Not surprisingly, there seems to be no measurable benefit or loss to FE from the "rear box cavity" design I tested this morning. This because the benefits would be small and the instruments and my approach to the test probably created too much noise in the "data."
The mod prototype (it's ugly):
I created a box like the one in post #3 above that has a flat decklid spoiler at the top, parallel sides extending downward along the trunk seam to a horizontal bottom midway across the rear bumper. All of it extending about 150mm off the back of the vehicle. Rigid. Stable. Attached with copious amounts of black duct tape. Pics:
Course and conditions:
I picked a 1 mile long hill, uninterrupted by lights and without heavily traffic.
The last "weather underground" report Friday night predicted wind at 2mph out of the SSE at 5am, shifting to 1 mph from the S by 8am. Temps were expected to be 52F at 5am and rising to 59F by 8am. Complete cloud cover throughout the morning.
Method/approach:
I drove the car 10 miles to the site to warm the tires and engine. I arrived at about 6am and was done by 740am, after which I went out for coffee with some friends. I used the Ultra Gauge speed sensor reading instead of the OEM gauge. I got the car to 20 mph or as close as possible--evenly coasting--by the time I crossed a specific crosswalk line on the road at the very top of the hill. Then I coasted, braking at the bottom, ahead of a light that was nearly always red.
I aborted any run where I could not get an consistent coasting speed at or as near as possible to 20.9 by the time I crossed the start line. The ultra gauge, I did not realize, seems precise but it not. Its speed readings jump by increments of about .7mph, which is more than 1% of the top speeds I achieved on the course.
To compensate for the rising temps and the effect they might have my speeds, I started without the box cavity attached for three runs (cooler ambient temps of maybe 52F), did six straight with the box cavity attached (presumably slightly warmer ambient temps) and then three more without the box cavity (during presumably warmest ambient temps of possibly 59F).
Results:
Numbers are speeds in mph at four different points coasting down the hill
A1 (no mod):
20.9 35.7 56.7 58.7 = +37.8
20.9 35.7 56.7 59.3 = +38.4
20.2 35.1 56.0 58.7 = +38.5
B (mod):
20.9 35.7 57.3 59.3 = +38.4
21.6 36.4 57.3 59.3 = +37.7
20.2 35.1 56.7 58.7 = +38.5
20.9 35.7 56.7 59.3 = +38.4
21.6 36.4 56.7 59.3 = +37.7
20.9 35.7 56.7 59.3 = +38.4
A2 (no mod):
21.6 36.4 57.3 59.3 = +37.7
20.9 36.4 57.3 59.3 = +38.4
20.9 35.7 56.7 59.3 = +38.4
Average final speeds: A (59.1) B (59.2)
Average speed change start/finish: A (+38.183mph) B (+38.183mph)
Conclusion:
Obviously, there was no change distinguishable from noise created by imprecision in the testing methods and gauges. The results are equivocal. Starting from a still position and using a GPS device would lessen the "noise." Hussain Ali's study of the box cavity I mentioned above showed a very small benefit, anyway. And a study by
Adrian Gaylard et al. has suggested the benefits were smaller for a coupe style body like mine. Ali also suggests that the diffuser and its angle mattered a lot.
Next Steps
Leave this mod idea behind. Do underbody panels, an adjustable angle diffuser, and wheel well skirts next. Get a GPS and do any future testing from a standing position. Maybe I'll redo this test later in the summer. A comparison before and after additional aeromods might be fun.