Typical faults of air conditioning systems and fault detection by ARX model and extended Kalman filter

Since faulty operation of heating, ventilating, and air-conditioning (HVAC) systems is detrimental to energy conservation, and maintenance experts are no longer able to detect faults due to the sophistication of current air-handling units (AHUs), automated fault detection and diagnosis (FDD) is increasingly important. In the present study, the results of a survey about typical faults that are commonly encountered in air-handling systems are summarized, and two methods of finding abrupt faults are described. To investigate the development of automated fault detection schemes, two methods to detect an abrupt fault are tested, and the effectiveness of the methods is analyzed. Both are based on a mathematical model of system dynamics. The first one is an autoregressive exogenous (ARX) model and the second is based on an extended Kalman filter. It is shown that faults that are difficult to detect by a simple limit checker method can be detected in both cases on the basis of computer simulation by HVACSIM+.