PJM's New Ride-Through Rules: Why Data Centers Must Stop Tripping Off the Grid

A large data center disconnecting from the grid may protect its own servers: but it can create a much larger reliability problem. PJM’s proposed ride-through requirements respond to a growing concern: large computational loads are tripping offline during normally cleared transmission disturbances, shifting suddenly to backup generation and creating sharp imbalances between load and generation.

The issue became impossible to ignore after a July 22 event in northern Virginia, when approximately 3,800 MW of data center load disconnected from the PJM system. PJM described the event as the largest data-center load loss of its kind in the organization’s history. On September 8, PJM introduced proposed interconnection reliability standards for large computational loads, including data centers and cryptocurrency-mining facilities. The proposal is now moving through stakeholder review, with a PJM filing to the Federal Energy Regulatory Commission (FERC) planned for November 2026.

What PJM Is Proposing

PJM’s proposal would require covered large loads to remain connected during normally cleared transmission-system disturbances. Instead of abruptly disconnecting when voltage or frequency moves outside normal operating limits, a facility would need to ride through the event without creating additional reliability risks.

The proposal addresses both existing and future large loads, although existing facilities would likely receive a transition plan to modify operational parameters and controls. For new facilities, ride-through capability would become part of the interconnection process before the load is energized.

The proposal materials establish performance concepts that data center operators should begin evaluating now:

• During defined voltage disturbances, the facility must remain connected and may reduce consumption, but it may not drop the entire load.
• Once voltage returns to approximately 0.9–1.1 per unit, the facility would be expected to recover to more than 90% of its pre-disturbance consumption within two seconds.
• During qualifying frequency deviations, active-power consumption would generally need to remain within 10% of the pre-disturbance level.
• The facility would not be expected to transfer unexpectedly to on-site backup generation for a disturbance covered by the ride-through requirements.

These values are proposed parameters, not final tariff obligations. PJM still must complete stakeholder review and submit its filing to FERC. Operators should treat them as design targets for planning: not as a substitute for the final approved requirements.

Read the PJM proposal presentation and PJM’s explanation of the proposed reliability standards for the primary source material.

Why Data Centers Have Been Tripping Off

The problem is not usually a single defective UPS. It is the interaction of multiple protection systems operating at different speeds and with different objectives.

A transmission fault can produce a short-duration voltage depression, phase-angle shift, or frequency excursion. Protective relays, UPS rectifiers, static bypass systems, automatic transfer switches, generator controls, and facility-management software may all interpret that event differently.

Common causes of large-load disconnection include:

1. Protective settings that prioritize equipment isolation.
   A relay or UPS input stage may be configured to disconnect when voltage or frequency crosses a threshold intended to protect sensitive equipment. That may be appropriate for a local electrical fault but excessive for a temporary transmission disturbance.

2. Poor coordination between UPS and generator controls.
   A UPS may transfer to battery while an automatic transfer switch starts a generator. If the facility’s control logic is not coordinated, the entire data hall can move from grid power to on-site generation unnecessarily.

3. Static bypass or rectifier behavior.
   Online double-conversion UPS systems isolate the IT output from many input disturbances, but the input rectifier and upstream switchgear still interact with the grid. A UPS can protect the server load while its input protection causes the larger facility load to disappear.

4. Large blocks of identical equipment responding at once.
   Data centers often deploy many similar UPS modules, power distribution units, and server power supplies. If they share the same settings, a disturbance can trigger a synchronized response across hundreds of megawatts.

5. Insufficient disturbance testing.
   Conventional commissioning often verifies outage transfer, generator start, and battery runtime. It may not test a 100–500 millisecond voltage event while the facility remains connected to the grid and the UPS operates in its intended mode.

For grid operators, the result is a sudden loss of demand followed by a sudden transfer to generation. For facility operators, the event may expose a gap between local power protection and regional reliability.

Why Now: Latency, Redundancy, and Thermal Management

AI infrastructure is increasing the consequences of a power event. High-density computing loads can reach tens of kilowatts per rack, with specialized AI deployments moving substantially higher. A large facility may also operate hundreds of megawatts of load across multiple buildings, campuses, and utility service points.

That scale changes the meaning of latency. A power-control decision made in milliseconds can determine whether a facility rides through a disturbance or becomes a major grid event. Operators cannot rely on a manual response after the event begins.

It also changes how redundancy should be evaluated. A Tier III or Tier IV design may provide redundant UPS modules, distribution paths, and maintenance capabilities, but redundancy alone does not guarantee coordinated grid behavior. Two independent power paths can still trip simultaneously if they use the same poorly coordinated voltage or frequency thresholds.

Finally, thermal management creates additional constraints. If a facility changes operating mode, loses cooling capacity, or shifts load between electrical paths, the resulting thermal transient can become a second failure mechanism. Ride-through planning must consider UPS capacity, generator response, cooling controls, battery temperature, and server power behavior together.

What the Proposal Means for UPS Design

PJM’s direction does not make UPS systems less important. It changes how they should be integrated.

A UPS should provide continuity for the IT load without automatically converting every grid disturbance into a full-site islanding event. That requires a coordinated design covering:

• Input breaker and relay settings
• UPS rectifier ride-through capability
• Battery operating logic
• Static bypass thresholds
• Generator start and transfer delays
• Load-bank and commissioning tests
• Power-management communications
• Facility-level emergency controls

Operators should also verify performance at the actual operating point. A UPS advertised at 97% efficiency in a specific operating mode may perform differently at 20%, 50%, or 90% load. Efficiency, overload capability, battery autonomy, short-circuit behavior, and bypass transfer characteristics should all be evaluated against the site’s expected load profile.

The design objective is not simply to “stay on battery.” It is to maintain the required load without causing an uncontrolled transition to generators or a full facility trip. In some architectures, that may require revised relay settings, firmware updates, additional ride-through controls, or a change in how generator transfer logic is initiated.

The Ride-Through Compliance Roadmap

Facility managers, grid compliance teams, and data center developers can take the following steps today.

1. Establish the facility’s grid-facing load profile

Document the facility’s maximum import, normal operating load, minimum stable load, and expected expansion. Identify whether the site approaches the large-load thresholds used in PJM’s broader interconnection and resource-adequacy discussions. Do not assume that a commonly cited threshold: such as 50 MW: will automatically define the final ride-through obligation.

2. Map every automatic disconnection path

Create a single-line and control-sequence review that includes utility relays, medium-voltage switchgear, UPS systems, static bypasses, ATS equipment, generators, server power supplies, and building-management systems. Identify every condition that can remove load from the grid or transfer it to on-site generation.

3. Compare settings with PJM’s proposed performance envelope

Evaluate whether the facility can remain connected through the proposed voltage and frequency events. Model recovery to more than 90% of pre-disturbance load within two seconds after voltage returns to the 0.9–1.1 per-unit range. Review whether active-power consumption can remain within approximately 10% during applicable frequency deviations.

These are planning benchmarks while the proposal is under review. Final requirements may change.

4. Test the complete electrical system: not isolated components

A UPS factory test is not enough. Commissioning should evaluate the interaction between the UPS, switchgear, relays, generators, cooling systems, and IT load. Use staged disturbance testing, power-system simulations, and hardware-in-the-loop testing where appropriate. Confirm that the same event will not simultaneously trigger multiple redundant systems.

5. Create a transition and evidence plan

For existing facilities, maintain records of relay settings, firmware versions, UPS operating modes, generator sequences, disturbance events, and corrective actions. Assign ownership between the facility team, utility, interconnection consultant, UPS manufacturer, and grid-compliance group.

The objective is to demonstrate that the site can meet the final requirements when PJM establishes its transition process: not to wait until a compliance deadline exposes a design gap.

The Broader Regulatory Picture

PJM’s proposal follows a June 18 FERC order directing PJM to address gaps in its treatment of large loads. FERC specifically identified the absence of clear ramp-rate and ride-through requirements for large transmission customers.

The proposed PJM framework is separate from longer-duration resource-adequacy measures. Ride-through addresses how a facility behaves during short-duration faults and frequency excursions. Capacity and emergency-curtailment rules address whether the grid has enough supply to serve the load over a longer period.

Both issues matter. A data center may need to remain connected through a short event but still participate in emergency curtailment during a sustained capacity shortage. Operators should design these functions separately and ensure that one control scheme does not undermine the other.

As of September 15, 2026, PJM’s ride-through proposal is not yet a final tariff requirement or FERC-approved standard. The June 2026 FERC order, the PJM proposal, and future stakeholder decisions should be reviewed with qualified engineering and regulatory counsel.

Real-Time Solutions for Grid-Ready Infrastructure

The new standard for data center resilience is not simply uninterrupted server power. It is coordinated behavior across the facility, the interconnection, and the grid.

Ace Real Time Solutions designs and installs power protection systems that combine UPS equipment, batteries, distribution hardware, monitoring, and professional support. Our team can review your existing architecture, identify ride-through risks, and develop a solution design that supports uptime without creating unnecessary grid disturbances.

Visit Ace Real Time Solutions, review our power delivery and protection services, or submit an enterprise request for quote to request a power audit, technical specification review, or customized solution design.

The grid is becoming less tolerant of uncontrolled large-load disconnections. Data centers that prepare now will be better positioned for PJM’s final requirements, future NERC standards, and the next generation of high-density computing.

Frequently Asked Questions

What is PJM’s proposed large-load ride-through requirement?

It is a proposed interconnection requirement that would require large computational loads, including data centers, to remain connected during defined transmission-system voltage and frequency disturbances instead of disconnecting unexpectedly or transferring immediately to backup generation.

How does ride-through affect UPS design?

Ride-through requires coordinated settings and controls across UPS rectifiers, batteries, static bypasses, relays, switchgear, generators, and facility-management systems. A UPS must protect the IT load while avoiding an unnecessary facility-wide trip or automatic transfer to on-site generation.

When will PJM’s ride-through rules become mandatory?

As of September 15, 2026, the requirements remain a PJM proposal under stakeholder review. PJM has indicated that it plans to file the proposal with FERC in November 2026. The final effective date and requirements will depend on the stakeholder process and FERC action.