As power systems become increasingly reliant on distributed energy resources (DER) and variable renewable energy, uncertainty in operational forecasting is becoming more pronounced. This uncertainty has important implications for how last-line-of-defence reserves are activated and how system reliability is managed in real time. These challenges are particularly relevant for the future role of gas-powered generation (GPG) and similar peaking plants, which currently provide the majority of firming capability in Australia’s major electricity grids.
This context makes our recent conference paper — presented at the IEEE’s Australasian Universities Power Engineering Conference (AUPEC) in late September — particularly relevant to ongoing research at the UQ Gas and Energy Transition Research Centre.
The paper presents a case study of 22 January 2024, a day when the Queensland region of the National Electricity Market (NEM) shattered its previous electricity demand record with an unprecedented 8.6% increase in peak demand. This surge was driven by extreme heat, high humidity and a sudden collapse in behind-the-meter rooftop PV output as dense cloud bands moved across South East Queensland in the late afternoon.
During the event, the region’s excess available generation fell as low as 425 MW, and it may have dropped even lower if not for the activation of the PeakSmart demand-response program and several unscheduled distribution-network outages that unintentionally reduced load.
A central theme of the paper is the role of forecasting accuracy in shaping reserve activation decisions. Although the NEM’s Short-Notice RERT mechanism is designed to act as a safeguard under tightening conditions, its activation relies on accurate short-term demand forecasts issued at least three hours ahead. On the case study day, AEMO’s P30 pre-dispatch forecasts underestimated actual demand by up to 750 MW — well above the 99th percentile of historical forecast errors for the region. This under-forecasting contributed to an LOR2 warning being downgraded prematurely, meaning that reserve negotiations did not commence despite conditions later becoming markedly tighter. These findings suggest that as DER penetration grows, conventional forecasting and reserve-trigger methodologies may require enhancement to better account for rapid, weather-driven changes in net demand.
The analysis also highlighted that the current suite of reliability mechanisms are not always synchronised during operational time frames. On the case study day, distribution level demand response scheme, RERT triggers, and real-time price signals did not operate in a fully coordinated way. These misalignments increase athe risk that critical interventions may occur too early, too late, or not at all during periods of genuine system stress.
With Queensland being one of the world’s highest rooftop-PV jurisdictions, these findings offer broader insights into future operational forecasting challenges that may face other power systems with high DER penetration.
The full paper is now available via the IEEE Xplore portal, and a longer summary of the paper’s key findings is also available on WattClarity.com.au.
Author
Dan Lee
Principal Research Analyst
Gas & Energy Transition Research Centre
