Dynamic period routing for a complex real-world system: A case study in storm drain maintenance

Abstract

This paper presents a case study of a real world storm drain maintenance problem where we must construct daily routes for a maintenance vehicle while considering the dynamic condition and social value of drains. To represent our problem, a dynamic period vehicle routing problem with profit (DPVRPP) model is proposed. This differs from the classical period routing problem in a number of ways. Firstly, it is dynamic: during the planning horizon, the demands from damaged drains and residents reports arrive continuously. In addition, the drains condition is changing over time. Secondly, our objective is maximizing the profit, defined here as the drains condition with respect to its social value. This study is based on large-scale data provided by Gaist Solutions Ltd. and the council of a UK town (Blackpool). We propose an adaptive planning heuristic (APH) that produces daily routes based on our model and an estimation of changing drain condition in the future. Computational results show that the APH approach can, within reasonable CPU time, produce much higher quality solutions than the scheduling strategy currently implemented by Blackpool council.