Talk to me about aero testing protocols

Triathlon Forum
21

Speaking of your method…what do the results look like for the analysis shown on slide 6 here:

http://anonymous.coward.free.fr/...cda/indirect-cda.pdf

when you correct for changes in kinetic energy as you did with your method (so as to truly compare apples-to-apples, and not apples-to-oranges)?

The starting and ending speeds for that 15 minute segment were about 0.3 m/s apart.

22

Speaking of your method…what do the results look like for the analysis shown on slide 6 here:

http://anonymous.coward.free.fr/...cda/indirect-cda.pdf

when you correct for changes in kinetic energy as you did with your method (so as to truly compare apples-to-apples, and not apples-to-oranges)?

The starting and ending speeds for that 15 minute segment were about 0.3 m/s apart.
Perhaps, but what you plotted (and performed the regression upon) were the individual data points. So again, what happens when you correct the power for the changes in kinetic energy from one point to the next?

23

Speaking of your method…what do the results look like for the analysis shown on slide 6 here:

http://anonymous.coward.free.fr/...cda/indirect-cda.pdf

when you correct for changes in kinetic energy as you did with your method (so as to truly compare apples-to-apples, and not apples-to-oranges)?

The starting and ending speeds for that 15 minute segment were about 0.3 m/s apart.
Perhaps, but what you plotted (and performed the regression upon) were the individual data points. So again, what happens when you correct the power for the changes in kinetic energy from one point to the next?

Ah, that’s the point, isn’t it? The normal regression method either presumes constant speed and elevation, or else looks at starting and ending velocities and starting and ending elevations. I don’t make those assumptions. Slide 6 is an illustration of what happens when the assumption of fixed speed and elevation is violated. That’s an apples-to-apples comparison.

24

Andrew, if for example this morning with my 2x20 minute intervals out and back at ~95% of FTP I averaged 276 watts and 25.45mph on chip seal road what would be the mathematic formula to estimate CdA?

Kevin

25

Andrew, if for example this morning with my 2x20 minute intervals out and back at ~95% of FTP I averaged 276 watts and 25.45mph on chip seal road what would be the mathematic formula to estimate CdA?

Kevin

I’m not AC, but (off the top of my head) the equation is (basically):

Power = (1/2 x rho x CdA x Vwind x Vground^2) + (M x g x Crr x Vground), assuming a flat course, of course :wink:

26

Thanks Tom. Just to check rho is air density, M is distance covered and g is my weight plus bike?

Kevin

27

Thanks Tom. Just to check rho is air density, M is distance covered and g is my weight plus bike?

Kevin

rho is air density in kg/m^3 (typically ~1.2), M is mass (you plus bike) in kg, and g is the acceleration of gravity (9.81 m/s^2).

Oops…I just realized I made a mistake on the wind portion, here’s what the equation should look like:

Power = (1/2 x rho x CdA x (Vwind + Vground)^2 x Vground) + (M x g x Crr x Vground)

Vwind is wind speed relative to your direction of travel (i.e. a tailwind is negative and a headwind is positive).

28

Velocity is in meters per second I presume? Recalling some physics from undergrad, but that was 22+ years ago!

Kevin

29

Velocity is in meters per second I presume? Recalling some physics from undergrad, but that was 22+ years ago!

Kevin

Yes…now, you’ll notice that there are 2 “unknowns” in that equation (CdA and Crr), so for the estimation you’re trying to make, you’re going to need to assume a Crr. You’re on your own on that one :wink:

30

Thanks again Tom! I was using Maxxis Refuse with butyl tubes, from experience they are not the fastest rolling tire around. I think I will try analytic cycling and see if I can come up with an estimation.

Kevin

31

Thanks again Tom! I was using Maxxis Refuse with butyl tubes, from experience they are not the fastest rolling tire around. I think I will try analytic cycling and see if I can come up with an estimation.

Kevin

Hmmmmm^TM…I think I’d guesstimate something in the .006 to .007 range for those…if not worse!

32

Yes but when you are trying to get in the training on a shoulder scattered with broken glass from drivers returning to the reservation they are great!

thanks

Kevin

p.s. I you ever want to come out here for some stage race action at VOS or TBC consider yourself with an offer of a place to stay as long as you arent allergic to cats!

33

I came back with an average of 0.26 for the out and back, I dont know if that sounds reasonable for a 5’6" 140lb rider riding Leipheimer style (with slightly flatter forearms)

Kevin
.

34

Oops…I just realized I made a mistake on the wind portion, here’s what the equation should look like:

Power = (1/2 x rho x CdA x (Vwind + Vground)^2 x Vground) + (M x g x Crr x Vground)

Jim would be proud of you (sorry, inside joke).

35

The normal regression method either presumes constant speed and elevation, or else looks at starting and ending velocities and starting and ending elevations.

Right. So, I’m asking you what happens when you apply the latter approach to the data shown in that figure.

Slide 6 is an illustration of what happens when the assumption of fixed speed and elevation is violated.

In terms of the virtual elevation, yes. However, it doesn’t really address the impact of such violations when the data are analyzed using the regression approach.

36

I came back with an average of 0.26 for the out and back, I dont know if that sounds reasonable for a 5’6" 140lb rider riding Leipheimer style (with slightly flatter forearms)

Kevin

Depends on what you were wearing and your bike/wheels…

37

Tight fitting jersey, no aero helmet, Bike is Cannondale Slice carbon, HED V8 aerobars, 28 spoke training wheels with velocity aerohead rims and DT Revolution spokes.

Kevin

38

Right. So, I’m asking you what happens when you apply the latter approach to the data shown in that figure.

Knock yourself out, big guy. The data for that figure were from about km 5 through about km 11.2 of field-cda-challenge.csv. Show your work.

39

Tight fitting jersey, no aero helmet, Bike is Cannondale Slice carbon, HED V8 aerobars, 28 spoke training wheels with velocity aerohead rims and DT Revolution spokes.

Kevin

Yeah…fairly reasonable…

40

A couple of things to note:

  1. Even if you have an absolutely calm day and completely flat course, pulling out drag area can be tough. Why? You’ve got a few unknowns in the power equation. Assuming no acceleration, P=.5rhoCdAVt^2Vg+massgCrr*Vg. Unless you do diligent coast down tests to get a better fix on Crr, the best you can do is come up with (Cda, Crr) pairs which make a good fit. I’ve found the best use of field testing to be to isolate changes in drag due to equipment or position changes. Fix Crr (by using the same tires and pressure for all tests) and get the relative change in CdA across tests.
  2. Living in a flat and windy area, I’ve found I can get good results by using a loop open to the elements. Looking at the velocity around the loop, the maxima and minima will correspond to tail and head wind sections, 0 degree yaw. From those 2 data points I can extract a reasonable estimate for the wind speed. Using the same methodology as 1) above, I come up with my family of (CdA, Crr) pairs and look for relative changes in CdA. By doing multiple laps and various speeds/power I can get a degree of confidence about the predicted wind speed. This has provided good results even on days with reported winds around 15 mph.
  3. You can get OK results with coast down tests. If you know how to do Runge-Kutta integration you can perform a kinematic simulation of the coast down. I actually did some of this back in November when FD was talking about his aero tuck position for PCs. You don’t even need a powermeter for this - just a bike computer or Garmin that you can download time and speed data with.