Model-based condition monitoring of a HVAC cooling coil sub-system in a real building

A comparison of the performance of two fault detection and diagnosis methods applied to a cooling coil subsystem in an air-handling unit installed in a real building is presented. Both methods employ a rst principles based reference model of the target system. One scheme carries out diagnosis using expert rules and the other recursively re-estimates selected parameters of the system model that correspond to particular faults. The procedures and information required to con- gure the schemes for condition monitoring are discussed. The results of testing the methods on an HVAC cooling coil subsystem in a commercial of ce building in the UK over an entire cooling season are reported. Both methods were able to both detect faults and provide some diagnosis. The expert rule method, however, appears to be more robust. Issues associated with the con guration and implementation of both methods are discussed in terms of performance and cost.