Considering that during this 15,000 km test the combined vehicle weight was 2500 to 5500 kg, speed limits vary from 45 to 130 km/h, altitudes from sea level to 10,000 ft and a prevailing westerly wind I just alternated on/off at each tank ~500 km. This then gave me, say, 500 km on and 500 km off at the Colorado Plateau, same at 80 mph Texas freeway speeds. Once I had convincing data I just left them on for the last 3000 km or so, since by then I was driving in heavy snow & mountain passes which gets a bit too variable.
TLAR: That Looks About Right
. These are about the size & angle of attack of aircraft VG's (400 km/h) at half spacing, installed at about the location AirTabs recommends. I didn't do the reynold's number conversion to see if the size was appropriate. I will bet double density would help.
Since then I also have tried premium gasoline which gives an 18% improvement (!) for a 10% cost increase. This is city driving and probably type-specific.
Per tank mileage variability was 9 to 25 l/100 km so the Signal to Noise Ratio here is pretty crappy. The fuel flow meter in the Saskatchewan Metro Videos would have speeded this up considerably.