Exponential versus linear amplitude decay in damped

oscillators
M. I. Molina
Departamento de Física, Facultad de Ciencias, Universidad de Chile, Casilla 653, Santiago, Chile
mmolina@uchile.cl

Abstract: We comment of the widespread belief among some undergraduate students that the
amplitude of any harmonic oscillator in the presence of any type of friction, decays exponentially
in time. To dispel that notion, we compare the amplitude decay for a harmonic oscillator in the
presence of (i) viscous friction and (ii) dry friction. It is shown that, in the first case, the amplitude
decays exponentially with time while in the second case, it decays linearly with time.

Key words: oscillator, viscous friction, dry friction

PACS: 01.55, 46.30M, 46.30P

1. Introduction
The problem of oscillatory motion is, without any doubt, one of the main topics in physics, from elementary up to
advanced courses. An understanding of this motion is also relevant in many areas outside physics, including
chemistry, biology, engineering, medical research and economics, to name a few. In physics, students encounter
oscillatory behavior in classical mechanics, electricity, optics and later, in quantum mechanics.

When considering adding some `realistic' features to a simple harmonic oscillator, such as damping effects, it is a
common practice in many textbooks, to consider only the case of an oscillator in the presence of a viscous force,
where the resistive force is −  v, where  is the viscous coefficient and v is the velocity of the oscillator. As shown
in many textbooks1, this leads to an exponential decrease of the oscillator's amplitude with time. One important case
which is not normally treated in standard textbooks is the oscillator in the presence of sliding (dry) friction, even
though this type of friction is present in all aspects of our life. Because of this, some students are led to believe that,
in the absence of driving forces, the motion of a harmonic oscillator in the presence of any friction, will decay
exponentially in time.

Here we try to dispell that notion by comparing the amplitude decay for a harmonic oscillator in the presence of
viscous friction and dry friction.

2. Viscous friction
A simple example of this is an object like a pendulum, or a block attached to a spring, that oscillates in the presence
of a drag force provided, for instance, by the surrounding air. At low speeds, the resistive force has the form −  v,
where  is the viscous coefficient and is v the velocity of the oscillator. In one dimension, Newton's equation reads:
d 2 x(t )

dx(t )
m 2
= − k x(t ) −  (1)
dt dt
We look for a solution of the form exp(−λ t ). After replacing in Eq.(1), we obtain an equation for λ:
λ2 + b λ +ω2 = 0 (2)

b = P and ω = k/m . Thus, we obtain two solutions for λ = − (b / 2) ± i ω 2 − (b / 2 ) , and the
2
where
oscillator's displacement has the general form x(t ) = exp(−(b / 2)t )[ A cos(Ω t ) + B sin(Ω t )] , where
Ω = ω 2 − (b / 2 ) . To keep things simple, let us take the initial conditions x(0) = x 0 and (dx/dt)0 = 0 . Also,
2

we will assume that b  ω , that is, we are in the ``underdamped regime'', where the system performs many
oscillations before coming to a rest. Thus, we obtain
x(t ) = exp[−(b / 2) t ]( cos(Ω t ) + sin(Ω t )) , (3)

and the oscillator amplitude is then exponentially damped with time. In particular, this means that the maximum
amplitude points x n where the velocity is zero, will also decay exponentially in time or cycle number n. Let us
show explicitly how x n depends on n: From (3), the oscillator's velocity v(t) = dx(t)/dt is found to be
ω x0
2
v(t) = exp[−(b / 2) t ]sin(Ω t ). (4)
Ω
Thus, v = 0 happens at t n = n π / Ω, and the amplitude An will be
An = | x(tn )| = exp[ −(b / 2) n π / Ω]. (5)
From (5) it is clear that the envelope of the oscillatory motion decays exponentially with the half-cycle number n.
This is readily apparent in the envelope of the oscillatory motion depicted in Fig.1.

3. Sliding (dry) friction

This motion has been analyzed by several authors with varying degrees of sophistication2,3,4,5. We will restrict
ourselves here to a very elementary treatment, which can be followed by a student with little or no calculus. The
most common example of this motion is that of a block resting on a rough table and attached to a horizontal spring.
Let us assume equal values for the static and kinetic coefficients of friction. Newton's equation in this case reads:
d 2 x(t )  dx 
m = − k x(t ) − sgn   µ m g , (6)
 dt 
2
dt
where µ is the coefficient of dry friction, and sgn(x) is defined to be +1 for x > 0, -1 for x < 0 and zero for x=0.
Thus, the frictional force is constant in magnitude but acts in the opposite direction to the velocity. Suppose we start
from rest with an initial amplitude A0 . For the block to move, the force provided by the spring k A0 must be
greater that the magnitude of the static friction, µ m g . This means, A 0 > µ m g ≡ A c . Let us assume that
A 0  A c , so that the system will perform several oscillations before coming to rest. After releasing the block
with initial amplitude A 0 , it will traverse a distance A 0 + A1 , where it will come to rest momentarily for the first
time. Conservation of energy requires that the initial potential energy (1/2) k A 20 be equal to the new potential
energy (1/2) k A12 plus the energy lost to friction µ m g (A 0 + A1 ) :
(1/ 2) k A02 = (1/ 2) k A12 + µ m g ( A0 + A1 ). (7)
After rewriting this as
(1/ 2) k ( A02 − A12 ) = µ m g ( A0 + A1 ) (8)
and, after simplifying, one obtains:
A1 = A 0 − 2 ( µ m g / k ) = A 0 − 2 A c . (9)
The block will swing back, provided k A1 > µ m g . If this is so, at the end of the cycle,
we will have:
A 2 = A1 − 2 A c = A 0 − 4 A c . (10)
This continues for a while, and after n of such half-cycles,
An = A0 − 2 n Ac . (11)
Eventually however, the oscillator's maximum amplitude will decrease below A c , and the spring will not be strong
enough to overcome the dry friction and subsequently, all motion will stop.

As Eq.(11) clearly shows, the decay of the maximum amplitude of the oscillator is linear with the half-cycle number
(or time), and is vividly illustrated in Fig.1. This stands in marked contrast with the previous case of the oscillator
subjected to viscous friction.
Now, as to the question of why most Physics textbooks do not consider dry friction and focus on viscous friction
only when dealing with a damped harmonic oscillator, it probably has to do with the rather awkward form of the
dry friction term: -sgn(dx/dt) µ m g , which is not amenable to a simple closed-form solution, as in the viscous
case.

1 1
DISPLACEMENT

DISPLACEMENT
0.5 0.5

0 0

 0.5  0.5

10 20 30 40 50 10 20 30 40 50
TIME TIME

Fig. 1. Amplitude decay of a harmonic oscillator in the presence of viscous friction (left) and dry friction (right).
( Parameters used: m = 1, k = 1, g = 1, γ = 0.1 and µ = 0.04 ).

References
1
Raymond A. Serway, Physics for scientists and Engineers with Modern Physics,
3rd ed. (Saunders, 1990), chap. 13.
2
I. R. Lapidus, ``Motion of harmonic oscillator with sliding friction'', Am. J. Phys. 38, pp. 1360-1361 (1970).
3
R. C. Hudson and C. R. Finfgeld, ``Laplace transform solution for the oscillator damped by dry friction'', Am. J.
Phys. 39, pp. 568-570 (1971).
4
C. Barratt and G. L. Strobel, ``Sliding friction and the harmonic oscillator'', Am. J. Phys. 49, pp. 500-501 (1981).
5
L. F. Zonetti et. al., ``A demonstration of dry and viscous damping of an oscillating pendulum'', Eur. J. Phys. 20,
pp. 85-88 (1999).