The longer the period of data collection, the more closely the sample mean would approach the population mean.

IOW, if anything limiting the test segment to only a little over 400 m would tend to reduce the correlation between the measured and predicted values. Yet, the R^2 was 0.97, i.e., steady-state modeling accounted for all but 3% of the variation.

More than anything I'm curious to know why the Flo wheels performed so poorly. Everything else on there (random Chinese wheels excepted) ended up about where I would assume they would. Also, Hambini seems to have largely confirmed that narrower = faster for tires which is what I've mostly seen as well from independent wind tunnel data.

Side note: I wish Schwalbe would update their 22mm "Ironman" tubeless tire with a faster compound.

I've been saying this since I joined ST... no relation to Reynolds personally. I never liked the design philosophy behind Zipps and their copycats, so it's funny to see everyone come around.

If you approach Albuquerque from the east, take a left and go about 100 miles, you end up where I live... ;)

I've never been that fond of the idealized environment of the typical wind tunnel test. Introducing realistic instability in the flow field plus time varying yaw of the article being tested makes a lot of sense to me. I was hoping for a very interesting discussion and hopefully an improved protocol for bike aero testing. But it devolved into paranoid ad hominim so severe that I believed Hambini was a total fraud. Lately it has improved a bit and he keeps posting more data. Maybe we'll get the full story eventually.

The most interesting thing I learned in that thread is that aero engineers in the UK are allowed to use restricted corporate wind tunnels for personal projects. Boggles my mind. I would have stayed in the business longer if they'd let us have that kind of fun ;)

We put it to you that you were simply using the air as your wind tunnel and not a fan.

If we (and we have) go out and do several 40 second blasts and plot them together the correlation is no where near 0.97 against a power meter using your equations.

40s is a very short interval and the fact it's in the middle of an airfield bares little resemblance to road conditions. Why do loads of users on here quote it as the basis of data to compare against ours beggars belief?

Regards
Sachin, Pete, Jeremy, Steve, Mark, Terry

---

HELLO HAMBINI FANS!!!

We simply tested in the weather conditions that prevailed that day (the only day available to us).

As it turned out, it was quite windy (and chilly), due to the passage of what Texans call a "Blue Norther" cold front. The measured ground level wind speeds are provided in the paper, but to put them in context, picture a large flag snapping in the wind, the landyard clanging against the flag pole, moderately thick tree branches swaying, etc.

If, as you claim, you have not been able to replicate our observations, I suggest that you write them up for publication in a peer-reviewed journal, so that they can vetted by many eyes.

Finally, as I pointed out before, the shorter the period of data collection, the greater the error in the measured (vs. actual) value. Thus, if anything lengthening the outdoor trials (or the wind tunnel runs, which IIRC were 30 s at each yaw angle) would be expected to improve, not reduce, the correlation between the turbulent (outdoors) and steady (wind tunnel) flow conditions.

Having had a discussion internally, we have decided to send our data for peer review. We think it would be good to assist our bicycle colleagues to improve their aerodynamics.

With regards your comments about the shorter duration giving a worse correlation. The following were noted:

- the wind by your own measurements was not particularly strong . Approx 7km/h even though anecdotal evidence is suggesting it was gusty.
- You were in an open area

If you take a spot measurement which for a 40 second burst is. It's going to give an answer that is more akin to a sedate environment such as a wind tunnel than If you were to ride for 10km and be exposed to the pressure differentials caused by temperature variations, passing vehicles, natural wind variation and rider adjustments.

What we find mind boggling is the cohort of users on here deem it to show a strong correlation between a wind tunnel and field testing and use it as the basis to dismiss transient effects. You yourself have commented you believe an error to be of 3%.

Thankyou for your time

Peter Williams PhD


I have used Sachin's forum account as the validation system is very slow.

---

HELLO HAMBINI FANS!!!

1. Wind speed was measured at 1 m off the ground. As others have pointed out, the wind speed at 10 m height (typically used in weather reports) would have been significantly greater.

2. Wind speed was measured several times during each run, and the average value used in further calculations. The between-run SD as given in our paper is therefore an underestimate of the within-run variability.

3. Based on the available dates, my conclusion is that the difference between steady and transient conditions appears to be 3% *at a maximum*, at least when testing the entire bike+rider system and using older equipment.

With respect to the latter, while I think that it would be interesting to repeat our study using a more modern bicycle, nothing you/your colleague hambini has posted has convinced me that testing under transient vs. steady state conditions would make a significant difference in this context.

Finally, I will say that I find hambini's persona here (and weightweenies) quite off-putting, exhibiting all of the negative qualites stereotypically associated with being British and/or an engineer.

Hello Andrew

In response to your comments.

1. If you are in an open area the difference in speed between 0.5m and 30m is considered to be negligible. This is in JAA and FAA guidelines for take off parameters, the exact paragraph is lost in my memory. You were on an airfield so it would apply.

2. Noted

3. Noted. I suspect that the weighting of the rider will be more of a constituent of the total drag than the wheels. However your study was conducted in conditions that do not correlate well with ours based on road data. The conclusion from our data is your wind tunnel data is lower than would be experienced on the road.

With regards personas. Everyone is different. The group of engineers in the Aero department here is well and truly multicultural and bring something to the table. I have watched this thread, indeed it is a topic of conversation for our lunch times. I have seen the types of questions that have been asked. Once an answer is given. The answer is systematically taken to task because of a small detail that was omitted. Requests to display data in formats that modern aero engineers are not accustomed to because they are considered dated are met with hostility. Once data is given, more data is requested, this self perpetuating cycle for someone who is providing it for free and without prompting is unjust.

The two answer giving forums on here and starbike have been shared internally for peer review... ironically. Whilst some of Sachin's antics are deliberately designed to wind individuals up. He has not given any information that is inaccurate. What is deemed as being evasive is quite simply Aerospace engineers working in units that bicycle engineers or those connected to the industry seem incapable of grasping, the majority of data was already included but required some manipulation to make it less technical.

In my lifetime I have seen the industry evolve and technology and practices being trickled down to other industries, the use of carbon being a good example. In the field of Aerodynamics that trickle does not seem to have occurred. I read an article in a UK cycling publication written by Simon Smart, he was formally at Red Bull Racing Formula 1, he commented that in his opinion the aerodynamics in the cycling industry were less developed than they should be. After following these two answer giving forums, I would tend to agree.

Thankyou for your time and I wish you a pleasant day

Peter

---

HELLO HAMBINI FANS!!!

As others have indicated, wind speed as measured at 1 m and 10 m would likely have differed by a factor of at least two. This is consistent with how I would describe the wind using the Beaufort scale, i.e., as a 4 or 5 (although when I ran back to the control building from the taxiway, I could definitely feel the wind pushing me along, which would make it more like a 6).

(Beaufort scale for those unfamiliar with it: https://simple.wikipedia.org/wiki/Beaufort_scale)

ETA: Here is a pic of Phil White (Cervelo founder) assisting with the field testing. Note the tousled hair and windbreaker wrapped over his hands to keep warm. He is kneeling next to the runway that we used - the black strip in the background is actually the grassy median separating the runway from the adjacent taxiway. Out of the image to the left is a line of trees next to the taxiway. I think that somewhere Jim has a picture of one of the jets that used the taxiway and runway during our testing, which IIRC shows the trees. I will see if I can dredge it up.




I believe the phrase that applies is "taking the piss." As I indicated, I find it both stereotypical and off-putting. You/your group may or may not have important contributions to make here, but his behavior is doing you no good.

For that matter, neither does reliance on appeals to authority, i.e., mentioning the opinions of others who reportedly agree with you.

Andrew,


Your dismissal of information presented is frankly staggering, quite arrogant and xenophobic.

It is with respect that I will pass no further comment

Thanks

Peter

---

HELLO HAMBINI FANS!!!

What information am I dismissing?

peter, at al, a few notes:

validation: everyone is vetted before validation. all validation happens within minutes or, at most, a very few hours. we do this because this is a very busy forum - one of the 2 or 3 largest in all endurance sport - and is the target of spammers dozens of times per day. but you don't see any spam here, because we try to make sure you're you, and that you belong here, before your account is validated. no one is in the validation queue, so, either you are validated or, if you aren't it's because you didn't pass muster. if you don't have a validated account i recommend you create the account once again and, in the question we ask you during the sign-up process, reference something about this discussion we're having. we have 90,000 registered users, so it's not hard, just, i need something in your account that tells me you're you and you belong here.

personalities: everybody hates andy coggan. except those who don't (which includes me, but i had to practice hard). my guess is that everybody hates hambini except those who don't. so, when you get hambini and coggan in the same forum, what do you think is going to happen in a hen house full of chickens, and 2 cocks? so, my advice, don't take things personally. otherwise, the urgency of your message is lost, and the cycling world loses your voice. i doubt thinks went any smoother in the weightweenies forum. this is cycling. this is the milieu. figure it out.

---

Dan Empfield
aka Slowman

Why, might I ask, are you singling me out? I was a late entry into this thread, have not had any disagreeable interactions with hambini or Peter (until his last somewhat bizarre message), and have not been involved in the exchanges on weightweenies.

IOW, I think you are pointing fingers in the wrong direction.

i just noticed the last couple of posts, and commented on them. but i have this sense that this is a moving discussion - like a progressive dinner - and you all are having the same argument from forum to forum. i just think that whomever are all the players - hambini, peter, the other 3 engineers there, you, tom anhalt - you all need to find a way to talk things through, just in case the hambini folks have something good to add to this discussion.

my sense is that the hambini crew are like so many others who've waltzed into cycling with zero regard for product or protocol that existed before they got here. but i might be wrong about that and even if i'm right about that sometimes folks learn to appreciate how incredibly precise and difficult this industry is (just try to make a crank, chainrings, a BB, and a frame, stick it all together, and see if the chainrings spin true; and then make that happen in a $69 complete bike).

but, hey! they're new here! and we need to be welcoming, and i know you guys! and i mean this in love: who do we think we're shitting here? the slowtwitch family is a nuanced group. so, we have 2 nuanced groups, with enough PhDs in this discussion to populate both starting lineups of a basketball game. it would be nice if we could all find a way to get past the idiosycracies, regardless of who is most idiosyncratic.

---

Dan Empfield
aka Slowman

The answer to that is long. The short answer is don't be offended. We want a octagon cage match performed with civility and professionalism. Slowman singled out because you're one of the (few) regular posters that we can throw into the octagon and expect to go toe to toe with someone who *appears* to have a legit aerospace background.

I hope it works out because there's a suggestion that Hambini could provide some significant value to the discussion of cycling aerodynamics.

I will simply repeat what I said above: I am a late entry into this thread, have only posted a few times, and have not been involved in the discussion on weightweenies.

IOW, your "sense", at least in my case, is wrong.

So I am clearly WAY late to the party here!!

So sad to me to see how little credit Zipp got for any of this... when we launched Firecrest in 2010, we did with its some amazing supporting data including 3 prior years of AIAA papers developed with Matt Godo... we were the first to talk about transient behavior, and then developed the models for handling stability..the first to extensively use the Aerostick to test and later validate handling improvements...etc, all of the other stuff mentioned here from every other single company has benefited from the work done to develop Firecrest (especially the work of Matt Godo.. we even became the first non-military winner of an Innovation Excellence Award at SuperComputing 2012), which makes me super proud, so hard to see it forgotten so quickly. Rant over.

I really like the way the Hambini crew have tried to put numerous variables into one single test.. it's an interesting attempt to blend real world data into something new and better. One of our secrets at Zipp was that our tunnel protocol went 0-30 and then built the curve stepping back to zero.. so it started with detached flow and measured when it all came back together rather than the other way around which led to shapes that seemed to be more forgiving of transients. Many wheels (and bikes) have large hysteresis when comparing 0-30 to 30-0

Past all of that, I have no horse in this race so my questions are all based on logistics:

1. What tunnel was this done in and who/how was it paid for? I see the it says weekend project for engineers, but with a 1600 second (26+ minute) protocol run time per velocity regime, you are looking at ~1hour per run time per setup assuming spin-up/spin-down time turn on of data acquisition, etc..then assuming a minimum ~30 minute de-fixture/change wheels/re-fixture, re-zero, and water the rider time this makes for 1:30 absolute minimum time required per wheel model if you're running it like an F1 pit crew.. I count ~30 fan-on hours here plus probably 50% more in fan-off time plus 2x that amount in post processing.. and that's assuming everything is perfect all the time. So that's a long way of asking if you are trying to monetize this in some way or if it was covered by other means?

2. Is the tunnel used for other things?
3. Did you have to tear it down and set it up each weekend or whatever to do this over a few months?
4. Was the same rider used for every test and if so, how did you find somebody who could/would ride a 26 minute tunnel protocol back to back at the required level of physiological stability required to net a +/-2.5% max error?
5. Is the tunnel temperature and pressure stable or is this accounted for in post processing? I'm assuming this took a few months to complete so ambient had to vary quite a bit?

If nothing else, I love seeing the discussion on these topics as we try to move the state of the art to the next level.
Josh

---

http://www.SILCA.cc
Check out my podcast, inside stories from more than 20 years of product and tech innovation from inside the Pro Peloton and Pro Triathlon worlds!
http://www.marginalgainspodcast.cc

Just... no. You can't seriously believe that. I don't want to go back to one ounce of that nonsense, so can we just move forward instead of reinventing the recent history?

Maybe engineering works totally differently in the UK, but when someone makes bold revolutionary claims, then other engineers are *really* going to want to see the details that back up those claims. The average Joe won't care because they don't understand the technical aspects anyway. But engineers who are interested in the topic get really excited about this stuff. And of course they'll want to go over the details thoroughly and will have many ideas and critiques.

Those details would be your field testing instrumentation, protocol, and data collected. And then how you replicated "real world conditions" in the tunnel, and the data collected. I know it's a lot of work, but if you are engineers you surely understand that "everyone has been doing it wrong (for decades), we did it right, here are the numbers, trust us" isn't going to cut it.

I'd be extremely surprised if you did everything perfectly on the first try. This was a hobby project after all, not a well funded research endeavor. So I hope you aren't too concerned about defending it on that level. Adjustments will surely be made. It's a learning process.

I for one have a very positive attitude regarding the concept of using unsteady flow or motion in wind tunnel testing. In the real world, wind is very "unsteady" and the bike and rider constantly making subtle corrections. Also the large reduction in drag at high yaw that is commonly reported in WT tests (for wheels, frames, and even riders on bikes) has never happened in the real world for me or most of the people I know.

Hi Josh. Some of your questions have been answered in the WW thread at least in a cursory way (but don't waste your time!). We haven't seen details of their protocol or setup however. Hopefully that will be reported eventually. It's a hobby project and they basically led with results. Full details might take awhile.

Apparently in the UK engineers are allowed to use their restricted corporate facilities for personal projects. Blows my mind, but.... I guess Airbus (or some other company?) paid for this indirectly. The specific tunnel wasn't identified.

I also wondered about the inclusion of a rider for these tests since that introduces a lot of variability. Hambini mentioned that they took care to ensure that the rider's position was consistent. I still think that is an issue, particularly when testing over a long period of time. They are estimating +-2.5% error which is quite a lot when you consider that the wheels are the only variable and they are small % of the total drag. For instance when you look at the 30km/hr chart here: https://www.hambini.com/...hich-one-is-fastest/ a wheel reported as 185W would have overlapping error bars with every wheel tested except for the Ksyrium.

I hope that they are advancing the "state of the art" as well. I like the concept anyway.

I like the test. I have a pair of Shimano C50 that I got for 500 USD 4 years ago that seem to be doing extremely well for the price tag.
I also like the bike bottom brackets you're making and selling on your site. I will get one of these next time mine is due for a change. I really like the idea of having only to swap the bearings later on.

I haven't really understood yet why your test protocol supposedly models reality in a better way than traditional approaches where one would determine the most relevant yaw angles seen in reality and then focus on these. Afterall race courses are seldomly routed through perfect loops with wind coming from the same direction during all this time. But then I'm only on page 8 on the weightweenie thread so far and probably I will get to the point where thats covered. Just because its been done in a certain way in the past doesn't mean that its the best way to look at it.

I absolutely don't think that you need to point out at every chance you get that you are an aerospace engineer. I get how you would lead with that during the introduction of your work, but reading it now in every second or third post of yours over various forums complete with pictures of military aircraft for illustration now that kinda seems totally stupid. And there is quite some condescending tone that people are getting annoyed by no point in denying that. Theres been quite some interesting points raised and scientific explanations given and those speak for themselves so no need for the crap between the lines.

Just to put some color on wind speed at 0.5m as it relates to cycling. On long rural rides, because it's more pleasant.

For standard reported numbers at 10m wind speed at 0.5m will be:

45% lower passing by a mowed airfield
60% lower spinning between a field of wheat and field of barley
75% lower passing by an oak
And 84% lower Just before you get to the pub.

I'll chime in because I think I understand the concept at least. Typical "real" wind has more variability than you see in the tunnel. It's constantly changing in direction and intensity. Plus the rider and bike move from side to side, and the wheel will pivot slightly as you move down the road. They are attempting to model this by moving the bike +rider back and forth in the tunnel a small amount, while simultaneously running through a varying yaw protocol. They don't stop at a particular yaw and wait for the drag to settle as it typically done. Hambini also mentioned louvers at one point but I don't know if they are manipulating the flow-field as well.

BTW, I'm WMW in the weightweenies thread but rruff everywhere else.

Thanks for chiming in Josh. Personally I don’t have a dog in this fight. As a neutral and somewhat educated observer I just found it interesting that Hambini’s results stacked up about the way I would have expected. I know Zipp has put a lot of R&D into their rims and came up with a shape that both Roval and Bontrager (D3) seemed to settle on as well after doing their own R&D. A few years hence and both Enve and Bontrager (XXX) produce new rim shapes that are remarkably similar to each other. As far as white papers go I place more faith in Trek/Bontrager than just about any other company (no offense to Zipp) so in the case of the XXX I believe Bontrager when they say it is both faster and more stable than their prior D3 rim shape. The fact that three others came up with a similar shape... seemingly independently... tells me it likely is a (slightly) better mouse trap.

That said, rims like the Enve 7.8 and 808 are so different in shape but so similar in performance it leads me to believe there’s more than one viable approach. The big question mark, in my mind, is why the Flos performed so poorly.

[quote GreenPlease The big question mark, in my mind, is why the Flos performed so poorly.[/quote]


Yeah. Even if the whole data collection/yaw distribution issue is problematic, that chart labelled "Steady State vs. Transient Drag Characteristics" is compelling purely as a standalone thing.

Also I used to scoff at Zipp's dimples. My scoffing is tapering off a bit now....