Experiments with a novel wind sensor
1st August 2025
I have been forever interested in the weather, and have had a Davis weather station on the roof for many years. So many, in fact, that it’s starting to need more and more maintenance. But also I have felt that with a bit of spare time I could build a weather station myself, and I’ve had an idea for a no-moving-parts wind sensor kicking around for a while. Now, with the time to spare, I’ve been developing the idea into a working model. This page briefly describes the design and the issues it has raised; a much longer description, for those who want more of the nuts, bolts, and bits and pieces of plastic can be found here:
Experience with a novel no-moving-parts anemometer
In essence although the unit as built does indeed measure the wind direction and speed, it’s not really a viable alternative to the traditional cup-and-vane anemometer or the higher-tech ultrasonic type. It has certain characteristics which would probably make it inadequate for any commercial realisation.
A
vertical rod, fixed at the base, has four strain gauges placed
orthogonally around its circumference near to the base. At the top is
fixed an omnidirectional “sail” or “paddle”. When the wind
blows on the paddle, the rod bends and the strain is measured on the
gauges, digitised and converted from an X-Y (North-South / East-West)
pair of values to polar coordinates and displayed on a compass rose
graph. Simple, eh?
I was quite surprised that such a fairly obvious technique didn’t appear to be commercially available as an alternative to the other types of anemometer. With no moving parts except the bending rod, it should be low maintenance and not particularly expensive to make. But building and running a couple of prototypes has shown that it’s not such an attractive idea as it seems.
There are two particular issues that show up when trying out the design. Firstly, because the maths shows that the strain at the base of the rod is proportional to the square root of windspeed, the reported value is at its most sensitive at low windspeeds or nil wind. Therefore, any drift or error is most noticeable in calm conditions; it’s almost impossible to see a consistent reading of 0mph. This is in contrast to the conventional cup anemometer where the cups simply don’t rotate when it’s calm. It’s necessary to temperature compensate the strain gauge output even though each pair can be well matched. Gross errors due to temperature variations can be compensated out, but even small errors from imperfect compensation are quite obvious.
The second problem is related to the flexing of the rod. If it’s flexible enough to give a reasonable degree of strain at the base – the prototype used 8mm dia PVC – then the mass of the paddle combined with the length of the rod causes the assembly to oscillate when the wind blows, rather than bend steadily in a fixed direction. A stiffer rod would oscillate less but also would give less deflection at the base. The oscillation means that a sensible reading can only be taken from an average of many individual measurements, over a longer period than the oscillation period. The prototype takes 32 readings over 2 seconds. Even then, because the rod oscillates partly from side to side, not just in the actual wind direction, the resultant display tends to be quite widely distributed around the actual direction.
For these reasons, and others such as the need for careful installation (to ensure the rod is vertical and the paddle is undamaged, as well as the need to provide for frequent zero re-calibration) it’s not really possible to say that this design is suitable for commercial development. Although the direction indication, when compared with a simple wind vane mounted close nearby, is somewhat “scattered” to put it kindly, the peak and average recorded windspeeds do agree pretty closely with the cup anemometer also mounted nearby.
After all this effort, I’ll keep the prototype running for a while to see what the long term reliability might be like. But the market dominance of the cup-and-vane and ultrasonic types doesn’t look like being challenged just yet.