CdA changes between headwind and tailwind - to be expected?

[Jesper]

A simple example to illustrate the headwind/tailwind difference: I'm riding with my winter setup in headwind spending total 300W, say, rolling resistance Prr=25W and aero drag is Pd=275W. But since I put too low Crr into AP (from my summer setup) it believes that my rr power is only Prr=15W, and that is aero drag is the remaining Pd=285W. Hence, the relative error on CdA readout will be 285W/275W = 1.036.

Now, I do a U-turn going to the tailwind, and the wind speed happens to be just so that I will ride at double speed when I spend the same 300W total, hence Prr=2x25W=50W (since it is proportional to my speed) and Pd=250W. But since I still have the lower Crr in my AP it will beleive that Prr=2x15W=30W and Pd=270W. So now the relative error on the CdA readout will be 270W/250W = 1.080.

Therefore, for this example there will be an apparant higher CdA in tailwind compared to headwind by a factor 1.080/1.036=1.042.

Now, I'm quite new to Isaac, but didn't I come across a feature to actually analyse the Crr from the data, so that such issue could be discovered? Please, correct me.

[Velocomp]

I think the analysis from Jesper is right on. In principle, if the "actual Crr" of the bike is higher than the "profile Crr" stored in AeroPod, then the tailwind CdA measurement will be high. In Jesper's example, the tail wind CdA would be about 4% higher than the headwind CdA.

However, while the headwind/tailwind principle seems correct, the real-world impact will be much less than 4%. Note that in the example used, the rider was in a headwind/tailwind situation such that the tailwind bike speed was 2x the bike speed in the headwinds. That's a really high wind! And also, there needs to be a meaningful difference between actual and profile Crr--about 67% in the example used.

Yes, it is possible to play-around with Crr in Isaac. You can use the command "Analyze/Tweak CdA, Crr..." to determine the effect of adjusted values on AP watts. I use this command all the time when I am working with beta firmware.

[Hein]

If Crr can only explain a small part of the effect, what about another slightly offset constant? If the stored system weight is to low, CdA should rise uphill and while accelerating. While coasting and going downhill CdA should go downwards.
As far as I understand Aeropod, accelerating and decelerating effects should add up to the measured speed.
If the system weight is set to low, you need more power than expected to go uphill. Since Crr and system weight are fixed in the formula and ground-speed, wind-speed and power are measured, CdA is the only free parameter to vary. As a result CdA has to climb. This effect should apply to all situations where acceleration is relevant.
If the system weight is set to high, the opposite effect should appear.
With every rotation of the crank, there is a small phase of acceleration and a small phase of drafting. If those patterns change with head and tailwind, this could result in different CdA readings.

[EHB]

I’ve wondered before if effective weight varies with airspeed & if the effect is significant. Just imagining a cyclist I can easily see an aerofoil type effect at higher speeds, with higher pressure underneath the rider than above the back. I can’t remember the formula for CRR but I think it is linear with downwards force. If cycling at (air) speed significantly reduced the force I’d expect CRR to significantly fall & (assuming CRR guesstimate is held constant) then CdA would appear to go up.

[Velocomp]

If Crr can only explain a small part of the effect, what about another slightly offset constant? If the stored system weight is to low, CdA should rise uphill and while accelerating. While coasting and going downhill CdA should go downwards.
As far as I understand Aeropod, accelerating and decelerating effects should add up to the measured speed.
If the system weight is set to low, you need more power than expected to go uphill. Since Crr and system weight are fixed in the formula and ground-speed, wind-speed and power are measured, CdA is the only free parameter to vary. As a result CdA has to climb. This effect should apply to all situations where acceleration is relevant.
If the system weight is set to high, the opposite effect should appear.
With every rotation of the crank, there is a small phase of acceleration and a small phase of drafting. If those patterns change with head and tailwind, this could result in different CdA readings.

If system weight is too low you're correct that, on uphills, AP watts would be a bit low, which in theory could CdA to rise some. However, on uphills the effect of wind is much less--most of your applied power is going to overcome gravity. In fact, if bike speed is less than 12 mph, we don't attempt to measure CdA, because wind watts are so low.

On the flats, the only situation where system weight has an impact is when you are changing bike speeds (accelerating). In a TT, where CdA is important, bike speeds are pretty constant on the flats, so weight errors would not be much of a factor.