Basilic is right--from an engineering perspective, Stormur, you're overcomplicating the issue. There's an important distinction between "all the variables I can think of" and "all of the variables that matter."
TL;DR: Rolling resistance tests like Tom Anhalt's and the pendulum method are very useful; the variables they neglect don't matter much or at all.
Rolling resistance is well understood within the engineering world but things are different in the bike-technical-enthusiast world. In bias-ply tires--and all bike tires are bias-ply, even the Maxxis Radiale*-- most of the rolling resistance comes from the interlaminar shear strains between the plies. The elastomer (rubber, latex, whatever) turns these strains into heat via hysteresis. Cotton-cased tubular tires typically have very little latex between the casing layers, so they tend to have low rolling resistance.
(Random side note: high-TPI tires don't have low rolling resistance because they're supple--suppleness is a side effect. They have low rolling resistance largely because there's less free volume between the threads to hold high-hysteresis latex or rubber).
Radial tires do not experience significant interlaminar shear, so their casings lose less energy to hysteresis and therefore have much lower rolling resistance than a bias-ply tire of comparable construction.
OK, on to Stormur's variables. Some of the things he lists matter, but many do not and some are redundant.
stormur wrote:Any test of rr made has certain margin of reliabilty caused mainly by procedure
This is true of every experiment ever conducted. It's something to keep in mind, but it's not a problem at all.
stormur wrote:To measure REAL crr values it would take far more time & effort & will generate cost of it being comercially ( for magazine ) unjustified. And what even worse , results ( difference between models ) can be very surprising .
Well, this all depends on what you mean by "REAL CRR values." Getting the absolute right value isn't really of interest; no one cares if all your values for power consumption are high by 0.01 Watts. What's really of interest is relative measurements, and these are pretty easy to get with a high degree of repeatability. Both Tom Anhalt's tests and the pendulum method are pretty rigorous and seem to generate similar results for the same tire models.
In other words, you can get very good results quite cheaply. This comes down to the distinction between accuracy and precision. Tom Anhalt's tests and the pendulum tests are quite precise, and, for rolling resistance, if you have good precision (relative measurements between tires) then accuracy (exact wattage consumed by a given tire) isn't very important. Here's more info: http://en.wikipedia.org/wiki/Accuracy_and_precision
stormur wrote:1. closed track with defined 3 types of tarmac : from glass-flat to harsh, surface dry + wet .
Texture doesn't matter. The value of rolling resistance is higher on a rough surface, but if tire A is faster than tire B on smooth surface, it will be faster than tire B on a bumpy road as well. This is why it's fine to use a smooth steel drum and apply the results to real-world roads: you're just trying to measure how much energy is absorbed for a given amount of deflection. Again, all that matters is relative values between tires.
I know Wheel Energy in Finland sometimes tests with textured surfaces. But I haven't seen any data that show that, as above, the difference between two tires changes depending on surface texture. If Wheel Energy (or anyone else) have those data for bike tires, I'd love to see them. Here's a link to some of their results:http://velonews.competitor.com/2014/12/ ... nce_355085
Relative values between tires might change depending on wet or dry conditions. It might be worth investigating if you were running a pro team that focused on the classics (which tend to be wet), but this is a second-order effect.
stormur wrote:2. air temperature, tests made in 3 different ( 10-20- 30*C for example )
This might matter (a little) in extreme temperatures. Hysteresis in elastomers can vary nonlinearly with temperature. But between 10-35 degrees C, relative measurements should be pretty similar. Again, a pro team racing classics (which tend to be cold) might want to check this out. This is also a second-order effect.
stormur wrote:3. air pressure : lets say 1013HPa, 980 and 1045
I think you're referring to ambient barometric pressure. This doesn't matter at all. What matters is the gage pressure of the tire (what it's pumped to). A tire pumped to 8.3 bar (~120 psi) over ambient pressure is 8.3 bar above ambient pressure no matter what that ambient pressure value is.
stormur wrote: 4. height ( over the sea level ) : 500, 1000, 1500
This is essentially the same variable as #3, and it doesn't matter for the same reasons.
stormur wrote:5. humidity : low , mid, dry
Nope, humidity doesn't matter. A cotton casing inadequately coated in latex *might* absorb some water from the atmosphere, but any change in CRR values would be swamped by moisture retained the last time it was ridden in the rain. This is a third-order effect at best.
stormur wrote: 6. load ( imaginable rider + bike weight : 70-80-90kg f.e. )
This doesn't matter at all
. In fact, the very concept of a coefficient of rolling resistance requires that energy dissipated be directly proportional to the weight on the wheel. If you think "load" is a nonlinear variable, then you've rejected the very concept of a CRR.
stormur wrote: 7. tire temperature ( cold and warmed up )
This is the same as #2. Tires do warm up when you ride them, but by a tiny amount. They dump so much heat to the wind that tire operating temperature doesn't change very much in the course of a ride. If anyone out there has an infrared thermometer and can provide data to contradict me, I'd love to know about it. (Tire temperature matters a lot for auto racing, but that's a different kettle of thermodynamic fish).
stormur wrote: 8. rim variability ( elasticity , profile width )
This is also irrelevant in practice. I'm not sure you realize this, but what you're suggesting is that it's somehow plausible that a GP4000S II will be significantly faster than a Pro Race 4 on a Kinlin rim but slower than a Pro Race 4 on a Hed Belgium rim.
If rolling resistance is a mysterious, semi-magical phenomenon to someone, then that suggestion might seem true. But even a rough understanding of why some tires roll faster than others will make it clear that a tire that rolls faster on a Kinlin rim will also roll faster on a Hed Belgium rim. It may be true that a Pro Race 4 on a Belgium rim is faster than a GP4000S II on a narrower Kinlin rim. But for a given rim, the relative CRR values will hold. E.g., if a Pro Race 4 has 7% more rolling resistance than a GP4000S II on a Kinlin rim, it will also have ~7% more rolling resistance when both tires are mounted to a Hed Belgium rim.
I'm not making fun of anyone or suggesting that they're ill-informed; I'm just trying to de-mystify the physical phenomena involved.
stormur wrote:9. front/ rear wheel load ( different weight balance for tt/ tri / road )
Nope. Irrelevant and redundant; see #6
stormur wrote:10. Various tire pressures for all tests for certain tire with factor of butyl & latex tube
Also irrelevant and redundant; see # 3. Butyl vs. latex matters a lot, but we only know that because it's been well established with the very tests you decry as inadequate.
stormur wrote:11. resistance on straight and cornering
Eh, in theory these could be different. But very few people are getting dropped in crits because their rolling resistance goes through the roof when they lean over for a corner. At any rate, this is easy to test using a pendulum rig with cambered wheels.
stormur wrote:12. same downhill & uphill ( weight load variable )
Irrelevant and redundant; see (again) # 6. Do you really think all of these incarnations of the same variable are somehow different?
stormur wrote:13. season ( air density ) & weather
Irrelevant (#3) and redundant (#2, #3, #5, #7 ...)
stormur wrote:14. what was used to pump tire : air, Co2, N … (?)
This is a third order effect at best. What you're really talking about here is viscous damping of the gas used to inflate the tire. In theory, there are tiny differences, but in practice, these are swamped by all of the first-order variables that do
matter, such as inflation pressure, section width, casing construction, etc.
stormur wrote:15. high and low cadence ( different load- rear tire bend on stroke )
This doesn't matter. If your cadence happened to match the resonant frequency of the bike/rider/tire system, you'd be bunnyhopping down the road. Besides, we've already established a direct relationship between load and rolling resistance--there's even a coefficient that describes this relationship.
Sounds - at least - crazy
But let's be "precise" - to the end . with or without any sense of it .
It doesn't sound crazy, but it does sound like a non-scientist's view of what science might be like. There's no shame in this, of course. Also, in the most technical sense, your list doesn't address precision; it addresses accuracy. But again, this is a subtle distinction that's lost on even some scientists and engineers.
list can be very long… no one will make it . ever. It cost money. Not even 1 manufacturer will agree to share so enormous costs of that kind of test, not knowing result
If you made a list of everything that has an effect on rolling resistance, it wouldn't just be long: it would be infinite. But the list of variables that matter for a useful test is quite short. Both the pendulum test and Tom Anhalt's tests on BikeTechReview capture them well. Anyone, including manufacturers, can perform tests like these at minimal cost. Including some of the second- and third-order variables mentioned above would increase cost a little without much benefit.
I'm an engineer (as you've probably guessed) and engineering is often described as "applied science." This is a useful description. A big part of applied science is knowing which parts of the science one should apply and which parts don't matter much. If you leave out something important, your results will be wrong. If you include things that aren't important, you get bogged down in the quagmire of variables Stormur seems to dread.
But if you can discern the variables that (a) are measurable and (b) matter, you can strike the right balance. One might think of this as a 90/10 rule: 90% of your result quality comes from 10% of the variables. Things like wind drag, rolling resistance, "speed wobble" and frame/crank/component flex are quantifiable and quite accessible to physicists and engineers. If they weren't--if these things were as unknowable as some people seem to believe--we'd be living in a permanent stone age.
P.S. I seem to have written a bit of a dissertation here. There's a reason for that. When I was a junior racer back in the mid-eighties and early nineties, I wondered about this stuff all the time. No one I knew--not teammates, not national-level coaches, not even the framebuilders I talked to--had any idea about why
bicycles work. One well-respected coach warned me that my radially-laced 28-hole front wheel would "probably fold up in a corner" under my crushing 130-pound weight.
I was desperate in those pre-internet days for any informed thoughts on these subjects, and I would have killed for Jobst Brandt's Usenet posts (or this forum) back then. These things are eminently knowable and understandable, and I find that exhilarating.
I became an engineer because I was tired of not knowing whether a stiff frame was more efficient than a flexy one (it's not, really) or whether aero spokes build a stiffer wheel than butted spokes (they don't; the only thing that matters is spoke cross-sectional area). I didn't write this to wag my finger at anyone or make them feel dumb. I write posts like this for the contemporary incarnation of the 13-year-old kid I used to be. So yeah, I'm flying my freak flag high.
* Panaracer made a true radial bike tire in the '80s, sold as OEM tires on some Miyata models. Allegedly, they felt odd and squirmy, as though they were underinflated even at full pressure. This is because radial tires lack torsional stiffness, which bias-ply tires have in spades. The radials on your car have a steel belt not to protect against punctures but to provide torsional stiffness. cf.: http://www.bikeforums.net/classic-vinta ... tires.html