There is no defensible universal “break point.”
With the explicitly assumed 170 GW normal-conditions ceiling, the first hourly crossing occurs between 17 and 25 GW of flat added load across nine weather years. Change the ceiling and the range changes one-for-one.
Every replay year crosses the illustrative ceiling.
At 30 GW of flat incremental load, the model records 20–97 shortage-screen hours depending on the weather year.
Year-by-year trace.
| Weather year | Observed peak after QC | First integer-GW crossing | 30 GW events | 30 GW max gap |
|---|---|---|---|---|
| 2016 | 152.9 GW | 18 GW | 15 | 12.9 GW |
| 2017 | 146.6 GW | 24 GW | 4 | 6.6 GW |
| 2018 | 151.5 GW | 19 GW | 15 | 11.5 GW |
| 2019 | 152.3 GW | 18 GW | 8 | 12.3 GW |
| 2020 | 145.4 GW | 25 GW | 11 | 5.4 GW |
| 2021 | 149.6 GW | 21 GW | 15 | 9.6 GW |
| 2022 | 148.5 GW | 22 GW | 10 | 8.5 GW |
| 2023 | 147.6 GW | 23 GW | 4 | 7.6 GW |
| 2024 | 153.1 GW | 17 GW | 19 | 13.1 GW |
The source snapshot is unchanged. 10 one-hour values are interpolated in memory: six isolated anomalies and four missing values, all listed in results.json.
The assumed ceiling moves the answer almost one-for-one.
| Scenario ceiling | First-crossing range | Median crossing |
|---|---|---|
| 160 GW | 7–15 GW | 11.0 GW |
| 165 GW | 12–20 GW | 16.0 GW |
| 170 GW | 17–25 GW | 21.0 GW |
| 175 GW | 22–30 GW | 26.0 GW |
| 180 GW | 27–35 GW | 31.0 GW |
These are scenario ceilings, not PJM ICAP or UCAP values. The table exists to keep the model's most important assumption visible.
The capacity auction made higher-fidelity “headroom hunting” unnecessary.
PJM's current procurement is already 6,831 MW UCAP below its reliability requirement. The hourly model still matters—but for designing conditions and fallbacks, not for declaring spare accredited capacity.