Optimal management of offshore wind farms and battery storage with ensemble forecasts

The transition to renewable energies requires managing the inherent variability and uncertainty of wind energy generation and maintaining grid stability. Especially offshore wind farms, despite their high energy potential, create significant challenges as their power output fluctuations can overwhelm grid capacity, necessitating costly curtailment that wastes valuable renewable energy. Compounding these challenges, prevailing operations typically still rely on deterministic forecasts that provide single-point predictions without quantifying uncertainty, and address forecasting limitations and grid congestion as separate problems. We present an integrated control of co-located battery energy storage using ensemble-based probabilistic forecasts within a stochastic optimization framework to simultaneously tackle both forecast uncertainty and grid limitations in offshore wind installations. Our real-world validation demonstrates that this integrated approach reduces curtailment by up to 15% compared to deterministic methods, while improving market revenues by 0.1% to 5% depending on intraday price volatility. Critically, we find that neither storage alone nor probabilistic forecasting approaches achieve these benefits - only the combination of storage providing the physical buffering to address grid constraints with uncertainty-aware ensemble forecasts triggers the full economic potential. These results challenge the prevailing reliance on single-point forecasts for battery dispatch optimization and indicate that co-located storage require stochastic-based dispatch strategies to realize its theoretical benefits. As wind capacity expands globally, shifting from deterministic to probabilistic operations offers a practical path to maximize renewable utilization while maintaining grid reliability and economic viability. This work tackles grid congestion and forecast uncertainty by co-optimizing a wind farm and battery via a stochas- tic model using ensemble forecasts. The resulting synergy unlocks the battery's full potential, significantly reducing costly wind curtailment far more effectively than conventional methods.