Column Generation Heuristic for a Rich Arc Routing Problem

In this paper we address a real world optimisation problem, the Rail Track Inspection Scheduling Problem (RTISP). This problem consists of scheduling network inspection tasks. The objective is to minimise total deadhead distance. A mixed integer formulation of the problem is presented. A column generation based algorithm is proposed to solve this rich arc routing problem. Its performance is analysed by benchmarking a real world dataset from the French national railway company (SNCF). The efficiency of the algorithm is compared to an enhanced greedy algorithm. Its ability to schedule one year of inspection tasks on a sparse graph with thousand nodes, arcs and edges is assessed. 1998 ACM Subject Classification J.m [COMPUTER APPLICATIONS]: Miscellaneous