This paper presents an approach for the estimation of the stochastic inputs for the simulation of railway operations on the basis of real data. The use of stochastic process-times appears particularly relevant on complex networks, where the interactions among trains and their behaviour can vary significantly compared to the planned timetable.
The method was developed to allow the quick preparation and validation of the simulation models for the most important parts of the Italian network . It has already been tested on networks with up to 1000 trains/day before becoming widespread for assessing the impact of infrastructure and timetable improvements as well as for estimating ex-ante the reliability of timetables. The primary advantage of the approach is the very low effort required to define the data sets on the basis of data collected automatically, as well as the simple generation of realistic simulation scenarios. These characteristics were the key factors for the use of the approach in large scale.
