Originally Posted by
pm124
Laboratory conditions exist because they eliminate variability, thus providing much more accurate results than real world tests. Real world tests exist because laboratory conditions aren't variable enough, and thus are not generalizable to the real world. Fortunately, a happy medium exists; testing similar routes with a power meter.
Don't take this the wrong way, but your roll down tests are not going to be even close to providing you with information on differences in rolling resistance. Your cheapest bike rolled best downhill simply because it was heaviest. Now, if you repeat the experience in a vacuum and get the same results, I'd be more inclined to believe them!
Roll down tests can be done with lab conditions. The problem with rolling resistance tests done in labs is that they rarely reflect the rear world. The tires are usually tested on smooth drums that are around 6" in diameter. These bear no reality to road surfaces. The bicycle and rider are removed from the equation, even though those provide dampening and suspension.
I've been involved in the rolling resistance testing for Bicycle Quarterly. We use the same bicycle and rider for all tests. We use the same test track and time the center of the rolldown (ignoring timing from the start and end when the rider may wobble). The rider does the route 3 or 4 times and is timed by two people independently. We always use reference tires for the first and last rolldown to make sure that our times are consistent in a day. The timing is done at 5am when there is no detectable wind (using a very accurate wind meter) and in a certain temp range. The exact procedure and results are listed in BQ issue Vol 5 No 1.
http://www.vintagebicyclepress.com
I think that this gives much more accurate results than a drum test. It also takes a significant amount of time.