"And an excessively strained glass lam in tension, seems to fail when the epoxy in the lam separates from the glass fiber?"
Perhaps that's partly what I saw. But the glass fibers themselves were failing too. If you fold your hands by interlocking your fingers and then straighten your fingers so they stand up a little, that's what the glass fibers were doing. Did they 'come unglued' and then break, or vice versa? I don't know.
These bows had seen decades of regular use before failing.
No glue joint issues in those bows. They stayed together. Yes, there were other root causes concerning how the fadeout was executed, it acted like a hinge of sorts, creating much strain of both types in a very specific, isolated spot, and it seems to me at least that if glass was weaker in compression resistance, that it would have failed there in that spot in compression and not tension.
Has anyone ever seen glass fail only in compression? As in crushing, fretting etc... without wood or glue joint failure?
Jeff,
thanks for the followup of your observations. I should mention my "excessively strained" observation was of a test specimen, newly constructed and strained to complete failure, well beyond normal design stresses. As for the glass fibers failing in your repair bows. I have no doubt that long term usage at some stress near the working limit will cause some breakdown in the glass fibers. Also, if you take a strand of glass from new e-glass roving and dissolve away the sizing, it can be easily seen what appears as a "continuous" fiber is in fact only a collection of shorter strands.
Is the belly a drag on the effort of the back? Does the belly need to push harder? Would that be an improvement in performance? Or, is it better to emphasize more tension in the back?
I will venture a wild ass guess that if there were more drag or hysteresis in compression or tension, it would have been confirmed by now. The technology of fiberglass has not changed all that much in 70 years. Having
all areas of the working limb, front and back, working as close to the limits
you choose in your design, would seem to be the way to optimize.