Page 1 of 89

_________________________________________________________________________________________________________________________________________________________________________
Course Title: FINITE ELEMENT METHOD I
Type of course: undergraduate, graduate
Field of study (Programme), specialization
Aerospace Engineering, Power Engineering, Robotics, Computer Aided Engineering
Format Lectures: Laboratory: Private study:
(Teaching methods) 2hours/week 1hour/week 2hours/week
Lecturers (course leader): Grzegorz Krzesiński

Objective: To supply the basic knowledge and skills required for understanding and simple practical applications of FEM

Contents (lecture’s programme):

1. INTRODUCTION TO FINITE ELEMENT METHOD
2. APPLICATIONS OF FEM
3. FINITE DIFFERENCE METHOD (FDM), BOUNDARY ELEMENT METHOD (BEM) AND FINITE ELEMENT METHOD (FEM)
4. BEAMS. RITZ-RAYLAYGH METHOD and FINITE ELEMENT METHOD
5. BARS AND SPRINGS
6. TRUSSES AND FRAMES
7. TWO AND THREE- DIMENSIONAL LINEAR ELASTOSTATICS
8. CST TRIANGULAR ELEMENT
9. 8-NODE QUADRILATERAL ELEMENT . NUMERICAL INTEGRATION
Computer lab: Introduction to practical problems of FE modeling in ANSYS/ 2D and 3D linear stress analysis/ Static analysis of simple shell structure/
Discretization error and adaptive meshing
Abilities: After completing the course the students will be able to build simple FE models and will know the possible applications and
limitations of the method in mechanics of structures.
Assesment method: Assesment based on tests and results of computer lab work (reports).
Practical work:Project/laboratory classes, where students will build and analyse the results of simple FE models of elastic structures
Recommended texts (reading):
[1] Huebner K. H., Dewhirst D. L., Smith D.E., Byrom T. G.: The finite element method for engineers, J. Wiley & Sons, Inc., 2001.
[2] Zagrajek T., Krzesinski G., Marek P.: MES w mechanice konstrukcji. Ćwiczenia z zastosowaniem programu ANSYS, Of.Wyd.PW 2005
[3] Bijak-Żochowski M., Jaworski A., Krzesiński G., Zagrajek T.: Mechanika Materiałów i Konstrukcji, Tom 2, Warszawa, Of. Wyd. PW, 2005
[4] Saeed Moaveni: Finite Element Analysis. Theory and Application with ANSYS, Paerson Ed. 2003
[5] Cook R. D.: Finite Element Modeling for Stress Analysis, John Wiley & Sons , 1995
[6] Zienkiewicz O.C., Taylor R.: The Finite Element Method.- different publishers and editions
 Page 2 of 89
_________________________________________________________________________________________________________________________________________________________________________

1. INTRODUCTION TO FINITE ELEMENT METHOD

The FEM is a numerical procedure that can be used to solve a large class of engineering problems including mechanics of structures, heat transfer,
electromagnetism, fluid flow and coupled fields problems (e.g. electro-thermal).
The simplest description.
The method involves dividing the geometrical model of the analysed structure into very small, simple pieces called finite elements, connected by
nodes. The behaviour of of the element is described by adequate physical laws. An unknown quantity (e.g. temperature, displacement vector,
electrical potential) is interpolated over an element from the nodal values using specially defined polynomials (called shape functions). The
procedure leads to the set of simultaneus algebraic equations with the nodal values being unknown.
During the solution process the nodal values (DOF- degrees of freedom of the model) are found. Then all interesting quantities ( strains, stresses)
are calculated within the elements. Finally the results may be presented in the required graphical form (the typical form of presentation is a contour
map)
MPa
MPa 0
0 -15
50 -30
100 -45
150 -60
200 -75
250
-90
300
350 -105
400 -120
450 -135

Finite Element Mesh Von Mises stress distribution Contact

u pressure between the shaft and the rotor disk

v
3D finite elements u

FEM was developed in 1950's for solving complex problems of stress analysis - mainly for aeronautical industry. The development of the method
was connected with the progress in digital computers and numerical techniques.
Today the method is considered as the most powerful analysis method for problems described by partial differential equations.
 Page 3 of 89
_________________________________________________________________________________________________________________________________________________________________________
FEM is one of the approximate methods for solving continuous problems of mathematics and physics

Approximate methods – flow chart

Real object
Real result wr

geometry, boundary (initial) conditions

material properties, laws pf physics

Methematical model Exact solution

(continous) of the mathematical model ws

discretization FINITE DIFFERENCE M. BOUNDARY ELEMENT M. FINITE ELEMENT M.

approximation FDM BEM FEM

Discrete Exact solution of the

discrete model wd Partial differential Boundary integral Functionals
model equations equations minimalisation

numerical calculations

Numerical result - wn

Discretization of the continuous problem – numerical estimation of the integral

 Page 4 of 89
_________________________________________________________________________________________________________________________________________________________________________

f(x)
The example – wooden board fi
Different models for the problem
f3
f2

f1
f0

one dimensional
a) model belki model - beam
N
q0
m h h h h
x0=a x 1 x 2 x i =a+ih b x

two- dimensional
b) model płyty model - plate
N
p0
m2

c) model trójwymiarowy
three dimensional bryły
model – solid volume

N
0
m3

number of DOF
 Page 5 of 89
_________________________________________________________________________________________________________________________________________________________________________
BASIC STEPS IN THE FINITE ELEMENT METHOD (FE modeling)
Preprocessor (preprocessing phase)

In the preprocessing phase the mathematical problem is described and presented in the numerical, discrete form:

Steps:
Description of :
the analysed domain (geometry of the analysis object)
the material properties
the boundary conditions (loads and constraints)
the meshing (dividing the domain into the finite elements of the required density distribution)

FE model of the bolted joint of the high pressure vessel

entire connection, representative part of the structure and its discretization, FE nodes with load symbols
 Page 6 of 89
_________________________________________________________________________________________________________________________________________________________________________
Processor (solution phase)

In this phase the user of the FE program defines the type of analysis (static, linear or nonlinear, dynamic, buckling ...... ) and other
details describing the method of calculations and solution process.
The FE program performs the calculations and writes the results in the adequate files.

Postprocesor

In this phase it is possible to present the interesting results in different forms: plots, graphs, animations, listings etc.
The user can create contour maps, tables, graphs and generate the reports.

Von Mises equivalent stress distribution (MPa). Contour map

 Page 7 of 89
_________________________________________________________________________________________________________________________________________________________________________
The results of FE analysis
Deformed model compared to undeformed structure

Displacement vector ( ux, uy , uz )

Stress state components within the model
{σx , σy , σz ,τxy , τyz ,τxz }

Z
σz

τ yz
τ xz
τ yz

τ xz

σy
τ xy τ xy

σx Y

Strain state components

{εx , εy , εz , γxy , γyz , γxz }
Principal stresses .

Equivalent stress distribution according to an arbitrary criterion

Any other entity defined by the user (ANSYS – APDL commands)
 Page 8 of 89
_________________________________________________________________________________________________________________________________________________________________________
2. APPLICATIONS OF FEM
Finite Element Analysis of Critical Central Connection Elements
of W7-X Stellator Coil Support System

The objective of Wendelstein 7-X project is the stellator-type fusion reactor. In this device plasma channel is under control of magnetic field coming from magnet
system of complex shape, made of 70 superconducting coils symmetrically arranged in 5 identical sections. Every coil is connected to central ring with two
extensions which transfer loads resulting from electromagnetic field and gravity.
During operation at a service temperature (ST) of 4K the superconducting coils of the W7-X magnet system exert high electromagnetic loads. Therefore, the
detailed analysis of the coil - central support connections, the so called Central Support Elements (CSE), is a critical item for W7-X. Each coil is fastened to the
CSS by two central support elements (CSE).
The aim of this work was to analyse mechanical behaviour of the bolted connections using detailed 3D finite element models (including bolts , washers, welds
etc). The Global Model of the structure, analysed by Efremov Institute in Russia, provided information about the loads acting on the connections.

Design analyses of the support structure elements: nonlinear simulations including contact with friction, plasticity,assembly stresses, submodelling technique
and using parametric models (14 bolted connections ) . The work performed for Institute of Plasma Physics, Greifswald , Germany.

The results of the numerical simulation help to check the magnitudes of displacements and stresses for different loading scenarios and
some modifications of the considered structures.
 Page 9 of 89
_________________________________________________________________________________________________________________________________________________________________________
material 1 --- 1.4429 steel: shim, wedges, ring,
1250

1125

1000

875

750

625

500

375

250

125

(x10**-2)
0
0 1 2 3 4 5
.5 1.5 2.5 3.5 4.5
EPS

Stress-strain curve for material 1 (1.4429 steel) corresponding to the temperature

293 K (red graph), 77K (violet), 4K (blue)
 Page 10 of 89
_________________________________________________________________________________________________________________________________________________________________________
Structural analysis and design of the "KLAUDIA" flight simulator
The FE model of the initial platform design has showed the
structure to be too flexibile. To find better solution the
simplified FE model has been built, easy for modifications.
The model has enabled quick verification of new concepts. The
final detailed FE model has confirmed the improvement of the
design. The fully nonlinear FE submodels have been built to
check the stress level in the main joints. Vibration
characteristics (natural frequencies and mode shapes) of the
structure have been found
The FE model was built using shell, solid, beam, mass and link
elements. The project was done for MP-PZL Aerospace Industries ,
Poland

Initial FE model

Modified (improved) design

submodel of the joint

 Page 11 of 89
_________________________________________________________________________________________________________________________________________________________________________
FE analysis of thin-walled elements' deformation
during aluminium injection moulding
Numerical simulations have been performed to model the process of filling the mould by hot aluminium alloy. The analysis has enabled
improvements of the element stiffness diminishing geometrical changes caused by the process. Fluid flow simulation with transient
thermal analysis including phase change have been performed, followed by the structural elasto-plastic calculation of residual effects.
The project performed for Alusuisse Technological Center, Sierre, Switzerland.

Temperature distribution (cooling effect) and displacements

FE model Velocity field during

injection
Residual stress distribution

FE model of the die

Velocity and temperature distribution
 Page 12 of 89
_________________________________________________________________________________________________________________________________________________________________________
FE analysis of a high pressure T-connection

The aim of the analysis was to find out stress and strain distribution in a T-
connection caused by high internal pressure (2600 at) and temperature gradients.
External cooling, assembly procedure (screw pretension), contact and plasticity
effects have been included. The project done for ORLEN petrochemical company
°C
190
205
220
230
250
265
280
295
310
Temperature distribution

MPa
0
50
100
200
300
400
FE model 500
600
700
Von Mises stress
 Page 13 of 89
_________________________________________________________________________________________________________________________________________________________________________

FE analyses of rotor disks

The aim of the analysis was to asses the right shape

details of the rotor to avoid high stresses and to find its
vibration characteristics. MPa
0
50
100
150
200
250
300
350
400
450
Von Mises stress distribution

FE mesh

MPa
0
-15
-30
-45
-60
-75
-90
-105
-120
-135

The mode shape for the natural frequency Contact pressure between the shaft
of 2203Hz and the rotor disk
 Page 14 of 89
_________________________________________________________________________________________________________________________________________________________________________

Thermo-electrical analysis of aluminium reduction cells

The analyses were performed to find temperature field and The influence of geometry, material properties and boundary
electrical potential distribution inside the reductant cell used conditions on the phenomena that take place in the bath and liquid
in the process of aluminium production. The project done aluminium is investigated. The analysis enabled to correct the design
for Alusuisse Technological Center, Sierre, Switzerland. and to improve efficiency of the processes.

Anode and cathode blocks

FE model (quarter of the cell)

Electric potential distribution

Temperature distribution
 Page 15 of 89
_________________________________________________________________________________________________________________________________________________________________________

FE analysis of the turbine blade locking part defects (imperfections)

Experiments show the presence of defects like surface scratch, or micro-crack in the region of blade locking part of the turbine
disks. Such imperfections may result in crack initiation and propagation. A segment of the turbine disk together with a blade has
been modelled (including contact). Half-elliptical crack has been introduced in the sub-model. Stress intensity factors and Rice
integral values have been calculated.

FE model Von Mises stress distribution

in the vicinity of the crack
 Page 16 of 89
_________________________________________________________________________________________________________________________________________________________________________

FEM Analysis of the Winch Frame and Boom of the snow groomer

The aim of the analysis was to check the stiffness and stress level of the new design of the
structure. Numerical model consisted of FE shell elements supplemented by brick, beam,
link and mass elements. In regions of special care sub-models were used involving contact
elements. The results suggested essential changes of design. The project done for
PLUMETTAZ S.A., Bex, Switzerland

Displacements

Initial model

Von Mises stress

FE model
Modified design
 Page 17 of 89
_________________________________________________________________________________________________________________________________________________________________________

CAD/CAE study of a New Design of Truck Frame

FE model Von Mises stress

CAD project
 Page 18 of 89
_________________________________________________________________________________________________________________________________________________________________________
Finite element method in bone-implant system strength analysis

The three-dimensional FE models of the living tissues-implant systems can deliver the valuable information about
mechanisms of stress transfer and healing processes after the orthopaedic surgery. In the presented example some different variants
of the hip stem were considered to find the best solution, which should reduce stress concentration within the bone tissues. The model
of the femur was built using the data obtained from CT scans. The considered load corresponds to one leg stance of a man weighting
800N.

α1

F1
trabecular bone
cement
stem α2
cortical bone

contour points from the CT scans F2

Finite element model of the femur endoprosthesis : body weight BW=800N, F1=2.47 BW, F2=1.55BW, α1=28°, α2= 40°
 Page 19 of 89
_________________________________________________________________________________________________________________________________________________________________________

FE model of the spine stabilizer and the von Mises stress distribution within the frame

Von Mises Equivalent stress (MPa)

FE model and selected results of numerical simulation of mandibular ostheosynthesis
 Page 20 of 89
_________________________________________________________________________________________________________________________________________________________________________
3. FINITE DIFFERENCE METHOD (FDM) BOUNDARY ELEMENT METHOD (BEM)
AND FINITE ELEMENT METHOD (FEM)
Draft presentation for solving Poisson’s equation in 2D space
,
Poisson's equation is a partial differential equation with broad utility in electrostatics, mechanical engineering and theoretical physics.

∂ 2 u ∂ 2u
+ 2 + f ( x1 , x2 ) = 0
∂x1 ∂x2
2
,
For vanishing f, this equation becomes Laplace's equation.

We consider a Dirichlet boundary condition on Γu and a Neumann boundary condition on Γq:

x2
(y) n2
n
n1

u ( x ) = u0 , x ∈ Γu Ω
Γq
∂u ( x )
q( x) = = q0 , x ∈ Γq Γu
∂n

x1
(x)
where u0 and q0 are given functions defined on those portions of the boundary.
In some simple cases ( shape of the domain Ω and boundary conditions) the Poisson equation may be solved using analytical methods.
 Page 21 of 89
_________________________________________________________________________________________________________________________________________________________________________
Finite Difference Method
Finite-difference method approximates the solution of differential equation by replacing derivative expressions with approximately equivalent
difference quotients. That is, because the first derivative of a function f (x) is, by definition,

then a reasonable approximation for that derivative would be to take

(difference quotient)
for some small value of h. Depending on the application, the spacing h may be variable or held constant.
The approximation of derivatives by finite differences plays a central role in finite difference methods
In similar way it is possible to approximate the first partial derivatives using forward , backward or central differences

y
h h
xi = xo + ih,
ui,k+1 ui ,k = u ( xi , yk )
u i+1,k yk = yo + kg ,
uui,k-1
i-1,k u i,k ui+1,k
u i,k+1
g
∂u ∆u ui ,k +1 − ui ,k
yk a) ≈ = ,
uui,k-1 g ∂y ∆y g
i-1,k
∂u ∆u ui ,k − ui ,k −1
b) ≈ = ,
∂y ∆y g
∂u ∆u ui ,k +1 − ui ,k −1
y c) ≈ = .
0 ∂y ∆y 2g

xo , yo - reference x 0
point x i function
of the grid, u ( x, y ) - unknown x
 Page 22 of 89
_________________________________________________________________________________________________________________________________________________________________________
Differences corresponding to higher derivatives

∂ 2u ∆ 2u ui +1,k − 2ui ,k + ui −1, k

≈ = , ∂ 4u ∆ 4u ui + 2, j − 4ui +1, j + 6ui , j − 4ui −1, j + ui −2, j
∂x 2 ∆x 2 h2 ≈ =
∂x 4 ∆x 4 h4
∂ 2u ∆ 2u ui ,k +1 − 2ui ,k + ui , k −1
≈ = .
∂y 2 ∆y 2 g2

Using the finite differences we can approximate the partial differential equation at any point (xi, yj ) by an algebraic equation .
In the case of Poissons equation:

2 ( i +1, j
u − 2ui , j + ui −1, j ) + 2 ( ui , j +1 − 2ui , j + ui , j −1 ) + f ( xi , y j ) = 0 .
1 1
h g
If h = g i f ≡ 0 (Laplace equation) we get
u +u +u +u
ui , j = i +1, j i −1, j i , j +1 i , j −1 .
4
N grid points in the domain Ω , N equations, N unknows

[ A] {u}= {b}
discrete form of boundary conditions
a) b)
In the case of irregular boundary shape
hu0 + δ u2
δ a) assumed u1 = instead of u = u0
2 1 0 2 0 1 h +δ
δ
hu0 − δ u2
h

h b) assumed u1 = instead of u = u0
h −δ
 Page 23 of 89
_________________________________________________________________________________________________________________________________________________________________________
y y

x x
c)
y

d) y

θ r

e) h

6
g
l

1 0
2

5
 Page 24 of 89
_________________________________________________________________________________________________________________________________________________________________________

Boundary Element Method

Uses the boundary integral equation ( equivalent to the Poisson’s equation with the adequate b.c.)

∂u( x ) ∗
c(ξ )u(ξ ) = − ∫ u( x ) q∗ (ξ , x )d Γ( x ) + ∫ u (ξ , x )d Γ( x ) + ∫ f ( x )u∗ (ξ , x )dR( x )
Γ Γ
∂n Ω x2

c(ξ ) - coefficient equal to 1/2 on the smooth contour, 1 inside the domain Ω
n2
n
Ω n1
r2
1 1 r
x
Kernel functions u = (ξ , x ) =
∗
ln   , (36)
2π  r 
r1 Γq
r = ( x1 − ξ1 ) + ( x2 − ξ 2 ) .
2 2
Γu ξ

∂u ∗ (ξ , x ) Γu Γq =Γ
q∗ (ξ , x ) = . (37)
∂n
x1
∂u ∗
∗ ∂u ∗
q = ⋅ n1 + ⋅ n2 ,
∂x1 ∂x2 ∂r xi − ξi ri
= = .
−(r1 ⋅ n1 + r2 ⋅ n2 ) ∂xi r r
q∗ = ,
2π r 2
∂u ( x )
The boundary integral equation states the relation between u ( x ) and its derivative in normal direction q ( x ) = on the boundary Γ.
∂n
 Page 25 of 89
_________________________________________________________________________________________________________________________________________________________________________

The numerical approach

1. Discretization of the boundary (LE boundary elements)
x2
2. Approximation of u ( x ) and q ( x ) on the boundary
(e.g. u(Pi), q(Pi) constant on boundary elements)
3 . Building the set of linear equations
Pj
LE LE
1
u ( Pi ) = ∑ ∫ u ∗ ( Pi , x )q ( Pj )d Γ j − ∑ ∫ q ∗ ( Pi , x )u ( Pj )d Γ j
2 j =1 Γ j j =1 Γ j
r

+ ∫ f ( x )u ∗ ( Pi , x )dR i = 1, 2,..LE
Ω

1 LE LE Pi
u ( Pi ) = ∑ U ij∗ ⋅ q( Pj ) − ∑ Qij∗ ⋅ u ( Pj ) + f i , i = 1, 2...LE . f i = ∫ f ( x )u ∗ ( Pi , x )d Ω( x )
2 j =1 j =1 Ω
x1

1
{u} = U ∗  {q} − Q∗  {u} + { f } .
2
LE linear equations with the unknows u ( Pj ) (if the point Pj Œ Γ q ) or q( Pi ) (if Pi Œ Γu )

Finally: [ A] { y}= {b}

The solution {y} represents unknown boundary values of u and q.
The matrix A – full, unsymmetric
4. Solution - provides complete information about the function u ( x ) and its derivative q ( x ) on the boundary
Boundary Element Method reduces the number of unknown parameters (DOF of the discrete model) in comparison to FDM and FEM ( domain
methods).
 Page 26 of 89
_________________________________________________________________________________________________________________________________________________________________________
Finite Element Method
nodes elements
Equivalent problem of minimising of the fuctional:

1  ∂u   ∂u  
2 2
x2
I (u ) = ∫   +  − 2 f ( x , x )u d Ω − ∫ q0ud Γ, węzły elementy
2 Ω  ∂x1   ∂x2 
1 2
 
 Γq
domain
with the Dirchlet b. c.
u ( x ) = u0 , x ∈ Γu obszar
e

1. Discretization of the solution domain Ω into elements Ωi , i=1,LE,

contour
connected in the nodes
LE
kontur
Ω = ∪ Ωe i Ωi ∩ Ω j = 0 i ≠ j,
i =1

2.Approximation of function u ( x ) within the finite element in the form of polynomials dependent on the unknown nodal values ui
x1
u(x1,x2)
LWE
approximation of the
u ( x1 , x2 ) = ∑ N ( x , x )u
i 1 2 i
i =1 u5
function u(x,y) over
the element Ωe
LWE – number of nodes of the element x2

ui , i = 1,...,LWE - nodal values of the approximated function, u6

5
u4
Ni(x1,x2) – shape functions 6 4
u3
u7
3. Discrete form of the functional e
3
7 u2
u8
1  ∂u   ∂u  
2 2
LE LK
I (u ) ≅ ∑ ∫    +  − 2 f ( x , x )u d Ω − ∑ ∫ q0ud Γ j 8 2
i =1 2 Ωi  ∂x1  ∂x2 
1 2 i
  
 j =1 Γ j
1 u1 LWE=8

x1
 Page 27 of 89
_________________________________________________________________________________________________________________________________________________________________________
In each element
∂u LWE ∂N i
=∑ ui ,
∂x1 i =1 ∂x1
∂u LWE ∂N i
=∑ ui .
∂x2 i =1 ∂x2

In this way the functional I is replaced by the function of several unknows ui , i = 1, 2,..., LW , where LW denotes the number of nodes of the finite
element mesh. In the matrix form :
LW

 k11 k12 k13 … k1LW   u1   b1  m

   
 k21 k22 k23   u2   b2 
1   elementy
I (u ) ≈ u1 , u2 , u3 ,..., uLW   k31 k32  u3  − u1 , u2 , u3, … , uLW   b3  zerowe
2  
  …  …
  
…
 k LW 1 k LW LW   uLW   bLW 


I≈
1
u  K [ ]{u} − u {b} .
2 1× LW LW × LW LW ×1 1× LW LW ×1

Necessary (and sufficient) condition of the minimum:

∂I
= 0, i = 1,… , LW . matrix: sparse, symmetrical, positive defined, banded
∂ui
Hence

[ K ]{u} = {b} , (+ Dirichlet b.c.)

Set of the simultaneuous equations with unknown nodal values of the investigated function.
 Page 28 of 89
_________________________________________________________________________________________________________________________________________________________________________

4. BEAMS
RITZ-RAYLAYGH METHOD and FINITE ELEMENT METHOD

Principle of minimum potential energy.

The potential energy of an elastic body is defined as

Total potential energy (V)= Strain energy (U) − potential energy of loading (Wz)

In theory of elasticity the potential energy is the sum of the elastic energy and the work potential:
1
V = U − Wz = ∫ σ ijε ij d Ω − ∫ X i ui d Ω − ∫ pi ui d Γ
2Ω Ω Γ

Ω – domain of the elastic body, Γ – boundary, σ ij – stress state tensor, ε ij – strain state tensor ,
ui – displacement vector, pi – boundary load (traction), Xi– body loads
The potential energy is a functional of the displacement field. The body force is prescribed over the volume of the body, and the traction is
prescribed on the surface Γ . The first two integral extends over the volume of the body. The third integral extends over the boundary.

The principle of minimum potential energy states that,

the displacement field that represents the solution of the problem fullfills the displacement boundary conditions and inimizes
the total potential energy.

V = U − Wz = min! ,
 Page 29 of 89
_________________________________________________________________________________________________________________________________________________________________________

Total potential energy of the beam loaded by the distributed load p   :

N
m
l l
1
V = ∫ EI ( w′′) 2 dx − ∫ pwdx ,
20 0

where the function w(x) describes deflection of the beam

Ritz method

1.The problem must be stated in a variational form, 3. The parameters ai are determined by requirement that the
as a minimization problem, that is: functional is minimized with respect to ai
find w(x) minimizing the functional V(w) ∂V
= 0, i = 1,..., n .
∂ai
2. The solution is approximated by a finite linear combination of the
form:
n
[ A] {a}= {b}
wɶ ( x) = ∑ aiϕ i ( x)
i =1 4. The solution provides ai, and the approximate solution
where ai denote the undetermined parameters termed Ritz n
coefficients, wɶ ( x) = ∑ aiϕ i ( x) .
and φi are the assumed approximation functions (i=1,2...n). i =1
The approximate functions φi must be linearly independent
Hence the approximate internal forces in the beam
and
Mɶ q ( x) = EIwɶ ′′( x),
3. Finally functional V is approximated by the function of n variables
ai Tɶ ( x) = − EIwɶ ′′′( x).
V=V(a1, a2 , a3 , ....an )
 Page 30 of 89
_________________________________________________________________________________________________________________________________________________________________________
EXAMPLE
N 
Find the deflection of the cantilever beam under the load p0   using the analytical solution of the differential equation and compare it to the
m
approximate solution using Ritz method ~
with the function w( x) = a1 + a2 x + a3 x 2 + a4 x 3 .

w(x)

p0
x

Exact analytical solution

M ( x) p0
w′′( x) = g M q ( x) = (l − x ) 2 ,
EI 2

w( x )
w( x ) = 0 , d =0
dx
Solution
p0
w( x) = (6l 2 − 4lx + x 2 ) x 2 ,
24 EI
p p0 l2 /8 EI
Max. deflection w(l)=

~ ( x) = a + a x + a x 2 + a x 3 has to satisfy the displacement boundary conditions

The approximate solution w 1 2 3 4

wɶ ( x = 0) = 0, wɶ ′( x = 0) = 0 .
Thus
~ ( x) = a x 2 + a x 3 .
w 3 4

EI l3 l4
V= (4a32l + 12a3 a4l 2 + 12a42l 3 ) − p(a3 + a4 ) .
2 3 4
 Page 31 of 89
_________________________________________________________________________________________________________________________________________________________________________

∂V EI p0l 3
∂a3
=
2
(8la3 + 12l a4 ) − 3 = 0,
2 W(x)
~
W(x)
-2 pl
* 10 EJ 12.5

∂V EI p0l 4
∂a4
=
2
(12l a3 + 24l a4 ) − 4 = 0.
2 3
8.35
W(x)
~
2 W(x)
5 p0l pl
a3 = , a4 = − 0
12.5
4.427
8.203
24 EI 12 EI
4.167
1.318
1.172
Finally the approximate solution is
l l 3l x
l
4 2 4

5 p0l 2 2 p 2

wɶ ( x) = x − 0 x3 ,
Mg(x) * pl
~
Mg(x) 0.5
24 EI 12 EI Mg(x)
~
Mg(x)
5 pl 0.417 0.292
Mɶ q ( x) = p0l 2 − 0 x, 0.167
12 2 0.281 0.42
0.125
− p0l
Tɶ ( x) = . l
4
l
2
3l
4
l
x

2 0.031 -0.083
T(x) * pl
~
T(x) l l 3l
l
4 2 4
Graphs presenting exact (bold line) and approximate (dashed line ) solutions x
of the cantilever beam: 0.5
displacement, bending moment, shear force ~
T(x)
1 T(x)
 Page 32 of 89
_________________________________________________________________________________________________________________________________________________________________________
Finite Element Method approach

Approximation : local, with nodal displacements w1 , w2 , θ1 and θ 2 as unknown parameters

w1=q 1 w2 =q 3 Positive directions:

w( ) upward for translation
1 =q2 2 =q4 counter clockwise for rotation

1 2
le

Simple beam finite element

Lets assume first the polynomial approximation within the element
w(ξ ) = α1 + α 2ξ + α 3ξ 2 + α 4ξ 3

with four unknown parameters α i .

The required new parameters : nodal displacements w1 , w2 , θ1 and θ 2 (degrees of freedom – DOF of the element)

 q1   w1 
q   θ 
{q}e =  2  =  1  .
4

Nodal displacement vector w(ξ ) = ∑ N i (ξ )qi .

q3  w2  i =1
q4  θ 2  w(ξ ) =  N (ξ )  {q}e ,
e
 Page 33 of 89
_________________________________________________________________________________________________________________________________________________________________________
Relation between α1 , α 2 , α 3 ,α 4 and q1 , q2 , q3 , q4

q1 = w(0) = α1 ,
displacement and node 1
dw
q2 = (0) = α 2 ,
dξ slope at node 1

q3 = w(l ) = α1 + α 2le + α 3le2 + α 4le3 , displacement at node 2

dw
q4 = (l ) = α 2 + 2α 3le + 3α 4le2 . slope at node 2
dξ  
1 0 0 0
In the matrix form  
α1    q 
 q1   1 0 0 0  α1  α   0 1 0 0  1
q     2    q2 
 2  0 1 0 0  α 2    = −3 −2 −1  q3
α3   2
3
 =  .  
 q3   1 le le2 le3  α 3  α 4   le l le2 le   q4 
   e
q4   0 1 2le 3le2  α 4  2 1 −2 1
  l3
 e le le3 le2 
The approximate deflection may be presented in the form
ξ2 ξ3
N1 (ξ ) = 1 − 3 +2 ,
le2 le3
α1   q1  ξ2 ξ3
α  q 
N 2 (ξ ) = ξ − 2 + ,
3  2   le le2
w(ξ ) = 1, ξ , ξ , ξ    =  N1 (ξ ), N 2 (ξ ), N 3 (ξ ), N 4 (ξ )   2  ,
2

α 3   q3  ξ2 ξ3
N 3 (ξ ) = 3 −2 ,
α 4   q4  le2 le3
−ξ 2 ξ 3
N 4 (ξ ) = + 2.
le le

The functions N i (ξ ) are called shape functions of the beam element.

 Page 34 of 89
_________________________________________________________________________________________________________________________________________________________________________

N i (ξ ) describes deflection of the beam element , where qi = 1 , and for j ≠ i q j = 0 (see graphs).

N( ) N( )
1 3

1 1

e e

N( ) N( )
2 4

tg =1 tg =1

e e

Shape functions of a beam element

w(ξ ) =  N (ξ )  {q}e ,
w′(ξ ) =  N ′(ξ )  {q}e ,
w′′(ξ ) =  N ′′(ξ )  {q}e .

Total potential energy of the beam element of the length le

l l
EI e e

Ve = U e − Wze = ∫ ( w′′(ξ )) dξ − ∫ p (ξ ) w(ξ )dξ −

2
. i i −
Pw M jϑ j
2 0 0
 Page 35 of 89
_________________________________________________________________________________________________________________________________________________________________________
l l
EI e EI e
Ue = ′′ ξ ′′ ξ ξ =  q  e { N ′′}  N ′′ {q}e dξ =
2 ∫0 2 ∫0
w ( ) w ( ) d

 N ′′ N ′′ N1′′ N 2′′ N1′′ N 3′′ N1′′ N 4′′ 

 1 1 
le 
EI N 2′′ N1′′ N 2′′ N 2′′ N 2′′ N 3′′ N2 N4 
′′ ′′
=  q  e ∫  dξ {q} .
0  N ′′ N ′′ N 3′′ N 2′′ N 3′′ N 3′′ N 3′′ N 4′′ 
e
2
 3 1 
 N ′′ N ′′ N 4′′ N 2′′ N 4′′ N 3′′ N 4 N 4 
′′ ′′
 4 1

 le le le le

 ∫ N1′′ N1′′d ξ ∫ N1′′N 2′′dξ ∫ N1′′N 3′′dξ ∫0 N 1
′′ N ′′d ξ 
4
0 0 0 
 le le le le 
 N 2′′ N1′′d ξ
1
U e =  q  e [ k ]e {q}e ,  ∫0 ∫ N ′′N ′′dξ ∫ N ′′N ′′dξ
2 2 2 3
′′ ′′
∫0 N 2 N 4 dξ 
2 [k ]e = EI  l
0
le
0
le le 
 N ′′ N ′′d ξ 
e

∫ 3 1 . ∫ N 3′′N 2′′dξ ∫ N ′′N ′′dξ

3 3
′′ ′′
∫0 N 3 N 4 dξ 
0 0 0

 le le le le

 ∫ N 4′′ N1′′d ξ ∫ N 4′′N 2′′dξ ∫ N ′′N ′′dξ
4 3
′′ N ′′d ξ 
∫0 4 4 
N
 0 0 0

Matrix [ k ]e is named stiffness matrix of beam element. After integration

 6 3le −6 3le 
 2
−3le le2 
2 EI 3le 2le
[ k ]e = 3
le  −6 −3le 6 −3le  .
 
2
3le le −3le 2le2 
 Page 36 of 89
_________________________________________________________________________________________________________________________________________________________________________

The external work done by the traction p:

le le le

W = ∫ p(ξ ) w(ξ )dξ = ∫ p(ξ )  N (ξ )  {q}e dξ = ∫  N1 (ξ ) p(ξ )dξ , N 2 (ξ ) p (ξ )dξ , N 3 (ξ ) p (ξ )dξ , N 4 (ξ ) p(ξ )dξ  {q}e dξ ,
p
ze
0 0 0

 q1 
q  le
 
Wze =  F1 , F2 , F3 , F4   2  =  F  e {q}e
p e e e e
Fi = ∫ N i (ξ ) p (ξ ) dξ
e
e q ,
 3 0
 q4 

Fi e - equivalent nodal forces

F1=3p0 le/ 20 F3= 7p0 le/ 20
p0le F2=p0 le2/ 30 F4= -p0 le2/ 20
F1e = F3e =
p0 e p0 e 2
2 2 2 2 p0le2
p
0 e p 0 e F2e =
12 12 12
1 2
− p0le2
1 2 F4e =
p0 12 P0

Equivalent nodal forces corresponding to the constant and linear distribution of p0 load
(kinematically equivalent or work-equivalent !)

Total potential energy of the beam element

q  e [ k ]e {q}e −  q  e{ F }e .
1
Ve = U e − Wze = 
2 1×4 4×4 4×1 1×4 4×1
 Page 37 of 89
_________________________________________________________________________________________________________________________________________________________________________

The conditions for finding the minimum of Ve:

∂Ve
=0, i = 1, 2,3,… , n
∂qi

6 3le −6 3le  q1   F1 
2
−3le
2 q   F 
2 EI 3le 2le le  2  2
  = 
[ k ]e {q}e = {F }e . le −6 −3le −3le
3
6  q3   F3 
3le le
2
−3le 2l e
2  q4   F4 
e e

Set of linear equations for one element model of the considered cantilever beam:
 Page 38 of 89
_________________________________________________________________________________________________________________________________________________________________________

 F1 
6 3l −6 3l  0   F 
   2 
2 EI 3l 2l
2
−3l 2
l 0  p l 
 = 0 
l 3 −6 −3l 6 −3l  q3   2 
3l l 2 −3l 2l 2 q4   − p0l 2 
 
 12 
Constraints q1=0 and q2=0 may be taken into account by
 F1 
F 
 
the transformation of the set of equation to the form [ A]  2  = {b} or by reductionof the problem to
 q3 
 q4 
2 EI pl
3
(6q3 − 3lq4 ) = 0 ,
l 2
2 EI − p0l 2
( −3lq3 + 2l q4 ) =
2
,
l3 12
1 p0l 4
q3 =
8 EI
Solution is:
1 p0l 3
q4 =
6 EI
 3 1  p0l 2  −2 1  p0l 3 5 p0l 2
2 2
pl
Finally the deflection function from the one element model is w(ξ ) =  −  ξ + +  ξ = ξ − 0 ξ3
 8 6  EI  8 6  EI 24 EI 12 EI
The same result as obtained in the case of Ritz method – why?
 Page 39 of 89
_________________________________________________________________________________________________________________________________________________________________________

Dividing the beam into LE elements

q1 q3 q5 q7
 q1   w1  q2 q4 q6 q8
q   θ  1 1 2 2 3 3 4
 2  1  e e e

 q3   w2  p0
   
 q  θ  P
global nodal displacements vector {q} =  4  =  2  . e
=
3
 q5   w3  M
 q6  θ 3 
   
q7   w4 
 q  θ 
 8  4 
N=8 nodal diplacements (degrees of freedom of the FE model)

Strain energy U e of each of the elements

q [ k * ] {q} ,
1 1
Ue =  q  [k ]e {q}e e = 
2 1×N N ×N e N ×1
2 e
1x4 4x4 4x1

 k ∗  =  k ∗  =  k ∗  =
1 2 3

element 1 with the global DOF : element 2 with the global DOF : element 3 with the global DOF :
q1, q2 , q3, q4 q3, q4 , q5, q6 q5, q6 , q7, q8
 Page 40 of 89
_________________________________________________________________________________________________________________________________________________________________________
LE
1  LE ∗  1
U = ∑ U e =  q   ∑  k   {q} =  q  [ K ]{q} .
e =1 2  i =1 e
 2
1
V = U − Wz =  q  [ K ]{q} −  q  { F } ,
2
∂V
= 0, i = 1, 2,3,… , n
∂qi

[ K ]{q} = {F } . + displacement boundary conditions (constraints)

For each element the internal forces M,T are calculated separately:

 q1 
q 
 ′′ ′′ ′′ ′′   
M q (ξ ) = EIw′′(ξ ) = EI N1 , N 2 , N 3 , N 4  2  ,
  q3
 
 q4  M (ξ ) = 12 (ξ − le )q + 6 (ξ − 2 l )q − 12 (ξ − le )q + 6 (ξ − le )q  EI ,
e q  3 1
le2
e 2
le3
3
le2 3 
4
 le 2 3 2
 q1 
q  T (ξ ) = −  12 ( q − q ) + 6 ( q + q )  EI .
   l3 1 3 l2 2 4 
T (ξ ) = − EIw′′′(ξ ) = EI  N1′′′ , N 2′′′ , N3′′′ , N 4′′′   2  .  
  q3 e e

 
q4 
e
 Page 41 of 89
_________________________________________________________________________________________________________________________________________________________________________
For the case of 3-element model shown in the figure the final set of linear equations is

k111 k121 k131 k141 0 0 0 0

1
 q1   F1 
q   F 
1 1 1
k 21 k 22 k 23 k 24 0 0 0 0
 2  2
1
k31 1
k32 1
k33 + k112 1
k34 + k122 k132 k142 0 0  q3   F3 
1
k41 1
k42 1
k 43 + k212 1
k44 + k222 k 232 k242 0 0    
 q4   F4 
 = 
0 0 k312 k322 k332 + k113 k342 + k123 k133 k143  q5   F5 
 q6   F6 
0 0 k412 k422 k 432 + k21
3
k442 + k22
3
k 232 k243    
0 0 0 0 q7   F7 
k313 k323 k333 k343 q   F 
0 0 0 0 3  8  8
k413 k423 k 43 k443

6 3le –6 3le 0 0 0 0  F1 
3le 2l 2
−3le l2
0 0 0 0  0   F2 
e e
0  
   p l 
–6 −3le 12 0 –6 3le 0 0 
0 e

 q3  0 
3le le2 0 4le2 −3le le2 0 0   
2 EI q4   p0le 
 = 
le3  q5   M 
0 0 –6 −3le 12 0 –6 3le
 q6   pl 
0 0 3le le2 0 4le2 −3le le2   P + 0 e 
q7   2 
0 0 0 0 –6 −3le 6 −3le q   − p l 2 
 8  0 e

3le le2 −3le 2le2  12 
0 0 0 0
 Page 42 of 89
_________________________________________________________________________________________________________________________________________________________________________

FEM calculations:
1. Generation of stiffness matrices [ k ]e for all elements
4x4
2. Assembling the element matrices to obtain the global stiffness matrix [K ]
{F }
NxN

3. Finding the equivalent nodal force vector

{q}
Nx1

4. Imposing the boundary conditions and the solution of the final set of linear equations – finding all nodal displacements
5. Calculation of the internal forces (bending moment, shear force) and the stresses within the elements Nx1

The example

Final set of equations (3 active DOF)

12 0 3l  q3   − P 
2 EI     M1 M2
0 4l 2 l 2  q4  =  M 1  .
l3
3l l2 2l 2  q6   M 2  P
1 1 2 3

A B 2 C
 q3   w2  7l 2 3l −12l  − P 
    l  
q4  = θ 2  = 3l 15 −12  M 1  .
q  θ  96 EI −12l −12 48  M 
 6  3   2

(exact solution – why?)

 Page 43 of 89
_________________________________________________________________________________________________________________________________________________________________________

5. BARS AND SPRINGS

Finite element of a bar under axial loads:

u2 − u1
Assuming nodal displacements u1 i u2 we have u (ξ ) as the linear function: u (ξ ) = u1 + ξ.
le
After some operations u (ξ ) may be presented in the standard form as dependent on the nodal displacements and shape functions:

 ξ ξ q 
u (ξ ) =  1 −  u1 + u2 =  N1 (ξ ), N 2 (ξ )   1  =  N  {q}e , p( )
 l l  q2 e F1 F2
where u1 u2
u( )
1 2
 q1   u1 
{q}e =   =   is the vector of nodal displacements
e
 q2 e u2 e N 1( )

 N  =  N1 (ξ ), N 2 (ξ )  is the vector of shape functions 1

ξ ξ 1 2
N1 (ξ ) = 1 − , N 2 (ξ ) = ,
le le N2( )

1 2

Tension bar element with 2 nodes and 2 degrees of freedom and its shape functions
,
 Page 44 of 89
_________________________________________________________________________________________________________________________________________________________________________

Strain energy of the element:

l l
1 e EA e
U e = A∫ σ (ξ )ε (ξ ) dξ = ∫ (ε (ξ )) 2
dξ .
2 0 2 0
Taking into account that

du  ′ ′   q1 
ε (ξ ) = = N1 , N 2   .
dξ    q2 
e
we have

l N ′ 
EA e  1   ′ ′   q1 
Ue =   N1 , N 2    dξ =
2 ∫0

 q , q 
e 
 N 2    q2 e
1 2
′

EA
le 
N1′ N1′ N1′ N 2′   q1  1
=  q1 , q2  e ∫  dξ   =  q  [ k ] {q} ,
0  N ′N ′ ′ N ′  q2 e 2
e e e
2 N
 2 1 2 2 
where
EA  1 −1
[ k ]e =
le  −1 1  ,
is the stiffness matrix of the rod element (symmetric, singular, positive semidefinite)
 Page 45 of 89
_________________________________________________________________________________________________________________________________________________________________________
Equivalent nodal forces
N
The forces equivalent to the distributed load p(ξ ) 
 m 
.

le le
q 
W = ∫ p (ξ )u (ξ )dξ = ∫  N1 (ξ ) p (ξ ), N 2 (ξ ) p (ξ )   1  dξ =
p
ze
0 0  q2  e
 le le
 q 
=  ∫ N1 (ξ ) p (ξ ), ∫ N 2 (ξ ) p (ξ )dξ   1  .
 0 0   q2 e
In result:
le
q 
W =  F , F   1  , where
p
ze 1
e
2
e
Fi = ∫ N i (ξ ) p (ξ ) dξ
e
,
e q
 2 e 0

Fi e - the nodal forces equivalent to the distributed load p (‘work-equivalent’ or ‘kinematically’ equivalent)

Next steps of FE modelling are similar as in the case of the beam element. Finally we get the system of linear quations :

[ K ]{q} = { F } .
The right side vector {F } contains the external forces acting on nodes of the model (active nodes and reactions).
The system is solved after taking into account all boundary conditions;
When the vector of nodal displacements is determined the stresses within each of elements are computed:
q  E (q2 − q1 )
σ = Eε = E  N1′ (ξ ), N 2′ (ξ )   1  = .
 
 q2 e le

Example.
 Page 46 of 89
_________________________________________________________________________________________________________________________________________________________________________
Solve the presented below rods using FE models consisted of 2 elements

a) b)
P p0

a a

p 0 (l-a)
q 1 =0 1 q 3 =0 q 1 =0 2 q 2 =0
q2 2

1 3 1 1 2 2 3
2 P
Stiffness matrices of the two finite elements

EA  1 −1 EA  1 −1
[ k ]e [ k ]e =
2
=
1

a  −1 1  l − a  −1 1  .  1 1 
 a − 0 
 a   q1   F1 
1    
EA  −
System of simultaneous linear equations 1 1 1
+ − q2  =  F2 
 a a l −a l − a   
  q F
 1 1  3  3
−
 l −a l − a 
.
After including the boundary conditions q1 = q3 = 0 and F2 = P (case a) we have
 Page 47 of 89
_________________________________________________________________________________________________________________________________________________________________________

P (l − a )a
q2 = ,
EAl
− P (l − a )
F1 = ,
l
− Pa
F3 = .
l
where F1 and F3 are the nodal forces (reactions).
In the case b the nodal force in the second node is:
p0 (l − a )
F2 = ,
2
p0 (l − a ) 2 a − p0 (l − a ) 2 − p0 a(l − a)
q
Thus 2 = , F1 = , F3 = .
2lEA 2l 2l
The reaction in the first node R1=F1
And the reaction in the third node
p0 (l − a )l − p0a (l − a ) p0 (l − a )l − p0 (l − a )(l + a )
R 3 = F3 − = − = .
2l 2l 2l 2l
R1 + R3 = − p0 (l − a ) .

FE solution in the case a is the exact one but in the case b the approximate (why?)
 Page 48 of 89
_________________________________________________________________________________________________________________________________________________________________________
Spring element

q1=u1 q2=u 2 F=k * u=k(u2-u1)

k

1 2 [k]e=k 1 -1
-1 1

Finite element of a spring

Strain energy

1 1 1
Ue = F ∆u = k (∆u )2 = k (u2 − u1 )(u2 − u1 ) .
2 2 2
1 k − k   u1 
Ue =  1 2  
u , u  ,
−k k  u2  k −k 
2
[ ]e 
k =
1 −k k  , (stiffness matrix of a spring)
Ue =  q  e [ k ]e {q}e ,
2
In the same way may be derived the stiffness matrix for the twisted shaft:

GI s  1 −1
[ k ]e =  −1 1  ,
le  
where GI s is a torsional stiffness and the nodal displacements correspond to the rotation of the end cross-sections.
The FE models of the elastic structures can be built dividing the structure into finite elements of different types ( beams, tension bars, springs etc.)
 Page 49 of 89
_________________________________________________________________________________________________________________________________________________________________________
Example:

Find the finie element system of equations [K ]{q} = {F }for the structure presented below
q7
2 4
P1 p 0 P2 3
q1 q2 q3 q4 q5 q6

1 1
1 1 2 2 3
k1 k2
4 5
5 6
q7 q8 q9

Solution 4
FE model may be created using 2 beam elements , one rod element and 2 spring elements. The total number of degrees of freedom is 9
The stiffness matrices of the beam elements
6 3l1 −6 3l1
2 EI 3l1 2l1
2
− 3l1 l12
[k ]1
= [k ]
2
= 3
l1 − 6 − 3l1 − 3l12
e e .
6
3l1 l12 − 3l1 2l12
Degrees of freedom of the first element are q1 , q2 , q3 , q4 , and for the second q3 , q 4 , q5 , q 6 .
The stiffness matrix of the rod element (with the degrees of fredom q3 and q7 ). is
− 1
[k ]3e = EA 
1
l 2 − 1 1  .
The stiffness matrices of the springs:
1 − 1
[k ]e4 = k1   [ k ]e = k2  −1
5 1 −1
1  and corresponding degrees of freedom are q8 , q1
− 1 1 
and q9 q5 .

 Page 50 of 89
_________________________________________________________________________________________________________________________________________________________________________
The FE system of equations [ K ]{q} = {F } for the assuming numbering of the degrees of freedom:

1 2 3 4 5 6 7 8 9

1  − p0l1 
 2 − P1 
2  
 q1   − p0 l12 
q   
3  2   12 
 q3   p0l1 
4    
q4   0
   
5  q5  =  − p0 l1 − P 
q   2 2

6  6  2

 0   p0l1 
   12 
7 0  
 0   F7 
8  F8 
 
9  F9 

– Coefficients of the stiffness matrix of the element No 1 (beam)

– Coefficients of the stiffness matrix of the element No 2 (beam)

– Coefficients of the stiffness matrix of the element No 3 (rod)

– Coefficients of the stiffness matrix of the element No 4 (spring)

– Coefficients of the stiffness matrix of the element No 5 (spring)

 Page 51 of 89
_________________________________________________________________________________________________________________________________________________________________________
[K ] may be written in the form
k111 + k224 k121 k131 k141 0 0 0 k124 0
1 1 1 1
k21 k22 k23 k24 0 0 0 0 0
1
k31 1
k32 1
k33 + k112 + k113 1
k34 + k122 k132 k142 k123 0 0
1
k41 1
k42 1
k43 + k212 1
k44 + k222 k232 k242 0 0 0
[K ] =
9×9
0 0 k312 k322 k332 + k22
5
k342 0 0 k125
0 0 k412 k422 k432 k442 0 0 0
3
0 0 k21 0 0 0 k113 0 0
k214 0 0 0 0 0 0 k114 0
5
0 0 0 0 k21 0 0 0 k115
 Page 52 of 89
_________________________________________________________________________________________________________________________________________________________________________

6. TRUSSES AND FRAMES

Trusses - structures made of simple straight bars (members), joined at their ends (nodes).
External forces and reactions to those forces are considered to act only at the nodes and result in forces in the members which are either tensile or
compressive forces. Other internal forces are explicitly excluded because all the joints in a truss are treated as articulated joints.

The examples of 2D truss and 2D frame

Frames are the structures with members that are rigidly connected - e.g. with welded joints. The members of frames can be loaded by concentrated
and distributed forces. As a result they carry all possible internal forces (normal and shear forces, bending moments and torsinal moments).
 Page 53 of 89
_________________________________________________________________________________________________________________________________________________________________________
TRUSSES
2D trusses

Relation between the nodal displacements in local (element) coordinate systems and in global coordinates

 q  e =  q1 , q2  e along the axis of the rod  qg  = u1 ,υ1 , u2 ,υ 2  e in x,y coordinate system
e

y qi = ui cos α + υ1 sin α . (i = 1,2)

v 2 q2  u1 
 q1  cos α sin α 0 0  υ1 
u2   =   ,
e q2 e  0 0 cos α sin α  u2 
υ2 
e
q1
v1
u1
{q}e = [Tk ]{q q }e
x
Finite element of a plane truss

Strain energy of the element

c2 sc − c 2 − sc

U e =  q  [ k ]e {q}e =  qq  e [Tk ] [ k ]e [Tk ] {qq }e − sc − s 2

1 1 T T EA sc s2
   k g  =
2 1×2 2×2 2×1 2 1×4 4×2 2×2 2×4 4×1 e le − c 2 − sc c2 sc (*)

− sc − s 2 sc s2
 qq   k g  {qq } ,
1
Ue =
2  e  e e s = sin α , c = cos α
The stiffness matrix of the truss element in global coordinate system
Example.
 Page 54 of 89
_________________________________________________________________________________________________________________________________________________________________________
Find the displacement vector of the node 4 of the simple 2D truss for the case β 1 = β 2 and the horizontal force P (γ = 0) .

q6 q5 l
Element 1 nodes 1 and 4 slope angle α 1 = β 1 length l1 = .
3 cos α 1
3
q8
P
l
q4 2 Element 2 nodes 2 and 4 slope angle α2 = 0 length l2 = .
q3 q7 cos α 2
2 2 =
1 4 l
Element 3 nodes 3 and 4 slope angle α 3 = −β 2 length l3 = .
1
cos α 3
q2
q1
1

[k ] , [k ] , [k ]
Rozwiązanie
1 2 3
.The stiffness matrices of the three elements ij e ij e ij e are defined by (*).
The system of FE equations:
k111 k121 0 0 0 0 k131 k141  0   F1 
1
k21 1
k22 0 0 0 0 1
k23 1
k24 0  F 
   2 
0 0 k112 k122 0 0 k132 k142  0   F3 
   
0 0 k212 k222 0 0 k232 k242  0   F4 
 = 
0 0 0 0 k113 k123 k133 k143 0
   F 5 .
0 0 0 0 3
k21 3
k22 3
k23 3
k24  0   F6 
   
+ k332 + k333 + k342 + k343  q7   P cos γ 
1 1
k31 k32 k312 k322 k313 k323 1
k33 1
k34
1
k41 1
k42 k412 k422 3
k41 3
k42 1
k43 + k432 + k43
3 1
k44 + k442 + k44
3  q   P sin γ 
 8  
Taking into account that q j = 0 for j = 1, 6 the set of quations may be reduced to
 Page 55 of 89
_________________________________________________________________________________________________________________________________________________________________________

 3 ci2 3
si ci 
∑ ∑  q7   P sin γ 
i =1 li i =1 li
EA  3  =  
 si ci 3 2    .
si   
∑ ∑   q8   P cos γ 
 i =1 li l
i =1 i 

Assuming β1 = β 2 = β γ =0
EA 1 + 2c 0   q7   P 
3

  =  
l  0 2s 2 c   q8   0  ,
where c = cos β , s = sin β .
Then
Pl
q7 =
EA(1 + 2c3 ) ,
q8 = 0 .
The normal forces in the elements are calculated from the nodal displacements in the local (element) coordinate systems
 Page 56 of 89
_________________________________________________________________________________________________________________________________________________________________________
3D truss element in the coordinate system x,y,z

z w2
cx2 cx c y cx cz −cx2 −cx c y −c x cz
v2 u1
v1 cx c y c 2y c y cz − cx c y −c 2y − c y cz
z u2 w1 EA cx cz c y cz cz2 −c x cz − c y cz −cz2
2 {q}e= u  k g  =
w1 v1
y
v2
2 e le −cx2 − cx c y − c x cz cx2 cx c y c x cz
y x
w2 −cx c y −c 2y − c y cz cx c y c 2y c y cz
u1
1 − c x cz − c y cz −cz2 c x cz c y cz cz2
x

where cx = cos α x , c y = cos α y , cz = cos α z .

 Page 57 of 89
_________________________________________________________________________________________________________________________________________________________________________
FRAMES
2D frame element in the local coordinate system
The stiffness matrix of a frame element assembled from the stiffness matrices of the beam element with four degrees of freedom and the rod
element with 2 degrees of freedom:
1 2 3 4 5 6
 EA EA 
 l 0 0 − 0 0 
le
q3 q6  e 
 12 EI 6 EI −12 EI 6 EI 
q2 q5  0 le3 le2
0
le3 le2 
 
q1 q4  6 EI 4 EI −6 EI 2 EI 
1 2  0 le2 le
0
le2 le 
[ k ]e =
 − EA EA


e
 l 0 0 0 0 
le
 e 
 −12 EI −6 EI 12 EI −6 EI 
 0 le3 le2
0
le3 le2 
 
 6 EI 2 EI −6 EI 4 EI 
 0 le2 le
0
le2 le 
 

The stiffness matrix of a frame element in local coordinate system

The element with 6 DOF , the deformation defined by the functions u (ξ ) i w(ξ ) in the local c.s. It is called also 2D beam element.
 Page 58 of 89
_________________________________________________________________________________________________________________________________________________________________________
2D frame element in the global coordinate system xy

y q6 2
q1 u1
v2 q4 q2 v1
q5
q3 1
u2 {q}e= q {q g }e= u
q3 le 4 2
1
q5 v2
v1 q1 q6 2
q2
u1

The vectors of DOF of the frame element in the local c.s. {q}e and in the global c.s. {qg}e

The relation between the displacement of a node 1 in local (element) coordinate system and in global coordinate system
 q1 = u1 cos α + υ1 sin α ,

 q2 = −u1 sin α + υ1 cos α ,
q = θ .
 3 1

 q1   u1 
q  υ 
 2  1
 q3  θ1 
  = [Tr ]   = [Tr ] ⋅ {qg }e ,
q4  u2 
 q5  υ2 
   
q6 e θ 2 

where the transformation matrix [Tr ] is

 Page 59 of 89
_________________________________________________________________________________________________________________________________________________________________________

c s 0 0 0 0
−s c 0 0 0 0 

0 0 1 0 0 0
[ r]  0
T =
0 0 c s 0

 .
0 0 0 −s c 0
 
 0 0 0 0 0 1 
Strain energy of the element

 q  e [ k ]e {q}e =  qg  e [Tr ] [ k ]e [Tr ]{qg }e ,

1 1
Ue =
T

2 2
U e =  qg  e  k g  {qg }e ,
1
2 e

 k g  = [Tr ] [ k ]e [T ] ,
T
where e
is the stiffness matrix of the 3D frame element in global c.s.
 Page 60 of 89
_________________________________________________________________________________________________________________________________________________________________________
3D frames (beams)

x'
z' 1
y' w2
w'1
z'
v'1 1 v2
y'
The local (element) coordinate system is connected with the axis of the element. 1

The element x’is oriented along the element. u2

u'1 x'
The y’ axis is automatically set parallel to the global xy plane 1 2

(If the element is perpendicular to the xy plane the x’ is located T

x' y' z' x' y' z'
{q}e= u '1 , v 1' , w 1' , 1, 1, 1, u '2 , v 2' , w2' , 2, 2, 2
in parralel to the global y axis)
The element input data include:
- the node locations
z w2
- the cross-sectional area
v2
- 2 moments of inertia about the principal axes of the section
w1 x
- the parameters defining shear stiffness in the principal directions 1 u2
y' v 2
and the torsional stiffness z' 1

x'
y z
1
y
1 1
u1
x
T
x y z x y z
{q}e= u 1 , v 1 , w 1 , 1 , 1 , 1 , u 2 , v 2 , w2 , 2 , 2 , 2
 Page 61 of 89
_________________________________________________________________________________________________________________________________________________________________________
x'
z' 1  EA 
w2  l macierz symetryczna 
y'
w'1  e 
z'
v'1 1 v2  12 EI z ′ 
y'
 0 
1
 le3 
u2  12 EI y ′ 
u'1 x'  0 0 
1 2  le3 
 
T  0 GI s 
x' y' z' x' y' z' 0 0
{q}e= u '1 , v 1' , w 1' , 1, 1, 1, u '2 , v 2' , w2' , 2, 2, 2  le 
 
 6 EI y ′ 4 EI y ′ 
 0 0 − 2
0 
l le
 e

w2  6 EI z ′ 4 EI z ′ 
z  0 2
0 0 0 
l le
[ k ]e = 
e
v2  EA 
EA
w1 x
− 0 0 0 0 0 
1 u2  le le 
y' v 2  12 EI z ′ 6 EI z ′ 12 EI z ′ 
z' 1  0 − 0 0 0 − 0 
x'  le3 le2 le3 
y z  
1
 0 12 EI y ′ 6 EI z ′ 12 EI y ′ 
1 y
u1 0 − 0 0 0 0
1  l 3
l2
le3 
x  e e

 GI s GI s 
T  0 0 0 − 0 0 0 0 0 
{q}e= u 1 , v 1 , w 1 , x
, y
, z
, u 2 , v 2 , w2 , x
, y
, z le le
1 1 1 2 2 2  
 6 EI y ′ 12 EI z ′ 6 EI y ′ 4 EI y ′ 
 0 0 − 0 0 0 0 0 
 le2 l le2 le 
 6 EI z ′ 2 EI z ′ 6 EI z ′ 4 EI z ′ 
 0 2
0 0 0 2
0 − 0 0 0 
 le le le2 le 
 

3D beam element and the corresponding stiffness matrix in the local (element) coordinate system x′y′z′
 Page 62 of 89
_________________________________________________________________________________________________________________________________________________________________________

7. TWO AND THREE- DIMENSIONAL LINEAR ELASTOSTATICS

The finite elements of trusses and beams are, due to specific assumptions and simplifications, one –dimensional. All field problems of stress analysis
are in fact three-dimensional. In some limited cases the mathematical description of the problem may be formally reduced to two dimensional models (plane
stress state, plane strain state, axisymmetry) or ore even one dimensional as discussed bef.
Consider a linearly elastic body of volume Ω, which is bounded by surface Γ.
Data:
y (x2)
Ω –the analysed volume (domain),
Xi
pi Γ –the boundary,
pi –boundary tractions [N/m2].,

X i –body forces [N/m3].

ui , ij , ij
prescribed displacements ui on on the part
e of the boundary Γ

Unknown internal fields:

u i –displacement field,
ε ij – strain state tensor,
x (x1) σ ij – strss state tensor,
z (x3)
The body is referred to a three (or two) dimensional, rectangular, right-handed Cartesian coordinate system xi , i=1,3 (or x,y,z). The body is in
static equilibrium under the action of body forces Xi in Ω, prescribed surface tractions pi and prescribed displacements ui on on the boundary Γ
The three unknown internal fields are displacements ui , strains ε ij and stresses σ ij . All of them are defined in Ω.
 Page 63 of 89
_________________________________________________________________________________________________________________________________________________________________________
Component notation (Einstein indicial notation) for Cartesian tensors

The notation is used in rectangular Cartesian coordinates. In this notation, writing ui is equivalent to writing the three components u1, u2, u3 of the
displacement field u.

The Einstein summation convention is a tensor notation, which is commonly used to implicitly define a sum. The convention states that when an
index is repeated in a term that implies a sum over all possible values for that index.

Three examples:

∂ui ∂ui ∂u1 ∂u2 ∂u3

=∑ = + +
∂xi i ∂xi ∂x1 ∂x2 ∂x3

∂ui ∂ui ∂u ∂u ∂u
nj = ∑ n j = i n1 + i n2 + i n3
∂x j j ∂x j ∂x1 ∂x2 ∂x3

aij xj = bi i,j=1,n denotes the set of n linear equations

The indication of derivatives of tensors is simply illustrated in indicial notation by a comma.

∂f
f ,i =
∂xi
The comma in the above indicial notation indicates to take the derivative of f with respect to the coordinate xi .
∂ui ∂ui ∂u1 ∂u2 ∂u3
Examples: ui ,i = =∑ = + +
∂xi i ∂xi ∂x1 ∂x2 ∂x3
∂u ∂ui ∂u ∂u ∂u
ui , j n j = i n j = ∑ n j = i n1 + i n2 + i n3
∂x j j ∂x j ∂x1 ∂x2 ∂x3
1 if i = j 
The Kronecker delta is a convenient way of expressing the identity in indicial notation: δ ij =  
 0 if ≠ j 
The Kronecker delta follows the rules of index notation: Aik = δ ij A jk
 Page 64 of 89
_________________________________________________________________________________________________________________________________________________________________________
Strain state:

y ∂u x
ux + dy
3 extensional strains 3 shearing strains ∂y
∂u x ∂u x ∂u y
εx = γ xy = + ∂u y ∂u x
∂x ∂y ∂x u B’ C’
α= β=
∂u y ∂u ∂u x β ∂x ∂y
εy = γ xz = x + z γ xy= α + β
∂y ∂z ∂x D’
∂u ∂u ∂u y A’ α
εz = z γ yz = z + B C ∂u y
∂z ∂y ∂z uy uy + dx
dy ∂x
γxy , γyz, γzx - engineering shearing strains A D
dx

x
The strains may be written in the form of symmetric matrix assuming that
εxy= γxy/2 , εyz= γyz/2 , εzx= γxz/2. In this case the strains components form the symmetrical strain tensor.
The components of the strain tensor εij are often written in the form of symmetric matrix.

ε ij =
1
2
(
ui , j + u j ,i ) ( εij=εji ) - kinematic equations
 Page 65 of 89
_________________________________________________________________________________________________________________________________________________________________________
Stress state : stress tensor σ ij
Constitutive equations ( 3D Hook’s law)
σz (εz)
ν
ε x = 1 [σ x − ν(σ y + σ z )] ( )
 E 
σx = εx + εx + εy + εz
E  1+ ν  1 − 2ν
τzy  ν
τzx 1
[
ε y = σ y − ν(σ x + σ z ) ] 

σy =
E 
1+ ν εy +
1 − 2ν
(
εx + εy + εz )
E
τxz
σ z = E ε z + ν ε x + ε y + ε z

σy (εy)
1
[
εz = σz − ν σx + σy ( )] 
1+ ν  1 − 2ν
( )
E 
τxy τyx ⇒
γ xy = 1 τ xy 
τ xy = G ⋅ γ xy
σx (εx) G 
1 
γ yz = τ yz 
τ yz = G ⋅ γ yz
G 
σx= σxx σy= σyy σz= σzz 1 
γ xz = τ xz 
G E  τ xz = G ⋅ γ xz
E-Young’s modulus, G= - shear modulus, ν- Poisson’s ratio
2(1 + ν )

 v  1  v 
σ ij = 2G ε ij + δ ij ( ε kk ) ε ij =  σ ij − δ ijσ kk  (εkk= ε11 +ε22 +ε33)
 1 − 2v  2G  1+ v 
Strain energy density:
σ xε x + σ yε y + σ zε z + τ xyγ xy + τ yzγ yz + τ zxγ zx 
1
U'=
2

U’= ½ εij σij

1
σ ijε ij d Ω − ∫ X i ui d Ω − ∫ pi ui d Γ =min ,
2 Ω∫
Principle of the total potential energy: V = U − Wz =
Ω Γ
 Page 66 of 89
_________________________________________________________________________________________________________________________________________________________________________
Matrix notation
Matrix notation is a modification of direct tensor notation in which everything is placed in matrix form, with some trickery used if need be. The
main advantages of the matrix notation are historical compatibility with finite element formulations, and ready computer implementation in
symbolic or numeric form.
The representation of scalars, which may be viewed as 1 × 1 matrices, does not change. Neither does the representation of vectors because vectors
are column (or row) matrices. Two-dimensional symmetric tensors are converted to one-dimensional arrays that list only the independent
components (six in three dimensions, three in two dimensions). Component order is a matter of convention, but usually the diagonal components are
listed first followed by the off-diagonal components.

σ x  εx 
σ  ε 
 y  y
σ  ε 
For the strain and stress tensors this “vectorization” process produces the vectors σ =  z  , {ε } =  z  ,
τ xy  γ xy 
τ yz  γ yz 
   
τ zx  γ zx 

The relation between the strains and the displacement components in matrix notation:

{ ε ( x, y , z ) } = [ R ] { u ( x, y , z ) } ,

[R] is called symmetric gradient matrix in the continuum mechanics literature.

For 3 dimensional case :
 Page 67 of 89
_________________________________________________________________________________________________________________________________________________________________________
∂ 
 ∂x 0 0
 
0 ∂
0
σ x  εx   ∂y 
σ  ε   
 y  y 0 ∂
0 ux   u 
σ  ε  ∂z 
σ =  z , {ε } =  z  , [ R] =  ∂ ∂ , {u} = u y  = υ 
τ xy  γ xy   0    
τ yz  γ yz   ∂y ∂x   uz   w 
     ∂ ∂
τ zx  γ zx  0 
 ∂z ∂y 
∂ ∂
 0 
 ∂z ∂x 
In 2D case
∂ 
 0
σ x  εx   ∂x 
  u 
σ = σ y  , {ε } =  ε y  , 
[ R] =  0
∂
, {u} =  .
τ  γ   ∂y  υ 
 xy   xy  ∂ ∂
 
 ∂y ∂x 
 Page 68 of 89
_________________________________________________________________________________________________________________________________________________________________________
Hook’s law:

1− v v v 0 0 0
v 1− v v 0 0 0
v v 1− v 0 0 0
1 − 2v
{σ } = [D ]{ε }, [ D] =
E
(1 + v )(1 − 2v )
0 0 0
2
0 0

1 − 2v
0 0 0 0 0
2
1 − 2v
0 0 0 0 0
2
.

Plane stress state (σ z = 0, τ yz = 0, τ zx = 0 ) Plane strain state (ε z = 0, γ yz = 0, γ zx = 0 )

1 v 0 1− v v 0
E
[ D] = 2 v 1 0 [D] = E
v 1− v 0 .
1− v (1 + v)(1 − 2v) 1 − 2v
1− v 0 0
0 0 2
2
Strain energy density

1
U' = ε  {σ }
2
1
V = U − Wz = ε  {σ }d Ω − ∫  X  {u}d Ω − ∫  p  {u} d Γ
Total potential energy :
2 Ω∫ Ω Γ
 Page 69 of 89
_________________________________________________________________________________________________________________________________________________________________________
Finite element method for 2D and 3D problems of theory of elasticity:
LE
The domain Ω is divided into the subdomains (finite elements) Ωi : Ω = ∪ Ωi Ωi ∩ Ω j = 0 i≠ j .
i =1

v
2D elementy
elementsdwuwymiarowe
u

LWE=3
6DOF LWE=6
12DOF LWE=4
8DOF LWE=8
16DOF
w
3Delementy
elements
trójwymiarowe
v
u

LWE=4 LWE=10 LWE=8 LWE=20

12DOF 24DOF 24DOF 60DOF
LWE - liczba węzłów elementu

2D and 3D finite elements

Displacement field over the element is interpolated from the nodal displacements:
{u} = [ N ( x, y, z )] {q}e ,
where {q}e - nodal displacements vector , [N ] - shape functions matrix.

For example for the simplest trangular element with 3 nodes and 6 DOF the relation is
 Page 70 of 89
_________________________________________________________________________________________________________________________________________________________________________
 u1 
υ 
 1
u ( x, y )   N1 ( x, y ) 0 N 2 ( x, y ) 0 N 3 ( x, y ) 0  u2 
 =   where Ni are the linear functions
υ ( x, y )   0 N1 ( x, y ) 0 N 2 ( x, y ) 0 N 3 ( x, y )  υ2 
u3 
 
υ3 
Shape functions Nij are usually polynomials defined in local (element) coordinate systems.
Displacements, strains and stresses within each element are defined as the functions of the coordinates of the considered point and the nodal
displacements
{u} = [ N ] {q}e ,
{ε } = [ R ] {u} = [ R ][ N ] {q}e = [ B ] {q}e , [B] – strain-displacement matrix
{σ } = [ D ] {ε } = [ D ][ B ] {q}e .
The strain energy of the element Ωe is:
1
Ue = ε  {σ } d Ω e .
2 Ω∫e

1
 q  e [ k ]e {q}e .
1
Ue = [ ] [ D ] [ B ] {q}e d Ωe , Ue =
2 Ω∫e
T

 q 
 e
B
2
Where

[ k ]e = ∫ [ B ] [ D ][ B ] d Ωe = ∫  B∗  d Ωe ,
T

Ωe Ωe

is called the stiffness matrix of the element (symmetrical, singular, semi-positive defined) with the range equal to the number of DOF of the
element. Matrix [B] depends on the position within the element so the integration requires the special numerical techniques.
 Page 71 of 89
_________________________________________________________________________________________________________________________________________________________________________
Total strain energy of the structure is the sum of the finite elements energy:
LE
U = ∑ U e . (LE- number of finite elements in the model)
e =1

1
Using the global nodal displacement vector {q} U=  q  [ K ] {q} ,
2 1×n n×n nx1
where n is total number of DOF of the model and [ K ] is the stiffness matrix of the model.

The next step in FEM algorithms is finding the equivalent nodal forces {F} corresponding to the distributet loads {p} and {X}.
The total potential energy of the model is:

1
V = U − Wz =  q  [ K ]{q} −  q { F }
2 1×n n×n n×1 1×n n×1 ,
The minimum is determined by the conditions
∂V
= 0,
∂qi

[K ]{q} = {F }. (to be solved using neccesary displacement boundary conditions)

The strain and stress components in each finite element are found using the relations

{ε } = [ B ] {q}e , {σ } = [ D ] {ε } = [ D ] [ B ] {q}e
 Page 72 of 89
_________________________________________________________________________________________________________________________________________________________________________

8. CONSTANT STRAIN TRIANGLE (CST)

3
u ( x, y ) = ∑ N i ( x, y ) ⋅ ui
i =1
3
Ni (xi, yi )= 1, Ni (xj, yj )= 0 for i≠j
v( x, y ) = ∑ N i ( x, y ) ⋅ vi
i =1

u1 
v 
 1
u   N1 ( x, y ) 0 N 2 ( x, y ) 0 N 3 ( x, y ) 0  u2 
 =   {u} = [ N ]{q}e
v   0 N1 ( x, y ) 0 N 2 ( x, y ) 0 N 3 ( x, y )  v2 
u3 
 
v3 e
 Page 73 of 89
_________________________________________________________________________________________________________________________________________________________________________

1 1 1 1 1 1
Ai ( x, y )
N i ( x, y ) = 1
Ae = x1 x2 x3 A2 =
1
x1 x x3
Ae 2
y1 y2 y3
2
y1 y y3

ai = x j yk − xk y j
1
N i ( x, y ) = ( ai + bi x + ci y ) bi = y j − yk
2 Ae
ci = xk − x j

ε x ( x, y ) 
  u( x , y ) 
{}  y
ε = ε ( x , y )  [ ]
= R  = [ R ]  N ( x , y )  {q}e
  3×2 v( x , y )  3×2 2×6 6×1
γ
 xy ( x , y )  2 × 1

{ε } = [ B ]{q}e
3×6 6×1
 Page 74 of 89
_________________________________________________________________________________________________________________________________________________________________________

Strain- displacement matrix [B] :

 ∂N1 ∂N 2 ∂N 3 
 0 0 0 
 ∂x ∂x ∂x 
 ∂N1 ∂N 2 ∂N3 
[ B ] = [ R ][ N ] =  0 0 0 
 ∂y ∂y ∂y 
 ∂N1 ∂N1 ∂N 2 ∂N 2 ∂N 3 ∂N3 
 
 ∂y ∂x ∂y ∂x ∂y ∂x 

b1 0 b2 0 b3 0
1 
[ ]
B =  0 c1 0 c2 0 c3 
2 Ae
c1 b1 c2 b2 c3 b3 
With constant coefficients for each finite element.
CST – constant strain triangle! - linear displacement field within elements and
constant strains and stresses

{σ } = [ D ]{ε }
{σ } = [ D ][ B ]{q}e
 Page 75 of 89
_________________________________________________________________________________________________________________________________________________________________________
STRAIN ENERGY OF THE ELEMENT

1 1
U e = he ∫ ε  {σ } dAe = Ae he  q  e [ B ] [ D ][ B ]{q}e
T

Ae
2 2

1
Ue =  q  e [ k ]e {q} e
2
The stiffines matrix of the CST element [ k ]e
1
[ k ]e = Ae he [ B ] [ D ][ B ]
T

2 6×3 3×3 3×6

The strain energy of the entire model (N degrees of freedom)

1
U=  q  [ K ]{q}
2
where {q} is the total nodal displacement vector. [ K ] matrix – symmetrical, semi-positive defined , singular
1
V = U − Wz =  q  [ K ]{q} −  q {F } = min!
2 1×n n×n n×1 1×n n×1

Global nodal forces vector {F } is assembled from the equivalent nodal forces of all elements

Minimum of V with respect to {q} → [ K ]{q} = {F }

 Page 76 of 89
_________________________________________________________________________________________________________________________________________________________________________

Nodal forces of the Ωe element equivalent to the body load  X  :

Wzx = ∫  X  {u} d Ω
Ωe
e = ∫  X  [ N ]{q} d Ω
Ωe
e e =  F x  {q}e ,
e

 F x  =
e ∫  X  [ N ] d Ω
Ωe
e ( e.g. F1X = ∫ X ( x, y ) N ( x , y )d Ω
Ωe
1 1 e )

Nodal forces equivalent to the surface traction p acting on the edge Γep of the element Ωe

Wzp = ∫  p  {u} d Γ = ∫  p  [ N ]{q} d Γ =  F p  {q}e ,

p p
e e e e
Γ ep Γ ep

 F  e = ∫  p  [ N ] d Γ
p p
e .
Γep

The total stiffnes matrix K is singular – the system of linear equations is modified by taking into account the current displacement boundary
conditions.

 F  =  F1 , F2 , F3 , F4 , F5 , F6  =  F X  +  F p 
e e e

Ωe

Γ ep
 Page 77 of 89
_________________________________________________________________________________________________________________________________________________________________________

a) b)

c) d)

e) f)

The 2D model of a tensioned plate (under external loads being in equilibruim ). The correct and incorrect constraints
(constrained rigid body motion,unconstrained deformation)
 Page 78 of 89
_________________________________________________________________________________________________________________________________________________________________________
Finite element program:

Preprocessor
Information describing
- the geometry,
-the material properties ,
-the loads, the displacement boundary conditions.
Discretization of the model using the chosen type of finite elemets
(e.g. CST)

Processor
Assembling the stiffness matrix using the stiffness matrices of all
finite elements
- Building the set of simultaneous equations with included
boundary conditions (displacement b.c. and equivalent
nodal forces)
- Solution of the set of equations – calculation of all nodal
displacements
Calculation of strain and stress components within all finite
elements

Postprocessor
Graphical presentation of the results (contour maps, isolines ,
isosurfaces, graphs, animations)
Listings, tables
User defined operations on the received results
 Page 79 of 89
_________________________________________________________________________________________________________________________________________________________________________

RESULTS OBTAINED USING CST ELEMENTS - AVERAGING

Example –2D FE model of the cantilever beam
Finite element mesh

Y
Vertical displacement distribution Z X

Y
Z X

-5.29 -4.114 -2.939 -1.763 -.587743

-4.702 -3.526 -2.351 -1.175 0
Bending stress (σx ) distribution
(element solution)

Y
Z X

-2356 -1310 -263.47 783.007 1829

-1833 -786.708 259.769 1306 2353
 Page 80 of 89
_________________________________________________________________________________________________________________________________________________________________________
stress vectors in the CST elements
Node n
{σ } ≠ {σ } ≠ {σ } ≠ ...
1 2 3
1
{σ}1 2
{σ}2

{σ}i
Y
Z X i

-2356 -1310 -263.47 783.007 1829

-1833 -786.708 259.769 1306 2353

Averaged presentation (named nodal solution)

Averaged stress vector
at node n
{σ}nav = Σ {σ}i / k
( k=7)

Y
Z X

-1553 -860.368 -168.061 524.247 1217

-1207 -514.214 178.093 870.401 1563
 Page 81 of 89
_________________________________________________________________________________________________________________________________________________________________________
9. 8-NODE QUADRILATERAL ELEMENT. NUMERICAL INTEGRATION

The technique used for the formulation of the linear triangle can be formally extended to construct quadrilateral elements as well as higher order
triangles. But it is connected with some difficulties:
1. The construction of shape functions satisfying consistency requirements for higher order elements with curved boundaries becomes
increasingly difficult.
2. Computations of shape function derivatives to evaluate the strain-displacement matrix.
3. Integrals that appear in the expressions of the element stiffness matrix and consistent nodal force vector can no longer be carried out in
closed form.
The 8-node element is defined by eight nodes having two degrees of freedom at each node: translations in the nodal x (u) and y directions (v). It
provides more accurate results and can tolerate irregular shapes without much loss of accuracy. The 8-node are well suited to model curved
boundaries.

η
y 3 ( x3 , y 3 )
7
4 1 7 3 4
6 ( x6 , y 6 )

8 6 8

-1 1 ξ
1
( x1 , y1 ) 5
( x2 , y 2 ) y
( x5 , y 5 ) 2
1 -1 5 2

( ξ ,η ) → ( x , y )
( −1, −1) → ( x1 , y1 ) (1, −1) → ( x2 , y2 ) (1,1) → ( x3 , y3 ) ( −1,1) → ( x4 , y4 )
( 0, −1) → ( x5 , y5 ) (1, 0 ) → ( x6 , y6 ) ( 0,1) → ( x7 , y7 ) ( −1, 0 ) → ( x8 , y8 )
 Page 82 of 89
_________________________________________________________________________________________________________________________________________________________________________
8
x (ξ ,η ) = ∑ N i (ξ ,η ) xi
i =1
8
N2(ξ, η) N6(ξ, η) y (ξ ,η ) = ∑ N i (ξ ,η ) yi
i =1
η η

N1 (ξ ,η ) = −
1
4
(1 − ξ )(1 − η )(1 + ξ + η )
1
N 5 (ξ ,η ) =
2
(1 − ξ 2 ) (1 − η )
ξ ξ
N 6 (ξ ,η ) = (1 + ξ ) (1 − η 2 )
1 1
N 2 (ξ ,η ) = − (1 + ξ )(1 − η )(1 − ξ + η )
4 2
N 7 (ξ ,η ) = (1 − ξ 2 ) (1 + η )
1 1
N 3 (ξ ,η ) = − (1 + ξ )(1 + η )(1 − ξ − η )
4 2
N8 (ξ ,η ) = (1 − ξ ) (1 − η 2 )
1 1
N 4 (ξ ,η ) = − (1 − ξ )(1 + η )(1 + ξ − η )
4 2

Shape functions N2 and N6

 x1   x1 
y  y 
 1  1
 x2   x2 
   
y2  y2 
 x   N1 0 N2 0 N3 0 N4 0 N5 0 … N8 0     
x
 = x    = [ N ]  x3  = [ N ]{ xy}
y  0 N1 0 N2 0 N3 0 N4 0 N5 … 0 N 8   3  y y 
y
 3  3
⋮  ⋮ 
x  x 
 8  8
 y8 e  y8 
8 8
u (ξ ,η ) = ∑ N i (ξ ,η ) ui v (ξ ,η ) = ∑ N i (ξ ,η ) vi
i =1 i =1
 Page 83 of 89
_________________________________________________________________________________________________________________________________________________________________________
 u1 
v 
 1
u2 
 
v2
u   N1 0 N2 0 N3 0 N4 0 N5 0 … N8 0    u 
 = u    = [ N ]{q}e
v   0 N1 0 N2 0 N3 0 N4 0 N5 … 0 N8   3  v 
v
 3
⋮
u 
 8
 v8 e

∂ 
 0  ∂N ∂N 2 ∂N 3 ∂N 8 
 ∂x   1 0 0 0 0 
 ∂x ∂x ∂x ∂x
…
 
∂
{ε } = [ R ]{u} =  0   N (ξ ,η )  {q} = [ B ]{q} 
[ B] =  0
∂N1 ∂N 2 ∂N 3 ∂N 8 
3×1 3×2 2×1  ∂y   2×16  16×1e 3×16 16×1 ∂y
0
∂y
0
∂y
… 0
∂y 


∂ ∂  ∂N1 ∂N1 ∂N 2 ∂N 2 ∂N 3 ∂N 3 ∂N 8 ∂N 8 
   
 ∂y ∂x   ∂y ∂x ∂y ∂x ∂y ∂x ∂y ∂x 
…
3×2

Partial derivatives of shape functions with respect to the Cartesian coordinates x and y are required for the strain and stress calculations. Since the
shape functions are not directly functions of x and y but of the natural (local) coordinates ξ and η, the determination of Cartesian partial
derivatives is not trivial.
We need the Jacobian of two-dimensional transformations that connect the differentials of {x,y} to those of {ξ,η} and vice-versa

 ∂x ∂y   8 ∂N i 8
∂N i 
 ∂ξ ∑ ⋅ xi
∂ξ   i =1 ∂ξ
∑ δξ ⋅ yi 
[J ] =  = i =1
 =  J (ξ ,η ) 
 ∂x ∂y   8 ∂N i 8
∂N i 
⋅ xi ⋅ yi 

 ∂η ∂η   ∑
i =1 ∂η
∑
i =1 η 
 Page 84 of 89
_________________________________________________________________________________________________________________________________________________________________________
Matrix J is called the Jacobian matrix of (x,y) with respect to (ξ,η), whereas J-1 is the Jacobian matrix of (ξ,η) with respect to (x,y).
J and J−1 are often called the Jacobian and inverse Jacobian, respectively. The scalar symbol J means the determinant of J: J =|J|=det J.
Jacobians play a crucial role in differential geometry.

∂N i ∂N i ∂ξ ∂N i ∂η  ∂Ni   ∂ξ ∂η   ∂N i   ∂N i 
= ⋅ +  ∂x   ∂x ∂x   ∂ξ  
−1  ∂ξ 

∂x ∂ξ ∂x ∂η ∂x  ∂N  =  ∂ξ 
∂η   ∂N i 
= [ ]  
J
∂N i ∂N i ∂ξ ∂N i ∂η  i   ∂N i 
= ⋅ +  ∂y   ∂y ∂y   ∂η   ∂η 
∂y ∂ξ ∂y ∂η ∂y

 ∂N i   ∂x ∂y   ∂N i   ∂N i 
 ∂ξ   ∂ξ 
∂ξ  ∂x   ∂x 
 =   = [ J ]  ∂N 
 i   ∂x
∂N ∂y   ∂N i   i
 ∂η   ∂y   ∂y 
 ∂η   ∂η

{ε } =  B (ξ ,η )  {q}e
1 1
Ue = ∫ ε  {σ } dxdy = ∫  q  e [ B ] [ D ][ B ]{q}e dxdy
T

Ωe ( x , y ) )
2 2 Ωe ( x , y )

∫
A( x , y )
f ( x, y ) dxdy = ∫ξ η f (ξ ,η ) det [ J ] dξ dη
A( ,
dxdy = det [ J ] dξ dη
)

 B (ξ ,η )  [ D ]  B (ξ ,η )  det  J (ξ ,η )  d ξ dη {q}e
T
U e = 1/ 2  q  e ∫
Ωe (ξ ,η ) 16×3 3×3 3×16
 Page 85 of 89
_________________________________________________________________________________________________________________________________________________________________________
1
Ue =  q  e [ k ]e {q}e
2
1 1

[k ]e = ∫ [ B ] [ D ][ B ] dxdy = ∫ ∫  B (ξ ,η )  [ D ]  B (ξ ,η )  det  J (ξ ,η )  dξ dη ∫  B (ξ ,η ) 
T T

Ωe ( x , y ) −1 −1 16 x 3 3×3 3×16 16×3

Nodal forces of the Ωe element equivalent to the body load:

Wzx = ∫  X  {u} d Ω
Ωe
e = ∫  X  [ N ]{q} d Ω
Ωe
e e =  F x  {q}e ,
e

 F x  =
e
Ωe
∫  X  [ N ] d Ω . e

X

F = Ae X



y F/3
F/12
F/3
F/3
F/3
F/3 F/3 F/3

F/12
F/12 F/12
x

Work-equivalent nodal forces for uniform constant body load in the case of CST element and 8-node quadrilateral element
 Page 86 of 89
_________________________________________________________________________________________________________________________________________________________________________

Finite element method results: continuous displacement field and discontinous stress field

Displacement component (e.g. u(x,y)) interpolation over two finite elements

∂u ∂u
=
∂x′ 1 ∂x′ 2 (ε )1 = (ε x′ )2
x′

∂u ∂u
⇒ (ε )1 ≠ (ε y′ )2
y′
⇒ (σ ) ≠ (σ )
ij 1 ij 2

2 ≠
∂y ′ 1 ∂y′ 2
1 y′
x′
 Page 87 of 89
_________________________________________________________________________________________________________________________________________________________________________

Numerical Gauss integration in FE algorihms

The use of numerical integration is essential for evaluating element integrals of isoparametric elements. The standard practice has been to use Gauss
integration1 because such rules use a minimal number of sample points to achieve a desired level of accuracy. This property is important for
efficient element calculations because we shall see that at each sample point we must evaluate a matrix product.
[k ]e = ∫ [ B ] [ D ][ B ] dxdy = ∫  B (ξ ,η )  [ D ]  B (ξ ,η )  det  J (ξ ,η )  d ξ dη
T T

Ωe (ξ ,η ) Ωe ( x , y ) 16×3 3×3 3×16

The numerical intergration have to be also performed for finding the equivalent nodal forces.

One dimensional integration

In general: F(x)
b n

∫ F ( x ) dx = ∑ αi Fi ( xi ) + Rn
a i =1

Introducing the new variable -1 ≤η≤ 1

x=
( a + b ) + b − a ⋅η dx =
b−a
η
2 2 2
x1 x2 x3 x4 x5 xn
b−a b−a
b 1 1

∫ F ( x ) dx = ∫ f (η )
a −1
2
dη =
2 −∫1
f (η ) dη
η
x

The Gauss integration

x=b
 d 2n f 
1 n x=a
∫ ( )
−1
f η dη = ∑i =1
wi f ( i) n
η + R Rn = 0  dη 2 n 
 
η=-1
x1 η=1

Here n ≥ 1 is the number of especially defined Gauss integration points, wi are the integration weights, and ηi are sample-point
abcissae in the interval [−1,1]. The use of the interval [−1,1] is no restriction, because an integral over another range, from a to
b can be transformed to the standard interval via a simple linear transformation of the independent variable, as shown above.
The values ηi and wi are defined in such a way to aim for best accuracy. Indeed, if we assume a polynomial expression, it is
easy to check that for n sampling a polynomial of degree 2n – 1 can be exactly integrated .
 Page 88 of 89
_________________________________________________________________________________________________________________________________________________________________________

Table below shows the positions and weighting coefficients for gaussian integration.
f(η )
Abscissae and weight coefficients of the gaussian Quadrature
n çi (i=1,n) Wi (i=1,n)
1 0 2
2
−1/ 3 1

+1/ 3 1 η1 η2 η3 η4
3 − 0.6 5/9
0 8/9 1 η
-1
5/9
+ 0.6
4 -0.861136311594953 0.347854845137454
-0.339981043584856 0.652145154862546
+0.339981043584856 0.652145154862546
+0.861136311594953 0.347854845137454
5 η
Remarks: The sum of weighing coefficients is always 2 1
The integration gives the exact solution for polynomials of 2n-1 degree.
8 5 40 5 5 25
5 5 25 w6* = × = w9* = × =
w3* = × = 9 9 81 9 9 81
9 9 81
Numerical integration – rectangular region:
1 1 1
 n 
j ∑ wi f (ξi ,η j ) =
n n
1
∫−1 −∫1 ( ) ( i ) i -1
f ξ ,η d ξ dη ≈ ∫−1  ∑
i =1
f ξ , η w

dη ≈ ∑
j =1
w
i =1 5 8 40 8 8 64 5 8 40
w2* = × = w5* = × = w8* = × = ξ
= ∑∑ wi w j f (ξi ,η j ) = ∑ w f (ξk ,ηk )
n n m 9 9 81 9 9 81
* 9 9 81
k
i =1 i =1 k =1
8 5 40 5 5 25
5 5 25 w4* = × = w7* = × =
w1* = × = 9 9 81 9 9 81
9 9 81
 Page 89 of 89
_________________________________________________________________________________________________________________________________________________________________________

RESULTS OBTAINED USING 8-NODE ELEMENTS - AVERAGING

Example –2D FE model of the cantilever beam (compare to the results corresponding to discretization with CST elements)

Y Y

Z X Z X

-3045 -1691 -336.56 1018 2372

-2368 -1014 340.577 1695 3049
Bending stress σx distribution (element solution)
ANSYS 11.0
MAY 18 2010 ANSYS 11.0
PRES-NORM 10:19:31 MAY 18 2010
PLOTPRES-NORM
NO. 5 10:20:11
PLOT NO. 6

MAY 18 2010

PLOT NO. 3

Y Y
Z X Z X

Y Y
-743.468 -413.194 -82.921 247.352 577.626 -743.468 -413.194 -82.921 247.352 577.626
Z X Z X -578.331 -248.058 82.216 412.489 742.762 -578.331 -248.058 82.216 412.489 742.762

σx element and nodal solution σy element and nodal solution