The Study On Stochastic Closed-Loop Supply Chain Network Design Problem

Closed-loop supply chain network is one of the effective ways to realize sustainable de-velopment strategy.The increasing complexity of closed-loop supply chain requires decision makers to pay their attentions to many uncertain factors such as customers' demand,cost and the number of returned products in the network.At the same time,the existence of these uncertain factors makes enterprises face higher risks in their making-decision processes.Therefore,how to control risks effectively in designing of closed-loop supply chain network is very important.In the existing research,many decision makers consider the design of closed-loop supply chain network based on risk-neutral attitude.Their aims are usually to maximize returns or minimize costs.Nevertheless,a large number of uncertainties make risk-neutral attitude inconsistent with reality.Hence,in order to avoid the risks effectively in the real closed-loop supply chain net-work,this thesis studies the closed-loop supply chain network under stochastic environment by using value-at-risk and conditional value-at-risk respectively.Firstly,a stochastic chance constrained closed-loop supply chain network design model is established under value-at-risk criterion in this thesis.When the joint distribution of stochastic demand and transportation cost is known,the original model is transformed into a computa-tionally tractable optimization model by introducing binary auxiliary variables.Further,when the exact probability distribution is not fully known,random variables are used to characterize uncertain demand,transportation costs and the number of returned products.Based on mean-conditional value-at-risk criterion,a distributionally robust closed-loop supply chain network design model is established,in which decision makers can obtain more flexible supply chain network configuration schemes by using ambiguity sets to characterize impreciseness probabil-ity distributions.In order to solve the proposed distributionally robust optimization models,the dual theories are used to transform the models into deterministic mixed integer linear program-ming models,and the models are solved by commonly used optimization software.Finally,the validity and practicability of the model are further illustrated by numerical experiments and results analysis.The main contributions of this thesis can be summarized as the following four aspects:(1) A stochastic chance constrained closed-loop supply chain network design model under value-at-risk criterion and a distributionally robust optimization model under mean-conditional value-at-risk criterion are built;(2)For the stochastic closed-loop supply chain network model,when the random variables obey the finite joint discrete distribution,the stochastic programming model is transformed into a deterministic model by introducing binary auxiliary variables;(3)For the distributionally robust closed-loop supply chain network model,the uncertain discrete proba-bility distribution is characterized by using Box and Polyhedral ambiguity sets,and the robust counterpart models are transformed into computationally tractable models by dual theories;(4)The validity and practicability of the model are verified by numerical experiments.