Rockfall alarm concept for railway safety

REQUEST

Rockfalls pose a threat to human infrastructure below cliffs. Sensitive and reactive alarm systems are needed for rail traffic safety because small rockfalls (≈ 0.01 m3) impacting the railroad may cause train derailment.

Sercel, a pioneering company in the measurement, monitoring and understanding of the subsoil, has collaborated with the SNCF network teams to respond, by committing to a 4-month sprint, to the need to have in the event of falls of block on the railway track, by minimizing the rate of false positives, a system based on a network of accelerometers should make it possible to locate the block with decametric precision and possibly give an indication of the trajectory of the blocks.

SOLUTION

Sercel has deployed a solution measuring, using a network of seismic sensors, vibrations during falling blocks.

This is with the aim of confirming its already existing detection and localization algorithms, quantifying false positives, verifying the quality of block localization and determining the density of sensors necessary for optimized precision.

The cliff has been monitored by a dense array of MEMS accelerometers (SERCEL, 350 WiNG seismic nodes). Three lines of 82 sensors with 1.5 m inline offset were installed along the 80 m long cliff. Two lines were installed on each side of the single rail track, and the third one was installed between the two steel rails. The sensors were sampling at 1 kHz, allowing for a spectral analysis at higher frequencies, useful for accurate rockfall energy characterization.

A rock fall was simulated and filmed in order to compare the events that took place with the sensor measurements.

CONCLUSION

Rockfalls can be detected by seismic sensors, allowing rockfall alarm systems powered by near real-time seismic processing. We show the high-sensitivity to small rockfalls (0.01 m3) of a dense seismic array. We developed a workflow for near real-time processing of a continuous seismic stream recorded by the array, with a philosophy similar to micro-seismic monitoring. The algorithms rely on few parameters to ease calibration.