That is fascinating...having previously been an R/C airplane enthusiast I'm surprised I never heard of this effect. You got the complicated explanation here however. Try this one.
To accomplish this on a sailboat, you would pretty much have to sail around from behind a wall and back again in a circle before you lost much hull speed and repeat. Ignoring the fact that the water drag is huge, it's not going to be easy to maintain any momentum while doing all those tacks and jibes!
[img]http://www.liftzone.com/~ade/DS/images/whatis/julian.jpg" width="576" height="311[/img]
If we assume a wind speed of 10mph, and ignore any losses from model drag, turning etc., we can see that a model heading downwind over the top of the hill with an airspeed of 30mph and a ground-speed of 40mph (30mph airspeed plus the 10mph windspeed) crosses the shear boundary into the still air where immediately the airspeed becomes equal to the ground speed i.e. 40mph. You will notice the model wobble slightly as it accelerates.
Now if the model makes a 180 degree turn in the 'dead air' on the back of the hill and again crosses the shear boundary on its way back out towards the front of the hill, its airspeed where it crosses this boundary will immediately rise to 50mph (40mph groundspeed plus 10mph windspeed - now a head wind).
If the model now immediately turns back 180 degrees to the downwind direction where it began, its ground speed will be :
50mph airspeed + 10mph windspeed = 60mph
A gain of 20 mph in a single 360 degree turn!
Now imagine this with a wind speed of 50mph!
Obviously it is not quite so simple as there are factors which affect the net speed gain per circuit. Not least the drag of the model, the size of the circuit and the interference from the pilots thumbs.
