Thanks for all the replies.

Perhaps I'll give it a try using both a GSC-10 and a hub-based one, connected to different head units and compare the data.

GreenPlease, did you ever try out your double wheel-magnet idea in the end and did it improve the data notably?

I've finally found a good half-pipe location, so I'm hoping to do a good bit of testing over the off-season, just need some calm weather now!

Yes I did try it out. Did it improve the data? Eh... not really. It's probably not worth doing tbh.

 
Did a Chung test with 3 different configurations today (intervals 2, 4, and 6). Air density was 1.212. Road was chipseal. I'm guessing Crr of .005.

Can anyone comment on whether this data is clean enough to draw conclusions from? And if so, what? I usually have better data, but today there was an occasional tuft of wind.


-- jens

---

My latest book: "Out of the Melting Pot, Into the Fire" is on sale on Amazon and at other online and local booksellers

Just quick-and-dirty look, but that look sez the middle interval had lower overall drag.

I can take a longer look tomorrow.

Do you have a .fit file? I have java code that reads fit files and computes CdA for each interval. It's very easy to use and, while I could analyze your worksheet instead, it would take a lot more work.

  
It's attached. You may have to negotiate some garbage. (e.g. at one point, I dropped the chain and coasted to the start of the loop to put it back on.)

cheers! Jens

---

My latest book: "Out of the Melting Pot, Into the Fire" is on sale on Amazon and at other online and local booksellers

I don't know your weight so I assumed about 180lbs. I also assumed a pedal or crank based power meter. Rho=1.212, CRR=0.005. Based on those inputs I got 0.176 for lap 2, 0.165 for lap 4, and 0.164 for lap 6. These are pretty low. If you weigh less than 180 they would go up a bit but my guess is either you were riding with traffic or your power meter was reading a bit low. Average power for lap 4 was only 125w, and for lap 6 was only 130w. Also note that these calcs are only valid if (1) the elevation that you end the lap is exactly the same as the elevation that you started the lap, and (2) there is no wind or the wind at least approximately averages out (such as on a loop or out and back).

 
Thanks for checking! I'm exactly 185lbs with the bike. I was doing an exact loop with same starting and endpoint. No close-by traffic.

The powermeter is a powertap, which might account for the low CdA numbers. I was expecting a CdA of 0.18 to 0.19. The bad news is the data is probably worthless. I didn't mention that the first and last intervals were the some setup. So it looks to me like there was drift throughout the session.

You might be interested -- what I was testing was a Hed Jet+9 front with Conti SS 20C vs. a Enve 8.9 with Corsa Speed tire. It was really a sanity check to make sure neither setup was a disaster.

---

My latest book: "Out of the Melting Pot, Into the Fire" is on sale on Amazon and at other online and local booksellers

Updated the parameters for (a) your weight, (b) the Powertap, and (c) Corsa Speed tires, which should give you CRR around 0.0035 on normalish imperfect pavement, as long as the pavement isn't terrible. (I've tested this in many ways.) This gives me Lap 2=0.203, Lap 4=0.194, and Lap 6=0.193. These are much more reasonable/believable numbers, but you still have a big difference between 2 and 6, which means something went wrong since they are supposed to be the same.

I test a lot, and in my experience I get within about 0.001-0.002 with tests of length ~7mins. Your tests are quite a bit longer than that. Your first test is 20% higher power so maybe that's part of the reason, though they are all very low power so not sure that would matter.

BTW if you want the software PM me and I'll send you a link. I never prettied it up for public consumption, but it's extremely easy to use and only requires that you have java installed, which most people do.

I can’t speak to the ENVE setup but I can tell you the Jet 9+ with the 20mm SS is fast. REALLY FAST. PM Grill, he has data.

If you are testing different wheels AND tires, how do you justify assuming a constant Crr??

  
Awesome. Thanks. I'm still guessing Crr of closer to .005. I used to do regression field tests down on Bayshore road (Sundays 4-5 am of course). That was .0035 Crr. Now I'm riding on Oregon chipseal, which must be worse. But even assuming it is .0035, if I add shoe covers and put my best skinsuit on I should be able to get into .18 territory.




I have tested both tires on the rollers. The Crr difference was .0001 So we're talking less than half a watt difference for the front tire at the speeds I was going (22mph).




It has done real well for me so far. I used the Enve wheel in the last race, and the CdA wasn't quite as good as expected. So I thought I'd test to see if there were any dramatic differences. I have to get a better test environment. The problem is there are too many corners so I have to softpedal a lot to avoid braking.

---

My latest book: "Out of the Melting Pot, Into the Fire" is on sale on Amazon and at other online and local booksellers

As long as you are comfortable with your assumptions...

Jens what do you think the optimal pressure is for the jet+ 20c supersonic combo onsmooth roads?

I've bent a little to the Josh + Tom A. school of rolling resistance and have been running around 95 (I used to 115+). But that's really based on my trust in them, rather than any data I've gathered myself.

---

My latest book: "Out of the Melting Pot, Into the Fire" is on sale on Amazon and at other online and local booksellers

Would be interesting to test that at 80, 90, 100 & 110 psi outside and see if you can see any difference in the results.

Based on Tom A.'s roller data, a change in pressure across that range would be expected to alter Crr by up to ~0.0005. A difference of that magnitude is measurable using field testing, at least using the regression approach:

http://www.trainingandracingwithapowermeter.com/...st-results-part.html

I haven't tested the GPSS in this way, but I did test once Vittoria CS 23's at a wide range of pressures on a fairly rough chip seal TT course and here's what I found: (I weigh 180 so figure 200lbs total weight)

Pressure: CRR
110/120 0.0043
100/110 0.0037
90/100 0.0037
80/90 0.0038

I’m really curious about putting the 20c tire on a wide rim like a jet +. I have measured it and it actually measures 25mm wide, perfectly inline with the rim. I’m convinced it’s the fastest tire for the jet + but just not sure if I should treat it the same as a 23mm tire that measures 27mm on a Jet + (conti TT), or should I put in a little more pressure. But as Josh showed in his graph, and your data seems to agree with, you’re better off under inflating than over inflating.

I'm getting a higher Crr: .00505
For interval 2 I'm getting a mean estimated CdA of .1906
For interval 4 I'm getting .1827
For interval 6 I'm getting .1845

The last couple of laps of interval 2 and the first few laps of interval 4 are pretty noisy.

Wind is thine enemy.

(A ~0.006 m^2 difference in CdA for the same set-up? Toss the data.)

I'm not so sure. The natural question is why parts of the data were noisy but other parts were more stable. That could tell you whether the entire test is trash and unsalvageable, whether some portion of it can be used but other parts need to be dropped, or whether some lesson can be learned. For example, it looks like Jens was getting slightly better at handling the corners in the latter part of the test than in the first part.

Several years ago when I was down at Carson observing a test, the CdA suddenly went wacky. Then one of the guys noticed that the outside door to the loading dock had been opened. That threw off the measurements. We took a break and waited until the door closed, and a few minutes more for things to settle down and then resumed the test. The rider's CdA went back to what it was just before it went wacky. We didn't have to toss out the entire test, we identified something that caused a systematic (as opposed to random) error in the measurements, and just tossed that one piece. That was an easy explanation, and not all explanations are going to be so easy and obvious.

That's precisely the sort of thinking that leads many a scientist down the proverbial primrose path.

Jens has already identified one known issue with the experimental conditions, i.e., too much wind. Furthermore, this wasn't a billion dollar space mission where you would want to try to salvage anything you could. The wise approach is therefore just to scrap the entire test.

I hate to say it but I'm kind of with Andy on this one. I didn't snoop the data too closely, but there are some data issues in there too, dropped data and repeated values. I cleaned up a few of them before calculating, but I'm sure there are more. Also he's using GPS for distance/speed, and the two fields don't match, so pick which one to use -- I used the distance one. In my experience that's not as precise as a speed sensor. Also, if he's using Vittoria CS 23's there's almost no way CRR=0.00505. I've tested them thoroughly inside and out and 0.0035 is a good guess for an imperfect but otherwise decent road. That's a huge difference. The fact that you get such a big difference off of such long laps means something really different was going on across those laps. It's not statistical noise.

That said, if I've issued you a data challenge, I fully expect you to rise to it :-)

GPS for speed? That's a deal-killer right there.