Posters

Abstract

Light Detection and Ranging (LiDAR) and other Remote Sensing Devices (RSDs) are frequently used in the wind energy sector because of their various benefits over traditional met masts, which include portability, flexibility, the capability to measure at extended heights, quicker permitting procedures, and lower equipment and maintenance costs. However, RSDs have limitations in the accuracy of their measurements in complex terrain due to flow heterogeneity. Lately, Reynolds-averaged Navier-Stokes (RANS) Computational Fluid Dynamics (CFD) simulations have been increasingly used to calculate a post-measurement correction of the data to address these limitations. The CFD models use the topography and roughness of the area surrounding the LiDAR as inputs to simulate how the local landscape affects the airflow around the device and, consequently, allow the user to calculate the measurement bias due to the device reconstruction mechanism and correct the data. A large research study was conducted in Brazil to assess the accuracy of stability-dependent CFD correction factors for WindCube LiDAR data across 15 sites with varying terrain complexity. Initial assessments utilized Meteodyn correction factors derived from neutral stability simulations. In this phase, 11 of the 15 LiDARs displayed errors of less than 1% in comparison with their co-located met masts. However, one specific LiDAR and met mast pair showed poor results, with a 5.5% error. A second phase of the study concentrated on the lower accuracy observed in the correction for this particular campaign and investigated the role of thermal stability in the results. Advanced CFD tools incorporate modified RANS equations and boundary conditions to carry out simulations that account for thermal stability, thereby enhancing the post-CFD correction once the site's dominant thermal stability conditions have been identified. The consideration of thermal stability reduced the bias of the corrected data from 5.5% to 2.3% in this critical case, and led to a mean absolute speed error of around 0.8%, considering all 15 cases.