Free Vibration of a Shear Building

In Part I we analyzed and obtained the dynamic response for structures modeled as a single-degree-of-
freedom system. Only if the structure can assume a unique shape during its motion will the
single-degree model provide the exact dynamic response. Otherwise, when the structure takes more
than one possible shape during motion, the solution obtained from a single-degree model will be at
best, only an approximation to the true dynamic behavior.
Structures cannot always be described by a single-degree-of-freedom model and, in general, have
to be represented by multiple-degree models. In fact, structures are continuous systems and as such
possess an infinite number of degrees of freedom. There are analytical methods to describe the
dynamic behavior of continuous structures that have uniform material properties and regular geome-
try. These methods of analysis, though interesting in revealing information for the discrete modeling
of structures, are rather complex and are applicable only to relatively simple actual structures. They
require considerable mathematical analysis, including the solution of partial differential equations
which will be presented in Part IV. For the present, we shall consider one of the most instructive and
practical types of structure which involve many degrees of freedom, the multistory shear building.

7.1 Stiffness Equations for the Shear Building

A shear building may be defined as a structure in which there is no rotation of a horizontal section at
the level of the floors. In this respect, the deflected building will have many of the features of a
cantilever beam that is deflected by shear forces only, hence the name shear building. To accomplish
such deflection in a building, we must assume that: (1) the total mass of the structure is concentrated
at the levels of the floors; (2) the slabs or girders on the floors are infinitely rigid as compared to the
columns; and (3) the deformation of the structure is independent of the axial forces present in the
columns. These assumptions transform the problem from a structure with an infinite number of
degrees of freedom (due to the distributed mass) to a structure that has only as many degrees as it has
lumped masses at the floor levels. A three-story structure modeled as a shear building (Fig. 7.1a) will
have three degrees of freedom, that is, the three horizontal displacements at the floor levels. The
second assumption introduces the requirement that the joints between girders and columns are fixed
against rotation. The third assumption leads to the condition that the rigid girders will remain
horizontal during motion.

© Springer Nature Switzerland AG 2019 173

M. Paz, Y. H. Kim, Structural Dynamics, https://doi.org/10.1007/978-3-319-94743-3_7
174 7 Free Vibration of a Shear Building

Fig. 7.1 (a) Single-bay model representation of a shear building. (b) Free body diagram

It should be noted that the building may have any number of bays and that it is only a matter of
convenience that we represent the shear building solely in terms of a single bay. Actually, we can
further idealize the shear building as a single column (Fig. 7.2a), having concentrated masses at the
floor levels with the understanding that only horizontal displacements of these masses are possible.
Another alternative is to adopt a multimass-spring system shown in Fig. 7.3a to represent the shear
building. In any of the three representations depicted in these figures, the stiffness coefficient, or
spring constant ki, shown between any two consecutive masses is the force required to produce a
relative unit displacement of the two adjacent floor levels

Fig. 7.2 (a) Single-column model representation of a shear building. (b) Free body diagram
7.1 Stiffness Equations for the Shear Building 175

For a uniform column with the two ends fixed against rotation, the stiffness or spring constant, k, is
given by
12EI
k¼ ð7:1aÞ
L3
and for a column with one end fixed and the other pinned by
3EI
k¼ ð7:1bÞ
L3

Fig. 7.3 (a) Multimass-spring model representation of a shear. (b) Free body diagram

where E is the material modulus of elasticity, I the cross-sectional moment of inertia, and L the length
of the column.
It should be clear that all of the three representations shown in Figs. 7.1, 7.2 and 7.3 for the shear
building are equivalent. Consequently, the following equations of motion for the thee-story shear
building are obtained from any of the corresponding free body diagrams shown in these figures by
equating to zero the sum of the forces acting on each mass. Hence
m1 €u1 þ k1 u1  k2 ðu2  u1 Þ  F1 ðtÞ ¼ 0
m2 €u2 þ k2 ðu2  u1 Þ  k3 ðu3  u2 Þ  F2 ðtÞ ¼ 0 ð7:2Þ
m3 €u3 þ k3 ðu3  u2 Þ  F3 ðtÞ ¼ 0

This system of equations constitutes the stiffness formulation of the equations of motion for a three-
story shear building. It may conveniently be written in matrix notation as
½Mf€ug þ ½K fug ¼ fFg ð7:3Þ

where [M] and [K] are the mass and stiffness matrices given, respectively, by
176 7 Free Vibration of a Shear Building

2 3
m1 0 0
½M ¼ 4 0 m2 0 5 ð7:4Þ
0 0 m3
2 3
k1 þ k2 k2 0
½K  ¼ 4 k2 k2 þ k3 k3 5 ð7:5Þ
0 k3 k3

and {u}, f€
ug and {F} are, respectively, the displacement, acceleration and force vectors given by
8 9 8 9 8 9
< u1 = <€ u1 = < F1 ðtÞ =
fug ¼ u2 , f€ug ¼ €u2 , fFg ¼ F2 ðtÞ ð7:6Þ
: ; : ; : ;
u3 €u3 F3 ðtÞ

It should be noted that the mass matrix, Eq. (7.4), corresponding to the shear building is a diagonal
matrix (the non-zero elements are only in the main diagonal). The elements of the stiffness matrix,
Eq. (7.5), are designated stiffness coefficients. In general, the stiffness coefficient, kij, is defined as the
force at coordinate i when a unit displacement is given at j, all other coordinates being fixed. For
example, the coefficient in the second row and second column of Eq. (7.5), k22 ¼ k2 + k3, is the force
required at the second floor when a unit displacement is given to this floor.

7.2 Natural Frequencies and Normal Modes

The problem of free vibration requires that the force vector {F} be equal to zero in Eq. (7.3). Namely,

½Mf€ug þ ½K fug ¼ 0 ð7:7Þ

For free vibrations of the undamped structure, we seek solutions of Eq. (7.7) in the form
ui ¼ ai sin ðωt  αÞ, i ¼ 1, 2, . . . , n

or in vector notation

fug ¼ fag sin ðωt  αÞ ð7:8Þ

where ai is the amplitude of motion of the ith coordinate and n is the number of degrees of freedom.
The substitution of Eq. (7.8) into Eq. (7.7) gives

ω2 ½Mfag sin ðωt  αÞ þ ½K fag sin ðωt  αÞ ¼ 0

or factoring out sin(ωt–α) and rearranging terms

 
½ K   ω2 ½ M  f a g ¼ f 0 g ð7:9Þ

which for the general case, is set for n homogenous (right-hand side equal to zero) algebraic system of
linear equations with n unknown displacements ai and an unknown parameter ω2. The formulation of
Eq. (7.9) is an important mathematical problem known as an eigenproblem. Its nontrivial solution,
that is, the solution for which not all ai ¼ 0, requires that the determinant of the matrix factor of {a} be
equal to zero; in this case
7.2 Natural Frequencies and Normal Modes 177

 
 ½ K   ω2 ½ M   ¼ 0 ð7:10Þ

In general, the expansion of the determinant in Eq. (7.10) results in a polynomial equation of degree
n in ω2 which should be satisfied for n values of ω2. This polynomial is known as the characteristic
equation of the system. For each of these values of ω2 satisfying the characteristic Eq. (7.10) we can
solve Eq. (7.9) for a1, a2, . . ., an in terms of an arbitrary constant. The necessary calculations are better
explained through a numerical example.

Illustrative Example 7.1

The structure to be analyzed is the two-story steel rigid frame shown in Fig. 7.4. The weights of the
floors and walls are indicated in the figure and are assumed to include the structural weight as well.
The building consists of a series of frames spaced 15 ft. apart. It is further assumed that the structural
properties are uniform along the length of the building and, therefore, the analysis to be made of an

Fig. 7.4 Two-story shear building for Illustrative Example 7.1

Fig. 7.5 Multimass-spring model for a two-story shear building of Illustrative Example 7.1. (a) Model, (b) Free body
diagram
178 7 Free Vibration of a Shear Building

interior frame yields the response of the entire building. Determine (a) the natural frequencies and
corresponding modal shapes, (b) the equations of motion with initial conditions for displacements
u01, u02, and for velocities u_ 01 , and u_ 02 , respectively, for the first and second stories of the building.
Solution:

(a) Natural Frequencies and Modal Shapes

The building is modeled as a shear building and, under the assumptions stated, the entire
building may be represented by the spring –mass system shown in Fig. 7.5. The concentrated
weights, which are each taken as the total floor weight plus that of the tributary walls, are
computed as follows:

W 1 ¼ 100  30  15 þ 20  12:5  15  2 ¼ 52, 500 lb

m1 ¼ 136 lb: sec 2 =in
W 2 ¼ 50  30  15 þ 20  5  15  2 ¼ 25, 500 lb
m2 ¼ 66 lb: sec 2 =in

Since the girders are assumed to be rigid and fixed at the two ends, the stiffness (spring constant)
of each story is given by Eq. (7.1a) as
12Eð2I Þ
k¼
L3
and the individual values for the steel column sections indicated are thus

12  30  106  248  2
k1 ¼ ¼ 30, 700 lb=in
ð15  12Þ3
12  30  106  118  2
k2 ¼ ¼ 44, 300 lb=in
ð10  12Þ3

The equations of motion for the system, which are obtained by considering in Fig. 7.5b the
dynamic equilibrium of each mass in free vibration, are
m1 €u1 þ k1 u1  k2 ðu2  u1 Þ ¼ 0
m2 €u2 þ k2 ðu2  u1 Þ ¼ 0

In the usual manner, these equations of motion are solved for free vibration by substituting
u1 ¼ a1 sin ðωt  αÞ
ðaÞ
u2 ¼ a2 sin ðωt  αÞ

for the displacements and

€u1 ¼ a1 ω2 sin ðωt  αÞ

€u2 ¼ a2 ω2 sin ðωt  αÞ

for the accelerations. In matrix notation, we obtain

    
k 1 þ k 2  m 1 ω2 k2 a1 0
¼ ðbÞ
k2 k2  m2 ω2 a2 0

For a nontrivial solution, we require that the determinant of the coefficients be equal to zero, that is,
7.2 Natural Frequencies and Normal Modes 179

 
 k 1 þ k 2  m 1 ω2 k2 
 ¼0 ðcÞ
 k2 k 2  m 2 ω2 

The expansion of this determinant gives a quadratic equation in ω2, namely

m1 m2 ω4  ½ðk1  k2 Þm2 þ m1 k2 ω2 þ k1 k2 ¼ 0 ðdÞ

or by introducing the numerical values for this example, we obtain

8976ω4  10, 974, 800ω2 þ 1:36  106 ¼ 0 ðeÞ

The roots of this quadratic are

ω21 ¼ 140
ω22 ¼ 1082

Therefore, the natural frequencies of the structure are

ω1 ¼ 11:83 rad= sec
ω2 ¼ 32:89 rad= sec

or in cycles per second (cps)

f 1 ¼ ω1 =2π ¼ 1:88 cps

f 2 ¼ ω2 =2π ¼ 5:24 cps

and the corresponding natural periods:

1
T1 ¼ ¼ 0:532 sec
f1
1
T2 ¼ ¼ 0:191 sec
f2

To solve Eq. (b) for the amplitudes a1 and a2, we note that by equating the determinant to zero in
Eq. (c), the number of independent equations is one less. Thus in the present case, the system of
two equations is reduced to one independent equation. Considering the first equation in Eq. (b)
and substituting the first natural frequency, ω1 ¼ 11.8 rad/sec, we obtain

55, 960a11  44, 300a21 ¼ 0 ðfÞ

We have introduced a second sub-index in a1 and a2 to indicate that the value ω1 has been
used in this equation. Since in the present case there are two unknowns and only one equation,
we can solve Eq. (f) only for the relative value of a21 to a11. This relative value is known as the
normal mode or modal shape corresponding to the first frequency. For this example, Eq. (f) gives
a21
¼ 1:263
a11

It is customary to describe the normal modes by assigning a unit value to one of the amplitudes;
thus, for the first mode we set a11 equal to unity so that
180 7 Free Vibration of a Shear Building

a11 ¼ 1:000
ðgÞ
a21 ¼ 1:263

Similarly, substituting the second natural frequency, ω2 ¼ 32.9 rad/sec into Eq. (b) we obtain the
second normal mode as

a12 ¼ 1:000
ðhÞ
a22 ¼ 1:629

It should be noted that although we obtained only ratios, the amplitudes of motion could, of
course, be found from initial conditions.
We have now arrived at two possible simple harmonic motions of the structure which can
take place in such a way that all the masses move in phase in the same frequency, either ω1 or ω2.
Such a motion of an undamped system is called a normal or natural mode of vibration. The
shapes for these modes (a21/a11 and a22/a12) for this example are called normal mode shapes of
simply modal shapes for the corresponding natural frequencies ω1 and ω2. These two modes for
this example are depicted in Fig. 7.6.

Fig. 7.6 Normal modes for Illustrative Example 7.1 (a) First mode (b) Second mode

We often use the designation first mode or fundamental mode to refer to the mode associated
with the lowest frequency. The other modes are sometimes called harmonics or higher
harmonics. It is evident that the modes of vibration, each having its own frequency, behave
essentially as single-degree-of-freedom systems.
(b) Equations of Motion
The total motion of the system, that is, the total solution of the equations of motion, Eq. (7.7),
is given by the superposition of the modal harmonic vibrations which in terms of arbitrary
constants of integration may be written as
0 0
u1 ðtÞ ¼ C1 a11 sin ðω1 t  α1 Þ þ C2 a12 sin ðω2 t  α2 Þ
0 0
ðiÞ
u1 ðtÞ ¼ C1 a21 sin ðω1 t  α1 Þ þ C2 a22 sin ðω2 t  α2 Þ
7.3 Orthogonality Property of the Normal Modes 181

0 0
Here C1 and C2 as well as α1 and α2 are four constants of integration to be determined from four
initial conditions which are the initial displacement and velocity for each mass in the system. For a
two-degree-of-freedom system, these initial conditions are
u1 ð0Þ ¼ u01 , u_ 1 ð0Þ ¼ u_ 01
ðjÞ
u2 ð0Þ ¼ u02 , u_ 2 ð0Þ ¼ u_ 02

For computational purposes, it is convenient to eliminate the phase angles [α1 and α2 in
Eq. (i)] in favor of other constants. Expanding the trigonometric functions in Eq. (i) and
renaming the constants, we obtain

u1 ðtÞ ¼ C1 a11 sin ω1 t þ C2 a11 cos ω1 t þ C3 a12 sin ω2 t þ C4 a12 cos ω2 t

ðkÞ
u2 ðtÞ ¼ C1 a21 sin ω1 t þ C2 a21 cos ω1 t þ C3 a22 sin ω2 t þ C4 a22 cos ω2 t

in which C1, C2, C3 and C4 are the new renamed constants of integration. From the first two
initial conditions in Eq. (j), we obtain the following two equations:

u01 ¼ C2 a11 þ C4 a12

ðlÞ
u02 ¼ C2 a21 þ C4 a22

Since the modes are independent, these equations can always be solved for C2 and C4.
Similarly, by expressing in Eq. (k) the velocities at time equal to zero, we find
u_ 01 ¼ ω1 C1 a11 þ ω2 C3 a12
ðmÞ
u_ 02 ¼ ω1 C1 a21 þ ω2 C3 a22

The solution of these two sets of Eqs. (l) and (m), allows us to express the motion of the
system in terms of the two modal vibrations, each proceeding at its own frequency, completely
independent of the other, the amplitudes and phases being determined by the initial conditions.

7.3 Orthogonality Property of the Normal Modes

We shall now introduce an important property of the normal modes, the orthogonality property. This
property constitutes the basis of the most important method for solving dynamic problems, the Modal
Superposition Method of multi-degree-of-freedom systems. We begin by rewriting the equations of
motion in free vibration, Eq. (7.7) as

½K fag ¼ ω2 ½Mfag ð7:11Þ

For the two-degree-of-freedom system, we obtain from Eq. (b) of Illustrative Example 7.1

ðk1 þ k2 Þa1  k2 a2 ¼ ω2 m1 a1
ð7:12Þ
k2 a1 þ k2 a2 ¼ ω2 m2 a2

These equations are exactly the same as Eq. (b) of Illustrative Example 7.1 but written in this form
they may be given a static interpretation as the equilibrium equations for the system acted on by
forces of magnitude ω2m1a1 and ω2m2a2 applied to masses m1 and m2, respectively. The modal shapes
182 7 Free Vibration of a Shear Building

may then be considered as the static deflections resulting from the forces on the right-hand side of
Eq. (7.12) for any of the two modes. This interpretation, as a static problem, allows us to use the
results of the general static theory of linear structures. In particular, we may use of Betti’s theorem
which states: For a structure acted upon by two systems of loads and corresponding displacements,
the work done by the first system of loads moving through the displacements of the second system is
equal to the work done by this second system of loads undergoing the displacements produced by the
first load system. The two systems of loading and corresponding displacements which we shall
consider are as follows:

System I:

Forces ω21 m1 a11 , ω21 m2 a21

Displacements a11 , a21

System II:
Forces ω22 m1 a12 ω22 m2 a22
Displacements a12 , a22

The application of Betti’s theorem for these two systems yields

ω21 m1 a11 a12 þ ω21 m2 a21 a22 ¼ ω22 m1 a12 a11 þ ω22 m2 a22 a21

or
 
ω21  ω22 ðm1 a11 a12 þ m2 a21 a22 Þ ¼ 0

If the natural frequencies are different (ω1 6¼ ω2), it follows that

m1 a11 a12 þ m2 a21 a22 ¼ 0

which is the so-called orthogonality relationship between modal shapes of a two degree-of-freedom
system. For an n-degree-of-freedom system in which the mass matrix is diagonal, the orthogonality
condition between any two modes i and j may be expressed as
X
n
mk aki akj ¼ 0, for i 6¼ j ð7:13Þ
k¼1

and in general for any n-degree-of-freedom system as

fagiT ½Mfagj ¼ 0 for i 6¼ j ð7:14Þ

in which {ai} and {aj} are any two modal vectors and [M] is the mass matrix of the system.
As mentioned before, the amplitudes of vibration in a normal mode are only relative values which
may be scaled or normalized to some extent as a matter of choice. The following is an especially
convenient normalization for a general system:
7.3 Orthogonality Property of the Normal Modes 183

aij
ϕij ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð7:15Þ
fagiT ½Mfagj

which, for a system having a diagonal mass matrix, may be written as

aij
ϕij ¼ sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð7:16Þ
P n
mk a2kj
k¼1

in which ϕij is the normalized i component of the j modal vector. For normalized eigenvectors, the
orthogonality condition is given by

fϕgiT ½Mfϕgj ¼ 0 for i 6¼ j

ð7:17Þ
¼1 for i ¼ j

Another orthogonality condition is obtained by writing Eq. (7.9) for the normalized j mode as

½K fϕgj ¼ ω2j ½Mfϕgj ð7:18Þ

Then pre-multiplying Eq. (7.18) by fϕgiT we obtain, in view of Eq. (7.17), the following orthogonality
condition between eigenvectors:

fϕgiT ½K fϕgj ¼ 0 for i 6¼ j

ð7:19Þ
¼ ω2j for i ¼ j

Illustrative Example 7.2

For the two-story shear building of Illustrative Example 7.1 determine (a) the normalized modal
shapes of vibration, and (b) verify the orthogonality condition between the modes.
Solution:
The substitution of Eqs. (g) and (h) from Illustrative Example 7.1 together with the values of the
masses from Illustrative Example 7.1 into the normalization factor required in Eq. (7.16) gives
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð136Þð1:00Þ2 þ ð66Þð1:263Þ2 ¼ 241:31
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ð136Þð1:00Þ2 þ ð66Þð1:629Þ2 ¼ 311:08

Consequently, the normalized modes are

1:00 1:00
ϕ11 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:06437, ϕ12 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:0567
241:31 311:08
1:263 1:629
ϕ21 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:0813, ϕ22 ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 0:0924
241:31 311:08
The normal modes may be conveniently arranged in the columns of a matrix known as the modal
matrix of the system. For the general case of n degrees of freedom, the modal matrix is written as
184 7 Free Vibration of a Shear Building

2 3
ϕ11 ϕ12    ϕ1n
6 7
½Φ ¼ 6
4 ϕ21 ϕ22    ϕ2n 7
5 ð7:20Þ
ϕn1 ϕn2    ϕnn

The orthogonality condition may then be expressed in general as

½ΦT ½M½Φ ¼ ½I  ð7:21Þ

where [Φ]T is the matrix transpose of [Φ], and [M] the mass matrix of the system. For this example of
two degrees of freedom, the modal matrix is
" #
0:06437 0:0567
½Φ ¼ ðaÞ
0:0813 0:0924

To check the orthogonality condition, we simply substitute the normal modes from Eq. (a) into
Eq. (7.21) and obtain
" #" #" # " #
0:06437 0:0813 136 0 0:06437 0:0567 1 0
¼
0:0567 0:0924 0 66 0:0813 0:0924 0 1

We have seen that to determine the natural frequencies and normal modes of vibration of a
structural system, we have to solve an eigenvalue problem. The direct method of solution based
on the expansion of the determinant and the solution of the resulting characteristic equation is
limited in practice to systems having only a few degrees of freedom. For a system of many
degrees of freedom, the algebraic and numerical work required for the solution of an
eigenproblem becomes so immense as to make the direct method impossible. However, there
are many numerical methods available for the calculation of eigenvalues and eigenvectors of an
eigenproblem. The discussion of these methods belongs in a mathematical text on numerical
methods rather than in a text such as this on structural dynamics. One of the most popular methods
for the numerical solution of an eigenproblem is the Jacobi Method, which is an iterative method
to calculate the eigenvalues and eigenvectors of the system. The basic Jacobi solution method has
been developed for the solution of standard eigenproblems (i.e., [M] being the identity matrix).
The method was proposed over a century ago and has been used extensively. This method can be
applied to all symmetric matrices [K] with no restriction on the eigenvalues. It is possible to
transform the generalized eigenproblem, [[K] – ω2[M]] ¼ {Φ} ¼ {0} into the standard form and
still maintain the symmetry required for the Jacobi Method. However, this transformation can be
dispensed with by using a generalized Jacobi solution method (Bathe, K. J. 1982) which operates
directly on [K] and [M].
Examples 7.1 and 7.2 can be solved using MATLAB program. The function of MATLAB are used
to solve eigenproblem using built-in function, eig(K, M). The natural frequencies and normal modes
are estimated using the following MATLAB codes. The outcomes are the natural frequencies (natural
periods) and normalized modal matrix. This MATLAB code adopts the framework proposed by
Anderson and Naeim (2012).
7.3 Orthogonality Property of the Normal Modes 185

clear all
close all

%%%%-GIVEN VALUES-%%%%

%%%Define Mass Matrix

M = [136 0; 0 66]

%%%Define Stiffness Matrix

K = [30700+44300 -44300;-44300 44300]

%%%Solve for eigenvalues (D) and eigenvectors (a)

[a, D] = eig(K, M)

[omegas,k] = sort(sqrt(diag(D)));

%%%Natural frequencies
omegas =sqrt(D)

%%%Natural periods
T =2*pi./omegas;

T1 = 2*pi./omegas(1,1);
T2 = 2*pi./omegas(2,2);

%%%{a}1before changing the unity in the first DOF.

a1 = a(:,1);

%%%{a}2before changing the unity in the first DOF.

a2 = a(:,2);

%%%Change the {a} w.r.t. the unity in the first DOF.

a11 = 1;
a21 = a1(2,1)./a1(1,1);
a12 = 1;
a22 = a2(2,1)./a2(1,1);

a =[];

%%%Calculate the {a}

a(:,1) = [a11, a21] %[a11,a21]
a(:,2) = [a12, a22] %[a12,a22]

%%%aMa = {a}'*[M]*(a}
aMa = a'*M*a; %Eq.7.14

%%%Normalization factor
norm_1 = sqrt(aMa(1,1));
norm_2 = sqrt(aMa(2,2));

%%%Normalized eigenvectors
nom_phi(:,1) = 1./norm_1.*a(:,1); %Eq.7.16 for the first mode
nom_phi(:,2)= 1./norm_2.*a(:,2); %Eq.7.16 for the 2nd mode
nom_phi
%Check the orthogonality condition for Mass Matrix
orth_M = nom_phi'*M*nom_phi; %Eq.7.17

%Check the orthogonality condition for Stiffness Matrix

orth_K = nom_phi'*K*nom_phi; %Eq.7.19
186 7 Free Vibration of a Shear Building

7.4 Rayleigh’s Quotient

Several iterative methods for the solution of an eigenproblem make use of the Rayleigh’s quotient.
The Rayleigh’s quotient may be obtained by pre-multiplying Eq. (7.18) by the transpose of the modal
vector fϕgjT . Hence,

fϕgjT ½K fϕgj ¼ ω2j fϕgjT ½Mfϕgj

The property of the mass matrix [M] being positive definite1 renders the product fϕgjT ½Mfϕgj 6¼ 0,
thus, it is permissible to solve for ω2j :

fϕgT ½ K  fϕg
ω2 ¼ ð7:22Þ
fϕgT ½ M  fϕg

in which for convenience the sub-index j has been omitted.

The ratio given by Eq. (7.22) is known as the Rayleigh’s quotient. This quotient has the following
properties: (1) It provides the eigenvalue ω2j when the corresponding eigenvector {ϕ}j is introduced in
Eq. (7.22). (2) When a vector {ϕ} different from an eigenvector is used, then Eq. (7.22) provides a
value ω2 that lies between the smallest eigenvalue, ω21 and the largest eigenvalue ω2N . (3) Finally, if a
vector {ϕ} that is an approximation to eigenvector {ϕ}j, correct to d decimals is used, then the value
of ω2 obtained from Eq. (7.22) is accurate to 2d number of decimals as an approximation to ω2j .

Illustrative Example 7.3

Use Rayleigh’s quotient to calculate an approximate value for the first eigenvalue of the structure in
Illustrative Example 7.1 beginning with the approximate eigenvector for the first mode {ϕ}T ¼ {1.00
1.50}, then iterate using Eqs. (7.12) and (7.22) to converge to the eigenvalue and eigenvector for the
first mode.
Solution:
The substitution of the given vector {ϕ}T ¼ {1.00 1.50} and the matrices [K] and [M] into the
numerator of Eq. (7.22) results in
 

75, 000 44, 300 1:00
f1:00 1:50g ¼ 42, 025
44, 300 44, 300 1:50

and also into the denominator

 

136 0 1:00
f 1:00 1:50 g ¼ 284:5
0 66 1:50

which substituted into Eq. (7.22) yields

42, 025
ω2 ¼ ¼ 147:9
284:5

1
Matrix [A] is defined as positive definite if it satisfies the condition that for any arbitrary nonzero vector {υ}, the
product {υ)T [A]{υ} > 0.
7.5 Summary 187

The use of the calculated value ω2 ¼ 147.9 together with a1 ¼ 1.00 into the first Eq. (7.12) yields
a1 ¼ 1:00 and a2 ¼ 1:24

A second iteration of Eq. (7.22) with {Φ}T ¼ {1.00 1.24} yields

ω2 ¼ 140:02

This value of ω2 is virtually equal to the solution ω1 ¼ 140.02 obtained for the first mode in
Illustrative Example 7.1
Another popular iterative method to solve an eigenproblem, that is, for structural dynamics, to
calculate natural frequencies and modal shapes, is the Subspace Iteration Method.

7.5 Summary

The motion of an undamped dynamic system in free vibration is governed by a homogenous system
of differential equations which in matrix notation is
½Mf€ug þ ½K fug ¼ 0

The process of solving this system of equations leads to a homogenous system of linear algebraic
equations of the form
 
½ K   ω2 ½ M  f a g ¼ f 0 g

which mathematically is known as an eigenproblem.

For a nontrivial solution of this problem, it is required that the determinant of the coefficients of the
unknown {a} be equal to zero, that is,
 
 ½ K   ω2 ½ M   ¼ 0

The roots of this equation provide the natural frequencies ωi, (i ¼ 1, 2, . . . n). It is then possible to
solve for the unknowns {a}i, in terms of relative values. The vectors {a}i corresponding to the roots
ω2i are the modal shapes (eigenvectors) of the dynamic system. The arrangement in matrix format of
the modal shapes constitutes the modal matrix [Ф] of the system. It is particularly convenient to
normalize the eigenvectors to satisfy the following condition:

fϕgiT ½Mfϕgi ¼ ½I  i ¼ 1, 2, . . . , n

where the normalized modal vectors {ϕ}i are obtained by dividing the components of the vector {a}ij
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
by fagiT ½Mfagj .
The normalized modal vectors satisfy the following important conditions of orthogonality:

fϕgiT ½Mfϕgj ¼ 0 for i 6¼ j

fϕgi ½Mfϕgj ¼ 1 for i ¼ j

and

fϕgiT ½K fϕgj ¼ 0 for i 6¼ j

fϕgi ½K fϕgj ¼ ω2i for i ¼ j
188 7 Free Vibration of a Shear Building

The above relations are equivalent to

½ΦT ½M½Φ ¼ ½I 

and

½ΦT ½K ½Φ ¼ ½Ω

in which [Ф] is the modal matrix of the system and [Ω] is a diagonal matrix containing the
eigenvalues ω2i in the main diagonal.
For a dynamic system with only a few degrees of freedom, the natural frequencies and modal
shapes may be determined by expanding the determinant and calculating the roots of the resulting
characteristic equation. However, for a system with a large number of degrees of freedom, this direct
method of solution becomes impractical. It is then necessary to resort to other numerical methods
which usually require an iteration process.

7.6 Problems

Problem 7.1 Determine the natural frequencies and normal modes for the two-story shear building
shown in Fig. P7.1.

Fig. P7.1

Problem 7.2
A certain structure has been modeled as a three-degree-of-freedom system having the numerical
values indicated in Fig. P7.2. Determine the natural frequencies of the structure and the
corresponding normal modes. Check your answer using MATLAB.
7.6 Problems 189

Fig. P7.2

Problem 7.3
Assume a shear building model for the frame shown in Fig. P7.3 and determine the natural
frequencies and normal modes.

3 K/ft
u2

10´
2 K/ft 2 K/ft
u1

For all columns

12´
EI = 106 (K-in.2)
20´ 20´

Fig. P7.3

Problem 7.4
Assume a shear building model for the frame shown in Fig. P7.4 and determine the natural
frequencies and normal modes.

Fig. P7.4
190 7 Free Vibration of a Shear Building

Problem 7.5
Consider the uniform shear building in which the mass of each floor is m and the stiffness of each
story is k. Determine the general form of the system of differential equations for a uniform shear
building of N stories.

Problem 7.6
Find the natural frequencies and modal shapes for the three-degree-of-freedom shear building in
Fig. P7.6.

m3 = 50 Ib . sec2/in/in u3

12 ft W10 X

m2 = 100 Ib . sec2/in/in u2

12 ft W12 X

m1 = 50 Ib . sec3/in/in u1

W14 X82
12 ft

15 ft

Fig. P7.6

Problem 7.7
Use the results of Problem 7.6 to write the expressions for the free vibration displacements u1, u2, and
u3 of the shear building in Fig. P7.6 in terms of constants of integration.

Problem 7.8
Use MATLAB to determine the natural frequencies for the six-story uniform shear building modeled
as a column shown in Fig. P7.8.
7.6 Problems 191

u6
m = 12 kip.sec2/in

u5 E = 29,000 ksi
I = 3200 in4

6 × 12 ft = 72 ft
u4

u3

u2

u1

Fig. P7.8