Gregory's ServerDemonstration of how vibration can cause an improperly torqued fastener to come loose
 23 comments
 @stux The nut is rising, therefore gaining potential energy. I would've naively expected the vibration to increase the entropy of the system by tending to let the nut follow gravity, yet the nut progresses away from it. Does it matter which way the motor is turning, I wonder? I don't think I fully understand the mechanics here. Clearly, when it's tight, any vibration of the nut will oscillate between a tighter and looser position, and the looser position will win -- it's clearly higher entropy. It'll get looser each cycle. But once it gets "fully loose" on the threads, it's pretty hard to believe the nut can "know" the macroscopic fact that it is "heading for freedom". I feel like this is one of those problems in which the "obvious answer", though correct, is not quite as obvious as it seems.  @TerryHancock @stux It's because the up and down movement of the nut is not symmetric. It's like shaking a box of cereals: the big stuff rises because big and small pieces don't move the same way up and down.   @stux I guess this is why people’s bike rack bolts are always 1 thread away from falling off!  @stux That's why in some setups screws that have the opposite direction are being used... Alongside Locktite which increases friction and thus prevents vibration-based unscrewing... I used my iPhone to video something being cut on the table saw, and the screws on my metallic small rig case came loose. Now I have some Loctite holding the replacement screws in place. @jmcleod @stux it doesn’t take just one. worst case for one screw would be something like a screw from a pitot sensor coming lose and somehow destroying an engine. even if that happens, the plane is certified for that flight so you will be able to land on the remaining engine(s) and the pilots will have slightly less reliable airspeed indications on one side. that’s a bad day and a ruined trip, but everyone still gets to go home.  |
@stux fucking hell!