Originally Posted by
gugie
This is one of those areas that would be worthly of a university project. You could have an experienced frame builder do several sets each of simulated dropout braze-ons, and do destructive testing with equipment to measure forces. Back this up with some FEA, determine the failure mechanisms for each method. My retirement plan includes upping my torchwork time. My wife and I plan on being in an area that has a University with a good engineering project, this is the type of project I'd love to be involved with. Published results would help guide framebuilders worldwide in best known methods.
When I taught engineering design a couple of years ago, a project cost $3000. I can use a fatigue machine, but I'm not sure that's really the failure mode that builders worry about. I think that probably just hitting a brazed on mount with a hammer and looking for cracks in a microscope would be as good of a test as anything. Just to qualify that, I am totally against that kind of testing for most brazing. I have seen some incredibly inept brazing survive such a test. I have my suspicions about eyelets that fail. I suppose I should do some destructive testing of my work. Not sure where you would publish it,maybe in an open source journal.
Originally Posted by
USAZorro
I do not understand there to be a difference in strength between "factory" mounts and "after-market" mounts. If, for example, I were to have after-the-fact attachment points added by someone competent, I would trust them as much as I would an attachment that was there as the result of building the frame. Is there a difference?
Just to be clear, we are talking about eyelets like the ones that come on dropouts. Just about every other kind of rack attachment is beyond safe. I'm pretty sure the difference between built-in and brazed on is ductility. It's also not too hard to braze something like an eyelet and not get full penetration with complete bonding in the center. At least that's my working theory on why people have had brazed on eyelets fail.