Theory of Vibration

Multiple Degrees of Freedom (MDOF)

System (9)
Stodola Method
Lecture 20

M.Sc. in Earthquake Engineering

Thapathali Campus, Institute of Engineering
2020
 Stodola Method
Iterative method of analysis, which is usually
formulated in matrix notation and hence also called
the Matrix iteration Method.

 1 
 2 I   a m vˆ  0
 
k    2 m vˆ  0
premultipl ying by
1
a ,
 2

 1 
 2 a k   a m vˆ  0
 
1
vˆ  a m vˆ (1)
 2
 Stodola Method
Dynamic matrix,
D  a m (2)
1
vˆ  Dvˆ (3)
 2

Equation (3) will be satisfied only by those vectors

which represent a true vibration mode shape.
To initiate a process, a trial shape {v1(0)} must be
assumed; it should represent the best possible
estimate of the first mode shape; the amplitude is
arbitrary. The subscript 1 used with this notation
identifies the first mode shape; the subscript (0)
shows that this is the initial shape assumption.
 Stodola Method
When this shape is introduced on the right hand side
of equation (3), a new shape will be obtained by
premultiplying it by [D]; thus

1
v   Dv 
1  0 
 2 1 1 (4)
Actually (4) cannot be written directly because ω is
not known; thus the product of [D] and the assumed
shape will be designated v 1  , that is,
v     Dv   
1
1
1
0 (5)
in which v   is proportion al to the computed shape,
1
1

1
being the unknown proportion ality factor; thus
 2
1
 Stodola Method
When this shape is introduced on the right hand side
of equation (3), a new shape will be obtained by
premultiplying it by [D]; thus
1
v     v   
1 1
(6)
 2 1 1
1
If it is assumed that the computed amplitude is the
same as the initially assumed amplitude, an equation
equivalent to (6) can be used to evaluate the
frequency. Considering the displacement at any
arbitrary point k leads to
1  1
v k1  v k01 (7)
 2
1
0 
v
or,  12  k11 (8)
v k1
 Stodola Method
If the assumed shape were a true mode shape, then
the same frequency would be obtained by taking the
ratio (8) for any coordinate of the structure. In
general, the derived shape v 1  will differ from v 0   ,
and a different result will be obtained for each
displacement coordinate. In this case, the true first –
mode frequency lies between the maximum and
minimum values obtainable from equation (8):

 v k01   v 0  
 1    12   k11 
v  v  (9)
 k 1  min  k 1  max
 Stodola Method
As is obvious, a better approximation of the
frequency can be obtained by an averaging
process. Often the best averaging procedure
involves including the mass distribution as a
weighing factor. Thus, writing the vector
equivalent of equation (7), premultiplying both
sides by v11 T m , and solving for  12 , gives

 2

v  m v 
1
1 T 0 
1

v  mv 
(10)
1  1 T   0
1 1
 Stodola Method
If v
1
1
 and its derived shape v  
1
2
were used in
equation (8) or (10), resulting frequency
approximations would be better than those
computed from the initial assumption. By
repeating the process sufficiently, the mode
shape approximation can be improved to any
desired level of accuracy.
After s cycles,
 
s  1  s 1 
v1  2 v1  (11) 
1
 s 1 
v k1
  s 
2
1
(12)
v k1
 Example: Stodola Method
The Stodola method of matrix iteration method is
demonstrated by calculating the first mode
shape and frequency of the three story building
frame as shown in the figure below.
Although the flexibility matrix of the given
structure could be obtained easily by inversion
of the stiffness matrix derived earlier, it will be
derived here for demonstration purposes by
applying a unit load to each degree of freedom
successively. By definition, the deflections
resulting from these unit loads, shown in the
figure below, represent the flexibility influence
coefficients.
 Example: Stodola Method

(a) structural system (b) flexibility influence

coefficients (x 3,600).
Example: Stodola Method

11 5 2
1   in./kips
a   k 
1
 5 5 2
3600  
 2 2 2

11 7.5 4
1  
D  a m   5 7.5 4 s 2

3600  
 2 3 4
Example: Stodola Method
D  v   
1
0
v   
1
1

v    v    v    v   
1
1
1
2
1
2
1
3

v    v    v    v   
1
3
1
4
1
4
1
5
 Example: Stodola Method
Note that the factor 1/3600 has not been considered
in the phase of analysis because only the relative
shape is important. The shapes have been
normalized by dividing by the largest displacement
component.
From equation (12), using the largest displacement
component, the first mode frequency is found to be

4 
v 11 1.000
  5 
   211
2
1
v 11 1
3600
17.04
 1  14.5 rad/sec
 Example: Stodola Method
in which it will be noted that the factor 1/3,600 has
now been included with the value of v115  .

After one cycle, using equation (9)

0 
 
2
1 min
v 11
 1 
3600
22.5
 160
v 11

0 
 
2
1 max
v 31
 1 
3600
9
 400
v 31
 Example: Stodola Method
A very good approximation can be achieved after this
first cycle by applying the averaging process of
equation (10)
1
22.5 24.75 18.00 13600
1 65.253600
 
2     218
 22.5
1
1077
16.5 
22.5 24.75 18.00  
 9 
 