Not sure if this was posted before but seems MIT did a comparison on vibrations of frame based on materials https://web.mit.edu/2.tha/www/ppt/Bike-ISEA.pdf

Seems AL transmits more vibrations than carbon and steel transmits the least. The reverse is that more energy is needed to speed up riding steel and AL requires the least energy. Interesting as i see in other posts that frame material makes little difference and is only marketing gimmick

Yes, so the question is are those measure differences significant in a real riding situation? That is where the other articles say, "No."

Interesting looking study, but with one significant flaw. Each material is represented by one embodiment, one bike. Is that the stiffest that carbon can be? Is that the absolute best design for aluminum to work with its strengths and minimize its weaknesses. One bike of each kind could never make a valid test. Better to see the differences between multiple bikes of the same material.

The middle bike is an aero design and doesn't compare well to the other two, just design-wise. Basically a useless experiment. So-called scientists should know better. Heck, even engineers should know better.

I'm a scientist (Astrophysics to be precise) and I love looking at stuff like this, especially for the red-flags and supposed "trends" that aren't there. For example, take a look at this page:



Those three graphs show a supposed "trend", with the blue lines are the errors in the measurement. Not only do they deceive by scaling the graphs from not-zero to show the maximum apparent difference, but if you believe their error analysis is correct, the fact is that all three of those graphs show no measurable differences between any of the frame materials in any of the tests.

What would an aero design mean to an indoor experiment, that makes no sense as there is no wind except your hot air? All three bikes are the same manufacturer also

As a scientist you do know the red bar is the average with the blue bars as the high and low values... And the high and low values are discarded. LOL jokers, the study is from mit where did you go to school.

No I don't know that actually, and that is not how I would interpret those graphs if this were a talk given in my field. Why do you assume that is what they are out of interest? Those look to be error bars, not "hi/lo" measurements, but then again, I'm not someone who works in "sports innovation" so I don't know what the standards are in that field.

Even if these are "max/min" rather than 1-sigma error, those graphs do not show an obvious trend. Where do they quote the results of statistical test that proves the trend exists? Answer: they don't, they show some tweaked graphs and state a trend exists, with no proof that it actually does.

Is there a published paper for this study, maybe they give more details there?

FYI I'm not saying that the trend isn't there, I'm just saying they haven't demonstrably measured it.

Why the **** is that relevant exactly? Do MIT studies get a pass on not actually proving their results because they're MIT?

The error bars are greater than the measured differences, aren't they?

Although the graphic (incorrectly) refers to a "trend of requiring less effort", the discussion bullet point is "no significant difference in rider performance" was measured.

Hey, take it down a notch dude. You opened up the conversation, and you got someone's opinion. Let's discuss, but don't get all huffy about something you have no understanding of.

No, the aero performance of the aero frame is not the problem. That wide, flat aero downtube imparts crazy stiffness to the frame, especially at the head tube and bottom bracket. It affects comfort, power transmission, road grip, tons of stuff. Way beyond the material effect.

Ph.D. from Princeton, thank you very much. And it wasn't in Fine Arts.

****

Heeere we go.

How soon before someone asserts MIT doesn't know what they're doing?

Oh wait, we already got that.

Given that the primary objective was to quantify vibration response of different frames, I don't think that any of the jumps you guys are making in assuming what it means about material performance in general, are valid. It's clear the scope of the test was very, very narrow, and not intended to do anything other than speak to the ability to 'see' (i.e. test or demonstrate) different vibration responses in frames.

I find this study to be essentially meaningless so far. If you want to build an Al frame that's comfortable you make it with relatively small diameter tubing and slightly thicker walls (such as the old Vitus and Alan frames). OTOH, if you want an Al frame that's stiff at the sacrifice of comfort you make it with large diameter tubes and thinner walls (such as the early Klein and Cannondale frames).

The Vitus/Alan aluminum frames were touted as being more comfortable but relatively 'whippy' compared to typical steel frames of the time. So Klein, and shortly thereafter Cannondale, designed their aluminum frames in the opposite direction - for high stiffness. But they then got the reputation for being harsh. Such design choices can be far more significant that the inherent material properties of Al vs. steel vs. carbon fiber.

This test just looked at three individual Cervelo frames and their conclusions only apply to the design choice made by that one manufacturer in making tubing sizing and shape decisions that influence the stiffness vs. comfort tradeoff.

The study objective appears to be to eventually (with much more extensive testing) see if performance is impacted by vibration levels and I think it may be able to give some insight in that area. It could also give valuable data about the vibration characteristics about specific bike frames or other components. But it doesn't seem to be properly designed to evaluate the effects of inherent material properties of different frame materials, nor is that listed as an objective of the study.

Another weird aspect of the presentation is that the authors imply that differences in rider performance (if they were to exist) would be the result of differences in transmitted vibration through the frame. In fact both vibration transmission and differences in rider performance would be two dependent variables that might be correlated, but without any hint of a cause-effect relationship. There are many other dependent variables that could be measured with regard to the three frame materials. At this point who could know which of these would be the cause of performance differences. Frankly, as a scientist, this study is embarrassing to me. How could folks with such credentials get things so wrong.

Medieval Supply Chain Operations?



What kind of meat head uses error bars to represent the range of data? What a misleading [strike]chart[/strike] study.

I dunno what the degree was for. I'm too old to remember. Something about Methotrexate I think. Maybe a synthetic study to provide it in pure form. Just guessing though.

I watched "Good Will Hunting" last night. How do you like them apples?

Gotcha beat. I stayed at Holiday Inn Express. BUSTED!

Do you know what a hypothesis is? Do you know all hypotheses are not valid? And do you know how you prove or disprove an hypothesis?

We shouldn't be too hard on these results based on a Powerpoint presentation. The fact is that what's on the screen isn't necessarily the whole story, as the scientist has to stand up and talk about it too, and will often address points and clarify things that are not explicitly stated in the slides. This is why I would be interested to read an actual refereed paper on their work.

Yes, but the authors didn't know how. They are mistakenly suggesting that if two dependent variable were to correlate, then one would necessarily be the cause of the other. Whether they showed it or not isn't the point. The point is that it is a false premise logically. They didn't even look for the other dependent effects that could be at work there. Neither did they vary the levels of vibration transmission through a single frame to test the effect on rider performance. Ideally you would start with one frame and a machine that would impart vibration to it. You would vary the level of vibration applied. Consistent with the frame's properties that would vary the level of vibration transmitted. Then you would measure rider performance at each applied and transmitted vibration level. Finally, if you found a cause-effect between the vibration independent variable and the performance dependent variable, you could develop the "hypothesis" that transmitted vibration levels caused by frame differences (keeping the input vibration constant) would also cause differences in performance and test that with different frames.

This is horrible science pure and simple.

No, the point of the study was not suggest that, as you say above, "if two dependent variable were to correlate, then one would necessarily be the cause of the other." Did you read it? The primary goal was to show ability to quantify vibration response of different frames, and from there to hypothesize on energy cost impact of frame stiffness upon rider performance.

Nowhere did they claim or assume correlation implies causation, and in fact stated explicitly (in big assed, bold font) that "Single subject design precludes generalizationof this result "

You are wrong. On page 4 they say that they want to develop a system to characterize the EFFECT of transmitted road vibration in road bikes and to do that they propose to measure the vibration and rider performance. It is clear they are attempting to measure the EFFECT of the vibration on performance. That thing which brings about an EFFECT is a CAUSE. They obviously are proposing that road vibration may cause changes in rider performance. But they don't keep all the other variables under control by using the same single bicycle and applying different levels of vibration through it. They use three different bicycles and measure the vibration and performance as two dependent variables. They try to keep some other variables fixed, but it isn't possible to know every possible variable. It is a clumsy way to go about the study. First they needed to measure the effect of different levels of vibration through one specific frame (any one) on performance. Then they could have (follow up study) measured the differences in transmitted vibration through different frames once again at a range of transmitted vibration levels. If the performance effect for a given transmitted vibration level was the same on any frame and significant, then it would be possible to claim transmitted vibration as a cause of the performance effect. Alternatively, a statistical design of experiments could have been used to collapse the two experiments into one and to attempt to separate and quantify the many different equipment related causes of rider performance effects. Apparently that would have required an experimental skill level far beyond this team's capability.

Bottom line is that it is easy to design perform an experiment but quite difficult to design and perform a meaningful one.