Design and protection strategies for critical infrastructure systems and supply chains

基础设施系统和供应链是社会功能的基础，涉及现代生活的方方面面。这些重要系统的例子包括货物和服务分配网络、应急服务、金融服务、运输和电信网络、电网和供水系统等。不幸的是，最近的世界事件表明，即使是最强大的供应链或基础设施系统也可能极易受到不可预见事件的影响。2001年的世贸中心恐怖袭击，2004年12月的海啸，以及2005年袭击美国墨西哥湾沿岸的卡特里娜飓风和丽塔飓风，都是最引人注目的证据，证明我们的关键基础设施在面对自然灾害和恐怖袭击时是多么不稳定和脆弱。其他灾难性较小但较为频繁的事件，如工厂火灾、工业事故、罢工和停电，也可能造成代价高昂的影响，有时甚至是有害的影响，特别是在供应线丧失能力或被摧毁的情况下，供应线提供的是基本服务或货物（如医院、药品、疫苗）。鉴于所有这些风险和危险，在过去几年中，人们更加认识和关注如何提高可能受到潜在外部干扰的基础设施和供应系统的安全和可靠性。针对可能的自然或恐怖主义破坏性行为进行规划是一项巨大的财务和后勤挑战，特别是如果考虑到当今物流系统的规模和复杂性，众多系统元素之间日益增加的相互依赖性，各种威胁和危害，以及保护大量系统组件所涉及的高昂成本。由于通常不可能确保所有资产的安全，因此设计系统方法来识别关键元素，优化关键系统组件的保护，甚至构建固有健壮的新系统是很重要的。优化技术可以在这方面发挥重要作用。在过去的几年中，已经开发了一些优化模型，用于识别供应链配置，这些配置在中断方面是可靠的，并且在基础设施投资方面具有经济效益（设计模型）。此外，最近已经提出了新的模型，可以用来提高已经到位的基础设施系统的可靠性，对于完全重新配置将过于昂贵（保护模型）。该项目的总体目标是在这两个与可靠性相关的研究领域开发一套新的优化模型。特别是，将开发新的建模方法，以克服现有模型所特有的一些过于简化的假设，捕捉系统可靠性规划中产生的额外复杂问题，并反映可能的应用程序设置的多样性。作为一个例子，未来的建模工作应该着眼于合并以下问题的各种组合：(1)系统的不同操作协议（基于成本的vs.基于标准的供应模型）；(2)不同类型的危害（自然灾害的预期模型与有预谋攻击的最坏情况模型）；(3)不同的底层模式（设施选址、网络、多级供应模式）；(4)随机方面，如可能损失的随机数和部件特定的失效概率；(5)不同程度的设施保护（受保护的组件可能完全不受攻击和故障的影响，被破坏的概率较低，或根据保护投资水平保留部分操作能力）；(6)多目标（运营成本、销售损失成本、系统效率、覆盖范围、风险水平）。精确和启发式技术也将被开发来解决新提出的模型。

Infrastructure systems and supply chains are at the basis of societal functions and reach into every aspect of modern life. Examples of these vital systems include goods and service distribution networks, emergency services, financial services, transportation and telecommunication networks, electrical power grids and water supply systems among others. Unfortunately, recent world events have demonstrated that even the strongest supply chains or infrastructure systems can be extremely vulnerable to unforeseen events. The World Trade Centre terrorist attack in 2001, the Tsunami in December 2004, and hurricanes Katrina and Rita, which hit the US gulf coast in 2005, are only the most dramatic evidence of the precariousness and vulnerability of our critical infrastructure to natural disasters and terrorist attacks. Other less catastrophic but more frequent incidents, such as plant fires, industrial accidents, labor strikes and power outages, can also have costly and, sometimes, harmful implications, especially if the incapacitated or destroyed supply line provides essential services or goods (e.g., hospitals, drugs, vaccines).In view of all these risks and hazards, there has been a heightened awareness and concern over the past few years for increasing the security and reliability of infrastructure and supply systems that may be subject to potential external disruptions. Planning against possible disruptive acts of nature or terrorism is an enormous financial and logistical challenge, especially if one considers the scale and complexity of today's logistic systems, the increasing interdependence among numerous system elements, the variety of threats and hazards, and the prohibitive costs involved in securing large numbers of system components. Since it is generally impossible to secure all assets, it is important to devise systematic approaches for identifying critical elements, optimize the protection of key system components and even build new systems which are inherently robust. Optimization techniques can play a significant role in this respect.Over the last few years, a few optimization models have been developed for identifying supply chain configurations that are both reliable with respect to disruptions and economically cost-effective in terms of infrastructure investment (design models). Additionally, new models have been proposed recently which can be used to improve the reliability of infrastructure systems that are already in place and for which a complete reconfiguration would be too costly (protection models).The overall objective of this project is to develop a suite of new optimization models in both of these reliability-related research areas. In particular, new modeling approaches will be developed to overcome some of the overly simplified assumptions which characterize existing models, to capture additional complex issues arising in systems reliability planning, and to reflect the diversity of possible application settings. As an example, future modeling efforts should aim at incorporating various combinations of the following issues: (1) different operational protocols of the systems (cost-based vs. standards-based supply models); (2) different types of hazards (expected models for natural disasters vs. worst-case models for premeditated attacks); (3) different underlying models (facility location, network, multi-echelon supply models); (4) stochastic aspects such as random numbers of possible losses and component-specific failure probabilities; (5) different degrees of facility protection (protected components may be completely immune to attacks and failures, have a lower probability of being disrupted, or preserve part of their operational capabilities depending on the level of protection investment); (6) multiple objectives (operational costs, lost-sales cost, system efficiency, coverage, risk level). Both exact and heuristic techniques will also be developed to solve the newly proposed models.