Science and Bicycles: Frame Stiffness

Science and Bicycles: Frame Stiffness

Lightweight frames, made from high-strength tubing, were thought to be stiffer than ordinary frames. They performed better for most cyclists, but the conclusion that stiffer frames were better was erroneous. Eventually, this led the makers of steel bicycles down the wrong path and may have hastened their demise.
When the plague ravaged Europe during the 14th century, people noticed that the disease arrived together with herds of cats. They concluded that the cats brought the disease and responded by killing as many cats as possible. Only much later was it discovered that the plague was transmitted by fleas that lived on rats. Cats arrived with the plague only because they preyed on the rats. So killing the cats actually made things worse, by allowing the rats to multiply.
This reminds me of bicycles and frame stiffness. For decades, riders have assumed that stiffer frames were more efficient, and thus faster. At the same time, physics tells us that a lighter bike is faster uphill and in accelerations. Stiffness and weight are conflicting goals: To make a frame stiffer, you need more material. To make it lighter, you need less material.

There was a apparent solution to this problem: high-strength steels allowed tubing makers to use less material without giving up strength. Reynolds 531 and other modern steel tubes enabled builders to craft lightweight frames. Even though these frames were light, they felt very stiff when the builders bent the tubes to align the frame. Many old builders have told me: “The lightweight Reynolds 531 was twice as stiff as ‘drainpipe’ (ordinary steel tubing). And the superlight 753 was three times stiffer than 531.”
The riders usually found that the “stiffer” bikes made from high-strength tubing performed better. This “confirmed” that stiffer bikes were faster:
1) Observation 1 (builders): high-strength tubing = stiffer frame
2) Observation 2 (riders): high-strength tubing = better performance
3) Conclusion: stiffer frame = better performance

However, Observation 1 is incorrect: The builders mistook yield strength for stiffness.
Yield strength determines how far you have to bend a frame until it no longer springs back, but “takes a set.” High-strength tubing has a higher yield strength.
Stiffness determines how much force is required to flex a frame a certain distance. All steels have roughly the same stiffness (modulus of elasticity). The only reason the high-strength frame is harder to bend is that you have to flex it further until you reach the yield point, where it does not spring back any longer.
Even though the high-strength tubing didn’t make the frames stiffer, the higher performance was no fluke. For most riders, a Reynolds 531 frame performed better than a “drainpipe” frame, and a Reynolds 753 frame performed better yet. The observation was correct, even though the explanation was incorrect.
For riders and builders, it didn’t matter that their “stiffer” frames in reality were more flexible (because the tubing walls were thinner), as long as they performed better. All was well for decades: Builders knew how to make well-performing frames, and riders loved riding them. Whether they really were stiff or not didn’t matter.
Things changed when new materials became more prominent during the 1980s: aluminum, titanium and carbon fiber. At first, most bikes made from these new materials were not very stiff, but they were very light. Many riders liked them.
How could the makers of steel bicycles counter the new competition? It was hard to compete on weight, so they focused on stiffness instead. To make a stiffer bike without a huge weight penalty, you increase the diameter of the tubes, and you get a great increase in stiffness with only a small increase in weight. That is how oversize steel tubing became popular.

Some makers also offered tubes in all kinds of shapes for improved stiffness in certain directions (above). There were even spiral-shaped reinforcements inside the tubes to increase stiffness (below).

It is ironic that by making their bikes stiffer, the makers of steel bikes may have hastened their demise. (In effect, they killed the cats instead of going after the rats and their fleas.) Instead of making their bikes perform better, they actually returned to the ride characteristics and performance of stiff frames from the early days of cycling.
The makers of aluminum bikes could make their bikes even stiffer, because aluminum’s lower density makes super-large tubing diameters possible with very little weight penalty. (In the medieval analogy, the aluminum bike makers invented a more efficient way of killing cats.) Their bikes also fell by the wayside. (Even Cannondale, the pioneer of oversized aluminum tubing, now uses carbon fiber for their top-of-the-line models.)
Carbon fiber and titanium became the preferred materials for high-end bikes. Ironically, the last steel bikes ridden by top-level amateur racers were made from the superlight (and super-flexible) Reynolds 753 tubing. I remember seeing them under riders whose sponsors would have loved to give them a new bike…
In recent years, it has become clear that many riders perform better on flexible frames, apparently because it allows them to apply their power more efficiently. Many riders and builders extol the virtues of a “lively” frame made from flexible tubing. When we tested different frame tubing in a double-blind test (Bicycle Quarterly Vol. 6, No. 4), we found that two of three riders preferred the most flexible frame both for constant efforts and for all-out sprints. (The third rider could not tell the – very small – differences between the frames in our test.)
Of course, the real story is more complex. There is more to bicycle performance than overall frame stiffness. Frames can be too flexible for a given rider and application. Some riders may even prefer very stiff frames. However, it is clear that the old mantra of stiffer = more performance is not true for most riders.
The main conclusion for me is: False explanations of real phenomena may work for a while, but eventually they will lead you down the wrong path. Whether it’s trying to combat the plague or producing bicycle frames with better performance, we need to examine the real explanations beyond the current beliefs. That is why Bicycle Quarterly does fundamental research to determine what really makes a bicycle perform, rather than just try and figure out how to make bikes stiffer.

Share this post

Comments (26)

  • Lee RIngham

    Hi Jan:
    Another well thought out article. A perfect case in point for your argument, in my mind, is the racing history of Sean Kelly. A feared and renowned sprinter, he rode the very flexible Vitus 979 aluminium frames for much of his career. Somehow he managed to win an awful lot of races, sprints and the green jersey in the Tour multiple times.
    All on one of the most ‘noodle’ frames ever made!

    October 3, 2011 at 9:36 am
    • Jan Heine, Editor, Bicycle Quarterly

      We’ve never measured the lateral flex in a Vitus. We have measured an Alan cyclocross frame, and we were surprised when it was about as stiff as a Columbus SL frame. We expected more flex… A Reynolds 753 frame is more flexible overall than the Alan.
      That said, Kelly clearly didn’t worry much about frame stiffness, and in fact, when he switched to steel frames (which may have been stiffer), his winning ways came to an end. (His age may have had something to do with that, rather than frame stiffness.) For Kelly fans, we have one of his last pro bikes (a Concorde he rode for PDM) in our book The Competition Bicycle.

      October 3, 2011 at 9:56 am
      • andy albershardt

        the Concorde was made in the CIOCC factory for Rentmeester a large Dutch bike distributor who owned Veltec Sports back then. Veltec later was served cease and dismissed papers by lawyers for Concord Bikes in Ohio. Their argument was the names were too close and the bike industry was too small. So Rentmeester/Veltec discontinued the frame. Kelly on PDM team was the last year of the Concorde. I was at Veltec when it all went down then in the early 80s.

        October 8, 2011 at 3:27 am
        • Jan Heine, Editor, Bicycle Quarterly

          Thank you for adding that interesting historic information. Do you know whether the team bikes also were made by Ciocc, or whether they came from a smaller specialist framebuilder?

          October 8, 2011 at 6:09 am
      • Brian Wilcox

        I have an NOS Vitus 979 that I am willing to provide for testing, if you’re interest. (It was built up as a show bike “but was never ridden.” It certainly has very low miles.) I’m in Portland. Let me know if you’re interested and we’ll arrange something. My plan is to build it up some day if I can ever get to within 20 pounds of my old racing weight….
        I rode a 979 back in the 80s and loved it. I never understood the whole “stiffness” argument except that friends on oversized-tube Cannondales would be bet to death by the end of a long ride. You explanation of “planing” makes perfect sense, and the first time I read it I felt like it made perfect sense.

        October 9, 2011 at 11:12 am
  • Paul

    Very thoughtful and well-written article. I have most recently been riding my early 1990’s Koga Miyata with new 26″ Compass tires. It is the bike I can most easily get in synch with, although it is made with oversized steel tubing. My Kogswell Model P (which seems “flexier”) does not respond as well for me. With more time on each, I could probably determine whether these issues are geometry-related, tire-related, or indeed “flex” related.

    October 3, 2011 at 12:43 pm
  • Rolly

    Does crank length play into the flex of a frame, given that a longer crank would act as a longer lever (and possibly stronger if that length makes enough difference)? My 853 frame, the first oversized tubed (just slightly though) road frame I’ve ever owned, took a while to get used to. Sometimes I feel faster and sometimes I feel I was faster on the smaller diameter tubed frames I owned in the past. I really noticed it at first and for a while I prefered the feeling of my girlfriend’s smaller-tubed frame. Since reading about your planing theory I’ve been wondering if a longer crank would help me find that old familiar flex feeling.
    Interesting article! I haven’t heard/read the term ‘modulus of elasticity’ since welding school.
    — Rolly

    October 3, 2011 at 5:59 pm
  • Jan Heine, Editor, Bicycle Quarterly

    We haven’t studied crank length, but neither Mark nor I notice much difference in frame flex between his bike with 175 mm TA cranks and my Singer with 170 mm TA cranks. Of course, the 5 mm length difference amounts to only 2.9%…

    October 3, 2011 at 6:18 pm
    • Rolly

      I figured it was wishful thinking. Thanks for the insight.
      Do you think the stiffness of the BB spindle would be a similar variable to frame stiffness/tube diamter?

      October 4, 2011 at 4:43 pm
  • Mark Williams

    Jan, on this subject of frame stiffness, I’ve occasionally wondered if the phenomena you’ve experienced that you’ve labeled “planing” might be akin to your cats and plague example. I don’t doubt the phenomena you’ve experienced, but find it difficult to understand how LATERAL and VERTICAL frame flex can be stored up and released as HORIZONTAL forward motion. I’m tempted to wonder if the more gentle effect on the body of a more resilient frame may lead to greater speed in some cases, rather than that the frame is innately faster. But you’ve described an effect of higher speeds even on short attacks up hills, so I’m mystified.
    My other question on frame stiffness would be, doesn’t the impact on speed relate mainly to the type of riding, and to the body characteristics of the rider? I can see how less frame stiffness may lead to faster times in for long distance endurance cycling; but it would seem less likely, say, in criteriums on flat ground that finish with a bunch sprint.
    My own racing experience: when I switched from a Colnago to a Rossin (which was stiffer), my racing results improved. I’m 6′, 175 lbs and muscular. If I weighed less, or had a different build, I would guess that I’d be more effective on frames with a lower level of stiffness.
    Anyway, this is an intriguing topic, and it’s great for cycling that you’re taking the time to consider it.

    October 3, 2011 at 7:02 pm
    • Jan Heine, Editor, Bicycle Quarterly

      When you look at Gary Houchin-Miller’s finite element modeling, you can see how stored frame energy drives the bike forward. (We published it in Vol. 5, No. 4.) Your example shows that frame stiffness depends on rider power output and riding style. It appears that Columbus SL tubing was developed for European racers. Interestingly, Jacques Anquetil used a Reynolds 531C frame (.8-.5-.8 mm walls) for massed-start races, where acceleration and sprinting was paramount, but a superlight frame (.6-.3-.6 mm walls) for time trials, where a constant effort was required.
      In our double-blind tests, the .9-.6-.9 mm wall frame worked equally well for me as the .7-.4-.7 mm wall frame during super-hard uphill sprints (800+ Watt). However, as I got tired, the more flexible frame began to show its superiority. Stronger or heavier riders, or those pedaling at different (lower?) cadences may have different preferences.

      October 3, 2011 at 8:47 pm
      • Mark Williams

        Thanks. I can see that I need to renew my BQ subscription (smile).
        So apparently, there is mechanical effect from frame flex that increases forward motion immediately, AND a reduction in stress effect on the body (less jarring) that can add additional speed in longer events ridden at constant effort.
        Have you by any chance investigated the relative impact of these two effects on greater speed? Of course it would depend on the ride, the rider etc, but overall, which one is stronger? What’s the balance between the two?

        October 4, 2011 at 10:06 am
        • Jan Heine, Editor, Bicycle Quarterly

          The reduction of stress on the body is mostly due to other factors, like the tires (and pressure) and the fork design. Possibly, the seatstays can affect comfort as well, but the main frame is a fairly stiff structure for “vertical compliance.” I doubt a lighter-weight main triangle is going to affect that much. Obviously, if you are comfortable, not only can you put out more power, but you also enjoy cycling much more.
          However, I find fascinating it that even if you can “tough it out” on a super-stiff frame and super-hard tires, you aren’t faster, because they simply are less efficient.

          October 4, 2011 at 10:17 am
  • Michael Richters

    I find the history lesson on bicycle frames quite interesting, and thank you for that. It’s a little quibble, but I regret to point out that your beautiful analogy to the Black Death is rather less accurate. There was no correlation of an increase in the cat population with the arrival of the plague, nor does this make any sense. Cats had been persecuted in Europe prior to the arrival of the plague because of an association with witchcraft. This allowed the rat (and flea) population to increase, thus amplifying the spread of the disease. No doubt the plague was also associated with witchcraft, and there were almost certainly cats killed in a misguided attempt to protect people from the plague, but the vast majority of the extermination of European cats occurred before the arrival of the plague. See the following article (with references):
    Of course you are quite right that the culprit in both cases was misidentified, resulting in counterproductive action, and that’s really the main point…

    October 3, 2011 at 10:53 pm
    • Jan Heine, Editor, Bicycle Quarterly

      Thank you for pointing this out. I am not expert on cats, but I wonder how people associated cats with the plague if the cat numbers did not increase. Of course, it could just have been coincidence: People thought cats were evil, so when the plague arrived, they started seeing cats everywhere, whereas before they didn’t notice them. Or it may be that people cast their cats loose to avoid being persecuted with them, and the feral cats followed the rats, as my source suggested.
      It’s an interesting subject, albeit not really suitable for this blog. In any case, there is no doubt that blaming cats for the plague and killing them made things worse.

      October 4, 2011 at 6:06 am
      • Michael Richters

        Apologies for the digression. I am an inveterate nit-picker.

        October 4, 2011 at 6:56 pm
        • Jan Heine, Editor, Bicycle Quarterly

          Don’t misunderstand, I appreciate the added information – most readers and I like to learn more. I just don’t want this to turn into a cat discussion.

          October 4, 2011 at 9:14 pm
  • Franklyn

    I have a cat, and I have to say it’s a shame that they killed them. He likes to hunt bugs, but I don’t think he could handle a rat.

    October 4, 2011 at 9:32 pm
  • Lovely Bicycle!

    Fascinating post and a lot to think about. I’ve been mulling this over for a couple of days, and here is what doesn’t quite make sense: If the tubing considered stiff by the framebuilders was in fact flexible, then how would you explain that the cyclists themselves experienced it as “stiff”? Or, did they not?

    October 5, 2011 at 1:16 am
    • Jan Heine, Editor, Bicycle Quarterly

      Asking a cyclist whether a frame is stiff is asking the wrong question. A cyclist experiences secondary effects: The frame climbs well. It accelerates well. Cyclists are told that this means that the frame is stiff, so they begin to equate good climbing and acceleration with stiffness, but stiffness is not what they usually experience.
      There are some rare instances where you can actually experience stiffness. In college, we sometimes carried a second rider for short distances on the saddle of our bikes, while we rode out-of-the-saddle. This worked great on my “drainpipe” Peugeot, but when I tried it again later on my “531” Mercian, the bike was almost unrideable. For the first time, I actually felt “stiffness” (or the lack thereof), and I felt cheated: I had spent a lot of money on my Mercian with its high-end “stiff” tubeset, yet it was obviously less stiff than the Peugeot. However, the Mercian performed better, so I wasn’t too upset.
      There is little doubt that a Reynolds 753 frame with 0.3 or 0.4 mm walls is less stiff than a Reynolds 531 frame with 0.5 or even 0.7 mm walls (same tubing diameters), yet many builders and riders felt that the Reynolds 753 was stiffer. However, the builders equated yield strength with stiffness, while the riders equated performance with stiffness.

      October 5, 2011 at 7:02 am
      • Lovely Bicycle!

        Well, speaking of Mercian – I now have one. A “Sportive” frame with track ends. When ordering, I tried to understand what the difference was between their Reynolds 631, 725, and 853 tubing options. I was told by every person I asked (cyclists and builders) that the higher the number, the lighter and stiffer but less comfortable the tubing. Based on this feedback I chose the 631. The ride does feel very comfortable, but the frame is on the heavy side by contemporary standards. I now wonder whether those I spoke to were incorrect in their descriptions of the 725 and 853 tubing.

        October 5, 2011 at 9:10 am
    • Richard James

      I wish the producers of production and semi-customized bike frames published the butting used for each frame size, and hope BQ will help drive greater openness. Geometry, outer diameter, and tire clearances are shared in marketing materials and I think Jan has shown that these parameters exert far more influence on the rider experience than tube butting. Yet butting remains something of a trade secret.
      I am aware that “butting” is not a well defined term, and while the butting of an uncut tube can be easily measured the actual butting of a tube mitered and joined in an actual bicycle frame is hard to measure and will have different parameters. I would suggest that knowing the butting of the uncut tube and the actual tube length will reveal the most incluential information (and be easy to capture). Leaving the length of each particular butt as unknown contributes less to the finished result. My obsevations in this area are that for CAD (not BikeCAD) designed frames the exact butt lengths are often not defined (as used on blueprints used to order batches of contract-built frames. These blueprints seep out onto the Internet with some regularity), and that many framebuilders simply follow a standard practice rather than attempt to use butt lengths to achieve particular ride characteristics (other than leaving “enough” thick section to support S&S couplings, when required).
      Anyone with a hacksaw and a few hundred dollars (i.e. a competitor who wants to knock off a particular design) can know for certain what choices were made for a specific frame size. As R&D costs go this is not very much money. Thus maintaining secrecy around materials selection is no impediment to competitors.
      Keeping this information proprietary harms only the customer who wants to experiment and inform his opinion. Sharing materials parameters would certainly help interested riders avoid making fallacious assumptions Jan has described. For one, I would prefer to buy 3-4 production frames with known parameters as experiments before ordering a custom frame.
      As things are today the best way to experiment is to acquire many custom frames, which is 2-10x as expensive (considering Rivendell at the top and QBP/Surly/Salsa at the bottom of the spectrum). Even if it is financially feasible, the months of lead time required to turn a custom order into a new frame will ensure that such research will require years of sustained effort (if one is not prepared to order all possible permutations in one go).
      Am hopeful that BQ will someday drive greater transparency in this industry. Apologies for the long comment

      October 6, 2011 at 5:59 am
      • Jan Heine, Editor, Bicycle Quarterly

        Even if companies published their butting specs, that doesn’t mean that the actual frames are built to that spec. Whether it’s Surly, Breezer, Jamis or others, most factory-built bike we have tested had a geometry that was significantly different from what the makers had spec’d. The exception was Kogswell (which was not spot-on, but closer), but Matthew Grimm told me many times how hard it was to get the Taiwanese frame shops to make frames to his specifications. This means that even the published geometries have to be taken with a grain of salt.
        For custom builders, the best ones do consider the butt lengths. However, I would never spec this as a customer, but rather communicate what I want for ride characteristics. If you want a flexible frame in a relatively small size, then the builder will select tubes that keep the butts as short as possible. If you want a stiff frame for rough roads in a large size, the builder might choose somewhat longer butts at the front of the frame to spread the load from fork movements over a large section of the tube.

        October 6, 2011 at 6:42 am
  • Garth

    How does handlebar width factor into this?
    Something I have noticed, going from moustache to “porteur” bars is that with less leverage, I did less out of the saddle acceleration. This difference in wider and narrower bars was even more succinct when I then sampled 44cm Noodle bars and then went to 42cm Nitto Randonneur bars. Whereas with the Noodles it was nice to be able to stand out of the saddle to get good leverage to crank away from a stop sign, the much narrower at the hoods Randonneurs simply don’t give the same leverage. Instead of a “side to side” lateral cranking, I find myself staying in the saddle and instead “pulling” on the bars. (I prefer the Randonneur bars because of the angle of the brake levers is much more comfortable – city riding causes one to keep their hands on the hoods, ready to brake! I am looking forward to a 44cm version, though, because with downtube shifters it’s easier to be able to crank away in my cruising gear than to downshift, then upshift.)
    I feel like the wider bars allow me to take advantage of the planing in the frame, whereas the narrower bars don’t.
    Jan also mentioned in an earlier article about wider handlebars usage being related to high trail geometry, and I think this is also a factor that I am experiencing.
    (63 cm Heron Wayfarer, 6′ 140 lbs)

    October 5, 2011 at 7:05 pm
  • Mark C

    I raced a Vitus 979 in the early 90’s. A classic find in the back of the bike shop I used to hang around and help out at. It had friction gears and campo components while Shimano’s STI gear lever brakes were just hitting the race scene. I bought it and took it home and fully restored every part to working order. I fondly remember the flex in the bottom bracket and I had a theory that it would flex back against my force being driven into the cranks especially in a sprint or on a climb. I won a few races on that bike and wish I still had it today. Forward 20 years I have got back into bike riding and love brevets and randonneuring events. I now ride a custom steel frame Llewellyn (Australia) and have racks, fenders and a bag. All things I never would have dreamt of in the 90’s and yet riding is so much more enjoyable on fit for purpose machine. Keep up the good work Jan.

    October 9, 2011 at 6:39 pm
  • TSW

    I wonder if the sensation of stiffness versus that of a good hill climber or sprinter, or going over bumps easily, is related to how the bike bears the load of the rider in an oscillatory way. For instance, I would think that during a single revolution of the pedal, the forces on the bike vary from one pedal to the other, to the saddle (or not, if pedaling out of the saddle), and the handlebars, the forces shifting to all these points of contact. Not to mention the road supporting the wheels, the friction of the rear wheel propelling the bike forward, as well as gyroscopic and centripetal forces while negotiating a curve.

    October 12, 2011 at 3:14 pm

Comments are closed.