Assessing the value of typical meterological years built from observed and from synthetic data for building thermal simulation

Typical Meteorological Years obtained from observed meteorological records have become the de facto data source when evaluating thermal performance of buildings. However, this data source has various drawbacks, and alternative TMY assembling methods based on statistical and stochastic models seem to have been perfected to a point where they become sound alternatives. Numerical simulations of test cells were performed for a mid-latitude temperate climate, Lisbon. Thermal performance was evaluated using as input long term observed time series (control situation), long term stochastic data, and TMY obtained by classic and by stochastic methods. It is found that stochastic data seems to provide a good source of meteorological data, indeed more flexible and adequate than those series provided by the currently usual approach.