[aerohead]

I've looked at the wind tunnel testing report by ARC, posted at Rocketail.com.
Some observations.
1) while actual Reynolds number for tractor trailers exceeds 5-million, ARC is limited to 1.1-million.
2) 'wing' performance is very Reynolds number dependent.
3) 'wing' performance is predicated upon free-flight conditions.
3-B) from a simple, turbulent boundary layer thickness calculation, the boundary layer at the rear of the 2016, Wabash, 53-foot trailer is 9.24-inches in depth, up to the 99% inviscid flow at infinity velocity.
4) ARC's baseline Cd for the tractor trailer drifts around a bit at zero-yaw.
5) At supercritical Reynolds number the drag coefficient would be constant up to transonic flow.
6) the testing was conducted with a top-mounted, nearly-full-span ' TOP A ( No Flex) wing mounted to the trailer van.
7) this 'TOP A ( No Flex) wing is never mentioned by Rocketail.
8) although the 'TOP A and side wings clearly add frontal area to that of the trailer van, there is no accounting for it in the drag summary tables.
9) ARC suggests a 2.20% improvement in mpg at 65-mph. A high of 2.7% at 75- mph.
10) Diesel fuel has a thermal expansion coefficient of 0.02093.
11) the average winter-to-summer temperature variability in the USA is 45-degrees F.
12) At a delta-45-degrees, a 100-gallon saddle tank could experience a natural volumetric variation to 'FULL', of 2.093-gallons at the pump, depending upon ambient temperature, above-ground fuel storage/ below-ground storage, fuel turnover rates, and how recently fuel was received from the refinery.
13) If 'volume' is the metric used to gauge mpg, mass equivalency would be required to ascertain actual energy consumed.
14) 22-separate on-road tests of Rocketail open up a virtual Pandora's Box of uncertainties with respect to all quanta.
15) If Messilla Valley truckers aren't disciplined enough to even operate their 'Trailer Tails' one would wonder as to their discipline during the Rocketail investigation.