EVALUATION OF THE IMPACT OF NETWORKING ACTIVITIES ON THE QUALITY OF RESULTS OBTAINED WHEN APPLYING SYSTEM IDENTIFICATION TECHNIQUES TO IN-SITU MEASUREMENTS FOR ASSESSING ENERGY PERFORMANCE OF BUILDING COMPONENTS

The application of system identification techniques to the energy performance of buildings and building components requires a very high level of knowledge of physical and mathematical processes. This factor, combined with the quality of the data, the description of the monitoring environment and procedure, together with the experience of the user of the analysis software itself, can end up in varying results from different users when applying different models and software packages. The objective of a number of activities in recent years has been to develop benchmark test data sets for assessing user performance.
Past international system identification competitions (94 & 96), demonstrated the spread in results that can be expected with regards to application of different models and techniques to the same data. Furthermore, activities of the PASLINK EEIG have tried to consolidate and strengthen the level of knowledge in system identification techniques when applied to the energy and environmental performance of buildings and building components. Workshops have been implemented to assist in the development and application of training instruments which aim to promote the levels of expertise within the grouping and to ensure that data analysis meets the minimum required quality levels. This paper compares the spread in results obtained during the previous competitions to that obtained during the workshops carried out by the PASLINK EEIG following ten years of networking activities in the field.
The objective is to identify the extent to which the networking activities have strengthened the position of the individual teams working in the field and to identify the areas where quality assurance is met and, furthermore, where further improvements can be made. A direct comparison of the quality of results obtained on test results from the previous decade is made with the resent results of data analyzed by following the networking is given and account is taken for changes in software and tools as well as the composition of the individual teams.