Seasonal Effect on the Optimization of Rail Defect Inspection Frequency

Broken rails are the most common cause of severe freighttrain derailments on American railroads. Reducing the occurrence of broken-rail-caused derailments is an important safety objective for the railroad industry. The current practice is to periodically inspect rails using non-destructive technologies such as ultrasonic inspection. Determining the optimal rail defect inspection frequency is a critical decision in railway infrastructure management. There is a seasonal variation in the occurrence of broken rails that result in train derailments. This paper quantifies the effect of this seasonal variation on the riskbased optimization of rail inspection frequency. This research can be incorporated into a larger framework of broken rail risk management to improve railroad transportation safety. INTRODUCTION Derailments are the most common type of freight-train accidents in the U.S., accounting for 60% of accidents and correspondingly 68% of cars derailed [1]. Broken rails are the most frequent cause of severe derailments [1-5]. The importance of broken rail prevention has been widely recognized [3-11]. The majority of broken rails are caused by fatigue growth of internal rail defects due to cyclic loading by the passage of trains [12]. Ultrasonic inspection is the primary rail defect detection technology used by American railroads. However, no feasible detection technology is capable of detecting all types and sizes of rail defects. Consequently, some defects remain undetected until growing to a critical size where thermal and applied stresses result in a broken rail. Fortunately, the majority of broken rails can be identified by visual inspection or track circuits before the passage of a train [2]. Consequently, only a small percentage of broken rails, approximately on the order of one in 100, result in train derailments [13]. The risk of broken rails and corresponding train derailments may vary by season [2, 14]. However, limited prior research has been conducted to understand the seasonal effects on the optimization of rail inspection frequency. The primary objective of this paper is to identify and quantify this effect and develop season-specific rail inspection schedules to minimize train derailment risk in a cost-efficient manner. To meet this objective, we first review an engineering model to estimate the number of broken rails. Next, we analyze the three-fold seasonal effects in terms of traffic, broken rail occurrence and the corresponding train derailment likelihood. Finally, a Paretooptimality model is developed to optimize the rail defect inspection schedule within different seasons. BROKEN RAIL RISK MODEL The U.S. Department of Transportation (U.S. DOT) Volpe Transportation Systems Center developed an engineering model to estimate the number of broken rails between two successive inspections given inspection interval and rail age [14]: