Chap 1 Introduction

AN INTRODUCTION TO THE
FINITE ELEMENT METHOD
J.P. PONTHOT
Aerospace & Mechanical Laboratory/LTAS-MN2L
1
University of Liège, Belgium
An introduction to Finite Element Method
Mechanics without Finite Elements is

like Rock’n Roll without Electricity
Juan C. SIMO
2
What Pierre Kroll thinks about my course (at least in 2011):
"E pur si muove " (and yet it moves), Galileo Galilei
2
CHAPTER 1:
INTRODUCTION TO THE FINITE ELEMENT METHOD
2
CURRENT APPLICATIONS OF THE FINITE

ELEMENTS IN MECHANICS
(Solids and Fluids)
64
Where the Course Fits:

Numerical Method in Mechanics
Computational Mechanics
The field of Mechanics cam be divided into 3 major areas:
Theoretical
Mechanics Applied
Computational
N.B. Mechanics is the branch of Physics concerned with the behavior of

bodies (movement and deformation) when submitted to forces
6
Computational Mechanics
Branches of Computational Mechanics can be distinguished according to the

physical focus of attention
Nano and Micromechanics
Continuum Mechanics:
Computational Solids and Structures
Mechanics Fluids
Multiphysics
Systems
7
Computational Solid and Structural Mechanics
A convenient subdivision of problems in Computational Solid and Structural

Mechanics (CSM) is:
Statics
Computational
Solid and Structural
Mechanics (CSM)
Dynamics
(“Theory of Vibrations”)
8
Computational Solid Mechanics (CSM)
A further subdivision of problems in CSM Statics is
Linear (the topic of this course)

(ELASTICITY AND SMALL STRAINS)
CSM Statics Nonlinear

•Material (“Advanced Solid Mechanics”)
•Geometry (“Large deformations of solids”)
•Contact (“Large deformations of solids”)
9
STRUCTURAL ANALYSIS
PDE- Partial Differential Equations

BC- Boundary Conditions
10
CONTINUUM MECHANICS
Local Partial Differential Equations (PDE) & Boundary conditions

(BC) describing:
•KINEMATICS
•CONSERVATION LAWS PDE.
•MATERIAL BEHAVIOR
•LOADS
•SUPPORTS BC
Infinite number of Degree Of Freedom (DOF)
Almost impossible to solve analytically for arbitrary shapes
11
MATHEMATICAL DISCRETE MODEL
Numerical approach (Computational Mechanics) 

Discrete approach
Algebraic equations
Finite number of DOF
PDE are solved approximately
Virtual analysis
FEM (Finite Element Method) is a toolkit
for such a numerical approach 12
MATHEMATICAL MODEL DEFINITION
Traditional definition
Scaled fabricated version of a physical system
(think of a car or a train model)
Simulation oriented definition

A model is a symbolic device built to simulate and predict
aspects of behavior of a system
Virtual analysis
13
STRUCTURAL ANALYSIS
Based on classical Mechanics of Materials (MoM)
IDEALIZATION
Limited to simple concepts, bar, trusses, beam, plate,
shell…
Limited combinations of structural concepts
COMPUTATION
“exact” solutions (but very simplified models…)
LIMITATIONS ADVANTAGES
Quasi-static cases Generality
Small displacements “formulas”
Elasticity (isotropic) Feeling for a designer
Mechanics of materials
(MoM)
14
STRUCTURAL ANALYSIS
Based on classical Mechanics of Materials (MoM)
PRE-COMPUTER AGE
(BUT STILL VERY USEFUL FOR PRE-DESIGN !!!!)
If current FEM solutions differs from MoM

by an order of magnitude
(or even a factor 2!!),
FEM is most probably wrong!
15
STRUCTURAL ANALYSIS
COMPLEXITY ARISES FROM:
•Complex geometrical shapes
•Complex material behavior
•Complex loading
•Boundary conditions (contact,…)
•Coupling with other phenomena (multiphysics)
•Design criteria
PRESENT CAD* SOFTWARE AREAS HELP ONLY IN

HANDLING THE GEOMETRICAL ASPECTS
IDEALIZATION
16
*CAD = Computer Aided design
THE IDEALIZATION PROCESS FOR A SIMPLE STRUCTURE
IDEALIZING
=
SIMPLIFYING!
Discrete unknowns= nodal displacements 17

THE DISCRETIZATION PROCESS
LOADS : Assumed to be known

BAR ELEMENT : Behavior to be determined
REACTIONS : To be determined
JOINTS OR NODES : Discrete unknowns = nodal displacements
18
COMPUTATIONAL SOLID MECHANICS

For the numerical simulation on the computer we must now
chose a spatial discretization method (transform PDE to
algebraic equations):
Finite Element Method (FEM)
Finite Difference Method (FDM)
Boundary Element Method (BEM)
Discrete Element Method (DEM)
Particle Finite Element Method (PFEM)
Finite Volume Method (FVM)
Spectral Method
Mesh-free or Meshless Method (EFG…)
eXtended Finite Element Method (XFEM)
Discontinuous Galerkin (DG) 19
WHAT IS A FINITE ELEMENT?

Tries to build an approximate (convergent) solution to the problem
Archimedes’ problem (circa 250 B.C.) rectification of
the circle as limit of inscribed regular polygons
Structure Idealization Element level

(approximations) (n elements) 20
MODEL PROBLEM
The basic idea is to divide the body (structure) into “finite elements”
(often called elements) of simple shapes -Think of a LEGO game –
connected by ”nodes” and try to obtain an approximate solution.
First step is to generate a Mesh or Grid that approximates the geometry
The process of making mesh = MESH GENERATION

CAD is most helpful in mesh generation! (However the link between CAD and FEM is still a bottleneck)
21
BASIC IDEAS OF FEM (1)

(Sometimes there are exceptions!)
Initial model is based on Continuum mechanics and is a set of
PDE +BC (e.g. heat conduction or equilibrium equation)
The body is divided into finite ( infinitesimal) elements of

simple shape (e.g. quads or triangles in 2D, tetrahedron or
hexahedron in 3D)
The vertices of the elements are called the nodes (this is not an
absolute rule).
The loads (heat fluxes) are applied at the nodes and we will try
to determine displacements (temperatures) at the nodes.
22

A field quantity such as temperature T (but it can also be the
displacement vector…) is interpolated by a polynomial over an
element (this is of course another approximation!!)
Values of the field at the nodes are called nodal values. They
become the basic unknowns or DOF (Degree of Freedom) of
the problem. Thus, there is now a finite number of DOF.
23

The elements are connected by nodes which insure that
the temperature or displacement field (still unknown!) is
continuous from one element to the adjacent one.
Finite Element = small piece of structure

Adjacent elements share the DOF at connecting nodes
• Now, if we can determine nodal values for T or u , the field is

easily interpolated anywhere inside the element (polynomial
24
interpolation!!)
Some examples of meshes:

When joined together elements can fill spaces
of almost arbitrary complexity
a) Blade of an aerospace engine

c) Drilling tool (Rocca, Sarrate & Huerta)
21
b) Woven fabric in a composite material (Boisse et al.)
Automatic meshers become more and more powerful:
Mesh of Romain Boman generated from a picture thanks to our mesher GEN4.
22

FEM provides a systematic methodology to evaluate T at the nodes by
transforming the initial PDE equations (infinite number of unknowns) to a
finite set of simultaneous algebraic equations (finite number of unknown, the
nodal values) formally written
Kq=g
where
1. q is an unknown vector containing all nodal temperatures
(dimension = number of DOF = NDOF)
1. K is the heat conductivity matrix (material and geometry dependent!)
2. g is the “action” vector resulting from imposed heat sources and fluxes
Accurate solutions (i.e. evaluation of K-1 )

 thousands of nodes are generally needed  COMPUTER !!
(1O6 DOFs  common 109  special applications)
27
THE KIND OF RESULTS YOU CAN OBTAIN
28 .)
Temperature field in a pump (Courtesy of Samtech S.A
THE KIND OF RESULTS YOU CAN OBTAIN
Front view Back view
Temperature field in a rocket engine (Courtesy of Samtech S.A

25
.)
FUNDAMENTAL CONCEPTS
The system of simultaneous NDOF algebraic equations
K q = g  q = K -1g
will play a central role all along this course. In general
1. K reflects the discretized properties of the media (geometry and material

properties). It is sometimes tedious to obtain, but it has to be defined.
2. g is the “action” vector describing the “generalized loads”. In this course,
we will assume that the loads are known.
3. q is a vector containing the NDOF nodal unknowns.
K & g are computed element by element (sometimes, it is easy, sometimes

not…) and then assembled into the structural system.
30
STRUCTURAL MECHANICS
Kq=g  q = K -1g
3 unknowns per nodes in 3D (2 in 2D):

the displacement components (ux, uy, uz)
1. K is the stiffness matrix that depends on the

geometry, the Young’s modulus and Poisson’s ratio
2. g is the load vector resulting both from volume loads
(self weight) and surface tractions.
3. q is the vector containing the nodal unknowns
(NDOF = 3 x number of nodes)
31
CONCEPTS-SUMMARY
FEM is a systematic way to approximate the solution of an infinite number of
PDE by transforming them into a finite number of algebraic equations
•FEM uses the concept of piecewise polynomial interpolation over a mesh

•By connecting elements together, the field quantities (displacement,
temperature…) become interpolated over the entire structure in a piecewise
continuous fashion.
•A set of simultaneous algebraic equations at the nodes has to be solved.
K q = g  simplest example
K = spring stiffness
g = applied load
q = end displacement 32
MODELING = SIMPLIFYING
Geometry, Material & Loads
Idealization
Actual structure Simplified CAD model

(holes, stiffeners & rivets) (Geometry and loads) 33
MODELING = SIMPLIFYING
Geometry, Material & Loads
Discretization
Idealized structure Discretized structure

CAD model Bar (8) & Membrane (6) elements
12 nodes, point loads
36 unknowns (nodal displacements)
30
Alternatively, Beam (8) and Shell (6) elements could have been used
VERIFICATION & VALIDATION OF THE FEM MODEL
DOF=DEGREE OF FREEDOM
NB. Prototypes are still necessary, but they appear in the last stages of design
35
FEM TRIANGLE
P.D.E.
FUNCTIONAL ANALYSIS
APPROXIMATION THEORY
VARIATIONAL METHODS
……..
SOFTWARE
MODELING
PHYSICAL PROBLEMS
COMPUTER
ENGINEERING “ART”
NUMERICAL ANALYSIS
MATERIAL DESCRIPTION
ALGORITHMS
C.A.D. MATERIAL SCIENCE
…….. 36
ADVANTAGES OF THE FEM

Can readily handle very complex geometry:
-The heart and power of the FEM !!!
Can readily handle very complex material behavior
Can handle a wide variety of engineering problems

-Solid mechanics -Dynamics -Heat transfer
-Fluid Flow (CFD) -Electrostatic problems -Electromagnetic problems
-Biomechanics -Fluid Structure Interaction
-Multi physics (FSI)
Very general & powerful

Little limitations but the cost
(Manpower is the major expense)
Sound Commercial Software tools 37
ADVANTAGES OF THE FEM
Can handle complex restraints

-Statically indeterminate structures can be solved without pain!
Can handle complex loading

-Nodal loads (point loads)
-Element loads (pressure, thermal, inertial forces)
-Time or frequency dependent loading
-Contact & friction
Today: Millions of engineers use FEM worldwide

on a daily base. 38
DISADVANTAGES OF THE FEM
Fancy, colorful
contours can
be produced by
any model,
good or bad!!
Temperature field in a pump (Courtesy of Samtech S.A.)

35
A general closed-form solution ( ), which would

permit one to examine system response to change in various
parameters, (i.e. sensitivity analysis) is not produced
ENGINEERING JUDGEMENT
STILL ESSENTIAL 40

 Black Box Syndrome & Overconfidence
 CAD, FEM & Computers are ALL MIGHTY
What they produce is correct.
ENGINEERING JUDGEMENT STILL ESSENTIAL

41

Hopefully under control!
The FEM obtains only ‘approximate’ solutions.
The FEM has ‘inherent’ errors:
Geometry is simplified
Material behavior is simplified
Field quantity is assumed to be a polynomial over an element.
This is not true in general.
Numerical techniques to obtain (Gauss quadrature) and invert K

finite number of digits ( round-off error)
ill conditioning 42

Hopefully under control!
Mistakes by users can be fatal
•DATA HANDLING
•Use of inconsistent units (e.g. E=200 GPa, Force=100 lbs)
•Bad idealization / wrong assumptions (element type, BC’s…)
•Distorted, skewed or bad aspect ratio elements in the mesh
•Coarse mesh
•no solution verification
THE USER MAY PUSH THE PROGRAM BEYOND ITS

RANGE OF VALIDITY!
WEAK KNOWLEDGE OF THE METHOD!!
ENGINEERING JUDGEMENT REMAINS ESSENTIAL 43


WEAK KNOWLEDGE OF THE METHOD AND
ITS LIMITATIONS CAN LEAD TO THIS!!
Collapse of a terminal at Paris-Charles-de-Gaulle airport, 2004
ENGINEERING JUDGEMENT REMAINS ESSENTIAL

40
RESPONSIBILITY OF THE USER
Figure!!!
BC: Hinged supports

Load: Pressure pulse
Unknown: mid point displacement
Results obtained from ten reputable FEM Displacement (mm) versus time (ms)
codes and by users regarded as experts.*
*P. Symonds & T. Yu Counterintuitive behavior in problem of Elastic-Plastic beam Dynamics.
45
ASME Journal of Applied Mechanics, Vol 52, No 3, 1985, page 517-522
BASIC FEM MODELING RULES
•Use the simplest elements that will do the job

•Never, never, never use complicated or special elements unless you are
absolutely sure of what you are doing
•Use the coarsest mesh that will capture the dominant behavior of the
physical model, particularly in design situations
3 word summary: Keep It Simple
46
A BRIEF HISTORY OF FEM

•FEM was first developed in the 1950’s in the aerospace industry
•Major industrials involved: Boeing & Bell Aerospace (long time vanished) in the
USA; Rolls-Royce in the UK, Snecma & Aérospatiale in France
•Most of the academic community first view FEM very skeptically, some
prestigious journals refused to publish papers on FEM.
FEM was first seen as an engineering trick!
47
A BRIEF HISTORY OF FEM (2)

•In 1956 M.J. Turner, R.W. Clough , H.C. Martin and L.J. Topp* published one of
the first papers that laid out the major idea (work actually performed in 1953 and
presented at a conference in 1954):
-they established the procedure of the element formulation and assembly
-they did NOT use the term ‘Finite Element’
•Ray Clough was a Prof. at the Berkeley and worked at Boeing for summer job. In
1960, he wrote a paper** where he coined the term ‘Finite Element’
•In 1960, Argyris and Kelsey published a book*** which was a collection of their
papers published in 1954 and 1955.
•First book on FEM by Zienkiewicz in 1967
*Stiffness and deflection analysis of complex structures, J. Aeronaut. Sci., 23, 805-823, 1956.
** The Finite Element Method in Plane Stress Analysis, Proceedings of 2 nd ASCE Conference on Electronic
Computation, Pittsburgh, PA, September 8-9, 1960. 44
***Energy Theorems and Structural Analysis, Butterworths, London
A BRIEF HISTORY OF FEM (3)
John Argyris (Stuttgart) Ray W. Clough (Berkeley) Olgierd Zienkiewicz (Swansea)
Argyris or Clough, who’s the father of

the Finite Element Method?
45
A BRIEF HISTORY OF FEM
•Subsequently (late 60’s), mathematicians (re)discovered a paper by COURANT
(1943)*, he used triangular elements to solve vibration problems. In this paper, he
wrote « We imagine a mesh of triangles covering the domain… »
many mathematicians have claimed that this was the original discovery of the
method.
•In 1973, FEM was provided a rigorous foundation (Strang & Fix, An Analysis of the
Finite Element Method, Prentice Hall)
Strang on YouTube: https://www.youtube.com/watch?v=WwgrAH-IMOk
•Mathematicians showed that (provided some conditions) for linear problems, finite
elements solutions converge to the exact solution of the PDE as the number of
elements is increased ( we also say that the mesh is refined)
Nowadays millions of engineers worldwide use the FEM to predict the behavior of
the structural, mechanical, thermal, material, electro-magnetical, biomechanical
and chemical systems! 50
*Variational methods for the solution of problems of equilibrium and vibrations. Bull. Am. Math. Soc., 42, 2165-86, 1943
FEM HAS PENETRATED MANY INDUTRIES
51
FEM HAS TRANFORMED THE ENGINEERING

WORKPLACE IN THE LAST 50 YAERS
•In the 1960s
-Drawing tables
-Simple Formulas (MoM)
-Prototypes were physically made and tested.
•Now, FEM had led to tremendous reductions in design cycle time in many
industries
WHY?
•Exponential growth in the speed of computers

•Decline in the cost of computational resources
52
EVOLUTION OF THE SPEED OF COMPUTERS
49
EVOLUTION OF THE SPEED OF COMPUTERS

•Gordon Moore’s law ( a founder of Intel in the 1990s) ‘
‘ The speed of computers doubles every 18 months’
•1 billion times faster in 40 years!
•June 2018 top computer Summit/IBM, Oak Ridge National lab. 266 PetaFlops
•June 2020 top computer FUGAKU/Fujitsu, RIKEN, 415 PetaFlops (7.3 million cores)
•2020 : China & France to reach EXA flop (1012 Mflops)(USA: 2021) delayed…
Mflops= Million Floating point

operations per second
Exaflop = 1000 billions of Mflops
= (1 million) 2 Mflops
54
COMPUTER COST EVOLUTION

•1Mflops in 1968 = $8 000 000
•1Mflops in 2003 = $0.5
(I Tflops = $500 000)
•Value of currency has been divided by 7 (inflation)
In 35 years, cost of computer power has decreased by

a factor of 100 millions!!
Microsoft joke: if the automobile industry had made the same

progress, your car would cost less than 1 cent!
Automotive joke: True, but if computer industry would manufacture

cars, it would lock up several times a day, and you
would need to press ‘start’ to stop your engine!!
55
(Data according to Hughes-Belytschko Nonlinear FEM short course)
HARDWARE EVOLUTION: Hard-disk 5Mb in 1956
52
HARDWARE EVOLUTION: 1Gb in 1988 and in 2005
53
FEM IN THE 1970’s IN LIEGE
The Aerospace laboratory of the University of Liège started to

develop FEM software in the 70’s:
B. Fraeijs de Veubeke Guy Sander
SAMCEF Software
(Système d’Analyse des Milieux Continus par Eléments Finis) 58
The Aerospace laboratory of the University of Liège

started to develop FEM software in the 70’s:
SAMCEF Software
(Système d’Analyse des Milieux Continus par Eléments Finis)
Main pioneering industrial partners:

Snecma & Aérospatiale
(Now: SAFRAN & EADS)
Commercialized since 1986 by

SAMTECH S.A., a spin-off company of the University
250 Engineers in the group!
55
56
57
58
59
60
61
62
63
CURRENT APPLICATIONS OF THE FINITE

ELEMENTS IN MECHANICS
(Solids and Fluids)
64
64
Walloon Share in Civil Aircraft Parts
Source: Skywin, Aerospace cluster of Wallonia www.skywin.be 64

SAMTECH software Applications in Aeronautics
Courtesy of D. Granville, Samtech S.A.

SAMTECH software Applications in Space

SAMTECH software applications in Automotive Industry

SAMTECH software applications in Machine Tool

Samtech software applications in Design & Verification of
Wind Turbines

Examples courtesy of SABCA S.A.
Fuselage parts
Main landing gear bay (Airbus A340)

Forward lower shell (Airbus A350) 70
Examples courtesy of SABCA S.A. : MLGB A350
77
Courtesy of Samtech S.A. Airbus A340
78
Courtesy of Samtech S.A. Airbus A340
73
Courtesy of Samtech S.A. Inflatable structure
74
Courtesy of CENAERO
Turbulent Flow Around a Nose Landing Gear

CFD = Computational Fluid Dynamics
75
Courtesy of CENAERO
Nose Landing Gear Geometry

76
Courtesy of CENAERO
Surface Mesh (800k nodes) 77

Courtesy of CENAERO
Surface Mesh (800k nodes) 78

Courtesy of CENAERO Volume Mesh (12M nodes)
79
Courtesy of CENAERO
Volume Mesh (12M nodes)
16mm
8mm
12 BL
(∆=0.5 mm)
(12 layers,
80
Courtesy of CENAERO
Instantaneous Vorticity Contours (12M)
81
Courtesy of CENAERO
Instantaneous Vorticity Structures (12M)
82
Finite element mesh generation for biomedical applications
Surface meshes
V. d’Otreppe, LTAS-MN2L
FIRST CONCLUSIONS
&
WARNINGS
90
ESSENTIAL REMARKS
 FE Modeling techniques are mostly deterministic: all correct geometrically,
physical, material, parameters must be given (INPUT of the problem). They
are generally not known.
 Optimization and Sensitivity analysis can (should) be performed to cover
variations (uncertainties) on input data.
 CAD should be connected easily to FEM. It is still a bottleneck!
 Always adapt model to the details to want to capture. Keep your model as
simple as possible…. But not too simple!
 Be aware of the limitations of the methodology.
Engineering judgment is still necessary at the computer age!!
91
As proposed by the project As understood by the project As designed by the senior analysts As produced by the programmers As corrected by the Business
sponsor managers Consultants
As documented in the paperwork As installed at the user’s site What has been charged on the As technical support has been What the user actually needed.
invoice contracted
92
93
BOOKS FOR FEM

Books for Linear FEM
Basic level (reference) :
ZIENKIEWICS & TAYLOR

The Finite Element Method, Vols I, II & III
Butterworth-Heineman, 2000
A comprehensive upgrade of the 1977 edition. Primarily an encyclopaedic

reference work that provides a panoramic coverage of FEM, as well as a
comprehensive list of references. Not a textbook. More and more editions to
appear.
88
•J. FISH & T. BELYTSCHKO

A First Course in Finite Elements
Wiley, 2007
•BECKER A.A.
An introductory guide to Finite Element analysis
Professional Engineering Publishing, 2004
•COOK R.D.
Concepts and Applications of Finite Element Analysis – Second Edition
John Wiley and Sons, 1981
•COOK R.D., MALKUS D. and PLESHA M.E.

Concepts and Applications of Finite Element Analysis – Third Edition
•COOK R.D.
Finite Element Modeling for Stress Analysis 89
•THOMSON E.G.
Introduction to the Finite Element Method
•AKIN J.E.
Finite Element Analysis for Undergraduates
Academic Press, 1986
•FELIPPA C.A.
Introduction to Finite Element Method
Ebook !!!
90
Intermediate level:
•HUGHES T.J.R.
The Finite Element Method
Prentice-Hall, 1987
Recently reprinted by Dover.
It requires substantial mathematical expertise on the part of the reader.
•BATHE K.J.
Finite Element Procedures
Prentice Hall, 1996
•MACNEAL R.H.
Finite Elements: their design and performance
Marcel Dekker, 1994 91
Mathematically oriented:
•STRANG & FIX

An analysis of the Finite Element Method
Prentice-Hall, 1973
Most readable mathematical treatment for non-mathematicians,
although outdated in several subjects.
Most fun (if you like British “humor”):

•IRONS & AHMAD
Techniques of Finite Elements
Ellis Horwood, 1986
92
BOOKS FOR NONLINEAR FEM
•BELYTSCHKO T., LIU W.K. & MORAN B.

Nonlinear Finite Elements for Continua and Structures
Wiley, 2000
93

Chap 1 Introduction

Uploaded by

Copyright:

Available Formats

Chap 1 Introduction

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chap 1 Introduction

Uploaded by

Copyright:

Available Formats

AN INTRODUCTION TO THE

FINITE ELEMENT METHOD

Mechanics without Finite Elements is

What Pierre Kroll thinks about my course (at least in 2011):

"E pur si muove " (and yet it moves), Galileo Galilei

INTRODUCTION TO THE FINITE ELEMENT METHOD

CURRENT APPLICATIONS OF THE FINITE

Where the Course Fits:

N.B. Mechanics is the branch of Physics concerned with the behavior of

Branches of Computational Mechanics can be distinguished according to the

Nano and Micromechanics

Computational Solid and Structural Mechanics

A convenient subdivision of problems in Computational Solid and Structural

Computational Solid Mechanics (CSM)

A further subdivision of problems in CSM Statics is

Linear (the topic of this course)

CSM Statics Nonlinear

PDE- Partial Differential Equations

Local Partial Differential Equations (PDE) & Boundary conditions

Infinite number of Degree Of Freedom (DOF)

Almost impossible to solve analytically for arbitrary shapes

MATHEMATICAL DISCRETE MODEL

Numerical approach (Computational Mechanics) 

Finite number of DOF

PDE are solved approximately

MATHEMATICAL MODEL DEFINITION

Simulation oriented definition

Based on classical Mechanics of Materials (MoM)

(BUT STILL VERY USEFUL FOR PRE-DESIGN !!!!)

If current FEM solutions differs from MoM

PRESENT CAD* SOFTWARE AREAS HELP ONLY IN

THE IDEALIZATION PROCESS FOR A SIMPLE STRUCTURE

Discrete unknowns= nodal displacements 17

THE DISCRETIZATION PROCESS

LOADS : Assumed to be known

COMPUTATIONAL SOLID MECHANICS

WHAT IS A FINITE ELEMENT?

Structure Idealization Element level

The process of making mesh = MESH GENERATION

BASIC IDEAS OF FEM (1)

The body is divided into finite ( infinitesimal) elements of

BASIC IDEAS OF FEM (2)

BASIC IDEAS OF FEM (3)

Finite Element = small piece of structure

• Now, if we can determine nodal values for T or u , the field is

Some examples of meshes:

a) Blade of an aerospace engine

Automatic meshers become more and more powerful:

BASIC IDEAS OF FEM (3)

Accurate solutions (i.e. evaluation of K-1 )

Temperature field in a rocket engine (Courtesy of Samtech S.A

1. K reflects the discretized properties of the media (geometry and material

K & g are computed element by element (sometimes, it is easy, sometimes

3 unknowns per nodes in 3D (2 in 2D):

1. K is the stiffness matrix that depends on the

•FEM uses the concept of piecewise polynomial interpolation over a mesh

Actual structure Simplified CAD model

Idealized structure Discretized structure

VERIFICATION & VALIDATION OF THE FEM MODEL