Utilizing Data Centers for Inertia and Fast Frequency Response Services

Abstract

This research evaluates data centers as an emergency source of virtual inertia and fast frequency response, using PMUs to detect disturbances. The performance of the proposed method is validated using DIgSILENT PowerFactory simulation, calibrated using a real frequency event that occurred in the Irish power system. Wind generation is significant in the Irish system and significantly higher levels are required to reach renewable energy targets. Wind power, like photovoltaics, are mediated by power electronics that do not inherently respond to frequency variation. This research addresses problems with the drop in system inertia and the availability of primary frequency response on systems with high non-synchronous infeed. Demand response has the potential to replace these services. Typically, a large number of domestic, or light industrial, loads are considered for such services, but these present challenges in terms of monitoring and control. This research focuses on the potential of large load data centers that incorporate uninterruptable power supplies as standard, therefore a demand response does not have a direct effect on operation.