在台灣，超高樓之開發行對於周圍環境之流場的改變，因此環境影響評估作業準則中，風場（行人舒適度）應進行相關之模擬分析與風洞試驗。 本研究主要利用計算流體力學軟體（fluent 6.3）與風洞試驗進行比較。主要分成兩個部分： （1）兩棟獨立高樓。就風洞試驗：主要使用熱膜探針與Irwin probes進行量測；數值模擬分析：使用標準紊流模式（standard κ-ε）、經修正紊流模式（renormalization Groups κ-ε）、LES (large eddy simulation)、DES（detached eddy simulation）。以DES模擬結果最好，平均風速誤差為9％，有效風速的誤差為17％。 （2）選定台中市七期重劃區。採用standard κ-ε紊流模式、RNG κ-ε紊流模式與LES紊流模式進行模擬預測。數值模擬預測與風洞試驗的結果： （a）16個方向無因次化平均速度的數值模擬（standard κ-ε紊流模式）預測與風洞試驗之相關性（correlation）介於0.53～0.91，平均值為0.81；（b）62個數值模擬測點與風洞試驗相互比較之下，其平均誤差約為20～30％；（c）若使用西安大略大學(university of Western Ontario)舒適度評估準則[2-24][2-41]，數值模擬預測與風洞試驗的趨勢是一致的，但風洞試驗的結果則較為保守；（d）LES紊流模式模擬結果會優於standard κ-ε與RNG κ-ε紊流模式且LES紊流模式的預測會較為保守；必須考慮模擬所需要的時間。（e）若使用丁育群、朱佳仁所建議的舒適度評估準則[2-41]，都符合宜人的狀況，因此數值模擬預測與風洞試驗的結果變得更好；但此評估方法是較為不保守，容易造成大樓興建後會產生較高的風速，因而造成行人感受到不舒適的情形發生。 The scientific and technological development allowed higher and higher buildings to be constructed causing serious wind acceleration on the ground. High-rise buildings tend to change the pedestrian wind environment and caused discomfort to the pedestrians or even pose threat to their safety in this area. Therefore, the Environmental Impact Assessment Act in Taiwan and Taipei Public Space Management Practices stipulated that the wind tunnel experiments for the assessment of the pedestrian comfort must be carried out prior to the construction of the high-rise buildings. This study utilized the computational fluid dynamics (CFD) software for simulation and prediction, and compared the results with the wind tunnel experiments. The study involved two isolated buildings and the wind field evaluation of complex terrain in cities. In the wind tunnel experiments, the ground level wind velocity was measured by hot-film probes and Irwin probes. In the numerical simulation analysis, the standard κ-ε turbulence model, RNG κ-ε turbulence model, LES turbulence model and DES turbulence model were adopted. For the two isolated buildings, the error of the mean velocity of the detached eddy simulation and the wind tunnel experiments was 9% and the error of the effective velocity was 17%. In the complex urban terrain of the Taichung, for the steady-state of the standard κ-ε turbulence model and the wind tunnel experiments in 16 directions, the correlation coefficient ranged between 0.51 and 0.92 and the average was 0.81.The mean error of the dimensionless wind velocity was 20~30%. If the comfort criterion of the university of Western Ontario for pedestrian wind were adopted, the tendencies of the CFD prediction and the wind tunnel experiments were consistent to a considerable extent. The LES turbulence model was the best prediction of standard κ-ε turbulence model and RNG κ-ε turbulence model, but it would take a lot of time to simulate. If used the comfort criterion of the Yu-Chun Ting and Chia-Jen Chu, the ranked evaluation of all locations were pleasant. This comfort criterion would not be conservative and could cause the high velocity after the tall building was constructed.