[kramarczyk]

I agree that boundary conditions are everything in FEA. That's why we didn't use the absolute numbers for our decision, we looked at relative differences instead. This shot is one of a series that we conducted trying to determine an appropriate web width. We started with a tube without any weight reduction and noted the stress and deflection then proceeded to take out weight. We got lucky with the geometry and didn't have any significant singularities pop up and the models convereged pretty quickly for us.
We plotted max stress vs. weight and max deflection vs. weight and looked for the knee in the graph. We had a safety factor of 10 throught the analysis, but I'm not sure if the software was basing it on ultimate or yield strengths though, so I can't draw much of a conslusion from it. Some design choices should keep the rail from taking a point load hit (i.e. bumpers) so I am pretty confident that we'll have dropped about 40% of the frame weight with no loss in performance.