Summary
- Echelon says a live fault ride-through test has been completed successfully at its DUB13 facility.
- Huawei, TNEI, NeoDyne, and EirGrid participated in work intended to keep large data centre demand stable during short grid faults.
- Operational data from the test is being shared for further validation as Ireland develops requirements for large demand facilities.
Echelon Data Centres has completed a live fault ride-through test at its DUB13 facility in Ireland, demonstrating how a large data centre load can remain stable through a short electricity-grid disturbance and restore its normal demand within milliseconds after the fault clears.
The work involved Echelon, Huawei, TNEI, NeoDyne, and EirGrid. Echelon said the test showed that data centre demand could endure supply fluctuations and recover rapidly, with operational data being shared for validation and further learning.
NeoDyne, which supported the wider energy-centre project, said its role included the integrated control and power-management architecture used at the site. That system includes controls designed for resilient load sharing and island-mode operation.
Fault ride-through is becoming an important engineering issue for Ireland because a data centre’s response to a grid fault can affect more than the facility itself. Short voltage disturbances are normally handled inside a critical facility through UPS systems, controls, and backup-power architecture, protecting servers from any interruption in external supply.
The grid operator, however, also has to manage what happens to the data centre as an electrical load. If several large facilities abruptly remove their demand from the grid during the same short disturbance, the collective change can create another system-balancing problem after the original fault has cleared.
Keeping a large load predictable
Fault ride-through requirements are intended to make the behaviour of major electricity users more predictable during those events. Rather than transferring away from the grid for every short voltage deviation, qualifying facilities can be required to remain connected through defined disturbances and recover their demand in a controlled way.
That changes part of the relationship between a data centre and the power system. Critical electrical infrastructure still has to protect the IT load, but the facility’s controls also have to account for the effect that switching decisions have on the wider network.
Echelon has been developing fault ride-through behaviour as part of a broader programme with grid and engineering partners. Earlier company material discussed work at DUB10; the recently reported successful live test was carried out at DUB13.
The distinction matters because the test moves the work from a general technical programme into observed behaviour at another operating facility. Echelon has not published a complete engineering data set from the exercise, so independent assessment of every performance parameter is not yet possible.
The company has instead said that test data will be shared for validation. That process will determine how closely the behaviour matches the parameters being developed for large demand users and what adjustments may be required before the approach is applied more widely.
Resilience works in both directions
Ireland is a particularly relevant market for this work because data centres represent a large and concentrated class of electricity demand. As more facilities connect, the collective electrical behaviour of those sites becomes increasingly important to system operators.
Grid resilience has traditionally been discussed from the customer side: how a data centre survives the loss of grid power and keeps servers online. Fault ride-through adds the reverse question — whether the data centre can protect itself without creating an abrupt change in demand that makes the wider system harder to control.
Meeting that requirement can involve UPS configuration, control logic, generation systems, switching sequences, protection settings, and the interaction between utility and facility equipment. Existing sites may also differ substantially from new campuses designed with grid-support requirements in mind from the outset.
That makes live testing useful because simulation alone cannot capture every interaction between controls, power electronics, electrical plant, and the external network.
The DUB13 exercise does not establish a universal technical solution for Ireland’s data centre fleet, but it provides another operational data point for EirGrid and industry participants developing fault-response requirements. The next challenge is to translate successful site-level testing into behaviour that can be reproduced across facilities with different electrical architectures and generations of equipment.
