Model Predictive Control for Fast Frequency Control in Hybrid Renewable Plants: Increasing Robustness through Enhanced Model Observability

Guaranteeing frequency stability in low-inertia power systems requires converter-based renewable energy sources to provide rapid response under tight operational constraints. This paper presents a two-fold contribution to the coordinated fast frequency control of grid-connected hybrid renewable plants consisting of grid-forming battery energy storage, and grid-following photovoltaic generation and electrolyzer loads. First, a Model Predictive Control (MPC) strategy is developed to coordinate these heterogeneous assets. Extensive robustness evaluations across an uncertainty space demonstrate that the proposed MPC achieves a 48-50% median improvement in tracking accuracy and a 33-48% reduction in energy storage requirements compared to traditional PI-based and decentralized strategies. Second, the MPC’s performance is enhanced through a novel prediction model that formulates system states by the converter’s output power instead of angle differences as in Tie line power flow-based models. Unlike the traditional approach, this model avoids the need for augmented disturbance states and achieves full observability by formulating load disturbances as process noise. The proposed model further reduces median tracking error by 18% and achieves a 9% median energy storage reduction in the robustness analysis. The performance of the controllers is evaluated through Hardware-in-the-Loop experiments for various contingencies. A variance-based global sensitivity analysis reveals that parametric uncertainty, specifically concerning grid inertia, grid impedance and power ramp limitations, dominates performance impacts over structural model-plant mismatch. The results demonstrate that the adoption of MPC offers substantial benefits for hybrid renewable plant control, while the proposed prediction model further maximizes control performance.