Collaborative Proposal: A Multilayer Capital Budgeting Model for Comparative Analyses of Infrastructure Networks

摘要：用于基础设施网络比较分析的多层资本预算模型本项目研究由交通网、水网、能源网、电信网、金融网和通用数据流网络组成的大规模多层基础设施网络的动态资本预算问题。具体而言，该项目的目标是开发一个多层动态网络资本预算模型，该模型可用于量化基础设施网络协调规划和设计可能带来的成本节约和效率提高。过去，人们只是孤立地考虑基础设施网络。因此，该项目是发展基础设施网络设计综合理论和新一代基础设施决策支持系统的第一步，这些系统能够识别和促进各个网络层之间的协同作用。反映物理、金融、经济和信息相互依赖的约束被用来耦合网络层。网络活动进一步受到流量守恒、资源和非负性约束的约束。状态动力学识别活跃在不同网络层上的个体代理的不同博弈行为，因此可以考虑关于网络上完全和不完全经济竞争性质的不同假设。目标是使净经济效益的现值最大化。这一目标与前面提到的约束和状态动态相结合，创造了一系列被表述为最优控制模型的微分博弈。这些模型可用于确定基础设施资本投资在时间和空间上的最有效配置。在这个项目中使用的研究方法结合了定性分析和非传统的解决方案技术。即，既考虑了经典数值方法，也考虑了组合优化/基于agent的仿真（ABS）模型。特别地，运用微分对策模型对ABS模型进行了验证。这方面的一项关键任务是研究模型动力学对参数值的敏感性。虽然实际的城市基础设施系统的案例研究超出了本初步研究的范围，但我们还将开发一个实验设计来验证优化/ABS模型。我们还将描述如何使用该模型为中等规模城市的基础设施系统制定最佳容量扩展计划。

ABSTRACTProposal: CMS-0116107PI: Terry FrieszInstitution: George Mason UniversityDate: July 25, 2001Abstract: A Multilayer Capital Budgeting Model for Comparative analyses of Infrastructure NetworksThis project addresses the dynamic capital budgeting problem for large scale, multi-layer infrastructure networks, comprised of transportation networks, water networks, energy networks, telecommunications networks, financial networks, and genera data flow networks.Specifically, the objective of this project is to develop a multi-layer dynamic network capital budgeting model that can be used to quantify the cost savings and efficiency enhancements that might accrue from the coordinated planning and design of infrastructure networks. In the past, infrastructure networks have only been considered in isolation. As such this project is the first step toward developing both a comprehensive theory of infrastructure network design and a new generation of infrastructure decision support systems capable of identifying and promoting synergies among individual network layers.Constraints reflecting physical, financial, economic, and information interdependencies are used to couple network layers. Network activities are further constrained by flow conservation, resource and non-negativity constraints. The state dynamics recognize the alternative gaming behaviors of individual agents active on the various network layers, so that different assumptions regarding the nature of perfect and imperfect economic competition over networks can be considered. The objective is to maximize the present value of net economic benefits. This objective is combined with the aforementioned constraints and state dynamics to create a family of differential games that are formulated as optimal control models. These models can be used to determine the most efficient allocation of infrastructure capital investments over both time and space.The research approach used in this project combines qualitative analysis and nontranditional solution techniques. That is, both classical numerical methods and combined optimization/agent-based simulation (ABS) models are considered. In particular, the differential game model is used to validate the ABS model. A key task in this regard is the investigation of the sensitivity of the model dynamics to parameter values. Although a case study of an actual metropolitan infrastructure system is beyond the scope of this initial research, we will also develop an experimental design for validation of the optimization/ABS model. We will also describe how the model may be used to develop an optimal capacity expansion plan for the infrastructure systems of a medium size city.