For updated version, please click on

http://ocw.ump.edu.my

Finite Element Analysis

Formulation of a Spring Element

by
Dr. Gul Ahmed Jokhio
Faculty of Civil Engineering and Earth Resources
[email protected]
 Lesson Outcomes

• At the end of this lecture, the student should be

able to:
– Understand the 8 steps of Finite Element Analysis
– Apply the 8 steps of Finite Element Analysis to 2DOF
structural systems
– Formulate stiffness matrix for a spring element
– Use the spring element formulation to analyze
structures consisting of spring type members
– Extract and interpret the results obtained from finite
element analysis for spring elements
 Example Structure (Spring)

• Consider the structure shown in Figure 1

• It consists of two springs connected end-to-end with a
6kN force pulling each spring
• The left end of the structure is ‘fixed’
• The stiffness of each of the springs is given as ‘21
kN/m’
• Now, we will use the 8 steps of FEA to analyze this
structure

Figure 1
 Step 1(a): Discretization

• This structure can be conveniently discretized into 3 nodes

and 2 elements with element 1 connected to nodes 1 and 2
and element 2 connected to nodes 2 and 3
• It is conventional to identify nodes with numbers written
close to the nodes
• The elements are also identified with numbers; to
distinguish element numbers from node numbers, these
are generally enclosed in circles
• The discretized structure is shown in Figure 2

Figure 2
 Step 1(b): Element Type

• To model this structure, it is best to use one

dimensional spring elements with 2 nodes as shown in
Figure 3
• The element type, therefore, is ‘linear spring’
• The following assumptions are made for this element:
• The spring obeys Hooke’s Law
• Resists forces only in the direction of the spring

Figure 3
 Step 1(b): Element Type (Continued)

• Points 1 and 2 are the ‘nodes’

• f1 and f2 are the forces in local x-direction
• u1 and u2 are the local nodal displacements
(also called nodal degrees of freedom)
• k is the spring constant or stiffness of the
spring
• x is taken positive going from node 1 to node
2
 Step 2: Displacement Function

• A linear displacement function can be assumed for this

element, shown in Figure 4 and given below:
• 𝑢 = 𝑎1 + 𝑎2 𝑥
• It can be expressed in matrix form as:
𝑎1
• 𝑢= 1 𝑥 𝑎
2
Figure 4
Step 2: Displacement Function (Continued)

• We can input the nodal values to determine the

coefficients:
• 𝑢 0 = 𝑢1 = 𝑎1
• 𝑢 𝐿 = 𝑢2 = 𝑎2 𝐿 + 𝑢1
𝑢2 −𝑢1
• 𝑎2 =
𝐿
• Therefore, expressed in nodal terms the
displacement function becomes:
𝑢2 −𝑢1
• 𝑢= 𝑥 + 𝑢1
𝐿
 Step 2: Displacement Function (Continued)

• Expressing in matrix form:

𝑥 𝑥 𝑢1
• 𝑢 = 1−𝐿 𝐿 𝑢
2
• Or, alternatively;
𝑢1
• 𝑢= 1 𝑁 𝑁 2 𝑢
2
• where,
𝑥
• 𝑁1 = 1 −
𝐿
𝑥
• 𝑁2 =
𝐿
• These are called the ‘shape functions’ Figure 5

• 𝑁1 + 𝑁2 = 1
 Step 3: Strain-Displacement and Stress-
Strain Relationships
• Strain-Displacement Relationship:
• Total element displacement (change in length)
can be expressed mathematically as the
displacement of the end-point minus the
displacement of the starting point:
• 𝛿 = 𝑢 𝐿 − 𝑢 0 = 𝑢2 − 𝑢1
• Assuming constant strain throughout the
element:
𝛿 𝑢2 −𝑢1
• 𝜀= =
𝐿 𝐿
 Step 3: Strain-Displacement and Stress-
Strain Relationships (Continued)
• Stress-Strain Relationship:
• For a spring element, this relationship can be
expressed in terms of Force/Deformation
• 𝑇 = 𝑘𝛿
• 𝑇 = 𝑘 𝑢2 − 𝑢1
• Where, T is the force in the spring taken as
positive if tensile
 Step 4: Element Stiffness Equations and
Matrix
• Considering the force in the spring to be positive if
tensile and its value being T:
• 𝑓1 = −𝑇; 𝑓2 = 𝑇
• 𝑇 = −𝑓1 = 𝑘 𝑢2 − 𝑢1
• 𝑇 = 𝑓2 = 𝑘 𝑢2 − 𝑢1
• 𝑓1 = 𝑘 𝑢1 − 𝑢2
• 𝑓2 = 𝑘 𝑢2 − 𝑢1
𝑓1 𝑘 −𝑘 𝑢1
• =
𝑓2 −𝑘 𝑘 𝑢2
 Step 4: Element Stiffness Equations and
Matrix (Continued)
• This is the system of equations that represents a spring
element
• The element stiffness matrix can be extracted from this
system of equations as:
𝑘 −𝑘
• 𝑘 =
−𝑘 𝑘
• It is the local stiffness matrix. Note that it is a
symmetric square matrix
• Since both the springs in the example structure are the
same, we can use this stiffness matrix for the both of
elements
 Step 4: Element Stiffness Equations and
Matrix (Continued)
1 1
𝑓1 21 −21 𝑢1
• 1
=
𝑓2 −21 21 𝑢 1
2
2 2
𝑓2 21 −21 𝑢2
• 2
=
𝑓3 −21 21 𝑢 2
3
• Where, the superscripts (1) and (2) represent the element numbers
• These can be expressed as element equations as:
1 1 1
• 𝑓1 = 21𝑢1 − 21𝑢2
1 1 1
• 𝑓2 = −21𝑢1 + 21𝑢2
2 2 2
• 𝑓2 = 21𝑢2 − 21𝑢3
2 2 2
• 𝑓3 = −21𝑢2 + 21𝑢3
 Step 5(a): Assembly of Structural Stiffness
Matrix
• The structure has 3 nodes
• Assuming the global forces at these nodes to
be 𝐹1 , 𝐹2 , 𝐹3 :
• To satisfy equilibrium:
1 1 1
• 𝐹1 = 𝑓1 = 21𝑢1 − 21𝑢2
1 2
• 𝐹2 = 𝑓2 + 𝑓2
1 1 2 2
• 𝐹2 = −21𝑢1 + 21𝑢2 + 21𝑢2 − 21𝑢3
 Step 5(a): Assembly of Structural Stiffness
Matrix (Continued)
2 2 2
• 𝐹3 = 𝑓3 = −21𝑢2 + 21𝑢3
• Compatibility requires that:
1 2 3 1
• 𝑢2 = 𝑢2 = 𝑢2 ; 𝑢3
= 𝑢3 ; 𝑢1
= 𝑢1
• Therefore:
• 𝐹1 = 21𝑢1 − 21𝑢2
• 𝐹2 = −21𝑢1 + 21𝑢2 + 21𝑢2 − 21𝑢3 =
− 21𝑢1 + 42𝑢2 − 21𝑢3
• 𝐹3 = −21𝑢2 + 21𝑢3
 Step 5(a): Assembly of Structural Stiffness
Matrix (Continued)
• Re-arranging the element equations:
• 𝐹1 = 21𝑢1 − 21𝑢2 + 0𝑢3
• 𝐹2 = −21𝑢1 + 42𝑢2 − 21𝑢3
• 𝐹3 = 0𝑢1 − 21𝑢2 + 21𝑢3
• Matrix Form:
𝐹1 21 −21 0 𝑢1
• 𝐹2 = −21 42 −21 𝑢2
𝐹3 0 −21 21 𝑢3
 Step 5(a): Assembly of Structural Stiffness
Matrix (Continued)
• Compact form:
• 𝐹 = 𝐾 𝑑 ; where:
𝐹1 𝑢1
• 𝐹 = 𝐹2 ; 𝑑 = 𝑢2 ; and
𝐹3 𝑢3
21 −21 0
• 𝐾 = −21 42 −21
0 −21 21
• This is the assembled stiffness matrix for the
structure
 Step 5(a): Alternative Method (Direct
Stiffness)
• Element stiffness matrices 𝑢1 𝑢2 𝑢3
can be written with rows 21 −21 0 𝑢1
• 𝑘1 = 𝑢2
and columns labelled with −21 21 0
0 0 0 𝑢3
the corresponding degrees
of freedom (DOF) 𝑢1 𝑢2 𝑢3
0 0 0 𝑢1
𝑢1 𝑢2 • 𝑘2 = 𝑢2
0 21 −21
• 𝑘 1 = 21 −21 𝑢1 0 −21 21 𝑢3
−21 21 𝑢2 • Adding the two:
𝑢2 𝑢3 𝑢1 𝑢2 𝑢3
• 𝑘 2 = 21 −21 𝑢2 21 −21 0 𝑢1
𝑢3 • 𝐾 = 𝑢2
−21 21 −21 42 −21
• These can be expanded as: 0 −21 21 𝑢3
 Step 5(b): Boundary Conditions

• The system of equations for the structure is:

𝐹1 21 −21 0 𝑢1
• 𝐹2 = −21 42 −21 𝑢2
𝐹3 0 −21 21 𝑢3
• The stiffness matrix is singular i.e. this system has no
solution
• To get a solution we need to apply the boundary conditions
• The known boundary condition is:
• 𝑢1 = 0, since node 1 is fixed
• Also, we know that: 𝐹2 = 𝐹3 = 6𝑘𝑁
Step 5(b): Boundary Conditions (Continued)

• Applying this condition:

𝐹1 21 −21 0 0
• 𝐹2 = −21 42 −21 𝑢2
𝐹3 0 −21 21 𝑢3
• 𝐹1 = −21𝑢2
• 6 = 42𝑢2 − 21𝑢3
• 6 = −21𝑢2 + 21𝑢3
• The last 2 equations can be expressed in matrix form as:
6 42 −21 𝑢2
• =
6 −21 21 𝑢3
• Note that homogeneous boundary conditions can be directly applied by
removing the rows and columns corresponding to the zero displacements
from the system of equations
Step 6: Solution of the System of Equations

• The solution of this system of equations is straight forward

• 𝑢2 = 0.57𝑚
• 𝑢3 = 0.86𝑚
• These are the structural displacements. The displacements
for the individual elements can be expressed using
compatibility:
1
• 𝑢1 = 𝑢1 = 0
1 2
• 𝑢2 = 𝑢2 = 0.57𝑚
3
• 𝑢3 = 0.86𝑚
 Step 7: Element Deformation and Forces

• For spring elements we are using deformations and

forces instead of strains and stresses
• 𝛿 (1) = 𝑢2 − 𝑢1 = 0.57 − 0 = 0.57𝑚 (elongation)
• 𝛿 (1) = 𝑢3 − 𝑢2 = 0.86 − 0.57 = 0.29𝑚 (elongation)
• Tension in each of the elements can be determined as:
• 𝑇 = 𝑘𝛿
• 𝑇 (1) = 21 × 0.57 = 11.97 𝑘𝑁 ≅ 12𝑘𝑁
• 𝑇 (2) = 21 × 0.29 = 6.09 𝑘𝑁 ≅ 6𝑘𝑁
 Step 8: Interpretation of Results

• In the previous step, we obtained the deformation and the tension

in each of the elements
• Interpretation of these results is not directly related to Finite
Element Analysis
• Whether the deformations are within the allowable limits and
whether the tension would cause the material to fail is up to the
engineer to decide
• Some of the students might have noticed that the tension obtained
in the elements is only approximately equal to the values suggested
by simple equilibrium
• This is because only 2 significant digits after the decimal were
considered while solving the system of equations
• Increasing the number of significant digits will bring the solution
closer to the exact solution, however, it will be exactly the same
only if the computing machine being used has infinite precision
 Author Information

Dr. Gul Ahmed Jokhio

is a Senior Lecturer at FKASA, UMP. He completed his PhD from
Imperial College London in 2012.