Compliance and Model Validation of a Grid Forming BESS Replacing Synchronous Generation: Real-World Integration into Australia’s Northern Grid

Battery Energy Storage System (BESS), with overload capability of 45.6 MW, was commissioned for the application of GFM technology in this environment [1]. This paper documents the transition from the design phase presented in CIGRE 2024 [2] to the empirical realities of operational integration in 2026.

The project undertook a testing regime to validate the GFM control strategy, confirming the capability of the plant to operate across its full active and reactive power envelope (±45.6 MW,

±35 MVAr) and adherence to dispatch targets. A critical operational finding was the divergence between the requirements in the Network Technical Code (NTC) [3] and the optimal tuning required for the specific network characteristics. While the GFM unit could meet fast rise times, such aggressive tuning induced oscillatory behaviour in Reactive Power Control mode.

Consequently, the performance standards were bifurcated based on control mode, with stringent dynamic requirements retained for Voltage and Control and more relaxed limits for Reactive

Power Control. A critical aspect of integration involved tuning the "Virtual Machine" to satisfy requirements for Inertia and Contingency FCAS [4]. To emulate the synchronous fleet it replaced, the GFM control loop was initially configured with an inertia acceleration time constant of 8.8 s (H = 4.4

MW.s/MVA). This was later retuned (to a H of 6.6 MW.s/MVA, i.e. an increase of 50%, while two damping gains were reduced by a factor of 3.3.) to match measured Frame 6 gas turbine behaviour [5]. Analysis of system separation events confirmed plant response was consistent with these settings, successfully providing active power injection during frequency excursions.

Additionally, the 45.6 MW overload capability [2] during under-frequency events was verified, with the system managing thermal constraints by automatically ramping back to continuous ratings after 60 seconds. Post-commissioning model validation revealed a distinct dichotomy between replicating field events in the Root Mean Square (RMS) and Electro-Magnetic

Transient (EMT) simulation domains.

The RMS model validation was highly successful. When benchmarked against a major system separation event on 04 September 2025, the model showed excellent correlation with field data.

Accuracy was achieved after correcting a discrepancy in measurement latency; the frequency meter delay was increased from 0.12 seconds to 0.30 seconds to match the physical plant response. Conversely, EMT model validation presented novel challenges. Initial attempts to use the standard voltage playback method for validating asymmetric faults resulted in power oscillations due to the high sensitivity of GFM controls to signal noise. To address this, a

"Synthesized Waveform" methodology was utilised based on RMS data. This approach successfully filtered the noise and produced a stable response, demonstrating a technically robust alternative for validating GFM performance against unbalanced events in the EMT domain.

The project has proven to be a successful and robust implementation of Grid Forming technology [6], providing valuable insights for the future integration of inverter-based resources.