Using a multi-parameter monitoring methodology to predict failures in the cryogenic plant of the cold neutron source at Australia's OPAL reactor

A 5 kW Brayton-cycle helium refrigeration plant provides cooling at 20 K to the Cold Neutron Source (CNS) at Australia's OPAL Reactor. During several years of operation to the present day, the plant has experienced an unusually high number of turbine and compressor failures. The root cause for some of the failures is known, but for others remains to be determined. All of the failures were catastrophic without any prior warning from standard industrial monitoring based on singular process variables such as temperature, pressure and vibration. The failures and the down time they caused have been very costly. As the operator of the plant, we have developed a multi-parameter monitoring (MPM) methodology to track the performance of the plant. The methodology utilises indicators obtained from a combination of process variables based on their thermodynamic relations. By studying the historical trends of appropriate indicators, especially during the past failures, we have found some indicators that would be able to improve our predictive capability so that we can avoid similar failures in the future.