Vehicle sharing and workforce scheduling to perform service tasks at customer sites

Abstract

Most of the research done in the Vehicle Routing Problem (VRP) assumes that each driver is assigned to one and only one vehicle. However, in recent years, research in the VRP has increased its scope to further accommodate more restrictions and real-life features. In this line, vehicle sharing has grown in importance inside large companies with the aim of reducing vehicle emissions. The aim of this thesis is to study different situations where sharing vehicles brings an improvement. Our main study focuses on developing a framework that is capable of assigning different workers to a common vehicle, allowing them to share their journey. We introduce a mathematical programming model that combines the vehicle routing and the scheduling problem with time constraints that allows workers to share vehicles to perform their activities. To deal with bigger instances of the problem an algorithm capable of solving large scenarios needs to be implemented. A multi-phase algorithm is introduced, Phase 1 allows us to solve the non-sharing scheduling/routing problem whose aim is to find the best schedule for workers. Phase 2 will merge the allocated workers into common vehicles when possible, while Phase 3 is the improvement procedure of the algorithm. The algorithm is tested in three different settings; using workers as drivers, hiring dedicated drivers, and allowing workers to walk between jobs when possible. Results show that sharing vehicles is practicable under specific conditions, and it is able to reduce both the number of vehicles and the total distance, without affecting the performance of workers schedule.