Sustainable Management of Infrastructure Assets and Networks Under Uncertainty

由于近40%的市政道路状况良好或更差，加拿大在维护其老化的基础设施方面面临着巨大的挑战。加拿大的道路系统价值3300亿美元，是其价值最高的市政资产类别。联邦、省和市政机构越来越依赖路面管理系统，这是一种量化的决策支持工具，以生成可持续的设计、施工和维护战略。在确定具有成本效益和环境可持续的路面管理计划时，需要考虑一些因素，如资产的衰退率、交通模式的演变以及未来财政预算的可用性，这些因素本身就是未知的。尽管存在这种现实，但在实践中发现的主流范式假设了一个已知的未来，这简化了分析，使复杂的问题变得容易处理。然而，这一假设导致选择静态的资源分配政策，使规划者在情况以不可预见的方式演变时缺乏调整决策的能力。这项研究计划旨在通过开发一套概率软件工具来显著改进当前的路面管理实践，加拿大各地的市级和省级机构可以利用这些工具为其道路资产选择强大、适应性强的资源分配策略。具体地说，我建议开发新的软件工具，以增强规划者用于推进大型基础设施项目的当前项目级模型，并推进网络级系统，该系统支持设施网络的分配决策。对于每一种情况，规划者必须在成本、环境影响和基础设施耐久性之间做出平衡的决定。由于这些路面管理工具必须能够根据实际情况的大小和复杂性进行计算缩放，因此我们的目标是生成关于替代启发式方法的计算性能和保真度的新知识。我的研究项目已经成功地与美国的三个州交通部(科罗拉多州、明尼苏达州和弗吉尼亚州)合作，创建了概率资产管理工具，使这些机构能够以显著降低的成本实现其目标。这些不同的合作举措表明，纳入不确定性为改善基础设施系统的管理创造了新的机会。我们现在的目标是促进我们对这些概率框架的基本理解，并利用它们来改善各种目标和规模的决策。该项目将为本科生和研究生培养未来的工程师和研究人员，负责指导加拿大的基础设施资产。

With nearly 40% of municipal roadways in a fair or worse condition, Canada faces tremendous challenges to maintain its aging infrastructure. Worth $330 billion, Canada's roadway systems is its highest valued municipal asset class. Federal, provincial, and municipal agencies increasingly rely on pavement management systems, which are quantitative decision-support tools, to generate sustainable design, construction, and maintenance strategies. The determination of a cost-effective and environmentally-sustainable pavement management plan requires the consideration of factors such as an asset's rate of deterioration, the evolution of traffic patterns, and availability of future fiscal budgets that are inherently unknown. Despite this reality, the prevailing paradigms found in practice assume a known future which simplifies the analysis and makes a complex problem tractable. However, this assumption leads to the selection of static resource allocation policies, leaving planners ill-equipped to adapt their decisions when circumstances evolve in unforeseen ways. This research program aims to dramatically enhance current pavement management practices by developing a suite of probabilistic software tools that municipal and provincial agencies across Canada can leverage to select robust, adaptable resource-allocation policies for their roadway assets. Specifically, I propose to develop novel software tools that enhance the current project-level models that planners use to move forward with large-scale infrastructure projects, and to advance network-level systems, which support allocation decisions across a network of facilities. For each of these contexts, planners must make decisions that balance between cost, environmental impact, and infrastructure durability. Because these pavement management tools must be able to scale computationally for the size and complexity of situations found in practice, we aim to generate new knowledge on the computational performance and fidelity of alternative heuristic approaches. My research program has successfully collaborated with three state departments of transportation in the United States (Colorado, Minnesota, and Virginia) to create probabilistic asset management tools that enable the agencies to achieve their objectives at significantly reduced costs. These different collaborative initiatives have illustrated that the incorporation of uncertainty creates new opportunities to improve the management of infrastructure systems. We now aim to advance our basic understanding of these probabilistic frameworks and leverage them to improve decision-making across various objectives and scales. This project will prepare undergraduate and graduate students to be the future engineers and researchers charged with directing Canada's infrastructure assets.