Vous êtes sur la page 1sur 15

Vibration with more (than one) degrees of freedom (DOF)

a) longitudinal vibration with 3 DOF

b) rotational (torsional) vibration with 3 DOF

c) bending vibration with 3 DOF

d) the 2D (planar) vibration of the flexibly supported rigid body with 3 DOF
The free longitudinal undamped vibration wit 2 DOF

ka kb kc
m1 m2

x1, v1, a1 x2, v2, a2

m1, m2 - the two masses, the two rigid bodies,


x1, x2 - the two independent coordinates, the two DOF,
v1  x 1 , v 2  x 2 - the two velocities, derivatives of corresponding coordinates,
a 1  v 1 , a 2  v 2 - the two accelerations, derivatives of corresponding velocities,
ka, kb, kc - the single stiffnesses.

The free-body diagram :

FSa FSb FSc


m1 m2

x1, v1, a1 x2, v2, a2

FSa, FSb, FSc - the single spring forces :


FSa  k a   a
FSb  k b   b
FSc  k c   c
The spring deformations are :
 a  x1 lengthening, the tension spring force,

 b  x 2  x 1 lengthening, the tension spring force,

 c  x 2 shortening, the pressure spring force.

The equations of motion are :


m1  a 1   Fi  FSa  FSb
m 2  a 2   Fi  FSb  FSc

m1  a 1  k a  k b   x 1  k b  x 2  0
m 2  a 2  k b  x 1  k b  k c   x 2  0
or :
m1  x1  k a  k b   x 1  k b  x 2  0
m 2  x 2  k b  x 1  k b  k c   x 2  0
with substitutions :
k 11  k a  k b
k 12  k 21  k b
k 22  k b  k c

m1  x1  k 11  x 1  k 12  x 2  0
m 2  x 2  k 21  x 1  k 22  x 2  0
finally in matrix form :
 m1 0   x1   k 11 k 12   x 1  0
0       
 m 2  x 2  k 21 k 22  x 2  0

or
Ma  K x  0
where :
m 0 
M 1
m 2 
- is the square mass matrix,
0

 a   x 
a   1   1 - is the column matrix (vector) of accelerations,
a 2  x 2 

k k 12 
K   11  - is the square stiffness matrix,
k 21 k 22 

x 
x   1 - is the column matrix (vector) of displacements,
x 2 
0
0  - is the column matrix (vector) of zeros.
0

Both M and K are symmetrical with respect to the main diagonal (k12=k21, m12=m21=0) and
have positive values on the main diagonal (m11, m22, k11, k22 > 0). Further M is diagonal (non-
zero values only on the main diagonal, zeros out of the main diagonal).
The supposed solution is :
x 1 t   C1  sin  t  
x 2 t   C 2  sin  t  

further :
x1t   C1   2  sin  t  
x 2 t   C 2   2  sin  t  

where C1, C2 - amplitudes of the corresponding coordinates,


 - natural circular frequency.

Put into the equations of motion :


 m1  C1   2  sin  t    k 11  C1  sin  t    k 12  C 2  sin  t    0
 m 2  C 2   2  sin  t    k 21  C1  sin  t    k 22  C 2  sin  t    0
The sin(0·t+) member can be canceled, the C1 and C2 unknowns can be factorized :
k 11 
 m1   2  C1  k 12  C 2  0
 
k 21  C1  k 22  m 2   2  C 2  0

The system of two algebraic equations with two unknowns is homogenous (zeros on the right
side). The trivial solution is :
C1 = C2 = 0
The trivial solution represents no motion. We search the non-trivial solution, representing
vibrating bodies :
C1  0, C2  0

The first equation gives the ratio between amplitudes C1 and C2 as :


C1  k 12

C 2 k 11  m1   2
while the second equation gives the same as :
C1 k 22  m 2   2

C2  k 21
This simply says that :
 k 12 k 22  m 2   2

k 11  m1   2  k 21
or :
k 11  
 m1   2  k 22  m 2   2   k 12    k 21 
k 11  m1   2
 k 22  m2   2
 k 12  k 21  0

This result can be expressed as :


k 11  m1   2 k 12
 
 k 11  m1   2  k 22  m 2   2  k 12  k 21  0 
k 21 k 22  m 2   2

This determinant is called frequentional determinant.

The substitution can be used : 2  


where  is called “eigenvalue”.

Then :
k11  m1    k 22  m 2    k12  k 21  0
or :
m1  m 2  2  k11  m 2  k 22  m1     k11  k 22  k12  k 21  0
what is called frequentional polynom.
The roots are two eigenvalues :

b  b2  4  a  c
1,2 
2a
where a  m1  m 2
b  k11  m 2  k 22  m1
c  k11  k 22  k12  k 21
and subsequently :
1,2  1,2

are two natural circular frequencies, and finally :


1,2
f1,2 
2
are two natural frequencies or eigenfrequencies.

The two eigenvalues 1,2, two natural circular frequencies 1,2 and two natural frequencies f1,2
are sorted in the ascending order, 1 < 2, 1 < 2, f1 < f2.
Once, the eigenvalue  is calculated, the amplitudes C1 and C2 can be calculated.
But the equations :
k 11  m1     C1  k 12  C 2  0
k 21  C1  k 22  m 2     C 2  0
are linearly dependent.
Simply second equation is only a certain multiple of the first one. It does not give the second
k 21 k  m2  
information. (The multiplier is  22 )
k 11  m1   k 12

The equations give only the ratio C1/C2 :


C1  k 12 k  m2  
  22
C 2 k 11  m1    k 21
not the final amplitudes C1 = ?, C2 = ?
For this reason the above presented results are not marked as amplitudes C1, C2, but v1, v2.
The infinite number of solutions exists, for example :
v1  k 12 v 2  k11  m1  
or
v1  k 22  m 2   v 2  k 21
or
k 11  m1  
v1  1 v2 
 k 12
or
 k 21
v1  1 v2 
k 22  m 2  
or
 k 12
v1  v2  1
k 11  m1  
or
k 22  m 2  
v1  v2  1
 k 21

or any other multiple of this results. The ratio v1/v2 is always the same.
These v1 and v2 are not the final amplitudes. The final amplitudes are their certain multiples
(the multiplier , see below). Because they refer to the coordinates (displacements) x1, x2,
they will be organized in column matrix :
 x   C  sin  t   0 
x   1   1 
x 2  C 2  sin  t   0 

 v    k 12 
v   1   
v 2  k 11  m1   

 C1   v1 
    
C 2  v 2 
The value of the multiplier  is determined from the initial conditions.

v 
The column matrix v   1  is called “mode shape” or “eigenvector”.
v 2 

 v    k 12 
As explained above, v   1     is only one from infinite number of solutions,
v 2  k 11  m1   
any other solution is any it’s multiple. The numerical values are changed, but their ratio is
kept. This fact leads to the process of normalization of mode shape. To normalize mode shape
means to multiply it by a certain number. The numerical values of the mode shape are
changed, but their ratio is kept.

Often used kind of normalization is normalization “to unit”. This means, that the highest
value in the mode shape is 1, others are less in the proper ratio. The largest value in the mode
shape is found and whole mode shape is divided by this value. The largest value, divided by
itself, is = 1, others, divided by the largest, are < 1 in proper ratio.
Example :
 v  123 
Mode shape before normalization : v   1    
v 2  456
the largest value is v2 = 456,

1 123   123
456  0.27
mode shape after normalization : v      456   
456 456  456  1 
Other often used kind of normalization is normalization “to mass matrix”. The denominator is

calculated as v T  M  v where M is the mass matrix, v is mode shape (before


normalization), vT is transponed mode shape (written as row matrix). All members in the
mode shape are divided by this denominator. For the mode shape after normalization
vT  M  v  1

Because two values of  exist, two sets of results exist. They are organized in so called
“modal matrix” or “matrix of mode shapes” :
v v12 
V   11
 v 21 v 22 

The first row corresponds to the first coordinate x1, the second row corresponds to the second
coordinate x2. The first column corresponds to the first natural frequency 1, the second
column corresponds to the second natural frequency 2.

v v12  x1
V   11
 v 21 v 22  x2

1 2

The single columns are single mode shapes.

The numerical example :

ka kb kc
m1 m2

x1, v1, a1 x2, v2, a2

m1 = 1 kg m2 = 2 kg ka = 3 N/mm kb = 2 N/mm kc = 1 N/mm


m 0  1 0
M 1 
m 2  0 2
kg
0

k  k b  k b   k 11 k 12   5  2 N  5000  2000 N
K a   
  kb k b  k c  k 21 
k 22   2 3  mm  2000 3000  m

K    M  v  0
k 11  m1   k 12
0
k 21 k 22  m 2  

k11  m1    k 22  m 2    k12  k 21  0
m1  m 2  2  k11  m 2  k 22  m1     k11  k 22  k12  k 21  0

a  m1  m 2  2 kg 2

b  k11  m 2  k 22  m1  13  103 kg  N  m 1

c  k11  k 22  k12  k 21  11  10 6 N 2  m 2

a  2  b    c  0


b  b 2  4  a  c 13  10 

3
13 10  3 2
 4  2  11  10 6
2a 22
1000
1,2  s 2
5500

31,6
1,2  1,2  s 1
74,2

1,2 5,03
f1,2   Hz
2 11,8

K    M  v  0
k 11    m1 k 12   v1  0
     
 k 21 k 22    m 2  v 2  0

1  1000 s 2  2  5500 s 2

5000  1000  1  2000   v1  0 5000  5500  1  2000   v1  0


          
  2000 3000  1000  2 v 2  0 
  2000 3000  5500  2 v 2  0

4000  v1  2000  v 2  0  500  v1  2000  v 2  0


 2000  v1  1000  v 2  0  2000  v1  8000  v 2  0
the second equation is simply first equation the second equation is simply first equation
dividing -2 times 4
4000  v1  2000  v 2  0  500  v1  2000  v 2  0

v1  1 simply choice v1  1 simply choice


4000  500
v2   v1  2 v2   v1  0,25
2000 2000
the first mode shape, referring to the second mode shape, referring to
the first natural circular frequency the second natural circular frequency
1 = 31,6 s-1 2 = 74,2 s-1

 v  1 v   1 
v   1    v   1   
v 2  2 v 2   0,25
normalization to unit :
the maximum value in v is 2

1  12  0,5
v     2   
2  2   1 

The mode shapes are organized in the modal matrix (the matrix of mode shapes).
0,5 1 
V 
 1  0,25
The rows in modal matrix refer to the coordinates, the columns - the mode shapes, refer to the
natural frequencies.

0,5 1  x1
V 
 1  0,25 x2

1 2

The mode shape interpretation.


The first mode shape, vibration with the natural circular 1 = 31,6 s-1.
0,5
v 
1
x1t   0,5  sin1  t  1  x 2t   1  sin1  t  1 

ka kb kc
m1 m2

x1, v1, a1 x2, v2, a2

Both bodies vibrate in the same direction.


The amplitude of the body m2 is twice the amplitude of the body m1.
The second mode shape, vibration with the natural circular 2 = 74,2 s-1.
 1 
v 
 0,25
x1t   1  sin 2  t   2  x 2t   0,25  sin 2  t   2 

ka kb kc
m1 m2

x1, v1, a1 x2, v2, a2

Both bodies vibrate in the opposite direction.


The amplitude of the body m2 is one quarter of the amplitude of the body m1.

The final vibration, the time function of the coordinates, is the linear combination of the mode
shapes :
x 1 t   0,5  1  sin1  t  1   1   2  sin 2  t   2 
x 2t   1  1  sin1  t  1   0,25   2  sin 2  t   2 

the velocities :
v1 t   x 1t   0,5  1  1  cos 1  t  1    2   2  cos  2  t   2 
v 2 t   x 2 t   1  1  cos 1  t  1   0,25   2   2  cos  2  t   2 

where 1 and2 are the coefficients of the linear combination, and 1 and2, together with
phase shifts 1 and2 are integrational constants, based on the initial conditions.
The initial conditions are four : t = 0 ... x1(t=0) = x10, v1(t=0) = v10, x2(t=0) = x20, v2(t=0) = v20
In the example : x10 = 10 mm, v10 = 1 m/s, x20 = 8 mm, v20 = 0,2 m/s
x 10  0,01  0,5  1  sin1    2  sin 2 
x 20  0,008  1  sin1   0,25   2  sin 2 

v10  1  0,5  1  1  cos 1    2   2  cos  2 


v 20  0,2  1  1  cos 1   0,25   2   2  cos  2 
and the solution :
1 = 15,72 mm 2 = 12,03 mm
1 = 36,4° = 0,636 rad 2 = 26,3° = 0,459 rad
finally :
x 1 t   C11  sin1  t  1   C12  sin 2  t   2 
x 2 t   C 21  sin1  t  1   C 22  sin 2  t   2 

where : C11 = 0,5·1 = 7,86 mm C12 = 1·2 = 12,03 mm


C21 = 1·1 = 15,72 mm C22 = -0,25·2 = -3,01 mm

x1 x1t   7,86  sin1  t  0,636  12,03  sin 2  t  0,459

0 0.2 0.4 0.6 0.8 1

x 2t   15,72  sin1  t  0,636  3,01  sin 2  t  0,459


x2

0 0.2 0.4 0.6 0.8 1

The modal coordinates


Let us go back to the system.

ka kb kc
m1 m2

x1, v1, a1 x2, v2, a2

The equations of motion are :


m1  x1  k a  k b   x 1  k b  x 2  0
m 2  x 2  k b  x 1  k b  k c   x 2  0

Let us analyze the specific case, in which ka = kb = kc = k and m1 = m2 = m.

k k k
m m

x1, v1, a1 x2, v2, a2


The equations of motion are :
m  x1  2  k  x 1  k  x 2  0
m  x 2  k  x 1  2  k  x 2  0
The coordinates x1 and x2 are called “primal coordinates” (defined primary) or “physical
coordinates” (they have direct physical meaning - displacement of the bodies m1 and m2).
Add and subtract both equations.
First equation + second equation :
m  x1  x 2   k  x 1  k  x 2  0
m  x1  x 2   k  x 1  x 2   0
First equation - second equation :
m  x1  x 2   3  k  x 1  3  k  x 2  0
m  x1  x 2   3  k  x 1  x 2   0
Now use the substitution :
y1  x 1  x 2
y 2  x1  x 2
y1  x1  x 2
y 2  x1  x 2

The equations of motion are then :


m  y1  k  y1  0
m  y 2  3  k  y 2  0
The new coordinates y1 and y2 are called “principal coordinates” or “modal coordinates”. The
modal coordinates are the linear combination of physical coordinates.
In the above described example (ka = kb = kc = k and m1 = m2 = m) the modal coordinates
have the following direct physical meaning :
y1 = x1 + x2 - twice the coordinate of the center of mass
x1  x 2
(the coordinate of the center off mass is )
2
y2 = x1 - x2 - distance between the two bodies.
This situation (the modal coordinates have direct physical meaning) is very unusual. Usually
the modal coordinates have no physical meaning, simply the linear combination of physical
coordinates.
Modal transformation
The modal coordinates are defined through modal transformation :
x  Vy
where x - column matrix (vector) of physical coordinates,
y - column matrix (vector) of modal coordinates,
V - modal matrix (the matrix of mode shapes).

In the above described example, where :


m1 = 1 kg m2 = 2 kg ka = 3 N/mm kb = 2 N/mm kc = 1 N/mm
and the modal matrix is :
0,5 1 
V 
 1  0,25
the modal transformation is :
x 1  0,5  y1  1  y 2
x 2  1  y1  0,25  y 2
and :
2 8
y1   x 1   x 2  0,222  x 1  0,889  x 2
9 9
8 4
y 2   x 1   x 2  0,889  x 1  0,444  x 2
9 9

That is clear, that the modal coordinates y1 and y2 have no direct physical meaning.

The equations of motion are :


M  x  K  x  0
Use the modal transformation :
x  Vy
x  V  y
The equations of motion are then :
M  V  y  K  V  y  0
Now multiply the equation by the transponed modal matrix VT (from the left side) :
V T  M  V  y  V T  K  V  y  V T  0  0
Define so called “modal mass matrix” and “modal stiffness matrix” as :
~
M  VT  M  V
~
K  VT  K  V
The equations of motion are then :
~ ~
M  y  K  y  0
This operation is often called “switch to modal space”, because y are “modal coordinates”.

In the above described example, where :


m1 = 1 kg m2 = 2 kg ka = 3 N/mm kb = 2 N/mm kc = 1 N/mm
and :
0,5 1  0,5 1 
V  VT   
 1  0,25  1  0,25
the modal mass matrix and modal stiffness matrix are :
~ 0,5 1  1 0 0,5 1  2,25 0 
M  VT  M  V         
 1  0,25 0 2  1  0,25  0 1,125
~ 0,5 1   5000  2000 0,5 1  2250 0 
K  VT  K  V        
 1  0,25  2000 3000   1  0,25  0 6188

Both matrices are diagonal (zeroes out of the main diagonal). It can be proofed, that it is not
an accident, but the rule.
This says, that the two equations of motion in the modal space are independent.
~ ~
M  y  K  y  0

2,25 0   y1  2250 0   y1  0


 0 1,125  y    0     
6188 y 2  0
   2 
2,25  y1  2250  y1  0
1,125  y 2  6188  y 2  0
This represents the two independent cases with 1 DOF. We can easily find the natural circular
frequencies as :

2250 6188
1   31,6 s 1 and 2   74,2 s 1
2,25 1,125
Compare with the above presented solution.
This approach, switch to the modal space, to the modal coordinates, and solve independent
DOF, becomes useful especially in the area of forced vibration.

Vous aimerez peut-être aussi