Emergency Response Models For Water Supply Systems With Vulnerability Analysis

Water supply system is one of the most critical infrastructure systems,the destruction or disruption of which during incidents would have debilitating impacts on the security,economics and social well-being in urban cities.In recent years,urban cities have subjected to emergencies including natural disasters and man-made incidents more frequently than ever before.Take Shanghai as an example,this city is periodically influenced by natural disasters such as blue-green algae blooms and salt tides,threatening the normal water supply to the millions of residents in this mega-city.Moreover,the urban water supply system generally constitutes a complex network with interconnections,such structure amplifies both its vulnerability to external incidents and the difficulty of quick recovery from incidents.This thesis is motivated by the challenging problem that faced by city emergency managers:designing an effective and efficient response plan particularly for the water supply system in order to promptly restore the disrupted public services when non-routine incidents occur.There are three main challenges in this problem,the first is the interdependencies of water supply systems,the second is the uncertain and dynamic nature of environmental incidents,and the third is the integration of the emergency response problem.All challenges make it difficult for city managers to make accurate,real-time response decisions under emergencies.In this thesis,the interactive water supply system is modeled as a complex network in order to represent the interconnected and dynamic features.The network is generally constituted by reservoirs,water treatment plants,pump stations,tanks,pipelines and the connections with other systems(such as the power system).In order to enable quickresponse to emergencies,decision makers have to make three decisions: determine a set of critical components in the water supply system that guarantee quick restoration under the limitation of rescue capacities,assign the selected tasks to the rescue teams with different capacities,and determine the schedule of each rescue team to realize the timely restoration.Corresponding to these essential processes of quick response,our thesis constitutes three parts of research.Firstly,we execute the vulnerability assessment for the water supply system to pre-identify and pre-select the critical components in the system.Secondly,on the basis of the pre-selection results,we formulate two models of resource(mainly refers to the rescue teams)assignment and rescue scheduling from the perspective of rescue-team managers and that of top city managers(generally refer to the managers who are in charge of the whole rescue process).The two perspectives also correspond to different decision-making scheme.The pre-selection results of the first part-vulnerability assessment-provide a guidance for the rescue team assignment and scheduling,which make the emergency response models easier and quicker to solve.Therefore,this part is the prerequisite study of the following two parts and serves as a fundamental step to investigate the problem of emergency response.The optimal results of these three applications collaboratively contribute to a decision-making framework for the emergency response managers who are in charge of urban water supply systems.Our work is divided into six chapters.Chapter 1 outlines the background,motivation,and organizations of our research.Chapter 2 is a literature review concerning the emergency response problem for the infrastructure systems and the available methods to solve it.In Chapter 3,we develop a multi-criteria vulnerability assessment approach named the stochastic Analytical Network Process-Game Cross Evaluation(ANP-GCE)approach to select the vulnerable components in the water supply system.This approach is able to deal with both uncertainties and subjectivities of experts' judgments under varied scenarios of incidents.It provides decision makers a whole assessment process including 4 steps:(i)developing an ANP criteria framework,(ii)proposing a weight calculation approach by integrating GCE with stochastic ANP,(iii)calculating the weighted vulnerability values,rating and visualizing the vulnerability grades accordingly,and(iv)performing assessment simulations and sensitivity analysis for possible scenarios.A case study of Shanghai water supply system demonstrates the effectiveness of the proposed approaches,which also provides management insights for the improvement of system performances against incidents.Based on the vulnerability assessment in Chapter 3,we formulate two emergency response models standing on the viewpoints of different city managers.In these models,vulnerability assessment is performed to select which tasks/components should be repaired to guarantee the optimal whole network efficiency,and then put the results into the models.In Chapter 4,a two-stage rescue team assignment and scheduling model is developed from the perspective of the rescue teams.In this model,the first-stage problem aims to minimize the total losses,whereas the second-stage problem aims to minimize the makespan of the rescue process.We apply a heuristic algorithm which is in polynomial complexity to solve the second-stage model and employ CPLEX to solve the first-stage model.Moreover,the efficiency and effectiveness of the algorithms are demonstrated by computational results.In Chapter 5,we develop a real-time emergency response model from the perspective of top-city managers,in which the decisions of rescue team assignment and scheduling can be integrated.Reducing the timespan of rescue process and minimizing total losses are two important objectives in this model.In addition,we introduce the time-space network and rolling-horizon approach to capture the dynamic and uncertain features of the emergency response phase in this chapter.To solve the proposed model,an improved Benders decomposition algorithm is developed within a branch-and-cut framework,the efficiency of which is validated by extensive computational tests.Chapter 3,4 and 5 constitute a whole decision-making framework for the urban city managers and emergency managers that could facilitate their formulation of response plans in anticipation of incidents.The contributions of this thesis are summarized as follows.Firstly,we improved the efficiency of decision-making in case of emergencies by integrating vulnerability assessment into the emergency response strategy.Secondly,we develop the emergency response models for water supply systems by considering the uncertain and dynamic features of incidents,which also deal with the interdependencies and dynamics of the water supply network.The optimal solutions of these models afford not only greater resilience of infrastructure systems but also damage mitigation against incidents.Thirdly,we develop efficient algorithms for both emergency response models in order to solve large scale problems quickly in real cases.It is demonstrated that both developed algorithms ensure significant speed-ups compared with the standard branch-and-cut algorithm in CPLEX.Moreover,several management insights for emergency decision makers are provided based on the computational results and sensitivity analysis in this thesis.