Optimizing the Construction Process of Industrial Maintenance Decision Models by using Predictive Reliability Analysis and RCM Dynamism

Abstract

A stochastic process, unlike a deterministic process, instead of having one possible reality of how a process can evolve over time, has an indeterminate future evolution. This uncertainty is described by the probability distributions. Failures events in a physical asset follow a random order based on a probability distribution, which is often represented in the reliability analysis as a tool for the maintenance management, whose results would offer decision criteria in order to optimize maintenance strategies. Current reliability analysis (RA) employ data dimensions based on age and failure probability in order to obtain a twodimensional graphic representation (probability density function of failure, PDF) that provides a connection between them and suggesting results that could improve the making-decision criteria. However, if the plan is to optimize maintenance management, it is necessary to assure a record of all the significant available variables to generate solid and coherent samples, in order to build up precise decision models beginning with a multidimensional reliability analysis. The following proposal states that the proper consistency of a sample initiates with the degree of dynamism which the body of knowledge RCM (Reliability Centered Maintenance) links to the daily work orders management, under the CMMS (Computerized Maintenance Management System) or EAM (Enterprise Asset Management), which are not thought to be in favor of a reliability analysis. Work orders will be converted in definition stages for the life cycles and for the FMEA (Failure Mode and Effects Analysis) records upgrading as new failure modes occurrences appear. In addition to CBM (Condition Based Maintenance) strategies complicity, which will add in the sample relevant predictable variables over the occurred failure modes, which usually are omitted or poorly involved in the two-dimensional PDF graphic, and therefore, are not often considered in the common reliability analysis. Quality of the sample is therefore assured and routinely prepared with the age and condition attributions, which gives a potential predictive model with satisfactory results about the probabilistic forecast, being able to sustain the maintenance strategies employed at the appropriate time and proper component, in order to guarantee, the return on investment (ROI) growth in the industrial plant.