The Sensitivity of Chiller Performance to Common Faults

In the development of automated fault detection and diagnostics (FDD), it is important to identify the most appropriate sensors for the faults that can occur. This paper presents data and results that provide a basis for development of fully automated FDD applied to chillers. Faults that can be detected and diagnosed with relatively low cost sensors were experimentally simulated in the laboratory and the sensitivities of different measurements to each fault were identified. The fault testing led to a set of generic rules for the impacts of faults on measurements that could be used for fault diagnoses. The impacts of the faults on cooling capacity and coefficient of performance (COP) were also evaluated. Based upon the results, all of the faults were found to be significant at the levels introduced and should be detectable and diagnosable by an FDD system using low cost measurements. The data set obtained during this work is very comprehensive and can be used to design and evaluate the performance of FDD methods.