I think another useful way of looking at it is most of the energy stored in the frame is returned along the peak force vector. If that peak force vector is tangential to the pedal arc (as it is when the peak force is applied at the 3 o'clock pedal position), then that work can be returned into the drivetrain. If that force vector is perpendicular to the pedal arc (6 o'clock pedal position), it is not returned in a useful way.

I believe the only way to know for certain, is to do a test on a sloppy frame versus a stiff frame. Use the same Look power meter pedals, SRM crank power meter and Power Tap wheel together on each of the bikes, do a number of flat out sprints and compare the different reading from each power meter unit between the stiff and sloppy frame.

Sounds like a nice job for bicycling magazine.

Other than that, we only can debate and guess.

Until then, I can do with bigger stronger leg muscles. :)

As I believe was discussed earlier in this thread:

When the frame is deformed, the action of the return does not go to the drivetrain (as noted). What it does is lift the opposite leg. Since your legs are not weightless, you save a small amount of energy when the frame lifts your leg for you. You also aren't using energy in the return. So, the energy isn't wasted.

We should also note that with almost any modern frame, the amount of deformation is very small; some frames are already at maximal stiffness, around 70 N/mm for the BB -- meaning they are stiff enough that no human will produce any measurable deformation. The amount of energy lost to friction is also quite small.

So even if that 90s era bike does somehow result in a small loss of energy, the difference in modern frames, especially those designed for racing, is going to be negligible in terms of power transfer.... no matter what the marketing materials claim.

Not to be rude, but I don't think we need a rehash of something which can be accessed with a click. I've seen your point, I don't agree, I've given my reasons earlier.

That type of approach was already discussed.

Some people (notably Mr Ratliff) categorically reject the idea that such empirical testing is valid.

I'd say that the difference is probably so small, it's with the margins of error of most commercially-available equipment.

We could go another 20 pages, and still not come to a conclusion. You've been warned. ;)

What about heat?

Close. But not a categorical rejection at all.

I just think the large number subtraction problem will make the testing extremely difficult to control. 300W (+-2%) in, minus 298W (+-2%) out equals... what? Unless you have a serious effort at calibration, you can get anything from 13W loss to 10W gain through the drivetrain. You'd have to spend a lot of time convincing me of your calibration technique to make any experimental numbers make sense. You would probably have to spend a lot of time with your powermeter on an independent dynamometer.

Nobody's convinced me they've taken their experimental method seriously. They think they can just slap a powermeter onto two different frames and come up with numbers that mean stuff. I don't think they can because they haven't displayed any knowledge of the testing problems with their method.

How about this then. Assuming for the sake of argument that the flex returns energy to the drivetrain propelling the bike forward, build a machine that flexes the frame, mimicking how a rider would flex it, and see if that moves the bike.

Sure, why not.

Any flexing is going to generate more heat than a theoretical non-flexing frame would. You could probably calculate the difference if you had enough data, but never measure it. It's an interesting thought experiment, but probably has no real world practical use unless the frame in question is so comicly flexible that it has physiological affects.

So if you are using a non-optimally stiff frame, you are contributing more to the heat death of the universe than you need to be. The universe wants its tiny fraction of a second back.

...have you guys really been discussing this since 2007 ?

No there were several iterations before then.

It's like the cycling version of The Song that Never Ends
https://www.youtube.com/watch?v=9ffL573XI50

http://1.bp.blogspot.com/--EpkZv9qsU...2011%2B9-7.jpg

:D

Let me put it this way. If an additional 20 N/mm of flex was going to produce (note: made-up number) a 5 watt difference, I'm fairly confident you could devise a reasonably accurate test using a single power meter.

Even if amateurs lack the discipline and/or skills to run a convincing test, bike manufacturers and independent labs ought to have the requisite expertise and the resources to run proper tests. Yet what do we have? Relatively simple deflection tests, with no actual quantification of watts saved.

IIRC you've pointed out that manufacturers wouldn't want to release such proprietary data. But this is belied by a) the fact that they do, in fact, release exactly that kind of data WRT aerodynamic tests, and b) independent testing labs have no such incentives to zip their lips.

Plus, as already noted, this debate has been going on for years. And yet, there is no proof.

As a result, I'm highly skeptical that frame stiffness has any measurable effect on power transfer. I strongly suspect that all it really does is improve the handling, change the ride feel, and *cough* stroke the ego. ;)

A single powermeter with two frames is still a big number subtraction problem.

But the bigger point is that any deflection will show a power loss, and it is still best for power transfer to have a stiff frame. If there is no local minimum you are trying to hit, there is no reason to quantify any property other than stiffness. In other words, if stiff is better than less stiff, than the objective is to make the frame stiff. I don't think you need a powermeter to make this argument. At best, you can say that "less stiff" is no different than stiff, and most don't really believe this.

...so all the "stiffer is better, TWSS," jokes have already played out, I guess. Ah well, late to the party again.

Really? A 2% margin of error voids any experiment? What's an acceptable margin of error for such experiments? What about wind tunnels, don't they have this same issue? Won't the interaction of a rider's position and/or body shape increase the margin of error for aerodynamic tests? Or, if the difference between a flexy frame and a stiff frame is less than the margin of error of a typical high-quality power meter, then why should we care about that particular quality?


But you have no proof that "any deflection will show a power loss" -- hence the problem. People have been discussing this for years and no one has a single shred of proof. Why should I accept this claim, when I have a perfectly good explanation and demonstration (the bucket test) that the return of the flex raises the pedal?

Or: How many watts does it actually take to deflect some of these frames? According to 3rd party tests, with a 100lb load the BB of a Roubaix flexes .838mm, and a Domane flexes .686mm. How much power are you really losing with that 0.152mm difference? Am I really supposed to believe that such a tiny difference is going to have a measurable effect on power transfer, let alone be noticeable by a human rider pouring imprecise and varied amounts of power to a drivetrain?


Thanks for the tautology. ;)

Again, there are valid non-power reasons to want a stiff frame, depending on your uses and objectives -- and a recognition that modern racing frames are almost all equally stiff, no matter what the marketing says. At the same time, there are perfectly good reasons to make fun of guys spending all their time measuring stiff tubes. :D


I beg to differ. ;)

The big number subtraction arises when the result of the subtraction is on the same order as the uncertainty in the measurement. A 2% uncertainty is certainly fine. A 100% uncertainty is not okay. If you subtract two large numbers, each with an uncertainty of 2%, and the expected result of the subtraction is less than 2% one of the large numbers, then you have a problem.

Physics is physics. If you need your own rational for deciding what physical reality is, I suggest you go out and get it. Don't be like a 2 year old and demand people do things for you. I like this topic because it is interesting to me. But I am not really looking to change your mind about things.

First, there is quite a bit more than 100lbs on the pedals. Second, there are other things than just the bottom bracket that flex under power. Draw your own conclusions from these observations. And also, humans are extremely sensitive to small changes.

Again, not trying to convince you to buy a competition bike. There are other reasons for cycling, yes, and there are other variables that trump power transfer for some applications. Buy the bike that suits your needs.

_____________________________

BUT... don't try to argue that an indirect power path (routing energy into the frame and back out) is equivalent to a direct path (routing energy directly from the crank to the wheel through the chain). At best, the indirect path doesn't penalize, but there are likely some situations where it is a penalty. And in competition, if it matters a little, it matters.

Ok, put gobs of carbon onto the crank area to keep it plenty stiff. Then shave off the thin walls everywhere else on the frame.

That's what most bikes look like now.

I don't actually believe that the bike will move at all without force on the pedals or other impedance on the rear wheel, except maybe a bit from the slight momentum of the cranks if the flex is just right to spin them. Which would disprove the "spring returns everything" theory.

But that's the problem with thought experiments - they aren't real unless someone actually does it.

FWIW all the Tacoma Narrows bridges (including Galloping Gertie) were/are suspension bridges.

Some times my frame is so stiff I can barely get out of bed. This does not make me a faster rider. Which leads me to my point; The rider neuromuscular application of power through a flexing joint and bone structure is so variable as to render the frame question somewhat moot (this is likely a pretty easy bio mechanical engineering test as in establish some sort of flex measuring, use accelerometers, strai. Gauges, power meters & hit the rollers and likely has been done...controlling for the rider is tougher). Someday, somewhere, a rider will appear (on a pale horse?) with a mechanical heart and carbon fiber legs and all hell will break loose.

What happened? So many names no longer around.

Stiff is better....just ask my wife.

was this ever done?

i was going to wait 2 months to complete the 2 year anniversary but I just couldn't help myself.