From: Terry Blanton
Grimer seems to think it work: http://www.besslerwheel.com/forum/viewtopic.php?p=112238#112238 If I understand the essence of his argument, it is that there are interlocking mechanisms involving progressively higher derivatives of acceleration wrt time (so-called "jerk" and "jounce") which is a hypothesis the he may picked up here on vortex, as it was floating around many years ago. http://en.wikipedia.org/wiki/Jerk_%28physics%29 Two of the systems A & B are lossy but efficient (like pendulums) but there some kind of an overlap between the two - C which is a higher order derivative that does not obey CoE. Essentially, C allows the system to show net positive torque sufficient to overcome the small losses in A & B. The downside, from what I can gather from the operation of a smaller prototype is that this large system is expected to barely self-power with essentially only a few kW of excess. In that regard it is reminiscent of the Wheel of Aldo Costa http://www.youtube.com/watch?v=rsBplmMDcRQ Which probably did not work (or worked only in a breeze) . but then again, that wheel did not benefit from jounce :-)