Finite Element Method

- FEM is a set of numerical algorithms to solve ODE’s and PDE’s

- FEM (or FEA) has application in all engineering field
- But the method is most frequently used in Mechanics

Main Applications: Field Problems

- Elasticity, Plasticity, Viscoelasticity, Hyperelasticity, etc.
- Heat transfer
- Fluid Mechanics
- Electromagnetic

KINES 384
 What is a field problem?

A field problem includes:

• Geometry of the field
• Field variable(s)
• Differential equation(s) governing the field

A Review Problem: An Elastic Field

KINES 384
 Example: A 2-D elastic field
Overview
• Geometry is subdivided
• Reaction forces
• Applied loads

KINES 384
 General steps to make FEM work?
1. Mesh Generation: The complex region is subdivided into smaller
subdomains called “ELEMENTS”. Apex of an element’s sides is a
“NODE”.

Then we use interpolation functions

to approximate the unknowns in
terms of nodal values:

KINES 384
2. Material and Physical Properties: (elastic
modulus, density, Poisson’s ration, heat
conduction coefficient,…).
3. Loading Conditions: (concentrated and
distributed forces, moments, heat
source,…).
4. Kinematic Boundary Conditions:
(constraints and fixed points).
FEM Applied to Solid Mechanics Problems
• A FEM model in solid mechanics can
be thought of as a system of assembled
springs. When a load is applied, all
elements deform until all forces
balance.
• F = K.d
Create elements
of the beam • K is dependant upon Young’s
modulus and Poisson’s ratio, as
well as the geometry.
• Equations from discrete elements are
assembled together to form the global
Nodal displacement and forces stiffness matrix.
• Deflections are obtained by solving the
assembled set of linear equations.
• Stresses and strains are calculated from
the deflections.

KINES 384
KINES 384
KINES 384
meshed model
KINES 384
 Boundary Conditions for a Solid Mechanics Problem

• Displacements DOF constraints usually specified at model boundaries to

define rigid supports.

• Forces and Moments Concentrated loads on nodes usually specified on

the model exterior.

• Pressures Surface loads usually specified on the model exterior.

• Temperatures Input at nodes to study the effect of thermal expansion or

contraction.

• Inertia Loads Loads that affect the entire structure (ex: acceleration,
rotation).

KINES 384
KINES 384
KINES 384
• Model of a Disk and a Brake Pad:

KINES 384
 Some Examples:
• Foot Model:

• Solid Model:

• FE Mesh (inside the foot):

KINES 384
• Foot, Ankle & Shank Model:

KINES 384
 Mathematical Concepts and Steps
• Discretization of the region (degrees of freedom)
• Generating element stiffness matrix

kel uel  f el
• Applying nodal loads and boundary conditions
• Assembling elements stiffness matrices
• A set of linear algebraic equations with unknown deformation

K u   f 
• Solution of the equations and computation of displacement
field, stresses and strains.
KINES 384
Example: A linear spring model (direct or stiffness method)

For this model:

• Two-node element
• One degree of freedom/node (1-D displacement)
• Two DOF/element
• Force-displacement equation:

  d 2 x  d1x T  K .

KINES 384
Generating Element Stiffness Matrix (direct or stiffness method)

• Stiffness matrix relates forces to nodal displacements

• Since spring is in equilibrium:

T2  T1  T
F1x  T  k (d 2 x  d1x )  kd1x  kd2 x
F2 x  T   k (d 2 x  d1x )  kd1x  kd2 x
kd1x  kd2 x  F1x

 kd1x  kd2 x  F2 x

KINES 384
 In Matrix Notations:

 k  k  d1x   F1x 
 k    
 k  d 2 x   F2 x 

• The LHS matrix is the element stiffness matrix (ESM).

• The element stiffness matrix is:
1. Square
2. Symmetric
3. Normally non-singular (non-zero determinant)
In this case the determinant is zero. Why?

KINES 384
 How does assembly of ESM work?
Consider the following two-element system:

 k1  k1   k2  k2 
 k  k k 2 
 1 k1   2
d1 x d2x d2x d3x
Now we need to expand these matrices for the overall system.

KINES 384
Expansion:

 k1  k1 0 0 0 0
K1   k1 k1 0 K 2  0 k2  k2 
0 0 0  0  k2 k2 
d1x d2x d3x d1x d2x d3x

Then the Global Stiffness Matrix (GSM) will be:

 k1  k1 0 

K  K1  K 2   k1 k1  k2 
 k2 
 0  k2 k2 

KINES 384
This results in a set of Linear Algebraic Equation:
 k1  k1 0  d1x  F1x 
 k     
 1 k1  k 2  k2  d 2 x   F2 x 
 0  k2 k2    F 
 3x   3x 
d

The GSM is square and symmetric but singular!

The reason is that the system is in rigid body mode.
It must be tied down with a boundary condition.

d1x  0

KINES 384
Incorporating the boundary condition removes singularity:

k1  k 2  k 2  d 2 x   F2 x 
 k    
 2 k 2  d 3 x   F3 x 

d 2 x 
Vector of unknowns =  
d 3 x 
KINES 384
Let us consider the analysis of the following loaded system of
linear springs (each with 2 nodes).
This system has 4 linear spring elements and there are 4
global nodes. Node 1 is fixed but node 4 can slide under the
(known) applied load F. We are interested to compute the
nodal displacements at nodes 1, 2, 3, and 4.
For element 1 which is made by nodes 1 and 2, we can write
the following equilibrium equation:

k11(1) k12(1)  d1(1)   F1(1) 

 (1) (1)   (1) 
  (1) 
k21 k22  d 2   F2 
For elements 2, 3, and 4 we have:

k11( 2) k12( 2)  d1( 2)  F1( 2)  k11(3) k12(3)  d ( 3)


 1   1   F ( 3)

 ( 2) ( 2)   ( 2) 
  ( 2)   ( 3)   ( 3) 
( 3)   ( 3) 
k21 k22  d 2  F2  k21 k22   d 2    F2  
k11( 4) k12( 4)  
d1 
( 4)
 F1 
( 4)

 ( 4) ( 4)   ( 4) 
  ( 4) 
k21 k22   d 2   F2  

And the connectivity matrix is:

1 2 2 3 
 2 3 3 4
 
One column for each element
We use d1, d2 , d3 , d4 from now on to denote the global degrees of
freedom. Let us re-write each of these equations to include in
the displacement vector all nodal displacements of our system.
k11(1) k12(1) 0 0 d1   F1(1)  0 0 0 0 d1  0 
 (1)         
k21
(1)
k22 0 0 d 2   F2(1)  0 k11( 2) k12( 2) 0 d 2   F1( 2) 
0      
0 0 0 d3  0  0 ( 2)
k21 ( 2)
k22 0 d3   F2( 2) 
   
0 0 0 0 d 4  0  0 0 0 0 d 4  0 

0 0 0 0 d1  0  0 0 0 0  d1  0 
     
0 k11(3) k12(3) 0 d 2   F1(3) 
0 0 0 0  d 2  0 
0      ( 3)    
( 3)
k21 ( 3)
k22 0 d3   F2  0 0 k11( 4) k12( 4)  d3   F1( 4) 
   
0 0 0 0 d 4  0  k22  d 4   F2( 4) 
( 4) 
0
( 4)
0 k21
The corresponding RHS and LHS of equations will be
added together. The overlapping members of matrices
will sum up and the final equation for the system will
be:
The total forces acting on nodes 2 and 3 are zero:

and forces at nodes 1 and 4 are:

Considering constraint at node 1:

The reaction force will be calculated using the main equation: