Thursday, August 14, 2014

The Null Hypothesis

Recently a new aerodynamics cycling product came to my attention, and the company behind it makes some extraordinary claims about the speed improvement attainable. As Marcello Truzzi would say, extraordinary claims require extraordinary evidence. So let's have a look at one such claim.

The product is the Nullwinds Upper Wheel Fairings. Here's their weblink:

The idea is pretty simple, add some fairings that cover the upper part of the wheels and bingo, instant aerodynamic improvement. Well that's not so remarkable, it's pretty common to attain an aerodynamic improvement through use of fairings. It's also why such things are banned in competition for cyclists and triathletes, but that's not the issue here as Nullwinds is targeting this to non-competitive riders looking for a speed advantage.

OK, that's fine, we all could use a boost.

Just so it's clear, an aerodynamics improvement means the drag coefficient of the bike and rider is reduced so that you require less power to sustain the same speed, or for the same power you can ride faster. Nice.

So let's examine one such scenario as listed on their website as being a Strong Headwind Test:

NOVICE RIDER February 9, 2014 
1. Test Summary 
The best available data taken on February 9, 2014, indicates that the use of our Upper Wheel Fairings on a typical road bike with a novice rider under strong headwind conditions yields gains in average speed exceeding 20 percent (22.2 percent was recorded). (The full report is available for download.) Power measuring tests in severe headwinds were conducted on identical multi-speed road bikes configured with and without wheel fairings. A novice cyclist was the rider. Data was recorded using an i-bike Newton power meter. 

2. Implications The results confirm that the use of our Upper Wheel Fairings can dramatically increase headwind penetration speeds of a novice rider under strong headwind conditions. Gains exceeding 20 percent are possible.

So, Nullwinds claim a novice rider riding into a strong headwind will be able to achieve a speed gain of more than 20% by putting these fairings onto their bike.

Well to Nullwinds credit they have at least published some information in an attempt to back up their claims. They:
  • did some testing to attempt to demonstrate the effectiveness of their product (tick)
  • attempted to establish some testing controls (tick, but they were not so successful as we'll see later)
  • published data for some of those tests (tick - more detail in pdf here)
  • claimed some impressive results (hmmm, no tick)
Unfortunately, Nullwinds missed an important step before the final one, which was to examine their own data before making their extraordinary claim. So let's do that step for them.

The details of their testing protocols and measurements are outlined in the document and I won't repeat them here, just summarise: They used two identical bikes each with an ibike Newton bike computer/power meter as a data logger, one bike fitted with the Nullwinds Upper Wheel Fairings, and the other without. They asked a novice rider to ride into the wind over a designated section of pretty flat road, doing a run or runs on each of the bikes, and to keep their effort level about the same for each run.

All the bike/rider data was recorded by the ibike, charts are shown in the pdf document along with other information such as weather conditions, details about the venue and tests controls. I'll list all the important details below.

Wind: Headwind of 23mph (10.3 m/s)
An attempt was made to ride each bike in similar wind conditions, so I'm going to take their word for it. You can read details of how they managed that in the document. Whether this is the actual headwind faced by the rider is hard to know, they are relying on the ibike Newton to provide the data.

Power: 149.4W
They reported 149.4W for the rider on the non-faired bike. I'll crunch the numbers to see what reduction in CdA is required to attain the claimed speed improvement at the same power. I'll also come back to this, as the power output reported for the faired bike run was not the same as for the non-faired bike run. Power is of course being reported by an ibike Newton, so who knows how reliable the data really is, but nonetheless let's assume that's the actual power and check the numbers to see if it makes sense (turns out it does, more or less, if you believe the wind speed data).

CdA: 0.372m^2 (non-faired bike)
They report a coefficient of drag area of 0.372m^2 for the non-faired bike. I've no reason to question whether that's correct or not, it's a plausible number for a novice on a standard steel framed road bike. We are of course testing relative changes due to the fairing in any case, and we'll just have to assume the rider maintained the same or very similar position on the bike.

Crr: 0.0054
They report a coefficient of rolling resistance of 0.0054 and again I've no reason to suspect that's wildly wrong as it sounds plausible for road bike on a road. I will keep that constant (as they did).

Gradient: +0.29% (unfaired) and +0.55% (faired)
This one is tricky as they report a different average road slope for each test. +0.29% non-faired test and +0.55% for the faired bike test. While the test was conducted over the same 1.5-mile stretch of road, they chose slightly different 1-kilometre sections from each run's data to make the comparison. They did this to choose a section which provided the same average headwind speed.

Mass: 188lbs (85.3kg)
They report 188lbs. I don't know if that's bike + rider or just rider but I'll assume that's total mass, and there was no mass change between the rides. On flat terrain, the outcomes in terms of impact on speed are quite insensitive to changes in mass anyway.

Air density: 1.108kg/m^3
They report 70F (21.1C) and 1020hPa for their calculations, no humidity reported but weather report they provided shows that to be between ~30% and 50%. I'll use 40% (the air density calculation is very insensitive to changes in humidity anyway). They don't report elevation but the road used was right next to Fox Airfield in California and the airfield is reported to be at an elevation of 2351 feet (717 metres) above sea level. That gives an air density of 1.108kg/m^3.

So with those power and other assumptions, using the model by Martin et al, you'd expect a rider on an non-faired bike to attain a speed of 3.32m/s = 11.93 km/h = 7.42 mph

They reported an average speed on the non-faired bike run of 7.2mph. So on the whole, the numbers seem to be in the right ball park.

OK, so what improvement in aerodynamics, that is, what reduction in CdA would be required, all else the same, to attain the claimed speed increase of 22.2% (i.e. from 7.42 to 9.07 mph)?

The CdA required at same power would be 0.244m^2.
That's a reduction in CdA of nearly 0.13m^2, a 34% reduction!

That's the equivalent of removing all of the air drag of the entire bike and some of the rider!

Houston, we have a problem.

Now here's the kicker: the faired bike run reported average power of 202.9W, some 53.5W (+35.8%) more than during the non-faired bike run. Nullwinds also reported the rider's heart rate was 10% higher for the faired bike run than the non-faired bike run.

It's no wonder the rider went faster on the faired bike.
They simply rode harder.

So knowing that, what did Nullwinds report the faired bike CdA to be?

0.369m^2, a drop of only 0.003m^2 or just 0.8% less than for the non-faired bike.

It's a real marketing bugger when the actual size of your "benefit" is quite a bit less than the error in measurement, and doesn't sound anywhere nearly as impressive as a 20+% gain in speed.

Sorry Nullwinds, your claims of big speed improvement attainable as a result of using your Upper Wheel Fairings are not plausible. Unless perhaps there are secret stashes of EPO hidden behind them.


Anonymous said...

Lazy reader here.

I haven't gone to your source, but rode iBike power meter for a few years.

The iBike doesn't measure torque to calculate rider power, it uses wind speed (pitot tube), wheel speed, and grade with a cheap accelerometer (about the quality of a Wii controller).

The CdA, Crr, total weight are plugged in, and the same equation you're using reports power.

To demonstrate the effectiveness of the fairing, you need a power meter measuring an independent value - like torque on the drive wheel.


Alex Simmons said...
This comment has been removed by the author.
Alex Simmons said...

Hi Charlie

Thanks, yes I'm well aware of how the ibike works.

In a wind testing scenario, you really need to measure both (real) power and relative wind speed (along with bike speed).

Even so, they saw a speed increase but either failed to examine why that was the case or they deliberately glossed over the real reason, i.e. the rider simply rode harder, and instead claimed it was due to their fairing. That's either foolish or disingenuous.

Given they claim all sort of flaws with the science and application of wind tunnel testing for cycling aerodynamics, you'd think they'd get some basics of their own testing right.

Cheers, Alex

djconnel said...

So the device improves power by N%! Awesome. Where can I get it?

Pete said...

they certainly didn't follow the iBike instructions for Cda:

"The Newton measures two different but related kinds of aerodynamic drag coefficients,
based on two different ways of measuring applied forces:
1) Snapshot CdA™. Whenever you’re coasting the applied force is known exactly (it’s
zero!). The rider holds his position still and the Newton measures opposing forces
while coasting (say, while coasting down a hill). The aerodynamic drag coefficient is
measured as a “point-in-time” snapshot. Snapshot CdA is very similar to what is
measured in a wind tunnel.
2) Continuous CdA™. In this situation both an Newton AND a wireless direct-force
power meter (DFPM) are on the bike, providing applied-force information to the
Newton. The DFPM provides continuous, accurate applied force data to the Newton.
The Newton compares DFPM data to opposing force data and computes drag
coefficients continuously, in real time. The restrictions of Snapshot CdA do not
apply: you can be pedaling, changing ride position, even changing clothing.
Continuous CdA provides constant updates on your CdA, and provides the basis for
a measurement called Time Advantage™.

Unknown said...
This comment has been removed by a blog administrator.
Unknown said...
This comment has been removed by a blog administrator.
Unknown said...
This comment has been removed by a blog administrator.