Scalable SCADA-Based Calibration for Analytical Wake Models Across an Offshore Cluster

Abstract

A multi-level hyperparameter optimization framework is performed to calibrate analytical wake models in the context of multiple wind farms within the Belgian-Dutch offshore cluster. The calibration, applied on the TurbOPark model with Gaussian wake profile, is performed on different scales. Initially, calibration focused solely on internal wake effects, followed by a calibration for individual wind farms, considering both internal and external wake effects, and finally performing the calibration using SCADA data from multiple wind farms within the concession zone, with and without accounting for blockage. It was observed that calibrating wakes for freeflow wind directions using internal wakes only results in tuning parameters similar to the calibration involving both intra-and-inter farm wake effects. Minor variations are noted between wind farms, with dependencies on wind speed and wind direction across all cases. When the tuning parameter is calibrated using SCADA data from multiple wind farms, a significant reduction in the tuning parameter was observed, compared to calibration that focuses on one wind farm per calibration. Analyzing the model residual error for wind coming from the north-west reveals that the calibrated wake model effectively accounts for cluster wake effects. Furthermore, a consistent and substantial presence of heterogeneous inflow perpendicular to the wind direction originating from the south-west is observed, which is not captured by the modeling framework, which assumes homogeneous inflow. The observed trend of inflow heterogeneity suggests that the discrepancy cannot be attributed solely to blockage effects.