xy  xy  dV
The potential energy can be expressed in terms of moments
and curvatures as:
U
1
2 A
 Mx x  My  y  Mxy xy  dA

Derivation of a Plate Bending Element Stiffness
Numerous finite elements for plates bending have been
developed over the years, references cite 88 different
elements.
In this section, we will introduce the basic12-degree-of-freedom

rectangular element shown below.
The formulation will be developed consistently with the stiffness

matrix and equations for the bar, beam, plane stress/strain
elements of previous chapters.

Consider the 12-degree-of-freedom plate element shown in the
figure below.
Each node has 3 degrees of freedom – a transverse

displacement w in the z direction, a rotation x about the x
axis, and a rotation y about the y axis.

w i 
 
The nodal displacements at node i are: d    xi 
 
 yi 
where the rotations are related to the transverse
displacements by: w w
x  y  
y x
The negative sign on y is due to the fact that a negative

displacement w is required to produce a positive rotation
about the y axis.  di 
d 
 j
The total element displacement matrix is: d    
dm 
 d n 

Since the plate element has 12 degrees of freedom, we select a
12-term polynomial in x and y as:
w ( x, y )  a1  a2 x  a3 y  a4 x 2  a5 xy  a6 y 2  a7 x 3
 a8 x 2 y  a9 xy 2  a10 y 3  a11x 3 y  a12 xy 3
The function given above is an incomplete quartic polynomial;

however, it is complete up to the third order (first ten terms),
and the choice of the two more terms from the remaining five
terms of the complete quartic must be made.
The choice of x3y and y3x ensure that we will have continuity in
the displacement among the interelement boundaries.

 a8 x 2 y  a9 xy 2  a10 y 3  a11x 3 y  a12 xy 3
The terms x4 and y4 would yield discontinuities along the

interelement boundaries.
The final term x2y2 cannot be paired with any other term so it is
also rejected.

 a8 x 2 y  a9 xy 2  a10 y 3  a11x 3 y  a12 xy 3
The displacement function approximation also satisfies the

basic differential equation over the unloaded part of the plate.
In addition, the function accounts for rigid-body motion and

constant strain in the plate.
However, interelement slope discontinuities along common

boundaries of elements are not ensured.

To observe these discontinuities in slope, evaluate the
polynomial and its slopes along a side or edge.
For example, consider side i-j, the function gives:
w
w ( x, y )  a1  a2 x  a4 x 2  a7 x 3  a2  2a4 x  3a7 x 2
x
w
 a3  a5 x  a8 x 2  a11x 3
y
The displacement w is cubic while the slope w/x is the same
as in beam bending.
Based on the beam element, recall that the four constants a1,
a2, a4, and a7 can be defined by invoking the endpoint
conditions of wi, wj, yi, and yj.

w
x
w
 a3  a5 x  a8 x 2  a11x 3
y
Therefore, w and w/x are completely define along this edge.
The normal slope w/y is cubic in x: however; only two
degrees of freedom remain for definition of this slope while
four constant exist a3, a5, a8, and a11.

w
x
w
 a3  a5 x  a8 x 2  a11x 3
y
The normal slope w/y is not uniquely defined and a slope
discontinuity occurs.
The solution obtained from the finite element analysis using this
element will not be a minimum potential energy solution.
However, this element has proven to give acceptable results.

The constant a1 through a12 can be determined by expressing
the 12 simultaneous equations linking the values of w and its
slope at the nodes when the coordinates take their appropriate
values.
 
 w   a1 
   1 x y x 2 xy y2
x 3 2
x y xy y 2 3
x y 3
xy  a2 
3
 w     
  0 0 1 0 x2 2 xy 3 y 2 3
 x 2y 0 x 3 xy 2   a3 
 y  0 1 0 2x  y 0 3 x 2 2xy y 2 0 3 x 2 y  y 3    
 w    
  a12 
 x 
or in matrix form as:    P a
where [P] is the 3 x 12 first matrix on the right-hand side of the

above equation.

Next, evaluate the matrix at each node point
 w i   1 xi yi xi 2 xi y i yi 2 xi 3 xi 2 y i xi y i 2 yi 3 xi 3 y i xi y i 3   a1 
   2 2 3  
 xi  0 0 1 0 xi 2y i 0 xi 2 xi y i 3y i xi 3 xi y i 2  a2
 
 yi  0 1 0 2 xi y i 0 3 xi 2 2 xi y i y i 2 0 3 xi 2 y i  y i 3   a3 
  2 2 3 2 2 3  
w j   1 x y xj xjy j yj xj xj y j xjy j yj x j3y j x j y j 3   a4 
      
    
 yn  0 1 2 xn y n 3 xn2 2 xn y n y n2 3 xn 2 y n  y n  a12 
3
0 0 0
In compact matrix form the above equations are: d   C a
Therefore, the constants {a} can be solved for by: a  C  d 

1

Substituting the above expression into the general form of the
matrix gives:
   P  C  d
1
or    N  d
where [N] = [P][C]-1 is the shape function matrix.

Step 3 - Define the Strain (Curvature)/Displacement and
Stress (Moment)/Curvature Relationships
Recall the general form of the curvatures:
 2w  2w 2 2w
x   y    xy  
x 2 y 2 xy
The curvature matrix can be written as:

 x   2a4  6a7 x  2a8 y  6a11xy 
   
 y    2a6  2a9 x  6a10 y  6a12 xy 
  2a  4a x  4a y  6a x 2  6a y 2 
 xy   5 8 9 11 12 
or in matrix form as:    Q a


The [Q] matrix is the coefficient matrix multiplied by the a’s in
the curvature matrix equations.
 a1 
0 0 0 2 0 0 6 x 2y 0 0 6 xy 0   a2 
   
   0 0 0 0 0 2 0 0 2 x 6 y 0 6 xy   a3 
0 0 0 0 2 0
 0 4 x 4y 0 6 x 2 6 y 2    
 
a12 
Therefore:
   Q a     Q  C  d    B  d
1
or
where: B   Q  C 
1

The moment/curvature matrix for a plate is given by:
 Mx   x 
M   My   D   y   D  B d
 
M   
 xy   xy 
where the [D] matrix for isotropic materials is:
1  0 
Et 3  
[D ]   1 0

12 1   
2  
0 0 0.5 1    

The stiffness matrix is given by the usually form of the stiffness
matrix as:
[k ]   [B ]T [D][B ] dxdy
The stiffness matrix for the four-node rectangular element is of a

12 x 12.
The surface force due to distributed loading q acting per unit

area in the z direction is:
[Fs ]   [Ns ]T q dx dy

For a uniform load q acting over the surface of an element of
dimensions 2b x 2c the forces and moments at node i are:
 fwi  3
  qcb  
f xi   c 
f  3 b
  yi   
with similar expression at nodes j, m, and n.

The element equations are given by:
 fwi   k11 k12  k1,12   w i 
f   k k22  k 2,12
  
  xi   21   xi 
 f yi   k31 k32  k3,12   yi 
   
 fw j   k 41 k 42  k 4,12  w j 
          
    
f yn  k12,1 k12,2  k12,12   yn 
The remaining steps of assembling the global equations,

applying boundary conditions, and solving the equations for
nodal displacements and slopes follow the standard
procedures introduced in previous chapters.

Plate Element Numerical Comparisons
The figure to the right shows
a number of plate element
formulations results for a
square plate simply sup-
ported all around and sub-
jected to a concentrated
vertical load applied at the
center of the plate.

The results show the upper
and lower bound solutions
behavior and demonstrate
convergence of solution for
various plate elements.
Included in these results is

the 12-term polynomial
plate element introduced in
this chapter.

The figure on the right shows
comparisons of triangular plate
formulations for the same centrally
loaded simply supported plate.
From both figures, we can observe a

number of different formulations
with results that converge for
above and below.
Some of these elements produce

better results than others.

The figure below shows results for some selected Mindlin plate
theory elements.
Mindlin plate elements account for bending deformations and
for transverse shear deformation.

Computer Solution for a Plate Bending Problem
Consider the clamped plate show below subjected to a 100 lb
load applied at the center (let E = 30 x 106 psi and  = 0.3).
The exact solution for the displacement at the center of the

plate is w = 0.0056PL2/D.
Substituting the values for the variables gives a numerical value
of w = 0.0815 in.

The table below shows the results of modeling this plate
structure using SAP2000 (the educational version allows only
100 nodes) compares to the exact solution.
Number of Displacement % error
square elements at the center (in)
4 0.09100 11.6
16 0.09334 14.5
36 0.08819 8.2
64 0.08584 5.3
256 0.08300 1.8
1,024 0.08209 0.7
4,096 0.08182 0.3
Exact Solution 0.08154 --

The figures below show non-node-averaged contour plot for the
normal stress x and y.

The next set of plots shows the non-node-averaged moments
Mx and My.

The next set of plots shows the shear stress xy and the node-
average shear stress xy.

The next set of plots shows the twisting moment Mxy and the
node-average twisting moment Mxy.

Both sets of plots indicate interelement discontinuities for shear
stress and twisting moment.

However, if the node-average plots are viewed, the
discontinuities are smoothed out and not visible.

A C-channel section structural steel beam of 2-in. wide flanges,
3 in. depth and thickness of both flanges and web of 0.25 in. is
loaded as shown with 100 lb. acting in the y-direction on the
free end. Determine the free end deflection and angle of twist.

A C-channel section structural steel beam of 2-in. wide flanges,
3 in. depth and thickness of both flanges and web of 0.25 in. is
loaded as shown with 100 lb. acting in the y-direction on the
free end. Determine the free end deflection and angle of twist.

C-channel displacements are shown

Maximum stresses in the C-channel (no stress averaging):

Maximum stresses in the C-channel (stress averaging):

Stresses in the x-direction is:

Shear stresses in the xy-direction is:

Problems
21. Do problems 12.1 and 12.5 on pages 590 - 598 in your
D. Logan using SAP2000.
End of Chapter 12

A First Course in The Finite Element Method - Lecture Slides PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

A First Course in The Finite Element Method - Lecture Slides PDF

Uploaded by

Copyright:

Available Formats

CIVL 7/8117 Chapter 1 - Introduction to FEM 1/26

A First Course in Finite Elements

• The finite element method has become a powerful tool

• Applications range from deformation and stress analysis of

A First Course in Finite Elements

• With the advances in computer technology and CAD

• Several alternative configurations can be tried out on a

• All of this suggests that we need to keep pace with these

A First Course in Finite Elements

• The material properties and the governing relationships

• An assembly process, considering the loading and

• Solution of these equations gives an approximate behavior

A First Course in Finite Elements

A First Course in Finite Elements

A First Course in Finite Elements

A First Course in Finite Elements

A First Course in Finite Elements

A First Course in Finite Elements

A First Course in Finite Elements

A First Course in Finite Elements

A First Course in Finite Elements

A First Course in Finite Elements

• Basic ideas of the finite element method originated

A First Course in Finite Elements

• In the early 1960s, engineers used the method for

A First Course in Finite Elements

• Hence, even though the finite element method was being

• With the advent of the computer, the solution of thousands of

A First Course in Finite Elements

• Mathematical foundations were laid in the 1970s.

• New element development, convergence studies, and

• Today, developments in distributed or multi-node

A First Course in Finite Elements

• Typically, for the structural stress-analysis problem, the

• For many structures, it is difficult to determine the

A First Course in Finite Elements

• There are two general approaches associated with the

• One approach, called the force, or flexibility method,

• To obtain the governing equations, first the equilibrium

• The result is a set of algebraic equations for determining

A First Course in Finite Elements

• The second approach, called the displacement, or

• The governing equations are expressed in terms of nodal

A First Course in Finite Elements

• These two approaches result in different unknowns (forces

• It has been shown that, for computational purposes, the

A First Course in Finite Elements

• A displacement function is associated with each finite

A First Course in Finite Elements

• The total set of equations describing the behavior of each

A First Course in Finite Elements

The basic steps or building blocks of any application of FEM

A First Course in Finite Elements

Whether the FEM is used for

(1) setting up and solving a problem by hand,

These basic steps are important in setting-up and solving

Step 1 - Discretize and Select Element Types

• Step 1 involves dividing the body into an equivalent system

• The total number of elements used and their variation in size

• The elements must be made small enough to give usable

• Small elements (and possibly higher-order elements) are

Step 1 - Discretize and Select Element Types

• Primary line elements consist of bar (or truss) and beam