New Bicycle Pedal Designfor On Road Measurementsof Cycling Forces

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

iOURNM OF APPLIED BIOMECHANiCS, 1996.

12, 130-142
© 1996 by Human Kinelics Publkshers, Inc,

A New Bicycle Pedal Design for


On-Road Measurements of Cycling Forces

Corka Alvarez and Jordi Vinyolas


A pedal is presented thai was designed specifically for the evalualion of cycling
technique of different cyclists in real conditions. Most of the instrumented force
pedals referred to in research literature have been designed for laboratory use, where
pedal weight and dimension have not been considered critical characteristics. By
means of this new instrumented force pedai, which is externally identical to one of
the most popular clipless pedals, cycling forces under real conditions are mea.sured.
The interest in this pedal lies in the fact that it can be used in road trials tike
the sprint or climb, where tridimensional movements must be considered. Some
measurements obtained in hill climb cycling, such as maximum normal force and
force distribution during crank revolution, are also presented and discussed.

The study of cycling, hitherto, has beeti divided into two independently siudied
areas. The first area has focused on factors resisting advance. The second area, bio-
mechanics, has been subdivided into two subjects: injury and power optimization. It is
clear, however, that the study of cycling calls for an interdisciplinary approach. It is
obvious, for example, that neither mechanical nor aerodynamic factors can be studied
independently from physiological aspects, since the cyclist's position on the bicycle
or the force employed for every revolution depends on the particular physiological
characteristics of each cyclist.
In order to evaluate interactions between the cyclist and the bicycle, the forces
and moments that the cyclist produces must be tneasured. This paper focuses on the
system of measurement of pedal forces. Research literature offers several references
concerning the measurement of forces on a bicycle pedal during cycling. Hoes. Binkhorst,
Smeekes-Kuyl. and Vissers (1968) measured ihe forces on a plane produced on the
pedal by means of extensometric gauges. Dal Monte (Dai Monlc. 1983: Dal Monte.
Manoni, & Fucci. 1973) pioneered the measurement of forces and their transmission
via a telemetry system. More recently, other dynamometric pedals have been presented
using extensometric gauges: Hull and Davis (1981). Lafortune and Cavanagh (1983).
Newniiller, Hull, and Zajac (1988). and Menard (1992). Olher researchers have employed
piezoelectric elements: Ericson. Nisell, and Ekholm (1984). Broker and Gregor (1990).
Wheeler. Gregor. and Broker (1992). and Ugarte and Padilla (1992). Wootten and Hull
(1992) focused on causes of injury in cycling, which led to a complex dynamometric
pedal design.

Gorka Alvarez is with the Mechanical Engineering Department. University of Navarra,


P.O. Box 1674. 20(X)9 San Sebasiidn. Spain. Jordi Vinyolas is with the Centro de Estudios e
Investigaciones T&nicas. P.O. Box 1555. P° Manuel de LardizSbal, 15, 20009 San Sebasti&i.
Spain.

130
Bicycle Pedal Design 131

Figure I — General view of Ihe dynamometric bicycle.

Until now, most studies on pedaling technique have been carried out in laboratory
conditions using bicycles on rollers or treadmills. Although rollers do not constrain
lateral motion of the bicycle, it is difficult to maintain the typical balancing movement
of the bicycle in the standing climb. Motorized treadmills provide a flat surface upon
which subjects can ride naturally (Stone & Hull. 1993). ln this case it is possible lo
simulate various trials in the laboratory, modifying the opponent force to the bicycle
advance by adjusting the inclination and speed of the treadmill; however, tbe cyclist
does not experience the wind resistance and we believe that it is always better to take
measurements in conditions that are as real as possible.
Our objective, therefore, has been to design and build a dynamometric bicycle
with which cyclist effort under real conditions can be measured. Figure I shows a
general view of the bicycle, in which pedals, handlebar, and seat forces are measured.
In this paper only dynamometric pedal design—and not that of the whole dynamometric
e^—will be described in detail.

Methodology
Only two forces in each pedal are measured. The reference system defining the orientation
of pedal loads is illustrated in Figure 2. The two force.s evaluated are contained in a
plane that is perpendicular to the pedal axis. The axial force (F,) within the pedal and
the two moments have been omitted, while the moment in the axial direction has to be
zero.
The resultant pedal force can be resolved into two orthogonal components, one
of which is perpendicular to the plane defined by the pedal (F,), while the other is on
a plane perpendicular to the pedal axis (F^). Due to these forces the pedal shaft must
withstand bending moments, which in tum produce strains. Therefore, a strain gauge
solution for ihe sensorization ofthe pedal is adopted.
A completely novel design was investigated, based on a Time pedal. Figures 3
and 4 show the differences between a traditional pedal and the new pedal, ln a conven-
U2 Alvarez and Vinyolas

Figure 2 — Pedal reference system.

ball bearings

pedal shaft

Figure 3 — Location nt ball bearings and sbaft flxation in a traditional pedal.

tional pedal, the pedal shaft is screwed onto the crank and Is supported by two ball
bearings inserted in the pedal. In the new design, the pedal shaft is modified and rigidly
fixed to the pedal. The bearing location has been changed and is now in the crank. Tlie
relative positions of pedal and crank are measured by means of a thin rod (fixed at one
end to the crank and to a potentiometer at the other) placed within a hollow pedal shaft.
The forces on the pedal are measured by means of the strains on the pedal shaft
using strain gauges. Two complete Wheastone bridges are mounted in each pedal to
measure forces F, and Fj (Figures 2 and 4). With this design, ihe sensitivity of the sensor
device can be adjusted, but the maximum forces thai the pedal has to support also must
be taken into account (since the force involved when starting can be over three times
the cyclist s weight; Gregor, Broker. & Ryan, 1991).
Strain gauge position can be observed in Figure 4, Figure 5 shows bridge configura-
tions. Bridge outputs ei and e? are independent of the distance x; these outputs do not
depend on the force's position. Tltis can be proved by expressing ei as follows;

= - ^ (e, - ei - £3 + 64) (1)


Bicycle Pedal Design 133

crank

all bearings

= 0.032 X sO.02

potentiometer fixed
to pedal
rod fixed to crank
0 0,005
0.094
Figure 4 — Location of bail bearings, strain gauges, and potentiometer set up for relative
pedal/crank angle measurement in the new instrumented force pedal.

Figure 5 — Bridge configurations.

where

(2)

e: = -£4 = F (3)
2EI
and where e = output voltage from bridge. K = gauge factor. V = excitation voltage to
bridge, e, = gauge unitary deformation i, F = applied force, x = force application distance,
d = .shaft diameter, E = Young's module, I = inertia moment, and i = separation between
gauges. By substituting Equations 2 and 3 in Equation 1,
f I — l\ • r (4)
where K' is independent of distance x and has the following expression:
KVld
(5)
1S4 Alvarez and Vinyotas

The notninal re.sistance of the gauges is 350 W. The bridges were balanced by
connections between terminals using a manganine wire of known resistance (50 W/m)
instead of conventional wires.
Theoretically, the cross-sensitivity between measurement channels of pedals must
be zero. However, as these effects can appear due to inaccurate construction, some tests
were carried out. Output values from both sensors (bridge) were recorded when normal
and tangential forces were applied to the pedal independently.
It was observed that the sensors" behavior was almost linear, and cross-sensitivities
were low. A sensitivity matrix from ihe sensor data was evaluated, which leads us to
the real applied pedal force:

(6)

where

~ U":,5j " I 18.593 35I.07OJ'


The data in Equation 7 consider the signal amplification made by the telemetry
equipment: ±3 mV of input gives ±5 V of output. In the same way it is necessary to
consider that bridges in gauges are fed with 6 V. If transducers" sensitivities are given
as output per unit load with a gain of unity and a bridge excitation of 1 V. the following
direct sensitivities are obtained:

• Normal force: 2.85 IO"' mV/NV


• Tangential force: 3.13 IO"' mV/NV

In addition, the intluence of where the force is applied on the measurements was
also calculated. Output values obtained from sensors due to load application in three
different positions—interior, centered, and exterior—were tabulated. As it is only pos-
sible to apply a load within the limits of the pedal surface, using the sensitivity matrix
calculated for centrally applied load, the maximum error that can be Introduced is less
than 2%.
The natural frequency of the pedal was measured experimentally. A impact trial was
undertaken and the acceleration response measured by an accelerometer. The calculated
natural frequency was 200 Hz. and thus the relative dynamic error was small.
In addition to measuring pedal forces (Figures 6 and 7), it is essential to measure
the relative positions between each crank and pedal as well as the crank position in
relation to the bicycle frame. The relative angle between the pedal and the crank is also
measured with a device that includes a potentiometer, two pulleys, and a timing belt.
The transmission ratio between pulleys is 1:1. The potentiometer, therefore, signals an
absolute crank position in relation to the bicycle frame, thus avoiding the necessity of
zero-positioning the apparatus each time the bicycle is used. Figure 8 shows the crank
angle measuring device. The potentiometers used present a linearity of ±0.5% and a
maximum output smoothness of 0.!%. The input is ±5 V and the range 340 ± 4°. so
the measured angle presents atypical "sawtooth" wave form. The lack of signal through
an interval of about 20° is easily solved by software that calculates the unmeasured
values by linear interpolation, as done by Hull and Davis (1981).
The signals from each pedal that correspond to forces and angle are transmitted
by means of a cable attached to the cyclist's leg with an elastic strip. This cable connects
the pedal to the telemetry system mounted behind the seat.
Bicycle Pedal Design 135

Figure 6 — Dismantled instrumented force pedal.

Figure 7 — Reassembled instrumented force pedal refitted to crank.

The dynamometric pedals are mounted on a bicycle that also has a sensorized
handlebar and seat. The.se signals are transmitted by a telemetry system to a recording
system on a vehicle that follows the bicycle. The telemetry solution offers a number of
advantages over other solutions where the recording system is included on the bicycle:
The bicycle weight is minimized, signals can be recorded and visualized at the same
time, and there is no limitation on trial measuring time (batteries allow an autonomy
of approximately 2 hr).
A small-sized housing containing a compiete 16-channel data acquisition and
amplification system—including bridge excitation, filters. PCM encoder, and a telemetry
transmitter with approximately 10 mW power output—is located behind the bicycle
136 Alvarez and Vinyolas

Figure 8 — Crank angle measurement system device.

seat. Power supply is provided by a battery pack with a lifelime of 2 hr. The amplification
stage on telemetry has a gain of 1666.66, so that for an input of ±3 mV the output is
±5 V. The amplification system has a nonlinearity <0.19c and a gain error <0.2%. Filters
are of 6 poles Butterworth type wilh a cutoff frequency of 150 Hz.
Data are transmitted to a control station connected to a portable FC equipped with
recording software. This software simultaneously records the 16 channels on the PC
hard disk (2 of which can be visualized on the PC screen during the recording process).
Useful incidents can be introduced into the recording file, such as gear changes and
variation of cyclist position.

Sample Collection

This .section includes two test results, obtained in hill climb cycling, of tbe complete
system, consisting of dynamometric bicycle and recording system. Anexhaustive analysis
is beyond the scope of this paper. These results are presented and cotnpared with results
found in the reference literature as an illustration of the validity of this system, The
cyclist was a first-year professional with considerable amateur experience, who weighed
75 kg and was 1.83 m tall.
The duration of the tesi was approximately 20 min. but this article presents just
two small trials from the total measurement. Results were measured on a hill climb with
an average slope between 8 and 9%. Both [rials were realized with a gear ratio of 42:17.
The first trial corresponds to pedaling while sealed and the second while standing. The
cadence was 65 rpm while seated and 61 rpm while standing. The average output power
in both cases was 390 W and 415 W, respectively. The results obtained from pedaling
seated and standing are compared below.
Figures 9 and 10 show the nonnal and tangential components of the righl pedal
for an average crank revolution on both trials. Slanding presents greater values for both
Bicycle Pedal Design 137

150

100 • • • \

/
50 /y.
/ /
\

2 -50
\

-100
\

V /
' • ' /

-150
90 180 270 360
Crank Angle (degrees)
Figure 9 — Average tangential forces during a Kill climb, seated (solid line) and standing
(broken line) (12 cycles).

200

-1200
0 90 180 270
Crank Angle (degrees)

Figure 10 — Average normal Forces during a hill elimb, seated (solid line) and standing
(broken line) (12 cycles).

force components. In this case the normal force is about twice that of pedaling seated,
but fewer differences in tangential components are observed.
Subsequent analyses can be carried out using the measurements recorded during
the test. It is interesting to obtain, for analysis purposes, a typical pedaling cycle of
360° by averaging several pedaling revolutions produced in similar conditions.
138 Alvarez and Vinyolas

left pedal right pedal

Figure II — Average Torce distribution while climbing seated with a gear ratio of 42:17 at
65 rpm (12 cycles).

left pedal right pedal

Figure 12 — Average force distribution while climbing standing with a gear ratio of 42:17
at 61 rpm (12 cycles).

Figures 11 and 12 show the total force distribution during an average pedaling
cycle (the vector has the same direction as the real force exerted and its module is
proportional to its magnitude) and pedal position. Results are obtained for both left and
right legs. The figures show that the pedaling technique is totally different in both cases.
Moreover, the maximum normal force is produced at approximately 95° when the cyclist
is seated and 145° while standing (where 0° occurs when the pedal is in its highest
point of trajectory, top dead center).
Plain terrain measurements wilh different gear ratios were also done. Only one
section will be mentioned in order to validate the results that have been compared with
those obtained by other authors. Figure 1.^ shows normal and tangential components to
the pedal for mean crank revolution, and Figure 14 shows the total force distribution
and pedal position during a pedaling cycle. These measurements were done with the
same person, on plain terrain, with a gear ratio of 52:15 and a cadence of 87 rpm; an
average power of 470 W was obtained.
Bicycle Pedal Design 139

100
\
0
\ •

-100 i

in
/
0 -200 \ ../.
1
1 -300 \ /
7
J \
/
-400

-500 \
: /
V
-600
0 90 180 270 360

Crank Angle (degrees)


Figure L3 — Average forces cycling in plain terrain, normal (solid line) and tangential
(broken line) (17 cycles).

left pedal light pedal

Figure 14 — Average Force distributinn in plain terrain with a gear ratio of 52:15 at 87
rpm (17 cycles).

Discussion
Pedal Design
Following is a brief discussion conceming some of the design solutions of the pedal,
the inherent limitations of this particular design, and possible measurement inaccuracies.
As these pedals are to be used under real cycling conditions, their design has to
take into account two fundamental constraint.s: weight and size, and adoption of the
automatic fastening .systems (Time system).
In laboratory conditions, the size of dynamometric pedals is not critical since
measurements are obtained in stationary conditions and pedaling is maintained in a
vertical plane, without tilting the bicycle. Our objective, however, focuses on the study
1 "Itl Alvarez and Vinyolas

of road trials such as the sprint or climb where pedal contact with the road is always
a possibility, where movements are not stationary, and where the frequency of pedaling
changes continuously. In these situations no sensors or auxiliary sensing device can be
added to ihe underside of the pedal, a soiution that has been widely used for the
instrumentation of pedals.
The dynamometric pedal must not add excessive weight or affect the way the
cyclist takes curves. Funhennore it has to incorporate a widely used shoe/pedal interface
that allows the cyclist to feel as if riding his or her own bicycle.
Tlie solution adopted for the dynamometric pedal design takes into account that
the pedal is to be used to evaluate the mechanical efficiency of the cyclist and so leaves
aside other interesting topics such as factors that can provoke injuries during cyciing.
Incorporating strain gauges in traditional pedals requires less effort than in modem
automatic pedals, basicaiiy because the latter are more compact. One of the main
problems that must be addressed when placing sensors in a pedal is how to obtain sensor
signals from the rotating element, bearing in mind that the pedal shaft is fixed lo the
crank and that the ball bearings are located inside the pedal. The measurement problem
of rotating elements is usually solved using slip rings. However, in this case, due to
the smali size of the pedal and the complexity of this design, this solution remain.s
impracticable.
It was observed that the sensors' behavior is almost iinear. and cross-sensitivities
are low. Nevertheless, the real applied pedai force was obtained by evaluating the
sensitivity matrix.
The influence of the appiying force point on the measurements is small. Because
of this, the sensitivity matrix used was the one obtained when applying a centered load.
The Introduced error accepting those hypotheses is similar to that obtained with other
pedai designs. For exampie. Stone and Huil (1993) obtained a root-mean-square error
of 6.7 N (FjJ and 14.1 N (FJ when different load combinations were applied.

Sample Results
Pedaling style and force magnitudes, obtained on plain terrain, have a similar shape
(with greater force values, because more power is produced with less pedaling rate) to
the measurements obtained by Cavanagh and Sanderson (1986) for a pedaling rate of
100 rpm and an average power output of 4(K) W. The force distribution in the present
study, especially that of the normal component, is analogous to that obtained by Hull
and Jorge (1985). although force magnitudes are bigger in the present study. This is
due to [he difference in produced power in the two experiments. In the Hull and Jorge
(i985) study, the test was done with a gear ratio of 52:i5 and 80 rpm. but power was
not mentioned. Maximum normal force is about 0.74 BW (body weighl). which is similar
to the 0.8 BW for 90 rpm and 434 W mentioned by Gregor et al. (1991) for pedaling
while seated.
Maximum normal force while standing occurred at a crank angle of 145° which
is similar to that obtained by Stone and Hull (1993). 140°. These authors reported that
2 of their 3 subjects reached a maximum tangential force at a crank angle nearly
coincident with the angle at which the maximum value of the nornial force occurred.
But the third reached the maximum of the tangentiai force at 22°. which is simiiar to
our finding (maximum at about 20").
Gregor ei al. (199i) aiso mentioned normal load values of 1.6 BW for standing
cyciing with 83 rpm and 370 W. which indicates the load ratio between standing and
sitting. Stone and Hull (1993) obtained valuesof the normal force of 989 N (1.27 BW)
Bicycle Pedal Design 141

for standing with 73 rpm and 430 W of power output. In our case we obtained a
maximum of normal force of 1,130 N {1.5 BW) for standing.

Final Comments
A new instrumented force pedal based on a clipless commercial pedal has been developed.
The goal of this design has been the measurement of forces on the pedal in real conditions
such as experienced on a hill climb or during a sprint, when the bicycle/cyclist movement
can no longer be considered planar.
The measurements obtained with this pedal complement the measurements that
can be taken with a laboratory bicycle on rolling elements. Moreover, tbis pedal is part
of a complete instrumented bicycle equipped with a telemetry system, which can be
considered a very useful tool for cycling studies and analysis. One of the main advantages
of this measurement system is that the cyclist can use his or her own bicycle. Only a
few components need to be substituted by their equivalent interchangeable sensing
elements (pedals, handlebar, and seat stick). This does not imply any geometric or
functional change: The cyclist continues to feel as if riding his or her own bicycle.
Some results obtained by measuring the performance of a first-year professional
cycli.st on this dynamometric bicycle have been presented. These results show how the
new pedal design extends force measurement beyond laboratory conditions.

References
Broker, J.P., & Gregor. R.J. (1990). A dual piezoelectric element force pedal for kinetic analysis
of cycling. International Journal of Sport Biomechanics, 6, 394-403.
Cavanagh. P.R.. & Sanderson. D.J. (1986). The biomechanics of cycling: Studies of the pedaling
mechanics of elite pursuit riders. In E.R. Btirke (Ed.), .Science of cycling (pp. 91-122).
Champaign. IL: Human Kinetics.
Dal Monte. A.. Manoni. A.. & Futci, S. (1973). Biomechanical study of competitive cycling: The
forces exercised on the pedals. In S. Cerquiglini. A. Venerando, & J. Warlenweiler (Eds.).
Medicine and Sport: Vol. fi. Biomechanics III (pp, 4.34-4.19). Basel: Karger.
Dal Monie, A, (1983). La valiitazionv funzionale delt'atleta [The functional evaluation of ihe
athlete]. Firenze, Italy: Manualli Sansoni.
Ericson, M.O., Nisei!, R., & Ekholm. J. (1984). Varus and valgus loads on the knee joint during
ergometer cycling. Scandinavian Journal of Sports Science, 6, 39-45.
Gregor. R.J.. Broker. J.P.. & Ryan, M.M. (1991). The biomechanics of cycling. EiT/r/.sfaHrfSpo;-/
Sciences Reviews. 19. 127-169.
Hoes. J,J,. Binkhorst. R.A.. Smeekes-Kuyl. A.E.. & Vissers. A.C. (1968). Measurement of forces
exerted on pedal and crank during work on a bicycle ergometer at different loads, Inter-
nationale Zeitschrift fur Angeuandtc Phvsiolo)>ie. Einschliessiich Arheiisphvsiologie,
26, 33-42.
Hull. M.L., & Davis, R.R. (1981). Measurement of pedal loading iti bicycling: 1. In.strumentation.
Journal of Biomechanics. 14, 843-855.
Hull. M.L., & Jorge, M. (1985). A method for biomechanical analysis of bicycle pedalling, ./owr/ia/
of Biomechanics, 18, 631-644.
Lafortune. M.A., & Cavanagh. P.R. (1983). Effectiveness and efficiency during bicycle riding. In
H. Matsui & K. Kobayashi (Eds.). Biomechanics VIII B (pp. 928-936). Champaign, IL:
Human Kinetics.
Mcnard, M. (1992). Detemiination de la perfonnance optimale du coureur cycliste [Determination
of optimum cyclist performance!. ' " Proceedings of Jornadas Internacionales de Bio-
mecanica de Ciclismo (Doc. 5). San Sebastian, Spain.
142 Alvarez and Vinyolas

Newmilter. J.. Hull. M.L.. & Zajac. F.E. (1988). A mechanicaliy decoupled two force component
bicycle pedal dynamometer. Journal of Biomechanics, 21. 375-386.
Stone, C . & Hull. M.L. (1993). Rider/bicycle interaction loads during standing treadmill cycling.
Jotirnat of Applied Biomechanics. 9, 202-238.
Ugarte. A.. & Padilla. S. (1992). Una nueva apmximacion para \n caracterizacion de ciclisias en
un entorno de laboralorio [A new approach for describing cyclists in laboratory conditions).
ln Proceedings of Jorttadas Iniernacianales de Biomecanica de Ciclismo (Doc. 4). San
Sebastian, Spain.
Wheeler. J.B..Gregor, R.J.. & Broker. J.P. (1992). A dual piezoeleciric bicycle pedal with multiple
shoe/pedal interface compatibility. International Journal of Sport Biomechanics. 8. 251 -
2.58.
Woolten, D.. & Hull, M.L. (1992). Design and evaluation of a multi-degree-of-freedom foot/pedal
interface for cycling. Internationat Journal of Sport Biomechanics, 8. 132-164.

STATEMENT OF OWNERSHIP. MANAGEMENT. AND CIRCULATION OF


THE JOURNAL OF APPLIED BIOMECHANICS (ISSN 1065-8483). as required
by 39 U.S. Code 3685:

The Jotirnal of Applied Biotnechanics (ISSN ]{)65-8483) Is published four limes


a year (quarterly). Subscription fees are .S4{l per year tor individuals and S90 per
year for institutions.
The owner of the Jotirnal of Applied Biomechanics is Human Kinetics Pub-
lishers, Inc., whose office of publication is ai 1607 N. Market St., Champaign.
IL 61820-2200. The editor is Robert J. Gregor. PhD. Department of Health and
Perfonnance Sciences, Georgia Institute of Technology. Atlanta. GA 30332-0110.
The publisher is Rainer Martens, whose address is Box 5076, Champaign, IL 61825-
5076. There are no bondholders, mortgagees, or other security holders.
Average number of copies printed per issue (net press run) during the preced-
ing 12 months is 1480: number of copies nearest to filing date is 1512. Average
nutnber of copies of each issue distributed after mass mailing to .subscribers is 0;
number of copies nearest to filing date is 0. Average number of copies of each
issue distributed in mass mailing to subscribers is 993; number of copies nearest
to filing date is 1036. Average number of copies of each issue distributed free is
64; number of copies nearest to filing date is 64.

You might also like