About frame stiffness – what is stiff?

During the many discussions with several frame stiffness testing centers and providers my understanding that frame stiffness tests are largely done in a somewhat inexact manner, mostly lacking any statistical validity, was reconfirmed. A quote from one engineering lab in particular sums up the situation nicely:

“And, to be perfectly honest, the bicycle industry has not been particularly demanding or even interested in a higher level of scientific rigour…”

So with that in mind I decided on our future steps which will produce highly credible, repeatable results. For the present however we decided to fall back to the magic of statistics and cohort analysis. We have stiffness data from one of our manufacturing partners and they happen to make several brands of frames that were extensively tested by the Tour Magazin over the years. Helpfully, Tour data shows useful consistency over time. We also know our own frame stiffness data since each of our carbon frames is tested for stiffness as part of our basic QC procedure. Thus with the other brand’s frames serving as a cipher key to correlate Tour test data with our own test data to within 5% accuracy (confidence interval actually) we can now present a comprehensive list of published Tour data for 19 competitors, and for three of our frames: Velocite Magnus, Velocite Geos, and Velocite Selene.

Table 1

The data in the table was sourced from Tour Quarterly online publication, from Tour Magazin print edition for November 2011 and April 2013. Only the data for German market specific online brands was excluded to increase relevance and chart legibility. Some current frame models are missing from the chart, but that is due to lack of Tour test data, not deliberate omission.


Chart 1 shows torsional stiffness as measured at the head tube, ranked from lowest (BMC Racemachine RM01) to highest (Velocite Magnus). This is the traditional way of ranking stiffness when other manufacturers talk about how stiff their frames are. Torsional stiffness at the head tube influences how well the bike will steer, all other things being equal. We can see that the best regarded brands (Specialized, Cannondale, Cervelo) all rank pretty high which is good to see as it shows that there is a degree of competition in the market based on product performance. It also shows that at the high end of the market, riders may get what they are paying for.

Coming to Velocite products, Velocite Magnus exceeds the torsional stiffness of the second stiffest frame by 17% while the Velocite Selene ranks third and Velocite Geos comes second from the end in a field of what are regarded as top end frames.

However as the second chart will show, this is not the entire story. What is not normally emphasised by the bike brands is stiffness at the bottom bracket, the pedaling stiffness. This measure describes how efficient the frame is in moving you forward. In simplified terms, the stiffer the frame is at the bottom bracket, the quicker it responds to your pedaling inputs and the more energy efficient it is. While the frame itself is just about perfectly elastic meaning that when looked at in isolation the frame flexing does not actually lose energy, the energy loss occurs as soon as you place a rider on the bike due to the rider being just about the perfect hysteresis material (damper, energy sink). Thus the stiffer the frame is at the bottom bracket, the quicker it will respond and the more efficient it will be in converting your energy into forward motion instead of wasting it.


This second chart (Chart 2) therefore shows a somewhat more complete picture. All the Velocite road frames rank significantly higher than any other high end frame. In fact the Velocite Magnus is cumulatively twice as stiff as the least stiff of the high end frames studied and 51% stiffer than the next stiffest non Velocite frame. This stiffness advantage is not marginal, barely perceptible or within the test protocol margin of error, it is significant and large. You can feel the difference immediately when you ride any of our bikes. What are the drawbacks of this extraordinary torsional stiffness? There are none, except for the higher frame manufacturing costs which we absorb. Despite their stiffness none of our frames are uncomfortable, none experience any rear wheel skipping and none of them are fragile. This exceptional performance is essentially free to the rider.

Why don’t other brands pursue pedalling stiffness? They in fact do and they spend large portions of their technology descriptions, press releases and training materials describing how they focus on gaining the maximum pedalling stiffness. Some even invent new super wide bottom bracket standards in the pursuit of pedalling efficiency. Other brands on the other hand make claims that “beyond certain point, there is no difference” – this is entirely incorrect due to simple physics. Frames operate in a perfectly elastic domain, well below their yield or non-linear behaviour meaning that every rider regardless of their strength is flexing the frame in the same ratio of force vs. deflection or N/mm.

We also did not achieve this performance leadership by making our frames heavy. For instance the Magnus size L frame (57cm) is just 1180g, painted and with all fittings. This compares very favourably with actual (not claimed) weights of the other frames studied.

To account for frame weight, for some time now many brands have been using an abstract measure of “stiffness to weight ratio” (STW) to describe the performance of their frames. The STW ratio is derived by dividing torsional stiffness at the head tube by the frame’s weight in kg. I call this an abstract measure as STW ratio (also known as specific modulus) is only useful to characterise mechanical properties of materials being used. STW ratio is entirely without merit when it is used to imply performance benefits in structures such as frames or wheels. Riders do not experience stiffness to weight ratio, they experience stiffness and weight separately. This is in contrast to the power to weight ratio which is directly related to how far and how fast you will go.

Nevertheless, high STW ratios have become the new marketing battleground with each new exotic and inevitably ultralight frame vying for the STW ratio crown. Thus here is a table listing the stiffness to weight ratios of tested frames, ranked in order of increasing STW ratio as measured at the head tube.

Table 2

Looking at the table above you can see the foundations of the performance claims made by the market leaders. Our first contender, the Velocite Magnus falls to the 5th spot in the STW ratio at the head tube table and the otherwise excellent alloy Selene now owns the last spot in this ranking due to its higher weight.

Let’s then have a look at what happens when the table is sorted according to the STW ratio as measured at the bottom bracket. This STW ranking is occasionally used by some manufacturers to establish dominance over their intended competitor.

Table 3


Things look a lot different now when the pedalling efficiency and responsiveness are ranked according to the frame weight. The Velocite Magnus now dominates this ranking convincingly, even against the very expensive, limited run and still very exotic ultra light frames. In fact the Magnus STW ratio is 30% higher than the next non Velocite competitor while the Geos STW ratio is 21% higher. Selene now also returns to the top part of the table with a solid 4th spot.

So what does the high STW ratio tell us when looking at modern bicycle frames? Nothing. Sorry. Please always look for absolute numbers, not a ratio of stiffness per unit weight unless you are choosing which material to use in your next project.

In conclusion, while attempting to maintain some objectivity it is clear that the Velocite Magnus, Velocite Geos and Velocite Selene are very special frames. They are either the stiffest outright or even dominate the synthetic STW ratio measure when it comes to their stiffness at the bottom bracket (pedalling efficiency). They also achieve this performance dominance not through excessive weight, but by focus on real engineering, material choice and advanced manufacturing methods.

To find out more or to buy your own Velocite frame or bike please click below:

Velocite Magnus

Velocite Geos
Velocite Selene


  • Tristan

    I have a magnus and I loved it. However velocite can improve on the paint scheme , as some ppl do buy due to aesthetics after considering the capability and pricepoint of the frame.

    • Thanks Tristan! Yes, the 3rd generation Magnus that is arriving soon has a different paint scheme and we have plans to offer it in more colours over time.

  • Proud to be a Magnus ( 1st gen ) frame owner!

  • I appreciate a stiff bike, but I am not sure about one claim made in this post which is that the rider is a good hysteresis material. While the riders body is very effective at soaking up vibration and could absorb the energy returned to it from frame flexing, I would expect that the tension created by the rider pedaling during a high output effort (when stiffness matters) would lead to this energy being returned to the pedal stroke. Can you elaborate on this claim?

    • While pedaling, the bike is under constant tension as a result of rider’s mass acting downwards, pulling on the handlebars and pushing on the pedals. While mass loading of a structure will dampen oscillations, especially if the mass is… um what’s the word… not hysterical (maybe) but having properties of hysteresis, the human body will also dampen any other stresses or forces that it comes in contact with.

      Thus while pedaling, the frame is being worked on by the rider’s legs, both legs while loaded by the rider. One leg is pushing, the other lifting or in some cases pulling. The force and stresses thus inevitably travel up both of the rider’s legs, whether they are pulling or pushing. Where in the body most of the damping of the torquing of the bottom bracket occurs I am not sure, perhaps it is in the hip flexors, or the upper body that needs to compensate for the leg trajectory misalignment, or for hyper-extension of the driving leg, but there is also constant passive hysteresis since the human body is far from elastic and as mentioned any force that it comes in contact with will be damped.

      As to returning of the tension, yes most of the frame flex will be returned back to the frame. Unfortunately this returned energy is not directly useful for forward propulsion as it is released out of plane of forward motion. The only way to move the bike forward is to apply forward force to the chain. Thus even though the flexible frame (all frames, even ours, for now) will return most of the energy spent on bending it, it will only be converted into forward motion (ie. “restored”) if the rider is simultaneously pushing onto the pedal. However, since pushing onto the pedal implies work and thus spending of more energy this return energy from the frame resetting to its resting state is not aiding the forward motion directly.

      How much does all this matter and is my hypothesis entirely correct? I have no idea actually. This is part of our ongoing research. We want to quantify how much does stiffness actually matter. I want to have valid empirical data on the bicycle dynamics, but to get sufficient data for some really solid conclusions will take at least another year.

      For now there is one entirely clear benefit to frame stiffness that will also benefit from being quantified, but every rider can notice it right now, and that is the lack of “wind up”. There is much less time loss while accelerating with very stiff frames like the Magnus since they are not flexing so much around the bottom bracket. You can even visualize this by doing a simple static test: apply both brakes, place cranks so that a pedal is at 3 or 9 o’clock, then apply strong but even force to the pedal and observe how much the crank is rotating while both wheels are static. Of course some of this is due to the tire compressing, wheels flexing, but a lot of it has to do with the bottom bracket torsional stiffness.

      • Collin Zeng

        If stiffness is so critical to power transfer, why is it that if you put a power meter at the rear hub and crank there is no measured power loss according to an engineer who has tried this before? The meters themselves were accurate to plus or minus 3%.

        • Can you show me the study or a link to this test so that I can see how it was done? Just using the information you gave, 6% power measurement discrepancy is actually quite a lot.

          • Collin Zeng


            I misquoted; it’s actually +/- 1.5% for a total of 3% discrepancy, not +/- 3%, which would give a 6% discrepency.

          • Thanks for that Collin. There is still no reference to the data or the test performed. Did Damon mention his source or reference it somewhere?

          • Collin Zeng

            No source. Perhaps it was internally sourced. In any case I’m inclined to believe Damon, if only because if what he said were not the case, that would be easy enough for anyone with a hub and crank based power meter to demonstrate. Now if he had said something such as “riding a Cervelo lowers the tides of the Atlantic Ocean” then I’d really be scratching my head and asking for some evidence before cautiously accepting the statement as “likely true.”

            If someone could attach both a hub and crank based power meter to a bike and demonstrate that there is a consistent power reading difference between the two I might be convinced stiffness matters wrt power transfer.

  • It’s interesting to see how poorly the stiffness data correlates with “received wisdom” from subjective testing. The Scott Foil, for example, is often referred to as being painfully stiff, while the Cannondale EVO is claimed to be one of the most comfortable top end frames available. More a reflection of geometry, perhaps, or component selection (seatpost particularly)?

    • It is a multi factor issue. First, testing protocols are not all the same and I do not mean that just the methodology varies, but actual testing rigour from setup to assembly of the testing rigs. Then there is the consumer wisdom of what stiff means. By and large a stiff frame is equated with the bike feeling buzzy or even harsh. This is of course wrong as the main factors contributing to this feel are indeed the seatpost, but also wheels, tires and tire pressure and not the frame’s torsional stiffness at the head tube or the bottom bracket.

      Scott Foil seatpost is a truncated airfoil shape and thus offers very high bending moment of inertia (all “airfoil” or elliptical seat posts suffer the same fate), while Cannondale SuperSix EVO uses a 27.2mm round profile seatpost that can be made to be as flexible as needed just by changing the layup, think of a fishing rod. We stick with 31.6mm as we also like the high torsional stiffness at the seat post to encourage whole body involvement when riding. Getting the 31.6mm round profile to deflect is fortunately not too difficult either.

      Lastly, we have third party data showing that even the most flexible frames contribute very little in terms of perceived comfort, and that if the frame is made to deliver the rider comfort through rear triangle flexing such a frame does not ride very well.

  • I was just wondering, does anyone independent of Montague have any data on the Torsional Stiffness at the Headtube and at the Bottom Bracket of a Montague Paratrooper X5? On one of their websites they state ‘Our trend-setting frame materials are then welded together in our state of the art welding facility making a frame that is stiff both laterally and in torsion’ but I have no idea of figures and it is not one of the frames listed here: http://www.g-boxx.com/pdf/Frame-Manufacturer-Overview-1.pdf

    I appreciate your time in reading and replying to my comment – thanks in advance!

    • No, but the single beam design is not the most effective for resisting torsion. With compact bike designs, such as the Montague, the main structural design requirement is strength, not stiffness. Beams are great for managing the planar (in this case vertical) loads and are thus a good way to make a frame that is strong enough to support significant weight, but are not the best at managing torsion.

      • I see, fair enough – thank you for explaining this to me! To be fair though, I expect that they must be fairly good at resisting torsion since the US Paratroopers throw everything they can at them – and that with hub motors putting extra strain on the spokes and creating instability!
        On another note, more closely related to your article, I have been in touch with Stewart Staibik at Rohloff and he says that the procedure for all Gates Carbon Drive frame stiffness tests is the same worldwide.
        He goes on to say:
        ‘The testing jigs are all produced by Universal Transmissions here in Germany and the protocol and test method are clearly defined in their literature. As such the issues mentioned in the report are not applicable and we can be sure the test results are fair by removing as many variables as possible.’
        Just thought you might find that interesting!

        • Gates are measuring the stiffness of the rear triangle of the frame, not the torsional stiffness of the entire frame.

          This test is different to what is used to measure torsional stiffness of bicycle frames, serves a different purpose and delivers different results. It is however very good that it is standardised. Standardisation is one of the main reasons for our decision to reach out to EFBe as the variation in results obtained by different people or companies performing an identical test is too great to be entirely useful.

          Also keep in mind that strength is not the same as stiffness so the fact that beam type frames can be heavily loaded and not fail is not directly related to their ability to resist torsion.

  • Dan

    This is a truly fascinating post, really great stuff. I have been really impressed by the transparency of your company. Being a data analyst myself I hate how companies pick unimportant variables to shout about, and do not use trade standards.

    I was wondering did Tour supply you with the dataset, or did you build it from a subscription to their magazine? I would be interested to see your research repeated with Tour’s latest reviews.

    • Thanks Dan. We’ll share some more information about stiffness shortly once our complete bicycle FEA model is complete. We’ll also organize some of our general findings about bicycle aerodynamics and post that too – for example “Does front brake shape and location affects aerodynamics ?” Yes it does.

      Regarding the data, it came from published data. Either from print magazines or the Tour online PDFs that they occasionally provide.

      We’ll update this chart with new data once our new frame is finished, and once the FEA model is done which will allow us to establish how stiffness in each major structural element of the frame (or bike) affects the stiffness of a bike as a whole.

      • Dan

        Thanks Victor, I look forward to it.

  • Pingback: Planing - Page 3()