Understanding the Evolution of the U.S. Electricity Grid Taking into Account Uncertainty for Improved Management of Costs and Environmental Impacts

管理电网的演变是一项重要的经济和环境挑战。美国的发电每年产生超过3000亿美元的收入，但却产生了39%的二氧化碳、69%的二氧化硫和超过50%的汞排放。像其他基础设施一样，电网的寿命很长，这加剧了锁定效应：资本投资一旦投入，就会持续几十年，可能会推迟采用先进的新技术。本项目考察了今天的电力基础设施和政策决策对未来电网结构的影响，使用工程和经济建模来理解基础设施决策、电力政策和技术进步之间的复杂关系。这项研究的结果对电网规划者和政策制定者都很有价值，他们都对美国电力系统做出重要的长期决策。这项研究模拟了美国未来的电网基础设施，考虑了电力需求的变化、不断发展的技术和政策干预。国家电网在美国大陆被划分为10个区域，每个区域的建设决策都是在20年内最小化总服务成本的基础上做出的，受到可再生能源投资组合标准等政策要求的限制，以及满足当前和未来负荷的需要。该项目的目标是了解电力均等化成本、碳排放和不同电网建设的公共投资等结果如何在不同的政策情景下发生变化。网格建模是一项复杂的任务，这项工作在两个方面改进了之前的工作。首先，采用改进的搜索方法（遗传算法）对成本最小化和满足可再生能源组合标准进行协同优化。其次，模型的核心包含了不确定性，考虑了电网建设如何依赖于不确定的未来燃料价格、资本成本、补贴/税收政策和电力需求。

Managing the evolution of the electricity grid is a critical economic and environmental challenge. Electricity generation in the U.S. produces more than $300 billion in annual revenue, but accounts for 39 percent of carbon dioxide, 69 percent of sulfur dioxide, and over 50 percent of mercury emissions. Like other infrastructures, the electric grid is long-lived which exacerbates lock-in effects: capital investments, once made, last for decades and may delay the adoption of superior new technologies. This project examines the effects that today's electricity infrastructure and policy decisions will have on the future structure of the electricity grid, using engineering and economic modeling to understand the complex relationships between infrastructure decisions, electricity policies, and technological progress. The results of this research will be valuable to both electricity grid planners and policy makers, both of whom make important long-term decisions about the U.S. electricity system. This research simulates the future grid infrastructure in the U.S., accounting for changes in electricity demand, evolving technologies, and policy interventions. The national grid is divided into 10 regions in the continental U.S., with build-out decisions in each region made by minimizing total cost of service over a 20-year horizon, constrained by policy requirements such as the Renewable Portfolio Standards and the need to meet current and future load. The goal of the project is to understand how outcomes such as levelized cost of electricity, carbon emissions, and public investments for different grid build-outs will change across policy scenarios. Grid modeling is a complex task and this work improves upon earlier efforts in two ways. First, improved search methods (genetic algorithms) are used to co-optimize minimal cost and the meeting of Renewable Portfolio Standards. Second, uncertainty is included at the core of the model, accounting for how grid build-out depends on uncertain future fuel prices, capital costs, subsidy/tax policy, and electricity demand.