GhostRider62

I extracted the sealant. 35 ml front and 38 ml rear. Since I had originally installed 45 ml of sealant about 8 weeks ago, I removed the front tire to inspect. There was quite a bit of hardened sealant coating the tire, explaining where the 10 ml difference went. I put the sealant back in.

I did 8 runs total with GP5000 TR S tires on Farsports 56 Venoux wheels and GP 5000 clincher on Flo60 wheels. ABBA and ABBA. First set at 90 psi and the last set at 78 psi.

The only result of curiosity was the 78 psi runs on the tubeless setup, the Crr was much lower than the 90 psi runs whereas the clincher tires were consistent at 78 psi or 90 psi.

Looking more carefully at the road surface, it is no longer very smooth and the rolling resistance is clearly higher than in the past, just surprising how much. It seems that the tubeless tires do not perform as well compared to the clincher version at 90 psi but they both compare about equal at 78 psi. This would take more testing than I am willing to perform but it might suggest that the casing of tubeless tires are more sensitive to the impedance breakpoint. Both the clincher and tubeless performed about the same overall with the exception of the two 90 psi runs on the tubeless.

So, I do not see Crr of 0.006ish on worn pavement to be outside the realm of possibility. YMMV. That is my conclusion. Thanks for all the ideas.

livedarklions

If it's ok to ask--for a dope like me, what's the practical application of that conclusion? Do you change anything in your setup or does it confirm you've got things in hand? This is out of genuine curiosity because I'm really impressed by how you're working this out. I don't know enough to argue with you even if I tried.

RChung

These were your coast down runs? How long were those runs, and what was the total elevation drop?

When I do coast downs, rather than test for farthest roll out from a standing start, I start at different speeds and record second-by-second speeds to a particular spot (like, a manhole cover, or a tree, or a driveway). The wider range of starting speeds makes it easier to isolate the aero drag from the rolling drag, and I can generally get better precision on the coefficients.

PeteHski

As a UK rider I get subjected to a wide range of "worn" pavement and, subjectively, the difference in rolling resistance is pretty huge across that range. Some of our roads are seriously slow. But I guess your idea of "worn" could be quite different.

GhostRider62

Wind was an issue today. 8-10 mph. I would not normally test in such wind. I found a 1.3 mile section along the river where the cliff/hill shielded the wind. I went up the river and turned without touching the brakes and then finished on an upslope without touching the brakes. According to the slope/elevation chart in GoldenCheetah, the differential was 23 feet.

My coastdown test method does not use VE. I simply coast until I stop. I do not pedal. Since I start at the same exact place on the hill, I leave with the same potential energy on each run and if wind is not a factor and if I hold a good position, the frictional loss differentials from tire to tire will be what determines how far I coast before stopping. Peak speed is usually 10.5 mph. On calm days, my stopping point that I mark in chalk will be within 10 feet of each other for one set of tires whereas the difference from one tire make/model to another tire can be 4-500 feet (lousy tires vs the best). This coast down is a very coarse way of ranking tires qualitatively. The only time it did not predict results and for which I scratched my head for years was the narrow Conti Supersonic vs the Vittoria Speed tubeless. The Speeds rolled out better but my TT or short 3-5 minute Strava segments were always faster on the Supersonics. (the supersonics are more aero on my wheels and at high speed, the rolling resistance is less important.....of course as you know)

Thank you for your input. I really do not have any significant conclusion from this exercise other than to look a little more closely at tire pressure on the tubeless setup and that I had not noticed the deterioration in the road surface.

PeteHski

That's good enough for me, thanks for confirming what I thought might be the case.

GhostRider62

In the winter, they coat this highway in a thick layer of "salt". Upon closer examination with my face up close, the surface seems to have the tar layer between the stones removed like a pock mocked teenagers skin. There is a 300 foot section of brand new pavement where a flood ravaged the road and going over it, the difference is night and day. My normal country lanes are very, very bumpy and very hilly, most of my riding has been there last year and this year This route along the river has always been my smoothest, so, I never really noticed that it had taken a beating. I knew the Crr there because I tested so much to try to break 20 minutes for 10 miles on my bent (did it handily). WRT the upcoming TT, it doesn't matter because everyone will be on the same road. I was hoping to do just under 23 minutes on my upright, but there is no way for me on that surface. As a former F1 engineer and cyclist, I am sure you can get my feelings on that.

son_of_clyde

Regarding Crr determination, any reason not to just solve the least squares approximation for x = (x1, x2)' (Ax = b (corresponding to scalar equation Pc = x1*v + x2*v^3), with A being n x 2 coefficient matrix or v and v^3 data, b being n x1 vector of Pc data, x1 = Crr*N (normal force), x2 = 0.5*rho*Cda, Pc = power corrected for drive train loss)? Basically do n tests holding power at a different constant for each test while recording v. This way no prior knowledge of Cda is necessary. With n sufficiently large, seems like this could be pretty accurate.

GhostRider62

Practically, I like to know my Crr in order to evaluate aerodynamic changes using the Chung VE method. My coast down test is just a down and dirty way of checking that my tires are not damaged. The old casings on Compass and GP4000 used to get ruined on gravel roads and visual inspection was not good enough, so, this became one way I checked if I ruined tires or also always a check when I put new tires on (using a control set of wheels to compare....always)

The relative rankings of tires stays the same in my experience. The power absorbed is lot more in practice than in theory in the real world as many already know.

Practically speaking I will probably limit my GP5000 TR S tubeless tire pressure to 80 psi but when I can get more calm days, I confirm what pressure is best from my fat derriere on these roads on those tires. In practical terms, optimization might give me almost an hour more sleep on a 600k brevet if indeed the wattage differentials hold up with more testing.

GhostRider62

Because holding power constant is not easy for this human and N sufficiently large costs a lot of money.

Lastly, I am very lazy. But I am honest.

I know the Chung method works. I am also confident in my simple coastdown method. Usually, I am just coast down testing to make sure i did not get a bad tire or that I did not damage it. The absolute value of Crr isn't the primary driver. I am comparing one set of wheels to a known control (my TT wheels).

Things that could have been considering in a more in depth investigation.

1. Narrow/Wide chain ring teeth
2. Frankenstein drivetrain. SRAM 1x chainring, YNB 12 speed chain, Rotor 12 speed cassette
3. Farsport wheel bearings vs NTN LLB in the Flo60 wheels
4. Aero difference in the wheels
5. Dinky DUB BB bearings pressed into Cervelo shell (more sensitive to misalignment)

I might mount both wheels on another bike and repeat testing just to rule out these components on the bike and therefore conclusively say the road sucks.

RChung

Good thought. However, people have tried this for many years; you can, too, if you like. The issues are that it's hard to keep Pc and v constant, plus your structural equation mis-specifies the full drag dissipation model: you don't have terms for changes in either kinetic or potential energy, and you're assuming air speed va = ground speed v. In particular, Crr scales like slope, so the LS approach requires either really careful measurement of slope or a dead flat (and I mean absolutely dead flat, like velodrome flat) surface. If you don't meet the severe test demands, the Gauss-Markov conditions aren't met and you can't guarantee that the LS estimator is either unbiased or minimum variance. So with the LS approach, you can occasionally end up with non-positive x1 or x2. I use a more robust estimation method with an approach sort of closer in spirit to a constrained maximum likelihood. Under ideal conditions (zero wind, zero acceleration so constant speed, zero slope (or, really constant slope) and constant power) the estimation method I use simplifies down to the LS approach, but when conditions aren't ideal my way gives better results. Plus, n trials is a pain. If you structure the estimation right, the data collection goes much faster.

son_of_clyde

I should have specified assumptions of zero wind speed (va = vg), known (ideally no) grade, and zero mean sensor noise - aside from maybe the last, not so easy in the real world I guess. I'll read up on your method, as I'm not familiar.

PeteHski

Yeah I think the variations in rolling resistance across differing road surfaces are very significant. We have a wide mixture of road types over here. The difference between fresh smooth tarmac and a worn, chip-sealed surface is to me night and day. The slowest I find is badly worn chip seal, which sounds similar to what you describe above. It creates a nasty, energy sapping, high frequency vibration through the whole bike and the holes from missing chips create larger shocks, further slowing the bike. I find that wider, lower pressure tyres help a lot on these roads, with little or no downside on the smoother roads. I'm currently running 30 or even 32 mm tubeless around here at 60-65 psi. As it happens my current 30 mm tyres are Conti GP500S TR and to me they seem fine. Certainly doesn't feel like a slow tyre, quite the contrary actually. Not that I've attempted to measure Crr objectively. These particular tyres, at least in this width, seem to do a good job of minimising the surface "buzz" you get from chip seal roads, which can only be a good thing.

As you mentioned F1, I remember we used to take surface sample profiles at each track as again there is a huge variation in track surfaces around the world. People might think of race tracks being super smooth, but in reality some of them can be pretty harsh. But we weren't looking to measure rolling resistance in this case. It was more about evaluatiing grip level and most importantly for us, tyre wear. Some circuits cause far higher tyre wear than others and much of that is down to the micro surface profile.

GhostRider62

Part of what I am seeing probably relates to a large load variation F/R distribution from my 50/50 recumbent where I had optimized tire pressure to what I am now riding, which has something like 30/70 F/R balance. Having gained weight doesn't help. Crr is usually considered a constant but it varies by temperature, speed and load for any given tire. I suspect I was over the "impedance break point" in the tubeless tires AND that breakpoint is lower than on the recumbent due to different load distribution F/R. Accurate tire pressure measurement is also an other question. I have four devices. Two measure the same, the other two measure 5 and 11 pounds higher than the other two. Which is correct.

Two of my bikes can take wide tires, this Cervelo S3 can take 27mm actual but not more. I think a bigger consideration is proper tire pressure for the given size. Often people will quote Crr for 32 mm vs 25 mm at the same pressure, say 90 psi. But, that is way too high to get the proper tire drop and suspension effect for the 32 mm tire. When both are inflated to the proper pressure to get the "Frank Berto" 15% drop, Crr will be similar at least on a drum per the linked tests. It would be an interesting experiment for me to mount some wide supple tires and directly compare with 25 mm GP5000 tires on that road. I do have two Powertap G3 hubs.

https://www.bicyclerollingresistance...000-comparison

HTupolev

It's unclear to me how Bierman evaluates that equal drop against a flat surface is "equal comfort." At least under static testing, the spring rate of a tire depends on the curvature of the object being pressed against it, with the deflection on sharper shapes being less dependent on tire width. This seems like something that really needs to get answered with vibration sensors, or at the very least by proxy by characterizing how tire width affects breakpoint.

GhostRider62

I can't defend their testing protocol. What would you say about the same pressure in a 25 mm tire vs 32 mm tire, on normal roads (not Pave)? If you follow the 15% drop rule, you get about the correct pressure to maximize comfort. When the breakpoint is exceeded, where does this additional lost energy go? Seems to me much of it goes into your tissue wit some into the tire casing. The proxy for me is comfort. I run the narrowest tire that I can run, the one that is comfortable. Pressure can be a speed tradeoff. The optimal width for me is 28 mm but that is too tight for my bike, so, I run 25 mm.

The only "studies" that I recall that looked at vibration and where it was absorbed were at two Universities. The results were not compelling, but anyone who does long distance endurance events would believe them.

HTupolev

As someone who hasn't precisely tested the issue, I'm not sure. Generally I do pump wider tires lower in like-for-like riding.

What do you mean? Adequately-low pressure to avoid discomfort from vibration is surface-dependent.


I'm not a physicist, but it makes intuitive sense that it would disproportionately tend to be dissipated in parts of the system that have both significant damping and flexibility.

GhostRider62

My guess is force/displacement of the tire is linear within the working range of the tire despite reading otherwise. The vertical displacement of the system is mostly absorbed by the human body in my opinion. There is a long standing guide to air your tires so that they drop 15% when loaded. This gives the best speed/comfort. I had found that some tires like the Compass extra legere perform better at slightly less drop. But what I mean is selecting tires and pressure to match local road conditions. On bumpy gravel roads, I need my 38 mm Herse tires at 40-45 psi. On local roads 25 or 28 psi GP5000 at appropriate pressure work just as well as 35 mm tires in terms of comfort but are much faster for the type of riding that I do.

L134 

My own long standing question, which I've never asked before, is how in the world are people measuring this at home? I've always been pretty much a 28-32mm thumb tester. I'm not questioning the validity of the guide but wondering about the practicality of it.

GhostRider62

Is it the ruler or the math? 15% of 32mm is just under 5 mm drop. You have to have a helper measure for you. A more practical way is to use Silca's tire pressure calculator, it is pretty good. One thing it gets wrong is weight distribution, it not even close to 52/48 on a road bike or at least in my experience. If it were so, the front and rear tire would have similar wear rates and when putting you and your bike onto scales, both wheels would show about the same weight.

L134 

I think it is the helper lying on the floor with a ruler part with a large suspicion that the margin for error is quite large. The tire pressure calculators are more practical but, as you mention, how accurate are they? And, wouldn't one really require a separate calculation for each tire/tube combination if one is really striving for an accurate 15% drop? It has been my impression that the 15% drop was some figure tire manufacturers gave to Berto (of whom I am a fan) but I don't ever remember reading what criteria the tire manufacturers were using in determining that recommended drop. Was it safety, low rolling resistance, longevity or what? I'm not saying I think the 15% drop guideline is not right, I'm just kinda wondering how practical it actually is. Out of curiosity, have you figured out what drop is ideal for your purposes and, if so, what have you settled on? I guess I would also be wondering if that ideal number varies from tire/tube combination to tire/tube combination?

tomato coupe

When you're cruising along at 250 W, all the force propelling you is transmitted to the road through that tiny contact patch under your rear tire. That's responsible for most of the increased tire wear.

RChung

I don't subscribe exactly to the 15% drop rule but, if I did, I'd measure it by the change in rollout between an unloaded and loaded tire. That is, I'd measure the radius of a wheel with an unloaded tire and calculate the "theoretical" rollout distance. Then if I were using 25mm tires, I'd calculate what a 15% drop in tire size would imply for a new "loaded" rollout. Then I'd measure the loaded rollout in my usual method (roll out 4 revolutions, and measure with a 10m long tape). I wouldn't need a helper for any step of that.

But I don't use the exact 15% rule, so I don't do it that way. I do what Tom Anhalt does: lower tire pressure until I am in danger of a pinch flat or bottoming out on the roughest road that I usually ride, then go up a little from there to account for surprise potholes or cracks in the road.

livedarklions

Wow, thanks! I always thought it was the weight difference but what you claim here seems to check out with Prof. Google.

freetors

I think for many people the actual weight distribution is much more rearward. I haven't actually gotten out the scales to confirm the distribution, but on most of my bikes I can get away with 5-10psi less in the front for about the same amount of squish. I wouldn't be surprised if my personal distribution is more like 60/40