Login required to started new threads

Login required to post replies

Prev Next
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [ In reply to ]
Quote | Reply
I did some testing early this morning ,as the wind was low (UK).
Was testing 40cm bars vs 44 and surprised to find the 44s ' faster'
Looking at the weather data,the temperature went up by 2--3degC,and the solar radiation doubled (from ~50w/m2 to ~100) during the time I was on the road.
I set off from the house, so tyres were 'warm' then presumably cooled outside, then I switched the bars (inside the house),and went back out.
I didn't do ABA ,so am unsure if this is a real difference,and wonder if anyone has experience of w/m2 affecting crr?
( I compensated for ambient temp change)
Interestingly the garmin recorded a swing of ~9 deg C over the time I was out,compared to 2 deg from the weather station.
Thoughts welcome!
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [kevinkeegan] [ In reply to ]
Quote | Reply
That's a big swing in temperature. Depending on what the starting conditions were, a 9 deg C. increase in temp can mean 2-3% decrease in air density, and also a decrease in Crr. The smaller the change in position or equipment, the more careful you have to be in tracking changes in conditions. Was the difference between the 40 and the 44 on that order of size, or was it larger than that?

[edited to add:] This method is sensitive enough to detect changes in the weight of air. Fortunately, if you keep track of the temperature and barometric pressure (humidity makes a small amount of difference, too) you can calculate the weight of the air and account for it. That's pretty cool.
Last edited by: RChung: May 12, 19 20:21
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [RChung] [ In reply to ]
Quote | Reply
The swing between setups I get as 3.8% (road bike,drops,from 40cm to 44 cm bars,44 being faster
Using the weather station data (about a mile away)- the air density reduced by a max of 0.85%

Looking back at the data:
The weather station recorded a +3 degree (celsius) temp change outside.
(It also showed zero wind at the beginning,going up to between 1 & 2 kmh by the end of testing)
The Garmin showed a 6 degree drop during '40' testing,but no drop on '44' testing ( the sun had come out a bit more)

I am wondering that if the Garmin showed this,were the tyres experiencing something similar ?

I do have an infrared thermometer (at work),that I will try out if I get a chance
Last edited by: kevinkeegan: May 12, 19 22:16
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [kevinkeegan] [ In reply to ]
Quote | Reply
kevinkeegan wrote:
I do have an infrared thermometer (at work),that I will try out if I get a chance
Better to use something that measures air temp (and isn't in direct sunlight), not surface temp (highly influenced by emissivity and sun exposure).
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [MattyK] [ In reply to ]
Quote | Reply
MattyK wrote:
kevinkeegan wrote:
I do have an infrared thermometer (at work),that I will try out if I get a chance

Better to use something that measures air temp (and isn't in direct sunlight), not surface temp (highly influenced by emissivity and sun exposure).
Right, air temp is really what to measure.

Did you [kevinkeegan] zero your torque between the runs? I do that when there are wide swings in temperature -- even though some PMs say they compensate for (or are immune to) temperature variation, it's good practice to do so. And, of course, some PMs are better for this sort of thing than others.
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [RChung] [ In reply to ]
Quote | Reply
 

Did you [kevinkeegan] zero your torque between the runs? I do that when there are wide swings in temperature -- even though some PMs say they compensate for (or are immune to) temperature variation, it's good practice to do so. And, of course, some PMs are better for this sort of thing than others.[/quote]
I didn't zero torque.Am using powertap wheel,and did plenty of coasting.
The air temperature didn't actually vary much(from 6 to 9 degC)
Will zero torque the next time
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [kevinkeegan] [ In reply to ]
Quote | Reply
kevinkeegan wrote:
...
I didn't do ABA ,so am unsure if this is a real difference
...

I hope this doesn't come off as too negative. But if you didn't do ABBA and convince yourself you can reproduce your results it is pretty hopeless for any of us to try to draw conclusions. I wasted quite a bit of time/runs back when I was doing this kind of testing working with poor protocols and thus having junk results (it has been a few years since I did any aero testing). I wish I had been urged more strongly to do ABBA testing as the very first thing and not move on to anything else until I was sure I had that nailed.
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [jbank] [ In reply to ]
Quote | Reply
jbank wrote:
I wish I had been urged more strongly to do ABBA testing as the very first thing and not move on to anything else until I was sure I had that nailed.

If possible do ABABAB etc...

Seems to me that a good first step is to see what SD or CV you can acheive when you are not changing anything.
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [rruff] [ In reply to ]
Quote | Reply
I don't think it's always necessary to do ABABAB... testing. Sometimes it's clear after an AB comparison that something is wrong and you should either postpone to another day|venue|PM|speed sensor| or else measure your rho more carefully or else convert the test session to practice runs to refine your protocols.

Even a bad run can provide you with good information -- just sometimes what you learn is what not to do the next time.
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [AndyF] [ In reply to ]
Quote | Reply
Hello,

I'm planning to do my first aero testing in the following days. Given the density of traffic where I live, I have decided (unless you have any reasons against it) my best bet is to do my testing in an outdoor 185m concrete velodrome and follow the black line. From my testing, I'd like to be able to distinguish CdA and Crr because ideally I'd like to be able to reasonably compare CdA values from one testing session to the next one (that is, one position should give a close enough CdA value testing in separate days). I haven't found a clear enough protocol, but I think the consensus is:

  • Record the whole test in one file, and every run as a lap within the file.
  • The runs should be long enough, I think 10 laps of the track is okay (1.85 km).

    • Should they be longer? Is it better to make them shorter?
  • Each of the runs should be building speed, let's say from 15 km/h to 45 km/h (Shen method)

    • Do this have to be precise (ie. I increase 3 km/h every 3 laps) or is it okay if I just build the speed but without being particularly precise about the specifics?
    • Should the steps of the speed build be long enough? For example, keep the same speed every 3 laps or build each lap
    • Do I have to come back down again to 15 km/h ?
  • Do alternate runs with the set ups you want to test. For example, A/B/C/C/B/A
  • Analyze every run (a lap in the file) as an individual ride in golden cheetah and play with CdA and Crr until I get a flat virtual elevation, but:

    • How flat is flat enough?
  • The velodrome is around a soccer/rugby pitch, I highly doubt I'm going to find any other cyclist but, do I have to worry about people playing affecting my results?
  • My area is a bit windy, what is considered still enough? Below 10 km/h wind? 15 km/h?



Thanks a lot in advance, really looking forward to hear from you.
Last edited by: dense: Aug 27, 19 1:48
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [dense] [ In reply to ]
Quote | Reply
185m is okay, as long as you can follow the black line; sometimes on very short tracks the turns are so tight that you may go wide on the transitions -- just do the best you can.

You don't have to build your speed by equal increments on a set schedule: the raison d'etre for this method is cuz I couldn't follow a constant speed or power method, so I needed to come up with something that was more practical to follow.

You don't have to come back down in speed -- but you can if you want. That would give you another way to check the CdA/Crr estimates.

The VE profile is, above all, a diagnostic of fit. Although you're actually riding on a velodrome, the virtual amplitude of the laps should be pretty small but even and easily discernible. If not, then it was too windy.
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [RChung] [ In reply to ]
Quote | Reply
Thanks for the clarifications, hopefully it goes well!
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [dense] [ In reply to ]
Quote | Reply
For a 185m long track I would estimate the curve radius to be around 17m. With 45km/h on the black line you will have a lean angle of around 45° thus you center of gravity will go down and up about 50cm twice a round. When you get the speed from a wheel sensor Virtual Elevation a la Chung will overestimate your elevation variation, i.e. you will clearly see the half laps in the virtual elevation profile.
VE is a quit good and robust method to solve the underlying equation of motion. It will underestimate Cda a little on a short track with tight turns.
To get both Cda and Crr from a single fit of some laps in GC Virtual Elevation is very optimistic. As I know VA in GC doesn‘t account for higher rolling resistance in the curves and some other tricky problems arising from cycling in turns are not considered. Above all, the dependance of power from Cda is higher than from Crr at reasonable speeds making it difficult to get a reliable fit of a model. But may be I am just too stupid to do it.
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [BergHugi] [ In reply to ]
Quote | Reply
Might be a bit too optimistic, I have no idea. Worst case scenario I hope to be able to make A/B comparisons. Anyway, this week I'm only going there to try see how everything works. I'll let you know.
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [BergHugi] [ In reply to ]
Quote | Reply
BergHugi wrote:
To get both Cda and Crr from a single fit of some laps in GC Virtual Elevation is very optimistic. As I know VA in GC doesn‘t account for higher rolling resistance in the curves and some other tricky problems arising from cycling in turns are not considered. Above all, the dependance of power from Cda is higher than from Crr at reasonable speeds making it difficult to get a reliable fit of a model.

To separate the instantaneous CdA from the instantaneous Crr is hard, so I usually just think about the CdA and Crr over lap-length intervals, averaging over both the straights and the turns. In addition, there are small non-zero yaw effects on a velodrome (since you're going 'round and 'round in one direction) that will bias the estimates. But these are relatively small effects and, if you need to know, there are ways to get around them (though they're a pain in the butt to do so you really have to want to go to that extra effort). The different dependence of aero drag and rolling drag with speed is the reason why you want a wide range of speeds -- the thing you have to worry about is dipping into a realm where the Re number is changing.

I agree that GC's Aerolab needs a refresher. If I could code in a credible way I'd've done it myself.
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [BergHugi] [ In reply to ]
Quote | Reply
BergHugi wrote:
Above all, the dependance of power from Cda is higher than from Crr at reasonable speeds making it difficult to get a reliable fit of a model. But may be I am just too stupid to do it.

You're right on. Making Crr a variable while trying to determine CdA differences doesn't work well. It's far better to make an educated guess on Crr and insure that it's the same for configuration A as B.

Also, I did several repeatable tests this summer that indicate that Crr is highly speed dependent (increasing with speed). This was by testing slow and fast runs on ~1% grade with a strong tailwind (~15km/hr), to remove the affect of aero drag as much as possible. It could be that something else was wonky, but increasing Crr with speed was the only thing I found that fit the data. I was measuring airspeed, BTW.
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [dense] [ In reply to ]
Quote | Reply
dense wrote:
Might be a bit too optimistic, I have no idea. Worst case scenario I hope to be able to make A/B comparisons. Anyway, this week I'm only going there to try see how everything works. I'll let you know.

If swapping between A and B is easy, then do it several times.

Don't get discouraged if your first time doesn't work out. Good controlled testing isn't easy, it takes some know-how and practice.
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [rruff] [ In reply to ]
Quote | Reply
rruff wrote:
Also, I did several repeatable tests this summer that indicate that Crr is highly speed dependent (increasing with speed). This was by testing slow and fast runs on ~1% grade with a strong tailwind (~15km/hr), to remove the affect of aero drag as much as possible. It could be that something else was wonky, but increasing Crr with speed was the only thing I found that fit the data. I was measuring airspeed, BTW.

Depends on the surface, I suspect. I've done tests on out-and-back courses (both "half-pipe" and relatively flat). On the flat courses, I vary the speed widely over the same spots on the course in the different directions; on half-pipes, I go slower on the climbs than on the descents over the same spots on the course. If Crr were highly speed dependent on those courses, the mirrored VE profiles wouldn't match. In either case, the VE profile is a diagnostic: different types of mismatches in the VE reflect different types of errors.
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [rruff] [ In reply to ]
Quote | Reply
rruff wrote:
Don't get discouraged if your first time doesn't work out. Good controlled testing isn't easy, it takes some know-how and practice.

Yeah. My kid is trying to learn to play the guitar and you can't expect to sound like Clapton the first time out. VE is easier than learning to play the guitar but, like almost all things worth doing, practice helps.
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [RChung] [ In reply to ]
Quote | Reply
RChung wrote:
In either case, the VE profile is a diagnostic: different types of mismatches in the VE reflect different types of errors.

Would be really cool to have software that would identify and diagnose those anomalies...

The road was pretty decent but lumpy asphalt (not chipseal) with some gravel and tar patches. I got Crr ~.0030 at 20km/hr and ~.0045 at 40km/hr. Seems like a speed dependence like that would be hard to separate from CdA without conditions like I had (uphill, tailwind), since the drag vs speed relationship is so similar.
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [rruff] [ In reply to ]
Quote | Reply
rruff wrote:

Would be really cool to have software that would identify and diagnose those anomalies...


Well, if you do loops or out-and-backs over that road at varying speeds you can try plotting the segment CdA/Crr against position on the course to see whether variation in either is specific to location or to speed. That's what I do. But I use my own routines to do that.
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [RChung] [ In reply to ]
Quote | Reply
Had a go at testing after finally finding what I thought was a decent spot. Since I'm new to this, was wondering if anyone had any critiques/insights. I didn't have time to do A/B etc. So I just repeated laps of a half pipe to get familiar with the protocol/analysis.

Does the half pipe look too short? It felt like I was either descending pretty rapidly or slowing rapidly. Didn't feel like a time trial where I'm on close to flat ground for extended periods. But maybe that is ok? I was thinking maybe my posture on the bike may be a bit different.

I thought my turn-arounds were pretty poorly executed while riding, but now looking at the plots it doesn't seem so bad (?). The road was decent, but there were a couple spots where there were seams or rough spots. I guess for that you would assume the crr is just a bit worse?

Thanks!


Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [brbbiking] [ In reply to ]
Quote | Reply
600 meters one-way isn't too short, but I'm a little confused by the scale on the y-axis: is that also meters?
Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [RChung] [ In reply to ]
Quote | Reply
Oops, apparently I had imperial units selected in my profile so it was somehow mixing units. Anyway, changed to metric in my profile and rebooted GC, and here's the new plot:


Quote Reply
Re: Platypus Thread: Aero Virtual Elevation Testing Protocol [brbbiking] [ In reply to ]
Quote | Reply
Yesterday I did some runs in an outdoor velodrome. Do the larger undulations look like wind? From other posts, I understand there will be some effect from the cornering, but I'm thinking those are the smaller ones? Is it a reasonable run for a velodrome, or does it look like I was encountering too much wind?

Thanks for any insights.

Edit: New thought: The larger undulations may be straying further from the black line and hitting more banking on the shorter sides of the track? They also tend to increase in size as I was going faster (later in a run), which would make sense.


Last edited by: brbbiking: Sep 26, 19 7:31
Quote Reply

Prev Next