BikeTechReview

Jan Heine and Mark Vande Kamp have an interesting article in the August 2009 edition of Bicycle Quarterly. I recently purchased the last four copies.

The magazine is interesting, attempting to be scientific and analytic but not always completely succeeding. Compared to the bicycle press, enormously refreshing: at least it inspires thought.

In the article, "minimizing suspension losses", it compares different tires and different forks on both smooth asphalt on on rumble strips. An attempt was made to ride @ 31 kph and measure the required power. Then an "analysis of covariance" was done to account for the effect of speed differences from the target. The claim is two runs were done for each treatment, although since a single powertap wheel was done for all measurements, this means tires were swapped in each case, so I suspect the replicates was done together.

Anyway, here's results fro smooth pavement, with a cheap Trek unicrown fork

Pasela 35 mm @ 3.5 bar: 198 watts
Grand Bois 25 mm @ 5.2 bar: 209 watts
Grand Bois 25 mm @ 6.6 bar: 209 watts
Bontrager Racelight hard-case 27 mm @ 6.6 bar: 232 watts

The following were all done with the Bontrager tires:

RockShox 1998 Ruby road fork: 203 watts
Singer 531 steel fork: 213 watts
Trek fork (with bar tape): 237 watts
Trek fork (no bar tape): 242 watts

Note the Trek + Bontrager don't agree in the two cases, presumably due to run-to-run variation.

Now this seems sort of nuts to me. The Rockshox and 35 mm tires @ 3.5 bar is the least power over smooth pavement?

Needless to say, on the rumble strips, the difference was far greater.

Pasela 35 mm @ 3.5 bar: 313 watts
Grand Bois 25 mm @ 5.2 bar: 386 watts
Grand Bois 25 mm @ 6.6 bar: 468 watts
Bontrager Racelight hard-case 27 mm @ 6.6 bar: 479 watts

RockShox 1998 Ruby road fork: 428 watts
Singer 531 steel fork: 426 watts
Trek fork (with bar tape): 459 watts
Trek fork (no bar tape): 479 watts

The conclusion was that a classic, compliant steel fork with fat low-pressure tires is substantially faster on rough surfaces than a relatively rigid steel fork with high-pressure narrow tires.

His argument is that the human body acts as a damper which dissipates energy internally when it is subjected to acceleration, and attempts to model rolling resistance by looking only at the tire will therefore substantially underestimate the total power loss, even if the tire is loaded with the full force it would experience with a rider. That muscle, fat, bones, and joints generate internal friction which places a premium on isolating the body from road vibration.

On the rumble strips, I'll accept his assertion. But on the smooth pavement? Just think how much faster Fabian Cancellera would be if he's only switch to 35 mm tires.... I'm having issues with this one. On the other hand, this magazine is in part Jan Heine's vehicle for trying to convince everyone of the advantages of steel randonneur bikes with 32 mm tires and stiff front racks.

One quick comment:

Jan looks like a fit guy of moderate height. Let's say 75 kg for rider, 2 kg for clothing, 11 kg for bike = 88 kg total. Then on smooth pavement the difference in Crr between the Bontrager (60 tpi, Kevlar belted) and the 35 mm Pacelas on smooth pavement would be 0.45%

% units
2433 units, 71 prefixes, 33 nonlinear units

You have: 34 watts / (31 kph * 88 kg * gravity)
You want: %
* 0.45752663

Bearing in mind "suspension losses" discussed in the article are a bit different from a classic "rolling resistance".

You mean that the *difference* in Crr is .0045 between those two... which is about the same as the Crr of a good tire on smooth roads. So if the Pasela was .0045, the Bontrager would be .0090, or double.

The concept of vibration losses is pretty old around these parts. Al did some roller tests with bumps, but I don't think there was much of an extra loss noticed... less than I expected anyway. There is plenty of anecdotal evidence that it is there, and you don't need to be too clever to realize that rumble strips would really slow you down. But some of JH's data has me scratching my head too. It'd sure be nice to get some more field test data like this, with tires we are more likely to use... maybe with a surface that is less extreme than a rumble strip, though that is convenient because you can use pavement that is side by side.

Ron Ruff wrote:


Here was my early afternoon experiment:

[video]http://www.biketechreview.com/images/bumpy_rollers_090610_lores.wmv[/video]

hard link if you don't want to download a plugin:

www.biketechreview.com/images/bumpy_rollers_090610_lores.wmv

higher pressure was lower rolling resistance for the zaffiro and the bontrager tire I tested from 60-140psi.

No: 0.45%. 34 watts at 31 kph is huge.

Anyway, I determined one procedural error: they discard outlier power points from the data stream (up to 2 from approximately 14 in the time taken to cross the test area). Powertap power points are not independent: if you pick up an extra power stroke one sample, good chance you lose one on another. It's important to keep all data in a given interval.

Anyway, hard to believe that explains things, unless the wind picked up during the Bontrager portion, which is quite plausible given the speed (31 kph): a wind around 2 kph (60 cm/sec) should be sufficient.

Principal difference rollers versus road:

• Rollers are rear-wheel only
• Rollers have roughness only in direction of travel

Doesn't seem to me that either of these effects should be huge, but they are differences.

djconnel wrote:
No: 0.45%. 34 watts at 31 kph is huge.[/quote]

I think .0045 and .45% are the same.



Hmmm.....



Many potential sources of confusion. Do you have rumble strips nearby, Dan?

kraig wrote:higher pressure was lower rolling resistance for the zaffiro and the bontrager tire I tested from 60-140psi.[/quote]

That sounds similar to what Al saw... but was the resistance higher with the bumps, and was the reduction in Crr at high pressure less with the bumps vs without?

Ron Ruff wrote:

I agree: 0.045% was written, not 0.0045.



Ideally the PT integrates power over a fixed time interval. Errors may be measurement errors or true variations in power. Variations in power from one 1.26 second interval to the next are expected to be substantial due to crank phasing. So that's real, and shouldn't be discarded.



Many potential sources of confusion. Do you have rumble strips nearby, Dan?[/quote]

Fortunately not ... plenty of potholes, and even some cobbles, but these are less repeatable due to trajectory variation. I might try something, however, just for fun.

djconnel wrote:


I think a couple more biggies are how the front and rear wheels are supported...the front fork stand I use doesn't have the same boundary conditions, and I'm not so sure what happens during a dynamic impact when using dual rear rollers. It doesn't intuitively seem to work out quite right.

My results also seem to speak directly to how pressure and vibration losses in the human body/sack of meat in the seat relate.

I will say that the feel with increasing pressure on the bumpy rollers was similar to my experience during my road testing - low pressure felt like I was riding in boss hog's Cadillac De Ville!

in addition, the similarities between my road testing, roller testing, and bumpy roller testing all seem to converge towards the same data driven result-> pressure that is low could very well be slow.

Ron Ruff wrote:


reported resistance was higher with the bumps. It was pretty much an upward shifted curve...though, I pretty much did single run tests with limited repeats, so, take the results for what they are worth.

if you have rollers, it's a pretty easy mod that shouldn't damage things. i'd encourage folks to give it a shot.

Ron Ruff wrote:


cattle guards and rumble strips are similar in that that type of excitation (and subsequent response), while definitely interesting, don't really seem super relevant to "real life" roads...do they?

kraig wrote:


Well, Jan says rumble strips represent roads encountered in brevets. The Seattle series must be especially tough.

Here's some plots from Al's data. Not much signature of an advantage to low-pressure tires here:

If it was done in the Seattle area there are 2 types of rumble strips used:
Most common - about 3 inches wide by 3/4-1 inch deep at the center rounded cutout with about 6-9 inches between
Less common - About the same profile in reversse = sticking up. plastic/rubber glued on to road.
In either case the general roads around here are fairly good unless you really want to ride on gravel or seek out rumble strips.

Either way riding over it is more like cyclocross through a cow pasture than riding on the road so I can see where cyclocross tires and pressures would do well.

I think the difference from the roller measurements comes from the bumps being big enough that to ride effectively the arms need to be relaxed to absorb the impact in addition whatever the tire is doing on the front. With a relaxed body things just kind of float, but if your upper body is rigid the bumps are big enough that hitting the bump is translated to a backwards force. At this level the rider's position and body tension have as much or more impact than the fork or tire. I think this is also why there is a difference showing up with taped vs no tape where there is much more padding in the gloves and palm of the hand but no tape might require more tension to hold on.

For brevet type riding (and cyclocross) i.e. very long and/or very rough giving up a bit of rolling resistance for comfort and being able to stay relaxed is probably a net gain as once you start getting fatigued and tensing up it starts a downward spiral.

Lindy wrote:


Good insights to noodle over!

I'd agree that the "loss model" for large amplitude, low frequency excitation (rumble strips/cattle guards/bots dots) might lead to a different optimal tire (i.e, fatty low pressure tires) than the "loss model" for low amplitude high frequency excitation (the kind of roads I'm interested in/intriqued by) might lead to.

I think Jobst Brandt and Jim Papadopolis went round and round on this on rbt over the years, no? Did they ever see eye to eye on this?

-k

djconnel wrote:


Cool plots! - things seem to be very similar to the results I obtained.

Well, I'm backing to testing this stuff too. Unfortunately, I don't have any rigorous controlled data, just stuff from my logs. If anything jumps out at me. It is that the differences may not be that big -- at least at Mountain bike speeds. Here are some data points from the "loop road" (Forest service 2060/2080) near Ashland, OR. This is a Forest Service road with about 3/4 gravel and decomposed granite, and the rest pavement. Measurement was from Powertap. I tried to wing some Crr numbers, using a power calculator. Clearly, the absolute values are meaningless, but the relative ones might provide some insight.



Rider+bike Weight 84KG
===================================================================
Mt. Ashland Climb (4 miles paved; rest bumpy)
Elevation 6581 - 2125 = 4456
Distance: 17.40 mi
4.9% grade
Weight 84KG


8/21/10 96:55@256 watts 65psi Fast fred rear+latex; ritchey speedmax front+butyl = .0083 Crr
8/24/10 95:58@265 watts 45 psi Fast fred rear+latex; ritchey speedmax front+butyl) = .0098 Crr
9/4/10 93:14@270 watts 40 psi Furious Fred rear+latex; fast fred front+latex) = .00855 Crr



The following 2 datasets are subset of this longer ride:

===================================================================
First 5.5 miles of Mt. Ashland -- includes 1 mile paved, the rest sandy or gravel
2125 ft to 4372 ft = 2247
Weight 84KG
Distance = 5.5 mi
7.75% grade


8/21/10 38:26@291 watts 65psi Fast fred rear+latex; ritchey speedmax front butyl = .0100 Crr
8/24/10 38:50@281 watts 45 psi (fast fred rear+latex; ritchey speedmaxfront+butyl) = .0080 Crr
9/4/10 38:04@291 watts 40 psi (Furious rear+latex; fast fred front+latex) = .0091 Crr
9/7/10 36:52@298 watts 65 psi (Furious rear+latex; fast fred front+latex) = .0080 Crr



===================================================================

Miles 4-5.5 (nearly all big gravel and very bumpy)
Elevation gain: 4372-3463 = 909
Distance 1.64 miles 10.5% grade

8/21/10 14:35@285 watts 65psi Fast fred rear+latex; ritchey speedmaxfront butyl = .0075 Crr
8/24/10 14:44@278 watts 45 psi (fast fred rear+latex; ritchey speedmax front+butyl) = .0056 Crr
9/4/10 14:19@288 watts 40 psi (Furious rear+latex; fast fred front+latex) = .0065 Crr
9/7/10 13:38@299 watts 65 psi (Furious rear+latex; fast fred front+latex) = .0048 Crr

Nice: basically no signal on tire pressure.

I'm sort of amazed at how dramatic and conclusive Jan Heine's results were, whereas here all the tests are fishing through the noise to try and see trends.

Obviously if a tire becomes spherical it doesn't fit into holes as well, and the road doesn't appear as rough. That's not a tire pressure thing, but rather a geometrical thing. But it requires transverse roughness, not only longitudinal.

It's become very clear to me from playing with the numbers, that the rolling resistance number is easily overwhelmed by other measurements. Suppose, for example, you lose a Kg from aspiration, perspiration &etc, between your first and last test runs. On my test course (~7.6% grade, 10mph), that single lost Kg equates to a 10% difference in calculated Crr!

Another key issue is, you MUST use a specific start and end point rather than relying on the odometer/speedometer. With a 26X2.25 tire, the tire pressure can make a big difference in the effective tire circumference, greatly skewing the odometer/speedometer readings. The only way to use the speedometer to accurately measure speed/distance would be to do a weighted rollout at different pressures.


That being said, here is a compilation of my best recent comparisons.


-jens

---------------------------
White Rabbit climb

Elevation gain: 3181 - 2125 = 1056
Distance: 2.63 miles
Grade: 7.6%
Weight 84Kg


17:25@297 65 psi Fast fred rear+latex; ritchey speedmaxfront butyl = .0081 Crr
17:44@287 45 psi fast fred rear+latex; ritchey speedmax front+butyl) = .0070 Crr
17:26@295 40 psi Furious rear+latex; fast fred front+latex = .0075 Crr
16:50@300 65 psi Furious rear+latex; fast fred front+latex = .0056 Crr
16:32@310 38 psi Furious rear+latex; fast fred front+latex = .0069 Crr

jkheycke wrote:


Jens, back in the saddle again! nice!

Also, another thing I've noticed is that out of the saddle efforts can muck things up too.

I think I reported some crr and cxa values for this in the forum somewhere...IIRC, it was not insignificant.

Gravel is a toughy too, i reckon - especially if it is deep enough that the "gravel line" is different for different pressures (I'm thinking something along the lines of "water line" for boat hulls).

anyway, welcome back into the fold!