BikeTechReview.com

I got a wild hair today while opening up for tomorrow's 180 degree hairpin turn extravaganza (otherwise know as the "manhattan beach grand prix") and explored a couple new "half pipe" field testing venues...

Here's the work per lap bootstrap method plot I come up with for venue #1:

[attachment=1:3n9bqr0m]<!-- ia1 overlapping_bootstrap_062709_rainbow_1.png<!-- ia1 [/attachment:3n9bqr0m]

and venue #2:

[attachment=0:3n9bqr0m]<!-- ia0 overlapping_bootstrap_062709_rainbow_2.png<!-- ia0 [/attachment:3n9bqr0m]

not a real fan of #2 as it's exposed to some wind (note the poor r^2, but hey - the eyeball method of leveling the VE plot wouldn't really have picked this up, eh?)

I did one lap slow, then two laps fast, then one lap slow. The pavement is chip-seal -> super rough compared to most of the racing done in the industrial park crits here in SoCal. I maintained the same "hoods" position on my road bike for both venues, and same tires/pressure. Way different CxA's and Crr's eh?

I just don't believe the results from venue #2 -> venue #1 holds some promise, as the CxA is in the ballpark of what I have seen for myself in the tunnel for that position and beta=0.

Venue#2 was a complete disaster today (way too much wind for the way I am reducing the data -> maybe if I incorporated ambient wind into things it might turn out better - I'm just not there yet). However, venue#1 was way more sheltered in the hills, and the results seem reasonable.

So, for me, venue #1 can be taken as an additional "field vs roller" data point. FWIW, I have measured 10 of the same tires that Al has measured (not exactly the same, but the same condition, model, brand, pressure) on my 4.5" aluminum rollers. Of the 10, I have not measured a single one to have a higher Crr than what Al has measured. In the aggregate, my measured Crr is about 20% lower than Al's.

Here is what I've observed for two different tires at 120 psi on rough chip-seal roads -> 50% increase in Crr compared to my roller data -> so, if I measure 0.0025 on my rollers, I would measure 0.00375 on rough chip seal roads. If I were using Al's data, this means that a good place to start on super rough, chip seal roads might be 1.3 times what is reported by Al.

Additionally, in the past I’ve tested on “normal” roads and found that the rolling resistance only increases about 30% or so from my roller based testing, which would mean that Al’s Crr data would be underpredicting the on-road Crr for the tires he has listed by roughly 10%.

These data I’ve collected are no real biggie, I reckon, unless some folks would like their field testing to be accurate in addition to being precise.

So, in summary, my recommendation based on my own personal field testing using a combination of SRM/powertap data is the following (if you are not going to vary the speed across laps while using the "chung method" and the various online spreadsheets available on the interwebs) :

1) On rough chip seal roads multiply Al’s reported Crr values by 1.3
2) On regular roads multiply Al’s reported Crr values by 1.1

Nice!!!! I'll use the 10% number from now on in my calculations. The assumption force ? speed × weight is only approximate, anyway, so a fixed 10% adjustment is probably good. Interesting question whether sew-ups or clinchers would be more affected, however. I suppose it depends on what fraction of the power loss occurs at locations relatively isolated from details of the road.

What are your thoughts on the effect of drive train efficiency on these results? Did you shift during the test?

djconnel wrote:


thanks! Yeah, the whole "on-road Crr" deal is a can of worms, I think. I've seen some weird stuff on flatter venues while using the "chung method" when it comes to Crr... however, I reckon folks are going to be doing these types of things regardless of any second order details, so I figgered I'd share my experience. The data I have is not extensive, but it is a place to start, I reckon.



ah, yes, for the plots shown, I took a fixed percentage of my SRM power -> I think it was 2 or 3%. In the past, I've gone through the excercise of running things through a srm-pt script that takes into account gear combos etc... I'm not sure that's any better considering all that's involved and the uncertainties at play. The fixed offset is the caveman shortcut! )

Yes, I did shift during these tests - including on venue #1 flipping it into the big ring after the first slow lap (but leaving it there on the last slow lap).

Again, I'd take these as preliminary recommendations as a place to start and, for me, they will probably evolve over time as I gain more experience/data. The data are what they are!

)

I had a sweet 1/2 day at work today in anticipation of the USA 4th of July holiday - that, combined with some post "hey, let's run up a 380m mountain as fast as I'm able to" soreness, led to me trying to pedal the bike a bit to loosen things up and gather some more data on the whole "on road Crr" deal...

It was pretty much a total waste of time from a big picture "results" perspective, as the wind was 1-3m/s on this half-pipe venue. I did three back to back to back runs of 1 lap slow, two laps fast, one lap slow anyway, though, in the spirit of caveman-ism and Huck Finnery!

run#1 on the hoods with tires at 140 psi. run #2 On the hoods with tires at 100 psi. and then I repeated run #1.

Here's what the work per lap bootstrap method came up with for CxA and Crr results:




Like I said, it was pretty much a throw away day from a "results" perspective, but I did gather some other data. Tire temps using an IR gage! )

air temp was 28c
road temp was 35c

run #1 post tire temp: 33.3c
run #2 post tire temp: 34.0c
run #3 post tire temp: 33.4c


so, if we take that temp data at face value, it seems as if the 140 psi case had lower Crr. Anyway, the data are what they are, so to speak, and wind doesn't help, eh?

I'll try to drag my butt out of bed tomorrow morning and give it another go!

-k

Good day conditions-wise this morning -> a bit foggy, chillier, and low/flat light, but little to no wind.

Again, three runs - hoods 140 psi, hoods 120 psi, hoods 140 psi (slow lap, 2 fast laps, slow lap) on the road bike -> same as yesterday:

WLB method results:



fixing the CxA:



and for kicks all 12 laps lumped together and WLB'd:

CxA=0.333 +/-0.005
Crr=0.00600 +/- 0.00021

This was an interesting plot to look at, in that it made clear that the wonky-ness/variability was a result of the slow laps -> the fast laps looks very consistent from run to run and tire pressure to tire pressure.

wind: 0-1 m/s

Air Temp:16c/18c

Road Temp:22/23.6c

Tire temp (front/rear) after:
run#1: 19.0/20.1
run#2: 19.2/20.6
run#3: 19.5/20.8

(kinda cool that you can measure a f/r temp differential -> makes sense since the rear wheel carries more load, though, huh!)


FWIW, the lumped together data is consistent with the recommendations above for fixing Crr based on Al's chart (i.e, for chip-seal/rough roads like this one, multiply Al's #'s by 30%) -> though, the CxA of the lumped together data does seem a bit low for me.

So, all in all, another perplexing day of field testing using this methodology. It is what it is, though!

Similar day conditions-wise this morning -> a bit foggier than yesterday.

Again, three runs - hoods 140 psi, hoods 120 psi, hoods 140 psi - today, though, I mixed up the speed progression. I went slow/fast/slow/fast today. Also, on the last fast lap of the day, I had a car pull out in front of me, so I have taken the liberty of throwing out that lap, which kind of mucked things up a bit (that was the 140psi repeat run)

WLB method results:



fixing the CxA:



and for kicks all 11 laps lumped together and WLB'd:

CxA=0.336 +/-0.003
Crr=0.00602 +/- 0.00018

The r^2 on this days methodology was much worse using th lumped together approach -> .95 or so, but there didn't appear to be any bias in the data and the bootstrapping of things doesn't seem to suggest any worse estimate of the variability -> not sure what to make of this at this point. Though, pretty good day to day repeat when all things are lumped together, huh?! Definitely within the std deviation estimates - so that's a good thing for me to see.

wind: 0-1 m/s

Air Temp:16.2c/16.2c

Road Temp:21.3/23.1c

Tire temp (front/rear) after:
run#1: 17.1/18.6
run#2: 17.9/19.6
run#3: 18.7/19.9


So, not sure what to make of the 140 vs 120 psi deal on rough road thus far -> individually looking at the runs, and using the WLB method it's a dog's breakfast. Fixing CxA and looking at Crr over the last two days we have 120 psi = 0.0061 and 140 psi = 0.0058

Happy 4th of July to those in the USA!!!

Happy "first day of the tour" to everyone else! )

kraig wrote:


Well, I was back at it again this morning...and uncovered a couple of things.

Based on the preceding testing in this thread, it should be clear that 4 laps on this course is not sufficient to get things to repeat. The data would suggest that 12 laps across days can get things to repeat. Well, I'm not really into doing 12 laps (that'd be 20-25 minutes of riding or so...) for each configuration of interest, my patience just isn't at that level. )

so, i tried two trials of 6 laps each today - same configuration, same approach -> slow/fast/slow/fast/slow/fast. And I did this sequence twice without changing anything. The difference this time was that I had both the SRM Pro and the wired PT Pro installed. The first thing I noticed right off the bat is that the SRM is basically reading the same as the PT. I have had this SRM factory calibrated twice. Both times it was within a couple % - but I have my doubts about the cal prior to the latest one I had done last fall..as prior to sending it in, I told the SRM crew (this was a couple years ago) that the PM was "reading a bit low compared to a PT I owned"... oh well, it is what it is, and a couple % isn't anything that is going to change how I globally approach things, so to speak.

This additional PT info might have contributed to why things seemed low to me (as noted above) for the data presented in this thread -> the SRM might have actually been reading a bit low with the offset I was plugging in after the fact.

Anyway, the other thing I observed was that the first run for the SRM was really noisy compared to the PT -> an r^2 of .98 or something compared to a 0.99 for the PT. And the values obtained on that first run were kind of whacky -> higher than previously measured CxA and lower than previously measured Crr. The second runs for both devices seemed to be within a couple std. deviations of each other using the WLB method.

Note to self: when using an SRM, ride around for more than 2 minutes and amp things up a bit prior to setting the zero offset! I really don't think it's a zero offset deal, cuz it changed less than 3 hz before/after the first run -> I checked. But still, the data looked really poor -> so I'll do a "fast" lap prior to getting things going next time around.

Armed with that bit of new information I stumbled on this morning, I re-ran the 140 psi lumped together data without reducing the SRM data by 3% and using the work per lap bootstrap method:



that seems to be moving in the right direction from a CxA perspective!

So, this mornings testing netted the following based on my PT pro hub with tire pressure at 120 psi:



pt pro lumped together both runs:




and with the raw file from the SRM pro:



srm pro lumped together both runs:




fixing CxA at 0.354 for both lumped cases:



Here were the ambient conditions:

wind: 0-1 m/s

Air Temp:19.1c/18.4c

Road Temp:21.8 before first run/22.5c after first run/22.5c after last run

Tire temp (front/rear) after:
run#1: 19.4/20.4
run#2: 19.6/20.9

Here is what the road surface looks like, with my hand as a length/scale reference:

[attachment=0:3dntj3ek]<!-- ia0 rainbow_surface.jpg<!-- ia0 [/attachment:3dntj3ek]

So, geez - I guess that is the really long way of saying that the simultaneously acquired SRM/PT data seems to suggest that the on-road Crr for these tires at 120 psi on this road with me pedaling, is somewhere in the 0.00605 +/- 0.000X range...which is still consistent with the initial recommendation for adjustment on Al's Crr data.
A couple of other takeaways for me:
1) tires seem to reach a temperature that is somewhere between the road surface temp (conductive heat transfer), and the ambient air temp (convective heat transfer)
2) if using an SRM Pro - pedal around more than 3 minutes and "get on it" for a bit to get things to settle out
3) pray for no wind (if you are into that kind of thing - otherwise, do what the golf guys do to see if there is any wind at play)
4) you could probably convince yourself that you are more precise than you actually are using this method of data reduction.
5) I really do have to much time on my hands! ;-) )

If anyone is still reading this thread and would like the raw PT/SRM files from this session drop me an email and I'll send them to you, with the only condition being that you don't send the files to anyone else. I trust y'all on this one.
-k

Hey Kraig --
Just wanted to let you know that at least one person is following this thread closely. Neat stuff! I may ask for those files a little later. At the moment, I'm re-writing some of my tools, and they're all a mess.

Some quick questions/comments:

- I've lost track a bit. So the on-road Crr appears to be ~0.006, but what was the roller Crr for those tires? I.e., what kind of correction factor are you finding for that rough road?

- You mentioned that the CdA you'd been finding seemed low, though possibly from SRM mis-calibration. Another possibility comes to mind. What are the turns like on this course? I've been trying to play with how turns affect the estimates. On the one hand, leaning means that the path traveled by the COD is shortened, which would lower the CdA estimate. On the other hand, turns presumably raise Crr by some unknown amount, which would raise the CdA estimate. I don't have a good handle on the tradeoff yet.

- It's interesting you mentioned fritzy data from the SRM. I don't have an SRM, but just this week I did a run where I got some funny results. When I checked the raw PT records, it turned out that the readings had frozen for 3 records. It also looked like the 2-3 records before the freeze were suspect. Dropouts are a bummer, as they remove a whole lap of data. But when I did remove them, results were more in line with what I would expect.

-- Adam

ahaile wrote:


I measured this particular set of tires to have a Crr in the 0.004-ish range on my rollers, IIRC...things still seem to be sorting out pretty consistently with my thoughts in the first post:

kraig wrote:


I do have a course around here that has an even rougher road surface than the one pictured earlier, but man, I don't think it is very typical...nor do I have a warm fuzzy about the data I generated on it previously after playing around a bit more seriously on this course.


out and back ~880m loop - slow speed turnarounds are at like 10kph and the higher speed turnarounds are at like 16kph, IIRC.



I haven't been looking for that - so, I'll keep an eye out!

I might be mistaken as I am a bit confused at times,
but ignoring run2's for a minute it looks a bit like the run3s were consisently lower CRR than Run1s?
Maybe warming up the tyres up might reduce CRR a bit??

With a view to to short prologues/track pursuits does this mean warming up on the tyres could potentially be worthwhile?
I have a routine of warming up on training wheels and just putting the race wheels in right before the start. Starting to think I should warm up on race wheels.
I can't really understand the info presented there enough to draw this conclusion for sure.

I have a feeling I may have missed a previous post /thread about field testing and the first few runs. ???

DMC.

DMC wrote:


I definitely see a temperature effect on Crr with the roller testing. As ambiant temperature increases the Crr decreases - approximately 1% per Degree C. There are really 2 effects - ambiant temperature and the temperature increase of the tire as it rolls along at speed with load on it. Also on rollers the tires heat up more than on a flat road. Actually the lower the Crr the less the tire heats up. Bottom line IMHO is that warming up on the trainer with race wheels can't hurt although if you have to go through a UCI bike check right before starting the tires will probably be down to ambiant temperature again. I've found that it takes ~ 3 minutes for the tires to lose most of the heat. At least you will not have the risk of goofing something up when putting the race wheels on. The trainer won't hurt your race wheels and if you can time it right you will have a slight advantage of a "hot" rear wheel for a few minutes at the start of the race.

DMC wrote:

I did some Chung field testing earlier this year,
<!-- l <a class="postlink-local" href="biketechreview.com/forum/viewtopic.php?f...p;p=24546#p24546<!-- l

and found that the Crr dropped by ~ one third as the tyres(Vred Fortezza Tricomp,butyl tubes..IIRC)warmed up over ~20-30 mins.
It was cold,~2 deg C,and the wheels were on the car roof on the way to the venue.....

I saw something similar on Vitt Open Corsas/butyl tubes in much better weather(and wheels in car en route)recently,but I didn't quantify it..it was a lot less IIRC

AFM wrote:


FWIW, the data I've gathered (n~20) on my rollers with a couple different tires suggests the slope (Crr/ambient temp in C) is pretty similar at the ends of the spectrum -> i.e - veloflex vs vittoria zaffiro. I'm coming up with about 0.4%/degree C slope, but the data has a lot of scatter.

DMC wrote:


Well, this thread is basically just some of my notes with the hope that the transparency in my data collection/process development might spur some discussion...as a result, it's organization is less than stellar!

run1 vs run3: I did some further data reduction at a later date (digging on the tire pressure deal) - and while fixing CxA for a given day I can make four comparisons of first vs last runs (which were the same configuration) 2 had higher Crr on first run and 2 had higher Crr on the last run. So, 50/50 deal when reduced like that.

My personal opinion is that I think the data is a bit too noisy to draw many hard conclusions about anything at this point - having said that, though - it seems like the data is wanting to tell me that there is no real penalty for running 140 psi on the road pictured above vs 120 psi or 100 psi - and perhaps a slight benefit as one would predict based on roller data.

sjdocman wrote:


That would be an interesting one to have tire/road/ambient temperature data for! Maybe it will get that cold here in SoCal over the winter - errr... or maybe I should be testing Crr right now in the garage where the temps in there have been getting to 45+c and up to 40c outside.

ooh, that's getting a little warm! d'oh!

A link I got from the Wattage list:
http://www.slowtwitch.com/Tech/What_s_i ... _1034.html

Higher pressures, rougher roads. Data are a bit questionable, but one does appear to see a breakpoint:



(Aaargh: png for plots, not jpg.... okay, I'm over it...)

Dan

Thanks for the link, Dan.

Considering the source of your link, I think I'll take the high road on this one.

I trust the greater "bike racing nerd" community ) (hey, that's me!) to ask the right questions of the intrepretation and presentation of the data presented in the article Dan linked to. As for me, I'll let my data and methods speak for themselves, y'know, cuz it's pretty transparent! y'know it's a work in progress and all... It reminds me of the great quote I saw at OHSU: "the obstacle of discovery is the illusion of knowledge" deal and all...

If others have a preferred/superior method of executing these types of tests, I'm all ears and willing to repeat these protocols in the best interests of understanding the effects at play here. Please share these superior and preferred protocols in this thread...I have a good track record of testing suggestions (the Lim/Haile/Chung method come to mind) to the best of my ability, so, please, share your thoughts, folks!

As this thread clearly indicates, I am currently unable to reproduce the results you have linked to, Dan, using the chung method/haile method/WLB method using multiple venues (i.e road surfaces), multiple protocols, multiple tires, and multiple pressures...and the CV's I seem to be able to achieve for Crr across days/within days/across configs/whatever...while fixing CxA appear to be in the 1-8% range (n=20+), YMMV, of course!

I would encourage folks to add to this thread with their own field testing experience! )

(edit) folks that have access to the BTR insiders forum can see a plot which experimentally "tests" (well, really, "tested", cuz that data was posted quite some time ago) what is implied by the plot you have posted, Dan. i.e - that on "real" roads (or, in the insider case, "super duper rough chip seal"), "low" psi is better than 140 psi.

Is the difference between roller tests and real world tests consistent? Or are there tires that have better/worse relative performance in the real world compared to the roller test?

Anachronism wrote:


good question! I have pretty good experience with a variety of clinchers that Al has tested and I've personally tested these brands/models both on the road and on my own rollers at a variety of inflation pressures...and my takeaway is that there more than likely wouldn't be a reversal of Al's roller data out on the road for you with tires that have a large difference in Crr on his chart.

If you are trying to compare tires in Al's table that are close in Crr...say, within, oh, off the top of my head, 20% of each other I think reversals (lower Crr tire on Al's table is actually higher on the road for you) could be possible depending on the particular tire you have in your hands...I've seen significant tire to tire variation in Crr within a particular tire brand/model. Tire mass, in this case, is a good proxy for within brand/model Crr.

If you do your own roller testing with tires you currently own, well, then I think roller testing is good for choosing tires for use on the road...Even when the rollers are bumpy:

biketechreview.com/forum/1-general-discu...-in-human-body#26444


...I've been able to draw similar conclusions with respect to tire and inflation pressure both on the rollers and on the road.

my experience with tubulars is limited, but since the gluing process is another variable when it comes to Al's chart and an individual's use on the road, things become a bit more noisy and a bit more unpredictable, I reckon.

Also, don't forget that tire aerodynamics can play a role as well in terms of which tire is faster for you out on the road...and man, field testing these kinds of things (tire aero and tire Crr) is tricky business -> my recommendation would be to be very skeptical of any field testing you see reported on the internets (including here!)...be especially skeptical of field tests with limited runs that do not span multiple days and lots of repeats. It's a tough bit of work teasing out these differences that are small-ish.