Integrated Optimization Of Production And Distribution In Closed-Loop Food Supply Chain With RTIs

Closed-loop supply chain(CLSC),as an important branch of supply chain,has received increasing attention in recent decades.However,CLSC for perishable food products(Closedloop food supply chain,CLFSC)that is more complex than classic CLSC due to food characteristics has been seldom studied.When returnable transport items(RTIs)are introduced into CLFSC(CLFSC-RTI),its complexity increases on account of the interdependencies between the perishable food and the RTIs that are used to ship them.Therefore,in spite of its growing applications in practice,CLFSC-RTI has been far less studied.This thesis aims to develop new models and methods for the integrated optimization of closed-loop food supply chain production and distribution planning problem with RTIs.To this end,three new problems are investigated.Firstly,an integrated optimization of production and distribution in closed-loop food supply chain with RTIs(CLFSC-RTI)is studied.This problem involves a single manufacturer and a single retailer.Dynamic customer demand is considered,product outsourcing is permitted and RTI purchasing budget is limited.Product starts to deteriorate onced produced.The objective is to maximize the total profit of the holistic supply chain.The problem is first formulated as a mixed integer linear program(MILP)and then is proved to be NP-hard.To solve the problem,an improved kernel search-based heuristic is designed to divide the complex original problem into a sequence of sub-problems by identifying decision variables subset.A real case study deriving from a food manufacturer in China shows the applicability of the proposed model and method.Numerical experiments on randomly generated instances demonstrate that the proposed heuristic can yield high-quality solutions with much less computation time compared with the commercial solver CPLEX and an existing promising heuristic named relax-and-fix in literature.Secondly,a bi-objective integrated optimization of production and distribution in closed-loop food supply chain with RTIs(BCLFSC-RTI)is investigated.The considered two conflicting objectives are the maximization of the total profit and the minimization of the carbon emissions in the transportation activities,simultaneously.The problem considered multiple retailers and product selling prices are distinguished according to retailers’ scale and district.For this complex bi-objective problem,a bi-objective MILP is first proposed for its modelling.Then several valid inequalities based on problem analysis are derived to reduce search space for the sake of accelerating resolution procedure.Aiming to obtain its approximated Pareto front,we developed an iterative ε-constraint method to solve it.In each iteration of the ε-constraint method,an improved kernel search-based heuristic is employed to solve the transformed single objective problems.A real case study from a slaughterhouse’s fresh meat supply chain illustrates the proposed method can improve the current strategy for a 7-day planning.And computational results based on various randomly generated instances show that the efficiency of the proposed method is superior to that of a state-of-the-art commercial optimization solver CPLEX while providing good approximation.Finally,an integrated optimization of inventory-routing problem in closed-loop food supply chain with multi-type RTIs(CLFIRP-RTI)is addressed.In this problem,vehicle routing problem is considered and different types of RTIs with distinguished product protective levels are used.An appropriate integer linear program(ILP)is proposed to formulate the problem,and it is proved to be NP-hard.Then,the problem is further extended to a bi-objective one by introducing new parameters and a new objective function that to maximize the total product production level by RTIs.A small-scale case study based on a fresh cherry supply chain derived from the literature are carried out to validate the effectiveness and correctness of the proposed model.Employing equidistant ε-constraint method,we obtain the Pareto front for the case study to provide managerial insights.Finally,a kernel search-based heuristic is developed to solve large-scale problem within reasonable time.Experiments on randomly generated instances are conducted to assess the heuristic’s performance.Computational results demonstrate that the proposed heuristic significantly outperforms CPLEX in terms of computation time while providing high-quality solutions.