You are on page 1of 10

Vibrations of Structures

Module III: Vibrations of Beams

Lesson 21: Modal Analysis

Contents:

1. Introduction

2. Modal Analysis of Uniform Beams

3. Orthogonality Relations

Keywords: Modal analysis, Uniform beams, Simply-supported beam, Can-


tilever beam, Eigenfunctions, Orthogonality
Modal Analysis

1 Introduction

Consider a Rayleigh beam described by

ρAw,tt + (EIw,xx ),xx − (ρIw,xtt ),x = 0. (1)

Assuming a modal solution in the form w(x, t) = W (x)eiωt , the differential

equation of the eigenvalue problem is obtained as

−ω 2 [ρAW − (ρIW 0 )0 ] + (EIW 00)00 = 0. (2)

This may be represented as a general eigenvalue problem of the form

−ω 2 M[W ] + K[W ] = 0, (3)

where

d2 d2
    
d d
M[·] = ρA − ρI [·], and K[·] = 2 EI 2 [·]. (4)
dx dx dx dx

We will solve this equation for uniform beams with certain boundary condi-

tions.

2 Modal Analysis of Uniform Beams

Consider the eigenvalue problem of a uniform Rayleigh beam described by

−ω 2 [ρAW − ρIW 00 ] + EIW 0000 = 0. (5)

2
Substituting in (5) a solution of the form

W (x) = Beβ̃x , (6)

where B and β̃ are constants, one can write

EI β̃ 4 + ω 2 ρI β̃ 2 − ω 2 ρA = 0.

Thus, β̃ has four solutions give as β̃ = ±β1 , ±iβ2 , where


 1/2
1 p
β1 = √ − ω 2 ρI + ω 4 ρ2 I 2 + 4ω 2EIρA , (7)
2EI

 1/2
1 2
p
and β2 = √ ω ρI + ω 4 ρ2 I 2 + 4ω 2 EIρA . (8)
2EI

Thus, the general solution of (2) is obtained as

W (x) = B1 cosh β1x + B2 sinh β1 x + B3 cos β2 x + B4 sin β2x, (9)

where Bi , i = 1, . . . , 4 are real constants to be obtained from the boundary

conditions.

For a uniform Euler-Bernoulli beam, using the solution (6) in the differ-

ential equation of the eigenvalue problem

−ω 2 ρAW + EIW 0000 = 0 (10)

we obtain
r
ω 2 ρA
−ρAω 2 + EI β̃ 4 = 0 ⇒ β̃ 2 = . (11)
EI

3
Therefore, we have the four solutions β̃ = ±β, ±iβ, where

β = (ω 2ρA/EI)1/4. (12)

Now, one can write the general solution (for ω 6= 0) of the eigenvalue problem

(10) as

W (x) = B1 cosh βx + B2 sinh βx + B3 cos βx + B4 sin βx, (13)

where Bi , i = 1, . . . , 4 are real constants of integration which are determined

by the boundary conditions of the problem, as discussed below.

Uniform Simply-Supported Beam

The boundary conditions for the eigenvalue problem are given by

W (0) = 0, W 00 (0) = 0, W (l) = 0, and W 00 (l) = 0. (14)

Using the first two conditions from (14) in (9) yield B1 = B3 = 0. The last

two boundary conditions in (14) yield

B2 sinh β1 l + B4 sin β2 l = 0, and B2 β12 sinh β1l − B4 β22 sin β2 l = 0.

For non-trivial solutions of (B2, B4) one must have

sinh β1 l sin β2l = 0 ⇒ sin β2l = 0 (15)

which is the characteristic equation for the problem, whose solutions are

obtained as

β2 = , n = 1, 2, . . . , ∞. (16)
l
4
Substituting this expression of β2 in (8), and solving for ω yield the circular

natural frequencies of a simply-supported uniform Rayleigh beam as


s
2 2
nπ 1 EI
ωnR = 2  , n = 1, 2, . . . , ∞. (17)
l ρA
1/2
I
1 + n2 π 2 2
l A
The eigenfunctions of a simply-supported uniform Rayleigh beam can be

written as
nπx
Wn(x) = B sin , n = 1, 2, . . . , ∞, (18)
l
where B is an arbitrary constant. These eigenfunctions are orthogonal, and

can be normalized to make them orthonormal.

In the case of a simply-supported uniform Euler-Bernoulli beam, we have

the same expression for βn given by (16). Therefore, circular natural frequen-

cies are obtained by


s
2 2
nπ EI
ωnEB = , n = 1, 2, . . . , ∞. (19)
l2 ρA

The eigenfunctions are same as those for the simply-supported uniform Rayleigh

beam.

The two non-dimensional natural frequencies ωnR l/cL, and ωnEB l/cL, where
p
cL = E/ρ, are compared for first few modes in Fig. 1. For lower modes,

the two frequencies tend to match. However, divergence is observed at higher

modes due to higher curvature and consequently the effect of rotary inertia

in the Rayleigh beam.

5
ωn l/cL
25
Euler-Bernoulli
20
15
10
Rayleigh
5

1 2 3 4 5
Mode, n

Figure 1: Comparison of natural frequencies of a simply-supported


p Rayleigh beam and an
Euler-Bernoulli beam for a fixed slenderness ratio sr = l A/I = 10

Uniform Cantilever Beam

Consider a uniform Euler-Bernoulli cantilever beam with boundary conditions

given by

W (0) = 0, W 0(0) = 0, W 00 (l) = 0, and W 000 (l) = 0. (20)

Substituting the solution form (13) in these boundary conditions yields

B1 + B3 = 0, (21)

B2 + B4 = 0, (22)

B1 cosh βl + B2 sinh βl − B3 cos βl − B4 sin βl = 0, (23)

and B1 sinh βl + B2 cosh βl + B3 sin βl − B4 cos βl = 0. (24)

6
3 cosh z

1 −1/ cosh z cos z

Π 2Π 3Π 4Π
z
-1

Figure 2: Graphical representation of the solutions of the characteristic equation of a can-


tilever beam

For a non-trivial solution of the (B1, . . . , B4 ), we must have


1 0 1 0
0 1 0 1
=0
cosh βl sinh βl − cos βl − sin βl
sinh βl cosh βl sin βl − cos βl

⇒ cos βl cosh βl + 1 = 0, (25)

which is the characteristic equation of a cantilever Euler-Bernoulli beam.

The solutions of the characteristic equation (25) are visualized graphically by

circles in Fig. 2. For higher modes, (25) may be approximated as cos βl = 0.

The solution can be expressed as


r  
ρA 2n − 1 1
βn = ω n = π + en
EI 2 l

 2 s
2n − 1 1 EI
⇒ ωn = π + en ,
2 l2 ρA

where en are small (and rapidly diminishing) correction terms obtained as

e1 = 0.3042, e2 = −0.018, e3 = 0.001, . . ..

7
W1 (x)
1

1
-1 x/l

W2 (x)
1

1
-1 x/l

W3 (x)
1

1
-1 x/l

Figure 3: First three eigenfunctions of a cantilever beam

Eliminating B3 and B4 from (23) using (21) and (22), we obtain

sinh βn l + sin βn l
B1 = − B2 := αn B2 . (26)
cosh βn l + cos βn l

Therefore, taking B2 = 1, one possible solution is given by

B1 = α n , B2 = 1, B3 = −αn , and B4 = −1, (27)

which yields the nth eigenfunction as


 
sinh βn l + sin βn l
Wn(x) = sinh βn x − sin βn x − (cosh βn x − cos βn x). (28)
cosh βn l + cos βn l

The modes of vibration corresponding to the first three eigenfunctions are

shown in Fig. 3.

8
3 Orthogonality Relations

Consider the eigenvalue problem of the Rayleigh beam described by (3).

For two different modes j and k, one can write (3) as

−ωj2 M[Wj ] + K[Wj ] = 0 and − ωk2 M[Wk ] + K[Wk ] = 0.

Multiplying the first equation by Wk , the second equation by Wj , subtracting

one equation from the other, and integrating the result over the length of the

beam gives
l
[((EIWj00)0 − ωj2 ρIWj0 )Wk − ((EIWk00)0 − ωj2 ρIWk0 )Wj ] 0 +
Z l
0 l
00 0 00
[EIWk Wj − EIWj Wk ] 0 + (ωj − ωk ) [ρAWk − (ρIWk0 )0]Wj dx = 0.(29)
2 2
0

For the boundary conditions considered above, the boundary terms in (29)

disappear. Hence, we obtain the orthogonality relation


Z l
[ρAWk − (ρIWk0 )0 ]Wj dx = 0, j 6= k, (30)
0

In the case of an Euler-Bernoulli beam, (30) simplifies further to


Z l
ρAWk Wj dx = 0, j 6= k. (31)
0

One may normalize the eigenfunctions with respect to an inner product such

that
Z l
M[Wk ]Wj dx = δjk , (32)
0

where δjk represents the Kronecker delta function. The eigenfunctions so nor-

malized form an orthonormal basis. As a consequence of this orthonormality,

9
one can easily write
Z l
K[Wk ]Wj dx = ωk2 δjk .
0

10

You might also like