Article ID Journal Published Year Pages File Type
569598 Advances in Engineering Software 2014 12 Pages PDF
Abstract

•Speed up factors are calculated for various idealized and real topography.•Both 2D and 3D topography are studied and their differences highlighted.•Various turbulence models are implemented and tested for predicting wake flow.•LES simulation at full scale dimensions are conducted using wall functions.•Validation of result against available literature and field observations.

The effect of topographic features on wind speed and wake turbulence is evaluated by conducting Computational Fluid Dynamics (CFD) simulations using an in-house CFD program that features various turbulence models. The simulation results are assessed by computing Fractional Speed Up Ratio (FSUR) along longitudinal lines at different elevations. Such information is useful for evaluating wind loads on long span structures and micro-siting of wind turbines on complex terrain. Simulations are conducted on both idealized and real topographic features in both 2D and 3D domain. The turbulence structure behind hills is examined using several turbulence models such as the mixing-length, standard k–∊k–∊, RNG k–∊k–∊, realizable k–∊k–∊ and Smagorinsky LES models. All turbulence models predicted FSUR values on upstream side of hills adequately; however, the performance of simple turbulence models, such as mixing length, is found to be insufficient for characterizing wakes behind hills. RANS turbulence models gave results close to one another; however, those models that incorporate modifications to account for adverse pressure gradient conditions performed better at wakes behind hills. LES conducted at full scale dimensions, and using wall functions, failed to give results that are comparable to the other turbulence models. Re-conducting the simulations at model scale dimensions, hence at relatively small Reynolds number, and without using wall functions gave results that are comparable to those found in the literature. Therefore, use of wall functions can degrade quality of results in LES of high Reynolds number flows of practical interest.

Related Topics
Physical Sciences and Engineering Computer Science Software
Authors
, ,