Evaluation of a CFD-based Wind Model Optimized for ABL Flows: Comparisons with Observations from a Tall Isolated Mountain

Abstract:

Near-surface wind predictions are important for a number of applications, including transport and dispersion, wind energy forecasting, and wildfire behavior. Researchers and forecasters would benefit from a wind model that could be readily applied to complex terrain for use in these disciplines. Unfortunately, near-surface winds in complex terrain are not handled well by traditional modeling approaches. Computational fluid dynamics (CFD) models are increasingly being applied to simulate atmospheric boundary layer (ABL) flows, especially in wind energy applications; however, the standard functionality provided in commercial CFD models is not suitable for ABL flows. Appropriate CFD modeling in the ABL requires modification of empirically-derived wall function parameters and boundary conditions to avoid erroneous streamwise gradients due to inconsistences between inlet profiles and specified boundary conditions. This work presents a new version of a wind model, WindNinja, developed for wildfire applications in complex terrain. The new version offers two options for flow simulations: 1) the native, fast-running mass-consistent method available in previous versions and 2) a CFD approach based on the OpenFOAM toolbox and optimized for ABL flows. The model is described and evaluations of predictions with surface wind data collected from a recent field campaign at a tall isolated mountain are presented. CFD models have typically been evaluated with data collected from relatively simple terrain (e.g., low-elevation hills such as Askervein and Bolund) compared to the highly rugged terrain found in many regions, such as the western U.S. Here we provide one of the first evaluations of a CFD model over real terrain with ruggedness approaching that of landscapes characteristic of the western U.S. and other regions prone to wildfire. A comparison of predictions from the native mass-consistent method and the new CFD method is provided.