PJM wants data centres to ride through grid faults instead of dropping off
Victoria Shinder2 min readPJM Interconnection has proposed reliability standards that would require large loads - data centres and crypto-mining facilities in particular - to stay connected through ordinary transmission disturbances rather than disconnecting from the grid. The proposal was introduced at PJM's Planning Committee on 8 September, and the operator says it plans to file solutions with the Federal Energy Regulatory Commission in November.
The trigger was an event on 22 July, when nearly 4,000 MW of data-centre load in northern Virginia disconnected unexpectedly and switched to on-site backup generation. PJM and Dominion operators had to respond quickly to imbalances between load and generation and to spikes in transmission voltage and system frequency. According to David Souder, PJM's executive director of system operations, the grid did not suffer unresolvable reliability impacts that day, but it was the third measurable event of its kind in two years. "This type of operation is unsustainable and needs to be addressed," he told the committee.
The proposed standards set minimum voltage and frequency ride-through requirements for existing and future large loads: facilities would have to remain connected through normally cleared faults without creating further reliability risks. PJM staff say they will work with existing facilities on a transition plan to adjust their settings. The work also answers a FERC order of 18 June that required PJM to identify gaps in how it evaluates the reliability impacts of large loads; FERC wrote that it was concerned the tariff did not specify ramp-rate or ride-through requirements. PJM notes that MISO, ERCOT, Dominion, AEP and NERC are working on similar requirements.
What it means
Generators have long had ride-through obligations because a power plant that trips during a fault makes the fault worse. Data centres were connected as ordinary customers, but at gigawatt scale they behave like generators in reverse: 4,000 MW vanishing in seconds is a disturbance in its own right, and it is caused by protection settings designed to protect the servers, not the grid.
The standards will shift cost and engineering onto data-centre operators, whose backup systems are set to transfer at the first sign of trouble. How strict the thresholds are, and how quickly existing sites must comply, are the questions to watch when PJM files in November.