Vibration loosens everything eventually
How cyclic motion undoes fasteners, and what actually prevents it.
Vibration loosens a fastener eventually because the tiny amount of clearance every thread relies on to turn freely also lets the nut and bolt shift by a microscopic amount relative to each other under repeated vibration, and if that shifting motion is not resisted, it walks the fastener in the loosening direction a little further with every single cycle, an effect that compounds steadily over enough repetitions even though no individual cycle looks dangerous on its own.
A picture frame by a slamming door
A picture frame hanging near a heavy front door shows the same one-directional drift. Every time the door slams, the frame takes a small jolt through the wall and its hanging wire, too small to see it move on any one occasion. Week after week, though, it ends up hanging crooked, rotated a little further the same way each time, although nobody has touched it. No single slam explains it. The same small jolt, repeated often enough and nudging the frame the same way every time, adds up to something plainly visible.
Friction lost for an instant, every cycle
A tightened bolt exposed to random vibration stays where it was left only because of friction, in its threads and between the bearing face of its head or nut and the surface it clamps. Nothing else holds it in that rotational position. Vibration disturbs that friction repeatedly, because the rapid back-and-forth motion through the joint, especially sideways slip between the clamped parts, momentarily releases the grip between the mating surfaces at the peak of each cycle.
During that brief release, the joint's own preload, still pulling the bolt's threads against the nut's under spring-like tension, has just enough freedom to turn the fastener by an almost imperceptible amount. The turn consistently goes in the loosening direction. A thread is a ramp wrapped around a cylinder, and a stretched bolt pulling on that ramp always pushes the nut slightly toward sliding back down it, so every moment of lost grip lets the nut slip a little the same way. The effect accumulates cycle after cycle, like the frame's drift, instead of mixing tightening and loosening in equal measure.
How far a nut has to turn back
The distance involved is remarkably small. A 10 mm steel bolt 50 mm long, correctly tightened, is stretched by about a tenth of a millimetre, while its thread advances one and a half millimetres per full turn. Turning the nut back by about a fifteenth of a turn, roughly 25 degrees, is therefore enough to give back almost all of that stretch and with it almost all of the clamping force. If vibration moved the nut by a hundredth of a degree per cycle, a drift nobody could see or measure on any single cycle, a part vibrating fifty times a second would lose its preload in under a minute. Real rates vary enormously, and most joints lose it far more slowly, but the arithmetic shows why invisible increments add up to a joint that has come apart between one inspection and the next.
Resisting the drift instead of hoping friction holds
Because the cause is a repeated loss of friction, the countermeasures work by removing the freedom to rotate or by gripping mechanically. A thread-locking compound fills the thread clearances with a cured solid so the nut has no room to turn. A serrated or properly designed lock washer bites into the mating surfaces to resist rotation mechanically, a kind of grip vibration finds much harder to defeat than smooth-surface friction. Correct preload is the first and most basic defence, since a bolt tightened to its proper clamping force presses its surfaces together harder, so a larger vibration is needed to make them slip at all. An under-tightened bolt starts with very little of that margin.
That is why a joint exposed to sustained vibration, an engine mount, a component beside a spinning shaft, or a fastener on anything moving fast over rough ground, is seldom left relying on friction alone however carefully it was torqued. Torque sets the starting friction and says nothing about how well that friction survives the repeated disturbance the joint will meet in service, so choosing a secondary countermeasure to suit the joint's vibration environment is a deliberate design decision.
Isolated joints and short inspection intervals
The drift depends on vibration reaching the joint with enough amplitude and enough cycles. A fastener isolated from the source, mounted through a rubber bush or on a structure far from whatever is shaking, may never loosen meaningfully however long it stays in service, much as a frame on the far side of the house never tilts however often the front door slams.
It also matters less on a joint that is taken apart and reinspected frequently anyway, since even a vibration-prone fastener rarely loosens badly between one short inspection and the next. The choice between a permanent countermeasure and a disciplined recheck schedule often comes down to how accessible the joint is and how often it is already serviced for other reasons.