Wind data representativeness to improve natural ventilation analysis andCFD simulations in buildings. Application case: Medellín, Colombia

In natural ventilation studies wind speed and direction are regularly reduced to mean and most frequent annual values, using data from wind charts or EPW files with a single location as the representative condition of an entire city. This study improves wind characterization using data from several meteorological stations calculating the representativeness of each paired values (speed and direction) depending on its annual frequencies. Classifying paired records by year, month, and daily time slots in eight Medellín city sectors revealed the great improvement the natural ventilation studies could obtain using local climate data sources. The SIATA weather station network in Medellín offers historical wind records still unused in architecture despite the complex topography and its deep valley conditions. Massive data classification to calculate a representativeness index helps to select the best input values to evaluate building ́s ventilation. The method could be easily adopted to locations where wind patterns are diverse and alternative data with higher spatial or temporal resolution is available.