Noise and Vibrations

(BDC4013)
Determination of Natural Frequencies and Mode Shapes

DR MUHD HAFEEZ B ZAINULABIDIN

MOHD NO RIHAN B IBRAHIM
Universiti Tun Hussein Onn Malaysia
Necessity to Use Computational Method

 In two degrees of freedom system, solving

the natural frequencies can be conducted
by simply calculating the root of the second
order polynomial.
A 4  B 2  C  0

By assuming An 2  Bn  C  0

Then the natural frequencies can be found

2
Classical Methods
 Standard Matrix Iteration Method
 Dunkerly’s Method
 Rayleigh’s Method
 Holzer’s Method

3
Standard Matrix Iteration

Considering a general
equation of motion  M x   K  x  0

Assuming harmonic xi (t )  X i sin(t )

motion

Equation to solve  2  M  X    K  X   0

4
 Standard Matrix Iteration
(two possible solution)
 2  M  X    K  X   0

1 1
multiply K  multiply M 
1 1 1 1
 2  K   M   X    K   K  X   0  2  M   M  X    M   K   X   0
1 1
 2  K   M  X    I  X   0  2  I  X    M   K  X   0
1 1 1
K   M  X   2  
X  M   K  X    2  X 


Converge to lowest nat freq Converge to highest nat freq

5
Standard Matrix Iteration
(Solution procedures to obtain the lowest nat freq)

(1) Identify matrix [K] and [M]

(2) Calculate [K]-1

(3) Define the initial trial vector {X} and convergence criteria

(4) Multiply [K]-1 [M] {X} = {Xnew}

(5) Normalized the result {Xnew}/largest Xnew

(6) Check the convergence , use for a new trial {X}

(7) When it is converged

1
n 
X normalized

6
Example 1

 Find the natural frequencies

and mode shapes of the
system as shown for
k1=k2=k3=k and
m1=m2=m3=m by the matrix
iteration method.
Solution
1 1 1 
1 
 Flexibility matrix [a]=[k]-1= k
1 2 2 
1 2 3
1 1 1
 Dynamical matrix is k 1m  m 1 2 2
k
1 2 3

 Eigenvalue problem:
1 1 1 
    k 1
DX  X where D  1 2 2 and    2
m 
1 2 3
Solution

1st natural frequency:

1 3
     
 Assume X 1  1 Hence X 2  D X 1  5
1 6
  
 By making the first element equal to unity:

1.0000 
we obtain    k
X 2  3.01.6667 , 1  3.0, 1  0.5773
2.0000 m
 
Solution

 Subsequent trial eigenvector can be

 
obtained from X i 1  Di X i

 Corresponding eigenvalues are given by 1  X 1,i 1
 st

where X 1,i 1 is the 1 component of X i 1
before normalization.
Solution
 The various λi and are shown below:

 The mode shape and natural frequency

converged in 8 iterations.
Solution

2nd natural frequency:

 1  1T
 Deflated matrix D2   D1   1 X X m
1.00000 
 1
 Let the normalized vector X   1.80194 
2.24698
 
 where α must be such that
T
1.00000  1 0 0 1.00000 
 1T  1 2   0 1 0 1.80194 
X mX   m 1.80194    
2.24698 0 0 1 2.24698
 
  2 m9.29591  1
Solution

0.32799
 α=0.32799m-1/2 , hence  1 1 / 2  
X  m 0.59102
0.73699
 

T
1 1 1  0 . 32799   0 . 32799  1 0 0
  
D 2   1 2 2   5 . 04892  0 . 59102   0 . 59102 
0
 1 0 
1 2 3   0 . 73699   0 . 73699   0 0 1 
  
 0 . 45684 0 . 02127  0 . 22048 
  0 . 02127 0 . 23641  0 . 19921 
  0 . 22048  0 . 19921 0 . 25768 
Solution
1
 
 Let X 1  1
1

 By using the iterative scheme, we obtain

 0.25763   1.00000 
    
X 2   0.05847   0.25763 0.22695 
 0.16201  0.62885
   
 2  0.25763
Solution
 Continuing the procedure,

 1.00000
k   
, X  2   0.44496
 Hence λ2=0.64307, ω2=1.24701 m  0.80192
 
Solution

3rd natural frequency:

 Use a similar procedure as before.

 Before computing [D3], need to normalize

 0 . 73700 
 2   2   
X to give X   0 . 32794 
  0 . 59102 
 
Dunkerley’s Formula

 It gives the approx value of the fundamental

freq of a composite system.
 Consider the following general n DOF system:
1
 k    2 m  0 or  2 I   am  0

 For a lumped mass system with diagonal
mass matrix, the equation becomes:
1 0 ... 0  a11 a12 ... a1n  m1 0 ... 0
   ... a2 n   0 m2   
1 0 1     a21 a22
 2  0
     0         0
    
 0 ... 0 1   an1 an 2 ... ann   0 ... 0 mn 
Dunkerley’s Formula
1
 i.e. 
2
 a 11 m 1 a12 m 2 ... a1 n m n
1
a 21 m 1   a 22 m 2 ... a2nmn
2 0
  
1
a n1 m 1 an2m2 ...   a nn m n
2

 Expanding:
 1 
n
 1 
n 1

 
 2   a11m1  a22 m2  ...  ann mn  2 
   
 (a11a22 m1m2  a11a33m1m3  ...  an1,n 1ann mn1mn
n2
 1 
 a12 a21m1m2  ...  an 1,n an,n1mn 1mn ) 2   ...  0 (E.1)
 
Dunkerley’s Formula

 Let the roots of this equation be 1/ω12,

1/ω22,…, 1/ωn2. Thus
n n 1
 1 1  1 1   1 1   1   1 1 1  1 
 2  2  2  2 ... 2  2    2    2  2  ...  2  2   ...  0 (E.2)
  1   2    n      1 2 n   

 Equating coefficients of (1/ω2)n-1 in (E.1) and

(E.2): 12  12  ...  12  a11m1  a22 m 2  ...ann m n
1 2 n

 In most cases, 1 1
2
 2 , i  2,3,..., n
i 1
Dunkerley’s Formula
1
 Thus i2  a11m1  a22 m2  ...  ann m (Dunkerley' s formula)
 Can also be written as 1  1  1  ...  1
2 2 2 2
i 1n 2 n nn

where ωin=(1/aiimi)1/2=(kii/mi)1/2
Dunkerly’ Formula
(calculation procedures)
(1) Identify k11, k22, knn, m1, m2, mn

(2) Calculate natural frequency of the individual component

knn nn : natural frequency of a SDOF system

nn 
mn consisting m n and spring of stiffness k nn

(3) Predict the fundamental natural frequency of the system

1 1 1 1
     n : fundamental (lowest) natural frequency
n 2 112 22 2 nn 2

21
Example 2

 Estimate the fundamental natural frequency

of a simply supported beam carrying 3
identical equally spaced masses, as shown
below.
Solution

3 3
3 l 1 l
a 11  a 33  , a 22 
256 EI 48 EI
m1=m2=m3=m
1  3 1 3  ml 3 ml 3
2
     0.04427
1  256 48 256  EI EI
EI
1  4.75375
ml 3
Problem 1

 Estimate the fundamental natural frequencies

of the system shown below. Given
k1=k2=k3=k and m1=m2=m3=m
Solution
1 1 1
Flexibility matrix 1
a  1 2 2
k
1 2 3

Apply the related equation’s to solve it….

Does the value of nat. freq. smaller & larger

than the exact value of nat. frequency…?
How many percent ?
Rayleigh Method
 This method predicts the fundamental
(lowest) natural frequency
 This method based on energy method

1 1 T
T  mx 2 T  x  M  x
2 2

1 1 T
V  kx 2 V    K  x
x
2 2

26
Rayleigh Quotient
1 T 1 T
T  x  M  x V  x  K  x
2 2

 x   X  sin(t )
 x    X  cos(t )
1 T 1 T
Tmax     M  X   2
X Vmax     K  X 
X
2 2

Tmax  Vmax

T
2  X   K  X 
  T
   M  X 
X

27
Rayleigh Method
(Calculation procedures)
 Identify [K] and [M]
 Select any trial vector mode {X}
 Predict the fundamental natural frequency
based on the Rayleigh Quotient

2 
 X   K  X 
T
 X   M  X 

28
Example 3

 Estimate the fundamental

frequency of vibration of the
system as shown. Assume that
m1=m2=m3=m, k1=k2=k3=k, and

1
  
the mode shape is X  2
3
 
Solution
 2 1 0 
k   k   1 2  1
 Stiffness matrix  0 1 1 
1 0 0

 Mass matrix m  m 0 1 0

0 0 1


 Substitute the assumed mode shape into R X  
2 1 0  1 
 
1 2 3k  1 2  1 2
  0 1 1  3 k k
 
R X  2 
1 0 0 1 
 0.2143   0.4629
m m
 
1 2 3m 0 1 0 2
0 0 1  3
Holzer’s Method

 A trial-and-error method to find natural

frequencies of systems
 Requires several trials
 The method also gives mode shapes
Holzer Method

I11  kt1 (1   2 )  2 I11  kt1 (1   2 )

I 22  kt1 ( 2  1 )  kt 2 ( 2   3 )  2 I 2  2  kt1 ( 2  1 )  kt 2 ( 2  3 )
I   k (   )
3 3 t3 3 2  2 I 3  3  kt 3 (  3   2 )

n
Assume 2
 i   i cos(t   ) 
i 1
I i i  0

32
Holzer Method
(calculation)
2  2 I11
 I11  kt1 (1   2 ) 2  1 
kt 1
 2 I 2  2  kt1 ( 2  1 )  kt 2 ( 2  3 )
kt 23  kt 2  2  kt1 (2  1 )   2 I 2 2
kt 1  2 I 22
3   2  ( 2  1 ) 
kt 2 kt 2
 2 I11  2 I 2  2
3   2  
kt 2 kt 2
2
3   2   I11  I 22 
kt 2

33
Holzer Method
(summary calculation)
Torsion Translation

1  1 X1  1
 2 I11  2 m1 X 1
2  1  X 2  X1 
kt 1 k1
2 2
3   2   I11  I 22  X3  X2   m1 X 1  m2 X 2 
kt 2 k2
 2  i 1   2  i 1 
i  i 1   k k 
kti 1  k 1
I  X i  X i 1   
ki 1  k 1
mk X k 

i  2,3, n i  2,3, n

34
Holzer Method
(calculation procedures)
 Set initial ω=0 and set the sweep increment of ω with a value Δω
 Station 1:
X1=1 (or Θ1=1), calculate M1=ω2m1X1 (or ω2I1Θ1)
 Station 2:
Calculate X2 (or Θ 2), calculate M2=M1+ ω2m2X2 (or ω2I2Θ2)
 Station 3:
Calculate X3 (or Θ 3), calculate M3=M2+ ω2m3X3 (or ω2I3Θ3)
 Station n:
Calculate Xn (or Θ n), calculate Mn=Mn-1+ ω2mnXn (or ω2InΘn)

35
Example 4

I1=2 kg m2 Calculate the natural

I2=4 kg m2 frequencies and mode
shapes
Kt=4 MNm/rad

36
37
38
Example 5

 The arrangement of the compressor, turbine

and generator in a thermal power plant is
shown below. Find the natural frequencies
and mode shapes of the system.
Solution
 This is an unrestrained torsional system.
 The table below shows its parameters and
the sequence of computations.
Torsional Systems
 The graph below plots the torque Mt applied
at the last disc against the chosen ω.

 The natural frequencies are the ω at which

Mt=0.
 The amplitudes Θi (i=1,2,…,n) are the mode
shapes of the system
Problem 1

I1=2 kg m2 Calculate the natural

I2=4 kg m2 frequencies and mode
shapes
I3=2 kg m2
kt1=3 MNm/rad
Kt2=2 MNm/rad

42