System identification and model-predictive control of office buildings with integrated photovoltaic-thermal collectors, radiant floor heating and active thermal storage

Abstract

The present study explores efficient integration approaches of photovoltaic-thermal systems coupled with corrugated transpired solar collectors (building-integrated photovoltaic-thermal, BIPV/T), Heating, Ventilation and Air Conditioning (HVAC) systems and thermal storage devices, to enable optimal collection and utilization of solar energy in high performance buildings. The objective is to (a) develop models that capture the relevant system dynamics and are computationally efficient for subsequent use within model-predictive control (MPC) algorithms; (b) evaluate the energy saving potential of the integrated system and the predictive controller in comparison with baseline operation strategies. An open plan office space at Purdue’s Living Laboratory is used as test-bed, in which the BIPV/T system preheats ventilation air, while also, it is coupled with the building through an air-to-water heat pump and a thermal energy storage (TES) tank that serves as the heat source for the radiant floor heating (RFH). A detailed energy prediction model developed in TRNSYS is considered as a true representation of the building and it is used to identify the parameters of low-order linear time-invariant state-space models. Both gray-box and subspace state-space system identification (4SID) methods are investigated. A simulation study is performed using TMY3 data for West Lafayette, IN during the heating period. The results show that implementation of a deterministic MPC algorithm for the optimal set-point trajectory of the TES tank can reduce the electrical energy consumption of the heat pump by 34.5%. For the BIPV/T configurations tested, the energy saving of the integrated solar system can be up to 45% compared to the baseline operation of the radiant floor heating. The study also investigates the impact of forecast uncertainty for the horizontal solar irradiance on the performance of the predictive controller, with the results showing considerable impact on thermal comfort conditions when the prediction error is higher than 38%.