Don't have an account yet? You can create one. As a registered user you have some advantages like theme manager, comments configuration and post comments with your name.
FixedGearFever - :: View topic - Aerodynamic Drag Testing - How to do it.
Hi folks, this is long but hopefully of interest/use to some of you.
I guess it qualifies me as a nerd!
OK - I did some aerodynamic drag (CdA) and rolling resistance (CRR) testing at the velodrome the other day. It’s one of the benefits of having a power meter described in Andy Coggan’s and Hunter Allen’s book, so I thought I’d share my initial findings and provide a tool to help others do it if so inclined. Firstly though, many thanks to the good Dr Coggan who helped me get this right, provided advice on the process and checked my calculations (which I eventually got right)!
Why?
I want to improve my pursuit performance (and those of others I may be coaching/assisting), so over time I intend to do tests to determine if changes, either to position or a new piece of equipment, represent an improvement in aerodynamics or not (and in the process eliminate other variables such as changing fitness, rolling resistance and weather conditions). As far as I know, apart from expensive wind tunnel testing, this is the only objective means to do this. A stopwatch and lap times simply won’t do it. I also like learning about this stuff and so I thought I’d give it a crack. Turns out it’s not that hard to do.
This is most useful for those interested in the pursuit and track time trial events (actually it works just as well for road/TT bikes as well if you are into those events). The method enables you to measure your own CdA & CRR, which you can then track as changes are made. The lower you can get these coefficients the better (as long as you remain legally positioned and don’t sacrifice much ability to produce power), since you will go faster without having to increase power output. Nifty, huh?
Data collection (actually doing the efforts on the bike) is the hardest part as it is a little time consuming (and depending on fitness can be exhausting!). The calculations are pretty straightforward (now I know how) and can be done readily in Excel. I’ve attached initial results based on my standard mass-start track bike set up as an example, including the calculations and a couple of charts, should anyone else want to do similar testing. You can just plug your own data into the table.
Please refer later post - I have attached an updated version.
Some lessons:
1. Collecting the data probably needs 2-3 hours.
I tried to do this exercise and get in some VO2 Max efforts as well since track time was at a premium and I’m still getting ready for upcoming championships. That was a mistake as I simply ran out of time to complete either properly. You will also need to record weather conditions, most particularly barometric pressure and temperature. I used a portable electronic weather station.
2. Ideally you want power & speed data from about 15-20 runs of at least 1 minute each and over a range of speeds.
I got good data even if it is only one pass at 8 different speeds. Ideally you would want to do 2 or 3 of these sets to eliminate any erroneous data points. There is a limit to the resolution of the data, hence it is important to realise the limitations when drawing conclusions on changes (e.g. remember power meters themselves are only accurate to within a couple of percent). And make sure the power meter is properly zeroed each time (a process that takes a few seconds at push of a button or two)!
3. You need flat and still conditions.
Flat is easy in a velodrome. I’m lucky as I have ready access to an indoor velodrome. Even so, there were quite a few more riders at the track than I had anticipated, including some derny action, so I had reduced opportunity to make sure there was as little wind disturbance as possible (not to mention recovery time for legs that were not feeling the best on the day) and making sure I was not hogging track time (we're all good buddies though). If you must do it outdoors, then pick a really calm day.
4. Stay smooth!
You need to maintain as constant a speed as possible during each run, keep the starting and finishing speeds the same and stay at the same height on the track (black line is easiest, no rolling down the track to get up to speed). Start/finish at the pursuit line on the straights is best. Trying to ride a perfect line and constant speed for at least 1 minute is quite challenging to do (try it – and I mean within 0.5km/h) over a range of speeds from 20km/h up to 45+km/h. Slow speeds may need to be ridden on the blue band if track grip is an issue.
5. Use the power meter software to extract data later
Using Cycling Peaks (or whichever power meter software you have) is really helpful to extract the data, especially to pick out starting and finishing points of each interval so that the start and finish speeds are equal. Once you have that, then all you need is the average speed and power for that pass. If start/finish speeds are not equal, then an additional calculation is needed to adjust for the difference in kinetic energy between the start and finish of the interval. It isn’t hard to do the adjustment if needed (I got worked it out in the end) but it is much easier/quicker to use the software to find start/finish points that are at the same speed.
6. The non-linear relationship between power and speed
See Chart 2 – while most know it intuitively, the chart shows real data demonstrating how much harder it is to increase speed the faster you are going. In terms of the power output difference, going from 30 to 35km/h is completely different to going from 50 to 55km/h. Hence small speed gains made by sprinters/top level pursuiters are hard won gains, often resulting from significant increases in power (or improvements in aerodynamic drag).
7. But wait - there's more!
As a bonus you also get a coefficient of rolling resistance from the data, so it is also possible to track changes to CRR as well. Woo hoo!
Happy to answer questions to the limit of my ability but I’m sure Andy will gladly chime in if needed!
Cheers
Alex
Last edited by alex on Tue Apr 25, 2006 3:15 am; edited 1 time in total
My only comment would be that the intercept of the relationship is your actual rolling resistance*, not the coefficient of rolling resistance. To derive the latter you need to divide the value of ~4 N by your mass and by g (i.e., 9.81).
*There will be a small contribution from hub bearing friction, but it's tiny. OTOH, if performing these tests using a device that measures power at the crank (e.g., SRM), drivetrain friction also plays a role, but it may have both linear and non-linear aspects so it's hard to say whether it would alter Crr or CdA more. My approach has therefore been to remove the influence of drivetrain friction by adjusting the power data downward prior to performing subsequent calculations.
acoggan •wrote• My only comment would be that the intercept of the relationship is your actual rolling resistance*, not the coefficient of rolling resistance. To derive the latter you need to divide the value of ~4 N by your mass and by g (i.e., 9.81).
Now I knew I should have asked that question before, since I thought the CRR measured in Newtons didn't seem right! Easily fixed - new version attached.
Please refer later post - I have attached an updated version.
If anyone is actually interested, I can also include in the table the calculations to modify the power for each run based on starting and finishing speeds. Just give me a holler!
Last edited by alex on Tue Apr 25, 2006 3:18 am; edited 1 time in total
Posted: Thu Apr 20, 2006 10:03 am Post subject: Cda calcs
First off, thanks a ton for including the speadsheet. As a fellow 'drome tester, that is a huge help.
The only thing that I would add to the testing protocol is to make sure you're doing complete laps. In the linked page, AC mentioned something about using duration vs. distance, but this obviously isn't going to work on a track, since the power requirements are different depending on where you are on the track. Typically, what I do is use 2 400M laps for the slower speeds (up to 20MPH), then use 3 laps averaging for speeds over that. To 'mark' the laps, I simply stop pedalling going into turn 1 and backtrack on CP.
I've gotten some really solid data on 6-7 runs per position on an outdoor track (on a really calm day).
JW •wrote• First off, thanks a ton for including the speadsheet. As a fellow 'drome tester, that is a huge help.
Happy to help and glad someone else finds it useful. Like I said, I will look to "upgrade" it to handle changes in kinetic energy should you be unable to find start/finish points at exactly the same speed. I might also document the "protocol" inside the sheet for ease of reference. Perhaps FGF could post it somewhere for people to access without the need for a topic search.
JW •wrote• The only thing that I would add to the testing protocol is to make sure you're doing complete laps.
Yep, agree and while I suggested starting/finishing on the pursuit line (i.e. middle of the straight), I would add that :
(i) 10-15 metres either side is OK on most tracks (250+m) assuming the straights are level
(ii) totally agree it would be more important to do this on an outdoor velodrome (I was indoors) as even on a calm day there maybe slight differences in the wind, so a full lap would pretty much cancel these out.
When Andy tested road/TT bikes on a flat stretch of road, he did passess in both directions, which cancelled out any slight differences in gradient and wind.
JW •wrote• In the linked page, AC mentioned something about using duration vs. distance, but this obviously isn't going to work on a track, since the power requirements are different depending on where you are on the track.
I should have said a MINIMUM duration rather than distance is recommended but that is more about ensuring a sufficient number of data points for each run since the power meters record data roughly every second. For my Powertap I get 47 data points per minute, so I suggested a minute as a minimum. It doesn't matter if runs are done over different distances (i.e. a different # of whole laps) or if you go for more laps (if you can stand it!), as long as you collect enough data from each run. So as the speed of each run goes up, you might find you have a much longer run than the one before as you want to complete a whole extra lap (as you descibe in your method below).
As to power differences depending on where you are on the track, well speed does vary as your centre of gravity moves up and down upon entering and exiting turns but not sure about power variations other than those caused by rider fluctuations in pedal pressure (and in the case of the Powertap, the frequency of torque sampling by the hub which can make power readings fluctuate a bit depending on cadence, but these will average out over time). Some tracks do have a bit of up'n'down at each end of the straight so that may cause you to vary power in an effort to maintain a constant speed.
JW •wrote• Typically, what I do is use 2 400M laps for the slower speeds (up to 20MPH), then use 3 laps averaging for speeds over that. To 'mark' the laps, I simply stop pedalling going into turn 1 and backtrack on CP.
Sounds good. I just used the interval marker function on the power meter just before the start point and after finishing point but yep I also did something like going up track or a power spike/drop so it was really obvious.
JW •wrote• I've gotten some really solid data on 6-7 runs per position on an outdoor track (on a really calm day).
Cool! Hope you found the best position for you. Out of curiosity, were the differences in calculated CdA of sufficient magnitude to choose an ideal position?
"Cool! Hope you found the best position for you. Out of curiosity, were the differences in calculated CdA of sufficient magnitude to choose an ideal position?"
Well, the only changes I've made have been equipment related, and had a difference of .016, with an r^2 of 99.96, so I was pretty confident in the result.
The position-related differnecnes that I tested gave me a difference of .005 (worse), with a similar goodness of fit, so I just used that testing to confirm that my position was better than the other positions that I tested!
I'm not sure what I'd consider 'sufficient magnitude' to actually change something though--good question!
JW •wrote• I'm not sure what I'd consider 'sufficient magnitude' to actually change something though--good question!
I think >1% to 2% is needed to be significant for just the one set of data in each position, but if you get multiple tests in each position and they show a consistent trend, then that would be good enough for me to run with the change, even if small.
Posted: Tue Apr 25, 2006 3:06 am Post subject: Aero Drag Testing Calculator - Version II
As promised, and requested by some, attached is an upgraded calculator for determining Cooefficients of Aerodynamic Drag (CdA) and Rolling Resistance (CRR).
Changes to previous version include:
- Inclusion of introductory and instruction notes
- Inclusion of new columns to enter starting and finishing speeds and the calculated adjustment to the average power due to any variance in kinetic energy between starting and finishing points of test runs.
- Inclusion of a calculator to determine Air Density based on temperature, barometric pressure and humidity (can also be done via analyticcycling.com).
Any feedback, edits, errors, ideas, etc gratefully received.
He's got a hell of a handicap start but I reckon if you lower the cda of your fingertips by 15% you should just be able to roll Panch0 by the 2000 mark
PeterB •wrote• He's got a hell of a handicap start but I reckon if you lower the cda of your fingertips by 15% you should just be able to roll Panch0 by the 2000 mark
Not a chance! Panch0's CFP (critical finger power) as tested by tracking his MMFP (mean maximal finger power) over various durations and input into the Monod Critial Power model suggests he has far greater finger endurance capacity than me!
I too am a nerd that rides a bike, and can't wait to try some of this testing myself.
Thanks for the spreadsheet!
One comment, I've noticed - at least with the SRM - that when you adjust the zero offset (ZO) run to run, it doesn't store a unique ZO with each interval. It just stores the last ZO offset as a global parameter on the data. The real time display shows the correct value as it is using the current ZO but when downloading data, all prior runs will also use the last ZO value.
You must make note of the ZO for each interval, create a new file from each interval in SRM native SW, and then apply the appropriate ZO offset.
trackie_jay •wrote• One comment, I've noticed - at least with the SRM - that when you adjust the zero offset (ZO) run to run, it doesn't store a unique ZO with each interval. It just stores the last ZO offset as a global parameter on the data. The real time display shows the correct value as it is using the current ZO but when downloading data, all prior runs will also use the last ZO value.
You must make note of the ZO for each interval, create a new file from each interval in SRM native SW, and then apply the appropriate ZO offset.
Jason.
Yeah, I was actually thinking about this - bit of a pain if you get my ZO drift
One other item to consider is the drivetrain losses since SRM (or ergomo) measures at crank (BB), which I was going to validate first before updating in the commentary in the sheet or provide an adjustment column in the sheet for it.
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum You cannot attach files in this forum You cannot download files in this forum
Forum FAQ
Search
Usergroups
Profile
Log in
Private Messages
Album
Pro
Pro
Masters World Champion
Cat 3
Cat 4
