The impact of PM2.5 pollution and urban heat island (UHI) on sustainable urban development and human health has been a major concern for governments. Based on Computational Fluid Dynamics (CFD) simulation and on-site monitoring, this study investigated the airflow velocity, PM2.5 concentration and urban heat island intensity (UHII) distribution in the urban area at different prevailing wind directions, from the perspective of urban design. In order to validate the accuracy of CFD simulation results, on-site monitoring of temperature and airflow velocity was conducted on the roof of a building. A low-dimensional model was adopted to improve the efficiency of CFD data analysis. Moreover, based on the results of the low-dimensional model, PM2.5 concentration and UHII in urban areas at different prevailing wind directions were assessed by clustering analysis. The findings showed that when α = 0° and 90°, the average PM2.5 concentration was 28.9–56.8% (8.76–27.66 μg/m³) lower and the average UHII was 23.4–34.2% (0.8–1.15°C) higher, compared to α = 30°, 45° and 60°. Meanwhile, the downwind area suffers from more serious PM2.5 pollution and UHII than the upwind area. These findings could potentially provide a guide for future building/urban design to improve the urban environment from the perspective of building and road construction.