Computational study on turbulent flow around a model of wind turbine rotor

• Main Author: Pakravan, Amir
• Format: Thesis
• Language: English
• Published: 2010
• Online Access: http://psasir.upm.edu.my/id/eprint/41143/1/FK%202010%2078R.pdf; http://psasir.upm.edu.my/id/eprint/41143/

There are different methods to study flow characteristics around a wind turbine; common methods are the Blade Element Momentum theory (BEM), wind tunnel tests and Computational Fluid Dynamics (CFD). The BEM approach has limitations on predicting velocity profiles around rotor disk and also it needs some initial data about the 2D airfoil section. Wind tunnel test also has its own problems and limitations and also it is so expensive. This study is concerned with simulating 3D flow around the rotor of a horizontal axis wind turbine using a commercial CFD package code FLUENT 6.3 and comparing experimental data from Sant (2007). Axial velocity was calculated on two measurement planes at downstream (Ya=3.5cm & 6cm) for different radial positions. Coefficient of lift force on the blades is also considered. All the simulations have been done by CFD software, code Fluent 6.3, for three different yaw angles (Ψ=0⁰,30⁰&45⁰). A Gambit pre-processor tool is used for mesh generation. A Moving Reference Frame (MRF) model in Fluent is used for modeling rotation of the wind turbine blades. Four turbulence models including Reynolds Stress, k-ε-Realizable, k-ε-standard and k-ε-s st are used and evaluated. The experimental data from (Sant, 2007) is used to compare the simulation results with the wind tunnel test. The k-ε-sst was the least expensive in terms of computational effort. K-ε- standard and k-realizable were similar to each other in case of computational effort and both took slightly more time in comparison with k-ε-sst. Reynolds Stress Model (RSM) needed about 40% time more than the other turbulence models to get the convergence. Almost in all of the positions for Ψ=0⁰, k-ε- Standard did the best prediction in comparison with the other turbulence models for calculating the axial velocities. A very good agreement was reached at Ψ=30⁰, all turbulence models performed a reasonable job for prediction of axial velocity at this position, except the k-ε- SST. None of the turbulence models did an acceptable prediction at Ψ=45⁰, K-ε-realizable and Reynolds Stress did a better job for prediction of on a blade in comparison with k-ε- Standard and k-ε- SST. Pressure and velocity contous are presented to show the flow behaviour around the rotor disk. According to the extracted results from the simulations all the turbulence models have predicted that on the upstream side at the vicinity of the blade surface by increasing the pressure along the blade from the hub to the tip blade velocity drops. Drastic changes in pressure at tip of the blade cause in significant changes for the velocity.