Detecting Deterioration of Compression Equipment by Normalizing Measured Performance Relative to Expected Performance
This paper presents the concept of comparing measured and expected performance in order to detect and assess deterioration of compression equipment. For a number of reciprocating and turbomachinery performance parameters, the paper describes methods to predict expected values as a function of conditions, options for expressing deviation between measured and expected values, and approaches to assessing the severity of such deviation. The study recognizes the fact that many engine- and turbine-driven compressors have data logging system but have not yet realized the full value of such systems. In many cases, measured parameters are still compared to a single limiting value instead of one which varies with conditions. The approach in this paper should help improve the situation. For a number of parameters, a reliable method of predicting expected values, described in the paper, is first to characterize expected behavior by mapping the performance over a range of operating conditions, and then to develop a regression model (with diagnostics such as standard error). An alternative, in the case of parameters such as reciprocating compressor bearing temperature, is to use the average of the other similar values. Normalizing (dividing each newly measured value by the corresponding expected value) and comparing the normalized value to unity represents one way to monitor for deterioration. Comparing the observed deviation against the statistical variance from the historical records, expressed as the standard error, provides an alternative method to assessed the significance of the deviation.