Two-echelon Location-routing Optimization With Delivery And Recovery Based On Customer Clustering

The demographic expansion of many cities around the world has not only continuously promoted the rapid development of city logistics,but also engendered substantial challenges to the optimization and management of urban logistics network.For the past decades,the impact of urban logistics activities on the environment has attracted more attention from governments,companies and researchers.To a large extent,the environment is mainly impacted by the level of pollution generated by freight transportation and low performances in waste products' recycling.In literature,several researchers have studied ways of protecting the environment through network optimization,but either focused on solving logistics facilities' location or vehicle routing problems.Aware that these optimization problems are interdependent,finding solutions to simultaneously solve location and vehicle routing problems can effectively optimize the total operating cost,contribute to environment protection,and above all facilitate more systematic decision-makings.Though several studies exist on the topic,less attention was given to the integrated optimization of a closed-loop supply chain with products recovery objective.In addition,research articles which introduced customer clustering mainly employed the latter to initialize routing solutions,but quite a few of them regarded customers' products preferences as clustering criterion for the location-routing optimization.With regard to these research gaps,this paper proposes the Two-Echelon Location-Routing Problem with Delivery and Recovery based on customer clustering(TELRPDR)to explore the economic and environmental efficiency of optimization solutions.Moreover,this research considers customer demand and recovery rates uncertainty to highlight the dynamic nature of modern logistics networks.After reviewing existing research in the field,closed-loop logistics networks are defined and illustrated,and an introduction of the traditional customer clustering procedure is provided.Further,the basic mathematical models for facility location and vehicle routing problems are discussed,and some typical multi-objective optimization approaches are summarized due to the multi-objective nature of the proposed model.Further,the investigated TELRPDR is introduced and mathematical modeling related sets are presented as well as parameters and decision variables.A bi-objective optimization model is proposed to minimize cost and environmental impacts.In fact,one of the main contributions of the designed mathematical model is the comprehensive environmental protection objective function which simultaneously maximizes product recovery rates and minimizes carbon emission.Besides,a customer clustering based Non-dominated Sorting Genetic Algorithm-II(INSGA-II)is designed to conduct local and global optimum searches.Customer clustering and the sweep algorithm are combined to initialize parent chromosomes,partial mapped crossover and swap mutation operators are used to generate offspring populations,and the non-dominated sorting and crowding distance comparison are applied to sort and rank optimization solutions.Subsequently,15 random datasets are compared and proved the performance and complexity of INSGA-II over existing multi-objective evolutionary algorithms(the standard NSGA-II and the multi-objective particle swarm optimization algorithm)on solutions quality and computation time.Finally,a practical experimentation of the model and algorithm is conducted on the location-routing problem of a beverage manufacturing company in Chongqing city.Analyses show that economic and environmental objectives are conflicting,and can cause difficulties to decision-makers in the choice of optimal results.However,the proposed clustering based INSGA-II algorithm alleviates the final decision difficulties.In brief,the investigated TELRPDR can not only reduce total cost and the environmental impact,but also contribute to a sustainable customer relationship management and facilitate multiple echelons networks optimizations.