The Nut That Walked Away: What the 2002 Potters Bar Derailment Teaches About Locking Fasteners

At 12:58 on a Thursday afternoon in May 2002, a commuter train heading north from London King's Cross derailed at Potters Bar station in Hertfordshire. Three of its four carriages left the rails. The fourth went airborne, struck the parapet of the Darkes Lane road bridge overhead, rolled through 360 degrees, and came to rest wedged beneath the station canopy roofs. Seven people died. More than 70 were injured.

Investigators from the UK's Health and Safety Executive (HSE) spent more than a year examining the wreckage, the maintenance records, and the hardware. Their conclusion, published in a 2003 progress report by the HSE Investigation Board, pointed to a failure most engineers and maintenance professionals will recognize the moment they read it. And one that still happens in workshops and rail yards everywhere.

The nuts had walked off the bolts.

What Failed, and How

The derailment happened at a set of railway points. What North Americans call a switch or a turnout. Points are the movable sections of track that guide a train from one line to another. They're held in position by a system of steel bars called stretcher bars, which keep the switch rails at the correct spacing and transmit motion from the point machine (the electric or hydraulic actuator that moves the rails).

At Potters Bar, the HSE investigation found that nuts securing the stretcher bar brackets to the switch rails were absent. They had backed off and eventually fallen away. The front and rear stretcher bars were no longer doing their job. The load shifted entirely onto the lock stretcher bar, a structural component not designed to carry that force on its own.

The lock stretcher bar then failed by fatigue, cracking through one of its bolt holes under repeated loading. When the rear wheels of the third coach reached those points, one switch rail sprang outward. The wheels on a single axle were simultaneously routed in two opposing directions. Three coaches derailed. One went airborne.

The HSE report also noted that other sets of points in the Potters Bar area showed similar evidence of nut loosening. And that some had already received attempted remedial fixes, suggesting the problem was recognized, at least in part, in that maintenance district. In 2011, Network Rail was fined £3 million in connection with the disaster following a prosecution by the Office of Rail Regulation. (BBC News, May 2011)

Why Plain Nuts Fail in Vibration

The mechanism behind this failure has a name: vibration-induced self-loosening. Every time a train passes over a set of points, it delivers a pulse of dynamic load. Vibration, shock, and lateral force. To the hardware underneath. Standard hexagonal nuts resist loosening through friction: the clamping force between the nut face, the washer, and the clamped surface keeps everything in place.

The problem is that under cyclic transverse loads. The kind that railroad switch hardware experiences hundreds of times every day. That friction breaks down. The nut doesn't suddenly spin off. It rotates just a fraction of a degree with each load cycle, gradually reducing clamp load until the joint has lost most of its preload. Once clamp load drops to zero, the nut is essentially free to continue backing off.

This is the fundamental problem with using a plain nut in any application where vibration or dynamic shock are part of normal service. Friction-based retention is a degrading system. In a high-vibration environment, it eventually fails.

The Junker test. A standard laboratory method for evaluating fastener self-loosening. Demonstrates exactly this behavior. Transverse vibration causes nut rotation even in properly torqued joints. The clamping force that held everything tight at installation slowly disappears with each cycle. A joint that passes a static load check on Monday may have lost half its preload by Friday if it sits in a high-vibration environment and nothing is done to mechanically prevent loosening.

The Right Fastener for a Vibration Environment

In railroad switch hardware. And in any application where vibration or cyclic loading are part of normal service. Fastener selection has to account for the full service environment, not just the static load.

Several fastener types are engineered specifically to resist self-loosening:

Prevailing torque nuts. These are the workhorses of vibration-resistant fastening. Polymer-insert locknuts (nylon insert locknuts, sometimes called "Nyloc" nuts) use a ring of nylon that deforms around the bolt thread, creating resistance to rotation independent of clamp load. All-metal prevailing torque nuts accomplish the same thing with a deformed thread section. Either type maintains grip even as clamp load fluctuates. Because their resistance to rotation doesn't depend on friction between clamping faces.

Castle nuts with cotter pins. Used extensively in aircraft, heavy equipment, and safety-critical joints, this combination provides a positive mechanical lock. A cotter pin through aligned holes in the bolt shank and slots in the castle nut physically prevents rotation. It cannot loosen without the pin shearing or being deliberately removed. This is what engineers call positive retention. It doesn't rely on friction at all.

Lock plates and tab washers. Common in European rail hardware and industrial equipment, these washers have tabs that bend up against the flat of a nut after tightening. The nut cannot rotate without bending the tab, which requires deliberate force and is immediately visible during inspection. A bent tab is a sign the fastener has been disturbed. Which is exactly the kind of field feedback that maintenance crews need.

Wedge-locking washers. Products like Nord-Lock style washers use opposing cam profiles cut into paired washers. When the joint tries to loosen, the cam angle forces the joint to expand axially, tightening the bolt rather than allowing it to back off.

The common principle across all of these: they don't rely on friction. When vibration has reduced clamp load, they still hold. That's the design requirement for any connection subjected to repeated dynamic loads.

Choosing for the Application, Not Just the Spec

The Potters Bar failure wasn't about the wrong grade of bolt or the wrong material. The steel itself wasn't the problem. The problem was the wrong type of fastener for the service environment. Standard plain nuts in a location where hundreds of train passes per day subjected every threaded connection to continuous vibration and lateral loading.

A plain nut is the correct choice in many static or low-vibration applications. It's cost-effective, easy to install, and works perfectly well when loads are primarily static and the joint won't see significant transverse movement. The error at Potters Bar wasn't picking a bad fastener in general. It was picking the wrong fastener for that specific application.

This is the selection decision that matters: matching the fastener type to what the joint will actually face in service. Load, vibration, environment, inspection interval, and consequence of failure all factor in. For a railroad switch that gets pounded by train traffic day after day, a positive-locking fastener isn't a precaution. It's the basic specification.

Right Fastener, Right Application

The HSE report on Potters Bar documented evidence of nut loosening across multiple sets of points in the area. The warning signs existed before the crash. The problem was recognized in maintenance records. The right hardware could have held everything in place.

A plain nut in a vibration environment is optimistic engineering. It works until it doesn't. And when it doesn't. On a railroad switch, an engine mount, a suspension linkage, a pressure vessel bracket, anywhere that dynamic loads are part of the job. The consequences can be severe.

At Potters Bar, seven people boarded a train for a routine commute and never came home. The root cause was threaded fasteners that weren't designed for the environment they were in.

The right nut, in the right application, stays put. That's the job. Make sure you're selecting the one designed to do it.

Sources: UK Health and Safety Executive Investigation Board, Train Derailment at Potters Bar 10 May 2002. A Progress Report, May 2003. BBC News, Potters Bar crash: Network Rail fined £3m, 13 May 2011. BBC News, Seven dead in Potters Bar crash, 10 May 2002.