Measuring the performance of railway track through large-scale trackside sensor deployments

Railway track performance is inherently variable along its length, even on straight plain line track. Current condition monitoring relies on snapshot train borne measurements of track geometry, which may be used to plan route level maintenance. Although this provides a global record of current track quality, it is difficult to infer the local condition of the trackbed and local mechanisms of deterioration that may be occurring. A more important indicator of local trackbed condition may be the trackbed support stiffness as seen by a train. The trackbed support stiffness influences the performance of the track and its variation could potentially be used to predict changes in longer term performance with trafficking by means of vehicle/track interaction models and settlement equations. Trackside measurements may be obtained using track mounted sensors such as accelerometers, geophones, deflectometers, high-speed video cameras or strain gauges. These may be interpreted to determine the trackbed support stiffness. However, many measurement locations are required to determine the variation in trackbed stiffness along a useful length of track. This requires a significant increase in the scale of typical trackside monitoring deployments, potentially generating large volumes of data and requiring a degree of automation for data processing. The availability of small, inexpensive, standalone micro-electromechanical systems (MEMS) accelerometers means that it is now practicable to instrument hundreds of sleeper-ends in a single deployment, covering far greater lengths of track than would be viable with more established and expensive trackside monitoring approaches. This paper shows how MEMS have been used to investigate the performance of longer sections of track. Data processing techniques are described, and insights into the actual variation of track system performance are discussed.