Empirical approach for estimating reduced order models of buildings

An energy audit is a standard process to support the decision-making in the area of energy management, aiming to improve buildings performance. Nevertheless, an energy auditor does not succeed as easily and quickly in identifying all the energy streams in a facility in order to decide whether retrofitting the audited object is beneficial. In fact, energy auditing of buildings is usually a time consuming and expensive process, due to efforts required for data collection and modelling of audited objects. However, the modelling phase might be made less demanding, thanks to the development of reduced models, conceived in the form of lumped parameters models. According to relevant literature, the detailed physical models are time consuming and often quite expensive; they are adopted when a very detailed evaluation is necessary. On the contrary, reduced-order models, also called grey-box models, showed their reliability to achieve a suitable description of the thermal response of buildings. The thermal parameters of the simplified models are usually extracted in real time; this allows estimating the thermal response of a building in its current state, whose information can be reused to make predictions about its expected behaviour. This paper attempts to derive a quick and efficient modelling procedure by identifying the unknown properties of a mathematical model based on partial observations of the heat dynamics of the building, that is, in our case, a small data collection of temperature measurements. Starting from a preliminary test performed in the machine laboratory of our department at the Università Politecnica delle Marche, the investigated empirical procedure for deriving grey-box models will be provided and the results will be analysed. The potential for energy retrofitting will be evaluated in order to define future developments.