Numerical Simulation Of Wind Turbine Wakes And The Study Of Wind Farm Layout Optimization

When the wind flow passes through the rotating wind turbine, a phenomenon of wind speed reduction and turbulence intensity increasement occurs due to the wind turbine wake effect. According to the former studies of other researchers, it was found that due to wind turbine wake effect the power losses in a wind farm range from 15% to 50%. So, it can be concluded that understanding of wake mechanisms is critically important for a wind power project. The micro-sitting optimization of wind farm is the prerequisite and foundation for a wind engineering project. During the process of micro-sitting optimization, the positions of the wind turbines are adjusted, with the main purpose of reducing the wake influence on the downstream turbines and increasing the efficiency of the wind power utilization as well as the economical benefit of the wind farm. Besides, with the rapid development of the wind power industry, some complex wind farms are got developed in recent years. The complex terrain effect and the various wind conditions make the micro-sitting within a complex wind farm more difficult. In this thesis, the simulation of wind turbine wake effect, the layout optimization of flat wind farm, the study of the flow distribution and the wind turbine layout design within complex wind farm are involved through the following aspects.(1) Modified AD/RANS methods are developed and validated to predict the wake flow behind wind turbines. Simulations are performed using the flow solver Ellip Sys3 D with the actuator disc(AD) methodology coupled with Reynolds Averaged Navier-Stokes equations(RANS) method. The under-prediction of wake width and a slow recovery of downstream wake are observed when compared with the experiment data, these are also reported in the previous researches. Based on the simplified theory of actuator disc model, the physical instinct of RANS turbulence model and the feature of atmospheric boundary flow, modifications are applied on the AD/RANS method, which further yields two modified AD/RANS models. To validate the proposed models, comparisons are made between numerical predictions and measurements for a single wind turbine, two wind turbine wake interactions and the the power output of a large offshore wind farm. Results show that the newly developed models can provide satisfactory results in most of the test cases.(2) Based on the classical Jensen model, an improved 2D_k Jensen wake model is developed and assessed using various test cases. Due to the limited computation resources, engineering wake models are still widely used in the wind power industry. The developed model is based on the Jensen model and further uses a cosine shape function to redistribute the spread of the wake deficit in the crosswind direction. Besides, a variable wake decay rate is proposed by taking into account both the ambient turbulence and the rotor generated turbulence compared to a constant wake decay rate in the Jensen model. The results are compared with field measurements, wind tunnel experiments, and results of advanced k-ω turbulence model as well as large eddy simulation(LES) data. Overall, it is found that the proposed wake model provides good predictions both in term of the shape and the velocity amplitude of the wind turbine wake, especially the far wake which is the region of interest for the wind engineering project. This developed model is expected to be used in the power output prediction of the real wind farm in the further studies.(3) The optimal placement of wind turbines within a flat wind farm is studied. An improved wake model is employed to predict the power production, a more sophisticated economical model is used to predict the investment of a wind power project. An advanced optimization algorithm and different optimization strageries are employed in this study. At last, several best configurations are presented for the wind farm developers to choose according to their requirements. In addition to the best layouts, annual energy output, the cost, the number of turbines and the efficiency of wind farm for each configuration are also included.(4) A study of the flow distribution and anoptimization in the placement of wind turbines in a complex wind farm is conducted. The flow around a three-dimensional hill on the top of which a wind turbine is positioned is simulated at first. And then the accuracy of decoupling simulation approach is tested. The decoupling method means that simulation of the wind flow around a dam terrain with two wind turbines positioned is artificially separated into two parts: the simulation of wake effects and the simulation of terrain effects, and then linear superposition is performed on these two parts to obtain the final velocity distribution. Through comparisons with the results from regular one-time coupling simulation, it is found that the decoupling method bring out an error of 20% for the velocity prediction. At last, a preliminary study is performed for the wind farm layout optimization in a cosine hill wind farm.