Professional Documents
Culture Documents
Abstract
In general, undergraduate experimental physics laboratories do not usually
have experiments designed to address rolling friction and to measure the
value of the rolling friction coefficient. This work explores an experiment,
which has the potential to arouse students’ curiosity about rolling friction by
addressing a counterintuitive aspect of the behavior of a sphere that rolls up,
stops and then rolls down on an inclined track. In fact, due to the difference in
the net static friction in the upward and downward movements caused by the
rolling friction term, the sphere’s acceleration when rolling upward is higher
than when rolling downward. This difference, which had been predicted
theoretically, was easily demonstrated graphically by video analysis, using
both rubber and mouse balls. For two steel balls, this difference was more
subtle, but it was still possible to detect it numerically. The experimental
setup employed typical laboratory equipment and the free video analysis
software Tracker to collect position and velocity data. The rolling friction
coefficient was calculated with good precision by means of an angular
coefficient of a fitted first-degree function involving appropriate variables.
September 2020 2 P hy s . E d u c . 5 5 ( 2 0 2 0 ) 0 5 5 0 1 0
Understanding the effect of rolling friction in the inclined track experiment
j N N
β
β Ref
i R
in the opposite direction to the rotational move- whereas the negative sign corresponds to rolling
ment [10–16]. This torque must necessarily be upward, because the normal force is shifted to the
caused by one component of the reaction force left from the CM.
acting orthogonallly to the surface and not neces- After some algebraic manipulation, we
sarily directed to the CM. This implies that the obtain:
application point of the reaction force is shifted
γmg mg d ()
a distance d in the direction of the translational f=− sinθ ∓ cosθ î (4)
γ +1 γ +1 R
motion.
In the case of a sphere rolling on an inclined g g d ()
plane (see figure 1), the reaction force applied by a= sinθ ∓ cosθ î . (5)
γ +1 γ +1 R
the surface is shifted, and can be decomposed into
two components: a component parallel to the sur- In the second term of equation (4), the negat-
face (the effective static friction) and a component ive sign corresponds to the sphere rolling down-
orthogonal to the surface (the normal force). ward, therefore in this case the magnitude of
The equations of the motion of the sphere are: the effective friction force is higher and thus the
acceleration magnitude is lower. Therefore, in the
() second term of equation (5), the negative sign also
mgsinθ − f = ma î , (1) corresponds to the sphere rolling downward.
If, instead of rolling on an inclined plane, the
() sphere rolls on an inclined track, the sphere will
N = mgcosθ ĵ , (2) be positioned on the rim of the groove and there
will be two tilted and symmetric normal forces,
a ( ) one on each side of the sphere, as illustrated in
− fR ± Nd = − I k̂ , (3) figure 2. These two normal forces will have the
R
same magnitude, which can be calculated using
where f is the effective static friction, d is the nor- the equilibrium condition and the geometry:
mal force shifted from the CM, R is the sphere’s
radius and I = γmR2 is the moment of inertia (the 2N1 cos β = mgcosθ (6)
sphere has γ = 2/5).
Note: in equation (3), the signs of the effect- mg R
ive static friction torque and the net torque are N1 = N2 = cosθ , (7)
2 Ref
always negative, both for upward and downward
rolling movements of the sphere. However, in the R
where cos β = Ref , R is the sphere radius, Ref =
second term of equation (3), a positive sign corres- √
R2 − L4 is the effective radius and L is the
2
ponds to the sphere rolling downward, because the
normal force is shifted to the right from the CM, groove’s width.
September 2020 3 P hy s . E d u c . 5 5 ( 2 0 2 0 ) 0 5 5 0 1 0
K Maslova et al
Again, due to deformation of the sphere, the if the limits k → 1 and d → 0 are taken simul-
line of action of these two normal forces is shif- taneously in equation (12), then only one term
ted a distance d in the direction of the translational remains, that for the acceleration magnitude due
motion. However, on the inclined track the normal to perfectly rigid surfaces, addressed in the stand-
forces are tilted by an angle β, so that only ( ) the ard undergraduate textbooks.
components of both normal forces in the ĵ dir- A characteristic of all pure rolling phenomen-
ological models is the experimentally confirmed
ection apply torque to the sphere. The magnitude
feature that the magnitude of the friction force
of the net torque is:
(or its torque) is proportional to the magnitude of
tnet = 2N1 cosβ d = mgcosθ d. (8) the normal force. This proportionality coefficient
is presumed to be constant, therefore it does not
Taking these modifications into account, the depend on the relative velocity of the CM and the
movement’s equations become: surface, nor the angle between the weight and the
() surface [10–16]. Therefore, it is used to define the
mgsinθ − f = ma î (9) friction coefficient as the ratio of the two friction
terms versus the normal force. Two friction coeffi-
cients are defined: the first is the ratio between the
a ( ) static friction term and the normal force, and the
− fRef ± mgcosθd = − I k̂ . (10) second one is the ratio between the rolling friction
Ref
term and the normal force:
In the second term of equation (10), the positive γ
sign corresponds to the sphere rolling downward, µs = 2 k2 tanθ (13)
k γ +1
whereas the negative sign corresponds to rolling
upward. Applying the same algebraic manipula-
tion, we get: k d
µr = . (14)
() k2 γ +1 R
γmg 2 mg d
f=− 2 k sinθ ∓ 2 k cosθ î
k γ +1 k γ +1 R The static friction coefficient itself depends
(11) on the inclined plane angle but the rolling fric-
tion coefficient does not depend on this angle. It
g g d () is conceivable that distance d could itself depend
a= sinθ ∓ k cosθ î , (12) on the angle and the velocity of the CM and, con-
k2 γ + 1 k2 γ + 1 R
sequently the rolling friction coefficient, as well.
where k = RRef is a non-dimensional constant. However, most authors assume a priori that for
small angles and velocities, the rolling friction
In the second term of equations (11)–(12),
coefficient is constant [10–16]. These assump-
the negative sign corresponds to the sphere rolling
tions can be confirmed by defining the variable Z:
downward and the positive sign corresponds to the
sphere rolling upward.
The effective static friction in equation (11) aup + adown
Z= . (15)
is composed of the static friction itself, which aup − adown
depends on the sine of the angle of inclination (the
first term) and the rolling friction that depends on Substituting equation (14) in equation (12) and
the deformation of the sphere and the cosine of the calculating equation (15), we obtain the fol-
angle of inclination (the second term). lowing simple relation which can be tested
As a coherence test for the mathematical experimentally:
models, if the limit k → 1 is taken in equation 1
(12) for the acceleration magnitude on an inclined Z= tanθ. (16)
µr (k2 γ + 1)
track, it then reduces to equation (5) for the accel-
eration magnitude on an inclined smooth plane, It should be noted that equation (16) is valid
because there is no longer a groove. In addition, only in pure rolling up to tan θmax . Furthermore,
September 2020 4 P hy s . E d u c . 5 5 ( 2 0 2 0 ) 0 5 5 0 1 0
Understanding the effect of rolling friction in the inclined track experiment
September 2020 5 P hy s . E d u c . 5 5 ( 2 0 2 0 ) 0 5 5 0 1 0
K Maslova et al
Figure 5. Parabolic fits for position versus time using the rubber ball at the lowest tilted track angle of 3.6◦ , graphs
(a) and (b). The graphs presented in (c) and (d) correspond to the highest tilted track angle of 15.7◦ .
Table 2. Anomalous points. using the free software Tracker. The upward and
s x t—acceleration downward accelerations were obtained in two
Case Angle obtained from parabolic fit ways, from the parabolic fit of the position versus
time graph and from the linear fit of the velo-
Ball ⊖ (degree) ∆a (ms−2 ) δa (ms−2 ) δa/∆a (%) city versus time graph. The experimental upward
(b) 15.8 0.021 0.008 38 and downward acceleration values were used to
(c) 15.7 0.146 0.005 3.4 obtain the Z-value using equation (15) and then
(d) 13.4 0.015 0.006 40
the rolling friction coefficient could be be calcu-
(d) 15.8 −0.01 0.02 −200
lated with good precision by means of a linear fit
of the angular coefficient and equation (16).
Table 3. The rolling friction coefficient for each ball in
contact with the aluminum track.
4. Analysis of the results
s x t—acceleration obtained from parabolic fit
Both the upward and downward accelerations can
Ball Slope δ slope µr δµr be obtained by video analysis using the parabolic
(a) 68 1 0.0104 0.0002 fit of the position versus time graph.
(b) 262 18 0.0027 0.0002 In the case of the rubber and mouse balls, the
(c) 80 1 0.0088 0.0002 differences between the upward and downward
(d) 355 35 0.0020 0.0002 accelerations are more pronounced and it is pos-
sible to visualize this difference directly from the
experimental data graphs. As an example, figures
therefore decided to let the ball gently roll down 5(a) and (b) present the parabolic fits (thin pink
the ‘V’ profile carton while its end was placed lines) for the rubber ball at the lowest tilted track
just above the aluminum groove. This procedure angle of 3.6◦ and figures 5(c) and (d) present the
allowed a smooth transition from the downward parabolic fit for the highest angle of tilt of 15.7◦ .
rolling movement in the ‘V’-profile carton to the It is important to note in figure 5(a) that
upward rolling movement in the track groove. the parabolic fit (thin pink line) corresponds to
The video analysis of the ball’s upward the upward rolling movement of the rubber ball
and downward rolling movement was performed (yellow experimental data on the left side of the
September 2020 6 P hy s . E d u c . 5 5 ( 2 0 2 0 ) 0 5 5 0 1 0
Understanding the effect of rolling friction in the inclined track experiment
20 200
Rubber ball Small Steel Ball
(a) (b)
18
175
Experimental data
16 Experimental data
Fit
Fit 150
14 Equation y = a+b*x
Adj. R-Squar 0.98856 0.96857
125
12 Value Standard Error
D Intercept –4.07918 1.41749
10 100 D Slope 262.10894 17.83888
Z
Z
8
Equation y = a+b*x 75
Adj. R-Squar 0.99071 0.99134
6
Value Standard Error
50
D Intercept 0.57199 0.06805
4
D Slope 68.0930 1.07873
25
2
0 0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.05 0.10 0.15 0.20 0.25 0.30
tan (theta) tan (theta)
15
100 Experimental data
Fit
10 Experimental data
Fit
50
5
0 0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.05 0.10 0.15 0.20 0.25
tan (theta) tan (theta)
Figure 6. Graphs of the Z-value versus tanθ for (a) rubber ball, (b) small steel ball, (c) mouse ball and (d) steel
ball.
graph). However, this parabolic fit lies above and (d)) than for the lowest angle of tilt (figures
the red experimental data on the right side of 5(a) and (b)).
the graph, which corresponds to the downward The Z-value (equation (16)) is more con-
rolling movement. Therefore, by a simple visual venient for calculating the rolling friction coef-
inspection, it is clear that the acceleration when ficient because it depends on the tangent of the
rolling upwards is higher than the acceleration angles. Since the tangent function has a high rate
when rolling downwards. Conversely, in figure of increase, even for small angles the Z-value
5(b) the parabolic fit (thin pink line) corresponds provides linear graphs (Z versus tanθ) with good
to the downward rolling movement (yellow exper- resolution and dispersion, providing accurate res-
imental data on the right side of the graph) for the ults for the upward and downward accelerations.
lowest angle of tilt. This parabolic fit lies below One can obtain the graph of Z-value versus
the red experimental data on the left side of the tanθ for all four cases (figure 6).
graph, which corresponds to the upward rolling The anomalous points observed in figures
movement. 6(b)–(d) are probably due to the precision of the
An interesting observation is that for the measurement of the experimental position data
highest angle of tilt, the parabolic fits are closer to and therefore may have no relation to the slipping
the experimental data on both sides (figures 5(c) phenomenon.
September 2020 7 P hy s . E d u c . 5 5 ( 2 0 2 0 ) 0 5 5 0 1 0
K Maslova et al
These anomalies may be explained by the fol- material has a rolling friction coefficient approx-
lowing reasoning: as the difference in acceleration imately one order of magnitude higher than steel
depends on the angle of tilt, a bigger angle implies (all balls were rolling on an aluminum track).
a smaller cosine and therefore a smaller difference However, even for the two steel balls, which have
between the accelerations. This is shown by equa- rolling friction coefficients much smaller than
tion (12): the rubber ball, it is possible to detect a reliable
value for the difference between the values of the
2g d upward and downward accelerations for angles
∆a = aup − adown = k cosθ.
k2 γ+1 R lower than 10◦ .
The definition of the variable Z, which
Therefore, when the values of the upward and includes acceleration data for upward and down-
downward accelerations are closer, their differ- ward rolling in a single expression, provides a
ence approaches the order of magnitude of their more systematic, elegant and precise approach to
own uncertainty and is no longer a reliable value. determining the rolling friction coefficient, com-
This problem can cause a spuriously very high pared with other procedures, such as the calcula-
Z-value if its denominator is a smaller number tion of the average of several individual rolling
than it should be. friction coefficients, one for each angle. In fact,
As an example, let us inspect case (b), the graph of Z versus tanθ yields a well-adjusted
a small steel ball, at 15.8◦ of tilt. The accel- first-degree function, whose fitting parameters
eration difference is ∆a = 1.927 − 1.906 = provide the rolling friction coefficient and its
0.021 ms−2 . If one compares this with the estim- uncertainty simultaneously.
ated uncertainty of δa = 0.008 ms−2 , the ratio
δa/∆a ~ 0.008/0.021 ~ 38%. Comparing this
with the same case for the smaller angle of tilt ORCID iDs
of 3.7◦ gives ∆a = 0.477 − 0.408 = 0.069 ms−2 , K Maslova https://orcid.org/0000-0001-9276-
and a ratio δa/∆a ~ 0.002/0.069 ~ 3%. Table 2 1218
summarizes the view of all anomalous points. V L B de Jesus https://orcid.org/0000-0002-
From equation (16) the angular coefficient is 6995-8378
given by µr (k21γ+1) for each ball. The rolling fric- D G G Sasaki https://orcid.org/0000-0002-
tion coefficient can be obtained from the linear fit 0087-6809
of each graph presented in figure 6. Each value
and its uncertainty are shown in table 3.
Received 28 March 2020, in final form 20 April 2020
Accepted for publication 11 May 2020
https://doi.org/10.1088/1361-6552/ab9217
5. Conclusions
The video analysis proved to be an efficient and References
easy technique for obtaining accurate position [1] Krasner S 1992 Why wheels work: a second
measurements and the respective velocities of dif- version Phys. Teach. 30 212
ferent balls rolling upward and downward without [2] Halliday D, Resnick R and Walker J 2013
sliding on a tilted track, for small angles (less Fundamentals of Physics 10th edn
than 15◦ ). Furthermore, the Tracker tool for fitting (Hoboken, NJ: John Wiley)
[3] Tipler P A and Mosca G 2007 Physics for
curves to experimental data, namely in the posi- Scientists and Engineers 6th edn (San
tion versus time and velocity versus time graphs, Francisco, CA: Freeman)
provides a very pedagogical visualization of the [4] Serway R A and Vuille C 2014 College Physics
difference between the upward and downward 10th edn (Boston, MA: Brooks Cole)
accelerations, which is an unexpected effect due [5] Young H D and Freedman R A 2014 University
Physics 13th edn (Boston, MA: Pearson)
to the rolling friction phenomenon. [6] Katz D M 2015 Physics for Scientists and
The magnitude of this difference is more pro- Engineers; Foundations and Connections 1st
nounced in the case of the rubber ball, because this edn (Boston, MA: Cengage Learning)
September 2020 8 P hy s . E d u c . 5 5 ( 2 0 2 0 ) 0 5 5 0 1 0
Understanding the effect of rolling friction in the inclined track experiment
September 2020 9 P hy s . E d u c . 5 5 ( 2 0 2 0 ) 0 5 5 0 1 0