Predicted and measured performances of near zero-energy houses: a comparison methodology

In an energy performance verification protocol, comparing predicted and measured data is a difficult task, especially due to uncertainties surrounding the building and its environment. This work presents the development of a global approach aiming to qualify "real" performance versus expected one, by exploiting measured data (including weather and occupant's behaviour measurements). Illustrated in its first iterative phase, this approach associates joint sensitivity and uncertainty analyses with a dynamic thermal simulation model in order to gradually identify and then reduce the main sources of uncertainty on the studied performance indicator. Lastly, temporal and correlation analyses are used to identify most suitable measurement periods for the estimation of the main influential building model parameters.