Why subscale testing?

Last year, one particular team's single biggest problem was in getting access to flying fields for UAV flight testing. Test sites for even manual control were at remote locations, and midway through the year even those sites became off-limits for automonous testing. Almost all the team's autonomous flight testing was basically shut down for several months prior to the IARC competition, which caused major problems.

By contrast, PAD testing was easy because the vehicle was slow, lightweight and made of a soft material (styrofoam). The added safety allowed the use of a wider variety of test sites.

This year we are experimenting with a similar approach for the UAV. At this writing, a slow, lightweight vehicle made of a soft material has made flight testing much easier and more convenient in many respects. In the future we expect that the widespread availability of very small sensors and actuators should allow much of the electronics to be exercised in subscale testing.

There is an obvious risk in this approach because of scaling issues. Reynolds number effects in particular make it more difficult to extrapolate full-scale behavior from subscale test results.

On the other hand, test results of a subscale vehicle tends to be conservative regarding performance. For example, for a given angle of attack, as Reynolds number decreases, the drag coefficient CD tends to increase and the lift coefficient CL tends to decrease. CLmax also tends to decrease. So the full scale vehicle will tend to have a higher CLmax and L/D (lift to drag ratio).

Another way of reducing the risk is to impose benign maneuverability requirements on the autopilot. By designing the mission profile in such a way as to allow constant altitude, and by avoiding high turn rates, we can make it easier for a design to meet maneuverability requirements. Extrapolation would be less of an issue.

I think the added convenience of subscale testing outweigh the risks involved. The risks can be kept manageable by tailoring the mission profile and by conducting full-scale testing in parallel to confirm the results of subscale tests.

-- Frank Manning