Return to Archive -by date - by topic - Archive 2002.
Points failure caused the Potters Bar derailment but why it failed is an on-going riddle for the accident investigators
It was pretty certain from the start that the derailment of the 12.45 WAGN service to Kings Lynn at Potters Bar on 10 May was caused my mechanical failure. And as soon as you saw the lay of points 2182A, what had failed was equally clear.
These facing points are part of a ladder crossing just south of Potters Bar. They link the down fast to the down slow.
Set for a through train on the fast line you would expect the left hand blade tip to be against the stock rail and the right hand blade tip to be clear of the rail to allow the wheels to run through. But immediately after the accident you could see both blades against their respective rails. In other words, the right hand blade was positioned as if the route had been set to the down slow, while the left hand blade was set for the fast line.
Since the position of each blade is detected and the route cannot be cleared until the blades are locked, the points must have been set for the fast line. Then, as the last coach of the Class 365 was passing, something broke, allowing the right hand blade to spring out. This deflected the rear bogie to the left, flicking the driving vehicle sideways.
As the cab end swung round to the left, the coupling to the third coach gave way, the shock derailing the trailing bogie of the third coach. The visible buckle in the track after the points show the forces involved. With the brake pipes severed, the three leading vehicles braked to a halt.
Now sideways on, the derailed vehicle collided with raised the beams of an overbridge. Debris falling onto the road below killed a pedestrian and damaged several cars.
This impact may have caused the vehicle to roll while hitting the ramps of the station's two island platforms. Several passengers were thrown out of the train at this point.
Sliding along the platform the coach demolished a waiting room, before it reached the main station structure where it finally stopped, jammed under the concrete canopy. Five passengers were killed and one of the injured later died in hospital.
At which point note the way in which the long welded aluminium extrusion Network body maintained its structural integrity. One bogie remained attached.
Immediately thoughts went back to the Ealing derailment when a battery box lid on a Class 50 came unlatched, dropped down and hit the mechanism of a set of points causing them to move under the train. But in this case, the blades of the points hadn't moved together.
On 2182A the switch blades were held together by three transverse stretchers.
At the blade tip end of the points the lock stretcher was a rectangular section steel bar, bolted to the webs of the switch blades over the slides. This lock stretcher was driven by the point motor. It was also connected to the detector rods which provide the mechanical interlock when the blades have moved fully over to the new position.
For the sake of clarity I will call the other stretchers tie bars' and number them 1 and 2 as seen from the blade tips.
These tie bars are rectangular section rods, say 30mm across, with the end turned to a circular section and threaded. The threaded ends are inserted into holes in steel plate brackets bolted to the switch blades.
Two pairs of nuts on each threaded section, a nut plus a thinner lock nut, clamp the tie bar to the bracket. Thus, by slackening the pair of nuts on one side of the bracket and tightening the other pair you can adjust the distance between the switch blades when setting up the points.
These are rugged components. The nuts are 30-50mm across flats. In addition to the lock nut, there is also a lock washer between the nut and the bracket. You would need two big spanners, probably an extension bar, and lot of muscle, to undo these nuts.
Accident investigators found that at one end of tie bar 1, the nuts on the rail side of the bracket had come off the end of threaded section. On tie bar 2, they found the same situation at the opposite side of the track.
These missing nuts meant that one end of each tie bar could slide backwards and forwards in its bracket without transferring any force to the other rail when in tension. And not only did the tie bars locate the rails, one of them was connected to the switch motor to provide a second drive for the points.
With the tie bars ineffective, all the switch forces were being carried by the lock stretcher, which eventually failed as the Class 365 was passing.
But how could these pairs of nuts have come off?
Since the Hatfield derailment, Railtrack's Eastern Region has been leading the drive to bring more responsibility for maintenance in house (see my article in the xx Modern Railways). As part of this regime, a team of engineers from Eastern Region and maintenance contractors Jarvis jointly inspected the Hatfield points on 20 March this year and gave them a clean bill of health. Indeed, the condition was considered to be above average.
On 3 April there was a facing point check. This is a rigorous check of operation, including ensuring that the detection and locking is working correctly. It was followed by a visual inspection by a patrolman on 24 April.
Then, on 1 May a team of three Jarvis engineers, all certificated under the Institution of Railway Signal Engineers, scheme carried out another facing point check. According to the inspection sheet forwarded to Railtrack only trivial items were noted.
Finally on 9 May, the points were again inspected visually by a patrolmen. Jarvis added that he was protected by a lookout, so we can assume that the points were inspected from the four foot. The patrolman also saw nothing wrong.
This history of inspections set the sabotage hare running. With such recent checks, surely nuts couldn't have dropped off. Particularly two sets of nuts. Perhaps someone with a knowledge of railways had undone the nuts in the night?
Then, on Saturday, Jarvis told the investigating team that during the 1 May inspection a set of nuts had been found to be detached on 2182A. The company said that they had been replaced and retightened during the inspection.
Why such an unusual occurrence had not been entered on the inspection report is, as yet, unknown. It was a potentially dangerous fault. Jarvis was quoted in the press as saying that since the fault was rectified, it was not listed.
But that still doesn't explain why two sets of nuts came off the ends of the tie bars near simultaneously. As Informed Sources went to press there were reports that the tie bars had been offset laterally. In other words the nuts on one end would be nearer to the end of the screw threaded section and vice versa.
In the worst case, if the lateral transposition were large enough, there would be insufficient thread projecting from the bracket to take the locking washer, nut and lock nut. If, and I theorise, if there was not enough thread for all three, without the lock nut it would be possible for the main nut to vibrate loose.
Determining how the tie bars might have become displaced is crucial to the investigation. One reason for moving the tie bars would be to replace the bush which insulates the tie bar from the rail where it passes through the bracket.
But to do this, you would simply take the nuts off the outer end of the tie bar, loosen the lock nut on the inside pair of nuts at the other end and turn them inwards towards the centre line of the track. This would unclamp the tie rod, allowing it to be slid sideways out of the bracket so that the bush could be replaced.
During this procedure you would leave the other two pairs of nuts in situ because when you had replaced the bush and slid the tie bar back into place, they would automatically locate the tie bar in its original position greatly simplifying and shortening the job. Contrarily for a correctly located tie bar to become displaced to one side, you would have to undo all four pairs of nuts and do them up again in new positions.
And when you came to do up the nuts, the shorter length of thread on the outside at one end would have been immediately apparent.
Published on 14 May, the HMRIs Initial Report was short on information, other than confirming that nuts were missing and the lock stretcher broke. The points have been lifted and sent to the HSE's Sheffield laboratory for forensic testing. One task will be to analyse the threads of the nuts the detached and the tie rod ends to estimate how long the nuts had been off.
To his credit Transport Secretary Stephen Byers handled the aftermath of Potters Bar maturely reassuring travellers that it was a one-off, something confirmed by Railtrack's subsequent inspection of over 800 points nationwide.
He was promptly flacked for complacency by the opposition who, predictably, called for a public inquiry. Heavens forfend. I suspect that this is one for the accident inspectors to solve, not lawyers to obfuscate over.