A robust model-based approach to diagnosing faults in air-handling units

This paper describes the development of a robust model-based approach to diagnosing faults in air-handling units that avoids false alarms caused by sensor bias but does not require application-dependent thresholds to be selected. The diagnosis is based on a semi-qualitative analysis of the measured data using generic fuzzy reference models to describe the behavior of the equipment, with and without faults. The scheme is applied to the cooling-coil subsystem of an air-handling unit, and the sensitivity of the diagnosis to sensor bias and fault size is examined. The results of the diagnosis are compared to those obtained using reference models that describe the behavior of a specific design. The scheme is also used to commission the cooling-coil subsystem of an air-handling unit in an office building. Results are presented that demonstrate the proposed scheme does not generate false alarms in practice. It is concluded that the accuracy of sensors currently used means it is likely that only large faults can be detected in practice and that more accurate measurements are required if a higher level of fault sensitivity is needed.