Contents | About | Contact
Article 08, issue 02
Measuring Airspeed on a Bicycle
Julian Gordon
February 22, 2005
Bicycle computers have reached a very sophisticated level considering their recent ancestry as the simple cyclometer. Top of the line models now measure speed, altitude, temperature, pulse rate, and power meters are also available (ref 1). However, at this time, the only practical way of measuring wind effects is in a wind tunnel. Wind effects and drag are important in determining how much energy one is expending, and clearly it would be advantageous to have real time-determination.
Airspeed determination is also critical to avionics, and methods of determination have been established since the �20�s. See for example ref. 2. I initially considered the avionics experience would be a valuable precedent, and the Pitot-static method has the advantages that (a) there are no delicate moving parts (b) and that bicycle computers already utilize an air pressure sensor in the form of an altimeter. However, I did not have the capability of putting these ideas together in the form of a practical device. Also, in order to evaluate a prototype, I needed the capability of downloading data into a PC in order to establish the validity of the methodology. Once the methodology is established, a real time display will be useful. I found a preliminary solution in the form of an anemometer sold for use in air-conditioning engineering. This device made by the company Extech has a rugged fan-type anemometer device and a display which incorporates a data-logger with capability of storing data and downloading to a PC ( fig. 1).
[Figure 1]
It is not optimal for use on a bicycle, but could at least prove useful for trying out some ideas and establishing proof of principle. I mounted the two components on the handlebars alongside my conventional CatEye bicycle computer to measure speed and distance in parallel. Thus, the anemometer gives a reading of apparent or indicated airspeed (Ias) and the bicycle computer routinely measures and displays groundspeed (Gs).
Ias is not the same as the airspeed some distance from the bicycle because to get this, the device would have to be mounted clear of any parts of the bicycle influencing the air flow. This is obvious from over 80 years of avionics practice, where the Pitot tube is mounted ahead of the leading edge of a wing and clear of any fuselage parts that may disturb the airflow. Nevertheless, if there were a monotonic relationship between Ias and the undisturbed airspeed one would be able to derive one from the other by applying a correction factor. Thus we call this airspeed the corrected airspeed or Cas. Ideally, in the absence of wind,
Cas = Gs [equation 1]
Such a condition seldom occurs in practice, especially not in the Chicago area.
I decided to attempt to average out the effect of wind by selecting a straight, relatively sheltered course of about 1 km, and making the assumption that wind effects would average out by being equal and opposite when traversing the course successively in opposite directions. Thus, I took the pairwise averages of the average Gs and average Ias in each direction, repeating this at speeds ranging from the lowest compatible with biking stability to the fastest I could go. In the following the averages are denoted as <Gs>, <Ias> and <Cas>.
I set out to determine whether the averaging gave <Gs> as a reproducible and monotonic function of <Ias>:<Gs> = f(<Ias>) [equation 2]
If that were the case, then the function can be used in a calibration procedure, and permit the determination of Cas from any observed Ias.
In practice, I obtained linear relationships between <Ias> and <Gs>, similar for similar bicycle types, and also reproducible when done under relatively calm conditions or relatively gusty conditions. This first set of data with three riders and three bicycle models is summarized in Table.
Regression slope Standard error Rider 1, Cannondale Hybrid 1.226 0.013 Rider 2, Trek Hybrid 1.256 0.006 Rider 3, Bianchi Road, gusty conditions 1.114 0.021 Rider 3, Bianchi Road, calmer conditions 1.083 0.010 [TABLE] Linear relationship between Ias and Gs (unitless)
[Figure 2]
In practice, it is then a simple procedure to obtain Cas by multiplying Ias by the slope of that linear regression. I had thus collected enough data with a non-optimal device to convince myself that, in principle, an anemometer device could be used for routine Cas determination.
About this time, a device came on the market, which could be more practical. This was a handheld weather station, the ADC ("Atmospheric Data Center") Summit sold in the US by Brunton and in Europe by the parent company Silva. See Fig 2. It measures temperature, altitude, barometric pressure, windspeed and even compass direction. It has the capability of collecting data and downloading to a PC via an ir port. It also comes with a lanyard strap. I removed the strap and used the slot to attach the device to the handlebars with cable ties. The advances over the Extech device are:
1. One unit to mount instead of two
2. More compact
3. Easy mounting.A disadvantage appears to lie in the fact, that, in order to face the frontal wind squarely, it needs to be mounted vertically. In this position, the display is almost unreadable. I therefore investigated the effect of tilt angle. Surprisingly, the regression curve was not significantly different between the vertical position, and a 45 degree tilted position, where the display is easily readable. These results are shown plotted in Fig 3. Finally, the bicycle-to-bicycle variation is much smaller than for the Extech, Fig 4.
[Figure 3] Effect of tilt
[Figure 4] Road and Hybrid bicycle
To a first approximation, it may be possible to have a universal Ias to Cas correction factor built into a device, so one would not necessarily have to perform the calibration procedure described for every individual bicycle. This would be analogous to the wheel circumference measurement for a bicycle computer. One can input standard values for specified tires, or, the more fastidious user can determine the circumference directly.
In summary, my experience to date shows no reason why one cannot have a bicycle computer that will measure airspeed by very simple means. My plans are to use it to determine drag coefficients when I get out again in warmer weather, and also to persuade some bicycle computer company to incorporate airspeed measurement as an adjunct to the other functions of a standard bicycle computer. Then, for example, a real time display of the difference between Cas and Gs will give an objective measurement of whether wind is favorably or adversely affecting performance.
The material in this paper has been filed as part of a US Provisional and an International Patent Application, which will be published later this year. The patent applications include more extensive data.
References:
1. Snyder, J and Schmidt, T. 2004. Determintion of Drag Parameters Utilizing a Bicycle Power Meter. Human Power eJournal. Article 5, issue 1.
2. Stewart, C.J. 1923. Measurement of Air Speed In Aeroplanes. J.Sci.Instrument 1:43-50 (PDF file available on line at http://www.iop.org/EJ/volume/0950-7671/1).
