The Modified Hull Oval Chainring

[willhs]

This is my first real post – I hope it’s insightful. After Bradley Wiggins won the TDF, I though it’d be fun to play around with ovalized chainrings. Anyway, I stumbled upon this paper on optimal chainrings:

http://www.noncircularchainring.be/pdf/Biomechanical%20study%20chainrings%20-%20release%202.pdf

The upshot of the article is on page 33, which states that the Osymmetric has a 0.67% power loss compared to a traditional circular chainring, but if it is optimally set up it can give a 2.5% improvement in power transfer. Most notably, the “best” configuration was one that was never in mass production, the Hull oval. And, in fact, it’s better if it’s in a different orientation. I searched around the entire internets, and no one had ever even tried to make a copy of it.

So, I had to. I looked at various options, and in the end I had Fibrelyte make it for me. At the time, the exchange rate was very favorable and the whole thing cost $140. Not bad for a one-off custom carbon part.

And here it is:

It’s roughly equivalent to a 55 tooth chainring during the power phase, and a 36 tooth chainring during the dead phase. Yeah, it’s crazy.

So, then I put it on my bike and tried it. I’ve never tried any of the other oval chainrings, so I can’t say how those feel, but I can definitely say that on this ring you “feel” like you are cranking during the power phase, but the dead space feels quite easy. But, after about half an hour it felt pretty natural. Before it felt natural, I figured out the fatal flaw with this chainring design: chain drops. As you can see from the picture, there is essentially a very long plateau on each side. As the chain comes around to meet the chainring, essentially the entire chain has to slap down on that plateau. Therefore, you need a pretty straight chainline – meaning, only the middle 3-4 gears can be used on the rear cassette. Apparently, in that paper they did not think of “will this chainring actually be usable on a bike?”

Fail.

Ultimately, that’s why I didn’t post about this sooner. The chainring can’t really be used unless you want to have a 3-4 speed bike. Maybe for a really flat time trial. Otherwise, some further modifications would need to be made, probably adding XX1 style tooth profiles and EGGring style ovalization. I figured I should probably post my “negative results” in case someone else comes along and wants to make the “best oval chainring.”

At least, I hope this provides entertainment value as one of the goofiest chainrings of all time.

[GreenPlease]

Very cool chainring in any event. Good to know Fiberlyte will do custom work.

[Morelock]

it looks cool at the very least

Something neat to play with.

[bobby11]

Might just the slightest bit of curvature added to the flat section help the chain problem? At least that way, the chain meets a progression of teeth one at a time rather than slapping down on a row of them at once. That might allow it to handle a bit more of a steep chainline. I get that it might compromise the principle a bit, but if it makes it workable, might be worth it.

[Cody_Beals]

Very cool. I love seeing tinkering like this.

Too bad about the shifting. Maybe this could still have applications for track cycling?

[willhs]

Well, I think if I rounded the edges it would essentially be an Osymmetric … But the orientation would be more optimal. I definitely think this ring would work on a track bike.

[DarkSpeedWorks]

Great attempt. Looks interesting. Any pics of how it mounts on your crank?

[Xavier]

The upshot of the article is on page 33, which states that the Osymmetric has a 0.67% power loss compared to a traditional circular chainring, but if it is optimally set up it can give a 2.5% improvement in power transfer. Most notably, the “best” configuration was one that was never in mass production, the Hull oval. And, in fact, it’s better if it’s in a different orientation. I searched around the entire internets, and no one had ever even tried to make a copy of it.

Unfortunately the paper is bogus because they assume constant pedal speed throughout the pedal stroke before they do the inverse dynamics and attempt to optimise for non-circular chainrings

Xav

[willhs]

@gabbiev: I definitely agree that Fibrelyte is incredible on service and support. I would recommend them as well.

@milesthedog: Yeah, I looked at that back when this all started, but the weight is just too much.

@DarkSpeedWorks: I took pictures back then… this was a little over a year ago… but I can’t find them now. It looked pretty crazy.

@Xavier: You make an interesting point. They performed the study assuming a constant pedal speed and constant force (see page 5). I think, if anything, this would underestimate the benefit. At a steady state (like a TT), I would think the pedals are pretty close to a constant velocity, but if anything the “power” phase of the stroke is where more of the power is delivered. This model doesn’t take that into account, even though the chainrings are obviously setup to utilize the power phase of the stroke. How would you set the experiment up differently?

[cwg_at_opc]

solution: 14spd Rohloff rear hub
.

[mauricemaher]

Could you PM me? I would like to buy it off you for testing.

Cheers,
Maurice

[willhs]

@cwg: My understanding is that internal hubs lose more than 3% in inefficiency compared to derailleur systems, so it probably isn’t worth going that route.

@Tom: Sorry, it has been a while since I read the study, but I just went through it again and it doesn’t assume a constant pedal velocity, it assumes a constant rate in terms of RPMs, but if you look on page 10, they are taking into account changes in velocity of the pedal. The model they use has been used on two other studies as well, and it is available for use in Wolfram CDF here. You have a great blog, btw.

I’m not suggesting that this study “proves” that oval chainrings are effective. They created a model to predict the theoretical optimal shape, sort of like using CFD software to predict the optimal wheel profile. Ultimately, the prototype needs to be created and tested in the real world to prove anything.

@maurice: Sure.

[Xavier]

it has been a while since I read the study, but I just went through it again and it doesn’t assume a constant pedal velocity, it assumes a constant rate in terms of RPMs, but if you look on page 10, they are taking into account changes in velocity of the pedal.

They assume a constant pedal velocity for round chainrings, and then use that assumption to optimise pedal velocity for a non circular chainring, at the same rpm. It may well be why it’s never become a published study, as any peer review process would flag that up and reject the paper (at least I would if asked to review it)

Xav

[willhs]

@Xav: Again, my point: when pedaling at a steady state on flat ground, the wheel, chain, and pedals should be going at a constant velocity if the chainring is round. So, their approach isn’t unreasonable. They have to find some way to directly compare 2 different pedaling systems. You are free to modify their model and test your own approach, if you think it is better. I’m still not sure how you would do it differently. If you have relevant background, it would be cool to hear your thoughts besides “I would reject this.”

A similar model was used in several peer-reviewed articles on non-circular chainrings, such as this one which showed a similar 2.7% improvement. (see the citations in that paper for a longer list of published, peer-reviewed articles). Again, I think these studies are neat for finding a theoretical “best shape,” which should be followed by more testing in the real world.

[Xavier]

@Xav: Again, my point: when pedaling at a steady state on flat ground, the wheel, chain, and pedals should be going at a constant velocity if the chainring is round. So, their approach isn’t unreasonable. They have to find some way to directly compare 2 different pedaling systems. You are free to modify their model and test your own approach, if you think it is better. I’m still not sure how you would do it differently. If you have relevant background, it would be cool to hear your thoughts besides “I would reject this.”

A similar model was used in several peer-reviewed articles on non-circular chainrings, such as this one which showed a similar 2.7% improvement. (see the citations in that paper for a longer list of published, peer-reviewed articles). Again, I think these studies are neat for finding a theoretical “best shape,” which should be followed by more testing in the real world.

Again, the pedals don’t go round at a constant velocity with a round chainring, here’s an example compared with the pedal speed of an Osymetric chainring (90rpm, 330w, ~50kph)

Similar with the original MetriGear data too:

Also http://www.ncbi.nlm.nih.gov/pubmed/18178104 :

This is true both outside and in - whenever it’s been measured and not modeled it’s been found to be the case. I’m not saying it’s a bad idea to explore, just bad execution in this case. If you assume constant pedal speed at 340-20deg and 160-200deg then the modeled input from the usable musculature will be overstated (as pedal speed will appear higher than reality)

Xav

[QKRTNU]

Track bikes do not have a derailleur and therefore can’t adapt to differing chain lengths. There are no ovalized rings that can be used on the track…

Don’t try this at home on a track bike.

[willhs]

@Xav: As you can see from that metrigear data, there are very small differences, and not necessarily coordinated with the pedal stroke. Some revolutions are at a relatively constant speed, some are varied. It would be rather difficult to simulate that. The study I just linked is actually very thorough in its discussion. It is a very good tour of the studies to date, and it addresses your point. Quote from page 1499-1500:

Another potential limitation is that we prescribed the
crank angular velocity similar to that of an isokinetic
ergometer, and therefore drive train dynamics were not
included in the model. This was necessary within our model
and optimization framework to assure the velocity profilesof
the optimal chainring shapes. However, previous studies
have shown that the crank inertial load has little effect on
pedaling coordination (Bertucci et al., 2005; Fregly et al.,
1996) and Soest and Casius (2000) successfully reproduced
experimental data while prescribing the motion of the
crank with a similar pedaling model. Thus, the isokinetic
nature of the present model would appear to have minimal
influence on the optimal chainring shape.
@Tom: This study also addresses your point – they discuss fixed hip location, also on page 1500. They discuss their model on page 1495, and it appears that the ankle is not fixed.

[Xavier]

@Xav: As you can see from that metrigear data, there are very small differences and not necessarily coordinated with the pedal stroke.

I disagree. What about the other data?

I also disagree with the Rankin/Neptune assumption on constant velocity throughout the whole pedal stroke - it’s far easier to model which is why people do it.

Xav

[QKRTNU]

I assume that was supposed to be in pink…

[damon_rinard]

Hi QKRTNU,

Most non-round rings can work on track bikes. I’ve done it with Rotor Q-rings and Shimano BioPace rings.

Will they work on my MTB singlespeed, or track bike?
Yes. MTB Q-Rings can be used for MTB singlespeed applications. We’ve also successfully tested and are currently producing select models of 144 BCD Track Q-Rings.

That’s from Rotor’s FAQ here: http://www.rotorbikeusa.com/support.html

Cheers,

Damon