A statistical approach for analyzing used oil data and enhancing maintenance decision making: Case study of a thermal power plant.

Abstract

A lubricant is an essential component for enhancing the equipment’s functionality and durability. For this reason, used oil analysis (UOA) is becoming an integral part of the plant’s lubrication program which is part of Condition Based Maintenance (CBM). By monitoring the lubricant’s condition through the UOA, organizations can optimize the equipment availability by reducing failure incidents of rotating elements. This paper advances the use of a predictive model of used oil analysis data with a view of assisting maintenance decision making of critical power

plant equipment. The steps of the proposed methodology include data pre-pro- cessing, principal component analysis (PCA) for dimension reduction, and logistic

regression analysis to build the predictive model, where the lubricant’s parameters

are compared against set thresholds, or limit values from which, indications of sig- nificant lubricant deterioration may be derived. The framework is applied to a ther- mal power plant case study. The novelty of the framework is towards providing

insights for maintenance decision making and moreover, highlighting critical used oil analysis parameters that are indicative of lubricant degradation. By addressing such critical parameters, organizations can better enhance the reliability of critical operable equipment.