The Virtual elevation method to measure Cda and Crr

[NorBike]

I am an engineer, working with roads, terrain-modelling and GPS survey.
I have access to needed equipment to measure heights and slopes in millimetre accuracy.

So, my experience is that the coastdown method is far from accurate enough to measure Cda and Crr. The same conclusion have I made with the Newton+/Isaac Cda descent measurement. Just stop wasting time with it...it's a totally waste of time. I have experience here...... :-/
As velocomp says, just use the Cda/Crr estimates from the setup in Isaac. You will get an acceptable results with it.

BUT..two more accurate ways to measure Cda and Crr is either calibrate with a power meter, if you know someone who owns it. I did it and got great results after some tweaking. Another way is the "Virtual elevation" method. This is a nightmare to get right, but then you get a very accurate Crr and Cda. There's a spreadsheet and lots of information about it on the internet. If you are serious about wasting time getting totally correct snapshot of your Cda and Crr, this is the best method, unless you own a SRM or a wind tunnel. But remember, this i just an snapshot of your riding position and the tire friction, and will vary when riding.

There are some factors that are very important when using this method.
1. Absolutely no wind.
2. Absolutely correct height loss.
3. Absolutely correct start/stop point every time you running the test.
4. Correct total weight
5. Correct Air density.
6. 1.sec speed logging and excel experience

1. Newton can measure the wind. Do out/back ride on the test road. Max 0-2km/t wind both ways and the Newton should measure the same both ways (one way negative, one way positive, or no wind)
2. This is tricky. I measure it with my company's pro GPS in cm accuracy, but I believe you can find an alternative if you have access to an accurate map that shows terrain heights in 1meter resolution, and also use these terrain lines as a start/stop point.
3. This is more tricky. You have to start/stop your clock in the precise same points every time. Newton sucks here.
4. Buy something accurate to measure your total bike and body weight
5. The Newton measures the air density. You find it if you export the ride file to *.csv, but Newton need 15-20minutes before it's accurate.
6. Your Ibike Newton can log your coastdown, but sucks in the start/stop measurement.

The newton helps you with nr 1 and 5 and 6. Nr.4, you have to weight yourself.
Then we have nr 2 and 3 left. Ibike can't measure height accurate enough, but an update could help us measure the absolutely correct start/stop point. For nr.3 you just have to find the exact height in 0.1-0.2m accuracy or it will destroy the Crr number.

I am lucky enough to have a totally straight road/test-track with a slope of 1.236%, measured with pro GPS. To get around the start/stop problem I can exclude the first and last couple of logs because the road has the same slope. My biggest trouble now is the wind, because I live in an windy area.

The virtual elevation method is a nightmare to get right, the coastdown method is a waste of time, the newton setup is easy and the fastest way to use your time out on training instead:-)

One other thing. The 5min auto tilt adjustment should be an on/off thing. If you have done something stupid before a ride it saves you, but doing lots if weird things.

Maybe the Newton in the future could include the Virtual elevation method and a correct stopwatch logger to measure/log the coastdowm….. and include a 5min auto tilt on/off....

[RChung]

Another way is the "Virtual elevation" method. This is a nightmare to get right, but then you get a very accurate Crr and Cda. [...] There are some factors that are very important when using this method.
1. Absolutely no wind.
2. Absolutely correct height loss.
3. Absolutely correct start/stop point every time you running the test.
4. Correct total weight
5. Correct Air density.
6. 1.sec speed logging and excel experience

VE can be pretty demanding if you want absolute accuracy.

1. If you have an accurate and precise way to measure wind speed *and direction* you can still use the method but, in general, you're right: you'll need zero wind if you want absolute accuracy.

2. Sometimes it's easier to solve for the (CdA,Crr) combinations as a function of elevation change. That is, suppose you think the test section has a total elevation change of X meters. Given your data, calculate what the CdA and Crr would be for a total change from X-1 to X+1 meters, at 0.1 m intervals. That will give you an idea of the sensitivity of the estimates to an error in height.

3. Sometimes it's handy to have a "feature" or two along the test section that you can then use to anchor the start and end. I have used a test section with "flat" spots along the course. Since speed and distance are easy to measure right, once you've anchored each run on a known spot you can count forward or backward in distance to locate the start and end points without having to worry about triggering the interval marker. I've taped the floor mat from my car down on the road and rolled over it -- the "bump" in VE provides an anchor point in location.

4. You'll want to vary your speed across test runs.

[lorduintah]

Given that the math (or complexity of the equations) involved in any of these analyses is not simple, that there are plenty of interactions of one component to one or more of the others, if all the calculations were laid out, I would personally opt for a designed experiment - with plenty of replicates.

Almost all examples of calibration I see regarding not just the iBike/Newton, but quite honestly any of the other power measuring tools - DFPMs - lead to a few of this or a few of that to get some output - without appreciation for the levels of interaction, without the consideration of reproducibility - either in the environment or the measurement tool. Some or most of those interactions are pretty obvious when you look at the equations involved. Has anyone looked at any Gauge R&R?

A bona fide DOE is likely the best way to capture the variations - not with one or so coast downs, not one or so calm days, or one or so riding positions or attire or an out/back ride. If you really spent the time to perform a statistically sound DOE, you could probably use you coefficients du jour - because they are not going to be the same day to day. And with a set of them a large variation in conditions could be employed.

But then - I am just as happy to dial in a reasonable set up and ride with the knowledge that as I improve, the readings I get will, too - even if I cannot take my results to the National Bureau of Standards. Let Isaac tweak a little, but in general - not worry about 350.2 vs. 348 watts.

Finally - it is entertaining to see how many users are really doing these deep dives into measuring and calibrating - and not using more of what at least some of them may have learned in one of those optimization classes in school.

FWIW ---

Tom