Shapely Curves, Girdle Free

The convention of mapping a wheel's aerodynamic drag against a range of Angles of Attack (AOA) produces a graph with curves that let you quickly compare the aero performance of multiple wheelsets. Less drag is better, so the curve that sits on the bottom is usually considered the aerodynamically fastest wheel. Usually however, it's not quite that simple. Different wheels produce curves of different shapes. Some wheels' curves dip down lowest at AOAs of 15 or 20 degrees, some hit their troughs at narrower AOAs. But the frequency of AOAs varies at different speeds (be thankful I'm the one writing this blog and not Dave - you'll have another 3000 words ahead of you when he takes it on). Narrow AOAs (0 - 5 degrees) are more common at higher speeds, which is also where aerodynamic impact is amplified. Wider AOAs (15 degrees and above) are more common at lower speeds, where the benefits of aerodynamics are mitigated.

What this means is that the wheel with the curve that dips down the lowest is not necessarily the fastest in the real world. When we calculate relative times for a 40K TT at 30mph, we use AOA frequencies that are appropriate to that speed. If we were to calculate a 40K TT at 20 or 25mph we would need to plug in different AOA distributions for each. When trying to interpret AOA curves, it's important to realize that the faster you go, the more important the left side of the charts become.

Here are the AOA sweet spots at different speeds:

20mph: 10 degrees and up (64% of AOAs occur here)
25mph: 5 - 15 degrees (66%)
30mph: 0 - 7.5 degrees (69%)

If you're looking for a wheel for holding a steady speed in a TT or triathlon, you'll want to pay more attention to the curve shapes in the middle and right side of the charts depending on how fast you ride. If you compete in road races or crits where the periods at higher speeds have a greater bearing on the results (sprints, attacks, breakaways, premes), a wheel's performance at narrower AOAs on the left side of the curve is more relevant to you.

When we sent the Rail down to the A2 wind tunnel in NC for testing, we sent a Zipp 404 Firecrest clincher along as a benchmark. Actually we sent the 404 as two benchmarks. Not only did we want to see how the Rail measured up against the Zipp under the exact same conditions (same tire, same protocol, same test session); we also wanted to see how our test of the Zipp compared to other public tests of the Zipp. What we found is that the conditions can have a profound impact on how a wheel tests in the tunnel, affecting both the shape and the placement against the Y-axis of the resulting AOA curves.

First, here is the shape of the 404 curve in the tests by Zipp, Bontrager and November. I've normalized them at the same point at a 0 degree AOA to allow you to see more accurately how the curves change at different angles.

Zipp claims that the 404 FC is optimized around a 15 degree AOA, which is borne out by the curve resulting from the tests in the conditions they selected, suggesting that the 404 performs better at slower speeds than it does at higher speeds. The curve produced by Bontrager's test of the Zipp 404 FC shows the trough shifted to the left with narrower AOAs. Based on Bontrager's test conditions, the 404 looks to be at its best in the middle AOAs that compromise the 25mph sweet spot. Finally, our own test of the Zipp moves the trough further still to the left, suggesting that the Zipp may actually be at its best at medium to high speeds.

Now of course what you are seeing here are not tests of different wheels, but tests of the same wheel under different conditions. There is no way to know all of the conditions present that may impact aerodynamic performance in each of these tests. Everything from humidity to temperature to spoke tension to valve stem length and decal placement can all have an impact. And even if all of that is the same, two different wheels of the same exact model and specs tested during the same session can also produce different results. So we cannot know all of what drives the curves above to look differently.

However, we do know the big changes present in each of these conditions, which fortunately for consumers is something they have choice over themselves - tire make and model, width and style. Bontrager selected the exact same 23mm Bontrager Race XXX Lite tubular tire for all of their tests, switching it from wheel to wheel. We did the same thing with a 23mm Vittoria Evo Corsa CX tire. Zipp says that the 404 is optimized for a 21mm tire and since they make an aerodynamics-forward 21mm tire of their own called the Tangente, it is a reasonable assumption the testing results they publish are with that tire. I'm sure they have disclosed their testing protocol somewhere but I can't find it. If anyone has seen, please share in the comments.

One assumption then from the curves above and what we know about the test conditions for each is that the tire make, model, width and style can alter the shape of the curves and therefore the wheel's performance at different AOAs (and by extension, different speeds). We don't know this for sure until we test a single wheel against a range of different tires, which was beyond the scope of this trip to the tunnel. So the impact of the tire on aerodynamic performance remains a hypothesis, though in our opinion a credible one.

Is it possible then that in addition to having an impact on the shape of the AOA curves, the choice of tire - particularly tire width - can affect the drag so that the AOA curves sit higher or lower on the Y-axis? Bontrager in their testing did find this to be true. They tested a set of their wheels against Zipps using the same 23mm tire and then the same 27mm tire and found that the wider tire created 20 to 100 more grams of drag depending on the AOA (have a look at their excellent white paper (PDF), page 29).

Here is the same data from the chart above showing the AOA curves of the Zipp 404 FC tests conducted by Zipp, Bontrager and November, only without the normalization at 0 degrees:

The Zipp 404 FC created more drag in our test conditions than it did in Zipp's. (Let's throw out Bontrager's for now since it's a tubular and not a clincher - you can't take your FC clinchers and run a Bontrager Race XXX Lite tubular tire on them to make them faster.) If Zipp did in fact test with a 21mm tire, it is somewhere between possible and likely that the disparity in drag of 30g - 100g you see in this chart is a result of the different tire width. Again, we do not know for certain because we did not test different tire widths ourselves at this session. The Rail's inside width is optimized for the 23mm tires the vast majority of riders and racers use. Testing it with a 21mm tire would be an exercise in curiosity but not practicality.

Next time I'll show how the Rail's AOA curves plot against the 404, and also against our current RFSC wheelsets. 

The Second Rail

Since we've announced the Rail, we've gotten a lot of questions about what other depths we'll be pursuing to complete the Rail lineup, so even though we don't yet have the prototypes for the next depth, we thought we'd share our plans.

At the outset of the Rail project, we were interested to quantify what exactly the overlap zones were between wheels.  We'd been selling the 38, 50, and 58, and it sometimes seemed like splitting hairs to choose between them.  The 85s didn't feel very much like splitting hairs.  Subjectively, Mike and I both thought the 38s felt noticeably faster then "regular" wheels, and that the 50s felt a little faster than the 38s, and maybe the 58s were a little faster than the 50s but the difference wasn't as big as it was between the 38s and 50s.  Though it turned out to be much more concept than action (I wound up using 38s for the dominant majority of all of my riding and racing as a long term test), I selected 50s as my race wheels based on these beliefs.  Surprise, surprise - our seat of the pants analysis was pretty accurate indeed.  50s are a bit quicker than 38s, and don't give much up to 58s. 

Being a 52mm wheel that has less aerodynamic drag than our previous 85, the 52mm Rail covers a lot of territory.  It's a better choice for any "pure speed, please" applications than any wheelset we've sold to date, yet its weight and depth make it suitable for any mass start race you're likely to do.  The Mt. Washington Hill Climb might call for a lighter choice, but for the kind of riding that most people do, the 52mm Rail is a great choice all season long.

If I keep referring to the new wheel as the "52mm Rail," there's clearly another depth on the board.  The 38s have earned a huge following.  They're stiff, light, snappy, accelerate like a shot, they're comfortable, they're far from slow, and you can ride them in a hurricane without issue.  And we think we can give them the Steve Austin treatment (better, stronger, faster) without the $6,000,000 fee.  How many people reading this actually ever saw an episode of "The Bionic Man"?  Good lord I'm getting to be like Methusala up in here. 

We are working on two prototype designs which we think will get us where we want to be with the Rail thirtysomething.  The tucks and tweaks to get the shape how we want it to be will dictate the ultimate depth, but we're aiming for the low 30s.  The goal is to have a wheelset that retains all of the positive attributes of the current 38s in a shallower, wider, and ultimately faster package. 

If we're able to accomplish that, we think we'll have the spectrum very well covered.  Tubular versions of both the 52 and the thirtysomething are planned as well.  The drawings for the thirtysomething are sort of kind of almost complete, and then we'll shuffle Sons of Canary Thunder (we're thinking 2 - the design of the shallower rim isn't as obvious as the deeper one was) off the the wind tunnel and go from there. 

Moneyball for Cyclists

Spoiler alert: the in-stock price of the Rail will be $1285.  The pre-order price is $1145.  The in-stock price is a 14% increase over our existing RFSCs.  The pre-order price (pre-order availability will be limited) is $40 more than the in-stock price of a set of RFSCs.  To make this comparison, I took the base price of the RFSC series, added the $40 ABG which will be standard on all Rails, and the $120 CX Ray upcharge since CX Rays will be standard on all Rails. 

So how do you bring an entirely new wheel, with post-season aerodynamics and all-star stiffness, ride quality and heat management, to life for such a small price jump?  If you're thinking to yourself "I think he's about to tell me," you're right.

Our design payroll is way under the salary cap.  Way.  Fortunately, I spent most of the first 30 years of my life working with and thinking about boats and sails.  What's that got to do with anything?  Well, the sails on a boat going 8 knots into an 18 knot breeze have a whole whole whole lot in common with wheels going 25 miles an hour into a 5 mile per hour breeze.  The centerboard and rudder on that boat, which have to move through the water effectively at anywhere between 4 and 25 knots, also have a lot in common with wheels.  To spend today's terrbile pun card, you could say that my background gave me a fluid intutitive understanding of how to achieve what we were after. 

It helped that our parameters are as well defined as they are.  By starting with the 18mm inside width, and knowing that that dictated a 25mm outside width, and knowing that we wanted to be just over 50mm deep, the design box was set.  It would be ridiculous to claim that the current state of the art in wheel design didn't have a huge influence, but that's just how it is with design - it's a continuum.  Although many wheel designs claim to have "thrown away" the NACA profile book, I thought it was an ideal (and free) starting point.  I've previously talked about why I started with the 0024 section, from which it was a fairly straightforward series of tucks and tweaks to get the points to come together in the way I wanted. 

In effect, the design was free.  Verifying its speed was not.  First, we had to commission a prototype, the handsome yellow devil you've seen us flog around so often lately.  Once we got that, we built it up and sent it to the tunnel.  Both of these were pretty hefty investments, which we viewed as absolute necessities.  A wind tunnel isn't a race course, but it's the best proxy that we've got, and the only real way available to prove the aerodynamic validity of a design.  The risk we bore was if my design sucked - first because although I was confident in it, you don't know until you know, and second we were dumb enough to tell everyone what we were doing before we did it.  Our failure would not be private.  And, oh yeah, if the design was a swing and a miss, the investment in the prototype and testing would be a waste. 

The results we got at A2 mark the significant departure that we took from what the other leading brands do.  Where they NEED to leave the tunnel with the fastest shape, we said "this is definitely a major league shape, to coax every last gram out drag out of it would be exponentially more expensive, we're going with this one."  Aerodynamic speed is a HUGE part of the game, but we knew that we had road feel in the bag thanks to our 18mm inner width, and critical other parts of our lineup were already set.  In the battle to keep costs from getting out of hand, knowing when to stop is key, so we released the mold for construction.  The mold is a huge capital expense, and like the prototype and testing costs, we plan to amortize it over a whole pantload of wheel sales. 

For the rest of the project, we've been more general manager than player.  Our rim supplier has proven themselves over I don't even know how many rims we've bought from them, and has been working hard on heat management.  We'll test and verify their effectiveness on that front, but we couldn't have done the work that they've done, so we're happy to leverage their expertise.  SwissStop has blown us away with the effectiveness of their new pads.  If the biggest carbon wheel companies are taking of the shelf SwissStop products to use as their own, why wouldn't we?  They're way more expensive, but the first time you pull the brakes you'll know what we've found.  They also have testing infrastructure and resources that we could no way replicate, so we're just doing the smart thing and renting their expertise to check our work.

Sapim spokes speak for themselves, and the wind tunnel proved to us that using CX Rays was the only logical choice for the Rail.  Done.  Hubs are a bit of a funny one.  We will continue to offer the option of Chris King, White Industries, and Powertap hubs, but our standard hubs are like that late round draft pick who starts putting up hall of fame stats.  Since we've been working with them, they've become darlings of the wheelbuilding crowd for their excellent geometry, top notch construction, and killer performance on the road.  The 11 speed version will continue where the 10 leaves off.  The build will be handled as it has been: here, by us.

Mike and I pretty much act as living test labs, ferreting out what doesn't work and chasing down promising leads.  That helps us stay locked down on the directions we want to go, but before we sell a single Rail, we'll test every parameter of the final version to prove the benefits of what we've put together. 

I wouldn't necessarily call it easy to put together a wheel with the Rail's verified performance on what amounts to a shoe string budget, but it can be done. 

Aerodynamic Drag of Lasers vs CX-Rays

One of the things we've learned is that offering choices invites questions. One we get all the time is the difference between Sapim Lasers and CX-Rays, which we offer in all our wheels. Or rather, the question really is whether CX-Rays are worth the extra money.

The answer we've always provided is that the spokes are the same weight but CX-Rays are purported to have some aerodynamic advantage. If you're looking for "every last watt of speed," they're the way to go. But we have never seen anything that quantifies the difference between the two in a wheelset. So we decided test Lasers against CX-Rays in the tunnel to give a more informed and specific response than "every last watt of speed."

It turns out, however, that we've been exactly right all along.

We sent two RFSC 38 (38mm) wheels to the wind tunnel, one built with 20 radial laced Sapim Lasers and the other with 20 radial laced Sapim CX-Rays. Here is how the wheels tested against a range of Angles of Attack (AOA):

At all AOAs, the wheel with the CX-Rays was a smidge faster, generating about 11 fewer grams of drag on average at 30mph. If you recall the calculations from yesterday's blog, you'll see that 11 grams of drag at 30mph is - yep - 1 aero watt. You really do save "every last watt of speed" with CX-Rays, and not a watt more.

You remember also from yesterday that aerodynamic impact is diminished at lower speeds. Here is the difference in aero watts between the Laser and CX-Ray wheels at 30mph, 25mph and 20mph. In these calculations, the average drag is calibarated by the frequency of different AOAs at different speeds, which is why at 30mph the difference between the two wheels is 1.8 watts instead of 1.

Most brands assume that if you're spending between $1K and $3K for a carbon wheelset, you're after that every last watt of speed and they make CX-Rays or other bladed spokes standard. The logic starts to break down with shallower alloys though, where the upgrade to CX-Ray spokes may net you a watt, but still leave you a handful or two behind your training buddy on deep carbon, or oblivious if you're training on your own. For alloys in particular, we think it makes a lot of sense to offer the choice so people are not paying for performance they don't need.

The Rail Prototype Wind Tunnel Data and Calculations

Our primary objective in sending 9 wheels to the wind tunnel was to see where the prototype of the Rail stacked up against our existing wheel offerings as well as the Zipp 404 Firecrest clincher. The challenge when you are sifting through the pretty substantial amount of data generated at the tunnel is how to present it in a way that it will be both meaningful and accurate. So we've developed a couple of charts which give you the executive summary, which we'll fill in with some context. 

The raw data is provided in grams of drag and aero watts, neither of which tell the complete story about how a wheel really performs. More commonly, you'll see wheels in the 40K TT standard, which shows how much time each theoretically saves over a benchmark wheel in a 40K TT at 30mph. That's what we show here:

Usually the benchmark used is a 32 spoke Mavic Open Pro, which is a ridiculous proxy for the "average" wheel on a TT bike but which makes it easy for a brand to claim they saved a big enough number of seconds to raise eyebrows. We compared everything to our FSW 23 wheelset, which is 21mm high and 23mm wide with 20 raidally-laced Sapim CX-Ray spokes. If you're concerned enough about aerodynamic performance to look at wind tunnel data, the FSW is much more likely to approximate the wheel you are currently using, or the one you would buy if you decide not to go with a deep carbon wheel. 

You will see from the data that under these test criteria, the Rail is faster than any of our current wheels, including the RFSC 85s which are 33mm deeper. It is also only 2 seconds slower than the Zipp 404 FC, despite having 8 more spokes in the prototype and being 6mm shallower. Our internal goal for deciding whether to proceed with the Rail was that it test faster than our current RFSC 58s and within a couple aero watts (about 6 seconds in this test) of the RFSC 85s. When comparing to the 404, we are not limiting the Rail to aerodynamics, as the improved road feel of the larger internal diameter is really the heart of the Rail's story. We also wanted to bring the depth down to 52mm to broaden its use range. So if we were within about 3-4 watts (9-12 seconds) of the Zipp at this stage we would have been satisfied. Coming in at faster than our 85s and within 2 seconds of the Zipp made our decision easy; we've already greenlighted the Rail for production.

The 40K TT at 30mph comparison is a standard, but an unfortunate one for consumers trying to fairly evaluate how different wheels would perform for them. Aerodynamics are amplified at higher speeds (as Dave points out here, particularly int he comments), but the average speed for most TTs and bike legs in triathlons is well under that. Wheel brands like the standard though because it shows a larger gap against benchmark wheels. To get a better sense of how each wheel performs at mortal speeds, have a look at the following chart:

The slower you are going, the less delta you see in the aero drag between each wheelset. If you want to do the math and figure out the 40K TT time savings at different speeds, use these calculations:

30mph: 11 grams of drag = 1 aero watt
25mph: 9 grams of drag = 1 aero watt
20mph: 7.5 grams of drag = 1 aero watt 

If 1 aero watt equals a difference 3 seconds in a 40K TT, the 39 second advantage of the 404s over the FSW alloys at 30mph drops down to 25 seconds at 25mph and 13 seconds at 20mph. 

Some people will ask about Angles of Attack (or you might ask about yaw in which case you mean Angles of Attack, as per Dave), knowing that they vary at different speeds. The faster you riding, the more likely you are to encounter lower angles of attack. At 30mph, over 80% of AOAs are at 10 degrees or below. At 20mph the sweet spot is between 10-20 degrees, comprising about 50% of all real-world conditions. The data above already reflects the distribution of AOAs at the different speeds.

We'll roll out more of our test data and findings over the next few days. You can sign up to receive future posts by email here if you don't want to keep hitting the refresh button. Until then, fire away with any questions in the comments.