The PJM Interconnection is evaluating potential changes to interconnection reliability requirements, including existing and future ride-through standards and practices for Large Loads, specifically computational loads such as data centers and crypto-mining facilities, in the wake of a July 22 event that saw nearly 4,000 MW of data center load trip offline unexpectedly in Virginia. It was the largest such event in PJM history.
The grid did not experience significant reliability impacts but did experience a high imbalance between generation and load and large swings in voltage and frequency. PJM and Dominion operators quickly took controlling actions, as data centers disconnected from the grid and switched to backup generation in reaction to a system fault, PJM said on Aug. 11.
Event Timeline
At the Aug. 6 meeting of PJM’s Operating Committee, PJM and Dominion discussed preliminary findings from the event, system impacts and work underway to develop standards to reliably integrate the operation of such computational loads on the bulk power system.
According to Dominion’s preliminary investigation, at 7:56 a.m. that Wednesday, a mechanical failure caused a 230 kV line to be automatically removed from service in Northern Virginia. As the fault was properly removed from the system, data centers in the Dominion zone unexpectedly disconnected from the grid. This caused spikes in voltage and system frequency.
The North American Electricity Reliability Corporation, or NERC, defines a fault as “an event occurring on an electric system such as a short circuit, a broken wire, or an intermittent connection.”
Upon the line’s failure, the first load drop of 2,970 MW took place when data centers started to transfer away, and system voltage increased, said Brian Starling, Director, System Operations Center for Dominion Energy.
When these first Large Loads transferred to backup generation, the loss of their loads prompted high system voltage disturbances. In response, a second wave of 1,099 MW of data centers disconnected and transferred to on-site backup generation. In all, approximately 3,800 MW of Large Load disconnected and transferred without warning, Starling said.
PJM observed overall load drop from 99,984 MW to 96,205 MW. The PJM Control Center quickly dispatched generation down to balance the large change between generation and load, to restore system frequency to 60 Hz and to maintain grid stability.
PJM recovered the Balancing Authority Area Control Error Limit within nine minutes, well within NERC standards of 30 minutes, and dispatched reactive power resources to lower system voltage. Area Control Error (ACE) is a measure of the imbalance between sources of power and uses of power.
Incidents Growing in Size
The July 22 Large Loads transfer was the largest experienced in PJM. Similar events in the Dominion zone caused sudden load transfers of about 1,500 MW each on Feb. 17, 2025, and July 10, 2024.
“This was a normally cleared fault,” Operating Committee Chair Emanuel Bernabeu said. “They should not disconnect from the grid. These data centers are too sensitive to the kind of voltage that they like, and we feel they are disconnecting too early.”
Bernabeu noted that PJM and other grid operators experienced similar ride-through challenges with the proliferation of inverter-based generation resources, mostly wind and solar, whose inverters were similarly set to levels that caused them to disconnect from the grid during minor disturbances. PJM and stakeholders approved voluntary ride-through guidelines for these generation resources in 2019.
Now the system is facing ride-through challenges on the load side. In next steps, PJM and Dominion will continue to jointly review the event and system impacts with an eye to stakeholder coordination for new standards.
“We are currently evaluating potential enhancements to reliability requirements, with consideration of existing and future industry ride-through standards and practices,” said Matthew Wharton, PJM Manager – Reliability Engineering.
FERC Ordered Reliability Standards in Process at NERC
Computational loads on the grid, which include data centers, crypto mining facilities and others, set their protection system to monitor for sudden change or abnormal voltage levels and may rapidly switch to on-site backup generation in response to grid events, even during routine disturbances that are correctly cleared by transmission protection systems.
This can create reliability issues for the grid because unexpected Large Load transfers from the PJM system have two effects on the system. First, system frequency rises above 60 Hz as a result of the imbalance between generation and load. Second, system voltage rises. Spikes in system frequency and voltage are monitored and controlled because they have potential to damage grid infrastructure and/or prompt further disturbances.
NERC has tentatively defined “computational load” as “load comprised of power demand from information technology equipment, such as servers, storage, and networking hardware.” As it stands, computational loads are not subject to NERC Reliability Standards, and computational loads are not a NERC-registered entity.
On July 16, the Federal Energy Regulatory Commission ordered NERC to define and create the first phase of enforceable Reliability Standards needed to manage reliability risks posed by computational loads.
The FERC order stated, “in light of the unprecedented load growth driven by data centers, it is imperative that NERC address the reliability concerns associated with computational loads in a timely manner and with greater certainty than provided by voluntary timelines.”
FERC ordered NERC to submit the first phase of new enforcement provisions by Dec. 31, 2026. However, this first phase, Wharton said, will not address the minimum voltage or frequency ride-through requirements for computational loads. Those requirements are expected to be included in NERC Reliability Standards in 2027.
The work to evaluate interconnection requirements for these Large Load facilities will take place in the Planning Committee because “you really need these requirements in place before the load comes on the system,” said David Souder, PJM Executive Director – System Operations.
