Recourse Policies in the Vehicle Routing Problem with Stochastic Demands

Download or read book Recourse Policies in the Vehicle Routing Problem with Stochastic Demands written by Majid Salavati-Khoshghalb and published by . This book was released on 2017 with total page pages. Available in PDF, EPUB and Kindle. Book excerpt: In the field of logistics, many practical problems can be formulated as the vehicle routing problem (VRP). In its broadest picture, the VRP aims at designing a set of vehicle routes to pickup or delivery goods through a set of customers with the minimum costs. In the deterministic VRP, all problem parameters are assumed known beforehand. The VRPs in real-life applications, however, involve various sources of uncertainty. Uncertainty is appeared in several parameters of the VRPs like demands, customer, service or traveling times. The VRPs in which one or more parameters appear to be uncertain are called stochastic VRPs (SVRPs). In this dissertation, we examine vehicle routing problem with stochastic demands (VRPSD). In this variant of SVRPs, the customer demands are only known upon arriving at the customer location and are defined through probability distributions. In this setting, the vehicle executing a planned route may fail to service a customer, whenever the observed demand exceeds the residual capacity of the vehicle. Such occurrences are called route failures; in this case the planned route becomes infeasible. There are two approaches when facing route failures. At the customer where the failure occurred, one can recover routing feasibility by executing back-and-forth trips to the depot to replenish the vehicle capacity and complete the service. In anticipation of route failures, one can perform preventive returns whenever the residual capacity falls below a threshold value. All the extra decisions, which are in the form of return trips to the depot in the VRPSD context, preserving routing feasibility are called recourse actions. To model the VRPSD, a recourse policy, governing the execution of such recourse actions, must be designed. The goal of this dissertation is to develop cost-effective recourse policies, in which the fixed operational conventions can govern the execution of recourse actions. In the first part of this dissertation, we propose a fixed rule-based policy, by which the execution of preventive returns is governed through the preset thresholds. We particularly introduce three volume based policies which consider the vehicle capacity, expected demand of the next customer and the expected demand of the remaining unvisited customers. Then, the integer L-shaped algorithm is redeveloped to solve the VRPSD under the rule-based policy. The contribution with regard to this study has been submitted to the Journal of Transportation Science. In the second part, we propose a hybrid recourse policy, which combines the risk of failure and distances-to-travel into a single recourse rule, governing the execution of recourse actions. We employ a risk measure to control the risk of failure at the next customer. When the risk of failure is neither too high nor too low, we apply a distance measure, which compares the preventive return cost with future failures cost. The contribution with regard to this study has been submitted to the EURO Journal on Transportation and Logistics. In the last part of this dissertation, we develop an exact solution methodology to solve the VRPSD under an optimal restocking policy. The optimal restocking policy derives a set of customer-specific thresholds such that the expected recourse cost is minimized. The contribution with regard to this study will be submitted to the European Journal of Operational Research.