| Optimal online supervisory control methods are developed, tested and implemented for the minimization of electrical demand of heating, ventilating and air conditioning (HVAC) systems. The systems considered are limited to variable-air-volume air handlers served with chilled water from a central cooling plant. The central plant consists of vapor compression cycle chillers and cooling towers for heat rejection. Chilled water is delivered to the air handling units via a primary/secondary pumping system with variable flow pumping on the secondary chilled water loop.; Optimal supervisory control is the selection of the systems setpoint that optimize a system. In the case of the HVAC systems considered, optimization is performed on the chilled water, condenser water and supply air temperature setpoints. In addition, methods for the selection of optimal supply air and secondary chilled water loop pressure setpoints are explored.; System-wide models of various forms and a sub-component based model are developed to predict the system energy demand. The models are minimized using simple solution-space search techniques. The controlled variable setpoints that minimize the models are then implemented on the actual equipment.; A building in Boulder, Colorado was used as the test site. The building was used as a laboratory to collect data for model development and testing, as well as for the actual implementation of optimal and near-optimal supervisory control schemes.; For many buildings, selection of constant setpoints is near-optimal. More generally, setpoint reset laws can be written and implemented to provide near-optimal setpoint selection. Model based optimal supervisory control can be directly implemented in an actual system for power minimization. Saving associated with optimal supervisory control at the test site are in the range of 6 to 9 percent of total demand. |
