A Combined Constraint-Based Search Method for Single-Track Railway Scheduling Problem

In this paper we present a three-phase structrue algorithm devised within a constraint programming paradigm to solve real-life single-track railway scheduling instances of problems. The combination of a hill-climbing, easing the process of finding iteratively improved solutions, and a branch-and-bournd strategies allows us to solve 21 real-life problems in a reasonable time, 19 of them to optimality. In addition, this paper discusses a group of practical constraints, incorporated into the software, that arise in real-life problems to which little attention has hitherto been paid. Results comparing the gain on using this approach on large instances problems are also presented.