Bicycle Quarterly Rolling Resistance Tests: No Surprises

anyway, testers gonna discuss spec armadillos gonna lose all
credibility

Tim McNamara wrote:
i leafed through the newest issue of BQ today, which has some further
tire tests and also some discussion of the statistical analysis they
used. I only skimmed the article and will read it in greater depth
hopefully in a few days.

Someone reminded me of some things I forgot to mention in my original
post. It is well known that for car tires, the rolling resistance
decreases significantly after a break-in period of about 200 miles.
This is due to changes in the physical/chemical structure of the
rubber as it gets worked into condition. I don't recall that effect
being discussed, and since both new and used tires were mixed in the
tests, it was not factored out.
Also, the rolling resistance changes significantly with
temperature, not due to pressure effects, but to changes in the
consistency of the rubber. A rise of 10degF as was reported during the
course of the day for the tests is significant. If the control runs at
the beginning and end of the day were nevertheless at the same
temperature, missing that fluctuation, they would ignore that
important effect.
o

I suspect that much of the underlying philosophy behind the BQ analysis
comes form Jim Papadopolous, who used to participate in r.b.t. Jim has
long maintained that wider, softer tires roll better on real roads
because they absorb and roll over small surface irregularities (such as
asphalt "grain") rather than being deflected upwards. There's been some
disagreement about the accuracy of that belief.

In article . com,
"41" wrote:

Tim McNamara wrote:
i leafed through the newest issue of BQ today, which has some
further tire tests and also some discussion of the statistical
analysis they used. I only skimmed the article and will read it in
greater depth hopefully in a few days.

Someone reminded me of some things I forgot to mention in my original
post. It is well known that for car tires, the rolling resistance
decreases significantly after a break-in period of about 200 miles.
This is due to changes in the physical/chemical structure of the
rubber as it gets worked into condition. I don't recall that effect
being discussed, and since both new and used tires were mixed in the
tests, it was not factored out.

Heine mentions something like in the source article.

Also, the rolling resistance changes significantly with
temperature, not due to pressure effects, but to changes in the
consistency of the rubber. A rise of 10degF as was reported during
the course of the day for the tests is significant. If the control
runs at the beginning and end of the day were nevertheless at the
same temperature, missing that fluctuation, they would ignore that
important effect. o

Heine also reported seeing this phenomenon and adjusted for it with a
set of reference wheels to use as a comparison, allowing them to develop
a correction factor to reduce the effects of temperature change.

You've re-read the thread, you say. Have you read the source article?
It's quite dense writing and much of what was in the article never made
it into the discussion, and only a few of the participants bothered to
read the original material so the rest tended to be talking through
their hat. It was rather bizarre at the time to be reading critiques
written by people who hadn't read what they were critiquing.

As I mentioned in another post, there is further information in the
newest issue of BQ including a discussion of the statistical analysis of
the data to determine reliability. I saw this at the LBS and am waiting
for my issue to arrive at home.

I've looked at the BQ test and results and they seem to be seriously
flawed.

I modeled their setup and got the following results:

...............................V1 (mph)..V2 (mph)...t (sec)...dt (sec)

Baseline .................18.17..........13.82......26.29
CdA= .36 vs .35.......18.14..........13.65......26.50......0.2 1
Bar. =29.9 vs 30.1...18.18 .........13.86......26.25.....-0.05
Head wind =0.5 vs 0..17.96........12.93......27.39 ......1.10
Temp= 60 vs 50........18.20.........13.93......26.16......-0.14
Crr=.006 vs .005.......18.03.........13.32......26.89.......0. 60
Weight = 162 vs 160..18.19.........13.88 .....26.22......-0.08

Tiny amounts of wind will really throw off the results, as will small
changes in rider position. Plus a change in Crr from .006 to .005 is
only going to give them a .6 sec time difference. Obviously, this sort
of test is not capable of discerning small Crr differences unless
*many* tests are done for each tire, and great care is taken to
minimize the effects of the many random variables.

On Mar 23, 1:30 pm, "41" wrote:
the values given for these tires from some
other test that has been oft-quoted in this group (I think done by Le
Cycle, on steel rollers as well) are as follows:

Tire Prr Crr Speed*
Vittoria Open Corsa Evo CX 27.1 W .00334 23.22
Michelin Pro2 Race 29.2 W .00360 23.14
Michelin Carbon 34.7 W .00428 22.94
Panaracer Stradius Pro 35.4 W .00436 22.91
Hutchinson Fusion 39.6 W .00488 22.76
Continenta l Ultra GatorSkin 40.3 W .00497 22.73
Continental Grand Prix 3000 46.6 W .00575 22.50
*185lb rider + bike 250W rider output CdA = .32m^2 (racing crouch,
normal road bike) Transmission efficiency = 96%


That looks like writing, but something is off. This is the correct
comparison. BTW, this is from Tour's Crr test.


Tire Crr Speed* Delta

Deda Tre Giro d'Italia 0.0038 23.08
Vittoria Open Corsa Evo CX 0.0039 23.05 0.03
Michelin Pro 2 Race 0.0042 22.96 0.12
Vittoria Diamante Pro Rain 0.0044 22.90 0.18
Michelin Megamium 2 0.0047 22.81 0.27
Pariba Revolution 0.0048 22.78 0.30
Michelin Carbon 0.0050 22.72 0.36
Panaracer Stradius Pro 0.0051 22.69 0.39
Schwalbe Stelvio Plus 0.0052 22.66 0.42
Schwalbe Stelvio Evolution Front 0.0056 22.54 0.54
Continental GP Force (rear) 0.0057 22.51 0.57
Hutchinson Fusion 0.0057 22.51 0.57
Schwalbe Stelvio Evolution Rear 0.0057 22.51 0.57
Continental Ultra GatorSkin 0.0058 22.48 0.60
Ritchey Pro Race Slick WCS 0.0058 22.48 0.60
Schwalbe Stelvio 0.0059 22.45 0.63
Specialized S-Works Mondo 0.0061 22.39 0.69
Continental GP 3000 0.0067 22.21 0.87
Hutchinson Top Speed 0.0069 22.15 0.93
Continental GP Attack (front) 0.0073 22.04 1.04


*
in MPH
185lb rider + bike
250W rider output
CdA = .32m^2 (racing crouch, normal road bike)
Transmission efficiency = 96%

----------------------------------------------

Yes, that's true. JH said

#higher tire pressure
#did results in slightly faster run times. But the differences were
too
#small to be statistically significant, much smaller than the drum
tests
#predict. The regression analysis looked at all the factors that
could
#explain a tire's speed, and tire pressure was not one of them

They say that roller test show that a drop in pressure from 105 to 85
psi resulted in a Crr increase of 15%, whereas their testing showed a
drop in speed of only 2%. Apples to oranges... Crr is not speed. An
increase in Crr of 15% *is* equivalent to a drop in speed of ~2%... so
they have verified the drum tests.

Someone reminded me of some things I forgot to mention in my original
post. It is well known that for car tires, the rolling resistance
decreases significantly after a break-in period of about 200 miles.
This is due to changes in the physical/chemical structure of the
rubber as it gets worked into condition. I don't recall that effect
being discussed, and since both new and used tires were mixed in the
tests, it was not factored out.

This has been noted with bicycle tires as well. A drop of ~10% in Crr
is common.

That is to say, everybody says the ride is best
when the casing and tread are so silken as to effectively disappear.
So, it really is the pneumatic effect (spread impulse out over time,
don't suck it into the tire) that takes the edge off the bumps, not
the hysteresis.

A good point. The smoothest riding tire should also be the one with
the lowest suspension losses. At high pressure this is likely to be a
tire with low Crr as well, but at lower pressures (MTB at 35 psi) I've
noticed that the damping effect of high hysteresis tires comes into
play, and they actually ride better.

Ron Ruff wrote:
I've looked at the BQ test and results and they seem to be seriously
flawed.

I modeled their setup and got the following results:

Tiny amounts of wind will really throw off the results, as will small
changes in rider position. Plus a change in Crr from .006 to .005 is
only going to give them a .6 sec time difference. Obviously, this sort
of test is not capable of discerning small Crr differences unless
*many* tests are done for each tire, and great care is taken to
minimize the effects of the many random variables.

Indeed - which is why

a) we were careful to minimize the effect of random variables, by
selecting days with favorable atmospheric conditions, etc.
b) we ran multiple tests for each tire
c) we report that we cannot distinguish very small differences between
tires. The current issue also has a table that shows which tires were
statisticially significantly different, and which were not. However,
the differences between tires are large enough that they can be
distinguished by our methods (see below).

Instead of hypothesizing, look at our raw data - we reported the
actual measurements both for a majority of the initial test runs, and
for all of the new runs. If you think you have an alternative
explanation for the data, let us know. To say "It can't be" is not a
valid critique. The data exists, and it needs to be explained.

There are two questions regarding our data:

1) Are we measuring actual differences between tires (explained
variance), or just random variations due to wind, rider position, etc.
(unexplained variance). We looked into that (see the current issue of
Bicycle Quarterly, Vol. 5, No. 3, p. 20). An analysis of variance
(ANOVA) showed that the ratio of explained to unexplained variance in
our initial test was more than 151 to 1. F(18, 36) = 151.9. The
resulting p is smaller than 0.0001, or in other words, there is less
than a 1 in 10,000 chance that we are just measuring random scatter in
the data.

(In case you wonder whether the author is qualified, check out his
credentials at http://www.vintagebicyclepress.com/contributors.html.
He had a minor in statistics in his Ph.D. studies.)

2) How good is our model in replicating real-world conditions? You may
argue that our road was rougher than the ones you usually ride. The
standing start also might have favored or disfavored heavier tires,
although we don't think it was an important factor. But these are
points we can argue. However, for our testing of tire pressure, the
standing start does not matter, as the tire weight remained the same
for the different runs. The fact that the run time changes only very
little with increasing pressure above a certain point indicates that
the overall resistance of the tire also changes very little with
increasing pressure.

Finally, somebody said that our tests just confirmed what was known
already. That is true - with the original tests in Vol. 5, No. 1,
there was an article on 1930s handmade French clinchers that seemed to
incorporate all the design features we found make a tire fast at
moderate speeds: great width, supple casing, soft rubber compound, and
as a result low pressure rating. Unfortunately, without real-world
testing, many, if not most people, including tire manufacturers, since
have lost that knowledge.

The real trade-off is between width, supple casings, and pressure. In
theory, it is quite simple:

- Wider tires roll faster (at least within the widths of the tires we
tested, 21-37 mm).
- Tires with supple casings roll faster.
- People used to believe that higher pressures also made tires roll
significantly faster.

However, you can't have all three - large width, supple casings and
high pressures - in the same tire. In the past, manufacturers
sacrificed the suppleness to be able to run high pressures. Most
makers used sturdier, less supple casings for their wider tires.
However, our results show that this tradeoff was wrong. Using the same
supple casings for wider tires, which then must run at relatively low
pressures, gets you faster tires.

As you know, I usually don't follow this group. If somebody comes up
with an alternative interpretation of our data, please contact me
directly. I will print your interpretation in the next issue of
Bicycle Quarterly. The goal of the magazine is to have an open
discussion, so I am looking for different points of view.

Jan Heine
Editor
Bicycle Quarterly
140 Lakeside Ave #C
Seattle WA 98122
www.bikequarterly.com

Tim McNamara wrote:
In article . com,
"41" wrote:


You've re-read the thread, you say. Have you read the source article?
It's quite dense writing and much of what was in the article never made
it into t he discussion, and only a few of the participants bothered to
read the original material so the rest tended to be talking through
their hat. It was rather bizarre at the time to be reading critiques
written by people who hadn't read what they we re critiquing.

I came at this from two directions: first, I was thinking of ordering
a few back issues of the VBQ because I wanted some of the older
articles. But in the contents pagee I saw they had some new articles
on tires. Also, since I can't get tanwall Avocet Duros or Duro Pluses
any more, I was wondering about those Grand Bois tires... although the
cost rules them out. So, I wondered: what was said about all this in
RBT? Are they worth ordering? Reading the threads, I couldn't believe
the collective amnesia of the group. We had already discussed
empirical results on the correlation between drum tests of various
surfaces, and between drum and road tests of various surfaces. I
pointed out that the statistical analysis of these things is
problematic, because of (a) the scatter, even between drum tests; and
(b) the presence of leverage points or outliers. The Rivendell tires
certainly seem to constitute same and I wonder if the statistical
significance would disappear if they were takien out.


As I mentioned in another post, there is further information in the
newest issue of BQ including a discussion of the statistical analysis of
the data to determine reliability. I saw this at the LBS and am waiting
for my issue to arrive at home.

I didn't realize that hadn't arrived yet. On the contents page it's
already listed as a back issue, so I thought it was already factored
into the discussion.i

Ron Ruff wrote:

I modeled their setup and got the following results:

..............................V1 (mph)..V2 (mph)...t (sec)...dt (sec)

Baseline .................18.17..........13.82......26.29
CdA= .36 vs .35.......18.14..........13.65......26.50......0.2 1
Bar. =29.9 vs 30.1...18.18 .........13.86......26.25.....-0.05
Head wind =0.5 vs 0..17.96........12.93......27.39 ......1.10
Temp= 60 vs 50........18.20.........13.93......26.16......-0.14
Crr=.006 vs .005.......18.03.........13.32......26.89.......0. 60
Weight = 162 vs 160..18.19.........13.88 .....26.22......-0.08

I note in addition that effects due to changes in trail would not be
so straightforward to calculate.


the values given for these tires from some
other test that has been oft-quoted in this group (I think done by Le
Cycle, on steel rollers as well) are as follows:

Tire Prr Crr Speed*
Vittoria Open Corsa Evo CX 27.1 W .00334 23.22
Michelin Pro2 Race 29.2 W .00360 23.14
Michelin Carbon 34.7 W .00428 22.94
Panaracer Stradius Pro 35.4 W .00436 22.91
Hutchinson Fusion 39.6 W .00488 22.76
Continenta l Ultra GatorSkin 40.3 W .00497 22.73
Continental Grand Prix 3000 46.6 W .00575 22.50
*185lb rider + bike 250W rider output CdA = .32m^2 (racing crouch,
normal road bike) Transmission efficiency = 96%


That looks like writing, but something is off. This is the correct
comparison. BTW, this is from Tour's C rr test.

I believe you meant, "that looks like my writing". Indeed it is:
http://groups.google.com/group/rec.b...79bf48e79872f?
hl=en&


They say that roller test show that a drop in pressure from 105 to 85
psi resulted in a Crr increase of 15%, whereas their testing showed a
drop in speed of only 2%. Apples to oranges... Crr is not speed. An
increase in Crr of 15% *is* equivalent to a drop in speed of ~2%... so
they have verified the drum tests.

Agreed. Señor Fogel's calculators show the same thing.


Someone reminded me of some things I forgot t o mention in my original
post. It is well known that for car tires, the rolling resistance
decreases significantly after a break-in period of about 200 miles.
This is due to changes in the physical/chemical structure of the
rubber as it gets worked into condition. I don't recall that effect
being discussed, and since both new and used tires were mixed in the
tests, it was not factored out.

This has been noted with bicycle tires as well. A drop of ~10% in Crr
is common.

This is a figure likewise often quoted for car tires.

Jan, thanks for tossing your hat in the ring again. I had a chance to
leaf through but not read the new article in the new BQ and was pleased
to see a discussion of the statistics. I'm looking forward to reading
it closely.

"It's stiff and inelastic, i.e. an energy sucker."
metal tires/wheels are energy suckers because metal wheels are
inelastic?
elastic tires are not energy suckers?

the entire (sorry) thrust (eeek) from the tyre industry since 1955
sumpthin is eliminating dragging the tyre's load induced elastic
bottom bulge from Monte Carlo to Paris.