Presentations

Abstract

Due to the rapid deployment of offshore wind projects worldwide, especially in the North Sea, there is a growing need for fast engineering models to simulate wake effects in order to understand and quantify the interactions among offshore wind plants. Such quantification supports wind farm operators in meeting current and future needs by providing insights into both intra-farm and external wake effects, which is critical for various operational and planning-related developments. However, wake modelling for operational wind farms presents significant challenges because of the combined mesoscale and microscale meteorological effects at each local site. Analytical engineering wake models rely on a fixed set of coefficients, which often fail to capture the actual wake losses caused by the farm itself, neighbouring farms, and future planned farms. One of the most crucial input parameters is the wake decay coefficient, which varies regionally and on the climate. Typically, engineering models use a fixed or interpolated value derived from a calibration against LES or RANS simulations, with the possibility to include it as a function of the ambient turbulence intensity. In this study, a statistical procedure for calibrating the wake decay coefficient is presented using full-scale operational SCADA data. The method has been validated on selected offshore wind plants that were not used for calibration. Since analytical engineering wake models can only simulate the steady-state flow dynamics, we also investigate the impact of Gaussian smoothing on the wake deficit with respect to wind direction to more accurately represent realistic wake behaviors. Our analysis reveals a relationship between ambient turbulence intensity and the model specific wake decay coefficient. Additionally, we perform a sensitivity analysis to investigate the impact of different turbulence and engineering wind farm models from the PyWake inventory, as well as the influence of metric selection on the calibration procedure. The calibrated models provide accurate modelling of wake losses at the cluster, farm, and turbine levels, supporting both forecasting and historical analyses of Belgian offshore cluster. In terms of forecasting, these models can be coupled with numerical weather prediction (NWP) simulations to facilitate maintenance planning, curtailment planning, and electricity market trading on day-ahead and intraday horizons. For historical analysis, the calibrated models offer a detailed breakdown and allocation of losses, helping operators identify underperforming wind turbines and conduct root cause analyses.