By simulating airflow using various polygonal configurations such as triangles and hexagons on actual building sites, the goal is to achieve better architectural designs. Environmentally conscious architecture maximizes the use of natural light and wind, leading to energy-efficient designs with significant advantages. Airflow dynamics are calculated based on the Navier-Stokes equations. Basic geometric shapes are probabilistically arranged using random number generation. The airflow behavior around buildings based on triangular prisms, quadrangular prisms, and hexagonal prisms is analyzed. A collective housing design is proposed based on triangular, quadrangular, and hexagonal prisms. The airflow of buildings arranged in four-unit configurations is examined. The differences in internal and external airflow due to four-unit arrangements of triangular, quadrangular, and hexagonal configurations are investigated. The steepest descent method is employed to identify buildings with strong airflow characteristics. This study proposes a method for generating architectural designs that optimize airflow within polygon-based collective housing layouts, using the steepest descent method applied to natural airflow environments. Finally, the effectiveness of the proposed designs is demonstrated through the actual architectural planning of collective housing layouts on a real site.