Not currently, but I think it would be possible to add that if you know how to code.

I don't have any recent (decades) coding experience, but I do it in a spreadsheet.

Any wind really increases the scatter, though. Would be nice to have an aerostick and good software for this.

Thanks. MapmyRide has elevation of the street so I think I can use the calculator to determine air density.

I assume I should stay in one gear?

---

Geoff from Indy
http://www.tlcendurance.com

Staying in one gear is good as there is no loss from any changes and Eta stays constant throughout the test.

The correct elevation tool in GC will be a help.

---

Developing aero, fit and other fun stuff at Red is Faster

James ("SkippyKitten") has it right: for best results, you should probably stay in one gear. Most modern derailleur systems don't have a lot of loss in close gears so *if you need to* you could switch one or two cogs but I wouldn't flip from one ring to another. (I also have a Power Tap, so I don't really worry too much about drive train losses).

Another thing I do is to soft-pedal when coast (and I do coast). For me, remembering to soft-pedal is easier than trying to remember exactly where to hold my feet while coasting.

But I would say both of these latter two things are details when you start to get good and serious. When you first start out, there's already a lot to remember. That's sort of why I recommend looking at a regular ride or two in Golden Cheetah, just so you can understand how it works and how careful you need to be to get precision. The first couple of times you do this you probably shouldn't expect too much precision -- that will come with a little bit of experience.

I created an easy to use air density calculator here:
http://www.aerocoach.com.au/air-density/

---

http://www.cyclecoach.com
http://www.aerocoach.com.au

I created an easy to use air density calculator here:
http://www.aerocoach.com.au/air-density/[/quote]
Great work Alex!

---

Hamish Ferguson: Cycling Coach

Very nice!

I should clarify, there are two standards for showing barometric pressure: when you get it from an official weather service, most likely the pressure has been standardized to sea level. However, sometimes, you'll find the pressure reported as "station" pressure. In that case, the altitude at the station where the pressure gauge is located is already included. The Golden Cheetah air density calculator uses "station" pressure, which is why it doesn't have altitude. Official barometric pressure that's been standardized to sea level doesn't vary much over long distances so, as Alex says, you can grab the barometric pressure from any nearby location. However, the altitude can vary quite a bit depending on terrain, so if you're using official reports you have to adjust for that.

I just did my first test, yesterday. I have an old wired powertap hub. I'm using the spreadsheet version because GC doesn't like the csv files for some reason. I think I found the spreadsheet from AlexS.

I think I have a pretty good test location. It's a 3/4 mile, bathtub section that runs north/south and is sheltered by trees on both sides. The main stretch is mostly flat, and the ends turn uphill allowing me to coast down and U-turn. There is also very little traffic. I saw a total of two cars yesterday.

I just did two simple runs (5 laps each). Each set of 5 laps took about 12 minutes to complete. One on the bull horns and one in the aerobars. Assuming a crr of 0.005, I got cda of 0.330 and 0.242, respectively.

My main question is how to use the spreadsheet to tease out crr? I'd like to establish a known baseline crr for my tire/test area/road conditions.

I've read the pdf, and this entire thread. I think I understand the concept, just not the practical mechanics.

At one end of my test location there is a short hill that I can climb. It's 50 ish feet tall, and about 1/4 mile long...it is Texas, afterall. Is that sufficient? I'm guessing not, since any error in elevation would propagate into crr.

Would I ride up/down this hill at two different speeds, and then adjust cda/crr to level the VE profile and match the elevation change?

If that's not sufficient, would it be better to do multiple (5, 10?) laps on my planned bathtub, at increasing speeds? If so, how much speed change is practically necessary and how slow can I start? Do I need to ramp up AND down... Or is monotonically, increasing sufficient?

Any recommended search method for finding crr/cda values graphically?

Thanks... Sorry for the barrage of questions.

The best way to get Crr right in my experience is, yes, to vary speed. Without accurate elevation data over the course, it’s better to do this over different runs/laps rather than as part of one run. This makes the ‘shape’ of the VE profile less important than the overall VE change.

Try and maximise the difference in speed while still maintaining pose so that CdA is less affected. Get accurate spot heights for top and bottom and record mass when testing.

Also, a consistent road surface is helpful.

---

Developing aero, fit and other fun stuff at Red is Faster

Ok. Thanks.

Is it better to ramp over several runs, or is a single (or 2-3) slow and max-aero-fast pass sufficient? Eg 2x16mph and 2x25mph?

Do you have a good search heuristic for finding crr and cda? Ie, when to inc/dec crr/cda based on VE shape?

Good info. What sort of a protocol would you proscribe for teasing out the fastest air pressure for a given tire/rider?

More runs always better, but you don’t have to keep at the same speed for particular runs. If it’s a half pipe course, then go with the flow - speed up and slow down with the terrain. If you want to coast I tend to soft pedal, partly to keep things the same CdA wise, partly to keep certain power meters happy. Powertap doesn’t have any issues there though.

Just to be clear, it’s best to hold the same position for all the runs, so on the basebar, even though this does let your body move a bit more.

---

Developing aero, fit and other fun stuff at Red is Faster

Thanks. 10psi steps are realistic. Use the pump or gauge you use on race day. Start at low pressure and increase (debatable). Pump slowly to avoid heating things up. Overcast days are more consistent for Crr tests in my experience. Setting front and rear togetheris easier.

The Red is Faster system does give me a bit of flexibility on courses, like only running in one direction. This means I can use nice long climbs and not care about the turning point. I also have TyreWiz which is handy for this sort of thing.

---

Developing aero, fit and other fun stuff at Red is Faster

So, the underlying issue is that if you need to estimate both Crr and CdA, you have two unknowns rather than one, so you need at least two equations. If you have a hill of unknown height, that's a thrid unknown so you need at least three equations. You can get those equations by varying speed up the same hill, or by doing loops on your bathtub at different speeds. The wider the range of speed, the more stable the estimates are going to be (that is, the more precise).

So let's say you're using the hill. You take 3 runs at widely varying speed or power. Then you pop the data into the spreadsheet and either solve analytically for the CdA, Crr, and hill height, or you can try different values to solve graphically. (I solve analytically then take the estimated values of CdA and Crr and produce overlapping VE profiles as a diagnostic of fit).

In theory yes, but only if both Crr and CdA are independant from speed. Is this allways the case? When is it true?

They don't have to be independent from speed. They just have to be different functions of speed.

Good point. In the range of speeds we *usually* ride in, on the kinds of surfaces that we *usually* use for TTs or road racing or velodrome track racing, we can *usually* get away with the assumption that Crr and CdA are invariant with speed. However, there are some surfaces where I've gone from paved road to a short segment of dirt or mud or cobblestones and I suspect that the Crr isn't invariant with speed (that's not a wild suspicion -- I've got a tiny bit of data on that). Likewise (and this may come up more frequently now that airspeed indicators are starting to hit the market) in gusty wind conditions, especially from different directions, we might expect that the Reynolds number will *not* be constant so CdA will not be independent of either ground speed or airspeed. There are ways to estimate the Crr and CdA under varying conditions but, from a practical point of view, when I'm trying to estimate CdA it's simpler not to use a test venue with cobblestones mixed in with smooth asphalt. Similarly, while I do occasionally test when there are gentle breezes I never test when the wind is highly variable in both speed and direction. (The reason I have only a tiny bit of data on venues with cobblestones is because once I noticed it I avoided using those venues. I recommend that you do, too). [Edited to add:] Listen to Josh Poertner's anecdotes about doing testing on certain sectors of the Paris-Roubaix course. It's very cool.



So, above I wrote "usually." A good thing about the VE profile is that it provides a good diagnostic for when the conditions were too variable to get a good estimate. There have long been methods that will allow you to solve for two unknowns from data; the value of VE is that it provides a pretty handy diagnostic of fit and (especially) misfit.

Hi, I'm new to this forum. I have a question about repeatability of CdA testing obtained using the VE method.

I've been a cyclist for about 25 years (mainly MTB’ing - sorry!). In the last few years, though, I've got into time trialling. I'm an aerodynamicist as a profession, albeit for an aircraft manufacturer, so technical side of time trialling really interests and appeals to me.

I want to focus on improving my position and CdA, which is in the 0.27-0.29 region (so there's room for improvement!). After some haphazard experimentation with field testing over the last couple of years, this summer I spent some time properly researching field test techniques, and came across Robert Chung's VE method, which I really like and am now using.

I have fully read this thread and have used all the useful information and comments to refine my test technique. However, I still have the feeling that the repeatability of my testing is quite poor. The repeatability I'm getting is around 2%, for CdA differences between two tests within the same session. I was wondering what level of CdA repeatability other people are getting when doing VE field tests?

Here is a few details about my test protocol. Sorry this is long, but I hope more information is better than not enough:
1) I use a 1-mile oval road on an industrial estate, near Bristol in England (https://www.strava.com/...62528?filter=overall). It has approximately 30ft of elevation change. The slowest speed is about half the fastest speed when ridden at constant power. No braking is ever required. There is no traffic at the weekend. The only downside is the asphalt is quite rough (slightly rougher even than average UK roads!).
2) I do 5 laps at a time, so each run is 5 miles & takes 13-14 minutes.
3) I use a Stages left-hand PM. I know this is not ideal, but it's all I've got.
4) I keep my power at 230-240W, which is tempo/sweet-spot for me in my TT position. I know it's not necessary to keep power constant, but my thinking is that any left/right leg imbalance will then be similar from one run to another if I keep the power similar. It also helps me correct occasional signal dropouts when I know I’m riding at constant power.
5) I now do three runs per session, in an A/B/A configuration sequence. Three runs are about my limit (~40 mins total), and it also gives me a useful workout! I've tried doing a 4th run, but fatigue was affecting my form and position.
6) I now use a GSC-10 speed sensor, and have GPS turned off (thanks to reading this thread).
7) I analyse the VE profile in Excel, mainly because I like to know what's being calculated and I can manually identify and correct the occasion zero due to dropouts. In any case, I have checked my Excel-derived CdA against Golden Cheetah, and the CdA agreement was within 0.001, so I'm confident there are no bugs in my spreadsheet.
8) I have done one session where the 2nd run was done slower at 100-150W instead. This slower run allowed me estimate that CRR should be 0.006 instead of the 0.005 that I initially estimated.
9) Completely calm days (<3-4mph) never happen in the West of England, unfortunately. On a ‘good’ day, the reported wind speed is about 5-10mph, which I understand corresponds to wind speed at 10m. I therefore target those 5-10mph days to do my testing, which happen about approximately 20-30% of the year. Most of the days in the year are more windy than that. I realise this wind speed is not good for VE testing, but I can't do a lot about it. My hope is that within a single session the wind speed will not change much, but it's just a hope. I am certainly not expecting good repeatability between tests on different days, because the wind will certainly be different.

There are one or two remaining things I can try to further improve things:
- Stay in the same gear. Currently I shift across a few sprockets to avoid the spinning too fast or grinding too slowly (which is more a personal preference than a real constraint). I am fairly consistent between runs I think.
- There's a chance my body position between the first and last run of the A/B/A sequence differs, due to fatigue, especially my head position. One thought I had is to attach a zip tie to the stem to 'tickle' my chin so I know my head's in the right position.
- An bike-mounted anemometer would be brilliant, to try to adjust my calculations for wind variations, but I can't anybody selling them at the moment. The cost may be an issue anyway.

Any other tips or suggestions would be appreciated.

The repeatability I've achieved for three sessions so far has been 0.018, 0.005 and 0.006. This is the difference in CdA between the first and third runs (i.e. in identical configurations). Only my most recent session was done using the GSC-10 sensor, though, so I am taking that 0.006 value as my only really valid repeatability data point so far. A 0.006 delta-CdA equates to just over 2%. Of course, I need to gather more data to see whether this was a 'lucky' repeat, or an unlucky one. With enough testing I could get a repeatability standard deviation, I guess.

Could other people that have done multiple VE field tests please let me know what level of repeatability that they get within the same session please? Is 2% typical, or do people achieve much better? This information would help me decide whether I need to keep working to refine my protocol, or accept it as it is.

Sorry for the long post.

I used to use Aztec West as an aero testing loop many years ago (pre Strava!) but found that inconsistencies in the road surface would really add up and unless you hit the exact same patches each time it'll introduce a lot of noise into the data. I wouldn't recommend using it.

Your methodology is good but you probably want to reduce lap distance so that you can get 5 data points per experiment for VE profiling. How about hopping over the bridge to Maindy velodrome in Cardiff? You'll get a lot done there in an afternoon and it would take less than half the time to collect the same amount of data. For a flat quiet road if you were to do out and back testing Pilning St would be okay but a bit more exposed.

---

AeroCoach UK
http://www.aero-coach.co.uk

2% really is not bad in my opinion especially given your environment. My suggestion would be to do shorter loops. Do one or two laps instead of three or four. IMO this will spread any variability due to wind over your sample. Personally I don’t shift when I test but I’m not sure how the Stages power meter behaves with regards to cadence so... I’d suggest experimenting with not shifting and then look at the power file to see if there are any weird spikes or dropouts.

First, and this is important, pay attention to suggestions that Xav makes. He's the real deal. That he used Aztec West is amazing, almost as amazing as that there’s a place in England called “Aztec West.” Is there an Aztec East?

Second, considering your overall situation (wind, road surface, level of experience, power meter) a delta of .006 m^2 on a base of 0.28 m^2 is not bad.

Third, whether at Aztec West or elsewhere, I would vary speed/power more. In fact, I think it would be preferable to reduce the number of laps from 5 to maybe 3 if you varied the speed/power; I think a protocol where you did the first lap at around 160 watts, then around 200, then around 240 would help you better identify both Crr and CdA *plus* it would help you identify whether the wind was changing. Fewer laps at lower avg power may also help you reduce fatigue and perhaps help you maintain your "form."(All that said, I usually toss the first lap of a test since for some reason that first lap is always noisier than subsequent ones, so I often will do 4 laps for the first set with the plan of tossing the first lap, then do 3 laps for subsequent sets).

As for your original question, I've analyzed data from experienced riders on calm days with power measured at the rear wheel on an outdoor course where the CV was around 0.5%.

Thanks Xavier. Amazing coincidence that someone has also used Aztec West before. It's a small world! I know what you mean about the road surface. I have been trying hard to follow the exact same line each lap, 3ft from the kerb, but it has been in the back of my mind that it may be contributing to the variability. I might give Pilning Street a go. It's part of one the local 10m TT courses I sometimes ride, and thinking about it now, I can see why it would be a good stretch of road for out-and-back testing. I wasn't aware of that Maindy velodome in Cardiff, so thanks, I'll check that out.

By the way, I really liked the Mark Florence time trialling podcasts that you've done. I haven't heard the entire series of podcasts yet, but all the ones where yourself or Andy Fronchioni have guested on are excellent.

That's great, many thanks. The advice is certainly consistent from everyone that I should do less laps. My legs and neck will appreciate doing only 3 laps at a time instead, so I'll try that! I'll also experiment with using a fixed gear and see how that goes.

That's a great suggestion, thanks Robert. Changing power each lap will help to prise apart the different drag contributors, but keeping it constant within each lap will still allow me to fix signal dropouts if they occur. Doing an extra lap, with the option to discard the first one makes sense too.

Getting a CV of 0.5% is impressive, considering it's an uncontrolled environment (outdoors). That's not too dissimilar to the variability I see from wind tunnel tests done as part of my day job. If I can get close to 1% CV, I'll feel more satisfied I think.