You are on page 1of 9

Multi-degree of freedom systems review

Free vibration analysis


m11 m12 x1 c11 c12 x1 k11 k12 x1 F1
m m x c c x k
21 22 2 21 22 2 21 k22 x2 F2

No damping

Equation of motion:
[ M ]
x [ K ]{x} 0

We know from theory of differential


A
equations that the solution is: x ( B)e j t

Thus we need to find the frequency and


the amplitudes A and B.

Characteristic equation: det([ K ] 2 [ M ]) 0

2 eigenvalues of [M]-1[K]
( ) eigenvectors of [M] [K].
A -1
B
: natural frequencies (number = number of
d.o.f.)
A
mode shapes (number = number of
i
B

d.o.f.).

General expression for free vibration:



x (t ) c1 1e j1t c2 2 e j 2 t

c1 1cos(1t 1) c2 2cos(2 t 2 )

Conclusions:

Vibration=mode1*c1+mode2*c2

Participation of each mode depends on


initial conditions

Steps in free vibration analysis

1. Find natural frequencies and mode shapes


2. Express displacements in terms of natural
frequencies and mode shapes:
x1 (t ) A A
c1 1 cos(1t 1 ) c2 2 cos(2t 2 )
x2 ( t ) B1 B2

3. Find constants and phase angles from


initial conditions

Example
x1 x2

k k k
m m

Second natural frequency:


3k
( 2) A B
m

Mode shape for above frequency:


A

A
Usually assume first entry is 1. Therefore,
1
mode shape is: 1
Masses move in opposite directions
t

First natural frequency:


k
( 2) A B
m

1
Mode shape for above frequency: 1

Masses move together.

t
Displacements of two masses are superposition
of displacements in the two modes:
x1 (t ) 1 k 1 3k
c1 cos( t 1 ) c2 cos( t 2 )
x2 ( t ) 1 m 1 m

Find c1, c2 and 1 , 2 from initial conditions

Orthogonality of mode shapes and


decoupling of equations of motion

Mode shapes are orthogonal wrt the mass


and stiffness matrices:

1T [ ]2 0

1T [ K ]2 0

Therefore:
[]T [ M ][ ] [ M' ]
[]T [ K ][ ] [ K ' ]

where [M'] and [K'] are diagonal matrices.


We can use the orthogonality property of the
mode shapes to uncouple the equations of
motion through the following
transformation:

x (t ) 1(t ) p1(t ) 2 (t ) p2 (t ) p(t )

This transformation expressed the unknown



displacements (t ) in terms of the basis
x

vectors 1(t ) and 2 (t ) . If we substitute the
expression for the displacement into the
equations of motion then we have two new

equations wrt the unknown coordinates p(t ) .
M [] p (t ) K [] p(t ) 0

These equations are equivalent to the


original system but they can be easily
decoupled.
M p (t ) K p (t ) 0
where
M []T [ M][]
K []T [K ][]

[M'] and [K'] are diagonal matrices, which


means that we have two uncoupled
equations with unknowns coordinates p1(t)
and p2(t).

Therefore, the equations of motion can be


written wrt to the new coordinate system can
be written as follows:
p1(t ) k11
m11 p1(t ) 0
p1(t ) k22
m22 p2 ( t ) 0

In some cases it is better to solve these two


uncoupled equations than the coupled
equations. To solve these equations we need
the initial conditions for coordinates p1(t)
and p2(t). These are obtained as follows:
p1(0) 1 x1(0)
[]
p2 ( 0 ) x2 ( 0 )
p 1(0) 1 x1(0)
[]
p 2 (0) x 2 (0)
Procedure for calculating free vibration
response by uncoupling the equations of
motion

1. Determine matrices:
M []T [M ][]
K []T [K ][]

2. Determine initial values of coordinates


p1(t) and p2(t) and initial values of the
derivatives.

3. Solve uncoupled equations for p1(t) and


p2(t):
p1(t ) k11
m11 p1(t ) 0
p1(t ) k22
m22 p2 ( t ) 0

where m'ii and k'ii are the diagonal elements


of matrices M' and K', using the initial
conditions found in step 2.

4. Recover displacements x1(t) and x2(t)


using the following equations:

x (t ) 1(t ) p1(t ) 2 (t ) p2 (t ) p(t )

You might also like