The Data Behind Zipp’s 858 NSW and 808 Firecrest
I promised a fulsome look into the testing Zipp performed on the new deep wheels it just launched. These are the 858 NSW (above) and the 808 Firecrest. Both these wheels were tested against the versions for sale immediately prior to the launch of these new wheels. The rolling resistance testing was performed on a Rolling Road; the effect of winds on the wheels was performed in a wind tunnel.
The prior versions of these wheel were tested with 25mm tires and the new wheels with both 28mm and 30mm widths of the same brand and model of tire. The reasoning is this: the prior wheels were optimized for 25mm tires; the new wheels for wider tires. The tests were performed with the tire widths most appropriate for each wheel. Yes, you can put a 25mm tire on the new 858NSW – as long as it’s a hookless compatible tire – but that’s not really the tire Zipp has in mind for you for this wheel.
The results are going to surprise and perhaps confound you. The results did not surprise me in the aggregate, but there were some eyebrow raisers in the particular, and I’ll give you a granular look at what in that data provokes questions.
When there is no explanation that fully satisfies my questions the answer is, “That’s science.” That’s the data. What I’m providing are the results.
The Rolling Road is like a huge treadmill on which you can ride a bike. In size and bulk it’s in between a running treadmill and a SkiMachine. The protocol is this: several runs (4, I believe), by a specific rider (same rider for all the testing you’ll see here), speed is constant, power is measured to keep the bike on the Rolling Road at that speed. There were 4 pressures for each wheel pumped into a given tire. For example, an 858 NSW with a 28mm tire was tested at 48psi, 58psi, 68psi and 73psi (or thereabouts). As I understand it, each one of those data points is generated from 4 runs. So, by averaging 4 runs using that tire on that wheel at that pressure you get what you see on this chart. A “run” takes 2 minutes: a 30sec spin up, 90sec of riding, then stop.
Wind tunnel testing was performed to try to discover 2 things: A difference in speed between the old and new wheels; and any difference in handling characteristics. As to that latter point, load cells were placed on the axles, on each side, to measure the torque on the wheel in a yaw at speed.
Finally, a word about field testing. A wheel is lighter or heavier (measured on a scale); it’s slipperier in the wind or it’s not (measured in a wind tunnel); there’s friction through contact between the tire and the road (measured on the Rolling Road); and then there’s all of at one time, which is what you measure during a field trial. Zipp has for many years conducted its field trials at Eagle Creek Park in Indianapolis. Zipp is circumspect about this testing, and all I can tell you is that the aero sensor tech that we’ve discussed on Slowtwitch from time to time is the basis for how they perform their field testing. They do not – as I understand it – feel they’re ready to publish studies using field trials as the operative technology. But they comfortable using field trials to validate what they see on the Rolling Road, or in the Wind tunnel. Therefore, all the numbers in these charts are either generated from the tunnel or the Rolling Road; none of the charts contain data resulting from field testing.
Rolling Resistance
Let’s begin with what is to me the most surprising and certainly the most controversial of the results: rolling resistance. In these tests, one make and model of tire was used for all the results in any graph you see here, but in various widths. The tires were either Goodyear or Continental, obviously hookless compatible because that’s the tech used in these wheels.
Here’s what does not surprise me about the chart above, which measures the rolling resistance of the 858 NSW: The data shows that the new wheel rolls better than the old, with the wider tire, and in the lower pressures. What you see here are 4 dots on each slope, and Zipp’s recommended tire pressure is the second from the left.
Here’s what does surprise me: The lower the pressure, the faster the tire performs. What we should see is a U-shape in those slopes. We should get to a pressure that’s just too low for good rolling performance. In no case did Zipp necessarily find the best pressure; it just found that the lower the pressure, the faster the system.
In fact, it found this to be the case with its prior wheel. The 25mm tire on that wheel performed better at 65psi than at 75psi or 85psi. This will not sit well with those who live and die by drum testing results, which show exactly the opposite. How does one square this circle? Maybe Zipp’s data is wrong. But we also know that drum testing has a flaw, which is, it cannot tell you when its reliable data becomes unreliable, that is, at which pressure system vibration overwhelms the value of high pressure. On paper at least, rider-on-bike (i.e., the Rolling Road) should solve that, in two ways: It shouldn’t suffer from the inability to determine break point; and it should pick up – to some degree – any (biomechanical) power losses resulting from muscle vibration.
The 808 Firecrest chart has its own strange phenomenon. Both the 25mm tire on the old wheel and the 28mm tire on the new seem impervious to pressure changes. While the 858 NSW is very sensitive to tire pressures, not so this wheel, until you consider the new wheel with a 28mm tire mounted. But just as curious: If you look at the 858 NSW, the fastest rolling pressures were lower than the current Zipp recommended pressure. But the fastest pressure on the 808 Firecrest, with the 28mm tire on, was at a higher pressure than the Zipp recommended. If Zipp is going to hang its hat on these results it’s going to need to rejigger its tire pressure recommendations.
What governs these numbers? Why should rolling resistance change at all between the old wheels and the new? Inner rim width; rim laminate; rim depth; they’ll all change the way the tire rolls. What I hear from wheel makers is that in their own testing 28mm and even 30mm is a fast rolling tire. But 25mm is a hard habit to quit, and there’s still divergence inside the wheel community as to which width is faster on typical roads that racers face. The smoother the road the narrow the tire, and Zipp’s view is that the most typical roads encountered favor a wider tire than 25mm.
Aero Performance
The charts presented above and below reflect data Zipp got from taking these wheels to the wind tunnel. What’s measured here is the delta – the difference – in test performance between the existing version of the 858 NSW and the new one. The red line is the baseline performance of the existing wheel. The blue line is how well the new version tests against it.
What you see “gained” is a reduction in the grams of drag. Realize that the units here changed from the units in the rolling resistance tests. In those tests what was measured was watts, and this wasn’t an extrapolation. It was a direct measure of how many watts it took Zipp’s rider to maintain his position on the Rolling Road using various set ups (different wheels, tire widths, pressures). In these aero tests what you see is also a direct measure. What the scale in the wind tunnel measures is what your bathroom scale measures: weight. Your bathroom scale measures how much weight presses down on the scale, and the force pressing against the scale is gravity. Wind presses against the bike in the wind tunnel, there's a scale in the system, and the unit of measure is grams.
The new 858 NSW is faster than the old but the lines come together in performance at 5° of yaw, remain the same at 10°, and the new wheel again gets a little better at a broader yaw. What you see is that the new wheel is better head-on than the old 858 NSW, which is kind of surprising when you consider that the wheel is wider.
However, as I read the notes on the test the one notable item is that 25mm tires were used throughout in the wind tunnel tests. (Not in rolling resistance, but wind tunnel.) I don’t have any wind data using a 28mm tire on either of these new wheels. While it would have made some sense to test this wheel with a 28mm tire, Zipp chose the 25mm size – in fact they used Zipp Tangente tires for their aero testing – because of all the historical data they have in the tunnel that used this tire.
The 808 NSW shows a slightly better performance until something between 5° and 7.5° of yaw, and then the old wheel is marginally better than the new. Again, both the old and the new wheels used a 25mm for testing.
Handling in a Wind
These graphs below display data I’ve been asking for for decades. Since the 1990s. What’s measured here is the torque applied to the fixture holding the wheel. The degree the wheel wants to turn when the wind hits it. If this were to get routinely measured by both bike and wheel companies it would be easier to make bikes that handled better in crosswinds.
What you see in the graph above with the new 858 NSW is that the wind in a more significant yaw (anything above 2.5°) produced less torque on the fixture, meaning, the wheel would be marginally easier to steer. But it’s 10g, so, we’re not talking a lot. In my opinion, wheel companies can only do so much here. It’s really up to frame companies to design frame elements (e.g., forks) that have more surface area behind the rake line. Would that work? I don’t know. Intuitively yes. I wrote about steering torque a half-dozen years ago. I thought the most interest stab at this was the fork on the Ceepo Shadow R.
The sawtooth profile on the 858 NSW is designed with steering torque in mind, as far as I can tell. The idea is that you ride a deeper wheel with the same pucker factor as a shallower wheel. Pucker you might in gusting sidewinds, but remain upright you will.
This is another surprise to me. Of course the sawtooth profile already existed, so I don’t know how much Zipp expected to gain on the steering torque test. But the 808 is the wheel that saw the biggest performance gain in steering, at least theoretically, over the prior version. As you see, the more the yaw increases the better the new 808 Firecrest is over the old.
But there is an answer that makes sense. The 858 NSW is actually a deeper wheel than the older version, by 3mm. The new 808 Firecrest went in the other direction. It’s 2mm shallower than the old. To me, that explains how the new 808 Firecrest is able to outperform (in the test, at least) the version is replaces, as regards steering torque.
What you see above is the data. Zipp chose just to release the data, or a compilation of it, even though the rolling resistance results are a refutation of the mode of testing upon which we’ve all relied for decades. There has been good agreement between those who’ve engaged in drum testing; but that argues for precision, not necessarily accuracy. Spitballing for a moment, the only rigs I can think of that are in the same genus as the Rolling Road (rider aboard smart rigs) are the Oreka Trainer and, perhaps, Wahoo’s ROLLR. If you wrapped the rear drum on a ROLLR with a textured surface, it might be interesting pursuing testing on that device.
I don't expect Zipp to sweat this too much. It stresses the phrase Total System Efficiency. Zipp doesn't care, that much, how aero the wheel is, in a vacuum. Or how light it is, or how well it steers, or how fast it rolls. It cares about how long it takes you to get from point A to B. This (intuitively) argues for field trials and if I had to guess I'd predict the next time these wheels are updated you'll be seeing field trial results.
Start the discussion at slowtwitch.northend.network