Handout 8

Modal Analysis of MDOF Systems with

Proportional Damping
The governing equations of motion for a n-DOF linear
mechanical system with viscous damping are:

+ D U
+K U =F
MU
(t )
(t )

(1)

and U
are the vectors of generalized displacement,
where U, U,
velocity and acceleration, respectively; and F( t ) is the vector of
generalized (external forces) acting on the system.
M, D,K represent the matrices of inertia, viscous damping and
stiffness coefficients, respectively1.
The solution of Eq. (1) is uniquely determined once initial
conditions are specified. That is,

at t 0 U (0) U o , U
(0)
o

(2)

Consider the case in which the damping matrix D is of the form

D M K

(3)

where , are constants2, usually empirical. This type of damping

is known as PROPORTIONAL, i.e., proportional to either the
mass M of the system, or the stiffness K of the system, or both.

The matrices are square with n-rows = n columns, while the vectors are nrows.
2
These constants have physical units, is given in [1/s] and in [s]
1

MEEN 617 HD 8 Modal Analysis with Proportional Damping. L. San Andrs 2013

Proportional damping is rather unique, since only one or two

parameters (at most), and , appear to fully describe the
complexity of damping, irrespective of the system number of
DOFs, n. This is clearly not realistic. Hence, proportional
damping is not a rule but rather the exception.
Nonetheless the approximation of proportional damping is
useful since, most times, damping is quite an elusive phenomenon,
i.e., difficult to model (predict) and hard to measure but for a few
DOFs.
Next, consider one already has found the natural frequencies
and natural modes (eigenvectors) for the UNDAMPED case, i.e.
+ K U = 0 ,
given M U

,
i

( i ) i 1,2...n satisfying

M i2 + K ( i ) = 0, i 1,...n .

with properties T M M ;

T K K

(4)
(5)

As in the undamped modal analysis, consider the modal

transformation

U(t ) q(t )

(6)

q ; U
q
( t ) , then EOM (1) becomes:
And with U
(t )
(t )
(t )

+ Dq K q = F( t )
Mq

(7)
which offers no advantage in the analysis. However, premultiply
T
the equation above by to obtain
T
T
T
T

q
+

D
q

q
=

F( t )

(8)

And using the modal properties, Eq. (5), and

MEEN 617 HD 8 Modal Analysis with Proportional Damping. L. San Andrs 2013

T D T M K T M T K

T D M K D

(9)

i.e., [D] is a diagonal matrix known as proportional modal

damping. Then Eq. (7) becomes

M q + D q K q = Q T F(t )

(10)

Thus, the equations of motion are uncoupled in modal space,

since [M], [D], and [K] are diagonal matrices. Eq. (10) is just a set
of n-uncoupled ODEs. That is,

M 1 q1 D1 q1 K1 q1 Q1
M 2 q2 D2 q2 K 2 q2 Q2

(11)

.....
M n qn Dn qn K n qn Qn
Or

M j qj D j q j K j q j Q j

Where n j

Kj

Mj

j 1,2... n

(12)

and D j M j K j . Modal damping

ratios are also easily defined as

Dj
2 K jM j

M j Kj
2 K jM j

; j=1,2,.n

(13)

For damping proportional to mass only, 0 , and

MEEN 617 HD 8 Modal Analysis with Proportional Damping. L. San Andrs 2013

M j
2 K jM j

2 n

(13a)
j

i.e., the j-modal damping ratio decreases as the natural frequency

increases.
For damping proportional to stiffness only, 0 ,
(structural damping) and

Kj

2 K jM j

(13b)

i.e., the j-modal damping ratio increases as the natural frequency

increases. In other words, higher modes are increasingly more
damped than lower modes.
The response for each modal coordinate satisfying the modal Eqn.
M j qj D j q j K j q j Q j , j 1,2...n is obtained in the same way
as for a single DOF system (See Handout 2).

First, find initial values in modal space qo j , qo j . These follow

from either
or

q o 1U o ; q o 1 U
o
q o M T M U o ,

(14)

q o M M U
o
1 T
1 T

qok
( k ) M U o , qok
( k ) M U
o
Mk
Mk
1

(15a)

(15b)

k=1,.n
MEEN 617 HD 8 Modal Analysis with Proportional Damping. L. San Andrs 2013

Free response in modal coordinates

Without modal forces, Q=0, the modal EOM is

M j qH j D j q H j K j qH j 0 Q j
with solution, for an

qH j e
where

j d j t

elastic underdamped mode j 1

C cos t S sin t
j

dj

C j qo j ; S j

if n j 0 (17a)

dj

d n 1 j2 , n K
qo jn qo
j

(16)

Mj

and

(17b)

See Handout (2a) for formulas for responses corresponding to

overdamped and critically damped SDOF systems.

Forced response in modal coordinates

For step-loads, QS j , the modal equations are
M j qj D j q j K j q j QS j
with solution, for an

qj e

j d j t

elastic underdamped mode j 1

C cos t S sin t q
j

(18)

dj

dj

Sj

n 0 (19a)

MEEN 617 HD 8 Modal Analysis with Proportional Damping. L. San Andrs 2013

where

qS j

d n 1 j2 , n
j

QS j
Kj

Kj

Mj

and

qo j jn j C j

; C j qo j qS j ; S j

(19b)

See Handout (2a) for formulas for physical responses

corresponding to overdamped and critically damped SDOF
systems.

For periodic-loads,

Consider the case of force excitation with frequency n j and

acting for very long times. The EOMs in physical space are

+ D U
K U = F cos t
MU
P

The modal equations are

M j qj D j q j K j q j QPj cos(t )
with solutions
for an elastic

(20)

mode, n 0
j

q j qtransient qss ( t )
e

j n j t

C cos t S sin t
j

dj

dj

(21)

Cc j cos t Cs j sin t
The steady state or periodic response is of importance, since the
transient response will disappear because of the dissipative effects
of damping. Hence, the j-mode response is:

MEEN 617 HD 8 Modal Analysis with Proportional Damping. L. San Andrs 2013

 QPj
qPS j
K
j

A j cos t j

(22)

Let f j be a jth-mode excitation frequency ratio. Then, define

n

Aj

1 f 2
2 2
j

fj

and

tan j

2 j fj

1 f
2
j

(23)

Recall that j is a phase angle and Aj is an amplitude ratio for

the jth-mode.
Note that depending on the magnitude of the excitation
frequency , the frequency ratio for a particular mode, say k,
determines the regime of operation, i.e., below, above or around
the natural frequency.
Using the mode displacement method, the response in physical
coordinates is

QPj

U j
A j cos t j

Kj
j 1

(24)

and recall that K j n j M j ( j ) K ( j ) and QPj ( j ) FP .

2

A mode acceleration method can also be easily developed to

give (* read addendum)
m

2 j

j 1

U U SP cos t

j q PS j

2
j 1 j

qPS j

MEEN 617 HD 8 Modal Analysis with Proportional Damping. L. San Andrs 2013

(25)

where U SP K Fp . Note that the mode acceleration method

cannot be applied if there are any rigid body modes (K is singular)

Frequency response functions for damped

MDOF systems.
The steady state or periodic modal response for the j-mode is:

QPj
qPS j
K
j

A j cos t j

(22)

Or, taking the real part of the following complex number

expression

QPj
qPS j
K
j
where
with i

Hj

i t
H j e

(26)

1
1 f j2 i 2 j f j

(27)

1 is the imaginary unit, and where

fj

is the j th

mode excitation frequency ratio. Then, recall from Eqs. (23)

Aj H j

1 f 2
2 2
j

fj

and j arg H j

(28)

Using the modal transformation, the periodic response UP in

physical coordinates is

MEEN 617 HD 8 Modal Analysis with Proportional Damping. L. San Andrs 2013

 QPj

A j cos t j
UP j
K

j 1
j

(24)

or take the real part of the equation below

Tj FP

it
Hje
U P q j q j j

Kj
j 1
j 1

(29)
n

j Tj j FP ei t

Kj
j 1

n

Now, the product j j matrix ( n n) . That is, define the

elements of the complex frequency response matrix H as
T

j p Tjq
H p ,q

Kj

1 f j2 i 2 j f j

(30)

p,q =1,2. n. The response in physical coordinates thus becomes:

U P = H FP ei t

(31)

or in component form,

n
U Pj H j ,r FPr
r 1

it
e ;

j 1,2..n

(32)

The components of the frequency response matrix H are

determined numerically or experimentally. In any case, the
components of H depend on the excitation frequency ().

MEEN 617 HD 8 Modal Analysis with Proportional Damping. L. San Andrs 2013

Determining the elements of H seems laborious and (perhaps) its

physical meaning remains elusive.

Direct Method to Find Frequency Responses in

MDOF Systems
Nowadays, with fast computing power at our fingertips, the
young engineer prefers to pursue a more direct approach, one
known as brute force or direct approach. Recall that the
equation of motion is

+ D U
K U = F cos t
MU
P
Or

+ D U
K U = Re F ei t
MU
P

(33)

i t

Assume a periodic solution of the form U = VP e

(34)
where VP is a vector in the complex domain. Substitution of Eq.
(34) into Eq. (33) gives

K i D 2 M VP FP

(35)

Define at each excitation frequency the complex impedance

(dynamic stiffness) matrix as:

K D K i D 2 M

(36)

And find the vector of physical responses (amplitude and phase) as

1

VP K D FP

MEEN 617 HD 8 Modal Analysis with Proportional Damping. L. San Andrs 2013

(37)

10

Since VP VPreal i VPimaginary , the physical response for each DOF

follows as:

U r VPr cos t r ;

r 1,2...n

VPr VP2r real VP2r imaginary ; tan r

VPr imaginary

VPr real

(38)

The direct method requires calculating the inverse of the

dynamic stiffness matrix at each excitation frequency. The
computational effort to perform this task could be excessive but for
systems with a few DOFs (n small).

MEEN 617 HD 8 Modal Analysis with Proportional Damping. L. San Andrs 2013

11

ORIGIN 1

PropDamping_STEPMDOF.mcd

FORCED RESPONSE of MDOF Linear system with

proportional damping
Original by Dr. Luis San Andres for MEEN 617 class /SP 2012, FA2001
The equations of motion are:

M d2Xdt2 + C dX/dt+ K X = F(t)

(1)

where M,K,C are nxn matrices of inertia, stiffness, and damping force coefficients,
and X, V=dX/dt, d2X/dt2,and are the nx1 vectors of displacements, velocity and
accelerations, respectively.
F(t) is a vector of nodal forces. At t=0, Xo,Vo=dX/dt are known.
For proportional damping, C = a M + b K, so the undamped mode shapes are still
valid. a & b are physical constants usually determined from measurements.
=================================================================

1. Define elements of inertia, damping & stiffness matrices:

1 0 0
2
M 10 0 2 0
0 0 1

2 1 0
7
K 10 1 2 1
0 1 1

n 3 # of DOF
example

a 0
b .001
C a M b K

2. Calculate the undamped naural frequencies and natural mode

shapes from the fundamental relationship:

2 M K = 0

(2)

If M is invertable, then define A M

and write Eq. (2) as:
sort ( eigenvals ( A) )

= A

Let:

(3)

(4)

(5)

<---- find eigenvalues

j 1 n
n
j

fn
j

j
n

The undamped
natural frequencies
are:

n ( 120.57 374.57 495.14 )

rad/s

fn ( 19.19 59.61 78.8 ) Hz

and undamped natural modes: j eigenvec A j

j
j

3. Plot the natural mode shapes:

This is the
matrix of undamped
modal vectors

1
k 1 0.5
0

k2

0.33 0.53 0.88

0.61 0.32
0.4
0.72 0.79 0.27

k3
0.5
1

mode 1
mode 2
mode 3
Modal matrices

4. Modal transformation of physical equations to (natural) modal coordinates

Using the transformation X = q
2

Mm

d q
2

C m

dt
where:

, the equations of motion (1) become uncoupled

in the modal space (principal coordinates):

dq
K m q = Q
dt

(6)

Mm M

Modal matrices

Km K

(7)

Cm C

Cm = a Mm b Km
T

and modal forces: Q = F

(8)

and with the initial conditions:

qo = Mm

M Xo

dqo
dt

= Mm

M Vo

(9)

k 1 n

Define the damping ratios and damped natural freqs. in modal space:
k

Cm
2 Mm

kk

kk

(10a)

2
d n 1 k
k
k

( 0.06 0.19 0.25 )

note all damping ratios < 1

.5

( 10b)

UNDERDAMPED CASE

The modal responses for arbitrary excitations are easily obtained for each natural mode (based on
response of simple 1DOF system).
And, the response in the physical coordinates is given by the superposition of the modal
responses, i.e.
X ( t) = q ( t)
( 11)
===============================================================================================
Modal matrices

Response to a STEP Load

Set STEP load vector:

2000
F 3000
1000

and set vectors of initial conditions

modal force is:

0
Xo 0
0

Q F

460.81

Q 675.13

3
3.21 10

0
Vo 0
0

Calculate response

calculate modal initial displs. & vels.

qo Mm

M Xo

q_doto Mm

M Vo

Evaluate response in modal coordinates:

j 1 n
qs
j

Qj
Km

: static

j j

Ac qo qs
j
j
j

displacement in modal space

As

q_doto j n Ac

p 1 Npoints

tp ( p 1) t
j n tp
j

q j p e
Calculate response

Ac cos d tp As sin d tp
j

qs

see your cheat-sheet

(1 DOF response)

Calculate response

Response in modal space

0

3 10

6 10

0
qo 0
0

0
q_doto 0
0

0.052

0.1

0.16 0.21
time(secs)

0.26

0.31

q1
q2
q3

The response in physical coordinates is: X ( t) = q ( t)

Let:

m 3
2 10

Response in physical coordinates

4 10

0
Xo 0
0

0.05

0.1

0.15 0.2
time(secs)

0.25

0.3

0.35

Check steady-state (t infinite) response:

points

0
Vo 0
0

Steady state response (long times)

X1
X2
X3

Xs N

Static response as t-> inf.

Recall I.C's

( 11)

<------ select number of modes for physical response

2 10

2.3 10 4
qs 4.37 10 5

1.13 10 4

7.86 10 6
1.85 10 4

8.29 10 5

4
Xs 2 10

4
1 10

4
Xs 2 10

4
1 10

ORIGIN 1

PropDamping_STEPMDOF.mcd

PERIODIC RESPONSE of MDOF system with proportional

damping
Original by Dr. Luis San Andres for MEEN 617 class /SP 2012
The equations of motion are:

M d2Xdt2 + C dX/dt+ K X = Fo cos(t)

(1)

where M,K,C are nxn matrices of inertia, stiffness, and damping force coefficients,
and X, V=dX/dt, d2X/dt2,and are the nx1 vectors of displacements, velocity and
accelerations, respectively.
F(t) is a vector of nodal forces - periodic. At t=0, Xo,Vo=dX/dt are known.
For proportional damping, C = a M + b K, so the undamped mode shapes are still
valid. a & b are physical constants usually determined from measurements.
1. Define elements of inertia, damping & stiffness matrices:

1 0 0
2
M 10 0 2 0
0 0 1

2 1 0
7
K 10 1 2 1
0 1 1

n 3 # of DOF
example

a 0
b .00025
C a M b K

find natural freqs.

0.33 0.53 0.88

0.61 0.32
0.4
0.72 0.79 0.27

1
k 1 0.5
k2

0
0

fn ( 19.19 59.61 78.8 )

k3

Hz

0.5
1

mode 1
mode 2
mode 3
Modal matrices

damped natural freqs. and

damping ratios

fd ( 19.19 59.55 78.65 )

Hz

( 0.02 0.05 0.06 )

Response to a PERIODIC Load

Set amplitude load vector:

2000
Fo 4000
6000

modal force is:

Qo Fo

Assume effect of initial conditions has vanished since periodic load

acts for very long time.

2.51 103
Qo 4.92 103

3
4.97 10

Calculate response

Vary the excitation frequency to determine the amplitude and phase of the FRF for each mode:
set maximum frequency (rad/sec) to display
max 4 n
calculations
n

(larger than highest natural freq)

n 495.14

k 1 log max 100

rad/s

k
excitation Frequency

k 10

100

freq

i 1 n

k
2

max 1.98 10
in Hz

Evaluate response in modal coordinates:

ri k

qs
i

k
n

frequency ratio

Qo
Km

ii

The MODAL complex amplitudes are:

and MODAL Phase Angles:

i k

See your cheat-sheet

(1 DOF response)

qs 1.26 10

qi k

1.75 10

qs

1 ri k j 2 i ri k
2

2 i ri k 180

if n k 0
2
i
1 ri k
2 i ri k 180

otherwise
180 atan
2
1 ri k

atan

Calculate response

3.19 10

Magnitude of Modal Amplitude (in m)

fd ( 19.19 59.55 78.65 )

Natural Mode Response Amplitude

0.1

0.01
1 10

1 10

1 10

10

100
Forcing Frequency (Hz)

1 10

q1
q2
q3
Natural Mode Response Phase angle

Phase Angle (in Degrees)

100

200

10

100
Forcing Frequency (Hz)

q1
q2
q3

1 10

( 0.02 0.05 0.06 )

Hz

The response in physical coordinates is: X ( t) = q ( t)

m 3

Magnitude of Modal Amplitude (in m)

Let:

<------ select number of modes for physical response

Physical Response Amplitude

0.1

( 0.02 0.05 0.06 )

0.01
1 10

1 10

fd ( 19.19 59.55 78.65 )

1 10

10

100
Forcing Frequency (Hz)

1 10

1 10

X1
X2
X3
Dynamic response in physical coordinates
200
Physical Phase lags (degs)

100

100

200

10

100
frequency (Hz)

X1
X2
X3

Hz

DIRECT METHOD

make

Magnitude of Modal Amplitude (in m)

and solve

KD K M k i k C
2

for each

1 Fo

Zk KD
k

Amplitude Physical Response - Direct

0.1

0.01
T

1 10

fd ( 19.19 59.55 78.65 )

1 10

1 10

10

100
Forcing Frequency (Hz)

X1
X2
X3

1 10

( 0.02 0.05 0.06 )

Hz

Derivation of Mode Acceleration Method for

MDOF systems (proportional damping or light
damping)
(Luis San Andrs, Lecturer. Based on homework delivered by Mr. Rahul Kar)

Problem Statement
Determine the system response of a MDOF system with proportional damping using the
Mode Acceleration method.
Solution
The differential equation governing the motion of a n-DOF linear system is:
[ M ] X + [C ] X + [ K ] X = P(t )

(1)

where [M], [K], [C] are the (nxn) matrices of (constant) mass, stiffness and damping
coefficients. P(t) is a vector of n-external forces, time dependent, and X(t) is the vector of
system displacements (physical responses). The physical damping is of proportional type,
i.e. [C] = a [M] + b [K]
The system described by (1) has a set of natural frequencies (i)i=i,..n and associated
modal (eigen) vectors (i )i=i,..n. Each pair (i i) satisfies the fundamental relationship
[K] i =i2 [M] i,

(2)

i=1,2,n

The physical response X(t) or solution to (1) can be found using modal analysis, i.e.
n

X (t ) = [ ] (t ) = i i

(3)

i =1

where [ ]= 1 2 ..... n is the modal matrix. Each of the components of the modal
response vector (t) is obtained from solution of the (uncoupled) equations:
M m i i + C m i i + K m i i = Qi

i=1,2,n

(4)

where Q = [ ] P , and (Km, Mm, Cm )i are the i-th modal stiffness, mass and damping
coefficients obtained from:
T

[M m ]= [ ]T [M ][ ]; [K m ]= [ ]T [K ][ ]; [C m ]= [ ]T [C ][ ];

(5)

In (3), using a number of modes m less than the n-DOF is known as the mode
displacement method, i.e.
m

X (t ) i i ;
i =1

m<n

(6)

The mode acceleration method aims to find exactly the system static response
should P be a vector of constant generalized forces. In this case, the mode displacement
method does poorly when just a few modes, m<<n , are used

To derive the appropriate equations, pre-multiply (1) by [K]-1 , i.e. the flexibility matrix
(obviously this operation precludes any rigid body motion), to obtain:
[ K ]1[ M ] X + [ K ]1[C ] X + X = [ K ]1 P(t )
and

X = [ K ]1 P(t ) [ K ]1[ M ] X [ K ]1[C ] X

m

i =1

i =1

(7)

from (6) it follows that X ii (t ) and X ii (t ) . Replacing these relationships
into (7) gives:
m

i =1

i =1

1
1
1
X (t ) [K ] P(t ) [K ] [ M ] i i (t ) [K ] [C ] i i (t ) ;

(8)

m<n

Lets work with the terms: [K ] [ M ] i and [K ] [C ] i . Since each pair (i , i) satisfies
the fundamental relationship
1

then

and similarly,

[K] i =i2 [M] i

(2)

[K]-1 [M] i = (1/i2) i

(9.a)

[K]-1 [C] i = (2i /i) i

(9.b)

where i is the i-th modal damping ratio defined as

i =

Cm i
C cri

; [C m ] = [ ] [C ][ ], C cri = 2 K mi M mi
T

1/ 2

(10)

Note that in the equation above, (Km, Mm)i are the i-th modal stiffness and mass
coefficients satisfying

K mi

Mm
i

1/ 2

=i

Replacing (9) into (8) gives the physical response of the system as:

X (t ) [K ]

m
2 i i

P(t ) 2 i (t )
i (t ) ;

i =1
i =1
i
i
m

m<n

(10)

which is known as the mode acceleration response method. The first term in the
response [ K ]1 P(t ) corresponds to a pseudostatic static displacement due to P(t).
Note that for P = Ps (constant), X=Xs=[K]-1Ps since all i=0. This simple check certifies
the accuracy of the mode acceleration method even when using few modes (m<n).

Reference:
MEEN 617 Handout #8 Modal Analysis of MDOF Systems with Proportional Damping, L.
SanAndrs, 2008.