Which plants meet the extra demand?
Virginia’s data centers drew 33.9 TWh of electricity in 2023. Researchers modeled that year’s grid with and without data center demand. Some of Virginia’s extra load fell to generators outside the state on its interconnected grid. In Texas, where data center demand was 21.8 TWh, most of the additional power came from within the state. To trace the resulting pollution, start with which power plants responded to the demand.[1]
Finding the generators is the start of the journey. In the model, emissions from Midwestern coal plants that took up the extra demand helped form fine particles in ammonia-rich air. Winds carried pollution away from its source. The study estimated high health burdens per person in Nebraska and Iowa and elevated rates in Kansas and South Dakota, where data center demand was relatively low. Electricity and pollution travel along routes that a map of server buildings cannot capture.[1]
What a siting decision needs to measure
The researchers estimated 559–740 premature deaths a year from air pollution associated with grid-supplied demand. The interval captures uncertainty in the relationship between exposure and mortality risk. Uncertainties in plant emissions and atmospheric modeling sit outside that range. The atmospheric simulation was built to see regional patterns. Exposure in a neighborhood next to a generating station calls for finer measurements.[1]
When assessing a site, ask the grid operator which power plants ramp up as the new load comes on. Before approving a connection, a grid operator can compare hourly generation schedules with and without the new load. The difference helps identify where to measure air pollution. If a different set of generators meets the extra demand, the same electricity consumption can have different consequences for nearby and downwind communities. A siting decision needs a map that follows the power and the wind.[1]