Assuming the change you are testing doesn't interact with your training gear.

Using training wheels is probably safe when testing hand positions. But I would wear the skin suit and aero helmet.

Skin suit is a good idea just to reduce noise in testing anyway.

---

---

Kat Hunter reports on the San Dimas Stage Race from inside the GC winning team
Aeroweenie.com -Compendium of Aero Data and Knowledge
Freelance sports & outdoors writer Kathryn Hunter

Cool thanks. Any thoughts on length of course?

A kilometer is fine.

If this is the very first time you're trying this, my guess is that your results will be kind of noisy. Do a couple of extra laps -- in general, it might take a lap or so for you to find a "line" you can hold consistently, especially if you're not sure how fast you can do the turnaround. This lets you toss a lap if it's too noisy. Make sure your halfpipe has clear line of sight at either end -- you don't want to take a chance of getting hit by a clueless car driver. Vary your speed and power during your runs. If you're using a crank- or pedal-based power meter, you might consider not shifting and doing your test in one gear.

Do you find it's easiest to keep all of your laps in one file. Or do you do a new file each time you take the turn?

I think a good length is about 5 minutes. If your lap is shorter than that, just do multiple laps until you get in the 5 minute range.

I'm going to disagree with Robert on one thing: in my opinion the line simply does not matter. It need not be the same each time and it doesn't matter what it looks like. Just don't use your brakes and try to find a spot with zero cars. What does matter: make sure you hit the lap button at exactly the same point each lap. A lot of people are tempted to do it at the turnarounds but I would recommend against this because a small error at the turnaround can lead to a large change in elevation. Do it on a flat part of the course, such as the bottom of the half pipe. The really important thing is that the elevation be exactly the same at start and finish, not even off by a foot. Finally, when you compute CdA (say, using aerolab), ignore the elevation profile and simply find the CdA that sets the beginning and end elevations to be exactly the same. Looking at the intermediate points can lead to bias in the CdA estimate. (Ignoring the intermediate points is also the reason the line doesn't matter.)

It is the intermediate data that allows you to tease apart CRR and CdA, though.

If you need to do that.

---

---

Kat Hunter reports on the San Dimas Stage Race from inside the GC winning team
Aeroweenie.com -Compendium of Aero Data and Knowledge
Freelance sports & outdoors writer Kathryn Hunter

Well it turns out that that doesn't work, at least not unless you're in a pretty close to perfect system -- constant pavement quality, no wind at all, etc. If you want to tease apart CRR and CdA in the presence of real world variations, you need to do multiple laps at different speeds and then find the CRR/CdA pair that works for all laps combined.

I definitely keep all my laps in one interval. I don't like fiddling with even one button during runs, and it will be abundantly clear where each turnaround is. You'll want to follow the same line each time -- sometimes it helps to pre-ride the test venue with a piece of chalk and mark the turns with a couple of "X"'s so you can make consistent entries and exits.

Here are a handful of recommendations: http://anonymous.coward.free.fr/.../recommendations.pdf

And check out the "platypus" thread.

Ah, yes. That's a good idea. I try to keep the line the same because I do an additional kind of calculation that's a lot simpler if each of the laps is the same length (that is, if a lap is, say, 1726 meters long, it can be handy for this additional analysis if I can rely on every 1726 meter long interval to be a lap) so that I know the elevation is the same. However, those are my own calculations, Aerolab doesn't make use of that, so if you're not doing the thing that I'm doing it may make sense to do what you're doing.

Thanks everyone for all the advice, we'll see how this goes!

I have a quick question on course selection.

Is it acceptable to ride a straight line course for 1k without a turnaround or is a loop an absolute requirement?

The road I am thinking about is nice and straight, gradual elevation change/false flat, with overhanging trees providing shelter from the wind.

That's acceptable. If there are no "elevation features" you'll want to make sure you hit the interval marker at the start and end. You will probably want to vary your power and speed -- I do a protocol where I steadily accelerate from walking speed up to maybe 25 mph then coast back down (while soft pedaling) to walking speed, then speed up again. Lather, rinse, repeat for your 1 km straight line course. Even better if you can do this in both directions.

Robert's being fairly easy going when he says even better if you can do it both ways. My anal answer is no it's not acceptable to do one direction only. The reason is that any wind or even one meter of elevation change throws things off. Record a lap one direction then turn around and record a lap the other direction too. Put the two together for your test lap. (Also you need to either have the entering velocity for the return be the same as the exiting velocity for the out lap, or you need to properly account for the difference.)

I'm teaching two classes this semester: an undergrad class and its "companion" grad-level class. In the grad class I'm pretty detailed and anal but in the undergrad class I try not to scare them off so I try to be easy going as long as they understand that what they're learning isn't the whole picture. But I do sometimes wonder if I've been doing them a disservice.

A ton of good information on the Platypus thread. Well worth the hour or two it will take to carefully read. Things like speedometer accuracy wouldn't occur to most people.

Speaking of this issue, does anyone know if the speedometer in a PT is very accurate?

It ought to be as long as you properly measure roll-out distance (loaded).

Hello,

I am unsuccessfully trying to create and upload a CSV file for the Chung Method CdA Estimation. I've exported the a TCX file from Garmin Connect, and used Excel to extract the speed and watts for a CSV file. However, the model can't read the Data File. I'm sure it is user error. Do you have any suggestions?

Scott

Use the Slowtwich search engine and type in "Platypus Thread"

---

AndyF
bike geek

Sounds like you're trying to use the Cyclingpowerlap.com page at http://www.cyclingpowerlab.com/cdaestimation.aspx?

There's an example data file (click on "example") and make sure your file has the right format.

Yep, that's it. I've compared the example file with my file and don't see any differences. Just two columns of numbers in a csv file.

I pasted my numbers into a downloaded example file, and then received the same parsing error message on the example file that I receive when trying to use my csv file. There must be something about the format of my numbers that is not right?

Sorry to waste your time with this. Just bugs me that I can't get it to work!

Scott

Hmmm. You might have better luck with Golden Cheetah. I think it can read your .tcx file.

It may not be you. Non-printing characters can break a csv import, so it could be your text editor or your spreadsheet or your OS. Or it could be the site's import algorithm.

Golden Cheetah will sort it for you

---

Mark E