Automation-assisted fault detection of an air-handling unit; implementing the method in a real building

Abstract

Building automation systems (BASs) are extensively utilized in fault detection and isolation (FDI) of heating, ventilating and air-conditioning (HVAC) processes. Usually a BAS, which is directly interfaced to the process only monitors or collects data for the FDI algorithm. Rarely both control and monitoring actions of the automation system are harnessed for FDI. However, in buildings such a diagnostic approach is possible and illustrated in this paper. The fault detection is based on an on-line diagnostic test (ODT), which is a series of control and monitoring actions applied to a process. Performing an ODT means exciting the automated process by using prescribed input signals, supervising responses and comparing results with a process model. All operations are performed on-line, during normal up state of the process and controlled by the automation system. The fault detection method is outlined for an air-handling unit (AHU) and implemented for its preheating process. The approach is demonstrated in a real building by programming the diagnostic algorithms in a BEMS, installed in a college building and further performing test runs. Faults are detected by comparing gathered data with a statistical model. Due to the difficulties in generating natural faults, a few artificial faults were introduced. The test runs show that the ODT is an uncomplicated diagnostic method for finding distinct and abrupt changes in a process but not for detection of slow degradations and gradual faults. Moreover, the ODT seems to be generic over faults and processes, requires no additional instrumentation and no more than domain knowledge for initiation.