Have losses due to frame differences other than aerodynamics been factored in? e.g. frame stiffness.

The tests "just" show that the P3C is a better bike than the P2K, but does not attribute the difference directly to aerodynamics...

---

___________________________
Chewie
Slowtwitch Aeroweenie since '06

That's one of the beauties of measuring power at the hub! That said, and as mentioned in that other thread, while I see the potential for frame stiffness to affect efficiency, I don't think it's ever been demonstrated.

I think a better question may be 'does frame material/design affect Crr'?

edit: so Tom, do you assume a constant Crr, or does the Chung method determine it for you?

I disagree. I don't see how a stiffer frame would account for the differences. Math is math after all.

I think (he can correct me if I'm wrong) that chewgl was suggesting that one frame may have less drivetrain loss than another. If Tom were measuring power at the crank, this could be an issue in determining the aerodynamic differences between frames. I think these differences though fall into the 'insignificant' category though, and it's moot since he's measuring at the hub.

Except...my "plain Jane" road bike is a Soloist...which by my measure is only slightly slower than the P2K.

BTW, I've commented to some friends that the P2K would probably make a pretty decent crit bike. It's got a steep seat tube and a relatively high BB. Perhaps that's what duty this particular P2K may end up performing in the future ;-)

I've seen Arnie Baker riding a P3C with drop bars in road races before...with a disk and a Troxel helmet too. Do you think he likes to ride off the front? :-)

---

http://bikeblather.blogspot.com/

The power measurement was "to the road", or more specifically "to the hub" since it was using the same PT SL 404 wheel. Everything between the hub and the road was identical.

IMHO, the evidence is pretty compelling that the differences measured were aerodynamically based.

---

http://bikeblather.blogspot.com/

No doubt that most of it was probably aerodynamic, but I'm wondering if any can be attributed to frame stiffness (yeah... something along the lines of drivetrain losses...).

---

___________________________
Chewie
Slowtwitch Aeroweenie since '06

The method requires you to assume a Crr and solve for CdA, or vice versa.

That said, if you KNOW the actual elevation differences AND you don't have any appreciable wind effects, it's possible to narrow in pretty quickly on the correct combination of Crr and CdA when using an "out and back" type course. Although there's virtually an infinite number of combinations that will "level" the calculated elevation plot, only one combination of Crr and CdA will "level" the plot, have the proper elevation change, AND be symmetric about the turnaround point. That's why I said above I was lucky to be able to nail down a Crr for my tire combo at a previous TT. The outbound leg and a portion of the inbound leg of the TT was under near zero wind conditions and I was travelling between 25 and 35mph, along with the leg having a known elevation change of ~50 meters.

However, if all you're interested in is comparing relative aerodynamic drag effects in a "head to head" type comparison like the one above, any reasonable Crr assumption will still give you the relative aero differences. You just may not be able to compare it to absolutely to testing on different days, or with different tires/wheels/pressures. Make sense?

---

http://bikeblather.blogspot.com/

My bad... power measured at the hub ==> losses measured probably come solely from aerodynamics.

I'd be curious to see how this would compare to measurements at the BB or crank: by comparing both, you could get some estimate of drivetrain losses due to changes in frame too.

That said: nice job Tom A!

---

___________________________
Chewie
Slowtwitch Aeroweenie since '06

I rode a P3SL in a road race last year. I pulled the 38 cm bars off my wife's road bike and used a 140 mm stem. When I was out on the hoods I was basically in my normal TT position as far as drop and extension, but of course my hands were much farther apart then normal, and I had only minimal forearm support from the bat-top.

It was a fast setup (I wore a skinsuit and silly pointy hat, as well). However, I have to say that a P3SL with narrow bars and long stem really does not handle particularly well and I don't think I would ride that setup in a crit.

Since Tom's riding on a recycled P2K, it's fun for me to note that this P3SL has also been recycled, and is now being ridden by someone who has reported (in another thread) large changes in CdA as compared to his former ride. It's a small world! :-)

Tom A: I've always wondered: how does the accuracy of the powertap (+/- 2%) figure into the calculation of errors for these field tests? Compared to an SRM Science (with +/- 0.5%)...

---

___________________________
Chewie
Slowtwitch Aeroweenie since '06

Ain't it cool?

That's not quite so easy. Part of the issue is that the PT and SRM measure in different ways (e.g., the PT is time-based), and we think that drivetrain losses may vary with chainspeed, chain tension, and cog size. In order to determine the effect of a frame, you'd need to be careful to partial out the other effects. Not impossible, but it is kinda tedious.

Yep, the evidence is fairly compelling. However, couldn't changes in frame stiffness cause a change in effective Crr? It's probably a second or third-order effect, but frame stiffness affects tire path, which affects effective Crr. I'd bet that for the normal range of frame stiffnesses the effect is insignificant, especially for the typical power laid down in field testing.

If you do a standard error propogation you'll see that speed is the factor that needs to be measured most precisely.

If the PT has had a static torque check, you can pretty much tell. The 1.5% quoted by Cycleops refers to accuracy, not precision. Because the PT is time-based while the SRM collects data by completed crank revolutions (and thus, is slightly less dependent on cadence), there's more record-by-record variability in the PT. However, when averaged over a few seconds, they appear to be quite close.

Yeah, that reminds me: you have to measure your rollout pretty carefully, and it helps to record in km/h. Speed gets recorded to the nearest tenth of a km or mile. I've toyed with figuring out how much of an error is introduced by that, but only in a half-assed kind of way.

Mr. Chung is probably best to answer this one, but my understanding is that the "integration and constrain" nature of the method tends to dramatically minimize the associated sampling errors (such as those reported accuracy levels you mention above). You're then basically dealing with systematic errors, which can be minimized and/or accounted for by other methods and/or experimental technique. See page 53 of Robert's .pdf presentation that was linked to in my original post.

---

http://bikeblather.blogspot.com/

With one magnet, anyway. However, if you put more than one magnet on the wheel (or axle, if using a PT), you run the risk of the reed switch miscounting wheel revolutions. You can almost always correct such errors after-the-fact, but it's a PITA.

So does signal averaging.

Thanks Guo-Liang. Coming from a self-professed "ST Aeroweenie", that means a lot ;-)

---

http://bikeblather.blogspot.com/

Very. In fact, I'm quite agog at how cool it is :-)

---

http://bikeblather.blogspot.com/

True 'dat.

---

http://bikeblather.blogspot.com/

Tom, haven't you said elsewhere your apparent precision -- well I prefer repeatability myself -- is about 0.001 m2 for your culvert? Ain't that down around 1W average power? Or maybe less at these test speeds?

IOW, I think that's getting down there!

Well...that's on my "preferred" half-pipe course. I seem to get pretty good repeatability there when testing identical configurations in the same session. Typically within +/- .001 to .002 m^2

Then again, maybe I'm just lucky ;-)

---

http://bikeblather.blogspot.com/