ETCS software error led to Lötschberg derailment
A FULL service of passenger and freight trains began operating through the 34·6 km Lötschberg base tunnel using ETCS Level 2 on December 9, marking another milestone in the protracted history of Europe's interoperable train control system.
The tunnel had been formally opened on June 15, when authority was granted for commercial freight to use the route while testing continued. From September 15 a trial service of Brig - Spiez Neat Express trips joined the freight workings as a prelude to the full daily timetable of 40 EuroCity and InterCity trains plus about 70 freight trains.
During the six months of interim operation two ETCS-related problems emerged. One was a software error in the new reversible-working functionality which, had a train needed to back out of the tunnel, would have allowed it to approach too close to a preceding train. Modifications are being made to around 450 locomotives and driving trailers, but meanwhile all passenger trains through the tunnel must have a driving cab at each end.
The second problem came to light when a southbound rolling motorway train derailed on a set of points at the north portal on October 16. Double-headed by a pair of Class 465 locomotives, the train from Freiburg im Breisgau in Germany to Novara in Italy was approaching the tunnel at around 03.30 when a BLS Lötschbergbahn dispatcher decided that a postal train waiting at Frutigen should precede it. He set a path for the postal train, having issued instructions via his control panel for the other train to stop. The Frutigen interlocking sent an immediate stop instruction to the Radio Block Centre, which should then have transmitted the command to the approaching train.
But at this precise moment the rolling motorway train was in the process of switching from Level 0 to Level 2, and had already received a movement authority for the section approaching the tunnel. The stop command was never received by the locomotive's on-board equipment, and the driver-machine interface continued to indicate that the train could proceed.
The points immediately outside the tunnel portal had meanwhile been changed to allow the postal train access to the main line, and that train departed under a clear signal. However, the signal suddenly changed to red and the driver immediately halted his train.
In the cab of the Class 465, the driver noticed that the target distance in his movement authority had mysteriously reduced, and he too brought his train to a stand. He contacted the control centre by radio, asking if he could continue over 'the rest of the route', to which he received an affirmative answer. According to a lucid account in Schweizer Eisenbahn-Revue, the dispatcher had understood the driver to mean the rest of the section as far as the virtual signal protecting the points where the postal train was to join the main line, while the driver had apparently meant the section beyond the points as his DMI continued to display a movement authority to the end of the following section. The driver restarted his train, which, travelling at about 40 km/h, ran through the points set for the postal service. The leading bogie of the first loco derailed, but fortunately there were no casualties.
Clearly, the accident could have been far more serious. That it happened at all begs several questions. How did the RBC manage to 'lose' a command? How could a path have been set for the postal train without checks first being made to ensure that no conflicting movement was in progress? The rolling motorway train should have sent back an acknowledgement via GSM-R to the RBC that it had received a stop command, but failed to do so as the command never arrived. Despite no acknowledgement being received, the RBC apparently permitted a conflicting movement to be set up.
Readers will be relieved to know that the software error that caused the Frutigen incident had been corrected by the end of November. The transport ministry was satisfied that it could not recur, allowing it to authorise full opening of the tunnel.