Atmospheric Regime-Dependent Wake Model Performance

Abstract

Accurate prediction of wind farm power output under varying atmospheric conditions remains a critical challenge for offshore wind energy planning. This study links wake-model performance explicitly to atmospheric regime using an event-based, time-domain analysis. SCADA data, lidar measurements, and meteorological reanalyses are combined to evaluate two classes of wake models—simplified engineering models and a high-fidelity large-eddy simulation (LES) model—under two contrasting offshore regimes: a shallow, stably stratified boundary layer and a deep, convective one. Representative 24-hour cases from two offshore wind farms (Moray East in Scotland and Mermaid in Belgium) are analysed. The results show that under convective boundary-layer conditions, both engineering models and the LES reproduce farm power with high accuracy, supported by reliable inflow representation and strong turbulent mixing that promotes rapid wake recovery. In this regime, engineering models capture mean wake losses well but remain limited in resolving spatial structure and short-term variability, partly due to heterogeneous inflow that violates their assumption of horizontal homogeneity. Under stable, shallow boundary-layer conditions, performance degrades markedly. Engineering models systematically overestimate farm power and substantially underestimate wake losses, while the LES, although more accurate, also overpredicts due to inflow errors inherited from the mesoscale forcing. These errors originate from misrepresentation of boundary-layer height, wind-shear structure, and low-level jets. The LES additionally indicates weak upstream flow deceleration consistent with global blockage, though observational data were insufficient to confirm this conclusively. Overall, the findings demonstrate that wake-model skill is strongly regime dependent: neither engineering models nor LES provide uniformly reliable predictions across all atmospheric conditions. A regime-aware modelling approach—combining appropriate model selection, calibration, and inflow characterisation—can substantially reduce uncertainty in energy-yield estimation and improve confidence in offshore wind-farm development.