I tried again to do kammback side panel angle testing. Interesting comment today...someone yelled out their window while driving by as I was adjusting something that I should, "take that time machine off of there". I certainly didn't expect to ever hear that comment.
All I got again was junky data (that's a technical term, of course). I would appreciate if anyone has input or suggestions of how to determine the best angles...because this isn't working. Possibly my methods are fine, but there's a fatal flaw throwing everything off somewhere, or maybe my whole strategy is flawed.
For this test, I got up to 60 mph and coasted to a stop in neutral. Again, I was planning to look for the longest coast distances starting from the same point every time. My first tests were done with the stock car. Then I added the cardboard prototype kammback panels and, holding the top panel angle at 0 degrees, I varied the side panel angles. The data were again inconsistent for the side panels and the trends were different depending on which direction I drove on the road. Then the final stock car tests at the end were nowhere close to the original stock car coasts I had done a while earlier. My car was well warmed up when I started (it had been running for over an hour).
Here's the data:
I was trying to attach the cardboard in the same positioning that I was planning for the final version, around the edge of the rear hatch, but that means that there's a step inward for the air to follow as it leaves the back edge of the side of my car and contacts the kammback sides. I knew this wasn't ideal, but I assumed I could still get decent results. Could that step be big enough (about 1/2-1") that flow trips to turbulent and renders the sides of the kammback almost useless somehow?
Maybe somehow with the shape of my car, only a top panel will really yield improvements, so leaving the top panel at 0 degrees and varying the side panel angles has done nothing but worsened air flow during testing? Maybe the side panels aren't really effective or contribute very little, so other factors I didn't notice during the test were what changed the outcomes each time?
I guess I could try building the kammback out over the headlights to the edge of the car. (I would have to use clear plexiglas for the side panels in the permanent version, of course). I could maybe see if I could get good data trying that?
Maybe tuft testing on the cardboard prototype instead of coasting?
Any thoughts as to why my data has opposite trends in each direction both times I did it? Any thoughts to help me out with future ways to do testing? I'm a little lost. Thanks.