Performance monitoring and fault diagnosis of vacuum pumps based on airborne sounds

This paper presents a cost-effective scheme of implementing condition monitoring (CM) for a vacuum pump station, which combines the airborne sound (AS) measured remotely with high efficiency of abnormality detection, with surface vibration (SV) measured locally with high diagnostic capability. In particular, AS measurement is employed to implement online and real time monitoring of a number of machines such as several vacuum pumps spread over a large area. Once there is any abnormality found, SV will be used to diagnose the faulty locations and severities. In this way the monitoring can be more cost-effective by avoiding the use of a high number of vibration measurements. Having gained the dynamics of vacuum pumps and had a failure mode and effects analysis (FMEA), this study details the implementation of this scheme based on the vacuum pump station of a paper mill. It demonstrates that airborne sound can show the relative spectral components for each vacuum pump to a certain degree of accuracy, allowing a quick discrimination of potential faults of these pumps. This demonstrates that the AS measurement is an appropriate technique to use for such circumstances where many machines need to be monitored but limited budget can be invested in the complicated monitoring systems.