Using Newton on a Touring Bike

[KenS]

I just returned from a week long bike tour using the Newton. The main reasons for using it were:
- To get a better idea of the grades as it was in hilly country
- To get a good reading of altitude to help with navigation and also to know when I was near the top of a big climb
- To have a rough idea of power output so I didn't overcook on the climbs
- To get a rough idea of calories burnt to make sure I replenished enough each day (nom nom nom)

I had the Newton pretty well dialled in on the bike when it was unloaded and on sealed road but the tour was also on some dirt road and with about 20kg of load.
I just upped the weight in the profile to cover the extra load.
Do you think that would be enough for the purposes above?
Or should I have done a full setup including a cal ride? And if so, if I do a cal ride would it need to be fully loaded or could I get away with just putting on the panniers with some light stuffing?
Has anyone else got any experience with this use of the Newton?

BTW I got some interesting braking numbers in Isaac from descending a long 12% downhill fully loaded on dirt with lots of sharp bends and steep dropoffs. Would have been interesting to see the braking power on the way down as it was all coming from my hands on the brakes

[racerfern]

There's no doubt the CdA increased with the bags so a cal ride with all the gear would have gone a long way. But too late now.

As far as the portions on dirt there's nothing that can be done about that unless you have a known day that is on dirt then you could estimate a profile for that part.

For the future, I would load the bike so your calride has the actual weight not some estimation.

[Velocomp]

There's no doubt the CdA increased with the bags so a cal ride with all the gear would have gone a long way. But too late now.

As far as the portions on dirt there's nothing that can be done about that unless you have a known day that is on dirt then you could estimate a profile for that part.

For the future, I would load the bike so your calride has the actual weight not some estimation.

You can use the Analyze/Tweak CdA to adjust the rides up to account for the panniers. You could even do some coast downs with the panniers on the bike to measure their influence...

[Russ]

I bet you can use the new analyze Cda in some of those downhill coasting sections, providing you did take your hands off the brakes some

That would give you a clue to the Cda especially if you did the coasting on some paved roads too.
On the paved road your Crr would be closer and so the Cda would be more accurate.

Then you would feed that Cda back in with the tweak tool.

You might be able to compare the Cda analyzed on the dirt to that on the road. Then take the those into one of the online power calculators and find the Crr that produces the power match for the dirt section but using the Cda from the road section... if that makes sense. You would find the Crr by trying different higher numbers until you bracketed in on a match.

Just some thoughts, have not tried some of this myself

Russ

[racerfern]

Russ is correct. Check this post that gives some insight as to CdA climbing although the real culprit is the brakes.

http://www.ibikeforum.com/viewtopic.php ... 786#p14786

[KenS]

I've got Isaac 1.0.9 for Windows but I don't see any Analyze CdA option
I did quite a bit of coasting downhill on paved roads so would be good to see the result.

[racerfern]

1. Select a downhill coasting portion but make sure to select a small section before or after the coasting section that has cadence value.

2) Click Tools > CdA Analysis on Selection... have fun.

[KenS]

The CdA Analysis on Selection is grayed out. Do I need a Newton+ for this to work? And if so is there an upgrade path from Newton to Newton+?

I like this approach a whole lot better than loading up 20kg of panniers and trying to get to 32km/h on a slight uphill to do a coast down 8 times over

[Velocomp]

CdA measurement is a feature of the Newton+. You can purchase an upgrade key at the iBike Store if you wish.

[KenS]

I can't find the Newton+ upgrade option on the store site. There is an indoor trainer upgrade and a Newton+ GT upgrade (I've already got the GT option) and an iPro to iAero upgrade but isn't that Gen3 only?

I must have more numbers to play with

[Velocomp]

This will do it:

http://store.ibikesports.com/upgrades-t ... re-upgrade

We'll get the name fixed on the webstore.

[KenS]

Got the upgrade and installed it.
Before looking at CdA with panniers I thought I'd better do a cal ride and coast downs to get my base CdA and Crr dialled in.
Also because last cal ride and coast downs I did I had the total mass a bit off and I was still learning how to do them
I figured I need coast downs as even the basic bike has handlebar bag, lights, racks and other stuff hanging off it that would affect the default CdA

Just wanting to check if I did everything right

So I weighed me and the bike - 92kg
Went to my cal ride location which is just far enough away to let the Newton acclimatise and is a quiet road with a good surface
It was a rare day of low wind with wind speed in single digits (km/h)

I did forget to do a tilt cal and wind cal before leaving which was not a big problem but I prefer to do those at home indoors. I did those when I arrived at the cal ride location.
Did the cal ride then 6 coast downs and finally a 6.4km (4 mile) out and back as per the instructions
Loaded everything into Isaac when I got home

When I looked at the coast downs with Crr "Setting in iBike" the Crr was 0.0131 and this gave a CdA of 0.178 which seemed pretty low given that I was on the hoods with a fairly relaxed (upright) position and with loosish clothing
So then I used the "Crr Adjusted" option which gave a Crr of 0.0067 and CdA 0.468
This seemed more reasonable although the Crr sounds a bit low for touring tyres (Vittoria Randonneur Pro)

Also, on the way home there is a bit of a downhill just before I get home which I coast along so I also got CdA numbers from the Newton which averaged about 0.34 and was rising as speed increased. So I guessed this was more evidence that the 0.178 CdA produced by the coast downs was wrong.
So for now I'm using Crr 0.0067 and CdA 0.468 and I'll see how those go

As a footnote some minor points:
- I was wondering where the 0.178 CdA came from. This looks more like the Aero numbers from the coast downs
- In Isaac, I kept seeing my weight in lbs rather than kg even when I chose "Metric". It fixed itself when I loaded up a cal ride file
- Isaac also says to do a 4 mile out and back after the final coast down but that doesn't seem to get used anywhere

Here's the ride files for the cal ride, 4-mile out and back and the coasting downhill section near home - I'm guessing the Cal Ride file contain the coast downs as well

121128_cal_data.ibcd4m

Cal ride

(283.7 KiB) Downloaded 595 times

28_11_2012_1219_iBike.ibr

4 mile out and back

(45.81 KiB) Downloaded 578 times

Coasting _28_11_2012_1244_iBike.ibr

Coasting

(6.29 KiB) Downloaded 611 times

[Russ]

Kens,

I tried several things with your cal ride file. See the comments in the jpg, cut and paste of those comments here:
Adjusted:
.441 .0073 for 8:14 only - highest firction
.523 .0061 for 8:17 only - highest Aero
Fixed:
.448 .0078 for 8:17 only crr from analytic cycling touring class.

when I tried the Cda analyze tool, did not get real analysis... just the default Cda value. Not sure why for that. So I put an export into Golden Cheetah of your coasting file and not sure I got anything definitive from that but the Cda .448 , Crr .0078 with the slope section zoomed in gave a fairly good match on the two lines. That said, I am a new novice with the aerolab, maybe someone else will weigh in on this...

Maybe a start,
Russ

[KenS]

Thanks Russ. This is certainly one of the dark arts
I'll have to start playing with Aerolab.
For now the .468 .0067 seems to be working although sometimes at speed seems to give high power readings

I found the CdA analysis interesting but I need to find a long downhill stretch to get longer coasting segments. I was thinking Velocomp needs to come up with a "brake switch" so it can tell when you are braking and turn off the CdA analysis
Would also be nice to be able to turn off the calories as well as cadence since I'm not interested in those as they come up on my Garmin. Interestingly, with cadence turned off it still comes on sometimes at 0rpm

[Russ]

Ken, you are welcome, such as it is

A few more thoughts... First, if your power is too high at high speed then likely your Cda is too low.

This is not a black art but there is a learning curve which I am still climbing. Fortunately I enjoy the climb

I recently went through this with my cross bike on which I installed Gatorskin Hardshell 700x32's. I have been running foldable Gatorskins on my road bike but these are different. These clearly had a higher rolling resistance and I could not find where anyone had published Crr values. I also was not able to find Crr values for your tires.

The Newton gave me a .0065 Crr from a cal/CD ride on smooth pave using 'adjustable' . But with the Cda analyze tool it became apparent that it was too low! So I went to cycling analytics with data from a smooth road uphill slow portion of a ride and worked the values until I had a power match, the Crr was around .0073 by memory. So I dialed that into my profiles.

Then I road seven times (the amount of time I had) around the same smooth course with some hills at various speeds. After exporting from iBike software as a csv file, I loaded that into Golden Cheetah and used the Aerolab, zooming in to six of the seven rides (the first was still in the five minutes after reset so I skipped it) and slid the sliders around until I got the best fit of the two lines. This yielded, I think, the best Cda/Crr pair yet. The Crr came out a bit higher still.

Another trick is to soft peddle while coasting so that iBike records the data instead of forcing zero power then play with the Cda/Crr values to find where that portion comes closest to zero power. Some here say that is no good but my experience with it says it seems to work.

If you want to upload more of a ride file, when I get more time I will be glad to play with it some more. It would also help if you give the full weather data ie temp, humidity, barometric pressure and dewpoint for the rho value.

Regards,
Russ

[Russ]

Ken or anyone interested,

I have just completed a set of measurements using various methods to arrive at CdA and Crr with confidence. In this case, with 700-32 Gatorskin Hardshells for which I could find no Crr data posted anywhere.

Before we get into it, going back to my cal ride and using this data fits perfectly, hang on to see how.
First I did several measurements using Chung and regression methods, see this link for details:
http://www.cyclingpowerlab.com/aerodynamicmodels.aspx

The problem for me, with these, was that you really should have your Crr known in advance as these are really designed to find CdA.

I also tried various slow steady hill climbs and put that data in this link:
http://www.analyticcycling.com/ForcesPower_Page.html
My idea with this was to reduce aero to a minimum to make Crr stand out more. This, especially when I selected portions of hill climbs where the slope and speed were steady and winds were minimum, worked well to yield a Crr value. The Cda was not so critical using this method. A 6% grade at about 8-10 KH steady cadence on smooth pave and steady slope worked well.

Next for a sanity check I used this method to measure Cda directly (assuming .88 drag factor):
http://www.cyclingpowerlab.com/cyclingaerodynamics.aspx
I got my wife to photograph me on the bike with a friend holding me up on one side. I forgot to tell the wife to make sure she was dead on in front so I got a bit of yaw angle in the picture. Turns out that was not so bad as it made me aware of the increased area that the wind sees when it is off center axis.
So I got from that a Cda of .388 to .407 with the approximate rear area of the front wheel subtracted and including the full area seen in the photo. The photo work in the software took a lot of time and some learning curve, but hey it is free software, too (not too bad, I might add)!

Then I went back into my last cal ride after I installed these tires and deselected the one obvious outlier. I used fixed Crr values from my analytic cycling runs and low and behold, the Cda came in at .407!

This is on the hoods. I am six feet and a bit stocky and on my cross bike.

[KenS]

I'm very interested in this Russ.
What I don't get is how you can work out Crr from riding uphill. Do you have a DFPM as well?

[Russ]

Ken,

I guess I make it sound a bit simpler and more bullet proof than it is/was
But basically, I iterated through this process and my last run with cycling analytic uphill analysis was compared to the ibike for a good match. I will try to upload pix at the end of this post.

This is also why I went (for me) to the extreme work with the photo I mentioned. A hindsight comment for that is it would probably make (with paint.net) the process of selecting the rider/bike object for pixel count analysis much easier with a nice even fairly light colored background such as a wall or a maybe better a sheet. Because I did not do that, I had to carefully erase all of the background !

Anyhow, the coastdown method does not require power data but the iBike makes a great data collection tool as it records everything you need at 1 second intervals nicely. The challenge is identifying the start points for the downhill run. So I started at exactly the same spot with my wheel magnet positioned one spoke from the sensor which gave me zero or appearing very close to zero point. Then I calculated the CVS file lined numbers and data to get the length of my run in zone. I had started coasting at the same spot as well so zero power shows up marking that spot, just add a little safe length for things to settle. Pasting the data from the speed columns for each run into excel works well to provide the file for uploading. For the Chung method, pasting speed and power (of course iBikes power may be chasing your tail a bit if you are going for different cda and crr values than the profile .

So nutshell, you still identified the potential flaw in the works. I guess I got a bit excited when I got the dead on match between the iBike cal file and the photo CdA and then very close match in the online stuff with the iBike profile. So it is a lot of work. I would point out that I was converging on the Crr value before I took the photo, from the use of these online tools and also Aerolab in Golden Cheetah.

Hope this is some help for you, I still want to go further and asses the Crr on rougher road in a similar way instead of just adding .001 or the like.

Merry Christmas!
Russ

12/25/2012 edited to correct ref in third paragraph from regression method to coastdown method.
Also redid pix to get rid of extra views, sorry - this is a learning curve for me

iBike uphill measurement section

iBike_figgy_12_18_2012.jpg (82.12 KiB) Viewed 34468 times

Analytic Cycling uphill section analysis

AC_figgy_12_18_2012.jpg (124.5 KiB) Viewed 34470 times

[Russ]

Sorry about the extra pix copies I think retrying in paint.net not sure how I did that.
Edit: fixed the extra pix for a small gift of disk space to Velocomp for Christmas!

Another site that helped with the weather data was this:
http://www.gribble.org/cycling/air_density.html

Russ

[Russ]

Kens and all,

I have done more analysis using analytic cycling...this time with the Terminal Velocity on a Slope tool.

This requires no ref to a power meter at all, and seems a tad less finicky than the coastdown method (so far). That said, I am still starting out with the Cda from the photo analysis, so using these coasts to find terminal velocity for Crr measurement.

Before I get into it, I must say that I appreciate the cal rides with the iBike much more
The iBike Newton did a fantastic job with my road bike, and once I figured my Crr for the cross bike, the Newton cal ride fell right into place. I am not sure why, without using fixed Crr, that cal seemed problematic. Perhaps that is for another post. I wonder, though, if having tires with higher than typical road bike Crr's gives the cal routines a bit of a challenge? I remember John posting something possibly hinting in that direction due to a relative lower volume of those? Anyhow.... onward.

So here are two runs, one selected for smooth pave and one for rough pave. This is a bit preliminary as I have ideas on how to get better measures of terminal velocity on some of my favorite hills but should illustrate the process. My idea for more, since I was using coastdown data from prior rides, is to go into the downhill as close to terminal velocity as I can so it settles for a longer period. Also note that it is a bit of luck of the draw, around here, to get a relatively low wind run, which helps a lot, just like cal rides.

I think, if you can get a variety of hill slopes that are good enough for these measurement and thus a variety of terminal velocities, then you can, with a bit of patience, iterate through the tool with only estimates of Cda and Crr and tinker the values until they best fit all of the hill slope speed measurements. I suppose if you want to take a transit out then you can improve the slope measurements a bit but iBike seems to work ok for this so far

Here are four pix, I hope , if I can do it right.
First two are the smooth iBike and Analytic Cycling tool pair
Second two are the rough pave. This is the first attempt at rough pave and I am not sure of the higher Crr value yet but it illustrates the point and method.

Merry Christmas,
Russ

Having trouble adding the fourth pix here, may need a follow on post....

[Russ]

I have not figured out the picture ordering thing. So rough seem to be top and this (bottom) with smooth in between

Fourth pix goes with above post:

[Russ]

A couple of quick tips with the analytic cycling tool above for terminal velocity.

The speed answer is the top point of the y axis. Multiply by 3600 to get meters per hour. I extended the time to 150 arbitrarily to extend the plot to make sure the T.V. is achieved. Setting the entry speed near the T.V. also increases resolution of the Y axis. I think this is fine as the only thing you really want is the actual terminal velocity to match to the iBike speed recorded on your hill coastdown.

Russ

[Zoltan]

Ken or anyone interested,

I have just completed a set of measurements using various methods to arrive at CdA and Crr with confidence. In this case, with 700-32 Gatorskin Hardshells for which I could find no Crr data posted anywhere.

Before we get into it, going back to my cal ride and using this data fits perfectly, hang on to see how.
First I did several measurements using Chung and regression methods, see this link for details:
http://www.cyclingpowerlab.com/aerodynamicmodels.aspx

The problem for me, with these, was that you really should have your Crr known in advance as these are really designed to find CdA.

I also tried various slow steady hill climbs and put that data in this link:
http://www.analyticcycling.com/ForcesPower_Page.html
My idea with this was to reduce aero to a minimum to make Crr stand out more. This, especially when I selected portions of hill climbs where the slope and speed were steady and winds were minimum, worked well to yield a Crr value. The Cda was not so critical using this method. A 6% grade at about 8-10 KH steady cadence on smooth pave and steady slope worked well.

Next for a sanity check I used this method to measure Cda directly (assuming .88 drag factor):
http://www.cyclingpowerlab.com/cyclingaerodynamics.aspx
I got my wife to photograph me on the bike with a friend holding me up on one side. I forgot to tell the wife to make sure she was dead on in front so I got a bit of yaw angle in the picture. Turns out that was not so bad as it made me aware of the increased area that the wind sees when it is off center axis.
So I got from that a Cda of .388 to .407 with the approximate rear area of the front wheel subtracted and including the full area seen in the photo. The photo work in the software took a lot of time and some learning curve, but hey it is free software, too (not too bad, I might add)!

Then I went back into my last cal ride after I installed these tires and deselected the one obvious outlier. I used fixed Crr values from my analytic cycling runs and low and behold, the Cda came in at .407!

This is on the hoods. I am six feet and a bit stocky and on my cross bike.

I had gone through the same analysis before I acquired my first power meter. I got almost the same CdA. I am 185.5cm tall and the tyre of my trekking bike is 700x35.

I did not use any software except Excel, so my contour was drawn to the inner side of the garage door by my kids, while my wife was holding the bike with me. The pencil was attached to the end of a long but narrow stick, so the small men were standing behind me and just to had to follow the required trajectory with the stick.

Then I applied the rectangle rectification or sthing like that. I mean I drew rectangles to approximate the parts of the unified area of the bike and the rider.

And I could collect some Crr data both from German websites and I also made a big pressure on the technical support of Continental, so finally they gave me a data. But you should keep in mind that Crr is a function not just of tire pressure and the smoothness and the material of the road surface, but also that of ambient temperature and ride speed.

That means that all these things are funny, but the whole thing is just a good fun.

[KenS]

Sorry I've been a bit quiet lately. Was wanting to try out the "elevation" method (Chung?) but realised on my last cal ride, after changing my rear tyre for a different model , I forgot to change the diameter - D'OH!!!!
I did another cal ride but we've been having gusty winds of 20km/h to 50km/h so the calride is of dubious value.
Nevertheless some things I discovered while trying to get the calculations right.
- Isaac outputs CSV with altitude to the nearest foot converted to metric so it is quite coarse - increments of 0.3 metres. So instead I use the gradient and incremental distance to determine the altitude change
- Whilst braking I tried back pedalling so I could ignore those parts but it seems due to some filtering that the braking effect still shows up after cadence drops to zero. However, I need to double check my technique on that
- Unfortunately, Isaac does not output dynamic CdA to the CSV, presumably unless you have a DFPM
Anyhow, I coasted down my street (with a big tailwind) and plotted elevation that corrects for power balance vs recorded elevation and also the difference in the incremental elevations. You can clearly see me braking on the first few data points

Calculated elevation

CdA graph.JPG (119.29 KiB) Viewed 34267 times

The good news is that the profile is the same as the one in Isaac although the Isaac one looks smoothed.

iBike dynamic CdA

iBike CdA.JPG (234.66 KiB) Viewed 34265 times

This makes me think I should just stick with the CdA analysis as recorded by Newton
But first I need to wait for a day with light winds so I can redo my calride, do some coast downs and then try down downhill coasting

[Russ]

Zolton,

That is a great story of your Cda measurement! I have two son who are now grown but I just picture that!

You wrote:
"But you should keep in mind that Crr is a function not just of tire pressure and the smoothness and the material of the road surface, but also that of ambient temperature and ride speed.

That means that all these things are funny, but the whole thing is just a good fun."

Yes and I might add load and, for the rear tire, power. Well, I ran another set yesterday afternoon just before sunset and it was about 7C down from 13C in the prior runs posted. The Crr values came out significantly higher. So I will prepare another profile for use when it is a colder ride. The most confounding thing is wind. It would be great if someone posted a spreadsheet that does the terminal velocity calculation but allows for wind values. I am still a spreadsheet novice

and
Kens,

Nice job and thanks for this tip:
"Nevertheless some things I discovered while trying to get the calculations right.
- Isaac outputs CSV with altitude to the nearest foot converted to metric so it is quite coarse - increments of 0.3 metres. So instead I use the gradient and incremental distance to determine the altitude change."

Happy New Year everybody!
Russ

[Zoltan]

Russ,

Later I will send you my excel spreadsheet in an email, just PM me your email address, because as i remember PM doesnt allow attachments.

You will be mad from that beauty.

I calculated the air density using a complex formula that I built up from two others found on web, so I could incorporate more variables into one formula than it was possible just relying on the web.

Alas the whole thing is not exactly about terminal velocity, but average velocity achieved on ascents, flats and distance to manually calculate three power numbers, but it can be used for your purpose, I guess. If not, at least you can see how much watts the wheel acceleration means.

[KenS]

The purpose of the terminal velocity method is to eliminate any effect from acceleration or deceleration. This is always hard to do unless you have ideal conditions with a long, constant slope, preferably straight, and no wind changes. My spreadsheet takes acceleration into account so it is not necessary to reach terminal velocity. Newton and Isaac do too.
Did you know that the new CSV file from Isaac also includes air density? But this does not take into account relative humidity

Basically my plan is to as follows:
- New cal ride and coast downs to get the initial Crr and CdA values
- repeated coasting downhill from a standing start (or close to it) being sure to back pedal whilst braking (plus a few seconds before and after)
- Once I get Crr dialled in do more downhill coasting with panniers lightly stuffed (to minimise impact on Crr) to get CdA with panniers
- Same again with loaded panniers to recalculate Crr under load. This will make the uphill return fun

As for the calride and coast downs the 7-day weather forecast for each day during the next week is "light winds increasing to 15-30km/h when KenS goes out for his calride"
This could take some time....

[Zoltan]

"My spreadsheet takes acceleration into account"

The same for mine. It is calculated by a software of mine from the gpx data.

[KenS]

I calculated the air density using a complex formula that I built up from two others found on web, so I could incorporate more variables into one formula than it was possible just relying on the web.

I also incorporated the air density formula from https://www.brisbanehotairballooning.co ... nsity.html
I deduced that the Rho value output to the iBike CSV file assumes a relative humidity of 50%. Living in a dry place I was worried that would affect the power calculation but once I included a correction for relative humidity is made almost no difference.

[Zoltan]

I calculated the air density using a complex formula that I built up from two others found on web, so I could incorporate more variables into one formula than it was possible just relying on the web.

I also incorporated the air density formula from https://www.brisbanehotairballooning.co ... nsity.html
I deduced that the Rho value output to the iBike CSV file assumes a relative humidity of 50%. Living in a dry place I was worried that would affect the power calculation but once I included a correction for relative humidity is made almost no difference.

Yep, no big influence of these variables.

As I wrote I created a really really complexed formula for air density which is:

100*((6.1078*10^((7.5*(273.15+B21)-2048.625)/((273.15+B21)-35.85))*B25)/(461.495*(273.15+B21))+((B24*(1-0.0065*B4/288.15)^(9.80665*0.0289644/8.31432/0.0065))-(6.1078*10^((7.5*(273.15+B21)-2048.625)/((273.15+B21)-35.85))*B25))/(287.05*(273.15+B21)))

where cell B4 is (average) altitude in meters, B21 is temperature in Celsius, B24 is barometric pressure in hPa and B25 is relative humidity in %. As I remember I could find many formulae, but all of them used not more than three of these four variables. I am not good at physics, so did not understand why none of the formulae used all the four variables, so I tried to incorporate all the four and I succeeded. But even I was shocked that finally this long formula gave me plausible values.