High Speed Train in a Hurry: The 2015 Eckwersheim TGV Derailment
Eckwersheim is a town of 1339 people (as of January 2018) in the northeast of France, located in the administrative region of Grand Est (at the time of the accident called Alsace) 11km/7mi north of Strasbourg on the German border and 39km/24mi southwest of the German city of Baden-Baden (both measurements in linear distance).
The town has no train station, relying on the neighboring town of Vendenheim for its railway-connections and maintaining a network of bus stops instead. Just one km/0.6mi linear distance north of the town lies the eastern end of the new LGV Est (short for “Ligne à Grande Vitesse Est européenne” (East European High Speed Line)), a two-track 406km/252mi high speed rail line from Strasbourg to Paris which allows up to 350kph/220mph in some sections. Opening in sections between 2007 and 2016 the line, which is part of the Paris-Budapest rail corridor shortened the distance between Paris and Strasbourg from 502km/312mi to 438km/272mi and allowed much higher speeds.
Running south to north on the eastern side of Eckwersheim is the original Paris-Strasbourg rail line, a 502km/312mi double-tracked electrified main line opened in 1852. From the start the line was used for national and international express trains including the famous Orient Express. During construction of the LGV Est parts of the line, like the section east of Eckwersheim, were refurbished and upgraded for higher speeds, serving dual duty as both the original Paris-Strasbourg rail line and part of the new high speed connection. Following the high speed line towards the east the two tracks of LGV Est separate north-northeast of Eckwersheim as they climb an artificial embankment, crossing the Canal de la Marne au Rhine (“Marne–Rhine Canal”) in a wide right hand turn on two separate bridges. Continuing the turn southward the original railway line is crossed also before the tracks descend and merge into the original rail line on the eastern side of Eckwersheim. The southern (eastbound) corner has a radius of 945m/3100ft and the track is banked into the turn as much as 163mm/6.4in.
The train involved
Introduced into service in February 2008 TGV Duplex Dasye trains are the third evolution of the third Generation TGV, the french high speed train. “Duplex” refers to that generation’s unique feature, to increase passenger capacity despite maxed out train lengths these 10-part multiple units feature 8 bilevel passenger cars between their two power cars. This reliance on what essentially are locomotives also sets them apart from their German counterpart, the ICE, which has motors in every car in an underfloor-configuration. Regardless, the trains can not be lengthened/shortened during operation, as, apart from the power cars, the cars don’t have their own wheel sets but have the ends of two cars rest on a shared wheel set (a system called a “Jacobs bogie”).
As such every Duplex Dasye offers seating for 545 passengers split into two classes, 45% more than a single-story TGV, spread out over a total length of 200m/656ft (including the motor cars), relying on combining several trains if more capacity is required on routes that can handle the doubled train length. Weighting 380 metric tons empty each Dasye can put out 9280 kilowatts (equivalent to 12440hp), enough to reach a top speed of 320kph/199mph. The trains featured several improvements over the previous Generation of the TGV, including crumple zones at the end of each car and especially rigid passenger compartments despite a switch to aluminum and composite materials to bring the weight down. The “Dasye”-part of the name refers to motor cars equipped with asynchronous motors and ERTMS, a standardized signaling, train control and radio communication system intended to easy the operation of trains across European borders.
In November 2015 SNCF (French national railway) was using Dasye-train number 744 to run tests on the new LGV Est, which had only been finished (last track welded) in March 2015 and was meant to open on the third of April 2016. The train, dressed in a special livery to commemorate the new line, had arrived on site in September to run over 200 different tests and simulations to ensure everything was set for the opening. Some tests would see the train exceed the regular top speeds already programmed into the signal-system, for those tests parts of the safety-system would be turned of. For the tests part of the train’s interior had been removed, replacing standard seats with computers and desks for the engineers to monitor various sensors.
On the 14th of November 2015 TGV Dasye #744 was running several “limitation tests” on the eastern part of the new rail line, where the scheduled top speed would be exceeded by up to 10 percent to ensure a margin of error engineered into the rail line for regular operation can safely be used if need be. t 2:20pm the train was sitting at Gare Meuse TGV, a small station right on the high speed line in the commune of Les Trois-Domaines, 181km/112.5mi linear distance west of the later site of the accident. The crew was waiting for permission to depart eastbound, for which the ETCS-2 system, which usually limits train speeds, had to be shut off first. On board was a crew of 49 people, but, tragically, also 4 children of crew members, something the SNCF’s guidelines explicitly banned because, well, testing can be dangerous and allowing passengers would be an insurance nightmare. Presumably, some employees had been allowed to bring their kids and show them the super cool train go super fast without having to buy tickets.
The original schedule had called for a departure at 2:18pm, heading for an arrival at Strasbourg station at 3:17pm before the train would turn around and head back. Once given clearance to depart, Dasye 744 left the station 10 minutes behind schedule at 2:28pm using the right hand, southern track. Shortly after departure the train breached the regular top speed for the first time, reaching 352kph/220mph in a section meant for 320kph/200mph during regular operation. The train was rapidly approaching the town of Eckwersheim, where it would navigate the flying junction to merge from the newly built part of the rail line into the modernized section, changing the heading from east to south. To gain the necessary elevation for the bridges over the Marne–Rhine Canal and rail line the LGV Est’s track would climb approximately 7m/23ft as it hands into the long right hand turn, crossing the canal and, 830m/2720ft further on, the western track of the old rail line on two curved bridges integrated into the artificial embankment. Everything was running smoothly, the train’s sensors reported no unexpected data as it sped towards Eckwersheim.
To allow a safe navigation of the turn the section is limited to 160kph/100mph, during the test run the train was meant to navigate it at 176kph/110mph, a speed it was ordered to slow down to by the time it reaches kilometer point 403.809. Instead the data logger later showed the train entered the turn at 265kph/165mph, significantly faster. At that point the driver became a passenger, disaster was imminent and the driver probably knew it.
At 3:04:42pm the rear bogie of the leading motor car derails, the forces of the turn pull it outwards and tear it off the train. It strikes a concrete parapet at the beginning of the bridge over the canal, cracking the massive motor car open and tearing the rear bogie off. Oil leaks from the transformer and ignites, spreading a fire across the bridge and the banks of the canal. The derailing train rips apart, applying the brakes well beyond a point where they would’ve been any help. Most of the power car slides across the bridge over the canal, scraping along the left hand railing at an angle before going off the tracks and overturning when the railing stops holding it up. It comes to a rest 150m/490ft beyond the eastern end of the canal bridge, partway down the embankment.
Cars 2–7 derail before the bridge, even the added rigidity of the Jacobs bogie system can’t keep the forces from breaking the train apart. The heavy cars move down the embankment ahead of the bridge, their velocity carries them across the canal without even touching the water. Due to the lateral forces from the turn the cars spread out over the neighboring meadow, coming to a rest fanned out 80–130m/260–430ft beyond the far end of the canal bridge as they dig into the soft soil. Cars 8 and 9 barely make it across the canal, ending up on the eastern bank still coupled together but severely damaged. Car 9 inverts while airborne and hits the opposite bank of the canal with the roof first, crushing the car and leaving no chance of survival for the people inside. The rear motor car, having the longest way to travel after derailing, manages to stay upright and gets stuck largely in the canal, suffering a visible crease in the train-side of the body as it strikes the eastern bank. No one is caught in the fire caused by leaked oil, but 10 people are killed in the accident while 43 suffer injuries, 23 of which being listed as severely injured. The children had been riding in the leading driver’s cab, they survive the accident with minor injuries.
Dasye 744 disappears off the control center’s screens at 3:05pm (vaguely comparable to air traffic control loosing an aircraft’s signature on the radar screen), as the train’s crew repeatedly fails to respond to radio calls emergency protocols are initiated at 3:10pm, expecting a derailment and possibly a mass casualty event. The junction is shut down for any traffic, as is the Paris-Strasbourg rail line. At approximately the same time one of the companies involved in the construction of the new rail line receives a call from a surviving employee, informing them that the TGV had derailed. 100 police officers flood the scene, along with 104 firetrucks and ambulances with their crews. They are met with a chaotic sight, the wreckage looks like an upset child had a temper tantrum and threw a model railway around. Pieces of the high speed train, damaged train cars, ripped off bogies or wheels and other debris are strewn across the meadow on the eastern side of the track. And parts of the meadow and canal are still on fire.
Seven people are pulled from the leading motor car, three more than there should be in the driver’s cab at most. The fire department soon takes care of the fire, and shortly thereafter responders find the majority of the victims in cars 8 and 9, where almost no one survived. A severely injured survivor succumbs to his injuries the following evening and raises the death toll to 11. The survivors are taken away by ambulance and helicopter, one of them won’t get to leave the hospital until spring 2016. Among the dead are 4 employees of the SNCF, 5 technicians from the engineering firm responsible for the tests, and two guests of the former. One of the deceased is the director of the LGV Est project department at the SNCF’s infrastructure department. The derailment is the second fatal accident in the TGV’s history, and only the third derailment since the TGV’s introduction in 1981. Some sources point out that this accident should not be counted into the train’s safety-record, as it didn’t happen during commercial service.
Both an internal and a criminal investigation were launched within hours of the accident, with the public prosecutor’s office going as far as questioning the lightly injured train driver at the site of the accident right after he was rescued from the wreckage. He insisted to have adhered to the 176kph/110mph speed limit chosen for the test run. He, another driver who rode in the cabin and a third employee of the SNCF were suspended pending the results of the investigation. An important factor was that Paris had suffered a terrorist attack the previous day, sparking fears of a bomb having detonated on the train, but this could soon be ruled out. Investigators walk along and inspect the track behind the train, finding scratches on top of the left hand rail 200m, showing where the wheels started to climb out of the track. Approximately 100m ahead of the bridge the scratch marks move further left, showing that at least one wheel set ran on the gravel ballast at that point. These scratches are later linked to the leading motor car’s rear bogie.
The momentum kept the motor car going relatively straight (and connected to the train), but the rear did slowly start to “slip out”, eventually striking the bridge, obliterating a concrete parapet at the start of the railing and ripping the motor car in two. The remains of the bogie, along with pieces that used to be the rear section of the motor car are found on the bridge, partially covered in burned oil, with the bogie itself burring the destroyed concrete parapet.
Everything points to the train being lifted out of its track, after no evidence of an obstacle in the path of the train can be found attention turns back to the speed of the train. Each TGV Dayse has two data loggers, similar to an aircraft’s black box, one in each motor car with the leading one recording various data as the train travels down the track. The day after the accident investigators recover the leading motor car’s black box, the seal on the casing being torn causes suspicion but the data from the day of the accident appears to be valid. Matching the recorded data to the schedule it’s found out that the driver did start to decelerate, but 10.2 seconds too late, in which time the train traveled over a kilometer (0.62mi). There are 4 ways the control desk installed in the train involved lets the driver stop the train. A large red button can be hit, initiating an emergency stop, the brake-lever for the pneumatic brakes can be pulled all the way towards the driver, or either of the two levers for the electric traction brakes (reversing the engines into generators, simply put) can be pulled all the way towards the driver. Examination of the driver’s control desk shows that the whole time the train driver only used the standard brakes to control the train’s speed, the emergency stop button, which would’ve not avoided the accident but slowed the train down much faster, had not been touched. Furthermore, the driver had not used the pneumatic brake control to initiate an emergency stop by pulling it all the way towards him.
Interrogations of the surviving crew reveal that the driver was ordered to initiate the deceleration at kilometer 401 rather than at kilometer 400, relying on alleged experience from previous high speed tests that the train could decelerate better than initially expected. Probably considering the low weight of the nearly empty train to be an advantage the crew felt confident in their decision, not knowing that they had made a fatal error. The moved point of deceleration cost the train the missing seconds to slow down sufficiently. This made the accident unavoidable.
No one knows why three more people than officially allowed were in the driver’s cabin, but since the crew made matching statements that a clear order from the crew further back in the train led to the braking-point being moved the decision was made to not list distraction as the cause. As it stands the SNCF declared a miscalculation in deceleration-rates as the cause of the accident, with no criminal negligence being found in any survivor’s actions. Who exactly gave the order to move the braking-point could not be clarified, it’s unknown if it was a survivor or a victim of the following derailment. As such, there were no legal consequences
SNCF held a minute of silence on the day following the accident, and officials laid down flowers at a makeshift memorial at Strasbourg Station. On the 17th, a memorial service was held at the church in the town of Mundolsheim, just north of Strasbourg, attended by SNCF president Guillaume Pepy. A few months after the accident the site where the wreckage had come to a rest was turned into a memorial garden, with a large metal wall reminding visitors of what happened at the site surrounded by one stone pillar for each victim. Some visitors have since criticized the memorial for being disrespectfully small, as well as complaining about poor maintenance of the site.
The SNCF suspended all high speed testing until the end of the investigation, wanting to use the findings of the investigation to make their testing-procedures safer. The train was kept in storage for a few months before eventually being scrapped. The opening of the new line was delayed until July 2016, with full operations only starting by December of that year.
Some time after the accident an anonymous informant made public claims that messing with the data-loggers was not a rare occurrence, apparently train crews would delete mistakes to avoid disciplinary action.