風力發電是一個十分乾淨的綠色能源，台灣目前也在正在推行風力發電，不過很少從氣象的角度來討論風場與發電效益之間的關係。從觀測上我們可以利用光達(Lidar)等儀器來測得風力發電機所在之風場環境條件，同時也可以根據IEC(2005)的規範去評估風機運轉性能，不過觀測所設置的點有限，對於風機與大氣的交互作用可藉由模式了解更多。 本研究針對桃園沿海大潭發電廠附近的8支風力發電機，進行都卜勒光達的風場測量，觀測的時間為2015/12/17～2015/12/30。觀測的目標風機為Vestas V80，其高度為78公尺，葉片直徑為80公尺，Lidar設置的地點位於其西北方約160公尺。模擬的部分則是利用中尺度氣象模式(WRF)來進行模擬，使用大尺度渦流模型(LES)、MYNN邊界層參數化以及風電場參數化(Wind Farm Parameterization)方法來討論風機產生之尾流與大氣邊界層的交互作用，並以Lidar的資料作為風速場的驗證。模擬總共使用六層巢狀網格，在第五以及第六巢狀網格的初始與邊界條件使用one-way nested down製作，同時只在第六層巢狀網格100公尺解析度使用風電場參數化方法，總共植入8支風機。 風電場參數化的概念為因風機拖曳力使得大氣動能的損失，轉為電能以及亂流動能。比較使用風電場參數化(WF)與未使用(Ctrl)的結果，尾流風速在風機後方風速約減少2.0 m/s，同時尾流會具有較大的亂流動能，亂流動能收支主要是來自於垂直傳送、風切作用以及消散速率。邊界層高度在尾流處則略有提高，因MYNN邊界層參數化使用hybrid方法，分別找出虛位溫以及亂流動能所訂之高度，作出權重後定義出邊界層高度，亂流動能在垂直方向上的增加使得高度略微提高。在風速模擬的部分，雖然不同的參數化方法模擬結果相關性高於0.8，但是LES有較大的正偏差，在風速分布亦可見到，整體來說，使用風電場參數化後，風速分布在大於10m/s是較為接近觀測的。在風能評估方面利用風電場參數化所估算出來的風機功率曲線較接近原廠，與觀測差別最大在於高風速區，除了風機老化所造成的差異，還有偏角差異的問題。 ;Wind energy is a clean and renewable resource. Taiwan gives impetus to it nowadays. Wind energy is seldom discussed about the relationship between wind field and the power generation performance from the perspective of meteorology. According to the standard of assessing the power generation performance from IEC(2005), Lidar is often used to measure the wind condition. Because of the limitation of the site observation, the model is applied in this study to understand the interaction between the wind turbines and the atmosphere boundary layer. This study focuses on the eight wind turbines at Taoyuan Datan. The wind speed is measured by Doppler Lidar. There are four fronts passing through from 2015/12/15 to 2015/12/30. The postfrontal cold northeasterlies lead to the maximum wind speed happened. The type of target wind turbine is Vestas V80 with height of 78 meters and rotor diameter of 80 meters. Field measurements are performed using Doppler wind Lidar which lies 160 meters to the northwest of the target wind turbine to verify the model results. On the other hand, Large-Eddy Simulation(LES), MYNN planet boundary layer and Wind Farm Parameterization (WFP), which are based on the WRF model, are applied to discuss the interaction between the wake flow generated from wind turbines and the atmosphere boundary layer. Model design with six nesting domain are used for the multiscale atmospheric simulations. In addition, the initial and boundary conditions of the fifth and the sixth nested domain are made by one-way nested down, and WFP inserted eight wind turbines only is used in domain six. The idea of the WFP is about energy transition. The kinetic energy loss caused by the turbine drag force is converted into electric power and turbulence kinetic energy (TKE). The difference between WF (with WFP) and Ctrl (without WFP) are apparent behind the wind turbines that the wind speed decreases around 2m/s while TKE increase among the wake flow. The TKE budget of the wake flow is based on the three parts: vertical transport, shear production and dissipation rate. Comparing to Ctrl, the boundary layer height of WF is slightly higher due to the hybrid method in MYNN that consider both the virtual potential temperature and TKE. The simulated wind speed in different parameterizations have high correlation with the observation, but LES has a higher positive bias. Wind speed distribution in WF is closer to the observation, especially for wind speed over 10m/s. The power curve of the WF form WFP is similar to the manufacturer, but far from the observation in high wind speed region because of the attenuated performance and yaw misalignment.
