3-Phase heuristics for capacitated multiple-depot vehicle routing problem with separate backhaul and linehaul with a case study on corn residue management system. (August 2021)

• Record Type: Journal Article
• Authors: Phonin, Sirilak; Likasiri, Chulin
• Is Part Of: Computers & industrial engineering. Volume 158(2021)
• Publication Date: 2021-08
• Subjects: Capacitated vehicle routing problem -- Backhaul and linehaul customers -- Multiple Depots -- Separate routing; Engineering -- Data processing -- Periodicals; Industrial engineering -- Periodicals; 620.00285
• DOI: 10.1016/j.cie.2021.107395
• Languages: English
• ISSNs: 0360-8352

Abstract

Highlights: - Find routing for pickup and delivery customers with depots via 3-Phase heuristics. - Develop heuristics based on greedy algorithm and random procedure. - Compare results with 33 benchmark problems from 5 recent works. - Apply heuristics on northern Thailand's agricultural residue system. Abstract: The focus of this work is routing problems in which backhaul and linehaul customers are grouped to a depot, and where the total traveled distance of all routes is minimized. We propose an improved mathematical model for the multiple-depot capacitated vehicle routing problem. Under this modified model, each depot contains separate travel routes for backhaul and linehaul customers. We also propose a 3-phase heuristic based on greedy algorithm in the first 2 phases and on random search in the third phase to ensure reasonable execution time as well as to obtain a better local optimum. Solutions from the heuristic are compared to the optimal solution of small generated problems with up to 24 nodes. Then we compare the performance of the proposed heuristic with the most recent literature and discover that 3 out of 33 benchmark problems can be executed with better solutions. There are 10 problems whose objective function values are the same as the best solutions of the 5 referenced works while the remaining 20 problems are within 3% of those solutions. We also tackle 9 normal distribution benchmark problems with up to 1000 nodes with varying numbers of linehaul customers and depots. We apply the proposed heuristic and compare results with two recent heuristics on these 9 problems. We also solve, using the proposed heuristic, a real-world problem of a corn crop residue transportation system for 16 provinces in northern Thailand encompassing 974 corn fields, 127 co-ops, and 274 factories.