You're basically describing the Computrainer "spin-down" function for calibrating the rolling resistance.

Here's a fun fact I discovered when playing around with a second-hand CT setup gifted to me, and something I read in some old CT documentation: the number output by the CT head unit is actually the tangential force of the roller calculated during the spin down.

So...if one had a way to measure the force the CT roller is being pressed into the tire, a CT spin-down would be a pretty easy way to Crr of various tires. One would need to do the roller-to-flat calculations still, but it would be a fairly quick and easy test setup after that.

---

http://bikeblather.blogspot.com/

Excellent approach...now you've got me re-thinking the rig I've slowly been trying to put together...I was going to use an old Quarq prototype spider I had lying around as a power measurement, but the reaction arm measurement is a much more elegant solution.

---

http://bikeblather.blogspot.com/

Oh, Pierre and I had discussed a method to validate and assess the precision of his estimates, and that's what he was referring to.

Different publications follow different standards, so most publications will give examples of how they prefer website citations and references to be formatted. There have already been a few scientific papers that reference my work. A typical citation (but check your specific publication's style) might look like:

Author's Last Name, First Name Middle Name or Initial. “Title of Article (or section if using only part of the website.)” Name of Website, Name of organizational sponsor (not advertising sponsor), date published or updated, URL. Date of access (needed if no date is found.).

Conti SS20 front and Vit CS23 rear both at 90psi. The road is asphalt but pretty low grade, lumpy, with tar and gravel patches. At low speed I'd get a Crr of ~.0027 and a little over .004 at high speed (40km/hr+). The low speed runs were done using my own protocol and Pierre's app with good agreement. The course has a 1.2% slope and there were a couple days when I had a very strong tailwind on the uphill leg. This allowed me to check Crr at high speeds because I could go fast on the climb and still have low aero drag. Otherwise it's tough to tell because if Crr rises with speed it tends to mimic aero drag (increasing CdA matches the data well enough), and it's harder to separate the two.

Sounds good! I was referring to a popular Crr tester. Don't think his setup is as good as it could be.

Ya, roughly that amount. .0027 to .0042. As I mentioned, uphill with a tailwind was how I could reduce aero drag to a low enough amount to see this. It could be something else of course... but everything else I could think of checked out. This was a year ago. I quit using that road because of too many close calls with motorists not looking, and dogs cutting me off.

I guess you were couped up for a long time? Happy to be out now? I would have gone crazy if I was prevented from riding outside for a couple months...

How where you measuring air speed ?

Hmmm...yeah...that's quite a large apparent increase for not really that fast of a ground speed. I'm thinking if I saw something like that I would be poking around at the particular protocol and calculations.

---

http://bikeblather.blogspot.com/

Anemometer.

Not saying it's the same thing but I had funny experiences with a device I used to use in very specific wind conditions, ie very very strong tailwind and some specific yaw conditions.

I do have a weatherflow anemometer I need to get around to testing against another device to see how it behaves.

It's an out-back and I ride near the center of the road. The calibration is adjusted post testing based on distance averaged wind, to remove any bias.

 
Happy to be out, yeah. The first bike ride was a rebirth: after that we know even more how much we love cycling.

I am in Ron's camp and believe there is quite a big variation of Crr with speed, even if Tom test with “the work per lap” approach makes me doubt: I have also a lot of tests with Adam's sheet (what a great tool) with also a straight line... But, a straight line doesn't imply that Crr and CdA are constants….

Take what follows with a grain of salt, it may be hasardous science, suppose:

• we give us a model for rolling resistance, for instance Crr(V in kmh)=0.003*(1+(V-10)/30) according to Ron testimony with a variation of 50% between 10 and 40 kmh.
• we take a CdA correlation found in wind tunnel (“Aerodynamic drag in cycling: methods of assessment” , by changing the constant to match Tom's CdA@40 kmh (0.316m2): CdA(V in kmh)=0.442*(1-0.0043*V+3e-5*V*V)

If we apply the model and plot Y=f(X), with Y=Total drag=mgCrr+CdA*X according to X=wind pressure (=0.5*rho*V^2) and compare with the data of Tom test:

As you can see, in the 20-40 kmh test range, there is a straigth line wheareas CdA and Crr in the model are speed dependent. In other words, may be that the speed dependency in rolling resistance term and in the aero drag cancel out into the equation... I add my sheet for verification.

In my mind, we can't have accurate numbers for Crr and CdA by using the equation behind the virtual elevation method. We can only have very precise number for CdA once you guess Crr at targeted speed and keep the average speed on the lap +/-1 km/h (the model says a variation of 2km/h gives a change of 1% for CdA. 1% is typically the precision you may reach with a good protocol by using VE method).

If you really want accurate numbers for CdA at different speeds, you would need to discover the law Crr=f(speed, temperature, your tires for a given pressure, your road quality). Good luck.

---

Blog | Twitter| Bike CdaCrr app

Hmmm...I see what you're saying about how the plot I showed above might not be a good "diagnostic" of if the effects being discussed exist.

But, I also looked for the 2009 Grappe article (book?) referenced in that paper you linked to get details of the data collection for that plot shown above...can't find anything.

In the process though, I DID find other cycling aerodynamics papers that also had wind tunnel measurements varying with wind speed. For example, here's one that varied from 20-70 kph (~6-20 m/s), with basically little variation in the calculated Cd



From: https://www.researchgate.net/...bicycle_aerodynamics

So...at this point, I'm having a hard time accepting the Grappe plot without knowing more about it.

---

http://bikeblather.blogspot.com/

I am in this position too. Here a link in this case of the dependency of the cda with the speed: https://repository.tudelft.nl/...76-9908-b23dd6f035d8


"In aerodynamic research, the critical condition is typically characterized by the critical Reynolds number, where the Reynolds number is a flow similarity parameter depending on the size of the object, the flow velocity and the viscosity of the fluid. For a cyclist, the body size and the viscosity of the air are largely fixed and, hence, in this work the critical velocity, VC is considered a more practical parameter to indicate the condition that minimizes the cyclist’s drag area. If the critical velocity exceeds the race speed of the rider (VC > VR), surface roughness application can reduce the aerodynamic drag. Instead, when VC < VR the suit is better designed with a smooth fabric."
I think that we need to get only one unknown in the equation. Calculating real Crr and control all around it, we can calculate Cda with less error at a target velocity.
We should have one value of crr and cda for one specific velocity.

Hi all- I'm very late to this thread and don't have much to add but finding the discussion very interesting. I too have tried to separate CdA and CRR in field tests and had mixed results but was never sure exactly why. I've defaulted to testing them separately -- test CdA with known Crr, or Crr with known CdA, all benchmarked by a careful joint test but these are generally quite noisy.

On the Crr testing issue, if CdA and Crr are changing with speed, how about testing at identical speeds but adding/subtracting weight? Has anyone tried this approach? Once my shoulder heals I could give it a go with water bottles filled with lead. (Thinking out loud: lead would add significant weight so thinking that I would need to adjust tire pressure to hold Crr constant.)

It's a pretty nice chapter in a book, but it's in French. As the caption said, those are figures measured in a wind tunnel. Interestingly, the very next figure in that chapter is for similar track cyclists on a velodrome, including one of the same riders as in the figure above. Unfortunately, the details of how the CdA (SCx) was measured aren't given. Here's that figure.



A couple of things:
1) I hate when people put in smooth curves.
2) It's pretty impressive that they went out to 65 km/h.
3) One of the riders (C3) used two different bikes. The data for rider C3 and velo 2 look pretty noisy.

That's exactly what I've been proposing for a while as a validation and precision check (and what Pierre (bugno) and I were chuckling about a few posts above). I've done a couple of these "delta mass" tests with empty and full water bottles. I used to use sand but then switched to small nails. I didn't want to use lead weights both because I wanted to re-use my water bottles and also because lead shot is kinda expensive. (On the other hand, a penny weighs 3 grams so $5 of pennies weighs 1.5kg). For me, with my other parameters, 1.5kg in mass increases rolling resistance force by the equivalent of ~.0001 in Crr, or around 10cm in virtual elevation per km. So over a 3 km run on a windless day, I was expecting to see a change in VE of ~30 cm between empty and loaded tests, which is pretty noticeable. So I did that, but at varying speeds, and was observing a VE difference close to that. But then the wind came back up, so I didn't get to do more tests.

That summary is referring to small areas on the body of the cyclist, and doesn't show any values for a systemic change in the CdA with velocity.

I found the following from the same authors, where they are comparing their PIV (Particle Image Velocimetry) technique to actual wind tunnel balance measurements: https://link.springer.com/...07/s00348-019-2677-6

They didn't test over a very wide range of speed (only ~12.5-15 m/s free stream velocity), but it still looks fairly flat (with if anything, a "hump" in the middle)...also, even they admit the PIV technique is a "rather coarse instrument" for the job.

---

http://bikeblather.blogspot.com/

Is this a case where you'd want to use Gatorskins or Tufos? (To get a bigger "signal")...the larger the Crr, the larger the absolute increase in rolling resistance force per kg, no?

---

http://bikeblather.blogspot.com/

With that little detail, I sure wouldn't hang my hat on those charts...

---

http://bikeblather.blogspot.com/

Friends don't let friends ride Gatorskins or Tufos.

Please do and report. You may be amazed by the device. Even if, with strong wind and high yaw, there are some oddities.

---

Blog | Twitter| Bike CdaCrr app

The choosen scale on the graph does not reveal the difference well, but it seems there is a small decrease on the Cd road position between 20-40 kmh (~2.5%) which is much smaller indeed than the reduction in the Grappe graph (-15% in 20-40kmh range).

In the same time, the TT position has a constant Cd in the two studies.

But, recent wind tunnel tests (no details though) done by skinsuit/sock vendors in the 11-17 m/s range showed the same trend, Cd decreases with speed on TT position:

• 2-3% in the 40-50 kmh range during velodrome tests for skinsuits (as reported by rruff, with data plotted):
  

• 5% in the 40-60 kmh range during wind tunnel tests (socks, by Rule 28): https://www.bikeradar.com/.../rule-28-aero-socks/
  

• What about Cd with speed under 40kmh ? We have this study with wind tunnel data, reported by cyclistgo ("Simulation Model for Road Cycling Time Trials with a Non-constant Drag Area") with a 8-12% Cd reduction between 30-60 km/h and 4% between 30 and 40 km/h:
  

So... It may depend on the rider position, and some other things too (bike, gear). But I keep thinking, with a limited confidence, that a small variation of Cd with speed is enough to hide much larger variation of Crr in regression/VE methods.

It will be nice if Xavier or Damon could give us a glimpse of what they know better than us on the subject about relationship between Cd and wind speed.

---

Blog | Twitter| Bike CdaCrr app

I'll try and find some helpful graphs on all the wind tunnel testing we did on fabrics as part of the new Nopinz Flow skinsuit/clothing range. Friction drag is affected by airspeed far more than pressure drag, and we did have some fabrics we made that were good at some airspeeds but slower at others, so had to make decisions on what the best overall option was.

As more people focus on clothing I think you'll get two schools of thought (which can make it difficult when you're doing self aero testing and not as in control of your speed as you would be in a tunnel) - those that focus on a specific airspeed for their protocol and optimise around that, or those who optimise for a wider range. The new Zone3 suit is a good example, we tested it and overall over a wide range of speed it's better than everything, but it is possible to pick out 1 or 2 competitors that are as quick/quicker in very specific scenarios. I can't publish data on that though as it's all owned by Nopinz/Zone3 but will see what we have for the AeroCoach/Nopinz Flow stuff.

ps. when we've been doing our testing we haven't seen large changes in Crr based on rider speed.

---

AeroCoach UK
http://www.aero-coach.co.uk

Thanks, for your last sentence, and confirming, if I understand correctly, that we need to aero test at different speeds for a hilly or windy TT and at a specific speed for a non windy flat TT (or for track racing), cause Cd is speed dependent.

---

Blog | Twitter| Bike CdaCrr app