conducting a study on how crank stiffness effects speed

Questions about bike hire abroad and everything light bike related. No off-topic chat please

Moderators: robbosmans, Moderator Team

istigatrice
Posts: 849
Joined: Sat May 12, 2012 8:32 am
Location: Australia

by istigatrice

thought I'd get a thread up here to as this forum sees more traffic than the fairwheelbikes forum...

I'm in the process of planing my study on how crank stiffness effects a rider's speed. Reading a few articles has got me thinking on testing if the flex of these cranks results in power loss.

My plan is to use a custom crank in which I can vary the stiffness by changing the inserts in the arms (with weight constant). I will be using a power meter to keep my power output consistent and taking timed efforts up a local climb. I’ll look to repeat these efforts over several days, taking an average over the days. I’ll be testing the different stiffnesses one after the other to try and control variables like wind and humidity. Obviously things like windspeed and humidity can’t be controlled, but I’ll look to take that info into account by getting the local weather station’s data.

does anyone have any critique on this topic and perhaps some suggestions?
I write the weightweenies blog, hope you like it :)

Disclosure: I'm sponsored by Velocite, but I do give my honest opinion about them (I'm endorsed to race their bikes, not say nice things about them)

User avatar
prendrefeu
Posts: 8580
Joined: Wed Oct 31, 2007 10:32 pm
Location: Glendale / Los Angeles, California
Contact:

by prendrefeu

Great idea! I hope this happens, it's very interesting.

Off the top of my head:
-Can you account for variability in drivetrain loss elsewhere? Such as the chain? Take a look at the FrictionFacts reports. Maybe you want to ensure that the chain's wear is kept to a minimum across the tests?
-You're probably keeping the chainrings constant throughout, and checking the torque on the chainring bolts for consistency?
-Will you be able to check that your custom crank arms are actually affected in their stiffness through your changing of the inserts? Perhaps a set up similar to the FWB can verify the change in stiffness?
Exp001 || Other projects in the works.

by Weenie


Visit starbike.com Online Retailer for HighEnd cycling components
Great Prices ✓    Broad Selection ✓    Worldwide Delivery ✓

www.starbike.com



User avatar
Leviathan
Posts: 1282
Joined: Sat Mar 17, 2007 10:49 am
Location: Mallorca, Spain
Contact:

by Leviathan

Im not (any more!) an Engineer, but my 2c worth is: How much differences do you expect to see between (good quality) cranksets. If you run some preliminary numbers and make some basic assumptions (eg you wouldnt expect deformation of a crank to remove more than, n% of the total power output, so a 5% variation in stiffness across units would be a n.5% change... )
I SUSPECT the biggest issues would be your own personal power output- how do you intend to adjust for how much sleep you got, how youre feeling, health, etc etc.? I agree averaging will remove some of that bias but I wonder if youre just going to be left with noise as opposed to any specific conclusions?

eric
Posts: 2196
Joined: Tue Nov 11, 2003 9:47 pm
Location: Santa Cruz, California, USA
Contact:

by eric

Rather than the massive expense of making a custom crank, you could use stiff and non stiff commercially available cranks, and weight them to equalize the weight.

I suspect that the difference, if there is any, will be much lower than your experiment's margin of error. That is kind of high due to the experimental design- times up a climb vary a lot, even at "constant" power (which is difficult to do). For example if you're over your targer average the first half and under the second, the average will be the same, but will the total time be faster or slower? Before you go too far with this design, try doing some times up your test climb with identical gear and see what the error is.

I'd be interested to see if the theory that any flex in the crank is largely returned later in the pedal stroke (the spring model). That's been theorized but to my knowledge never measured or simulated. Perhaps it is still too difficult, but you'd think that an FEA program or a custom model would be able to do it. But answering that question would tell us the effect of the flex in the crank (and could be extended to all other bike parts).

User avatar
fa63
Posts: 2533
Joined: Mon Feb 19, 2007 7:26 am
Location: Atlanta, GA, US

by fa63

I agree field testing may not be the best way to quantify this. The differences will likely be within the accuracy of the powermeter, so you won't really know if you are measuring is a true difference due to stiffness. As mentioned above, it is also very difficult to maintain a consistent power output. Averaging over several runs may help but I am still not sure you will be able to draw solid conclusions from your field data.

This is where finite element analysis comes into play; you can change the stiffness of your system and see what the outcome is without the real world limitations. For what it is worth, Fairwheel Bikes did this using a strain energy approach in their latest crank shoot-out, and the results showed ~2W difference between a flexible crankset and one that is 50% stiffer. In comparison, I think SRM claims accuracy to within two percent, and assuming you are climbing at 200W there may be a 4 W difference in the readings just due to accuracy limitations.
Last edited by fa63 on Sun Jul 07, 2013 5:46 pm, edited 2 times in total.

Epic-o
Posts: 621
Joined: Tue Nov 20, 2007 7:18 pm

by Epic-o

As others have said, field testing is a no go for this. Even in a more controlled environment, I don't think that current powermeters have enough accuracy for this type of studies

istigatrice
Posts: 849
Joined: Sat May 12, 2012 8:32 am
Location: Australia

by istigatrice

thanks for all the input guys, lots of food for thought

I do hope that this test doesn't just boil down to the accuracy of power meters, but if it does, that is an acceptable conclusion for me. I'm trying to look into how stiffness effects speed, so if it doesn't effect it to a noticeable degree then that is fine.

I've seen the fairwheelbikes crank test and that's the inspiration for this study. Obviously a FEA will be much more accurate than what I am doing, but I would like to try and quantify this in the "real world" (not dismissing the FEA). If the accuracy of the power meter is the only thing that effects the speed, I believe a valid conclusion is that crank stiffness does not have a noticeable effect on speed. Is there any reasons why that isn't a valid conclusion?

Lastly; I've just had a thought for me to keep my speed or time constant and measure my power output... What are people's opinions on that? I think it may work too, however if I try to keep speed constant, I'll be relying on the accuracy of two instruments rather than 1 (which could work with or against me). In keeping time constant, the problem with that is I can surge in different places to get the same time, which could effect the average power output.

any other thoughts?
I write the weightweenies blog, hope you like it :)

Disclosure: I'm sponsored by Velocite, but I do give my honest opinion about them (I'm endorsed to race their bikes, not say nice things about them)

istigatrice
Posts: 849
Joined: Sat May 12, 2012 8:32 am
Location: Australia

by istigatrice

eric wrote:I suspect that the difference, if there is any, will be much lower than your experiment's margin of error. That is kind of high due to the experimental design- times up a climb vary a lot, even at "constant" power (which is difficult to do). For example if you're over your targer average the first half and under the second, the average will be the same, but will the total time be faster or slower? Before you go too far with this design, try doing some times up your test climb with identical gear and see what the error is.


Interesting point there, my times when doing repeats up this climb (Anstey Hill) usually hover in the low tens, 10:05, 10:02, & 10:30 :shock: (I got tired :lol: ) however I don't keep it in the same gear etc. For the testing I was considering a motor pace up, however that has issues again :(
I write the weightweenies blog, hope you like it :)

Disclosure: I'm sponsored by Velocite, but I do give my honest opinion about them (I'm endorsed to race their bikes, not say nice things about them)

istigatrice
Posts: 849
Joined: Sat May 12, 2012 8:32 am
Location: Australia

by istigatrice

Leviathan wrote:I agree averaging will remove some of that bias but I wonder if youre just going to be left with noise as opposed to any specific conclusions?


I was thinking the same thing, however if I am just left with noise, wouldn't "crank stiffness does not have a major effect on speed" be a specific conclusion?
I write the weightweenies blog, hope you like it :)

Disclosure: I'm sponsored by Velocite, but I do give my honest opinion about them (I'm endorsed to race their bikes, not say nice things about them)

Epic-o
Posts: 621
Joined: Tue Nov 20, 2007 7:18 pm

by Epic-o

verycreativeusername wrote:
I've seen the fairwheelbikes crank test and that's the inspiration for this study. Obviously a FEA will be much more accurate than what I am doing, but I would like to try and quantify this in the "real world" (not dismissing the FEA). If the accuracy of the power meter is the only thing that effects the speed, I believe a valid conclusion is that crank stiffness does not have a noticeable effect on speed. Is there any reasons why that isn't a valid conclusion?


Below the resolution of current powermeters doesn't mean that it doesn't have an effect on speed. I appreciate the work done by the Fairwheel guys but I don't think that a simple calculation of strain energy is the proper way to analyze the effect of crank stiffness. More complex models and a better understanding of energy paths is needed

spud
Posts: 1266
Joined: Sat Feb 07, 2009 5:52 am

by spud

I suspect you will have a difficult time as the sampling rate of the power meters will not allow enough resolution to capture differences throughout 360 degrees. As I understand it, Powertap captures once per second.

As we know, the generation of power during the pedal stroke is not a constant - torque changes throughout the rotation. I think stiff cranks feel better because they don't deflect as much during that peak torque, potentially allowing you to capture higher torque values during that .125 second interval (90 degrees rotation at 120 rpm).

I'd look to use a pedal based system with a super high sampling rate which can measure loads parallel to the instantaneous spindle vector, and develop some efficiency ratio between that composite, and power at the hub. I'm presuming that if there is a difference which can be measured, it is only at a very high output level.

User avatar
strobbekoen
Posts: 4426
Joined: Wed May 18, 2005 6:24 pm
Location: BELGIUM

by strobbekoen

Power meters and a field test will have too much variation in your data, exceeding the actual differences.
Would be fun to try, but I doubt the data will show any valuable information for crank stiffness.

verycreativeusername wrote:I was thinking the same thing, however if I am just left with noise, wouldn't "crank stiffness does not have a major effect on speed" be a specific conclusion?


No. Your environment variables change also for each test.

User avatar
kbbpll
Posts: 494
Joined: Wed Oct 20, 2010 5:56 am

by kbbpll

Then do it again with an elliptical chainring.

jooo
Posts: 1510
Joined: Wed Feb 02, 2011 3:48 am

by jooo

I don't understand why you are thinking of doing this in the field? Why not set up a bench test of some description? Surely you'll have far greater control over a large number of variables...

User avatar
djconnel
Posts: 7917
Joined: Wed Sep 26, 2007 1:57 pm
Location: San Francisco, CA
Contact:

by djconnel

First off, EXCELLENT! I really want to see results of this sort of test.

First, people advocate the strain energy approach. This isn't good enough: energy going into strain can be released in the drivetrain or into other reservoirs, including heat. Juanmoretime asserts, and I agree, that most will be released into the drivetrain. This will result in a delay in application of a few watts, but the watts aren't lost. The whole point of the experiment is to see if this is true.

For power, you won't get dissipation upstream of the measurement. So a powertap would be useless. By the time power gets to the wheel, it's going to move the bike forward. Maybe powertap for testing spoke flex, not crank arm flex. Same deal with the spider: if you lose energy in the crank arms, the spider won't even pick that up. So you want to use a pedal-based power meter. Even then, if there's a difference in internal body dissipation, you won't pick that up. You'd need to measure metabolic indicators like CO2 exhaled to detect that. But at least the pedals will tell you something about the effect of crank flex.

By the time you're ready to go, you'll be able to use a Garmin Vector, hopefully, so that's what I'd use.

You can use the powertap to compare with the pedal-based system. They should differ more if there's more transmission loss through the cranks.

I'd do the test on a trainer with a huge flywheel or even better, a treadmill. Most power meters work better when cadence doesn't vary much through the pedal stroke. That implies a high-inertia state, with round chainrings.

by Weenie


Visit starbike.com Online Retailer for HighEnd cycling components
Great Prices ✓    Broad Selection ✓    Worldwide Delivery ✓

www.starbike.com



Post Reply