CVT interest

In article <[email hidden]>,

Robert Chung said:

http://anonymous.coward.free.fr/rbr/bushnell.png

Robert, thanks for putting up the plots.

I see a curious trend on the avg_watts vs mph plot of the high climbing
density rides to show an unexpected sublinear relationship of power vs
speed. Probably don't have enough data to draw a conclusion. I also
see that I have a gap in the climbing density between 30 ft/mi and 60
ft/mi. I've included more data below that might illuminate the trends
better.

The following data are with the same bike and rider and a slightly
different fairing. The difference between the fairing used for the
first 10 rides (Fairing A) and (Fairing B) used for next six rides
probably aren't enough to confound the results. The only substantial
difference between these two fairings is that Fairing B lacks the front
wheel half-cover of Fairing A and is the one shown in the photo that
started this discussion.

http://tinyurl.com/etw5e

I've labeled these A through G. Maybe ride 10 can be relabeled as ride
0 to keep everything to single characters on the plots.

Ride A (2005/06/19)
Distance: 94.0 miles
Climbing: 6010 feet
Climbing density: 64 ft/mi
Total time: 7:59:24
Riding time: 6:12:41
Average speed: 15.1 mph
Maximum speed: 51.8 mph
Average power: 153 watts
Energy: 3438 kJoules

Ride B (2005/06/25)
Distance: 110.0 miles
Climbing: 4550 feet
Climbing density: 41 ft/mi
Total time: 9:41:20
Riding time: 6:57:56
Average speed: 15.8 mph
Maximum speed: 48.5 mph
Average power: 131 watts
Energy: 3290 kJoules

Ride C (2005/07/02)
Distance: 111.1 miles
Climbing: 7190 feet
Climbing density: 65 ft/mi
Total time: 10:47:11
Riding time: 7:48:52
Average speed: 14.1 mph
Maximum speed: 54.3 mph
Average power: 135 watts
Energy: 3798 kJoules

Ride D (2005/07/16)
Distance: 114.4 miles
Climbing: 7700 feet
Climbing density: 67 ft/mi
Total time: 9:59:40
Riding time: 7:49:03
Average speed: 14.6 mph
Maximum speed: 56.0 mph
Average power: 153 watts
Energy: 4295 kJoules

Ride E (2005/07/23)
Distance: 105.4 miles
Climbing: 2150 feet
Climbing density: 20 ft/mi
Total time: 11:38:23
Riding time: 6:18:39
Average speed: 16.5 mph
Maximum speed: 40.6 mph
Average power: 127 watts
Energy: 2884 kJoules

Ride F (2005/07/30)
Distance: 95.3 miles
Climbing: 5890 feet
Climbing density: 62 ft/mi
Total time: 7:03:41
Riding time: 6:00:33
Average speed: 15.8 mph
Maximum speed: 50.6 mph
Average power: 158 watts
Energy: 3431 kJoules

Ride G (2005/08/13)
Distance: 118.2 miles
Climbing: 6000 feet
Climbing density: 51 ft/mi
Total time: 10:37:52
Riding time: 7:14:20
Average speed: 16.3 mph
Maximum speed: 52.7 mph
Average power: 146 watts
Energy: 3808 kJoules

--
Bill Bushnell
http://pobox.com/~bushnell/

Bill Bushnell said:

I see a curious trend on the avg_watts vs mph plot of the high climbing
density rides to show an unexpected sublinear relationship of power vs
speed. Probably don't have enough data to draw a conclusion. I also
see that I have a gap in the climbing density between 30 ft/mi and 60
ft/mi. I've included more data below that might illuminate the trends
better.

I think what you were seeing was confounded because differences in the
amount of climbing weren't made clearly enough. Grouped by climbing.
http://anonymous.coward.free.fr/rbr/bushnell.png

Michael Press said:

In article <[email hidden]>,
dvt <[email hidden]> wrote:

Quoted message said:

Quoted message said:

Good chart, Robert. I was thinking of making my own chart, but I think
yours answers my question. There is one outlier in the data with little
climbing, low power output, and low average speed. I think that data is
all from Ride 4. Throw out that outlier and I think you'll see some
strong trends.

Do not throw away information. The outliers can tell you
something. Researching them is difficult, but worth the
time.

True outliers tend to be measuring errors, and can be validly ignored. Of
course, on the other hand, throwing out half your data as unrepresentative
outliers is called fraud.

Jasper

In rec.bicycles.tech Robert Chung said:

I think what you were seeing was confounded because differences in the
amount of climbing weren't made clearly enough. Grouped by climbing.
http://anonymous.coward.free.fr/rbr/bushnell.png

Robert,

These are excellent plots.

watts vs. mph:

For the black data (med-high climbing density) and to a lesser extent the red data
(high climbing density) the relationship appears to be slightly concave downward,
but I think this is just random error. I'd expect any curve to be concave upward. I
think it's safe for this range of data to use a linear fit for each of these four sets
of data, where each of the curves does not cross any other. At some point off the top
of the chart, I would expect the curves to show a concave upward trend.

At a given speed I suspect that the slopes of the curves are parallel for the black,
blue, and green data. The slope of the red data appears to be greater than that of the
black data. This might be because on the very hilly rides I scrub away more energy
into my brakes on the steeper and more technical downhills.

The blue data (moderate climbing density) are still too few, and I suspect with more
data we'd see a linear fit that lies entirely between the black and green data.

What I find most interesting is that for the green data (low climbing density), the
relationship of power to speed up to at least 21 mph is essentially linear. I'll need
to gather more data at higher speeds to spot more clearly the concave upward trend
due to wind resistance.

Other observations:

watts vs. climbing density: As climbing density increases my average power range
narrows. Not sure why this would be. Physiological or data bias?

watts vs. miles: Average power decreases with distance.

mph vs. climbing_density: Speed drops (and range narrows) with increased climbing
density.

mph vs miles: Speed drops with increased mileage.

climbing_density vs. miles: Seems that the most climbing dense rides are also my
longest. That's not always true, but I didn't have any data for short, climbing dense
rides using the same "banana bike" configuration.

Do you see anything else significant in these plots? Also, do you have any similar
data for other bikes/cyclists for comparison?

--
Bill Bushnell
http://pobox.com/~bushnell/

Bill Bushnell said:

I'd expect any curve to be concave upward.

Well, the cubic relationship between speed and power applies
instantaneously and what we're plotting here is averages over the entire
ride. Because we're looking at averages, the range of (average) speed is
quite a bit more limited than what you'd see if you looked within an
individual ride.

Quoted message said:

Also, do you have
any similar data for other bikes/cyclists for comparison?

Mostly what I have is race data, so nothing directly comparable.

Have you used your PT to estimate CdA for your bike? I haven't seen many
figures for faired bikes and using a PM with a faired bike simplifies the
positioning issue.

In article <[email hidden]>,

Jasper Janssen said:

Michael Press said:

In article <[email hidden]>,
dvt <[email hidden]> wrote:

Quoted message said:

Quoted message said:

Good chart, Robert. I was thinking of making my own chart, but I think
yours answers my question. There is one outlier in the data with little
climbing, low power output, and low average speed. I think that data is
all from Ride 4. Throw out that outlier and I think you'll see some
strong trends.

Do not throw away information. The outliers can tell you
something. Researching them is difficult, but worth the
time.

True outliers tend to be measuring errors, and can be validly ignored. Of
course, on the other hand, throwing out half your data as unrepresentative
outliers is called fraud.

They are not measuring errors until you have evidence of
measuring errors. Throwing away data is not bad science,
it is no science at all. If the best you can do is shrug
your shoulders and say `beats me', so be it. But outliers
must not be ignored while presenting data to support a
hypothesis.

Outliers are not good or bad. Advances in science occur
when the data does _not_ support the hypothesis; e.g.
non-conformity to theory of specific heat measurements at
low temperature.

--
Michael Press

Michael Press said:

Jasper Janssen said:

True outliers tend to be measuring errors, and can be validly ignored. Of
course, on the other hand, throwing out half your data as unrepresentative
outliers is called fraud.

They are not measuring errors until you have evidence of
measuring errors. Throwing away data is not bad science,
it is no science at all. If the best you can do is shrug
your shoulders and say `beats me', so be it. But outliers
must not be ignored while presenting data to support a
hypothesis.

Outliers are not good or bad. Advances in science occur
when the data does _not_ support the hypothesis; e.g.
non-conformity to theory of specific heat measurements at
low temperature.

Sadly, nobody else seems to know this[1] these days.

Last term I had a student who blamed "outliers" as the source of all
problems in all statistical calculations - as in, "when we get rid of
all this data that troubles us, then we'll get the conclusions we want!"

I'm afraid the student didn't make it up on their own, but had help from
prior instructors.

[1] That inconvenient data can't be justly dismissed as measurement
error without specific cause.

Mark

Mark said:

Michael Press said:

Jasper Janssen said:

True outliers tend to be measuring errors, and can be validly ignored. Of
course, on the other hand, throwing out half your data as unrepresentative
outliers is called fraud.

They are not measuring errors until you have evidence of
measuring errors. Throwing away data is not bad science,
it is no science at all. If the best you can do is shrug
your shoulders and say `beats me', so be it. But outliers
must not be ignored while presenting data to support a
hypothesis.

Outliers are not good or bad. Advances in science occur
when the data does _not_ support the hypothesis; e.g.
non-conformity to theory of specific heat measurements at
low temperature.

Sadly, nobody else seems to know this[1] these days.

Last term I had a student who blamed "outliers" as the source of all
problems in all statistical calculations - as in, "when we get rid of
all this data that troubles us, then we'll get the conclusions we want!"

I'm afraid the student didn't make it up on their own, but had help from
prior instructors.

[1] That inconvenient data can't be justly dismissed as measurement
error without specific cause.

Mark

Dear Mark,

I agree that outliers shouldn't be dismissed as measurement error
without specific cause, but being specific can be embarrassing.

I've been squeezing spokes and measuring spoke tension changes for
some time now, and I'm throwing out data left and right.

Specific cause: Realized at spoke 11 that my 32-spoke wheel had 19
drive-side spokes.

(Or else I miscounted while measuring spokes.)

Specific cause: Re-measured tension changes 4 times for 32 spokes.
Three graph lines match nicely and show that my measurements are
consistent. But the 4th graph line is insane.

(It does match the other three lines beautifully if spokes 1-16 are
re-numbered 16-1, but I don't want to admit that I went backwards on
one round of measurements.)

Specific cause: Each unsqueezed spoke is measured twice, once at its
midspan and again with the gauge next to the nipple. With a butted
spoke, measuring next to the nipple always produces a slightly higher
gauge reading. But sometimes the laws of physics are mysteriously
broken. Very rarely, a midspan reading of 23.2 will rise not to 23.9,
but much higher to 24.9. And every hundred readings or so, the
measurement will actually go the other way--a 23.2 midspan reading
will somehow drop to 22.9!

(True, I might have mistaken 23.9 for 22.9 or 24.9 while staring at
hundreds of tiny marks between 20 and 25, but it's more comforting to
think that Newton was wrong.)

Specific cause: Hmmm . . . why are all the drive-side measurements so
low? They shouldn't match the non-drive-side measurements, but they're
almost exact duplicates.

(It's as if I stupidly re-measured the same 16 spokes on the same side
of the wheel.)

Specific cause: My, look at how fast I'm getting at doing this!
There's no substitute for experience!

(And it really speeds things up when I forget to clamp the opposite
pair because then there's no 8-inch clamp sticking out from the wheel
in my face!)

Cheers,

C rl Fog l

Mark said:

Last term I had a student who blamed "outliers" as the source of all
problems in all statistical calculations - as in, "when we get rid of
all this data that troubles us, then we'll get the conclusions we want!"

Sounds like a member of the Bush family.

Michael Press said:

In article <[email hidden]>,

dvt said:

Robert Chung said:

Bill Bushnell wrote:
> I have collected
> ride statistics from rides in the local hills (SF Bay Area). I've
> included ten of these below. All of the rides below were ridden on the
> same bike (Rotator Ti Pursuit with 559/406 wheels) with the fairing

Quoted message said:

Quoted message said:

Quoted message said:

http://anonymous.coward.free.fr/rbr/bushnell.png

Quoted message said:

Quoted message said:

Good chart, Robert. I was thinking of making my own chart, but I think
yours answers my question. There is one outlier in the data with little
climbing, low power output, and low average speed. I think that data is
all from Ride 4. Throw out that outlier and I think you'll see some
strong trends.

Quoted message said:

Do not throw away information. The outliers can tell you
something. Researching them is difficult, but worth the
time.

Look at the data. It's apparent that Bill (the rider) was exerting
considerably less effort than other rides with similar climbing density.
Maybe he wasn't feeling good, maybe he was riding with a slower rider,
maybe.... Those might be good things to investigate if you want to know
while Bill was having an off day.

If you want to see Bill's "normal" capacity, you have to ignore the
"abnormal" days. If you're interested in the Bill's variability, then
you will want to include that data point. I didn't state that in my
earlier message, so I can see where confusion would arise.

--
Dave
dvt at psu dot edu

One of the most time-consuming things is to have an enemy. -E.B. White,
writer (1899-1985)

From: Bill Bushnell <[email hidden]>
Subject: Re: Banana Bike ride statistics [was Re: CVT interest]
Newsgroups: rec.bicycles.tech,rec.bicycles.racing
Followup-To: rec.bicycles.tech,rec.bicycles.racing
References: <[email hidden]> <[email hidden]> <[email hidden]> <[email hidden]> <[email hidden]> <[email hidden]> <[email hidden]> <[email hidden]> <[email hidden]> <[email hidden]> <quIhg.208769$5Z.17928@dukeread02> <[email hidden]> <wiVhg.210665$5Z.26484@dukeread02> <[email hidden]> <[email hidden]> <[email hidden]> <[email hidden]> <[email hidden]> <[email hidden]> <[email hidden]>

In rec.bicycles.tech dvt said:

Michael Press said:

In article <[email hidden]>,

dvt said:

Robert Chung wrote:
> Bill Bushnell wrote:
>> I have collected
>> ride statistics from rides in the local hills (SF Bay Area). I've
>> included ten of these below. All of the rides below were ridden on the
>> same bike (Rotator Ti Pursuit with 559/406 wheels) with the fairing

Quoted message said:

Quoted message said:

Quoted message said:

> http://anonymous.coward.free.fr/rbr/bushnell.png

Quoted message said:

Quoted message said:

Quoted message said:

Good chart, Robert. I was thinking of making my own chart, but I think
yours answers my question. There is one outlier in the data with little
climbing, low power output, and low average speed. I think that data is
all from Ride 4. Throw out that outlier and I think you'll see some
strong trends.

Quoted message said:

Quoted message said:

Do not throw away information. The outliers can tell you
something. Researching them is difficult, but worth the
time.

Quoted message said:

Look at the data. It's apparent that Bill (the rider) was exerting
considerably less effort than other rides with similar climbing density.
Maybe he wasn't feeling good, maybe he was riding with a slower rider,
maybe.... Those might be good things to investigate if you want to know
while Bill was having an off day.

Ride 4 was on May 31, 2004. Given the route profile, I can narrow it down to one of a
few possible routes I would have ridden. Given that it was Memorial Day, I can now
recall the exact ride, even though I don't have any notes written.

I rode from Sunnyvale, CA north along the San Francisco Bay mostly on bike paths, to
the Golden Gate National Cemetary in San Bruno. The wind was in my face north of
Foster City and past the SF Intl. Airport. Bike path riding tends to be slower, and I
do not usually like to work hard when riding into headwinds. The circumstances of my
visit to the cemetary may have played a role. The ride home was on Sneath Lane to
Skyline Blvd, then south through Woodside and Portola Valley. I recall that the ride
home felt easier even though it had all the climbing.

In any event the data for Ride 4 falls nicely in line with the other green data, the
low climbing density data, and does not appear to me to be an outlier. Reload the PNG
file at the link above to see this.

Quoted message said:

If you want to see Bill's "normal" capacity, you have to ignore the
"abnormal" days. If you're interested in the Bill's variability, then
you will want to include that data point. I didn't state that in my
earlier message, so I can see where confusion would arise.

Since we're examining the "speed vs power" characteristics of the bike, not the engine
performance, it's helpful to have data that represent a variety of conditions. I don't
see any ride in the set of 16 that I would consider to be an outlier.

--
Bill Bushnell
http://pobox.com/~bushnell/

In article <[email hidden]>,

dvt said:

Michael Press said:

In article <[email hidden]>,

dvt said:

Robert Chung wrote:
> Bill Bushnell wrote:
>> I have collected
>> ride statistics from rides in the local hills (SF Bay Area). I've
>> included ten of these below. All of the rides below were ridden on the
>> same bike (Rotator Ti Pursuit with 559/406 wheels) with the fairing

Quoted message said:

Quoted message said:

> http://anonymous.coward.free.fr/rbr/bushnell.png

Quoted message said:

Quoted message said:

Good chart, Robert. I was thinking of making my own chart, but I think
yours answers my question. There is one outlier in the data with little
climbing, low power output, and low average speed. I think that data is
all from Ride 4. Throw out that outlier and I think you'll see some
strong trends.

Quoted message said:

Do not throw away information. The outliers can tell you
something. Researching them is difficult, but worth the
time.

Look at the data. It's apparent that Bill (the rider) was exerting
considerably less effort than other rides with similar climbing density.
Maybe he wasn't feeling good, maybe he was riding with a slower rider,
maybe.... Those might be good things to investigate if you want to know
while Bill was having an off day.

If you want to see Bill's "normal" capacity, you have to ignore the
"abnormal" days. If you're interested in the Bill's variability, then
you will want to include that data point. I didn't state that in my
earlier message, so I can see where confusion would arise.

By definition the clusters are his normal capacity; a
tautology. The outlier is normal for whatever he was doing
that day.

--
Michael Press

On Tue, 13 Jun 2006 03:26:33 GMT, Ryan Cousineau <[email hidden]>
wrote:

[---]

Quoted message said:

An automatically shifted manual acts like an automatic to the user (or
can act like a push-button or sequential manual), but internally it has
the efficiency advantages of a manual: mainly, a mechanical clutch
instead of a fluid torque converter.

Torque converter boxes with mechanical lock-up on the highest gear
have been around here in Europe for going on 20 years now. They have
similar fuel economy figures to straight manual transmissions.

In article <[email hidden]>,

Andrew Price said:

On Tue, 13 Jun 2006 03:26:33 GMT, Ryan Cousineau <[email hidden]>
wrote:

[---]

Quoted message said:

An automatically shifted manual acts like an automatic to the user (or
can act like a push-button or sequential manual), but internally it has
the efficiency advantages of a manual: mainly, a mechanical clutch
instead of a fluid torque converter.

Torque converter boxes with mechanical lock-up on the highest gear
have been around here in Europe for going on 20 years now. They have
similar fuel economy figures to straight manual transmissions.

Lockup torque converters are at least as common on American cars (my
father's 1984 Oldsmobile had one). I'd believe the efficiency in the top
gear could be the same, but perfomance and economy in the other gears
(ie, city driving) will inevitably suffer.

-RjC.

--
Ryan Cousineau [email hidden] http://www.wiredcola.com/
"I don't want kids who are thinking about going into mathematics
to think that they have to take drugs to succeed." -Paul Erdos

In rec.bicycles.tech Robert Chung said:

Have you used your PT to estimate CdA for your bike? I haven't seen many
figures for faired bikes and using a PM with a faired bike simplifies the
positioning issue.

I've been meaning to do this for a while, and your question prompted me
to go out and collect some data.

I went out very early last Saturday morning to Crossman Rd. in
Sunnyvale, CA, a road that is wide and reasonably clean with no traffic
or wind at 6:30a. I did several runs on two different bikes at speeds
from 15 to 28 mph. What I did is to accelerate to the target speed,
then start the interval (using the PowerTap interval feature), hold that
speed to within 1/4-mph, then end the interval before braking. The
PowerTap records avg. power, avg. speed, and a bunch of other stats such
as duration and heartrate, that I noted on the sheet. The runs were
anywhere from 22 seconds to 1:49 each.

While doing the runs I noticed that I was using more power in one
direction than the other on a road that appeared to be flat. When I got
home I looked up the road on Topo! and discovered that it has a 7-foot
elevation difference from end to end over a 0.5 mile distance. I did
each speed run in both directions on the road. As best my eyes could
tell the slope was linear.

To calculate CdA I used the Power formula given in Whitt and Wilson (2nd
Ed.), at the top of page 157.

I put together an MS Excel workbook and a few charts with a cubic curve
fit through the origin for Power vs Speed. A quadratic function also
seems to fit the data well. The Banana Bike is a Rotator Pursuit with a
homebrew fairing (http://tinyurl.com/oplcf) under the tab labeled,
"Pursuit F3". The other bike I tested was a mostly stock faired and
socked Easy Racers Gold Rush (http://tinyurl.com/z5eq2), tab "Gold Rush
FS", probably the most commonly faired stock LWB recumbent model on the
market.

I was a bit unsure about choosing the coefficient of rolling resistance
as its choice seems to have a large affect on the estimate of CdA. I
ended up choosing Cr seeking to minimize Std. Dev. of the estimate for
CdA.

Summary

Pursuit F3: CdA = 0.19, Cd = 0.35
Gold Rush FS: CdA = 0.24, Cd = 0.33

As expected from the Power vs Speed data, CdA for the Pursuit was
smaller than for the Gold Rush, on account of the smaller surface area.
In fact, Cd was nearly the same for both bikes, although I suspect that
at speeds higher than 30-35 mph Cd might be higher for the Gold Rush on
account of the free edge of the sock fabric flapping in the wind,
creating additional turbulence.

The Excel workbook and all the details can be found at

http://tinyurl.com/plbrd

--
Bill Bushnell
http://pobox.com/~bushnell/

In article <[email hidden]>,

Bill Bushnell said:

Pursuit F3: CdA = 0.19, Cd = 0.35
Gold Rush FS: CdA = 0.24, Cd = 0.33

CdA units are m^2.

Quoted message said:

The Excel workbook and all the details can be found at

http://tinyurl.com/plbrd

For those who don't have MS Excel, I've created two PDF files of the two
Excel worksheets:

Pursuit (Banana Bike): http://tinyurl.com/zald3
Gold Rush: http://tinyurl.com/lyx74

--
Bill Bushnell
http://pobox.com/~bushnell/

Bill Bushnell said:

Summary

Pursuit F3: CdA = 0.19, Cd = 0.35
Gold Rush FS: CdA = 0.24, Cd = 0.33

The Excel workbook and all the details can be found at
http://tinyurl.com/plbrd

Thanks. Very interesting. I'm vaguely surprised that the CdA for the Gold
Rush is as high as it is. What kind of fairing is a Zzipper?

If your PT is a Pro or a SL and you have the cadence wire hooked up, it's
often easier to maintain a constant speed by watching the cadence. A
sheltered spot is important since even small amounts of (unmeasured and
unobserved) wind can screw up the estimate.

If you do find a flat windless spot (I've always lusted after Hangar 1 at
Moffet Air Station) and you do constant speed runs you can plot w/v
against v^2. If it's truly flat, windless, and constant speed then the
intercept and slope will be proportional to rolling resistance and CdA.
This isn't a robust method, however, and small errors will give you a
lousy estimate.

All that said, I don't think that constant speed runs on flat venues are
really necessary--in fact, I often think it'd be better to do this on
varying terrain with variable speed because the error in the estimate is
reduced. I like it that you ran your speed from 4.5 m/s to 12.5 m/s.
That's a nice wide range. If you do it on a dip or a hump (and you have a
long enough run-out so that you don't have to brake at the ends of the
test section) then you can vary speed, power, and power for speed--just
account for the accelerations. You can see an example of this with Dede
Demet's WC data here:
http://anonymous.coward.free.fr/wattage/altimeter/altimeter.html

(That estimate is only for proof-of-concept since it presumes that she
didn't use her brakes)

I'll probably do an on-road CdA estimate next month, when my schedule
frees up a bit.

In rec.bicycles.racing Robert Chung said:

Bill Bushnell wrote:

Quoted message said:

Quoted message said:

Summary

Pursuit F3: CdA = 0.19, Cd = 0.35
Gold Rush FS: CdA = 0.24, Cd = 0.33

The Excel workbook and all the details can be found at
http://tinyurl.com/plbrd

Quoted message said:

Thanks. Very interesting. I'm vaguely surprised that the CdA for the Gold
Rush is as high as it is. What kind of fairing is a Zzipper?

My measurement of CdA for the Gold Rush has a fairly high standard deviation, over
0.03, compared the the Pursuit's measurement with a std dev of 0.01. The air was very
still when I started. Maybe I got a few unlucky gusts of wind that I could not detect.
It was later in the morning, about 8a when I did the runs with the Gold Rush, 6:30a for
the Pursuit. I'll probably go out again sometime and take some more measurements.

http://www.zzipper.com/
They make the stock fairing that Easy Racers uses on their bikes.

Quoted message said:

If you do find a flat windless spot (I've always lusted after Hangar 1 at
Moffet Air Station) and you do constant speed runs you can plot w/v
against v^2. If it's truly flat, windless, and constant speed then the
intercept and slope will be proportional to rolling resistance and CdA.
This isn't a robust method, however, and small errors will give you a
lousy estimate.

With about 0.2mi of usable length Hangar 1 isn't long enough. I barely had enough
runway at 28mph to get a decent length interval, and I had 0.5 mi to play with.
Aside from the slope, which can be compensated for, the pavement wasn't as smooth as
would be ideal. Fresh asphalt would be best, and some of the roads in the area are
good in that respect, but they have other problems such as traffic, traffic lights, not
flat (usually bumps over the creeks). There might be a better road somewhere else in
the Si Valley flatlands.

Quoted message said:

All that said, I don't think that constant speed runs on flat venues are
really necessary--in fact, I often think it'd be better to do this on
varying terrain with variable speed because the error in the estimate is
reduced. I like it that you ran your speed from 4.5 m/s to 12.5 m/s.
That's a nice wide range. If you do it on a dip or a hump (and you have a
long enough run-out so that you don't have to brake at the ends of the
test section) then you can vary speed, power, and power for speed--just
account for the accelerations.

Unfortunately, I have don't have the "pro" model, just a more recent "standard" model
(yellow case), so I have no way of retreiving the data log, just the averages given on
the computer display. I do have a cadence meter on both the PT and my other bike
computer.

--
Bill Bushnell
http://pobox.com/~bushnell/

Bill Bushnell said:

For those who don't have MS Excel, I've created two PDF files of the two
Excel worksheets:

Pursuit (Banana Bike): http://tinyurl.com/zald3
Gold Rush: http://tinyurl.com/lyx74

The average heart rate column is identical on both sheets. Is that right
or did you miss a cut 'n paste operation?

--
Dave
dvt at psu dot edu

One of the most time-consuming things is to have an enemy. -E.B. White,
writer (1899-1985)

Bill Bushnell said:

Unfortunately, I have don't have the "pro" model, just a more recent
"standard" model (yellow case), so I have no way of retreiving the data
log, just the averages given on the computer display. I do have a
cadence meter on both the PT and my other bike computer.

The standard model should still let you download.

The Pro and standard hubs are identical (except for color): the difference
is in the head unit. The Pro has the option of using conventional wired
(rather than "virtual"😉 cadence: see
http://anonymous.coward.free.fr/wattage/rosetta/rosetta_details.html#speed
for a discussion of the PT's virtual cadence.