Anda di halaman 1dari 25

Week 11

1
Part 4 & 5. Mainly based on the text book by D.J. Inman
Week 11
2
4. More Than 2 DOF
Systems with any finite number of DOF can be analyzed by using
the modal analysis procedure explained before.
In each axis, we can also have
rotation besides translation.
If the forward motion is x
2
, then

2
is called roll,
3
is called
yaw and
4
is called pitch.
Week 11
3
More Than 2 DOF
M and K are n n matrices, and are n 1 vectors. x x
A Free Body Diagram of the system of the figure yields n equations of
motion in the form:

n = 1, 2, 3, , n
Week 11
4
Matrices and Vectors
Equations of motion in the matrix form:
Week 11
5
Approach to Solution
For such systems, the process stays the same just more
modal equations.
Steps as explained in the previous lecture.
There are now n natural frequencies,
which correspond to the eigenvalues
of the n n matrix M
1/2
K M
1/2

The eigenvalues can be found with
the same method:

which yields n
th
order polinomial in
Follow the steps given previously.
Week 11
6
Nodes of a Mode Shape
Note that for more than 2 DOF, a mode shape may have a zero
valued entry. This is called a node of a mode.
A node of a mode means there is no motion of the mass or
coordinate corresponding to that entry, at the natural frequency
associated with that mode.
They make great mounting points in machines
u
2
=
0.2887
0.2887
0
0.2887

(
(
(
( node
Week 11
7
Rigid Body Mode
A rigid body mode is the mode associated with a zero frequency.
E.g. as a train rolls down its track, the cars vibrate relative to one
another.
The system is not constrained and can move as a rigid body
unrestrained, rigid body motion.
Physically if this system is displaced we would expect it to move off
the page whilst the two masses oscillate back and forth
Week 11
8
Example: Rigid body mode
The free body diagram of the figure (previous slide) yields:
1 1 2 1 2 2 2 1
1 1 1
2 2 2
( ) and ( )
0 1 1 0
0 1 1 0
m x k x x m x k x x
m x x
k
m x x
= =

( ( ( ( (
+ =
( ( ( ( (


Solve for the free response given:
m
1
= 1 kg, m
2
= 4 kg, k = 400 N subject to
x
0
=
0.01
0

(
m and v
0
= 0
Week 11
9
Example: Rigid body mode
Following the steps from the previous lecture:
( ) ( )
1/ 2
1/ 2 1/ 2
2
1 2 1 2
1 0
1.
1
0
2
1 0 1 0
1 1 400 200
2. 400
1 1
1 1 200 100 0 0
2 2
4 2
3. det 100det 100 5 0
2 1
0 and 5 0, 2.236 rad/s
M
K M KM
K I

e e


(
(
=
(

( (

( (
( (
= = =
( (
( (



| |
(
= = =
|
(


\ .
= = = =
Indicates a rigid body motion
Week 11
10
Example: Rigid body mode
4. Calculate the eigenvectors and note in particular that they cannot
be zero even if the eigenvalue is zero.
= 0 100
4 0 2
2 1 0

(
v
11
v
21

(
=
0
0

(
4v
11
2v
21
= 0
v
1
=
1
2

(
or after normalizing v
1
=
0.4472
0.8944

(
Likewise: v
2
=
0.8944
0.4472

(
P =
0.4472
0.8944

0.8944
0.4472

(
As a check, note that:
| |
and diag 0 5
T T
P P I P KP = =
Week 11
11
Example: Rigid body mode
5. Calculate the matrix of mode shapes
6. Calculate the modal initial conditions
7. Now compute the solution in modal coordinates (Eq. 4.66 & 4.67)
and note what happens to the first mode.
Week 11
12
Example: Rigid body mode
Since e
1
= 0 the first modal equation is:
And the second modal equation is:
r
1
+ (0)r
1
= 0
r
1
(t ) = a + bt
r
2
(t ) + 5r
2
(t ) = 0
r
2
(t ) = a
2
cos 5t
Oscillation
Rigid body translation
Week 11
13
Example: Rigid body mode
7 (continued). Applying the modal initial conditions to the first modal
equation yields:




Similarly for the second modal equations we get:
Week 11
14
Example: Rigid body mode
8. Transform the modal solution back to the physical coordinate
system:
x(t ) = Sr(t ) =
0.4472 0.8944
0.4472 0.2236

(
0.0045
0.0089cos 5t

(
x(t ) =
x
1
(t )
x
2
(t )

(
=
2.012 + 7.60cos 5t
2.012 1.990cos 5t

(
(
10
3
m
Each mass is moved a constant distance and then oscillates at
a single frequency.
Week 11
15
Order of the Frequencies
It is convention to name the lowest frequency
1
so that
1
<
2
<

3
<
Order the modes (or eigenvectors) accordingly.
In term of computing the time response, this order does not make
a difference.
However:
When we measure frequencies, they appear lowest to highest
Physically, the frequencies respond with the highest energy in
the lowest mode (important in flutter calculations, run up in
rotating machines, etc.)
Week 11
16
5. Using Numerical Computation
Examples here are computed using Matlab. You can do the same
using any other computational software, e.g. Mathematica,
Matcad. If using Fortran or C++, then you have to build your own
functions.

First, take a look at the case where the system is dynamically
coupled:
A dynamically coupled system is one in which the mass matrix is
not diagonal mostly when we model in FEM
When the mass matrix is diagonal, finding M
1/2
and M
1
are all
simple calculations we have seen this in previous examples
Here we address what to do when M is not diagonal
Week 11
17
Dynamically Coupled System
The best way to compute the inverse of a matrix is to use
Gaussian Elimination. In Matlab this is: M\I.
The best way to compute the matrix square root is not to use GI,
instead use the Cholesky Decomposition.


L is a lower triangular matrix
Be careful: In Matlab, the command chol(M) returns an upper
triangular matrix.
I L M L LL M
T T
= =
1 1
) ( that so
I M(U) ) (U U U M
T T
= =
1 1
that so
Week 11
18
Dynamically Coupled System
Calculating Cholesky Decomposition in Matlab:
U = chol(M) %upper triangular matrix
L = U %transpose U = lower tri. Matrix
>> M = [5 2 0; 2 4 1; 0 1 3]
M =
5 2 0
2 4 1
0 1 3

>> U = chol(M)
U =
2.2361 0.8944 0
0 1.7889 0.5590
0 0 1.6394
>> L = U'
L =
2.2361 0 0
0.8944 1.7889 0
0 0.5590 1.6394

>> L*L'
ans =
5.0000 2.0000 0
2.0000 4.0000 1.0000
0 1.0000 3.0000
Week 11
19
Modal Analysis Using Cholesky
1) Calculate L from M
2) Calculate
3) Calculate the eigenvalue solution for
4) Normalize the eigenvectors v
i
and form the matrix P
5) Compute S =(L
T
)
1
P and S
1
=P
T
L
T

6) Calculate the modal initial conditions:

7) Use the modal initial conditions to compute r(t)
8) Multiply r(t) by S to get the solution x(t)=S r(t)
1 1
) (
~

=
T
L K L K
i i i
K K v v = :
~
0
1
0
1
) 0 ( , ) 0 ( x r x r

= = S S
Week 11
20
Example: Dynamically coupled
Using the mass and stiffness from a bar FEM.
>> M=[2 1;1 2]; K=[1 -1;-1 1];

>> inv(M)
ans =
0.6667 -0.3333
-0.3333 0.6667

>> U=chol(M)
U =
1.4142 0.7071
0 1.2247

>> L=U'
L =
1.4142 0
0.7071 1.2247
>> inv(L)*M*inv(L')
ans =
1.0000 0.0000
0.0000 1.0000

>> K_tilde=inv(L)*K*inv(L')
K_tilde =
0.5000 -0.8660
-0.8660 1.5000

>> [P,D]=eig(K_tilde)
P =
-0.8660 -0.5000
-0.5000 0.8660

D =
0 0
0 2
Week 11
21
Example: Dynamically coupled
Calculate S and inverse of S, check that P
T
P = I
>> S=inv(L')*P

S =

-0.4082 -0.7071
-0.4082 0.7071

>> inv(S)

ans =

-1.2247 -1.2247
-0.7071 0.7071

>> P'*P

ans =

1 0
0 1
Week 11
22
If Mass is Diagonal
Note the difference in calculating the mass (here we can use M
1/2
):
>> M=[9 0; 0 1]; K=[27 -3; -3 3];

>> Msq=sqrt(M) %only for diagonal M

Msq =
3 0
0 1


>> Kt = inv(Msq)*K*inv(Msq)

Kt =
3 -1
-1 3
>> [P,D]=eig(Kt)

P =
-0.7071 -0.7071
-0.7071 0.7071

D =
2 0
0 4
Week 11
23
Numerical Simulation of Time Response
Assume the initial conditions, then continue the steps:
x0=[0; 1]; xd0=[0; 0];

S=inv(Msq)*P;
Sinv=inv(S);

r0=Sinv*x0;
rd0=Sinv*xd0;

%Find the free response in modal coordinates
tmax = 10;
numt = 1000;
lambda = diag(D);

t = linspace(0,tmax,numt);
[T,W]=meshgrid(t,lambda.^(1/2));
Week 11
24
Numerical Simulation of Time Response
% Form the equation for time response from t=0 to t=10
R0 = r0(:,ones(numt,1));
RD0 = rd0(:,ones(numt,1));
r = RD0./W.*sin(W.*T) + R0.*cos(W.*T); %See slide 9 week 10plus

% Transform back to physical space
x = S*r;

% Plot results
figure

subplot(2,1,1) % in modal coordinate
plot(t,r(1,:),'-',t,r(2,:),'--')
title('free response in modal coordinates')
xlabel('time (sec)')
legend('r_1','r_2')

subplot(2,1,2) % in physical coordinate
plot(t,x(1,:),'-',t,x(2,:),'--')
title('free response in physical coordinates')
xlabel('time (sec)')
legend('x_1','x_2')
Week 11
25
Numerical Simulation of Time Response

Anda mungkin juga menyukai