Felt IA Wind Tunnel – Data Analysis

Last year, we were given a little flash drive with some very interesting wind tunnel data about the then-new Felt AR and its competition, which we analyzed. While we had to wait a little while, this year we finally got a look at how the IA performs in the tunnel, and we found the results a little surprising.


Test Methodology

Just like last year, Felt traveled out to test the IA in the San Diego Low Speed Wind Tunnel with two competitor’s bikes in tow this time, Cervélo’s P5 and Specialized’s Shiv Tri, as well as the DA with a Bayonet 3 fork, bringing the total to four bikes and five configurations (more on that in a minute). We asked about the depth of field in this test, and were told that it was simply a timing and resource issue for acquiring other bikes in time for this trip. According to Felt, they “selected a small sample of what we and the consumers believe to be the leading triathlon bikes in the industry.” Well, we can’t argue with that – the P5 and Shiv are both staples of transition rack envy for the past few years; even we’re guilty of it from time to time.

Test Setup

Felt notes that “All bikes tested were… setup for the same hypothetical rider with identical stack, reach, saddle height, etc.” but that there are some limitations in terms of adjustability between models, so the setups are “as close as possible.” We are trying to get the fit coordinates used, for rigor’s sake, and when we obtain those numbers, we will update accordingly. The component setup for each bike was also identical, and is detailed below. Yaw sweep is the common protocol, starting at -20 and stopping at +20, passing through 0 in 2.5 degree increments.

Frame Size 54
Wheelset Zipp 404 FC front Clincher, Zipp 900 Disc Clincher
Tires Continental GP4000s, 700x23c
Component Group Shimano Dura-Ace Di2 9000
Brakes Model specific
Seatpost Model specific
Saddle Fizik Ares
Stem Model specific
Handlebars Model specific


The Chart

Tunnel Data

Here’s the moment we’ve all been waiting for; the answer to whether Felt has dethroned Cervélo in the aero category is… the same you’d get from a Magic 8-Ball: “Answer Unclear.”


What is clear is that Felt has clearly taken on the Shiv Tri and soundly beat it at its own UCI-Illegal game. To our eyes, the closest the Shiv comes to the Felt is about 100g of drag, and that is out at +20, where you’re truly unlikely to ever ride for any significant length of time. Setting aside questions of nutrition and spare parts integration, if what you’re after is speed that’s unavailable to the Pro Tour teams (in any configuration), the IA is your best bet. The rule of thumb is that 10g of drag is worth about 1 watt, or a tenth of a second per kilometer. When you’re saving 100g, you’re ticking off down the road at 1 second per kilometer faster than your rivals, and that adds up very quickly.image004

The Aero Crown

Let’s talk about the P5 for a moment, because it is deserving of some serious praise, here. When testing against the competition, in a test conducted by the competition, it comes out ahead from -6 degrees to +7.5 against the absolute best the industry has to offer. We don’t care who you are, that’s something to sit up and take notice about, but only in terms of philosophy. Yes, we said philosophy.

The P5-Six is a bike to go balls-to-the-wall, pain cave performance, leave nothing on the table type of bike when it comes to the wind. They have built a bike for the podium hunters, squeezing every ounce of low-yaw efficiency they can out of a bike designed around the UCI’s rules. By optimizing for narrow yaw, Cervelo has said, effectively, “Slow riders need not apply.

What do we mean by that? We don’t mean anything by it, but the math certainly does. It says that your maximum yaw angle is equivalent to the arctangent of the wind you’re experiencing divided by your bike speed. If you’re riding at 22 mph and the wind is blowing at 5 mph (see here for calculating wind speed), your maximum yaw angle is 12.8 degrees, putting your average somewhere around 6.4, which is right at the transition point between the P5 and the IA. Add another two miles an hour to your bike? Your maximum yaw is 11.7, average yaw 5.85, and you’re into the P5’s territory for over half your ride. At 30mph, the time you spend at yaw angles that the IA is faster in is about 10% of your total ride. You see what we mean about speed?

Where the IA stakes its claim is evident from the “Time Analysis” table and accompanying note saying that Felt “used a weighted average with an emphasis placed on more common angles of attack such as 10, 12.5 and 15 degrees.” Here’s the thing, though; we’re not sure they needed to, because the drag differential between the IA and the P5-Six from -6 to +7.5 isn’t that great. We’re not saying it isn’t present, it is. And we aren’t saying that the P5 isn’t faster at those degrees of yaw, the chart clearly indicates that it is. What we are saying is that for mere mortals, from this chart, the IA is the bike to buy, because it offers a significant performance advantage out at angles that you’re likely to be riding at. For the record, we’re somewhere between that 22 and 24 number, which means we spend more than half our rides at yaw angles where the IA is the clear victor. If you’ve got a bike split like Wiggo, go for the P5. If not… the IA takes the throne as the king of the superbikes.

In talking with an engineer from Felt, he told us that Felt made a “made a conscious choice to target those higher yaw angles and we knew that we will give up a little at low yaws.” What the IA does is generate lift on the back half of the tube shapes at yaw, which counteracts the longer chord length’s increased drag by having the wind effectively “push” against the drag being experienced by the front half of the airfoil. This is the same principle behind high-drag, high-lift wheel shapes like Zipp’s Firecrest; you use the wind to help propel you forward and neutralize drag. It’s worked for Zipp, and now we’re seeing it in bike frames like the IA to great effect, as well. The IA, in contrast to the P5, takes a hit at low yaw precisely because of the additional drag being generated, which isn’t yet offset by lift. The difference, like we said, is one of philosophy of frame design.

Time Savings

Table 1.  Time savings at 230 Watts over 112 miles and Power savings at @30mph.

Time saved (s) Time saved (m) Delta Power
Felt IA 281.2 4.7 14.2
Cervélo P5 186.2 3.1 9.4
Specialized Shiv 22.8 0.4 1.1

As with the Felt AR’s time savings notations, we are presenting the table without comment. The figures are a result of Felt’s weighted averages as noted above.

Final Thoughts

We feel we should be clear about this: the IA and the P5 are both ridiculously fast machines, and neither are likely to “hold you back” in any sense. The IA was Rinny’s ticket to a world championship last year, and the P5-6 was under Frederik Van Lierde for the same bike course, so let’s not create any controversy where there is none – these are Kona-winning bikes and either one is much, much faster than we will ever hope to be. For those of us with day jobs, the IA is just that extra bit faster than the P5, and is available at price points (if you don’t mind mechanical groupsets) that the P5 isn’t. To crib from our last analysis, long live the Felt IA.


14 responses to “Felt IA Wind Tunnel – Data Analysis

  1. If Cervelo said slow riders need not apply, then you should say Felt said the fastest riders need not apply. That way there is no controversy in your philosophy.

    • We don’t believe the converse is true, which is what you’re asserting. The IA is optimized for higher yaw angles, that’s true, which you could still experience while being in the “fastest riders” group, but you’re more likely to approach the transition point between the two bikes instead of being squarely into the IA’s camp… for at least half of your ride, anyway.
      Take Kona this year (http://triathlete-europe.competitor.com/2014/10/11/measuring-the-kona-winds) as an example, or last year’s data from Mavic (http://www.engineerstalk.mavic.com/yaw-angle-measurement-in-real-conditions-on-kona-ironman-course/) – depending on the course, you can start to “need” a high-yaw optimized bike, even as a top-tier rider (Kienle’s average was just shy of 26mph) he will still see a significant portion of his ride out beyond the 7 degrees where the P5 reigns supreme. Remember, the arctangent of the wind speed divided by the pace you’re pushing should be the maximum yaw angle you will experience. The P5 is for timetrialists and no-wind courses; 30+ mph riders pushing at above threshold on calm courses. For those triathletes who ride as merely exceptional mortals or have course profiles that will see the wind attacking at high angles, the IA is likely the better choice.

      They are two different bikes for different riders, but saying that one is for “the slow kids” is disingenuous at best. One is simply better suited to super-high-speed bike legs and the other is for where conditions are less than ideal.

  2. Really appreciate that your analysis includes comments for real world riders. I am not a TT rider but a serious recreational rider. While I crave the very best it is good to know when the latest and greatest will just not benefit my capabilities. ie on a road bike aero advamtage is so speed dependent that I’m probably better off with lower weight options.

  3. Nice test, thanks. Would be really interesting to see Shiv TT in the comparision (which according to german Tour Mag test is much, much faster than the Tri-version).

  4. your conclusions are OK but your analysis and mental gymnastics you use to leap to them are totally whack. Please leave the praising of the awesome Felt IA and awesome Cervelo to people who actually know what the heck they are talking about. Please don’t let this comment offend you…. I optimized it for high yaw.

  5. Personally I think both now are outdated for the long course triathletes.
    The Scott Plasma 5 is just the tip of the ice berg in regards to where we should be heading….
    Drink bottles should be one with the frame otherwise all benefits are lost
    To me fuel and hydration and in being seem less are critical for aerodynamics for triathletes. The attached front drink bottle like on the Plasma 5 to me seems great… The question is now we have 600mls integrated how can we add another to keep us going for longer

  6. Is there any comparative data on how these bikes with large side areas are affected with crosswinds in relation to side loads or handling?

      • Hi John – we do not think anyone has put the Slice in the tunnel with either the IA or P5 so we cannot give you a truly accurate assessment how they will do comparatively in a cross wind.

  7. Pingback: 11-2-2014 WiR | AeroGeeks·

  8. Was the testing conducted with riders aboard? That would affect crosswind contact immensely. Cervelo claims that the P5, while actually a bit less aero than the P4 in some ways, is more aero than the P4 with a rider on board.

Leave a Reply