Hours Unserved vs. Dominion Share
Dominion can handle about 10–20% of DC load, not 70%. Moving from 70% to 20% at 5 GW eliminates 330 hours of unserved energy.
The grid-optimal distribution is 0% in Dominion—all data center load in the rest of PJM. But the practical threshold is 10–20% Dominion share, above which reliability degrades sharply.
Fleet Calibration Sensitivity
DOM CT capacity is calibrated to 5.0 GW (named plants total ~3.6 GW; remainder represents DR and uncounted peakers). We swept CT from 4.0 to 6.0 GW to test sensitivity. The qualitative finding holds: under all calibrations, 70% share is catastrophic.
| DOM CT | Firm Capacity | Mean Hrs (5 GW DC) | P90 Hrs |
|---|---|---|---|
| 4.0 GW | 23.7 GW | 876 | 1,593 |
| 5.0 GW (calibrated) | 24.7 GW | 545 | 1,095 |
| 6.0 GW | 25.7 GW | 313 | 684 |
50-draw stochastic MC per CT value. 3x range in absolute hours, but 0% reliability under all calibrations at 70/30.
DOM combustion-turbine fleet sensitivity: sweeping the local peaker fleet from 4.0 GW to 6.0 GW changes mean unserved hours from 876 to 313 at 5 GW DC — a ~2.8× spread — but the system remains 0% reliable under every calibration at the 70/30 concentration. Firm peaking capacity inside the DOM zone matters, but it does not rescue the siting pattern on its own.
CETL Sensitivity
The entire study hinges on the 6.6 GW transfer limit. How much does it matter?
| CETL | Mean Hrs (5 GW DC) | P90 Hrs | Reliable% | Note |
|---|---|---|---|---|
| 5.0 GW | 1,143 | 2,004 | 0% | |
| 6.6 GW | 544 | 1,094 | 0% | Current |
| 8.0 GW | 242 | 567 | 8% | |
| 9.6 GW | 78 | 213 | 46% | +HVDC |
CETL matters enormously: the HVDC reduces mean hours 7x (544→78). But even with $4.8B in new transmission, only 46% of stochastic scenarios are reliable at 5 GW DC. The deterministic analysis showed 16 hours (looked almost fine); the MC reveals the HVDC still fails in more than half of scenarios.
Transfer capacity (CETL) is the single largest lever. Increasing Dominion CETL from 6.6 GW to 9.6 GW (the HVDC case) cuts mean hours unserved from 544 to 78 — a ~7× reduction — but only reaches 46% reliability under stochastic MC. The deterministic analysis showed 16 hours and looked almost fine; MC reveals the HVDC still fails in more than half of scenarios.
Demand Noise Sensitivity
Demand noise (the uncertainty in how much load Dominion actually sees) has minimal effect on the findings. Varying noise from ±1% to ±7% changes mean hours by less than 10%. The spatial dynamics dominate demand uncertainty.
| Noise | Mean Hrs | P90 Hrs | Reliable% |
|---|---|---|---|
| ±1% | 536 | 997 | 0% |
| ±3% (default) | 545 | 1,095 | 0% |
| ±7% | 592 | 1,301 | 2% |