An adaptive opportunistic maintenance model based on railway track condition prediction

Maintaining a track line in a good condition is a continuous challenge since it has to deal with various track line heterogeneities that contribute to accelerate its degradation. As a result, railway tracks should be inspected regularly to detect geometry faults and to plan maintenance actions in consequence. A maintenance plan that minimizes track maintenance cost is highly desirable by infrastructure managers. This paper presents an adaptive maintenance scheduling based on track condition prediction. The degradation indicator is the standard deviation of the longitudinal (SDL) level that is sampled on every 200m-long track section. Standards define some thresholds on this indicator that correspond to different levels of severity and related penalty costs. From collected data, a degradation model that uses a random coefficient Wiener degradation-based process is built. A probabilistic model to simulate the recovery effect after the maintenance action (tamping) is also used. Based on this degradation and recovery models, a cost model is built to find the optimal time for tamping on a single track section. After that we use a Monte Carlo approach to assess the performance of the cost model for the whole track line, considering both calendar based and adaptive opportunistic tamping actions. The results show that the adaptive opportunistic maintenance strategy has a lower cost per unit of time than the systematic preventive maintenance.