Structural dynamics 8.

FINITE ELEMENT METHOD

The equation system obtained by f.e.m. is
 m  u    c  u    k   u
The generalised load vector and stiffness
matrix have been studied in the f.e.m.
course and are identical in static and
dynamic analyses.
  p(t )

The damping matrix is not obtained by finite

with element discretisation. The different ways of
introducing and dealing with damping have
been studied in lesson 6.

 m : mass matrix The resolution of the equation system, by

mode superposition or by direct time
 c : damping matrix integration methods has been studied in
lessons 4 and 6.
 k : stiffness matrix
The only remaining work is to present the
derivation of the element mass matrix. The
 p(t ) : generalised load vector
assembly of the element mass matrices to
the structure mass matrix is performed in
 u(t ) : displacement vector the same way as for the stiffness matrix.
 Structural dynamics 8.2

CONSISTENT ELEMENT MASS MATRIX

The equation of motion can be derived by
equating the work done by the externally applied
loads (external work) with the work absorbed by
inertial and dissipative forces (internal work) for
any virtual displacement (that is, for any
imagined small motions that satisfies
compatibility and essential boundary conditions.
u and  represent virtual displacements
and their corresponding strains.
Finite element discretisation provides
 u   N  d  u   N  d      B  d
Shape functions [N] are functions of space
while nodal d.o.f. d are function of time.

A linear elastic material is assumed

Wi          u
T T
 u 
  dv     E  
v
n
We    u  F dv    uT   ds    uiT  f i
T
The internal and external virtual works can
v s i 1
then be rewritten as

where F and  are the prescribed body

forces and surface tractions, fi and ui
T 
Wi   d     B T
 E  B  d   N T   N  d   dv 

v 
represent prescribed concentrated loads and
 n 
their corresponding virtual displacements at a We   d
T
  N T
 F dv    N T
   ds    N T
 f  i
total of n points,  is the mass density. v s i 1 
 Structural dynamics 8.3

The last equations in the preceding page are

This can be rewritten as
T 
Wi   d     B T
 E  B  d   N T   N  d   dv 

 m  u    k   u   p(t )
v 

 n 
We   d T
  N T
 F dv    N T
   ds    N T
 f  i
with
v s i 1   k     B T  E  B dv
v

 m     N T  N dv
Wi  We is true for arbitrary d, which gives v

   N  N dv  d
     B T  E  B dv  d   p
T
The mass matrix defined in the last equation
v v is called consistent element mass matrix.
n The word consistent emphasizes that this
 p    N  F dv    N   ds    N T  f i
T T
form has been derived using the same shape
v s i 1
functions as the element stiffness matrix.
 Structural dynamics 8.4
BERNOULLI 2D BEAM ELEMENT
shape functions

v1 v2
1 2
1 1
EI m mA
1 L 2 2 3 2
x x  x
1( x)  1  3    2    2 ( x)  x 1  
L L  L
v1(t ) 
 (t ) 
v ( x,t )   N  d  d   1 
v 2 (t )
 2 (t )  1 1

2
x x
3
x2  x 
 N   1( x )  2 ( x )  3( x )  4 ( x )   3( x)  3    2    4 ( x)    1
L L 
L L

 156 22L 54  13L   12 6L  12 6L 

 2
13L  3L2   2 2 
L
mL  22L 4L EI  6L 4L  6L 2L 
mij   m  i ( x )  j ( x ) dx  m    k  3
0 420  54 13L 156  22L  L   12  6L 12  6L 
 13L  3L2  22L 4L2   6L 2L2  6L 4L2 
   
 Structural dynamics 8.5
BERNOULLI 2D BEAM ELEMENT WITH AXIAL DEFORMATIONS

v1 v2 A bar element is superposed to the previous beam element.

1 2
EI A m
u( x,t )  1( x ) u1   2 ( x ) u2
1 u1 2 u2
L

The shape functions for the bar element are

x x
1( x )  1    2 ( x) 
L L

140 0 0 70 0 0   EA L 0 0  EA L 0 0 
 0 156 22L 0 54  13L   0 12EI L3 6EI L2 0  12EI L3 6EI L2 
   
2 2
 0 22L 4 L 0 13L  3L   0 6EI L2 4EI L 0  6EI L2 2EI L 
 m   mL     k    
420 70 0
 0 140 0 0    EA L 0 0 EA L 0 0 
3 2 3 2
 0 54 13L 0 156  22L   0  12EI L  6EI L 0 12EI L  6EI L 
 2 2   
 0  13L  3L 0  22L 4L   0 6EI L2 2EI L 0  6EI L2 4EI L 
 Structural dynamics 8.6
LUMPED MASS MATRIX
The rotational inertia is defined by considering
For certain applications it is better to use a
that a uniform slender bar of length L/2 and
diagonal (or lumped) mass matrix.
mass mL/2 is attached at each node.
Two obvious advantages are less storage
The associate inertia moment is
space and less processing time, especially in
case of an explicit time integration scheme. J  (mL / 2)(L / 2)2 / 3

PARTICLE MASS LUMPING

mL3
Example : Bernoulli 2D beam element
1
mL 24 mL 2
2 2
v1 v2 u1 u2
1 2 1 2
EI m mL3
mA
1 2 24
L

1 0 0 0 
 2
0 
mL 0 L / 12 0
An easy way of obtaining a lumped mass  m  
matrix is to replace the distributed mass by 2 0 0 1 0 
two particles of mass mL/2 at each node. 0 0 0 L2 / 12

 Structural dynamics 8.7
HRZ LUMPING

Different methods can be used to transform the consistent element mass matrix and obtain a
diagonal matrix. One of them is the HRZ lumping.
The idea is to compute only diagonal terms of the consistent element mass matrix, then scale
them so as to preserve the total element mass. Since there may be both translational and
rotational d.o.f., the method for an element of total mass M is

1. Compute diagonal coefficients mij of the consistent element mass matrix.

2. For each coordinate direction in which motion is described by the element d.o.f.
a. determine a number S by adding the mii associated with translational d.o.f. only.
b. multiply all coefficients mij associated with this direction by the ratio M/S.

Example : Bernoulli 2D beam element

1. 2.a. 2.b.
156 0 0 0  1 0 0 0 
 312
mL  0 4L
2
0 0  S mL  2
mL 0 L / 39 0 0 
 m   420  m  
420  0 0 156 0  2 0 0 1 0 
 0 M  mL
 0 0 4L2  0
 0 0 L2 / 39
 Structural dynamics 8.8
EXAMPLE
Mass and stiffness matrices for the structure
The natural frequencies and eigenmodes
Le  L / 2
of a simply supported beam are studied.
The beam is discretised in 2 elements.
 12 6Le  12 6Le 0 0 
 2 2 
6
 e L 4 Le  6 Le 2 Le 0 0 
 
EI   12  6Le 12+12  6Le+6Le  12 6Le

EIALm
 k = 3  2 2 2 2 
Le 6Le 2Le  6Le+6Le 4Le  4Le  6Le 2Le
 
 0 0  12  6Le 12  6Le 
mA  2 2 
 0 0 6Le 2Le  6Le 4Le 
1 2 3
1 3  156 22Le 54  13Le 0 0 
2  2 2 
 22Le 4Le 13Le  3Le 0 0 
v1 v2 v3  
mLe  54 13Le 156+156  22Le+22Le 54  13Le
 m = 
L/2 L/2 420   13Le  3Le  22Le+22Le
2 2
4Le  4Le
2
13Le  3Le 
2
 
 0 0 54 13Le 156  22Le 
 2 2 
 0 0  13Le  3Le  22 Le 4 Le 
 Structural dynamics 8.9

Boundary conditions

      v1        v1   p1(t ) 

                 0 
v1  v 3  0     1    1  
       2
v        2
v  0 
v1  v3  0           
       2        
 2  0 
p1(t ) , p3 (t ) unknowns       v3        v 3   p3(t )
          
       3         3   0 

1 2 3
1 3
2

v1 v2 v3
Equation system

 4L2e 13Le  3L2e 0   θ1   4L2e  6Le 2L2e 0   θ1  0

         
mL  13Le 312 0  13Le  v2  EI   6Le 24 0 6Le  v 2  0
   + 3    =  
420  3L2 0 8L2e  3Le2 
θ2  L  2L2e 0 8L2e 2
2Le θ2  0
e
         
 0  13Le  3L2e 2 
4Le  θ3 
  0 6Le 2L2e 2
4Le  θ3  0
 
 Structural dynamics 8.10
Solution of the eigenvalue problem (exact values in parenthesis)

EI EI EI EI
ω1 = 9.91 ω2 = 43.8 ω3 = 110 ω4 = 201
mL4 mL4 mL4 mL4
(9.87) (39.5) (88.9) (158)

 1   1  1   1
0.32 L   0  0.05 L  0
       
1    2    3    4   
 0   1   0   1
  1   1    1   1 

eigenmode 1 eigenmode 2 eigenmode 3

2  0
1  1  2  1
1  1  3  1 1  1  3  1
3  1 2  0

v 2  0.32L v2  0
v 2  0.05L