UC Berkeley

This dissertation opens with a discussion of three key transformations of the electricity grid in the 21st century: (1) integration of renewable electricity and electricity storage, (2) growth and increased public popularity of distributed, rather than centralized, electricity supply, and (3) increasing concerns about reliability in modern power systems. These trends broadly motivate a central question that links each chapter of the dissertation: what role can distributed energy play towards providing power locally and mitigating electric system reliability concerns?

My second chapter sets the conceptual stage for the remainder of the dissertation by discussing the microeconomics of power outages. Its general premise is that estimating the benefits of policies that enhance the reliability of the electricity grid is challenging. I outline a microeconomic framework that ties electricity demand to the industry metric meant to encapsulate power outage avoidance benefits, the "value of lost load'' or VoLL. This metric, which I argue is too often presented as a singular point estimate, is used in cost-benefit analyses (CBA) that inform infrastructure investment decisions, and the framework leads to insights about its inherent complexity. I describe the public good properties of electric reliability and review the historical development of VoLL literature, highlighting distinctions of applying the VoLL in CBA versus electricity rationing decision-making. Finally, I outline rarely-implemented revealed preference approaches to VoLL estimation that take advantage of recent adoption of backup power systems like solar-plus-storage and that offer attractive theoretical advantages over traditional estimation methods such as macroeconomic proxies and contingent valuation. I argue that the underlying uncertainty of VoLL estimation suggests that cost-effectiveness analysis should remain a cornerstone of electric reliability decision-making.

My next two chapters then apply different empirical strategies to estimate the VoLL by considering the adoption of backup power technologies. Chapter 3 implements an experimental stated-preference survey. The survey asks whether notification timing of power outages affect consumers willingness-to-pay (WTP) to avoid power outages and compares how ownership of backup power technologies differentiates the results. The setting is the recent major power outages in California, where utilities have begun using targeted demand-side shut offs of the grid to prevent their equipment from sparking wildfires. We find positive and statistically significant WTP to avoid power outages ($86/month or $11/kWh); however, we did not find that advanced notice of a power outage led to statistically significant reduced WTP. Heterogeneity analysis suggested that household wealth is negatively correlated with WTP, though household income was not a strong determinant of our respondents WTP. Customers who owned backup power at the time of the survey reported higher WTP.

Chapter 4 complements the stated-preference survey by describing a revealed preference approach to VoLL estimation. I estimate the causal impact of Pacific Gas & Electric's Power Safety Public Shutoff Program (PSPS) on the adoption of solar and storage technologies. The research design centers around a regression discontinuity that estimates the difference in adoption trends between neighbors who experience differing amounts of power outages and takes advantage of PG&E's differential use of nearby electrical feeder lines to mitigate wildfire risk. The VoLL is then estimated using the incremental costs of such investments and how they compare to a scenario analysis of customer expectations of future power outages. Overall, I calculate a VoLL range of $0.3 - $0.8 per kWh. I conclude this chapter with a discussion of limitations and challenges of applying the mitigation cost framework to estimate the VoLL.

Chapter 5 then aims to understand what the technical limitations are for solar-plus-storage systems to provide residential household electricity during long-duration power outage events in the United States. I develop a methodology to model the performance of behind-the-meter solar-plus-storage systems (PVESS) to provide backup power across a wide range of customer types, geography / climate conditions, and power outage scenarios, considering both whole-building backup and backup of specific critical loads. Such analysis provides new information about the conditions PVESS technologies can be relied on to serve load during long-duration outage events. The results indicate that PVESS can meet a limited set of critical loads that includes refrigeration, nighttime lighting, well-pumps, and internet / computer load for 93% of counties, though this percentage drops to 27% in a critical load scenario that includes heating and cooling. Backup performance is lowest in winter months where electric heat is common (southeast and northwest U.S.) and in summer months in places with large cooling loads (southwest and southeast U.S.). Winter backup performance varies by roughly 20% depending on infiltration rates, while summer performance varies by close to 15% depending on the efficiency of the central air-conditioning system. Differences in temperature set-points correspond to a 40% range in winter backup performance and a 20% range in summer performance. A back-of-the-envelope calculation combines the costs with the performance of PVESS in a set of historical case studies to offer comparisons of what VoLL is needed to justify PVESS investment.

Finally, in chapter 6, I explore concerns that the proliferation of PVESS technologies could lead to widespread grid defection. The cost declines of PVESS technologies have led to suggestions that a non-insignificant number of customers would disconnect from utility service and self-supply electricity. Prior research addressing this issue focused on average electricity tariffs, solar profiles, and demand without considering detailed customer heterogeneity. This chapter fills the gap by analyzing how electricity tariffs that shift cost recovery away from variable charges towards fixed charges influence a customer’s decision to disconnect from utility service. A linear optimization model is developed to size an off-grid PVESS. Technology cost and reliability parameters of the optimization model are then adjusted to calculate a range of off-grid costs. Unique rate structure information for over 2,000 utilities in the United States is then used to compare grid costs to off-grid costs. The results show limited ability for solar-plus-storage systems to economically substitute for grid services. However, 1% of households might disconnect from utility service in a scenario that accounts for future off-grid costs and updated tariff designs. We find that 3% of households in the Southwest and California have private economics that favor defection in this scenario, rising to as much as 7% in Hawaii. Grid defection could increase from 1% to 7% of United States households in a slightly reduced reliability scenario. These results indicate that utilities and regulators seeking to limit rooftop solar adoption by lowering variable charges face a possibility that the corresponding increase in fixed charges could lead to inefficient grid defection, though at relatively low penetration rates.