EST56 - Tópico 5b

EST - 56
DINÂMICA DE ESTRUTURAS E
AEROELASTICIDADE
Prof. Airton Nabarrete
Tópico 5-b: Análise Modal
1
Instituto Tecnológico da Aeronáutica
Contents
o Modal Analysis
o More then 2 DOF
o Nodes of a mode shape
o Cholesky decomposition
2
MODAL ANALYSIS
3
Introduction
 Say we have physical coordinates x and want to

transform to some other coordinates u
u1 = x1 + 3 x2  u1  1 3   x1 
⇒ = 1 −3  x 
u2 = x1 − 3 x2  u2    2
 Physical coordinates are not always the easiest to work
in
 Eigenvectors provide a convenient transformation to
modal coordinates
4
Normalized Equations of Motion
Review of the Equations of Motion (EOM):
Start with Mx(t ) + Kx = 0 where x is a

vector and M and K are matrices. Assume
initial conditions x 0 and x 0 . Rewrite as
x + Kx = 0 and let
1 1
M M

2 2

q
− 12
M x=q⇒x=M
1
2
q (coord. trans. #1)
5
Normalized Equations of Motion
Review of EOM (cont.):
− 12
Premultiply the EOM by M
~ (4.55)
− − −
+ q=q
 + Kq = 0
1 1 1 1
M


2
M 
q2
M K M 2 2

I ~ ~T
K=K
 Now we have the mass matrix as an identity matrix
 Guarantees real eigenvalues and distinct, mutually
orthogonal eigenvectors
6
Matrix of Eigenvectors
The matrix of eigenvectors can be used to decouple the
equations of motion:
If P orthonormal (unitary), PT P =I ⇒ PT =P −1
Thus, PT K P = Λ = diagonal matrix of eigenvalues.
 + K q =
Back to q 0. Make the additional coordinate
transformation q =Pr and premultiply by PT (coord. trans. #2)
r + PT K P=
PT P r + Λ=
r I  r 0 (4.59)
 Now we have decoupled the EOM i.e., we have n

independent 2nd-order systems in modal coordinates r(t)
7
Decoupled Equations of Motion
Writing out r1 (t )  ω12
1 0    0   r1 (t )   0 
 0 1     + 2  =
  (4.60)
equation (4.59)
   r2 (t )   0 ω2   r2 (t )  0 
yields
r1 (t ) + ω12 r1 (t ) =
 0 (4.62)
⇒
r2 (t ) + ω22 r2 (t ) =
 0 (4.63)
We must also transform the initial conditions

 r1 (0)   r10 
= r0  =  =  P T
=
q (0) P T
M 1/ 2
x(0) (4.64)
 r2 (0)   r20 
 r1 (0)   r10 
=
r0  =  =  P T
=

q (0) P T
M 1/ 2
x (0) (4.65)
 r2 (0)   r20 
8
Physical vs. Modal Coordinates
This transformation takes the problem from coupled
equations in the Physical Coordinate System x(t) in to
decoupled equations in the Modal Coordinates r(t) .
r1
(ω1)2
x1 x2
k1 k2 1
m1 m2
(ω2)2 r2
Physical Coordinates: 1
Coupled equations
Modal Coordinates:
Figure 4.5 (Inman) Uncoupled equations
−1
x= M Pr 2
9
Modal Transforms to SDOF
1
T
 The modal transformation P M 2 transforms
a Two DOF system into 2 SDOF systems.
 This allows us to solve the two decoupled SDOF

systems independently using the methods we have
seen before for One DOF.
 Then we can use the inverse transformation to obtain

the solution in terms of the Physical Coordinates.
10
Response in Modal Coordinates
The free response is calculated for each mode independently
using the formulas from chapter 1
ri 0
ri (t ) =sin ωi t + ri 0 cos ωi t , i=
1, 2
ωi
or (see Window 4.3 for a reminder)
ri 20 ωi ri 0
ri (t ) =
r + 2
sin(ωi t + tan −1
), i=
1, 2
i0
ω i
2
ri 0
Note, the above assumes that no frequency is zero
11
Response in Physical Coordinates
Once the solution in Modal Coordinates is determined (ri) then
the response in Physical Coordinates is computed:
 With n DOFs these transformations are:
x(t ) = S r (t )
(n x 1) (n x n ) (n x 1)
− 12
where S =M P
(n x n ) (n x n ) (n x n )
where, n = 2 in the previous slides (Two DOF)
12
Modal Analysis
Steps in solving via modal analysis:
Window 4.4 (Inman)
13
Example 4.3.1 via MATLAB
9 0   27 −3 0  0 
=M  =  , K  = , x(0) =  , x (0)  
0 1   −3 3  1  0 
 Follow steps in Window 4.4 (pg 300)

−1 −1 −1
1) Calculate M 2 2) Calculate K = M K M 2
˜ 2
» Minv2 = inv(sqrt(M)) »Kt =Minv2*K*Minv2

Minv2 = Kt =
0.3333 0 3 -1
0 1.0000 -1 3
14
Example 4.3.1 via MATLAB (cont.)
3) Calculate the symmetric eigenvalue problem for K-tilde
» [P,D] = eig(Kt);
» [lambda,I]=sort(diag(D)); % just sorts smallest to largest
» P=P(:,I); % reorder eigenvectors to match eigenvalues
lambda =
2
4
P=
-0.7071 -0.7071
-0.7071 0.7071
15
4) Calculate S = M^(-1/2) * P and Sinv = P^T * M^(1/2)
» S = Minv2 * P;
» Sinv = inv(S);
5) Calculate the modal initial conditions
» r0 = Sinv * x0;
» rdot0 = Sinv * v0;
16
6) Find the free response in modal coordinates

» tmax = 10;
» numt = 1000;
» t = linspace(0,tmax,numt);
» [T,W]=meshgrid(t,lambda.^(1/2));
Use Tony's trick

» R0 = r0(:,ones(numt,1));
» RDOT0 = rdot0(:,ones(numt,1));
» r = RDOT0./W.*sin(W.*T) + R0.*cos(W.*T);
7) Transform back to physical space

» x = S*r;
17
Plot results
» figure
» subplot(2,1,1)
» 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)
» plot(t,x(1,:),'-',t,x(2,:),'--')
» title('free response in physical coordinates')
» xlabel('time (sec)')
» legend('x_1','x_2')
18
Modal and Physical Responses
Free response in modal coordinates
4
Modal Coordinates:
r1
Independent 2 r2
oscillators
0
-2
λ1 = 2 ⇒ ω1 = 2 -4
0 1 2 3 4 5 6 7 8 9 10
⇒ T1 = 2π = 4.44 sec, π sec 4.44 sec
λ 2 = 4 ⇒ ω1 = 2 4
Free response in physical coordinates
⇒ T2 = π sec x1
2
x2
0
Physical
Coordinates: -2
Coupled oscillators
Note IC -4
0 1 2 3 4 5 6 7 8 9 10
19
MORE THEN 2 DOF
20
More then 2 Degrees of Freedom
Extending previous section to any number of degrees of
freedom:
Fig 4.7
Fig 4.8
21
Extending previous section to any number of degrees of
freedom:
Fig 4.8
A free body diagram (FBD) of the system of figure 4.8 yields the
n equations of motion of the form:
mi xi + ki ( xi − xi −1 ) − ki +1 ( xi +1 − xi ) = 0, i = 1, 2,  , n (4.81)
22
Writing all n of these equations and casting them in matrix
form yields: x(t ) + Kx(t ) =
M 0, (4.80)
 x1 (t )   
x1 (t ) 
 x (t )   
x (t ) 
=
where, x(t ) = 2  , 
x(t )  2 
     
    (4.84)
 n 
x (t )  
xn (t ) 
m1 0 0  0 k1 + k 2 − k2 0  0 
0 m
 2 0  

 −k
 2 k 2 + k3 − k3  
M =0 0    K = 0 − k3   
   
   mn −1 0     k n−1 + k n − kn 
 0  0 mn   0  − kn k n 
(4.82) (4.83)
23
Modal Analysis
For such systems as figure 4.7 and 4.8 the process stays the
same…just more modal equations result:
Process stays the same as section 4.3

r1 (t ) + ω12 r1 (t ) = 0
r2 (t ) + ω 22 r2 (t ) = 0
Just get more modal equations,
r3 (t ) + ω 3 r3 (t ) = 0
2
one for each degree of freedom
 (n is the number of DOF)
rn (t ) + ω n2 rn (t ) = 0
See example 4.4.2 for details
24
NODES OF A MODE SHAPE
25
Node of a Mode Shape
The second mode shape of Example 4.4.3 has a node:
 Note that for more then 2 DOF, a  0.2887 

mode shape may have a zero  0.2887 
valued entry u2 =  
 This is called a node of a mode.  0  node
 
 −0.2887 
 A node of a mode means there is
no motion of the mass or
(coordinate) corresponding to that
entry at the frequency associated
with that mode.
They make great mounting points in machines
26
Node of a Mode Shape
A rigid body mode is the mode associated with a zero
frequency:
Fig 4.12
 Note that the system in Fig 4.12 is not constrained and

can move as a rigid body
 Physically if this system is displaced we would expect it to
move off the page whilst the two masses oscillate back
and forth
27
Example 4.4.4
Rigid Body Motion:
The free body diagram of figure 4.12 yields

x1 =
m1  k ( x2 − x1 ) and m2 
x2 =
− k ( x2 − x1 )
 m1 0   
x1   1 −1  x1   0 
⇒    +k    =
0
0 m2   
x2   −1 1  2  
x
Solve for the free response given:

m1 = 1 kg, m2 = 4 kg, k = 400 N subject to
 0.01 0
x0 =   m and v 0 = 0 
 0  0
28
Example 4.4.4 (cont.)
Following the steps of Window 4.4:
1 0 
1. M −1/ 2 =  1 
0 
 2
1 0  1 0 
 −1/ 2 −1/ 2    1 − 1    400 −200 
= =
  −1 1  0 1   −200 100 

2. K M KM 400 1
0
 2  2
 4 − λ −2  
3. det ( K − λ I=
 ) 100 det  = 100 ( λ 2
− 5λ=) 0
  −2 1 − λ  
⇒ λ1 =0 and λ2 =5 ⇒ ω1 =0, ω2 =2.236 rad/s
Indicates a rigid body motion
29
Now calculate the eigenvectors and note in particular that they
cannot be zero even if the eigenvalue is zero:
 4 − 0 −2   v11   0 
λ = 0 ⇒ 100     =   ⇒ 4 v11 − 2 v21 = 0
 −2 1 − 0   v21   0 
1   0.4472 
⇒ v1 =   or after normalizing v1 =  
 
2  0.8944 
 −0.8944   0.4472 −0.8944 
Likewise: v 2 =   ⇒P = 
 0.4472   0.8944 0.4472 
As a check note that
=PT P I= 
and PT KP diag [ 0 5]
30
5. Calculate the matrix of mode shapes:
1
−1/ 2 0  0.4472 −0.8944  0.4472 −0.8944 
= = P    =  0.4472 0.2236 
S M
 0 1/ 2  0.8944 0.4472  
−1  0.4472 1.7889 
⇒S =  −0.8944 0.8944 
 
7. Calculate the modal initial conditions:
−1  0.4472 1.7889   0.01  0.004472 
=
r (0) S= x0     = 
 −0.8944 0.8944  0   −0.008944 
−1
=
r (0) S= x 0 0
31
7. Now compute the solution in modal coordinates and note
what happens to the first mode:
Since ω1 = 0 the first modal equation is

r1 + (0)r1 = 0

⇒ r1 (t ) = a + bt Rigid body translation
And the second modal equation is

r2 (t ) + 5 r2 (t ) = 0

⇒ r2 (t ) = a 2 cos 5t Oscillation
32
Applying the modal initial conditions to these two solution
forms yields:
r1 (0)= a= 0.004472
r1 (0)= b= 0.0
⇒ r1 (t ) =
0.0042
as in the past problems the initial conditions for r2 yield
r2 (t ) = −0.0089 cos 5t
 0.0042 
⇒ r (t ) =
 
 −0.0089 cos 5t 
33
8. Transform the modal solution to the physical coordinate
system:
 0.4472 −0.8944   0.0045 
x(t ) = Sr (t ) =    
 0.4472 0.2236   −0.0089 cos 5t 
 x1 (t )   2.012 + 7.60 cos 5t  −3

⇒ x(t ) =   =   × 10 m
 x2 (t )   2.012 − 1.990 cos 5t 
Each mass is moved a constant distance and

then oscillates at a single frequency.
34
Order the frequencies
 It is convention to call the lowest frequency ω1 so that

ω1 < ω2 < ω3 < …
 Order the modes (or eigenvectors) accordingly
 It really does not make a difference in computing the time

response
 However:
 When we are measuring 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.)
35
CHOLESKY
DECOMPOSITION
36
Cholesky Decomposition
Alternate Path to Symmetric Single-Matrix Eigenproblem:
 Square root of matrix conceptually easy, but

computationally expensive
− 12 1 − 12 − 12 ~
M  + M
M q
2
KM q = q
 + Kq = 0
 More efficient to decompose M into product of upper and

lower triangular matrices (Cholesky decomposition)
37
Let M = U T U where U is upper triangular

Introduce the coordinate transformation
Ux = q ⇒ x = U −1q ⇒ U T U
x + K x
  =0
−1

q U q
premultiply by U −T to get
Iq  K q =
 + U −T K U −1q =+
q 0
−1 T −1 T T
=
note that: U K U 
−T
=
U  K U  U −T K U −1T −T
38
 Is this really faster? Let’s ask MATLAB

1 1
M= M M 2 2
M=U U T
»M = [9 0 ; 0 1]; »M = [9 0 ; 0 1];
»flops(0); sqrtm(M); flops »flops(0); chol(M); flops
ans = 65 ans = 5
 sqrtm requires a singular value decomposition (SVD),

whereas Cholesky requires only simple operations
1
Note that M = U for diagonal M
2
39
Reading Recommendation
Example: 4.9.2 (Inman’s book);
40
Exercises
Problems 4.8 through 4.10, 4.13 and 4.15 (Inman’s book).
Problems 4.39, 4.42, 4.46, 4.47 and 4.49 (Inman’s book).
41

EST56 - Tópico 5b

Uploaded by

Document Information

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

EST56 - Tópico 5b

Uploaded by

Copyright:

Available Formats

EST - 56

Prof. Airton Nabarrete

Tópico 5-b: Análise Modal

 Say we have physical coordinates x and want to

Start with Mx(t ) + Kx = 0 where x is a

 Now we have decoupled the EOM i.e., we have n

We must also transform the initial conditions

 This allows us to solve the two decoupled SDOF

 Then we can use the inverse transformation to obtain

where, n = 2 in the previous slides (Two DOF)

 Follow steps in Window 4.4 (pg 300)

» Minv2 = inv(sqrt(M)) »Kt =Minv2*K*Minv2

3) Calculate the symmetric eigenvalue problem for K-tilde

4) Calculate S = M^(-1/2) * P and Sinv = P^T * M^(1/2)

5) Calculate the modal initial conditions

6) Find the free response in modal coordinates

Use Tony's trick

7) Transform back to physical space

Process stays the same as section 4.3

See example 4.4.2 for details

 Note that for more then 2 DOF, a  0.2887 

 Note that the system in Fig 4.12 is not constrained and

The free body diagram of figure 4.12 yields

Solve for the free response given:

Since ω1 = 0 the first modal equation is

And the second modal equation is

 x1 (t )   2.012 + 7.60 cos 5t  −3

Each mass is moved a constant distance and

 It is convention to call the lowest frequency ω1 so that

 It really does not make a difference in computing the time

 Square root of matrix conceptually easy, but

 More efficient to decompose M into product of upper and

Let M = U T U where U is upper triangular

 Is this really faster? Let’s ask MATLAB

 sqrtm requires a singular value decomposition (SVD),

Example: 4.9.2 (Inman’s book);

Problems 4.8 through 4.10, 4.13 and 4.15 (Inman’s book).

Problems 4.39, 4.42, 4.46, 4.47 and 4.49 (Inman’s book).

You might also like

» Minv2 = inv(sqrt(M)) »Kt =Minv2KMinv2