Statistical analysis of wind data using Weibull distribution for natural ventilation estimation

Incorporating natural ventilation in buildings can have significant benefits. Nevertheless, realistic wind conditions continually change. Airflow rate estimation methods that assume wind speed and direction as being constant are thus error-prone. To elucidate wind variability and its effect on airflow rate estimation, the Weibull distribution was used in the current study to represent high-resolution historical wind data, statistically, for a 1-year period. A preliminary investigation was carried out on categorizing the wind by its statistical characteristic, and two distinct wind speed distributions were found by applying the K-means algorithm in the categorization process. The yearly distribution of days in two different clusters showed seasonal changes. The wind data from the same clustering was grouped and utilized to estimate the airflow rate in the form of probability density. The results were in good agreement with the full-scale experiment measurements of a 10-day test in Austin, Texas. The proposed method could be used to characterize wind speed variances, especially with a small time step wind data and enables building designers to do a statistical analysis of airflow.