A vapour compression chiller fault detection technique based on adaptative algorithms. Application to on-line refrigerant leakage detection

The existence of faults in vapour compression chillers plays a significant role in terms of energy efficiency loss, performance degradations, and even environmental implications. In this paper, a dynamic model-based fault detection technique suitable for real-time implementation is proposed. The main objective is to obtain a reliable and automated tool for fault detection in vapour compression chillers, which can be applied in steady-state or transient operation. The fault detection methodology is based on comparing actual and expected performance using an adaptative model and operating variables dynamic thresholds. The technique has been successful applied for on-line refrigerant leakage detection with experimental tests involving the artificial introduction of the fault in a laboratory vapour compression plant, showing the results its capability of detecting incipient leakage failure conditions avoiding false alarms.