Accurate and Computationally Efficient Aggregated Modeling of Offshore Wind Farms for Grid Compliance

Abstract

Detailed electromagnetic transient models of offshore wind farms incorporate the individual wind turbine dynamics and controls of many wind turbines (WTs), resulting in substantial computational burdens. This limits their practicality for large-scale system studies. While aggregation techniques reduce this burden, they often compromise accuracy, particularly in capturing dynamic behaviour during power system disturbances. To overcome this challenge, this paper proposes a gaussian mixture model clustering with information-theoretic averaging to group similar WTs and derive representative aggregated models. A structured eigenvalue sensitivity analysis is conducted to identify the control parameters that have the greatest impact on system dynamics, reducing the dimensionality of the parameter space. Additionally, an improved multi-objective gradient descent optimisation strategy is developed to identify and provide optimal values of system and control parameters under varying wind conditions and fault scenarios to ensure accuracy. The proposed modeling applicability is further demonstrated through assessment on an HVDC connected offshore wind farm under large disturbance and weak grid conditions in the time and frequency domains. Finally, the aggregated model's performance is validated against the detailed benchmark to ensure U.K.’s grid compliance on modeling, confirming its ability to replicate key dynamic and steady-state characteristics with substantially reduced computational effort.