Pertuis english summary

publié le 18 février 2011

On Friday 9 November 2007, on the single-track line between Grenoble and Marseille, the train TER 17 417 made up of a diesel locomotive and four coaches was travelling from Briançon to Aix-en-Provence. At 8:11 p.m. at mileage point 365,848, in bend with a radius of 490 metres at a speed of 105 kph, between the stations of Manosque and Meyrargues, all four axles of the locomotive were derailed over a break in the outer rail. The front bogies of the first two coaches were also derailed, the other axles remained on the rails.
The vehicles remained in the vertical position and did not tip over below the railway line. There were no injuries amongst the passengers (approximately 150), the driver or the guard.
The track was damaged over a distance of 300 metres by the destruction of 500 twin block concrete sleepers ; the running gear and under-body equipment of the rolling stock were also damaged. The amount of the damage to the railway infrastructure was more than € 300,000 and the operation of the line was interrupted for three days.
The immediate direct cause was the undetected failure of a thermit rail weld in an area of continuous welded rails (CWR).

The following causal factors explain why this failure was not detected :
- this line is not fitted with track circuits.
- cracks under the rail flange are virtually impossible to detect by the ultrasonic rail testing vehicles.
- the detection of a rail failure by drivers is difficult. In this case, several drivers had driven over this break without noticing anything. The track environment in this area, due to the frequent alternation of sections of CWR track and normal rails, does not give drivers a stable aural reference.

Three recommendations were made in the report :
- Since several failures of thermit welds over a limited period of time (3.5 years and 4.5 years), in limited area (5.4 km and 1.7 km) in two areas of continuous welded rails on the Aix-en-Provence/Manosque line, additional investigations should be conducted to map the stability of other welds on these sections of CWR, in the top stretch of tracks in bends.
- Other lines of the national rail network with the same characteristics shall be inspected on the basis of experience feedback on the failures of rails and a weld health report shall be drawn up (lines without track circuit, areas of continuous welded rails on concrete sleepers, section in curves and the top stretch of rails).
- Drivers’ failure to detect the break which appeared after the failure of the weld is worrisome, since it is probable that more than one driver drove over this important failure. To improve drivers’ ability to detect such failures it would be useful to test a catalogue of sounds representing various track faults for various types of traction units, with drivers being trained on driving simulators.