Stochastic modelling of temperatures affecting the in situ performance of a solar-assisted heat pump: The multivariate approach and physical interpretation

Abstract

Box-Jenkins-based multivariate stochastic modelling is carried out using data recorded from a domestic heating system. The system comprises an air-source heat pump sited in the roof space of a house, solar assistance being provided by the conventional tile roof acting as a radiation absorber. Multivariate models are presented which illustrate the time-dependent relationships between three air temperatures—at external ambient, at entry to, and at exit from, the heat pump evaporator. Using a deterministic modelling approach, physical interpretations are placed on the results of the multivariate technique. It is concluded that the multivariate Box-Jenkins approach is a suitable technique for building thermal analysis. Application to multivariate model-based control is discussed, with particular reference to building energy management systems. It is further concluded that stochastic modelling of data drawn from a short monitoring period offers a means of retrofitting an advanced model-based control system in existing buildings, which could be used to optimise energy savings. An approach to system simulation is suggested.