The capacity market's central promise rests on a simple premise: pay generators to be available, and they'll show up when it matters. Summer 2025 stress-tested that premise across North American grids. The results should trouble anyone responsible for keeping the lights on.
PJM's natural gas combined-cycle fleet posted forced outage rates of 8.1% during June's heat event—a full 2.3 percentage points above the 5.8% reference case baked into capacity planning models. That variance isn't academic. It translates to roughly 4,100 MW of expected capacity that simply failed to materialize when temperatures spiked and demand surged. For context, that's equivalent to two large nuclear plants worth of phantom megawatts.
The timing couldn't be worse. FERC's resource adequacy dockets are active. PJM is mid-stream in capacity market reform proceedings. And this data lands as ammunition for intervenors who've long argued that nameplate capacity bears little relationship to stress-condition performance.
The Physics of Failure
Simple cycle gas turbines—units that exist specifically to provide peaking capacity—performed even worse than combined-cycles, posting 11.2% forced outage rates during the same period. Their underperformance during peak conditions defeats their fundamental purpose in the resource stack.
The Reliability Gap: Expected vs. Actual
The Reliability Gap: Expected vs. Actual. Source: FERC Summer Assessment 2025. CCGT plants failed at rates 40% higher than models predicted—a 4,100 MW gap.
The mechanisms are well understood, if poorly accounted for in planning models. Combustion turbine output falls as air density decreases with rising temperatures. A unit rated at 200 MW may deliver only 175 MW when intake temperatures exceed 95°F. This derating is predictable—yet often excluded from capacity counting.
Cooling water constraints compound the problem. Plants drawing from rivers or lakes face thermal discharge limits when water temperatures rise. Some units must curtail output simply to comply with environmental permits. Coal units showed 7.4% forced outages against planned rates of 5.1%, with fuel handling issues and cooling water limitations driving derates.
The contrast with non-thermal resources is stark. Nuclear plants maintained 2.3% forced outage rates, consistent with historical averages—delivering exactly as expected. Battery storage achieved 1.4% forced outages, though PJM's relatively small storage fleet (3,100 MW) limits statistical confidence.
Summer 2025 Forced Outage Rates by Technology
Summer 2025 Forced Outage Rates by Technology. Source: FERC Summer Assessment 2025. Peakers designed for stress conditions failed 8x more often than batteries.
CAISO Already Moved
California confronted this reality after the August 2020 rolling blackouts and reformed its capacity accreditation methodology accordingly. The new approach assigns capacity value based on Effective Load Carrying Capability calculated through probabilistic modeling—resources receive credit based on their expected contribution during loss-of-load-probability hours, not nameplate ratings.
The recalculation revealed significant discrepancies. Gas peakers previously credited at 100% received ELCC values of 75% to 85%. The methodology better reflects operational reality: resources cannot contribute to reliability if they aren't running during stress conditions.
Other jurisdictions have reached similar conclusions through different paths. The UK reformed its capacity market de-rating methodology in 2023, with Ofgem now applying stress event performance factors. Australia's National Electricity Market uses de-rating factors reflecting both forced outage rates and temperature sensitivity. India's Central Electricity Regulatory Commission applies availability-based tariffs, creating direct financial incentives for reliability improvement.
How ELCC Recalculation Changed Gas Peaker Credits
How ELCC Recalculation Changed Gas Peaker Credits. Source: CAISO post-2020 reforms. Gas peakers previously at 100% credit now receive 75-85% based on actual stress performance.
The Regulatory Window Opens
FERC Docket RM21-17 addresses resource adequacy reforms including accreditation methodology. Summer 2025's performance data strengthens the case for ELCC-based approaches that accreditation reformers have advocated for years.
PJM's pending capacity market filing would increase non-performance penalties for units failing to deliver during emergency conditions—a reform that gains urgency when 4,100 MW of expected capacity evaporates during a heat wave. ISO-NE's parallel reform proceeding shows that current UCAP methodology overstates thermal plant contribution by 8% to 12% during summer peaks.
For generation owners, the financial exposure is immediate. Units that underperformed during summer 2025 face capacity performance penalties potentially exceeding $100/kW. For load-serving entities, the data should inform procurement strategy: resources with strong stress-condition performance provide more reliable coverage than resources with identical nameplate capacity but weaker availability.
What Comes Next
The gap between planning assumptions and operational reality is no longer theoretical—it's quantified, seasonal, and growing. Climate projections suggest extreme heat events will become more frequent, widening the performance gap during precisely the hours that determine system adequacy.
Watch for PJM's accreditation methodology response in early 2026 and FERC action on the resource adequacy docket by mid-year. The commission faces a choice: continue crediting capacity that vanishes under stress, or align market rules with the physics that summer 2025 made impossible to ignore.
About the Author
Dr. Sayonsom Chanda
Dr. Sayonsom Chanda is an electrical engineer and senior scientist with more than a decade of experience in developing AI, ML, and other advanced computing solutions for the electric utility industry in US and India. He is also an energy policy thinker and a published author with more than 20 papers and 1 book.
