AT-15-015 -- Distributed Predictive Optimization of a Building HVAC System ✓ Most Recent

The use of advanced control strategies can result in thereduction of energy consumption in buildings, but a singlecentralized controller is often impractical.A distributed controlapproach that combines the ease and modularity of local controlwith the system-wide benefits of centralized control has significantpotential for positive impact. Earlier research effortsresulted in the development of a distributed optimization techniquereferred to as "neighbor-communication optimization"(NC-OPT). In this paper, the algorithm is applied to a multizonebuilding air-conditioning system.Adesign for the control architectureis developed wherein each zone is treated as a separatesubsystem, as are the variable-air-volume system and the waterchiller. This design is applied to a building modeled and simulatedusing EnergyPlus software. The NC-OPT approach isshown to successfully modulate the system setpoints to balanceenergy efficiency with occupant comfort, including calculatingnew setpoints for chilled-water temperature reset. This modulationis achieved without a centralized controller for thesystems and with communication only between neighboringsubsystems. Discussion of the results and plans for future workare provided.