I like these sets of picture because it makes it easy to understand the concept and its impact:





It shows area in front of the steering axis is the leading factor.

More importantly it shoes the Cp location is dependent on the sum of the pressure gradients front and aft of the steering axis. Basically, the side with the most red is the one getting pushed and that tells you which way the wheel wants to turn. The impacts of how the bike will react to that are well discussed earlier in the thread. The greater the front/aft ratio, the further the Cp will be from the steering axis, and that is one of the important bits.

Now that makes it very interesting because it is universally accepted a disc is harder to ride at high yaw. Looking at this we see that a disc will naturally correct to make a rider lean into the wind, which at first sight makes it puzzling.

The problem is having the Cp on the steering axis does fix instant steering torque, but the corrolary is that the further the Cp is from the steering axis, the greater the torque moment will be. Now the second important bit is that means as effective yaw angles rapidly change while you ride (hello hambini), the torque moment deltas that needs to be dealth with increase. Your handlebars get thrown around more and you constantly try to control for it. Now that explains why a font disc can be a pain.

As it applies to deep wheel design, the problem is the Cp location is also a moving target, it changes with effective yaw angle.

In 2011 we built a fixture to correlate the Godo CFD to wind tunnel data by attaching a fork to the sting (overhead balance) at the ARC here in Indy. This is the sister tunnel to the Mercedes tunnel in England run by Simon Smart and used by a bunch of brands, Zipp has since built a traditional balance and is doing current testing there.

We originally set out to develop an dedicated overhead balance that could be used on the rolling road surface of the tunnel and which would allow us to do some serious handling investigation, but it became too expensive and complicated for something that the general public would just say 'doesn't look real' and ultimately would have been too limited in test scope anyway. However, we did manage to test about a dozen wheels in one fork attached to this overhead balance where we could measure real time drag, side force and steering torque in the steering axis. From here we developed an equation to transform this data to/from the vertical Y axis data that we could see at the San Diego LSWT and built a CFD model of that system to compare to the wheel in fork in bike CFD we were doing with Godo though we only had the data for that particular fork with those wheels. By 2011 working with Godo we had a paper presented at the 2011 AIAA conference which you can read here: https://arc.aiaa.org/doi/10.2514/6.2011-1237

This paper shows some CFD work done to correlate CFD data to the ARC tunnel data using a Reynolds and Blackwell fork (which feels like forever ago just saying those names..). I'll see if we can post the picture, but we tested it and the Cp data from CFD and from San Diego correlated to the Cp data in a fork (the Reynolds) from the limited study at ARC. So to answer the question about data on Cp, yes it exists and correlates consistently within 5% or so to the CFD test data based on ~10-12 wheels tested in one fork.

However, with hindsight I'm convinced this is not exactly the fundamental question, which I think is the question that I think Slowman is after. This question is probably something like: how does front end system Cp and side force magnitude directly relate to handling in various cross-wind scenarios from the perspective of a real rider? Ideally we'd want a model for steady cross-wind, transient gusty cross-wind and the one I wanted Godo to build for me in CFD, the 'rolling wave' cross wind which would be analogous to a passing car. We would want to correlate both CFD and/or tunnel data to some sort of repeatable real world test

When I was doing wheels, we only really paid attention to the wheel Cp and worked on making the wheel Cp stable in numerous forks, and by numerous I mean like 3-4 of them as each additional fork added like hundreds of hours of CFD meshing, running and post processing time. I think there is still significant potential here in looking at the system dynamics of the frame/fork/wheel/ handlebar combination as all of them play a roll in the system behavior. I was always interested in the super skinny laterally ovalized grips of the original 3T Ventus as our CFD models showed that this design reduced steering torque when in the aerobars.. but it was offset by riders not liking to ride them in more than flat TT type conditions.. but it shows that there are some pretty large effects in play here outside of the wheel.

It was Richard Cunningham and Doug Milliken who helped turn me on to the contact patch/countersteer theory we have today.. I'm not saying it's bullet proof or even correct, but when we were at a low point trying to understand why our more highly rear Cp wheels (originally Firecrest was a sort of a project to make a rudder like front wheel) they stepped up and suggested this theory which made sense. From there forward we looked at lots of shallow wheels, all with Cp in front of steering, and wheels that people didn't like, all of which were Cp rear of steering or had Cp crossing the steering axis with yaw and went in the direction we went... but of course there are a LOT of variables unknown and unaccounted for, the biggest being fork interaction. Our work with Godo showed that the fork could have a significant damping effect on front wheel harmonics and also could move the wheel Cp around quite considerably, but again, we were really just focused on making a more neutral wheel with the idea that we couldn't control the fork that was being used.

The ultimate answer here needs to come from somebody independent who has the chops to make a true experiment out of this, like the crew at Alphamantis. We tried looking at steering input and response using 2 Alphamantis Aerosticks, one on the head tube and one on the handlebars plus an angular measurement sensor on the steerer tube.. but while we had some good directional data, mainly wheels that people thought were a handful in the wind really did show larger amplitudes of both initial and corrective steering inputs, the reality of outdoor testing for a variable like this is that the repeatability is very poor as its so hard to replicate the input conditions.

---

http://www.SILCA.cc
Check out my podcast, inside stories from more than 20 years of product and tech innovation from inside the Pro Peloton and Pro Triathlon worlds!
http://www.marginalgainspodcast.cc

Could you possibly setup powerful fans along a track, blowing sideways along one section? You'd have replicates from each lap, a constant wind speed, and the ability to turn it on/off at will. I'm sure it would make their normal measurement much noisier than it usually is, but it'd be way better than testing in the wild, no? -J

---

----------------------------------------------------------------
Life is tough. But it's tougher when you're stupid. -John Wayne

Hi!



Here's data from our original experiments done in 2012 by KQS while developing the LG Gennix TR1. Experiments were done at the NRC Windtunnel in Ottawa, Ontario, Canada.

Before anyone starts blasting me about the precision of this test (different handlebars, suspect tying of the force gauge, etc) I want to explain that this was our first test to determine whether steering torque was a real phenomenon and whether it can be affected by equipment design. The answer to both is yes.

This doesn't answer whether the Cp should be ahead or behind the steering axis or whether this makes the TR1 fork more or less stable. Our determination was that the fork geometry was unable to completely offset the side force component of a deep wheel. All it could do is reduce the magnitude of the unbalanced force (or steer torque) and that was our goal.

KevQ

KQS
http://www.kqbikes.com

Knight Composites
https://www.youtube.com/watch?v=MC6MJTG4t9A

Thinking more about it, the deeper the wheel, the deeper the fork blades and rake should be to move the Cp back over the steering axis. Logic has it Cp on steering axis results in near zero torque. If all else fails put some lead in the bullhorns :)

take a look at what kevin quan wrote just above you. it's not necessarily the depth of the fork blades, but where that surface area sits - in front of or behind the steering axis.

also, for those who're showing me simply wind tunnel or CFD data on wheels, remember that it's the fork, the wheels, the aerobar, the hydration system, the rider, it's everything that attaches to the steerer that gets hit by the wind, and the force of that wind x the lever, possible x the weight of whatever that thing is (imagine a hydration system sticking out 2 feet in front of the steering axis).

---

Dan Empfield
aka Slowman

It's an interesting topic! Problem is the effect of yaw on the wheel is so variable. According to Trek flow separation typically occurs first on the leading rim section. That means side force should drop as you go from lift to stall. Meanwhile the rear section of rim is doing something else... maybe still lifting or not... and will be effected by the fork. And whatever is happening will likely oscillate on some schedule...

I guess what you are wanting to know is... is it better add surface area behind the steering axis or in front? I'm no expert, but this is my take on it. If it's an airfoil, as yaw increases from zero, the side force gradually increases until separation, then it drops. When yaw declines and the rim (or fork or whatever is attached to the steering axis) behaves like an airfoil again, the side force will jump up again. If the flow is in that regime where we have attachment, the wheel should push the steering axis in the same direction the wind is blowing, but it will be sedate. When separation occurs in the forward part of the rim first, then the net force will shift to the opposite direction. If the wheel turns in that direction, then you'll probably get attachment and the net force will shift again, repeat...?

Any airfoil shapes you attach to the steerer are going to lift or stall at different times relative to the rim, so I don't think any place you put them is going to aid in controlling the bike. It would only help if you could make them behave exactly counter to the rim's behavior (or exactly the same but on the other side of the steerer).

i'm collecting info and data now from all the relevant wheel companies on their narratives for lift and stall. mostly, what are you doing to make that transition less abrupt? we can all deal with sideforce. what puckers us are abrupt changes on sideforce. the fastest wheel i ever rode in my field trails was a bontrager aeolus 90. it was also the scariest. just on a flat road. sail. stall. sail. stall. shitting my pants to a PR.

so, yes, you have the wheel's own discrete behavior inside the bike and then you have whatever the bike is designed to do. which makes this an impossible construct to test except just via field trials, or by someone who just is smarter than i am. or both.

---

Dan Empfield
aka Slowman

I'm not convinced that it can't be done in a tunnel. I think we can take a pretty fair stab at the sort of force variations that would be hard to control. On second thought maybe the rider's steering reaction to this force is the part you'd need to test in the field?

i now feel like i need a physics or engineering degree to determine which wheel to buy. Was about to get a Jet4. The 60 i use is fine on most days but i struggle with exactly what you described....abrupt changes in wind direction or wind speed (clearings in trees and crosswinds, trucks going past on breezy days etc) and figured on days like that something shallower was needed - and it could live the rest of the time on my road bike. From reading this thread i could just need a 60mm wheel with a different cp and getting the 46mm Jet4 may or may not be better than the 60 i have.

.....or did i completely misunderstand (would be the first time when things get complicated)

i'm not smart enough to know how to test pressure during a field test. but, change in direction during steering, that's pretty easy. if we assume that, for a given rider, the reaction to inputs remains constant, then this might be a reasonable way to test this.

---

Dan Empfield
aka Slowman

We actually agree :)

That's why I threw rake in there to act a a tunable variable that offsets the fork surface area relative to the steering axis. There are other steering considerations with rake which I'm sure you know much better than I do, but I just want to focus on Cp tuning.

I think we can agree the ultimate goal is to maintain the Cp on the steering axis through the whole yaw range.

So, just thinking out loud we know the Cp moves from back to front with increasing yaw on a lone wheel. I will postulate that generally the level at which it does it is proportional to the surface area of the wheel. We then need an arrangement of the whole steering system so that the pressure forces on the wheel can be offset equally by pressure forces elsewhere in the system, or dampened through inertia.

So how can we achieve that? I might be wrong, but I'll postulate once more and say that handlebars play a minimal role, or should be designed to do so since no part of it is behind the steering axis. It can't help, but poor design can exacerbate the problem. So a thin side profile is the target. That leaves the fork. As I mentioned before we can play with rake and profile. The more rake we go with, the smaller profile we need for the same impact on Cp, but it needs to be tuned to the depth of the wheel. We need to tune these two so it's set to progressively offsets the movement of the Cp of the wheel with yaw. What we might need to investigate at this point is the impact of fork stall on the moment of torque. If it's an issue, come up with a fork profile that doesn't stall easily. Some will laugh at me, but VGs (or texturing) should be investigated here.

Just not possible. Unless all the components are "bluff bodies" (slow, high drag) then you are going to experience transitions from lift to stall and back again, and a large variation in side force. Manufacturers have made gains in achieving a better balance between the front and back of the rim, but it's never going to be perfect.

I'm thinking an active aero element is necessary. One that purposefully counteracts what is happening to the rim.

Hello rruff and All,

Food for thought ... 'blue sky' speculation or 'spit balling' ....................

Modern aircraft have a control problem because of the long tube like fuselage .... and small control surfaces.

While more of a problem in high performance military aircraft .... it is also a problem for transport aircraft.

Aircraft like the F8U had roll retrictions because of yaw coupling .... for example only one max180 roll or one 360 roll depending on configuration .... exceeding those restrictions leading to the aircraft becoming unstable and trying to swap ends.

The angle of attack of the long tube in aircraft is often not along its central axis and when it is rotated about an non central axis the inertial coupling causes instability.

https://en.wikipedia.org/wiki/Inertia_coupling

The yaw coupling is called 'Dutch Roll' and caused crashes in some early Boeing 707 aircraft.

Before flight sims when flying the 707 we would practice Dutch Roll recovery during recurrent pilot training as once the instability was induced it continued to increase if no counter action was taken.

The trade off is larger control surfaces that cause more drag or restricting the flight envelope .... or augmenting the stability of the system with devices like yaw dampers.

We apply these policies to bicycles too .... restricting the use of disk front wheels or deep rim front wheels for some condtions ....

Perhaps we could introduce some stability augmentation utilizing the large rudder on the new Specialized TT bike.

Attaching a yaw damper to actuate a rudder (possibly mounted fore or aft on bike) or introduce a small steering input to the rear wheel when the bicycle has stabilty excurisons ..... could automatically smooth the effect of gusts .... similar to the effect of the yaw damper and stability augmentataion on aircraft .... and result in a bicycle that is difficult to upset in gusty cross winds.

---

Cheers, Neal

+1 mph Faster

That's what occurred to me, although I think it would need to be attached to the steerer to work well, since this is the sensitive area.

I just did a quick online search looking for wind induced steering torque on a bicycle and got nothing. Lots of articles on wheel side force vs yaw, but that is a different thing.

At the core of all this is the fact that due to the flexibility and interchangeability of the bicycle as a system, all these various companies are generally trying to bring system level improvements though development at the component level.

This is the opposite of aircraft, where the system is owned and refined heavily by a single entity and then limits are placed on the final use case.

It also points to the challenges we have with current testing, something Dan has pointed out here.. most companies continue to test at finite yaw angles in the tunnel starting at zero and yawing outward.. this gives good and generally repeatable numbers but doesn't account for the hysteresis of flow detachment and reattachment. This was something I really liked about what the Hambini people were trying to do.. clearly any test is going to have it's flaws and biases, but something like the Hambini test regime is going to lead to better designed with higher angles of flow reattachment and most likely better flow control on both front and rear wheel halves.

Remember it was just 8 years ago now that we started paying real attention to the rear half of the wheel, so all of this is pretty new to everybody. Before that, everybody was optimizing the front half and getting what they got on the rear. Now companies are at least looking at both.

As for Cp migration, it moves around quite a bit depending on wheel design. This was one of the last things I was working on when I left Zipp and it was becoming clear then that to really optimize this, we needed to own the entire system as even small interaction effects could be quite large. Here is a study we did on the then not released Scott TT frame and fork and found that the fork was actually causing the Cp of the wheel to begin moving forward after 20 degrees of yaw which is not something we saw in wheel only or some other forks.



I'd love to study this on the Canyon with that large surface area front hydration box. I've asked Jan if he can tell a difference in handling with/without it and he said 'not really' but seems that would be a great test bed for looking at optimal Cp location for the entire front system of the bike as you can easily add/subtract surface area.

---

http://www.SILCA.cc
Check out my podcast, inside stories from more than 20 years of product and tech innovation from inside the Pro Peloton and Pro Triathlon worlds!
http://www.marginalgainspodcast.cc

Thank you for sharing that. If I'm reading that correctly, the cp for the wheel moves steadily rearward until 20 degrees and then moves forward. For the frame, the cp moves steadily rearward until 25 degrees and then moves forward. For the full bike, the cp moves steadily rearward until it experiences a rather large move forward between 25 and 30 degrees (large relative to the other changes seen). I would assume it's this discontinuity that makes a bike feel "twitchy" in a cross wind. In my mind, a disc in the rear of that bike would keep the cp moving steadily rearward.

Do you have any thoughts on how the presence and interaction of a down tube would affect the center of pressure? One thing that has always bothered me is that we test in the tunnel with the front wheel and the down tube perfectly aligned when it is not aligned in the real world especially at yaw.

Anecdotally, running something like the Torhans 30 on my bike made handling in crosswinds much worse. Other thoughts:

• Tire selection likely plays a large role in CP behavior.
• I'd imagine a skinnier fork (fore/aft) would affect CP behavior less (correct me if I'm wrong)
• I'd imagine a wider fork would affect CP behavior less
• At higher yaws I'm sure drag from the spokes starts to become a significant factor in the behavior of the CP.

Anecdotally, I had an Aeolus 7 D3 that was one of the fastest and most terrifying wheels wheels I've ever ridden. I had two incidents where a strong crosswind hit me descending causeways in coastal Florida. Both times said crosswind blew me half a lane over and I was lucky to stay upright let alone not get hit by a car. After the second time I bought a 303 which was a night and day improvement in handling. Since then the deepest front wheel I've run has been a Jet 6+. Even though I never ran the two wheels back to back, I can say with confidence that the Jet 6+ is far more stable than the Aeolus 7 D3 despite the rather minor difference in depth.

Proof.
Pudding.
(However he's probably got lower yaw angles than most ;)

With all the words written so far in this thread it's a shame noone has yet done my proposed cardboard-and-duct-tape experiment. If so, you'd have an answer by now.

Great thread; thanks all for the education. It seems from the discussion that an unavoidable consequence of making a wheel which reduces drag in "normal" conditions which is then going to be installed into a bicycle made by someone else who may or may not have had that wheel in mind when they designed the bicycle is that sudden changes in yaw (even over relatively small ranges in some circumstances) can present the rider (who has a non-zero reaction time) with a correction to make. That correction is performed by a human brain acting largely on autopilot (I'd argue that only the decision to get on the hoods and drop anchor to minimise the consequences of a loss of control is a conscious one). That split second correction could initially correct or even amplify the problem. I'm not sure a computer can do the job of modelling this part.

With that in mind, I'm struck by this comment (above). Do manufacturers have a duty to solve for one or all of these on an eyes-open basis? For example, I know that my bicycle helmet is designed to absorb the energy from a fairly specific range of impacts which are most likely to befall me [I was once told that putting a melon in a helmet and seeing what happens if you drop if from a height of around 5 feet is as good a way as any of testing] but not others (if I'm going to ride into a tree at 50mph I'll be needing a motorcycle helmet). At some point anyone riding the wheel will ride past a gap in a wall or hedge with a strong wind blowing through it, and at some point they'll be overtaken at close range by a truck. So isn't the way to test for that to have some "typical" riders ride past a massive industrial fan whilst trying to stay on a line, and have a rider try to stay on a line whilst a truck goes past at a distance of (say) 8 feet at (say) 60mph and see what happens*? A wheel which passes the former test but not the latter would be fine for a velodrome, or a closed road race, but not riding on the open road.

It's heartening as a consumer of your former employer's products to see that you've thought about this a lot. I worry that your former employer is competing with manufacturers who would not be so diligent ...


* I'm not volunteering