Controlling the operation of HVAC (Heating, Ventilation, and Air-Conditioning) systems is arguably the most effective way to reach desired indoor conditions in buildings. Nevertheless, such control may involve complex dynamics when dealing with passive energy technologies. In this paper, we focus on maximizing the passive operation of HVAC in a novel low-energy building design by means of Model Predictive Control (MPC). The low-energy building design, located in The Green Village, consists of a thermal chimney and solar shades over all-glass facades to provide the required indoor air conditioning as passively as possible. The MPC controller is based on a transient grey box model and a hierarchical control architecture to satisfy thermal comfort while minimizing the active energy requirements. Using sensor data collected from the actual building in April and May 2021, the grey box model shows a good agreement with the measurements, since the variance accounted for is 90% in most cases. Moreover, via a comparative study among different MPC architectures we show that managing the distinct transient response of each component (shades and chimney) is the best for successful overall performance – e.g. considering linear agents for shading and nonlinear agents for ventilation. The hierarchical MPC architecture established outperforms the standard ones by 22.7% in terms of control performance. We also compare the proposed MPC approach against the rule-based control method currently implemented in the actual building, which indicates that MPC demands about 78% less active energy, highlighting the proposed optimization-based control approach.

Energy-saving devices are extensively sought in several fields, including heating, ventilation, and air-conditioning (HVAC) tasks in buildings. This paper investigates six model predictive control (MPC) strategies as a way to optimize the operation of a solar shaded, natural ventilated building located at TU Delft campus. Such building is based on an innovative combination of a thermal chimney and glazing walls for harvesting passive energy. The main challenge is dealing with the system dynamics. The predictive controllers we consider include both linear and nonlinear MPC, and four hierarchical MPC strategies. All controllers aim to minimize auxiliary power consumption and, consequently, increasing the energy savings. The six control strategies performance are evaluated using reference values for thermal comfort, while relying on simulations performed in MATLAB for calculations. The hierarchical MPC architecture which considers a hybrid structure with nonlinear tracker for ventilation and linear agents for heating purposes appears the most promising one.