Monday, May 3, 2010

Cda and Crr testing highracer vs highracer with a powermeter & Golden Cheetah AeroLab

I am usually too busy or ADD affected to actually go out and test, so I just dump the data I have into tools and see what I get. The crit I do on Thursday nights is not completely flat, but it is a loop. It was also very windy this Thursday. I spend the whole warm up ant the last 19 laps of the event riding pretty much alone, and then a 4 lap cool down.

This was on Carbent Raven (27 degree seat angle) with American Classic Hurricane wheels, used Continental GP 4000S tires (100#f/120#r), one with a butyl tube and one with Panaracer Greenlite tube. One bottle on a seat back cage, no seat bag. I was wearing Defeet wool Kneekers, a JBV Coaching LS jersey, and a Giro Atmos helmet.

This is getting me interested in going down to the same industrial park on a calm weekend and actually riding to collect data, not to train.

Anyway in the plot below you can see how well the tool (Roberts Chung's Aerolab in Golden Cheetah) works. The first level section is warming up alone, at a variety of speeds, but no coasting. The last level section is a 19 lap solo breakaway followed by 4 laps of cool down.

See the cool down laps slope up? My Cda goes from .268 to .286 there. Why? I was wearing a slightly heavier jacket, I had one more bottle, and I probably had one arm hanging or on my chest most of the time.

This is just a tease. The thing that really impressed me we how repeatable it is at a big variety of speeds and power levels. Maybe even bigger is that this was on the windiest day at the crit I have ever experienced! If it can look this steady lap to lap, I can't wait to see it on a calm day. The course has sections that are sheltered, some that are wide open and some where the wind can channel between buildings making it change direction.

At first I thought my Crr was different for the 2 level sections, but once I corrected for air density they were the same. Also, the course has 3 meters of elevation change per lap according to Aerolab shows an average of about 3 meters if you look at a bunch of laps. Good! In fact, I can look at the 1st 7 breakaway laps and see that my Cda was a bit lower, until I got smacked with a big gust of wind, probably from an angle. My Cda might have hit as low as .245 in the last 3 racing laps, due to a lull in the wind I guess.

I am working with Robert Chung to understand the data better, and how to interpret and apply it. An interesting question is what the drag of a reclined bent rider and bike does at different yaw angles. Are we more or less sensitive to that than our DF brethren?

Next I need to put the roof rack on my car so I can bring 2 bikes to the crit. Next up will be my 700c Bacchetta CarbonAero 2.0 with Hed Jet9s and a 22 degree seat angle, assuming is a bit calmer.

What I want to test later: bottle locations, wheels, clothes, and seat bags. 

Weather at the time in Camarillo.

Although you can use Golden Cheetah to download data from my Powertap powermeter, I used TrainingPeaks WKO+ 3.0 and imported the file into Golden Cheetah. GC is only supposed to support WKO+ 2.2 files, but it seems to work. 

Here is the link that Robert refers to in his comment, from someone else who has dome some Chung method aero testing.: Colin's Biking Bits


  1. GC can read WKO+ 3.0 files -- I'd just forgotten that it can because I've pretty much switched over so I don't really keep track of what it can import and what it can't.

    You don't really need to load your bikes into the car. I originally started analyzing the data this way because I didn't have a flat road that was suitable for conventional aero field testing. Being able to test on non-flat roads means you can almost always find somewhere to test as long as you don't have to hit your brakes or dodge cars, pedestrians, or other riders. You might be interested in reading some of the experiences of another rider who was trying to do aero testing:

  2. Thanks for the additional info Robert! Actually, where I live there are either traffic and lights or downhills I have to coast on, plus the 2 bikes have different seat angles, and only one is adjustable, so 2 bikes makes it easier. Camarillo is a lot flatter, and industrial parks are usually quiet on the weekend. I keep looking locally though, as driving to ride has always been something I like to avoid.

  3. Downhills you have to coast on? There's a protocol for that. Rather than measure max speed like Willie's been doing you can feed your data into Aerolab and fiddle with the sliders until the hill has the right profile.

    There's another protocol that may also work, depending on how much traffic and wind there is. You could climb that hill twice, once at low power and once at high (like, 300 watts and 150 watts). If you happen to also know the height of the hill from a topo map, you can find the Crr and CdA that make the two elevation profiles match (as long as there isn't much traffic or wind). This has the advantage of getting you estimates of both Crr and CdA from just two climbs.

  4. Actually, I have to brake on the downhills. However, you have me thinking of using my Tuesday night threshold workout for data next. 9 minutes flat/rolling followed by a 1000' climb. Very few cars and usually little wind, until right at the top. I can use the middle section of this ride.

    Now, as I have read all of Simon's work, how can I import a GPS track (.gpx from my Gramin eTrex Vista Cx) into Golden Cheetah. Do I need to convert it to a .csv file first?

    Also you have mentioned "pulling apart Crr and Cda" and I think I get it. I just need to make sure there is a some good speed variation during runs, correct? This will make it easier to adjust the slider for Crr and Cda correctly, right?

  5. Braking is bad but that's why I also suggested the "twice up the same hill at two different powers" protocol. You don't use your brakes going uphill. The two different speeds (you don't have to slavishly hold the power, you just want to get a good separation between the two runs) gives you enough speed variation to identify the right values for the sliders: when the sliders are right, both runs up the hill will match. I give an example on slides 37-41 here:

    I don't have a Garmin but I *think* GC can import some version of Garmin files -- maybe .tcx or .fit files.