I'd be interested in seeing some tests like that. I could see a bad chainline being measureable, but would be shocked to see losses to to flexibility. A simple way to test would be temp of the frame. If power is being lost then its going somewhere and it should be heat.

An good test would be a full supension MTB set squisy then locked rigid. Ride them on the road like Tom did and measure the loss. That should be an absolute worse case scenario for frame stiffness losses.

Styrrell

I know I'm going to hate myself for this but I'm going to respond very briefly.

1. Indirect doesn't mean inexact. Neptune was discovered by perturbations in the orbit of the planet named for you, or more precisely, Uranus (thanks folks, I'm here all week. Tell your friends). Its position was subsequently verified through direct observation. All field-based methods are indirect methods since they use measurements on other variables combined with known relationships between the variables to estimate CdA and Crr. Regression-based methods are often classified as indirect methods. There's nothing unusual about indirect methods. You just haven't spent any time thinking about the distinction before.

2. As you noted, the "not bad" came from page 3. What was page 3 about? Oh yeah, a particular data set proposed as a challenge. That challenge data set contained only speed and power. There was no info on weight, weather, or route, except that it wasn't flat. The "not bad" was in reference to this particular data set, which was quite extreme in its parsimony.

3. "Wind": Wind is always a problem with field tests. A good thing about this approach is that --> you can tell when the wind was strong enough to mess up the results <--.

Lots of people have asked about the accuracy of this method, including some of the people participating in this thread. It's just that they asked earlier, in other threads or in other places. The reason why you're the only one asking now is because you're late to the party.

iow, there's a door charge to the Chung Comedy Club and, Frank, you've not paid it :-)

The next time you're in town, I'm buying you a beer! (or, the beverage of your choice, if you don't like beer).



Yeah...that's a real shitty deal. Have you found out if it's going to be surgery or not yet?




Hmmm...we're about the same size, right? How about letting me borrow them for awhile? I could do another "Something borrowed..." test :-)

---

http://bikeblather.blogspot.com/

No, I'm quite familiar with the various Cervelo models...I went back & checked my records. I've got a private email from a reputable friend who was in a tunnel with a P2K and P3C, he quoted .5 - .8s/km @ 30mph. Maybe he was being conservative since he knew I was deciding whether to swap my P2K for a TTX?

Trek Whitepaper had P3C as a little more than .5s/km faster than a Scott Plasma, closer to .8s/km faster than a Madone (all @ 0' yaw). I assumed the P2K was no worse than a Madone...

Tom's work appears to double the difference, at least, which is why I wonder if something has been missed.

ROTFLMAO!!!

Say what? You compare anecdotal reports of performance improvements due to use of your product with measurements made using a powermeter, then state that they are the same technique??

Is that with or without a rider, and at what yaw angle(s)?

You tell me!

The Trek whitepaper data was, as you know, with a rider, and I said I was only quoting the 0' numbers . Looking back at it, it is roughly doubled at even just 10' yaw (if I assume the P2K is as bad at yaw as a Plasma).

Did I miss Tom's estimation of the apparent wind during his testing? If so, I apologize for bringing up the point, as it appears even minor wind could easily result in these sorts of differences once we make a couple assumptions.

I'll take you up on the beer soon... talk to j and maybe we'll add ya to the interbike list if you can swing a hallpass...

I'll loan you the bikes if you pay shipping and ride them in my position (you can move saddle/posts) and have the back in 4 weeks...

:P

-g

---

greg
www.wattagetraining.com

Sorry, I missed where you said that it was at 0 deg of yaw.

Anyway, differences in how CdA changes as a function of yaw could certainly explain the difference. Indeed, I recall John Cobb once commenting that the Hooker (which used narrow NACA-profile tubes, similar to the P2k) didn't perform as well in crosswinds as it did in a pure headwind. I still "sailed" pretty well upon it, but that doesn't mean that a better design (e.g., P3C) wouldn't be even faster under such conditions.

Indirect doesn't have to mean inexact, execpt the pertebations of Neptune told astronomers there should be something in this area. It did not tell them exactly where to look. Now that it is found, people know exactly where to point there telescopes each time. It is hard to imagine an indirect measurement being more accurate than a direct one (except when direct measurement is not possible). Hence the comment on accuracy.

Second, the measurement starts by using a device with a known accuracy and then introduces additional steps on top of that. The method, it seems to me, cannot be any more accurate than the power meter one is using and, because of the additional steps involved should be substantially less accurate. While others may have questioned the accuracy of the method in other threads I haven't seen anyone question the accuracy of this result, people seem to accept it as being true.

Even when methodology is potentially very accurate, errors can be introduced by those performing the methodology, despite their best efforts. I remember analytical chemistry very well. I could never get a very accurate result for whatever reason, despite what I considered my best efforts, while others could reliably do so. In that instance, the methodology is proven yet, subtle errors can be introduced throughout.

As I said, the methodology seems very inventive and I am impressed. But, until it is compared to known standards and its reliability when done by others less skilled is known we cannot necessarily trust this particular result as being anywhere close to the truth. That is what I am trying to say. If he saves 2 minutes off his next 40 k TT using that bike (and everyone else rides pretty much what they have done before, on average except, of course, those who have changed to this bike who shuld all be 2 minutes faster also) then I would take that as an independent confirmation the result is pretty accurate, at least in relative terms. If he doesn't, where are the inaccuracies in the result going to be explained?

As near as I can tell from watching the big people ride TT's, the bikes they are on make little difference in performance. This result says otherwise so is a little hard to believe. It is even better, as I understand, than the manufacturers claims. It just makes no sense to me that this much change can supposedly be due to the bike alone. It might be true, but I would like a little more rigor in the process before it is taken as gospel.

---

--------------
Frank,
An original Ironman and the Inventor of PowerCranks

Not only has this been done, the results of such studies have been published in the peer-reviewed scientific literature. The bottom line is as alluded to by others: at least under optimal conditions/in the right hands, field testing using a powermeter can be just as accurate as testing in a wind tunnel, and is nearly as precise. It does, however, have some disadvantages: 1) it's generally more time-consuming/less convenient, and 2) more importantly, you can't test at controlled yaw angles. The latter is critical, because cyclists are NOT bluff bodies.*

*A while back I was surfing the web for information on building a small wind tunnel, and came across an aerodynamic primer for model rocket hobbyists. It gave the Cd value of a long, multi-stage (fuselage of two diameters) rocket as being 0.7...IOW, the same as a cyclist in the aero position on an aerodynamic bicycle! Remember this the next time somebody starting spouting off about how non-aerodynamic a "floppy" human pedaling a bicycle is...

Maybe they're on the wrong bikes, or maybe you're just not a very good observer...

Anyway, here's what I know: when my wife was riding a P2T, she had to produce 410 W to ride 3 km in 3:51.X. Five years (and one DVT) later, she goes nearly as fast on a P3C while producing only 330 W. Conversely, when I was riding a Hooker my 40 km TT times were routinely in the 53:00-53:30 range, but when I retired that machine and started using a Javelin Arcole or P2T (my wife's bike) instead, the fastest I went was 54:12, despite producing just as much power as previously.

John, I hear what you're sayin'...for example, assuming that the P3C is slightly (let's say ~.05lbs @ 30mph) lower drag than a Hooker Elite at zero yaw (as Andy has reported) and that the P2K isn't appreciably different from the original P2 at that angle, the data Andy presented in the post below indicates the difference to be around 0.2 lbs of drag, or ~.010 m^2 of CdA.

http://forum.slowtwitch.com/...oker%20drag;#1109749

So yeah...my "zero yaw" number looks to be at least double than what was reported there.

All I can say is the data "is what it is"... :-)

In regards to the apparent wind conditions, the testing was done in the early morning hours with the intent of minimizing crosswind affects. However, as I discussed earlier in the thread, even though in the past I haven't found much problem with dealing with slight winds on my original course, on this course there seems to be more sensitivity to winds, however so slight, due to the lower average and max speeds encountered. I have to emphasize however, that the amount of wind encountered during this test was extremely light...in fact, I'm not even sure if it would have registered on a wind gauge.

Here's some more data that may help to serve as a "sanity check". Below is a plot from the first lap of a 40K TT I did a few weeks back on the P2K. As I mentioned earlier, the outbound leg of this lap provided an ideal opportunity to nail down a CdA and Crr since it was done in near zero wind conditions and the outbound speeds are high due to it being a generally downhill run. Knowing the actual elevation difference of the course was 50meters, and the fact that the wind didn't start picking up until ~1/4 to 1/3 of the way into the inbound leg, by choosing a Crr and CdA combo that matched the elevation difference and made the elevation profile symmetric about the turnaround point (up until the part where the wind started) I came up with a Crr = .0038 and a CdA = .228 m^2, which basically matches the "zero yaw" measurement from the field test.




Now, at the same race was another rider who is nearly the same size as myself and has a similar position on his bike. The main differences are that he rides a P3C and his wheels are slightly different (404 front vs. my TriSpoke, and tubulars vs. my clinchers) although that shouldn't have a huge affect on zero yaw performance. This other rider was, in fact, my 30s man so it's safe to say that we were both racing under VERY similar conditions. Using the same technique above, 2 things stand out. First, as expected, his Crr is higher due to the use of tubulars, and his CdA turns out to be .207 m^2.




So...knowing that, I don't find the difference I measured to be unreasonable, especially when considering Andy's data in the mix.

Oh yeah...the second lap of that TT was done with a steady wind. Using the Crr determined from the first lap, my calculated CdA dropped to .220 m^2 (just like in my testing when I thought there might be some wind present) and IIRC (I don't have THAT data in front of me right now) the other rider's CdA dropped to ~.190 - .195 m^2. I know that's not super compelling data since it's 2 different riders, but it WAS under near identical conditions for both of us. The fact that the general magnitudes of the changes were similar lends some confidence (in my mind at least) to the results in the P2K vs. P3C testing.

But, I'll admit...there's some question about how truly zero yaw my "zero yaw" measurement were. Then again, the data also points out how very little yaw it takes for the CdA to drop significantly on frames like these and further emphasizes how testing in a wind tunnel at only zero yaw can be misleading as to the differences between setups.

I'm thinking I might have to take the P3C over to my "preferred" course and do some runs prior to returning it to it's rightful owner. That's not going to happen this weekend though...since I'll be racing it :-)

---

http://bikeblather.blogspot.com/

Tom,

1. Small density difference between those files 1.178 vs. 1.19 (66 degF vs. 61F ??)
2. Which point on the course is at 613ft? I know very minor but as there's 50m delta E ... there's some variation in rho as well.
3. How does the pavement on this TT course compare to that of your test culvert?
4. For your original post (and I know this is obvious in hindsight) I would have swapped back to the P2k and repeated the baseline.

Thanks, good stuff to mull over!

Aah...yes. Thanks for pointing that out. I had forgotten that the rider who sent me that data had looked at it first using Alex Simmon's spreadsheet posted over on wattage. He had used weather data from a different weather station than I did, and the numbers in that file reflect that. I was trying to see how closely our spreadsheet calculations matched using HIS data.

Check this out...I just "copied and pasted" the weather data from my file into his. Guess what the CdA for him works out to be now? .205 m^2...can you believe it? :-)



Actually, that's the elevation of the weather station that's across the river valley from the course. You're right though, to be more exact I really should use the average course elevation, huh?



Pretty similar. They're both asphalt pavement. I'd say the TT course might be slightly rougher.




Well...I only had so much time...and as I related before, the second P3C run was beginning to suffer from "automotive interference" anyway. It only would've gotten worse as it got later....

---

http://bikeblather.blogspot.com/

I wold love to see a link to the study, if available. I do love the caveat, "under optimal conditions/in the right hands".

And, exactly what do you mean when you say it can be "just as accurate" and "nearly as precise". What is the difference?

---

--------------
Frank,
An original Ironman and the Inventor of PowerCranks

Hi guys. I only just came across this thread this morning. Very cool reading, well most of it, but Robert gets the joke of the month award.

Just on the air density calculation - I did reload the files for both the Chung and regression methods onto the Wattage forum files section on 5 May - the latest version has the altitude correction factor included.

---

_________________________________________________________________________________
Training Plans -- Power Meter Hire -- SRM Sales Australia -- cyclecoach.com -- My Blog -- Sydney Turbo Studio

Um, Frank? You know just a bit above where you wrote: "I remember analytical chemistry very well"? I'm thinkin' you don't remember it quite as well as you think you remember it.

Can you persuade him to ride the P2...

How do you know that your position on the P3C is exactly the same as on the P2K?

Um...

A ruler, a level and a little bit of time???


G

---

greg
www.wattagetraining.com

I hardly remember anything about analytical chemistry other than I wasn't very good at it even though I took as much care to be "accurate" as I knew how. Is there a problem with asking him how he is using these words? One statement implies an equivalence and the other doesn't.

---

--------------
Frank,
An original Ironman and the Inventor of PowerCranks