Fea Basic Course

Basic FEA Course
BASIC FINITE ELEMENT

COURSE
Prepared by: DTE/GIS/SDD-

DTE/GIS/SDD- Fernanda Mariana Nunes Ravetti
Gustavo Limongi Araújo
Date: March 2004
THE INFORMATION CONTAINED HEREIN IS THE PROPERTY OF EMBRAER S.A. AND IT SHALL NOT BE COPIED OR USED WITHOUT EMBRAER’S WRITTEN CONSENT.
Basic FEA Course
SUMMARY
1. Introduction
2. Matrix Algebra
3. Principles of Finite Elements Method
4. Bar Elements
5. Beam Elements
6. Structural Analysis – Methods / Elements
7. Finite Elements Computer Program
8. Final Considerations
9. References
Basic FEA Course
1. Short History of FE Method / Introduction
In a 1943 paper, the Large general-purpose FE software began to
mathematician Courant appear in the 1970s. By the late 1980s the
described a piecewise software was available on microcomputers,
polynomial solution for the complete with color graphics and pre- and
torsion problem. His work was postprocessors. By the mid-1990s roughly
not noticed by engineers and 40,000 papers and books about the FE
the procedure was impractical method and its applications had been
at the time due to the lack of published.
digital computers.
In 1950s, work in the aircraft

industry introduced the FE
method to practicing engineers.
The name finite element was coined in

1960 [Clough and Robson]. By 1963 the
A classic paper described FE work
mathematical validity of the FE method
that was prompted by a need to
was recognizes and the method was
analyze delta wings, which are too
expanded from its structural beginnings
short for beam theory to be reliable
to include heat transfer, groundwater
[Turner and Clough]
flow, magnetic field, and so on.part in an
approximate way).
Basic FEA Course
Example
Geometry of the part Model of the part
Basic FEA Course
Model with results
Basic FEA Course
As the Finite Element Method is a discretization method, it produces lots of algebraic
equations, starting from the equilibrium condition of each element. This system of
equations, for small elements of simple geometry (squares, triangles, bars, beams etc.), is
solved by the computer through Matrices.
The principles and concepts used to describe the engineering problem under consideration
most of the times are known, like:
Newton’s Laws Potential Energy Strain Energy
Thermodynamic Law Principle of Virtual Work Mass Conservation Law
Maxwell Equations Principle of Energy Conservation
The advantages of an analysis through FEM are evident. Aeronautic industries, when
manufacturing large aircrafts, could not wait the prototype to be ready and then to verify its
structural integrity. With the previous results of the FEM analysis, it is possible to change
loads, materials and geometries, and then to run a new computer analysis.
Basic FEA Course
ANALYSIS PROCESS WITH THE FINITE ELEMENT METHOD
Change in the
PHYSICAL PROBLEM physical
configuration
MATHEMATICAL MODELING
Definition: Improvement of
-Geometry the
-Material Properties mathematical
-Loads model
-Boundary conditions
SOLUTION – FINITE ELEMENTS Mesh refinement,

Choose: changing
-Element Type parameters, etc
-Mesh Density
-Parameters for the solution
Represent: Verification of the
-Loads precision through
-Boundary Conditions the solution
RESULTS INTERPRETATION
Design Improvement
Structural Optimization
Basic FEA Course
It is important to remember that it is impossible to predict the exact solution for a

problem, even if the model is the best one. The analysis always gives an
approximate solution.
A model must be:
Effective Reliable
Represents the problem with the The utilization of the chosen mathematical
necessary precision and low cost model implies in expected results
(time & money); (known by experience, theoretical or
experimental validation).
The next example (Ref.: Bathe,K.J.,Finite Element Procedures) illustrates this

concept. The first figure represents the real problem. The first step is to know what
phenomenon must be described when analyzing the part, that is, what behavior is
expect when applying the load to it.
Basic FEA Course
PROBLEM DESCRIPTION – PROPOSED MODELS
Basic FEA Course
Selection of the most adequate model conditions for the mathematical model:
material properties loading boundary conditions
In a preliminary analysis, it is assumed that the objective is to know the bending

moment in section AA and the deflection in the region of application of force W.
ANALYTICAL FORM OF moment and deflection of the beam:
Where:
Moment: M = WL = 27500 ⋅ N .cm
E = Elasticity Modulus
G = Shear Modulus
1 W ( L + rN ) 3 W ( L + rN ) I = Moment of inertia
Deflection: δ= + = 0.053cm
3 EI 5 6 AG A = Section Area
Basic FEA Course
In a second analysis, a 2D elastic model (PLANE STRESS MODEL)
MODEL is calculated
numerically (FEM). The results obtained are:
M = 27500 N.cm δ = 0.064 cm
To verify if the model is effective and reliable,

reliable the analytical results are
compared to the plane stress model results, and the conclusion is that THE PLANE
STRESS MODEL IS RELIABLE BECAUSE THE BENDING MOMENT IS
CALCULATED WITH 0% ERROR AND THE DEFLECTION WITH 21% ERROR.
Next, there is another example (Ref. Boresi,A.P.,Sidebottom, O.M., ”Mechanics of

Materials”), where the results of numerical models (FEM) with several meshes are
compared with the analytical solution (Elasticity Theory) for a known example from
literature.
Basic FEA Course
The model is a fixed beam with a parabolic load applied to one of its extremities. Its
exact solution for stresses σx and τxy is known by the Elasticity Theory.
Basic FEA Course
The 2 x 6 mesh was discarded because the results showed inconsistence. The
comparative results for the other meshes (4x12 and 8x24) and the comparison with
theory are presented next.
As expected, the 8 x 24 mesh brings closer results to the theoretical solution.

Although, if an error of 10% is allowable, the model with 4x12 mesh can also be
considered an effective model.
model
During the course, “parabolic interpolation elements” will be discussed. These
elements, in contrast to the previous analysis, enable the creation of more effective
models with fewer elements.
Basic FEA Course
2. Solution of Linear Equations
Here there is a linear system of equations:
A11 x1 + A12 x2 + + A1n xn = C1 A11 A12 A1n x1 C1

A21 x1 + A22 x2 + + A2 n xn = C2 A21 A22 A2 n x2 C2
=
Am1 x1 + Am 2 x2 + + Amn xn = Cm Am1 Am 2 Amn xn Cm
Putting the coefficient values in matrix form [A]{x}={C}, the known numerical
methods for the solution of this system are: Triangulation Method, Gauss
Method and Gauss-Siedel Method.
Method
Basic FEA Course
3. Principle of the Finite Element Method
3.1. Stiffness and Flexibility

3.2. Stiffness Matrix
3.3. Energy Functional
Basic FEA Course
3.1. Stiffness and Flexibility
Consider an elastic spring with an applied load P, as shown:
k
P
If:
k = spring stiffness (load necessary to cause a unit displacement);
u = displacement of the spring because of load P;
1
Then: P = k ×u and u= ×P
k
The flexibility is the opposite of the stiffness, then: u = f ×P
Basic FEA Course
EXAMPLE:
•If the structure is RIGID,

RIGID it means that k
has a high value. Suppose for a generic
example, that k=1000 and P=1000 for the
spring of the figure:
1 1
u= ×P = × 1000 = 1
k 1000
• If the structure is FLEXIBLE,

FLEXIBLE it means
that k has a low value. Suppose that
k=100 and P=1000 for the spring of the
figure:
1 1
u= ×P = × 1000 = 10
k 100
Basic FEA Course
The displacement is
If the stiffness is known obtained for any load P
This elementary problem is a description of how the structures behave themselves.
Thousands of interconnected springs
General structures Different stiffness values in several orientations
Complex load combinations
For a practical problem of linear analysis, it is possible to do an analogy between a

spring and a finite element,
element and, connecting them by their extremities, it is possible to
describe a complex structure.
Basic FEA Course
3.2. Stiffness Matrix
k
P1 u 1 1 2 P2 u 2
Supposing that it is desired to analyze an element like a spring, with loads P1 and P2
applied to its extremities, generating displacements u1 and u2:
If: P = k × (spring .displacement )
From de equilibrium condition:

P1 = k × (u1 − u2 )
P1 = − P2
P2 = k × (u 2 − u1 ) = − k × (u1 − u2 )
P1 k −k u1 k −k
In matrix notation: = Where [ k] = is the element
P2 −k k u2 −k k stiffness matrix.
{Pi} is the vector of nodal internal forces and {ui} is the vector of displacements. In this
case, as the spring is free, the matrix is singular and the system cannot be solved.
Basic FEA Course
In other example, the structure is composed of three elements, represented by three
springs, with stiffness ka, kb and kc respectively. One extremity is fixed, and the
representation of the effort on the other extremities is like the figure:
P1 P2 P3
ka kb kc
1 2 3
Original:
Deformed:
tração compressão tração
u1 > 0 u2 < 0 u3 > 0
For each node, there must be an equilibrium condition:
P1 = k a u1 − kb (u 2 − u1 )
P2 = kb (u2 − u1 ) − kc (u3 − u2 )
P3 = kc (u3 − u 2 )
Basic FEA Course
From the equilibrium condition:
P1 = k a u1 − kb (u 2 − u1 ) = (k a + kb )u1 − kbu 2
P2 = kb (u2 − u1 ) − kc (u3 − u2 ) = −kb u1 + (kb + k c )u2 − kc u3
P3 = kc (u3 − u 2 ) = −kc u 2 + k cu3
The correspondent symmetric matrix system is:
( k a + kb ) − kb 0 u1 P1
− kb ( kb + k c ) − k c u 2 = P2
0 − kc kc u3 P3
The same matrix can be mounted in another way, element by element, as follows:
Basic FEA Course
P1 P2 P3
ka kb kc
Point 0 1 2 3
tração compressão tração

u1 > 0 u2 < 0 u3 > 0
Matrix of element (a): Matrix of element (b): Matrix of element (c):
ka − ka kb − kb kc − kc
− ka ka − kb kb − kc kc
The common points between ka − ka 0 0 u0 P0

matrices are the points where
equal displacements occur. − ka k a + kb − kb 0 u1 P1
=
So, the mounting of the 0 − kb kb + kc − kc u2 P2
system matrix is:
0 0 − kc kc u3 P3
Point 0 is the fixed region, with displacement u0=0. For the analysis matrix, the first line and the
first column are excluded, and the stiffness matrix is equivalent to the one shown before.
Basic FEA Course
The example before shows how the global stiffness matrix of a finite element model
is found.
The stiffness matrix can be partitioned, and lines and columns correspondent to the
constraints (where the displacement is null or known from the input data) are taken
from the principal analysis.
For the cases that the displacement on the constraints is null, the global stiffness
matrix can be represented as:
KSS KSL uS PS If Point 0 is fixed, u0=0 and the first
column of the matrix is null.
ka − ka 0 0 u0 P0
Then:
− ka k a + kb − kb 0 u1 P1
=
0 − kb kb + kc − kc u2 P2 − k a u1 = P0
0 0 − kc kc u3 P3
And the reaction of Point 0 is
KLS KLL uL PL defined.
[K LL ]{u L } = {PL } Displacements are defined
[K SL ]{u L } = {PS } Reactions are defined
Basic FEA Course
3.3. Energy Functional

For the structural problem, the Finite Element Method (FEM) finds the stresses
through the minimization of the ENERGY FUNCTIONAL.
FUNCTIONAL
The ENERGY FUNCTIONAL takes into account all energies associated to the model,
according to the PRINCIPLE OF ENERGY CONSERVATION.
CONSERVATION The total energy of the
system must be zero (in other words, the functional must be zero).
The minimum value of the functional is obtained getting itself derivate with respect to
the displacement of the nodal points, equal to zero, as follows:
∂F
=0
∂u
Next, it is presented a simple example of this principle, using a spring.
Basic FEA Course
The spring has a known constant k. The variables to be determined are the
displacements of points 1 and 2 (u1 and u2 respectively).
k
P1 u 1 1 2 P2 u 2
The Energy Functional is the total energy, given by:
1
F= k (u1 − u 2 ) 2 − P1d (u1 − u 2 )
2
Doing the derivation with respect to the displacement (where u=u1-u2) and
equaling to zero, them resulting equation will be:
k (u1 − u 2 ) − P1 = 0
Isolating P1, the relation is the same of the beginning of this topic :
P1 = k × (spring .displacement ) = k × (u1 − u2 )
Basic FEA Course
4. Bar Elements
4.1. Obtaining the Stiffness Matrix (1D)
4.3. Example
Basic FEA Course
If a bar of length “l”, cross-section area “A” and elasticity module “E” is
subjected to axial forces “F1” and “F2”, as in the figure below, the relations
between force and displacement are:
AE
F1 = (u1 − u 2 ) F u F u
l 1 1 1 2 2 2 x
AE
F2 = − F1 = (u2 − u1 ) l
l
In the matrix form, the equations above can be rewritten as:
1 −1 [k] = Stiffness Matrix of the

F1 AE 1 − 1 u1 [k ] = AE element in Local Coordinates

= l −1 1
F2 l − 1 1 u2 F1 {pi} = vector of the forces in the
{pi } = elements and nodes
F2
{pi } = [k ]{ui } u1 {ui} = vector of the displacements
{ui } = of the elements and nodes
u2
Basic FEA Course

The stiffness matrix usually has no applications in the one-dimensional form because
the truss members (for example) are disposed inclined in several angles in relation to
the horizontal. There is a 2D matrix that describes the element in any angle of
inclination, called the Rotation Matrix.
[k0] = stiffness matrix for a 2D element in Global Coordinates
Observing the Coordinate System:

y
o
A
x
0 0 v
u = u cos α + v sinα y
vo
E
v = v 0 cos α − u 0 sinα B
u
In matrix form: xo
O C
u cos α sinα u0
= uo
v − sinα cos α v0
Basic FEA Course
The bar has two nodal points and the relation between the local and global
displacements is:
cos α sinα

0
u1 u 1 − sinα cos α
0
v1 v 1
0 0 0
= or
0 {ui } = {u }i
u2 u20 0
v2 v20
The relation between the forces is:
0 0
{pi } = i
{p }
0
The relation between the stiffness matrices is:
T
0 0 0
[k ] = [k ]
0 0
Basic FEA Course
4.3. EXAMPLE: Fy o ,v o
2 2
yo Fx ,u
2 2
For the example of the figure, it is Fy ,v
2 2
necessary to obtain the relations between
o o
forces and displacements, in the Fy1 ,v1 Fx2o,u 2o
Coordinate System (x,y):
y x
Fy ,v
1 1
Fx1 1 0 − 1 0 u1 o o
Fx ,u Fx1 ,u 1
Fy1 AE 0 0 0 0 v1 1 1
= xo
Fx2 l −1 0 1 0 u2
Fy2 0 0 0 0 v2 u 01 u 02 v 01 v 02
0
In Global Coordinates (x0,y0), the c2 cs
− c 2 − cs u 1
transformed stiffness matrix is: 2 u 02
AE cs
− cs − s s2
k0 =
[ ] l − c 2 − cs c 2 cs v01
Where c = cos(a), s = sin(a)
− cs − s 2 cs s 2 v02
Basic FEA Course
Exercise1
y For the given bar truss calculate the
displacements U1 and V1 by using the
V2 V3 V4 Finite Element Displacement Method.
U2 U3 U4
2 3 4
V1 θ1
θ2
1 U1 x
F1
θ1 = 135°
θ 2 = 45°
Basic FEA Course
Solution u1 v1 u2 v2
c2 cs − c2 − cs F1x 0,5 − 0,5 − 0,5 0,5
0 AE cs s2 − cs − s 2 F1 y − 0,5 0,5 0,5 − 0,5
k =
[ ] K12=EA1/L1 *
l − c 2 − cs c2 cs F 2 x − 0,5 0,5 0,5 − 0,5

− cs − s 2 cs s2 F2y 0,5 − 0,5 − 0,5 0,5
u1 v1 u4 v4 u1 v1 u 3 v3
F1x 0,5 0,5 − 0,5 − 0,5 F1x 0 0 0 0
K14=EA3/L3 * F1 y 0,5 0,5 − 0,5 − 0,5 F1 y 0 1 0 −1
K13=EA2/L2 *
F 2 x − 0,5 − 0,5 0,5 0,5 F 2x 0 0 0 0
F 2 y − 0,5 − 0,5 0,5 0,5 F2y 0 −1 0 1
u1 v1 u2 v2 u3 v3 u4 v4
F1x a1 + a3 − a1 + a3 − a1 a1 0 0 − a3 − a3
a1=A1/L1
F1 y − a1 + a3 a1 + 2a 2 + a3 a1 − a1 0 − 2a 2 − a3 − a3
a2=A2/L2
F 2x − a1 a1 a1 − a1 0 0 0 0
F2y a1 − a1 − a1 a1 0 0 0 0 a3=A3/L3
Kt=E/2 *
F 3x 0 0 0 0 0 0 0 0
F3y 0 − 2a 2 0 0 0 2a 2 0 0
F 4x − a3 − a3 0 0 0 0 a3 a3
F4y − a3 − a3 0 0 0 0 a3 a3
Basic FEA Course
Solution
• Aplicando C.C. u2=v2=u3=v3=u4=v4=0
A1
F1x 0 U1
= E L1
F1 y 0
A1 A2 V 1
+
L1 L 2
F1x * L1
U1 =
E * A1
F1 y
V1 =
A1 A2
+
L1 L 2
Basic FEA Course
5. Beam Elements
5.1. Derivation of the Deflection Equations

5.2. Solution of Beams (1D)
Basic FEA Course
5.1. Derivation of the Deflection Equations

To obtain the stiffness matrix of a
beam element, initially it is necessary
to derivate the deflection equations.
For the beam shown in the picture
beside, considering equilibrium of
forces and moments, the equations
for M1, Fy1, M2 e Fy2 are:
1 2 Where v and v are the displacements
6 EIθ 1 6 EIθ 2 12 EI on ‘y’ direction, and θ1 and θ2 are
Fy1 = − 2 − 2
+ 3 ( v1 − v 2 )
l l l rotations.
4 EIθ1 2 EIθ 2 6 EI
M1 = + − 2 (v1 − v2 ) Fy1 12 / l 3 − 6 / l 2 − 12 / l 3 − 6 / l 2 v1
l l l
M1 − 6 / l2 4/l 6 / l2 2/l θ1
6 EIθ 6 EIθ 12 EI = EI
Fy2 = + 2 1 + 2 2 − 3 (v1 − v2 ) Fy2 − 12 / l 3 6 / l 2 12 / l 3 6 / l 2 v2
l l l M2 − 6 / l2 2/l 6 / l2 4/l θ2
2 EIθ1 4 EIθ 2 6 EI
M2 = + − 2 (v1 − v2 )
l l l Stiffness Matrix
Basic FEA Course

Comparing to bars, the solution of the element stiffness matrix and loads vector for
beams contains a rotational component M and a correspondent rotational
displacement Θ.
3 kN
EXAMPLE:
v1 θ1 v2 θ2 1 2 3
l = 4m 4m 2m
12 / 64 − 6 / 16 − 12 / 64 − 6 / 16 v1

− 6 / 16 4/4 6 / 16 2/4 θ1 The stiffness matrix for (2) is given by
[k12 ] = EI
− 12 / 64 6 / 16 12 / 64 6 / 16 v2 the influence coefficients, corresponding

− 6 / 16 2/4 6 / 16 4/4 θ2 to to displacements v2 and θ2:
v2 θ2
v2 θ2 v3 θ3
l = 2m v2 64 + 12 / 8 6 / 16 − 6 / 4 v2
12 / 8 − 6 / 4 − 12 / 8 − 6 / 4 (2)
[K ] = EI 126 // 16 −6/4 4/4+ 4/2 θ2
−6/ 4 4/2 6/4 2/2 θ2
[k23] = EI v2 θ2
− 12 / 8 6 / 4 12 / 8 6/4 v3

1.688 − 1.125 v2
2/2 6/4 4/2
!

−6/ 4 θ3 ( 2)
[K ] = − 1.125 3 θ2
Basic FEA Course
The vector of the loads correspondent to
3 kN
the displacements v2 and θ2 is:
−3 v2
{qF } = 1 2 3
0 θ2 4m 2m
The resulting equations are:

The negative values shows that the
deflection is in the lower part and the
−3 1.688 − 1.125 v2
= EI arrow is in the counterclockwise
0 − 1.125 3 θ2 direction.
v2 1 3 1.125 − 3 1 − 2.37 3 kN
- +,
'* & (%)
$ # "
' & %
$ # "
' & %
$ # "
= =
θ2 EI 1.125 1.688 0 EI − 0.889
3.798 1 2 3
4m 2m
The values of the forces in nodes 1 and 3, and of the bending moments in nodes 1, 2
and 3 are obtained by the application of formulas of the Beam Theory, showed
before.
Basic FEA Course
V1
L1
The results of this problem are compared with C1
NASTRAN results. There is a model with 12 beam

123456 123456
elements, dimensions, loads and boundary 3000.
conditions equal to the problem of figure before.

Y
Z X
The cross-section of the idealized beam is a square
V1
of 50x50mm, and the material is a steel with L1
C1
E=210GPa (Elasticity Module) and v=0,3 (Poisson).
123456 123456
The comparison of results of deformation and of the
inclination angle in the load application point (node 9)
3000.
Y
are indicated in the table below. The graphic shows Output
Z SeX t: MSC/NASTRAN Ca se 1
De forme d(0.0239): Tota l Tra nsla tion
this values for distances through the life cycle.
0.0142
0.00787
COMPARATIVE TABLE 0.00152 0.
-0.00483
Theoretical NASTRAN -0.0112
Displacement (m) 0,02167 0,02169
-0.0175
Slope 0,008126 0,008128
-0.0239
-0.0239
1 2 3 4 5 6 7 8 9 10 11 12 13
Node ID
1: MSC/NASTRAN Ca s e 1, T2 Tra ns la tion
2: MSC/NASTRAN Ca s e 1, R3 Rota tion
Basic FEA Course

Analogously to bar elements, the stiffness matrix of a beam element can be extended
to Global Coordinates.
Fy o ,v o
2 2
Fx ,u Fx1 ( A / lI ) 0 0 (− A / lI ) 0 0 u1
yo 2 2
Fy ,v Fy1 0 12 / l 3 − 6 / l2 0 − 12 / l 3 − 6/ l2 v1
2 2
o o Fx2o,u 2o M1 0 − 6 / l2 4/l 0 6 /l2 2/l θ1
Fy1 ,v1 = EI
Fx2 (− A / lI ) 0 0 ( A / lI ) 0 0 u2
y x
Fy ,v Fy2 0 − 12 / l 3 6/ l2 0 12 / l 3 6 /l2 v2
1 1
: 8 8 8 8 8 8 8 89
4 . . .. 3 . . . . 2
1 . . .. 0 . . . . /
47 .5 .5 5 .. 35 .5 .5 .5 . 5 62
1 . . .. 0 . . . . /
o o
Fx ,u Fx1 ,u 1 M2 0 − 6 / l2 2/l 0 6 /l2 4/l θ2
1 1
xo
Stiffness Matrix
{pi } = [k ]{ui }
Basic FEA Course
Fx1 cos( α ) sin( α ) 0 Fx10

The relation between the
Fy1 − sin( α ) cos( α ) 0 0 Fy10
local coordinates and the
M1 0 0 1 M1
global coordinates is given =
Fx 2 cos( α ) sin( α ) 0 Fx 20
by: Fy 2 0 − sin( α ) cos( α ) 0 Fy 20
A ; ; ;; @ ; ; ; ; ?
> ; ; ;; = ; ; ; ; <
G E E E E E E E EF
A D ; B ; B B ;; @B ;B ;B ;B ;B C ?
> ; ; ;; = ; ; ; ; <
Note: c = cos(α ) s = sin(α ) M2 0 0 1 M2
The stiffness matrix of the element is [k0] = [kb0]+[kr0], where:
12 2
s c2 cs 0 − c2 − cs 0
l3
12 12 2 cs s2 0 − cs − s 2 0
− 3 cs c symmetric
l l3 0 0 0 0 0 0 0
6 6 4 r
[k ] = AEl
s − c 2 − cs 0 c2 cs 0
2
− 2c
0 l l l
b
[k ] = EI − cs − s 2 0 cs s2 0
12 12 6
S Q Q Q Q Q Q Q QR
P N N N N N N N NO
12 2
− 3 s2 cs − 2s s 0 0 0 0 0 0
l l3 l l3
12 12 6 12 12 2
cs − 3 c2 c − cs c
l3 l l2 l3 l3 Matrix for an inclined bar
6 6 2 6 6 4
M K K K K K K K K K K K K K L
J H H H H H H H H H H H H H I
s − c − 2s c
l2 l l l l2 l
Matrix for an inclined beam
Basic FEA Course
6. Structural Analysis – Methods /

Elements
6.1. Introduction / Methods
6.2. Element Types / Characteristics
6.3. Calculus General Procedure
Basic FEA Course
6.1. Introduction / Methods

The methodology showed here to the Linear Static Structural Analysis (for bar and
beam elements) is the DISPLACEMENT METHOD.METHOD
It is clear that one of the main problems of the FEM method is to find the STIFFNESS
MATRIX of the element as well as the global structure.
Although, these elements (bar and beam) are the simplest ones. The same theory
can be extended for PLANE and SOLID elements (that are much more complex), with
several shapes and several distinct properties, but also called FINITE ELEMENTS.
The METHOD OF THE FUNCTIONAL MINIMIZATION is the most adequate method

for the derivation of these matrices.
Basic FEA Course
6.2. Element Types / Characteristics

The main idea of the finite element formulation is obtaining the relation between nodal
displacements and the displacements in any point of the element, through the SHAPE
FUNCTIONS. Next, formulations for some types of elements are shown.
6.2.1. Bar / Beam Elements
The first step for the element formulation is relating the GLOBAL COORDINATES
with a system of NATURAL COORDINATES.
COORDINATES In the example below, u(x,y) is a vector
representing the element displacement functions. There is a displacement function,
as follows:
y
x=l
x=0 u = α1 + α 2 x
1 2 x, u
Where α1 and α2 are constants.
Basic FEA Course
The values of vector {αi} are obtained by the boundary conditions:
T
(1) If x = 0 then u = u1;
T
(2) If x = l then u = u2, in such a way that:
u1 1 0 α1 1 0
= or {ui } = [C ]{α i } where: [C ] =
u2 1 l α2 1 l
The matrix [C] relates the displacements with the polynomial parameters.
parameters
These parameters are obtained by the inverse matrix of [C]:
−1
1 0
{α i } = [C ] {u i } [C ]−1 =
−1/ l 1/ l
FUNCTIONAL APPLICATION - for the energy calculus, it is necessary to:
T
obtain vector {e} of deformations
T
obtain the relation between stresses and strains (Hooke’s Law)
T
calculate the variation of the internal energy
T
calculate the work of external forces
Basic FEA Course
The deformations vector are:
du α1
ε= {ε } = [0 1] or {ε } = [B]{α i }
dx α2
Where matrix [B] is the matrix that relates deformations and vector {α} with
the polynomial parameters.
parameters
By Hooke’s Law:
Law
σ = Eε or {σ } = [D]{ε }
Where [D] is the matrix of the material elasticity modulus.

modulus
Basic FEA Course
6.2.2. Virtual Work Approach
The virtual work method is a more general method than the displacement
one and can be used to derive stiffness matrices for one- and two-
dimensional elements. The virtual work approach for the known plane rod
element is presented here.
− T
y internal virtual work = {ε } {σ }d (vol )

x=l vol
x=0
1 2 x, u {σ } = [ D ][ B ][ C ] −1 {u i }
Plane rod element
{ε }T = [B]T [C −1 ]T {ui }T
internal virtual work = {u i }T [C −1 ]T [ B]T [ D][ B][C −1 ]{u i }d(vol)
vol
T
external virtual work = {ui } [ Pi ]
Basic FEA Course
External Virtual Work=Internal Virtual Work
[ Pi ] = [C −1 ]T [ B]T [ D][ B ][C −1 ]{ui }d (vol ) = K {ui }

vol
~ ~
T −1 −1 T
[K ] = vol
[ B ] [ D ][ B ]d ( vol ) [ K ] = [C ] [ K ][C ]
~ 1 0 1 −1/ l 0 0 1 0
[K ] = A [ E ][0 1] [K ] = Al
0 1 0 1/ l 0 E −1/ l 1/ l
~ 0 0 AE 1 − 1
[ K ] = Al [K ] =
0 E l −1 1
Basic FEA Course
6.2.2. Triangular Elements (CST->Tria 3)
v3
y,v Displacement Field: Strain Field:
(x 3 ,y 3 )
u3
u = β1 + β 2 x + β 3 y εx = β2
v1 v2
v = β 4 + β5 x + β 6 y ε y = β6
x,u
(0,0)
u1 (x2 ,0) u2 γ xy = β 3 + β 5
u1
Strain Field obtained from shape function:
v1
εx y23 0 y31 0 y12 0 Element Stiffness matrix:
u2
1
εy = 0 x32 0 x13 0 x21
2A v2 k = BT EBdV
γ xy x32 y23 x13 y31 x21 y12
u3 T
u = Nd k = B EBtA
v3
ε = ∂Nd
B = ∂N
Basic FEA Course
Stress along x axis in a beam modeled by CSTs Deformation of the lower-

left CST in the model
y 5a y,v a
P 3
x
P x,u
a
σx
v2
1 2
u2
Basic FEA Course
6.2.3. Triangular Elements (LST->Tria 6)
Displacement Field:
u = β1 + β 2 x + β 3 y + β 4 x 2 + β 5 xy + β 6 y 2
v = β 7 + β 8 x + β 9 y + β10 x 2 + β11 xy + β12 y 2
v3
y,v (x 3 ,y3 )
u3
Strain Field: v1 5 v2
6
4 x,u
(0,0)
ε x = β 2 + 2β 4 x + β5 y u1 (x2 ,0) u2
ε y = β 9 + β11 x + 2 β12 y
γ xy = ( β 3 + β 8 ) + ( β 5 + 2 β10 ) x + (2 β 6 + β11 ) y
Basic FEA Course
6.2.4. Bilinear Quadrilateral (Q4-> Cquad4)
y,v
a a Bilinear u and v are product of two linear
polynomials: (c1 + c2 x )(c3 + c4 y )
4 3
u = β1 + β 2 x + β 3 y + β 4 xy
x,u
v = β 5 + β 6 x + β 7 y + β 8 xy
1 2
Strain Field:
ε x = β2 + β4 y
ε y = β 7 + β8 x
γ xy = ( β 3 + β 6 ) + β 4 x + β 8 y
y
P
x
P M2 M2
M1
M1
Basic FEA Course
6.2.5. Tetrahedral Elements
Analogously, for a tetrahedral element, the shape functions are:
z w4
(x 4 , y 4, z )4
v
4 4 u4
x
y u = α1 x + α 2 y + α 3 z
w3
w1 v = α4 x + α5 y + α6 z
(x 3, y 3, z )3
(x 1, y 1, z )1 3
1
u3 w = α 7 x + α8 y + α9 z
u1 w2 v3
v1
2
(x 2, y 2, z )2
u2
v2
All matrices shown before and the Energy Method are used analogously for
the construction of [k] matrices.
Basic FEA Course
6.2.6. Higher Order Shape Functions
The shape functions presented here are LINEAR. It is also possible to use
elements of HIGHER ORDER FUNCTIONS (parabolic). For example,
elements with intermediary nodes have parabolic interpolation, with second
order polynomials.
The results of model analysis with these elements are more precise. On the
other hand, the size of the stiffness matrix (that is proportional to the number
of nodes) is much bigger.
TRIANGULAR QUADRANGULAR TETRAHEDRAL HEXAHEDRAL
Intermediary nodes for parabolic interpolation
Basic FEA Course
6.3. Isoparametric elements: Formulation
ξ =1/2
ξ=1
η η Quadrilateral may be nonrectangular
=1
ξ =-1/2
ξ =-1
Shape or interpolation function:
η =1/2 1
N1 = (1 − ξ )(1 − η )
4
ξ 1
N 2 = (1 + ξ )(1 − η )
4
Y,v η =-1/2
1
N 3 = (1 + ξ )(1 + η )
4
η =-1 1
X,u N 4 = (1 − ξ )(1 + η )
4
∂u ∂X ∂Y ∂u Element stiffness matrix:

∂ξ ∂ξ ∂ξ ∂X k = BT EBdV = BT EBt | J | dξdη
=
∂u ∂X ∂Y ∂u −1−1
∂η ∂η ∂η ∂Y
J
Basic FEA Course
6.4. Work Equivalent Nodal Loads
Element
q Structure
q
1 2
L 3L
1 2 qL/2 qL qL qL/2
qL/2 qL/2
L 3L
qB Fa L 2 1 qa
q Fb =
qA Fa Fb 6 1 2 qb
A B B L
A
L FaVa'+ FbVb' = v' (qds )
s
0
Basic FEA Course
6.5. Gauss Quadrature/Stress Calculation (1D)
I = 2φ1 φ Numerical Integration
1
1 n
→I ≈
I = φdξ  Wiφ
−1 i =1
ξ
-1 0 1 5 8 5
I = φ1 + φ2 + φ3
9 9 9
2 φ
I = φ1 + φ2 1
φ 3
1 2
b b
a a
ξ
ξ -1 0 1
-1 0 1 n = 1
→ξ = 0
3 b = .6
a= 3
n = 2
→ξ = ±
3 3
n = 3
→ξ = ± .6
Basic FEA Course
Exercise 2
Integrate the function below by using the Gauss
method (1D) described below considering:
1
I = x 2 + 2(dx)
−1
a) 1 integration point
b) 2 integration points
c) 3 integration points
d) Compare the previous results with the integral
value.
Basic FEA Course
6.6. Gauss Quadrature/Stress Calculation (2D)
3 3
ξ =− ξ=
3 η 3 Numerical Integration
4 3
2 η=
3 I = φ (ξ ,η )dξdη ≈ WiW jφ (ξi ,η j )
ξ
1 3 3
η=−
3
I ≈ φ1 + φ2 + φ3 + φ4
Gauss Point locations for integration of a
function φ = φ (ξ ,η ) in two dimensions
using order 2.
Basic FEA Course
6.7. Hourglass Mode
Complete Integration:
v
u N ≤ 2n − 1
2
N:polynomial order
n:integration rule
v
σx εx
u σ y = {D} ε y stress-strain Matrix
1
γ xy γ xy
Free DOF: 8 grids x 2DOF/grid=16-3 rigid body mode=13
3x3 rule 2x2 rule
Total stiffness rank=2x2x3=12

Total stiffness rank=3x3x3=27
1 zero energy mode!!!
Basic FEA Course
6.8. Stress Calculation
• Displacements are more accurate than stresses;
• Stresses are proportional to strains and strains are derivatives of
displacement;
• The most accurate values of stress are element center stresses and nodal
average stresses;
• Stresses of greatest interest usually appear at boundaries, but this is not
where stresses are most accurately computed.
q (uniform) q(uniform)
σx
u Exact
Exact
FE
FE
x x
Basic FEA Course
6.9. Stress Calculation
• Calculated stresses σ = EBd are often most accurate at Gauss Points;
• Stresses at nodes or at other element locations are obtained by
extrapolation or interpolation from Gauss Point Values;
• Element stress field is represented as bi- or trilinear in isoparametric
coordinates. For instance, in a plane element the σ x is represented as:
σ x = c1 + c2ξ + c3η + c4ξη

• Hexa-element polynomial of lower degree than contained in the B matrix
information has been discarded
Basic FEA Course
7. General Procedure of Calculus (for each
element)

Initially, the displacements matrix is obtained for each element, {ui } = [C ]{α i }
considering all the shape functions

The deformation vector is found by the derivate of {ε } = [B]{α i }
displacements – vector {ε};

The stress vector is found by application of Hooke’s Law – {σ } = [D]{ε }
vector {σ};

The VIRTUAL INTERNAL WORK, WORK the VIRTUAL EXTERNAL T
WORK,
WORK the VARIATIONAL and the STIFFNESS MATRIX are [k ] = [B ] [D ][B ]d ( vol )
vol .
calculated for the element.
T
Other operations that are necessary, like matrix inversion, are done as
presented before.
Basic FEA Course
8. Finite Element Program

A Finite Element computer program can be divided in:
• Pre-processor:
processor comprises a modeler, a module of material input data
and a mesh generator;
• Solver:
Solver finds all necessary matrices (including boundary conditions),
does coordinate transformations, numeric integrations, and solves the
analysis;
• Post-processor:
processor presents the results after the analysis, in an adequate
form to the user.
In the next page it is presented a flowchart of this program.
Basic FEA Course
Basic FEA Course
8. Final Considerations
The flowchart beside shows, by the
viewpoint of design, the number of
variables that influences the quality
of the analysis by finite elements.
For each step of the design, there is
an approximation caused by
assumed hypotheses or resources
limitations.
It would not be surprising if the
accumulated error (from these
hypotheses) is close to 50%. It
justifies the application of safety
factors.
Basic FEA Course
9. References
ALVES FILHO, A. 2000. Elementos Finitos: A Base da Tecnologia CAE, 1st
ed., Editora Érica Ltda, São Paulo, SP.
COOK, Robert D.1995. Finite Element Modeling for Stress Analysis, 1st
ed., John Wiley & Sons, Inc.
ROSS, C. T. F. 1985. Finite Element Methods in Structural Mechanics, 1st
ed., Ellis Horwood Limited, Chichester, West Sussex, England.

Fea Basic Course

Uploaded by

Copyright:

Available Formats

Fea Basic Course

Uploaded by

Document Information

Original Description:

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

Fea Basic Course

Uploaded by

Copyright:

Available Formats

Basic FEA Course

BASIC FINITE ELEMENT

Prepared by: DTE/GIS/SDD-

In 1950s, work in the aircraft

The name finite element was coined in

Geometry of the part Model of the part

Model with results

Newton’s Laws Potential Energy Strain Energy

Thermodynamic Law Principle of Virtual Work Mass Conservation Law

Maxwell Equations Principle of Energy Conservation

SOLUTION – FINITE ELEMENTS Mesh refinement,

It is important to remember that it is impossible to predict the exact solution for a

A model must be:

The next example (Ref.: Bathe,K.J.,Finite Element Procedures) illustrates this

material properties loading boundary conditions

In a preliminary analysis, it is assumed that the objective is to know the bending

ANALYTICAL FORM OF moment and deflection of the beam:

M = 27500 N.cm δ = 0.064 cm

To verify if the model is effective and reliable,

Next, there is another example (Ref. Boresi,A.P.,Sidebottom, O.M., ”Mechanics of

As expected, the 8 x 24 mesh brings closer results to the theoretical solution.

2. Solution of Linear Equations

Here there is a linear system of equations:

A11 x1 + A12 x2 + + A1n xn = C1 A11 A12 A1n x1 C1

Am1 x1 + Am 2 x2 + + Amn xn = Cm Am1 Am 2 Amn xn Cm

3. Principle of the Finite Element Method

3.1. Stiffness and Flexibility

Consider an elastic spring with an applied load P, as shown:

The flexibility is the opposite of the stiffness, then: u = f ×P

•If the structure is RIGID,

• If the structure is FLEXIBLE,

This elementary problem is a description of how the structures behave themselves.

Thousands of interconnected springs

General structures Different stiffness values in several orientations

Complex load combinations

For a practical problem of linear analysis, it is possible to do an analogy between a

From de equilibrium condition:

For each node, there must be an equilibrium condition:

The correspondent symmetric matrix system is:

tração compressão tração

Matrix of element (a): Matrix of element (b): Matrix of element (c):

The common points between ka − ka 0 0 u0 P0

[K LL ]{u L } = {PL } Displacements are defined

[K SL ]{u L } = {PS } Reactions are defined

3.3. Energy Functional

Next, it is presented a simple example of this principle, using a spring.

The Energy Functional is the total energy, given by:

P1 = k × (spring .displacement ) = k × (u1 − u2 )

F1 AE 1 − 1 u1 [k ] = AE element in Local Coordinates

4.2. Obtaining the Stiffness Matrix (2D)

[k0] = stiffness matrix for a 2D element in Global Coordinates

Observing the Coordinate System:

The relation between the forces is:

5.1. Derivation of the Deflection Equations

5.1. Derivation of the Deflection Equations

5.2. Solution of Beams (1D)

The resulting equations are:

NASTRAN results. There is a model with 12 beam