Research On The Fixed Partition Problem Of Distribution Customers For Outsouring With Replenishment Mechanism

In China,transportation business expenditure has always occupied more than 50%of the total logistics expenditure of enterprises.The high expenditure of transportation business has a great relationship with the investment of fixed assets.Therefore,many enterprises choose to outsource the transportation business to third parties to reduce this part of the expenditure and the occupation of fixed assets.In order to prevent the disadvantages caused by the complete outsourcing of all transportation businesses to third-party enterprises(the demand is fluctuating,so the quality of service is difficult to guarantee in long-term outsourcing),and to reduce the high fixed assets occupied by self-support logistics,many enterprises adopt the mixed mode of outsourcing and self-support transportation.In view of this situation,fixed partition problem of distribution customers for outsourcing with replenishment mechanism(FPPDCO-RM)is proposed.The FPPDCO-RM fully considers the service consistency and demand fluctuation in the real world.On the basis of the traditional vehicle routing problem(VRP),it adds the fixed partition of distribution customers for outsourcing and self-support replenishment mechanism in the distribution network.Once the outsourcing decision is determined,it will execute a period instead of one day or one trip.The transportation mode of "fixed partition for outsourcing+self-support replenishment"can provide a relatively stable transportation plan for carriers,and reduce the empty-loading ratio for self-support vehicles.For shipping enterprises,establishing long-term and stable cooperative relationship with carriers can maintain the stability of the fleet,ensure the quality of transportation and obtain a relatively favorable price,reduce operating costs,and join some of the self-support vehicles to cope with abnormal situations and fluctuations in demand,and ensure a high level customer service.Service consistency is also an important issue in the distribution problem.The method of fixed partition of distribution customers for outsourcing establishes a fixed relationship between downstream customers and drivers,increases the familiarity for both parties,and brings a better logistics experience for downstream customers.Mathematical model with the objective of minimizing the total transportation costs in a period is established.The FPPDCO-RM is splited into two sub-problems.Since the partitioning and replenishment are mutually influential,embedded two-phase genetic algorithm is designed to solve the FPPDCO-RM problem.The coding and decoding methods,initial solution generation,fitness calculation,descendant selection strategy,crossover mode,mutation mode,combined partitioning mechanism and termination criterion of the algorithm are designed in details.In the test part of the instances,33 instances based on Augerat instance A are constructed.Firstly,two small scale instances are solved by CPLEX,and the obtained results are compared with the best values obtained by embedded two-phase genetic algorithm(ETPGA),which verified the correctness of the proposed algorithm.Secondly,the ETPGA under static partitioning and dynamic partitioning is used to solve the instances,and the effects and stabilities are compared with the combined partitioning method proposed in this thesis.Subsequently,the embedded two-phase genetic algorithm is compared with the results of the two-phase algorithm of "first partitioning and then replenishing",which proves the good performance of the embedded two-phase genetic algorithm designed in this thesis.Furthermore,the effectiveness of the embedded two-phase genetic algorithm is verified by comparing with the results of CPLEX.Finally,the actual case analysis shows that the algorithm proposed in this thesis can greatly improve the scanning method that the enterprise used.Experimental results show that the proposed algorithm can effectively solve the FPPDCO-RM problem,and the correctness of the ETPGA algorithm is verified by the solution of two small scale instances.The static partitioning and combined partitioning method are used to solve the instances respectively and the average value of target function standard deviation percentage is 1.37%(static partitioning)and 1.24%(combined partitioning),respectively.The best value obtained by the latter is better than the former on average 1.88%,and the maximum improvement is 10.32%.Compared with dynamic partitioning,the average improvement is 0.36%,the maximum improvement is 10.66%,and the target average value of function standard deviation percentage is 1.96%(dynamic partitioning)and 1.24%(combined partitioning).In terms of solution stability,the combined partitioning method improves the average by 36.73%compared with the dynamic partitioning method.Compared with the two-phase algorithm of "first partitioning and then replenishing",it proves that there is an intrinsic relationship between partition and replenishment,when the outsourcing cost is minimized,the total cost is not necessarily the smallest.Compared with the two-phase algorithm of "first partitioning and then replenishing",the algorithm has improved by 1.86%on average and the maximum improvement is 18.14%.Using CPLEX to solve the small and medium scale instances,the average difference between the optimal value obtained by CPLEX and the best value obtained by the embedded two-phase genetic algorithm is 0.24%,among these instances,there are 22 instances,the best value obtained by the ETPGA algorithm is equal to the optimal value obtained by CPLEX,and the effectiveness of the algorithm is verified.The actual case analysis shows that the distribution method of "fixed partition for outsourcing+replenisment" can reduce the fixed assets occupation for enterprises.By comparing with the enterprise partitioning method,it is proved that the proposed algorithm has a greatly improvements,the total transportation costs for one week and half a month has decreased by 31.46%and 30.61%respectively.