tire pressure does not significantly affect rolling resistance.
wle
---------
1. Weekly Dispatch o^o o^o o^o o^o o^o o^o
We heard the buzz about a surprising new tire test in Bicycle
Quarterly, a nifty magazine published in Seattle and edited by Jan
Heine. The name was Vintage Bicycle Quarterly until recently, but
"vintage" has been axed because it implied the mag was about old,
collectible bikes and equipment.
Well, plenty of pages are devoted to arcane and interesting gear,
history and randonneur-style riding, but Bicycle Quarterly also
publishes cutting-edge material. The tire test is testimony, appearing
in the Autumn 2006 issue.
Heine gave RBR permission to summarize several major findings.
Interestingly, they confirm lots of what Uncle Al has been ranting
about for years regarding tire width and inflation pressure.
Some test conclusions will be particularly enlightening if you're
riding on narrow, high-pressure clinchers seeking more speed via lower
rolling resistance. Your skinny tires may not be as fast as you think.
For the full eight-page report on tire performance, order the Vol. 5
No. 1 issue from http://www.bicyclequarterly.com. The test included
nine 700C tires, seven 650B tires and two tubulars. The protocol and
results were reviewed by industry experts. These are eight findings:
---With roughly the same power output, the rider's speed can vary by as
much as 20% depending on tire choice. For example, the rider on the
fastest tire [in this roll-down test] moved down the road at
approximately 16.4 mph (26.2 kph) while the same rider on the slowest
tire went approximately 13.6 mph (21.7 kph).
---Many longtime riders believe tires with a cotton casing are faster
than modern casings made from nylon. Testing seems to confirm this. The
best-performing tire in the test, the Deda Tre Giro d'Italia 700x23C
(actual width 24.5 mm), has a cotton casing.
---Tire pressure has only a small effect on the rolling resistance of
most tires. Narrow 23-mm tires seem to roll fastest at pressures of 105
psi (7.2 bar) or more. However, running these tires at 85 psi (5.8 bar)
for improved comfort increased the test times only 2%. Wider 28-mm
tires are as fast at 85 psi as they are at higher pressures.
---Tubular tires perform worse at very high pressure. At 130 psi (9
bar), the narrow Clement Criterium rolled slower than it did at a more
comfortable 105 psi. The wider Clement Campione del Mundo rolled
slightly faster at 85 psi than at 105 psi.
---Wide tires do not roll slower at lower pressures. In fact, testing
indicated that a wide tire at lower pressures rolls faster than a
narrow tire at high pressures, if all other factors remain the same.
Even narrow tires can be ridden at comfortable pressures with only very
small concessions to performance.
---Tires rolled slightly slower with Michelin's relatively thick latex
tubes than with butyl tubes. Thinner latex tubes, like used in tubular
tires, may offer better performance, but when used in clinchers they
are more prone to punctures caused by friction between tire and tube.
Latex tubes do improve comfort.
---Perhaps the most important result of the test is that tire pressure
does not significantly affect rolling resistance. Wide tires in
particular do not need high pressures to roll fast. But because many
current wide tires are designed to handle high pressure, they have
strong casings that lack suppleness. This results in higher rolling
resistance than necessary.
---The test's findings point to a new direction for performance
bicycles. For most cyclists, wide, supple tires at low pressures offer
more speed, better comfort, increased versatility and improved safety
than today's narrow high-pressure tires. However, this type of wide,
fast tire currently is not available. Hopefully, these test results
will help persuade manufacturers to produce them.
______________________________________
sally
@m7g2000cwm.googlegroups.com:
> ---Tire pressure has only a small effect on the rolling resistance of
> most tires. Narrow 23-mm tires seem to roll fastest at pressures of 105
> psi (7.2 bar) or more. However, running these tires at 85 psi (5.8 bar)
> for improved comfort increased the test times only 2%. Wider 28-mm
> tires are as fast at 85 psi as they are at higher pressures.
Your posting gave very little real data except for the above 2% number.
2% sounds like a huge amount to me. In a 100 mile time trial race, you will
be 2 miles behind due to air pressure alone. That is a difference of several
minutes, which could easily be the difference between first place and last
place in the race. Unless your math is off by a full decimal place
somewhere, I'm not going to reduce my pressure to 85psi any time soon.
Phil Holman
"sally" <sa...@sally.com> wrote in message
news:Xns9876A0...@192.160.13.20...
I'm sure the test times were the result of rolling resistance only so
the 2% increase is realistic. In reality, rolling resistance will only
account for a smaller percentage of the total drag force on a rider so
it's not going to be as big as the difference which you infer.
Phil H
Larry Dickman
"wle" <w...@mailinator.com> wrote:
> ---With roughly the same power output, the rider's speed can vary by as
> much as 20% depending on tire choice. For example, the rider on the
> fastest tire [in this roll-down test] moved down the road at
> approximately 16.4 mph (26.2 kph) while the same rider on the slowest
> tire went approximately 13.6 mph (21.7 kph).
A 20% difference in speed among similar sized tires? I don't believe it.
wle
no, they are saying it;s all rolling resistance that makes tires 'roll
faster'.
===
---Perhaps the most important result of the test is that tire pressure
does not significantly affect rolling resistance. Wide tires in
particular do not need high pressures to roll fast. But because many
current wide tires are designed to handle high pressure, they have
strong casings that lack suppleness. This results in higher rolling
resistance than necessary.
===
i guess what i don;t understand is this:
if the rolling resistance is the only thing that is different, and it;s
such a small percentage of overall force [they mention 16mph as a
ballpark speed] then how can one tire be 'faster' than another?
do these 'supple' tires magically lower air resistance somehow?
what gives?
the last paragraph...
====
---The test's findings point to a new direction for performance
bicycles. For most cyclists, wide, supple tires at low pressures offer
more speed, better comfort, increased versatility and improved safety
than today's narrow high-pressure tires. However, this type of wide,
fast tire currently is not available. Hopefully, these test results
will help persuade manufacturers to produce them.
====
...makes it sound like a thinly veiled commercial for a new magic tire
shortly to be foisted on the public
wle
Ningi
It doesn't say they are similar sized. I also note that the fastest
tire in the test is a 700x23. That's pretty skinny.
Pete
wle
>
> A 20% difference in speed among similar sized tires? I don't believe it.
ya
doesn;t make sense
plus they talk about a 'roll down test' but then also mention
'different riders with the same power'
doesn;t roll down mean just coasting?
since they say different riders, and coasting, surely they can;t be
comparing riders of different weights?
i tried to write to the magazine with some questions
wle.
jobst....@stanfordalumni.org
> ---------
I think the testing method and instrumentation is probably the culprit
in these erroneous findings that go contrary to most RR tests done
without involving bicycles or riders but only rolling tires.
Please review the chart at:
http://www.sheldonbrown.com/brandt/rolling-resistance-tubular.html
All these tires nearly approach zero RR at infinite pressure with a
constant offset for tubulars that have rim glue losses. You'll notice
that this is a family of curves of identical shape.
Jobst Brandt
Werehatrack
results were questionable.
--
Typoes are a feature, not a bug.
Some gardening required to reply via email.
Words processed in a facility that contains nuts.
Kenny
Werehatrack wrote:
> "Latex tubes do improve comfort"? Once I got to that one, I knew the
> results were questionable.
After having used Michelin latex tubes for about a year, I believe I
can feel a noticeable improvement in ride quality versus using butyl
tubes. Sounds strange I know. Can't quantify it but in my mind it
exists.
Kinky Cowboy
>
>I think the testing method and instrumentation is probably the culprit
>in these erroneous findings that go contrary to most RR tests done
>without involving bicycles or riders but only rolling tires.
>
>Please review the chart at:
>
>http://www.sheldonbrown.com/brandt/rolling-resistance-tubular.html
>
>All these tires nearly approach zero RR at infinite pressure with a
>constant offset for tubulars that have rim glue losses. You'll notice
>that this is a family of curves of identical shape.
>
With a smooth roller, yes. Add some real world surface features, which
are often sized at ~20% of the tyre section height even for fairly
smooth pavement, and your "infinite pressure" pneumatic tyres will
roll like iron tyred cart wheels. Bicycle tyre tests which don't
involve bicycles or riders are only slightly interesting, and almost
completely useless.
Kinky Cowboy*
*Batteries not included
May contain traces of nuts
Your milage may vary
Luns Tee
<jobst....@stanfordalumni.org> wrote:
>Please review the chart at:
>
>http://www.sheldonbrown.com/brandt/rolling-resistance-tubular.html
Here's the same data plotted on a more appropriate scale for
extrapolating.
http://www.ocf.berkeley.edu/~tee/rbt/rolres.gif
>All these tires nearly approach zero RR at infinite pressure with a
>constant offset for tubulars that have rim glue losses. You'll notice
>that this is a family of curves of identical shape.
The limit at infinite pressure for the best tire is about half
that of the glued tubulars: significantly better, but certainly not zero.
The 28c Specialized Touring (white line) has an offset that's
nearly as large as the tubulars. I'm not familiar with this tire, but
would guess that it has a raised tread rib, and this loss is from squirming
of that rib.
-Luns
jobst....@stanfordalumni.org
>> I think the testing method and instrumentation is probably the
>> culprit in these erroneous findings that go contrary to most RR
>> tests done without involving bicycles or riders but only rolling
>> tires.
>> Please review the chart at:
http://www.sheldonbrown.com/brandt/rolling-resistance-tubular.html
>> All these tires nearly approach zero RR at infinite pressure with a
>> constant offset for tubulars that have rim glue losses. You'll
>> notice that this is a family of curves of identical shape.
> With a smooth roller, yes. Add some real world surface features,
> which are often sized at ~20% of the tyre section height even for
> fairly smooth pavement, and your "infinite pressure" pneumatic tyres
> will roll like iron tyred cart wheels. Bicycle tyre tests which
> don't involve bicycles or riders are only slightly interesting, and
> almost completely useless.
There is no point in testing riders or bicycles when the merit of tire
RR is being assessed. Maybe you can explain what advantage you see in
doing so.
RR comes from tread and casing deformation which is dependent on
inflation pressure. If these tests are done on rougher surfaces all
values will be higher but the characteristic will remain unchanged.
You are suggesting that the lab at IRC that constructed this tire
test, that is more credible and apparent than any others performed,
does not know how to correctly test tires.
I don't think you understand what is being tested and what the
mechanisms that cause rolling losses are. Beyond that, tread patterns
only make more RR as you can see in those curves. Those are the ones
that don't flatten as much as one would expect as inflation pressure
was increased.
Jobst Brandt
jobst....@stanfordalumni.org
>> Please review the chart at:
http://www.sheldonbrown.com/brandt/rolling-resistance-tubular.html
> Here's the same data plotted on a more appropriate scale for
> extrapolating.
http://www.ocf.berkeley.edu/~tee/rbt/rolres.gif
>> All these tires nearly approach zero RR at infinite pressure with a
>> constant offset for tubulars that have rim glue losses. You'll
>> notice that this is a family of curves of identical shape.
> The limit at infinite pressure for the best tire is about half that
> of the glued tubulars: significantly better, but certainly not zero.
> The 28c Specialized Touring (white line) has an offset that's nearly
> as large as the tubulars. I'm not familiar with this tire, but would
> guess that it has a raised tread rib, and this loss is from
> squirming of that rib.
Elegant!
Although the RR values at infinity seem high there must be an
explanation for their separation and high value. Are we seeing the
correct vertical scale? At infinite pressure, only thin tread rubber
is still flexing as tread squirm. That the logarithmic curves are
parallel rather than converging seems odd at first inspection.
Jobst Brandt
Luns Tee
<jobst....@stanfordalumni.org> wrote:
> http://www.ocf.berkeley.edu/~tee/rbt/rolres.gif
>
>>> All these tires nearly approach zero RR at infinite pressure with a
>>> constant offset for tubulars that have rim glue losses. You'll
>>> notice that this is a family of curves of identical shape.
>
>> The limit at infinite pressure for the best tire is about half that
>> of the glued tubulars: significantly better, but certainly not zero.
>
>> The 28c Specialized Touring (white line) has an offset that's nearly
>> as large as the tubulars. I'm not familiar with this tire, but would
>> guess that it has a raised tread rib, and this loss is from
>> squirming of that rib.
>
>Elegant!
>
>Although the RR values at infinity seem high there must be an
>explanation for their separation and high value. Are we seeing the
>correct vertical scale? At infinite pressure, only thin tread rubber
>is still flexing as tread squirm. That the logarithmic curves are
>parallel rather than converging seems odd at first inspection.
These aren't log-scale. The vertical scale is linear, and
exactly the same scale as in your graph - you can even check the data
points visually and see that they agree. The horizontal scale is
actually -1/pressure with the origin at the right, and the tickmarks
then re-labelled according to pressure.
1/pressure is just area per force, so dimensionally, the
k/pressure term is proportional to the area of the contact patch for a
given wheel load.
That the curves straighten out into lines in this scale shows that
the data fits a relationship of
resistance = offset + k/pressure
and that the curve for each tire can actually be captured by two
constants - the offset, and the scaling factor. I have a scatter plot
of what these constants are for the various tires, but it doesn't
really tell anything that can't be seen from the image I gave above.
-Luns
peter
> > http://www.ocf.berkeley.edu/~tee/rbt/rolres.gif
> These aren't log-scale. The vertical scale is linear, and
> exactly the same scale as in your graph - you can even check the data
> points visually and see that they agree. The horizontal scale is
> actually -1/pressure with the origin at the right, and the tickmarks
> then re-labelled according to pressure.
>
> 1/pressure is just area per force, so dimensionally, the
> k/pressure term is proportional to the area of the contact patch for a
> given wheel load.
>
> That the curves straighten out into lines in this scale shows that
> the data fits a relationship of
>
> resistance = offset + k/pressure
>
> and that the curve for each tire can actually be captured by two
> constants - the offset, and the scaling factor.
Interesting graph. I'd assume that for the clinchers the offset value
would be primarily determined by the losses due to the tread rubber at
the contact patch whereas the 'k' value would be primarily due to flex
losses in the casing and sidewall. In that case a combination of the
casing used in the Avocets and the tread rubber of the Michelin Hilite
Comp would seem to be a worthwhile tire design.
frkr...@gmail.com
jobst....@stanfordalumni.org wrote:
> someone writes:
>
> >> I think the testing method and instrumentation is probably the
> >> culprit in these erroneous findings that go contrary to most RR
> >> tests done without involving bicycles or riders but only rolling
> >> tires.
>
> >> Please review the chart at:
>
> http://www.sheldonbrown.com/brandt/rolling-resistance-tubular.html
>
> >> All these tires nearly approach zero RR at infinite pressure with a
> >> constant offset for tubulars that have rim glue losses. You'll
> >> notice that this is a family of curves of identical shape.
>
> > With a smooth roller, yes. Add some real world surface features,
> > which are often sized at ~20% of the tyre section height even for
> > fairly smooth pavement, and your "infinite pressure" pneumatic tyres
> > will roll like iron tyred cart wheels. Bicycle tyre tests which
> > don't involve bicycles or riders are only slightly interesting, and
> > almost completely useless.
>
> There is no point in testing riders or bicycles when the merit of tire
> RR is being assessed. Maybe you can explain what advantage you see in
> doing so.
>
> RR comes from tread and casing deformation which is dependent on
> inflation pressure. If these tests are done on rougher surfaces all
> values will be higher but the characteristic will remain unchanged.
Jobst, was the data linked above measured by rolling on a smooth drum?
IIRC, the test in [Vintage] Bicycle Quarterly was a roll-down test on
pavement. The justification given was that tests on perfectly smooth
surfaces miss certain effects, alluded to above. On a perfectly smooth
surface, no "tire suspension" (that is, flexibility in the contact
patch and the tire just above it) is needed, so rolling resistance
always decreases with increased pressure.
But on a rough surface, a tire with infinitely high pressure, like a
steel wheel coated with rubber, would launch the bike into thousands of
tiny vertical hops. Those consume energy, and represent lost forward
motion. In fact, the vertical bounces would probably consume more
energy than the tire hystresis, the main effect measured on a smooth
test surface.
Perhaps testing rolling resistance by an outdoor roll-down test isn't
perfect. I can see error being introduced by minor changes in rider
position, clothing, temperature, wind speed & direction, etc. But the
testers seemed aware of those problems, made efforts to control them,
and the data in the original article shows good precision.
Another point is, IIRC, they didn't describe the exact nature of the
asphalt. (Is there some roughness measurment in common use, similar to
the RMS method of describing metal finishing smoothness?) ISTM that
rougher asphalt might rank tires differently than smoother asphalt.
In any case, until shown otherwise, I'd expect rolling resistance
measured _somehow_ using an appropriate mass on an appropriately rough
surface - not smooth steel - would be more valid for real riding.
> You are suggesting that the lab at IRC that constructed this tire
> test, that is more credible and apparent than any others performed,
> does not know how to correctly test tires.
Well, it wouldn't be the first time a manufacturer made a big mistake!
- Frank Krygowski
frkr...@gmail.com
Werehatrack wrote:
> "Latex tubes do improve comfort"? Once I got to that one, I knew the
> results were questionable.
[Vintage] Bicycle Quarterly - largely a one-man operation, it seems -
does contain a lot of subjective judgements stated as fact. Many of
them seem questionable at best.
But ISTM that one can be skeptical about those judgements, yet not toss
out other data resulting from measurements made with decent technique.
- Frank Krygowski
jobst....@stanfordalumni.org
http://www.sheldonbrown.com/brandt/rolling-resistance-tubular.html
All RR tests by the auto industry are done on smooth drums, pavement
granularity only being undefined and not comparable from one tester to
the next. These tests need to be verifiable at different labs, so
they use standard smooth drums of known diameter. Roller roll-down
tests by Micheline at InterBike are also done on steel rollers.
> IIRC, the test in [Vintage] Bicycle Quarterly was a roll-down test
> on pavement. The justification given was that tests on perfectly
> smooth surfaces miss certain effects, alluded to above. On a
> perfectly smooth surface, no "tire suspension" (that is, flexibility
> in the contact patch and the tire just above it) is needed, so
> rolling resistance always decreases with increased pressure.
I think they need to assess what these "certain" effects are and how
they do not correlate to smooth drum testing. I propose that the
order of RR quality does not change with true pavement compared to a
steel drum, the losses coming from the same effects.
> But on a rough surface, a tire with infinitely high pressure, like a
> steel wheel coated with rubber, would launch the bike into thousands
> of tiny vertical hops. Those consume energy, and represent lost
> forward motion. In fact, the vertical bounces would probably
> consume more energy than the tire hysteresis, the main effect
> measured on a smooth test surface.
That is only true if the roughness is greater than the rubber
thickness and uneven at that. Bouncing does not in itself absorb
energy, it is the hysteretic losses in the tread and casing. There is
no purpose in testing tires on RR ballast for instance.
> Perhaps testing rolling resistance by an outdoor roll-down test
> isn't perfect. I can see error being introduced by minor changes in
> rider position, clothing, temperature, wind speed & direction, etc.
> But the testers seemed aware of those problems, made efforts to
> control them, and the data in the original article shows good
> precision.
The rider has far more influence on drag than any tire so the whole
experiment falls apart.
> Another point is, IIRC, they didn't describe the exact nature of the
> asphalt. (Is there some roughness measurement in common use,
> similar to the RMS method of describing metal finishing smoothness?)
> ISTM that rougher asphalt might rank tires differently than smoother
> asphalt.
You can't describe that. That is why you don't do that.
> In any case, until shown otherwise, I'd expect rolling resistance
> measured _somehow_ using an appropriate mass on an appropriately
> rough surface - not smooth steel - would be more valid for real
> riding.
I don't believe so. If you could show me where the energy goes that
makes the pavement test produce a different hierarchy among samples, I
would be interested.
>> You are suggesting that the lab at IRC that constructed this tire
>> test, that is more credible and apparent than any others performed,
>> does not know how to correctly test tires.
> Well, it wouldn't be the first time a manufacturer made a big
> mistake!
Well? Tell me where the big mistake might be. As you see, these are
the first RR tests that revealed the dirty secret of tubular rim glue
and explained why there is pressure sensitive road glue and hard non
resilient track glue.
Jobst Brandt
Tim Binns
>Frank Krygowski writes:
>
>All RR tests by the auto industry are done on smooth drums, pavement
>granularity only being undefined and not comparable from one tester to
>the next. These tests need to be verifiable at different labs, so
>they use standard smooth drums of known diameter. Roller roll-down
>tests by Micheline at InterBike are also done on steel rollers.
>
Couldn't that be interpreted to mean that the auto industry,
similarly, is just dodging the issue?
[snip]
>
>> Another point is, IIRC, they didn't describe the exact nature of the
>> asphalt. (Is there some roughness measurement in common use,
>> similar to the RMS method of describing metal finishing smoothness?)
>> ISTM that rougher asphalt might rank tires differently than smoother
>> asphalt.
>
>You can't describe that. That is why you don't do that.
>
The people who make it their busines to monitor road surface quality
certainly can, see:
http://www.wdm.co.uk/pdf/S_RAV.pdf
...where roughness and texture, as well as cracks and wheel ruts are
some of the quality parameters that are assessed.
carl...@comcast.net
wrote:
Dear Tim,
I like those test trucks, partly because they're painted yellow.
Do you have any links to output from the machines? That is, graphs or
numbers indicating roughness? I'm just curious what kind of reporting
they can give.
Cheers,
Carl Fogel
Luns Tee
peter <prat...@comcast.net> wrote:
>Luns Tee wrote:
>> > http://www.ocf.berkeley.edu/~tee/rbt/rolres.gif
>> That the curves straighten out into lines in this scale shows that
>> the data fits a relationship of
>>
>> resistance = offset + k/pressure
>Interesting graph. I'd assume that for the clinchers the offset value
>would be primarily determined by the losses due to the tread rubber at
>the contact patch whereas the 'k' value would be primarily due to flex
>losses in the casing and sidewall.
That's pretty much what I read of it, though I think the offset
may also have some component from the tire/rim interface too: the
elastomers of the casing in the clinch area get worked to squirm around
just as the tubular glue does, except on a smaller scale. I'd expect
this loss to be small relative to glue, or tread squirm losses, but I
can't say how much smaller.
>In that case a combination of the
>casing used in the Avocets and the tread rubber of the Michelin Hilite
>Comp would seem to be a worthwhile tire design.
While I don't know the construction of the Michelin tire, the
difference may be as simple as a matter of the tread thickness, in which
case this is just the compromise between RR and tread life.
If we knew the tread thicknesses for these different tires, I
would expect to see a good correlation between the tread squirm offset
and the thicknesses.
-Luns
O-V R:nen
> But ISTM that one can be skeptical about those judgements, yet not toss
> out other data resulting from measurements made with decent technique.
As it happens, this month's Tour magazine has some tests of various
types of inner tubes, and the latex one beat the competition in all
categories except for air retention (the need to refill). I guess the
category closest to "comfort" is rolling resistance:
- latex (80 g) 23.9 W
- polyurethane (60 g) 25.1 W
- butyl (50 g) 25.9 W, (75 g) 28.8 W, (104 g) 32.1 W
(What I thought was missing from the test was whether and how patching
affects these characteristics, and also whether there are any inherent
differences as to the likelihood of patch failure.)
Donald Gillies
purposes, above 20 miles per hour, wind resistance account for most of
the drag experienced by a bicycle. The other types of resistance
(drivetrain - 3-5% loss, and rolling resistance) are linear
w.r.t. speed, I believe.
Get into a TUCK and let air out of those tires !! *grin*
- Don Gillies
San Diego, CA
Antti Salonen
> As it happens, this month's Tour magazine has some tests of various
> types of inner tubes, and the latex one beat the competition in all
> categories except for air retention (the need to refill). I guess the
> category closest to "comfort" is rolling resistance:
>
> - latex (80 g) 23.9 W
> - polyurethane (60 g) 25.1 W
> - butyl (50 g) 25.9 W, (75 g) 28.8 W, (104 g) 32.1 W
I'm guessing these losses must also include the losses of some tyre,
which was used for testing purposes. Is this for a single tyre or a
pair? What was the tyre pressure, weight on the wheel(s) and speed?
-as
Mark Hickey
Equally likely that if you learned by reading these posts that it
turns out that latex tubes actually reduce the ride quality, that
you'd notice that as well.
A good example of the bicycle placebo effect was when big-tube
aluminum bikes were new, and following the experience with the Vitus
small-tube aluminum bike. The magazine writers in the US concluded
that big-tube aluminum bikes were harsh, and lo, riders in the US
agreed. At the same time, writers in the UK (who apparently didn't
get the memo...) concluded that, like the Vitus, the big-tube
aluminium (sic) bikes were soft and compliant. And lo, their readers
felt that too.
That's not to say that there isn't some physical difference in the way
the latex tubes react to deflection - but I would have a very hard
time believing that it would reach a level that is measurable on the
"butt-o-meter". ;-)
Mark Hickey
Habanero Cycles
http://www.habcycles.com
Home of the $795 ti frame
O-V R:nen
I guess the setup was the same as for their tyre comparison a while
back, all tests were made with a 23 mm Continental Supersonic at
7.5 bar and 30 km/h, the article doesn't say whether the figures are
for a single wheel or a pair.
Antti Salonen
> > > - latex (80 g) 23.9 W
> > > - polyurethane (60 g) 25.1 W
> > > - butyl (50 g) 25.9 W, (75 g) 28.8 W, (104 g) 32.1 W
>
> I guess the setup was the same as for their tyre comparison a while
> back, all tests were made with a 23 mm Continental Supersonic at
> 7.5 bar and 30 km/h, the article doesn't say whether the figures are
> for a single wheel or a pair.
It depends on the load also, but these might be for a pair. Anyway,
considering that the differences are several watts, it's easy to say
that for any kind of competition use latex tubes make a lot of sense.
Their only disadvantage there is irrelevant because people usually top
off their tyres before any kind of race anyway.
-as
carl...@comcast.net
>A Bicycle Quarterly reader pointed me to the discussion here, and I
>read it with interest. A few points from the authors of the test:
>
>1. The methodology of the test was carefully designed. It was reviewed
>by several experts in the field, including Jim Papadopoulos, Frank
>Berto and Andreas Oehler. The tests showed excellent reproducibility -
>the same tires tested at different times during the test to check this
>gave identical results. Each combination of tire/pressure/tube was
>tested at least 3 times to weed out erroneous results. Fortunately, we
>had very few - all wind-related, when a light gust of wind slowed the
>rider down. We felt the wind before the rolldown even had been
>completed. We only tested when there was no wind, of course.
[snip]
>Jan Heine
>Editor
>Bicycle Quarterly
>140 Lakeside Ave #C
>Seattle WA 98122
Dear Jan,
As far as I know, a 1 mph wind is undetectable, but will have
considerable consequences for such testing. Plug in a +1 or -1 mph
wind on these calculators:
http://www.kreuzotter.de/english/espeed.htm
http://w3.iac.net/~curta/bp/velocity/velocity.html
How do you determine "no wind" conditions over an open course that's
several hundred yards long?
What's the minimum speed of "no wind" conditions?
Cheers,
Carl Fogel
Phil Holman
"Donald Gillies" <gil...@cs.ubc.ca> wrote in message
news:ej1p33$cmh$1...@cascade.cs.ubc.ca...
> Wind resistance goes up as the CUBE of the speed, so for all practical
> purposes, above 20 miles per hour, wind resistance account for most of
> the drag experienced by a bicycle. The other types of resistance
> (drivetrain - 3-5% loss, and rolling resistance) are linear
> w.r.t. speed, I believe.
Wind resistance goes up with the SQUARE of the speed (Bernoulli's P =
1/2 rho v^2).
Phil H
Phil Holman
"wle" <w...@mailinator.com> wrote in message
news:1163119376....@h54g2000cwb.googlegroups.com...
>
> no, they are saying it;s all rolling resistance that makes tires 'roll
> faster'.
>
>With roughly the same power output, the rider's speed can vary by as
>much as 20% depending on tire choice. For example, the rider on the
>fastest tire [in this roll-down test] moved down the road at
>approximately 16.4 mph (26.2 kph) while the same rider on the slowest
>tire went approximately 13.6 mph (21.7 kph).
Well, at such low speeds on flat terrain, tire RR is a larger percentage
of total drag but at more realistic road speeds, say 20 mph and above,
the pecentage is much smaller. For example, using your previous numbers,
if you were to travel at 25mph on the slowest tire, you would go at
26mph on the fastest at the same power output.
Phil H
Peter Cole
Force goes up as square (as you say), power goes up as cube.
Tim McNamara
hei...@earthlink.net wrote:
> A Bicycle Quarterly reader pointed me to the discussion here, and I
> read it with interest. A few points from the authors of the test:
>
> 1. The methodology of the test was carefully designed. It was
> reviewed by several experts in the field, including Jim
> Papadopoulos, Frank Berto and Andreas Oehler.
When I read the article, I found it very interesting, but noted that the
results seemed to differ from other studies (you offered one comparison,
testing done for Tour Magazine). That is always a question mark in
science. I brought it up for discussion here looking for better
understanding of why some of your results appeared to contradict the
corpus of results from other research. The conclusion in the article
appeared to explain this as based on the differences between a tire
rolling on a steel drum versus on pavement. It's an interesting
question as to whether this makes a real difference in the ordinal
ranking of tires (I would have said "not" if asked before reading your
article).
> The tests showed excellent reproducibility - the same tires tested at
> different times during the test to check this gave identical results.
Not quite identical, looking at your article. Some of the tires have
quite a bit of variability, according to the graph on page 23. Note the
spread for the Deda Tre Giro d'Italia, Avocet Duro 700 x 32, and
Panaracer Col de Vie. The Avocet, for example appears to have ranged
over nearly 2 seconds elapsed time- that's a range of nearly 14%. The
fastest speed would have it ranked 6th, tied with the Clement Criterium
tubular, and the slowest speed would have it ranked 12th, just below the
Grand Bois. I can think of no intrinsic reason for this variability- it
seems to me that the variability must have resulted from extrinsic
factors. That was the issue that raised questions for me about the
results.
> Each combination of tire/pressure/tube was tested at least 3 times to
> weed out erroneous results. Fortunately, we had very few - all
> wind-related, when a light gust of wind slowed the rider down. We
> felt the wind before the rolldown even had been completed. We only
> tested when there was no wind, of course.
The concerns raised in the r.b.t discussion were IIRC mainly about
potential confounds. One such was variances in timing accuracy (the
photographs appear to show you timing with a wristwatch, for example)
which would be an issue given the very short timing period. The
difference in elapsed time ranged from 25.3 seconds to 30.6 seconds,
with tires being ranked by intervals as small as 0.1 seconds. Research
has tended to find that the average time lag in eye-hand coordination is
somewhere around 0.25 seconds. To my thinking this is the biggest
confound in your test, even using two people timing independently. But
then I am a psychologist so I tend to look to human factors for
explanation of differences.
You can see it for yourself using a simple timer. Just try to stop the
clock at 10 seconds even.
http://www.easysurf.cc/stimer.htm
If you anticipate by counting time in your head, you'll tend to be more
accurate IME. If you wait to see the 10 show up, you'll be off by
around 0.2-0.3 seconds. Another reaction time test that prevents
anticipation is:
http://getyourwebsitehere.com/jswb/rttest01.html
You'll notice the reaction times are significantly longer because you
can't readily anticipate the stimulus. However, in your test you could
see the bike rolling to the timing mark so anticipation was possible.
Another potential confound was whether the path traveled by the bike was
the same each time. A few small wobbles here and there could
significantly impact the outcome, since the time differences were very
small to begin with.
A third was the issue of air movement, even an imperceptible breeze of
0.5 mph having a predictable and measurable effect. I believe Carl
Fogel posted a link to the relevant page at analyticcycling.com that
shows this effect.
It would be interesting to run a statistical analysis on the data, to
calculate the error of measurement, degrees of freedom, etc. This would
establish the reliability and give some indication of the validity. The
main question is whether the intervallic differences are larger than the
error of measurement.
_Bicycle Quarterly_ has been the most interesting bike magazine I read
for a few years now. I was very pleased to see you try to address this
issue and hope that it can be honed and refined.
frkr...@gmail.com
> > drum?
>
> All RR tests by the auto industry are done on smooth drums, pavement
> granularity only being undefined and not comparable from one tester to
> the next. These tests need to be verifiable at different labs, so
> they use standard smooth drums of known diameter. Roller roll-down
> tests by Micheline at InterBike are also done on steel rollers.
I understand that using drums of similar smoothness will give the most
consistent results. It doesn't follow, though, that the results
correctly indicate real-world results.
> > IIRC, the test in [Vintage] Bicycle Quarterly was a roll-down test
> > on pavement. The justification given was that tests on perfectly
> > smooth surfaces miss certain effects, alluded to above. On a
> > perfectly smooth surface, no "tire suspension" (that is, flexibility
> > in the contact patch and the tire just above it) is needed, so
> > rolling resistance always decreases with increased pressure.
>
> I think they need to assess what these "certain" effects are and how
> they do not correlate to smooth drum testing. I propose that the
> order of RR quality does not change with true pavement compared to a
> steel drum, the losses coming from the same effects.
ISTM the order (as well as the magnitudes, of course) could change
significantly.
If all rolling resistance effects were linear wrt. deflection (that is,
wrt. surface roughness and tire pressure), I think your proposition
would have weight. But I don't think that condition has been proven.
ISTM that tire A could have lower rolling resitance than tire B at low
values of deflection, but that A could be worse than B at higher
deflections.
In the front view of a tire supporting a load, the tire sidewalls don't
form a circular arc; the bottom of the arc is flattened where the tire
contacts the road. ISTM that immediately left and right of the contact
patch is the minimum radius of curvature, i.e. the points where the
tire material flexes most.
By variations in tire construction and materials, I think tire A might
have less hysteresis loss than B when that max flex point is close to
the centerline, and more hysteresis loss than tire B when that max
flex point is moved further around to the sidewall. Think, for example
of two tires, one with thick, inflexible, lossy sidewalls and the other
with thin, super-flexible sidewalls.
>
> > But on a rough surface, a tire with infinitely high pressure, like a
> > steel wheel coated with rubber, would launch the bike into thousands
> > of tiny vertical hops. Those consume energy, and represent lost
> > forward motion. In fact, the vertical bounces would probably
> > consume more energy than the tire hysteresis, the main effect
> > measured on a smooth test surface.
>
> That is only true if the roughness is greater than the rubber
> thickness and uneven at that. Bouncing does not in itself absorb
> energy, it is the hysteretic losses in the tread and casing. There is
> no purpose in testing tires on RR ballast for instance.
>
> > In any case, until shown otherwise, I'd expect rolling resistance
> > measured _somehow_ using an appropriate mass on an appropriately
> > rough surface - not smooth steel - would be more valid for real
> > riding.
>
> I don't believe so. If you could show me where the energy goes that
> makes the pavement test produce a different hierarchy among samples, I
> would be interested.
I think in real life, bouncing does absorb energy. The losses in the
bike-rider assembly are not confined to the rubber in the tire. For
example, when the rider's mass is jolted upward, it consumes
significant energy. That energy is probably dissipated as a
combination of hysteresis and muscular work within the body. Even
things like the contents of one's handlebar bag get jostled in a way
that is not recoverable energy.
I think potential nonlinearity of loss w.r.t. deflection is a partial
explanation, and I think energy loss internal to the suspended mass
(mostly, the rider's body) is another part of the explanation. And
there may be more.
But if you want two examples that would obviously produce a different
hierarchy, just make one of them a rigid steel wheel with a rubber
coating!
- Frank Krygowski
carl...@comcast.net
>carl...@comcast.net wrote:
>> Dear Jan,
>>
>> As far as I know, a 1 mph wind is undetectable, but will have
>> considerable consequences for such testing.
>
>> How do you determine "no wind" conditions over an open course that's
>> several hundred yards long?
>>
>> What's the minimum speed of "no wind" conditions?
>
>We watched leaves of trees nearby for movement, plus watched dust we
>kicked up in the air. If the dust cloud did not move, then we assumed
>there was no wind.
>
>In the end, the fact that our results are reproducible gives us
>confidence that these factors do not play a great enough role to affect
>our results significantly. We did state in the article that we
>considered run times within 0.5 seconds to be basically identical. (Our
>run times were between 25.3 and 30.6 seconds for the course length of
>184 m.)
>
>Of our 155 measurements (timed runs that were not aborted due to wind),
>we have only 5 or 6 outliers, where there is more than a spread of 0.8
>seconds between the slowest and fastest run for the same configuration.
>We have 2 or 3 runs that are more than 1 second apart. In those cases,
>we did additional measurements. We aborted about 10 runs due to very
>light, but noticeable gusts of wind. When you consider that the
>differences between the fastest and slowest tires are 5.3 seconds, you
>realize that the run-to-run variability is relatively small. Before we
>did the experiments, we did a few test runs, because we were not sure
>that our methods would produce reproducible results. Only once we had
>established that, we proceeded to spend the time and effort on the
>tests.
[snip]
Dear Jan,
At about 1 mph, you won't see leaves rustling or smoke drifting:
http://www.unc.edu/~rowlett/units/scales/beaufort.html
The test apparently involved average speeds for a 184-meter run that
took from 30.6 to 25.3 seconds, presumably after coasting down a hill
to gain speed and then losing it on a flatter runout.
That works out to average speeds of 13.45 to 16.27 mph. If the riders
entered the timed section and lost speed, their entry speeds must have
been noticeably higher than the average speed for the 603.67 foot
timed section.
The testing took place over several mornings.
There is no hope of detecting wind changes of half a mile per hour
either way where you're standing, much less over a distance of several
football fields.
Even if the temperature was the same every morning (unlikely),
mornings are when temperatures rise. A 15F degree range of
temperatures seems reasonable.
The riders presumably varied a bit in weight, due to matters too
delicate to discuss here, plus what they were wearing.
What they were wearing, how they wore it, and indetectable changes in
posture can be expected to cause 1% variations in frontal area and the
wind drag profile (CV).
Let's see what kind of maximum variations are predicted for terminal
velocity coasting down a 2% grade with conservative +/- changes of 0.5
mph wind, 7.5 degrees F, 1% frontal profile, 1% rider weight, and 1%
CV:
http://w3.iac.net/~curta/bp/velocity/velocity.html
wind temperature frontal CV weight speed speed
km/h C/F m^2 kg km/h mph
0.80 16.0C 60.8F 0.404 0.909 79.2 26.6882 16.58
0.00 20.0C 68.0F 0.400 0.900 80.0 28.0833 17.45
-0.80 24.0C 75.2F 0.396 0.891 80.8 29.4959 18.33
This means that with good intentions the "same" rider on the "same"
course can coast into a 184-meter test section at the bottom of a 2%
grade anywhere from 16.6 to 18.3 mph from one day to the next, a 1.7
mph variation.
Cut that in half, and we still have a 0.8 mph variation.
Raise the entry speed (likely, just as it's likely that the grade was
more than 2%), and the potential variation rises.
If the runs involved testing the same tire several times in a row
early in the morning one day, that would reduce variation for that
tire, but not variation compared to another tire tested on another
morning.
That is, a fast morning could easily be compared in good faith to a
slow morning.
Cheers,
Carl Fogel
Tim McNamara
carl...@comcast.net wrote:
> Even if the temperature was the same every morning (unlikely),
> mornings are when temperatures rise. A 15F degree range of
> temperatures seems reasonable.
>
> The riders presumably varied a bit in weight, due to matters too
> delicate to discuss here, plus what they were wearing.
>
> What they were wearing, how they wore it, and indetectable changes in
> posture can be expected to cause 1% variations in frontal area and
> the wind drag profile (CV).
Carl, you might want to actually read the article. Those issues were
addressed and efforts made to minimize them. Only one rider was used,
wearing the same clothing each time, riding the same bike in the same
position.
carl...@comcast.net
Dear Tim,
It would be interesting to have the rider zero a cyclocomputer every
morning for two weeks on the same bike, cover it, roll down the same
hill, decide whether he felt the run was acceptable, and then uncover
the cyclocomputer and record its maximum speed.
The cyclocomputer's maximum reading would get around any human factor
in hitting the stopwatch button.
I expect that the maximum speeds would vary in a discouraging fashion,
due to the effect of wind speeds below our perception, plus variations
in temperature, barometric pressure, rider weight, tuck position, and
so forth.
I don't know if the test included another bike with unchanging tires
going down the hill every time to serve as a rough baseline, but I'll
be surprised if this basic check was performed.
I believe that the test was conducted in good faith, but I'd be very
surprised if the wind varied only 1 mph. The weather service calls it
calm if the wind drops below 3 mph.
Here's a link to the last 7 days of wind speeds recorded near Seattle:
http://www.wrh.noaa.gov/mesowest/getobext.php?wfo=sew&sid=KSEA&num=168&raw=0
Only one day in the last week reported brief "calm" periods in the
early morning, and it was raining then:
From 4:50am to 9:50am on Nov. 4th, the wind blew from 9 to 13 mph.
From 4:50am to 9:50am on Nov. 5th, the wind blew from 5 to 15 mph.
From 4:50am to 9:50am on Nov. 6th, the wind blew from 13 to 29 mph.
From 4:50am to 9:50am on Nov. 7th, the wind blew from 6 to 13 mph.
From 4:50am to 9:50am on Nov. 8th, the wind blew from 7 to 12 mph.
From 4:50am to 9:50am on Nov. 9th, the wind blew from 3 to 9 mph,
but it was "calm" (< 3mph) and raining at 5:40am and 5:50am
From 4:50am to 9:50am on Nov. 10th, the wind blew from 16 to 32 mph.
The testing may have been done during less windy weeks and somewhere
less windy than Sea-Tac, but the weather data suggests that no tests
were possible for the last week. It is unusual for the wind to drop to
less than 1 mph for any appreciable time over a 600-foot stretch of
road.
The tests are also measuring the increase in wind drag from what at
first seems like only slightly wider tires.
But a 700 x 32 is 9 mm wider than a 700 x 23, which means roughly four
9 mm wide strips (front and back of each tire), each about 600 mm
high--that's 4 x 9 x 600 = 21,600 mm^2 of extra frontal area, or 33.5
square inches, roughly a square 5.8 inches wide.
Arguably, the extra wind drag could be lumped in as part of the tire's
rolling resistance, but it would confuse the hell out of things.
Cheers,
Carl Fogel
carl...@comcast.net
Dear Tim,
The same clothes can be worn differently, particularly if it isn't a
tight-fitting skin suit.
That is, failing to tuck your shirt in as much will cause noticeable
drag changes at speed. I often notice that my shirt has loosened as I
descend a long stretch of highway on my daily ride.
Similarly, the claims for the same position aren't very credible.
Raising or lowering your chin an inch is going to make a difference in
wind drag, not just from the pure frontal area, but from the angle
that it tips your helmet.
Surely you're aware of how a very slight change in your head or the
curve of your back can change how the wind feels on your back when
you're coasting down a hill--the rippling of your shirt is the air
flow separating and a very bad thing indeed in terms of drag.
I'm willing to believe that the bicycle can be duplicated to an
impressive degree on repeated runs.
I'm much less willing to believe that the chin angle, amount the
shoulders are raised or lowered, knee spread, and angle of the crank
are going to be duplicated to the same degree.
Was the rider weighed, with his bicycle, before each morning's run,
day after day? I doubt it.
Consider that a 700 x 32 tire often weighs 150 grams more than a 700 x
23 tire. Two of them amount to 300 grams, or two thirds of a pound
just from the tires. If a 150-lb rider's weight varies 1% either way
during the testing, the potential variation is about 1 kg either way.
It's hard to credit a test that attempts to measure one small variable
amid a host of confounding factors.
Here's the maximum mph for the same bicycle and rider, trying to tuck
in and coast as fast as possible, on the same three hills for the last
two weeks. That's 42 test rides, about a quarter of the 155 test rides
in question.
last 14 rides:
hill hill hill fastest middle slowest
A B C
43.0 38.3 34.7 A B C
45.4 38.3 33.7 A B C
44.7 42.7 34.3 A B C
38.0 35.3 34.6 A B C
35.0 37.7 36.8 ? B C A
36.0 38.6 36.0 ? B tie tie
40.7 39.5 36.1 A B C
38.4 40.3 36.1 ? B A C
39.7 42.7 35.1 ? B A C
38.2 42.3 36.2 ? B A C
45.7 39.4 33.7 A B C
47.1 40.7 34.6 A B C
37.1 38.6 37.1 ? B tie tie
36.8 34.0 36.4 ? A C B
47.1 42.7 37.1 14-day high
35.0 35.3 33.7 14-day low
12.1 7.4 3.4 14-day variation
40.41 39.17 35.39 average last 14 rides
40.04 37.82 35.59 yearly average mph
Hmmm . . . the 2-week averages are within 0.37, 1.35, and 0.20 mph of
the averages for the year so far. That's reassuring.
But it's not too reassuring that the hills came out in the right
order, fastest to slowest, A-B-C, only 6 out of 14 times.
Nor is it reassuring that speed on the fastest hill varied from 35 to
47 mph.
Nor is it reassuring that speed on the slowest hill varied only 3.4
mph, even though they were ridden within 30 minutes of each other.
(The slowest hill is a steep S-bend down a gully and sheltered from
most of the wind. But it still varies 3.4 mph in 14 runs, about 10%.)
Is the variation due to the wind, the initial speed before coasting,
what the rider ate for breakfast, how quickly he got into his tuck,
how well he held it, how tightly his shirt was tucked in, the
temperature, the barometric pressure, or what?
If you didn't know that the same tires were used, what would you
conclude about the rolling resistance?
I agree with Jobst that a spin-down test on a smooth drum indoors is
likely to be far more accurate and reliable than a test that adds a
rider, his shirt-tail, whatever he ate for breakfast, whether he's
hunching his shoulders up against the cold, and whatever wind may be
blowing on a course several football fields long.
It just doesn't make sense to expect that adding lots of variables and
then trying to weed them out will produce good results.
I wish it were otherwise, but I wouldn't be surprised if the tuck and
the tightness of the clothing were about as closely controlled as the
wind.
As I've mentioned elsewhere, it would be interesting to have the same
rider attempt to duplicate the same runs over a few weeks on the same
bicycle and have him decide whether the runs were acceptable as far as
he could tell--and then uncover the speedometer and see what the
maximum speed was.
Cheers,
Carl Fogel
carl...@comcast.net
>
>carl...@comcast.net wrote:
>
>> The testing may have been done during less windy weeks and somewhere
>> less windy than Sea-Tac, but the weather data suggests that no tests
>> were possible for the last week.
>
>You are right. No testing would have been possible last week, as we had
>one front after another coming through. We'll probably have to wait
>until spring until we can do more tests... even though we have a few
>new tires we'd love to test. Maybe January, if we get some cold, still
>days. We'll have to adjust for temperature, of course.
>
>> Arguably, the extra wind drag could be lumped in as part of the tire's
>> rolling resistance, but it would confuse the hell out of things.
>
>The article is titled "The Performance of Tires," not "rolling
>resistance." We tested overall resistance, of course. In the end, that
>is what matters to riders. However, for the pressure tests, we can say
>that the aerodynamic resistance did not change significantly for the
>same tires at different pressures, so we probably were measuring
>rolling resistance.
>
>Jan Heine
>Editor
>Bicycle Quarterly
>140 Lakeside Ave #C
>Seattle WA 98122
Dear Jan,
Before you spend more time on such tests, you probably need a better
method of judging the wind speed than looking at leaves on trees.
You can buy digital ultrasonic wind-speed sensors, but I don't know of
any reasonably priced models that measure as low as half a mile per
hour. I enjoyed playing with my ultrasonic wind-speed sensor, but it
taught me about the limitations of the technology.
A cheaper method and more convincing method would be to set half a
dozen smoke pots along your test run, light them, and take a picture
of the smoke. I suspect that you'll find that the wind exceeds the
margin of error for your results, but I'd be happy to find that I'm
wrong.
Or you could try my previous suggestion and have the same rider on the
same bike roll down the same hill on the same tires for a few weeks
and see how much his maximum cyclocomputer speed varies when he thinks
that the wind was calm and his tuck was good.
Cheers,
Carl Fogel
peter
> >carl...@comcast.net wrote:
> >
> >> The testing may have been done during less windy weeks and somewhere
> >> less windy than Sea-Tac, but the weather data suggests that no tests
> >> were possible for the last week.
> >
> >You are right. No testing would have been possible last week, as we had
> >one front after another coming through. We'll probably have to wait
> >until spring until we can do more tests... even though we have a few
> >new tires we'd love to test. Maybe January, if we get some cold, still
> >days. We'll have to adjust for temperature, of course.
> >
> >> Arguably, the extra wind drag could be lumped in as part of the tire's
> >> rolling resistance, but it would confuse the hell out of things.
> >
> >The article is titled "The Performance of Tires," not "rolling
> >resistance." We tested overall resistance, of course. In the end, that
> >is what matters to riders. However, for the pressure tests, we can say
> >that the aerodynamic resistance did not change significantly for the
> >same tires at different pressures, so we probably were measuring
> >rolling resistance.
> >
> >Jan Heine
> Before you spend more time on such tests, you probably need a better
> method of judging the wind speed than looking at leaves on trees.
Given that they saw consistent results with the reference wheels they
tested at least twice on each day it appears they already have a
reasonable protocol for keeping the extraneous variables (incl. wind)
sufficiently constant to be able to see differences in the performance
of the tires being tested.
Tim McNamara
carl...@comcast.net wrote:
Again, Carl, I'd suggest you read the article. That would give you a
better idea about the potential confounds and their efforts to minimize
them.
carl...@comcast.net
Dear Peter,
Short of having a reference bicycle coasting down the other side of
the same 600 foot stretch of road at the same time, it's hard to see
controlling for different days.
I don't know of any easy way to measure half-mile-per hour wind
variations on the open road between runs, but that's all that it would
take to confuse the data badly.
Nor do I know of any way short of a wind tunnel to determine how
closely the rider manages to duplicate his drag from run to run.
I agree that it would be very interesting to have a way to compare
real-life downhill runouts with spin-down tests, but when the wind
speed is estimated by leaf-rustling over two football fields plus the
approach, the potential variation amounts to much of the observed
difference.
Unfortunately, consistent results in unblinded studies are nothing
new.
I waste a good deal of time measuring minor things and am painfully
aware of how unsure I am which way I'm shading things. As Tim McNamara
points out elsewhere in this thread, just hitting the stopwatch allows
surprising inaccuracies at the level involved in the test.
Add similar potential inaccuracies in the rider duplicating his wind
drag, right down to how far his shirt is tucked in, and things get
hazier.
The curse of most testing is removing the confounding factors. The
temptation is to leave them in and believe that they don't matter.
It's possible that this test produced accurate results. But unless we
know the wind speed and direction on each run to within less than a
quarter mile per hour, it's open to question even with robot riders
and automatic timing what the test was primarily measuring--how the
wind speed and direction changed from run to run or the effect of
different tires on rolling resistance.
Cheers,
Carl Fogel
jobst....@stanfordalumni.org
... about testing tires with a bicycle and rider, outdoors on an
uncontrolled surface.
I find the whole process faulty because neither the bicycle nor the
rider have anything to do with rolling resistance but have a lot to do
with aerodynamic losses and can influence results by expectation of
good or bad performance.
There is no reason to repeat runs three times if sufficient accurate
data is collected. As I said, the RR tests done by IRC are the best
and most credible ones I have seen. These curves don't need
testimonials or reviewers like Papadopoulos, Berto, or Oehler. They
stand on their own merit as anyone who understands what causes RR can
see.
> Jan Heine
> Editor
> Bicycle Quarterly
> 140 Lakeside Ave #C
> Seattle WA 98122
> http://www.bikequarterly.com
I don't doubt that the editor of this magazine should stand by the
published findings, but that doesn't convince me.
Jobst Brandt
peter
The fact that they were able to observe, and repeat, the time
difference between having the rider wear gloves and not wear gloves
seems to put a pretty good bound on just that.
>
> I agree that it would be very interesting to have a way to compare
> real-life downhill runouts with spin-down tests, but when the wind
> speed is estimated by leaf-rustling over two football fields plus the
> approach, the potential variation amounts to much of the observed
> difference.
There's a big difference between attempting to estimate wind speed and
looking for a period of essentially no wind. Depending on the type of
leaf, it can indeed detect winds of 0.5 mph or less. I would tend to
agree with you on the wind being a hard to control factor *if* their
results showed significant scatter. But the lack of such scatter
across many trials is good evidence that wind variation was adequately
controlled by their procedures - at least for the tests published to
date.
>
> Unfortunately, consistent results in unblinded studies are nothing
> new.
But this was not really an unblinded study. Although not fully
double-blind, the procedures used did protect against experimenter bias
influencing the results. The rider had no reasonable way of knowing
how the timing of a particular run would come out, so I don't find it
credible that he could somehow adjust his position to precisely
compensate for changes in wind/etc. and therefore make the results come
out as consistently as they did. And the person watching the rider
approach and clicking the stopwatch would have his attention on the
rider and the marked finish line and therefore wouldn't see the time on
the watch until after he pushed the button. Unless he deliberately
cheated and just watched the clock rather than the rider I don't see
how he could have made the results come out so consistent either.
> I waste a good deal of time measuring minor things and am painfully
> aware of how unsure I am which way I'm shading things. As Tim McNamara
> points out elsewhere in this thread, just hitting the stopwatch allows
> surprising inaccuracies at the level involved in the test.
>
> Add similar potential inaccuracies in the rider duplicating his wind
> drag, right down to how far his shirt is tucked in, and things get
> hazier.
>
> The curse of most testing is removing the confounding factors. The
> temptation is to leave them in and believe that they don't matter.
> It's possible that this test produced accurate results. But unless we
> know the wind speed and direction on each run to within less than a
> quarter mile per hour, it's open to question even with robot riders
> and automatic timing what the test was primarily measuring--how the
> wind speed and direction changed from run to run or the effect of
> different tires on rolling resistance.
Sure, but that's why they had the multiply repeated trials with the
reference tires and why they did multiple trials on each of the test
tires. If the confounding factors were significant, then these should
have shown substantial scatter of similar magnitude as the difference
between the various tires & pressures being tested. Frankly, I'm also
surprised that the data looks as good as it does, but it just doesn't
show the amount of random fluctuation that would be expected if the
variable wind and shirt-tail effects you mention were having a
significant effect. That leaves open either the possibilities of
deliberate data manipulation or an incredible coincidence that the
confounding factors always acted in the same way for a given tire model
but varied in consistent ways between tire models.
In the absence of evidence to the contrary, I'm willing to trust the
authors that they didn't deliberately falsify the data and the
coincidence hypothesis doesn't look credible given the number of trials
and degree of consistency.
jobst....@stanfordalumni.org
>> As far as I know, a 1 mph wind is undetectable, but will have
>> considerable consequences for such testing.
>> How do you determine "no wind" conditions over an open course
>> that's several hundred yards long?
>> What's the minimum speed of "no wind" conditions?
> We watched leaves of trees nearby for movement, plus watched dust we
> kicked up in the air. If the dust cloud did not move, then we
> assumed there was no wind.
Oh please stop. Here is a test that involves watching dust and leaves
for accuracy. Not that the wind was going in the right direction, but
that it needed to be included in the test. Get a laboratory with
measuring equipment. Car companies run engines on dynomometers at
various power levels long before they do any road testing to assess
the human interface.
Jobst Brandt
frkr...@gmail.com
carl...@comcast.net wrote:
>
Carl, I think you're overestimating the following differences.
>
> Dear Tim,
>
> The same clothes can be worn differently, particularly if it isn't a
> tight-fitting skin suit.
But would they be worn much differently if the testing crew was trying
to guard against that?
> Similarly, the claims for the same position aren't very credible.
> Raising or lowering your chin an inch is going to make a difference in
> wind drag
Measure one inch tilt at your chin. It's an easily perceptible
difference. ISTM that, if you're going to pretend a cyclist can't
sufficiently control that variable, you'll have to tell Lance Armstrong
and his compatriots that wind tunnel testing is useless. Given the
success that technique has apparently generated, I think they'll
disagree.
> I'm willing to believe that the bicycle can be duplicated to an
> impressive degree on repeated runs.
>
> I'm much less willing to believe that the chin angle, amount the
> shoulders are raised or lowered, knee spread, and angle of the crank
> are going to be duplicated to the same degree. ...
>
> Here's the maximum mph for the same bicycle and rider, trying to tuck
> in and coast as fast as possible, on the same three hills for the last
> two weeks. That's 42 test rides, about a quarter of the 155 test rides
> in question...
> ...
>
> Hmmm . . . the 2-week averages are within 0.37, 1.35, and 0.20 mph of
> the averages for the year so far. That's reassuring.
>
> But it's not too reassuring that the hills came out in the right
> order, fastest to slowest, A-B-C, only 6 out of 14 times.
>
> Nor is it reassuring that speed on the fastest hill varied from 35 to
> 47 mph.
>
> Nor is it reassuring that speed on the slowest hill varied only 3.4
> mph, even though they were ridden within 30 minutes of each other.
>
> (The slowest hill is a steep S-bend down a gully and sheltered from
> most of the wind. But it still varies 3.4 mph in 14 runs, about 10%.)
>
> Is the variation due to the wind, the initial speed before coasting,
> what the rider ate for breakfast, how quickly he got into his tuck,
> how well he held it, how tightly his shirt was tucked in, the
> temperature, the barometric pressure, or what?
>
> If you didn't know that the same tires were used, what would you
> conclude about the rolling resistance?
You're purposely witholding much information - especially, the wind
direction and velocity, and the temperatures, and perhaps the initial
velocity. My bet is that there was no attempt to ride in consistent
conditions.
Do you seriously think there would be that much variation if you
purposely chose weather conditions to be as consistent as possible? I
don't.
> I agree with Jobst that a spin-down test on a smooth drum indoors is
> likely to be far more accurate and reliable than a test that adds a
> rider, his shirt-tail, whatever he ate for breakfast, whether he's
> hunching his shoulders up against the cold, and whatever wind may be
> blowing on a course several football fields long.
It's likely to be more _precise_. That I'll give you. (Technically,
there is a difference between precision and accuracy.)
It could be said to be more accurate, if everyone agrees that what
you're attempting to determine is rolling resistance on a perfectly
smooth surface.
Is it more representative of what a rider on a real road feels, or
(more importantly) must overcome through his input power? I doubt
that.
Admittedly, if the Bicycle Quarterly tests were done haphazardly,
they'd be worthless. But it appears the testers were aware of
confounding factors and tried mightily to overcome them. I've become
aware of the technical expertise of some of those guys, and it's
impressive. I doubt you're thinking of anything they didn't think of.
It also appears they _did_ overcome the confounding factors. How else
to explain the precision - that is, consistency - of their results?
- Frank Krygowski
frkr...@gmail.com
jobst....@stanfordalumni.org wrote:
> Jan Heine writes:
>
> ... about testing tires with a bicycle and rider, outdoors on an
> uncontrolled surface.
>
> I find the whole process faulty because neither the bicycle nor the
> rider have anything to do with rolling resistance but have a lot to do
> with aerodynamic losses and can influence results by expectation of
> good or bad performance.
Well, the bicycle was a constant. Its aero losses fall out of the
equation, so to speak.
The rider could have unconsciously affected results, based on his
expectations. But it would take a remarkably sophisticated
subconscious to pull it off. "Hmm. I think this tire's slower, so I'm
going to hold my elbow out an extra inch or two, and do it almost
perfectly consistently on each of three different runs..." Seems
unlikely to me.
> There is no reason to repeat runs three times if sufficient accurate
> data is collected.
Usually, as in this case, multiple tests serve to tell if the variables
are adequately controlled. When repeatability is good, it's an
indication that variables are adequately controlled.
> As I said, the RR tests done by IRC are the best
> and most credible ones I have seen. These curves don't need
> testimonials or reviewers like Papadopoulos, Berto, or Oehler. They
> stand on their own merit as anyone who understands what causes RR can
> see.
Seems like we've got a "conflicting experts" situation!
Explain why you think tire losses have to vary consistently despite
differences in deflection. That is, why would the loss-vs-deflection
function be the same for all tires, despite differences in
construction?
I believe I could design a tire with low rolling resistance on smooth
surfaces, but relatively higher rolling resistance on rougher roads
(compared to a reference tire). Could you not do the same?
Hell, I could do it without taking the easy way out - that is, solid
metal "tires" with a thin rubber coating.
Don't we all agree those would roll wonderfully on a steel drum, but
badly on asphalt?
> > Jan Heine
> > Editor
> > Bicycle Quarterly
>
> I don't doubt that the editor of this magazine should stand by the
> published findings, but that doesn't convince me.
Well, we shouldn't throw charges of deliberate bias. ISTR you've been
involved in tire design - probably using steel drum tests, no? Someone
might take that as a hint of bias on your part.
Let's stick with the physics.
- Frank Krygowski
carl...@comcast.net
Dear Frank,
Jobst puts it better than I can:
[Jan wrote:]
"We watched leaves of trees nearby for movement, plus watched dust we
kicked up in the air. If the dust cloud did not move, then we assumed
there was no wind."
[Jobst replied:]
"Oh please stop. Here is a test that involves watching dust and
leaves for accuracy."
But I'll give in to the urge and try to explain better.
Consistency is hardly unknown in unblinded tests where various people
are thumbing stopwatches (harder than you'd think, as Tim McNamara
points out) and other people are trying to maintain precisely the same
bicycle tuck day after day (harder than you'd think, as any artist's
model can tell you).
I don't doubt their good faith. Indeed, I doubt that I'd do as well.
But if the wind varies half a mile per hour in various directions from
0 mph over a city block or so, the test has serious problems. That's
about one-eighth the speed of a brisk 4 mph walk down the street, far
less than needed to move tree leaves or cause noticeable movement with
dust or smoke.
So turn the matter around.
Do you know of any accurate way to measure wind speed and direction to
within 0.5 mph in calm conditions for a 600-foot stretch of road for
the 30 seconds in which we're interested?
I don't.
We can't feel a 0.5 mph wind on our face or hands. Most ultrasonic
wind speed instruments specify higher thresholds for readings.
At this point, many people mistakenly try breathing very gently on
their hands and think that they can detect the motion of the air.
But they're actually noticing the much more impressive temperature
increase as the hot, humid air from their lungs heats the skin.
A better test is to raise your hand slowly, sweaty palm down, and see
if you can feel the wind.
Half a mile per hour is about 9 inches per second, so count
one-thousand-one while moving your hand smoothly from the bottom to
the top the computer screen while counting one-thousand-one.
You probably can't feel a thing.
Now lick your finger and move it to test the same breeze.
Again, it's the temperature change that we feel as the spit on the
windward side of our finger evaporates under the wild rush of 0.5 mph
air.
Wave your finger briefly in the air to encourage things to dry on all
sides, and you won't be able to feel the same wind half a minute
later.
You can even just twist your head from side to side slowly. The end of
your nose is moving about 0.5 mph, but unless it's wet, you won't
notice it.
Of course, the wind may be doing something quite different a city
block away from you, and at various points between. If you've ever
seen smudge-pots on a grass air-field at dawn (hideous memory!), you
know how the wind moves, even in what you'd swear is a dead calm.
Or enlist the aid of a smoker for an early morning walk.
I'd like to have a credible coasting test, but this keeps reminding me
of the two accurate digital thermometers on the same side of my house,
both hanging under broad eaves about eight feet above the ground and
thirty feet apart. Next to each other in the same room, they give
identical measurements.
Right now, about seven hours after sunset, one says 33.8F, but the
other says that it's 32.9F thirty feet away.
Local variation, as you may remember, is one of my favorite themes.
Cheers,
Carl Fogel
carl...@comcast.net
Dear Peter,
You wrote, "Frankly, I'm also surprised that the data looks as good as
it does, but it just doesn't show the amount of random fluctuation
that would be expected if the variable wind and shirt-tail effects you
mention were having a significant effect."
Like you, I don't think that they falsified their data.
But unblinded tests involving several people that appear to produce
data better than seems possible, given obvious confounding factors,
are not unknown.
Nor is it unusual for people to assume that blinding cannot possibly
affect whatever testing is in question.
If my objections are mistaken or answered elsewhere, other people
should be able to demonstrate differences as small as gloves in
coasting tests.
This seems unlikely, since imperceptible changes in posture from day
to day should be larger than gloves.
What does it suggest if unblinded testers and riders can measure the
difference between rolling hundreds and hundreds of feet with and
without gloves, but can't say if there was a wind up to 1 mph, head,
tail, or side?
Lest there be any misunderstanding, I don't know whether Jan is right
or wrong, but I absolutely don't think that he's fudging any data. He
explained without hesitation that the test just assumed that there was
no wind if leaves on trees and dust kicked up didn't seem to move.
So I'm skeptical, but I'm willing to change my mind. Is anyone willing
to try to measure the coasting difference between gloves and no gloves
on a similar course?
Cheers,
Carl Fogel
Tim Binns
>Jan Heine writes:
>
>... about testing tires with a bicycle and rider, outdoors on an
>uncontrolled surface.
>
>I find the whole process faulty because neither the bicycle nor the
>rider have anything to do with rolling resistance but have a lot to do
>with aerodynamic losses and can influence results by expectation of
>good or bad performance.
Surely that's like saying, back in the day of the IRC tests...
"Well we know what causes energy loss in tyres: Tubs have the
thinnest cases with the least lossy rubber component and they run at
higher pressures, so they will have the lowest RR - no point doing
comparative tests..."
...which overlooks the viscous losses in tub glue - a part of the
system in which they function.
Aren't the rider and the road surface also important factors in the
system?
>
>There is no reason to repeat runs three times if sufficient accurate
>data is collected. As I said, the RR tests done by IRC are the best
>and most credible ones I have seen. These curves don't need
>testimonials or reviewers like Papadopoulos, Berto, or Oehler. They
>stand on their own merit as anyone who understands what causes RR can
>see.
>
Would you expect to see the same sort of findings if the RR tests
were repeated on a drum featuring an ISO standard roughened surface*,
with the wheel being applied to the drum via a sprung and damped mass
which approximates an ISO standard rider's** load?
* Which doesn't, of course, exist AFAIK.
** Neither does this...
carl...@comcast.net
[snip]
>The rider could have unconsciously affected results, based on his
>expectations. But it would take a remarkably sophisticated
>subconscious to pull it off. "Hmm. I think this tire's slower, so I'm
>going to hold my elbow out an extra inch or two, and do it almost
>perfectly consistently on each of three different runs..." Seems
>unlikely to me.
[snip]
>Let's stick with the physics.
>
>- Frank Krygowski
Dear Frank,
I'd like to stick with the physics, since the near-0 wind speed
question intrigues me, but endless examples should teach us that we
can't stick with the physics when people are doing the testing.
In addition to the rider, there are also the people thumbing the stop
watches and deciding if enough of a gentle gust of air warranted
tossing out data. I assume that they were all doing their level best
to be objective. I'm also willing to be convinced that they succeeded.
But the most notable thing about unblinded testing of tricky matters
is the sincere belief of its victims that it isn't going to bite them
on the ass.
I have embarrassing toothmarks from several incidents to remind me. I
still can't explain how my careful measurements could have been so
mistaken.
In the current thread about toasting spoke bends, I'd love to have
someone else barbecue a few spokes and come up with similar but
somewhat different results.
I'd feel reassured if someone else fumbled through the same process
and came up with roughly the same results, but disagreed with me on
some point and highlighted a difference that would otherwise never
occur to me.
But what I'm really hoping is that no one will aim a propane torch at
a bent spoke and prove that I must have measured too few spokes or had
bad luck or unconsciously mis-measured small changes and that the
spokes don't really move the way that I could swear they do.
Jan put a lot of work into his test, and I'd be pleased to see it
confirmed, despite my doubts--it suggests a whole slew of new
theories.
But the measured differences seem very small, and they seem to depend
on the wind not rising to half a mile per hour in any direction while
a rider duplicates his posture so well from day to day that the
testers can tell whether he's wearing gloves.
Cheers,
Carl Fogel
M-gineering
> All RR tests by the auto industry are done on smooth drums,
that's probably great if you're interested in the rolling resistance of
a tyre on a smooth drum, but I find rollerriding extremely boring.
Rollers do not simulate real conditions well. If you put a car on a
rolling road dynamometer the first thing to do is to pump up the tyres.
If you want to do high speed testing I believe it is usual to run the
tyre INSIDE a drum.
Still it would have been nice if the experiments had been carried out
with more precise measuring equipment, as in
http://www.fahrradzukunft.de/fz-0501/0501-03.htm
--
---
Marten Gerritsen
INFOapestaartjeM-GINEERINGpuntNL
www.m-gineering.nl
r...@stdout.org
hands on it.
People have been complaining about confounding factors, such as wind,
body position, clothing, etc. If the experiment is done well, it will
try to control for these things. But even if they're _not_ well
controlled, it's OK, as long as the error effects are random, and not
systematic.
There are two basic approaches to measurement. One is to get a
super-controlled environment and super-accurate measuring equipment.
You measure once, and that's the end of it. The other method is to
live with some noise, make multiple measurements, and use statistical
methods to find the true values. In both cases, you need to avoid
systematic error (confounds, eg., all tests with tire X were with an
unzipped jersey). This is the important part: with the statistical
method, random error (noise) is permitted. If there's a lot of noise,
then the precision of your estimate will be low, but the math will
reflect this.
Most people aren't familiar with the distinction between the two types
of measurement because we don't use the statistical measurements in our
daily lives. But it is essential to anyone who does science in a field
where it is impossible or prohibitively expensive to perfectly control
the conditions, and this seems like a perfect case.
If the error is random, then it's much more likely that the differences
are the result of the experimental manipulations (tire/pressure/tube)
than from accumulations of "lucky" random events. The probability can
be quantified, but I won't go into that here. (I don't know if the
authors did this or not, since I don't have a copy of the article.)
So to sum up, it's OK to have noisy data and imperfect measuring
devices, as long as you understand how to interpret the data. You
can't just say "factor Z wasn't perfectly controlled" and write off the
results completely. That said, it would be nice to have estimates of
variance so that we could tell just how much to read into the results.
If the article provides the raw data, then anyone could calculate these
values, I suppose.
Gary Young
If suspension is so important to speed, then why did the performance of
narrow tires degrade when their pressure was reduced? I don't have access
to the full study, but the summary above says, "Narrow 23-mm tires seem to
roll fastest at pressures of 105 psi (7.2 bar) or more. However, running
these tires at 85 psi (5.8 bar) for improved comfort increased the test
times only 2%."
Also, how do real-world conditions account for the fact that narrow
clinchers are faster at high pressures while narrow tubulars are slower at
high pressures? (From the summary: "Tubular tires perform worse at very
high pressure. At 130 psi (9 bar), the narrow Clement Criterium rolled
slower than it did at a more comfortable 105 psi.") It's difficult to see
how road conditions could contribute to those inverted results. If there's
something about the tires themselves that explains the results, wouldn't
the same pattern be observed on smooth rollers?
Gary Young
> A Bicycle Quarterly reader pointed me to the discussion here, and I
> read it with interest. A few points from the authors of the test:
>
> 1. The methodology of the test was carefully designed. It was reviewed
> by several experts in the field, including Jim Papadopoulos, Frank
> Berto and Andreas Oehler. The tests showed excellent reproducibility -
> the same tires tested at different times during the test to check this
> gave identical results. Each combination of tire/pressure/tube was
> tested at least 3 times to weed out erroneous results. Fortunately, we
> had very few - all wind-related, when a light gust of wind slowed the
> rider down. We felt the wind before the rolldown even had been
> completed. We only tested when there was no wind, of course.
> Temperatures also remained relatively constant during the test, an
> advantage of being in Seattle on an overcast day. The methodology is
> published in the article, for all to see. Same rider, same clothing,
> same position, same bike for all tests. We had doubts at first whether
> the rider could keep the same position (on the hoods, arms locked), but
> the reproducibility of the results indicates that he could.
Did you try to control for investigator bias? You're an advocate for
reviving an older style of bicycle that's fallen out of fashion (as the
inclusion of so many 650B tires in your sample indicates). Did your rider
and the other timekeeper share that preference? If not, did you
communicate to them that that was your preference? Was the rider aware of
what type of tire he was riding? Were you or your other timekeeper aware
of what tire was being tested on any given run?
Investigator and test-subject bias could account for the consistency of
your results. All it would take is for your rider to adopt a slightly
different position or for the timekeepers to click the stopwatch slower
or faster depending on what tire was being tested. (Note that I am not
suggesting bad faith on your part -- I believe it's very common for biases
to manifest themselves subconsciously.)
Granted, it's difficult to design a study in which the subjects and
investigators are both in a blind, but not entirely unachievable in this
case. You could, for instance, put really, really full-coverage fenders on
the bike and have a third investigator mount tires at random. Did you do
something of that nature? Also, I would have more confidence in your
results if the timing had been done purely by instrumentation or at least
someone with no stake in the outcome.
>
> 2. The pressure testing was done consecutively, by starting with high
> pressures, after three runs reducing it, etc. The tires slowed down
> significantly only after pressures were reached that correspond to about
> 15% tire drop (according to Frank Berto's data).
>
> 3. Jobst's assertion that infinite pressure approaches zero rolling
> resistance is true for steel rollers. On a steel roller, a bare rim
> (which deforms only very little as it rolls) would have the lowest
> rolling resistance. On real roads, it would not, otherwise, we'd still
> ride bare rims with a thin layer of rubber for traction. Jobst, have you
> measured a bare rim for comparison?
>
> Steel railroad wheels have lowest resistance on steel surfaces. On a
> steel drum, any bike tire that approaches the hardness of railroad
> wheels will perform best. Try rolling a railroad wheel (or an iron-clad
> cartwheel) on a real road!
>
> 4. When pneumatic tires were invented, speeds went way up, because the
> pneumatics added suspension to the bike. When your bike bounces, moving
> the bike upward takes energy. That energy no longer is available to the
> forward motion. When the bike falls back down, much of that energy is
> NOT returned to forward motion, but simply "squishes down" the tire,
> that is, causes additional hysteretic losses.
>
Then why did narrow clinchers go faster at high pressures?
> 5. Many have questioned the results that tire choice has a great impact
> on your speed, because they believe that wind resistance is all that
> matters. This of course depends on your speed. The German magazine TOUR
> tested 700C x 23 mm racing tires on Continental's steel drum, and found
> rolling resistance coefficients between 0.0038 and 0.0076. As these are
> from steel drum tests, they probably underestimate the differences
> between tires (a "supple" tire with low hysteretic losses also absorbs
> shock better, so has a double advantage on real roads). But even if
> taken at face value, and plugged into analyticcycling.com for speeds of
> 15 mph, you will see that they result in speed differences of about 10%,
> depending on your other parameters. We tested a wide variety of tires on
> a real road, so differences of 20% are not all that surprising. (BTW,
> Tour's fastest tire also was the fastest in our tests.)
>
> 4. I am amazed that Jobst concluded our test methodology was invalid
> without even having seen it. If the methodology is incorrect, or our
> interpretation of the data has problems, please do let us know. Then we
> can redesign the test. Instead, he'll probably find some place in this
> where I used incorrect terminology, and claim victory.
>
> On the other hand, the steel drum test methodology makes some
> simplifying assumptions, and it does not appear that the effect of these
> assumptions on the results has been tested.
>
> 5. The Bicycle Quarterly tests were performed by three people, at least
> two of whom have Ph.Ds. in research-oriented fields. We know how to
> design tests and how to interpret data. The methods and results were
> reviewed by experts in the field who are not associated in the bike
> industry. None of us have a stake in the outcome of the tests. (Well, I
> do sell a few 650B tires, but the tires I sell tested in the middle
> ground, even though they "feel" fast. So I didn't really gain anything
> from publishing the results.)
>
It's incredibly naive to think that the prospect of financial gain is the
only thing that could introduce bias into a study. How about vindicating
the type of bicycles you love? (By the way, I share many of your
preferences, but I'm still doubtful they don't slow me down.)
Tim McNamara
Tim Binns <binn...@gmail.com> wrote:
> Aren't the rider and the road surface also important factors in the
> system?
In practical daily experience, yes. However, if you want to know how
the tire performs you want to isolate its performance from those other
factors. This is of course standard scientific practice.
A guy riding a bike down the road is an extremely complex system to
describe mathematically- a chaotic system, when you consider the random
grained roughness of the road surface, the constant movement of the
rider, variation in local wind effects, etc. That all creates too much
noise to isolate the contribution of the tire's performance.
What Jan and his cohorts did was to try to minimize the non-tire
factors. They used a roll-down test on a soapbox derby course to
eliminate the vagaries of pedaling. They used one bike for all the
tires with a couple of wheelsets, which they calibrated against each
other using a standard reference set of tires. They used their standard
reference tires as a check against changes in conditions (and found, in
the process, that bike tires roll faster when they are warm compared to
cold).
The question they were trying to answer was how tires compared in
performance on asphalt rather than a smooth steel drum. Their
conclusion was that the steel drum testing method favors certain design
elements that may not be favored on the road, such as tire pressure and
hardness of the rubber. The question is whether the measured
differences are greater than the error of measurement or are greater
than chance. Repeatability does argue that the result are non-random.
I think that Jan and his associates made a great effort and I want to be
appreciative of that, while still admitting my uncertainty as to the
reliability and validity of the results due to the risks of confounds.
Some of their findings compare qualitatively with other studies. For
example, Jobst's IRC tests for Avocet showed that there is a point of
diminishing returns for reducing rolling resistance with higher
pressures. Bicycle Quarterly's tests suggested the same thing.
Jan, you mentioned in one of your posts that a statistical analysis of
the data was done. This wasn't discussed in your article (e.g.,
standard deviation, error of measurement, confidence intervals,
regressions and the like). Perhaps that might be something to offer as
an appended article on your Web site. A factor analysis could be very
interesting. It sounds as though you have enough data points to make
that worthwhile.
And, really, for people critiquing the study it would help if you
actually read the study first. Buy a copy of the magazine! I find it
to be the most interesting bike reading I do, far better than any of the
mainline bike magazines. There are, from my perspective, three
interesting bike magazines that actually give me something to think
about and don't publish "Seven Days to Great Legs" articles- (V)BQ by
Jan, Velovision by Peter Eland, and A to B Magazine by David and Jane
Henshaw. The rest of the bike publishing industry is pretty much dreck
IMHO. Cycling Plus is interesting sometimes, too, especially the
various ride descriptions.
Tim McNamara
"peter" <prat...@comcast.net> wrote:
> There's a big difference between attempting to estimate wind speed
> and looking for a period of essentially no wind. Depending on the
> type of leaf, it can indeed detect winds of 0.5 mph or less.
Quaking aspen are very sensitive to wind. I don't know if they grow
around Seattle.
Tim McNamara
> In the absence of evidence to the contrary, I'm willing to trust the
> authors that they didn't deliberately falsify the data
I have no doubt whatsoever in this. I've never met Jan but I've
corresponded with him and know people who have met him. I think he's an
honest guy and have no doubts that he accurately reported the results
they got. He also went so far as to have a degree of peer review for
the article. One of the reviewers, Jim Papadopoulos, helped rewrite
_Bicycling Science_ a couple of years ago. He used to post here, once
upon a time.
http://ruina.tam.cornell.edu/research/topics/bicycle_mechanics/overview_p
apers_and_links.htm
or
Frank Berto should be well-known to cyclists over the age of 30, anyway.
The other guy- Andreas Oehlberg or something like that- is unfamiliar to
me except seeing him mentioned once in _Velovision_.
Tim McNamara
r...@stdout.org wrote:
> Most people aren't familiar with the distinction between the two
> types of measurement because we don't use the statistical
> measurements in our daily lives.
Not formally but we do intuitively use statistics daily. We understand
intuitively the ideas of central tendency and margin of error, for
example. But most people when confronted with a t-test or linear
regression are out to sea as to what they mean.
One of Jan's posts indicated that such analysis had been done. It is
not in the article in the usual statistical fashion, that not being the
interest of his audience I suspect. One of his associates is apparently
a statistician. I have suggested in another post that that data could
be posted to Jan's Web site for interested parties to read.
BTW, getting a copy of the magazine is easy:
frkr...@gmail.com
M-gineering wrote:
>
> Still it would have been nice if the experiments had been carried out
> with more precise measuring equipment, as in
> http://www.fahrradzukunft.de/fz-0501/0501-03.htm
I don't read German, so I'm just assuming the tricycle cart shown is
what's used to measure tire rolling resistance. I've heard of such
rigs before. But I'm not convinced that they do a better job of
evaluating tires for real riding. Here's why.
When a rigid cart (as that appears to be) hits a little bump or
roughness, the only energy losses would be from tire hysteresis. At
least, that's all that occurs to me now. There is no damping in the
rest of the rigid cart.
OTOH, a human rider on a bike has lots of internal damping. ISTM that
the motion of the bike upward into the human body will cause lots of
internal squishing, sloshing, muscle work, etc. that dissipate that
mechanical energy immediately. Keeping the rider's mass stationary and
un-jolted would have great benefit.
On a somewhat related matter, I recall an article in Bicycling magazine
many decades ago. The author tested coasting speed down a dirt trial
with big bumps using a bike with primitive suspension (maybe a Yamaha
BMX bike??) and compared it to the same bike with suspension locked.
The suspended version was significantly faster over the bumps. And of
course, all downhill racing bikes these days come with large suspension
travel.
What we're discussing is the same thing, but with the bumps on a
smaller scale.
- Frank Krygowski
Johnny Sunset
Jan Heine wrote:
> ...
> My preferences are evolving based on what my research shows. However,
> while some are surprised that we found wider tires to roll faster, that
> is not inconsistent with previous results. So now I am torn between
> having my next custom bike made for 700C x 24.5 (actual width) tires
> that tested fastest both in our test as well as in the TOUR steel drum
> test, and waiting for the perfect, hand-made 650B x 38 mm tires....
Of course, the lowest rolling resistance tire may not be the fastest in
the real world if it produces a harsher ride. Not only will vibration
contribute to rider fatigue, but it also may contribute to the rider
deliberately slowing down due to decreased confidence in the handling
of the bicycle on poor surfaces.
This would be hard to prove, but intuitively it would be expected that
racers in competition would be less affected by a harsh ride than
touring, commuting and recreational cyclists.
--
Tom Sherman - Post Free or Die!
bfd
hei...@earthlink.net wrote:
> My preferences are evolving based on what my research shows. However,
> while some are surprised that we found wider tires to roll faster, that
> is not inconsistent with previous results. So now I am torn between
> having my next custom bike made for 700C x 24.5 (actual width) tires
> that tested fastest both in our test as well as in the TOUR steel drum
> test, and waiting for the perfect, hand-made 650B x 38 mm tires.
>
How likely is it that a "perfect, hand-made 650B x 38 mm tire" will
ever be made? Most people consider DUGAST tubular tires, which come in
both cotton and silk casing, to be "the best." Since DUGAST is french
and 650B is really a french size and has been around like forever, if
he hasn't made a 650B x 38 mm wide tire by now, when will he? Further,
who else makes "hand-made" tires?
Personally, I like Avocet Fasgrip 700x25 tires (formerly 700x28) and
pumped up to about 100psi find them to be "the best" for me. These
tires just roll well and at 100psi are very, very comfortable. Also,
with a 25mm wide tire, it handles better than most 23mm, the best of
all worlds!
Gary Young
<snip>
> And, really, for people critiquing the study it would help if you
> actually read the study first. Buy a copy of the magazine! I find it
> to be the most interesting bike reading I do, far better than any of the
> mainline bike magazines. There are, from my perspective, three
> interesting bike magazines that actually give me something to think
> about and don't publish "Seven Days to Great Legs" articles- (V)BQ by
> Jan, Velovision by Peter Eland, and A to B Magazine by David and Jane
> Henshaw. The rest of the bike publishing industry is pretty much dreck
> IMHO. Cycling Plus is interesting sometimes, too, especially the
> various ride descriptions.
I haven't read Velovision or VBQ in a couple of years. One of the reasons
I no longer read them is that both displayed a certain credulity--or at
least lack of rigor--when it came to the bikes under discussion. See the
following rbt threads for examples of what I mean:
http://tinyurl.com/ydzoz6 (Velovision)
That second thread concerns an article that appeared in the Rivendell
Reader, but it was written by Heine.
In my mind, the burden is on Heine to prove his scientific bona fides
before I'll shell out $10 for his magazine. Besides, as you've mentioned
in another posting, the magazine doesn't contain the raw data.
It is encouraging that Heine now sees the need for more rigor, though I
remain skeptical of his results.
By the way, I'm surprised that you didn't include the Rivendell Reader as
one of the magazines worthy of consideration. Is Grant no longer
publishing it?
jobst....@stanfordalumni.org
>> ... about testing tires with a bicycle and rider, outdoors on an
>> uncontrolled surface.
>> I find the whole process faulty because neither the bicycle nor the
>> rider have anything to do with rolling resistance but have a lot to do
>> with aerodynamic losses and can influence results by expectation of
>> good or bad performance.
> Well, the bicycle was a constant. Its aero losses fall out of the
> equation, so to speak.
So? Why include a bicycle in a test of tires?
> The rider could have unconsciously affected results, based on his
> expectations. But it would take a remarkably sophisticated
> subconscious to pull it off. "Hmm. I think this tire's slower, so
> I'm going to hold my elbow out an extra inch or two, and do it
> almost perfectly consistently on each of three different runs..."
> Seems unlikely to me.
It's not that complex. I see it often when coasting down hills while
watching the speedometer. Small changes in position make significant
changes in speed.
>> There is no reason to repeat runs three times if sufficient
>> accurate data is collected.
> Usually, as in this case, multiple tests serve to tell if the
> variables are adequately controlled. When repeatability is good,
> it's an indication that variables are adequately controlled.
As is evident in the IRC test, the consistency of curves and the few
anomalous data points are readily apparent, without wind effect or
rider position. These tests were done not for IRC but for Avocet,
that wanted a reading on a range of tires to see whether slicks have
lower RR than patterned tread.
>> As I said, the RR tests done by IRC are the best and most credible
>> ones I have seen. These curves don't need testimonials or
>> reviewers like Papadopoulos, Berto, or Oehler. They stand on their
>> own merit as anyone who understands what causes RR can see.
> Seems like we've got a "conflicting experts" situation!
What is the conflict?
> Explain why you think tire losses have to vary consistently despite
> differences in deflection. That is, why would the
> loss-vs-deflection function be the same for all tires, despite
> differences in construction?
Rolling losses come from flexing elastomers, the tread, the inter-ply
binders and the inner tube. A small amount arises from casing cord
stretch. All of these effects result essentially from bending the
tire body entering and leaving the road contact. A rougher surface
increases that bending just as lower inflation pressure does.
If the road surface contains pointed chip-seal aggregate then tread
squirm increases and does so for all tires regardless of tread
pattern. In contrast, typical hot mix asphalt roads have river bottom
(rounded) aggregate and are rolled to a planar surface, the reason why
users, automotive and bicycles, prefer this type of road.
Paved road irregularities are not large enough to cause lift-off or
even nearly so unless a truly bumpy (dirt like) road was chosen, which
I doubt. Therefore, the bounce scenario is invalid. I recall that
Jim Papadopoulos proposed the bounce scenario a while back here on
wreck.bike, something that is true for cattle guards with the
appropriate rib spacing but not for paved toads.
Pneumatic tires make that concept fall n its face. Pavement
irregularities are random and don't show up as saw-tooth patterns with
regular frequency. The best example of this is riding on tractor
cleat indentation in asphalt or even worse, wake-up ribs on roadsides.
> I believe I could design a tire with low rolling resistance on
> smooth surfaces, but relatively higher rolling resistance on rougher
> roads (compared to a reference tire). Could you not do the same?
I doubt it. You'll need to explain what you would do to achieve that.
> Hell, I could do it without taking the easy way out - that is, solid
> metal "tires" with a thin rubber coating.
That is not a pneumatic tire and it won't change with inflation
pressure. Tossing in ridiculous examples does not improve your point
of view on this.
> Don't we all agree those would roll wonderfully on a steel drum, but
> badly on asphalt?
Not funny!
>> I don't doubt that the editor of this magazine should stand by the
>> published findings, but that doesn't convince me.
> Well, we shouldn't throw charges of deliberate bias. ISTR you've
> been involved in tire design - probably using steel drum tests, no?
> Someone might take that as a hint of bias on your part.
Where is deliberate bias mentioned? I find fault with the
experimental process that seems to come from shying away from the cost
of proper testing.
> Let's stick with the physics.
I didn't see that in what you wrote.
Jobst Brandt
Solvang Cyclist
news:f-6dndk1gO--isvY...@giganews.com:
> By the way, I'm surprised that you didn't include the Rivendell Reader
> as one of the magazines worthy of consideration. Is Grant no longer
> publishing it?
>
Last weekend I rode in the Solvang Prelude and in the "goodie bag"
handed out at the end was a Rivendell catalog. It indicates that they
are in the 12 th year of publishing the quarterly (at best <sic>)
Rivendell Reader.
If it's anything like most of the articles and sidebars in their
catalog, I think I'll pass. They do have a few interesting and useful
products, but it seems most of the text is just daffy justifications for
their retro product mix. I certainly have nothing against what they sell
(in fact I will probably buy some stuff from them) but I find the text
to be a complete joke and certainly without any scientific basis. And,
yes I do know that some of it is intended to be tongue-in-cheek, but
other than for its high entertainment value, it's NOT worthy of
consideration. (IMHO of course.)
Cheers,
David
M-gineering
> M-gineering wrote:
>> Still it would have been nice if the experiments had been carried out
>> with more precise measuring equipment, as in
>> http://www.fahrradzukunft.de/fz-0501/0501-03.htm
>
> I don't read German, so I'm just assuming the tricycle cart shown is
> what's used to measure tire rolling resistance.
the tricycle isn't the interesting bit. They used a small calculator to
log the time for each revolution of the wheel, so they could deduce the
air and rolling resistance with fairly good accuracy after several
coastdown tests.
peter
> Rolling losses come from flexing elastomers, the tread, the inter-ply
> binders and the inner tube. A small amount arises from casing cord
> stretch. All of these effects result essentially from bending the
> tire body entering and leaving the road contact. A rougher surface
> increases that bending just as lower inflation pressure does.
>
> If the road surface contains pointed chip-seal aggregate then tread
> squirm increases and does so for all tires regardless of tread
> pattern. In contrast, typical hot mix asphalt roads have river bottom
> (rounded) aggregate and are rolled to a planar surface, the reason why
> users, automotive and bicycles, prefer this type of road.
> ...
> Paved road irregularities are not large enough to cause lift-off or
> even nearly so unless a truly bumpy (dirt like) road was chosen, which
> I doubt. Therefore, the bounce scenario is invalid.
No "lift-off" needs to occur for there to be energy losses. When I
ride on a typical chip-seal road I clearly feel vibrations in my hands,
feet, and rear end. Those vibrations are damped by my body and
represent lost energy that must result in the bike slowing down. Since
the vibrations are much more pronounced when riding on 20 mm wide tires
inflated to 120 psi than when riding on 32 mm wide tires at 75 psi they
will contribute more to the rolling resistance in the former case than
the latter. How significant these energy losses are compared to those
in the tire/tube material is not clear to me, but seems worth
investigating. Insisting on a testing regimen, like either the steel
drum arrangement or the German tricycle cart, that excludes this effect
may well lead to erroneous conclusions.
jobst....@stanfordalumni.org
What you are saying is that at higher inflation pressure you can feel
road texture better than at low pressure. Any losses that occur
jiggling the flesh of your body is not part of a tire rolling
resistance assessment but peculiar to each rider. Besides, whether
you can feel vibration or not does not mean the amplitude is
significant. Descending fast on smooth pavement, tires make a rushing
sound from smooth road texture, tiny as it is. That same resonance
can be felt by the rider if focused upon.
This has descended into hair splitting asusual... but what if...?
Jobst Brandt
Gary Young
>> Did you try to control for investigator bias? You're an advocate for
>> reviving an older style of bicycle that's fallen out of fashion (as the
>> inclusion of so many 650B tires in your sample indicates). Did your rider
>> and the other timekeeper share that preference? If not, did you
>> communicate to them that that was your preference? Was the rider aware of
>> what type of tire he was riding? Were you or your other timekeeper aware
>> of what tire was being tested on any given run?
>
> The rider was aware what type of tire he was riding, but not at which
> speed he was coasting.
So what? I don't have to be aware of my exact speed to know that raising my
torso a little will slow me down.
> The timekeepers did not look at the watch until
> the test was completed.
In other words, you did know beforehand which tire was being
tested? It would help your credibility with me if you would respond
directly to my points instead of trying to shift the discussion.
> Of course, all could have tried to influence
> the results, but a few things speak against this.
>
> 1. Our results show that modern, narrow racing clinchers roll faster
> than wide tires from various manufacturers. Both the 700C tires I
> usually ride and the 650B tires I sell rated below average.
>
If that was the result you anticipated going into the experiment, then it
doesn't really rule out investigator bias.
> 2. The tires that I'd like to
see aren't even available, so we could not
> test them. Or does anybody know of 38 mm wide tires with a cotton
> casing?
Oh come on. In the absence of your ideal, you have no preferences or
prejudices about the tires that are currently available? They're all just
an indistinguishable mass? Why do you sell 650B tires if you don't have
some preference for them? I suppose you could be trying to strike it rich
that way, but I'm doubtful that's your real motivation.
>
> 3. While randonneuring is not a race, it is somewhat competitive. I am
> interested personally in making my bike faster. I have ridden 700C x 23
> mm tires in several long events this year, rather than the 700C x 28 mm
> that I usually use.
That tells me that your motives are mixed, not that you have no bias. As
for them canceling out, wouldn't that depend on how strong each one is?
I want to reiterate that I'm not suggesting that you are uniquely biased.
These are influences that all scientists have to guard against.
>
>
>> Investigator and test-subject bias could account for the consistency of
>> your results. All it would take is for your rider to adopt a slightly
>> different position or for the timekeepers to click the stopwatch slower
>> or faster depending on what tire was being tested. (Note that I am not
>> suggesting bad faith on your part -- I believe it's very common for
>> biases to manifest themselves subconsciously.)
>
> It would be hard to be consistent with that.
>
I expect you're right if you were consciously trying to bias the results.
If it happens by way of an unconscious process, however, I'm not so sure.
Maybe Tim McNamara could speak to that issue. Also, I suppose an analysis
of the raw data could shed some light on this.
> Regarding the position: From experience, I know that the same full aero
> tuck is difficult to replicate. But a on the hoods, arms nearly locked
> position is much more consistent. The gloves on/gloves off was
> discovered after a few runs. The times were about 0.3-0.5 seconds
> slower. If it had not been for the obvious - no gloves on my rider's
> hands - we would have accepted the times as normal scatter. So
> basically, I need to clarify that in a double-blind test, we would not
> have discovered the lack of gloves, as it was at the margins of what we
> considered significant differences.
I may be misunderstanding, but how could you attribute the 0.3 to 0.5
second difference to gloves unless you compared runs on which the rider
used the same tires? The fact that the rider maintained a constant
position when the type of tire was not at issue doesn't really address the
question of whether he subconsciously varied his position when it comes to
the factor that was at the very heart of your study. In other words, I
would expect bias to play more of a role during the main event, the one in
which there is the most emotional investment.
>
> Statistical analyses: My co-author did all those, and found the results
> to be highly significant. We did regression analyses for pressure,
> width, weight, tire thickness (tread and casing), and found that of all
> those, only width and tire thickness were significant, but those factors
> alone could not explain the differences we saw.
I have only a layman's understanding of statistics, but you seem to be
saying that your results for tire pressure were not statistically
significant. Only width and tire thickness were significant? Don't you see
that as a problem?
>Looking at our tires and
> data, it became obvious that tread pattern plays a role. Some tires have
> small tread blocks that seem to act as micro-knobbies, while other have
> more continuous treads that do not appear to affect performance as much.
> The remainder of the differences, which still is quite large, appears to
> be due to tire construction - i.e., the materials used.
>> >
>> > 4. When pneumatic tires were invented, speeds went way up, because
>> > the pneumatics added suspension to the bike. When your bike bounces,
>> > moving the bike upward takes energy. That energy no longer is
>> > available to the forward motion. When the bike falls back down, much
>> > of that energy is NOT returned to forward motion, but simply
>> > "squishes down" the tire, that is, causes additional hysteretic
>> > losses.
>> >
>> Then why did narrow clinchers go faster at high pressures?
>>
>>
> Because you need high pressures to get the 15% drop with narrow
> clinchers. Here is the data from Berto, for our 85 kg rider-cum-bike
> weight: 23 mm: 95 psi; 27 mm 70 psi, 38 mm: 40 psi.
>
> We found that the 23 mm tires were 0.4-0.5 seconds faster at 105 psi
> than at 85 psi (2 tire combinations tested at those widths and
> pressures). Considering the accuracy of our measurements, that is not
> inconsistent with the figures above for tire drop.
>
> Our 37 mm tires showed the same speed at 55 psi and 35 psi, but were
> significantly slower at 25 psi.
>
> Our 27 mm tires were the same at 105 and 85 psi, but significantly
> slower (0.8 sec) at 55 psi, and even slower (1.8 sec) at 35 psi.
>
> Because the pressure results were unexpected, we had not looked at tire
> drop before we made the measurements. So we did not try to hone in on
> that 15% drop value.
>
I'm not sure I understand what you mean by the 15% drop. Elsewhere you've
suggested that as a tire's pressure approaches infinity (that is, it
becomes more like a railroad wheel), it will be subject to more bounce and
therefore be slower. Do narrow clinchers cease to bounce above a
certain pressure? If there truly is a sweet spot for pressure, then don't
you have to come up with another explanation (other than the bounce
theory) for what's going on?
> It is noteworthy that while the higher pressures on
clinchers did not
> yield _significantly_ faster times, all our averages were faster for
> higher pressures.
What do you mean by "not ... significantly"? That the differences were
statistically insignificant? Or that the differences were insignificant in
duration?
> For example, 27 mm tire from 85 to 105 psi was 0.2
> seconds faster. 37 mm tires from 35 to 55 psi: 0.1 second faster. With
> the exception of the tubulars, none were slower at higher pressures. We
> could have done a statistical analysis to see whether this effect was
> significant, but the differences are so small that we did not bother.
How do you explain the results for tubulars? Also, it seems to me that you
may be conflating two very different things -- statistical significance
and significance in terms of duration or speed. How can you have any
confidence in your findings that increased tire pressures yield little in
the way of speed if you don't know whether your results are statistically
significant? Again, I'm not a statistician, but it does seem to me that
just because a result is small in magnitude doesn't mean that it shouldn't
be subject to testing for statistical significance.
> For riders, choosing their tires wisely has a greater effect than
> pressure.
>
>> It's incredibly naive to think that the prospect of financial gain is
>> the only thing that could introduce bias into a study. How about
>> vindicating the type of bicycles you love? (By the way, I share many of
>> your preferences, but I'm still doubtful they don't slow me down.)
>
> My preferences are evolving based on what my research shows. However,
> while some are surprised that we found wider tires to roll faster, that
> is not inconsistent with previous results. So now I am torn between
> having my next custom bike made for 700C x 24.5 (actual width) tires
> that tested fastest both in our test as well as in the TOUR steel drum
> test, and waiting for the perfect, hand-made 650B x 38 mm tires.
>
> Finally, the steel drum tests have not been verified on the road. They
> make simplifying assumptions (perfectly smooth surface), but nobody has
> verified that these assumptions do not affect the results. In fact, our
> tests could be seen as that - an attempt to replicate the results of the
> steel drum tests on real roads. And while you can argue with individual
> results, such as whether tire A is 0.2 seconds slower than tire B, I
> know the pressure data holds up to statistical scrutiny.
Above you said first that pressure results were not statistically
significant and then that you didn't subject some of
your pressure results to significance testing. Can you clarify this?
> So you could
> say that the first test done to validate steel drum testing was a
> failure, in that it showed that factors influencing resistance on a
> steel drum are different from those affecting it on a real road.
>
> Beyond that, I think we have outlined clear, plausible mechanisms that
> explain our results, and that explain why steel drum tests
> insufficiently replicate the resistance of tires on real roads.
To my knowledge, you haven't explained why tire pressures are relatively
unimportant until tire pressures exceed the 85% mark, at which point they
seem to overcome the bouncing problem. Also, this effect is more
pronounced in narrow clinchers than in wide clinchers and is entirely
absent from narrow tubulars -- for reasons I don't see you explaining.
> The ball
> is in the court of the drum testers to show that their method is valid.
> I have seen no attempts to show this. The drum testers so far have
> resorted to authority, and to belittling the efforts of our tests.
Of course they've resorted to authority. The only alternative to
resorting to authority is for each of us to replicate every scientific
study for himself before reaching any conclusions. Is it a matter of
indifference to you whether a study is published in a peer-review journal
or not? Whether it's done by a professional or an amateur? You yourself
resort to authority when you talk about the qualifications of your
assistants.
> Just
> because we use simple means does not mean our tests are flawed.
It does if they're too simple to capture the relevant
information and distinguish causes. Besides I think I've suggested a
couple of fairly simple ways to improve your studies -- for instance,
keeping the rider and/or the timekeepers in the dark about which tire is
being tested on a given run.
> Just
> because IRC or Michelin use drum tests doesn't mean it's a good way to
> test tires. In fact, both the Michelin Pro2 Race and the Avocet Duro,
> which have been optimized on steel drums, scored much less well in our
> real-road tests, indicating that the current approach is not the best.
> (They both were fine tires, but not as excellent as steel drum tests
> would have us believe.)
>
> Even if our test were flawed, the drum tests are even more flawed, yet
> they are used as reference.
Oh come on. If your tests are valid, then you're probably right that the
drum tests are flawed. If your tests were flawed, on what basis do you
claim that the drum tests are even more flawed?
>
> The history of cycling is full of the disagreements between the
> "theoreticians" and the "practicians." In the 1890s, there used to be
> mathematicians, who tried to calculate and predict how a bicycle
> handled. In the end, the "practicians," who developed front-end
> geometries by trial and error, came up with better bikes - the ones we
> use today.
Do you have a reference for that? I'm inclined to disagree with you based
on Sharp's book on bicycles and tricycles. IIRC, Sharp singled out a lot
of production bikes for poor design and used his technical skills to show
the superiority of the double triangle frame.
> Their experiments may have been poorly controlled, but their
> results were better. As a rider, I care mostly how bikes behave on the
> road. My first question with all assumptions, whether riderless bikes or
> steel drum tests, is whether the results are relevant for real riding.
> And in many cases, while I understand that simplifying assumptions are
> necessary, I also know as a scientist that if they aren't tested, then
> you run the risk to get results of limited use in your studies.
>
> With bicycle handling, I don't see a way to model the input of the
> riders, but of course, the fact that a rider can make minute course
> corrections means that if a riderless bike exhibits poor stability with
> one geometry, that means very little on the road. I am perfectly willing
> to move the handlebars (instinctively) a tad every few seconds... With
> tires, however, I do not see any problems (beyond expense) to create
> steel rollers with a more representative surface. The ball is in the
> drum testers' court, so to speak.
>
> With that, I sign off. Got to get working on the next issue of our
> magazine.
>
> Jan Heine
> Editor
> Bicycle Quarterly
> 140 Lakeside Ave #C
> Seattle WA 98122
> www.bikequarterly.com
peter
> Peter Rathman writes:
> > No "lift-off" needs to occur for there to be energy losses. When I
> > ride on a typical chip-seal road I clearly feel vibrations in my
> > hands, feet, and rear end. Those vibrations are damped by my body
> > and represent lost energy that must result in the bike slowing down.
> > Since the vibrations are much more pronounced when riding on 20 mm
> > wide tires inflated to 120 psi than when riding on 32 mm wide tires
> > at 75 psi they will contribute more to the rolling resistance in the
> > former case than the latter. How significant these energy losses
> > are compared to those in the tire/tube material is not clear to me,
> > but seems worth investigating. Insisting on a testing regimen, like
> > either the steel drum arrangement or the German tricycle cart, that
> > excludes this effect may well lead to erroneous conclusions.
>
> What you are saying is that at higher inflation pressure you can feel
> road texture better than at low pressure. Any losses that occur
> jiggling the flesh of your body is not part of a tire rolling
> resistance assessment but peculiar to each rider.
Earlier you stated "If you could show me where the energy goes that
makes the pavement test produce a different hierarchy among samples, I
would be interested." Since then several posters have stated where the
energy goes - it is mainly dissipated in the rider's body. Every
cyclist can feel that there is mechanical energy input to and damped by
their body when riding on rough pavement surfaces and that the amount
of this energy varies depending on the tire width and pressure. Yet
you continue to insist that the only valid method of testing the drag
associated with different tires is one that deliberately excludes any
consideration of this mechanism for energy loss.
> Besides, whether
> you can feel vibration or not does not mean the amplitude is
> significant.
Agreed, as I had already stated. But that's why experiments to measure
rolling resistance including losses in the rider's body should be
encouraged so the significance of this factor can be evaluated.
jobst....@stanfordalumni.org
Oh shit! I guess I'll have to be wordier and write like a legal
document. So where does the energy go in the tire being tested.
From your explanation, a suspension seat post would improve tire
rolling resistance. I think this is getting entirely out of the
scientific assessment of RR. What you are proposing is an entirely
different parameter and not that of a reasonable tire test. In all
the miles I have ridden, RR plays no role in rough road riding and a
lot to do with touring over many miles of paved roads.
You might propose a rough road test with bumps and rocks that meet
your definition of rolling performance. Obviously high pressure
tires, that have so little RR that it is insignificant won't play a
role in this test.
This reminds me of the endless go-around about the effects of wind
drag on a rider, in which endless diversions about mechanical losses
in the drive train and tires was found to be lacking. The wind
analysis stands on the effects of wind. You can add in all the rider
effects you wish but in a different analysis. The same goes for RR.
Let's separate the variables and have a clear picture of what occurs.
>> Besides, whether you can feel vibration or not does not mean the
>> amplitude is significant.
> Agreed, as I had already stated. But that's why experiments to
> measure rolling resistance including losses in the rider's body
> should be encouraged so the significance of this factor can be
> evaluated.
I find that method of testing obscure and not representative of what a
tire does on good roads. People interested in these values are TT
riders and other competitive folks who do not care about rough road
effects but rather how well the tire rolls on a smooth road.
Jobst Brandt
r...@stdout.org
> > Statistical analyses: My co-author did all those, and found the results
> > to be highly significant. We did regression analyses for pressure,
> > width, weight, tire thickness (tread and casing), and found that of all
> > those, only width and tire thickness were significant, but those factors
> > alone could not explain the differences we saw.
>
> I have only a layman's understanding of statistics, but you seem to be
> saying that your results for tire pressure were not statistically
> significant. Only width and tire thickness were significant? Don't you see
> that as a problem?
"Statistical significance" is different from the ordinary use of the
word "significant". The two words can often go together, but they
don't mean the same thing.
The ordinary word just means "a lot". If you fed hormone X to kids and
they ended up 6 inches taller on average, that's significant in the
ordinary sense.
Statistical significance means something else: suppose you fed hormone
Y to kids and they ended up 0.5 inches taller on average. Given enough
measurements, the stats can tell you, for example, that it is 95%
probable that the differences in height are attributable to hormone Y
(assuming no confounds) and not random noise. The key is the number of
measurements, and how much noise is present in those measurements.
If there is an effect, but the effect size is small relative to the
noise, then it requires more measurements to reach statistical
significance. That's what's going on here. The results might only
have, for example, an 80% probability that the differences are due to
pressure differences (the remaining 20% represents the probability that
the measured differences are due to random factors). It doesn't mean
that the difference isn't real; it just means that there's not enough
data to say with authority whether the measured differences are due to
pressure or not. So no, their findings aren't a problem.
> > It is noteworthy that while the higher pressures on clinchers did not
> > yield _significantly_ faster times, all our averages were faster for
> > higher pressures.
>
> What do you mean by "not ... significantly"? That the differences were
> statistically insignificant? Or that the differences were insignificant in
> duration?
>
> > For example, 27 mm tire from 85 to 105 psi was 0.2
> > seconds faster. 37 mm tires from 35 to 55 psi: 0.1 second faster. With
> > the exception of the tubulars, none were slower at higher pressures. We
> > could have done a statistical analysis to see whether this effect was
> > significant, but the differences are so small that we did not bother.
>
> How do you explain the results for tubulars? Also, it seems to me that you
> may be conflating two very different things -- statistical significance
> and significance in terms of duration or speed. How can you have any
> confidence in your findings that increased tire pressures yield little in
> the way of speed if you don't know whether your results are statistically
> significant? Again, I'm not a statistician, but it does seem to me that
> just because a result is small in magnitude doesn't mean that it shouldn't
> be subject to testing for statistical significance.
Sometimes when you get your data, it's obvious that a test for
statistical significance isn't going to meet your 95% or 99%
probability threshold. That is, if there _is_ an effect, it's
obviously too small to be confident about. This happens when the
effect is small relative to noise. That's how he knows that that
pressure effect is small.
Just because you don't have enough data to meet that threshold, that
doesn't mean that there's no effect. It just means you don't have
enough data to be confident that there's a real difference. For
example, say you want to find out whether there's a height difference
between 12-year-old boys and girls. If you measure three boys and
three girls, you may find a difference in average height, but you can't
be very confident that this difference represents a real difference
between boys and girls in general, or if it's due to chance. If you
measure 100 boys and 100 girls, then you can be more confident that any
differences you find are real.
peter
I frankly don't care which tire dissipates the most energy internally
but rather in how my choice of tires will affect the total energy lost
while riding since it's the latter that will determine how fast I can
ride and how many miles I can cover in a day's worth of touring.
For that matter, the glue used to attach tubular tires is also not
strictly speaking part of the tire itself. So should effects related
to energy lost in the glue layer also be excluded from consideration?
I'd argue against that since the tire choice inevitably will affect the
need for such glue and associated energy loss and similarly, the tire
choice will affect how much energy is lost in vibration and impacts
transmitted to the rider's body while riding on typical road surfaces.
>
> From your explanation, a suspension seat post would improve tire
> rolling resistance. I think this is getting entirely out of the
> scientific assessment of RR. What you are proposing is an entirely
> different parameter and not that of a reasonable tire test. In all
> the miles I have ridden, RR plays no role in rough road riding and a
> lot to do with touring over many miles of paved roads.
Nothing in the cited test by Bicycle Quarterly nor anything that I've
posted here mentioned particularly rough roads and certainly nothing
about unpaved ones. OTOH, the paved roads on which I ride rarely have
much resemblance to the surface of a polished steel drum. You're the
only one bringing various strawmen into the discussion such as rocks,
cattle guards, etc.
> This reminds me of the endless go-around about the effects of wind
> drag on a rider, in which endless diversions about mechanical losses
> in the drive train and tires was found to be lacking. The wind
> analysis stands on the effects of wind. You can add in all the rider
> effects you wish but in a different analysis. The same goes for RR.
> Let's separate the variables and have a clear picture of what occurs.
>
> >> Besides, whether you can feel vibration or not does not mean the
> >> amplitude is significant.
>
> > Agreed, as I had already stated. But that's why experiments to
> > measure rolling resistance including losses in the rider's body
> > should be encouraged so the significance of this factor can be
> > evaluated.
>
> I find that method of testing obscure and not representative of what a
> tire does on good roads. People interested in these values are TT
> riders and other competitive folks who do not care about rough road
> effects but rather how well the tire rolls on a smooth road.
I'd expect TT riders to be more concerned about the added wind
resistance of wider tires since at their competitive speeds that would
be sufficient to overcome any slight rolling resistance difference.
The tests under discussion were done at speeds more representative of
touring cyclists and were done on a rather smooth paved road surface,
albeit not as smooth as polished metal. Rolling resistance generally
plays a bigger role for long-distance tourists and randonneurs than it
does for TT riders.
Phil Holman
<carl...@comcast.net> wrote in message
news:s6sal2lgnfjqlnd22...@4ax.com...
> On 10 Nov 2006 22:10:15 -0800, frkr...@gmail.com wrote:
>
>>
>>carl...@comcast.net wrote:
>>>
>>Carl, I think you're overestimating the following differences.
>>>
>>> Dear Tim,
>>>
>>> The same clothes can be worn differently, particularly if it isn't a
>>> tight-fitting skin suit.
>>
>>But would they be worn much differently if the testing crew was trying
>>to guard against that?
>>
>>> Similarly, the claims for the same position aren't very credible.
>>> Raising or lowering your chin an inch is going to make a difference
>>> in
>>> wind drag
>>
>>Measure one inch tilt at your chin. It's an easily perceptible
>>difference. ISTM that, if you're going to pretend a cyclist can't
>>sufficiently control that variable, you'll have to tell Lance
>>Armstrong
>>and his compatriots that wind tunnel testing is useless. Given the
>>success that technique has apparently generated, I think they'll
>>disagree.
>>
>>> I'm willing to believe that the bicycle can be duplicated to an
>>> impressive degree on repeated runs.
>>>
>>> I'm much less willing to believe that the chin angle, amount the
>>> shoulders are raised or lowered, knee spread, and angle of the crank
>>> are going to be duplicated to the same degree. ...
>>
>>
>>>
>>> Here's the maximum mph for the same bicycle and rider, trying to
>>> tuck
>>> in and coast as fast as possible, on the same three hills for the
>>> last
>>> two weeks. That's 42 test rides, about a quarter of the 155 test
>>> rides
>>> in question...
>>> ...
>>>
>>> Hmmm . . . the 2-week averages are within 0.37, 1.35, and 0.20 mph
>>> of
>>> the averages for the year so far. That's reassuring.
>>>
>>> But it's not too reassuring that the hills came out in the right
>>> order, fastest to slowest, A-B-C, only 6 out of 14 times.
>>>
>>> Nor is it reassuring that speed on the fastest hill varied from 35
>>> to
>>> 47 mph.
>>>
>>> Nor is it reassuring that speed on the slowest hill varied only 3.4
>>> mph, even though they were ridden within 30 minutes of each other.
>>>
>>> (The slowest hill is a steep S-bend down a gully and sheltered from
>>> most of the wind. But it still varies 3.4 mph in 14 runs, about
>>> 10%.)
>>>
>>> Is the variation due to the wind, the initial speed before coasting,
>>> what the rider ate for breakfast, how quickly he got into his tuck,
>>> how well he held it, how tightly his shirt was tucked in, the
>>> temperature, the barometric pressure, or what?
>>>
>>> If you didn't know that the same tires were used, what would you
>>> conclude about the rolling resistance?
>>
>>You're purposely witholding much information - especially, the wind
>>direction and velocity, and the temperatures, and perhaps the initial
>>velocity. My bet is that there was no attempt to ride in consistent
>>conditions.
>>
>>Do you seriously think there would be that much variation if you
>>purposely chose weather conditions to be as consistent as possible? I
>>don't.
>>
>>> I agree with Jobst that a spin-down test on a smooth drum indoors is
>>> likely to be far more accurate and reliable than a test that adds a
>>> rider, his shirt-tail, whatever he ate for breakfast, whether he's
>>> hunching his shoulders up against the cold, and whatever wind may be
>>> blowing on a course several football fields long.
>>
>>It's likely to be more _precise_. That I'll give you. (Technically,
>>there is a difference between precision and accuracy.)
>>
>>It could be said to be more accurate, if everyone agrees that what
>>you're attempting to determine is rolling resistance on a perfectly
>>smooth surface.
>>
>>Is it more representative of what a rider on a real road feels, or
>>(more importantly) must overcome through his input power? I doubt
>>that.
>>
>>Admittedly, if the Bicycle Quarterly tests were done haphazardly,
>>they'd be worthless. But it appears the testers were aware of
>>confounding factors and tried mightily to overcome them. I've become
>>aware of the technical expertise of some of those guys, and it's
>>impressive. I doubt you're thinking of anything they didn't think of.
>>
>>It also appears they _did_ overcome the confounding factors. How else
>>to explain the precision - that is, consistency - of their results?
>>
>>- Frank Krygowski
>
> Dear Frank,
>
>
> [Jan wrote:]
>
> "We watched leaves of trees nearby for movement, plus watched dust we
> kicked up in the air. If the dust cloud did not move, then we assumed
> there was no wind."
>
> [Jobst replied:]
>
> "Oh please stop. Here is a test that involves watching dust and
> leaves for accuracy."
>
> But I'll give in to the urge and try to explain better.
>
> Consistency is hardly unknown in unblinded tests where various people
> are thumbing stopwatches (harder than you'd think, as Tim McNamara
> points out) and other people are trying to maintain precisely the same
> bicycle tuck day after day (harder than you'd think, as any artist's
> model can tell you).
>
> I don't doubt their good faith. Indeed, I doubt that I'd do as well.
>
> But if the wind varies half a mile per hour in various directions from
> 0 mph over a city block or so, the test has serious problems. That's
> about one-eighth the speed of a brisk 4 mph walk down the street, far
> less than needed to move tree leaves or cause noticeable movement with
> dust or smoke.
>
> So turn the matter around.
>
> Do you know of any accurate way to measure wind speed and direction to
> within 0.5 mph in calm conditions for a 600-foot stretch of road for
> the 30 seconds in which we're interested?
>
Carl, you don't need to. Why would you be doing this test on a windy day
anyway? If you do enough runs where you alternate two tires, a
comparison of each 'mean' will go a long way to reducing the effect of
random wind velocity changes. The more runs, the higher the confidence
that any difference is meaningful.
Personally, I prefer the real world test than some academic measure
that doesn't necessarily apply to riding on the road. I remember letting
30 psi out of my tires on a freshly chip-sealed TT course to go faster
(Fort Smith, AR during Master's Nats). The rough road was too difficult
to ride at speed (too much vibration and jarring) at my normal pressure.
The losses incurred by jiggling, wobbling body parts are not
recoverable.
Phil H
Phil Holman
<jobst....@stanfordalumni.org> wrote in message
news:45567d99$0$34549$742e...@news.sonic.net...
I could suggest a new thread titled "how tire pressure can affect rider
performance". It seems quite a few posters care less about the exact
measurement of the precise definition of tire rolling resistance and
more about how this will affect their "on the road" performance or ride
quality.
Phil H
diann...@yahoo.com
> Still it would have been nice if the experiments had been carried out
> with more precise measuring equipment, as in
> http://www.fahrradzukunft.de/fz-0501/0501-03.htm
> --
> ---
> Marten Gerritsen
>
> INFOapestaartjeM-GINEERINGpuntNL
> www.m-gineering.nl
I've made a translation using Altavista and my school years german
studies, and I know I've got some errors, but here's a stab anyway.
Please fel free to correct the mistakes:
=========================================
The Case for Good Tires
Thomas Senkel, Uni Oldenburg
from Pro Velo 32
How quickly your bicycle goes depends on the one hand, on your own
power, and on the other hand on the retarding forces which must be
overcome. If you consider constant speed on a level road, then there
are only the rolling resistance and aerodynamic drag that consume the
rider's forces (the friction losses in the drive train can usually be
neglected).
Now there are at least two kinds of cyclists: one want to go as fast as
possible, and the other wants to ride with as little effort as
possible. In fact both depend on the same thing (the reduction of the
sum of all riding resistances), however the first category (the racers)
pays special attention to air resistance, which has been well
researched, fiddled and written about. However, below the speed of
approximately 4.5 m/s (16 km/h), rolling resistance is greater than air
resistance and therefore deserves more interest from the everyday rider
than has been paid so far. In particular, riding with a trailer and
load wheels the rolling friction constitutes the lion's share of the
total resistance.
In Diagram 1 the solid curves give the speed attainable for a given
power output for two different tires (based on a bicycle with rider of
m=100 kg and Cw*A=0,5 m²). From the dotted curve one can read off the
speed increase with a given power, if you lower the coefficient of
rolling resistance from Cr=0,00568 to Cr=0,00160. This is roughly the
difference between a rather bad and a very good result from measurement
of the coefficient of rolling resistance (in each case 20" tire with
the same pressure of 5 bars). At 75 Watts (the continuous power of an
average rider) one could then ride 12% faster. Even at 200 W power the
increase still amounts to over 6%.
So far the development of bicycle tires is left, to a large extent, to
the industry; an exception is the engineer Paul Rinkowski in Leipzig,
whose hand-made radial ply tires should be pointed out [ 5 ]. Some tire
manufacturers measure the rolling resistance of their tires on
laboratory test stands and use the results in optimization as well.
However the rolling resistance is only rarely quantitatively used as a
sales pitch; that may be related to the fact that because of the
variability of the measuring procedures used a direct comparison of the
data is not a given. Beyond that it seems to be nevertheless rather the
appearance of the tires, the steep lugs and schliddrigen Sliks, which
fall salespromoting to the eye.
Since a basis for the development of a theory of bicycle tire rolling
resistance measurements interested us (physicists and physics students
of the bicycle research work group) as much as possible different tire.
More than two years ago, in the context of a study on arbitrary roadway
surfaces [ 1 ], we perfected a measuring procedure that makes realistic
measurements possible (with high accuracy).
The Rollout Method
The measurements take place with a measuring tricycle particularly
designed for it. The ORM (Oldenburger Rollwiderstands-Meßgerät, or
Oldenburg rolling resistance measuring instrument) is pushed by hand
and rolls then unpowered over a measuring section.
Unpowered measuring tricycle ORM (Oldenburger
Rollwiderstands-Meßgerät)
With the help of a pocket computer the time for each wheel revolution
is measured and stored with each run. From the time differences between
every two wheel revolutions the negative acceleration can be computed,
and with the help of the traction resistance equation, the entire
traction resistance can be computed. At the end then over all
individual values are averaged. Since the air resistance and the
rolling friction of two of the tricycle's tires are known from
calibration measurements, the rolling friction of an individual unknown
tire can be determined.
In order to compensate for the influence of the elevation profile of
the test section, the test section is run in both directions, and the
measured values are evaluated in pairs. The load of the measured tire
amounts to about 55 kg, which corresponds to the usual wheel loads of a
two-wheeler. For correct appraisal, the accurate wheel circumference,
the wheel loads, the moment of inertia of the wheels and the
atmospheric pressure must be ascertained. In addition it must be as
windless as possible; therefore we accomplished the measurements in a
closed corridor, whose surface consisted of PVC lining on a high-duty
concrete cover.
How Rolling Resistance Develops
The rolling friction of a tire is given by Fr = Cr*m*g. The coefficient
of rolling resistance Cr indicates by what fraction of the weight the
rolling wheel is slowed. The larger the Cr and the heavier the bike
(and/or the riders), the more heavily it rolls. There is still no
assured theory about how the rolling resistance develops in the tire,
but one assumes that it consists mainly of two components:
Unreeling Resistance
The unreeling resistance results from the fact that the tire does not
affect the roadway in a mathematical point, but stands on an elliptical
surface. The unreeling procedure can be regarded as constant tilting
over an imaginary unreeling edge within the road-contact area. The
longer the road-contact area and the smaller the wheel diameter, the
larger the force required.
Pull Resistance
When unreeling, the tire constantly moves a rubber bulge over from the
bearing surface like a belly before itself. The tire is flex-leveled
due to the wheel rotation. The energy of the flexible deformation is
not completely returned by the absorption of the tire material. These
absorption losses are proportional to the sinking depth [ 2 ].
Results
Table 1 shows the coefficients of rolling resistance measured so far,
arranged according to tire size. One example of each tire was measured,
and the results are reproducible with the indicated error. However,
whether the results apply to all tires of the same type is
questionable; the extent of production variation is unknown, and also
the age and the degree of wear of the tires affect the rolling
friction. Nearly all measured tires were new. Naturally, this selection
of tires is to a large extent coincidental and far from complete.
Table 1
Naturally when buying a tire it is crucial to consider the rolling
resistance together with other important criteria e.g.:
Breakdown safety
Lifetime
Cushioning capacity
Riding dynamics
Road adherence
Price
In Diagram 2 one sees the dependence of the coefficient of rolling
resistance on the tire pressure for a Continental Top Touring 37-622.
At low pressures an increase in inflation pressure becomes much more
strongly apparent than with high pressures. Weakly inflated tires
cushion well, but have a clearly higher rolling resistance than with
nominal pressure. In contrast, at high pressure the difference between
e.g. 800 and 900 kPa (8 and 9 bar) is minor.
Diagram 1: Speed increase through better tires
Diagram 2: Pressure dependence of rolling resistance
For the dependence of rolling resistance on the tire diameter Schwalbe
Standard GW tires with Profil HS159 were at our disposal, which all
have the same structure and the same width of 47 mm. Diagram 3 shows
that rolling resistance is inversely proportional to the diameter.
Diagramm 3: Dependence of Cr on tire diameter
A remarkable and unexpected result is the fact that - with the same
design and same parameters - wide tires have less rolling resistance
than narrow tires (Table 1). This seems to contradict the conventional
wisdom, but on closer inspection it becomes clearer: With equivalent
pressure the road-contact area of both tires is of equivalent size.
However with the wide tire the ellipse is shorter and broader and thus
the unreeling resistance smaller. The reason narrow tires became
generally accepted in racing is primarily because of the better air
resistance, which becomes apparent at higher speeds, and the lighter
weight. For these reasons no high-quality (high-pressure) tires have
been produced in larger widths.
Naturally rolling resistance also depends on the road surface. For
typical cycle tracks made from concrete blocks or bitumen the values
are about 20-50% higher than on PVC floor. With very soft surfaces,
e.g. grass or sand, the deformations of the soil are the main cause for
the rolling resistance, which then amounts to a multiple of the values
on PVC. Also the rank order of of the tires can greatly change among
themselves, so Table 1 does not have significance for very soft
surfaces.
In order to achieve the least possible rolling resistance Cr, a tire
should have the following characteristics:
good flexibility
small contact patch
large cross section
For a small contact patch one needs:
high pressure capability
large width
high stiffness of the tire walls
large profile positive portion
These characteristics are partially contradictory; for example a
flexible tire also has lower stiffness [ 2 ].
Results
The measured values show that the tires have quite different rolling
resistances and it is therefore assumed that there is substantial
potential for research and innovation. In particular Rinkowskis
experiments with radial ply tires of small diameter and large width can
be cause - perhaps also for the tire industry - to be concerned with
new construction principles since the rolling resistance of these tires
amounts to nearly half that of conventional bias-ply tires despite the
poor production conditions.
Literature
Hauschild, A., Senkel, T.: Messung des Rollwiderstandes mit dem ORM;
Studienarbeit 1990, Universität Oldenburg
Gerdes, J., Wieting, P: Die Abhängigkeit des
Rollwiderstandskoeffizienten Cr von Reifenparametern; Studienarbeit
1991, Universität Oldenburg
Krieger, R: Die Fahrradbereifung; Pro Velo 24 (1991), S. 20 ff.
Reimpell, J.: Fahrwerktechnik Räder und Reifen; Vogel Fachbuch,
Würzburg 1986
Rinkowski, P: Dokumentation zur Herstellung von ZT-Radialschlauchreifen
aus Stahlcord, Leipzig 1975
======================================
Tim McNamara
Gary Young <garyy...@gmail.com> wrote:
> I expect you're right if you were consciously trying to bias the
> results. If it happens by way of an unconscious process, however, I'm
> not so sure. Maybe Tim McNamara could speak to that issue. Also, I
> suppose an analysis of the raw data could shed some light on this.
Well, if there was an unintended bias, I would expect the rest results
to support the tires the testers like, and to not provide results that
are unexpected. When you set up an experiment and get exactly the
results you expect, it's time to take a look at biases. In this case,
the tires that "felt" fast to ride were not necessarily fast against the
watch.
frkr...@gmail.com
> Frank Krygowski writes:
>
> >> ... about testing tires with a bicycle and rider, outdoors on an
> >> uncontrolled surface.
>
> >> I find the whole process faulty because neither the bicycle nor the
> >> rider have anything to do with rolling resistance but have a lot to do
> >> with aerodynamic losses and can influence results by expectation of
> >> good or bad performance.
>
> > Well, the bicycle was a constant. Its aero losses fall out of the
> > equation, so to speak.
>
> So? Why include a bicycle in a test of tires?
:-) Let's see: one reason might be because they are bicycle tires!
I think the disagreement is this, in a nutshell: You think the [V]BQ
test includes too much of the actual riding situation to be controlled.
Others (myself included) think smooth drum tests include not enough to
be truly representative.
>
> > The rider could have unconsciously affected results, based on his
> > expectations. But it would take a remarkably sophisticated
> > subconscious to pull it off. "Hmm. I think this tire's slower, so
> > I'm going to hold my elbow out an extra inch or two, and do it
> > almost perfectly consistently on each of three different runs..."
> > Seems unlikely to me.
>
> It's not that complex. I see it often when coasting down hills while
> watching the speedometer. Small changes in position make significant
> changes in speed.
I fully understand that. The tricky part is not merely changing
position subconsciously. The tricky part is changing position
subconsciously in such a precise way that successive runs display
excellent precision, as these did.
> >> As I said, the RR tests done by IRC are the best and most credible
> >> ones I have seen. These curves don't need testimonials or
> >> reviewers like Papadopoulos, Berto, or Oehler. They stand on their
> >> own merit as anyone who understands what causes RR can see.
>
> > Seems like we've got a "conflicting experts" situation!
>
> What is the conflict?
It _seems_ Papadopoulos, Berto, and Oehler believe there is reasonable
validity to the test - although of course, we haven't hear from them.
It seems Brandt thinks there is no validity to the test. Sounds like a
conflict to me!
>
> > Explain why you think tire losses have to vary consistently despite
> > differences in deflection. That is, why would the
> > loss-vs-deflection function be the same for all tires, despite
> > differences in construction?
>
> Rolling losses come from flexing elastomers, the tread, the inter-ply
> binders and the inner tube. A small amount arises from casing cord
> stretch. All of these effects result essentially from bending the
> tire body entering and leaving the road contact. A rougher surface
> increases that bending just as lower inflation pressure does.
>
> If the road surface contains pointed chip-seal aggregate then tread
> squirm increases and does so for all tires regardless of tread
> pattern. In contrast, typical hot mix asphalt roads have river bottom
> (rounded) aggregate and are rolled to a planar surface, the reason why
> users, automotive and bicycles, prefer this type of road.
>
> Paved road irregularities are not large enough to cause lift-off or
> even nearly so unless a truly bumpy (dirt like) road was chosen, which
> I doubt. Therefore, the bounce scenario is invalid.
I don't think that It has to be a literal bounce - that is, catching
air. As long as the bike and rider accelerate upward, there will be
hysteresis losses and other losses within the rider's mass.
> I recall that
> Jim Papadopoulos proposed the bounce scenario a while back here on
> wreck.bike, something that is true for cattle guards with the
> appropriate rib spacing but not for paved toads.
>
> Pneumatic tires make that concept fall n its face. Pavement
> irregularities are random and don't show up as saw-tooth patterns with
> regular frequency. The best example of this is riding on tractor
> cleat indentation in asphalt or even worse, wake-up ribs on roadsides.
I don't see why pavement roughness needs to be non-random to cause
losses.
I live in an area with a fairly weak tax base and frequent freeze-thaw
cycles. There is plenty of rough asphalt around here, and it remains
rough for years before being paved. At a certain point in its
"maturity" it frequently displays hundreds of random cracks.
I learned many, many years ago to avoid those patches of cracks. They
produce notably higher rolling resistance. Ditto for series of patched
potholes. Both are random. Both rattle bike and rider, but to
different degrees. Both slow the bike significantly.
It seems obvious to me that if I were on, say, thin-walled tires of 3"
cross section, and perhaps lower pressure, the transfer of energy from
the road to my body would be much less efficient. As in other matters
we've discussed, it's an impedance matching problem. The low stiffness
of such a tire would not efficiently transmit the high frequency bump
impulses; the tire would "float" over them better than a narrow, high
pressure tire.
This does not mean the big soft tire would have lower RR on all
surfaces. ISTM the problem is to match tire width and pressure to the
surface. Off-road 4-wheelers seem to use large, low pressure tires
because they encounter great roughness. Mountain bikers use narrower
tires with (usually) somewhat higher pressures. Bikes raced on board
tracks doublessly do best with very narrow, high pressure tires. The
question is, where in that spectrum is the typical rider's road
surface?
>
> > I believe I could design a tire with low rolling resistance on
> > smooth surfaces, but relatively higher rolling resistance on rougher
> > roads (compared to a reference tire). Could you not do the same?
>
> I doubt it. You'll need to explain what you would do to achieve that.
To design that tire, I would probably try a thin, hard rubber, slick
tread surface to minimize deflection, squirm and hysteresis losses in
response to the tiny irregularities of a smooth surface. But I'd build
thick, multi-ply sidewalls using low durometer rubber, thick and hard
fabric, and design in lousy adhesion between layers, so that larger
deflections would cause lossy materials to flex and eat energy. Such a
tire could be fine on a steel drum, lousy on a moderately rough road.
> > Hell, I could do it without taking the easy way out - that is, solid
> > metal "tires" with a thin rubber coating.
>
> That is not a pneumatic tire and it won't change with inflation
> pressure. Tossing in ridiculous examples does not improve your point
> of view on this.
It's not intended as a ridiculous example. It's intended to illustrate
the extreme case. In many cases, study of extremes can tell you the
direction more moderate changes will take you.
> > Don't we all agree those would roll wonderfully on a steel drum, but
> > badly on asphalt?
>
> Not funny!
Again, not intended to be funny! But if you won't accept that, don't
we all agree that a 500 psi, 1/8" cross section tire of extremely stiff
construction would likewise roll wonderfully on a steel drum, but badly
on asphalt? Substitute the small cross section and extreme pressure of
your choice.
If there is such a thing as a tire that's too narrow and high pressure
- something I think is obvious - then again, the task is to determine
what size and pressure and construction works best for normal roads.
I see no reason it has to be the same one that works best on a
perfectly smooth roller.
Again: It seems you think the [V]BQ test includes too much of the
actual riding situation to be controlled. Others (myself included)
think smooth drum tests include not enough to be truly representative.
But it seems the degree of control can be reasonably assessed by
examining the repeatability (or precision) of the results. And that
precision seemed quite good.
Why do you think the jostling of the rider's muscles and guts has no
effect?
- Frank Krygowski
Gary Young
So I was the one confounding the two senses of significance. I stand
corrected.
frkr...@gmail.com
> > makes the pavement test produce a different hierarchy among samples,
> > I would be interested." Since then several posters have stated
> > where the energy goes - it is mainly dissipated in the rider's body.
> > Every cyclist can feel that there is mechanical energy input to and
> > damped by their body when riding on rough pavement surfaces and that
> > the amount of this energy varies depending on the tire width and
> > pressure. Yet you continue to insist that the only valid method of
> > testing the drag associated with different tires is one that
> > deliberately excludes any consideration of this mechanism for energy
> > loss.
>
> Oh shit! I guess I'll have to be wordier and write like a legal
> document. So where does the energy go in the tire being tested.
>
> From your explanation, a suspension seat post would improve tire
> rolling resistance.
Perhaps what we're seeing is that the term "rolling resistance" is
overly limited, and inadequate to describe important real-world
effects.
But yes; a well-designed suspension seat post might decrease the energy
lost by vibration transmission to the rider's body. As I mentioned
earlier, suspended bikes roll faster over very rough terrain. These
mechanical forms of suspension are clearly valuable on roughness of
large enough magnitude.
On roughnesses of smaller magnitude - that is, the ordinary pavement we
normally ride - such suspension is provided only by the tires, and this
"tire suspension" is also clearly valuable.
The difference is merely a matter of scale! Just as a 2" cross section
mountain bike tire provides less than optimum suspension over big
bumps, a 23mm road bike tire provides less than optimum suspension over
certain smaller bumps. It seems the losses caused by this
less-than-optimum suspension _must_ be classes as rolling resistance -
unless we introduce a new category of energy loss into all bike energy
discussions. (See below.)
> I think this is getting entirely out of the
> scientific assessment of RR. What you are proposing is an entirely
> different parameter and not that of a reasonable tire test.
IIRC, in every discussion I've ever read of cycling energy consumption,
energy losses have been divided into three areas: Aerodynamic losses,
mechanical losses (i.e. bearings and chain), and rolling resistance
losses.
If the energy losses within the rider's body, caused by imperfect tire
suspension, are not to be grouped under "rolling resistance losses,"
where should we put them? They clearly don't fit the other two
choices!
If we don't put these energy losses into "rolling resistance," we need
to define a new category for energy loss. Admittedly, this component
of energy loss is problematic. I'd guess identical tires on identical
roads would give different results for a 130 pound, muscular rider of
5% body fat, versus a 200 pound softie with 20% body fat. We may need
to develop an ANSI standard bag of fat! But pretending these losses
don't occur at all - the effect of using only smooth drum data -
doesn't seem realistic.
> In all
> the miles I have ridden, RR plays no role in rough road riding and a
> lot to do with touring over many miles of paved roads.
I don't follow your meaning.
> You might propose a rough road test with bumps and rocks that meet
> your definition of rolling performance. Obviously high pressure
> tires, that have so little RR that it is insignificant won't play a
> role in this test.
Again, into which category would you put those losses?
>
> This reminds me of the endless go-around about the effects of wind
> drag on a rider, in which endless diversions about mechanical losses
> in the drive train and tires was found to be lacking. The wind
> analysis stands on the effects of wind. You can add in all the rider
> effects you wish but in a different analysis. The same goes for RR.
> Let's separate the variables and have a clear picture of what occurs.
Separating the variables is a very good idea, as long as no important
variables are discarded in the process. I think the standard model
discards a significant loss, which is generated by imperfect tire
suspension.
> I find [the VBQ] method of testing obscure and not representative of what a
> tire does on good roads. People interested in these values are TT
> riders and other competitive folks who do not care about rough road
> effects but rather how well the tire rolls on a smooth road.
I beg to differ! I'm a tourist and commuter, not a racer. But I'm
interested in those values.
And I'll note, anyone racing in my area _should_ be interested in RR on
rough roads. That's where our races are held, because those are the
kinds of roads we usually have.
- Frank Krygowski
Tim McNamara
Gary Young <garyy...@gmail.com> wrote:
> On Sat, 11 Nov 2006 09:26:42 -0600, Tim McNamara wrote:
>
> <snip>
> > And, really, for people critiquing the study it would help if you
> > actually read the study first. Buy a copy of the magazine! I find
> > it to be the most interesting bike reading I do, far better than
> > any of the mainline bike magazines. There are, from my
> > perspective, three interesting bike magazines that actually give me
> > something to think about and don't publish "Seven Days to Great
> > Legs" articles- (V)BQ by Jan, Velovision by Peter Eland, and A to B
> > Magazine by David and Jane Henshaw. The rest of the bike
> > publishing industry is pretty much dreck IMHO. Cycling Plus is
> > interesting sometimes, too, especially the various ride
> > descriptions.
>
> I haven't read Velovision or VBQ in a couple of years. One of the
> reasons I no longer read them is that both displayed a certain
> credulity--or at least lack of rigor--when it came to the bikes under
> discussion. See the following rbt threads for examples of what I
> mean:
>
> http://tinyurl.com/ydzoz6 (Velovision)
>
> http://tinyurl.com/yf8cyb
>
> That second thread concerns an article that appeared in the Rivendell
> Reader, but it was written by Heine.
Jan likes centerpull brakes and has bikes with brazed-on pivots for
them. Given where he lives, where mountain passes are available to
ride, brakes are important to him.
I just tried a bike today with brazed-on pivots, by an odd coincidence,
and found the centerpulls to be smoothly progressive and had great lever
feel. They were Dia Compes. I was impressed with them. The last bikes
I rode with centerpulls were my Schwinn Continental and my Viscount Pro,
both of which I had when I was in high school. They stopped fine, as I
recall.
Jobst's criticism of centerpulls and cantilevers, IIRC, is that of the
path of the brake pads being curved rather than nearly linear and normal
to the rim resulting in "sine error" or "cosine error" or some such
(displaying my lack of attention in high school geometry and trig and my
faulty memory in one sentence). As the pads wear on cantilever brakes,
they will eventually touch the tire sidewalls; as the pads wear on
cantilevers they will eventually slip under the rim and into the spokes.
Both of these are problems with potentially serious consequences.
I do not recall that Jobst has claimed these brakes don't work well in
terms of stopping the bike or lever feel (whereas Campy sidepulls have
long been criticized for poor stopping performance and high lever
effort. I've never noticed this myself, but then I am 6'4" and have
pretty strong hands).
> In my mind, the burden is on Heine to prove his scientific bona fides
> before I'll shell out $10 for his magazine. Besides, as you've
> mentioned in another posting, the magazine doesn't contain the raw
> data.
It contains some quantitative data, but not the statistical analysis.
And there is quite a bit of qualitative analysis. It's interesting
reading, even if you end up unconvinced. Thus far in these threads,
though, I seem to be one of the very few who has bothered to read the
article. That decreases the rigor in *our* discussion rather
significantly.
In terms of Jan Heine's personal scientific bona fides, Google is handy.
He has a Ph.D in, it appears, geological sciences. He was awarded a
NASA fellowship back in the mid-90s for studying climate change.
There's a nice photo of him with a core sample from 1995. He seems to
work or to have worked in the Department of Geological Sciences at the
University of Washington. He appears to be published in peer reviewed
journals. Not only that but he is fluent in three languages. Criminy,
I barely speak English.
> It is encouraging that Heine now sees the need for more rigor, though
> I remain skeptical of his results.
Actually, it's been pretty clear from issue one that Jan is fairly
rigorous in his thinking in general. He seems very organized and did
make an effort to have some peer review on the article from people with
engineering backgrounds. That said, the key to scientific research is
repeatability. Anyone care to duplicate the experiment or even part of
it?
> By the way, I'm surprised that you didn't include the Rivendell
> Reader as one of the magazines worthy of consideration. Is Grant no
> longer publishing it?
It is still published on its usual inconsistent basis and I still read
it and enjoy it. But the focus has shifted to mainly be about Rivendell
and its products, which is fine and don't get me wrong about that- I
think that is what the Rivendell Reader should be. It's still worth
reading IMO, but it's not a magazine so much as Rivendell's primary
marketing tool. Because of that I left it off my list. It is, however,
one of the four cycling periodicals to which I subscribe.
carl...@comcast.net
Dear Dianne,
Thanks--that reminds me of the Rinkowski tire, which is a tire option
on this calculator:
http://www.kreuzotter.de/english/espeed.htm
The rolling resistance coefficient is set to 0.00300. For comparison,
the calculator assumes 0.00600 narrow racing tires. 0.00550 for medium
width high-pressure slicks, and 0.00500 for wide high-pressure slicks.
Cheers,
Carl Fogel
Michael Press
<1163305753.0...@e3g2000cwe.googlegroups.com>,
frkr...@gmail.com wrote:
>
> jobst....@stanfordalumni.org wrote:
> > Peter Rathman writes:
> >
> > > Earlier you stated "If you could show me where the energy goes that
> > > makes the pavement test produce a different hierarchy among samples,
> > > I would be interested." Since then several posters have stated
> > > where the energy goes - it is mainly dissipated in the rider's body.
> > > Every cyclist can feel that there is mechanical energy input to and
> > > damped by their body when riding on rough pavement surfaces and that
> > > the amount of this energy varies depending on the tire width and
> > > pressure. Yet you continue to insist that the only valid method of
> > > testing the drag associated with different tires is one that
> > > deliberately excludes any consideration of this mechanism for energy
> > > loss.
> >
> > Oh shit! I guess I'll have to be wordier and write like a legal
> > document. So where does the energy go in the tire being tested.
> >
> > From your explanation, a suspension seat post would improve tire
> > rolling resistance.
>
> Perhaps what we're seeing is that the term "rolling resistance" is
> overly limited, and inadequate to describe important real-world
> effects.
I disagree strongly. Tire rolling resistance should be
narrowly defined so it can be measured simply,
accurately, and precisely. Any other dissipative
mechanism also needs to be narrowly defined so that it
too can be measured simply, accurately, and precisely.
Tire rolling resistance is the power dissipated flexing
the tire measured on a smooth surface.
Now define the dissipative mechanism you want to
discuss, but do not call it tire rolling resistance.
--
Michael Press
Gary Young
I didn't mean to revive the debate over centerpull brakes. I'm less
concerned with Heine's preference for centerpulls than with the reasons he
offered for his preference, reasons that seemed dubious to me when I
first started that thread and even more so at the
conclusion of the thread.
For instance, saying that centerpulls varied widely in mechanical
advantage and that flexiness in the upper part of centerpull arms doesn't
degrade their performance.
(By way of contrast, I think that in that same thread, Chalo offered what
seemed to me to be sound reasons why some riders might prefer centerpulls.)
Well, he certainly has more scientific credentials than I do (as
I've perhaps illustrated with my slip-ups in statistics in other parts of
this thread). Nonetheless, I'm mindful of the Linus Pauling factor --
competence in one area of science doesn't necessarily translate into
another area.
>
>> It is encouraging that Heine now sees the need for more rigor, though I
>> remain skeptical of his results.
>
> Actually, it's been pretty clear from issue one that Jan is fairly
> rigorous in his thinking in general. He seems very organized and did
> make an effort to have some peer review on the article from people with
> engineering backgrounds. That said, the key to scientific research is
> repeatability. Anyone care to duplicate the experiment or even part of
> it?
>
Having just been over to his website, I still question his rigor. For
instance, how rigorous can the Quarterly's peer-review process be if it
lets pass a review of a tire that reads like advertising copy:
http://www.vintagebicyclepress.com/images/vbqgrandbois.jpg
He acknowledges the difficulty of assessing rolling resistance while
riding, yet doesn't seem to question his subjective impressions of shock
absorption, cornering ability and road adhesion.
Michael Press
M-gineering <ikmotg...@m-gineering.nl> wrote:
> jobst....@stanfordalumni.org wrote:
>
> > All RR tests by the auto industry are done on smooth drums,
>
>
> that's probably great if you're interested in the rolling resistance of
> a tyre on a smooth drum, but I find rollerriding extremely boring.
>
> Rollers do not simulate real conditions well. If you put a car on a
> rolling road dynamometer the first thing to do is to pump up the tyres.
> If you want to do high speed testing I believe it is usual to run the
> tyre INSIDE a drum.
>
> Still it would have been nice if the experiments had been carried out
> with more precise measuring equipment, as in
> http://www.fahrradzukunft.de/fz-0501/0501-03.htm
It is time those deluded bunglers in the
Society of Automotive Engineers were alerted. Would you
do the honors?
--
Michael Press
Gary Young
Just after posting this message, I noticed that Heine posted a message
essentially repudiating that review -- to his credit I might add. I think
I'll just shut up for a while until I've paid closer attention to this
discussion.
carl...@comcast.net
Dear Phil,
The test assumed that there was 0 mph wind, based on looking at the
leaves on trees.
As far as I know, that means that the wind could be blowing from any
direction at up to 1 or even 2 mph. At such low speeds, the wind tends
to eddy.
If the wind failed to cooperate and lie still, the data can end up
with some runs having 1 mph tailwinds, some 1 mph headwinds, and some
1 mph sidewinds, which probably don't go neutral as easily as
expected.
That means that data points can be 2 mph off, judging by what several
calculators predict for winds below 1 mph.
What a reference bike run or two per session experiences doesn't mean
much--I'm interested in what the wind was doing as the rider coasted
for 30 seconds over about two football fields.
I suspect that a dozen smudge pots along the road would have led to a
lot more runs being thrown out, rather than just the few where the
wind became so noticeable that the testers felt obliged to toss out
their effort. (How do you decide if the leaves moved enough? Which
leaves? When? By the time that the leaves do move, the wind is getting
closer to 3 mph than to 0 mph.)
Other posters have commented that the data looks remarkably good.
The data may look good because the wind really was 0 mph, not varying
0.25 mph or 0.5 mph or even 2 mph here and there.
But it wouldn't be the first time that several unblinded participants
came up with remarkably good data in good faith that later turned out
to be unreproducible with more rigorous controls.
If you're going to a lot of trouble to make a lot of tests on a lot of
early mornings, it's going to be hard to decide by a seat-of-the-pants
method whether the wind is still enough to be called 0 mph.
When unconscious, unblinded bias is possible, the randomness of the
statistics often vanishes in the damnedest ways, baffling both the
testers who measured things in good faith and critics who not only
believe in the good faith of the testers, but can't figure out how the
hell the bias could have even operated.
I'd like to see the results confirmed by more tests with more rigorous
controls to remove doubt about the statistics.
To repeat what I've said elsewhere, what does it suggest that a test
comes up with measurements that can distinguish whether a rider was
wearing gloves or not, but can't say which way the wind was blowing to
within 1 mph?
The assumption of 0 mph wind from looking at leaves or kicking up a
little dust over a 600-foot timed section is questionable with the
kind of resolution suggested by the testers.
As Tim McNamara has pointed out, the hand-operated stop-watches raise
another question--again, not of the good faith of the testers, but of
what was actually measured.
Cheers,
Carl Fogel
carl...@comcast.net
[snip]
>I think the disagreement is this, in a nutshell: You [Jobst] think the [V]BQ
>test includes too much of the actual riding situation to be controlled.
> Others (myself included) think smooth drum tests include not enough to
>be truly representative.
[snip]
Dear Frank,
Good nutshell!
Cheers,
Carl Fogel
Tim Binns
>In article
>
[snip syuff about "systemic losses"]
>
>I disagree strongly. Tire rolling resistance should be
>narrowly defined so it can be measured simply,
>accurately, and precisely. Any other dissipative
>mechanism also needs to be narrowly defined so that it
>too can be measured simply, accurately, and precisely.
>
>Tire rolling resistance is the power dissipated flexing
>the tire measured on a smooth surface.
>
>Now define the dissipative mechanism you want to
>discuss, but do not call it tire rolling resistance.
I sort-of agree with you here, Jack.
Tyre RR is well--defined and can be tested accurately and repeatably
using the drum test a la IRC.
My only quibble is with the usefulness of this test: What is it
actually telling us that can be applied to our riding?
US Postal (or whomever) can stick Lance and his bike in a wind tunnel
and meddle with the frame and his position to get him as aero as
possible with him still being able to pedal effectively.
- Aerodynamics: Check!
Anyone with some basic mathematical ability can check out bearings and
chain drives in Kemp's or Machineries Handbook and get a reasonable
estimate for what energy is lost there. Or they can even repair to the
lab and measure the losses.
- Drivetrain overhead: Check!
And then we come to the bit where the rubber meets the road, wherin
IRC and suchlike perform rolldown tests using a smooth steel drum. And
they get a lovely set of nested curves, with the RR asymptotically
getting closer to zero as tyre pressure increases.
I agree, this is a good and accurate way to measure the losses
attributable to the flexing of the tyre casing. If I wished to select
a tyre in order to roll it, like an old-fashioned hoop, as far as I
could across a smooth homogenous surface, I'd be wishing I'd
bookmarked those results.
But bike tyres aren't used in this fashion (except by bored wrenches,
but that's another story). The tyre is part of a system,which also
involves the road surface and the sprung-damped mass of the rider,
neither of which are accounted for in drum tests. Are drum tests, for
all their useful comparative capabilities just measuring what we *can*
measure? A case of looking for your keys where the light is, perhaps?
- Rolling losses: Hm?!
Michael Press
<1163268233.8...@m7g2000cwm.googlegroups.com>,
"bfd" <bfd...@yahoo.com> wrote:
> hei...@earthlink.net wrote:
>
> > My preferences are evolving based on what my research shows. However,
> > while some are surprised that we found wider tires to roll faster, that
> > is not inconsistent with previous results. So now I am torn between
> > having my next custom bike made for 700C x 24.5 (actual width) tires
> > that tested fastest both in our test as well as in the TOUR steel drum
> > test, and waiting for the perfect, hand-made 650B x 38 mm tires.
> >
> How likely is it that a "perfect, hand-made 650B x 38 mm tire" will
> ever be made? Most people consider DUGAST tubular tires, which come in
> both cotton and silk casing, to be "the best." Since DUGAST is french
> and 650B is really a french size and has been around like forever, if
> he hasn't made a 650B x 38 mm wide tire by now, when will he? Further,
> who else makes "hand-made" tires?
>
> Personally, I like Avocet Fasgrip 700x25 tires (formerly 700x28) and
> pumped up to about 100psi find them to be "the best" for me. These
> tires just roll well and at 100psi are very, very comfortable. Also,
> with a 25mm wide tire, it handles better than most 23mm, the best of
> all worlds!
I favor the 700x25 also. They corner beautifully, and
the ride is comfortable as you say. They certainly do
not roll slower than other tires.
--
Michael Press
SocSecTr...@earthlink.net
carl...@comcast.net wrote:
> The assumption of 0 mph wind from looking at leaves or kicking up a
> little dust over a 600-foot timed section is questionable with the
> kind of resolution suggested by the testers.
This is all ridiculous. They tested the tires on multiple runs. If the
wind, clothing, riding positions, etc., were significant factors you
would expect the measurements to be all over the place, but they
weren't. No, your criticism, strangely enough, weakens the most
plausible criticism of the tests- that the rider was affecting the test
to get the results they expected. If these extraneous variables really
were present to a significant degree, it would mean that the rider was
hiding not only the real impact of the tested variables, but also
hiding the impact of the extraneous variables, all to distort the
measurements toward the desired result.
The extraneous variables are most likely reflected in the variation
that exists among measurements for a single tire. It takes more than
just pointing out that something could be wrong with the experimental
design; you have to show that these potential defects had an impact.
There is no evidence that their impact was of sufficient magnitude to
invalidate the conclusions drawn from the results.
Phil Holman
<SocSecTr...@earthlink.net> wrote in message
news:1163340838.6...@i42g2000cwa.googlegroups.com...
Phil H
Joe Riel
> The tyre is part of a system,which also
> involves the road surface and the sprung-damped mass of the rider,
> neither of which are accounted for in drum tests. Are drum tests, for
> all their useful comparative capabilities just measuring what we *can*
> measure? A case of looking for your keys where the light is, perhaps?
> - Rolling losses: Hm?!
Given that the "vibration induced losses" are dependent on many
variables (road, bike, rider, etc), it is hardly reasonable to claim
that they are a property of the tire. Possibly a useful exercise
would be to repeat a few of the tests on a bike with a full suspension
that can be completely locked out. Does such a beast exist? That
might suggest whether the effect is significant.
--
Joe Riel