2.

003SC
Recitation 11 Notes: Double Pendulum System
EIGENVALUES and EIGENVECTORS
Consider the following matrix equation,
Ax = x

(1)

where A is a matrix of size nxn , x is a vector of length n, and is a scalar

For a given matrix, A , the values of i and xi , i = 1, ..., n that satisfy the above equation are called (the
matrixs) eigenvalues and eigenvectors, respectively.
Eigenvalues and eigenvectors are a very important and valuable concept that arises in many technical
elds, especially vibrations.
Consequently, well-established, robust computational procedures exist for evaluating the eigenvalues and
eigenvectors of a matrix.
Connection to Vibrations
Recall the matrix form of the equations of motion for an n-degree-of-freedom system,
M x + Kx = 0
This can be re-written as
x + M 1 Kx = 0
or
M 1 Kx = x
Recall that, for harmonic motion
x = 2 x
So, the matrix equation has the same form as (1) above, i.e. can be seen to be an eigenvalue problem.
(M 1 K)x = ( 2 )x
where
A = M 1 K is the system matrix
the eigenvalues, i , are the natural frequencies, i 2
the eigenvectors, xi , are the natural modes

Double Pendulum System - Problem Statement

Consider a system of two masses and one spring as shown in the gure below. Note that 1 and 2 are
small-angle displacements.

The systems equations of motion are

g
k
k
+
1
2 = 0
m1
l
m1

k
g
k

1 +
+
2 = 0
2
l
m2
m2

1 +

For the special case where m1 = m2 = m ,

Write the equations of motion in matrix notation.
Find the characteristic equation
Find the natural frequencies and natural modes

Double Pendulum System - Solution

EQUATIONS OF MOTION IN MATRIX NOTATION
1 0
0 1

1
2

mk1
( gl + mk1 )
g
mk2
( l + mk2 )

1
2

0
0

(1)

or
M x + Kx = 0

Setting m1 = m2 = m, the equations of motion are

1 0
0 1

1
2

k
k
( gl + m
)
m
g
k
k
(l + m
)
m

1
2

0
0

(2)

CHARACTERISTIC EQUATION
Assume the two masses undergo harmonic motion, i.e. they oscillate with the same frequency, , albeit
dierent amplitudes, a1 , a2 .


1
2

1
2


=


=

a1
a2

a1
a2

cos(t )

(3)


cos(t )

(4)

Substituting (3) and (4) into (2), we obtain,





2
0
0 2

a1
a2


cos(t ) +

k
k
( gl + m
)
m
g
k
k
m
(l + m
)



a1
a2


cos(t ) =

0
0

Dividing by cos(t ), we obtain



2 + ( gl +
k
m

k
)
m

k
m
2
+ ( gl +


k
)
m

a1
a2


=

0
0


(5)

Setting the determinant equal to zero produces the CHARACTERISTIC EQUATION/POLYNOMIAL.

h 2h +

k
m

2
=0

where

h=

g
k
+
l
m

NATURAL FREQUENCIES AND NATURAL MODES

Applying the quadratic formula to the characteristic equation,
k
=h
=
m
2

g
k
+
l
m

k
m

or
r
1 =

g
l

2 =

g
k
+2
l
m

From the first row of (5),



k
g
+ +
l
m
2


a1

k
a2 = 0
m

we can obtain the formula for the natural modes,

a2
m 2 + 2k
=
a1
k
which we evaluate at each of the natural frequencies,

r
1 =

r
2 =

g
l

k
g
+2
l
m

a2
=1
a1

a2
= 1
a1

General Solution
In general, however, (i.e. for arbitrary initial conditions), the systems free response will contain BOTH
natural frequencies,

x1
x2


= A1

1
1


cos(1 t 1 ) + A2

1
1


cos(2 t 2 )

where A1 , A2 , 1 , 2 are determined by initial conditions.

This general response can appear to be very irregular, with little discernible pattern. When the natural
frequencies are close together, beating behavior can be observed.

MIT OpenCourseWare
http://ocw.mit.edu

2.003SC / 1.053J Engineering Dynamics

Fall 2011

For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.