Fault detection and diagnosis in liquid chillers

Abstract

Automated fault detection and diagnosis of refrigeration equipment is important in maintaining efficient performance, reducing energy consumption, and increasing the reliability and availability of these systems. The reducing costs of microprocessor technology and the incorporation of more sophisticated monitoring equipment on to even fairly small refrigeration plant, now makes the introduction of on-line fault detection and diagnosis on refrigeration equipment feasible and cost effective. This paper reports on the development of a fault detection and diagnosis (FDD) system for liquid chillers based on artificial intelligence techniques. The system was designed to monitor plant performance and to detect and diagnose faults through comparison with expected behaviour and previous experience of fault characteristics. The system operates on line in real time on a Java 2 platform and was initially used to detect refrigerant charge conditions. The results indicate that the FDD system developed is able to detect and diagnose fault conditions arising from low or high refrigerant charge correctly, using two parameters as detectors: condenser refrigerant outlet temperature and discharge pressure.