Global sensitivity analysis framework for supporting the design of railway systems

Railway designers and operators have the hard task of identifying effective solutions to achieve the established network performance levels trying to optimize as more as possible the allocation of economic and technical resources available. To this aim, a preliminary sensitivity analysis is strongly recommended to understand which design parameters are really worth focusing on, to improve a certain measure of performance, without wasting efforts for intervening on non relevant parameters. In this paper in fact, a variance-based sensitivity analysis framework is introduced for supporting decisional processes of railway system design. In particular such framework uses an “own-built” parallel microscopic simulation model to estimate network performances for all possible value combinations of the design parameters involved in the analysis. A practical application has been then carried out on a real mass transit system to understand how four different performance measures were influenced by four design parameters. Results show that the proposed framework is able to identify for a certain performance the variables which mostly affect it, addressing therefore the users towards the choice of intervention solutions which can lead to more efficient performance improvements.