ArticlePDF Available

# Sliding on the surface of a rough sphere

Authors:

## Abstract and Figures

A well-known textbook problem treats the motion of a particle sliding frictionlessly on the surface of a sphere. An interesting variation is to consider what happens when kinetic friction is present. This problem can be solved exactly.
Content may be subject to copyright.
326 DOI: 10.1119/1.1607801 THE PHYSICS TEACHER Vol. 41, September 2003
A
well-known textbook problem treats the mo-
tion of a particle sliding frictionlessly on the
surface of a sphere. An interesting variation
is to consider what happens when kinetic friction is
present.
1
This problem can be solved exactly.
A free-body diagram is sketched in Fig. 1 and de-
fines many of the relevant variables. The radial com-
ponent of Newtons second law is
mg cos
N = ma
c
. (1)
Solving for N, which will be needed to compute the
frictional force, we obtain
N = m
gcos
r
2
, (2)
where
(
) is the speed of the object. Now we insert
this into the tangential component of Newtons sec-
ond law,
mg sin
N = ma
t
, (3)
to obtain
g(sin
cos
) +
r
2
=
d
d
t
=
d
d

d
d
t
. (4)
Substituting for the angular speed, d
/dt =
/r, and
using the identity 2
d
/d
= d(
2
)/d
we get
d(
d
V
2
)
–2
V
2
= 2(sin
cos
), (5)
where the dimensionless speed is V
/rg
.
A numerical solution to this differential equation
can be readily obtained using finite-difference itera-
tion in a spreadsheet for a given initial speed and an-
gle.
2
The equation also can be solved analytically in a
calculus-based course as explained in Appendix A.
This solution is graphed in Fig. 2 for various values of
the two parameters
and V
0
, where V
0
is the initial
dimensionless speed at the top of the sphere. For ref-
erence, curve 1 plots the standard frictionless example
of
= 0 and V
0
= 0, showing that the particle flies off
the sphere at
= cos
-1
(2/3) = 48.2 with a dimension-
less speed of V = 2
/3
= 0.816.
There are two possible fates of the object. It will ei-
ther be brought to rest on the surface or it will eventu-
ally lose contact with the sphere. The first situation
occurs if there exists a value
3
of
between and 90°
Sliding on the Surface
of a Rough Sphere
Carl E. Mungan, U.S. Naval Academy, Annapolis, MD
Fig. 1. A block of mass m sliding on the surface of a
rough sphere of radius r, at the instant it is located at
angle
with respect to the vertical. The three forces act-
ing on the object are the normal force N, gravity mg,
and sliding friction
N, where
is the coefficient of
kinetic friction. The acceleration has been resolved into
centripetal a
c
and tangential a
t
components.
THE PHYSICS TEACHER Vol. 41, September 2003 327
for which V = 0, i.e., if a plot of V versus
hits the
horizontal axis in Fig. 2. Curve 2 shows a typical ex-
ample of this behavior. On the other hand, the object
flies off the sphere if N = 0. According to Eq. (2), this
happens when V
2
= cos
, i.e., if a plot of V versus
strays into the shaded region of the graph in Fig. 2.
(In particular, V
0
is constrained to be less than 1 if the
object is to even begin on the sphere. This gives phys-
ical significance to the speed rg
used to normalize
V.) For example, curve 3 shows friction initially slow-
ing down the particle but the increasing gradient of
the surface subsequently re-accelerating the mass
(which occurs in general whenever it does not come to
rest first, i.e., provided
is not too large). By starting
with just the right speed for a given value of
, the
particle can be slowed down arbitrarily close to zero,
so that the particle appears to “bounceoff the hori-
zontal axis in Fig. 2. For example, see curve 4, dis-
cussed in greater detail in Appendix B.
The motion of the block on the surface of the ball
can now be understood by considering its trajectory in
the V-versus-
parameter space of Fig. 2. A curve be-
gins at a point on the vertical axis at which V = V
0
with 0 < V
0
< 1. The plot proceeds rightward until it
ends when it contacts the horizontal axis along the
bottom or the gray region limiting its upward range,
whichever occurs first. It is an instructive exercise (cf.
Appendix B) to prove that it is impossible for the tra-
jectory to pass through the intersection point of these
two bounding curves, i.e., the particle can never reach
the equator. However, it can get arbitrarily close to
90° by skirting the shaded region all the way along (cf.
curve 5 in Fig. 2).
In summary, a rich variety of curves of
(
) are
possible for a point particle sliding on the surface of a
rough sphere. Generating and graphing these curves
can therefore prove a profitable method for students
to learn how to use a spreadsheet in introductory
physics. In particular, it can help them appreciate that
while it is helpful to try random values of the plot pa-
rameters (to generate curves 2 and 3 in Fig. 2, for in-
stance), a more focused approach is necessary to sam-
ple the full range of possible trajectories (e.g., curves 4
and 5) of a nontrivial dynamical system.
Acknowledgment
I thank John Mallinckrodt for the definition of the
dimensionless speed and for suggesting that I investi-
gate the limit as
and V
0
1 simultaneously.
Appendix A:
Analytic Solution of Eq. (5)
This linear, inhomogeneous, first-order ordinary
differential equation can be directly solved for V
2
us-
ing an integrating factor.
4
But an alternative ap-
proach can be used with students who have not yet
taken a differential equations course. One solution
can be found by substituting the trial form
V
1
2
= A cos
+ B sin
into Eq. (5) and separately
equating cosine and sine terms to find
V
1
2
= . (A1)
On the other hand, consider the simpler equation
obtained by setting the right-hand side of Eq. (5) to
zero. This can be rearranged into d (V
2
)/(V
2
) =
2
d
and both sides integrated to obtain
V
2
2
= Ce
2

, (A2)
where C is an arbitrary constant of integration. It is
not hard to see that V
1
2
+ V
2
2
is also a solution of Eq.
(5). (Since this sum contains one undetermined con-
(4
2
2)cos
6
sin

1 + 4
2
Fig. 2. Five trajectories of the particle in V-vs-
space
using the following parameters:
1:
= 0 and V
0
= 0;
2:
= 0.6 and V
0
= 0.5;
3:
= 0.3 and V
0
= 0.6;
4:
= 1 and V
0
0.71944 from Eq. (B4);
5:
= 100 and V
0
0.9999625 from Eq. (B1).
328 THE PHYSICS TEACHER Vol. 41, September 2003
stant and the original differential equation is first
order, this is in fact the general solution.
5
) Finally, the
value of C is obtained by fitting this general solution
to the initial conditions: The particle begins at angle
0
with dimensionless speed V
0
. In particular, sup-
pose that
0
= 0, i.e., the object starts at the top of the
sphere just as in the conventional frictionless problem.
Unlike this standard case, however, V
0
cannot be zero
because friction would then cause the particle to
remain in stable equilibrium (rather than slipping out
of unstable equilibrium) at the north pole of the
sphere.
6
The final solution therefore becomes
V
2
= + V
0
2
e
2

.
(A3)
Appendix B: Approaching the Equator
Setting V = 0 at
=
/2, Eq. (A3) can be solved for
V
0
2
to obtain
V
0
2
= . (B1)
(In order for this to be positive,
must be larger
than approximately 0.6034.) Substitute this back
into Eq. (A3) and put
=
/2
to obtain
V
2
= . (B2)
By design, this is zero at
= 0. However, it equals
(3

2)
when it is expanded to second order in
, i.e., V
2
is negative when
is infinitesimally small.
This means the particle cannot reach the equator: It
must stop at some smaller angle.
3
But how close to the equator can one get? In order
to maximize the final angle
, we want d
/d(V
2
) 0
at V = 0. Consequently
according to Eq. (5).
In this limit, Eq. (B2) becomes
V
2
= sin
2
3
cos
. (B3)
Therefore, the object comes to rest near
/2
1.5/
. For example, curve 5 shows what happens
when
= 100. The particle actually stops at 89.2°,
in good agreement with this prediction.
This is an unrealistically large value for
. The fi-
(4
2
2) sin
+ 6
(e
-

cos
)

1 + 4
2
4
2
2 + 6
e
-


1 + 4
2
(2 – 4
2
)(e
2

cos
) – 6
sin

1 + 4
2
nal angle
is maximized for a more reasonable value
of
= 1 when V
0
is chosen so that the trajectory of
the particle in Fig. 2 just grazes the horizontal axis [so
that both V 0 and d(V
2
)/d
= 0] at, say, angle
r
and then speeds back up and flies off the sphere at
.
The left-hand side of Eq. (5) is zero, thus implying
that
r
= 45, and Eq. (A3) can then be solved at this
angle to deduce that
V
0
2
= 0.4
1 + 2
e
-
/2
. (B4)
Subsequently the particle leaves the surface at angle
= 69.6with a dimensionless speed of V = 0.59,
plotted as curve 4 in Fig. 2.
References
1. H. Sarafian, “How far down can you slide on a rough
ball?” AAPT Announcer 31, 113 (Winter 2001). Sarafi-
an has analyzed this problem using Mathematica for a
special issue of the Journal of Symbolic Computation to
be published in late fall of 2003.
2. Many introductory textbooks now include an overview
of Euler’s method of numerical integration in a spread-
sheet such as Excel. For example, see P.A. Tipler and G.
Mosca, Physics for Scientists and Engineers, 5th ed. (Free-
man, New York, 2003), Sec. 5-4.
3. For many values of
and V
0
, there are two mathemati-
cal solutions of
for which V = 0. Only the smaller so-
lution has physical significance for
0
= 0.
4. Integrating factors are discussed in standard differential
equation texts, such as D.G. Zill and M.R. Cullen, Dif-
ferential Equations with Boundary-Value Problems, 3rd
ed. (PWS-Kent, Boston, 1993), Sec. 2.5.
5. Students who have taken an introductory course in dif-
ferential equations will recognize Eq. (A1) as a particu-
lar solution and Eq. (A2) as the complementary solu-
tion of the homogeneous equation corresponding to
Eq. (5). If even this alternative approach (without the
technical terminology) is too advanced, students could
still be challenged to verify that Eq. (A3) satisfies both
Eq. (5) and the initial conditions.
6. Another reasonable choice of initial conditions has been
adopted in W. Herreman and H. Pottel, “Problem: The
sliding of a mass down the surface of a solid sphere,”
Am. J. Phys. 56, 351 (April 1988).
PACS codes: 46.02C, 46.30P
Carl E. Mungan
is an assistant professor and coordinates
the classical mechanics course at the Naval Academy. His
current research interests are in organic LEDs and solid-
state laser cooling.
Physics Department, U.S. Naval Academy, Annapolis,
MD 21402-5026; mungan@usna.edu