Summary
The rapid growth of large-scale data centres is introducing new, fast-acting electrical demand characteristics that may bring challenges to power system operation and stability, in particular power system frequency control and stability. This paper examines how utility-scale data centre operation, driven by AI training and inferencing workloads, may result in sudden active power mismatch to power systems and thus may pose risk to frequency stability and control, mainly due to different modes of data centre operation or server load variability. A modelling framework is proposed to represent key data centre behaviours relevant to frequency control studies, including voltage-driven uninterruptable power supply (UPS) switching strategy and high-magnitude server load fluctuations. The data centre modelling framework is then integrated into representative model of the Australian National Electricity Market power grid to assess the system-level impacts and interactions across a range of inertia conditions.
Read more Read lessSimulation results show that hyperscale data centre operation can materially influence frequency stability, resulting in high rate of change of frequency (RoCoF), severe frequency nadir/zenith, and sustained frequency oscillations, particularly in low-inertia operating regimes, and may result in frequency-dependent cascading failures such as photovoltaic tripping or generation disconnections. The study also explores the role of utility-scale energy storage in moderating these frequency-related issues, providing insights for future system security planning and operational assessment.
Additional informations
| Publication type | Session Materials |
|---|---|
| Reference | C4_10492_2026 |
| Publication year | |
| Publisher | CIGRE |
| Country | Australia |
| Study committees |
|
| File size | 1 MB |
| Price for non member | 30 € |
| Price for member | 30 € |
Authors
ABBASI Abbas - Ausnet Services, Australia; TAO Juinn - Ausnet Services, Australia; BULLEN Amalie - Monash University, Australia; GHAZAVIDOZEIN Mehdi - Monash University, Australia
Keywords
