Tom Dalton wrote:
"At the risk of inciting all the real engineers on the list (I don't even play one on TV).... There is something called a fatigue limit. The fiatigue limit is is the magnitude of stress that a part can be cycled through indefinitely and still not fail. There is some sort of asymptotic relationship between stress level and number of cycles to failure for any steel part. At a low enough stress, the steel will never break. This is not the case for aluminum. There is no fatgue limit for aluminum, so any amount of stress, applied enough times, will eventually cause failure. I have no idea what the metallurgical mechanism is that causes the two classes of materials to behave differently. I also don't know that anyone else does. This fatigue limit stuff might just pertain to "practical" engineering experiece and be empirically derived. It's engineering after all."
As far as I can tell, the conclusions about the behavior of Al, frequently expressed in the bike literature, may not be well-founded. I expect publication soon of a different hypothesis relating Al failures to multiple strain events beyond the design limits. In short, it isn't the regular vibration that does in the handlebars, but the gross hits from jumping the MTB, etc.
I'm not a materials scientist or mechanical engineer, but advise folks to be a bit skeptical about reports of "fatigue" failures of Al w/o pre-existing cracks. I do think, though, that failures of Al tubes, etc. are more likely to be sudden; I've seen big cracks in steel that finally tore away instead of breaking cleanly. But, alas this is anecdotal...