應用對稱性守恆離散法於風機流場模擬

Abstract

本研究建立大渦流法(LES)計算框架，以渦黏度模型估算次網格尺度 剪應力，通過 Lagrangian-Averaging Scale-Independent Dynamic Wong-Lilly Model (LASI-DWLM)獲取紊流模型係數，並以致動盤模型(ADM)模擬風機造成的 影響，以模擬在大氣邊界層正常運轉的風機流場。本研究以有限體積法離散統御 方程式，通過二階近似在空間中計算動量方程式各項，並以二階 Adams-Bashforth 方法計算時間項。在空間中使用一階迎風近似，在時間上以一階正向歐拉法積分， 求解 LASI-DWLM 中的兩個附加的輸運鬆弛方程式。本研究以大氣邊界層理想案 例驗證所提出的 LES 框架，並通過參數研究評估 LES 框架的性能，發現大氣中性 分層條件下可獲得良好結果。對於在大氣邊界層中單獨運轉的風機流場模擬中， 本研究結果與先前研究相比，獨立風機後方的跡流速度有較大損失，但兩者在附 加紊流強度的表現上具有良好一致性。本研究進行基於中性大氣邊界層的紊流入 流的 Horns Rev 離岸風場的模擬。本研究使用移位週期邊界條件能夠顯著減少入 流條件導致的非均勻性，本研究發現目前使用的數值模型與先前研究相比改善紊 流行為預測以及風場功率的預測精度。

This study develops a large-eddy simulation (LES) framework for the flow simulation of wind turbines. The subgrid-scale stresses are modeled through an eddy viscosity model, where the Lagrangian-averaging scale-independent dynamic Wong-Lilly model (LASI-DWLM) is employed to obtain the coefficients of the turbulence model, and the influence of the wind turbines in the atmospheric boundary layer are modeled through an actuator disk model. The momentum equations are discretized via second-order scheme in-space and a second-order Adams-Bashforth scheme in-time. The LASI-DWLM in- troduces two additional transport equations, which are discretized via first-order upwind schemes in-space and first-order Forward Euler scheme in-time. The proposed LES frame- work is validated with idealized cases of the atmospheric boundary layer. A parametric study is done to identify the performance of the proposed framework, where satisfactory results are delivered for neutrally stratified conditions. The simulation of undisturbed wind turbines where turbulence effects are considered is done. The results of stand-alone wind turbines are compared to a reference study, showing a slightly larger wake veloc- ity deficit in our current model due to differences in the actuator disk model, but obtain good agreement in the behavior of the added turbulent intensity. In the simulation of the Horns Rev offshore wind farm, the usage of shifted periodic boundary conditions reduces the non-homogeneities introduced by the inflow conditions if non-shifted conditions were used instead. The results are compared with two references and the previous version of the current LES framework, where improved turbulent flow prediction is obtained with respect to the previous version, which leads to improved power prediction in reference to the SCADA data for the purpose of validation. The current framework respect to SCADA data have a relative error of 21%, while for the comparison of the current LES framework with that of the reference study, the maximum error is 14.7%.