Resource-Oriented Augmentation of a Train Timetable

In this paper, we present a practical approach for the integrated problem of train timetabling and rolling stock scheduling, namely the Resource-oriented Augmentation of a Train Timetable (RATT). As the integrated problem is known to be challenging for real-world instances, the RATT approach is a possible alternative to handle the problem, especially considering the Korean high-speed railway circumstances. In our RATT approach, we extend an existing timetable via train augmentation under the given number of rolling stocks and the managerial constraints of a railway company. To this end, we develop three approaches to solve a real-world problem from a Korean high-speed railway company. First, a generic mixed integer programming model based on the network flow structure is formulated. Second, a train generation technique is adopted to the generic model to enhance the model solvability. Finally, an iterative decomposition scheme is tested to handle large scale problems. Through computational experiments, we show that the train generation technique is essential for the RATT approach, and the decomposition scheme should be considered for large scale problems.