Road Bike Racing - Crazy aero test results

Ok, I didn't have much time, but I downloaded Alex Simmons' CdA-Crr test spreadsheet (http://alex-cycle.blogspot.com/2006/10/aerodynamic-drag-testing-using-power.html), and went out for some tests.

I found the flattest piece of road near here, wind was really low (calm to start, 1mph when finished), and I did tests both directions. Tests were in pairs, 2x20mph, 2x25mph, and 2x30mph, out and back, ~1 minute.

First position was with flat bars (flat forearms), and the 2nd position was with a pretty aggressive praying-Landis position (high enough that I could touch my helmet to my hands by dipping it slightly.

I think the wind played into it more than I thought, but the praying Landis position came back as being 65% less drag. The thing is that the praying CdA isn't believable:
Flat bars: 0.293
Praying : 0.191

This is with my road bike with forward post and clip-on bars, standard helmet, skinsuit, and 64 spokes. So, the .293 number is believable, but I don't know what to make of the .191. It is much less comfortable in the praying position, but I could feel that it was easier to go fast almost instantly.

Another hint that the wind played a role is that the Crr was twice as high for the praying position (0.007 vs 0.014). Still, if I set my weight to 0kg, (no rolling resistance), it didn't affect the CdA much on either setup.

Ok, I pulled the 30mph tests out of the spreadsheets and came out with much more believable numbers:
Flat CdA: 0.363
Praying CdA: 0.281

So, a ~22% improvement is more believable. Also, the absolute numbers seem more believable to me in general. I think my speed was too variable trying to hold 30mph for a minute. It's not too tough going that fast, but it's tough to be subtle with the speed control.

I can't believe I'm beating Pcad to this one.

Can you explain what the hell CdA/Crr is?

:)

...I think my speed was too variable trying to hold 30mph for a minute. It's not too tough going that fast, but it's tough to be subtle with the speed control...
Thats the difference between you and me. I don't think I couldn't hold 30mph for a minute even if you strapped a Saturn V rocket to my arse.

I can't believe I'm beating Pcad to this one.

Can you explain what the hell CdA/Crr is?

:)

CdA = coefficient of drag multiplied by frontal area. Cd would be just the coefficient of drag for a given shape moving a certain direction. You multiply by area to come up with the aero drag. So, I could calculate my Cd if I knew what my A was :)

Crr is the coefficient of rolling resistance. This is dependent on tire dynamics and rider weight, and also includes wheel bearing friction the way I'm measuring. If you use an SRM, Quarq, or Ergomo, then it will include drivetrain losses as well.


Thats the difference between you and me. I don't think I couldn't hold 30mph for a minute even if you strapped a Saturn V rocket to my arse.

Easy: blame your position ;)

I think my speed was too variable trying to hold 30mph for a minute. It's not too tough going that fast, but it's tough to be subtle with the speed control.

Have you looked into the Chung method. It doesn't require a flat road, constant speed, or lack of wind. All you need is a short to moderate loop that you can ride without braking. Wind should be constant (though not necessarily zero) and it helps if there is some change in altitude. There's a lot on it, and some Excel files to work the data, on the Wattage list.

Crr is the coefficient of rolling resistance. This is dependent on tire dynamics and rider weight, and also includes wheel bearing friction the way I'm measuring. If you use an SRM, Quarq, or Ergomo, then it will include drivetrain losses as well.



Makes me wonder, has anyone used both a SRM and a powertap together to test/compare drive-train friction?

As I'm sure you're aware, WR, the results at 30 MPH will be much less sensitive to noise than at 20 MPH, due to the sheer power #'s involved.

Makes me wonder, has anyone used both a SRM and a powertap together to test/compare drive-train friction?

Haven't seen it done, but it's definitely been considered by many. The problem is usually finding a SRM user that will let you steal his stuff.

Well that, and getting all the flexy-frame hysteresis accounted for.

Have you looked into the Chung method. It doesn't require a flat road, constant speed, or lack of wind. All you need is a short to moderate loop that you can ride without braking. Wind should be constant (though not necessarily zero) and it helps if there is some change in altitude. There's a lot on it, and some Excel files to work the data, on the Wattage list.

Yeah, I read about that method while I was studying Alex's. I'll give that a run sometime too. I'll be taking a look at my normal road drop position soon too. Maybe I can use Chung's for that test.

Power meters are kewl.


As I'm sure you're aware, WR, the results at 30 MPH will be much less sensitive to noise than at 20 MPH, due to the sheer power #'s involved.

Yeah, I was much more excited for the higher speed results to be in there. I just wasn't prepared for what it would take to maintain the speed. I had a +/- 10% variance in there. Once I saw something like 28mph, I'd get into it, go right past 30, and over-react coming back.

I should have dialed it in on a near-by hill to get the cadence feel for a gear, then I could use that to guide me in the test. As it was, I was doing these tests rapid-fire, just turning around and getting on it right away. So fatigue was preventing easy speed control too.

The 20mph tests were a bit of a joke, requiring only 170-200W. Much easier to control, but I felt like inertia had a huge role in it.


Well that, and getting all the flexy-frame hysteresis accounted for.

;)

Alex Simmons' CdA-Crr test spreadsheet

link?

Ok, I pulled the 30mph tests out of the spreadsheets and came out with much more believable numbers:
Flat CdA: 0.363
Praying CdA: 0.281

So, a ~22% improvement is more believable. Also, the absolute numbers seem more believable to me in general. I think my speed was too variable trying to hold 30mph for a minute. It's not too tough going that fast, but it's tough to be subtle with the speed control.

I believe that a lot more than the first post. .191 is probably going to be borderline impossible for you to attain without spending some $$$ on new stuff, etc.

link?

D'oh! Sorry, I'll throw it up in the OP as well :o: http://alex-cycle.blogspot.com/2006/10/aerodynamic-drag-testing-using-power.html


I believe that a lot more than the first post. .191 is probably going to be borderline impossible for you to attain without spending some $$$ on new stuff, etc.

Yeah, I was expecting 0.3 kinds of numbers, and the .191 pretty much discredited the whole thing for me. Tossing out the four tests with highly variable speed seems to have fixed it up though. I really was just interested in seeing if I should have the bars up or flat at this point.

I'm also so tall that I really don't think 0.2 is possible for me. Too much leg sticking down there. I wonder if there's a body-building program that can develop fins/farings for the backs of your hams and calves :lol:

I'm thinking about building a TT bike up though, on the cheap. Focus on used aero-ish frame, front wheel, low and tight cockpit, and wheel cover for the PTap. We'll see.

Thanks. Good lord, getting a PT is going to cost me a lot of time when I start running every test ever listed on the internet.

Thanks. Good lord, getting a PT is going to cost me a lot of time when I start running every test ever listed on the internet.

Testing is training (A. Coggan)

Or if you take my view... Racing is testing.

Or if you take my view... Racing is testing.

Yeah, I'd never hit my 1m or 5s numbers in a race though...

Get a motobecane for your TT rig.

Then you can dispel two bf myths. You can be badass on a steel frame and on a motobecane.

Now if your steel frame were a bd bike too, then that'd be a whole nuther level ;)

Yeah, I was much more excited for the higher speed results to be in there. I just wasn't prepared for what it would take to maintain the speed. I had a +/- 10% variance in there. Once I saw something like 28mph, I'd get into it, go right past 30, and over-react coming back.

My god! Ignoring the offset due to rolling resistance, you had to have been doing 700W for something like 40s. That's crazy sweet.

Force to overcome drag (A*Cd*v^2) goes with the square of velocity, and power (F*v) goes with the cube of velocity. So extrapolating from 200W at 20mph, at 30mph average power is about 700W, not counting drivetrain and friction losses. What is your 1 minute power?


P.S. - I'm a PT noob, just got the new freehub onto my new (used) PT last weekend. I've been out on it a bit but don't have enough data to get a good feel of real world numbers. Cool stuff.

Eh, every max number I have, for every duration, is from a race.

If you're going to build up a TT bike on the cheap, with a focus on what you want... Older Cervelo P2/P3 aluminum frame, a decent Vision Tech aluminum cockpit, and 9 speed Shimano. I doubt it'd be uber cheap, but you could probably pull it off for $700 if you got the right deals on the frame.

WR: what was the r^2?

Next time, add a slow run (15mph or even 10 if you can keep it steady) to anchor the Crr.

Any systematic difference between out and back runs (wind and/or slope)?

I prefer Chung testing, though do it on a windless or near windless day. Despite the claims, I see substantial wind effects with it. There are ways to get around that (won't go into it here) but it's easy just to pick a good day.

iBike works well too and is the most brainless.

Yeah... I did the Chung thing once. I didn't have much luck, but that was probably in large part due to the runs being low speed.

I want to suck it up and buy an iBike, but there's no way I'm laying down the cash for one of those things now (or probably ever, unless their price gets cut by more than 1/2).

My god! Ignoring the offset due to rolling resistance, you had to have been doing 700W for something like 40s. That's crazy sweet.

Force to overcome drag (A*Cd*v^2) goes with the square of velocity, and power (F*v) goes with the cube of velocity. So extrapolating from 200W at 20mph, at 30mph average power is about 700W, not counting drivetrain and friction losses. What is your 1 minute power?


P.S. - I'm a PT noob, just got the new freehub onto my new (used) PT last weekend. I've been out on it a bit but don't have enough data to get a good feel of real world numbers. Cool stuff.

The 30mph bits were in the 420-450W range... 1-minute mean-maximal power is 801W... ;)


Eh, every max number I have, for every duration, is from a race.

Yeah, that works great for TTs, but you really just can't produce your best effort in shorter durations in the middle of a road race (although you're with Racer Ex, who has his best numbers in races as well -- but I still don't believe those are his best numbers ;)). Now, if you raced match sprints, kilo, and pursuit, I'd believe you :)

If I busted out 801W for 60s in a road race, I'd be off the back when they caught me trying to recover at 15W. At the end of the race, I'm only good for maybe 550W or so.

In the race I won, vs. my personal bests:
5s: 1220W vs. 1570W test (middling Cat 5 vs. middling Cat 1)
1m: 550W vs 801W test (middling Cat 5 vs. Cat 1/domestic pro) <-- I know, I know, meaningless, but illustrates the difference
5m: 336W vs 457W test (middling Cat 4 vs. Cat 2/Cat1)

So, without true, taper/recover maximal attempts, you really don't know what your max powers are for 5s, 1m, and 5m. That's fine though, because you and your coach have the numbers you need to improve. I'm my own newbie coach, and recognizing my true ability profile was important, as it led me to the training changes to improve my FTP.


If you're going to build up a TT bike on the cheap, with a focus on what you want... Older Cervelo P2/P3 aluminum frame, a decent Vision Tech aluminum cockpit, and 9 speed Shimano. I doubt it'd be uber cheap, but you could probably pull it off for $700 if you got the right deals on the frame.

Yeah, that's the direction I'm leaning. Those Fuji Aloha aluminum frames come pretty cheap used too. We'll see.

Tonight, my legs fell asleep again :mad: I started off slow to try to prevent it, and all it bought me was a 5W lower average at the end when I was unable to ramp it. The aero changes and possibly the wind conditions combined to save me a couple seconds in the TT anyway, so that's cool :)

It's frustrating though, because I think I could pull off a 365W run in my road position, and not have my legs fall asleep. Next time, I'm just hitting it like a 10m time trial, let the legs fall asleep, sit up, work it out for a minute, then get back on it.

WR: what was the r^2?

Next time, add a slow run (15mph or even 10 if you can keep it steady) to anchor the Crr.

Any systematic difference between out and back runs (wind and/or slope)?

I prefer Chung testing, though do it on a windless or near windless day. Despite the claims, I see substantial wind effects with it. There are ways to get around that (won't go into it here) but it's easy just to pick a good day.

iBike works well too and is the most brainless.

Hmmm. r^2, I didn't even look at it. Pretty sucky, at .605 (praying), and .713 (flat). I'm guessing it should ideally be 1? What's r^2 for a wind tunnel test usually?

There was a 1mph wind that started up when I was testing praying. There was also a very slight rise (1m over 700m or so?). I still think it was solid enough to pick between the two positions.

The 30mph bits were in the 420-450W range... 1-minute mean-maximal power is 801W... ;)

WR, I must be missing something. Even if you were only doing 150W at 20MPH, when extrapolated to 30mph by the cube law, that's something like 500W. Doesn't follow aerodynamic laws. When cruising at 20, I put in 180-200W or so too. What am I missing? A slight tailwind both ways? 801W is insanely impressive... I've got a long way to go. :)


In the race I won, vs. my personal bests:
5s: 1220W vs. 1570W test (middling Cat 5 vs. middling Cat 1)
1m: 550W vs 801W test (middling Cat 5 vs. Cat 1/domestic pro) <-- I know, I know, meaningless, but illustrates the difference
5m: 336W vs 457W test (middling Cat 4 vs. Cat 2/Cat1)

Don't mean to be nosy, but what's your kg? I'm trying to get into perspective what're reasonable aspirations with all these cool new numbers available.

Eh, every max number I have, for every duration, is from a race.

If you're going to build up a TT bike on the cheap, with a focus on what you want... Older Cervelo P2/P3 aluminum frame, a decent Vision Tech aluminum cockpit, and 9 speed Shimano. I doubt it'd be uber cheap, but you could probably pull it off for $700 if you got the right deals on the frame.

I did $900 with 10 speed, almost exactly what you describe.

Hmmm. r^2, I didn't even look at it. Pretty sucky, at .605 (praying), and .713 (flat). I'm guessing it should ideally be 1? What's r^2 for a wind tunnel test usually?

There was a 1mph wind that started up when I was testing praying. There was also a very slight rise (1m over 700m or so?). I still think it was solid enough to pick between the two positions.

If you do the Excel yourself, or can modify what you've got, you can try to take a term for inertia. This might improve the correlation.

WR, I must be missing something. Even if you were only doing 150W at 20MPH, when extrapolated to 30mph by the cube law, that's something like 500W. Doesn't follow aerodynamic laws. When cruising at 20, I put in 180-200W or so too. What am I missing? A slight tailwind both ways? 801W is insanely impressive... I've got a long way to go. :)

I think the mild downhill was into the headwind, and the reverse on the way back? I dunno, this could be fallout from the reason I threw out the 30mph tests. They were really variable, and really only close enough to 30mph for 40s or so. Crappy attempts.


Don't mean to be nosy, but what's your kg? I'm trying to get into perspective what're reasonable aspirations with all these cool new numbers available.

80.9 kg. I'm a sprinter/kilo kind of guy (duh?), so my FTP is down in the 4.35W/kg range, and that number matters more than the others.

Just don't get hung up on other people's numbers. It's fine to see what some people can do on a given amount of training time, but don't let it distract your training focus on improving your own numbers according to your race limiters..

If you do the Excel yourself, or can modify what you've got, you can try to take a term for inertia. This might improve the correlation.

Yeah, that would be a pretty big change to the existing spreadsheet, since the Excel currently just takes averages and start/stop speeds (rather than taking samples).

The Chung method must take the samples, so I'll have to take a look at that.

I'd like to write one myself some time, but I've got too much going on to concentrate on anything for more than 15 minutes at a time it seems :rolleyes:

Well... If you're training at your max, you're never going to be recovered enough to race. But yeah, I expect my true 1m power is higher than what Cycling Peaks shows it to be.

I don't get the point of all of the testing, though. I say test maybe 4-5x a year, but I don't get the logic behind these weekly and monthly testing, though. You probably don't need to readjust your training zones that much, and really, who cares if your 1mp is or isn't higher? Results speak.

Well... If you're training at your max, you're never going to be recovered enough to race. But yeah, I expect my true 1m power is higher than what Cycling Peaks shows it to be.

I don't get the point of all of the testing, though. I say test maybe 4-5x a year, but I don't get the logic behind these weekly and monthly testing, though. You probably don't need to readjust your training zones that much, and really, who cares if your 1mp is or isn't higher? Results speak.

I agree, but I don't train at my max. Most of my weeks are three days of high tempo SST, one day of hill repeats, and either the TN'r or the TT. The 5m test is just my first hill repeat every 3-5 weeks (but usually my first is my slowest, so it's just on test days they I go all out). The 1m test is usually just integrated into an SST or endurance ride, and has little impact. 5s tests don't create any training stress at all, that I can tell. It's just a jump between light poles.

For me, I need data. Conducting tests every 3-5 weeks gives me true maximal numbers, which allows me to track my physiological adaptations and adjust my training accordingly. Also, occasionally going all-out tunes my RPE, and offers another training opportunity. Lastly, it just makes training easier (motivationally) when you see your numbers rise.

My testing isn't that invasive (e.g. the test is the first interval or first hill repeat), and it's infrequent (3-5 weeks).

Gotcha ;)

:) Now that I've got a power meter, my engineering personality is taking over, and I need data.

:) Now that I've got a power meter, my engineering personality is taking over, and I need data.

Need input!

http://i159.photobucket.com/albums/t141/chideltfrog/gas_station.jpg

... I don't get the logic behind these weekly and monthly testing, though. You probably don't need to readjust your training zones that much, and really, who cares if your 1mp is or isn't higher? Results speak.

Off the top of my head some tests someone might perform:

FTP, VO2max, Anaerobic, Neromuscular power about every four weeks. Not just to set levels, but also to see if improvement is as expected or has plateaued. Someone who is spending a lot of time doing anaerobic training should care if her 1 minute power is increasing. If it isn't, that shows the time and energy spent on this training is being wasted. Either the workouts need to be changed so there is improvement, or development has plateaued and training focus should change to something else. Either way, testing prevents wasted effort for no benefit.

TT start and turn around gearing and pace.

Position testing, road and TT.

Equipment testing, road and TT. With these tests a repeat of at least one earler test would be needed as a control and then multiple repeats to reduce error and increase confidence.

So it seems clear that depending on how many position and equipment choices one wants to explore, someone could spend a significant fraction of his riding time just performing tests. That's why the tsting should be viewed as a part of training and the testing designed not only to gather the data, but fit into the training plan.

Hmmm. r^2, I didn't even look at it. Pretty sucky, at .605 (praying), and .713 (flat). I'm guessing it should ideally be 1? What's r^2 for a wind tunnel test usually?

Ugh, with an r^2 that low I'm afraid the data is junk. Welcome to the joys of field testing ;). For the differences you're trying to detect, I'd want to see that at least over 0.9, even better over 0.95.

The r^2 is a goodness-of-fit metric for the regression used by this method, so it doesn't apply to windtunnels.

If you want to salvage anything from the data, I'd just assume a Crr and calculate the CdA from the 30mph runs. IIRC, you use a tire that's known to not be the fastest, right? Maybe 0.005 as a ballpark Crr. Plug it all into the equation of motion and see what CdA you come up with.

Yeah, that would be a pretty big change to the existing spreadsheet, since the Excel currently just takes averages and start/stop speeds (rather than taking samples).

The Chung method must take the samples, so I'll have to take a look at that.

I'd like to write one myself some time, but I've got too much going on to concentrate on anything for more than 15 minutes at a time it seems :rolleyes:

If you want, e-mail the power data file to me and I'll do it. Let me know how much you + the bike weighs.

WR, I must be missing something. Even if you were only doing 150W at 20MPH, when extrapolated to 30mph by the cube law, that's something like 500W. Doesn't follow aerodynamic laws. When cruising at 20, I put in 180-200W or so too. What am I missing?



Maybe the trouble is you only cube the wind drag portion of the power. So if you are at 150W at 20mph and 75% of that is wind drag, you get 150*.75*(30/20)^3 + 150*.25*30/20 = 380W + 56W = 436W. I don't know if 75% of power at 20mph going into wind drag is accurate, but that seems close to WR's numbers.

Ugh, with an r^2 that low I'm afraid the data is junk. Welcome to the joys of field testing ;). For the differences you're trying to detect, I'd want to see that at least over 0.9, even better over 0.95.

The r^2 is a goodness-of-fit metric for the regression used by this method, so it doesn't apply to windtunnels.

If you want to salvage anything from the data, I'd just assume a Crr and calculate the CdA from the 30mph runs. IIRC, you use a tire that's known to not be the fastest, right? Maybe 0.005 as a ballpark Crr. Plug it all into the equation of motion and see what CdA you come up with.

Good to know. I think I'll try Chung's test some time. I botched this one by not taking the constant speed requirement seriously enough. There's not really even a flat spot in the speed plot for the 30mph tests...


If you want, e-mail the power data file to me and I'll do it. Let me know how much you + the bike weighs.

I appreciate the offer, but I don't want to spread the time wasting around too much :)