Downhill Demolition Derby: The 2010 Braz (Austria) Derailment
Braz is a pair of two communities (Ausserbraz and Innerbraz) in the extreme west of Austria, located 32km/20mi northwest of Ischgl and 41km/25.5mi south-southeast of Bregenz. The communities are officially part of the city of Bludenz, despite being located 7km/4.3mi to the east of the city/5km/3mi to the east of the edge of the city (all measurements in linear distance).
Braz has a station on the Arlberg Railway (“Arlbergbahn”), a 136.7km/84.9mi electrified largely double-tracked main line connecting Innsbruck with Bludenz. Opening in sections between 1884 and 1886 the line is nowadays mostly used for freight services and express trains up to the “Railjet” high speed trains at speeds of up to 160kph/99mph. Despite the use of various bridges and tunnels some sections of the line are rather steep, reaching 34‰/3.4 percent on the western ramp down the Arlberg towards Bludenz.
The train involved
Z46676 was a freight train from Curtici in Romania to Valenton in France, heading through Hegyeshalom (Hungary), Vienna (Austria), Innsbruck (Austria) and Buchs (Switzerland). It consisted of 16 bilevel car-haulers type Laaeks owned by the French railway logistics provider STSVA. Laaeks car haulers are run in pairs of two two-axle cars, meaning the train consisted of eight pairs. Each pair (they are not meant to be separated) measures 33m/108ft in length at an empty weight of 33 metric tons. They are allowed to travel at 120kph/75mph empty, when loaded with 23 metric tons the top speed drops to 100kph/62mph. Depending on the size of the cars each pair can hold up to 14 cars. Their main use is long-distance transport of factory-new cars. The units making up Z46676 were ten years old at the time of the accident. Car 372–4, which would play a pivotal role in the accident, had last been inspected/maintained in May 2010, just weeks before the accident.
Pulling the train at the time of the accident was ÖBB (Austrian National Railway) 1116 173, a Siemens ES64U2 “Taurus” multipurpose electric locomotive introduced in 1999. Measuring 19.28m/63ft in length at a weight of 86 metric tons. Each Taurus locomotive has an output of 6.4MW/8583hp (with a temporary boost to 7MW/9387hp through four motors, allowing them to reach a top speed of 230kph/143mph. Siemens made 380 first generation Taurus between 1999 and 2006 which see service all over Europe with literally everything from regional passenger services to freight trains and international express trains. One of the locomotives’ four motors was out of order on the day of the accident, but despite this handicap the total braking-power was still well above the required value.
On the day of the accident Z46676 was loaded with 208 new compact cars by Romanian manufacturer Dacia intended for the French market, mostly “Sandero” hatchbacks and “Duster” SUVs. In total the train measured 548m/1798ft in length at a weight of 863 metric tons.
On the 16th of June 2010 at 2:33am Z46676 reaches St. Anton Station, pulled by two ÖBB Taurus locomotives. The leading locomotive, ÖBB 1116 548 had helped pull the heavy train up the hill to St. Anton and was now separated from the train to head back down the way it had come. With 1116 173 alone on the train a test of the pneumatic brakes was conducted, showing no suspicious results. At 2:41am the freight train departed St. Anton station for Braz station at the other end of the Arlberg’s western ramp. The train accelerated to 100kph/62mph, a speed it could hold despite the slight downhill gradient (16 ‰) with the three working motors slowing the train instead of driving it. The pneumatic brakes were only briefly needed to slow the train to 80kph/50mph ahead of passing through Langen station. The section of the Arlberg railway behind Langen station is the steepest of the railway, heading down through Braz into Bludenz at up to 32 ‰ gradient. Here the pneumatic brakes had to be used as well to maintain the speed limit.
The driver used a sawtooth-method of braking to avoid overheating the brakes. This means that rather than permanently keeping the brakes slightly applied the train is slowed well below the speed limit, giving the driver enough time to release and then reapply them without breaching the speed limit in the meantime. The data-logger showed that the driver performed this method five times after leaving Langen station, each time keeping the train perfectly under control. 5.78km/3.6mi ahead of the site of the accident the train passed over a row of old rails that had been stored on the track, between the rails the trains ran on. The locomotive’s control system reported a sudden loss of air pressure just as the locomotive cleared the old rails, triggering an emergency stop.
But instead of slowing down the train proceeded to steadily pick up speed, breaking the speed limit and going faster and faster despite the locomotive’s system having applied full braking-power. In his desperation the driver hit the emergency shutoff of the locomotive, cutting the motors and lowering the pantograph. This ensured full application of the pneumatic brakes and no energy being fed to the train, but disabled the motors’ braking-power. The driver didn’t know that most of the train was without pneumatic brakes and that his otherwise reasonable step had reduced the little braking-power he had. In his defense, he had acted in accordance with his training and the train wasn’t going to slow down either way. But now it picked up speed even faster.
At 3:06am the train entered a long left hand turn just a few hundred meters outside Braz station, travelling at 125kph/78mph instead of the permitted 60kph/37mph. The rear five train cars got overwhelmed by the centrifugal forces and derailed, mowing down several overhead catenary support poles as they tore off the rest of the train and flew down an embankment to the right of the tracks, spilling a few dozen cars along the way. Four seconds later the switch from a left-hand to a right-hand turn overwhelmed the rest of the train, with the locomotive and the leading eight cars derailing off the tracks to the left. The locomotive rolled onto its left side as its bogies were torn away, sliding to a stop less than 2m/6.5ft from a resident’s house while the eight derailed train cars piled up behind it, throwing some of their cargo into the surrounding gardens. Only one pair of train cars remained completely on the track with minor damage. The train driver was severely injured as he was thrown around the cab, but otherwise no one got hurt as the freight train managed to miss any houses, with the late hour making for empty roads.
Mister Vonbank was woken up by a deafening rumble shaking his house just after 3am, running into his garden he finds a towering wall of twisted metal separating him from the train tracks, dotted with more or less destroyed cars. Rounding the corner of his garage he finds an 84 ton locomotive-shaped lump of metal in his driveway, having barely avoided hitting his house. So barely, in fact, that some debris hit his car parked outside the garage. Other residents aren’t as lucky, with the (formerly) loaded cars being flung into various gardens, parked cars, greenhouses and sheds.
Mister Vonbank and several of his neighbors call the emergency services before two of them manage to gain access to the locomotive, pulling the train driver from the wreckage and taking him to the nearest accessible road to be picked up by the arriving ambulances. He’s taken to the local hospital where doctors attest a concussion and severe bruising “on most body parts”, but no broken bones. The ÖBB shuts off power to the affected section, making it safe for responders to approach the wreckage. Experts contain an oil-leak on the locomotive, but 2880l/761gal of transformer-oil as well as various liquids from the cars still seep into the soil.
The first daylight shows the scale of the accident and also allows responders to find the rear part of the train. All but two of the train cars are written off, and only 33 of the loaded cars are recovered in a usable condition. The report later lists a damage of 2 Million Euros/2.27 Million USD to cargo alone, the total damage excluding that to the adjacent properties is listed at 5 Million Euros/5.67 Million USD.
Investigators started examining the train as cranes started collecting the various spilled cars, recovering the data-logger from the locomotive and comparing it to the driver’s statement. It’s soon clear that the driver did everything right, and the locomotive was in perfect working order too, as were the train cars as far as the investigators can tell. After recovering the locomotive’s wheels investigators find the brakes fully applied and strong discoloration on the brake-discs, showing that the locomotive’s brakes were fully applied for a while as the discs must have gotten rather hot to change color.
Moving to the back of the train investigators find a lot of the train cars with their breaks released, and on those that have them applied it seems like they only did so when pneumatic lines were torn in the derailment. Documentation shows that they were in perfect working order during the test at St. Anton station, and the train also properly slowed down after leaving Langen station, as evident by the data-logger showing precise changes to the train’s speed during the five applications of the brakes in the sawtooth-method. Retracing the train’s journey in reverse investigators find impact-damage on the old stored rails a short distance uphill from the site of the accident, but no sign of the rails having been shifted and no corresponding damage to the front of the locomotive. Also, had the locomotive struck the old rails and been derailed it would never have reached the site of the accident.
Examining the tracks just past Langen station investigators venture inside the Blisadona-Tunnel and find scratches and impact damage on catwalk-railings between the rails of the track, evidently something on the train hit the ground this early into the train’s doomed downhill journey. Similarly, the cover of the operating-mechanism for a set of points ahead of a siding (officially called Dalaas station) at the town of Wald am Arlsberg shows damage from a small object hitting it. Just a short distance further down the line, at the valley-side end of the siding (km 116.633), investigators stumble upon a small metal clamp sitting by the side of the tracks. Suspecting that it belongs to the crashed train the investigators confiscate it before continuing their way down the line.
Another 4km/2.5mi down the tracks investigators find a small steel cable, and finally piece the puzzle together. The cable and clamp belonged to the first pair of car-haulers, finding them separately and quite a distance from the train meant their failure was the direct cause of the accident. The pneumatic brake lines can be disconnected in the middle of each pair as well as their ends. When hooked up in the middle of a pair the connecting-mechanism, which is attached to a metal plate, is hanging below the pair’s internal coupler, being suspended off a 6.42mm thick steel cable which is clamped into a loop and attached to the coupler. This cable, the one investigators found by the side of the tracks, keeps the brake line’s lowest point at least 140mm/5.5in above the top of the rails, a mandated minimum clearance.
Investigators trace the cable and clamp to the very first pair of car-haulers, which had overturned in the derailment and lost all its wheels in the process. Twelve of the pair’s sixteen brake pads could be recovered from the wreckage, six from each half. The ones belonging to the leading half showed extensive wear, having been completely rubbed through in a few cases, while the pads belonging to the rear half showed no unusual wear. This backed up the theory that the brake line had disconnected between car 1 and 2 of the freight train, leaving nearly all of it without brakes. With eight and a half pairs of freight cars pushing from behind there was nothing that would’ve stopped the train on a downhill track. But pneumatic train brakes are designed to “fail to safety”, meaning if the lines get disconnected the air pressure would drop, applying the brakes. An air-leak would not leave the train cars without brakes. Furthermore, investigators still had to figure out how the line could separate in the first place.
Investigators recovered several more safety cables and their clamps from the train, some by opening the clamps and some by cutting the cables to leave the clamps in their used condition. A laboratory examines the cables, finding spots with varying degrees of wear and/or compression. The inspection of the cables led to the conclusion that the clamps weren’t always attached with the same tightness, which is controlled by two small nuts. As a side-note the report points out that investigators found different kinds of nuts and clamps being used on the train involved in the accident, with only five of the eight pairs using identical clamps. But their character as well as the relatively low load (approximately 10kg/22lbs) meant that this had no effect on their performance.
Testing the clamps and further examining the cables the investigators found that there seemed to be no set torque the clamps were closed with, some of the compression spots being so light that the clamps were likely closed by hand rather than with any tools. Testing showed that closing the clamps “finger-tight” would hold the cable and with that the brake line, but the rumble and vibrations from the moving train would allow the cable to start slipping, dropping the brake line below the 140mm-threshold before, at worst, completely letting it fall as the looped cable opens. Eventually the connector hit objects between the rails such as the catwalk and the points’ operating mechanism. At last the connector struck the stored old rails, breaking it open. This caused a massive air-leak forward of the separation (locomotive and car 1), triggering the emergency stop. But investigators still had to find out why the rest of the train was unaffected. Going back to the wreckage of the leading pair of cars they soon found the reason. The brake line on the leading end of car 2 had been folded pack almost 180°, pinching it at the bend and keeping any air from escaping.
Normally the stiffness of the brake line and its weight should have easily made it swing back, delaying the loss of air pressure by a second at worst. But in this case the metal plate the connector is mounted to had been thrown at the underside of the train car at such an unlucky angle that it had gotten stuck there, keeping the line pinched as the train continued on its increasingly out of control way down the line.
While most of the metal plate showed equal and expected wear and dirt one of the two metal loops holding it to the brake line looked brand new, which was the last puzzle-piece investigators needed. Apparently, when the train car had been down for maintenance just weeks ahead of the accident, the metal plate had been removed from the train car, replacing a worn metal loop with a new copy. Since the pairs are not separated during regular operation it is to be assumed that whoever installed the new loop also reinstalled the plate and then the safety-cable, insufficiently tightening the clamp. In a way, the chain of events that led to the accident was set in motion by the time the train car left the maintenance facility in France, heading to Romania with a clean bill of health to pick up the new Dacias.
During this phase of the investigation a near-identical unrelated pair of Laaeks car haulers was also examined at Hegyeshalom station in Hungary. This particular pair had had its brake line connector completely removed, featuring an uninterrupted hose between the halves of the pair. This solution, which works with far less parts and easily offers enough ground clearance, has the drawback of being more difficult to handle if a pair has to be separated. Why certain pairs got the connectors while others got a single-piece solution could not be determined. From a safety-standpoint it is the preferable solution, with nothing that could break or get caught on anything.
Considering their findings investigators calculated the braking-performance of the train after the brake line separated and got pinched, concluding that the train would have needed 5.1km/3.2mi of level track to come to a stop. It was never going to make it to a level section that long. The calculation also proved that even with one motor out of order the locomotive’s motors alone still produced 150kN of braking-power, the limit for the locomotive. Furthermore, the electric brake via the motors isn’t even a requirement to run a train, only the pneumatic brakes are. As such there was no way for the regular procedures to spot the faulty clamp before it came apart.
The investigation concluded that the derailment was caused by an unknown worker at STSVA’s maintenance facility insufficiently tightening the safety-cable clamp by likely doing it by hand. Due to the construction and placement of the connector this enabled the connector to drop as the cable slipped, eventually hitting various obstacles in its path (since it hung lower than designed) and breaking. The following separation let the base plate of the connector swing up in a way that caused it to get jammed on the underside of the train car, pinching the brake line which prevented a loss of air pressure. There was no mandated minimum torque-value for the bolts of the clamp and the cable wasn’t part of a regular pre-departure inspection, allowing this simple act of negligence to slip through. Essentially, the train was doomed no later than as it departed St. Anton station for the steep downhill journey to Bludenz, although one could argue the defect and brake-failure could have happened at any point between the French maintenance facility, the loading-ramp in Romania and the site of the accident on the way back west.
It took a week for the tracks to be repaired and enough of the wreckage to be cleared to reopen the line. The recovery of most of the loaded cars proved to be a time-consuming endeavor, especially those that had tumbled onto a meadow as the rear part of the train derailed were at times difficult to access by cranes and trucks. Most of the train and its cargo was written off, only a handful of the loaded cars was spared the scrapyard and one pair of train-cars returned to service. The final report notes “extreme” damage to private property along the rail line, without giving an exact number.
ÖBB 1116 173 was recovered and stored at Bludenz before being taken to an ÖBB maintenance-facility at Linz, being photographed there by enthusiasts at different points, showing slow progress. At that point in late July it was sitting outside on its own wheels showing severe damage from the accident to its frame and body, at another (June 2012) an enthusiast spotted the damaged body and frame placed indoors at the facility. The ÖBB ended up using one of three new bodies they had bought from Siemens and repairing the locomotive, using an unknown amount of original components. It was spotted back in service 5 years after the accident and is still in full service today (February 2022). It was the last Taurus-locomotive the ÖBB rebuilt with a spare body after 1116 017 (collision at Szőny) and 1116 062 (derailed after an unmanned runaway at Tarvisio Boscoverde).
The train driver returned to work shortly after the accident, having been quickly relieved of any responsibility for the events of the night. In 2014, then 45 years old, he settled with the ÖBB out of court and was paid 50000 Euros/56570 USD. The documentation of the settlement notes that he was lucky to escape relatively lightly injured but that mental consequences of the accident cannot be underestimated. Another lawsuit by the driver against STSVA ended with an undisclosed result. The ÖBB themselves sued STSVA for 937000 Euros/1.06 million USD, maintaining the option for further lawsuits at a later date. At the same time seventeen lawsuits were filed against the company in France, likely related to the new cars lost in the derailment.
All in all it has to be said that the accident, while being absolutely avoidable, had a comparatively good outcome. One doesn’t want to imagine what could have happened had the freight train struck the surrounding houses, and/or if the freight train had contained cars with dangerous/flammable cargo (think of fuel, gas or chemicals).
This was the second railway accident to befall Braz in fifteen years, the first being a fatal derailment in 1995 which claimed four lives just 1km/0.6mi east of the site of the 2010 derailment.
The report closes with a lengthy list of recommendations, including:
- The Laaeks car haulers should be converted to the brake lines without a connector hanging down between cars.
- Worn safety cables should be replaced earlier than before, making them less likely to slip due to damaged threads or varying diameter.
- When reinstalling the connector the proper function of the clamp should be checked, this can be done with a simple kick. If the kick lengthens the cable the clamp isn’t tight enough
- Two rather than one clamp per cable.
- Introduction of a minimum torque-value of at least 3Nm/2.2 pound-feet.
- A lower height-limit for rails stored on the sleepers of an active rail line, reducing the chance of a significant impact in case of a defect on a passing train.