Many thanks to Steve Maas and his link to F.E.A. analysis on the bikethink.com website. I was stumbling around looking for a similar analysis that M.Rosenblatt & Son, Inc. had to do to find out why a class of Exxon supertanker was tearing up massive bearings supporting the propeller shaft far sooner than expected. All the problems regarding flex, Q-factor (resonance or the tongue in cheek tuning fork effect), ultimate strength, and the torsional forces on any series of bulkheads as well as the bearing support structure had to be modelled. It's the modelling that is the key to comprehensive answers using Finite Element Analysis. As you can well imagine, this modelling will probably be unique for every bike because in addition to the forces at any of the joints being a function of angles, lengths, thickness of tube, dimensions of lugs, and the classes of forces applied -- lateral, vertical, torsional -- , you have the odd idiosyncracies of the maker, namely lug profiles or lack of same, welded or brazed connexions, torsional variations due to top tube placement and bar/steering geometry design. The end result is something ideally suited to data processing, namely a matrix. Once you've amassed the data for these hundreds of nodes of force you can put the model through its paces, applying various forces at any number of points, modelling wheel forces and watch the interaction, input the various modulus characteristics of the tubes and see if anything happens, and just as easily model the bike as a resonating structure -- with some damping features as well -- to see if imperceptible flex is magnified in some fashion. You just might end up with surprising results as chaos theory would suggest. It's a whole lot of work, but you can create data packets as we've seen in bikethink and build to something comprehensive. The guys who did the modelling on the Exxon supertankers were John Daidola, Wasfi Youssef, and Nick Fotakis in case you'd like some insight.