Yes, they understand how to do carbon correctly. If they didnt, your bike would break. Different carbon layups are used on different areas of the bike to meet the desired goal

---

Make Inside Out Sports your next online tri shop! http://www.insideoutsports.com/

I'm abit confused, your question isn't entirely clear.

Aerodynamics is abit of an imprecise science. We DO understand it to a pretty high level, but saying we have it 'down'. is not really correct. The science of fluid dynamics with regards to drag has only really started to have serious advances in the past 100 years. Alot of what we know is from trial and error, especially if you're dealing with higher speeds than you typically see with bikes. I'm not purely an aerodynamics guys, just someone with abit of CFD experience, so this is what I can offer.

What we do have is that given the boundaries of an engineering problem, we do understand what will generally work- that's the reason why most aircraft look alike, most F1 cars look alike, and most aero bikes are starting to converge at a similar shape. The problem with Bike aero is that there is a very inhomogenous variable at play - the rider.

Given a specific tube shape, the aerodynamics SHOULD NOT change unless you are changing another variable- the speed, the yaw angle, or perhaps the surface roughness of the tube shape. We do understand within limits what changing certain factors will do- for example, chord length. Thats why we have a whole library of shapes- called NACA profiles. NACA isn't always the best for bike aerodynamics, though. For example, we know in general which areas of the bike are responsible for the most drag, and some general rules regarding fork blade design. This has been obtained through many, many iterations of testing and trial and error

Carbon layup affects the material properties, and not aerodynamics - i.e. the stiffness of the tube and how it performs under stresses. and yes, engineering is very specific. within limits, we can quite accurately predict what will happen when using 4 layers of say Toray 9000 carbon laid up at 45 degrees to each other, compared to 90 degrees, and et cetera. We can even say that a crack of under X length is safe to ride in a carbon bike (this varies from tube shape to tube shape, material, rider weight, etc.) . You would want engineering to be specific, after all. Engineers are paid to be specific ( a lesson I learnt from my university professor)

How do engineers gain this knowledge, you ask? well, through making educated guesses and trial and error. When designing a product engineers will make some general assumptions- for example- If I were designing a bike I would assume that it is being used in 'normal' temperature conditions of, say zero degrees to 60 degrees celsius of ambient temperature ( some allowance is always necessary for extreme conditions). I will also rely on historical test data- for example- the tensile strength of this material ( in our case, carbon), which must be documented. I will design my product based on these general 'rules', then build a prototype and test it. Nowadays, we used computers to test them, so it cuts down the number of prototypes we have to make. If the prototype performs to what we have predicted, well, then our design was good. if not, something went wrong, and we need to investigate.

The structural stuff is actually a bit more well understood than the aerodynamics actually.

Speaking of Newey, in F1 the analysis is actually aero-elastic where they are analysing the aerodynamics and the structure at the same time because they interact. Take of example the front wing (or even just elements of it), as the aerodynamic forces are generated, the wing will deflect (because no material is infinitely stiff). This deflection changes the shape of the wing, for example bending it downward. So now the wing is closer to the ground, which increases the aerodynamic forces, causing more deflection. Or it may be deflected backwards, causing it to rotate, which causes the wing to actually produce less downforce (and drag, which may be the goal because at high speeds they have more than enough downforce, but you do not just want to trim front downforce, you also then need to figure out how to trim rear downforce at similar speeds). The way it deflects and how much it deflects is tied directly to how the layers are arranged and what the layers are. It is understood, but it is very difficult to actually do. This is without actually making sure that you can build the wing and even harder make sure you can build multiple copies that behave the same way.

...said no composite structural analyst ever. Metals, sure, knock yourself out.

High performance composites (I'll say this is for the current high fiber volume material) have been around since, maybe, the 80's with relatively significant gains being made even in the current day (with respect to OOA processing as well as damage tolerance). Composites are pushed, maybe 40-50% of their limit in relatively normal applications on a bike if everything is loaded in plane. Remember, the materials we're talking about are 100-300 times stronger in one direction than every other direction (assuming, at least, an IM carbon uni - variation is for tension and compression). Hey, what material properties are the bike companies using? I seriously doubt they're paying $5M to characterize one material system at NCAMP. Maybe they're using material properties someone else tested - which is fantastic if equivalency testing was performed. Maybe they're just using ballpark numbers - 18Msi for IM uni, 0.006 strain at failure, etc... At which point, you really have to wonder if there is a real advantage to rolling in [super new cool material that makes a frame $1000 more expensive]


The first wind tunnel, 1909. Let's not mention the unbelievably complicated 30mph flow regime. No compressibility. High viscous effects. I mean, if Karmen et al. had not completely proofed out this flow regime in the 1930s how could we begin to consider it 80+ years later. Aerodynamics start to get a little more complicated at 200ish mph when compressibility occurs, but doesn't begin to be significantly complicated until, say, M 0.7 in the transonic regime (where computer models still are somewhat unreliable and testing is fairly important).

I understood none of that, however, it could go to my point (if I understood it, and it were relevant) as to why carbon is a good choice for bikes and the boats in the VOR

how is it that a material that clearly has a big problem with both impact and flex has become the go to material for frame / hull material?

It can be lighter. It can be easier to shape (aerodynamics). It can have very good strength characteristics, at least in the axes you care about most. Flex isn't always a bad thing, and you can design "flex" into the axes where you want flex. Has a good fatigue life vs. some metals, like aluminum.

Your post is full of contradictions.

First you call the 30MPH flow regime "unbelievably complicated." Then a bit later you say that aerodynamics doesn't begin to be significantly complicated until the "transonic regime" of "M 0.7".

I am genuinely the idiot in this discussion - for the time being (I hope)

So engineers working on frames at bike firms understand that one additional layer in the bottom bracket contributes X in terms of stiffness to Y in terms of another metric and that this is part of their undergraduate engineering degree's or postgraduate training?

Yes. I'm an EE, but work with ME's who do this. They typically use programs like SolidWorks that do finite element analysis. And, yes, learning those tools is now a staple of undergrad study. But there are other much higher-end programs. Or proprietary programs. They'll enter a layup into the tool, then set it to to simulate various kind of stresses. Then you start the program on a simulation. This can take a while, depending on the fidelity of the simulation. The result is a bunch of data on what sort of flexes occurred, how close you came to "breaking" the material, etc. The other posters are probably right in that this is just an approximation based on a lot of assumptions of the underlying materials. So you do have to still validate the design with real-world testing. But this makes the initial design much quicker. The programs can make animations of the stresses (usually exaggerated in scale so that small flexes look much larger in the animation for illustrative purposes). ME's love to show off these animations.

But yes, if an ME sees too much lateral flex in the below, or getting too close to the failure point, then he can choose to add additional layers to certain parts. Or choose to upgrade type of fiber to portions, etc.

It seems pretty legit to me. The ME's I've worked with have never "gotten it wrong," e.g. we've never had a part unexpectedly fail (except for manufacturing or assembly defects which is a whole other story).


of the stress, like this:

I think we are also discussing two different things in some ways, you are describing the bike industry and I was talking about the general understanding of the phenomenon. They are both very complicated, which is why I qualified my statement with "a bit". I would stand by my statement that fundamentally the fluid dynamics is less well understood, I mean turbulence is not really fully described. In fact we do not know if we can even describe it in all cases.

I do agree with you on the vast majority of what you say. Especially when you see company X describing how they now have improved their analysis ability and you see what they were doing before. So when 5 years ago they added some crazy expensive nano-filler to the resin, they did not really understand how the laminates interacted with each other before, so did they really need that crazy expensive filler before, could they even know that? Of course I think fundamentally they are more conservative with structural matters than aerodynamics, they may get sued if their frames break, but probably not if the aerodynamics did not work out.

Also wind tunnels existed before 1909, the Wrights had one in 1901.

I would doubt there is anyone that go, hmm we need .5% more stiffness in X direction and need 5mm more deflection in y direction, so I can add this ply in this orientation in this location and this shape to achieve this. They could make good guess and once it is modeled they can tell the difference and have high confidence that the physical testing will match their models. I do not think all bike companies can do it and not all frames are analyzed are down to that level either. There are some that could compare the results of two models and understand the change that layer had on the properties.

It is probably some post graduate training, maybe through a university or through some other provider. The basics will be taught in undergraduate, but the level of analysis you are talking about is not really an undergraduate thing unless it is a very specialized program or they are working on some extra-circular program.

Aerodynamics is inherently non-linear. Even the simplest assumptions of inviscid incompressible flow is non-linear, Euler flow. Turbulence is not deterministic, so you can only characterize it statistically. It isn't until you reach airplane cruise speeds that linear systems of equations can be used to predict lift, using potential flow theory. Mach numbers between 0.8 and 1.2 ish are also very nonlinear and difficult to simulate, but supersonic flow up to Mach 3 or so are actually easier than subsonic flow.

Most mechanical problems can be handled with linear systems of equations, and that's what you're taught in a course in finite element analysis, hopefully not how to be a CAD end user :). There are practical complications, but that's why everything is built with a factor of safety.

Optimization theory is more of a graduate level applied math topic that only trickles down to most engineers through easy to use software tools.

Must not have taken the elective course "Sarcasm in Engineering". While I'm sure you're super cool in real life, this post has thus far confirmed my perceived stereotype of SparkE's ;-)


Contact is inherently non-linear, but can be ignored in all but a couple critical areas
Loading given large displacement is inherently non-linear, but displacement on carbon road frames are generally small enough to ignore (except maybe the old Roubaix)
Plasticity is, by definition, non-linear (if a company is taking advantage of material plasticity in designs, please let me know so I don't buy their product)
Fatigue is, by definition, non-linear and should not be ignored
I would assume you are 100% correct, bikes are the product of completely linear analyses - whether that is applicable or the company has the capability to do anything else is a different story.

Collectively, academia/industry has agreed to simplifying assumptions. Doesn't everyone - actually looking to accomplish meaningful work in this flow regime - assume inviscid, incompressible, irrotational (although I suppose that's redundant after assuming inviscid), steady flow? So we test in the tunnel to gauge the disparity between computational and experimental results and attempt to quantify the effect our assumptions have on results, although 30mph isn't really what some people would consider tricky (ok, maybe the interactions are worthy of experimental verification and there are a couple areas where the inviscid assumption would be detrimentally non-conservative). It's not like the N-S equation was analytically solved before producing the newest superbike, although every marketing department would have you believe that...

Same story with structures. Do we assume homogenous material properties for composites? No. Orthotropic? Technically should be fine but if we're all going to be pedants about our field of specialty, no. Anisotropic? Yep, and having all those material properties with statistical significant which is representative of the individual's processing is expensive. What about element testing to qualify a certain design feature? Population statistics on the assembly? All those cost money and would sell fewer frames than a cool paint job by a long shot. On a good day we can get our models to correlate to test data in either strain or displacement, generally not both simultaneously (although the degree of anisotropy is not helping). At the end of the day, a [reputable] FEA code really is only good at predicting sensitivities given the right inputs (material properties). Remember, Solidworks and ProE/Creo/Wildfire are modeling software packages which include a FE module to sell licenses, which is nice if a small company doesn't want to spend money on a FE license and analysis tokens for an accepted good solver to produce results which don't garner the same wow factor as ultra sweet wind tunnel results. The redistribution of load given contact is very much a gray area in structures; we use three really good solvers (~$50k/seat/year each, not counting analysis tokens) for both implicit and explicit, linear and nonlinear analyses and only one solver really does an acceptable job (which does not begin to describe how computationally expensive that analysis job is) yet we and everyone else have contact everywhere - one of my recent projects modeled contact between components as not including it was too conservative, each job takes ~32 processor-hours on 50gb of RAM (even on 16 cores, iterating on a 2-hour lag is irritating). To get those results, a license had to be purchased for the pre/post processor (say $50k) and 16 tokens ($20k each), plus my burden rate. Or they could put out a gif like the one posted above from Creo Simulate, 99.99999% of the people would have starry eyes and I just click back on my browser without even reading - apparently a converged mesh is optional?

The level of knowledge probably differs. smaller companies with less engineering resources are probably making 'educated guesses' based on material properties from published manufacturers. Bigger companies with actual FEA engineers and more testing resources can actually measure this.

I'm an FEA engineer in the biomedical field. This sort of knowledge IS typically postgraduate work, although one could probably pick it up in industry with a couple of years of experience.

typically, the other posters (like chapparal) are right in saying that we typically only simulate/test as much as we need to, because it costs time and money for not much more benefit. If you need really accurate results of intricate simulations, these can take years of work ( and is delving more into the realm of academia than inudstry applications.)

There are some things left out for brevity and audience; in reality nothing is "linear". Sure, real material properties can be modelled with logarithmic strain, and some software can do non-linear analysis, but as you mention, most of the analysis should be linear as we assume these things aren't plastically deforming or breaking, or flexing to some larger-than-is-appropriate for linear assumptions ranges. Most classical treatments of finite element structures are linear, and "contact" forces (being non-holonomic for engineering or math geeks) are probably rarely modelled outside of say bearing related designs, and instead are input as constraints, which are simply boundary conditions that are nowhere near as complicated.

The contrast is that in aerodynamics, the linear assumptions are usually not correct. Nobody assumes inviscid flow for a whole domain, especially at lower Reynolds numbers, otherwise you wouldn't have boundary layers, wakes, lift, etc. CFD solvers have to "close" the Navier-Stokes equations with one of various boundary layer models, which have their own faults and costs, but in general it's a messy affair.

I do agree with your comment on FEA analysis that rarely has decent mesh properties though - It's probably the most important thing in any discrete analysis that people tend to abuse.

Holy gobbledegook. We are not talking big machines& missiles here but plain 'ol bikes - and the engineering budget would support a couple of people in design, pencils and napkins in hand. Sure the CFD programs are nice but at the end of the day I'll bet it's mostly empirical/design touch, then to tunnel or crude tunnels, or based on "getting rid of xxxxx". (where xxxx= take your pick!)

And it's over - there's not enough static stuff on a bike to be more much of a difference aero-wise. Weight is probably not even negotiable now.

That is why I thought innovation would come from other areas such as in the data/electronics and automation fields.

---

Training Tweets: https://twitter.com/Jagersport_com
FM Sports: http://fluidmotionsports.com

Engineering can get as complex as the budget and personnel allows. As for not being "big machines & missiles," what matters is a designer's ability and desire to make the best designs possible, in any arena - going that extra analytical mile to smash the competition :)

Can it be napkins and wax and downhill tests? Sure, and you can likely get somewhere with reasonable logic and not a lot of expertise. What's the saying? - "steel is cheaper than brains."

In the aero case, tunnel time and part designers may be cheaper than a few specialist experts, but eventually you'll get to a ceiling of performance. If time and competitiveness is a concern, the design teams with better analysis will win out over brute force iterations.

There was an article here on ST a couple of weeks ago, describing why your expensive frame (in the same mold) is so expensive. It showed an awesome picture of a sample frame (sorta) ready for being put in the mold. It may have a couple of hundred patches of carbon cut to the right shape and put in the right places in the right orientation. That's what engineering is all about.

It's relatively easy to generate an aero shape, but it you just made it with constant thickness of carbon it would either be a noodle/break, or be way too heavy. The engineers use Finite Element Analysis to determine how much of what type of carbon goes where in the frame. That way you can have a complaint frame without being a noodle and a rigid bike without being heavy and so on.

If you could ignore the R&D costs, then frames would be cheaper for sure, but high end bikes would still be more expensive because of A) more expensive materials and B) more time consuming to build up more pieces of carbon to put into the mold.

While I am a mechanical engineer, it's not in carbon frame design. I have some friends working for carbon yacht and mast builders that help me and I do repair damaged carbon frames, but only for my own use or for friends to use as indoor bikes. I'd hate for someone to get hurt on a bike I repaired not quite well enough. My timid foray into the carbon fibre world has at least lead me to have huge respect for carbon bike frame engineers. Even what you or I would call low end carbon bikes, are pretty well engineered.

---

TriDork

"Happiness is a myth. All you can hope for is to get laid once in a while, drunk once in a while and to eat chocolate every day"

Basically engineers have a set of tools and programs that help us determine the correct layup. First, you need a tube shape, as this has a tremendous amount of effect on stiffness. From there, you can do some FEA and determine your critical loads and load paths and beef up those as necessary with different layup orientations. Also, the nice thing about carbon is the ability to tune the frame. I.e. not all parts need to be uniform and you can add layers only where necessary. This cuts weight but generally costs more as you have to add labor to cut plys in a more precise manner.

Carbon layup is pretty well understood and really isn't all too hard at the end of the day if you understand the principles.

Also, important to add you could spend years fine tuning a fe mesh and altering the layup etc. Often it is the most budget and timeline friendly to converge quickly on a solution based on prior knowledge and tune as necessary depending on any abnormalities seen in the simulation. Then there is always the iterative process if someone higher up determines it doesn't feel right.....

 
Would it be correct to assume that modern carbon frames are just as or more structurally sound , not because of the carbon tubes properties themselves but the ability to make stronger / large joints ?

With traditional metal tube shapes the methods of joining of them together was / is pretty limited. I suspect joint failures to be much more cause for engineering concern than tube failures.

I think I am correct in saying the carbon has opened up the ability to move material towards the joints where it is needed and away from the tube members where it has less impact.

Any material you choose will have strengths and weaknesses. For a bike, metals have an advantage particularly if they are in a temper/heat treat that is ductile. I've ridden a bike with a dent in the top tube that caused me no concern whatsoever. Try that with composite! Composites can be tricky because of their damage (in)tolerance. Isn't it odd how a carbon frame can survive a crash at 25-30 mph but at other times breaks falling over at a gas station rest stop?

You can make robust joints with metal, so carbon has no advantage there. Of course, you won't find many frame designs today with beautiful steel lug work. Metallic frames tend to have welded sections that butt nicely to each other rather than into end fittings that are then welded.

Where composites really have an advantage is when the loading and response are directional in nature. The tubes on the bike primarily are taking loads along the lengths of those tubes, so by biasing fibers towards the length direction, you can achieve much better stiffness/weight and strength/weight ratios compared to metals.

Composites can be attractive for many reasons. If you have well defined loads and those loads are in a "safe" region relative to the allowables, they can be fantastic. But if you have an overload event (structures aren't designed/built for the worst ever loads they may see but some statistically based load that makes the likelihood of exceedance quite low), I'd much rather have ductility in my back pocket.

And the FIA has regulations regarding the stiffness of body elements. They don't want adaptive structures which can benefit the aerodynamic performance of the vehicle (i.e., reduce that drag on faster sections, provide more grip at slower speeds). The real kicker is when minor damage is sustained and the cars continue to keep pace or are faster with damaged aero elements. Mother Nature still knows more than we do in terms of fluid dynamics, structures, and the interaction of the two.

They do not want flex able structures and their rules state that the nothing can move, but no material is infinitely stiff. The real rule they need to meet are the load deflection tests the FIA perform at the races. They are allowed some deflection, so there is lots of effort to made to make the most of this allowed deflection.

Also, F1 has some real tricky non-steady state problems. Like the DRS system, where an element on the rear wing can be trimmed out to reduce drag, but then goes back to the high downforce/drag position once brakes are applied. The flow will not immediately attach back to the element, so there is some time before the downforce returns. So trying to simulate that how to get the flow back as soon as possible is real hard. Same thing with underbody flow, they stall the underbody flow at a certain speed to reduce drag, this also reduces downforce but at high speed they have way more downforce than they need, then when they are braking they need this flow to return as soon as possible so they have downforce for the corner. These non steady state problems are very difficult.

Actually, per 3.17 of the technical regulations, the FIA indeed does mandate stiffness requirements for body elements. The rules don't state that nothing can move. The "nothing can move" does reflect actuated surfaces, with DRS obviously exempt from that. But any "nothing can move" requirement would be in relation to active control surfaces, active suspension, etc. But you can get tremendous performance advantages by the inherent flexibility of the structure. Witness the Boeing Dreamliner. The wing deformations are not actively changed by internal actuation but rather the aerodynamic forces. My understanding is the wing was designed to flex to a more beneficial configuration during nominal flight for better efficiency. FIA doesn't want that beyond what they deem acceptable, thus the stiffness requirements per 3.17 of the technical regulations.

Would be interesting to see the extent of their aero analysis. Do they perform Monte Carlo analysis to effectively capture nominal and "3-sigma" "trajectories" or do they use point solutions. And just how good is that analysis in the case of being in the wake of one or more cars? Some things CFD and wind tunnels just can't get a completely accurate depiction of.