Introduction to the Stiffness

(Displacement) Method
Introduction
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 We next illustrate how to assemble the total stiffness matrix for a structure comprising an assemblage of spring elements by using 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.
After establishing the total structure stiffness matrix, impose boundary
conditions (both homogeneous and nonhomogeneous).

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

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 43

Step 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.

Step 2 - Select a Displacement Function

A displacement function u 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
a2
We can express u as a function of the nodal displacements d by evaluating u
at each node and solving for a1 and a2.

u ( x = 0) = d1x = a1
u ( x = L ) = d 2 x = a2 L + a1
CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 44

Solving for a2:

d 2 x d1x
a2 =
L

Substituting a1 and a2 into u gives:

d d1x
x + d1x
u = 2 x

In matrix form:

u = 1
L

x d1x

L d 2 x

Or in another form:

u = [N1

d1x
N 2 ]
d 2 x

where N1 and N2 are defined as:

N1 = 1

x
L

N2 =

x
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.

Step 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 are related by Hookes law:

T = k

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 45

where deformation of the spring is given as:

= u(L ) u (0)
= d 2 x d1x
Step 4 - Derive the Element Stiffness Matrix and Equations
We can now derive the spring element stiffness matrix as follows:

f1x = T

f2 x = T

Rewrite the forces in terms of the nodal displacements:

T = f1x = k d 2 x d1x

T = f2 x = k d 2 x d1x

We can write the last two force-displacement relationships in matrix form as:

f1x k k d1x
=

k
k
f

d 2 x
2x
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

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 46

Step 5 - Assemble the Element Equations

and Introduce Boundary Conditions
The global stiffness matrix and the global force vector are assembled using
the nodal force equilibrium equations, and force/deformation and compatibility
equations.
N

K = [K ] = k ( e )
e =1
N

F = {F } = f ( e )
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 = Kd
Step 7 - Solve for the Element Forces
Once the displacements are found, the forces in each element may be calculated from:

T = k

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 47

Example 1 - Spring Problem

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 d1x
=
d
f
k
k

3x 1
1 3x
For element 2:

f3 x k 2 k 2 d 3 x

=
k 2 d 2 x
f2 x k 2
Both continuity and compatibility require that both elements remain connected at
node 3.

d3x

(1)

= d3x

(2)

We can write the nodal equilibrium equation at each node as:

F1x = f1x

(1)

F2 x = f2 x

( 2)

F3 x = f3 x

(1)

+ f3 x

( 2)

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 48

Therefore the force-displacement equations for this spring system are:

F1x = k1d1x k1d 3 x

F2 x = k 2 d 3 x + k 2 d 2 x
F3 x = k1d1x + k1d 3 x + k 2 d 3 x k 2 d 2 x
In matrix form the above equations are:

0
F1x k1

k2
F2 x = 0
F k k
2
3x 1

k1 d1x

k 2 d 2 x

k1 + k 2 d 3 x

or

F = Kd
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:
CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 49

d1x d 3 x
k k1
k (1) = 1
k1
k1
For element 2:

d3x d2x
k k2
k (2) = 2
k 2
k 2
Write the stiffness matrix in global format for element 1 as follows:
(1)
(1)
1 0 1 d1x f1x

k1 0 0 0 d 2(1x) = f2(x1)

1 0 1 d 3(1x) f3(x1)

For element 2:
( 2)
(2)
0 0 0 d1x f1x

k 2 0 1 1 d 2( 2x) = f2(x2 )

0 1 1 d 3( 2x) f3(x2 )

Apply the force equilibrium equations at each node.

f1x (1) 0 F1x

( 2)
0 + f2 x = F2 x
f (1) f ( 2 ) F
3x 3x 3x
The above equations gives:

0
k1
0
k2

k1 k 2

k1 d1x F1x

k 2 d 2 x = F2 x

k1 + k 2 d 3 x F3 x

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 50

To avoid the expansion of the each elemental stiffness matrix, we can use a
more direct, shortcut form of the stiffness matrix. For element 1:

d1x d 3 x
k k1
k (1) = 1
k1
k1
For element 2:

d3x d2x
k k2
k (2) = 2
k 2
k 2
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:

d1x

d2x

0
k1
k2
K= 0

k1 k 2

d3 x
k1
k2

k1 + k 2

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: homogeneous boundary
conditions (the most common) occur at locations that are completely prevented
from movement; nonhomogeneous boundary conditions occur where finite

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 51

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 d1x = 0. The equations describing the elongation of the
spring system become:

0
k1
0
k2

k1 k 2

k1 0 F1x

k 2 d 2 x = F2 x

k1 + k 2 d 3 x F3 x

Expanding the matrix equations gives:

F1x = k1d 3 x
F2 x = k 2d 3 x + k 2d 2 x
F3 x = k 2d 2 x + (k1 + k 2 )d 3 x
The second and third equation may be written in matrix form as:

k 2 d 2 x F2 x
k2
k k + k d = F
2 1
3x
2 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 = k1d 3 x
For homogeneous boundary conditions, we can delete the row and column corresponding to the zero-displacement degrees-of-freedom.
Lets 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 d1x = . The equations describing the elongation of the spring system become:

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 52

 k1
0

k1

0
k2
k2

k1 F1x

k 2 d 2 x = F2 x

k1 + k 2 d 3 x F3 x

Expanding the matrix equations gives:

F1x = k1 k1d 3 x
F2 x = k 2 d 2 x k 2 d 3 x
F3 x = k1 k 2 d 2 x + (k1 + k 2 )d 3 x
By considering the second and third equations because they have known nodal
forces we get:

F2 x = k 2 d 2 x k 2 d 3 x
F3 x + k1 = k 2 d 2 x + (k1 + k 2 )d 3 x
Writing the above equations in matrix form gives:

k 2 d 2 x F2 x
k2
k k + k d = F + k
2 1
3x
1
2 3x
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 k1d 3 x

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 53

Example 2 -Spring Problem

Consider the following three-spring system:

The elemental stiffness matrices for each element are :

1

1 1
k (1) = 1000

1 1
4

1 1
k ( 2 ) = 2000

1 1

1 1
k ( 3 ) = 3000

1 1
Using the concept of superposition (the direct stiffness method), the global stiffness matrix is:

0
1000
0
1000
0
3000
0
3000

K=
1000
0
3000 2000

3000 2000 5000

0
The global force-displacement equations are:

1000
0 d1x F1x
0
1000
0
3000 d 2 x F2 x
3000
0

=
1000
0
3000 2000 d 3 x F3 x

3000 2000 5000 d 4 x F4 x

0
CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 54

We have homogeneous boundary conditions at nodes 1 and 2 (d1x = 0 and d2x =

0). Deleting the first two rows and the first two columns and substituting for the
known force at node 4 (F4x = 5000) gives:

3000

2000

d 3 x = 0

2000
5000 d 4 x 5000

Solving for d3x and d4x gives:

d3x =

10
in.
11

d4x =

15
in.
11

To obtain the global forces, substitute the displacement in the force-displacement

equations.

0
0 0
1000
F1x 1000
F 0
3000
0
3000 0
2x

=
0
3000 2000 10 11
F3 x 1000

F4 x 0
3000 2000 5000 15 11
Solving for the forces gives:

F1x =

10,000
lb
11

F3 x = 0

F2 x =
F4 x =

45,000
lb
11

55,000
lb
11

Next, use the local element equations to obtain the force in each spring.
For element 1:

f1x 1000 1000 0

=
10

1000
1000
f

11
3x
The local forces are:
CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 55

10,000
f1x =
lb
11

10,000
f3 x =
lb
11

A free-body diagram of the spring element 1 is shown below.

For element 2:

f3 x 2000 2000 10 11
=
15

2000
2000
f

11
4x
The local forces are:

10,000
f3 x =
lb
11

10,000
f4 x =
lb
11

A free-body diagram of the spring element 2 is shown below.

For element 3:

f4 x 3000 3000 15 11
=

f2 x 3000 3000 0
The local forces are:

45,000
f4 x =
lb
11

45,000
f2 x =
lb
11

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 56

A free-body diagram of the spring element 2 is shown below.

Example 3 -Spring Problem

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:

0
0
0
200 200
200
0
0
400 200

K=
400 200
0
0 200

200
0
0
200
400

0
0
0 200
200

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 57

The global force-displacement equations are:

0
0
0 d1x F1x
200 200
200
400 200
0
0 d 2 x F2 x

0 200
400 200
0 d 3 x = F3 x

400
200
0
0
200

d 4 x F4 x

0
0 200
200 d 5 x F5 x
0
Applying the boundary conditions (d1x = 0 and d5x = 20 mm) and the known
forces (F2x, F3x, and F4x equal to zero) gives:

0
0 d2x 0
400 200
200 400 200
0 d 3 x 0

=
0
200 400 200 d 4 x 0

200 200 0.02 F5 x

0
0
Rearranging the first three equations gives:

0 d 2 x 0
400 200

200
400 200 d 3 x = 0

0 200
400 d 4 x 4
Solving for d2x, d3x, and d4x gives:

d 2 x = 0.005m

d 3 x = 0.01m

d 4 x = 0.015m

Solving for the forces F1x and F5x gives:

F1x = 200(0.005 ) = 1.0kN

F5 x = 200(0.015) + 200(0.02) = 1.0kN

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 58

Next, use the local element equations to obtain the force in each spring.
For element 1:

f1x 200 200 0

=

f2 x 200 200 0.005

The local forces are:

f1x = 1.0kN

f2 x = 1.0kN

For element 2:

f2 x 200 200 0.005

f3 x 200 200 0.01

The local forces are:

f2 x = 1.0kN

f3 x = 1.0kN

For element 3:

f3 x 200 200 0.01

=
0.015
200
200

4x
The local forces are:

f3 x = 1.0kN

f4 x = 1.0kN

For element 4:

f4 x 200 200 0.015

=
0.02
200
200

5x
The local forces are:

f4 x = 1.0kN

f5 x = 1.0kN

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 59

Example 4 -Spring Problem

Consider the following spring system:

The boundary conditions are:

d1x = d 3 x = d 4 x = 0
The compatibility condition at node 2 is:

d2x

(1)

= d2x

(2)

= d2x

(3)

= d2x

Using the direct stiffness method: the elemental stiffness matrices for each element are:
1

k k1
k (1) = 1
k1
k1
2

k k2
k (2) = 2
k 2
k 2

k k3
k (3) = 3
k 3
k 3
CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 60

Using the concept of superposition (the direct stiffness method), the global stiffness matrix is:
1

k
0
k
0
1
1

k1 k1 + k 2 + k 3 k 2 k 3
K=

k2
0
k2
0
0
k3
k3
0

Applying the boundary conditions (d1x = d3x = d4x = 0) and the known forces (F2x
= P) gives:

P = (k1 + k 2 + k 3 )d 2 x
Solving the equation gives:

d2x =

P
k1 + k 2 + k 3

Solving for the forces gives:

F1x = k1d 2 x

F3 x = k 2 d 2 x

F4 x = k 3 d 2 x

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.
Thus, 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 axiCIVL 7117 Finite Elements Methods in Structural Mechanics

Page 61

symmetric stress element, the plate bending element, and the three-dimensional
solid stress element.

Total Potential Energy

The total potential energy 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; 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

The total strain energy is:

x

U = dU = (kx ) dx =
L

1 2
kx
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:
CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 62

 = Fx
Therefore, the total potential energy is:

p =

1 2
kx Fx
2

The concept of a stationary value of a function G is shown below:

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.

dG
=0
dx
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
d 1

d 1 +

p
d 2

d 2 + ... +

p
d n

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.
An admissible variation is one in which the displacement field still satisfies the
boundary conditions and interelement continuity.
CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 63

To satisfy p = 0, all coefficients associated with di must be zero independently,

therefore:

p
d i

=0

i = 1, 2,L, n

or

{d }

=0

Example 5 -Spring Problem

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.

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 64

The total potential energy is defined as the sum of the internal strain energy U
and the potential energy of the external forces :

p =U +
U=

1 2
kx
2

= Fx

The variation of p with respect to x is:

p =

p
x

x = 0

Since x is arbitrary and might not be zero, then:

p
x

=0

Using our express for p, we get:

p =
p
x

1 2
1
kx Fx = 500(lb / in ) x 2 (1000lb ) x
2
2
= 0 = 500 x 1000

x = 2.0in.

If we had plotted the total potential energy function p for various values of deformation we would get:

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 65

Lets 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 is:

p =

1
k d 2 x d1x
2

f1x d1x f2 x d 2 x

Expanding the above express gives:

p =

1 2
2
k d 2 x 2d1x d 2 x d1x f1x d1x f2 x d 2 x
2

Minimizing the total potential energy p:

p
d i

=0

i = 1 to 2

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 66

 p
d1
p
d 2

=0=

1
k 2d 2 x + 2d1x f1x
2

=0=

1
k 2d 2 x 2d1x f2 x
2

In matrix form the above equations are:

f1x k k d1x
=

k
k
f

d 2 x
2x
This result is identical to stiffness matrix we obtained using equilibrium.

Example 6 - Spring Problem

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)

1
2
k1 (d 3 x d1x ) f1x d1x f3 x d 3 x
2

The potential energy p for element 2 is:

(2)

1
2
k 2 (d 4 x d 3 x ) f3 x d 3 x f4 x d 4 x
2

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 67

The potential energy p for element 3 is:

(3)

1
2
k 3 (d 2 x d 4 x ) f2 x d 2 x f4 x d 4 x
2

The total potential energy p for the spring system is:

3

p = p

(e)

e =1

Minimizing the total potential energy p:

p
d1x

= 0 = k1d 3 x + k1d1x f1x

(1)

p
(3)
= 0 = k 3 d 2 x k 3 d 4 x f2 x
d 2 x
p
(1)
( 2)
= 0 = k1d 3 x k1d1x k 2d 4 x + k 2d 3 x f3 x f3 x
d 3 x
p
( 2)
(3)
= 0 = k 2 d 4 x k 2 d 3 x k 3 d 2 x + k 3 d 4 x f4 x f4 x
d 4 x
In matrix form:

0
0 d 1x
f1x
k1
k1

k 3 d 2 x
k3
0
f2 x

= (1)

(2)
k1 0 k1 + k 2 k 2 d 3 x f3 x + f3 x
(2)
(3)

0 k 3 k 2 k 2 + k 3 d 4 x f4 x + f4 x
Using the following force equilibrium equations:

f1x

(1)

= F1x

f2 x

(3)

= F2 x

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 68

f 3 x + f3 x
( 1)

(2)

= F3 x

f4 x

(2)

+ f4 x

(3)

= F4 x

The global force-displacement equations are:

0
0 d1x F1x
1000
1000
0
3000
0
3000 d 2 x F2 x

=
1000
0
3000 2000 d 3 x F3 x

3000 2000 5000 d 4 x F4 x

0
The above equations are identical to those we obtained through the direct stiffness method.
Problems:

5.

Do problems 2.2, 2.9, and 2.17 on pages 59 - 62 in your textbook A First

Course in the Finite Element Method by D. Logan.

CIVL 7117 Finite Elements Methods in Structural Mechanics

Page 69