Lee Todd, Vice President, Energy & Carbon Services, ABB Electrification Service, examines how EirGrid’s MPID345 grid code modification could require data centres to transition from passive consumers to taking an active role in supporting grid stability.

Ireland’s data centres are at the centre of a significant regulatory shift. They now consume around a fifth of the nation’s electricity supply, putting pressure on the national grid and net zero targets. At the same time, the sector contributes an estimated €7.3 billion to the Irish economy, while AI adoption is projected to add between €250 billion and €310 billion to GDP by 2035.

Data centres and new technology loads comprised approximately 24% of Ireland’s total electrical energy requirements in 2024, a share expected to reach 30% by 2032. EirGrid’s proposed grid code modification, MPID345, is a regulatory response to that growth.

Thinking about resilience

For decades, data centre electrical architecture has been built around a single overriding objective: protecting the IT load from the grid. During a disturbance, successfully transferring critical load onto stored energy or local backup generation is considered a positive outcome from an internal operations perspective. But from the transmission system’s perspective, the picture looks quite different.

When large loads shed simultaneously during a disturbance, the resulting power imbalance can drive frequency excursions and overvoltage conditions that snowball across the entire grid. This means the nominal frequency of 50 Hz can stray outside tightly controlled limits, while voltage levels can spike beyond acceptable thresholds. Both could cause equipment to trip offline, triggering further disconnections and potentially escalating into a cascading event.

The aggregate level of demand reduction observed during actual events in Ireland has grown sharply as data centre capacity has expanded – from 74 MW (16% of total data centre demand) during a fault in January 2022 to 387 MW (52%) during a fault in May 2025.

MPID345 is EirGrid’s response to this growing risk. It proposes three interlocking requirements for large demand facilities:

• Fault ride-through (FRT) – requires facilities to remain electrically connected to the transmission system during voltage disturbances.
• Active power recovery (APR) – requires restoration of at least 90% of pre-fault demand within 500 milliseconds of voltage recovery to 0.9 per unit.
• Rate of change of frequency (RoCoF) robustness – requires facilities to withstand frequency changes of up to ±1 Hz/s without disconnecting.

Historically, comparable requirements applied primarily to generators and other grid-connected assets, such as wind farms, solar farms, HVDC interconnectors and battery storage systems. Customers with large demand faced less stringent standards, given their inherent behavioural diversity and comparatively modest individual scale.

If adopted, MPID345 will end that difference. By proposing generator-class grid code requirements for large demand facilities, it signals a fundamental shift in how data centres would be expected to interact with the grid and participate in grid stabilisation.

The relevant boundary is the complete facility

One of the most important things operators need to understand is that compliance with the proposed MPID345 requirements is not assessed component by component. The relevant engineering boundary is the complete facility, as seen at the regulated grid connection point.

IT load, UPS systems, cooling and variable-speed drives, generators, storage, reactive compensation, transformers and protection systems can all contribute – positively or negatively – to the active and reactive power trajectory seen by the network. What matters is how all of those assets interact during the disturbance, transfer and recovery sequence.

This distinction has immediate practical implications. Conventional data centre electrical architectures were designed primarily for continuity of IT load. UPS systems provide seamless transfer and energy buffering, but their control logic is not designed to interact with grid dynamics at the point of connection.

A double-conversion UPS can keep IT equipment operating while its grid-side rectifier limits current, reduces input power, blocks or disconnects entirely, depending on its control and protection settings. The result is that the IT load remains online, but the data centre has effectively tripped offline from the grid’s perspective.

Operators should therefore establish the actual UPS voltage-time ride-through envelope, rectifier current limit, blocking and de-blocking behaviour, frequency protection settings, battery capability and input-power recovery ramp. In some installations, OEM-supported firmware or settings changes may improve capability, but software cannot remove fundamental hardware, current or thermal limits.

There is no single solution

Several engineering pathways can contribute to compliance, and the right approach depends entirely on the site’s existing architecture, the scale of the residual compliance gap and operational constraints.

The first question is not how much energy storage a facility needs, but how much incremental active and reactive power is required, for how long and how quickly. Different technologies address different portions of that duty.

The limitation of conventional technologies is not their individual performance, but their inability to simultaneously deliver fast active power, voltage support and frequency resilience within sub-second timeframes.

STATCOM systems provide strong reactive support but lack active power recovery capability. Synchronous condensers provide inertia and system strength but cannot satisfy APR requirements because they do not inject controlled active power during recovery windows. A battery energy storage system (BESS) can provide energy and flexibility, but repeated micro-events can accelerate degradation and impact long-term asset economics. UPS-only upgrades improve continuity but do not create coordinated, grid-supportive behaviour at the point of common coupling.

For some sites, changes to UPS controls and protection settings may close a significant portion of the compliance gap at relatively modest cost. Others may require dedicated storage, power-electronic compensation, generation coordination or more fundamental architectural changes.

The important principle is that additional equipment should be sized only after considering existing site capability that can be demonstrated safely, repeatedly and with validated controls.

Where the challenge becomes most acute

APR is the most stringent requirement of MPID345. Restoring 90% of demand within 500 milliseconds after voltage recovery requires a level of coordination and responsiveness that exceeds the capabilities of conventional systems.

Across a large facility, different UPS systems may recover their rectifiers at different rates; batteries may begin recharging; mechanical systems may be restarting; generators may be transferring; and protection timers may be resetting. The aggregate recovery trajectory at the grid connection point is the product of all those processes happening simultaneously – and managing that trajectory is as much a controls and orchestration challenge as an energy storage problem.

Recovery also does not end when pre-fault demand has been restored. After the event, recharge must be ramp-limited to restore readiness without creating a secondary disturbance. A second power step drawn from the grid in the minutes following a successful fault ride-through can itself become a grid stability event. Controlled, ramp-limited recharge is therefore an integral part of a compliant response strategy.

MPID345 compliance is not an energy problem – it is a sub-second power orchestration problem. The implication is clear: compliance cannot be achieved through isolated component upgrades. It requires coordinated system-level behaviour.

From compliance to operational resilience

MPID345 also sits within a wider regulatory shift that operators should understand in full. Under Ireland’s Large Energy User Action Plan (LEAP) and associated connection policy, new data centres will have to demonstrate how at least 80% of their annual energy demand will be met by additional renewable electricity projects generating in Ireland. Grid access is increasingly conditional on active contribution to energy system resilience – not just on capacity availability.

This is the context in which the energy trilemma – balancing security of supply, cost and decarbonisation – becomes a practical challenge rather than an abstract policy framework. The cheapest compliance solution is not necessarily the most resilient. The most resilient architecture may not make the best use of existing infrastructure. And simply adding more equipment is not always the most sustainable or cost-effective answer.

Operators who approach MPID345 as a site-level engineering assessment will be better placed to find solutions that serve all three legs of the trilemma simultaneously. Addressing these requirements early may also help facilities strengthen their operational resilience and plan for future expansion in grids where dynamic performance is becoming as important as installed capacity.