Effective marshaling and buffering strategies can reduce the likelihood of special dangerous commodity (SDC) cars being involved in a train derailment. The objective of these strategies should be to minimize the probability that an SDC car is located in a potential derailment block, subject to external rail corridor characteristics that affect derailments. A procedure is developed for predicting derailments for different railcar positions in a train, on the basis of the point of derailment and the number of cars involved. The number of cars involved in each derailment is assumed to be a function of the train operating speed, the cause of derailment, and the number of cars following the point of derailment. Canadian rail accident data for the period 1980-1985 are used to calibrate a probabilistic expression of number of cars involved in derailments. The Canadian accident data base is also used to estimate point-of-derailment probabilities for different railcar positions and derailment causes. Alternative marshaling and buffering strategies for SDC railcars are evaluated using a combinatorial approach. The results of this analysis indicate that SDC car derailments can be reduced appreciably by considering the derailment potential of different positions along a train forvarious rail corridor conditions. (A)
Abstract