Before you tar and feather us as snake-oil salesmen, please remember we, we made more than 60 computer and physical prototypes of our dimpled hubs and tested a dozen of these physical prototypes in the wind tunnel in determining the final production version. I'm not sure what else can be said about the data other than it just is what it is. Faster.

As for the CSC and Phonak dimples, we also made one physical sample of this before committing a tool and found no discernible difference from our standard dimpled wheel, which was as predicted since the dimples serve really only to create a turbulent boundary and the logo is within the dimple field, so the airflow over that part of the wheel is generally already turbulent. Also, the hard edges of the logos are theoretically very good at converting the boundary layer, so it appears to be a net wash in terms of the data, but it sure is cool, and is a great example of the advances we've made in this tool manufacturing technique.

We make all of our own tools and develop our own processes as well as make our own products, so there are hundreds of technologies that make our processes and tools special that are not necessarily marketable or sexy, but this is one of area where we felt that we could show our capabilities a bit, and I think it is quite cool. And the team riders also thought it to be quite cool as well, so in my opinion if we can do something that the riders and sponsors appreciate with no net penalty, why not?

Here's a couple links to some more dimple information in the tech section of our website:

• Racecar Engineering Article [pdf]
• ABLC™ (Aerodynamic Boundry Layer)
• Boundary layer

We have tons of stuff on this at the website, but here is an interesting third party article on the topic pasted below. As an aside, the 3M riblet film mentioned in the article (which requires much higher Reynolds number flows that we see in bicycles unfortunately, but we've tried it anyway) has been outlawed in pretty much every type of boat and aircraft racing worldwide, if we rode underwater we would all be paying a fortune for riblet films.

The key to making the surface work, is based on dimple pattern, dimple shape and depth of dimple. In the old days, many golfballs had positive protrusions on them, and that worked too, just not as well. Now companies spend millions trying to make the dimples on the balls more efficient, generally packing them tighter and trying odd shapes. There are actually balls that are illegal for tournament play because they fly too far in the PGA 'Iron Byron' test.

For the record, the Zipp dimples are icosohedral (20 sided) and a few thousandths of an inch deep, the dimples at the perimeter and towards the top of the wheel are the most important, and letters, actually form a very random and efficient pattern, but their location near the hub minimizes their importance and effect. We tried about 10 different patterns in the wind tunnel before arriving at the pattern currently in production, and they were definitely all not created equal, and yes, some of them were worse.

In the wind tunnel, the result is between 2 and 6 watts of efficiency improvement over the smooth disc between 5 and 30 degrees of yaw, and between 0 and 2 watts improvement from 0-5 degrees. As an average that's like 6-10 seconds per 40km, more if it's windy. We see even better results in the rims due to the sidewall curvature and rim shaping but that's another story. Lastly, the new Lexus LS400 uses a dimpled underbelly not so much for drag, but to minimize noise. Since it takes energy to create sound (we all don't want higher wattage speakers for nothing) this means it also operates at a higher efficiency though this clearly was not the primary goal, I think this is a really clever use of the technology.

Here's the article text: Providence, RI--More than a century ago, golfers noticed a curious phenomenon: Used golf balls flew better than new, smooth ones. Random patterns of dents on the used balls actually stabilized their flight and, in some cases, made them go farther. Eventually, the golf world incorporated that knowledge into the design of new balls, providing them with dimples.

Now, engineers may benefit from a similar phenomenon that offers potential improvements in disciplines ranging from aircraft design to pump operation. A recently published study in the journal Nature describes a technique for reducing skin friction drag by up to 13% on aerodynamic surfaces. Aircraft engineers say that if such reductions could be achieved on production aircraft, it would be a significant step forward. "When we try to reduce drag, we normally work in second decimal places," notes William W. Greer, vice president of engineering and quality assurance for Learjet (Wichita, KS). "So 13% represents a revolution to us."

Drag reduction has long been an area of concentration among aircraft designers, for good reason. Some estimate that a typical commercial aircraft can save between $100,000 and $120,000 a year in fuel costs by cutting overall drag just 1%. Worldwide, a 1% drag reduction could translate to fuel savings of more than $1 billion a year. What's more, the new technique could apply to a wide variety of other applications, including designs of ships, submarines, torpedoes, pumps, HVAC ducting, mixing machinery, and pipe flow systems.

Random pattern is key. The new reductions in skin friction drag involve the use of a simple aerodynamic surface geometry. Researchers Sture Karlsson, an emeritus professor of engineering at Brown University, and Lawrence Sirovich, director of the applied mathematics laboratory at CUNY/Mt. Sinai in New York, found that a random pattern of small bumps could lower skin friction drag. The size of the bumps, or so-called "chevrons," is determined by the thickness of the viscous wall layer on the aerodynamic surface, they say. Though the precise size of the chevrons is proprietary, they do describe them as being on the order of a "fraction of a millimeter."

The real key to the extraordinarily high reductions, however, is not the size of the bumps, but their overall pattern. Sirovich and Karlsson found that the chevron pattern must be random. When using a random pattern, they achieved skin friction drag reductions between 12% and 13%. In contrast, aligned bumps not only failed to lower the drag, they raised it by as much as 20%.

The theory behind their findings is complex, but it basically says that the random chevrons reduce "bursting" near an aerodynamic surface. Bursting, which is caused by low speed air streaks near the wall, is generally believed to be responsible for intense turbulence and, therefore, greater drag.

Sirovich and Karlsson aren't the first to employ surface geometry changes as a means of lowering drag. Researchers have been trying to find ways to "energize the boundary layer" for almost two decades. They've tried putting microscopic holes on aerodynamic surfaces to suck the boundary layer off, and they've applied grooves to wing surfaces to decrease turbulence. 3M Aerospace is currently testing Drag Reduction Film, which employs small riblets in the air flow direction. Used on an Airbus A-340, the film has been successful in reducing drag. Cathay Pacific Airlines applied it on one commercial aircraft that flies long routes, and it has achieved drag reductions of approximately 0.8%. "It has allowed Cathay to fly from Hong Kong to Toronto without a refueling stop," notes Tom Ihbe, 3M Aerospace market development supervisor. "Sometimes in the past, they had to refuel in Anchorage." Ihbe adds that the riblet film offered drag reductions ranging from 2-4% in wind tunnel tests.

Up to now, however, no one has achieved skin friction drag reductions as great as 12-13%. Because the figures are so high, some experts believe that airframe manufacturers might be inclined to invest substantial capital in it if the concept proves out.

Still, they warn, much remains to be learned about it. "Its success will depend on how it affects an aircraft's handling and sound characteristics," Greer says. "It will also depend on how maintainable it is and what it costs." One way to implement the concept, Greer says, is to program CNC systems to ma-chine the random chevrons into the wing skins during regular machining operations.

Business jet manufacturers might be willing to make such accommodations, not because of fuel savings, but because of the additional range the concept could provide. Greer says that the newest Learjet could extend its 2,000-mile range by as much as 120 miles by employing the concept. "For business jet operators, that's where the real advantage is," Greer says.

I know this is going to soud like a cop out but, the 650's are actually more expensive to produce than 700s because we make so few of them. Making dimpled 650 tooling is truly not an option due to price as it would take like 7 years to break even on the tooling for a dimpled 650 404 if we added $100 to the price. The dimpling roughly triples the price of the tooling and these are tools made in the US using US steel and labor so they are already about 10x the price of tooling made in China. This unfortunately creates a situation where we just cannot amortize these costs effectively with 650's selling at the current rates (and falling).

As for standard production, the tool changeover for a 650 rim is the same cost as for a 700, but making far fewer of them, the costs again do not amortize well. For example if it costs $4000 to changeover the tooling to make a 650 rim, program the CNC carbon cutting equipment, reprogram the CNC drilling machine, and so on, and you only then make 50 rims, it costs you $80 per rim in setup costs. If you then change to a 700c rim and have the same setup costs but then make 400 rims you only have $10 per rim in setup costs. These are totally made up numbers, but demonstrate the cost amortization differences, and again we are using US labor which is highly skilled, our people on average make more per hour than a Chinese worker makes per week, so a setup that takes a couple of employees a couple of hours is quite costly in terms of labor and lost production time.

We are committed to 650 production and feel that it is an important part of triathlon as well as a key factor in fitting small riders, so we have committed to making them even though they are not profitable. In light of so many companies killing off their 650 products altogether we feel that it is important that we stay in this business, but with the 650 market continuing to shrink it becomes harder and harder each year to justify staying in the 650 business much less advancing technology within that part of the business.

The dimpled hub photos seem to have stirred up some discussion here, so I figured I'd try to set it straight. We are unavailing a dimpled hub at the Tour that will be part of a new lineup that is replacing the Z series wheels for next year, more on that later. These hubs are very, very expensive to make and will not be appearing on standard Zipp wheels anytime soon.

I've included the real photo and the wind tunnel data below, so you guys can see what is going on. The dimpled hub is pretty cool because with the rims we find that we can tune the shape of the rim to achieve better aero performance at various wind angles, but there is always a tradeoff, so if we want to be more efficient at low wind angles we have to give up performance at the middle and high-yaw angle conditions and vice versa. But with this hub, we gained efficiency between 0-10 degrees (before the hub flange starts to shadow the hub body) with no adverse effects anywhere. For those of you already trashing the concept, you will be heartened to know that there is tons of data in aerodynamics handbooks dating back to the 1940's showing the effects of surface roughness and dimples on spheres and cylinders, so this is a very well documented phenomenon outside of cycling and the dimpled hub is considerably more analogous to the golf ball than the dimpled rim due to the similarity in shape.

We started with more than 20 dimple combinations with varying dimple depths, densities and diameters to find this one, and even the logo is cut into the part with a V shaped tool that leaves neat little air tripping corners such that we cannot see a difference between hubs with and without the logo. The testing was done by graduate students at Texas A&M without any Zipp people being present (for those of you who think we hang out down there and manipulate the data) and the the data below represents 4 run averages with each wheel to reduce uncertainty. We would have used Allied Aerospace as they have greater balance accuracy, but they cannot yet spin the wheel faster than 18mph, so the test would have greater accuracy, but be less realistic, so we chose to do numerous run averages to improve the data.

The jpeg file sort of fuzzes out the text, but the 808 with dimpled hub is the one with the dark red asterisk shaped data points. Interestingly, both of these wheels were tested with 19mm tires as we accidentally left tires out of the box so they used tires they had there, and you can see that neither wheel looks as good as it does with a 21mm tire, in the data either on our website, or in either Tour Magazine test, so one more reason NOT to run 19mm tires with our wheels.

My point that the concept of a dimpled cylinder in a free stream is a proven concept was just to make a point that this has been done before and tested in other applications so it isn't like we are shooting in the dark, but not to imply that a hub in a spinning wheel is equivalent to a cylinder in straightened wind tunnel flow, so I don't think that I nor anybody else was expecting this to mimic handbook or computer generated values (although none of us expected it to do quite so well either). I can certainly understand the skepticism, but we are talking about such small numbers here that this just doesn't seem that unrealistic to me. What is most likely happening here is that the reduced pressure drag behind the hub is resulting in cleaner airflow onto the rim which is giving us a beneficial secondary effect, but no I have not delved into this in depth to fully understand it and no computer model at this point can show the dynamics of the airflow onto the hub inside the spinning wheel to tell us how this is affecting things. As one of my engineers who worked for the last 5 years doing CFD and wind tunnel modeling for IRL and Champ Car racing teams here in Indy said, "sometimes you just have to accept that something seems to work and do it before somebody else does it and kicks your ass."

I opened the original data from A&M and have been really looking at it closely to see what is going on. This is where it gets tricky as people used to say that we manipulated the data to make our wheels look faster and all this, so we started relying on the tunnel to do all the data management and keeping the files locked, but now we have a situation where you guys are calling us out on this and we haven't even been able to really eat sleep and breathe the data. Anyway, using the 4 run averages, the drag reduction at zero degrees is 0.039lbs (17.7 grams). So the standard rule of thumb is that at 30mph (13.41m/s) 50 grams of drag is equal to 6.5 watts, so we get 2.3 watts savings. Now the margin of error of the tunnel for a 4 run average is nearly 0.02lbs (9 grams) so this gives is a spread of roughly 1-3 watts taking all of that into consideration (and I know that I said 2-3 in my original email which may also have been what was published originally, but the actual press release on this that went out to Bicycling and Pez (which went out over a month ago) said 1.5-2 watts so that is clearly my bad and should be taken as proof that either: a.) I am a lying, greedy, bastard; or b.) That we have 3 engineers managing more than 50 products and this one was finished and released nearly 2 months ago.

Anyway, Looking at the data, there is one outlier in the dimpled hub data that is a good bit lower than the others at zero degrees, and if I take that entire data set out, the maximum difference between the two becomes only .034 lbs (15.4 grams). Which slightly lowers the estimate to just under 2 watts, but still shows an average difference that it statistically significant and repeatable between the two wheels. The difference between the worst dimpled hub and best std hub is 0.017 lbs (7.7 grams) which would give us a difference of 1 watt.

If all goes as planned we will be back at Allied Aerospace (which has about a quarter the margin of error of A&M) later this year and will revisit the dimpled hub to see how well the data repeats as well as trying out some other new things, and if we have time we will try to get smoke on it to see what else is going on. Ultimately I think that this is a case of what John Cobb has said before: "sometimes 1+1=3 in aerodynamics". Which is just to say that changing one thing is sure to have unintended consequences (either good or bad) within your entire system, and at this point it appears that this is one case that the unintended consequences are good.

As I said before, we took a lot of crap from folks when we first posted the 808 aero data and showed it to mimic a disc up to 12.5 degrees, and now outside experiments have shown that to be completely true (actually better according to some independent data), so I feel that we definitely don't have a history of lying to anybody and certainly expect these products to be tested by both consumers (who seem to be frequenting the wind tunnel more and more these days) and magazines, so it really isn't beneficial for us to come here and be making crazy claims or outright lying to anybody when sooner or later it'll all be in Tour magazine anyway.

Wow, sorry for all the confusion, but the website seriously removed some copy from the press release we sent out to them. The original press release said that the dimpled rims save between 1 and 4 watts over their non-dimpled predecessors depending on wind angle. Then goes on to say that an efficiency gain of 1 watt is roughly the equivalent of removing 340 grams from the bike when climbing an 8% grade. I think in the attempt to condense the copy, they lost the meaning of the whole thing, so sorry for the confusion.

This 340 gram figure is something we started using when we developed the first ceramic bearings used in the bike industry, the savings of a ceramic bearing set over a cup and cone type bearing for a set of wheels is 1-1.5 watts, and since everybody is willing to spend all their money to save weight, but not necessarily to improve efficiency, we used a simple physics model to try to tie the efficiency gain to a weight savings.

The numbers are much funnier if you use a flat TT model, you get something like 3 kilograms offsetting 1watt, but this is essentially the model calculating the time penalty of accelerating the additional mass off of the starting line. The beaty of the 1-4 watt savings is that the shape of the wheels and the dimples have really allowed us to fine tune the drag profile of the wheels such that the peak energy savings occurs in the 10-20 degrees of yaw condition, which account for 80+% of the real world conditions, this means that 80% of the time the savings is closer to the 4 watt figure and even in the worst case is still roughly 1 watt more efficient than the previous generation.

There is quite a bit of discussion of this on our website in the technical section where you can download white papers describing all of this, and goes into detail on rim shape and reading aero graphs and so on. I guarantee you that no company in the world has put in the amount of time or money researching and perfecting these technologies (not to mention just coming up with them in the first place), plus the data used in the white papers was all compiled and averaged by the Texas A&M staff and uses averages of multiple runs, so we can guarantee that the data is accurate and repeatable, and not just something made up and thrown up on the website by some marketing guy.

For fun, I recommend using the <www.analyticcycling.com> to try plugging in some of these numbers and compare wattage with weight and inertia and so on. We are hoping to get all of this data into a new white paper which will have Cd values and wattages, inertias, etc, that can be used on the analyticcyling.com site as the data they have for us on the site currently is from 3 generations ago, so is not very representative.

The simple answer is that a disc becomes more important with increasing wind speed or wind angle. At low wind speeds and low wind angles, a disc is only very marginal over a deep section wheel. In fact at some very low wind angles like indoor on the track, we are seeing that a wheel like the 808 may actually be generally faster than a disc, but once you throw a decent wind into the mix, the advantage starts moving towards the side of the disc.

For our wheels specifically, a 909 (disc rear, 404 front) is worth about 24 seconds per 40k assuming an average wind of 8mph with rider speed of 30mph (the time savings will increase as rider speed decreases) compared to a pair of 404s. Now that is assuming our disc with a nicely matched tire, put a 23mm tire on the disc and you give most all of this savings back. Likewise, some disc designs that have an lip or hard edge where the outer hoop transitions to the disc body may give up as much as half of these gains, so there are a lot of factors that are really key to make the disc work ideally over deep sections. Now that same 909 wheelset is only about 12 seconds faster than a 606 wheelset (808 rear 404 front) and an 808 wheelset should be equal to if not 1-2 seconds faster than a 909 due to the faster front wheel and the rear wheels being near equal.

Now bump that 8mph wind up to 12 and the advantage moves about 4-6 seconds per 40k in favor of the disc, and this will also be true for decreasing rider speed which has the effect of increasing wind angle for a fixed wind speed.

Ultimately, if you are competing for a tight spot in a tight age group or whatever, the disc will earn you valuable seconds compared to deep section wheels (more with more wind) but you need to get the right tire on it and make sure you use it on windy days (which is where I often see people leaving the disc in the car).

Here’s a link to our engineering white paper on rim shape the testing Tour Magazine did on wheel aerodynamics:
Rim Shape: (PDF)

You’re spot on with your statement that there is a lot more to learn about in the area of aerodynamics – namly that there are always exceptions to the rule and really the only way to know is to test personally. Of the dozen athlete testing sessions I've been involved in over the past few years, there has been at least one person at every test that just seemed to blow all of the standard positioning thoughts out of the water.

Floyd is a great example, I was in the tunnel with him last February intent on getting rid of his crazy position, and had been at MIT a few months earlier with a couple of athletes and found the Landis position to be terrible with every one of them, then low and behold after about 4 hours the fastest position for Floyd by far... almost exactly what he had. We made a modified pad setup out of PVC pipe so that his arms were basically touching at the elbows and it got even faster.

The next day we tried all of the same things with Santiago Botero and he was markedly slower with his arms angled up every time. I think why this worked for Floyd is that he has pretty wide hips and very large hands combined with narrow shoulders, so I think that with his arms slightly wider or horizontal we were getting lots of flow across the front of the body. Botero had broad shoulder so when we widened him up about 2cm he want about 3-4% faster.

The key things that seem to not have exception, are removing all clothing wrinkles and folds, and keeping the aero helmet sealed to the back of the rider. I don't think I've ever seen somebody exhibit lower drag with a gap between helmet and back... but then I'm sure that probably works for somebody.

I was in the tunnel with him in March and all I can say is that I'm not exactly sure why, but it definitely does work. We tried the same position on Santiago Botero as well as Miguel Perdiguero and it was terrible for both of them, but with Floyd it works quite well. I think that Floyd has a different body shape than a guy like Zabriskie and he also has very large hands, but overall I think your theory on the airflow going around instead of under the body is probably pretty accurate.

Actually, the Mavic Cosmic Carbone is very similar in performance to a 303 and slightly slower than the 404. This is because like the HED Jet and Trek Aeolus it uses the patented Zipp rim shape, however, the Zipp patent discloses that shape in terms of a structural rim, and not a fairing rim. At the time we applied for the patent all the various federations were outlawing fairings of all sorts, so we never even thought to include a fairing design in our patent on that rim design as it wasn't anything we ever intended to make, although had we known that fairings would still be here 10 years later we certainly would have included it.

Anyway, the shape is much faster than a V or U shaped rim (and the canted brake surface patent used in the 808 is slightly faster still) but what you see with the fairing rims is that the wheels are generally only as aero as a rim of the same depth where the spokes exit the fairing. So a Cosmic Carbone is slightly faster than the old 38mm 303 and slightly slower than the new 44mm 303 because the spokes enter the fairing at about 41-42mm, but the cross wind force is the same as if the rim were the full 50mm depth. That is the main reason we have never been enamored with the fairing – you get only about 80% of the aero benefit, but 100% of the crosswind penalty.

The Cosmic also makes up a little ground by only using 16 spokes, which makes for slightly better aero, but reduces spoke life and makes the wheel unridable in the event of a spoke failure, where we are using 18 or 20 spokes depending on rim depth, resulting in a wheel which can be ridden with a broken spoke. One of our design criteria is to cut a spoke out of a perfectly good wheel and it has to spin freely without contacting the fork legs.

For more information on how rim shape check out the Rim Shape engineering white paper on our web page:
Rim Shape: (PDF)

The new tires are made in conjunction with Vittoria and the quality thus far has been really superior to the previous tire (not made by Vittoria).

The history of this tire was that we tested more than 40 high end tires on 404's and 808's and found that some of them were actually faster than the old zipp tire, which had generally been the fastest tire on our wheels for the last few years, and we narrowed the reason for that down to the degree of tread roughness, as well as tread thickness profile.

There seems to be a sweet spot somewhere between smooth and file-tread that really works well, so we computer modeled an equivalent surface roughness into a dimple pattern (that's all dimples are, just a form or roughening a surface in an orderly manner…). But the real excitement came when the Vittoria guys started working their computer design magic and realized that the dimples add essentially no stress into the tread, unlike traditional tread designs which have lots of sharp interior corners which act as stress risers, so the predicted wear and crack resistance with age of the tire was dramatically decreased.

This lower propensity to crack allowed us to go with a slightly lower rolling resistance rubber compound without any penalty, while the tire still showed excellent adhesion both wet and dry in testing, so we really think we have a winner here. 1-3 Watts is for a 404 or 808, more like 1-2 watts on the 303 depending on wind angle.

Unlike the previous tire this one will also be available in clincher.

A little history on the development: the previous Zipp tire was designed in 2002 and went on sale in 2003 and was the fastest tire we'd ever seen on our disc or 404 at the time, and the 808 and revised 404 shape were subsequently designed and developed in the wind tunnel using that tire, but in the mean time about 30 new tires have hit the market (plus we had access this time to some very specific specialty tires which we did not have access to before).

While our initial work on the earlier tire design was really primarily based around smoothing the tread to casing edge, our work this time around focused more on tread patterns and roughness as it seemed that tires with a slight tread lip but with perfect roughness could match or even beat the previous zipp tire, so we had more time and more experience this time around. In fact, it seems that at the perfect roughness a little tread lip actually helps the situation by acting as a boundary layer trip strip so you will notice this tire having a roughly 0.5mm lip compared to the 0.3mm lip in the previous tire. Remember, the previous tire was the first tire ever designed in the wind tunnel, so everything was essentially new and unknown as nobody had done it before, this time around we know quite a bit more and have a more capable manufacturing partner so this is just the natural evolution of things.

Any clincher tire 20-23mm wide will yield optimal aerodynamics with your wheelset. As for the tight fitting tire issue, we recommend using the thinnest rim strip you can find (not Velox, it's too thick). Using Rox, Michelin or our own Zipp rim strips will give you up to 1mm extra clearance when installing the tire.

Enjoy your wheels!

From all the data I've seen in the last 6 or 7 years, I have never seen any reason TO run a 19mm tire. Many wheels such as the Hed3 do work better aerodynamically with a 19 as that matches the rim width, but even with that wheel the aero advantage with a 19 seems slight compared to the increased rolling resistance, increased puncture and pinch flat likelihood, and poor ride quality, at best I would think that the advantages/disadvantages are a wash on these wheels. We do all of our tubular design around 21mm tires and clinchers around 23mm tires, and as such our wheels are faster aerodynamically with these tires than with 19's, so there is absolutely no advantage to run tires this narrow on our wheels.

Also, you are looking at a likely 3-5% decrease in rolling resistance in using a 21 over 19mm tire, and possibly even more improvement as you can run lower tire pressures with a 21 for even lower RR without increasing the pinch flat likelihood on poor road surfaces. Lastly, I think that for all triathlon events comfort and ride quality should really be considered as you may be costing yourself on the bike and the run by running 150psi in very narrow tires through higher rolling resistance and increased muscle fatigue due to high frequency vibration.

The quick and dirty rule of thumb is to just to an inverse tangent calc that goes tan^-1(wind speed/rider speed). This gives you the max possible wind angle for these two velocities as you are assuming the wind is perpendicular to the rider with this calc. So if you are keeping a 24mph pace and there is a 12mph wind on that day, then your max possible crosswind will be 26.6 degrees. So every net effective wind angle will be between 0 and 26.6 degrees of yaw.

We generally assume a sort of bell curve shaped distribution of wind angles, so that the assumption is that most of the day's wind will be in the middle third of the total yaw, or for this example will be between about 8 and 18 degrees of yaw. Of course there are situations like out and back courses that can be greatly simplified by doing some vector math and you may find wind in those situations to concentrate more towards zero or more towards the 26... it just depends on the wind and course directions.

Since what you are actually asking is the effective lateral force on the wheel in a crosswind, you can look here: (PDF)

The bottom graph is actual side force on the wheel relative to wind angle for the two wheels being discussed as well as numerous others.

The thing that you won't get just figuring the side areas of the two wheels is that the curvature of the surfaces has as much or sometimes more of an effect on side load than the pure surface area, so that a wheel with high surface area, and also high curvature like the 808 can have less drag as some wind angles than a wheel with more surface area but less curvature like a 3 spoke.

You are pretty much spot on with the explanation of wind vector and it's real world effects. We use data from the National Meteorlogical Society which quantify an average wind velocity anywhere in the country at any time is roughly 8mph. To tell us that for an average triathlete, this will result in a wind angle of 0-20 deg (assuming 21-22mph average speed). The key here is that the faster the rider, the lower the wind angle and the more the course moves around, the lower the wind angle. To acheive high angle winds, you need the cross-wind to be at 90deg, if the road curves or turns by 10 degrees or so, the apparent wind angle begins to come down quite quickly, so to quantify something for high angle aerodynamics you need a pretty specific course and wind condition, such as a perfectly straight coastal road with steady 20mph cross-wind, but the reality is that generaly the road is going to curv and twist a bit, which will begin to bring down the wind angles actually experienced.

For a rule of thumb, we say that roughly 75%-80% of conditions in the real world will be between 0 and 20 degrees of apparent wind angle with something like 15-20% of those between 0-10, leaving the spread from 10-20 accounting for more than 50% of conditions. The other 20-25% accounts for conditions between 20 and 90 (with likelihood of occurence diminishing with increased angle such that 90 is theoretically impossible if you are actually moving forward). This is why we have really been working to optimize rim shapes between 10 and 20 degrees, since that is where you do most of your riding. Here's a paper on it:

Rim Shape: (PDF)

As for the thought that deep rims might have higher side force than 3 or 4 spoke composite rims, we’ve found the opposite is true. All of the data we have for our own 808 and the Hed Deep show that both have lower side force than any 3 or 4 spoke composite wheel. By our data, the Deep has about 15% less side force than a Hed 3 and the 808 has 35% less side force (slightly less deep, high sidewall curvature, and dimples allowing for lower leeward pressure drag allow for this), while both deep section wheels are more aero in the 0-20 range.

The other aspects that is important is wheel stability, with deep section wheels having more consistent handling with varying wind angle due to linear side force characteristics, whereas some composite spoked wheels have non-linear side force such that small changes in wind angle can have disproportionately large increases in side force. This effect and wheel torque are something that we have studied for a while and really lost a lot of sleep trying to optimize in the 808 rim shape. You can see an actual side force graph from the wind tunnel at the end of that paper, and see how much less force any of these wheels have than a disc, but relative to each other, there are some big differences as well.

We have not been able to test the new Hed Stinger 90 or 100mm Blackwell wheel yet as they are so recently available, but during our 808 development we narrowed some 40 rim shape/depth combinations down to 8 wind tunnel models varying from 80-98mm deep with varying sidewall curvatures, as well as having original an original HED DEEP Jet style wheel as well as a structural carbon version of the DEEP. We found that the older DEEP rims with flat sides performed very similar to the new at the time 4th generation 404 wheels from about 0-20 degrees, which is to say slightly better than H3, and slightly better than the 404 and slightly better than the H3 from 20-30 degrees.

Of the 808 models we found that the rim width and sidewall curvature made more of a difference than the rim depth in the critical 0-20 degrees of yaw (~80% of the real world conditions) such that the 80mm profile we ended up choosing for production was faster than the 98mm rim with less 'bulge' in the shape. The 98mm deep model with larger sidewall curvature was actually slightly less aero from 0-15 degrees of yaw, but about 8% less drag from 20-30 degrees of yaw. Our decision at the time was that we need to really tune the rim for 0-15 or 0-20 degrees of yaw, not only because these conditions are most common, but because at higher yaw angles (read, higher cross wind speeds) people would be less likely to actually use the wheel. So the 808 ended up being not only the fastest in the most common conditions but had the least side-load/steering torque of all of the models we worked from..but we definitely had models both 88 and 98mm deep that were faster from 22-30degrees of yaw. That being said, I bet the Blackwell is seriously fast in high yaw conditions say past 20 or 25 degrees.

We have the link on our website to the Tour Magazine wind tunnel test in Germany where they validated our data that the 808 was actually slightly faster than a disc wheel from 0-12.5 degrees of yaw (making it the fastest non-disc they've tested in 10 years of annual wind tunnel testing), and from all the data we've taken over the years nothing (neither prototype nor production) has been able to match this yet.

What we tell all of our athletes is to just remember that it is when the wind is blowing that these sorts of wheels are most advantageous. So a pair of 606's may be worth 20 watts over a pair of Ksyriums in a 5mph cross-wind, but they are worth like 45 watts in a 12mph crosswind. So by choosing to ride these wheels in conditions where your competition chooses more traditional wheels is where they are truly most advantageous.

Plus if you train on them a bit you will get used to it. I have been using 606's exclusively on my road bike for almost 2 years and don't even notice them even in strong winds.

For more information on how Rime shape affects crosswind handling, click here: Rim Shape: (PDF)

This is actually a repeat of a similar test they did last year. Funny enough I was just talking with Marcus from Triathlon DE just Monday of this week about the test. What they do is have the riders ride at a constant speed for so many laps and then take the average wattage from that run. They do all the wheels at the same speed and use the wattage difference at the same speed to determine the savings, so those wattage numbers they are quoting are watts saved at the same speed.

What we have been discussing with them is that this test really only holds water for indoor track usage as the airflow is completely stagnant in at the track they use, so from an engineering perspective we can consider all of the tests to be conducted at exactly zero degrees of yaw. Unfortunately, zero degrees of yaw will statistically occur less than 1% of the time in the real world, so to really make use of this data we have to either add similar testing done outdoors or wind tunnel data.

At zero degrees of yaw, not only are most wheels very similar, but many aero wheels with large surface area are at a slight disadvantage due to the predominance of skin friction drag over pressure drag, but with just a few degrees of yaw, the pressure drag becomes an order of magnitude more important than skin friction drag.

If you look at our white paper 'A Note on Rim Width' : you will see that all of the data is pretty compressed at zero degrees and that a disc actually gives up a little to both the 404 and 808, this is due to the predominance of skin friction at this angle (note that the zipp disc has less drag here than any other discs with the same shape...thank you dimples!). With just a little wind angle, though, the disc, or 808 suddenly becomes vastly superior to less deep rims. This wind angle phenomenon is part of the cleverness of the Xentis wheel, which has the spokes angled relative to the wind centerline so that at zero degress the spokes are at 10 degrees or so to the wind.

This works excellent at zero degrees, but you can also have too much wind angle and at the higher angles the drag begins to increase again as the Tour magazine test determined: you can see in their tour article Download Translated PDF wind tunnel graph that the Xentis is slightly faster than either the 808 or the disc at zero, but by 4 degrees the disc and 808 have a significant advantage. Using the numbers at zero from the Tour magazine test we extrapolate a 1.2 watt savings at zero over the 808 and 2 watts over the 999 (the data in the Zipp white paper mimics this identically), which is almost exactly what Triathlon DE found in this test.

As with any testing, the problem is determining the limitations of the test itself. In my opinion the Triathlon test is dangerous as they don't really adequately explain the very limited nature of the usefullness of the data. But when we combine testing methodologies you can start to really have something that is worthwhile. I think that this test is very interesting in that it more or less validates one of the data points seen in wind tunnel data, but it is dangerous in that wheel decisions made from this type of test will be very limiting in the real world. This also shows the give and take necessary in the design of these products. We spend about 50 hours per year in the wind tunnel trying to tweak and refine these shapes, but the reality is that you almost always have to give up something to get something.

Lately we have been giving up very low and very high yaw angle performance to improve mid range performance, and I am convinced that this methodology is really the future of design since real world wind angles are generally nothing close to zero degrees. I'm currently writing another white paper on this describing the vector resolution and wind angle distribution in percentages that we are now using for design.

For more information on Rim Shape and the aerodynamic “sweet spot” see our engineering white paper on rim shape:
Rim Shape: (PDF)

A frame with a well designed cutout definitely allows the rear 808 to perform at its best, but from all the testing we have done and seen at high yaw angles the wheel/frame relationship is really only affecting the airflow over the front half or so of the wheel for the 808, but not changing much on the rear half of the wheel compared to a frame of similar design with no cutout, whereas the frame with the cutout can affect the airflow over the entire surface of a disc.

Overall we have seen that the drag differential between the 808 and disc when tested alone matches quite closely to the relative differences when tested in a bike without a cutout, and that the disc gains a slight edge in the bike with the cutout by a few watts over what was predicted by the drag differential of the wheels on their own (and this advantage seems to be limited to the higher yaw angles), and of course different frames with different cutouts all show different levels of improvement. I don't have any of our data in front of me at the moment for complete bike testing, but from all my experience I would say that the cutout generally favors the disc at higher yaw angles.

After years in this business and more hours in the wind tunnel than I care to remember I have to say that the 'you don't ride fast enough for (insert aero product here)' is really that person's way of saying 'I don't want you to have (insert product name)' for reasons none of us will understand. The aero benefits of most any aero product are reduced somewhat at lower speeds, but are still proportional to total drag, which means that you will go faster. Generally the time savings for a given aero product will be greater if you ride slower, as you are out there longer, but the notion that riding at lower speeds negates aerodynamics is pure siliness. There is some interesting stuff on this at the Cervelo website showing estimated time savings for frames, wheels, helmets, etc. at varying speed averages which you might find interesting, and we have reams of aero info under the tech section of our website as well which may be helpful:

http://www.zipp.com/technologies/aerodynamics/aerodynamics.php

Much of the rumors about the minimum speed for aero stuff to work center around the fact that we wind tunnel test at 30mph, so everybody thinks that this is a speed regime we are optimising around, but in actuality this is about the slowest you can run most wind tunnels and get very accurate data. Even 'low speed wind tunnels' like Texas A&M or MIT are really designed to run between 100 and 400 mph, so trying to run at 15 or 20, and even 30 is difficult. In single run testing, we have seen drag reductions at 15mph, and computer modeling of flow predicts benefits down below even 12mph, so you can rest assured that 18mph is more than fast enough to reap some benefit from this stuff.

Our patent on the dimpled rims actually discusses this exact issue at length as we learned that just like the golf ball, the rims will begin to show optimal flow characteristics at lower speeds with dimples when compared to smooth ones. A golf ball has the same drag profile at 150mph that a smooth ball would have at 300mph, but the surface treatment energizes the air and forces a turbulent boundary layer allowing the airflow to follow the contour of the ball making it appear smaller than it actually is. In the golf ball analogy, the smooth ball cannot fly more than about 150 yards no matter how hard you hit it because the drag is so high that the ball cannot be accelerated to a speed at which it will attain the lower drag flow, however the dimpled ball transitions to the optimal flow regime very quickly allowing the ball to continue accelerating before continuing it's flight at the low drag state. The same principle applies to the wheels, and response so far to the new products has been overwhelmingly positive.

I hope this helps, and I don't want to sound too much like a sales guy here. Any aero rim, frame, helmet, whatever is going to work at your speed so don't worry. And as for the mid-v rim issue, these rims/wheels are less aero than their deeper counterparts at high speeds and are similarly less aero at the lower speeds, so if you are going to drop the cash you are best served to go as deep as you are comfortable riding, and you will see the most benefit. Shallower rims only really serve to reduce cross wind effect in very blustery conditions or to reduce weight in conditions where that is critical, but are not more aerodynamically efficient at any speed or angle. I hope that helps answer your question.