I have no stake in the game, aside from looking to get a set of aero wheels, and doing my best to learn about bike aerodynamics to make an intelligent decision. Call me an armchair hobbiest.
Here is my outsiders take on things:
What interests me most about Hambini’s data is not the details, but rather that he is proposing an entirely new method to interpret/collect aero data that is proposed to be more representative of the real world. What he proposes, if correct, could make everyone at least reevaluate protocols... or risk being left measuring aero data that is removed from reality (yes, no one test will truly represent reality).
The problem here, is that he approached the problem a little too casually when it came to initially sharing the data, and critiquing some of the brands/designs. You don’t go throwing bricks through windows without being ready for a fight, and to back up your claims with hard data and protocol details. Is the world of science, your data needs to be independantly verifiable to gain true validity. Maybe few marketing departments give that level of data/methadology... but if Hambini is going to dump on data/products of others (calling them out), his data/methadology needs to hold weight.
Irrespective of what he actually did or did not do in regards to protocol, it was not communicated in detail when the results came out. I do not blame others for asking a lot of questions, or being skeptical. I also do not think Hambini quite expected the response he got... he was a relitavly newcomer to the scene, and making big claims would not be left without challenge. A weekend hobby project was being analyzed like a whitepaper... except there was no whitepaper.
In my eyes, Hambini further hurt his credibility by being narcissistic in his own defense. I am not saying that mud was not slung both ways, but his behaviour in response to challenge was telling in my eyes.
So a few observations on the results (and i think it has some use if you have some faith, just like most data).
- this is impressive work, and i commend it. Sure it is not perfect, but it is up to the reader to interpret the level of value. And there is plenty of value here... even if i would not take it as gospel.
- 2.5% error. I doubt that most tests of this nature will do much better, so i will not criticize. I share concerns of others that the amount of wheels tested, over a number of days over a rather long protocol (~26min per wheel)... leaves a lot of room for measurement error... simply because a human on a bike is involved. As hambini said himself, rider position can make a far larger difference than the small differences in wheels being measured. Rider position was within +/- 1cm via laser (if memory serves). So i take that as up to 2cm variation. That can be a lot comparative to what is being measured with the wheels... especially if multiple parts of the human anatomy have the same variation. A test dummy has it’s own limitations... but may be the lesser evil.
- many people discussed accuracy and repeatability. The next wheel i would like to see tested would be a repeat of a wheel tested on day 1. A consistent end result would be very telling.
- Flo. I have read both sides of the story. Hambini apparently contacted flo before the data went public? Why? To gloat? To say their product sucked? Whatever that initial private exchange was... hambini came out pretty hard on flo. I am sure something personal was said... product criticized, credentials questioned, egoes flared. Who knows. All i know is Hambini can get nasty when provoked, and i could not figure out why he was so hard on flo (maybe he did not single them out in the original print of his article/data, and only went after them post lawyer letter?). When i looked at the best and the worst of wheels of a given depth... flo wheels were only 2.5% or so worse than the best... margin of error. Why the hate? To me the zipp nsw was the big underperformer.
- tires. 23cc conti gp4000’s were used, all at 8.25bar. This was a controlled factor, but in reality, no two tire widths would be the same in the test. Wheel Internal diameters would all be different. And what if a wheels aerodynamics were designed to work with a tire that was the same width or narrower than the rim? the 23cc gp4000 measures closer to 24.8mm at 7bar. At 8.25bar it is approaching closer to 26mm. It is well known that tires wider than the rim can have some negative aero consequences... hambini made a big deal of it. Near 26mm tires may have hurt the 24mm flo wheels if optimal aero performance was designed for a tire closer to 24mm wide. Other narrow wheels (ff, and shimano) were effected less, or not at all, by a wide tire. Maybe it was serendipity, maybe they were designed that way. But i want to find the best tire for the wheel i want, or vice versa. Maybe the flo wheels would have done just as well as the others of similar depth, had 1bar of air been removed and the tries been narrower. Maybe shimano wheels would have done even better had the tires been narrower. Who knows, but i will bet that every narrow rim in the test (i.e. actual tire width was wider than rim width) will perform a little better with a narrower tire. And remember, at 30km/hr, the flo 45 was only 2.4% slower (5w) than the best of the 45mm wheels (which is less than the margin of testing error), despite a tire that could have been up to 2mm wider than optimal, Again, why the hate?
- to build on the above, it would be interesting to know the measured tire widths on each wheel, and the actual widths of each wheel in the tests. In a perfect world, the test would be done with a tire narrower, same width, and wider than the wheel... and then the reader could see which tire width to aim for if they own that wheel (or how far they could push the envelope with the tires... as wider is beneficial for many reasons). I can guess the results... the wheels with wider tires (wider than the wheel) will not perform as well aerodynamically. Silica has some really nicely presented data on this.
And that is is my problem with hambinis data... the tire width issue is a confounder... even if a small one. Remember, the margin of error is 2.5%, and the “worst from first” in hambinis study (for a given rim depth) is in a similar order of magnitude. Some wheel makers performance was called out by the author for that level of difference.
- to go further, to be positive... what is also very cool about hambini’s data, is that it may also be interpreted in a way that helps you identify wheels that perform quite well despite tires being wider than the rim! There are a few pretty narrow rims that do just fine. Something in their design presumably makes them a little less prone to the wide tires, and that may be a good thing for folks with callipers that don’t perform as well on wider rims. Hopefully those cases are not due to measurement error...
- hambini has said he will not further discuss his methadology. I can understand his frustration... but again, don’t go throwing stones if you live in a glass house.
Overall, I think Hambini’s work has a huge potential to help educate buyers, and industry folks alike. He seems like a smart enough guy to be able to take his protocol, improve it (for example, using a dummy instead of a human rider), and hopefully represent (or repeat) his data in the form of a white paper. If his work can be produced independantly, than maybe we have a basis for a game changer as a more accurate testing protocol (in relation to fluctuating airstreams).
My theory is that Hambini has no interest in sharing the gritty details because 1. He did it for interest initially over a weekend and had no intention on investing the time on a hobby project: this is not his day job, and now 2. Because he is now refining the process, possibly to use on a more commercial scale. Why give that info away for free when someone like Trek may pay him for it? Who knows, i hope for the latter.
Sorry for the long post, but i figure that the select few people that got this far in the thread, might be patient enough to read it.
Here is my outsiders take on things:
What interests me most about Hambini’s data is not the details, but rather that he is proposing an entirely new method to interpret/collect aero data that is proposed to be more representative of the real world. What he proposes, if correct, could make everyone at least reevaluate protocols... or risk being left measuring aero data that is removed from reality (yes, no one test will truly represent reality).
The problem here, is that he approached the problem a little too casually when it came to initially sharing the data, and critiquing some of the brands/designs. You don’t go throwing bricks through windows without being ready for a fight, and to back up your claims with hard data and protocol details. Is the world of science, your data needs to be independantly verifiable to gain true validity. Maybe few marketing departments give that level of data/methadology... but if Hambini is going to dump on data/products of others (calling them out), his data/methadology needs to hold weight.
Irrespective of what he actually did or did not do in regards to protocol, it was not communicated in detail when the results came out. I do not blame others for asking a lot of questions, or being skeptical. I also do not think Hambini quite expected the response he got... he was a relitavly newcomer to the scene, and making big claims would not be left without challenge. A weekend hobby project was being analyzed like a whitepaper... except there was no whitepaper.
In my eyes, Hambini further hurt his credibility by being narcissistic in his own defense. I am not saying that mud was not slung both ways, but his behaviour in response to challenge was telling in my eyes.
So a few observations on the results (and i think it has some use if you have some faith, just like most data).
- this is impressive work, and i commend it. Sure it is not perfect, but it is up to the reader to interpret the level of value. And there is plenty of value here... even if i would not take it as gospel.
- 2.5% error. I doubt that most tests of this nature will do much better, so i will not criticize. I share concerns of others that the amount of wheels tested, over a number of days over a rather long protocol (~26min per wheel)... leaves a lot of room for measurement error... simply because a human on a bike is involved. As hambini said himself, rider position can make a far larger difference than the small differences in wheels being measured. Rider position was within +/- 1cm via laser (if memory serves). So i take that as up to 2cm variation. That can be a lot comparative to what is being measured with the wheels... especially if multiple parts of the human anatomy have the same variation. A test dummy has it’s own limitations... but may be the lesser evil.
- many people discussed accuracy and repeatability. The next wheel i would like to see tested would be a repeat of a wheel tested on day 1. A consistent end result would be very telling.
- Flo. I have read both sides of the story. Hambini apparently contacted flo before the data went public? Why? To gloat? To say their product sucked? Whatever that initial private exchange was... hambini came out pretty hard on flo. I am sure something personal was said... product criticized, credentials questioned, egoes flared. Who knows. All i know is Hambini can get nasty when provoked, and i could not figure out why he was so hard on flo (maybe he did not single them out in the original print of his article/data, and only went after them post lawyer letter?). When i looked at the best and the worst of wheels of a given depth... flo wheels were only 2.5% or so worse than the best... margin of error. Why the hate? To me the zipp nsw was the big underperformer.
- tires. 23cc conti gp4000’s were used, all at 8.25bar. This was a controlled factor, but in reality, no two tire widths would be the same in the test. Wheel Internal diameters would all be different. And what if a wheels aerodynamics were designed to work with a tire that was the same width or narrower than the rim? the 23cc gp4000 measures closer to 24.8mm at 7bar. At 8.25bar it is approaching closer to 26mm. It is well known that tires wider than the rim can have some negative aero consequences... hambini made a big deal of it. Near 26mm tires may have hurt the 24mm flo wheels if optimal aero performance was designed for a tire closer to 24mm wide. Other narrow wheels (ff, and shimano) were effected less, or not at all, by a wide tire. Maybe it was serendipity, maybe they were designed that way. But i want to find the best tire for the wheel i want, or vice versa. Maybe the flo wheels would have done just as well as the others of similar depth, had 1bar of air been removed and the tries been narrower. Maybe shimano wheels would have done even better had the tires been narrower. Who knows, but i will bet that every narrow rim in the test (i.e. actual tire width was wider than rim width) will perform a little better with a narrower tire. And remember, at 30km/hr, the flo 45 was only 2.4% slower (5w) than the best of the 45mm wheels (which is less than the margin of testing error), despite a tire that could have been up to 2mm wider than optimal, Again, why the hate?
- to build on the above, it would be interesting to know the measured tire widths on each wheel, and the actual widths of each wheel in the tests. In a perfect world, the test would be done with a tire narrower, same width, and wider than the wheel... and then the reader could see which tire width to aim for if they own that wheel (or how far they could push the envelope with the tires... as wider is beneficial for many reasons). I can guess the results... the wheels with wider tires (wider than the wheel) will not perform as well aerodynamically. Silica has some really nicely presented data on this.
And that is is my problem with hambinis data... the tire width issue is a confounder... even if a small one. Remember, the margin of error is 2.5%, and the “worst from first” in hambinis study (for a given rim depth) is in a similar order of magnitude. Some wheel makers performance was called out by the author for that level of difference.
- to go further, to be positive... what is also very cool about hambini’s data, is that it may also be interpreted in a way that helps you identify wheels that perform quite well despite tires being wider than the rim! There are a few pretty narrow rims that do just fine. Something in their design presumably makes them a little less prone to the wide tires, and that may be a good thing for folks with callipers that don’t perform as well on wider rims. Hopefully those cases are not due to measurement error...
- hambini has said he will not further discuss his methadology. I can understand his frustration... but again, don’t go throwing stones if you live in a glass house.
Overall, I think Hambini’s work has a huge potential to help educate buyers, and industry folks alike. He seems like a smart enough guy to be able to take his protocol, improve it (for example, using a dummy instead of a human rider), and hopefully represent (or repeat) his data in the form of a white paper. If his work can be produced independantly, than maybe we have a basis for a game changer as a more accurate testing protocol (in relation to fluctuating airstreams).
My theory is that Hambini has no interest in sharing the gritty details because 1. He did it for interest initially over a weekend and had no intention on investing the time on a hobby project: this is not his day job, and now 2. Because he is now refining the process, possibly to use on a more commercial scale. Why give that info away for free when someone like Trek may pay him for it? Who knows, i hope for the latter.
Sorry for the long post, but i figure that the select few people that got this far in the thread, might be patient enough to read it.
Last edited by:
Rocket_racing: Dec 16, 18 8:57