Overview of the L16 Analysis of the Wind Data

In order to determine which effects were significant, as well as to determine which effects (if any) showed any interaction, we structured 16 of the tests to do a full factorial experiment varying four factors (an L16 experiment). The factors were:

Azimuth Angle of Attack (AoA) at 0 and 45 degrees
Elevation (El) at 30 and 60 degrees (actually measured as Zenith angle, which is 90-El)
Upwind Vent Gate (UVG) Open and Closed
Downwind Vent Gate (DVG) Open and Closed
The experiment was split over the two nights, but the division of experiments was done with optimal blocking, so that systematic differences between the two blocks of tests have no effect on the analysis.

A complete list of the tests, by block is available.
Also some comments on the stability of the wind conditions.

Results

Generally, what the results show for these values is that the following effects are statistically above the experimental noise for most of the sensors (pressure sensors and wind velocities inside the dome):

  • AoA
  • UVG
  • AoA/UVG interaction
  • DVG
  • UVG/DVG interaction
Of these, the largest effect by far is from the UVG position. Interestingly, the elevation angle was not a statistically significant effect (except at the wind sensor located at the secondary mirror) for any of the metrics investigated.

For the outside wind sensors (on top of the dome), the only significant effects were the AoA and the wind velocity itself. The blocking information in the test contains information about the mean effects of varying the wind, and this information is only confounded with 3-way interactions, we can't use it for the external sensors, because a hardware error resulted in zero readings for some of the tests.