Strategic Physical Withholding of Renewable Energy Generators
Author(s)
Irvine, Paul M.
DownloadThesis PDF (752.8Kb)
Advisor
Knittel, Christopher
Xie, Le
Terms of use
Metadata
Show full item recordAbstract
Renewable generators may have incentives to strategically withhold energy output in electricity markets, either to exercise market power or to avoid congestion pricing caused by transmission constraints. Although academic work often treats renewables as not downward dispatchable, renewable generators technically can, at least in principle, reduce their output by self-curtailing. This paper shows that a firm with a large, diverse portfolio could find it profit-maximizing to withhold renewables over conventional thermal generators once it accounts for constraints on ramp rates and minimum generation, as well as the costs associated with starting-up generators and the probability of detection on generator type by market monitoring authorities. Long-term forward contracts like pay-as-produced Power Purchase Agreements (PPAs) can blunt incentives to exercise market power by insulating individual generators from wholesale prices; however, since generators under PPAs typically bid into the wholesale market and influence competitive prices, they may actually encourage renewable withholding if contract prices are sufficiently low and the parent firm’s portfolio is exposed to wholesale prices. To screen for renewable withholding, this paper proposes three methods: (1) examining the distribution of aggregate output across export interfaces for suspicious bunching, (2) testing deviations from ex-ante forecasts, and (3) identifying the time intervals where generators encounter model structural changes compared to a benchmark presumed free of withholding. Together, this work prepares academics and regulators to more accurately model the behavior of renewable generators in electricity markets and to screen for potential market abuses.
Date issued
2025-05Department
Massachusetts Institute of Technology. Department of Electrical Engineering and Computer SciencePublisher
Massachusetts Institute of Technology