Rail Safety Investigation Report - Final

Shortly before 2118:171¹ on 3 February 2003 a driverless empty suburban train, numbered as 5264, rolled away from Broadmeadows Station under the influence of gravity and subsequently ran largely downhill for 16.848 kilometres to Spencer Street Station. At about 2133 train 5264 collided with the stationary Bacchus Marsh train 8141 at platform two on Spencer Street Station. The estimated speed at impact with the Bacchus Marsh train was 75kph. The leading car of train 5264 was extensively damaged, as was the locomotive of the Bacchus Marsh train. Both trains were derailed as a result of the impact, the Bacchus Marsh train being forced back some 22 metres in the process. The driver of the Bacchus Marsh train and a V/Line employee jumped clear seconds before impact.

Emergency services, including the police, fire and ambulance services attended the scene. There was no fire, no fuel spill or trapped persons. Eight passengers in the two occupied carriages of the Bacchus Marsh service were injured. Four of the injured persons were treated on site and four were conveyed to local hospitals. None of the injuries were serious.

Train 5264 rolled away from Broadmeadows Station due to a release of brakes as the driver was using the station amenities in the process of changing ends in preparation for the return journey to Melbourne. All passenger doors were open, all carriage saloon lights were illuminated and the passenger indicator display on Broadmeadows platform was displaying the correct information for the intended journey of train 5264. The gradient between Broadmeadows Station and Spencer Street Station is predominantly falling, there being an overall height difference of 116 metres between these two stations. The runaway train reached speeds in excess of 100kph and passed through level crossings and pedestrian crossings well in excess of design speed.

The investigation team has determined that the release of brakes was due to the manner in which the driver's controls were isolated and that the unplanned movement was due to the park brake not being applied.

The investigation team determined that it was not possible for Metrol officers² to control or stop the runaway movement. Metrol officers did not know whether or not there were passengers on the train and this constrained their assessment of options such as seeking to derail the train or route it into a siding. Metrol officers were forced to formulate contingency measures 'on the run'. The efforts of Metrol officers in this regard were severely hampered by the lack of visual indication of the train position and a system of voice communications that is not optimised for contingency broadcasts. For the majority of the journey, Metrol officers were relying on third party information being relayed from station and signalling personnel in the field.

In addition, the speed of the runaway train was significantly in excess of normal scheduled services. Consequently, many of the people involved experienced difficulty in adjusting their mental model of where they expected the train to be between Broadmeadows and its final destination at Spencer Street Station, compared with third party reported sightings.

Train 5264 came within a second of being placed on a collision course with the previous Broadmeadows to Flinders Street Station train, number 5262, in the vicinity of North Melbourne. Train 5262 had between 30 and 40 passengers on board.

The investigation team determined that Metrol officers made a conscious decision to route train 5264 into the unwired precinct of Spencer Street Station. If they had not done so, the train would most likely have continued into the 'heart' of the Melbourne suburban network towards Flinders Street Station.

No advance warning of the impending high-speed arrival of train 5264 was conveyed to railway personnel or members of the public at Spencer Street Station. The investigation team found that a number of factors played an important role in this failure.

The investigation established that neither train maintenance nor track maintenance was a factor in the accident. Safety systems designed to stop trains in the event of unauthorised movement are foot and hand pilot valves (often referred to as dead man's handle and pedal), trackside signals and train stops. These devices are effective only when the driver's controls are activated and a driver is in attendance. In this instance the driver's controls were isolated and no driver was in attendance.

Fatigue and the medical condition of the driver of train 5264 and relevant Metrol and field employees were not factors in the accident.

A number of remedial and positive safety actions have been taken or are under way through the Victorian Department of Infrastructure.

The report's recommendations in section 6.2 relate to:

• engineering an automatic application of the park brake when the driver's controls are isolated;
• reviewing and consolidating procedures for changing ends;
• mandating the application of the park brake when the driver's cab is vacated in all instances;
• auditing of driver's actions, isolating driver's controls;
• voice communications across the Melbourne network to be critically examined;
• visual indications in the form of 'real time' display of train movements at the Metrol control centre;
• general training for Metrol staff on train characteristics;
• maintenance on suburban electric trains;
• updating the 'Emergency Response Plan Trains Division';
• training in relation to the 'Emergency Response Plan Trains Division'; and
• minimising the potential for roster induced fatigue.

¹ This is the time recorded by the POTS transponder located at the Up end of Broadmeadows platform. The POTS system tracks train position and records data via interaction between train and track transponders. For a description of the POTS system see section 3.14.5 of this report.

² Metrol officers - network control officers, train controllers, signallers and so on who staff the metropolitan train control centre.

Download final report [PDF 1 MB]