Hello Mark,
Wow! Quite a post and a lot to cover. I won't try and touch everything in this one, as it has become far too long.
On an overview of what technology is required, depends on what someone wants to achieve: accuracy, time required for trustworthy measurement, ease of use, etc.
With Streamlines, we are aiming to provide is an accurate and reliable system for measuring CdA that does not require an aerodynamicist to use. It will use a 1km out and back course, that can have mild undulations, a gentle incline, or flat, and can be used in calm or windy conditions. We believe that there are many professionals who can then use this equipment and provide a service.
In addition, it can serve as the R&D platform and verification standard for the development of simpler more economical devices and methods.
marcag wrote:
You are the third of maybe 5 or 6 vendors to go on the Michael’s podcast and put a certain emphasis on the performance of the air sensor portion. IMO, no sensor can ever be too good, but I personally struggle with when "enough is enough". I think that threshold is different in cycling than F1 or Motosport and since you play in both understand your approach. We spent a lot of time trying all kinds of designs to measure wind speed and yaw and found while it’s important, the incremental benefit of one system to another is marginal compare to other aspects of the system. In our opinion, air Speed measurement is not the limiting factor in these systems.
...
We even saw that the inexpensive anemometer used in the CDA/Crr app gave reasonable performance when well positioned.
I will put forward two reasons why the probe type is important.
1) If barometric pressure is being used as part of the elevation measurement, a precise value for the static pressure is critical.
Elevation difference = Barometric difference / (air density x gravity)
A pitot tube is very precise when pointed directly into the air, but as yaw angle increases, there is a small error in dynamic pressure measurement, but a much larger error in static pressure measurement.
From
https://www.unitedsensorcorp.com/pitot-properties.html "Note that yaw and pitch angle affect the readings exactly the same. The errors in total and static pressure increase quite rapidly for angles of attack higher than 5°, but they tend to compensate each other so the probe yields velocity and weight flow readings accurate to 2% up to angles of attack of 30°"
In other words both the total and the static pressure measured at angles greater than 5°, have significant errors, but the error largely cancels out for dynamic pressure measurement - but not for static pressure which is required for barometric elevation.
As an aside, a kiel type probe (
https://en.wikipedia.org/wiki/Kiel_probe) does not suffer any loss in total pressure at yaw angles up to around 50°, but if paired with a prandtl type static pressure measurement, it then leads to larger errors in dynamic pressure as static pressure decreases with yaw, the total stays constant.
In addition to this, a bicycle sees large scale turbulence created from the wind passing over the ground which manifests as a random noise in the wind angle.
To quote from the same pitot-properties page -
"Under some conditions of high intensity, large scale turbulence, could make the angle of attack at a probe vary over a wide range. This probe would presumably have an error corresponding to the average yaw or pitch angle produced by the turbulence."
One of my bosses in F1 used to say "Everyone focuses on the total pressure, but that's the easy one, it's the static that everyone ****s up." Because in any moving vehicle, the static measurement always relies on compensation or calibration due to the onset angle and the presence of the vehicle moving through the air. With the Velosense probe, this compensation is built in with an extremely accurate wind angle (I will cover rider/bike compensation later).
To summarise, pitot tubes suffer a mild (2%) error in dynamic measurement from stable yaw angle changes. These errors in dynamic pressure increase when the angles fluctuate rapidly, which is often seen in cycling. But even more so, the errors in static pressure are much larger and become very problematic, meaning that static (barometric pressure) cannot be measured accurately unless the vehicle is in smooth air.
2) The wind angle makes a big difference to the drag of a TT bike. The sail effect from deep section wheels and aero frames can reduce drag by more than 10% at a 10° yaw angle. Fortunately for angles of 10° and less, it can be measured in wind tunnels and in outdoor testing and is quite consistent for a given package.
On a windy day, we often see yaw (absolute averaged) moving +/-3° degrees between runs; 10° across a block repeat is not uncommon. This change in angle can easily provide an 5%-10% error, so it must be accounted for and normalised when comparing CdA between configurations.
On a calmer day, onset angle differences may only be a few degrees, but at 1% per degree, it cannot be ignored.
I will agree that elevation is the largest source of error and the noisiest signal in terms of real time measurement. However, if static pressure can be measured accurately in real time, the major problem with barometric pressure is that it drifts in a slow but unpredictable fashion. If the same road is traversed, and we know our position along the road, we can map the barometric pressure, or any other signal such as an accelerometer, back to each increment on the road. With more than one pass, the elevation can be determined to such a degree that it is not only removed as a source of error, but then becomes one of the most trusted elements.
Which is what I think you were alluding to when you said?
marcag wrote:
... Once you have accurate elevation all kinds of cool things can happen.
Thank you for the post, I have a lot more to say about simple sensors, body positions, racing, etc. But I will put them up in a separate reply as this one has grown far too long.
John Buckley
https://streamlines.aero