Vehicle and crew scheduling in public transport

Abstract

This thesis examines two scheduling applications arising in public transport: crew scheduling problems where crew members are assigned to trips and vehicle scheduling problems in which vehicles are assigned to trips. It is unlikely that efficient exact algorithms will ever be discovered which can solve large instances of the crew scheduling problems which are studied here. As a result, a large number of heuristic algorithms have been developed. In this thesis a new general purpose heuristic algorithm is presented. This algorithm is based upon repeated matching and can be used to solve crew scheduling problems which are modelled as set partitioning problems. The Repeated Matching Algorithm can also be used to obtain heuristic solutions to a large class of set partitioning problems. As a result of the need to solve many large matching problems, this thesis also studies algorithms for assignment and matching problems. New initialization strategies are proposed for the assignment algorithm used, and two new matching algorithms are presented. The general public transport vehicle scheduling problem with multiple depots has been investigated by many researchers. Though most proposed algorithms are heuristic, a few algorithms which can exactly solve small instances have been developed. In this thesis a new exact algorithm for public transport vehicle scheduling is presented. This algorithm is applied to both bus and locomotive scheduling problems. Exact solutions for problems far larger than any which have previously been solved exactly are also presented. Finally, the problem of combined vehicle and crew scheduling is considered. The practical constraints of a real world problem are discussed and a possible procedure for solving such problems is outlined.