A Hybrid MPC Optimization approach for HVAC Direct Load Control*

Direct load control posits the existence of a dynamic relationship between electric utilities and customers, enabling the utility to manage customer consumption as it deems appropriate. Thus, supply can be disconnected where necessary provided such action complies with customer-imposed requirements. Proper execution of a fully automated management plan will require a host of online tools including instrumentation, models, optimization algorithms and automation systems.This study develops a novel methodology that solves the online direct load control problem from an automation based formulation by means of hybrid model predictive control, using HVAC (Heating, Ventilating and Air Conditioning) systems as a particular case. The methodology models the thermal behavior of a standard zone in a building constrained by the customer's maximum temperature requirement.Our simulation results show that hybrid model predictive control is a viable alternative to solving the direct load control problem and can manage electricity demand while satisfying constraints set by consumers. It was also found that the controller operation is robust to simulated disturbances and uncertainties in its parameters.