CdA changes between headwind and tailwind - to be expected?

RolfK
Posts: 26
Joined: Fri Nov 02, 2018 4:47 am

CdA changes between headwind and tailwind - to be expected?

Post by RolfK »

Hello,

I've been doing my first tests with my aeropod. First of all, the calibration process works a lot better than the powerpod-process, very nice.

Resultswise, I'm a little puzzled over the pattern of the results - the absolute level of the CdA-numbers don't bother me much at this time.

I've done laps on an old airport runway, each lap is a little over 5 km long. The wind was 4-5 m/s, mostly headwind and tailwind.

The puzzle: First I did four laps counterclockwise, and I noticed that the CdA was significantly lower into the headwind than with tailwind. To be sure that the difference in fact was caused by wind direction and not the slight difference in asphalt texture between the two long straights of the runway, I then did two laps going the opposite way around. However, same pattern, lower CdA into the headwind.

Now, across all six laps, the *average* CdA's vary within an acceptable band, all things considered, but I don't understand the pattern - and I didn't expect it. Should I be concerned? It does seem to put restrictions on which courses will work for testing.

I've constructed a couple of charts in WKO4 that visualize CdA, based on a .TCX-export of the ridefiles from Isaac, see below.

I've also attached both riding and calibration files, as well as a matching Garmin .fit file - I couldn't get Isaac to merge the GPS-information into the ride files.
WKO4 - custom analysis pack
WKO4 - custom analysis pack
Screenshot 2018-11-03_20-26-30-384.jpg (115.19 KiB) Viewed 15569 times
Attachments
iBike_11_03_2018_1054_4_km_HiDef.ibr
Calibration ride
(48.33 KiB) Downloaded 193 times
iBike_09_17_2018_0000_32_km_HiDef.ibr
Test ride, six laps
(498.75 KiB) Downloaded 200 times
RolfK
Posts: 26
Joined: Fri Nov 02, 2018 4:47 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by RolfK »

The Garmin-file
Attachments
2018-11-03-11-03-52_fit.ibr
(118.3 KiB) Downloaded 209 times
Velocomp
Velocomp CEO
Posts: 7793
Joined: Fri Jan 11, 2008 8:43 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by Velocomp »

1) Export your Garmin file in .tcx or .gpx format. This can be merged into Isaac.

2) Your AP appears to be properly calibrated

3) I also see the headwind/tailwind difference. Is this a loop route, or are you going to each end of the runway, then turning around?
John Hamann
RolfK
Posts: 26
Joined: Fri Nov 02, 2018 4:47 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by RolfK »

Thanks, I will give try to export the file and merge.

It is a loop, I am doing laps and coasting as little as possible, but one of the two curves connecting the "taxi-runway" w. "the starting-runway" was not wide enough to permit going full gas thru it given the wind direction. The Garmin-file can explain the course a little clearer.
RolfK
Posts: 26
Joined: Fri Nov 02, 2018 4:47 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by RolfK »

Hello again,

I've merged the GPS-info into the ride file; the TCX-format worked.

Having done that, I noticed an interesting point, also pointed out in another thread: there is a small "climb" / "descent" on the course, depending on the direction taken, not more than 5 meters of ascent / descent, but this is clearly visible in the CdA-numbers:
Screenshot 2018-11-11_14-48-37-314.jpg
Screenshot 2018-11-11_14-48-37-314.jpg (40.29 KiB) Viewed 15531 times
This leads me to wonder if there is a general issue with elevation changes? If it's the small altitude difference from one end of the course to the other (probably not more than 10 meters of elevation) that is cause of the fluctuations in CdA, not the headwind / tailwind?
Attachments
iBike_09_17_2018_0000_32_km_HiDef_GPS.ibr
Test rides with GPS data merged in
(545.74 KiB) Downloaded 193 times
Velocomp
Velocomp CEO
Posts: 7793
Joined: Fri Jan 11, 2008 8:43 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by Velocomp »

What you're seeing at the places marked is AeroPod "resetting" CdA to the default value, every time the lap marker is clicked. Of course, if your current CdA differs from the default CdA, there will be an abrupt transition. We may need to change the way CdA transitions each time the lap marker is clicked.

It looks like AP is measuring a maximum elevation change of 7m, from the lowest point in the course to the highest point. Is this consistent with your other information?

I very much appreciate your careful analysis; it will help us make a better product.
John Hamann
RolfK
Posts: 26
Joined: Fri Nov 02, 2018 4:47 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by RolfK »

Oh. My starting point of the laps was very near the "climb", so a spurious correlation? But how does AeroPod know that I am hitting the lap-button on my Garmin - is that transmitted via ANT+ or is it the ConnectIQ App working magic under the hood?

In any case, yes, this lap-CdA-reset is definitely not good for data quality. What were the reasons for implementing it?

Still, there could be confounding effects between altitude diff and wind direction diff (the 7 meters sounds correct).
Velocomp
Velocomp CEO
Posts: 7793
Joined: Fri Jan 11, 2008 8:43 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by Velocomp »

Yes, when you hit the lap marker in your Garmin, it resets AeroPod too, by ANT+!

The way we implemented this is a bug. What we want to do is: when you hit the lap marker, the average CdA from your previous lap is shown for a few seconds on your screen, then the CdA goes "live" again.

We will fix this in the next FW release.
John Hamann
RolfK
Posts: 26
Joined: Fri Nov 02, 2018 4:47 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by RolfK »

That makes sense - I'll look forward to that feature.

I guess the headwind / tailwind / elevation puzzle will take a little longer to figure out.
Velocomp
Velocomp CEO
Posts: 7793
Joined: Fri Jan 11, 2008 8:43 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by Velocomp »

RolfK wrote: Sun Nov 11, 2018 10:25 am That makes sense - I'll look forward to that feature.

I guess the headwind / tailwind / elevation puzzle will take a little longer to figure out.
Yes, we are thinking about this one too!
John Hamann
roadrunner
Posts: 6
Joined: Fri Nov 02, 2018 9:08 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by roadrunner »

FYI: I've seen the same regarding cda-difference in tailwind - headwind, on out and back runs.

At first I thought it was down to me being better at actually focusing on position and staying low during headwind, but the figures seems to show to large of a difference. I've also seem to get lower cda-numbers when going uphill, which doesn't really make sense.

Really interesting to hear about the lap-button function. Good to know about current setup and I love the planned functionallity.
Velocomp
Velocomp CEO
Posts: 7793
Joined: Fri Jan 11, 2008 8:43 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by Velocomp »

roadrunner wrote: Mon Nov 12, 2018 5:28 am FYI: I've seen the same regarding cda-difference in tailwind - headwind, on out and back runs.
Please post a ride file.
John Hamann
roadrunner
Posts: 6
Joined: Fri Nov 02, 2018 9:08 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by roadrunner »

THE GPS is unfortunately not synced as I forgot to start my GPS tracker. But attach the Garmin file as well if that helps.
Attachments
activity_3143009352.tcx
Garmin file
(3.31 MiB) Downloaded 197 times
iBike_11_06_2018_1218_34_km_HiDef.ibr
Ride
(581.57 KiB) Downloaded 205 times
iBike_11_06_2018_1212_4_km_HiDef.ibr
Calibration
(50.65 KiB) Downloaded 190 times
Velocomp
Velocomp CEO
Posts: 7793
Joined: Fri Jan 11, 2008 8:43 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by Velocomp »

roadrunner wrote: Mon Nov 12, 2018 3:24 pm THE GPS is unfortunately not synced as I forgot to start my GPS tracker. But attach the Garmin file as well if that helps.
I married-together your files, then merged the GPS data with the bigger file. It is attached.

I was able to use the GPS map to see where your out-and-back (headwind/tailwind) section was located.

The "out" portion measures CdA = .246
roadrunnerout.png
roadrunnerout.png (522.95 KiB) Viewed 15474 times
The "back" portion measures CdA = 0.235
roadrunnerback.png
roadrunnerback.png (432.01 KiB) Viewed 15474 times
It looks like wind conditions changed some between the out and back, so I think these numbers are pretty close. For the entire ride from minute 6 to minute 49, CdA was 0.240--very very close to the AeroPod cal ride measurement of 0.240!
Screen Shot 2018-11-12 at 3.55.27 PM.png
Screen Shot 2018-11-12 at 3.55.27 PM.png (444.28 KiB) Viewed 15474 times
NOTE: you've got your HR strap paired to AP, so I used the Tools/CdA Analysis feature in Isaac to determine your CdA numbers
Attachments
Roadrunner_11_06_2018_1212_38_km_HiDef.ibr
(466.89 KiB) Downloaded 187 times
John Hamann
roadrunner
Posts: 6
Joined: Fri Nov 02, 2018 9:08 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by roadrunner »

That’s awesome! Many thanks for the help and analyse.

Gonna do a re-run on same course tomorrow. Hopefully will see some more interesting numbers. 😃
ruud_g
Posts: 5
Joined: Wed Oct 10, 2018 4:10 pm

Re: CdA changes between headwind and tailwind - to be expected?

Post by ruud_g »

“n any case, yes, this lap-CdA-reset is definitely not good for data quality. What were the reasons for implementing it?”

Yes. This is also what I have seen in my rides. A steep curve from the lap button which then messes up the data and then the AP trying to get back to the reasonable numbers which sometimes do and sometines don’t appear.

In the attached file you see results of 3 rides under ideal circumstances (almost no wind, flat on course with no traffic, and just a velodrome round). You don’t want to see those different bumps nor the long tails in these distributions. Caused by a.o. the lap button and sensitivity of the recalibration of the AP at the moment. You do seem to see a pattern but it can of course definitely be improved.
Attachments
0F043CD8-E10E-4362-BC68-0D020DE2F5DF.png
0F043CD8-E10E-4362-BC68-0D020DE2F5DF.png (39.66 KiB) Viewed 15457 times
DavidTriska
Posts: 24
Joined: Sat Oct 13, 2018 4:50 pm

Re: CdA changes between headwind and tailwind - to be expected?

Post by DavidTriska »

hey ruud_g, that graphical analysis is great! how did you extract the raw numbers from isaac?
ruud_g
Posts: 5
Joined: Wed Oct 10, 2018 4:10 pm

Re: CdA changes between headwind and tailwind - to be expected?

Post by ruud_g »

Hi David. I used the csv files which are generated with Isaac. I labelled the time periods from I knew I rode in a certain time period and used Python (library Seaborn) to make the voilin plots. This gives me exact the right angle to analyse the distribution of the data for the positions
Velocomp
Velocomp CEO
Posts: 7793
Joined: Fri Jan 11, 2008 8:43 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by Velocomp »

This is a great presentation of the data.

The reason for the "long tails" in the TT and Straight Up measurements is because of the feature/bug that resets CdA to the default value, every time the lap button is clicked. Since the TT and Straight Up measurements deviate the most from the default CdA, this puts long tails on the values.

We will be updating AeroPod FW soon to address this bug.
John Hamann
ruud_g
Posts: 5
Joined: Wed Oct 10, 2018 4:10 pm

Re: CdA changes between headwind and tailwind - to be expected?

Post by ruud_g »

Thanks john
DavidTriska
Posts: 24
Joined: Sat Oct 13, 2018 4:50 pm

Re: CdA changes between headwind and tailwind - to be expected?

Post by DavidTriska »

John that sort of presentation for CdA analysis would be perfect for a future version of Isaac - the spotters among us love it!
Velocomp
Velocomp CEO
Posts: 7793
Joined: Fri Jan 11, 2008 8:43 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by Velocomp »

DavidTriska wrote: Wed Nov 14, 2018 4:24 pm John that sort of presentation for CdA analysis would be perfect for a future version of Isaac - the spotters among us love it!
I'm already thinking about how to do this...
John Hamann
RolfK
Posts: 26
Joined: Fri Nov 02, 2018 4:47 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by RolfK »

Just a quick follow-up, I rode a few laps on a different course today.

First of all, I'm impressed with the overall consistency of the CdA estimates across the laps. As you can see, the three laps resulted in CdA values of 0.258, 0.256 and 0.257, even though I had to coast a little on the third lap because of a truck (these laps where done in a quiet industrial zone, but one truck had to leave saturday morning). I did not press the lap-button during the laps, to avoid the artefact of CdA-reset.

Also, the estimated power numbers are very, very close to the DFPM numbers; on average less than 1 watt difference. Again, impressive.

However, I see the same pattern re. headwind and tailwind on this course as on the other one. I've attached the ibr-file.

Lap analysis in WKO4
Lap analysis in WKO4
Screenshot 2018-11-17_15-04-14-535.jpg (129.53 KiB) Viewed 15334 times
Attachments
iBike_11_17_2018_1143_23_km_HiDef_GPS.ibr
(440.32 KiB) Downloaded 135 times
RolfK
Posts: 26
Joined: Fri Nov 02, 2018 4:47 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by RolfK »

Another view; the pattern is fairly consistent across the three laps. Noticeably, near the end of lap one, CdA is is unchanged for more than 1 km / almost 2 minutes.
Screenshot 2018-11-17_22-25-25-918.jpg
Screenshot 2018-11-17_22-25-25-918.jpg (48.42 KiB) Viewed 15314 times
Velocomp
Velocomp CEO
Posts: 7793
Joined: Fri Jan 11, 2008 8:43 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by Velocomp »

The two sharp corners at the west end of your route may also be a contributing factor to the CdA "bump"; the corners are only 1 minute apart.

Otherwise, I agree this is really solid data!

The reason your AP watts are so close to your DFPM is because your "default" CdA in the AP file is 0.255--extremely close to the measured values on your ride.
John Hamann
RolfK
Posts: 26
Joined: Fri Nov 02, 2018 4:47 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by RolfK »

Thank you. Yes, the corners may have an impact, but the overall tendency remains. CdA climbs with tailwind, falls with headwind. It would be interesting to see CdA-numbers without smoothing, to see how abrubt the change really is. Please, make CdA smoothing configurable, like smoothing of power numbers is :)

"Noticeably, near the end of lap one, CdA is unchanged for more than 1 km / almost 2 minutes." I think this is worth looking into, a completely steady CdA for almost 2 minutes seems unrealistic.

I will try a different course as soon as weather permits. Winter is closing in over here, so it isn't getting easier to find useable conditions, unfortunately.

Capture.PNG
Capture.PNG (218.95 KiB) Viewed 15293 times
Velocomp
Velocomp CEO
Posts: 7793
Joined: Fri Jan 11, 2008 8:43 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by Velocomp »

To the left of your Isaac graphs are icons showing power, cadence, etc.

Right click with the mouse over the icons and select Advanced Stats. You will find that data presentation far more useful.
John Hamann
Jesper
Posts: 3
Joined: Wed Nov 21, 2018 4:06 pm

Re: CdA changes between headwind and tailwind - to be expected?

Post by Jesper »

On the headwind/tailwind issue just wondering: Since many of us have switched to winter tyres with higher Crr, wouldn't this give apparantly higher CdA, but more too high in tailwind and less too high in headwind?
Velocomp
Velocomp CEO
Posts: 7793
Joined: Fri Jan 11, 2008 8:43 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by Velocomp »

Jesper wrote: Thu Nov 22, 2018 1:20 am On the headwind/tailwind issue just wondering: Since many of us have switched to winter tyres with higher Crr, wouldn't this give apparantly higher CdA, but more too high in tailwind and less too high in headwind?
This is a surprisingly tricky question!

Crr determines the number of watts produced by frictional rolling resistance. Watts caused by frictional Crr are proportional to speed; for example, (and this is made up), if frictional watts are 10W at 10 mph, they would be 20W at 20 mph.

Note that the value of frictional watts calculated by AeroPod is dependent on the value of Crr selected by the user.

Let's suppose your winter tires have higher Crr, but you don't adjust AeroPod profile settings to account for the higher Crr. And let's suppose the increase in Crr is high enough such that it can be noticed. And finally, let's suppose everything else on your bike ride is the same (bike speed, wind speed, slope, acceleration). Here is what will happen:

1) With winter tires, to hold the same bike speed you will have to pedal a bit harder, in order to overcome the increased frictional forces. The DFPM will measure slightly higher applied watts, because Crr has increased.

2) AeroPod will NOT measure higher opposing frictional watts, because it's internal Crr setting is the same--from your summer tires.

3) So, AeroPod watts reading will be slightly lower than the DFPM reading. AeroPod will therefore INCREASE measured CdA by an amount sufficient to account for the higher watts reading reported by the DFPM. So, the effect of higher Crr on CdA is that CdA value goes up a bit. And I mean--just a small amount. Why? Because the opposing force of wind is MUCH greater than the opposing force of friction. Therefore, the impact of changes in friction, say from summer to winter tires, have a relatively small impact on total opposing force, relative to the impact of wind.

All of the above is why we say, multiple times in the instructions, that we assume Crr is constant, and that you should do CdA testing on roads where the pavement is not changing!

OK, now to the next question: what effect would headwind/tailwind have on the "Crr effect" of winter tires?

The answer is, not much. Why? It's for the same reason as above: wind forces are much greater than frictional forces. Headwinds and tail winds will affect bike speed (assuming you're holding the same level of power). With headwinds, slower bike speeds--but only slower enough to overcome the even stronger force effect of wind. Frictional force will go down a bit in a headwind (slower bike speed), but the reduction in frictional force, which varies linearly with bike speed, will be overwhelmed by the increase in wind force, which varies as the square of bike speed.
John Hamann
GooseRider
Posts: 58
Joined: Wed May 10, 2017 10:52 am

Re: CdA changes between headwind and tailwind - to be expected?

Post by GooseRider »

John,

Thank you for good explanation. Just to add to the argument, in a tail wind, the speed will go up, frictional forces will go up but wind resistance is smaller. Thus, could errors in Crr estimation create the difference between head and tail winds? In principle there should be no difference, right?

Also, why wouldn't the most accurate CdA measurement happen on steep downhills? The power meter would read zero (no inaccuracy) and it is all down to accuracy of wind measurement and slope measurement (both under Aeropod control). Also, the higher speeds of downhill would create very large wind resistance making wind force much bigger than friction force.

Thanks,

Morten
Post Reply