Heriot-Watt has a small scale wind tunnel, good for things up to about 12 inches in height or width if I remember right; it's been a while. Not so good for real-world testing of aero positions though.
Who watched Adam Hart-Davis on Science Shack in 2001? That was when he compared an upright bike with a part-faired recumbent trike and a fully-faired trike. Bob Dixon and Mike Burrows were on hand, and Bob wrote a report of the day on the IHPVA mailing list (quoted via here):
"Readings were taken across a wind speed range of 20 to 50mph and the drag expressed as Newtons. Here's a sample of the drag:
Speed / Bicycle / Part -faired trike / Fully faired trike
20mph 16.0 8.0 5.3
30mph 36.0 15.5 8.3
40mph 64.0 31.0 15.7
It's interesting to note that a part-faired trike (nose cone and rear fairing, plus wing mirror, but no mudguards or wheel fairings) is generating only half the drag of a conventional upright; this figure gets even better as speed is increased.
Even more impressive is the fully-faired trike which is half the drag of a part-faired trike, therefore a quarter of the drag of a conventional upright bike. The fully faired trike was totally enclosed, with wheel fairings and small side ventilation apertures.
Tyre rolling resistance doesn`t form any part of these calculations, in theory the bicycle should roll easier than a trike. The bike was a mono-forked mountain bike, and Adam sat on/in each machine for the tests. The part faired trike was [Bob's] own Windcheetah and the fully faired trike was Mike Burrows own Windcheetah racer.
It's a sobering thought that the drag on Mike's racer at 40mph is still less than the upright at 20mph. The tunnel is an old RAF facility capable of generating 150mph wind forces."
That wind tunnel may or may not now be part of Manchester University.