PJM's grid operators had 162,401 MW of demand staring them down on June 20, and their planning models hadn't seen it coming. That peak—nearly 4% above summer projections—forced the Eastern Interconnection's largest grid operator to declare an Energy Emergency Alert Level 1 before July even arrived. The RTO avoided blackouts. But the margin between managed stress and cascading failure was thinner than anyone in the control room would have liked.
For the executives, lawyers, and policymakers tracking FERC's extreme weather dockets, the June event delivers something more valuable than another crisis averted: hard data showing precisely where planning assumptions diverge from operational reality.
The Gap Between Paper and Performance
On paper, PJM entered summer 2025 with an 18.5% reserve margin—comfortable by historical standards. During the June heat wave, actual reserves collapsed to single digits on consecutive afternoons. The culprit wasn't demand forecasting alone. Generators simply didn't show up.
Forced outage rates across PJM's thermal fleet ran 2.3 percentage points higher than the reference case underpinning the 2024 Resource Adequacy Study. For natural gas combined-cycle units specifically, outage rates exceeded 8%—against a planning assumption of 5.8%. That delta represents thousands of megawatts that existed in capacity filings but vanished when the grid needed them most.
The 162,401 MW Stress Test
The 162,401 MW Stress Test. Source: PJM Operations. June 20 demand exceeded summer projections by nearly 4%, triggering the first Energy Emergency Alert before July.
The numbers carry weight beyond one week's operations. FERC Docket AD25-7, examining extreme weather preparedness across RTOs, now has fresh ammunition for intervenors arguing that current accreditation methods systematically overstate thermal generator availability during stress conditions. The June data doesn't suggest the planning construct is imperfect. It demonstrates the construct is broken.
What Actually Held the Line
Three factors prevented the EEA1 from escalating to load shedding—and each carries implications for market design.
First, demand response. PJM dispatched 4,200 MW of load management resources, and they cleared within 45 minutes of the first signal. That activation speed compares favorably to ERCOT's February 2021 debacle, where demand response lagged by hours. The June event validates years of investment in automated DR systems, but it also reveals how close PJM came to needing that full 4,200 MW buffer.
Planning Assumptions vs. Reality
Planning Assumptions vs. Reality. Source: PJM 2024 Resource Adequacy Study. Gas plants failed at rates 38% higher than planning models assumed—thousands of phantom megawatts that existed on paper but vanished under stress.
Second, solar. Utility-scale and distributed solar contributed 8,400 MW during the June 20 afternoon peak. Strip that contribution out, and PJM crosses from emergency management into rolling blackouts. The same resource class that critics label "unreliable" delivered when thermal plants couldn't.
Third, batteries. PJM's 3,100 MW storage fleet discharged at 94% of rated capacity during the critical 4-8 PM window. Natural gas plants averaged 87% of rated output during the same hours. Batteries delivered what they promised. The gas fleet fell short. That performance differential isn't a rounding error—it's a market design problem hiding in plain sight.
A Global Pattern Emerges
PJM's June isn't an isolated event. It's the latest data point in a global pattern that grid planners ignore at their peril.
What Saved the Grid
What Saved the Grid. Source: PJM Operations, June 20, 2025. These three resources—often labeled 'unreliable'—prevented the emergency from escalating to rolling blackouts.
India's grid operator faced parallel conditions during May 2025. Peak demand in the Northern Region hit 89 GW against 112 GW of installed capacity—a 26% buffer that looked adequate until coal plant availability cratered to 71% on fuel constraints and deferred maintenance. Delayed monsoon slashed hydro generation 23% below normal. India avoided blackouts only through emergency imports from Bhutan and Nepal.
China's State Grid, facing the 2024 Sichuan drought, took a different approach entirely. Rather than waiting for conditions to deteriorate, grid operators pre-positioned mobile generation assets and activated interprovincial transfer agreements before the crisis peaked. That proactive stance reduced load curtailment duration by an estimated 40%.
The contrast is instructive. Reactive systems survive extreme weather by luck and heroics. Proactive systems build margin before they need it.
Batteries Beat Gas When It Mattered
Batteries Beat Gas When It Mattered. Source: PJM Operations. During the critical 4-8 PM window, batteries delivered 94% of rated capacity while gas plants managed only 87%. The 'unreliable' technology outperformed the 'reliable' one.
The Regulatory Reckoning
CAISO already applies a "stress test" modifier to thermal capacity values—units with poor historical performance during heat events receive reduced credit toward resource adequacy requirements. PJM has resisted similar reforms, defending its capacity performance construct as adequate.
The June data erodes that defense. FERC Docket EL24-99, addressing capacity performance penalties in PJM, will likely see the event featured prominently in briefs arguing for steeper non-performance charges or alternative accreditation methods. PJM's Independent Market Monitor is expected to release a post-event assessment by August 2025, and early signals suggest recommendations for structural changes to the capacity construct.
For asset owners, the implications are immediate. Generation units that failed to perform during the June event face capacity performance penalties potentially exceeding $2,000/MWh for each hour of underperformance. The June data creates quantifiable litigation exposure for thermal generators with poor forced outage records.
What This Sets Up
The June 2025 heat wave revealed something more important than a single reliability close call: it exposed the systematic gap between resources that count toward capacity requirements and resources that actually deliver under stress. Batteries outperformed their ratings. Gas plants underperformed theirs. Solar contribution prevented blackouts. Demand response activated as designed.
That performance data will shape docket outcomes across multiple proceedings over the next 18 months. Watch for PJM's market monitor report in August, intervenor briefs in AD25-7 through fall, and the commission's eventual ruling on thermal accreditation methodology. The June event didn't break the grid. But it may well break the planning constructs that pretended the grid was safer than it actually is.
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.
