You are on page 1of 33

CHAPTER 9

FREE VIBRATION OF MDF SYSTEMS

Expected outcome:
1. Students can determine modal natural frequencies and mode shapes.
2. Students can expand a displacement vector by modal expansion.
3. Students can find displacements of an MDF system in free vibration.

9.1 MDF system without damping

Let us consider the two-story shear building in Fig. 8.1 (or 9.1a), whose equation of
motion was formulated in Chapter 8, and assume that it is an undamped system for simplicity.
The equation of motion for this system in free vibration [no external force, p  t   0 ] is

 + ku = 0
mu (9.1)

The free vibration displacements of two floor masses are the solution of Eq. (9.1) with initial
conditions:

u  u  0 and u  u  0  (9.2)

If the building is let to vibrate with an arbitrary initial displacement position (Fig. 9.1b)
and initial velocity is zero, the displacement of each mass will be as shown in Fig. 9.1d. The
free vibration of each mass seems complicated (Fig. 9.1d). The motion is not simple harmonic
as the free vibration of an SDF system, and deflected shape of this building changes with time
as the ratio u1  t  / u2  t  varies with time t .

Suppose the initial displacement instead has an appropriate shape, which is a natural
mode shape of vibration of this two-story shear building as Fig. 9.2b. The undamped structure
would undergo simple harmonic motions without changing the deformed shape proportion,
where the ratio u1  t  / u2  t  remains equal to two at all time t . It maintains that mode shape

and both floors will vibrate in phase by reaching their maximum displacements at the same
time. These vibrations have the same period T1  2 1 and same angular frequency 1 of the

first mode vibration.

9-1
2 (c)
q2 q1

qn 0 t

-2
m c u2(0) a b c
u2 2
a
k
2m u2 0 t
u1 b
Rigid beams u1(0)
2k -2
2

(a) (b)
u1 0 t

-2
0 T11 21
2T 3T31
(d)
Figure 9.1 Free vibration of an undamped MDF system due to arbitrary initial displacement:
(a) two-story shear frame; (b) deflected shapes at time instants a, b, and c; (c)
modal coordinate qn(t); and (d) displacement histories

1 (c)

q1 0 t

-1
1 abc d e

u2 0 t

Rigid beams -1
1 T1  2 / 1

u1 0 t

-1
0 T11 21
2T 3T31
(d)
Figure 9.2 Free vibration of an undamped MDF system in its first natural mode of vibration:
(a) two-story shear frame; (b) deflected shapes at time instants a, b, c, d, and e; (c)
modal coordinate q1(t); and (d) displacement histories

9-2
Let us try to impose another initial displacement shape of the second mode shape as
shown in Fig 9.3b. The undamped shear building again undergoes simple harmonic motions
without changing the deformed shape proportion, where the ratio u1  t  / u2  t  equal to -1 at all

time t . In contrast to the first mode shape, where both floors move in the same direction, in
this second mode shape, the two floors move in the opposite directions. The location where
displacement remains zero is called “node.” These shapes are called “natural modes of
vibration.” The number of modes increases as number of degrees of freedom increases.

(c)
1

q2 0 t

-1
bd
1 ace

u2 0 t

Rigid beams -1
T2  2 / 2
1

u1 0 t

-1
0 2T
T121 4T
2 21
2T 6T321
3T
(d)
Figure 9.3 Free vibration of an undamped MDF system in its second natural mode of
vibration: (a) two-story shear frame; (b) deflected shapes at time instants a, b, c, d,
and e; (c) modal coordinate q2(t); and (d) displacement histories

For each mode, all floors move like a simple harmonic motion with the same frequency
called “natural circular frequency of vibration” ( n ), or “natural cyclic frequency of vibration”

( f n ) and the time to complete one cycle is called “natural period of vibration.”

2 1
Tn  fn  (9.3)
n Tn

9-3
The lowest natural frequency is known as the fundamental natural frequency, denoted
by 1 . It corresponds to the longest natural period, which is called fundamental natural period

( T1 ) of vibration. This fundamental natural period of building in unit of seconds is

approximately number of stories divided by 10.

9.2 Natural vibration frequencies and modes

Displacement vector u  t  , which is the solution of system of ordinary differential

equations (Eq. 9.1), is assumed to be composed of products of mode shape vector n , invariant

with time, and modal coordinate qn  t  , which is a scalar quantity varying with time t .

u  t   n q n  t  (9.4)

Modal coordinate qn  t  varies with time whereas the mode shape vector n does not. In free

vibration, the modal coordinate qn  t  oscillates like a simple harmonic wave with frequency

equal to the natural frequency. It is assumed to be of the form:

qn  t   An cos  n t  Bn sin  n t (9.5)

Substituting Eq. (9.5) into Eq. (9.4), we get

u  t   n  An cos  n t  Bn sin  n t  (9.6)

Next, we substitute Eq. (9.6) into the equation of motion (Eq. 9.1) and take derivative of u  t 
  t  . Then, we have a vector equation:
to replace u

 n2mn  kn  qn  t   0 (9.7)

This vector equation (Eq. 9.7) can be satisfied by two ways:

(1) scalar qn  t   0 but this does not lead to useful results; it is called trivial solution.

(2) vector n2mn  kn  0 or kn  n2m n

In the latter case, it is called a matrix eigenvalue problem. The matrices k and m are
known and we need to determine  n2 and n that satisfy this vector equation.

9-4
kn  n2mn  0 k   m
2
n n 0 (9.8)

Then, we factorize the left-hand side by taking common factor, vector n , out of parenthesis.

Now the left-hand side vector is a product of matrix k  n2m and vector n . This Eq. (9.8) is

considered as a linear system of equations, where k  n2m is called coefficient matrix and

vector n is the unknown of equations. When the right-hand side is a vector of zero, such a

linear system of equations is called homogeneous system, which has non-trivial solutions (non-
zero solutions) only when the determinant of coefficient matrix k  n2m is zero.

det  k  n2m   0 (9.9)

This determinant is a polynomial of degree N in terms of  n2 . This Eq. (9.9) is called

“characteristic equation” or “frequency equation.” This Nth-degree polynomial equation has


N real and positive roots, which mean  n2 that makes the polynomial equal to zero, because

matrices k and m are symmetric and positive definite.

The N roots for  n2 are known as eigenvalues, or characteristic values, and they

correspond to the natural frequencies of N modes of vibration.

For each value of  n2 corresponding to each mode, a mode shape vector n can be

solved from the Eq. (9.8), but such solutions are not unique. Many possible vectors obtained
by scaling a solution n by any scalar can also satisfy the equation (Eq. 9.8). This vector n

is called natural mode of vibration, or mode shape vector. It is also known as eigenvector
corresponding to the eigenvalue  n2 .

Modal and spectral matrices

We can put many mode-shape vectors together in a matrix to have them in compact
form called “modal matrix” Φ
 11 12 ... 1N 
 22 ... 2 N 
Φ   jn    21

      (9.10)
 
N 1 N 2 ... NN 

9-5
The N eigenvalues  n2 can be assembled into a diagonal matrix Ω2 , which is known as

“spectral matrix.”

12 
 
 22 
Ω 
2
(9.11)
  
 
  N2 

From the eigenvalue problem in vector form:

kn  n2m n (9.12)

It is possible to write them in matrix form:

kΦ  mΦΩ 2 (9.13)

Orthogonality of modes

The natural mode-shape vectors corresponding to different natural frequencies can be


shown to satisfy the following orthogonality conditions.
When n  r ,

nT kr  0 nT mr  0 (9.14)

This can be proved by considering Eq. (9.8):

kn  n2m n (9.15)

Multiply Eq. (9.15) from the left by rT

rT kn  n2rT mn (9.16)

The above Eq. (9.16) contains scalar quantity, so we can transpose both sides and obtain.

nT kr  n2nT mr (9.17)

Leave Eq. (9.17) and suppose we start from Eq. (9.8) in the rth mode,

9-6
kr  r2mr (9.18)

Multiplying Eq. (9.18) from the left by nT , we get

nT kr  r2nT mr (9.19)

Subtracting Eqs (9.17) and (9.19), we have

 2
n  r2  nT mr  0 (9.20)

Equation (9.20) means that if we are considering two different mode n  r , then n2  r2 and

we can divide n2  r2 from both sides of Eq. (9.20). We then have

nT mr  0

By similar manipulation, we can also prove that when we consider two different mode n  r ,

nT kr  0

The orthogonality condition implies that the matrices

K  ΦT kΦ and M  ΦT mΦ (9.21)

are diagonal where the diagonal elements are

K n  nT kn M n  nT mn (9.22)

These K n and M n are called “generalized stiffness” and “generalized mass,” respectively.

Because k and m are positive definite, K n and M n are always positive, and they are related

by

K n  n2 M n (9.23)

which can be shown by

K n  nT kn  nT n2mn   n2 nT mn   n2 M n (9.24)

Interpretation of modal orthogonality

Consider a structure vibrating in nth mode with displacements

9-7
u n  t   qn  t  n (9.25)

 n  t   qn  t  n and inertia forces are


The corresponding accelerations are u

 f I n  mu n  t   mn qn  t  (9.26)

Next consider displacements of the structures in its rth mode

u r  t   qr  t   r (9.27)

The work done by inertia force is

 f I n ur    nT mr  qn  t  qr  t 


T
(9.28)

which is zero because of modal orthogonality.

Another implication is that the work done by equivalent static forces associated with
displacement in nth mode through the rth mode displacement is zero.

 f S n  ku n  t   kn qn  t  (9.29)

The work done is

 fS n ur   nT kr  qn  t  qr  t 
T
(9.30)

which is zero because of modal orthogonality.

Normalization of modes

From theorem in linear algebra, the eigenvector multiplied by a scalar is also the
eigenvector corresponding to the same eigenvalue. Therefore, only the shape of mode,
proportion of displacements at all DOFs, is essential. A mode can be scaled without changing
the shape. This process is called “normalization.” Scaling factor to normalize mode-shape
vectors could be selected in many ways. Sometimes, the mode-shape vectors are scaled such
that the largest element equal to unity. For a building structure, mode-shape vectors are often
scaled such that value in the vector at top roof DOF is equal to 1. Another popular choice is to
scale or normalize the mode-shape vectors such that generalized mass M n of all modes are

equal to 1.

9-8
M n  nT mn  1 or M  ΦT mΦ  I (9.31)

where I is the identity matrix, or unit matrix (all entries on the main diagonal are equal to 1).
These mode-shape vectors are called a mass orthonormal set. This set of normalized mode-
shape vectors results in generalized stiffness equal to square of modal natural frequency for
each mode.

K n  nT kn  n2 M n  n2 or K  ΦT kΦ  Ω 2


(9.32)

Example 9.1
(a) Determine the natural vibration frequencies and modes of the system of Fig. E9.1a using
the first set of DOFs shown.
(b) Repeat part (a) using the second set of DOFs in Fig. E9.1b.
(c) Show that the natural frequencies and modes determined using the two sets of DOFs are
the same.

Solution

(a) The mass and stiffness matrices for the system with the first set of DOFs were determined
in Example 8.2:
m 2 1 1 0 
m kk
6 1 2  
0 2 

 k  mn2 / 3 mn2 / 6 
k  n2m    (a)
 mn / 6 2k  mn / 3
2 2

Eq. (a) is substituted in Eq. (9.9) to obtain the frequency equation:

m 2n4  12kmn2  24k 2  0

The solutions of this frequency equation are


12  6  2 3  mk  2.536 mk 
22  6  2 3  mk  9.464 mk (b)

The modal natural frequencies are 1  2.536k m and 2  9.464k m . Next, the natural

mode-shape vectors are determined by substituting n2  12 in Eq. (a), and then Eq. (9.8) gives

9-9
 0.155 0.423 11  0 
k     (c)
 0.423 1.155  21  0 

Now select any value for one unknown, say 11  1 . Then use the equation in either the first or

second row of this vector equation to obtain 21  0.366 . Next, substituting n2  22 in Eq.

(9.8), we get

 2.155 1.577  12  0 


k     (d)
 1.577 1.155  22  0 

Select 12  1 and use either rows of the vector equation. We get 22  1.366 . In summary,

the two mode-shape vectors are


1   1 
1    2    (e)
0.366  1.366 

The displacement shapes of these vectors are shown in Fig. E9.1c.

First mode
1  2.536k / m

Second mode
 2  9.464 k / m

Figure E9.1

9 - 10
(b) The mass and stiffness matrices of the system described by the second set of DOF were
derived in Example 8.3:

m 0   3k kL / 2 
m 2  k 2  (f)
 0 mL /12   kL / 2 3kL / 4 
Then
3k  mn2 kL / 2 
k  m  
2
 (g)
n
 kL / 2  9k  mn2  L2 /12

Eq. (g) is substituted in Eq. (9.9) to obtain

m 2n4  12kmn2  24k 2  0

This frequency equation is the same as obtained in part (a); obviously, it gives the same solution
ω1 and ω2 of Eq. (b).

To determine the nth mode-shape vector, we substitute Eq. (g) into Eq. (9.8) and obtain

kL 3k  mn2
 3k  m  
2
n tn 
2
 n  0 or  n  
kL / 2
tn (h)

Substituting for 12  2.536k / m and 22  9.464k / m in Eq. (h) gives

L L
 1  0.464t1  2  6.464t 2
2 2

If t1  1 , then  1  0.928L , and if t 2  1 , then  2  12.928L . In summary, the two

mode-shape vectors are


1  1 
1    2    (i)
0.928 / L  12.928 / L 

The displacement shapes of these vectors are shown in Fig. E9.1d and we can see that they are
having the same shapes as mode shapes in part (a).

(c) The same natural frequencies were obtained using the two sets of DOFs. The mode shapes
are given by Eqs. (e) and (i) for the two sets of DOFs. These two sets of results are plotted in
Fig. E9.1c and d and can be shown to be equivalent on a graphical basis. Alternatively, the

9 - 11
equivalence can be demonstrated by using the coordinate transformation from one set of DOFs
T
to the other. The displacements u  u1 u2 are related to the second set of DOFs,
T
u  ut u by

u1  1  L / 2  ut 
    or u = au (j)
u2  1 L / 2  u 

The displacements u in the first two modes are given by Eq. (i). Substituting the first mode in
T T
Eq. (j) leads to u  1.464 0.536 . Normalizing the vector yields u  1 0.366 , which is

identical to 1 of Eq. (e). Similarly, substituting the second mode from Eq. (i) in Eq. (j) gives
T
u  1 1.366 , which is identical to 2 of Eq. (e).

Example 9.2

Determine the natural frequencies and modes of vibration of the system defined in
Example 8.5 as shown in Fig. E9.2a. Show that the modes satisfy the orthogonality properties.

EI
1  3.15623
mL4

EI
2  16.2580
mL4

Figure E9.2

9 - 12
Solution

The stiffness and mass matrices were determined in Example 8.5 with reference to the
translational DOFs u1 and u2:

 mL / 4  48EI  2 5
m k
 mL / 2  7 L3  5 16 
 
Then
48 EI  2   5 
k  n2m  (a)
7 L  5 16  2 
3 

where
7 mL4 2
  (b)
192 EI

Substituting Eq. (a) in (9.9) gives the frequency equation

2 2  20  7  0

which has two solutions: λ1 = 0.36319 and λ2 = 9.6368. The natural frequencies corresponding

to the two values of λ are obtained from Eq. (b):

EI EI
1  3.15623 2  16.2580 (c)
mL4 mL4

The natural modes are determined from Eq. (9.8) following the procedure shown in Example
9.1 to obtain
 1   1 
1    2   
0.3274  1.5274 (d)

These natural modes are plotted in Fig. E9.2b and c. Next, we can show that the two mode
shapes are orthogonal by substituting Eq. (d) in Eq. (9.14) to obtain

mL 1   1 
1T m2  1 0.3274    0
4  2  1.5274 
48 EI  2 5  1 
1T k2  1 0.3274    0
 5 16  1.5274 
3
7L

This verifies that the two natural mode shapes in Eq. (d) are orthogonal.

9 - 13
Example 9.3

Determine the natural frequencies and modes of vibration of the inverted L-shape frame
defined in Example 8.6 as shown in Fig. E9.3a. Normalize the modes to have unit vertical
deflection at the free end.

Solution

The mass and stiffness matrices were determined in Example 8.6 with reference to DOFs u1
and u2:
3m  6 EI  8 3
m k
 m  7 L3  3 2 
 

The frequency equation follows Eq. (9.9), in which, after substituting m and k, evaluating the
determinant, and defining
7 mL3 2
 
6 EI (a)
can be written as

3 2  14  7  0

The two roots are λ1 = 0.5695 and λ2 = 4.0972. The natural frequencies corresponding to the

two values of λ are obtained from Eq. (a):

EI EI
1  0.6987 2  1.874 (b)
mL3 mL3

The natural modes are determined from Eq. (9.8) following the procedure used in Example 9.1
to obtain
 1   1 
1    2   
2.097  1.431 (c)

These mode shapes are shown in Fig. E9.3b and c.


In computing the natural mode shape vectors, value at the first DOF had been arbitrarily
selected as unity. These mode shapes might be normalized to have unit value at DOF u2 by
dividing 1 in Eq. (c) by 2.097 and 2 by −1.431. This normalization gives another

representation of mode shape vectors:

9 - 14
0.4769  0.6988
1    2    (d)
1  1 

EI EI
(a) (b) 1  0.6987 (c) 2  1.874
mL3 mL3

Figure E9.3

Example 9.4

Determine the natural frequencies and modes of the system defined in Example E8.1 as
shown in Fig. E9.4a, a two-story frame idealized as a shear building. Normalize the mode shape

vectors such that generalized mass Mn = 1 for all modes.

Solution

The mass and stiffness matrices of the system, determined in Example 9.1, are

 2m   3k k 
m k (a)
 m   k k 

where k  24 EI c / h3 . The frequency equation follows Eq. (9.9), which, after substituting for

m and k and evaluating the determinant, can be written as

 2m     5km 
2 4 2
 2k 2  0 (b)

The two roots are 12  k / 2m and 22  2k / 2 , and the two natural frequencies are

k 2k
1  2  (c)
2m m

9 - 15
Substituting k  24 EI c / h3 gives

EI c EI c
1  3.464 2  6.928 (d)
mh3 mh3

The natural modes are determined from Eq. (9.8) following the procedure used in Example 9.1
to obtain
1 2  1
1    2   
1  1 (e)

These natural modes are shown in Fig. E9.4b and c.

To normalize the first mode, M1 is calculated using Eq. (9.22), with 1 given by Eq.

(e):

 2  1 2  3
M 1  1T m1  m 1/ 2 1     m
 1  1  2

To make M1 = 1, divide 1 of Eq. (e) by 3m / 2 to obtain the normalized mode,

1 1 
1   
6m 2 

For this 1 it can be verified that M1 = 1. The second mode can be normalized similarly.

Rigid beams

First mode Second mode


1  k / 2 m 2  2k / m
(a) (b) (c)

Figure E9.4

9 - 16
Example 9.5

Determine the natural frequencies and mode shapes of the cantilevered beam defined
in Examples 8.7 and 8.9 as shown in Fig. E9.5a. The story height h is 10 ft.

Solution

With reference to the lateral displacements u1 and u2 of the two floors as the two DOFs, the
mass matrix and condensed stiffness matrix were determined in Example 8.9:

2  EI  54.885 17.508
mtt  m   kˆ tt  3 
 1 h  17.508 11.607  (a)

The frequency equation is


det kˆ tt   2m tt  0  (b)

Substituting for m tt and kˆ tt , evaluating the determinant, and obtaining the two roots just as in

Example 9.4 leads to

EI c EI c
1  2.198 2  5.850 (c)
mh3 mh3

It is of interest to compare these frequencies for a frame with flexible beams with those for the
frame with flexurally rigid beams determined in Example 9.4. It is clear that beam flexibility
has the effect of lowering stiffness and natural frequencies, consistent with intuition.
The natural modes are determined by solving

 kˆ tt 
 n2m tt n  0 (d)

with ω1 and ω2 substituted successively from Eq. (c) to obtain

0.3871 1.292 
1    2    (e)
 1   1 

These vectors define the lateral displacements of each floor. They are shown in Fig. E9.5b and
c together with the joint rotations. The joint rotations associated with the first mode are
determined by substituting ut  1 from Eq. (e) in Eq. (b) of Example 8.9:

9 - 17
 u3   0.4426 0.2459  0.4172 
u   0.4426 0.2459  0.3871  
 4 1     1 0.4172 
        (f)
u5  h  0.9836 0.7869  1.0000  h  0.4061
u6     0.4061
 0.9836 0.7869 

Similarly, the joint rotations associated with the second mode are obtained by substituting
ut  2 from Eq. (e) in Eq. (b) of Example 8.9:

u3   0.3258
u   
 4  1  0.3258
     (g)
u5  h  2.0573
u6   2.0573

First mode Second mode

1  2.198 EI / mh3 2  5.850 EI / mh3

(a) (b) (c)

Figure E9.5
Example 9.6

Figure E9.6a shows the plan view of a one-story building supported on three frames, A,
B, and C, with roof considered as a rigid diaphragm. The roof weight is uniformly distributed
and has a load intensity of 100 lb/ft2. The plan dimensions are b  30ft and d  20ft , the
eccentricity is e  15ft , and the height of the building is 12 ft. The lateral stiffnesses of the
frames are k y  75 kips/ft for frame A, and k x  40 kips/ft for each of frame B and C.

Determine the natural periods and modes of vibration of the structure.

Solution

Weight of roof slab is w  30  20 100  6000 lb  60 kips

9 - 18
Total mass is m  w / g  60 / 32.2  1.863 kips-sec2 /ft

Moment of inertia, or rotational mass, of the roof slab is

m  b2  d 2  1.863  302  202 


IO    201.825 kips-ft-sec2
12 12

Lateral motion of the roof diaphragm in the x-direction is governed by:

mux  2k x u x  0 (a)

Thus, the natural frequency of x-lateral vibration is

2k x 2  40 
x    6.553 rad / sec
m 1.863

The corresponding natural mode for translation in x-direction is shown in Fig. E9.6d. The
coupled lateral (uy)-torsional (uθ) motion of the roof diaphragm is governed by

m  uy   k y ek y  u y  m  1 
           u (t )
 I O  u   ek y k  u   I O  0  gy

The mass matrix is


1.863 
m
 201.863
The stiffness matrix has four elements:

k yy  k y  75 kips/ft
k y  k y  ek y  1.5  75  112.5 kips
d2
k  e 2 k y  kx  8168.75 kips-ft
2
Hence,
 75.00 112.50 
k 
112.50 8168.75

With k and m already determined, the eigenvalue problem for this two-DOF system is solved
by standard procedures to obtain:

Natural frequencies: ω1 = 5.878 and ω2 = 6.794 rad/sec

9 - 19
0.5228 0.5131 
Natural modes: 1    and 2   
 0.0493 0.0502 

These mode shapes are shown in Fig. E9.6b and c. The motion of the structure in each mode
consists of translation of the rigid diaphragm coupled with torsion about the vertical axis
through the center of mass.

Frame A

(a) Plan view of one-story frame

(b) First mode (c) Third mode (d) Second mode


1  5.878 rad/sec 2  6.794 rad/sec x  6.553 rad/sec

Figure E9.6

9 - 20
Example 9.7

Consider a special case of the system of Example 9.6 in which frame A is located at the
center of mass (i.e., e  0 ). Determine the natural frequencies and modes of this system.

Solution

For symmetric-plan system, the three equations of motion are:

d2
mux  2k x u x  0 muy  k y u y  0 I O u  k x u  0 (a)
2

The first equation of motion indicates that translational motion in the x-direction vibrates at the
natural frequency x . This motion is independent of lateral motion uy or torsional motion uθ

(Fig. E9.7c).

2k x 2  40 
x    6.553 rad/sec
m 1.863

The second equation of motion indicates that translational motion in the y-direction vibrates at
the natural frequency  y . This motion is independent of the lateral motion ux or torsional

motion uθ (Fig. E9.7b).


ky 75
y    6.344 rad/sec
m 1.863

The third equation of motion indicates that torsional motion oscillates at the natural frequency
 . The roof diaphragm would rotate about the vertical axis through its center of mass without
any translation of this point in either x- or y-direction (Fig. E9.7a).

 20  40  6.296 rad/sec
2
d 2kx
  
2IO 2  201.825 

Observe that the natural frequencies ω1 and ω2 of the unsymmetric-plan system


(Example 9.6) are different from and more separated than the natural frequencies ωy and ωθ of
the symmetric-plan system (Example 9.7).

9 - 21
(a) First mode (b) Second mode (c) Third mode
  6.296 rad/sec  y  6.344 rad/sec x  6.553rad/sec

Figure E9.7

9.3 Modal expansion of displacement vectors

Any set of N independent vectors can be used as a basis for representing any vectors of
N components. Vector u is written as a linear combination of N linearly independent mode-
shape vectors r where each vector is scaled by a scalar qr .

N
u   r qr  Φq (9.33)
r 1

where qr are scalar multipliers called “modal coordinates.” If we are given a displacement

vector u together with a set of N mode-shape vectors r , and we want to find modal expansion

of displacement vector u , we need to determine scalar qr for all modes. We then multiply

Eq. (9.33) by nT m from the left, and obtain

N
nT mu   nT mr  qr (9.34)
r 1

Because of orthogonality of mode shapes, only the term r  n in the summation is non-zero.

nT mu  nT mn  qn (9.35)

Therefore,
nT mu nT mu
qn   (9.36)
nT mn Mn

9 - 22
This modal expansion is employed to determine initial modal coordinates qn  0  and qn  0 

when initial condition u  0  and u  0  are given, so that the solution displacement u  t  can

be solved for MDF systems.

Example 9.8

For the two-story shear frame in Example 9.4, determine the modal expansion of the
T
displacement vector u  1 1 .

Solution
T
The displacement vector u is substituted in Eq. (9.36) together with 1  1
2 1 and
T
2  1 1 , from Example 9.4, to obtain

 2m  1
1
1  
2
 m  1 2m 4
q1   
 2m   2  3m / 2 3
1
1
1   
2
 m  1 

 2m  1
1 1  
 m  1 m 1
q2   
 2m  1 3m 3
1 1    
 m  1

Substituting qn in Eq. (9.33) gives the desired modal expansion, which is shown in Fig. E9.8.

Figure E9.8

9 - 23
9.4 Free vibration of undamped MDF systems

Given initial conditions u  0  and u  0  , the displacement vector u  t  of undamped

MDF systems in free vibration is to be determined. Consider the equation of motion

  ku  0
mu (9.37)

Assume that the displacement vector u  t  is represented by modal expansion

N
u  t    r qr  t  (9.38)
r 1

where modal coordinates qr  t  varies with time. Substituting Eq. (9.38) into Eq. (9.37), then

we obtain
N N

 mr qr  t    kr qr  t   0


r 1 r 1
(9.39)

Multiplying nT from the left, we get

N N

 nT mr qr  t    nT kr qr  t   0


r 1 r 1
(9.40)

Many terms in the summations are zero by mode orthogonality. There remains only one non-
zero term when r  n ; thus, Eq. (9.40) is simplified to be

M n qn  t   K n qn  t   0 or qn  t   n2 qn  t   0 (9.41)

where M n and K n are generalized mass and stiffness for mode n as defined earlier in Eq.

(9.22). This Eq. (9.41) is a scalar equation in term of unknown modal coordinate qn  t  for

mode n and it is called “modal equation for mode n.” The solution of modal equation is the
same as Eq. (2.10) because Eq. (9.41) is the same as equation of motion for an undamped SDF
system with mass M n and stiffness K n , and natural frequency n .

qn  0 
qn  t   qn  0  cos nt  sin nt (9.42)
n

9 - 24
The initial values qn  0  and qn  0  can be obtained from modal expansion of the given initial

conditions u  0  and u  0  as

nT mu  0  nT mu  0 


qn  0   qn  0   (9.43)
Mn Mn

Once qn  t  is determined from Eq. (9.42) using initial values from Eq. (9.43), it can be

substituted into Eq. (9.38) to obtain displacement vector u  t  of the undamped MDF system

in free vibration.

N
 q  0  
u  t    n  qn  0  cos nt  n sin nt  (9.44)
n 1  n 

Example 9.9

Determine the free vibration response of the undamped two-story shear frame in
T
1
Example 9.4 due to initial displacement u(0)  1 and zero initial velocity u  0   0 .
2
Solution

The initial displacement and velocity vectors are

1  0 
u  0   2  u  0    
1  0 

For the given u(0), qn(0) are calculated following the procedure of Example 9.8 and using n

from Eq. (e) of Example 9.4; the results are q1(0) = 1 and q2(0) = 0. Because the initial velocity

u (0) is zero, q1 (0)  q2 (0)  0 . Inserting qn (0) and qn (0) in Eq. (9.42) gives the solution for
modal coordinates as
q1  t   1cos 1t q2  t   0

Substituting qn (t ) and n in Eq. (9.38) leads to

9 - 25
u1 (t )   12 
u t        cos 1t
u2 (t )  1 

where 1  k / 2 m from Example 9.4. These solutions for q1 (t ) , u1 (t ) , and u2 (t ) had been

plotted in Fig. 9.2c and d. Note that q2 (t )  0 implies that the second mode has no contribution

to the response, which is all due to the first mode. Such is the case because the initial
displacement is proportional to the first mode and hence orthogonal to the second mode.

Example 9.10

Determine the free vibration response of the undamped two-story shear frame of

and zero initial velocity u  0   0 .


T
Example 9.4 due to initial displacement u(0)  1 1

Solution

The calculations proceed as in Example 9.9, leading to q1  0   0 , q2  0   1 , and

q1 (0)  q2 (0)  0 . Inserting these initial values in Eq. (9.42) gives the solutions for modal

coordinates:
q1  t   0 q2  t   1cos  2 t

Substituting qn(t) and n in Eq. (9.38) leads to

u1 (t )  1 
     cos 2t
u2 (t )   1

where 2  2 k / m from Example 9.4. These solutions for q2(t), u1(t), and u2(t) had been

plotted in Fig. 9.3c and d. Note that q1  t   0 implies that the first mode has no contribution

to the response and the response is due entirely to the second mode. Such is the case because
the initial displacement is proportional to the second mode and hence orthogonal to the first
mode.

9 - 26
Example 9.11

Determine the free vibration response of the undamped two-story shear frame of Example 9.4
T
1
due to initial displacements u(0)   2 and zero initial velocity u  0   0 .
2
Solution

Implement modal expansion of the given initial condition u  0  and u  0  , we obtain

q1 (0)  q2 (0)  1 and q1 (0)  q2 (0)  0 . Substituting these initial values in Eq. (9.42), we get

the solution for modal coordinates

q1  t   1cos 1t q2  t   1cos 2t

Substituting qn(t) and n in Eq. (9.39) leads to

u1 (t )   12  1 
     cos 1t    cos 2t
u2 (t )  1   1

These solutions for qn(t) and uj (t) had been plotted in Fig. 9.1c and d. Observe that both natural
modes contribute to the response due to these initial displacements.

9.5 Free vibration of damped MDF systems

If the damping matrix c of the MDF system is such that C  ΦT cΦ becomes a diagonal
matrix, then the system has classical damping. The equations of motion can be uncoupled and
solved by modal analysis as in Section 9.4. If C  ΦT cΦ is not a diagonal matrix, the system
is non-classically damped and must be solved by numerical method; and its eigenvalues will
be complex numbers. This course only considers classically damped systems and the
uncoupled modal equation is

M n qn  Cn qn  K n qn  0
(9.45)

where generalized damping coefficient for mode n is

Cn  nT cn (9.46)

Then, the modal damping ratio is

9 - 27
Cn
n  (9.47)
2 M nn

Dividing Eq. (9.45) by M n , we have the modal equation

q n  2 nn qn  n2 qn  0 (9.48)

The solution qn  t  of Eq. (9.48) will be the same as free vibration of a damped SDF system

(Eq. 2.17).

 q  0    nn qn  0  
qn  t   e nnt  qn  0  cos nD t  n sin nD t  (9.49)
 nD 

where

nD  n 1   n2 (9.50)

The displacement vector u  t  of the damped MDF system is

N
 q  0    nn qn  0  
u  t    n e nnt  qn  0  cos nD t  n sin nD t  (9.51)
n 1  nD 

Example 9.12

Determine the free vibration response of the damped two-story shear frame of Fig.
 6c 2c 
E9.12a with damping matrix c    and c  km / 200 due to initial displacement
 2c 2c 
T
u(0)  1
2 1 .

Solution

 6c 2c   0.5 
generalized damping coefficient C1  1T c1  0.5 1      1.5c
 2c 2c   1 
The initial value qn(0) corresponding to this u(0) were determined in Example 9.9: q1 (0)  1

and q2 (0)  0 ; q1 (0)  q2 (0)  0 . The differential equations governing qn(t) are given by Eq.

(9.49). Because q2 (0) and q2 (0) are both zero, q2 (t)  0 for all times. The response u  t 

9 - 28
involves only one term where n  1 in Eq. (9.51). Substituting the aforementioned values for
T
q1 (0) , q1 (0) , and 1  1
2 1 gives

u1 (t )   12   11t  1 
 
   e  cos  t  sin  t 
u2 (t )  1   1D
1   2 1D

 1 

C1 1.5 200 km
where 1  k / 2 m from Example 9.4, and  1    0.05 according to
2M 11  2   32m  2km

Eq. (9.47). This solution is plotted in Fig. E9.12d.

Note that q2 (t)  0 implies that the second mode has no contribution and the response

is due entirely to the first mode. Such is the case because the initial displacement is proportional
to the first mode and the system has classical damping.

(c)
1

q1 0 t

-1
a b c de T1D  2 / 1D
1

u2 0 t

-1
1

u1 0 t

-1
0 T11 21
2T 3T31
(d)
Figure E9.12

9 - 29
Example 9.13

Determine the free vibration response of the two-story shear frame of Fig. E9.13a with
T
c  km / 200 due initial displacement u(0)  1 1 .

Solution

 6c 2c   1
generalized damping coefficient C2  2T c2  1 1      12c
 2c 2c   1 
The initial values qn(0) corresponding to this u(0) were determined in Example 9.10: q1 (0)  0

and q2 (0)  1 ; qn (0)  0 . The differential equations governing qn(t) are given by Eq. (9.49).

Because q1(0) and q1 (0) are both zero, q1 (t )  0 for all times. The response involves only one

term in Eq. (9.51) when n  2 . Substituting for q2 (0) , q2 (0) , and 2  1 1 gives

u1 (t )  1   22t  2 


    e  cos 2 D t  sin 2 D t 
u2 (t )   1  1   22 
 

C2 12 200km
where 2  2 k / m from example 9.4, and  2    0.10 according to
2 M 22  2  3m    2k
m

Eq. (9.47). This solution is plotted in Fig. E9.13d.

Note that q1 (t )  0 implies that the first mode has no contribution and the response is

due entirely to the second mode. Such is the case because the initial displacement is
proportional to the second mode and the system has classical damping.

9 - 30
1 (c)

q2 0 t

-1 bd T2 D  2 / 2 D
a ce
1

u2 0 t

-1
1

u1 0 t

-1
0 T11 21
2T 3T31

0 0.5
T2 1 1.5
3T2 2 2.5
5T2 3
(d)
Figure E9.13

9 - 31
Exercises

1. Determine the natural vibration frequencies and modes of the system in terms of the DOFs
shown in Fig. P9.1.

Figure P9.1

2. For two-story shear building with rigid beams shown in Fig. P9.2:
a) Determine natural vibration frequencies and modes in terms of m, EI, and h.
b) Verify that the modes satisfy the orthogonality properties.
c) Sketch the modes associated with its natural frequencies by normalizing each mode so
that the roof displacement is unity.

Rigid beams m/2 Rigid beams m/2

h EI EI EI EI
m m

h EI EI EI EI

2h 2h

Figure P9.2 Figure P9.3

3. The structure in Fig. P9.2 is modified so that the columns are hinged at the base as shown
in Fig. P9.3. Determine the natural vibration frequencies and modes and compare them to
results in Problem 2. Comment on the effect of the column support on the vibration
properties.

9 - 32
4. If the structure in Problem 2 is displaced as shown Fig. P9.4 and released, determine the
free vibration response of the structure and comment on the relative contributions of the
two vibration modes to the response that is produced by the two initial displacements (Fig.
P9.4a and b).
a) assume no damping for each mode
b) assume 5% damping for each mode

Rigid beams m/2


2cm 1cm

h EI EI
m
1cm -1cm

h EI EI

2h (a) (b)

Figure P9.4

5. For the system defined in Problem 3 of Chapter 8 (Fig. P8.3) with m  40, 000 kg ,
k  260 kN/m , and b  7.5m .
a) Determine natural frequencies and modes.
b) Sketch these modes by normalizing each mode so that generalized mass M n has a unit

value.

9 - 33

You might also like