Fault detection of cooling coils based on unscented Kalman filters and statistical process control

Buildings account for about 40% of energy consumption in the U.S. The Heating, Ventilation and Air Conditioning (HVAC) systems account for 57% of energy used in commercial and residential buildings in the U.S., and between 4% and 20% of HVAC energy is wasted because of sudden faults and component degradation. To reduce energy consumption and improve occupancy comfort, HVAC fault detecting is important. In our recent work, a synergistic integration of Kalman Filter and Statistical Process Control (SPC) was used for fault detection of chillers based on a gray-box model. This paper builds on that framework to detect cooling coil faults. Major enhancements in several aspects are made, including the establishment, improvement and validation of a (nonlinear) gray-box model and the extension of method to the nonlinear case by using the unscented Kalman filter to provide prediction and SPC control limits. The method is tested by using a simulation model of a building on the UConn campus. Numerical results show that the method can detect both sudden and gradual faults of cooling coils with high detection rates and low false alarm rates.