GERMAN Railway's carefully cultivated image of speed, safety and luxury built up since it launched IC-Express services on June 2 1991 is in tatters. The derailment at 10.59 on June 3 of ICE884 Wilhelm Conrad Röntgen at Eschede on the Hannover - Hamburg main line and its disastrous aftermath shocked DB and other operators of high speed trains. With a death toll of 100 and about 50 people seriously injured, it was Germany's worst railway accident since 1945.

ICE884 had left München on June 3 at 05.47 and had called at Hannover at 10.30; the 300 to 350 people on board expected to reach Hamburg at 11.52. Shortly after the train had passed through Celle, some passengers became alarmed by loud noises and rattling, but this quickly ceased, suggesting that the problem had righted itself. About 2min later, the noises returned, with disaster striking immediately.

Piecing together the available evidence, the sequence of events would seem to be as follows. Around 5 to 6 km before the accident site a tyre becomes damaged or broken on one wheel of the third axle of the first trailer car. As the train approaches Eschede at the line speed of 200 km/h, it passes over a set of points about 300m before a road bridge. The damaged wheel strikes the pointwork, forcing an 8m long check rail upwards. This pierces and derails the first two trailers, causing the following cars to follow suit. While the first three trailers remain more or less upright, the fourth overturns, ending up on the east side of the four-track alignment. As these vehicles pass the bridge, one strikes the piers, causing the structure to collapse on the rear of the fifth car. The following restaurant car and trailer are crushed by the bridge deck, which now forms an immovable obstacle across all four tracks. The remaining cars, still being pushed by the rear power car, smash into the obstruction with devastating force, jacknifing against the bridge.

Over 1000 people took part in the rescue operation, which lasted until the weekend of June 6-7; trains did not run through Eschede again until June 9. Several attempts had to be made to lift the 300 tonne bridge to release the crushed vehicles - one report suggested that the restaurant car had been squashed to a height of just 160mm.

Examination of aerial photographs showed that the first three trailers were travelling on the inner of the two tracks on the east side of the alignment. The fifth car appears to have been on the outer track, which was severely damaged north of the bridge. South of the bridge this track appeared totally destroyed, while the rails of the main northbound track were bent towards the west. The rear power car's position and the bent rails suggested that it was on the outer track. The leading power car had uncoupled from the rest of the train, coming to a halt 2 km further on.

On June 4 DB announced that all ICE and IC trains would be subject to a temporary speed limit of 160 km/h, and that all 59 ICE1 trainsets would be withdrawn for special checks at the dedicated ICE workshops in München and Hamburg. The available fleet of 43 ICE2 sets plus IC and EC rolling stock was drafted in to help cover for the ICE1s, with some vehicles hired from Swiss and Austrian Federal Railways. Despite this, there were major delays and cancellations. Once the checks had been completed, the ICE1s were returned to service and given the all-clear to operate at their maximum speed of 280 km/h on the Neubaustrecken. But on June 13 DB withdrew the trains again in the wake of reports of broken wheels on the Hamburg S-Bahn and Hannover light rail cars. This time DB decided to refit the whole ICE1 fleet with monobloc wheels or resilient wheels with factory-new tyres before returning them to service.

Causes

Early reports suggested that the crash had been caused by a car falling off the bridge in front of the train, but this was quickly discounted. DB Board Member Peter Münchschwander on June 4 ruled out signalling checks as the cause of the accident - two staff had been checking a lineside cabinet by the bridge, and the car was theirs. They are believed to be among the victims. DB also said no maintenance was in hand on the affected track at the time, although relaying was under way on the route about a month earlier.

On June 4 marks were found on sleepers and trackside components about 5 km south of the crash site, indicating dragging or falling equipment. That evening investigators announced that debris recovered from the track was part of a wheel tyre, and this immediately shifted the focus of the search for the cause of the tragedy.

The main enquiry is in the hands of the Federal Railway Office (Eisenbahn Bundesamt, EBA), but DB announced on June 8 that a special committee led by Research & Technology Director Roland Heinisch was being set up to examine all aspects of railway safety. The results will influence future operating strategies and further development of high speed services.

The ICE involved, set 51, had accumulated around 3·5 million km in service and had just undergone a 20000 km maintenance examination, which included a wheel inspection - a process also carried out every two or three days. As originally set up, the wheel checks included an ultrasonic scan of the wheels, a stroboscopic check for profile abnormalities, and measurements to check for roundness. The process was largely automated, with a central computer analysing data from the ultrasonic and optical wheel scans. However, we understand that the ultrasonic modules in Hamburg and München 'were not in permanent use'. This is because the read-out was not 'absolutely definite' as serious cracks could be obscured by numerous small cracks on the wheel surface. Hardening of the outer surface is also thought to reduce penetration of the ultrasonic signals. DB was well aware of the problem, and more advanced ultrasonic wheel checking equipment was due to begin tests in the week beginning June 29.

EBA chief Horst Stuchly confirmed on June 4 that a wheel break was the likely cause of the accident, although other causes were still being investigated. He referred to a broken tyre, and the possibility of a fatigue failure or 'external effect' causing the break.

While the root cause of the catastrophe remains uncertain, there are some pointers for the investigators. ICE1 sets originally had monobloc wheels, but were refitted with resilient wheels from 1992 onwards to mitigate vibration and a drumming sound affecting the trailer cars. It is known that the ICE1s suffered from out-of-round wheels, and DB's Minden research centre had already produced a report on this phenomenon. There appears to be some linkage between this and track stiffness, and some experts suggest that the out-of-round phenomenon may be related to excitation of a loaded track resonance, with the unsprung mass bouncing on the track stiffness, possibly coinciding with a torsional resonance. Out-of-round wheels are likely to cause enormous dynamic loads, generating excessive bending of the resiliently-mounted wheel rims, and possibly leading to premature fracture.

• CAPTION: Top: The crumpled remains of ICE884, with the front part of coach 5 just visible to the top left, and the rear power car to the right of the mound of debris Associated Press RTC
• Fig 1 (above): Position of the wreckage in relation to the bridge and pointwork indicates how factors combined to amplify the effects of the derailment