Bicycle weight and commuting time: randomised trial
BMJ 2010; 341 doi: https://doi.org/10.1136/bmj.c6801 (Published 09 December 2010) Cite this as: BMJ 2010;341:c6801All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
Considering Parts I & II, what then are the most effective ways
for Dr Groves to reduce his commuting times, for the same power output?
For constant effort on dry paved roads and without braking, the two main
energy sinks are rolling resistance and drag. At lower speeds rolling
resistance dominates, while at higher speeds drag does. At an average
speed of 15 mph over hill and dale, both are important factors. The
following table summarizes the effects of some basic variations in the
corresponding parameters. Each row of the table shows the effect of one
change, all other parameters remaining at baseline.
Baseline parameter values: bicycle = steel, total rider power output
= 130w, rider weight = 76 kg, Crr = 0.007, frontal area = FA = 0.5 m^2,
drag coefficient = Cd = 0.9, road grade (average) = -0.13% out, +0.13%
back, headwind velocity = 0, distance = 21.75 km each way, temperature =15
deg C, altitude = 400 ft, transmission efficiency = 95%.
Parameter Total Time TimeSaved AvgSpeed
(hr:min:sec) (mph)
Baseline 1:48:02.4 _________ 15.02
CRP bike 1:47:21.0 0:00:41.4 15.11
Crr=0.006 1:45:52.2 0:02:10.2 15.33
Crr=0.005 1:43:46.2 0:04:16.2 15.64
FA=0.4 1:41:25.8 0:06:36.6 16.00
Cd=0.8 1:44:28.8 0:03:33.6 15.53
Thus the most effective of these changes does not require any
purchase but only to ride in a more aerodynamic position. Next would be to
consider the rolling resistance tests mentioned above and choose a less
hypertrophied tire, if necessary keeping a better lookout for road
hazards. These two moves alone could easily allow him to sleep in more
than five minutes, and moreover get home another five minutes early for a
nap before supper.
For valuable information on many of these topics and many more,
including discussions of specific tire models, the reader is advised to
examine the well-worn archives and FAQ of the Usenet newsgroup
rec.bicycles.tech, such as may be found at
<http://draco.nac.uci.edu/rbfaq/FAQ/index.html> or
<http://www.sheldonbrown.com/brandt/>.
Competing interests: No competing interests
To continue the discussion (and the section numbering) from Part I:
(3) Examining the photograph of the two bicycles, it is evident that
relative to saddle height, the handlebar height on the CRP bicycle is
lower than on the steel bicycle. Dr Groves has kindly confirmed this and
by his measurements the tops are roughly 8.5 cm lower (the differing bar
shapes reduce the difference at the drops). No doubt this is one of the
factors making the CRP bicycle uncomfortable. For this sacrifice, such an
adjustment should make the body position more aerodynamic. The wheels on
the CRP bicycle are also more aerodynamic, by virtue of having many fewer
spokes (they are therefore also much less reliable, and effectively not
user-repairable), as well as thinner tires. The steel bicycle has full
front and rear mudguards while the CRP has only a vestigial rear one,
likely again to the advantage of the CRP bicycle. The shapes, sizes, and
configurations of the bicycle frame's tubes, and of its components, also
affect the drag force; see
<www.sheldonbrown.com/rinard/aero/aerodynamics.htm>.
4. Examining the photograph further, it also appears that the bottom
bracket (crankset axle & bearing assembly) of the steel bicycle is
lower than that of the CRP bicycle (Dr Groves confirms it is 2 cm lower).
As has been known since at least 1939 [1], the geometry of a lower bottom
bracket, as of that of a longer wheelbase, smooths the geometry of the
rider's travel over a bump, making the ride both more energy efficient and
more comfortable. Increased comfort on a ride, in basic position and also
in response to bumps, helps the rider to maintain effort over long
distances.
[1] Davison AC. Long or short wheelbase? Cycling, May 17 1939: 699-
700. <http://www.classicrendezvous.com/Events/Long-or-short-
wheelbase_1939.pdf>
5. Rolling resistance is not the same as friction. A rolling bicycle
wheel does have some friction, but precisely because it is rolling, not
sliding, this friction is minuscule even for the rear wheel, with
effectively the entirety of rolling resistance being due to hysteretic
losses in the sidewalls and tread, as the contact-induced deformation
makes its way round and round the tire. This means that the rolling
resistance is greatly affected by the thickness and viscoelasticity of not
just the sidewalls but also the tread, as well as any reinforcing
materials.
Dr Groves assumes that tires of the two models used, Schwalbe
Marathon 700x32mm on the steel and Schwalbe Marathon Plus 700x25mm on the
CRP, have identical rolling resistances and further that their common
coefficient of rolling resistance (Crr) is 0.0045, giving a power loss of
26 watts on the steel and 24.8 watts on the CRP. This is incorrect on all
counts:
(1) The inflation pressure was not mentioned. This can have a
dramatic effect on rolling resistance, although at higher pressures the
effect levels off. At the same pressures, all else being equal, wider
tires have less rolling resistance but more drag. However, it is not
normal to run a 32mm tire at the same pressure as a 25mm tire. For Dr
Groves' weight, about 85-90 psi would be a good starting point for a 25,
and about 55-60 psi for a 32. See
<http://www.precisiontandems.com/photos_files/tirechart.jpg>,
<http://www.bikequarterly.com/images/TireDrop.pdf>. As a
complication, the actual widths of many tires can be several millimetres
more or less than their nominal widths. Even for the same model, some
manufacturers use thicker cords or rubber for tires of the wider widths;
others may use thinner rubber, while still others keep both the same
across a wide range of sizes.
(2) The tires used by Dr Groves are amongst the heaviest, most
durable, and most puncture resistant available, of the kind one might
consider for cycling in Waziristan or Mogadishu. Most published rolling
resistance tests are of much faster tires, and these have shown that even
amongst expensive racing and training models of the same sizes and
occupying overlapping market niches, Crr values on polished steel drums or
plastic rollers- which are much lower, and have much less scatter, than
those found for pavement- can range from less than 0.003 to almost 0.007.
See:
<http://www.rouesartisanales.com/article-1503651.html>,
<http://www.terrymorse.com/bike/rolres.html>,
<http://biketechreview.com/tires/rolling-resistance/475-roller-
data>.
It seems unreasonable to assume that a Schwalbe Marathon Plus, a
heavy-duty tire with, according to Schwalbe, 1 cm (!) of rubber between
the tube and the road, weighing 590 grammes in 700x25mm, should have a Crr
of 0.0045 on English roads, when the Schwalbe Stelvio, a racing tire
weighing 223 g in 700x22.5mm, tests out to have a Crr of 0.0059 on a
polished steel drum. Further, despite their similar names, the two
Marathon models are of radically different internal construction, and also
of different sizes, both of which affect the Crr. While the tires on the
steel bike are older, and increased wear generally reduces rolling
resistance by thinning the tread, Dr Groves has advised that the Marathons
had only 400 miles more than the Marathon Pluses, likely not a significant
difference.
Dr Groves relied on Schwalbe literature- presumably the chart found
at <http://smtp.schwalbetires.com/tech_info/rolling_resistance>- to
conclude that both tires have the same Crr, but this is only a coarse
classification and can't be relied upon. However, in my calculations, for
the sake of comparing the effect of cycle weights alone I also disregarded
any differences and likewise kept the two Crr's constant. I did though use
a more reasonable baseline value of 0.007- still likely an underestimate,
if only given the road conditions. Increased rolling resistance, as with
reduced drag, makes the effect of weight more pronounced, but in either
case the effect is slight.
This does emphasize another crucial difference between our two
methods: with calculation, we can do a sensitivity analysis, such that
disregarding the real difference in a confounding factor betters the
accuracy and reliability of the fundamental conclusion; while in the
experiment, such disregard worsens them.
6. We are told that "lighter rims can confer a significant advantage,
but only if there are a significant number of points of speed change on
the journey." This would be correct if by "significant advantage" it were
meant the tiny number of tenths of a second that could conceivably mean
the difference between winning and losing a 40 km criterium race, the type
of race with the most and hardest speed changes. In reality even at full
throttle the maximum acceleration of a bicycle is so relatively minuscule
that there is no difference in effect between weight on the frame or the
rims. In fact lighter-rimmed wheels are often slower, because extra weight
often goes to a more aerodynamic shape. See
<http://biketechreview.com/reviews/wheels/63-wheel-performance>.
In Part III calculations are made to show how the main confounding
factors may be exploited to reduce commuting times.
Competing interests: No competing interests
The bicycling community includes many medical doctors, and some have
wondered whether their expensive carbon-reinforced plastic (CRP) framed
bicycles really give them any speed advantage. In 2010 one of them, Dr
Groves, became curious enough to use the methods of his profession to
investigate the matter, and we should be grateful to both Dr Groves and
the BMJ for the provocative and useful article that resulted. After six
months, 56 journeys, and 1500+ miles of riding over hill and dale in
English weather, on a 27 mile out and back commute, he found his average
speed to be 15.03 mph using his 29.75 lb, GBP50 steel-framed bicycle, and
14.95 mph using his 20.9 lb, GBP1000 CRP-framed bicycle. The average round
trip time advantage during the winter was 2 min 35 sec in favour of the
steel bicycle, while in the summer it was 1 min 04 sec in favour of the
CRP bicycle. The overall average difference was 32 seconds in favour of
the steel.
The bicycling community also includes many mathematicians,
physicists, and engineers, and most if not all of them have given some
thought to the effects of lowered bicycle weight on speed. Perhaps since
the dawn of the bicycle age, some of them became curious enough to use the
methods of their profession to investigate the matter in more or less
detail. In recent decades the state of the art made the endeavour
practical enough that a number of them, after a few evenings at home at
their desks writing computer programs, made quite thorough solutions
available to all via the internet, for arbitrary pairings of bicycling
circumstances. A few examples are:
<http://bikecalculator.com/veloMetricNum.html>,
<http://www.analyticcycling.com>,
<http://www.whitemountainwheels.com/SpeedPower.html>,
<http://sportech.online.fr/sptc_idx.php?pge=spen_esy.html>,
<http://www.grennan.com/BikePower/> [source code in C]
Using for example the calculating engine at bikecalculator.com with
the relevant data contained in Dr Groves' article, and reasonable values
for the parameters, I found an overall average round trip time advantage
of 41.4 seconds in favour of the CRP bicycle. This is much the same as the
1:04 advantage for the summer months found by experimentation, an
agreement that may owe as much to the consistency of Dr Groves' cycling
efforts as it does to the heritage bequeathed to us by Galileo, Newton,
Euler, and their many successors.
To be sure, this value is an approximation to the actual one for Dr
Groves' circumstances, because I didn't bother to use the detailed
topography, but only the average slope in each direction; because the
other parameter values are also approximations; and because I ignored
traffic considerations. Nevertheless, for the given parameter values and
unlike for the methods of medical research, it has the benefit of
unambiguously demonstrating the effect of weight alone, all other things
being equal. Or for that matter the effect of any other parameter alone or
in combination, all else being equal.
The differences between our methods can be summarized thus: in
science we rely on, and therefore demand, thorough knowledge of the
situation; while in medicine the hope is that a randomization procedure
will cover up for ignorance.
The latter works passably well for random errors but not for
systematic ones. Just because systematic errors (confounding factors) are
missed does not mean they are not present. This is why most of
epidemiology as it is practised today- perhaps as it ever will be-
resembles an elaborate hoax. Dr Groves does far better than most
epidemiologists by making a serious effort to account for the confounding
factors. I take the opportunity here to call attention to a few more, and
at the same time to correct a few errors in his presentation. I continue
this in a second part, and close in Part III with a table showing the
effects of various ways of exploiting the confounders to lower his
commuting times.
1. We are told that in traversing a hill, the extra work done uphill
on the heavier bicycle is recovered as extra kinetic energy on the
downhill, because energy is conserved. In fact the bicycle is a
dissipative system, not a conservative one. Energy is lost due to rolling
resistance, drag, and braking; and even to other sinks such as jiggling of
the viscoelastic rider and his luggage. All other things being equal, on
the uphill a lighter bicycle should essentially always be faster, while on
the downhill a heavier bicycle may be either faster or slower, depending
on the balance between rolling resistance and drag, as well as such things
as the need to brake on sharp curves.
2. Drag, a force, is proportional to the square of the velocity, not
the cube. The power lost to drag is proportional to the cube.
3. We are told that the power required to overcome drag is
independent of which of the two bicycles is ridden, and that for either
one at 15 mph, it is 170 watts. This is incorrect on all counts.
(1) Using the parameter values listed in the article, the power
formula 0.5 x 1.2 x 6.7^3 x 1 x 1 gives 180 watts, not 170.
(2) Dr Groves gets this formula from the standard drag-force
equation, using a value of 1 for his drag coefficient (Cd), and a value of
1 m^2 for his frontal area (A), their product (CdA) being 1 m^2. He might
consider that a frontal area of 1 m^2 would be that of a person of height
2 metres and constant width 0.5 m from top of head to soles of shoes
(rather tall and strangely thick), standing fully erect. This seems
unreasonable for a thin crouched-over cyclist, and indeed the CdA values
typically reported for bicyclists are in the range of under 0.3 m^2
(racing time trialists) to about 0.8 m^2 (fully upright on a city bike;
see <http://www.sheldonbrown.com/rinard/aero/formulas.htm>). For my
baseline calculations, considering the drop-bars on the two bicycles, I
used values of Cd = 0.9, A = 0.5, for a lower CdA of 0.45 (m^2). This is a
conservative assumption as reducing drag makes the effect of weight more
pronounced. The value of 1 used by Dr Groves was from a citation for a
"touring bike"; this may mean one with handlebar bag, panniers, and so on.
The bicycles shown in the photograph accompanying the article are better
described as traditional audax (the steel) or racing (the CRP).
A robust method for estimating CdA using a power meter is presented
in <http://anonymous.coward.free.fr/wattage/cda/indirect-cda.pdf>,
while nowadays one can ride with a bicycle computer that calculates it on
the fly: <http://www.ibikesports.com/products_iaero.html>.
More on aerodynamics and other matters in the second part.
Competing interests: No competing interests
Scoolboy error is actually calling the trial randomised. If you know
which bike you are on the placebo and nocebo effects will dominate any
actual variation.
If you repeat as double blind trial results will still be garbage because
speed will be limited by the guide dog
Competing interests: Triathlete
My own experience of commuting a shorter distance across Sheffield is
entirely compatible with Dr Groves findings, with the optimum journey
times being achieved on a mid range road bike; my 3 speed Brompton is
hopeless up the hills and the 2.5 inch tyres on my carbon mountain bike
have too much rolling resistance.
However, whilst Dr Groves clearly identifies a number of the factors
which may effect the length of commute and the amount of effort required
to achieve this his conclusion regarding the usefulness or otherwise of a
carbon frame is limited to the effect on journey time.
Of equal importance, based on the increased mass of the steel frame
together with the likelihood of it being fitted with lower quality
components, achieving shorter journey times presumably requires more
energy and will therefore deliver greater health benefits.
Perhaps this line of reasoning should be employed when trying to
explain to other family members why they really should be perfectly happy
with a much less expensive bike than mine?
Competing interests: Currently seeking spousal approval for (another) completely unjustifiable carbon bike
This article was an interesting piece, even if written in a tongue
and cheek objective. Yet bike commuting is a serious matter for public
health and transportation planning in nearly every city, as BMJ has
published on this matter previously. So I'm not surprised if this study
will garner many polarized opinions and technical comments. Perhaps that
was the real objective?
At any rate, I must agree with many of the commentaters that there
are numerous holes in the study:
When riding a new carbon bike, one naturally protects one's interests
by riding more cautiously. Carbon fails; steel bends. Hence unless the
writer is finacially indiferrent, there's a bias inherent in the type of
bike ridden.
Using the median is generally preferrable when speaking about time
periods. For instance, in the analysis of waiting lists, medians are
reported to remove the skew effect of outliers.
Were "track stands" performed? A full "dab" stopping technique is
usually not performed on a high end bikes.
"Non-clips" along with running shoes? Efficiency of high end bikes
is greatly increased by pedalling technique. A pounding style, further
marred by the hyper-extension of the achilles due to the flexible shoes,
is poor form on a high end bike. The full package should, perhaps, be
evaluated: tweed- commuting versus racer-commuting.
Tester's physical ability was not standardized. On a track, in
controlled conditions, was the tester faster or more effecient on the
carbon bike? There is a learning curve associated with high-end, twitchy,
bikes. Likely the tester went along this curve as the season progressed,
with lower times resulting.
Stopping power was not analyzed, nor the number and type of stops
tracked. Commuters must stop instantaneously to avoid death. But the
bike setup, braking power, attention, skill and many other factors affect
this. Random and regular stops were not included.
Winter? Lycra-clad cyclists have a "wet" bike. And bikes accumulate
water in the frame and in the tires, changing the weight and rolling
weight. This would create a bias if the same bike is ridden the next
day.
Will the tester, or reader, be persuaded to commute again the next
day based on the bike chosen? Since the carbon bike requires a more
aggressive cycling position, how likely will a L5-S1 bulge or herniation
occur over time, thereby curtailing one bike-to-work career?
Speed is an immaterial, if not dangerous, goal during an urban
commute. It's not about how fast one gets to and from work; rather, it's
about one's physical and psychological state during and after the voyage.
A study on the effect of route choices available to commuters according to
the vehicle chosen would be more informative. For example, chosing grid-
like car routes on a carbon bike versus watershed trails on a mountain
bike. Educating the reading public away from car-centric mentality for
route choice may lead to more satisfied, and safer, bike commuters.
J.
Competing interests: Bike commuter for 15 years; Owner of 4 types of commuter bikes, excluding tandem; 1 - 4 hour daily commute; member of a old-timers bike racing club (FOG); back injury attributed to cycling.
So Dr. Groves has figured out that a high-end bike confers no
significant performance advantage in a test of low-performance demand. Is
anyone surprised?
First of all, what appears "insignificant" on the clock might appear
otherwise when analyzed spatially. Groves reports that he rode at an
average velocity of 15 mph, or 22 feet per second. If we were to have Dr.
Groves race himself during his 27-mile commute on both bikes, record the
performances on digital video and superimpose them (as they do in
telecasts of the Olympics), we would see the carbon-framed Groves cross
the finish line 704 feet ahead of his steel-framed self. In the world of
competitive cycling, where the top ten finishers are often grouped within
a 3 or 4 meter span, 704 feet would be considered a huge performance
advantage.
But this would be better analyzed in a high-performance context. If
Dr. Groves absolutely positively had to make it to work in under 1:30, and
would be fined by his employer for each second he was over, the advantage
of the higher-performance machine would very soon become apparent.
You don't need a Pinarello to pedal to work at a leisurely pace any
more than you need a Ferrari to drive to work at city speed limits. But
in a situation of high-performance demand, the advantages of high-
performance equipment WILL emerge.
Competing interests: No competing interests
As both a scientist and a recent convert to the MAMILs I am delighted
that the BMJ has finally recognised the importance of the study of
bicycles to international health. However, I agree with several other
commentators that the article by Dr. Groves is seriously flawed and should
not have been published. First reports in a new field tend to be the most
cited regardless of subsequent evidence from better-designed studies which
contradict them. In this regard, I find it appalling that BMJ has chosen
to allow open access to this particular piece of work. Now, rather than
being confined to a relatively small number of technical experts with the
knowledge, skills and appropriate training to evaluate the weaknesses of
the claims being made, the work is readily accessible to a general public
seriously lacking these attributes. Unfortunately, among those to whom
the author's conclusions are available are spouses, partners and family
members of cyclists. The majority of these people are a) greatly
influenced by pseudo-scientific claims made by both print and electronic
media, and b) extremely reluctant - and in my case, absolutely refuse - to
listen to an alternative viewpoint no matter how well-reasoned and
scientifically sound. Will the editors accept responsibility for the
thousands of us throughout the world who will be told "You don't need
another bike, there's an article in BMJ that proves it"?
I find it ironic that the author should compare the uptake of new cycling
technology to the uptake of new pharmaceuticals when the major error he
commits is identical to one so frequently encountered in drug trails:
namely the measurement of surrogate outcomes. Frankly, who cares whether
you save four minutes riding from Sheffield to Chesterfield? The more
important questions are how does the rider feel about the journey on one
type of bicycle versus the other; what impact does the bicycle have on the
rider's self-esteem and feeling of worth; how are the rider's interactions
with colleagues, friends and family impacted by the type of machine he/she
rides; etc?
Until researchers shift their focus from the purely mechanistic to the
psycho-social aspects of cycling many MAMILs will be condemned to riding
junk.
Competing interests: Only owns one bike (at the moment).
How I love the 'n of 1' trial where I can be both subject and
investigator. Congratulations, Dr Groves, on your motivation to faithfully
conduct and write up this study. The rapid responses clearly indicate that
similar research questions have been asked and trials performed in the
setting of cycle commutes the world over, but the rest of us have lost the
race to publication.
I, too, recently purchased a 1000 pound (sterling) carbon frame
bicycle on a ride to work scheme after previously commuting on a steel
frame bike that is about 4kg heavier. My round trip commute is 24 miles. I
find that there's no appreciable difference in the time that it takes me
to commute BUT my knees feel less sore whilst I am riding the lighter bike
and I've noticed that my quadriceps seem less bulky (observed, not
measured) since I began riding it. I found the new bike far less
comfortable than the old at first but after my husband, a bike 'anorak',
made some adjustments that involved indulging in his favourite hobby -
buying bike parts on eBay - the ride became enjoyable again.
Said husband owns a bike that, like the human body, never consists of
the same components from one month to the next. He trades bike parts on
eBay convinced that finding the perfect mix of parts will lead to the
creation of the perfect bike. Husband does not commute on his creation, in
fact his bike leaves the house 6 or fewer times a year, most often to be
ridden to a local bike shop, but he is very proud of its lightness. James
Ward wonders about the effect of this study on marital harmony. I can say
that in a study of 1 married couple this wife felt great righteous
satisfaction on showing the study to her husband who labours under the
(false and untested) belief that the lighter a bike is the better it is.
However, as her husband's belief in his rightness was curiously unaffected
by the findings of the study, and positions remain entrenched, the overall
effect on marital stability was negligible.
Competing interests: owner of several bicycles and frequent cycle commuter; not involved in the selection of this paper for publication
Re: Bicycle weight and commuting time: randomised trial
Dr.Groves' elegant study has clearly stimulated a lot of interesting
discussion here. Others have already pointed out the limited
generalizability of his results. I wish to point out a few common sense
(without any level I evidence) with regard to the choice of bicycle for
commuting:
1. On snowy and icy days, a full carbon bike with skinny tires just
won't get you to work unless you intend to carry it all the way.
2. A bicycle commuter will never take the elevator! Carrying the extra 4kg
weight to climb a number of flights of stairs may sway you to take the
carbon bike.
3. The steel frame bike with a set of 28C tires may just be the right bike
to ride through the trail instead of the open road saving time and
possibly your head.
4. Nearly 2/3 of all commutes are less than 3km. Whatever time gain on the
correct choice of bike is negligible when considering other important
factors such as theft proofing your bike.
5. Aesthetic is important, even commuting to work. I don't know anyone who
would put a luggage rack on a full carbon bike, nor anyone who wears the
full spandex outfit while riding an old steel frame bike - that's just not
done in any civilized society!
6. If you can only have one bike - get a full carbon! Life isn't about
commuting back and forth to work. If you set a higher goal in life, such
as to do an Ironman, you'll need the carbon bike!
Competing interests: I use a steel mountain bike only for snowy winters, a Scandium cyclo-cross bike for commuting all other times, and a full carbon bike for triathlon training and racing. Still happily married!