Drag Measurement - Coastdown Testing

[@gerritgovaerts8443]

Done that a decade ago on my tail faired recumbent low racer bike with some succes (numbers looked acceptable) but I did not think about including a term for rotational inertia . Thx for that ! My Crr estimate was checked by doing a low speed coastdown (from 10 kph) when aero drag is negligible (certainly so on a very aero recumbent bike with a very reclined body position ->low A) . Crr came out as approx 0.0035 which is very low for 20 inch bike tire and even low for a normal 28 inch road race tire , but this was a very special limited edition race tire (not for sale unless you had special access to Schwalbe) with no punction protection whatsoever and a very limited life (about 1000 km) .

[@AirShaper]

Hi Gerrit, cool, thanks for sharing your experience! You must have been quite precise with your measurements to come up with those values. What GPS (or other position tracking) did you use?

[@gerritgovaerts8443]

@@AirShaper no GPS , but a special app and interface (with sensor) on a Palm device (PDA's back then were popular before the age of Smart Phones) . It actually tracked every revolution of my front wheel with a very accurate time stamp . I could transfer that recorded CSV file to my PC and then do all the calculations (regression stuff) in Excel. It did require a very accurate roll out measurement of the circumference of the wheel under load

[@AirShaper]

@@gerritgovaerts8443 Ah that's fantastic, indeed a very precise way of measuring the distance/velocity (by derivation). Interesting note on the changing circumference under load!

[@jaganadithya7347]

Fgrav (gradient force) is supposed to be just m*g*sin(theta) right?

[@namenotshown9277]

did you do a coast down test on aptera in real world?

[@musikSkool]

How could rolling resistance be linear? The wheels turn faster at higher speed, so wouldn't they deform more? Hmm, maybe their deformation per degree of rotation actually reduced at speed so it becomes a constant. I guess, but I suspect the error bars are suspiciously larger than the difference between a linear and non-linear rolling resistance.

[@AirShaper]

Indeed, it isn't linear! It is at best "sort of linear", which allows you to filter out aerodynamic forces & rolling resistance forces. But if you want to extract more accurate values, you'll need a non-linear model for rolling resistance. And you're right, as you normalize by the vehicle speed, the rolling resistance coefficient can also go down in function of speed.

[@gtx4290]

Hm, this equation isnt really usefull, because what is "v" in rho*v*S*Cx/2? starting test velocity? ending test velocity? Mean velocity during test? Same thing for "a"... mean acceleration?

[@AirShaper]

Hi GTX, you have to plot the measured deceleration over time. Then, you can build a theoretical curve to predict the deceleration over time, based on the forces (using F = m * a). Those forces are calculated using the equations shown in the video (and thus they vary in function of speed, and the speed drops over time). Once you have those equations, you can play with the coefficients Cd and Crr to find the best match with the measured curve. Bit hard to explain in a comment, and this is the most basic version! To learn more, you could perhaps check this paper: www.researchgate.net/publication/262726078_Coast_Down_Test_-_Theoretical_and_Experimental_Approach