Large-load reliability
Energy GridSeptember 11, 20267 min read

PJM’s Ride-Through Proposal Turns Data-Center UPS Settings Into a Grid Compliance Test

After 3.8 GW of Northern Virginia data-center load transferred to backup generation during a normally cleared line fault, PJM proposed specific voltage, frequency, and power-recovery requirements. The operating question is whether a campus can protect its equipment without exporting a larger disturbance to the grid.

By Nawaz LalaniPublished September 11, 2026
More in Energy
Source trail

4 primary links in this brief

The full citation trail is inside the article so readers can verify the signal.

Read the citations
Topic path

AI Power Demand Guide

Track the recurring load, utility, and grid-readiness signals behind this story.

Open the guide
Daily product

Get the Grid Brief

The email version turns the newest AI power, markets, and infrastructure stories into a shorter morning read.

Subscribe free
At a glance
  • PJM is moving the data-center power-quality debate from a broad warning into an engineering compliance test.
  • The event shows why load disappearance can be as operationally important as load growth.
  • PJM’s proposed rule is unusually useful because it translates the risk into numbers.
Article details
Section
Energy
Read time
7 min read
Data included
PJM’s proposed large-load performance test
Editorial diagram showing a transmission fault, a data center remaining connected through voltage and frequency disturbances, and PJM compliance thresholds for active-power recovery
Image note
PJM’s proposal converts data-center electrical behavior into a measurable grid obligation: remain connected through defined voltage and frequency excursions, then restore at least 90% of pre-disturbance power within two seconds.
Data snapshot

PJM’s proposed large-load performance test

The proposal links facility protection behavior to explicit grid-facing consumption and recovery targets.

Test layerProposed behaviorOperator diligence
FrequencyMaintain active power within 10% of the pre-disturbance level in the ride-through rangeValidate aggregate UPS and protection response
VoltageStay connected across the required envelope, with limited proportional power reductionModel the full campus, including cooling and auxiliaries
RecoveryReturn above 90% of pre-disturbance power within two seconds after voltage normalizesTest recovery timing and coordinated load pickup
CurrentLimit voltage-sag overcurrent to 150% of normal maximum consumptionConfirm equipment and switchgear tolerances
ImplementationTransition existing sites while writing tariff and manual requirementsTrack retrofit, testing, exemption, and compliance terms

Source: PJM LCL Ride-through Requirements presentation, September 8, 2026. Requirements remain proposed.

PJM is moving the data-center power-quality debate from a broad warning into an engineering compliance test. On September 8, the grid operator proposed voltage and frequency ride-through requirements for large computational loads, including rules for how much active power a facility must continue consuming during a disturbance and how quickly it must recover afterward. The proposal follows a July 22 event in Northern Virginia in which roughly 3,800 megawatts of data-center load unexpectedly left the grid and transferred to on-site backup generation after a normally cleared 230-kilovolt line fault.

The event shows why load disappearance can be as operationally important as load growth. Dominion reported an initial transfer of 2,970 MW, followed by another 1,099 MW as the first wave of disconnections pushed voltage higher. PJM system load fell from 99,984 MW to 96,205 MW. Operators dispatched generation down, deployed reactive-power resources, and restored the balancing-area control error within nine minutes. The grid remained stable, but PJM says this was the third measurable event of its kind in two years and the largest it has experienced.

A campus can protect its own equipment and still export a larger disturbance to the grid.

PJM’s proposed rule is unusually useful because it translates the risk into numbers. During specified frequency deviations, a large computational load should remain within 10% of its pre-disturbance active-power consumption. During voltage disturbances, a facility may reduce consumption in parts of the proposed envelope, but after voltage returns between 0.9 and 1.1 per unit it generally must return above 90% of its previous load within two seconds. The proposal also caps overcurrent during a voltage sag at 150% of normal maximum consumption and allows cooling load to trip when voltage falls below 0.35 per unit.

Those thresholds turn what can look like an internal UPS or protection-setting choice into a grid-facing performance obligation. Data centers tune transfer schemes to protect servers, cooling equipment, and power electronics from electrical conditions outside narrow tolerances. But if several gigawatts share similar settings, a routine transmission fault can trigger a synchronized move to backup generation. The campus protects its equipment while the bulk system suddenly inherits excess generation, rising frequency, and rising voltage.

The new diligence layer therefore sits between utility interconnection studies and facility commissioning. A developer will need validated dynamic models for UPS systems, switchgear, cooling loads, backup generation, and control logic; evidence that the full campus can ride through the required envelope; and a process for proving that protection changes do not invalidate the study. A passing component certificate is not enough if interactions across thousands of power supplies cause the aggregate campus to behave differently.

The recovery requirement creates a second design problem. Remaining connected during the fault is only half the obligation; the site also has to bring active power back in a controlled, predictable way. An immediate step change can stress equipment or create another system excursion, while a slow return can leave the grid carrying surplus generation longer than planned. PJM says ramp-rate and load-restoration requirements are a separate workstream targeted for the first quarter of 2027, which means today’s ride-through envelope is the beginning of the operating contract rather than the finished rulebook.

Existing campuses are the hardest part. PJM wants minimum requirements for both existing and future large loads and says it will collaborate with operating facilities on a transition plan. New projects can specify compatible equipment and settings before procurement. An energized campus may face firmware limits, warranty restrictions, coordination with tenants, and physical equipment changes. The transition terms—deadlines, exemptions, testing methods, and responsibility for retrofit costs—will determine how quickly the reliability benefit becomes real.

The timetable matters for buyers and investors. PJM plans an education session on September 15, proposed manual and tariff language review on October 6, and a FERC filing in November. A broader PJM design envelope is scheduled for development and stakeholder review in 2027. Projects entering equipment selection or utility negotiations now should not wait for the final tariff text to ask vendors whether their systems can meet the proposed envelope and produce the required models.

This update is materially different from the earlier Grid Report thesis that large AI loads were becoming power-quality counterparties. The July story identified the emerging contract problem from prior events and standards work. PJM has now supplied a concrete performance proposal and a real 3.8 GW case: stay online across defined excursions, recover consumption on a clock, and prove the campus behaves as modeled. That converts a forecasted diligence risk into an active compliance program.

There are still limits. The September presentation is a proposal, not final tariff language; PJM may revise the curves and transition rules; and national NERC standards remain under development. The July event also ended without unresolvable reliability impacts. But waiting for a blackout would be the wrong threshold. When one correctly cleared line fault can move nearly four gigawatts of load without warning, protection settings are no longer merely a facility preference. They are part of the grid’s reliability architecture.

Sources

PJM Inside Lines, “PJM Proposes Reliability Standards to Manage Large Load Disconnection Events,” published September 10, 2026: https://insidelines.pjm.com/reliability-standards-to-manage-large-load-disconnection-events-proposed-by-pjm/

PJM Interconnection, “PJM LCL Ride-through Requirements: PJM Interconnection Proposal,” presented September 8, 2026: https://www.pjm.com/-/media/DotCom/committees-groups/committees/pc/2026/20260908/20260908-item-05---1-large-load-ride-through---presentation.pdf

PJM Inside Lines, “PJM, Dominion Review Large Load Transfer Event,” published August 11, 2026: https://insidelines.pjm.com/pjm-dominion-review-large-load-transfer-event/

Federal Energy Regulatory Commission, Order Instituting Proceeding Under Section 206 of the Federal Power Act, Docket EL26-67-000, issued June 18, 2026: https://www.pjm.com/-/media/DotCom/documents/ferc/orders/2026/20260618-el26-67-000.pdf

Author and standards

By Nawaz Lalani

The Grid Report is written by Nawaz Lalani and focuses on source-backed coverage of AI infrastructure, grid power demand, automation systems, and market signals.

Related reporting
Get the brief

Follow the signal, not just the headline.

Get the daily Grid brief for source-backed coverage on AI power demand, infrastructure timing, automation, and market signals.