Perhaps the point of failure is not located inside the mechanics at all. I could imagine that dust is not necessarily always 100% non-conductive and might have an impact on pcb traces if their routing is very close (sporadic shortening).
Another interesting test would be heating up those microchips and see if potentially lose pins might be reflowed by this procedure.
Another interesting test would be heating up those microchips and see if potentially lose pins might be reflowed by this procedure.