Development of a remote refrigerant leakage detection system for chillers and VRFs

Abstract

In Europe and Japan, owners of large refrigeration and air-conditioning equipment, such as chillers and VRFs, are required by law to carry out regular inspections for refrigerant leaks. There are two methods of regular inspection: the direct method, which uses visual inspection and leak detectors equipped with gas sensors; and the indirect method, which uses equipment operating data to  estimate leaks. However, large equipment requires many inspection points and direct inspection is time-consuming and labor-intensive, placing a heavy burden on both the equipment owner and the inspector. On the other hand, the European F-gas regulation provides an incentive to halve the number of inspections if a permanent leakage detection system is installed, and similar incentives are being considered for other countries regulations. The authors developed a highly accurate refrigerant leakage detection system using machine learning techniques that can be used to meet incentive requirements. The details of the technology and the accuracy of the detection system tested on chillers and VRFs are discussed in this paper.