bgarrood wrote:
Hi Tom,
I'm Barney, cofounder of Velosense, and I was interested to see your data. I dont know the details of the sensor you used, but I think worth a few comments.
Of course wind varies massively by geographical position and current weather, so we can't say what the norm is, but we have seen substantial swings in yaw angle - we are putting together a graph similar to yours to show this, and will be similar to what we showed Dan at Interbike.
Clearly we arent measuring the same conditions, so we can't compare our results to yours directly, but worth considering what else may cause differences in our results.
Hi Barney,
That example I posted above was just a snippet of some data I have and was selected to show how it takes quite dramatic handlebar movements (turning into, and then out of, a low-speed, 180 degree turn) for the Aerostick to see what I would consider "wild" swings of measured yaw angle. It wasn't intended to be a summary of the typical ride data.
bgarrood wrote:
For our current development we are sampling at about 27Hz, which is as fast as we can log over Ant currently (Ant+ on a production unit would be substantially slower than this, perhaps 2-4Hz). This data seems to well resolve typical gusts, which can cause <1s fluctuations in wind speed and angle. At 27Hz we are getting multiple points through each cycle of the waveform. It looks like your data is at a much lower rate, perhaps 1Hz? If so, perhaps the Alphamantis sensor is internally doing some averaging across each time step, meaning it will smooth out these higher frequency fluctuations (if they indeed exist).
Yeah...that's what I was trying to explain. I don't know the details of the inner workings of the Alphamantis setup, and so those questions I have about averaging vs. downsampling (or whatever) are present. I also don't know if the app I'm using to record the results is doing so either. In the end, all I get is a 1Hz recorded .csv file.
bgarrood wrote:
This data is really byproduct of our main thrust of being able to accurately measure drag on the road, but it has (to our surprise) highlighted how large the fluctuations are that you can see, and so the importance of having a sensor whose yaw angle range covers those fluctuations if you are going to accurately measure the average wind.
As I said, John will post some of our data on this forum, and we will keep looking at this aspect of our data and posting our findings as we go. We are always interested to hear your thoughts.
That will be interesting to see, although what has been mentioned recently here on the forum is that "steady-state" wind tunnel data does quite a good job at modeling "outdoor" power demands, even when the outside riding is done under quite variable conditions.
As taken from this classic published study:
http://cdmbuntu.lib.utah.edu/utils/getf ... e/5200.pdf As Andy Coggan likes to point out, that data implies that any additional drag caused by those yaw variations (and any separation/reattachment effects) would only account for ~3% of the total power demand at most.
In then end though, I completely agree that better measurement tools are always...well...better :-)
http://bikeblather.blogspot.com/