4.

1 Undamped free vibration of MDOF systems

Subject CSE40418: [ M ]{ x}  [ K ]{ x}  0

Advanced Structural Analysis with initial conditions:
x (t ) t  0  x (0)

Phase I: x (t ) t  0  x (0)
Structural Dynamics (4) The general solution is given by a superposition of the response
in individual modes:
n
Prof. Y.Q. Ni (Prof. Ngai) {x (t )}   {a} j ( A j cos  j t  B j sin  j t )
Office: ZS930 (CEE) j 1

Phone: 2766 6004 where Aj and Bj are 2n constants to be determined.

E-mail: [email protected] n
{x (t )}   {a} j  j ( A j sin  j t  B j cos  j t )
j 1
The Hong Kong
Polytechnic University The Hong Kong
Polytechnic University

Effective stiffness of columns with different base connections:

By setting t = 0, we have
n
x (0)   {a} j A j
j 1
n
x (0)   {a} j  j B j
j 1

A modal expansion of any displacement vector x(t) has the

form:
n
x (t )   {a}k qk (t )  [Φ ]{q}
k 1

where qk(t) are modal coordinates or normal coordinates

and q  q1 q2  qn T .

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

1
 n 4.2 Undamped forced vibration of MDOF systems
{a}Tj [ M ]{ x}   ({a}Tj [ M ]{a}k )qk (t )  ({a}Tj [ M ]{a} j ) q j (t )
k 1
[ M ]{x}  [ K ]{ x}  {F (t )}
so
{a}Tj [ M ]{ x} {a}Tj [ M ]{ x} Introducing the linear transformation of coordinates (modal
qj   expansion or modal decomposition)
{a}Tj [ M ]{a} j Mj n
x (t )   {a}k qk (t )  [Φ ]{q}
where Mj is called modal mass. yields k 1

Thus [ M ][Φ]{q}  [ K ][Φ]{q}  {F (t )}

{a}Tj [ M ] x (0) T
in1{a} j [ M ]{a}i Ai {}Tj [ M ][Φ ]{q}  {}Tj [ K ][Φ ]{q}  {}Tj {F (t )}
q j (0)    Aj
Mj Mj
Because
{a}Tj [ M ] x (0) in1{a} j [ M ]{a}i i Bi
T
{}Tj [ M ][Φ ]  M j {}Tj [ K ][Φ ]  K j
q j (0)     jBj
Mj Mj therefore

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

M j q j  K j q j  Pj (t ) j  1, 2,  , n
It is therefore obtained that:
q j (0) where
A j  q j (0) Bj  n
j Pj (t )   aij Fi (t ) a1 j F1 (t )  a2 j F2 (t )    anj Fn (t )
i 1
So the solution is: In the above, a set of n uncoupled or independent equations
n q j (0) of the motion in terms of the modal coordinates qj. They are
{ x (t )}   {a} j [q j (0) cos  j t  sin  j t ] readily solved.
j 1 j
or, n When the mode shapes are normalized to mass,
{ x (t )}   {a} j q j (t )
j 1 {}Tj [ M ][Φ ]  1 {}Tj [ K ][Φ ]  2j
where the equation becomes:
q j (0)
q j (t )  q j (0) cos  j t  sin  j t q j  2j q j  Pj (t ) j  1, 2,  , n
j
where
which is analogous to the free vibration response of n
Pj (t )   ij Fi (t ) 1 j F1 (t )   2 j F2 (t )     nj Fn (t )
SDOF systems. i 1

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

2
An n-storey shear frame subjected to ground excitation: Example 1:

Formulate the modal equations for the 2-storey shear

[ M ]{ x}  [ K ]{ x}  [ M ]{1}ys (t ) frame subjected to ground excitation and determine
the participation factors.
Introduce the modal expansion:
n
x (t )   {a}k qk (t )  [Φ ]{q}
k 1
where [] is the modal matrix obtained in the solution of
corresponding eigenproblem ([ K ]  2 [ M ]){}  0 .

[ M ][Φ ]{q}  [ K ][Φ ]{q}  [ M ]{1}ys (t )

{}Tj [ M ][Φ ]{q}  {}Tj [ K ][Φ ]{q}  {}Tj [ M ]{1}ys (t )

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

Using the orthogonality property of the eigenvectors results in Solution:

modal equations:
m1u1  ( k1  k2 )u1  k 2u2   m1 ys
q j  2j q j   j ys (t ) j  1, 2, , n
m2u2  k 2u1  k 2u2   m2 ys
where the modal participation factor j is given by
The natural frequencies of the system are solved to be 1
and 2. The corresponding eigenvectors of the system are
 in1 mi ij
j   n 2
{}1  {11 21}T and {}2  {12 22 }T .
 i 1 mi ij
Modal expansion (transformation):
and for mass-normalized mode shapes by
u1 (t )  11q1  12 q2
 j  in1 mi ij
u2 (t )  21q1  22 q2

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

3
 Solution:
m111  m221
q1  12 q1   2
ys (t ) m 0  136 0 
m111  m2221 [ m]   1  
 0 m2   0 66
so the participation factor:
m1  11  m 2  21 k  k  k 2   75000  44300
1   [k ]   1 2 
2
m1  11  m 2  221   k2 k 2   44300 44300 

 F (t )   0 
[ F (t )]   1    
m   m222  F2 (t )  F0 sin t 
q2  22 q2   1 12
2
ys (t )
m112  m2222
Based on the eigen-analysis,
so the participation factor:
m112  m222 [ k ]   2 [ m]  0
2   2
m112  m2222 we obtain 1 = 11.83 rad/sec and 2 = 32.89 rad/sec.

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

Example 2: With 1 = 11.83 rad/sec,

A 2-storey shear building is subjected to a harmonic force
F(t) = F0sint = 10,000sin20t acting on the second floor.
([k ]  12 [m]){ }  0
k1 = 30,700 lb/in and k2 = 44,300 lb/in. m1 = 136 lbsec2/in we obtain the 1st mode shape as
and m2 = 66 lbsec2/in. The natural frequencies are
obtained as 1 = 11.83 rad/sec and 2 = 32.89 rad/sec.  
Determine the steady-state response. {}1   11 
21 
With 2 = 32.89 rad/sec,

([ k ]   22 [m]){ }  0

we obtain the 2nd mode shape as

 
{}2   12 
22 

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

4
 Therefore the mode shape matrix is 22 F0 1
q2 (t )  sin t
   K 2 1  ( / 2 ) 2
[ ]  {}1 {}2    11 12 
21 22  The response is obtained by
The modal masses are: 2
x (t )   { } j q j (t )  [Φ]{q}
M 1  { }1T [ m]{ }1 M 2  { }T2 [ m]{ }2 j 1

Therefore the 1st floor response:

The modal stiffnesses are:
2   F 1   F 1 
K1  { }1T [k ]{ }1 K 2  { }T2 [k ]{ }2 x1 (t )   1 j q j (t )   11 21 0 2
 12 22 0  sin t
j 1  K1 1  ( / 1 ) K 2 1  ( / 2 ) 2 
The modal forces are:
and the 2nd floor response:
P1 (t )  {}1T {F (t )}   21 F0 sin t   F 
2 1   F 1
x2 (t )   2 j q j (t )   21 21 0 2
 22 22 0 2
sin t
P2 (t )  {}T2 {F (t )}  22 F0 sin t j 1  K 1 1  ( / 1 ) K 2 1  ( /  )
2 

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

The response is therefore expressed in modal expansion: 4.3 Damped forced vibration of MDOF systems
2
x (t )   {} j q j (t )  [Φ]{q} [ M ]{ x}  [C ]{ x }  [ K ]{ x}  {F (t )}
j 1
Introducing the linear transformation of coordinates (modal
and the modal equations are: expansion or modal decomposition)
n
M j q j (t )  K j q j (t )  Pj (t ) j  1, 2 x (t )   {a}k qk (t )  [Φ ]{q}
yields k 1
or
M 1q1 (t )  K1q1 (t )  21 F0 sin t [ M ][Φ ]{q}  [C ][Φ ]{q}  [ K ][Φ ]{q}  {F (t )}
M 2 q2 (t )  K 2 q2 (t )  22 F0 sin t
{}Tj [ M ][Φ ]{q}  {}Tj [C ][Φ ]{q}  {}Tj [ K ][Φ ]{q}  {}Tj {F (t )}
Based on undamped SDOF harmonic vibration theory, the It is noted that the orthogonality property of the modes
steady-state responses are:
21F0 1 {}Tj [ M ]{}i  0 
q1 (t )  sin t  for i  j
K1 1  ( / 1 )2 {}Tj [ K ]{}i 0 

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

5
causes all the components except the jth mode in the mass and Conditions for damping uncoupling:
stiffness terms to vanish. A similar reduction is assumed to apply
to the damping term, that is, if it is assumed that When the damping matrix is of the form:

{}Tj [C ]{}i  0 for i  j [C ]  [ M ]  [ K ]

then the coefficient of the damping term will reduce to where  and  are arbitrary constants (proportionality factors),
the orthogonality condition will be satisfied.
{}Tj [C ]{} j
Thus Damping matrix that meets the above equation is called
proportional (classical) damping or Rayleigh damping.
M j q j  C j q j  K j q j  Pj (t ) j  1, 2, , n
or, When the above equation is satisfied, we have

q j  2 j  j q j  2j q j  Pj (t ) / M j j  1, 2, , n C j  {}Tj [C ]{} j  {}Tj [ M ]{} j  {}Tj [ K ]{} j

where
 modal mass
 M j  K j  M j   2j M j  (   2j ) M j
M j  {}Tj [ M ]{} j

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

C j  {}Tj [C ]{} j  2 j  j M j  modal damping C j  2 j  j M j  (  2j ) M j

K j  {}Tj [ K ]{} j  2j M j  modal stiffness So constants  and  can be determined based on the measured
modal damping ratios 1 and 2 of the first two modes:
Pj (t )  {}Tj {F (t )}  modal loading
1 
 j  (   j )
If the modal shapes are mass-normalized, 2 j
{}Tj [ M ]{} j  1 or
that means, 212 (12   21 )
M j 1 
22  12
Thus, in the case of normalization, we have
q j  2 j  j q j  2j q j  Pj (t ) 2( 22  11 )

22  12
which is a set of n uncoupled differential equations.

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

6
Example 3:
0.0414 0 0 
[m]   0 0.0388 0  kip-sec2/in
The three-storey shear frame building shown below is  
subject to the harmonic forces given by  0 0 0.02588

 F1 cos Ω1t   299.286  209.78 0 

 
{F (t )}   F2 cos Ω 2t  [k ]   209.78 258.969  49.184 kips/in
 
 F cos Ω t   0  49.184 49.184 
 3 3 

An eigen-analysis yields the following frequencies and mode

where F1 = 8 kips, F2 = 10 kips, F3 = 18 kips, 1 = 10 rad/sec, shapes:
2 = 20 rad/sec, 3 = 50 rad/sec. Determine the steady-
state response {x(t)}.  25.076   1.0 1.0 1.0 
 
{}   53.578  rad/sec [ ]  1.303 0.860  1.000
 
110.907  1.947  1.685 0.183 
 

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

The mode shapes for the structure are shown below.

Solution:

The mass and stiffness matrices for the structure are

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

7
The modal mass matrix is determined as

0.2054 0 0  for j = 3,
[ M ]  [ ] [m][ ]   0
T
0.1436 0 
  q3 (t )  0.00808 cos 10t  0.01036 cos 20t  0.00414 cos 50t
 0 0 0.08117 
The modal stiffness matrix is determined as
The response {x(t)} in physical coordinates is determined
129.166 0 0  as
[ K ]  [ ] [k ][ ]   0
T
412.216 0 
  3
 0 0 998.42 { x (t )}   {} j {q j (t )}
j 1
The modal force vector is determined as
 0.1018 cos10t  0.2911cos 20t  0.6570 cos 50t 
{P(t )}  [ ]T {F (t )}  
 0.1052 cos10t  0.3925 cos 20t  0.3755 cos 50t 
8 cos10t  13.03 cos 20t  35.046 cos 50t  0.1110 cos10t  0.4970 cos 20t  0.7835 cos 50t 
   
  8 cos10t  8.6 cos 20t  30.33 cos 50t 
 8 cos10t  10.01 cos 20t  3.294 cos 50t 
 

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

Example 4:
The expressions for steady-state response, qj(t), in modal
(normal) coordinates are determined as (from the solution For the three-storey shear frame building shown below, the
of SDOF systems): modal damping ratios for the first two modes of vibration
have been determined to be 1 = 0.05 and 2 = 0.06.
 Pj    Calculate 3 on the assumption of Rayleigh damping. The
1
q j (t )    cos t
 K j  1  ( /  ) 2  mass and stiffness matrices for the structure are
  j 
Thus for j = 1,

q1 (t )  0.07365 cos 10t  0.27723 cos 20t  0.09118 cos 50t

for j = 2,

q2 (t )  0.02011 cos10t  0.02424 cos 20t  0.56992 cos 50t

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

8
 Solution:
0.05823 0 0 
[m]   0 0.04658 0  kip-sec2/in
  The Rayleigh damping ratio for the jth mode is given by
 0 0 0.03494
1 
j (   j )
 137.782  111.883 0  2 j
kips/in
[k ]   111.883 286.70  174.817 
  For 1 = 13.02 rad/sec, we have
 0  174.817 174.817 
1 
1  (  13.02  )  0.05
The eigen-solution for the system yields the frequencies and 2 13.02
mode shapes as For 2 = 55.36 rad/sec, we have
 13.02   1.0 1.0 1.0 
  1 
{}   55.36  rad/sec, [ ]  1.143  0.364  4.12 2  (  55.36  )  0.06
  2 55.36
101.44 1.183  0.939 3.90 
 
The above two equations can be re-expressed as

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

The mode shapes are shown below.

  (13.02) 2   1.302
  (55.36) 2   6.644
Simultaneous solution of the above equations yields
  0.9893   0.001845
Therefore 3 is determined as

1  1 0.9893
3  (  3 )  (  0.001845  101.44)  0.0984
2 3 2 101.44
and the modal damping matrix is determined as

2 11M1 0 0  0.2187 0 0 

[C ]   0 2 22 M 2 0  0 0.6325 0 
   
 0 0 2 33M 3   0 0 27.531

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

9
 Example 5: The mass matrix and stiffness matrix of the structure are:

Consider a two-storey shear building frame as shown m 0  2 0

[M ]   1   m0  0 1 
below. W1 = 2W0 and W2 = W0 where W0 = 5,000 kN.  0 m 2  
k1 = 2k0 and k2 = k0 where k0 = 4.0105 kN/m. The
vibration modes of the structure are obtained as {1} = k  k 2  k 2   3  1
[K ]   1  k0 
{1 2}T and {2} = {1 1}T, respectively. The damping   k2 k 2  
 1 1 
ratio is 0.02 for the first mode and 0.01 for the second
mode. Calculate the steady-state dynamic response of The eigenvalue equation is
the structure at the two floors when a harmonic [K   2 M ]x  0
excitation p(t) = p0sint = 2.5104sin0.51t (kN) is
horizontally acted on the second floor, where 1 is the 3k 0  2m0 2  k0 
 x  0
natural circular frequency of the first mode of the   k 0 k 0  m0 2 
structure.
For a nontrivial solution, equating to zero the determinant
of the coefficient matrix gives

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

p (t) m2
x2 3k 0  2m0 2  k0
0
 k0 k 0  m 0 2
k2

5 k 0 2 k 02
m1
x1 4    2 0
2 m0 m0
k1 The solutions of the above equation are
k0 2k 0
12   22 
2m0 m0
So the first and second natural frequencies of the structure are
Solution:
k0 2 k0 2 k0 g 2 4.0  10 5  9.8
The masses of the first and second floors are: 1      19.799 rad/sec
2m0 2 m0 2 W0 2 5000
W1 2W0 W2 W0
m1    2m0 m2    m0 2k 0 k k g 4.0  10 5  9.8
g g g g 2   2 0  2 0  2  39.598 rad/sec
m0 m0 W0 5000

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

10
The governing equation of the damped forced vibration is The steady-state solution of q1 is given by

q1s (t )  q 01 sin( t  1 )
[ M ]{ x}  [C ]{ x }  [ K ]{ x}  {F (t )}
where
where p0 1
q 01 
 0   0  3m0 (12   2 ) 2  ( 2 11 ) 2
{F (t )}    
 p (t )   0  t
p sin p0 g 1

3W012 [1  ( / 1 ) 2 ]2  [2 1 ( / 1 )]2
By introducing the modal expansion
2.5  10 4  9.8 1
  0.0555 m
 x (t )  2 3  5000  19.799 2
{ x (t )}   1      j q j (t ) (1  0.5 2 ) 2  (2  0.02  0.5) 2
 x2 (t ) j 1
2 11 2 1 / 1
we can obtain the following uncoupled modal equation 1  tan 1  tan 1
12   2 1  ( / 1 ) 2
q j  2 j  j q j   2j q j  Pj (t ) / M j j  1, 2 2  0.02  0.5
 tan 1  1.53o
1  0. 5 2

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

where
For the second mode,
M j  {}Tj [ M ]{} j
 2 0  1 
Pj (t )  {}Tj {F (t )} M 2  {}T2 [ M ]{}2  1  1m0      3m0
0 1  1
For the first mode,
 0 
P2 (t )  { }T2 {F (t )}  1  1    p0 sin  t
2 0 1   0 sin  
M 1  {}1T [ M ]{}1  1 2m0 
p t
    6m0
0 1 2 so

 0  p0
P1 (t )  {}1T {F (t )}  1 2 q2  2 2 2 q 2   22 q 2  P2 (t ) / M 2   sin  t
  2 p0 sin  t 3m0
 0  t
p sin
so where 2 = 21 = 39.598 rad/sec, 2 = 0.01, and /2 =
p /(21) = 0.25.
q1  2 11q1  12 q1  P1 (t ) / M 1  0 sin  t
3m0
The steady-state solution of q2 is given by
where 1 = 19.799 rad/sec, 1 = 0.02, and /1 = 0.5.

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

11
 q 2s (t )  q02 sin( t   2 ) *1.2 Free vibration (F(t) = 0)
where
m x(t )  c x (t )  k x (t )  0
p0 1
q02  
3m0 ( 22   2 ) 2  ( 2 2 2 ) 2 1.2.1 Undamped free vibration (c = 0)
p0 g 1 The SDOF system without damping (c=0) specializes to

3W0 22 [1  ( /  2 ) 2 ]2  [2 2 ( /  2 )]2
m x(t )  k x (t )  0
2.5  10 4  9.8 1
 2
 0.0111 m
3  5000  39.598 (1  0.25 2 ) 2  (2  0.01  0.25) 2 Free vibration is initiated by disturbing the system from its static
equilibrium position by imparting the mass a displacement y0
2 22 2 2 / 2 and a velocity v0 at time zero, i.e.,
2  tan 1  tan 1
22   2 1  ( / 2 ) 2 x(t ) t  0  x(0)  y0
2  0.01  0.25
 tan 1  0.31o
1  0.252 x (t ) t 0  x (0)  v0

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

Consequently the response expression is obtained as

v0
So the steady-state dynamic response of the structure is x (t )  y 0 cos t  sin t

 x s (t ) 1 1 where k
{ x s (t )}   1s    1 q1s (t )   2 q2s (t )   q1s (t )   q2s (t ) 
2  1 m
 x2 (t )  
The response also can be expressed as
Thus
x(t )  C sin(t   )  C cos(t   )
x1s (t )  1  q1s (t )  1  q2s (t )
where the response amplitude C is
 0.0555 sin(9.9t  1.53o )  0.0111sin(9.9t  0.31o ) (m) 2
v 
C  A2  B 2  y02   0 
x2s (t )  2  q1s (t )  1  q2s (t ) 
and the phase angles  and  satisfy
 0.1110 sin(9.9t  1.53o )  0.0111sin(9.9t  0.31o ) (m)
y0 v0 / 
tan   tan  
v0 /  y0
The Hong Kong The Hong Kong
Polytechnic University Polytechnic University

12
 *2.1 Forced vibration (F(t)  0)
1.2.2 Viscously damped free vibration (c  0)

The SDOF system with damping (c0) in free vibration is

m x(t )  c x (t )  k x (t )  F (t )

m x(t )  cx (t )  k x (t )  0 2.1.1 Undamped harmonic vibration (c = 0)

with initial displacement y0 and velocity v0 at time zero, The SDOF system without damping (c=0) specializes to

x(t ) t 0  x(0)  y0 x (t ) t 0  x (0)  v0 m x(t )  k x (t )  F0 sin  t

where F0 is the excitation amplitude and  is the exciting
Dividing the equation of motion by m gives frequency. The initial conditions are

x(t )  2x (t )  2 x(t )  0 x(t ) t  0  x(0)  y0

where  k m as defined before, and x (t ) t 0  x (0)  v0

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

After determining the constants A and B by imposing the initial

The system oscillates about its equilibrium position with a
conditions, we have
progressively decreasing amplitude.

v0  y0 v F  
x(t )  [ y0 cos Dt  sin Dt )] e  t x(t )  y0 cos t   0  0 sin t
D  k 1  (  ) 2 

 
 
transient
where D   1   2 .
F 1
 0 sin  t
Alternatively, this expression can be written as k 1  (  ) 2
 
steady state
x(t )  Ce  t cos( D t  )
where where the steady state response is

(v0  y0) 2 (v0  y0) F0 1 F0

C y02  tan   x s (t )  2
sin  t  sin  t
2D D y0 k 1  (  ) k  m 2

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

13
2.1.2 Damped harmonic vibration (c  0)

The damped SDOF system in harmonic vibration is

m x(t )  cx (t )  k x (t )  F0 sin  t

with initial conditions
x(t ) t 0  x(0)  y0 x (t ) t 0  x (0)  v0

The solution can be expressed as

x (t )  x c (t )  x p (t )
The complementary solution xc(t) is

xc (t )  e  t ( A cos  D t  B sin  D t )

The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

The steady-state response can also be expressed as

x p (t )  u0 sin(  t  )  (u st ) 0 Rd sin(  t  )
where
F0 F0
u0  
( k  m  2 ) 2  (c  ) 2 k [1  (  / ) 2 ]2  [2 (  / )]2

F0
(ust )0 
k
u0 1
Rd  
(u st ) 0 [1  (  / ) 2 ]2  [2 (  / )]2

c 2 (  / )
  tan 1  tan 1
k  m2 1  (  / ) 2

End of Lecture 4 The Hong Kong The Hong Kong

Polytechnic University Polytechnic University

14