Handling uncertainties in vehicle routing problems through data preprocessing

This paper presents a global preprocessing methodology for handling uncertainties in operations management. Beyond theoretical considerations on solution feasibility, the methodology provides practitioners with a Monte Carlo simulation-based framework for effective risk management. The main strength of this methodology is being easily applicable to almost any decision problem. Application field of the paper is a real-life workforce management problem for which we propose several mixed integer formulations as well as dedicated solution algorithms. Extensive numerical tests on real-life instances assess the benefit from preprocessing schemes when performed as recommended by our approach, and thus prove its practical relevance.

► Problem: uncertain task durations in a vehicle routing problem with time windows. ► We propose preprocessing schemes ensuring task feasibility with a targeted risk level. ► Method for tuning the risk level for a globally optimal economic gain. ► Extensive testing on real-life data: such preprocessing is relevant in practice.