Summary
- MPID345 introduces fault ride-through, frequency tolerance, and active-power recovery requirements for large demand facilities.
- Ireland’s concentrated data centre load can amplify a transmission fault when multiple sites reduce grid demand simultaneously.
- ABB proposes ultracapacitor storage as one response, although compliance will depend on site-specific modelling and controls.
Irish data centres may be required to ride through transmission faults and rapidly restore demand under proposed grid-code rules developed in response to the growing system effect of large, power-electronic loads.
EirGrid and SONI’s MPID345 modification would require affected facilities to tolerate defined voltage and frequency disturbances, withstand rates of change of frequency of up to plus or minus 1Hz per second, and return to at least 90% of pre-fault active-power demand within 500 milliseconds after system voltage recovers to 90% of nominal.
ABB has set out an ultracapacitor-based architecture intended to support that response, combining short-duration energy storage with fast controls and coordination across the data centre’s power systems.
A protected site can still disturb the grid
Data centres have traditionally been designed to isolate sensitive equipment from poor power quality, transferring load to UPS systems or standby generation when a disturbance passes defined limits. That approach protects the IT environment, although several large sites responding in the same way can create a second system event.
During a transmission fault, voltage falls and both generation and demand may reduce. Generation generally returns quickly after the fault clears, while some power-electronic demand remains disconnected or continues operating from internal backup sources. The resulting imbalance can push system frequency upwards.
Ireland’s relatively small synchronous electricity system makes concentrated demand particularly visible. More than 2,000MVA of data centre and other new-technology demand is contracted through EirGrid, with additional capacity connected or contracted through ESB Networks and SONI.
Data centres and other new-technology loads accounted for about 24% of Irish electrical energy requirements in 2024 and could reach 30% by 2032 under the median scenario cited by the system operator. During periods of low national demand, the proportion represented by operating data centres can be considerably higher.
EirGrid has recorded growing aggregate data centre demand reductions during transmission events. As more campuses energise, identical protection behaviour across several facilities can produce a larger and faster change than the system was originally designed to absorb.
Ultracapacitors cover the transition interval
ABB’s proposed approach uses ultracapacitors because they can charge and discharge rapidly over repeated short-duration events. The storage layer can support the facility while protection systems, UPS equipment, generators, and grid conditions move through the fault and recovery sequence.
Ultracapacitors do not replace the full standby-power system. Batteries and generators remain necessary where the facility must continue operating through a prolonged outage. Their role is concentrated in the first seconds, when controls must maintain internal stability and avoid a delayed or abrupt return of demand to the network.
Compliance will depend on the complete electrical architecture rather than a single storage product. Connection voltage, transformer arrangement, UPS topology, static transfer systems, generator controls, IT power-supply behaviour, and protection settings all influence how the site responds.
Existing facilities may present the greater challenge because their electrical systems combine equipment from different generations and manufacturers. A retrofit can require additional storage, revised protection, new communications, control changes, and staged testing within an operating data centre.
The proposed rule does not necessarily require a facility to draw exactly the same power throughout the voltage dip. It can transfer load internally, provided that it remains within the required fault envelope and restores its grid demand at the specified rate after voltage recovery.
That recovery must be fast enough to support system stability without exposing the IT load to an uncontrolled transfer. Electromagnetic-transient studies, digital models, hardware-in-the-loop testing, and agreement with the system operator will become part of the connection and compliance process.
New facilities can incorporate these duties from the first design stage. Protection zones, storage capacity, controls, metering, model validation, and test points can be coordinated before equipment orders are placed. Retrofitting the same behaviour after commissioning carries greater operational and outage risk.
Connection applications are therefore moving beyond maximum demand and redundancy descriptions. System operators increasingly need to know how a data centre behaves during voltage dips, frequency excursions, backup transfers, and recovery, including its reactive-power response and interaction with the wider network.
The proposed requirements also alter the relationship between facility resilience and grid resilience. Early disconnection may no longer be the safest overall response where data centres form a large share of national demand. Controls must protect computing equipment without imposing a destabilising load change on the network supplying it.
ABB’s architecture provides one possible route, rather than a universal compliance design. Final performance will depend on the equipment installed at each site, the approved dynamic models, the control sequence, and witnessed testing against the adopted grid code.
Ireland is addressing the issue earlier than many European markets because data centre demand already occupies a large share of the electricity system. Similar requirements are likely to spread as individual campuses reach several hundred megawatts and large demand becomes a system-security concern in its own right.
