Background
Carrbridge is a village of 708 people (as of 2024) in northern Scotland, located 32km/10mi southeast of Inverness and 105km/65mi west of Aberdeen in the Scottish Highlands.
The village has a station on the Highland Main Line, a mostly single-track unelectrified main line connecting Iverness with Perth. Most of the line dates back to the early 1800s, with today’s route and expansion being reached around 1898. The line runs through fairly remote, mountainous landscape, with trains going up to 452m/1494ft above ground on the 190km/118mi journey. Opposing services on the line are usually timed to reach stations at the same time, allowing them to pass each other through sidings at the stations.
The Train Involved
4N47 was a freight service from Inverness to Mossend Yard at North Lanarkshire consisting of 10 type FKA twin container cars, meaning 10 pairs of semi-permanently coupled cars, making for a 20 cars long train. The train was pulled by DB Schenker Rail 66048, a Class 66 locomotive loaned from its German owner to the British Stobart Rail Freight logistics provider (who also owned the containers on the train), a Warrington-based english provider rail freight provider. The BR (British Rail) Class 66 is a six-axle diesel locomotive for freight trains, produced by EMD, introduced in 1994. The type, like it’s predecessor (Class 59) is based on American locomotives but with a distinct “ducked” design to fit within the smaller British size-limits. Each Class 66 measures 21.39m/70ft in length at a weight of 129 metric tons. A turbocharged 139.6l/8520in³ two-stroke V12 diesel engine produces 2400kW/3200hp as it drives a generator providing electricty to the six traction motors. They have a top speed of 120kph/75mph, but usually run much slower with a heavy freight train on the back.
The Accident
The 4th of January 2010 was a cold winter day in Scotland, not particularly unpleasant but not nice either. Temperatures just above freezing saw arond 35cm/14in of settled snow along the rail line with mild snowfall and visibility varying from moderate to poor. A red weather alert had been issued at 6am that morning, listing the combination of severe frost, freezing fog and moderate snowfall as contributing to conditions that could risk the operation of the Scottish rail network.
The day was 66048’s first day back in commercial service after a two-month maintenance overhaul which included a close examination of the braking-system after the locomotive’s wheels, traction motors and brake cylinders had been replaced. The locomotive passed the conclusive safety examination on the third of January, allowing it to reenter service the next day. The train (then designated 4H47) arrived at Inverness with loaded containers at 11:00am, where the containers were swapped out with empty ones and the locomotive was moved to the other side for the return trip. The driver was also replaced at this point, while the “technical rider” (an assistant) remained the same.
The train now carried 20 empty shipping containers and was calculated to require a braking force of 279 metric tons. The available braking force was calculated at 462 metric tons, making the train more than capable of stopping itself. The train departed Inverness to the south at 2:52pm after conducting a brake test, running 1 hour and 56 minutes late. The train line steeply climbs pretty much as soon as it departs Inverness, snaking its way up the 400m/1312ft Slochd Summit.
The train almost immediately stopped after departure, braking to a standstill from 14kph/9mph due to a red signal, before continuing its journey behind a passenger train. The driver lightly applied the brakes as the train passed through Moy at 3:45pm at 80kph/50mph, with the ten second application bringing no reduction in speed. This was 28 minutes after the brakes had last been operated. Another application followed 4 minutes later as the train approached Tomatin, climbing the hill at 82kph/51mph. This application of the brake, again, had no influence on the train’s speed.
4N47 crested Slochd Summit at 3:57pm, travelling at 43kph/27mph. The driver reduced the throttle-setting to 0 two minutes later and applied the brakes for 21 seconds, only for the speed to slightly increase during this time. The train was now catching up to the passenger train in front, which was standing at signal AC336 (marked “F” in the graphic above). The stopped passenger train meant that the pre-signal AC336R showed a yellow “expect stop”/”caution”-message, to which the freight train’s driver reacted by shutting off the throttle again and applying the brakes. He increased the degree of brake application over the next 40 seconds as the brakes failed to reduce or even maintain the train’s speed, leading to the speed peaking at 111kph/69mph. The driver then started to also use the pneumatic brakes, having previously only used mechanically linked brakes. The train proceeded past AC336R with next to no reduction in speed, at which point the train driver triggered an emergency stop, applying the train’s full brakes as the train was travelling at 107kph/67mph. The train started to decelerate at 4:03pm, 81 seconds after the brakes had been applied, just as it passed over a speed sensor of the train control system (OSS), which registered the train’s excessive speed on approach to the red AC336. This would have triggered an emergency stop if the driver hadn’t already done so. The passenger service up ahead had luckily proceeded further down the line by this point.
The driver laid on the horn to warn people who may be near the tracks at Carrbridge station, as he believed the train was not going to stop before reaching the station. The train passed AC336 at 100kph/62mph at 4:04pm, from where the downhill gradient slightly lessened from 1.7% downhill to 1% downhill. Carrbridge Station was fitted with so-called trap points, a set of points that could protect trains at the station by directing an unauthorized train into a short dead-end siding away from the station.
4N47 reached the points at the entrance at 90kph/56mph, being directed first towards the siding and then, running over the trap points, into the run-out. The heavy locomotive obliterated the buffer stop at the end of the run-out and proceeded down the embankment into a wooded area as most of the freight cars derailed on the destroyed track behind it. 66048 missed a couple of trees and mowed down various plants before ever so slightly hitting a more solid tree on the upper right hand corner of its cab, bringing it to a stop. Behind it, six pairs of container cars had been derailed, throwing their cargo all over the place including into some locals’ gardens. Nobody was hit by the derailing train, and the two-man crew survived with minor injuries.
Aftermath
The locomotive crew climbed off their stranded locomotive after the accident, reported the accident to the dispatch center and placed clips on the rails that reported the section of track as “occupied” to the signaling system, along with attaching detonators as a secondary safety measure. Detonators are roughly coin-sized explosives that explode with a loud bang (but neglectable damage) when run over by a train. The bang is louder than the running noise of just about any locomotive, providing a reliable “stop immediately”-command to locomotive crews.
Local residents had notified emergency services, but responders found themselves with relatively little to do (for a train crash) and thus soon left the site to the investigators. The direct cause of the derailment was as obvious as it could be, with 4N47 running over trap points designed to derail an unauthorized train after it ran a red signal. The points were linked to the signaling system and automatically adjusted to the “derail”-position when the signal is red. This is to keep unauthorized trains (be it a runaway or purposeful ignorance of signals) from proceeding down the descending rail line.
The bigger question was why the train had gotten out of control, especially with the locomotive just returning from an extensive overhaul and examination. Investigators found high amounts of snow and ice in suspension- and braking-components of the locomotive and freight cars, even those which had remained on the rail line. In many cases the suspension spring, brake pads, brake head (the part moving the brake pad onto the wheel) and bogie were covered in snow and ice, with some cars also showing snow and ice on the brake pad’s wheel-side surface. The brake heads on the freight cars are attached to the mechanical brake linkage by a hanger pin at the top, pivoting the brake pad against the wheel with a very small motion. If this pivot-motion was blocked (by ice on the lower part of the pad, for example) there would be enough leverage to pivot at least a small part of the upper pad against the wheel, still providing some deceleration depending on the condition of the pad and wheel. The friction between the pad and the wheel would heat up the wheel, which could gradually melt off snow and ice and thus improve friction, increasing deceleration as more and more of the actual pad can be pressed against the wheel.
The scenario of ice and snow having to gradually melt away while the brakes are applied explains why the deceleration became slightly less insufficient during longer brake-applications ahead of the accident, such as when the driver approached AC336. Braking was suspected to be further limited by snow and ice accumulating on the brake-linkage of the freight cars, inhibiting the movement of various rods and the pneumatic cylinders, but the post-accident conditions didn’t allow for reliable testing as the derailment may have shaken off ice and snow while the deteriorating weather conditions applied new ice and snow as the train cars sat in the wreckage.
The report notes that one measure to reduce the amount of snow that can be kicked up into the bogies and brake-linkages of passing trains are line proving locomotives, maintenance trains fitted with miniature snow ploughs (MSP). MSPs are small snow ploughs approximately 450mm/17.7in high, which clear snow off the rails as the locomotive moves along. They’re not comparable to large, independent snow ploughs or snow blowers railways may also use. The line between Inverness and Carrbridge had been navigated by locomotives with MSPs several times prior to the accident, the last time just 45 minutes before the derailment occurred. This had carved a cleared path into the lying snow, but a comparison of the profiles of the FKA-cars and MSPs showed that the plow was noticeably narrower than the freight cars.
The comparison of the profiles implies that the “valley” cut into the snow by the MSP might leave the FKA cars’ springs and axle boxes (the axle-stump outside the wheel where the bearing is attaching the axle to the bogie) to be dragged through lying snow. The continued usage of MSPs over several days also increased the height of the snow right outside the valley compared to that of the surrounding area, as snow would get repeatedly pushed aside without the weather conditions causing any notable amount to melt away. The train had done the opposite trip just six hours prior, meaning it would have at that point “adapted” the shape of the cleared area, but the falling snow in the hours before the derailment may have increased snow-levels back to where they could once again touch the suspension components.
Lying snow being kicked up by passing trains (often called “snow smoke”) and accumulating on brake and suspension components is an internationally understood risk to braking performance. Snow can also melt as it is compressed under a train’s wheels, causing a water spray to land on the cold metal components where it quickly turns to ice.
The amount of snow and waterspray kicked up by trains is related to a train’s speed, with the photo above showing 4N47 at approximately 48kph/30mph. The train would eventually reach as much as 111kph/69mph according to the data-logger as it descended beyond Slochd Summit, but the investigation assumed that the function of the brakes had been degraded so far by this point that the increase in added material didn’t make a difference to the issue.
The report notes that the reduced brake performance alone wasn’t the cause of the train running out of control, such effects are well documented and understood and could have been handled. Drivers are required to conduct a running brake test (checking brake-performance without actually stopping) as soon as they can after departure, regardless of outside conditions. This check was effectively replaced by 4N47 stopping at a signal almost immediately after departure, which proved that the brakes were working as required. However, the guidelines also require that additional running brake tests be conducted ahead of entering a steep descent (such as the one past Slochd Summit) as well as conducting running brake tests every 3–5 minutes when conditions see falling snow or lying snow that is kicked up by the train. The test, which is to use the full operative brake force (the strongest brake-application short of an emergency stop) has to see a reduction of speed by at least 16kph/10mph. Locomotive-hauled trains of a certain weight are allowed to expand the interval in order to first clear particularly steep section where the test would bring the train to a stop.
A freight train suffered reduced brake performance in late December 2022 at Carstairs, Scotland, causing it to run a red signal and almost collide with two passenger trains. Here, too, the cause of the lackluster brake performance was traced back to ice and snow accumulating on braking system components. DB Schenker, the owner of the locomotive involved in the derailment at Carrbridge, reacted by issuing a reminder of the importance of brake tests, which was provided to the employee facility at Millburn Yard on the 31st of December 2009. They also asked their traffic controller to remind drivers of the importance of those tests in winter-y conditions.
The driver of 4N47 didn’t make any running brake tests that met the requirements of a successful test during the ascent towards Slochd Summit, stating to the investigation that he worried about the train coming to a stop and struggling or failing to restart motion as there were no sufficient level or even downhill section along the path. He was aware of the risk posed by snow and ice, which is why he occasionally slightly applied the brake during the ascent while leaving the engines at full throttle, aiming to “drag” the brakes and thus generate heat that could melt away snow and ice.
The driver reported to the investigation that he had slowed the train several times during the ascent, none of which were backed up by the onboard data-logger. The investigation suspected that he would do his improper brake-applications a few times when the shape of the rail line (curves, inclines) caused him to feel like the brakes were slowing the train.
This false confidence led to the driver disregarding the train actually picking up speed when he lightly applied the brakes with the throttle off just after cresting Slochd Summit, as he did so with the intent to warm up the brakes rather than actually slowing the train. This ineffective application occurred just 2.5 minutes before he would realize brake failure on the approach to AC336R, making it probable that the brake performance was already extremely redued by this point. A full application of the brakes at this point, generating the same deceleration the train achieved as it ran the red signal, would have allowed the train to stop ahead of the run-out, avoiding the derailment. However, the mixture of the driver’s improper brake tests and wrongly assumed/falsely recalled deceleration led to him not seeing a reason to do so at that point.
The report further criticizes the rule-exemption for the brake tests allowing drivers to delay the interval of running brake tests when they feel that it might bring the train to a stop, since a stretched interval (with no given maximum length) can allow the brake performance to deteriorate catastrophically even if the tests are conducted properly. Witnesses told the investigation that some drivers will delay repeated brake tests “effectively indefinitely”. It’s further noted that the improper brake tests conducted by the driver of 4N47 are somewhat common, in ignorance of the fact that the brake pads may be wet (or get wet from the melting snow), reducing their effectiveness even further when the brake-application is too light/short to evaporate the water and thus causing the pads to freeze over during the interval ahead of the next application. A frozen brake pad, even if there is no ice on the wheel or pad, will not provide nearly the same resistance as a thawed, dry one.
Another point of criticism was aimed at Network Rail, the owner of most of Great Britain’s railway infrastructure. The run-out beyond the trap points at Carrbridge station had measured 27 meters/88ft when a train like 4N47 was calculated to have required around 600m/0.4mi to come to a stop. It’s the shortness of the run-out that caused 4N47 to obliterate the run-out before careening into an adjacent wooded area, where the train got tangled in the undergrowth and made contact with several trees, bringing it to a halt. The destruction in the path of the locomotive caused the derailment of the freight cars, throwing some of them into adjacent gardens of houses that were built as recent as the early 2000s.
The run-off had been built after a fatal accident in 1940, and had remained pretty much unchanged since then. Network Rail explained that the run-off was where it was in order to keep the station safe and be as long as it topographically could be, disagreeing with the investigation on the adjacent houses (which were built between 1898 and the early 2000s) being at risk of being struck by derailing trains or debris. A passenger train had previously been directed into the run-out at low speed in January 1994, overshooting the end of the run-out by a few feet and coming to a stop upright. It was this experience that likely caused Network Rail to feel that the run-out was up to the requirements. As such, the run-out was reconstructed effectively just as it was ahead of the accident. The only difference are several trees which were cut down to recover the wreckage, now making it easier for a train to potentially “punch” through the wooded area and reach the station’s access road.
Six of the FKA twin container cars had derailed during the accident, with some of them (and their containers) being damaged beyond repair and written off after the accident. DB Schenker 66048 was also written off, being stripped for parts before being sold for scrap in 2016. DB Schenker retained the bogies as spares while it’s rumored that EMD UK pulled some of the powertrain components before sending the rest of the locomotive to the scrapyard after 11 years in service.
_______________________________________________________________
The original report can be found right here.
_______________________________________________________________
I have chosen to stay off Medium’s monetization offers to keep these stories as accessible as possible, but you can support me with a small tip via “Buy me a coffee” if you feel like it.
_______________________________________________________________
A kind reader is posting the installments on reddit for me, I cannot interact with you there but I will read the feedback and corrections. You can find the post right here.
