Integrating complex human dimensions into residential demand flexibility program design

Abstract

The United States residential buildings sector with its over 135 million customers contributes 48% of peak energy demand and about 1 billion tons of carbon emissions. Peak energy demand reduction via demand flexibility (DF) is critical to increasing grid-integrated renewable energy and reducing carbon emissions. Yet, DF does not meet its peak demand reduction goal by over 60% due to low participation rates (<8% in the residential sector) and high attrition rates (>30% override rates). Complex diverse human behavior and evolving comfort priorities, especially with climate change, further complicate predicting peak energy demand and DF adoption rates. Anticipating and predicting DF overrides and accounting for different customer types in peak demand estimates are also critical to successful DF program design. Existing residential occupant behavior and building energy models (BEM) relevant to peak demand predictions and DF technology prioritization do not suffice in reducing peak demand uncertainties, understanding how occupants interact with DF, or predicting how different residential customers might adopt DF. This dissertation fills these gaps by delineating the key factors that impact what people do to drive peak demand (behavior). It also answers why people override signals from DF programs (interactions) and how these might affect DF estimates (modeling) based on population demographics. The methods in this dissertation combine econometric/statistical and BEM approaches via supervised machine learning models, statistical significance tests, big data analytics, and large-scale physics-based building stock energy model development to create new DF-relevant datasets. For instance, low-income groups may refrain from DF enrollment due to price shocks (up to 8 times greater peak energy burdens relative to high-income groups) if a uniform price-based demand flexibility scheme is signaled. The overarching goal is to influence robust DF program designs that can fully integrate clean energy grid solutions while including underserved communities in the United States.