Introduction to Finite Element Model

QAZI Samiullah (PhD)

Assistant Professor
CIVIL Engineering Department
UET Peshawar
 Objective

Understand the basics of Finite Element

Model

2
Introduction to FEM

3
Introduction to FEM

4
Introduction to FEM

5
Introduction to FEM

6
 Introduction to FEM

• FEM = Finite Element Method

• FEM is a numerical approximation method,

which is used for the calculation and optimization
of the structural behavior of mechanical parts !

• Finite Element
=
Discrete structural description of continua with
help of mathematical formulations

7
 Introduction to FEM

• 1950´s: Application of matrix methods for structural analysis

• 1950´s: Introduction of the displacement- and stiffness method
for complex aerospace structures
• 1960: Birth date of the name “Finite Elements“
• 1970´s: FEM applications mainly in the aerospace and the
automotive industry (NASA -> NASTRAN)
• 1980´s: Introduction of powerful computer graphics
• 1997: FEM is the standard tool for structural analysis

8
 Introduction to FEM
• Finite Elements have simple geometric shapes,
like triangle, rectangle, cube.

• The points at the corners are called Node or Grid point.

Nodes connect the elements.

• The description of the structural behaviour inside an element is

done by the calculation of the nodal displacements (=
discretization) in combination with
special shape function (= mathematical functions used
for integration across the element region)

9
 How does FEM work?

Calculate the area of a circle

 Solve a number of simple problems, add them all up and

get the answer of a complex problem
 Divide a complex problem into simple ones
 Divide complex geometry into simple objects which we
can understand (Lines, Squares, Cubes)
 Use the computer to do millions (and millions, …) of
numerical operations
 Use modern hardware equipment to present the results
graphically
10
 Mathematical Model

Formulation of simple
elements

K = Spring stiffness (Ea/L).

u

U = Spring elongation
F
F = Spring force

K*U=F
11
Mathematical Model

Dividing of the geometry into simple

elements and assembling all elements
[K] = Stiffness matrix of the part
(Sum of all elements)
{U} = Components of the displacements
of the single nodes of the part
{F} = Components of the loads of
the single nodes of the part
[K] *{U} = {F}

Solving the matrix equation with

thousands of unknowns

12
 Theoretical Background
 A given problem is discretized by dividing the original
domain into simply shaped subdomains, the so called
elements.

Y uy
ux

[ k ]e { u }e = { f }e
X
element level

 Each element it quite simple, and the program can figure

out its mechanical properties quite easily
 By summation of all the element contributions one gets
the whole model behavior

13
 Theoretical Background

 Each element is connected to its neighbour only at

a number of points, called nodes

 Each node has 6 independent possibilities to move:

3 translational and 3 rotational

 These independent possibilities to move are called

degrees of freedom (DOF’s)

14
 Theoretical Background

Ry ,My
Translations
Tx, Ty, and Tz (1, 2, and 3)
Ty ,Fy
Rotations
Rx ,Mx
Rx, Ry, and Rz (4, 5, and 6)
Tz ,Fz
Tx ,Fx X
Forces
Fx, Fy, and Fz
Rz ,Mz
Moments
Mx, My, and Mz Z Cartesian Coordinate System

15
 Steps in an FE Analysis

Geometry

Elements Displacements
Stresses
Loads Forces
Strains

Analysis
Solver Analysis Results
Model

Contour Plots
Constraints X-Y-Plots
Listings
Materials

Properties

16
 FEM Program algorithm

 Represent continuous model as a collection of

elements and connections
 Formulate element stiffness matrices [k]
 Assemble all element stiffness matrices to a global
stiffness matrix [K]
 Generate load vector [F]
 Solve matrix [F]=[K][x] with respect to [x]
 Calculate element stresses and strains

We shall illustrate this using a simple example...

17
 Continuous Model

F = 1000 N

A = 20mm2
E = 210000 N/mm2 L = 50 mm

A = 50mm2
L = 50 mm
E = 210000 N/mm2

18
 Continuous Model

We make one-dimensional elements with two degrees of

freedom

Node 3
A= 20mm2
E = 210000 N/mm2 Element 2
L = 50 mm
Node 2
A = 50mm2
E = 210000 N/mm2 Element 1
L = 50 mm
Node 1

19
 Element stiffness matrices

For one element, the stiffness relation is

 AE AE 
 L 
k L 
AE AE 
 
 L L 

Each of the element matrices then becomes:

 84  84  210  210

k1     1000 N / mm 2
k2    1000 N / mm2
 84 84   210 210 

20
 Global stiffness matrices

The Global stiffness matrix is assembled by combining the

matrices at the appropriate degrees of freedom for each
node

 84  84 0 0 0 0 
K   84 84 0 1000 N / mm  0 210  210 1000 N / mm
 0 0 0 0  210 210 

 84  84 0 
K   84 294  210 1000 N / mm
 0  210 210 

21
 Apply boundary conditions

Node 1’s DOF is constrained, and is therefore a nulled,

giving the effective stiffness matrix

 x1  0
x   x2  x   x2 
 x3   x3 

22
 Load Vector

The only force is acting at Node 3, giving the Load Vector

 0 
F   0  N
 1000

23
 Solve the matrix equation

The simple equation is now solved, giving the

deformations
F   K  x
 0   84  84 0  0
 0    84 294  210   x 1000 N
     2
 1000  0  210 210   x3 
 x1   0 
 x    0.012 mm
 2  
 x3   0.017

24
Calculate element forces and stresses

Using well known formula, the strains and

stresses are calculated for each element

L  0.012  0
1    0.00024
L 50
L  0.017  (0.012)
2    0.00010
L 50
 1  E  1  50.4 N / mm2
 2  E   2  21.0 N / mm2

25
 Introduction to FEM

• What an FEA user has to specify :

• Choose the element type

• Creation of the Finite Element Mesh
• Specification of the material properties
• Specification of the element properties
• Verification of the Mesh quality
• Application of the loads and boundary conditions
• Specification of the options of the desired analysis type
• Request of the desired results
• Interpretation of the results

26
 The Software Returns:

Static or time-dependent:
 Deformations
 Stresses
 Strains
Others:
 Eigenfrequencies
 Stability/Buckling Load Factors
Other Types of Analysis will Return other Results

27
 Element Types
Standard Elements

First order

Line Triangle/Quadrilateral Hexahedral Tetrahedral

( tri/quad) ( hex ) ( tet )
Second order

Line Triangle/Quadrilateral Hexahedral 10-noded Tetrahedral

28
 Element selection criteria

Which element type should be used?

 Beams, plates or solids?

 Quadrilateral or triangle?
 Hexahedral, pentahedral or tetrahedral ?
 Higher or lower order?

The choice of the element type is strongly dependent

on the structure to be analyzed

29
 Beam Elements

• 1 - Dimensional elements
• 6 DOFS
• First and second order
• Cross section constant or varying along the element length
• Transfers axial forces, torsional and bending moments
• Neutral line can be moved away from the node (“Offset”)
• Degrees of freedom may be released at the end of the
elements
• Assumes non-deforming cross section when deforming

30
 Typical Beam Models

Trusses

Frames

Stiffeners

31
Examples of Beam Models

32
 Shell Elements

 2 - Dimensional Elements
 5 DOFS (No “drilling” DOF)
 Thickness constant or varying across element
 Triangle and Quadrilateral basis shape
 First and second order
 Neutral line can be moved away from the nodes (“Offset”)
 Assumes constant thickness when deforming

33
 Typical Shell Models

Thin shells, Thin plane or curved

thinwalled metal sheets
pressure vessels

Aircraft components

Automotive parts

34
Examples of shell models

35
 Examples of shell models

Courtesy of Adtranz

Courtesy of Adtranz Courtesy of Adtranz

36
 Solid Elements

 3 - Dimensional elements
 3 DOFS (no rotational DOFs)
 Tet and cube basic shapes
 First and second order

37
 Typical Solid Models

Thick walled
pressure vessels

Thick plates
and consoles

Cast iron parts and

fittings

38
 Examples of Solid Models

Courtesy of Volvo Car

39
Example:
Deciding between 1D, 2D and 3D
element models
The Model
 We have the following model:
It consist of rectangular beams that
are 100 mm wide and 200 mm high
with thickness 5 mm. The longest
beams are 2000 mm and the total width of the
model is 1000 mm. The centerline of the crossing
bars are located 550 mm from each end.
 The beams are welded together, but the weld fillets
are to be neglected due to their small size
 The structure carries a load of 2000 kg evenly
distributed over the two cross bars
The problem

 We want to investigate the maximum overall

deflection and maximum stress
 As a guideline for meshing, we do not want
element aspect ratios greater than 5

Now lets do 1D, 2D and 3D calculations and see what

effort they take and what answers they yield!
Beam Model

Element size 100 mm

Shell Model

Element size 33 mm
Solid Model

Element size 25 mm
 Comparing the three analyses
1D 2D 3D
Number of nodes 60 9072 66015
Number of elements 60 3024 33003
DOF 336 44640 196317
Disk usage for solving 1 MB 195 MB 124 MB
Memory usage for solving 16 MB 21 MB 1530 MB

Modelling time 20 min 30 min 60 min

Solving time 2 sec 70 sec 2000 sec
Postprosessing time 15 min 2 min 1 min
Total time 35 min 33 min 94 min
Maximum deflection 0.0984 mm 0.143 mm 0.135 mm

The calculations were conducted using MSC.PATRAN V8.5, MSC.NASTRAN

V70.5. Hardware were HP Kayak 450 MHz, 256 MB RAM, MS/NT 4.0.
 Comparing the three analysis
Beam Model
Minmal modelling effort, accurate calculation results
Does not calculate beam cross section deformation (which happens in this
case)
Postprocessing somewhat awkward and time consuming, could present errors

Plate Model
Higher modelling effort, good calculation results
Must have control of what is “top“ and “bottom“ of elements

Solid Model
Highest modelling effort, good calculation results
Does not give any significant additional results compared to plate model
Geometric Model
Geometry

The geometry defining the model in question can be:

 Imported into the FE program

 Created using the FE program
Geometry

 It is of greatest importance that small features not

important for the calculation of results the user is
looking for should be neglected

 A CAD MODEL is not the same as a CAE MODEL!!!

Coordinate Systems
 Different systems available
 Default is the global Cartesian system

Z Z Z
q

z z r
r Y
x Y Y
X X q X F
y

Cartesian Cylindrical Spherical

Discretization of the geometric
model (Meshing)
Element Type
Before meshing the user must chose between different
element types:
 Beam, Shell or Solid
 First or second order
 Triangle or quadrilateral basis shape

When this is done, meshing of the structure can begin

Element Size

Before the mesh routine is activated, the user specifies

the element size. This can be done in a number of ways:

 Global element size

 Preferred number of elements along an edge
 Varying element length along an edge
 Allowed deviation from geometry
Automatic meshing routines
Unmeshed… and meshed geometry
Using meshing refinement

Examples of:
 Curvature based
mesh refinement
 Desired number
of elements along
edges - uniform
and varying
Defining Model Loads
Definition of Loads
The loads are the reasons why we do calculations on
the models. We want to calculate the response on
the model due to the loads

The loads can be in terms of:

 Force
 Displacement
 Pressure
 Temperature
 Heat flux
 etc etc
Definition of Loads
 Loads can be applied to points , surfaces or bodies
 They can be constant, or time dependent

F F
F

t t
t
Defining Model Constraints
Constraints
 By constraining our model we model the interface
of the model to the rest of the world
 Constraints can be given to both displacements
and rotations
 If the model is not constrained, it will yield
numerical problems (“singularities”) during solving,
as shown below

Unstable Stable
Constraints

Fixed Connection

X
Transfers vertical and
No displacement horizontal forces and
or rotation possible moments
Constraints

Pinned connection

X
Rotation possible,
no displacements
Horizontal and vertical
reaction forces,
no reaction moments
 Constraints

Sliding and pinned connection

X
Rotation and horizontal
displacement possible,
no vertical displacement Vertical reaction force,
no reaction moments or
horizontal forces
Defining Materials
Material Model Selection

 Isotropic
 Orthotropic (also "Composite"- plates/shells)
 Anisotropic
 Hyperelastic
 Manual data import
 User-modifiable material libraries
Linear Material Assumptions
Conditions in the range of validity
 Theory of Elasticity of linear static calculations:
E (Hooke’s Law)
 Linear stress-strain-curve
 (Hooke’s Law) (Valid as long as the
F resulting stresses are below yield)

 Displacements are very small

u
compared to the size of the part or its
thickness

F=ku  The principal of superposition is valid !

x = E  x
+ =
Linear Material Properties
 Young‘s modulus, E

 Shear modulus, G E
G
2  (1  ν)
 Poisson ratio, n
 Relationship between axial and transverse
strain:
y
n =
x

P x P
 Non-linear Materials

 K
Outside the range of validity of linear
static calculations:
 Nonlinear geometric effects (stability,
Eo
large displacements)
  Nonlinear material behaviour (elasticity
and/or plasticity)
Stress, 
Tension  Time- and temperature-dependent
Limit
material behaviour (creep)
YieldLlimit
 Contact with/without friction
 The principle of superposition is no
Linear longer valid !
Range / =E

+ 
Strain, 
0.2% Offset
Defining Element Properties
Properties

 In addition to the element shapes extra

information is needed in order to describe the
model completely.
 The extra information needed is dependent on
element type, but almost always material is an
example of a property that needs to be input
 The extra element information is input as element
properties
Beam Element Properties

Beam elements needs to input:

 Material
 Cross sectional Area (A) - transfers axial forces
 Moments of Inertia (I) - transfers bending and shear
 Torsional Moment of Inertia (J) - transfers torsional
forces
Shell Element Properties

Shell Element needs to input:

 Material
 Thickness
Solid Element Properties

Solid elements needs to input:

 Material
Conducting the analysis
Checking the model

Before the analysis is run, the model should be

checked. A few items might be good to verify:

 Check for coincident nodes and/or elements

 Check the element quality
 Check the mass or volume of the model
 ???
Types of Analysis
Linear Static Analysis

 Calculation of stresses
 Calculation of deformations

”Linear” means that the calculated results are linear

proportional to the applied loads.

”Static” means that the loading is not time dependent

or the time dependence is negligible.
Finding the answer..

 After the model is created in the preprocessor, the

problem is sent to the solver

 The solver calculates the answer to the problem, and this

can be viewed and interpreted in the postprocessor
Postprocessing Calculation Results
Postprocessing of Results

Postprocessing can be done as:

 Plot of Deformed Structure

 Contour Plots
 Animated Display of Deformations and Contours
 X-Y Plots
 Result Listings
Questions?

81