A customized genetic algorithm for solving multi-period cross-dock truck scheduling problems. (October 2017)

• Record Type: Journal Article
• Authors: Khalili-Damghani, Kaveh; Tavana, Madjid; Santos-Arteaga, Francisco J.; Ghanbarzad-Dashti, Mahdokht
• Is Part Of: Measurement. Volume 108(2017)
• Page Start: 101
• Page End: 118
• Publication Date: 2017-10
• Subjects: Cross-docking -- Truck scheduling -- Supply chain transportation -- Evolutionary computation -- Genetic algorithm; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2017.05.027
• Languages: English
• ISSNs: 0263-2241

Abstract

Graphical abstract: Highlights: We define a multi-period cross-docking model with due dates and temporary warehouse. Trucks face time windows and a balanced workload during the multiple planning periods. The model minimizes the operational time of cross-docking in all planning periods. A genetic algorithm (GA) is proposed to solve the mixed-integer programming problem. The performance of the GA is compared with that of an exact branch and bound method. Abstract: Cross-docking is a logistics strategy for direct distribution of products from a supplier or manufacturing plant to a customer or retail outlet with little or no handling and storage time. The classical cross-docking models are used to find the optimal inbound/outbound truck schedule that minimizes the total operational time. We propose a new multi-period cross-docking model with multiple products, due dates, variable truck capacities, and temporary warehouse. The problem is formulated as mixed-integer programming and an evolutionary computation approach based on a genetic algorithm (GA) is designed to solve it. The structure of the chromosomes, the operators, and the constraint handling strategy are specifically designed for multi-period problems. Several test instances have been generated to compare the performance of the proposed GA with that of a branch and bound solution procedure. Moreover, a comprehensive statistical analysis is conducted to illustrate the performance efficacy of the proposed GA relative to the branch and bound algorithm. This analysis reveals that the GA provides a substantial decrease in the computational burden when compared to the branch and bound algorithm.