A bolted joint does not hold because the bolt is tight. It holds because the bolt is stretched, and that stretch — preload — clamps the parts hard enough that friction between them carries the load. When preload is lost the joint moves, the threads ratchet, and the fastener walks out. Everything in fastener locking is an attempt to keep preload from disappearing.

This catches teams out because the failure has two mechanisms and most products get the wrong fix. One is loss of clamp load with no rotation: gasket creep, plastic relaxation, thermal cycling, or embedment. The other is true rotational self-loosening, and it has one dominant cause.

Transverse Vibration Is the Enemy

Axial vibration rarely loosens anything. Transverse vibration, where the clamped parts slip sideways, loosens fasteners quickly and reliably. Gerhard Junker demonstrated this in the test that now carries his name: when the faces slip, friction under the bolt head momentarily drops to nothing and the thread's helix angle converts stored torsional energy into rotation. A few thousand slip cycles and the nut is finger loose.

The consequence is that the best anti-loosening measure is not a washer but enough preload that the joint never slips. A joint clamped to 75 percent of proof load resists transverse force far beyond what the same joint at 30 percent preload survives, and most field loosening traces back to a joint that was never properly tightened.

What the Locking Methods Actually Do

Chemical threadlockers

Anaerobic adhesives cure in the absence of air between engaged threads, filling the clearance and bonding the parts. They prevent rotation, seal against moisture, stop fretting corrosion, and are the most cost-effective locking method in production.

  • Low strength (purple, grade 222) for screws under M6 and adjustment fasteners.
  • Medium strength (blue, grades 242 and 243) is the general-purpose choice: holds under vibration, removable with hand tools, and grade 243 tolerates light cutting-fluid residue.
  • High strength (red, grades 262 and 271) needs heat above about 250 degrees C to remove. Use only where the joint is permanent.
  • Pre-applied patches and dry films arrive on the fastener and cure at assembly, costing a few cents per part and removing the operator variability that ruins liquid application. For volume production this is almost always the right form.

Two limits: anaerobics do not cure well on stainless, aluminum, or plated surfaces without a primer, and most grades top out around 150 degrees C. They also stress-crack many resins, so plastic joints need threads in plastic parts and usually threaded inserts instead.

Wedge-lock washers

A pair of washers with radial cams on the mating faces and serrations outside. The cam angle exceeds the thread's helix angle, so any attempt to rotate the fastener makes the pair climb its own ramps and increases clamp load rather than releasing it. These are the only washer type that reliably passes a transverse vibration test, they are reusable, and they work where adhesives cannot. They cost $0.20 to $1.50 per pair and need a hard, flat bearing surface.

Prevailing-torque nuts and what does not work

Nylon-insert nuts, all-metal distorted-thread nuts, and thread-forming locking screws create friction independent of clamp load, resisting rotation even after preload is lost. Nylon inserts are cheap, single-use in practice, and limited to about 120 degrees C. Both types are good secondary retention and neither substitutes for preload.

Split and tooth lock washers are the most widely specified and least effective options in the category. Under transverse vibration they perform at roughly the level of a plain washer, because a split washer flattens completely at any reasonable preload and then contributes nothing. Where failure is a safety event, use positive locking instead — safety wire, cotter pins, or staking.

Writing a Torque Spec People Can Follow

Torque is a poor proxy for preload: 80 to 90 percent of applied torque is consumed by friction under the head and in the threads, so the same torque produces preload varying by ±25 percent with plating, lubrication, and finish. That variability is why the spec has to state its conditions.

  1. State the target torque and a tolerance band, typically ±10 percent.
  2. State the condition — dry, lightly oiled, or with threadlocker applied. Lubricated threads reach much higher preload at the same torque, and a dry spec applied to a lubricated fastener yields a stretched or snapped bolt.
  3. State the tightening sequence and pass count for multi-fastener joints: a cross pattern at 50 percent then 100 percent, so the parts seat evenly.
  4. Specify the tool and its calibration interval, and note whether re-tightening after a heat cycle is required.

All of this belongs on the assembly drawing rather than in someone's memory, alongside the other content in what a manufacturing drawing must include. Where preload really matters, angle-control tightening removes the friction guesswork entirely.

Design Choices That Reduce the Problem

  • Increase grip length. A longer bolt stretches more for the same preload, so embedment or creep costs a smaller fraction of the clamp load. Short, thin fasteners are the ones that go loose.
  • Use fewer, larger fasteners. Four M6 screws at proper preload beat eight M4 screws at whatever the operator felt like.
  • Add a shear feature. A dowel pin, a shoulder, or knurling carries the transverse load so the bolts do not resist slip by friction alone. This eliminates more loosening problems than any washer.
  • Do not clamp soft or creeping material. Use a metal compression limiter through plastic or gasket layers so the fastener clamps metal to metal.
  • Match thermal expansion across the stack, or accept that a steel bolt clamping aluminum changes preload substantially over a wide temperature range.

Verify It, Do Not Assume It

Transverse vibration testing to the Junker method sorts locking schemes in hours at most independent labs. Where that is not practical, run the assembled product on a shaker through its real vibration profile and thermal cycling, mark every fastener with a torque stripe, and inspect for rotation. Fastener retention belongs in the same program as your other durability work, per reliability testing for a new product and designing a product to survive a drop test. Loosening is also often the first stage of a fatigue failure, since a slipping joint loads the fastener in bending — the mechanism behind many breakages in material fatigue in product design.

Getting Joints Right Before Production

Projects House specifies bolted joints as engineered joints: preload targets, a locking method matched to the environment, torque specs written for the assembly line, and validation testing that proves it. Send your assembly through the contact form.