Available online at www.sciencedirect.

com

Procedia Engineering 34 (2012) 479 – 484

9th Conference of the International Sports Engineering Association (ISEA)

Cycling comfort on different road surfaces

Christin Hölzela*, Franz Höchtla, Veit Sennera
a
Technische Universitaet München, Department Sport Equipment and Material
Parkring 35, 85748 Garching bei München

Accepted 02 March 2012

Abstract

The increased awareness for the environment and steadily higher costs for gasoline makes bicycles more and more
attractive for short-distance traffic. Due to this the infrastructure of cycling pathways gets more important not only
for safety reasons but also for the aspect of cycling (dis)comfort. Comfortable cycling requires smooth rolling at
lowest possible energy input. To better understand this relationship, the objective of the study was to determine
rolling resistance and the resulting accelerations due to external agitations for different surfaces.
Rolling resistance was quantified using a one degree of freedom pendulum (transversal rotation). By setting it at an
initial angular displacement the measured decay of oscillations is a direct measure for the rolling resistance. To
investigate the vibrations, an accelerometer was applied under the seat of a racing bicycle. The cyclist had to roll over
a distance of 15 m at 3 different velocities with no pedaling. Four different surfaces were investigated. An effective
value factor was defined for an overall description of the observed data.
The covered distance of the pendulum is directly related to the rolling resistance and varied from 22,1 m on concrete
slabs to 10,2 m on cobblestones. Asphalt and self-binding gravel rank in between. This ranking however changes
when analyzing the vibrations. Lowest effective value factors were measured on asphalt (0, 05) whereas concrete
slabs (0, 17), self-binding gravel (0, 21) and cobblestones (0, 57) are far behind. Comfort decreases with higher
velocities.
The results of this study allow suggestions about the best surface for cycling pathways thus making the systematic
design of comfortable bicycle lanes easier. Designing bike lanes for comfort is an important issue to persuade more
people to use the bike, especially the elderly population.

© 2012 Published by Elsevier Ltd. Open access under CC BY-NC-ND license.

Keywords: Cycling comfort; rolling resistance; cycling pathways

* Corresponding author. Tel.: +49 089 289 10351; fax: +49 089 289 15389.
E-mail address: hoelzel@lfe.mw.tum.de.

1877-7058 © 2012 Published by Elsevier Ltd. Open access under CC BY-NC-ND license.
doi:10.1016/j.proeng.2012.04.082
480 Christin Hölzel et al. / Procedia Engineering 34 (2012) 479 – 484

1. Introduction

Due to higher gasoline costs and the increased awareness for the environmental protection cycling gets
more attractive to people, which is a great development because more people using bicycles come along
with less people using cars and therefore less traffic within the cities. In relation to this the infrastructure
of pathways gets more important especially to enhance the cyclists comfort and safety. Cycling comfort
requires smooth rolling with low energy inputs as well as a good ride quality as a synonym for few
transmitted vibrations.
The rolling resistance (Rr) is one of the main sources of energy dissipations for bicycles. Cycling
comfort increases with lower rolling resistances because lower resistances stand for lower energy inputs.
Several methods have been used to determine the Rr in former studies. Van der Plaas used the coast-down
method where a bicycle rolls down a constant grade slope [1]. The required distance to stop the bicycles
movement is measured and the coefficient of rolling resistance (cr) can be directly calculated out of the
results by assuming that it is the predominant force during the test [1]. This method of examining the cr
requires smooth and absolutely even surfaces and is therefore not suitable for field tests. Three other
methods, large diameter rotating drums, a tricycle configuration and a rotating oscillating weight, are
known and were commented by Wang et al. [2] in 2004. He chose a method proposed from Hill in 1990
[3] first. Hill measured the Rr using an oscillating eccentric weight. Therefore two bicycle wheels are
rigidly fixed to an eccentric weight. Due to the weight the system is oscillating for and backward with a
given initial angle. Based on the advantages of the system a similar appliance was used in this study.
Ride quality is an often used term to quantify the quality of a bicycle due to its frame geometry,
especially its stiffness and damping. Published research has focused on material testing of road bicycle
frames and on examining the vibration response of different bicycles without a rider [4]. There is less
literature to find about transmitted road vibrations in cycling. Hastings et al. 2004 measured the effect of
transmitted road vibration on the cycling performance [5]. They used a specially fabricated clamping
system where the accelerometer was attached to and made standardized tests on a treadmill.
The aim of the current study was not to define the properties of different bicycle frames, or to
investigate the effect of transmitted vibrations on the performance but to examine the transmitted
vibrations on different road surfaces. Therefore a racing bicycle was equipped with a vertical
accelerometer under the seat.

2. Methods

Rr was quantified using a one degree of freedom pendulum which was developed for this study (Figure
1 a). Therefore two bicycle racing wheels are rigidly connected by an axis.

V-velocity
M-rotating A-acceleration
Fig.1. (a) one degree pendulum; (b) racing bike with accelerometer and data acquisition hardware
 Christin Hölzel et al. / Procedia Engineering 34 (2012) 479 – 484 481

An eccentric mass, consisting of a steel-plate, is welded on the axis. The sensor signals from the
acceleration sensor, which is fixed at the end of the plate, are amplified by data acquisition hardware
(National Instruments). The hardware, including analog digital conversion, was connected to a mini-pc
confirmed at the steel-plate. A data logging software records the sensor signals with a sampling rate of
500Hz. The pendulum had a total weight of 27 kg which seemed to be a good compromise between
transportability and inertia. The pressure of the wheels was set to 8, 5 bar. For all measurements the
pendulum was rotated to an initial angle of 86, 2 degrees and released. It oscillated back and forth until it
came to a complete stop. The whole procedure was performed for at least six times on one surface. For
data mining the recorded voltage signals are transferred to acceleration over time signals. The
acceleration curve is then classified by maxima und minima. By converting the acceleration data into
orientation angles of the pendulum the covered distance is calculated. The last data mining step is
examining the cr with the loss of energy. At the Beginning of the movement the pendulum has a potential
energy (Epot) which decreases due to the Rr until the movement stops.

‫ ݐ݋݌ܧ‬ൌ ݉ ‫݄ כ ݃ כ‬ (1)
‫ݐ݋݌ܧ‬ ൌ ‫ ݈݈݋ݎܧ‬ൌ ܿ‫ݏ כ ݃ כ ݉ כ ݎ‬ (2)
‫ݐ݋݌ܧ‬
ܿ‫ ݎ‬ൌ (3)
݉‫ݏכ݃כ‬

To examine riding quality a racing bike was prepared with an accelerometer and the data acquisition
hardware from the pendulum (see Fig. 1). The sensor was attached under the seat whereas the hardware
was fixed within the frame. Sampling rate was configured at 500 Hz and wheel pressure was set to 8, 5
bar. One rider (1, 90 m, 90 kg) rolled over the testing track of 15 m, without pedaling, while he kept his
velocity (10 km/h, 15 km/h, 20km/h) constant. Therefore the surface had to be as even as possible to deny
influences like acceleration or delay. At least three trials per velocity were measured. For data mining the
voltage signals are transferred to acceleration signals by scaling and calibrating. Furthermore the analysis
of the signals followed the VDI-norm 2057 to characterize the vibrations influence on the human body.
Therefore the frequencies of the acceleration signals are valuated with a weighting curve ܹ௄ which
considers the influences of vertical oscillations on a sitting person. The important parameter to interpret
the results is the effective value factor (awT) of the frequency valuated acceleration (aw(t)) which could be
calculated out of the frequency valuated signals. This parameter allows conclusions about the influence of
the oscillations on the human being in a seating position (Table 1).

ଵ
ଶ ‫ݐ‬ሺ݀‫ݐ‬ሻ்
ܽ௪் ൌ ට ‫׬‬଴ ܽ௪ (4)
்

Table 1. Effective values of the frequency valuated acceleration

effective value awT of the frequency

description of the perception
valuated acceleration aw(t)
<0,01 not perceptible

0,015 perception threshold

just perceptible
0,02
good perceptible
0,08
strong perceptible
0,35
very strong perceptible
<0,35
482 Christin Hölzel et al. / Procedia Engineering 34 (2012) 479 – 484

Data collecction for the pendulum

p as well
w as for the racing
r bike toook place on foour different suurfaces
(Fig. 2): asphhalt, concrete slabs, cobblestones and self-binding graavel. Influencees of weatheriing are
considered byy determining less
l used and battered
b surfaces of asphalt aand concrete sllabs. To capturre local
influences surrfaces were meeasured at two different placees. The points w
where the meaasurements tookk place
were the samee for the penduulum and for thhe racing bike. Temperaturess and wind keppt constant due to one
testing day.

Fig.2. (a1) new asphalt

a surface (a2) battered asphalt surface; (b1) new
w concrete slabs (b22) battered concreete slabs; (c) cobbllestones;
(d) self-binding gravel
g

3. Results

The followwing table givees an overview w about the ressults of all meeasurements, inncluding the ccovered
distances, thee rolling resisstances and th he effective values
v with thhe descriptionss of the perception.
Regarding to the mean valu ues the penduluum covered thee longest distaance on concreete slabs, follow wed by
asphalt, cobb blestones and self-binding gravel. The detailed d distannces could bee seen in Figgure 3.
Comparing th hese results to table
t 2 shows good agreement with the staatement that thhe covered distance is
directly relateed to the cr beccause it could be
b seen that sm mall rolling ressistances are eequal to long ccovered
distances (Co ompare Table 2: 2 for examplee self-binding gravel got thee lowest distannce of 10,2m aand the
highest cr of 0,013225).
0
Considerin ng the effectivee values the ressults are differrent (Fig. 4). H
High effective vvalues describee many
and strong vib brations. Asphhalt has the low
west effective value
v of 0,05 aand therefore jjust good percceptible
vibrations. Assphalt is followwed by concrete slabs with an a awT of 0,17,, self-binding ggravel with ann awT of
0,21 and cobb blestones with the
t highest vallue of 0,57.

Table 2. General Results

surfaces rolling resistance riding qualitty

covered distance rolling resistan

nce effective vallue
subjectivve perception
[m] coefficient (cr) (awT)

asphalt (less used) 19,8 0,0071804 0,05416666 7 good perrceptible

concrete slabbs (less used) 22,1 0,0063459 0,16605882 4 strong peerceptible
cobblestoness 13,8 0,0103079 0,57 very stroong perceptible
asphalt (batttered) 15,6 0,0089477 0,10277777 8 strong peerceptible
concrete slabbs (battered) 21 0,0066998 0,18611111 1 strong peerceptible
self-binding gravel 10,2 0,013225 0,21277777 8 strong peerceptible
 Christin Hölzel et al. / Procedia Engineering 34 (2012) 479 – 484 483

Covereddistancesondifferentsurfaces
25
covereddistance[m]

20

15

10 22.1 21 19.8 lessusedsurfaces

15.6 13.8
5 10.2 batteredsurfaces
0
concreteslabs asphalt cobblestones selfͲbindinggravel

Fig. 3. Covered distances of the pendulum on different surfaces

Comparisonoftheeffectivevalues
0.6

0.5
effectivevalues

0.4

0.3 0.57
0.2 lessusedsurfaces

0.1 0.17 0.19 0.21 batteredsurfaces

0.10
0 0.05
asphalt concreteslabs selfͲbindinggravel cobblestones

Fig. 4. Effective values, investigated with the racing bike, on the different surfaces

Additionally to the less used surfaces, the parameters of battered asphalt and concrete slabs are
analyzed. Concrete slabs show less degeneration (-5 %) than asphalt (-21%) regarding to the covered
distance (Table 3). Considering the effective values the results are different again. Although the effective
value of battered asphalt increases about 90% whereas the awT of concrete slabs increases about 12%,
asphalt shows lower absolute values than concrete slabs in any case (Table 3).

Table 3. Changes between the covered distance and the effective value of less used and battered asphalt and concrete pavement

surfaces covered distance [m] effective value awT

asphalt (less used) 19,8 0,054
change Ļ -21% Ļ 90%
asphalt (battered) 15,6 0,103
concrete slabs (less used) 22,1 0,166
change Ļ -5% Ļ 12%
concrete slabs (battered) 21 0,186
484 Christin Hölzel et al. / Procedia Engineering 34 (2012) 479 – 484

The analysis of the effective values regarding different velocities showed a linear dependency of the
awT with increasing velocities. The awT for the velocities 10, 15 and 20 km/h were measured. The effective
values up to 30 km/h are calculated.
effectivevaluesofthedifferentsurfacesindependancyofthevelocity
1.200
cobblestones
1.000
selfͲbindinggravel
effectivevalue

0.800
concreteslabs
0.600
(battered)
0.400 concreteslabs(less
used)
0.200 asphalt(battered)
0.000 asphalt(lessused)
9 14 19 24 29
velocities[km/h]
Fig. 5. Effective values of different road surfaces regarding different velocities

4. Discussion

Contemplating the comparison of concrete slabs and asphalt, shows greater effective values for
concrete slabs whereas asphalt has a higher rolling resistance. Both of the other surfaces are far behind,
regarding the awT as well as the cr, and therefore inadvisable for cycling pathways with high comfort.
Regarding the weathering concrete slabs perform much better than asphalt in both parameters.
Nevertheless cycling on asphalt is much higher comfort, especially because of the fact, that the awT of less
used concrete slabs (0,17) is still much higher than the awT of battered asphalt (0,10). Finally that means
that despite its worse performance (+ 90%), asphalt is much better than concrete slabs overall.
Another interesting conclusion is that the effective values and therefore the riding quality are linear
dependent from increasing velocities. For asphalt for example the awT increases from 0,035 (10km/h) over
0,05 (15km/h) to 0,09 (25km/h). Due to this perception is changing from good perceptible to strong
perceptible while the comfort decreases. Finally it could be said that there are two good opportunities for
cycling pathway surfaces. With Regards to the comfort asphalt is the most useful surface. Nevertheless
concrete slabs are a good alternative due to their easier repairing properties. Summarizing, this study
seems to be a good base to decide about cycling pathway design considering different demands.

References

[1] Van der Plaas R. Rolling Resistance of bicycle tires. Bike Tech 1983; 2:8-12.
[2] Wang E, Macedo V, Reid J. A method for quantifying rolling resistance of bycicle tires. In: Hubbard M, Mehta RD, Pallis JM,
editors. The engineering of sport 5, Vol. 2. Sheffield: International Sports Engineering association; 2004, p. 132-138.
[3] Hill B. Measurement of rolling resistance using an eccentrically weighted oscillating wheel. First international symposium on
surface chracteristics, State College, PA; 1990.
[4] Thibault J, Champoux Y. Rider influence on modal properties of bicycle frames. Canadian Acoustics/Acoustique Canadienne
2000; 28(3): 44-45.
[5] Hastings A Z, Blair K B, Culligan F K, Pober D M. Measuring the effect of transmitted road vibration on cycling performance.
In: Hubbard M, Mehta RD, Pallis JM, editors. The engineering of sport 5, Vol. 2. Sheffield: International Sports Engineering
association; 2004, p. 619-625.