waterrockets 

A teammate and I did some aero testing on Wednesday during our lunch ride. It was calm when we departed, so we planned to do coast-downs in various positions, comparing max speed.

Unfortunately, the wind did start up, and the first couple coast-downs were clearly invalid.

So, rather than waste the ride, we decided to head to a more flat course and try something new: One of us would ride in a repeatable static position (locked straight elbows, in the drops) and target 350W. The other, who was testing positions, would match speed from about six feet off to the side. Then we would compare power. Test runs would be 30 seconds, and we'd look at the average.

Well, on the bike, we were getting a fair bit of variability with identical positions, which was a little discouraging. We figured that the data on computers at home might tell us something useful, so we completed our tests, hoping for the best.

So, at home, we trimmed off the first and last 5" of each run and found that we were w/in 1% consistency on expected power. That wasn't enough to clarify his aero bar position decision (tipped up vs. flat), but it was enough to see the difference for me on the road bike between drops/locked elbows, drops/deep bend elbows, and IAB.

We're going to increase test duration and number of tests to see if we can separate the positions a bit more. We're trying to keep the speeds similar to competition speeds (25-27 for road positions, and 27-30 for TT positions).

Another analysis I'm going to do is to see if I can get an approximated CdA out of the numbers to see if we get more separation between the positions.

Anyway, I get a 10% power savings between locked straight elbows and a deep bend while in the drops. I usually relax in the drops with a 10-15 degree bend or so, which would probably put me at a 5-7% benefit for working down low.

IAB was about 2% more power required compared to deep bend in the drops, (which has my shoulders about 1.5" lower than IAB). The advantage here is that it's much more sustainable for long pulls. For short pulls though, and for establishing a break, it's good to know that I'm getting a good benefit by hunkering way down in front. I was thinking that would be the case, but there is a point that it would be too far down, and I don't know where that point might be.

Enthalpic

Have you checked out the Rchung method? His virtual windtunnel aka aerolab is pretty cool.

echappist

WR, thanks for sharing. Just to make sure i understood everything, the ranking from lowest to highest drag goes: drops, elbows locked <IAB<<elbow straight on drops, is that correct?

Also, would you be testing for elbows locked, hands on the hoods?

waterrockets 

I have, but it I haven't seen the latest stuff. I was looking at it in 2008 or so, and it seemed a little complicated, and if I remember correctly there were talks of having a mobile weather station to capture conditions or something. The method is very intriguing, but what we did today was pretty simple, so if we can get useful results, then it might be another route. We'll see where it goes, and I'll re-check the Rchung method again.

Close, it's elbows locked in the drops, not the hoods. So there's a 10% difference even with my hands in the drops, if I just bend my elbows deep.

My point though wasn't my results, but my method. It's an easy test, so it may hold some promise if we can bump up the consistency enough to see the difference between stuff like having a tt helmet or not.

aicabsolut

Interesting, WR. That sounds better than coast downs since you can add that factor of sustainability (comfort) in there. Too bad I don't have a power meter or I'd go do this tomorrow (my first classic ITT of this season is Sunday).

kleinboogie

Get GC 2.0 and use Aerolab. It saves a bunch of time. Still have to get the other numbers but the interface is easy. GL

waterrockets 

Thanks kleinboogie and Enthalpic, I'll go recon the GC/Aerolab deal.

kleinboogie

Aerolab resources:

Wiki: https://bugs.goldencheetah.org/wiki/g...heetah/Aerolab

Crr: https://biketechreview.com/tires/roll...75-roller-data, probably want to do your own tests but this can get you started.

Rho: Take the weather data (barometric pressure, air temperature, course altitude and relative humidity) from the time you'll ride, or an estimate from a similar day, say from weather.com and find an air density calculator that doesn't take 15 minutes to figure out how to use. I haven't found one but this one has a lot of data (https://wahiduddin.net/calc/density_altitude.htm). It has a calculator but I'm not sure how accurate it is since it doesn't consider altitude.

In the end you'll have enough information to run the optimalcycling.com program, if you can get it to work. https://www.bikeforums.net/showthread...-of-Qatar-2011

GL

SpongeDad

I think it's worth it to get good barometric data from approximate time of ride. You can usually do it with web access with a phone, so it's pretty convenient.

Racer Ex 

Calcing CdA is just running data you already have through a math formula.

FWIW you can certainly see benefit/not using this protocol, but it's going to be difficult to get a handle on fine differentiations. 1% variation on same/same is actually pretty big, my coast downs seldom show more than 0.4 % and often I've had 3 runs at 0% based on 0.2 km/hr speed increments.

Nate552

I've done quite a bit of Virtual Elevation testing and will throw in my $.02. I think you can get some good information and ideas from using this but I think it's also important to understand the protocol (for both coast downs and VE) and it's limitations. I prefer the virtual elevation over coast downs. For me, I found it more accurate, easier to accomplish, and I am able to get 10-12 runs in fairly quickly to help me weed out noise. I also prefer the "bowl" course because of the terrain around me. You don't have to have constant power or speed for the VE test either, I actually vary speed and power purposefully. The nice thing is that when you are looking at the results in aerolab you can quickly see if you have a valid test. If you do, your VE "hills" will look almost identical. If you have some "noise" that's easy to spot too, you VE "hills' will have different shapes. You need to be careful with your testing protocol though. Wind matters! To get a truly valid test there better not be any wind. I mean none. Zero. Nothing. Now, for me that's paramount. I'm often looking for smaller changes when I'm doing VE tests so any breath of wind and I'm not trusting the test. If you're looking for big changes, well, some noise is ok I guess. (But you probably already know what the result of a "big change" would be anyways, so why test it?). Cars also, will cause an issue. A car passing you during a test will mess it up. The car will also dirty up the air for a few minutes too. So if you're looking for real solid results, throw out the run when a car passes you. I'll mark it on my PT cpu with an interval count during my turn around so later I can remember that the run is invalid. I also recommend gather air information from the closest weather station. (I like wunderground.com for this) Ideally, you'd test near an airport so you'd have real accurate air data, but that's not always feasible. If you get good air data, crr info, and good tests, you can actually get a reasonable idea of your cda. But like Racer X said, it's just running the numbers through a math equation. On limitations, don't forget that this only works for 0 yaw. People seem to forget that. From a TT perspective, I have yet to do a TT that was at 0 yaw (indoor track is another matter). I usually see an effective yaw of 3-8 degrees, and sometimes more, especially in the DFW area where the wind can be quite strong. So, my testing may or may not be accurate for what my racing conditions are. Why does that matter? Because what's fastest at 0 yaw is not always fastest at 5, or 10, etc. To illustrate, I spent 3 VE sessions on different days testing 2 helmets. I knew both were good for me, but I only had tunnel data on one of the two (on me). So after all the sessions (with some runs being tossed due to the slightest breeze popping up) I came to the conclusion that my original helmet was fastest, but just barely. But because this was VE, the test only reflected 0 yaw. A couple months later, while at the tunnel we had a few extra minutes so I jumped in a tested the 2 helmets. First, I thought it'd be interesting to see if my VE testing had gone well, and two I wanted to of course know which was fastest. Sure enough, the VE test proved correct and my original helmet was fastest.....at 0 yaw. But at 5-20 the other helmet beat it. Hmmm....So, in real world TT conditions the helmet that tested slower using VE (or coast downs) was actually the fastest. Just a little story to help illustrate some of the limitations of VE or coast downs. I'm attaching a few pics to illustrate a good vs bad test.

ericm979

There's an interesting discussion of yaw angle here: https://djconnel.blogspot.com/2011/02...rt-frames.html

Racer Ex 

FWIW, I'm going to stop using "coast down" which is a totally inaccurate term, and start using TVR (Terminal Velocity Run) which is much more accurate.

RChung

Nice summary.

RChung

I've never tried this because it requires two people and two power meters but I've heard of people doing this. In theory it's a good way to control for wind and terrain though since I've never seen the data produced by these kinds of runs I can't say for sure. But it was a good idea to try.

Nate552

Thanks. (Props from the master!) I think I've read most everything you've published on the internet on VE.

RChung

My condolences. You'll never get those minutes back, you know.

waterrockets 

I was thinking though that it doesn't really correct so well for the terrain because I weigh more than my test partner. The power data don't know that I guess it's a small impact, but we were testing on a bit of a descent to get the speed up and make the air a bigger factor.

I haven't had time to get back to this thread since last week. I was planning to do a quick CdA calculation on the test data, just letting hills and winds fall where they may. I figure comparing the two of us, it might bump the accuracy up a bit (looking at ~CdA vs power). When I get some time to beat on this a bit, I'll post back if I find anything interesting.

It's already interesting that the deep bend elbows give me as much of benefits as I was seeing. Good mental note to make for bridging and attacking. Kind of a "duh" thing on that one, but I had no idea I'd be looking at 10%. At bridging power, that's going to be around 40W.

RChung

Not too much of a problem, as long as you (now) know your weight and that of your test partner, and have both data files. VE will take your weight and the terrain into account.

I haven't had time to get back to this thread since last week. I was planning to do a quick CdA calculation on the test data, just letting hills and winds fall where they may. I figure comparing the two of us, it might bump the accuracy up a bit (looking at ~CdA vs power). When I get some time to beat on this a bit, I'll post back if I find anything interesting.

It's already interesting that the deep bend elbows give me as much of benefits as I was seeing. Good mental note to make for bridging and attacking. Kind of a "duh" thing on that one, but I had no idea I'd be looking at 10%. At bridging power, that's going to be around 40W.[/QUOTE]

SpongeDad

In response to Nate's post on yaw and VE limitations, would it not be correct that if you were doing comparison tests on a loop course with consistent wind, that the relative cost/benefit of a position or equipment change would be accurate (because each iteration would be exposed to same wind at various yaw angles) even though the absolute VE-derived CdA might be off for each run?

Nate552

The wind wouldn't be constant enough. There is no such thing as a constant 5 mph wind in nature.

SpongeDad

True enough. VE on its own isn't going to be a controlled windtunnel session. But in the context of "aero testing on the cheap," if we modify my hypothetical doing say 4 laps in one position and then 4 laps in the next, have we averaged (a sloppy word here) the variables out enough be able to make reasonable determinations?

At a gross level - hands on hoods vs hands in drops vs arms in clip on bars - the answer is clearly yes. I've done it on a 1.2 mile loop that definitely has some cross wind by mid day. Now whether I could tell the difference btw helmets or a 1cm drop in bar height, that's a different story. (I'm planning on a crack of dawn / still air session to see if I can tease that out.)

I will be very interested to see how Chung on a Stick performs and how it compares to iBike once it's released.

Nate552

Well, there are a few problems with that. When there is wind, it screws up your virtual "hills." (See post where I attached a screen shot of a test where there ended up being just a little wind). If you change your cda, the hills move, but you have no idea what the "screwed up hills" should look like since they are not identical. If you set your cda at .250 they look all wompy. Then you set it at .260 and they still look all wompy. So which is it? .250 or .260? You don't know. If you've used VE in Aerolab you'd have a better idea of what I'm talking about. Also, with the inconsistent wind, what yaw would you really be testing? 5? 8? You wouldn't know, and the point of testing is to compare two apples, not an apple and an orange. Say you did a run of 4 laps and the wind speed was 5 mph from the SE. Then, you changed helmets and did another 4 runs and the wind was 3 mph, but changed direction ever so slightly. Well, your effective yaw for the first test, on the out leg was maybe 5 degress, and -5 on the way back (I'm just making stuff up). Then the second test the effective yaws were 2.5 and -2.5. So, even assuming the wind speed during each tests didn't change, you've only compared a helmet at 2.5 yaw to another helmet at 5 yaw. Well, that's not really a good test. And the reality is, that a 5 mph wind, is not a 5 mph outside. It's 2, then 6, then 3 then 4, etc. Ever played with a anemometer? Get one and go outside on a day that weather.com says is a 5 mph day and watch it. It'll be all over the place speed, and direction wise. Like Rchung has stated in his documentation, zero wind. None. Nada.

waterrockets 

Would be cool to have a bluetooth helmet-mounted anemometer and wind direction indicator that could report into a phone file. It would only show apparent wind, but that could be accounted for in the algorithm easily.

Nate552

That'd be cool alright. I'm not sure how easy that algorithm would be though. HED used to have an apparent yaw calculator on their website. I found a few bugs with it and emailed them about it awhile back. They admitted it was "buggy" and have since pulled it down. I'm not sure if it was because it was too hard to get it to work or what.
And Chung on a stick will be a nice advancement, but I still have questions about how much impact the device will have on wind flow over the rider, and how far out in front of the rider will it need to be to have true readings? If you watch a video of smoke in a tunnel you can see the smoke stream start to "bend" before it hits the rider's hands, wheels, etc. So IMO it would have to be out in front of you, not like how the iBike sits between your arms.

You can see a video I shot while in the tunnel showing smoke here
https://cobbcycling.com/aero-video.cfm (At the bottom)

And here's a nice pic from the A2 tunnel that shows a string bending in front of the rider.