Why bolts loosen under vibration
A nut does not unscrew because it is shaken. It unscrews because the joint slides sideways, a few hundredths of a millimetre at a time, and each slide lets the thread ratchet back a fraction of a turn. Gerhard Junker showed this in 1969 and the test rig he built for it is now a standard. Press play to watch the joint slip and the clamp load fall, then change the locking method and see what actually holds.
What actually stops it
- Enough preload. Clamp load high enough that friction stops the joint sliding in the first place. This is the first and cheapest defence, and the one most often missing.
- A device that works when the joint slips — wedge-locking washers, prevailing-torque nuts, thread-locking adhesive, or a positive mechanical lock.
- Not a split spring washer. It is flat by the time the bolt is tight and locks nothing.
- Design out the slip — dowels, shear keys, fitted bolts, or longer bolts that tolerate more movement.
What Junker found
Before Junker, self-loosening was blamed on vibration along the bolt's axis — the nut being bounced. His experiments showed that axial vibration barely matters. What loosens a nut is transverse movement: the clamped parts sliding sideways relative to each other, at right angles to the bolt. When they slide, the friction under the nut face and in the threads is momentarily used up resisting the slide, and there is nothing left to resist the thread's own tendency to unwind under the preload. The nut turns back a tiny amount on every cycle. Over enough cycles the clamp load reaches zero.
The rig he built to demonstrate it — a bolted joint with one plate driven back and forth across the other while the bolt tension is measured — became the Junker test, written into DIN 65151 and ISO 16130. It is still considered the most severe vibration test a bolted joint can be given, and it is how locking devices are compared.
The split washer problem
The lockwasher serves as a spring while the bolt is being tightened. However, the washer is normally flat by the time the bolt is fully torqued. At this time it is equivalent to a solid flat washer, and its locking ability is nonexistent. In summary, a lockwasher of this type is useless for locking.NASA Reference Publication 1228, Fastener Design Manual, R. T. Barrett, 1990
That is not a controversial opinion; it is the finding of every Junker test ever run on one. A helical spring washer goes solid long before the bolt reaches its preload, and from that point it is a flat washer with a slit in it. It does nothing until the joint has already lost its clamp load — at which point the joint has already failed. If a drawing calls for one, the honest reading is that the drawing has inherited a habit, not a calculation.
How the methods compare
On a Junker rig the behaviour of the common methods is consistent enough to describe, though the numbers depend on the joint, the preload and the amplitude. Treat the curves in the animation as the shape of each result, not as data.
| Method | What it does | Under transverse slip |
|---|---|---|
| Plain nut, correct preload | Relies on friction alone | Loosens once the joint slips. Holds only if preload prevents the slip. |
| Split spring washer | Flat when tight; nothing | Indistinguishable from a plain nut. |
| Prevailing-torque nut (nylon insert, deformed thread) | Resists rotation with drag in the thread | Slows the loss markedly; preload still falls. Keeps the nut on the bolt after the clamp load has gone. |
| Double nut, correctly jammed | Locks the two nuts against each other | Effective when done right; sensitive to the order and torque of the two nuts. |
| Wedge-locking washer pair | Cams lock the nut against rotation using the preload itself | Holds. The standard demonstration on a Junker rig. |
| Thread-locking adhesive | Fills the thread clearance and bonds | Holds, within its temperature limits. Single use. |
| Positive lock — castle nut and pin, tab washer, wire | Physically prevents rotation | Holds. Does not preserve preload if the joint settles. |
Questions we are asked
Are split lock washers really useless?
For locking, yes — that is NASA's word, not ours, and Junker rig tests bear it out. The washer is flat at full torque and behaves as a plain washer. It can slightly delay a nut falling off a bolt that has already lost its clamp load, but by then the joint has failed. Use a plain hardened washer and get the preload right, or use a device that actually locks.
Does a nylon insert nut stop loosening?
It slows it. The drag in the thread resists rotation, so the nut backs off more slowly and stays on the bolt long after the clamp load has gone. That is genuinely useful — a retained nut is safer than a lost one — but it is not the same as keeping the joint tight. Under sustained transverse slip the preload still falls.
Why does a longer bolt loosen less?
A longer bolt stretches more for the same preload, so a given amount of joint settlement or embedding costs it a smaller share of its clamp load. It also tolerates more transverse displacement before the head or nut face starts to slide. Where the geometry allows, a longer grip length is one of the cheapest anti-loosening measures there is.
Can I re-use a prevailing-torque nut?
Not indefinitely. The insert or deformed thread wears each time it is run on and off, and the prevailing torque falls with each use. Aviation practice replaces them every use; industrial practice commonly allows a handful. If the nut runs on by hand, it is finished.
Related
Torque is not preload · Why bolts fail · Flange tightening sequence · Back to the Bolt Lab