Facility location optimization with pMP modeling incorporating waiting time prediction function for emergency road services

Abstract

Emergency road services providing on-site rescue for disabled vehicles are essential to maintain personal mobility in society. To improve the service level, the time until recovery completion must be minimized. Location optimization of a service base shop is considered an effective approach for achieving this aim. Various similar problems have been investigated as the p-median problem (pMP), aiming at minimizing the sum or average of travel times between demand points and the nearest facilities. For emergency road services, one must additionally consider unique effects on dispatch preparation time and travel time (together referred to as waiting time) caused by vehicle trouble situations and shop characteristics such as repair capability and business forms. To address this problem, we constructed a pMP model combined with a Bayesian regression-based predictive model for waiting time along with the factors unique to this problem. We applied the proposed method to an actual record of emergency road services from Bridgestone Corporation and attempted to determine the optimal shop locations to deliver services in the shortest time. The results show that the proposed method reduced the waiting time by up to 7% compared with the conventional method considering only travel distance.