Login to VeloClub|Not a member?  Sign up now.
Does bicycle frame stiffness matter
  • Jan Heine

    Thanks for posting this interesting podcast! For those interested to read more, the original post about the double-blind “planing” experiment is here: https://janheine.wordpress.com/2014/11/23/what-is-planing/

    • Il_falcone

      Jan, I read your post when you published it and found your findings very interesting, yeah even somewhat amazing then. Kudos for what you did then! This podcast hasn’t really offered more on this topic than you already did then. Which is a bid sad since at least in your tests the difference seemed to be quite significant despite Damon Rinard obviously having a hard time to accept that the effect can be so big.
      Although I know that this will be costly with regards to the actual costs and also the effort I think what we’ll really need is a test similar to what you have already done but with more riders who ideally wear mobile spiroergometry devices in addition to carefully calibrated power meters. Otherwise I think the sceptics and believers in stiffer = faster will always call it vodoo and will prevail preventing at least the mass manufactures from offering frames with tuned stiffness which you would liket them to do.
      As a side-effect there will probably also be some stiff (pun intended) resistance by magazines against your findings / theory since testing bikes has become so much easier for them since it has been widely accepted that you have to run them through a lab and do a (ever increasing) number of measurements in order to rate them. The “testers” at the “biggest European road bike” magazine that is also mentioned in the Podcast don’t really ride those bikes for most of the tests they do unless they really like a certain bike and are curious to find how it really rides. But with most of the test bikes you as a manufacturer get them back at some point with an absolutely clean like new chain and tires that have never seen any road surface and brake pads without any sign of wear or use.

      • Jan Heine

        You make good points. The big German magazine is trying to be objective, but unfortunately, this leads to simple numbers that are reproducible, but not necessarily meaningful. We approached the issue from a different angle, using the methods of medical research (double-blind tests rather than engineering measurements) to show that the rider’s power output can vary significantly when riding different frames.

        We showed that this effect exists and is real, and that some riders are more sensitive than others. Two of three testers study could tell the very slight differences between the test bikes in our study. The third thought he felt differences, but they didn’t track the different bikes. We then took the two testers who could tell the bikes apart and measured their power outputs under full efforts in sprints against each other. And we found that on the “faster” bikes, the riders put out more power. On the slower ones, they put out less, no matter how hard they tried.

        It would be fun to enlarge the study… One thing we could do is use bikes with greater differences – our third tester probably would be able to tell those bikes apart, too. But what does that tell us? Nothing new – it’s to be expected that the larger the differences between bikes, the more riders can tell them apart. That is why we kept the differences between our test bikes very small – to make it harder for the testers to tell them apart. It was a risk, but if they could tell them apart, we could show that even very small differences in frame flex can make a significant difference.

        Expanding the study to find out what frame stiffness works well for what type of rider would be awesome. Unfortunately, that would basically be an epidemiological study that requires hundreds of testers to have meaningful (i.e., statistically significant) results. I doubt anybody in the bike industry has the resources for such a study…

        I think what we already can say now is
        a) stiffer isn’t better, and
        b) that the right stiffness can be good, and it’s worth trying many bikes for yourself to figure out whether some work better for you than others.

        • OnTheRivet

          Double blind testing with 3 participants is the epitome of “bad” science, sorry.

          • Mark Hespenheide

            My advisor for my Master’s thesis (admittedly in Geology, not medicine or engineering, but bear with me) told me that ‘science advances at its margins’, which is to say that we don’t push the boundaries of what we know by more study of things that are well understood. Those studies lead to refinement, not jumps in advancement.

            Double-blind testing with three participants may not be the best science, but I think it’s “worse” science to dismiss it out of hand. A real scientist, when confronted with data that doesn’t fit their understanding, should think — ‘hey, that’s interesting; I wonder what we could learn about that.”

          • Jan Heine

            It all depends on what you are trying to figure out. If you want to show that it is _possible_ to discern small differences in frame stiffness, one tester is enough – if that tester has a positive result. It’s like proving that the Sasquatch really exists – you only have to catch one, not a whole herd.

            It’s different from a medical double-blind test, where some patients will get healthy even without doing anything. In our study, we had two “positive” testers and we ran enough tests that the positive results (identifying the different frames correctly) were not possible through pure guessing and luck. (The chance was less than 1 in 2000.)

            If you want to show that this matters to the majority of the population, then you need hundreds of testers. So it’s important to understand the limitations of the study – it showed is that _some_ riders are faster on more flexible frames, because they can put out more power. If we said “two out of three cyclists…” then you’d be correct in saying that we need more testers before we can say anything about the prevalence of the phenomenon in the general population.

  • Robert Merkel


    I don’t have trouble believing that bicycles are pretty efficient springs, but I’m not convinced by the hypothesis for the planing mechanism leading to higher outputs. If “lengthening the power stroke” is the explanation, could the same effect be achieved by drilling holes in your cranks to introduce additional flex? Has anyone tested this?

    Anyway, as far as the limitations of the original study, yeah, it’s not definitive. So in an ideal world one/some of the major bike manufacturers might considering ponying up some cash to conduct additional research with more people and a larger set of frames.

    • Wily_Quixote

      I have replaced my cranks with slinkys.

  • Wily_Quixote

    At the 30 minute mark much was made of a blind test measuring power. I wonder how power measured at the BB has any relationship to power lost in frame flex? Surely, power lost at the BB would be the sum of crank flex and friction losses in the BB.
    For example, a track sprinter is not slower on a full suspension DH bike, as compared to a track bike, because less power is measured at the BB on the DH bike it is because energy is lost in the suspension. The energy lost in the suspension does not feedback to the BB.
    I wonder how frame flex can rob energy at the BB when the energy loss is post hoc. A true test would be examining speed at a given power output at the BB and all other variables (cadence, temperature, tyres, wind speed, aerodynamics etc) being equal, or at least adjusted for.
    Unless I misinterpreted Jan at that point.

    • Jan Heine

      We didn’t try to measure power lost to flex. We tried to measure the riders’ maximum power outputs on different frames. The rider’s max. power output wasn’t constant, but varied significantly depending on the frame stiffness. Damon tried to quantify the power lost to frame flex by using two power meters – one in the crank and one in the rear wheel – but couldn’t find any power losses…

      • Wily_Quixote

        I think that we need to be wary of extrapolating this data too widely. If riders are exerting more power through a flexible bike than all it means is that they are choosing to put out less power through a stiff bike. It is a function of preference not physical properties of the frame material directly influencing how power may be applied.
        The frame properties can not influence the capacity of the rider to push on the pedals except in circumstances where suspension locks out the chain (i.e. the chain becomes a rigid arm), or in very obvious cases (i.e. a slingshot bike) where a frame has discernable (and deliberately engineered) flex along the long axis of the bike. I think that if your extrapolations were correct we would see the slingshot bike as the preferred bicycle for those non-UCI riders wishing to exert maximal power through the BB

        Given that both tested frames are still stiff (as compared to a FS bike or a bike designed for springlike energy return i.e a slingshot bike) I think that your experiment was not rigorous enough to draw firm conclusions from.

        if you assert that there is a springlike resonant property of more flexible frames to match cadence (i.e. even out the dead spot) this doesn’t explain how a ‘flexible’ frame matches different cadences, especially as this forward flex is not even detectable (albeit by the crude measurement detailed in the podcast).

        i am not a materials engineer but i think that even a bike engineered to return energy (such as the slingshot) had difficulty with riders propelling the bike in an uneven cadence (let alone lateral flex). So it is reasonable to assume that in ‘planing’ there is an optimal range of frame flexibility where a bike might be too rigid (as per your test bike) or too flexible (as per a slingshot) it seems curious to me that the flexible test bike has those exact properties at all cadences.

        I call this implausible with the data that you have shown. This needs to be tested much more rigorously before we can conclude that a flexible bike permits/encourages a rider to exert more power.

        • Jan Heine

          You are right – you need a certain stiffness, not too much, not too little. We did test one bike with a rubber elastomer in the rear triangle. We had three elastomers, plus a rigid piece to lock out the suspension completely. Only the middle setting offered great performance and “planing”. Too soft, and you got suspension bob, which robs energy because it’s out of phase with the rider’s pedal strokes. Too hard, and nothing moved…

          • Wily_Quixote

            I don’t think that this is the right experiment. Suspension bob does not mean that power to the BB is less, just that energy is lost in the suspension post hoc.

            Anyway,a mechanical lever operating a pedal on a test bike in a jig is the way to test it. If the mechanical lever cannot apply the same degree of mechanical power to the BB due to (lack of) frame flex on a stiff frame, or (in a flexy frame) if the lever meets resistance and/or the frame then flexes unduly because it is out of phase with the lever, I will believe that planing exists.

            In the experiments you cite, just because some riders don’t pedal as hard on a stiff bike does not mean that there is a planing effect. It just means that some riders don’t pedal as hard on a stiff bike. There are psychological and biological explanations for this, which is why you have to take humans out of the equation and hand this to a mechanical engineer who can look at frame resonance and application of constant mechanical effort through the BB at a variety of cadences in a jig that measures frame strain and flex.

            Until that happens this is all conjecture: you may be correct but I have no reason to believe you yet.

            • Jan Heine

              With suspension bob, I don’t think it’s so much that power is being lost in the suspension (i.e., the shocks get warm), but that it disrupts the rider’s pedal stroke. On the bike with the elastomer in the rear, pulling a trailer reversed the working of the suspension – instead of compressing with each pedal stroke, the trailer’s inertia extended it. It felt terrible, and it was much harder to climb with the trailer than when carrying the same weight on the bike. So it’s biodynamic and not engineering issues we are dealing with here.

              • Wily_Quixote

                I agree, power through the BB can not be affected by any property of the frame as any distortion (compression or stretching) of the frame complex is achieved post-hoc.

                Therefore, the riders in your study chose unconsciously to put less power through the BB.

                perhaps elite riders on stiff bikes have had sufficient neuromuscular training on stiff bikes to apply maximal power through stiff frames.

    • John Barry

      I am skeptical of your assertion that the suspension would reduce power at the bottom bracket. The points of contact on the bicycle are bars, pedals and saddle, and unless something is causing these points to move relative to each other, there should be no loss due to suspension. Yes, bouncing might disrupt a rider’s rhythm, but that would be an indirect variable, unrelated to what is being measured. (i.e., not frame flex) Bumpiness of the ride would be a factor to be controlled for in this sort of experiment, rather than being the focus of the study.

      In contrast, frame flex does cause the points of contact to move in relation to each other. It’s a significant distinction, imo.

      • Wily_Quixote

        If you read my post again you will see that I didn’t state that energy is lost on a suspended bike at the BB. I don’t think that this is physically possible.

        I stated that any energy lost at the BB is within the BB complex – i.e. friction and crank flex. I did ask Jan specifically:
        “I wonder how frame flex can rob energy at the BB when the energy loss is post hoc.”

        • John Barry

          I misinterpreted your 2nd sentence. Apologies. I do think this will be a bear for engineers to measure, as my experience (I have a 1972 Fuji Finest that sometimes feels as though it “planes” on climbs) causes me to believe there’s a dynamic interaction between riding motion and the feel of the frame flexing. Perhaps robots will become sophisticated enough to perform the analysis in our lifetimes.

          • Wily_Quixote

            No probs.

            Yeah, I don’t necessarily disagree with Jan – I just don’t think that there is sufficient evidence to assert a planing effect.
            Jan has ignored, so far as I can tell, all the human factors involved.

            I think that we should be clear that Jan’s work (at least as posted on his blog) simply proves that riders who apply more power through the bottom bracket climb faster – which agrees with standard physics (and is something that all riders know). he also found that riders who rode a more flexible bike chose unconsciously to apply more power as measured via a powermeter.

            There is no suggestion that frame properties allowed this power production (as power application is independent of the frame) and there is no suggestion that efficiency is increased or power output in other applications (steady state riding or flat sprinting) is affected by frame flexibility. The power production must have been a decision of the rider, albeit unconscious.

            These results have yet to be reproduced, so far as I am aware, nor reproduced without the human factors excluded. Which is why a test in a lab with a piston activating the crank instead of a human would reveal whether planing is wishful thinking or a mechanical advantage conferred by frame properties.

            • John Barry

              The phenomenon which I have experienced happens exclusively on climbs where I am able to sustain a steady cadence where I pull on the bars with each pedal stroke. Here is what I perceive to be happening. When I apply forward/downward force on the right pedal, I am pulling on the right side of the handlebar. This slightly flexes the frame, so that the distance between the right handlebar and the right crank axle decreases slightly. When the force that I have applied on the right side is relieved, the frame returns to its unstressed position. If I happen to be applying the force to the left pedal while this rebound is happening, a portion of the force of the rebound can be applied to the left pedal stroke. While this might not increase the actual power that would be measured at the bottom bracket on any single pedal stroke, the result is that I expend less effort – which makes the task of pedaling uphill noticeably more pleasant.

              • Wily_Quixote

                ‘the result is that I expend less effort’ – which means that you’ll be going slower.

                Jan’s research found that frame spring returns frame flexion energy from the lateral plane to the axial plane. This energy return is not massive but implies that energy lost to frame flexion is not lost as heat in the frame. Jan’s planing hypothesis is that this spring return is coincident with the pedal stroke which allows/leads/ prompts the rider to push harder on the pedal.

                In all cases if you are expending less energy you are going slower, because springiness is not meaningfully contributing to overall speed, and, even if it is, stiff frames still undergo this energy return – it’s just that the energy return is less because less energy is lost to the frame.
                If you hop off your bike it doesn’t ‘slinky’ down the road for another 100m.

                • John Barry

                  The expending less effort is relative to a bicycle that doesn’t flex – to achieve the same output (at least) at the bottom bracket. I really think this has more to do with the body’s efficiency than the bicycle’s. An imperfect analogy would be running with shoes. A pair of shoes that has more rise in the front than in the heel would be less efficient than a pair with the heel slightly higher than the front. The output measured by the resulting forward speed, or friction generated between the sole and the ground wouldn’t tell the whole story, because the body would actually have to work harder to achieve the same results wearing the low heeled shoes.

                  The frame that has a slight bit of flex in the “right direction”, at the “right time”, is like the pair of shoes with the slightly raised heels, and the bike with no flex, is like the other pair of shoes. I don’t think the concept is difficult or wrong, but proving it is the big challenge.

                  • Wily_Quixote

                    You’re not describing mechanical efficiency here but a biomechanical problem. If you have a raised toe running shoe it means that the foot cannot contribute as well to power application and the lower leg cannot develop enough force, because muscle power development depends on muscle length. This is not the same ‘planing’ effect that Jan describes.

                    With a ‘planing’ bike the problem is not that a cyclist can develop more power, or that in a stiff bike more energy is lost ( the stiff bike is no less efficient) it’s that they unconsciously choose to apply more power – at least the 3 riders in Jan’s experiment did when climbing.

                    It would be curious to see whether cyclists who always use stiff bikes under high power, I.e. track sprinters , have the same unwillingness to apply power on a stiff frame when climbing.

                    • There are clearly 2 issues here:

                      There is the mechanical question of how much energy storage/loss can occur in a frame, and does that stored energy get returned in such a way that it contributes to forward propulsion. I was doubtful that it would, although the FEA analysis on Jan’s blog which detailed not just the lateral flex but also the twisting action of the BB axis and compression of the drive side chainstay has given me cause to reconsider. This is a question that I think might be best investigated in a lab setting, as you suggest.

                      The second issue is a biomechanical one. My interpretation of what Jan is proposing is that the more flexible bike allows a well matched rider to put out more power. You seem to assume that the only way this could happen is through the rider choosing to put out more power, presumably due to some conscious or unconscious inspirational feel of the more flexible bike. I would propose that there is another possibility here though, which Jan alluded to when he was talking about the sensations the riders experienced in their legs, which might be better termed the muscular Rate of Perceived Exertion (RPE).

                      He suggested that the riders legs felt less fatigued relative to their cardio RPE at the end of a long high-intensity effort when riding the more flexible bike. This effect could be caused by a lengthening of the power phase with a lower peak but similar area under the curve (AUC). Since peak contractile force may be reduced due to the deflection of the frame, the rider may experience a reduction in muscular fatigue for a given power output, in a manner similar to pushing a smaller gear at a higher cadence vs. a big gear at a lower cadence. Power output may be the same, but biomechanical efficiency, and the sustainability of the effort can differ greatly. This could mean that the riders are able to increase their power output further on a more flexible frame, not because they simply feel inspired by it, but instead because the “rounding” of the power spikes keeps their legs from becoming exhausted.

                      If the frames flexibility was net neutral from a mechanical perspective, or even if a more flexible frame did indeed lead to a small energy loss as conventional wisdom has long suggested, then the more flexible frame could still be faster if the biomechanical gain due to the reduced peak contractile force was significant enough.

                    • Lyrebird_Cycles

                      Late to the party but there’s another possibility: It has long been known that the efficiency of a kangaroo’s bipedal motion is influenced by the energy recovery from the elasticity of the Achilles’ tendon. Similarly running on a track tuned to the rebound sequence of the runner’s footstrike can be more efficient.

                      A secondary effect is that the energy absorbed and released by the tendons can play a role in ensuring that the sarcomere shortening velocity remains in its optimal range: see https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3130454/

                      The tendons and the moving mass of the body part (in this case the leg) form a resonant system with a characteristic frequency: you can test this yourself by sitting in a chair with your forefoot on the ground but your heel raised and “bouncing” your knee vertically. You will naturally do this at one frequency and if you consciously try to raise or lower that frequency the bounce will reduce. Try it with the other leg, the frequency should be very similar. Now stand up and “pogo” on both feet, again balls only. The frequency is lower because the ratio of mass / spring constant is higher.

                      It is entirely possible that “planing” is simply a description of the situation where the resonant frequencies of the driven part of the bike / rider system happen to align with the driving part.

                    • Wily_Quixote

                      I think that this is jan’s contention. I wonder how a bike can be ‘tuned’ to different pedalling frequencies, though. What happens if you are spinning fast and the spring return is out of phase with your cadence?
                      How does the flexible bike know that you’re spinning tempo on the flat or wrenching the bars in a sprint or climb?

                      I am not a mechanical engineer but it sounds dubious to me that resonance could occur at the convenience of the rider at any possible cadence.

                    • Wily_Quixote

                      Interesting hypothesis, but I wonder if this is possible that the few mm of longitudinal flexion in the frame could lead to a muscle group reaching peak contractility at a point less than its optimal point.

                      This small distance is likely to be the same, or less, than the change of position that the test riders would have experienced on the test frames.

  • Ted Durant

    When a boat planes the coefficient of drag is substantially reduced, so the boat goes faster for a given amount of thrust. Nothing like that is happening on a bicycle. A better word to describe the phenomenon Jan identifies comes from another boating term, “swing”, which rowers use to describe a rowing shell whose flex characteristics synchronize with the rowing stroke.

    • James Huang

      Jan has fully acknowledged that the “planing” term is far from an ideal analogy. Ultimately, all I believe he’s trying to convey is some sort of physical state a bicycle can achieve (in this case, some sort of harmony between frame flex and a rider’s pedaling output) whereby speed and efficiency are optimized.

  • Ronald Hillberg

    Ask a pole vaulter. It’s matching the right stiffness and the right frequency of return to the riders weight and riding style to get the right outcome. I don’t like when they say flex doesn’t use energy. If i go flex a bow I will get tiered quickly. The energy isn’t lost in the bow but in my muscles. The body isn’t very elastic but with the right setup it can be efficient and even if the energy is close a more forgiving ride will be valuble.

  • Jonas/BETA

    Interesting discussion., But I think people are always looking at the wrong place when talking about energy loss because of stiffness. The energy lost in the frame is minimal, since the damping/hysteresis in most frame materials is very small. So no surprise Damon did not measure any significant loss of power between crank and hub. The question remains, what is happening with the energy that went into deforming the frame? If we are talking about BB stiffness, the energy mainly comes back as a lateral force (integrated over lateral deformation, to be correct), which is not very useful for our propulsion. I am quite sure it is our body acting as an (active) damping element in the system, so if you want to look at energy loss, you need to look into the body.
    Vertical deflection on the other hand could maybe have a positive effect on power output, virtually allowing for a longer pedal stroke. So I am wondering whether the positive effects felt in Jan’s experiment were not coming from vertical deflection rather than BB stiffness. When I was working for BMC, I went from the GF01 (good BB stiffness, vertically very soft) to the TMR01 (very stiff in all directions), and the TMR always felt inefficient to me. I was quite sure it was too stiff. But then I rode the SLR01 (even stiffer than TMR01, but good vertical compliance), and it felt like the most efficient bike ever.


Pin It on Pinterest

June 24, 2017
June 23, 2017
June 22, 2017
June 21, 2017