Effect of various uncertainties on the performance of occupancy-based optimal control of HVAC zones

Abstract

Model Predictive Control (MPC) has emerged as a potential control architecture for operating buildings in a more energy efficient manner. We study through simulations the effect of several sources of uncertainty that arise in the implementation of MPC on the energy consumption, thermal comfort, and indoor air quality (IAQ). These include occupancy profile, measurement errors and mismatch between the plant and its model that the control algorithm uses. Simulations are carried out for two extreme cases: a winter day with no solar load and a summer day with high solar load. The study shows that increasing fluctuations in occupancy, errors in measuring occupancy, and model mismatch have the strongest impact on the energy consumption. However, measurement errors in outside temperature and solar load does not have significant impact. Therefore, it is possible to improve the controller performance by using more accurate occupancy sensors. Furthermore, implementation cost can also be reduced by eliminating the sensors and prediction algorithms for predicting outside temperature and thermal loads without compromising the controller performance. Even with these uncertainties, MPC delivers 12-37% reduction of energy use over conventional control methods without affecting thermal comfort and IAQ.