Probabilistic analysis of braking performance in railways

To increase safety and efficiency in the management of railway traffic, a new speed control system, named SCMT, is currently being developed by RFI and Trenitalia for the Italian Railways. Other innovative speed supervision systems are being developed in Europe, such as the ETCS/ERTMS, which will also be installed on the new high-speed line Roma-Napoli. All traffic management systems are generally based on a set of supervision curves relating the allowed velocity of the train to the running distance, in order to ensure the respect of speed restrictions on the line by 'soft' or 'hard' intervention such as an acoustic and visual warning to the driver (soft) and/or service or emergency braking (hard) in the case of train velocity exceeding the permitted one. To elaborate this set of supervision curves, the on-board unit requires train deceleration depending on time and speed as basic information about the braking behaviour of the train. The implementation of a speed supervision system requires a preliminary definition of braking models that allow the conversion of the general parameters affecting the braking performances of trains (such as a braked weight percentage, goods/passenger brake position, brake equipment, train length, etc.) into a basic deceleration profile as a function of time, during the deceleration rise phase, and of speed, during fully developed braking. The deceleration used to evaluate braking curves is obtained by applying a proper safety margin to the nominal deceleration value (which depends on train characteristics). In this paper a probabilistic analysis of train deceleration is carried out, starting from probability distributions of parameters affecting the braking. For the major parameters, the probability distribution was determined on the basis of technical knowledge and experimental results. The aim of this work is to determine the probability that the real deceleration is lower than the nominal value multiplied by a given safety margin.