CAREER: Bridging the Gap between Engineering Simulation and Reality of Home Energy-Efficiency Improvements via Big-Data Analysis

CBET 1652696 PI：邱月明本项目旨在将家庭能效改善的实证分析转化为准确、可推广和可扩展的过程。该研究计划将(1)开发一个大数据驱动的能效因果影响评估框架，并为实现的节能提供可靠的统计证据；(2)研究人类环境生态系统中的各种因素(例如，居住者行为)如何与节能技术相互作用；(3)评估节能基准信息是否可以改变能效的有效性；(4)基于节能的时机，量化对电网、经济激励和环境的影响；以及(5)改进工程节能模拟模型。该教育计划将(1)通过咨询小组吸引一群能源从业者；(2)通过开放获取工具教育公众关于能源效率的知识；(3)通过跨学科研讨会促进相关领域研究人员之间的交流；以及(4)培养具有跨学科思维和有效技能的未来工程师和科学家。关键的科学贡献预计来自大规模建筑能源数据集以及融合了不同学科的知识和进步的新的计算能效评估框架。为了构建有效的基准能源使用，该框架使用了控制组、安装前周期、时变协变量的控制、灵活的固定效应和面板回归。由于智能电表正在成为标准，这个评估框架来得正是时候。该项目使用了来自亚利桑那州凤凰城的综合数据集，其中包括2013年每隔15分钟的客户级别的能源需求数据-目前为48,000个住宅客户，确保了统计上稳健和具有代表性的结果。对客户级能效特征、技术、居住者行为、建筑属性和人口统计数据进行多年的家电饱和调查，再加上拟议的评估框架，应该可以克服现有评估研究中的主要缺陷，包括不适当的基准能源结构、选择偏差和忽略的变量偏差。该项目还应该揭示能源效率的复杂影响异质性。对于环境和经济效益的评估，特定技术影响的日内数据将比以前使用日均数据进行的研究提供更准确的结果。该项目还将完善能效性能的相关工程建模技术。

CBET 1652696 PI: Qiu, YuemingThis project aims to transform empirical analysis of home energy-efficiency improvements into an accurate, generalizable, and scalable process. The research program will (1) develop a big data-driven energy-efficiency causal impact evaluation framework and provide reliable statistical evidence of the realized energy savings; (2) examine how various factors in the human-environmental ecosystem (e.g., occupant behaviors) interact with energy-efficient technologies; (3) evaluate whether an energy-saving benchmarking message can alter the effectiveness of energy efficiency; (4) quantify the impact on power grid, economic incentives, and environment based on timing of energy savings; and (5) refine engineering energy savings simulation modeling. The education program will (1) engage a group of energy practitioners through an advisory group; (2) educate the general public on energy efficiency through an open-access tool; (3) facilitate communication among researchers across related fields through an interdisciplinary workshop; and (4) train future engineers and scientists with an interdisciplinary mindset and effective skills.Key scientific contributions are anticipated to stem from both a large-scale building energy dataset as well as a new computational energy-efficiency evaluation framework that incorporates knowledge and advances from various disciplines. To construct valid baseline energy use, the framework uses control group, pre-installation period, control of time-variant covariates, flexible fixed effects, and panel regressions. This evaluation framework is timely now that smart meters are becoming the norm. The project uses a comprehensive dataset from Phoenix metropolitan Arizona that includes 15 min-interval customer-level energy demand data from 2013-present for 48,000 residential customers, ensuring statistically robust and representative results. Multi-year appliance saturation surveys on customer-level energy efficiency features, technologies, occupant behaviors, building attributes, and demographics, together with the proposed evaluation framework, should overcome key shortcomings in existing evaluation studies, including inappropriate baseline energy construction, selection bias, and omitted variable bias. This project also should uncover the complex impact heterogeneity of energy efficiency. For evaluation of environmental and economic benefits, intraday data of technology-specific impacts will offer more precise results than previous studies using average daily data. The project will also refine relevant engineering-modeling techniques of energy efficiency performance.