Background
Melbourne is a city of 5.2 million people (as of 2023) in southeast Australia, located in the state of Victoria 659km/409mi southwest of Canberra (the Australian capitol) and 654km/406mi southeast of Adelaide (both measurements in linear distance).
Melbourne is home to a sprawling commuter rail network, including the Craigieburn Line from Craigieburn Station in the north to Flinders Street Station, Australia’s busiest train station right in the heart of the city. The 27km/17mi long double-tracked electrified rail line is constructed in 1600mm broad gauge (compared to the standard 1435mm track width) and serves 21 stations as of April 2024.
The Trains Involved
Train 5264 was a commuter rail service from Broadmeadows to Flinders Street, to be performed by train 393M-1048T-394M, which had also done the trip from Flinders Street to Broadmeadows as train 5859. The train was a first-series Comeng-Train, a three-car electric multiple unit (EMU) consisting of two motor-cars (393M and 394M) with an unpowered trailer car (1048T) in the middle. The Comeng trains, named after their manufacturer Commonwealth Engineering, were introduced in 1981 and measured 71.2m/234ft in overall length at an empty weight of 141 metric tons. They offer 295 seats in a single-class configuration and are allowed to carry as much as 696 passengers in total, reaching speeds of up to 115kph/71mph.
Sitting at Spencer Street Station in Melbourne was a four-car “Country Train” operated by VLine (written as V/Line by the company themselves). The train consisted of four H-type passenger cars, a four-axle regional passenger car introduced in 1956 as part of Harris-type EMUs before being rebuilt into unpowered train cars in the 1980s. The cars can seat 54 to 87 passengers each, depending on configuration, with the four cars involved in the accident offering a combined capacity of 247 passengers. The four train cars had a combined length of 71.26m/234ft at a weight of 133.5 metric tons.
The Country Train was pulled by VLine N-Class number 463 “City of Bendigo”, a six-axle multipurpose diesel locomotive made by Clyde Engineering in Australia starting in 1985. Unit number 463, just like its train cars and the Comeng-train, was built for the Australian broad gauge of 1600mm, while other units of its type were also made for 1435mm regular gauge as Australia uses both standards. Each N-Class locomotive measures 18.87m/62ft in length at a weight of 123 metric tons and can reach 130kph/81mph.
The Accident
The Comeng arrived at Broadmeadows station at 9:14pm on the third of February 2003, terminating the train service number 5859. Both the train and the driver were meant to also provide the return trip as train service number 5264 after a short break. The driver stepped out of the cab of 393M (which had been leading on the trip into Broadmeadows station) for some time to discuss with the station master whether or not he had to move the train from the platform to a storage track so that the platform could be used by another train in the meantime before reentering the cab in order to enter the next trip’s information into the train’s onboard computer, setting up the passenger information screens. The dispatcher also set the path and signals for the trip at that point. The driver then “isolated” the cab (“turning it off”, essentially, as it wouldn’t be used for driving on the return trip) in accordance with standard procedures before he left the cab. Usually, at this point, he would have walked down the length of the train to “activate” the cab on the other end of the train (aboard 394M) ahead of the return trip, but since he had some minutes to spare he instead entered the station building, heading to the staff toilets.
The Comeng started to slowly roll away as the driver entered the building, passing the end of the platform at 3kph/1.9mph with the lights on and the doors open. A passenger who had intended to take that train complained to the station master that it had departed early without warning, leading the somewhat confused station master to try and reach the driver on the radio. He failed to get a reply and thus headed into the station building, running into the driver who was just walking out to the platform. Both men realized that the train had departed without permission, and the driver started to run after the train but failed to catch up to it due to its 3 minute head start. He used a trackside phone to report the runaway to the signal box at Broadmeadows station, whose staff in turn made an emergency call to the traffic control center (“Metrol”) at 9:21pm, declaring the emergency. The traffic controllers contacted Broadmeadows’ station master, telling him to monitor the surveillance cameras to provide updates on the train’s location. Most of the stations on the line were unstaffed, being remotely observed by the remaining station masters via CCTV cameras. Thus, the next call went to the station master at Essendon, a station considerably further down the line, who had access to the next set of CCTV cameras. There was no way for the traffic control center to get live updates on the train’s position directly.
The Comeng reaches Glenroy station at 9:24pm, just ahead of the line’s steepest downhill section. The dispatcher had set a path and subsequently turned the signals green for the train ahead of the runaway, anticipating its scheduled departure, but there was one exception. A level crossing just a few feet beyond Glenroy station was designed to close its barriers as a train stops at the station, with the station’s exit-signal remaining red until the barriers were down. However, since the Comeng didn’t stop it flew right through the level crossing, without the barriers being down, luckily not hitting any cross-traffic.
Metrol was starting to grasp the size of their problem around that time, and one massive problem stuck out even beyond “runaway train within an urban area”. Nobody knew if there were people on the train. Cellphones weren’t nearly as common as they are today, and the train had been sitting at the platform for several minutes with the doors open, so it wasn’t unlikely that some passengers had decided to await departure in the train rather than stand on the platform next to it. This potential made the usual ways of stopping runaway trains, intentional derailing or, alternatively, even intentional collisions, even less popular choices. Metrol called Electrol, the power supply management center, at 9:26pm, telling them to turn off the overhead wires between Glenroy and Newmarket. They might have acted on the possibility that the train was somehow being driven by its motors, in which case a disabled overhead wire would have stopped the train. They had to shorten the section meant to be shut off when they were informed about a slow commuter train (service 5262) up ahead which they didn’t want to trap in the path of the runaway, but a shortened section was shut off. Unfortunately, gravity doesn’t care much about electricity, so the Comeng continued to accelerate on the steep downhill section ahead of Pascoe Vale, reaching speeds in excess of 100kph/62mph. There were several further level crossings in this area, but they luckily had time to activate and close, even if with shorter than usual warning times. The speed limit in this area was meant to be 80kph/50mph, but this was fortunately not due to sharp turns but due to the spacing of signals, which the train was reliably ignoring anyway.
The Comeng proceeded through the end of the steep decline as it reached Strathmore station, but the following uphill section wasn’t long enough to have the train run out of momentum, leading to it blowing through Essendon station at approximately 75kph/47mph. Metrol’s staff realized that turning of the power had done exactly nothing in helping to stop the train, and since they were now also out of any uphill tracks they had to find a solution as to what to do with the train fast. They still couldn’t tell if anyone was onboard the train, not knowing that the passenger who had complained about its early departure had been both the first passenger to arrive for the trip and probably the luckiest passenger to ever miss a train.
By this point there were two main options as to what to do with the train. One was to have it pass through Spencer Street station on the suburban through-tracks, from where it would have continued further towards the downtown area. There was a lot of trackage there (points and curves) only meant for low speeds, but if it would manage that it would likely eventually run out of momentum on the level, sometimes slightly uphill track. That route involved having to guide the train through the busy Flinders Street station, the unlikeliness of managing that was probably why that option was only briefly considered.
The other option was to have it enter Spencer Street station and direct it to the unelectrified Country Train platforms, which were dead-end tracks. A guaranteed stop to the runaway train’s journey, but also a guaranteed crash. Metrol still chose that option, calling up Spencer Street 1, the signal box in charge of the station. The signalman told them that he would only have an empty platform track available once the Country Train to Adelaide leaves, and even then it would “probably end up hitting the end of the track”. This concern was met with a statement by the Metrol worker along the lines of “that’s the point”. The signalman still accepted the plan he was presented with, not that he had much of a choice. It does appear that the people involved never quite grasped the speed of the train and/or the severity of the situation, with Spencer Street’s signalman saying they wouldn’t have an empty track until after the next scheduled departure instead of going to move a train out of one of the platform tracks due to this being an emergency. The phone call ended when the runaway train was about 7 minutes from Spencer Street station, not a lot of time but enough to move a train to the storage tracks and maybe even evacuate it first.
Another problematic situation was also developing due to the train’s higher than expected and widely underestimated speed, in that it was starting to catch up to the slower train 5262 up ahead. That train was another three-car Comeng unit, carrying 40 passengers. It had originally been meant to go through the East Suburban track at North Melbourne station (see sketch above), which was switched to the Main Suburban track as the runaway, which was meant to use the East-track, was closing in. The old path was cancelled at 9:29pm, but a safety-system meant that a new path couldn’t be set for 90 seconds. This had been programmed into the system to avoid accidents from points being shifted right in front or even underneath a moving train. Train 5262 thus came to a stop at a red signal just past Kensington Mill station, where its line merged into the tracks entering North Melbourne station.
The signalman at Kensington station, located about 1km/0.6mi behind the train, contacted Metrol right then, saying the runaway had just gone through his station “like a tyrant”. The 90 second blockade on 5262’s signal expired just then, and the driver, as he started to pull away from the signal, spotted the train coming around the corner behind him at high speed, on his track. Seeing the headlight-less train coming out of the dark must have been a terrifying sight. The points ahead of 5262 had a 40kph/ speed limit for turning, but the driver still applied full throttle and made it through them at approximately 65kph/mph. The signal behind 5262 switched to red as it was passed, but a Metrol-worker could still operate the points to switch them back to “straight” right in front of the runaway train. The red signal obviously did nothing to stop the train, so the runaway blew right past it and reached the points mere moments after 5262 had passed them. The signaling-system still had the path set for the Main Suburban track, even if the Metrol worker had ordered the points to be switched, and it appeared at first as if the runaway had followed the train in front. Metrol also couldn’t establish a radio contact with 5262’s driver for several agonizing seconds, fearing that they had been too slow and that disaster had struck. However, 5262’s driver then finally responded, reporting his train stopped at North Melbourne’s Main Suburban platform and saying the runaway had just sped past him on the East Suburban track. Later calculations would show that the points had obeyed the Metrol worker’s input and switched to “straight ahead” less than a second before the runaway passed them, getting staggeringly close to a catastrophe.
The signalman at Spencer Street 1 had received no further communication from Metrol as they had been occupied with the near miss at Melbourne North, and he still didn’t really seem to know what to do with the inbound runaway train which was still going at an estimated 80kph/mph.
- Platform 1 held the “Overland” passenger train, meant for a 9:40pm departure to Adelaide.
- Platform 2 held the 9:35pm service to Bacchus Marsh
- Platforms 3 and 5 held passenger trains waiting for a departure early the next morning
- Platforms 4 and 6 held EMUs meant to depart within the next few minutes
- Platform 7 held a row of passenger cars without a locomotive attached
There were also a few empty non-platform dead-end tracks, but those would see the train crash into the concourse beyond the buffer stops at full speed, something that could have catastrophic results of its own. The signalman doesn’t even get to try and solve his own depressing version of the famous trolley problem when he is called by Metrol at 9:31pm, being told that the runaway is “very close” and at high speed. Just how close becomes abundantly clear when it comes into view while he’s still receiving that statement. His response goes on record as follows:
“I’ve got him, I see him, [expletive] hell he’s coming up hard!”
The signalman had just deleted the set paths for upcoming departures in order to decide and set a path for the incoming runaway, but the 90 second block left no time for that. The runaway thus barelled into the station, forced its way through a set of points in the wrong direction and zeroed in on the 9:35pm service to Bacchus Marsh sitting at platform 2. That train was occupied by the driver, his assistant and 16 passengers (all in the two forward cars), as departure was just over 2 minutes away. Alarmingly, nobody at the station knew about the inbound runaway, no alarm had been raised, no evacuation attempted.
The runaway train was less than a minute from reaching the platforms, and Metrol called (apparently) the only person they could get a hold of at the station itself, a Connex-employee on the platform between tracks 9 and 10. This employee, while on the phone with Metrol, saw an empty train slowly roll into track 14. He figured that that must be the runaway he was informed about, so he made no effort to raise an alarm or contact the country-platforms on the other side of the station.
Time was up anyway.
The two men aboard N463 had received no warning ahead of the Comeng train appearing in their path, perhaps not even registering that the electric train was hurling towards them on an unelectrified track. They leapt off the locomotive and someone, perhaps the conductor on the platform, shouted at the passengers to “watch out” a second before, at 9:33pm, the runaway train slammed into the stationary locomotive with a deafening bang. The rigid locomotive, supported by the train behind it, made a rather effective buffer, being shoved back just 22m/ft as the leading end of the Comeng train broke apart on its leading end. It was only after it came to a standstil that it could be confirmed to be empty, while 8 of the 16 passengers aboard the stationary train were injured, 4 of which requiring hospitalization.
Aftermath
Police helicopters flew along the path of the runaway train later in the evening, using onboard thermal imaging cameras to see of the runaway train hadn’t struck anyone on its unsecured, unilluminated journey through Melbourne. They found nothing, so trains were gradually allowed to start running again around 11:00pm.
N463 was lifted off its forward bogie in the accident, but remained structurally intact with limited damage beyond shattered windows on the driver’s cab. In contrast, the Comeng had its driver’s cab crushed with the leading wall being pushed back through the cab and entrance-area all the way to the forward doors.
The first suspicion by investigators, which was also circulated in the media, was a purposeful act of sabotage or even malice, but the police quickly shut those theories down as there wasn’t the slightest sign of either. Someone in the media als found out that the train had had several reported “brake faults” in the weeks leading up to the accident, but all of those ended up being about brakes not releasing rather than not applying.
The accident also came just 3 days after the Waterfall train derailment had claimed 6 lives when a train ran out of control after the driver succumbed to a heart attack as the train derailed, which incapacitated him in the train’s final minutes which left him unable to control the train’s speed. It didn’t sit well with the public when news broke that lackluster safety equipment played a role in both accidents, with the investigation into the accident at Melbourne being also criticized for low transparency and insufficient information of the public about progress and findings.
The Australian Transport Safety Bureau released their report in December 2003, finding the initial cause of the runaway to be the driver’s failure to set the parking brake when leaving the train unattended, which is what allowed the train to start rolling on the slightly uneven platform track. But that alone shouldn’t have allowed the accident to happen, as railways are meant to be set up in a way that keeps a single failure from causing a catastrophic event. And it is true some railways, even in the modern day, fail to fulfill that demand, but the Comeng and the line it was on were actually built with several systems to keep such an even from occurring. So why did all these systems fail to do their job?
Firstly, there is the so-called “driver’s pilot valve”, more commonly referred to as the dead man’s handle or dead man’s switch. Depending on the train it’s a handle or pedal in the driver’s cab that requires regular operation or consistent applied pressure to ensure that a driver is present and alert. An automatic stop is triggered if the driver fails to operate the system at the required interval/in the required way. The system failed to avoid the Waterfall-accident mentioned above because of a design flaw on that train, but at Melbourne it never even got a chance to stop the train as it’s only operational in an “active” cab, that being one that is currently used for driving. It would obviously be quite impractical for the system to require operation in a cab that is not leading the train. The driver of the runaway train had “isolated” the cab he left, deactivating it, telling the train that this cab would not be used to control the train. However, he then went inside the station (and the train started moving) before activating the other cab, meaning both cabs were “isolated”/deactivated and their dead man’s switches were thus both not operational.
Secondly, the trains on the line are equipped with an Automatic Train Stop (ATS). The system consists of a small lever on the front of the train and a moving arm by the side of the tracks at each signal. The trackside arm raises up when a signal is turned red, bringing it into a position where it will be struck by the lever hanging off a train running the signal. Shifting that lever back would then open a valve on the pneumatic system of the train, activating the brakes. But that lever is folded up when a cab is isolated, only folding down beneath the operational cab. So that system, too, was not turned on as the train rolled away unintended.
There really was no technical defect that played a role in the accident, every system train had functioned (or, not functioned as the case may be) as it was meant to. It was also not against regulations to have a train sit at the platform with brakes released, but the driver should have applied one of the brakes available (either the parking brake or the standard brakes used in operation) when he chose to walk away from the train rather than just walking down the length of it to the other cab. Either that or he should have delayed his trip into the station building until after he had activated the other cab so that that cab’s ATS or dead man’s switch at least watched over the train, if no parking brake was applied.
The report also noted that staff downline from the runaway train could have better anticipated the train if they had been informed better about its status and progress, which was inhibited by the lack of a train position sensing system on most of the line. This left only the CCTV-coverage of different stations to track the train, along with sigthings of it by trackside personell, information that then had to be forwarded to Metrol. Better tracking of the train and its speed might have allowed an earlier reaction regarding 5262, for example, possibly avoiding the terrifyingly close near-miss the two trains had at Melbourne North.
That near-miss also occupied the attention of the Metrol-workers at a critical time, further worsening the forwarding of information to Spencer Street 1, cutting down how much time they had to react. An earlier warning to Spencer Street 1, perhaps even with more precise information, would have allowed a warning to be broadcast on Radio Channel 1, a channel received by all local trains, allowing them to evacuate their trains and the platforms.
The report closes with 22 recommendations, the most significant ones being the advice to install a consistent train position sensing system on the line and installing automatic parking brakes on the trains. The latter recommendation was fulfilled by May 2007, with the Transport Minister testifying to a parliamentary committee that all trains operating for Melbourne’s Metro-system now automatically activated their parking brakes if a driver’s cab was isolated. It was also stated to the committee, though, that installation of a train position sensing system wasn’t possible at that point as other parts of the infrastructure had to be modernized first.
The accident, along with the tragedy at Waterfall just days earlier, directly led to the prioritization and subsequent passing of the 2006 Rail Safety Act in the state of Victoria. The act, which was later upgraded in 2010, introduced a list of measures to improve operational safety from requirements for company structures as well as alcohol and drug testing for rail workers to a broad range of sanctions and penalties to be handed out to the private rail service providers if they failed to live up to the requirements. Most important, though, was the introduction of an independent agency to oversee operational safety of trains and buses in the state, separating the oversight from both political and economical influences.
In a bit of a surprising development neither train was sent to the scrapyard after the accident. N463 had suffered relatively minor damage and was soon back to regular service. The type received an overhaul in the 2010s, giving them new traction motors for an increased top speed of 130kph/80mph over the prior 115kph/71mph. Their standard gauge versions were retired from regular service in 2022, but there appears to be no end in sight for the broad gauge versions (like N463) as of April 2024. They might see a transfer into freight operations soon, but total retirement doesn’t appear to be in sight.
The Comeng-train was also repaired and returned to service, despite the severe damage it had suffered in the accident. It looked almost just like any other train of its type, but if one looked at it from above there was a visible seam above the forward doors of 394M. The damage had in part been repaired with spare panels delivered along with the train way back when they were new, and the new panel simply looked different than the rest of the roof after 20+ years of service. The unit continued to serve until being retired in June 2022, finally getting sent to the scrapyard as part of the Comeng-type’s retirement.
_______________________________________________________________
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.
