Posters

Abstract

In complex/rugged terrain it is possible to observe large differences in predicted wind speed-up when adopting different aerodynamic-flow model families, such as linearized flow models (engineering orographic models), RANS-CFD solvers, and LES-based approaches. For a rugged site where we have multiple wind farms in operation, several met masts and Lidars, and multiple repowering and greenfield developments, we compared commercial RANS-CFD tools, few commercial WRF/LES tools, and a linearized engineering flow model in terms of orographic and roughness speed-ups. This case study highlighted significant discrepancies among the models, and area-related bias between model predictions and measured references. It is important—especially in projects located in complex terrain—to assess the site-specific uncertainty due to modelling that, in addition to the typical list of contributors to total AEP uncertainty, explicitly includes the flow-model uncertainty associated with horizontal and vertical wind extrapolation across the domain. In repowering and extension projects, the availability of long-term SCADA data and multiple on-site measurements provides a unique opportunity to assess this uncertainty directly from operational evidence, rather than relying solely on generic assumptions. We worked on a method to define site-specific uncertainty quantitatively, based on a statistical analysis of compatibility between model-based predictions and available measurements.  Across the tested sites, we observed that model accuracy often depends on factors that go beyond the typical uncertainty drivers usually considered, such as measurement distance, the height difference between measurements and hub height, orographic complexity, vegetation, and atmospheric stability characteristics. Instead, performance appears to be intrinsically site-specific: at some sites the models are able to reproduce the recorded measurements with high accuracy, while at others the error remains persistently high. This method investigates a possible method for the quantification of such site-specific uncertainty e.g. through a statistical compatibility test. It was also implemented a statistical method to exploit the potential of SCADA wind data for defining the wind field and speed-ups over the repowering area where the irregular grid of SCADA measurements is located.