Multi-Storm Monte Carlo Assessment of Geomagnetically Induced Currents in the Virginia Mid-Atlantic Power Grid

Abstract

Severe geomagnetic storms can induce geomagnetically induced currents (GICs) in extra high voltage power lines, which can cause thermal heating in transformers or mis-operation of substation components such as static VAR compensators. While GIC risk has been evaluated at the continental scale, and in some cases privately by power operators, there is a lack of systematic assessment of regional exposure and its dependence on individual storms. Rapid data-center growth in Northern Virginia is increasing demand on the Mid-Atlantic high-voltage grid. This study assesses the intensity and spatial distribution of GIC exposure across 688 extra high-voltage substations in Virginia, Pennsylvania, Maryland, and Delaware during three historic geomagnetic storms. Virginia's substations are prioritized throughout this assessment; the three surrounding states are included so that the interconnected transmission network is represented accurately, without border effects on the currents computed within Virginia. The geoelectric field at the earth's surface was simulated during each of these storms and the ground currents were then estimated for each substation using the Coupled Space Weather Impact Model (C-SWIM) based on the Lehtinen-Pirjola algorithm, using a 1,000-run Monte Carlo ensemble. Across Virginia's substations, the 95th-percentile peak ground current reached 55 A for the Halloween 2003 storm, 88 A for the May 2024 Gannon storm, and 142 A for the March 1989 storm (ensemble range 123–167 A). The largest exposure, however, was limited to one cluster of substations in Chesterfield, Henrico, and Hanover counties around Richmond, Virginia, whose position as the most exposed substations remained relatively constant in all three storms (rank correlation p = 0.985–0.997). To weigh these currents against their consequences, the resulting outages were priced using the Value of Lost Load. A March 1989-class storm could interrupt the order of 1,550 MW of load, close to a tenth of Virginia's electricity demand, at a modeled cost of roughly $186 million per day. Protecting Virginia's grid therefore depends less on storm-by-storm forecasting than on the targeted hardening of the small, fixed set of substations that dominate the risk in every storm.