Predictive Analytics and Diagnostics Drive Effectiveness in Condition Based Monitoring

Large industrial internal combustion engines employ a wide range of analytical techniques in order to improve their availability and fuel efficiency, and to decrease their maintenance expense. When viewed in context of reliability centered maintenance programs, most industrial engines rely on one of the following systems: • Time-based preventive maintenance analyses, • Continuous monitoring systems that have first-principles deterministic models, or • Run until failure operating philosophy. Predictive analytics using empirically-derived pattern recognition algorithms can enhance problem detection and increase the effectiveness of an organization’s monitoring program. With robust early detection that has few false alarms, equipment operators can remove time-dependent randomness from their ability to discover problems. Skilled equipment analysts and technicians will no longer spend time devoted to analyzing healthy equipment. When a company has numerous engine installations, scalability and cost-effectiveness becomes increasingly more important. Empirically-derived pattern recognition speeds fleet deployment across a wide range of equipment types and models. This reduces the dependence on the expertise that is required to establish and maintain a first-principles based system. Predictive analytics enables a bridge between the depth of coverage of a permanently installed full analytical system and the versatility of portable analyzers and preventive maintenance analyses. Equipment operators can use early detection from predictive analytics to focus their technicians on analyzing the right equipment at the right time. This paper will describe the predictive analytics modeling philosophy around frame mechanics, combustion, and emissions. One or more case studies will show the processes of even detection, diagnostics, collaboration, and information consolidation. The extension of predictive analytics is predictive diagnostics which combines detection with the context of how equipment operates. Equipment operators can extend run times and maintenance intervals by using predictive analytics as a foundation. The observations, diagnoses, and feedback will then roll up into a total asset management system and bridge major gaps that occur in many reliability programs. Predictive diagnostic methodologies enable equipment owners to extend run times and to decrease maintenance effort with confidence.