Aero testing
 

Folks,

After reading this article, about parsing out CdA from power data, and this sister article about parsing crr and CdA, I've been inspired to do some testing. However, I can't find anything very specific about the appropriate size of samples for this testing. Some of the other sites I've found say useful things like "enough data points," or "with enough samples," but fall short of giving anything related to meters, or minutes for creating these data pools. One paper said they used a 470m flat stretch of road, or suggested "maybe two laps" of a velodrome.

I had a rough idea for a testing procedure that was something like 2 trials at 3 speeds, in both mass start position and aerobar position. From there I'd like to tweak some equipment changes, and maybe experiment a bit with elbow width, as that seems to be a hot topic. I'd go to the Springs track, as the bubble keeps the wind at bay, and the conditions are normalized. Is a kilo per test speed adequate sample size? Should I look at 4 or 5 laps per trial? 6? Planned speed tests were 40, 45, and 50kph, as those seem to encompass most of my race night velocities.

I know of a few here are engineering types, and I know a few of you are more versed in stats than I, so I ask some advice if you all would be so kind. Please and thank you!

Not an engineering type, but one thing I've found that makes testing significantly more consistent is a power meter. Most of the time trialists and pursuiters I know rely on velodrome laps combined with power to dial things in. Personally, I've found that works as well, because rates of perceived effort vary for me based on numerous factors, so I struggle to retain consistency between efforts.

I agree with SyntaxMonstr that you need to correlate power and speed. Plan to keep either power or speed more or less fixed for your testing period.

I'd probably divide your trials into two phases: Start & Steady State, with the bulk of your analysis dealing with the steady state condition (either fixed power or fixed speed, whichever you prefer).

Although, for some things like wheel weight, perhaps initial acceleration is also important.

I think 3 conditions (40/45/50) is unnecessarily complicated, and you'll probably get better data if you reduce your efforts to two conditions, say 40 KPH and 50 KPH, and thus have more trials or laps in each condition.

I'm not convinced that you would universally need to even look at two speeds. A single speed should probably be sufficient, and I'd choose the fastest speed you can get repeatable data.

However, it might depend on what factors you're playing with (hopefully changing as few things at a time as possible).

Something like bar type may be primarily a wind resistance configuration, however, you may also find yourself positioning your body differently at low and high efforts. Anyway, some things may only need to be tested at high effort.

Other aspects like tire choice or pressure might have both friction/rolling resistance and wind components and might be tested best with multiple speeds.

my advice for chung/aerolab stuff.

If you're going to be doing it on an indoor track, just spend your first time trying to get good, repeatable data... don't jump straight into a vs. b testing.

Do a run as close to race pace as you can without it messing up your form (your goal is to eliminate anomalies) I almost always throw out my first and last lap and take the middle... so 6+ laps would be ideal. The more, the better. (up to the point that you start breaking down)

For your first time, repeat that 4-6 times (so 4x6 laps or something like that) and see if when you plug the data into goldencheetah if you get similar "hills and valleys" between each test. If you do, next time start testing a/b/a/a/b/b/a/b, etc. It takes time to get it right. I've scrapped entire sessions because points don't line up (maybe I was moving around too much, other riders on the track, temperature changed without me noticing, whatever)

Always try to keep a "baseline" run... something you go back to at the start of every time you test that you already have (trusted) data points on, that way you can look back against that and say "hmm, last time I did a baseline run it lined up at .xxx cda, this time it looks like it's .yyy, what was the difference in my protocol?" (and that allows you to adjust your results from that session to more accurate. Again, if you trust your protocol from before)

Good luck, you can learn a lot with the excellent tools out there given to us for free.

*edit* also, while it won't be as accurate as it would if you knew your exact crr, just plugging in a steady number (so long as you use the same tyres pumped up to the same pressure, in similar temp) into GC's crr section is a lot less work than trying to figure out crr and cda. While it'll mean your exact CdA # won't be as accurate, it's still more than sufficient for A vs. B testing imo.

Agreed.

Also: take notes. Lots of them. When I started doing a little bit of AeroLab testing I realized that if I wanted to remember what the test was about, I had to write down what gear I was using. Every single piece of it: "Road bike, 38cm traditional bend drops, seatbag, 2x 16oz water bottles, SS bibs, jersey, armwarmers, Bont shoes" and so on and so forth.

Where you will get good consistent payoff is organization beforehand. Come up with a game plan before you even head out about what you will be testing in what order, and what "recipe" you will be using. This does two thing: 1) it helps minimize mistakes of data entry, the more automatic you can make things, and 2) it will force you to really think about what exactly you want to test, narrowing things down, and keeping you and your results from getting sidetracked.

I did something similar to this over 10 years ago back when all we had for track were either SRMs or PowerTaps. Being a junior, I had the PT. It was done on an outdoor track, at about 80% race speed (which was the max speed for that particular track) on a pretty much windless, cloudless day. I had the testing protocol line-up ready to go for weeks until the perfect weather came. Headed out to the track with everything I wanted to test and more. In a 2 hour session I was able to dial down my mass start/sprinting position, TT/kilo position, wheels (but not tires, it was an asphalt track), helmets, gloves, and skinsuits.

I had made my notes/sheets in microsoft excel and printed them out. All I had to do when I got to the track was follow the directions for each run, and input the times and wattage. I did some warm ups before the protocol with an un-aero set up, and an aero set up, just to see where/what range my numbers would fall in, and then went off the protocol after that. We had brought a portable weather station with us (the kind you can find at most big box stores) to make sure that the conditions stayed consistent.

We broke testing down into 2 periods which were dictated by position (Sprint/mass start and Kilo), tested various positional changes, and then tested clothing/helmets, followed by wheels. The clothing equipment changes were done after we had found the fastest position and dialed the bike into that position. This was all done on 32 spoke wheels, and once the clothing tests were done, then the wheels were thrown in with the new fastest combo. Once the Sprint position was tested, we repeated the whole thing with the Kilo set up. I did my laps up on the stayers line because the track was shallow, and bumpier at the black line. It also made for a smoother line to follow and allowed more consistency in the results.

When we were done, we had a little notebook of about 20 pages worth of notes and results. This wasn't full up full pages, but I had made a separate page for each change, with room for notes on the bottom half of each page. We did 2 lap runs (500m) for each "heat", flying start so that you crossed the line with a consistent speed, and did 3 heats for each set up. This gave 6 laps to average out the data. All laps were recorded, as well as the average, which was our "Final" number. 3-4 laps were done before the first heat after a change at a light pace to loosen the muscles up. All in all it was about 30km of riding to get the testing done, so about an elite level points race.

Having that notebook set up made decisions/references after the fact so much easier. It also made sure that we weren't entering data from one set-up into another's.

It's also good to remember to not just test aero (by using power), but to test power by using position. Basically the reverse of the aero test. Pick a position, set up on the trainer, and do some efforts at race pace to see where your optimal power is, and where it starts to really drop off. If your refined aero position is somewhere in between these, you can then go back and forth within these parameters to find your ideal position. You may find that for someone that does both Kilo and pursuit, they may settle on two different positions that take into account both of these factors.

The nice thing about testing this way, is when you want to refine/repeat in the future, there is much less to look at, and the adjustments tend to be finer as well.

Find someone familiar with Design of Experiments - that way you can cut down on the number of tests. Perhaps someone at a local university - students often are looking for a project/term paper topic by combining the variable you mention.

Also I've seen an Olympic track team ride a track bike where the rear wheel had an integral motor controlled by an iPhone - if you could such a wheel then perhaps you could concentrate on obtaining aero data - just thinking "out of the box."

Actually Design of Experiments (DOE) technique allows two or more variables at a time. DOE is used extensively in industry.

Some interesting devices coming on the market (supposedly this year) that should help remove some of the variables involved in aero testing outside the wind tunnel, such as this one.

This, at first DOE seems counter intuitive but is quite effective.

Intersting link jsk. Real time CdA would be huge.

Anyone worked with Aero Analyzer in Best Bike Split?

^ Notio is soon releasing their setup to test in real time. I messed around with a prototype last year and the results were pretty close to what I got in A2 wind tunnel. (I did test it on an outside track - Rock Hill)

There is also an app, cdacrr by Bungo... I've messed around with it a little bit, I think you could get good results out of it testing A/B

BBS Analyzer is...accurate-ish, until it isn't.

Excellent point, and I've wondered about that when I was experimenting with my Funny Bike and Recumbent Trike (for fun).

One could potentially get the perfect aero position, but not have rider efficiency.

Perhaps try some maximum effort laps to verify one can still hit the same top speeds.

Of course, that will also be difficult to measure for a few confounding effects:

• The Placebo Effect (new bike effect). Try something new, it must be better, and that can potentially influence performance.
• Or the opposite... liked the old one :)
• Plus training. It may not be as efficient simply because one hasn't trained and habituated to the new specs, so it could get better with practice (but always keep in mind alternatives if it doesn't get better).

This is the reason I mentioned further refinement. As long as one keeps getting better, then the incremental steps taken shouldn't be a hindrance to the rider (when it comes to time invested making changes and refinements). Usually it takes some time to adapt to a new position, especially if it's a radical change. It may need to be done in steps, so knowing where one produces the most power, and starting their, then working towards their most aero, may be the most efficient way to arrive at their ideal position.

Basically refining along a line of constant improvement instead of pendulum-ing back and forth between two points until a happy medium is achieved.

aero testing - Float Aero
 

Maybe consider using float-aero - he will guide you expertly through independent aero testing.