Assessing the co-heating test as reference for heat loss coefficient estimation: In-depth analysis on an artificial dataset

In order to bridge the building energy performance gap it is necessary to quantitatively and qualitatively characterize the actual energy demand of buildings and the thermal behaviour of their envelopes. Current European legislation prescribes the assessment of the energy efficiency of the building stock by means of energy performance certification, which labels the building performance based on theoretical data and specifications reflecting the design phase. At the current state, in most European countries such certificates provide the only insight into the overall performance of the building stock, while the need of characterizing the actually achieved building behaviour is recognized. The most common testing procedure used with the aim of characterizing the as-built thermal performance of commissioned dwellings is the co-heating test. It consists of a prescribed measurement procedure where a vacated house is submitted to an elevated indoor temperature. During the experiment, the indoor and outdoor environment, as well as the energy demand needed to maintain the stable indoor temperature are precisely monitored. Up to date, this testing procedure is acknowledged as the most reliable in practice, however, the limitations of the methodology are not yet completely explored and unambiguously defined. Thus, the aim of this work is to explore the limits of the co-heating testing procedure on artificially generated experiments which provide a completely defined environment and reference targets. To assess the impact of different aspects, such as the parameters of the statistical models or monitoring period and length, the artificial datasets provide the ideal conditions since the impact of measurement inaccuracy can be avoided and the expectations of the methodology itself can be evaluated. The main findings complement the prevalent research and indicate the possibility of assessing the building performance in the 10% bounds from the target, however considering idealised monitoring data. The research shows that using global horizontal instead of south vertical irradiation can provide overall more accurate results, especially in transient months. Finally, when assessing suitable periods for performing tests in a moderate climate, only 20% of the whole year can be considered to yield satisfactory results.