Uncertainties in airtightness measurements: regression methods and pressure sequences

Prignon, Martin [UCL] Van Moeseke, Geoffrey [UCL] et al.

This paper presents the results of a series of 30 fan pressurization tests in reproducibility conditions performed within a period of 10 days in October 2017. The tested unit is a newly constructed unoccupied apartment in Brussels. These results make possible to compare different regression methods and evaluate the impact of pressure stations chosen for these regressions. The comparison of three different methods shows that the Weighted Line of Organic Correlation and the Iterative Weighted Least Square reduce the airflow rate standard deviation and give a better uncertainty calculation than Ordinary Least Square. Further work should focus on the generalization of these results in applying these methods to other sets of reproducibility tests. The comparison of different pressure sequences shows that the European standards are coherent in the imposed requirements and the given recommendations. In addition, when performing a regression method, one should avoid low-pressure stations as much as possible in order to reduce the airflow rate standard deviation at low- and high-pressure. Even if this work shows and confirms ways to improve the reliability of calculated combined standard uncertainties, there is still an important difference between these and the observed standard deviation. The investigation of such a difference should be a priority for further researchers focusing on airtightness uncertainties.