Mytbusting: Tubes and Ride Quality

[Daniel_Y]

Wouldn’t be great to have a model:

I have slowly been pecking at this holy grail model with analysis, but I think its time to start refining the analysis with some real-world datapoints.

Stiffness, Diameter, Wall Thickness

I coded up a quick analysis that compares the stiffness to weight of different tube diameters and wall thicknesses.

One important note: the calculation is for straight tubes with different wall thicknesses, no butts. How this analysis translates to butted tubes is up for debate. I can see arguments for both ways:

• most of the stress happens in the thick butts
• the tube is mostly the thin butt (by length)

Either way, the absolute numbers may not represent butted tubes, but the relative relationships should follow the same trend.

In front impacts, the end butts experience the most stress and strain:

Analysis:

I generated this diagram that shows the stiffness vs weight with a parameter sweep of diameters and wall thicknesses:

Example:

Let’s look at a 34.9x.7 tube:

The 38x.5, 34.9x.7, and 31.8x.9 tube share the same stiffness

Tube	Weight	Diff
38x.5	173g	+0g
35x.7	240g	+67g
32x.9	280g	+40g

BFS has an nice selection of 38mm downtubes:

Tube	Butt	Weight
Velospec	38x747	405
Tange	38x969	475
Tange	38x1.0/.7/1.0	578

Personally, I would not feel comfortable spacing the Velosepc 38x747 tube. According to the analysis, a 35x969 tube probably has the same stiffness, and weights at most 30g more for the same length. The thinner tube will be more dent resistant and easier to work with.

Conclusions

I think the current conventional wisdom is that fancier butted tubing results in better “ride quality”. But I think the ride quality people feel is a less stiff frame (which IMO is a good thing). Less stiffness can be achieved by using smaller diameter tubes with thicker butts.

There are some other real-world considerations:

• Larger diameter tubes are easier to dent
• Thin wall tubes are easier to dent
• Thin tubes are harder to join, which is the majority of the frame failures
• High quality tubes (heat treated and extra butted) have better QC (in our experience)

Unanswered Questions:

• Does ride quality = frame stiffness (and visa versa)?
• What bending mode contributes to ride quality the most?
  • torsional stiffness at the headtube?
  • torsional stiffness a the BB?
  • vertical stiffness of the frame?
• What work has been done to understand steel frame stiffness?

What questions do you have? What have your experiences been with tube selection? What tubes would you like to see?

[BS_Industries]

This is awesome! looking at the plot makes intuitive sense, reminds me of the whole 31.8mm vs 35mm handlebar debate.

I would be curious about full frame stiffness in turns as well and how that factors into perceived ride quality:

Torsion at the head tube and rear axle from the wheels when leaned into a turn with rider weight applied to the BB, as well as lateral stiffness of the entire fame. In my mind this is representative of “tracking” through a turn.

Of course this brings a ton more variables into the equation since it will also be driven by full frame geometry.

[mark_pmw]

Ride quality is so subjective that it may be impossible for two riders to come to the same conclusion. Doing analysis on a bare frame may yield some values that can help with frame life or crash resistance. By the time it’s built into a bike, you have to consider how all the components, particularly suspension and tires, effect ride quality. I believe changing tire pressure or fork adjustments will change ride quality far more than swapping tubes, within similar parameters.

The biggest change in ride quality I have ever experienced was going from a steel frame/fork road bike to a titanium frame/CF fork road bike. The steel bike was painful after a couple of hours, the titanium one was good all day. Conventional wisdom says the ti bike was less stiff, but I never did an objective analysis on the frames. In this case, my definition of ride quality was comfort, but other riders may have very different definitions.

Interesting subject, looking forward to hearing more.

[Mark]

There’s an article out there somewhere where the same frame geo was recreated using many different tubesets.

Some people preferred tubing that was on the higher end, but others preferred tubing that some would consider lower end.

[anon91558591]

I’ve always felt you want lots of lateral flex for cornering reasons. Every time I build a really stiff bike for myself I end up hating it and losing traction unexpectedly and lowering tire pressure to try to compensate.

I don’t even like stiff wheels, though (no carbon for me!) so I might be a weirdo.

-Walt

[drwelby]

 

[Luniz82]

@Daniel_Y

Unanswered Questions:

• Does ride quality = frame stiffness (and visa versa)?
• What bending mode contributes to ride quality the most?
  • torsional stiffness at the headtube?
  • torsional stiffness a the BB?
  • vertical stiffness of the frame?
• What work has been done to understand steel frame stiffness?

I can maybe contribute to this from my experience working for a simulation software company a few years ago. The big selling point of this software was that it could do optimization of all sorts. Topology optimization might be the best known discipline, but you can also do size/shape/parameter/composite layup/whatever you like optimization, as long as you have a functional model of your engineering problem AND a set of goals to optimize for.

So the team I was in, we all had a soft spot for bicycles somehow, and consequently tried pitching the software to bike companies. Turns out, most of them don’t really have any of these goals. By far the most important driving factor for bike development is aesthetics, maybe followed by the requirement to passing some ASTM/ISO/safety test. So the only times we have seen any real FEA (not even optimization, just analysis) in any of these design offices, the models replicated the test benches.

For us, this was a real surprise to see, since the software and techniques were widely used in other industries. Not just automotive and aerospace, but also consumer goods, packaging, kitchen appliances… One of the most complex multi-discipline optimization projects I have seen was for a shampoo bottle.

By the way, I still have access to this software, and I also now work for a company that sticks strain gauges to all sorts of things, mainly railway equipment. I think it would be possible to sneak a bike frame in there at some point. Maybe we can put some real world data behind that nebulous “ride feel”, that’d be awesome

[Daniel_Y]

@BS_Industries

Torsion at the head tube and rear axle from the wheels when leaned into a turn with rider weight applied to the BB, as well as lateral stiffness of the entire fame. In my mind this is representative of “tracking” through a turn.

@anon91558591

I’ve always felt you want lots of lateral flex for cornering reasons. Every time I build a really stiff bike for myself I end up hating it and losing traction unexpectedly and lowering tire pressure to try to compensate.

For offroad bikes, I found the lateral and torsional loads to be the biggest difference in feel:

• In the vertical plane (up and down), the frame is so structurally efficient so it does not flex much
• in the forward plane (front on collision), the frame is not as stiff, but the fork is an order of magnitude less stiff, so the fork will dominate the feel.
• in the lateral/torsional plane (sideways loading from a turn), the frame is the least stiff, and the fork does not contribute.

@Luniz82

Turns out, most of them don’t really have any of these goals. By far the most important driving factor for bike development is aesthetics,

Sad, but not surprised .

@Luniz82

company that sticks strain gauges to all sorts of things, mainly railway equipment. I think it would be possible to sneak a bike frame in there at some point.

That would be great, a bike frame could pass as a rail shaped object

[Daniel_Y]

Thanks, that is a great call out. That is a really cool test that they did. I wonder how it was funded.

I combined the scans to a PDF incase that site gets lost in the depths of the internet:

Magnificent 7 - Feb 1996.pdf (2.4 MB)

From the article:

I picked the Thron as the most shock absorbing. And I lumped Aelle, Cromor, Brain and EL-OS together. To be honest, I couldn’t feel a difference between an Aelle frame - with straight guage tubing and weighing in at 4lbs 12 oz - and a EL-OS frame - with double-butted, oversize thin-wall Nivacrom tubing and only 4 pounds of heft.’

Aelle was straight .8mm tubing. EL-OS was really thin 747 butts. That lends credibility to the earlier analysis: smaller diameter straight gauge tubing has the equivalent stiffness as larger diameter butted tubing.

the bike I liked the best, which I also thought was the stiffest, was bike number six - the Neuron frame. The one I thought was the softest was number two - the SLX frame.

Looking at the tubing chart, that kinda tracks. SLX uses 25.4TT 969 and 28.6DT 969. It has the smallest diameter tubing with the thinnest butts, so it makes sense it was the softest.

Interesting read, thanks for posting it!

[manzanitacycles]

@Daniel_Y

the bike I liked the best, which I also thought was the stiffest, was bike number six - the Neuron frame. The one I thought was the softest was number two - the SLX frame.

He thought the Neuron was stiffest. It was one of only 3 that used ovalized chainstays and larger diameter seat stays despite having thinner walled main triangle tubes compared to some of the others. Based on his impressions, seems like the rear triangle has more impact on feel than the main triangle.

[Alex]

I hadn’t read that article in a long time, thanks for sharing it.

Bicycle Quarterly did a similar test a decade ago that I was involved in. True Temper donated the tubes, Jeff Lyon built the frames, and I think the parts came from Jamis or Free Range Cycles. For that test there were 4 bikes, one 7/4/7 OS, one 9/6/9 standard and I think two 7/4/7 standard. I don’t think the article is online, but Jan and Mark were able to reliably tell the bikes apart by racing each other up hills (they are well matched riders). I was not reliable at telling them apart and am slower than either of them. I wish there had been a 9/6/9 OS frame in the mix because I have a bias against heavy oversized tubing and would have liked to have tested it. The test was double blind and run by Hahn Rossman who swapped the stem caps between each ride (so we never knew which bike was which — bike #1 was a different bike on ride A than it was on ride B). It was fun to be a part of it, but also a lot of work.

[photon]

@manzanitacycles It is counterintuitive, but if you calculate the area moment of inertia for a 30x16 mm elliptical cross section for bending about its major axis, the geometry is about as stiff as a 19 mm (3/4") tube of the same wall thickness and only about 60% as stiff as a 22.2 mm (7/8") tube of the same wall thickness. This doesn’t help make sense of the article’s conclusion, and there could be more complicated effects in a rear triangle than can be explained by simple hand calcs, but for small diameter tubes individual millimeters can make a big difference to their bending stiffness.

[Daniel_Y]

@Alex

Bicycle Quarterly did a similar test a decade ago that I was involved in. True Temper donated the tubes, Jeff Lyon built the frames, and I think the parts came from Jamis or Free Range Cycles.

Any idea which issue it was? I would love to read about it

@photon

if you calculate the area moment of inertia for a 30x16 mm elliptical cross section for bending about its major axis, the geometry is about as stiff as a 19 mm (3/4") tube of the same wall thickness and only about 60% as stiff as a 22.2 mm (7/8") tube of the same wall thickness.

Thanks for pointing that out. Realistically, I don’t think the author can tell the difference between seatstays (they even acknowledge that). They were able to bin all the “regular” diameter tubing bikes and one extra thin OS diameter bike, which I think is already good enough.

I imagine the difference in stiffness of the SS’s is below the margin of error of their tire pressure, so the author is just playing around in the noise of the experiment.

I think that circles back to the complicated nature of “Ride quality”. Numbers and calculations get you some theoretical results, but how a rider feels matters the most.

[manzanitacycles]

That’s nice to know. I was more interested in his impressions. He thought a certain model was the stiffest even though on paper it might not have been.

[Daniel_Y]

@anon91558591

I’ve always felt you want lots of lateral flex for cornering reasons. Every time I build a really stiff bike for myself I end up hating it and losing traction unexpectedly and lowering tire pressure to try to compensate.

What combination of tubes do you feel are pushing into the “too stiff” category?

[poopieprancer]

It’s very interesting thinking about how the different tube diameters can be comparably stiff by varying the thickness. What I’d really like to see is a model that can create equally stiff feeling frames in all sizes.

Let’s say just focusing on torsion. First you’d need to come up with the metric to optimize for. Angular deflection between HT and BB seems tempting, but doesn’t account for the handlebar width and flex variables. Perhaps arc length that the handlebar ends travel through during torsion could give a more realistic proxy of the feel. A frame could feel fine with narrow handlebars, ala drop bars, but feel like a total noodle with wide MTB bars. I experienced this on my 80s Miata road bike, which feels pretty decent with narrow drop bars, but super flexy when I swapped to 780mm flat bars (duh). My intuition says there’s probably a pretty universal amount of handlebar end arc travel that people find pleasing regardless of their size.

[Alex]

The BQ test was done in this issue:

BQ 24 (Summer 2008)

The focus here was on performance, not comfort.

[DEVLINCC]

Bare with me here. Opinionated thread reply

Everything said above is valid and worth talking about. Up to a point. There are so many variables and also different peoples interprtations of teh feedback a bike gives. That being said I also think unless you want to push steel to it’s absolute limits, what ever that limit is decided as, then a lot of the research is not going to give you a huge benfit. It’s fun to do on the side though.

My take is, as frames increase in size you increse the tube diamter and/or wall thickness and teh starting point for that is going to vary depending on the intended use. Daniel’s plot shows how you can balance the two against each other. What the plot doesn’t show is the combinations of wall thickness and butt lengths for a given tube over say 500mm. That would make for an even more interesting plot and see how playing with the combinations can affect the end result.

As @poopieprancer mentioned, metrics. What are we measuring? How are we measuring it? What meric is more important than another? What’s the threshold of go/no go? What’s the base component spec we are using to evaluate it? With out any of these determined, then each builder is going to have to figure it out based on some shared info and a bunch of experience and intuition. There is no one formula for producing a bike of a particular torsional stifness and ride quality.

It’s an interesting discusion and there is some great info here. Very easy to get into the weeds though.

[Nifty]

I like coming back to this article on Sheldon Brown’s site. It’s old, sure, but I really appreciate the way they break the strain energy distribution by loadcase/tube/mode:

https://www.sheldonbrown.com/rinard/fea.htm

I’m aware of very capable DOE tools (my day job is as a FEA engineer for a global auto company) that could easily handle the kinds of design variable and response matrices that could give some immense data for understanding frame static/dynamic stiffness and strength (especially for such discrete variables as steel tubing), but “linking” those responses to ride quality is where things still seem a little vague and subjective.

[Daniel_Y]

@DEVLINCC

What the plot doesn’t show is the combinations of wall thickness and butt lengths for a given tube over say 500mm. That would make for an even more interesting plot and see how playing with the combinations can affect the end result.

Agreed, the contribution of the butts are all pieces of a puzzle that need to be figured out. A lot of engineering relies on the assumption that complex behaviors are linearizable. That is what allows you to break down complex problems (total frame stiffness) into individual pieces.

For example: is the effective stiffness of a butted tube simply proportional to the lengths of the butts?

I_{total} = I_1L_1/L_{total} + I_2L_2/L_{total}+I_3L3/L_{total}

Or maybe, I_2 dominates because it is the center butt?

Or maybe the whole tube can be approximated by I_{total} = (I_1 + I_2 + I_3)/3

I bet the latter would be pretty darn accurate (within 15%).

You could easily fit these dumb models with empirical data from measuring tubes deflection with weights. I would love to do it, but I live in a studio apartment

@DEVLINCC

As @poopieprancer mentioned, metrics. What are we measuring? How are we measuring it? What meric is more important than another? What’s the threshold of go/no go? What’s the base component spec we are using to evaluate it?

For a torsional stiffness metric, with the BB clamped, you can measure the angular deflection at the headtube from a known mass and lever.

With the angle, you can determine whatever metric you care about: grips, contact patch, etc… by multiplying the cosine(angle) by whatever characteristic length you care about

@DEVLINCC

With out any of these determined, then each builder is going to have to figure it out based on some shared info and a bunch of experience and intuition. There is no one formula for producing a bike of a particular torsional stifness and ride quality.

From my perspective, I am not trying to quantify what ride quality should be. I think ride quality is subjective, but stiffness can be calculated/simulated and validated to a model.

Like @Alex says, he found he really enjoys regular-diameter road frames. I have no idea how to scale his ideal stiffness to a frame for someone 6’6 and 260lbs. With a model, you could have a pretty good enough guess! We already sorta do this with our Hummingbird Model to spec tubes, but that study was extremely manual, and not generalizable to other bikes.

I agree that components, wheels, tire pressure, and subjective opinions of stiffness are all over the place. But I design frames, and so I try to understand what I have control over so I can build better bikes!