Performance evaluation of heuristic and ILP-based algorithms for multi-period planning of SBVT-enabled transport networks

Optimization algorithms are a valuable tool for decision makers looking at different technological options and future traffic estimates. In order to quickly and accurately analyse multiple what-if scenarios, these algorithms need to balance solution quality and scalability. In this paper, we look at the specific case of multi-period planning in a transport network featuring SBVTs. We analyse the performance of an algorithm based on two ILPs running in sequence, and its suitability to a multi-period scenario, discussing the challenges of offloading complexity away from the ILPs onto pre or post-processing stages. We also show experimental results on how the use of symmetry breaking constraints can significantly improve the algorithm´s running time in the most complex cases. Finally, the applicability of the ILP-based approach is measured against a simpler sequential heuristic algorithm, showing that specific problem instances can do with simpler methods, while others require a global optimization.