Anda di halaman 1dari 20

INTRODUCTION TO FINITE

ELEMENT METHOD

Professor Dr. Mongi Ben Ouezdou

M. Ben Ouezdou, University of Nizwa, 2011 1

Basic Concept

• Building a complicated object with simple


blocks,
or

• dividing a complicated object into small


and manageable pieces.

M. Ben Ouezdou, University of Nizwa, 2011 2

1
Basic Concept

• The structure is considered as an


assemblage of a finite number of individual
structural components called “elements”.

• These elements can be put together in a


number of ways,
represent complex geometry.

M. Ben Ouezdou, University of Nizwa, 2011 3

Basic Concept
FEM based on “Principle of discretization”
=
procedure in which a complex problem of
large extent is divided (discretized) into
smaller equivalent units.

M. Ben Ouezdou, University of Nizwa, 2011 4

2
Application
• Application of this idea can be found everywhere in
everyday life and in engineering.
• Examples:

Lego (kid’s play) aircraft


M. Ben Ouezdou, University of Nizwa, 2011 5

Examples

beam bridge

M. Ben Ouezdou, University of Nizwa, 2011 6

3
Application to buildings

Steel frames

Concrete building
M. Ben Ouezdou, University of Nizwa, 2011 7

Why Finite Element Method?


• Design analysis: hand calculations and
experiments are sometimes difficult to
conduct Computer simulation.
• FEM can model complex geometry,
variable cross sections, various geometries,
different properties, …
• FEM can be closely integrated with CAD
applications.

M. Ben Ouezdou, University of Nizwa, 2011 8

4
Advantages of the FEM

1- Incorporate diff properties of each element.


2- No restriction for the shape of the medium;
hence arbitrary and irregular geometries
cause no difficulty.
3- Accommodation of any type of BC.
4- Handle non-linearities, time-dependant Pb.
5- Valid for any engineering Pb.

M. Ben Ouezdou, University of Nizwa, 2011 9

Some History of the FEM


• 1943: Courant (Variational Methods);
• 1956: Turner, Clough, Martin and Trop (Stiffness);
• 1960: Clough (“Finite Element”, plane problems);
• 1970s: Applications on mainframe computers;
• 1980s: Micocomputers, pre- and postprocessors;
• 1990s: Analysis of large structural systems.

M. Ben Ouezdou, University of Nizwa, 2011 10

5
Some Terminology
• FE: “small” elements (1D, 2D, 3D) obtained by
subdividing the given domain to be analyzed.
• Nodes or nodal points: intersections of the sides
of the elements.
• Nodal lines and nodal planes: interfaces
between elements.
• Linear elements: FE with straight sides.
• Higher order elements: FE with curved sides.
• Primary unknowns: nodal displacements
• Secondary unknowns: strains, stresses,
moments, shear forces, etc.

M. Ben Ouezdou, University of Nizwa, 2011 11

Basic procedure
• Step 1: Discretise the continuum: subdivide into elements: generate a mesh;
• Step 2: Select element displacement functions;
• Step 3: Calculate element properties: stiffness matrix [k].
• Step 4: obtain element load vector [F];
• Step 5: Assemble element properties (element stiffnesses global stiffness,
load vector).
• Step 6: Incorporate B.C. (set the element to the ground so disp = 0 or finite):
the stiffness matrix developed in step 5 will be modified to realize the
condition that disp of some coordinates = 0 or finite.
[F] = [K] {u} and {u}=[K]-1{F}. [K]: global stiffness matrix, [F]: vector of known forces
and {u}: displacements.
• Step 7: Determine displacements, strains and stresses
• Step 8: Check and iterate to eliminate precision errors if present.

M. Ben Ouezdou, University of Nizwa, 2011 12

6
Discretization
=
Process of separating the length, area or volume
into discrete (separate) parts or elements.
structure

1-D elements 2-D elements 3-D elements Axisymmetric


elements

M. Ben Ouezdou, University of Nizwa, 2011 13

1-D elements
• Used for beams or frames

Node 1 Node 2 3
1 2

Basic element H-element


1

Curved element
2

M. Ben Ouezdou, University of Nizwa, 2011 14

7
2-D elements
Used ,for plane-wall, diaphragm, slab, shell, etc.
1- Triangular elements

3 3
3
5
6
6 5

1 2 1 1 2
4 2
4
3 nodes 6 nodes 6- nodes curved triangle

- Triangular elements are the most used ones


- Curved elements for 2-D domain with curved boundaries

M. Ben Ouezdou, University of Nizwa, 2011 15

2-D elements
2- Quadrilateral elements

M. Ben Ouezdou, University of Nizwa, 2011 16

8
3-D elements
Used for the analysis of solid bodies (stresses under
foundation, contact stress under point loads, etc).

Tetrahedron Hexahedron Curved 3D element

Problem: Complex visualization and stiffness matrices size can be enormous.

M. Ben Ouezdou, University of Nizwa, 2011 17

Axisymmetric elements
Used in problems that are axisymmetric in nature.

Can achieve huge simplification in axisymmetric problems.

M. Ben Ouezdou, University of Nizwa, 2011 18

9
Guidelines for discretization
• Discretization is a major decision making step in FEM.
• Simple structures: no problems.
• Most real structures: difficulties in
– processing of subdividing the structure;
– Numbering the nodes;
– Assigning coordinates to each node;
– Relating the structure coordinate numbers to elements numbers and
their coordinate number.
• In most FEM Software: discretization is handled automatically
by the “preprocessing” module of the software.

M. Ben Ouezdou, University of Nizwa, 2011 19

Examples of discretization

5
1 2 3 4 5 3 4
1 2

1 2 3 4 1 2 3 4

Point loads: Stepped beam:


5 nodes and 4 elements 5 nodes and 4 elements

M. Ben Ouezdou, University of Nizwa, 2011 20

10
Examples of discretization
Material 1

Nodal line

Change in loading

Material 2

Change in material

M. Ben Ouezdou, University of Nizwa, 2011 21

Size of elements when discretize

• In recent past, the number of elements was limited due to


capabilities of both hardware and software. But nowadays a
sufficient fine discretization of a whole structure can easily be
produced very Quickly by graphical preprocessors.
• Most new FE softwares provide automatic mesh generation;
• But this tool should not be used in an uncaring manner:
engineering knowledge is still required.
– An inadequate modeling of apparently irrelevant details (e.g. small
cantilever slab or opening in a slab) can lead to faulty result and
unsafe design.
– A sufficiently FE mesh should be used in regions of high deformation
pr stress gradients.

M. Ben Ouezdou, University of Nizwa, 2011 22

11
FE Equation

M. Ben Ouezdou, University of Nizwa, 2011 23

Element stiffness matrix of a prismatic bar:


Direct method
ƒ1, u1 A,E ƒ2, u2
x
Node 1 Node 2

[F] = [k] [u]

Load vector stiffness displacement

ƒ1  EA  1  1  u1 
ƒ     1 1  u 
 1    2
M. Ben Ouezdou, University of Nizwa, 2011 24

12
Element stiffness matrix of a prismatic bar:
Formal approach
ƒ1, u1 A,E ƒ2, u2
x
Node 1 Node 2

Use conservation of energy: strain energy = work done by nodal forces

[F] = [k] [u]

 B   1 1 
where

k T
E B dV : element stiffness matrix, with B  
  
1  1
k  EA 
V

  1 1 
Same result !

M. Ben Ouezdou, University of Nizwa, 2011 25

Global stiffness matrix for a prismatic bar


E, 2A E, A
Example:
P x
1 2 3
1 2
Find stresses in the 2 bars
ℓ ℓ

Solution: Use two 1-D bar elements


u1 u2 u2 u3
1  1 EA  1  1
 1  2EA  k2 
  1 1    1 1 
Element 1: k Element 2:

u1 u2 u3
 2 2 0 
EA 
Global stiffness matrix K  
  2 3  1

 0  1 1 

M. Ben Ouezdou, University of Nizwa, 2011 26

13
Global FE equation
 F1   2  2 0  u1 
  EA   
F2     2 3  1 u2 
F    0  1 1  u 
 3    3
Use B.C. and Loads condition: u1 = u3 = 0. and F2 = P.

 F1   2 2 0  0 
  EA   
P    2 3  1 u2 
F    0  1 1   0 
 3   
P
P EA
3 u2  u2 
3 EA

u u E  P   P
 1  E 1  E 2 1    0  
   3 EA  3A
u3  u 2 E  P  P
2  E 2  E   0    
   3 EA  3A
M. Ben Ouezdou, University of Nizwa, 2011 27

v02
Bar element in 2-D x, u02

in global coordinates E, A
2


y, v01 Y, v
1 α
x, u
Element stiffness matrix in u0 1

global coordinates

u1 v1 u2 v2
c 2
cs  c  cs  u1
2
c = cos α
 
s 2  cs  s 2  v1
k    2
EA  cs
s= sin α
  c  cs c 2 cs  u2
 
  cs  s s 2  v2
2
cs
M. Ben Ouezdou, University of Nizwa, 2011 28

14
Element stiffness matrix of a beam:
Direct method
F1, v1 F2, v2

E, I 2
1
x
M1, θ1 M2, θ2
Element stiffness equation:

v1 1 v2 2
 F1   12 6  12 6   v1 
M   6 4 2  6 2 2   
 1   EI    1
 F2   3  12  6 12  6  v2 
     
M 2   6 2  6 4   2 

M. Ben Ouezdou, University of Nizwa, 2011 29

Element stiffness matrix of a beam:


Formal approach
Strain energy stored in, the beam element

Stiffness matrix for the simple beam element: k



k   BT EI B dx
0
with the strain-displacement matrix B:
 6 12 x 4 6 x 6 12 x 2 6x 
B   2  3   2  3   2
    
2
    
 12 6  12 6 
 6 4 2  6 2 2 
EI
obtain the same result k 3  
  12  6 12  6 
 
 6 2  6 4 
M. Ben Ouezdou, University of Nizwa, 2011 30

15
Local Stiffness matrix of a general 2-D beam element:
u1 v1 1 u2 v2 2
 EA EA 
  0 0  0 0 

 12 EI 6 EI 12 EI 6 EI 
 0 0  
 3 2 3 2 
 0 6 EI 4 EI
0
6 EI
 2
2 EI 
 2    
k  
EA EA
 0 0 0 0 
   
 0 
12 EI

6 EI
0
12 EI 6 EI 
 2
 3 2 3  
 6 EI 2 EI 6 EI 4 EI 
 0 0  2 
 2    
M. Ben Ouezdou, University of Nizwa, 2011 31

Element Stiffness matrix of a general 2-D beam


element in a global coordinate system:
v02
y, v x0, u02

2
E, A M2, θ2


y0, v01
α
1
u01 M1, θ1

x, u

M. Ben Ouezdou, University of Nizwa, 2011 32

16
u1 v1 1 u2 v2 2
 EA 2 12 EI 2 EA 12 EI 6 EI  EA 2  12 EI 2  EA 12 EI 6 EI 
  c  3 s 
cs  3 cs
 3
s

c 3 s
 
cs  3 cs
 2 
s
 EA 12 EI EA 2 12 EI 2 6 EI  EA 12 EI  EA 2  12 EI 6EI 
 cs  3 cs s  3 c c cs  3 cs s  3 cs c
     2     2 
 6 EI 6 EI 4 EI  6 EI EA 2  6 EI 2 EI 
 s c s c 2 s
k   2
 2
 2    
 EA 2  12 EI 2  EA 12 EI  6 EI EA 2  6 EI 2 EA 12 EI  6 EI 
 c 3 s cs  3 cs s c 2 s cs  3 cs s
     2     2 
  EA cs  12 EI cs  EA s 2   12 EI c 2  6 EI EA  12 EI
c cs  3 cs
EA 2 12 EI 2
s  3 c
 6 EI 
c
  3  3 2     2 
 6 EI 6 EI 2 EI  6 EI  6 EI 4 EI 
 s c s c 
  2
2  2 2  

M. Ben Ouezdou, University of Nizwa, 2011 33

Computation of nodal displacements


This step needs the use of the computer (mainly if stiffness matrix exceeds 5 x 5)
because it needs to invert the matrix.

{F} = [k] {u} {u} = [k]-1 {F}

Solution: enforcing 0 displacement BC and solve by:


- Unit diagonal method;
- Large diagonal method;
- Row column delete method.

M. Ben Ouezdou, University of Nizwa, 2011 34

17
Calculation of primary unknowns
Features of the assembled stiffness matrix
1) The stiffness matrix has its non-zeros terms along its
main diagonal (terms distant from the diagonal are 0).
2) Stiffness matrices are symmetric: advantage in storing
the matrices.
Half band width Values to be stored

5 3 4
-Reduction of the 5 0 6 2
3 4 0 0
required storage memory 3 6 7 0
2 0  7
-Reduction of the  4 3 5
solution time 4 7 4 0
3 5 k  
k   9 6 4
0 2 3 9
4 6
7 3 0
0 0 5 6 7 3  
  5 0 0
0 0 0 4 3 5
M. Ben Ouezdou, University of Nizwa, 2011 35

Calculation of primary unknowns


Solution of equations {u} = [k]-1 {F}

direct scheme: iterative scheme:


Linear problems Non Linear problems
(Gaussian elimination) (Jacobi, Gauss Seidal)

The most known methods of solutions are:


- Cholesky’s square root methods;
- Halfband Gauss elimination solution technique;
- Skyline technqiue
- Frontal solution technique.

M. Ben Ouezdou, University of Nizwa, 2011 36

18
Calculation of secondary unknowns
nodal displacement {u}

strain {ε}

stresses {σ} : (Hooke’s law)


Standard software give the output in a tabular form and on graphical form.

Results include:
- Refined colored graphics;
- Direct stresses σx, σy;
-Shear stresses xy; maximum shear, etc.

M. Ben Ouezdou, University of Nizwa, 2011 37

Summary:

Computer implementations
• Preprocessing (build FE model, loads and
constraints);
• FE Analysis solver (assemble and solve the
system of equations);
• Postprocessing (sort and display the results).

M. Ben Ouezdou, University of Nizwa, 2011 38

19
Conclusions:
Procedures of FEM in Structural Analysis
1) Divide structure into elements with nodes;
2) Describe the behavior of the physical quantities on each
element;
3) Assemble (connect) the elements at the nodes to form an
approximate system of equations for the whole structure;
4) Solve the system of equations involving unknown quantities
at the nodes (e.g., displacements);
5) Calculate desired quantities (e.g., strains and stresses) at
the selected elements.
N.B.: be aware of the limitations of the FEM: such as loads
application is imposed (no moving loads), and do not
misuse the FEM (it is a numerical tool).
M. Ben Ouezdou, University of Nizwa, 2011 39

References
1- Yijun Liu, “Introduction to Finite Element Method”,
Lecture notes, University of Cincinnati, Ohio, USA,
1998.

2- R. Vaidyanathan, P. Perumal, “Comprehensive


Structural Analysis”, 2nd ed., Laxmi Publication ed.,
New Delhi, 2008.

M. Ben Ouezdou, University of Nizwa, 2011

20

Anda mungkin juga menyukai