Dynamic truck and trailer routing problem for last mile distribution in disaster response

Abstract

Purpose The purpose of this paper is to investigate the application of the dynamic vehicle routing problem for last mile distribution during disaster response. The authors explore a model that involves limited heterogeneous vehicles, multiple trips, locations with different accessibilities, uncertain demands, and anticipating new locations that are expected to build responsive last mile distribution systems. Design/methodology/approach The modified simulated annealing algorithm with variable neighborhood search for local search is used to solve the last mile distribution model based on the criterion of total travel time. A dynamic simulator that accommodates new requests from demand nodes and a sample average estimator was added to the framework to deal with the stochastic and dynamicity of the problem. Findings This study illustrates some practical complexities in last mile distribution during disaster response and shows the benefits of flexible vehicle routing by considering stochastic and dynamic situations. Research limitations/implications This study only focuses day-to-day distribution on road/land transportation for distribution, and additional transportation modes need to be considered further. Practical implications The proposed model offers operational insights for government disaster agencies by highlighting the dynamic model concept for supporting relief distribution decisions. The result suggests that different characteristics and complexities of affected areas might require different distribution strategies. Originality/value This study modifies the concept of the truck and trailer routing problem to model locations with different accessibilities while anticipating the information gap for demand size and locations. The results show the importance of flexible distribution systems during a disaster for minimizing the disaster risks.