Statistical approach for finding sensitive positions for condition based monitoring of reciprocating air compressors

Reciprocating air compressors are one of the most popular and widely used machines in industry today. Timely detection of fault occurring in these machines becomes very critical since it influences the plant performance by virtue of system reliability, operating efficiency and maintenance cost. Monitoring of faults by identifying sensitive positions through sensory output forms vital part of everyday manufacturing. Health monitoring of a reciprocating air compressor using PC based data acquisition system and timely identification of potential faults can prevent failures of the entire system. Various transducers, data acquisition DAQ hardware and relevant software forms the basic components of the health monitoring system. Prior to acquiring the data for health monitoring and consequently the fault diagnosis, it is crucial to locate the sensitive positions on the machine. This paper proposes a scheme to determine the sensitive positions on a machine in healthy condition, based on computation of statistical parameters such as Peak value, Standard Deviation, RMS value, Variance and Cross Correlation.