Skip to content
Study 2: Beyond 19 GW → Question 1

How Does Concentration Change the Breaking Point?

We split PJM into two zones: Dominion (Northern Virginia) and rest-of-PJM. Dominion gets 70% of data center load but has only ~13% of generation capacity. The zones connect through a 6.6 GW transfer limit (CETL). We ran 200 stochastic draws per DC load level—sampling weather years, gas forced outage rates, demand noise, and import derate.

Question 1 — Simulation + Monte Carlo

Deterministic (2024 profiles)

DC AddedSingle-Node70/30 DOM ZoneProportional (13%)
0 GW0 hrs6 hrs6 hrs
3 GW0 hrs142 hrs14 hrs
5 GW0 hrs374 hrs22 hrs
10 GW0 hrs2,468 hrs64 hrs
Finding 1

PJM's 19 GW system-wide tipping point masks a 10x local problem. At 70/30 concentration in Northern Virginia, the Dominion zone hits 142 hours unserved at just 3 GW of data center load—while the system-wide model shows zero reliability issues.

Stochastic — 200-Draw Monte Carlo

Under Stress, the Answer Gets Worse

The deterministic answer was already bad. The stochastic answer is worse. At 0 GW DC, only 62% of draws meet the LOLE reliability standard. Dominion is already marginal under stress conditions—even without data centers.

Stochastic MC: DOM hours unserved distribution (200 draws, 70/30)
DC AddedDOM MeanDOM P90DOM P95DOM Reliable%Single Reliable%
0 GW25.9 hrs79.7167.062.0%94.0%
3 GW200.5 hrs455.5688.16.0%88.0%
5 GW530.6 hrs988.21,396.50.0%78.0%
7 GW1,122.8 hrs1,858.52,581.30.0%66.5%
10 GW2,755.3 hrs4,642.55,414.30.0%54.0%

200 draws: weather year (2016-2024), gas CCGT EFOR (Beta, mean 7%, max 40%), gas CT EFOR (mean 10%, max 45%), coal EFOR (mean 5%, max 20%), demand noise (Normal, ±3%), import derate (Beta, mean 10%, max 80%). Same calibration as Study 1.

Finding 2

Stochastic analysis reveals the cliff: at 70/30 concentration, Dominion hits 530 mean hours unserved at 5 GW DC load. Distributing load proportionally (13% Dominion share) reduces this to 26 hours—a 20x improvement from spatial awareness alone.

Demand Trend

DOM Demand Trend

Dominion's share of PJM demand grew from 12.1% (2016) to 15.2% (2024)—a 25% increase in 8 years. Almost certainly data centers. The zone that was comfortable a decade ago is now at the edge.

Finding 3

Dominion's share of PJM demand has grown from 12.1% (2016) to 15.2% (2024)—a 26% increase in 8 years. The 70/30 data center concentration would accelerate this to ~30%, pushing a zone designed for 13% of PJM load to carry twice its share.