Emergency Logistics Network Optimization Considering Secondary Disasters

In recent years,catastrophic disasters or public health incident have been causing sub-stantial casualties and socioeconomic impacts.So we should concentrate more on the emergency rescue problems.However,due to the suddenness and uncertainty of disasters,there are still major shortcomings in actual emergency rescue work.Moreover,real-life cases show that the impact of secondary disasters or even continuous disasters caused by many major disasters may exceed that of primary disasters.Thus,the impacts of second disaster on emergency rescue should not be ignored.Studies show that logistics-related activities account for up to 80%in the emergency rescue process.Therefore,an efficient emergency logistics network(ELN)can speed up the logistics process,and thus alleviate the suffering of victims.However,considering the suddenness,uncertainty,and dynamics of disasters,it is more difficult in emergency logistics network optimization(ELNO).Therefore,how to cope with the impact of uncer-tainties and subsequent disasters on the ELN and establish an ELN with high robustness is an urgent problem to be solved in emergency relief.Based on this,this paper inves-tigates the ELNO problem considering secondary disasters under uncertainty,aiming to improve the response speed and service capability of emergency logistics networks in post-disaster situations.Stochastic optimization,robust optimization and distributionally robust optimization are used to deal with the uncertain parameters with different infor-mation completeness of probability distribution in the ELNO.Moreover,algorithms with accelerated strategies are propose to improve the computing speed.The main contents include the following three aspects:(1)The stochastic optimization of ELN with known information of probability distri-bution of uncertain parameters is studied.A stochastic optimization model considering secondary disasters is constructed,and a conditional probability scenario tree is proposed to describe the uncertain parameters and the relationship between primary and secondary disasters,and on this basis,the pre-disaster facility location and inventory decision,pri-mary disaster and post-secondary disaster resource allocation decisions are optimized.We also extend the problem to a multi-period ELNO model,and makes decisions on facility location,inventory and resource allocation plans based on dynamically updated disaster information.For the scenario-based mixed integer planning model,the Benders decom-position algorithm with effective inequalities,strengthening the cutting plane and other acceleration strategies are proposed.The effectiveness of the modeling method consid-ering subsequent disasters on the service capability of the ELN are verified through nu-merical experiments,and the obvious advantages of the improved algorithm in terms of computational efficiency are also verified.(2)The robust optimization of ELN with unknown information of probability distri-bution of uncertain parameters is studied.A conditional probability scenario tree based on range estimation is proposed to describe uncertain parameters through a combination of random scenarios and uncertain sets,and a robust optimization model based on ran-dom scenarios is established to design emergency logistics networks.On this basis,the model is extended to the robust optimization problem of ELN under multiple cycles for the dynamic nature of disaster situations,and a robust optimization model of multi-cycle ELNO based on prediction-decision making-dynamic adjustment is proposed.According to the characteristics of the model,the progressive hedging algorithm and rolling hori-zon algorithm with corresponding acceleration strategies are applied to solve the model.Numerical studies illustrates that the robustness of the model,and the information-update-based multi-stage model can improve the rescue capability of ELN.(3)The distributionally robust optimization of ELN with incomplete information about the probability distribution of uncertain parameters is studied.The ambiguity set is con-structed by extracting the information related to the probability distribution from the ex-isting uncertain parameter data.A basic distributionally robust model is first proposed,and the model is transformed based on duality theory and a model approximation method based on the linear decision rule to improve its processability.Then the model is extended to a multi-stage optimization model,and the parameters of the ambiguity set are dynam-ically adjusted according to the updated information.Numerical experiments first verify the stability of the solution obtained by distributionally robust optimization model.And the increase in the size of the data related to uncertain parameters can improve the accuracy of the ambiguity set in describing uncertain parameters and thus develop more accurate solutions.In addition,the results show that the multi-stage distributionally robust model outperforms the static decision model in terms of demand satisfaction rate and the total rescue cost,which reflect the performance of the ELN in face of continuous disasters.This paper concentrates on improving the response speed and rescue capacity of the ELN considering secondary disasters.Therefore,we establish a more systematic ELN optimization method,aiming to enrich the theoretical methods of emergency management and provide decision suggestions for emergency management from the practical point of view.It is hoped that this research can provide reference for decision makers to make scientific and rapid ELN optimization decisions.