Chapter 4 Multiple Degree of

Freedom Systems

Extending the first 3 chapters to

more then one degree of freedom
Two Degrees of Freedom (4.1)

x1 x2

m1 m2
k1
k2
 Free-Body Diagram

k2(x2 -x1)
k1 x1 m1 m2

x1 x2
 Equations of Motion
m1 x1 (t )  k1 x1 (t )  k 2 x2 (t )  x1 (t ) 
m2 x2 (t )  k 2 x2 (t )  x1 (t ) 
Rearrangin g :
m1 x1 (t )  ( k1  k 2 ) x1 (t )  k 2 x2 (t )  0
m2 x2 (t )  k 2 x1 (t )  k 2 x2 (t )  0
 Initial Conditions
• Two coupled, second -order, ordinary
differential equations with constant
coefficients
• Needs 4 constants of integration to
solve
• Thus 4 initial conditions on positions
and velocities
x1(0)  x10 , x1(0)  x10 , x2 (0)  x20 , x2 (0)  x20
Solution by Matrix Methods

 x1 (t )   x1 (t )   x1 (t ) 
x(t )    , x (t )    , x(t )   
 2 
x ( t ) 
 2 
x ( t ) 
x 
 2  ( t )

m1 0  k1  k 2  k2 
M   ,K  
 0 m2    k2 k 2 

Mx  Kx  0
 Initial Conditions

 x10   x10 
x(0)   , x (0)   
 20 
x 
 20 
x
 Solution:
jt
Let x(t)  ue
  unknown
j  1, u 0,
 - M  K ue
2 j t
0
 - M  K u  0
2
 Changes ode into algebraic
equation:

- M  K u  0 
2

two algebraic equation in 3 uknowns

u1 
u= , and 

u2 

 Condition for Solution:
inv - 2 M  K  exists  u  0

  - M  K  does not exist

1
Require u 0 2

or det- M  K   0
2

One equation in one unknown 

 Back to our specific system: the
characteristic equation
det- M  K   0 
2

 
 m1  k1  k2
2
k2 
det 0

 k2  m2  k2 
2

m1m2   (m1k2  m2 k1  m2 k2 )  k1 k2  0
4 2

Quadratic in 2 so four solutions

21 and 22or +1 and + 2
 Calculating the corresponding
vectors u1 and u2
A vector equation for each square frequency
(  M  K)u1  0
2
1

And:

(  M  K)u2  0
2
2

4 equations in the 4 unknowns (each

vector has 2 components, but...
Numerical example (4.1.5 & 4.1.6)

• m1=9 kg,m2=1kg, k1=24 N/m and k2=3 N/m

• Characteristic equation becomes
4-62+8=(2-2)(2-4)=0
2 =2 and 2 =4 or
1,3   2 rad/s, 2,4  2 rad/s
Each value of 2 yields an expression or u:
 Computing the vectors u
u11 
For  = 2, let u1 
2
then we have
1

u12 

(- M  K)u  0 
2
1

27  9(2) 3 u11  0 

 

 3 
3  (2) 
 12  
u 0 

9u11  3u12  0 and  3u11  u12  0
2 equations, 2 unknowns but DEPENDENT!
continued

u11 1 1
  u11  u12 from both equations :
u12 3 3
only the direction, not the magnitude can be determined !
This is because : det( 12 M  K )  0.
The magnitude is arbitrary. Suppose u1 satisfies
( 12 M  K )u1  0, so does au1 , a arbitrary :
( 12 M  K )au1  0  ( 12 M  K )u1  0
Likewise for the second value of 2:
u21 
For  = 4, let u2 
2
then we have
2

u22 

(-12 M  K)u  0 
27  9(4) 3 u21  0 
 

 3 
3  (4)u22 
 0 

1
9u21  3u22  0 or u21   u22
3
Note that the other equation is the same
What to do about the magnitude!
Several possibilities, here we just fix one element:

Choose:  3  1
u12  1  u1 

1
 

Choose:
 3 
1
u22  1 u2 
 1
  
 Thus the solution to the
algebraic matrix equation is:
 3 
1
1,3   2, u1 

1 

1
 
3
2,4  2, u2 

1 

 Return now to the time response:
We have four solutions:
x(t)  u1e j1t ,u1e j1t ,u2 e  j2 t ,u2 e j2 t 
Since linear we can combine as:
x(t)  au1e  j1t  bu1e j1t  cu 2e  j2 t  du2 e j2 t
 x(t)  ae  j1t  be j1t u1  ce j2 t  de j2t u2
 A1 sin( 1t  1 )u1  A2 sin( 2t  2 )u2

where A1 , A2 , 1 , and 2 are constants of integration

determined by initial conditions
Physical interpretation of all that
math!
• Each of the TWO masses is oscillating
at TWO natural frequencies 1 and 2
• The relative magnitude of each sine
term, and hence of the magnitude of
oscillation of m1 and m2 is determined
by the value of A1u1 and A2u2
• The vectors u1 and u2 are called
mode shapes
 What is a mode shape?
• First note that A1,A2, 1 and 2 are
determined by the initial conditions
• Choose them so that A2 = 1 = 2 =0
• Then: x1 (t)  u11 
x(t)   A1 sin 1t
 
x2 (t)  
u12 
• Thus each mass oscillates at (one)
frequency 1 with magnitudes
proportional to u1 the1st mode shape
 Mode shapes:

1/3 1

Mode 1: m1 m2

-1/3 1
Mode 2: m2 m2
Example 4.1.7 (continue on)
1  0
consider x(0) =   mm, x (0)   
 0  0
 A1

 x1 (t )   sin 2t  1   A2
sin 2t  2 

 x (t )   3
 
3

 A1 sin 2t  1  A2 sin 2t  2  
 2 

 x1 (t )  
  A1
 
2 cos 2t  1 
A2 
2 cos2t  2 
 x (t )   3
 
3

 A1 2 cos 2t  1  A2 2 cos2t  2  
 2 
At t=0 we have

 
1 mm 
 
A1
sin 1  
A2
sin 2 
 0  3 3
   A1 sin1   A2 sin2  
 
2 sin1   2 sin2 
A1 A2
0 

0  3 3

 A1 2 sin1   2 A2 sin2  
 
4 equations in 4 unknowns:
3  A1 sin 1   A2 sin 2 
0  A1 sin 1   A2 sin 2 
0  A1 2 cos1   A2 2 cos2 
0  A1 2 cos1   A2 2 cos2 
Yields:

A1  1.5 mm, A2  1.5 mm, 1  2  rad
2
Solution:
x1 (t)  0.5cos 2t  0.5cos 2t
x2 (t)  1.5cos 2t 1.5cos 2t
4

mm 2

x1 t

x2 t 0 5 10 15 20

t seconds
 Solution as a sum of modes

x(t)  u1 cos 1t  u2 cos 2 t

Determines how the second

frequency contributes to the
Determines how the first response
frequency contributes to the
response
 Things to note
• Two degrees of freedom implies two
natural frequencies
• Each mass oscillates at with these two
frequencies present in the response
• Frequencies are not those of two
component systems
k1 k2
1  2   1.63, 2  2   1.732
m1 m2
4.2 Eigenvalues and Eigenvectors
• Can connect the vibration problem with
the algebraic eigenvalue problem
• This will give us some powerful
computational skills
• And some powerful theory
• All the codes have eigensolvers so
these painful calculations can be
automated
 Change of coordinates
For a symmetric, positive definite matrix M:

m1 0  1  1 m1 0  1 / 2  1 m1 0 
M   ,M   1 
,M  1

 0 m2   0 m2 
  0 m2 

Let x(t )  M 1 / 2q(t ) and multiply by M 1 / 2
1 / 2 1 / 2 1 / 2 1 / 2
M
 MM  
q ( t )  M
 KM  q(t )  0
I identity ~
K symmetric
~ ~
q(t )  Kq(t )  0 where K  M 1 / 2 KM 1 / 2
The real symmetric eigenvalue problem

jt ~
Assume q(t )  ve (t )  Kq(t )  0
in q
jt ~ jt
  ve  Kve
2
 0, v  0 or
~ ~
K
 v  v 
2
K
 v v v  0
vibrationproblem real symmetric
eigenvalueproblem
Important Properties of the Real
Symmetric Eigenvalue Problem
• There are n eigenvalues and they are all
real valued
• There are n eigenvectors and they are all
real valued
• The set of eigenvectors are orthogonal
• The set of eigenvectors are linearly
independent
• The matrix is similar to a diagonal matrix
 Square Matrix Review
• Let aik be the ikth element of A then A is
symmetric if aik = aki denoted AT=A
• A is positive definite if xTAx > 0 for all
nonzero x (also implies each i > 0)
• The stiffness matrix is usually
symmetric and positive semi definite
(could have zero eigenvalue)
• The mass matrix is positive definite and
symmetric.
 Normal and orthogonal vectors

x1  y1  n
x  
,y   , inner product is x y   xi yi
 T

x  y  i 1
 n   n 
x orthogonal to y if x T y  0 
if set is both : orthonormal
x is normal if x x  1
T

The norm of x is x T x
 Example 4.2.2-4.2.4
~ 1 / 2 1 / 2  1 3 0  27  3  1 3 0
KM KM     3 3   0 1
 0 1   
~  3  1
K 
  1 3 
~ 3 -  - 1 
det(K  I )  det      6  8  0 
2

 -1 3 -  
which has roots : 1  2  12 and 2  4  22
 ~
( K  1 I ) v1  0 
3  2  1   v11  0
  1 3  2   v    0 
   12   
1
v11  v12  0  v1    
1
v1   (1  1)  1   
2 1
2

1 1
v1  1
2 The first normalized eigenvector
Modes and Eigenvectors
u1  v1 and u2  v 2
1/ 2 1/ 2
xM q u  M v
1 3 0 1 1 3 
u1   

 0 1 

1
  1 

 1 1  T 1
v2  , v1 v 2  (1 1)  0
2 
1
 2
1
v v1  (1  1)  1
T
1
2
1
v 2 v 2  (1  (1)(1))  1
T

2
 v i are orthonormal
Alternate method to normalize a
vector
Fix the missing element in x then compute:

x
T
x x
is normalized
 Orthogonal Matrix
P  v1 v2  called a matrix of eigenvectors

 v T
v v T
1 v2
 1 0
P P T
T 1 1
T   I
 v 2 v1 v 2 v 2  0 1 Called orthogonal
T ~

T ~ ~

P KP  P Kv1 Kv 2  P T 1v1 2 v 2 
1v1T v1 2 v1T v 2  1 0 
 T 
    diag ( 2
, 2)
2

1v 2 v1 2 v 2 v 2   0 2 
T 1

P also called a modal matrix

 Example 4.2.4
~ 1 1 1   3  1 1 1 1 
P KP 
T
1  1  1 3  1  1
2   2 
1 1 1  2 4 
 
2 1  1 2  4
1 4 0 2 0
     
2  0 8  0 4 
 Matlab commands
• To compute the inverse of the square
matrix A: inv(A) or use A\eye(n)
where n is the size of the matrix
• [P,D]=eig(A) computes the
eigenvalues and normalized
eigenvectors (watch the order). Stores
them in the eigenvector matrix P and
the diagonal matrix D (D=)
 More commands
• To compute the matrix square root use
sqrtm(A)
• To compute the norm: norm(x)
• To compute the determinant det(A)
• To enter a matrix:
K=[27 -3;-3 3]; M=[9 0;0 1];
• To multiply: K*inv(sqrtm(M))
 4.3 - Modal Analysis
• Physical coordinates are not always the
easiest to work in
• Eigenvectors provide a convenient
transformation to modal coordinates
– Linear combination of physical coordinates
– Say we have physical coordinates x and want to
transform to some other coordinates u
u1  x1  3x2 u1  1 3 x1 
     

u2  x1  3x2 u2  1 3x 2 
 Review of Eigenproblem
Start with M x  K x  0 where x is a
vector and M and K are matrices. Assume
initial conditions x 0 and x 0 . Rewrite as
1 1
M M
2 x  K x  0 and let
2

q
1
 12
M x  q  x  M q (coord. trans. #1)
2
 Eigenproblem (cont)
 12
Premultipl y by M to get
 12 1
 12  12 ~
MM q  
M KM q  q  Kq  0
2
  
I ~ ~T
KK

• Now we have a symmetric, real matrix

• Guaranteed real eigenvalues and distinct,
mutually orthogonal eigenvectors
 Eigenproblem (cont)
• Seek eigenvectors by various methods
~
K pi  i pi for the i th mode, or for the system
1 0
 | | |  
2

K P   P     p1
~
p2 
 pn  
 


 | | |  
0 n 
 Eigenvectors = Mode Shapes?
Mode shapes are solutions to M u = Ku
2

in physical coordinates. Eigenvetors are

characteristics of matrices. The two are
separated by a simple transformation,
but they are not synonymous.
 Eigenvectors vs. Mode Shapes
The eigenvectors of the symmetric PD matrix K˜
T
are orthonormal, i.e., P P  I. Are the mode shapes
 12
orthonormal? Using the transformation x  M q,
 12 1
the modes shapes U  M P  P  M U. Now, 2

1 1
P P  U M M U  UT MU  I. Thus, the mode
T T 2 2

shapes are orthogonal only w.r.t. the mass matrix.

 12  12
Similarly, U KU  P M KM P  
T T

K˜
 Eigenproblem (cont)
If P orthonormal (unitary), P T P  I  PT  P 1
T ~
Thus, P K P    diagonal matrix of eigenvalues.
~
Back to q  Kq  0. Make the additional coordinate
transforma tion q = Pr and premultipl y by P T
T ~
T
P P r  P K Pr  I r   r 0

• Now we have decoupled the EOM i.e., we

have n independent 2nd-order systems in
modal coordinates r(t)
Decoupled EOM

1
x  M Pr
2
•
Initial Conditions
Must transform the initial conditions to modal
coordinates
• Easy since we left a clear path

rP qP M x S x
T T
  1
2 1

 12
Thus, S = M P and
1 1
r0  S x0 and r0  S x0
• Free Response
Calculate for each of the n equations
independently

ri 0
ri (t )  sin i t  ri 0 cos i t
i
where
th

ri 0 and ri 0 are the i positions in r 0 and r 0
and i = i is the i natural frequency
th
 Response in Physical
Coordinates
• With n DOFs and m time values:
x(t)  S r(t)
n m n n n m
where
1

S M P 2

n n nn n n
 Example 4.3.1
9 0  27  3 1  0
M   ,K    ,x (0)   ,x (0)   
 0 1  3 3  0 0

• Follow steps in Window 4.4 (pg 183)

1 1 1
1) Calculate M 2 2) Calculate K˜  M 2
KM 2

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

Minv2 = K_tilde =
0.3333 0 3 -1
0 1.0000 -1 3
 Example 4.3.1 (cont)
% 3) Calculate the symmetric eigenvalue problem for K_tilde
[P,D] = eig(K_tilde);
[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
 Example 4.3.1 (cont)
% 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;
r_dot0 = Sinv * v0;
 Example 4.3.1 (cont)
% 6) Find the free response in modal coordinates
t_max = 10;
num_t = 1000;
t = linspace(0,t_max,num_t);
[T,W]=meshgrid(t,lambda.^(1/2));

% Use
R0 = r0(:,ones(num_t,1));
R_DOT0 = r_dot0(:,ones(num_t,1));

r = R_DOT0./W.*sin(W.*T) + R0.*cos(W.*T);

% 7) Transform back to physical space

x = S*r;
 Example 4.3.1 (cont)
% 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')
 Modal and Physical Responses
Independent
oscillators

Note IC

1  2  1  2  T1  2  4.44sec, 2  4  1  2  T2   sec
 Mode Summation Approach
• Based on the idea that any possible time
response is just a linear combination of the
eigenvectors
~
Starting with q  K q  0
n n
let q(t ) = 
i 1
q i (t )  
i 1

 ai e

j i t
 bi e
j i t v
 i

 two linearly independent solutions for each term.
n
can also write this as q(t )   d sin t   v
i 1
i i i i
 Mode Summation Approach (cont)
Find the constants di and i from the I.C.
n n
q(0)   di sin i v i and qÝ(0)   di  i cos i v i
i 1 i 1

Assuming eigenvectors normalized such that v vi  ij

T
j
n n
v Tj q(0) = v Tj  di sin  i v i =  di sin i v Tj v i  d j sin  j
i 1 i 1

Similarly for the initial velocities, v Tj qÝ(0) = d j  j cos  j

 Mode Summation Approach (cont)
Solve for d i and i from the two IC equations
v Ti q(0)  v T
1 i i q(0)
di  and i  tan
sin i v Tj q (0)
IMPORTANT NOTE about q(0) = 0
if you just crank it through the above expressions
you might conclude that d i  0, i.e., the trivial soln.
Mode Summation Approach (cont)

n
Instead, return to q(0)   d sin  v
i 1
i i i and

realize that i  0 instead. Get d i from the

expression T
vi q (0) = i d i cosi
 Mode Summation Approach (cont)
• Watch out for rigid-body modes!!

if 1 = 0,
q1 (t )   a1e  j 0 t  b1e j 0 t v i  a1  b1 v i
 
does not give two linearly independent solutions.
Now we must use the expansion
n
q(t )  a1  b1t v1  
i 2

 ai e

j i t
 bi e
j i t v
 i

and adjust calculatio n of the constantsfrom the
initial conditions accordingly.
 Example 4.3.1 by MSA
 3 0 1 1   1  1
From before, we have M   
2
and V =  1 1 
 0 1  2  
1  3 1 0
Appropriate IC are q0 = M x0 =  , q0 = M v0 =  
2 2

0 0
 T
1 i i q(0)
v  T
1 i i q(0)
v 1  - 2 
i  tan  tan     -  
v Ti q (0) 0 2   2 
 d1  3
v Ti q(0) 2 
di    
2
sin i d 2  3 2 
2
 Example 4.3.1 by MSA (cont)
q1 (t ) 3 2    1 1 3 2    1 1
  sin  2t     sin 2t    
q2 (t) 2  2  2 1 2  2  2 1 
free response of DOF 1
1

0.5 modal
MSA

-0.5

-1
0 1 2 3 4 5 6 7 8 9 10
time (sec)

free response of DOF 2

3

-1
modal
-2 MSA
-3
0 1 2 3 4 5 6 7 8 9 10
time (sec)
 Alternate Path to Symmetric
Single-Matrix Eigenproblem
• Square root of matrix conceptually easy,
but computationally difficult
 12 1
 12  12 ~
M M q  M
2
K M q  q  Kq  0
• More efficient to decompose M into
product of upper and lower triangular
matrices (Cholesky decomposition)
 Cholesky Decomposition
Let M  U U where U is upper tria ngular
T

Introduce the coordinate transformation

Ux  q  x  U 1q  U T U x  K x  0
 
q U 1q
premultipl y by U T to get
T 1 ~
I q  U KU q  q  K q  0
 


note : U T
KU   U  K U 
1 T 1 T T T T
 U T KU 1
 Cholesky (cont)
• Is this really faster? Let’s ask MATLAB
T
MU U
1 1
MM M 2 2

»M = [9 0 ; 0 1]; »M = [9 0 ; 0 1];
»flops(0); sqrtm(M); flops »flops(0); chol(M); flops
ans = 65 ans = 5
1
Note that M = U for diagonal M
2

• sqrtm requires a singular value decom-

position (SVD), whereas Cholesky requires
only simple operations
 Large Systems (103->? DOF)
• Full eigenproblem very expensive
– Plenty of ways to get selected eigenpairs
• Also common to discretize by FE in space
and by finite differences in time

M d  C d  K d  F  FEM EOM

M an 1  C vn 1  K d n 1  Fn 1
t 2
d n 1  d n  tvn  1  2 an  2 an 1 
2
vn 1  vn  t1   an   an 1 
 Power Method
• Simple iterative way to extract the dominant
(largest ) eigenpair
• Additional EV via matrix deflation
1
recast Ku   Mu as
2 1
2
u  K Mu
and iterate on Au   u
where A  K M and   1
1
 2
 Power Method (cont)
Guess u0  1 1 1 T

up  Aup1 for p  1,2,

up 1, j
 1  lim for j  any component
p  u
p, j

stop when u p  u p1

Ref: Meirovitch; Principles and Techniques of Vibrations
 Viscous Damping in MDOF Systems

• Two basic choices for including damping

– Attribute some amount to each mode based on
experience, i.e., guess
– Estimate damping due to viscoelasticity using
some approximation method (MSE)
– Model the damping mechanism directly (hard!)
 Modal Damping
• Decouple system based on M and K, i.e.,
use the “undamped” modes
• Attribute some zi to each mode of the
decoupled system

ri  2z ii ri  i2 ri  0

 ri (t )  Ai e z ii t
sindit  i 
 Modal Damping (cont)
• Can also use modal summation approach
• Again, modes are from undamped system
n
q(t)   di e z i i t sin  it  i vi
i1

where
v Ti q(0)  T
di i q(0)
v
di  and  i  tan 1
sin i v i qÝ(0) + z i i v i q(0)
T T
 System-Level Damping
• Can we decouple the system with the same
coordinate transformations as before?

Mx  Cx  Kx  0
1 1
Ir  P
 T
MCM 2
 P r  r  0
2 T 

diagonaliz able?
• In general, these can not be decoupled since K and
can not be diagonalized simultaneously
 Proportional Damping
• It turns out that CM-1K = symmetric is a necessary
and sufficient condition for C to be diagonalizable
by the eigenvectors of the “undamped” system,
i.e., those based on M,K
• best known example is “proportional”damping

C   M   K  linear combination of M and K.

CM K   M   K M K   K   KM K
1 1 1

both symmetric
 Proportional Damping (cont)
Write the systemas Mx   M   K x  Kx  0
1
2
  ~ ~
q  M x  Iq   I   K q  Kq  0
q  Vr  Ir   I    r  r  0
 
diagonal!
Thus, the damping ratios in the decoupled systemare
  i
2z ii     i2  zi  
i 2
Forced Response
 Forced Response (cont)
 F1 (t ) 
 F (t ) 
 2 
Mx  Cx  Kx  F(t )   
  
 Fn (t ) 
Assume C diagonalizable for now, i.e.,
~ ~ 1 ~ 1 1
q  Cq  Kq  M F where C  M CM 2
2 2
 Forced Response (cont) ~
Decouple the systemwith the eigenvalues of K
   

Ir   2z ii  
r   r  P T
M
1
2
F
  i 
   
so the i equation would be ri  2z ii ri  i ri  f i (t )
th
  2

• Responses to harmonic, periodic, or general forces

as in Chapters 2 and 3
• Note that the modal forcing function is a linear
combination of many physical forces
 Forced Response (cont)
T 1
R(t)  P M 2 F(t) for the decoupled EOM.

• An excitation on a single physical DOF may

“spread” to all modal DOFs
• Is it possible to drive a MDOF system at one of
its natural frequencies and not experience
resonant response? YES!. Watch.

Let F(t)  b g(t), where b is some spatial vector

and g(t) is any fuction of time. What if b happens
to be related to the i th
mode shape by b  Mui ?
 Forced Response (cont)
The force in the decoupled EOM s becomes
1 1
R ( t )  PT M 2 M u i g ( t )  PT M 2
u i g ( t )  PT p i g ( t )
1
rememberin g the transformation q  M x. 2

But through orthogonality, PT pi  e i , where

e i  0 0  0 1 0  0T
 
1 in the i th position only
Since only the i th mode is excited, g (t ) can contain
any or all of the other natural frequencie s.
•
General Damping Matrix
What do we do if C is not diagonalizable?
Integrate EOM directly? (Newmark )
• Let’s push ahead with SDOF technique

xÝ CxÝ K x  0  Ý
MÝ xÝ M 1 CxÝ M1 Kx  0
1
(why bother with q = M 2 x since C not diagonalizable)
Let x  ue t  2  M1 C  M 1K ue t  0
• Both  and 2 present implies not std EV problem
 First-Order Form

x  M 1C x  M 1K x can be rewritten as

 x  d  x   0 I x x
    1 1     A 
 x dt  x   M K  M C   x   x 
 y1   x 
Define the state vector y       such that
 y2   x 
y  Ay is our first - order system
 First-Order Form (cont) t
Seek a solution of the form y  z e
t t
Aze   ze  Az   z,
the standard eigenproblem, only
now matrix A is rarely symmetric

• Doubled system size to 2n

• Likely to get complex eigenpairs
 First-Order
 ui  Form (cont)
Az  z  zi   where ui  i mode shape
th

i ui 
 0 I  ui   ui 
Azi   1 1    i  
 M K  M C i ui   i ui 
yields  i ui  i ui
 M 1 K ui  M 1C i ui  2i ui
 M u  Ci ui  K ui  0
2
i i
 First-Order Form (cont)
• The “fundamental theorem of algebra”
guarantees that the eigenvalues and
eigenvectors of A will either be real or occur in
complex conjugate pairs
• Refer to section 9.5 to see the following

 i,i 1  z i i  j  i 1  z i2  i  j  i
from this the natural frequency and damping can be found to be
i
 i,i 1   i  i and z i,i 1 
2 2
where  i  Re(  i ),  i  Im( i )
 i  i
2 2