Tuned Mass Dampers

A tuned mass damper is a system for damping the amplitude in one oscillator by
coupling it to a second oscillator. If tuned properly the maximum amplitude of
the first oscillator in response to a periodic driver will be lowered and much of the
vibration will be transferred to the second oscillator.
This is used, for example in tall buildings to limit the swaying of the building in the
wind. People are sensitive to this swaying, so by adding a tuned mass damper the
building sways less and the damper, which no one can feel, vibrates instead.
I found a nice powerpoint presentation at
http://www14.informatik.tu-muenchen.de/konferenzen/Jass06/courses/4/Stroscher/S
troscher.ppt.

m2
x2

k2 c2

m1
x1

p0 cos(t) k1 c1

In the figure, the spring system m1 , k1 , c1 is the oscillator to be damped (say a

building) and m2 , k2 , c2 is the damping oscillator. (say a reasonably large mass
attached to the building).
Note: x2 is the absolute position of m2 . This is often replaced by the relative position
of m2 with respect to m1 , i.e., with what we would call x2 x1 .
Assuming that the damping force is proportional to velocity and there is a periodic
force p0 cos(t) on m1 it is easy to work out the differential equations governing the
motion of the system.
We simplify slightly by letting c1 = 0 and get the following equations. (x01 is the time
derivative of x1 .)
m1 x001 + k1 x1 + k2 (x1 x2 ) + c2 (x01 x02 ) = p0 cos(t)
m2 x002 + k2 (x2 x1 ) + c2 (x02 x01 ) = 0

Our goal is to grind through the calculations and find explicit expressions for the
periodic solutions to these equations. The computation is similar to what wed see if
we converted to a 4-by-4 system of first order equations, complexified and looked for



a periodic solution of the form eit K for some 4-vector K.

1
Tuned Mass Dampers 2

Replacing the first equation by the sum of the two equations gives

m1 x001 + k1 x1 + m2 x002 = p0 cos(t)

m2 x002 + k2 (x2 x1 ) + c2 (x02 x01 ) = 0

Now, we find the periodic solution in the form

x1 = a cos(t) + b sin(t)
x2 = c cos(t) + d sin(t)

Substituting these into the differential equations gives the following algebraic system
of equations.

k1 m1 2 0 m2 2 0 a p0
k1 m1 2 m2 2

0 0 b 0
=

k2 c2 k2 m2 w2 c2 c 0

2
c2 k2 c2 k2 m2 d 0

Call the coefficient matrix M . We can write and invert M in block form.
   
0 1 A B
Let W = , then M = , where
1 0 C D
A = r1 I, B = r2 I, C = r3 I s1 W , D = r4 I + s1 W ,
r1 = k1 m1 2 , r2 = m2 2 , r3 = k2 , r4 = k2 m2 2 , s1 = c2 .
Since A and B commute with everything in sight we get

(AD BC)1
  
1 0 D B
M = .
0 (AD BC)1 C A

Now define r and s by, AD BC = (r1 r4 r2 r3 )I + s1 (r1 + r2 )W = rI + sW

1
(AD BC)1 = 2 (rI sW ).
r + s2
Putting this together we get

a rr4 + ss1
rs1 + sr4

b p o
=
c r 2 + s2 rr3 + ss1

d rs1 sr3

The amplitude of x1 is A1 = a2 + b2 and that of x2 is A2 = c2 + d2 . We see that
p0 p0
A1 = (r42 + s21 ) A2 = 2 (r2 + s21 )
r2 +s 2 r + s2 3
Tuned Mass Dampers 3

We write out A21 and A22 explicitly in terms of the parameters.

c22 2 + (k2 m2 2 )2
A21 = p20
[(k1 m1 2 )(k2 m2 2 ) k2 m2 2 ]2 + c22 2 (k1 m1 2 m2 2 )2
c22 2 + k22
A22 = p20
[(k1 m1 2 )(k2 m2 2 ) k2 m2 2 ]2 + c22 2 (k1 m1 2 m2 2 )2

It seems fairly standard to write this in terms of the following constants (the names
are from the Stroscher powerpoint)
k1 k2
eigenfrequencies: 12 = , 22 =
m1 m2
m2
mass ratio: =
m1
c2
damping ratio: =
2m2 2
p0
static deformation: u1,stat =
k1
It is easy enough to write the amplitudes in terms of these parameters. We get

A1 w12 4 2 22 2 +(22 2 )2
u1,stat
= 2 2 2 2 2 2 2 4 4 2 2 2 2 2 2 2 2 2 2 2 2 2 4 2 2 2
(1 ) (2 ) + 2 22 (1 )(2 )+4 2 ((1 ) + 2 (1 ))