Background
Salzburg is a city of 155021 people (as of January 2020) in the north of Austria, 117km/73mi east-southeast of Munich and 138km/86mi northeast of Innsbruck (both measurements in linear distance).
Located right on the Border with Germany the city is an important traffic-hub, holding an airport, connections to two motorways (A1 and A10) as well as Salzburg Main Station, which connects 4 major railway lines by means of 9 platforms and sees 25000 travelers each day. The station serves everything from suburban rapid transport (“S-Bahn”) to international express trains, especially to Switzerland and Germany.
The trains involved
Since December 2017 the ÖBB (Austrian national railway) offers all their overnight express trains under the “Nightjet”-brand (shortened NJ for numbering), having introduced the brand the prior year. The trains use rolling stock from Austria, Germany, Switzerland, Italy and the Czech Republic, offering connections as far as Hamburg (North of Germany), Switzerland, Italy, Budapest (Hungary), Warsaw (Poland), Slovakia, Slovenia and Croatia as well as connections within Austria. The trains carry standard passenger cars with seating as well as Couchette cars and sleeper cars. Some connections also offer transport for passengers’ cars and motorbikes.
The night of the accident NJ467 from Zürich to Vienna was being extended at Salzburg main station by combining two sets of passenger cars while the passengers remained onboard. NJ467 had come from Switzerland earlier in the evening and consisted of 6 passenger cars carrying 171 passengers and a conductor. Sitting at the back of the parked group (which had its brakes applied) was ÖBB WLBmz 61 81 7694 218–2, a bilevel sleeper car made for the DB (German national railway) in 1995. Each of these distinctly dark blue sleeper cars offers 16 2-bed compartments, along with 2 3-bed compartments, being the equivalent of first class on a normal train. Each bilevel sleeper car weights 61 metric tons empty and measures 26.9m/88ft in length.
NJ467 had a scheduled wait at Salzburg main station in order to be extended, with the ÖBB integrating NJ236 (Venice-Vienna) into the train, adding the cars to the waiting train. For this the locomotive was removed from NJ236, meaning the front of the train was now an ÖBB-owned “Eurofima” four-axle second class passenger car (numbered Bmz 73 81 2191) measuring 26.4m/86.7ft in length at 42 metric tons (empty). These cars, which offered 66 seats in a second-class configuration were a standard part of the Nightjet trains, despite not necessarily wearing the dark blue Nightjet-livery. This is done because the cars get swapped between Nightjet- and regular services, while the sleeper and couchette cars are only used for Nightjet trains.
Pushing the set of 7 cars (total weight including the shunting-locomotive 469 metric tons) was ÖBB series 1163 001, a four-axle electric locomotive made in 1994. The series 1163 was acquired as part of the ÖBB’s effort to use electric locomotives for shunting wherever possible, and was developed to bring the noise levels down compared to its predecessor (series 1063). The locomotives are noted for having a centered driver’s cab with 4 identical control desks, allowing full control no matter what corner of the cabin the driver is standing in. Each 1163 weights 75.5 metric tons at 16.4m/54ft long and can reach up to 120kph/74mph, more than enough for shunting work and the occasional light freight train.
The accident
On the 20th of April 2018 at approximately 4:35am NJ467 was parked on track 5 of Salzburg main station waiting to receive 7 passenger cars that had come from Venice as NJ236. As the new cars were to be added to the front of the train the locomotive had been removed and placed on a different track (it doesn’t play a role in the ensuing events). The train cars were kept in place with their own brakes and were occupied by 171 passengers and a conductor. Because of the time a lot of the passengers were asleep or at least in their seats/beds. At the same time NJ236 had arrived from Venice, and once its locomotive had been removed also ÖBB series 1163 001 was attached to the leading end of the 7 cars instead to take them to track 5 and attach them to the waiting 6 cars. As the driver of the locomotive had to remain inside the cab even when pushing the train a shunting assistant entered the rear car and stood behind the closed doors at the gangway-connection (where, once hooked up, passengers could move between the cars). She was in radio-contact with her coworker on the locomotive, relaying information about the train’s position and distance to the waiting cars and ordering input to the throttle and brakes.
The lights inside the station were on, creating a near-daylight situation with great visibility while the roof above the tracks and platform kept possible precipitation from interfering (although there was none at the time of the accident). As both trains carried passengers during the shunting-operation the logbook noted that the operation should be performed with extraordinary care, avoiding needless jolting or noise. The speed limit throughout the operation was set at 25kph/15.5mph, with the shunted train being expected to slow down eventually and cover the last few meters at a crawl. A new conductor was scheduled to check the train after the cars had been combined and perform the brake-test before starting her shift with the trip to Vienna. The conductor stood outside the trains at the receiving end of NJ467 during the operation to watch the proceedings. At 4:40am NJ236 began approaching the waiting train cars, entering the platform-area at 25kph/15.5mph. The shunting-assistant had been on duty since 7pm the previous day and later stated to have taken part in 20–30 shunting-operations during her shift. The driver had been on duty for 7 hours after a 22-hour rest-period, being scheduled to go home by 6:35am. She carried all necessary certifications and licenses and had 9.5 years of experience working with the series 1163 locomotive. As the pushed train neared the waiting cars the shunting assistant began calling out the distances, with the driver repeating them as confirmation. “400 meters out” (approximately 1310 feet) got a confirmation, “200 meters out” (approximately 656 feet) got one also. “100 meters out” (328ft) was met with silence, no response from the locomotive.
As the train approached the waiting cars the conductor on the platform recalls thinking that the train cars were moving “unusually fast” and she could see the shunting assistant talking on her handheld radio. The shunting assistant frantically radioed her coworker several times, urging her to apply the brakes. The idea of being gentle about it went out the window, they needed to stop right now. Five radio calls to the locomotive, first calling out distances and then also urging to slow down were recorded in just 77 seconds. None were answered or reacted to.
At last the conductor saw the shunting assistant disappear from the doors at the end of the carriage as the assistant went looking for the emergency brakes. At this point the trains were approximately 25m/82 feet apart. This coincides with the last recorded radio call, marking 20m/65ft and urging the driver to stop already. Knowing a collision was imminent the conductor turned her back to the train for protection, ending her phase of being an eyewitness to the event.
At 4:42am NJ236 struck the waiting set of train cars at 24kph/15mph, pushing the 343 metric ton obstacle back 13m/42.6ft despite all brakes being fully applied. Passengers aboard both trains report being jolted out of their seats and beds, and a number of passengers were struck by luggage falling from overhead storage compartments.
Aftermath
The driver of the shunting locomotive later recalled hearing the 200m/656ft command and confirming it, before suddenly finding herself sitting in the stationary locomotive. She originally thought she was standing outside the station waiting for the points to be adjusted so she could begin pushing the train into the station, it reportedly took her a moment to notice injuries to her knee, hand and cheek. Feeling slightly dizzy but not remembering what happened she laid down in the driver’s cabin, propping her feet up on the seat. Only then did she recognize the shunting assistant radioing her, asking what had happened/how the collision had happened. Slowly realizing that she’d had an accident the driver got up and left the locomotive to try and help with whatever had happened at the other end of the train. By that time the conductor and several witnesses and passengers had notified both the ÖBB and the Austrian emergency services, who soon arrived at the station. Both volunteer and full time fire departments along with the police and medical services responded to the accident, sending a total of 92 people. The shunting assistant and conductor used emergency releases to force the doors of the cars open and went looking for people requiring assistance to leave the train before the driver could be reached. Everyone had survived the collision, but 59 people (including the driver and train staff) were injured, two of which severely.
Curiously the cars that had hit each other appeared to be largely intact apart from minor creases and waves in the outside skin. The most severe damage was suffered by car 11, in the middle of the parked group which visibly buckled. The car was deformed as a whole, shown by an entire window being ejected from its frame a few meters away from the main damage.
By 7:50am the rescue-effort was finished, with investigators taking over the site. The police questioned the conductors, shunting assistant and train driver along with another railway worker (who had been on his way over to couple the trains together), and started to piece together the chain of events that led to 232 passengers getting a very rude awakening, with some having to end their vacation in a hospital rather than making it to Vienna (which was just over 2 hours away). The driver admitted that she’d had a “dull”/nauseous feeling in her stomach as she’d waited for the points to be adjusted, but that it quickly went away and she felt “perfectly fine” as 400m/1310ft distance to the waiting cars were called out. Her total blackout from shortly after confirming 200m/656ft until after the collision was presumably caused by a random circulatory insufficiency leaving her unable to control the train as it pushed into the station, possibly the worst time for this to happen. The question that was to be answered was how the train could keep moving at speed while the driver was unresponsive. The series 1163 is fitted with a standard “SIFA”-system, a dead-man switch ensuring the driver is alert and in control of the locomotive. A pedal under the control desk has to be regularly pressed/released to keep the locomotive operational. If the driver doesn’t release the pedal it takes 900m/0.55mi for a buzzer to go off, another 75m/246ft later a louder horn sounds. If the driver still doesn’t release the pedal for another 75m/246ft an emergency stop is initiated with the traction power being killed. In the opposite way, if the pedal isn’t pressed in the first place, it takes 75m/246ft for the horn and after another 75m/246ft the locomotive stops on its own. The buzzer-part is left out in that case, meaning a stop is initiated after just 150m/492ft. Had the driver not passed out with her foot on the pedal the train could’ve come to a stop in time, or at least slowed down significantly, assuming she lost control right after confirming the 200m/656ft distance-notice. The correct operation of the SIFA-system had been checked the day before the accident.
In the end it was clear that, as bad as that sounds, the cause was an extreme case of bad luck. The driver lost control of the train due to something she couldn’t control (the report states she last had experienced such a blackout as a teenager). The slight dull feeling she had for a few seconds in her stomach wasn’t seen as enough of a reason to abandon the operation/require to be relieved. As it was she lost control at the most unfortunate moment, as the train was headed towards the waiting cars with too little space for the safety-system to stop it. Had she lost control as she pulled the cars out of the station the runaway train would’ve tripped the signaling system upon exiting the station and come to a safe stop. It would’ve caused more backup/delay as more than one track would’ve been blocked, but there would have most likely been no collision. With the railway worker who’d been on his way over to couple the trains together confirming that he’d overheard the radio-calls from the shunting assistant there was no fault to be placed on the assistant, with the driver losing control she was essentially nothing but a passenger. It’s not part of the training to teach shunting assistants were to find the emergency brakes on every train car, so she couldn’t be blamed for needing to look for it. In 2019 the ÖBB announced that they would have shunting assistants install a pneumatic kill switch on their end of an occupied train car being shunted, which lets the assistant initiate an emergency stop. The device comes with a long ripcord, pulling on it has the same effect as using the emergency brake would have but without the assistant needing to go looking for it. The ÖBB decided that the approximate 5 minutes it takes to install the device and place the ripcord are well worth making an accident like this far less likely to occur.
The accident is a sad flaw on the Nightjet-trains’ otherwise spotless safety-record. Never before or since the accident has anyone aboard a Nightjet-train gotten hurt.
_______________________________________________________________
