dbohemian

Here is a paper that I have gotten permission from its author to publish.

This is only for you tech heads, closet engineers etc.

There has been very little actual research done concerning tandem design. This is a paper by Geoff Turnpenny of England that compares the following tandem designs in steel for stiffness and overall stress.

1. Open frame conventional tubing
2. Double Diamond
3. Reverse Internal
4. Marathon frame
5. Direct Lateral
6. Open Frame (oversized tubing)

At least this is an excellent start. The finite analysis comparisons could be infinite, certainly comparing various materials and individual designs to one another, but the concepts given in this paper generally translate well to the same design considerations in other materials.

The link on my site is here.

https://www.bohemianbicycles.com/Tand...ss%20paper.htm

The basic break down?

1. The conventional marathon design has the least stress and is the stiffest of the traditional designs, but is also the heaviest
2. The direct lateral that is prevalent today is a very efficient use of material and did well overall
3. The very oversized open framed compact tandem is the least stressed stiffest and lightest tandem design.

Dave Bohm
Bohemian Bicycles

TandemGeek

I'm sorry to say, I became somewhat confused when I saw all of the different frame types evaluated used essentially the same, compact frame design that is commonly found on the open frames. I expected the marathon, double diamond, direct internal, and uptube frames to have been modeled using much taller rear seat tubes and a more horizontal top tube. Moreover, the rear stoker compartment and overall wheelbase also seemed dispropotionately short for a conventional upright tandem for adults.

The mystery was solved once I read section 6.3 and it was revealed that the study was undertaken to support the development of a "kid-back" frame for a 168 lb captain and a 56 lb stokid:

So, at this point, I'm not sure I know more than I did before reading this paper. Interesting, but not an epithany by any stretch given how the parameters were skewed towards a very small kid-back tandem design.

dbohemian

You are correct. When in engineering school, one must give a reason for a particular paper and have it approved by the governing body. This is what was needed for this particular course.

With that being said, it does not negate a single thing. Yes, the dimensions are slightly skewed, but they are the same for all the frames in question and therefore, still scientifically valid.

Longer seat tubes for instance would have increased loading and made the frames heavier, but once again we were not comparing a long seat-tubed bike against a short seat tubed bike.

In fact, if the study had included designs of frames were the geometry varied that would have completely invalidated the entire study. How does one find the most efficient geometric design if one keeps changing the envelope which that designs encompasses? I.E. If I was studying the most efficient bridge design then I would want to keep my parameters the same (bridge length, height etc) and vary only the design within that.

In my opinion this is the most thorough tandem stress analysis paper that I am aware of and validates current engineering practice with tandems (which by the way, is less than nothing with almost all manufacturers)

This is also why I gave the caveat that of course it could have been much more thorough. Then again, that would require mass numbers of FEA cases and would have been an engineering text on its own as opposed to a paper.

dbohemian

To add one more thing. The second FEA stress analysis show here:

https://www.bohemianbicycles.com/bohm...m%20stress.JPG

is actually an exact representation of the exact tube lengths, diameters and butting of the tandem that was shown at the NAHBS show. Although comparison is difficult you can compare the general stress gradients to the tandem designs shown in the main paper to see the different levels of stress concentrations and compare overall loading.

My tandem design even with a drastically increased length and stoker compartment, the increase in tubing diameter and the short length of tubing (seat tubes) creates a FEA analysis with far lower stress concentrations than any other in the study.

TandemGeek

I don't disagree, but what I don't know from reading the paper is how much you can extrapolate from the analysis? Do the results remain linear as you increase the rider weights and bike dimensions to those that more closely approximate full-size tandems ridden by adults?

What happens when you load the actual frame characteristics from a Paketa, Co-Motion Robusta, or Calfee? How far from the baselines to they diverge, etc...

Again, I don't disagree with the results, I'm just not sure what I can learn from it. After all, Co-Motion and Santana have both used direct internal frames for years with very similar geometry, but to completely different effect for a variety of reasons.

dbohemian

That is a very good question. For an FEA analysis to be valid the question asked has to be very defined. The idea for this paper originally formed with the question; “Will an open framed tandem experience stress levels that will cause a failure in the product?" I was contacted by Mr. Turnpenny about my oversized, open framed design. I had done simplistic beam calculations and the numbers were good but I was excited about an actual FEA analysis of the design. This would have been the paper essentially but the course required that the actual product being tested be one that was not represented on the market yet. (Of course an engineering student would be working for a company making new product and have to justify that)

So with the question above in mind, this is answered pretty well in the study. Yes, as far as this FEA analysis is concerned an open framed tandem, utilizing oversized steel tubing will experience frame loads that are well within normal ranges and should have a very long fatigue life. For a framebuilder/manufacturer like me that is a very important question. Will a frame design that I am selling to the public be reliable? Am I building something that will experience such stresses that a catastrophic failure is inevitable? This is the most important information for me and having information like this corresponds and validates the previous simpler beam calcs and gives me confidence that I am selling a product that is a least a valid design from a stress standpoint.

FEA analysis is a very time consuming, involved endeavour which is why you don't see very many FEA studies of bicycle frames. Companies in the bike world often do not have the resources or the time to complete a thorough FEA on every frame they build and go with what they know because, honestly it plain works without all the effort. In order to model some of the frames you mentioned, I would have solid model each one with the appropriate tube geometry, butting, size and then set the appropriate parameters in the FEA software (of which my 7500 dollar software can only do a simple FEA, the software used on this project for FEA was more in the order of 25k) Carbon fiber is near impossible for a simple FEA analysis in that its structure is not homogenous like metal materials and all those plies and orientations have to be accounted for. I would say that each model and analysis would take between 40-60 man hours minimum to perform. Can you see why this is not done very often?

I admit that would be very neat to model other frames and compare them to a base frame or my frame, just because I would like to know but it would not have a lot of relevance except to build up marketing hocus pocus. Let's say I modeled a Co-motion, Which size? What tubing? Does the bike have the same purpose? Does it need to be super stiff? For instance, I can guarantee beyond all doubt that the Co-motion Macchiato is a much higher stressed, more flexible frame than the Bohemian Version 2 but then again, they were not built for the same purpose. The Macchiato is a full on race machine were weight is of primary importance, Extra stress levels and a slightly reduced lifespan were the tradeoffs. The Bohemian was not, it was designed to carry a lot more weight and handle dirt road and adventure riding if necessary. Comparing the two is like comparing a Porsche 911 VS a Porsche Cayenne.

Sadly, FEA is not the grail for saying one frame is better than another. As you mentioned, two similar designs can feel very different. I guess that part of it still eludes engineers

All the best,

Dave Bohm
Bohemian Bicycles

mrfish

Hmmm... Having gained an M.Eng degree and also having written a dissertation on fatigue cracking in rotary steam tube dryers based on Finite Element analysis, I feel qualified to comment.

1) The study seems sound and is a good preliminary analysis of the problem, but with quite some simplifying assumptions. I don't know what the time frame for the project was, but I suspect the FEA did not take long. Modelling a bike using a small number of beam elements, very simple restraints and load inputs with one load case is the sort of thing that can be done in a day. It's not the final word in accuracy as it makes some very big assumptions on loading, restraints and how the tubes join together. However the model seems perfectly fit for purpose if the aim was to make broad comparisons between different tubing layouts and refine the design concept along the lines 'roughly how big would the tubes have to be to compensate for loss of the marathon tube'.

2) As suggested by the Visteon engineer, then next logical step is to the frame using plate elements and include the bottom bracket shells and tube intersections. Other load cases and more realistic restraint cases should also be investigated to find those worth further analysis. This could include frontal impact or standing starts in terms of elastic analysis. A good example of a more detailed approach is available at Jim Henderson's personal website (UK champion hillclimber, D.Phil Engineer and FEA truck dynamics engineer). This type of model would probably help refine the design concept and should provide more accurate results than (1), and could also be used to investigate tandem interest items such as potential advantages of o-o-p pedalling or single sided drivetrains. However it would take a lot longer to build and compute than (1)

3) I am a bit confused by the end of the write-up. Is the write-up in the same order as the design process? At best the report is unclear, at worst there is a suspicion that the decision to produce frame 7 was made because frame 7 was already available.

4) FEA is not the end of the story - The FE model is only as good as the inputs to the model - bottom line is that the model can only find weaknesses in the design that are included in the model.

Frames don't just fail where stresses are highest, but normally where the detail design gives an area with different stiffnesses, an area where the geometry concentrates stress (stress riser) or where manufacturing techniques are not properly implemented, e.g. overheating a braze-on part or wrinkling the carbon fibre lay-up. Thus the FE model is always an approximation of the truth, and should never be expected to find 'the answer'. An interesting example of the level of effort which some manufacturers put into their frames is available in the Cervelo website where they discuss how they redesigned their chainstay bridge.

dbohemian

Agree entirely with your assesment. Some clarification.

1. Yes, these are simplistic models. I should have clarrified that to do a full analysis is quite a bit more involved time wise. Incorporating variable tube geometry, multiple inputs and such as you explained would be add complexity. Not a bad thing, just takes time.

2. Version number 7 did not exist at the time of the report but certainly this was the final goal and may have influence design decisions.

Thank you for the input.

P.S. Roland Della Santa, maker of frames to Greg Lemond did fatigue testing of his own years and years ago and says that chainstay bridges (at least on steel bikes) improve longevity greatly.

rjberner

All good stuff:

1) Interested in how you can say this? Care to share? CoMo has done some very different things with their "deeply machined head tube" and tubing design for the Macchiato.

2) The paper showed how some bench testing showed very similar results to the model, yes? Which tells me that the models are not far from the truth.

3) We engineers love to critique other engineers or designs. If products were left up to a committee of engineers, they would struggle to make it to market. (And I are one.) I quess that's why God invented Project Managers and entrepreneurs who have to say "Enough already! Build the damn thing.

Rincewind8

I was a bit disappointed in the FE analysis. Using beams to model tubes is very much simplifying the problem. I'm not even sure I would consider a single simple static loading as a validation for a complex and dynamic loading pattern as it happens on a tandem.

Time constraints as a argument for not doing a shell (2D) element mesh is quite weak, since a 3D drawing existed and meshing that is very easy and fast. Running a 200 elements beam model (from the graphs it looks like the model had around 200 elements) for a static load takes... a few seconds on a reasonable fast computer. Running a 5000 elements shell model for a static case might take a several minutes. Running it for a dynamic load definitely takes longer, but I'd assume one could keep it within a reasonable time frame (maybe 2 hours?!).

dbohemian

Ha Ha! Your observations about engineers are right on the money! I was just being emphatic about showing that we need to compare apples to apples when talking about engineering practice.

I think Co-motion is a great company and they would ones I would purchase from if I was not a framebuilder, so no insult was intended.

Different bikes all together. Probably some specifics on the Bohemian tandem shown at the show are in order. Steel, all tubes 2'' in diameter (even seat tubes) mostly butted 8.5.8 some 9.6.9. 1.5' steerer standard 2.0 head tube diameter, six-two inch S&S couplers, and 45cm seat tube lengths. Frame weight on the bohemian is probably twice what the Macchiato is (13.5lbs with couplers 8.5 without). Tube diameters are quite a bit larger throughout except for the boom tube.

All you guys are engineers, I don't have to tell you that a 2" steel tube is going to be significantly stiffer than a 1.5 or 1.75 aluminum tube (unless it is so thick that we call it pipe) and we know they didn't do that because it is svelte race machine. If you like, I can call Dwane at Co-motion and ask. We are friendly. I was trying to point out that a FEA comparing these two bikes doesn't mean much at all. They were built for two different purposes.

dbohemian

Hey, it sounds like some of you guys know how to do this. There is very little information out there about tandems stresses. Why don't some of you get us some real solid info?

Dave

Hermes

Engineers are everywhere (I chase electrons).

My experience as a cyclist is that aerodynamics "generally" but not "always" trumps weight (one gets the benefit of lower aero load uphill, flat and downhill) in selecting wheels, frames and components. Notwithstanding that, aero performance on the bike is dominated by team position and size. Sooooooooo, when the finite element analysis discussion is concluded, what impact does the larger "very oversized" tubing have on aerodynamics? Is there any wind tunnel testing to support the "very oversized open frame compact tandem" is superior or equal to more traditional frames with laterals? If there is an increased aero load, then the advantage of lower weight may be negated except for long slow climbs.

Rincewind8

I can mesh something and run a FE analysis, but I can't do any testing (lack of equipment). So if you give me the geometry data with wall thickness (preferably in a common CAD format) and frame material data and provide me with some load input data...

dbohemian

So, that sounds like a plan. I can provide you the information you would need.

Now the question is what are we going to compare?

dbohemian

Good question, but I don't have any experience or knowledge about this possibility. As you mentioned most aerodynamic benefit is with rider positioning, instictively I would think that we we are talking large tubes at most we are discussing a .5-.75 inch difference and as long as all tubes were round there would be a minimal difference in the aerodynamic properties of the tubes, but that is just a guess.

Dave

Rincewind8

Someone on the T&H mailing list pointed out the following article:

https://www.pezcyclingnews.com/?pg=fu...01&status=True

It's not about testing of tandems, but it is mentioned, that wind tunnel testing is quite expensive.

BloomingCyclist

Thank you very much for sharing the article. Such completely open testing with publicly shared results are rare. It would be fantastic if the other posters who have access to finite element analysis software and the expertise to correctly enter the parameters would do so. If it's all done and reported in an open manner then the "peer review" process would no doubt be spirited but it would be a spirited debate involving data and the correctness of entered parameters along with the connection between the data and perceptions instead of just debating perceptions. ..Again, thanks for sharing.

Bloomington, IN

Hermes

The cost is USD 975 for your turn in the tunnel as part of a group - expensive: maybe, maybe not, it depends on your objectives and potential reward. It is more than I would spend to tune up for the next race, but it would be totally cool to do it.

zonatandem

Until we come up with a more aerodynamic human bodies, there will always be variances in actual road tests.

bikeriderdave

It is unfortunate, then, that Ms.(?) Turnpenny did not apply more realistic loading, which might have generated more useful results. When a tandem team stands on the pedals, the primary loads are NOT normal to the face of each bottom bracket shell (BB). Rather, direct loading more closely resembles a moment at each BB about the boom tube axis, with tension on the timing and drive chains also creating moments about (approximately) vertical axes at the front and rear BBs. Had the author used real world loading, the results may well have been different, perhaps closing the gap between the various configurations and amost certainly favoring the Direct Lateral design over the Marathon.

It's true that most bicycle manufacturers are quite small and do not have engineering resources to devote to tandem frame analysis. But you should not assume that the bigger guys (Trek, Cannondale) have not generated proprietary results. I'd also be surprised if the smaller outfits (e.g., Co-Motion, Santana) had not worked with grad students on similar studies over the years.

dbohemian

You are correct, I should not assume.

We framebuilders are much like engineers in that if we did not feel that we had something to add or could improve upon something then we would not be in this game.

Certainly, current tandem designs work; they are reliable and have been getting better as far as fit and handling are concerned.

There may be a chance to do further study on this with more accurate loading and more appropriate designs. I hope we can do this and I believe through my own engineering research that we will find similar results.

IMHO I think many designs could be greatly improved and I don't personally believe any serious engineering has gone into a majority of designs, that instead this has been an evolutionary change (i.e. tandems getting OS tubing, then the direct lateral and improved fitting) which is the way most things are done in the bike world.

For instance, from my own calculations and simple FEA analysis it is my belief that torsional rigidity is greatly increased through the use of larger seat tubes, but of course this is difficult because we have to work with current front derailleur that limit what size seat tube a designer can use. So even if a company found through analysis and testing that a tandem frame could be improved by this, there would be no way to adjust for it unless they were also willing to redesign and manufacture a new front mech.

All frames are a compromise between what tubing is available, what parts fit on those frames, ancient standards (i.e. 1.5'' diameter BB shells, etc) how much something costs to manufacture, how easy it is to manufacture and will the buying public buy such item or will they be scared off from purchasing by what they don't understand (people love familiarity)

Given some time I think we can expound upon these results. At least it’s a start and it adds more than just the apocryphal information we are all exposed too.

Dave B

zonatandem

Dave:
Braze-on adjustable front derailluers work fine.
Have a 'glue-on' on our c/f Zona.

dbohemian

they work fine up to 1.5'' diam tubes. Any bigger and no go. Carbon is different as you know in that its strength is not only geometric but can be varied by layup design so size is not as critical.

Not that I think it is for everyone but that is one reason I like the Rohloff on a tandem. From a designers point of view not having to worry about all that muck (like triple chainrings, cassettes and two deraillieurs) really makes my life easy

And not to start another thread here, maybe I can start a new topic, but boy am I loving the Rohloff on this tandem. Other than the 7-8 shift hangup issue, I can't believe how nice it is to ride with it. Never would I go back.

TandemGeek

But at what price and to what effect?