Exact resolution of a simultaneous vehicle routing and crew scheduling problem in long-haul transport

TL;DR

This paper develops exact integer linear programming models for a complex long-haul vehicle routing and crew scheduling problem, incorporating flexible truck-driver assignments, multiple transportation modes, and multi-day planning, with extensive computational validation.

Contribution

It introduces three novel ILP formulations and valid inequalities for the simultaneous routing and crew scheduling problem, addressing flexible truck-driver matching and multi-modal transport.

Findings

Formulations outperform existing methods in computational tests

Valid inequalities significantly improve solution efficiency

Models effectively handle multi-day, multi-modal routing scenarios

Abstract

This work focuses on exact methods for a Simultaneous Vehicle Routing and Crew Scheduling Problem in long-haul transport. Pickup-and-delivery requests with time windows must be fullfiled over a multi-day planning horizon. Unlike some classic approaches, the correspondence between trucks and drivers is not fixed and they can be exchanged in some locations and at any time. Drivers can also travel for free as truck passengers or take external taxis for an additional cost. The objective is to minimise the truck and taxi travel costs and the penalties for late deliveries. Routes for trucks and drivers are represented separately as directed paths in certain digraphs and then synchronised in time and space. Three compact Integer Linear Programming formulations are proposed and many families of valid inequalities are described. Extensive computational experiments are conducted on randomly generated instances. The formulations are experimentally compared and the effectiveness of the proposed valid inequalities as cutting planes in a branch-and-cut algorithm is evaluated.