Risk-Informed Life Extension Decisions Using Unstructured Maintenance Data

Life extension (LE) decision of complex, long-life, capital-intensive assets is a challenge due to availability of structured maintenance data. Data is often unstructured and text-based, making it difficult to perform systematic reliability assessments useful for assessing the remaining useful life (RUL). This study presents a methodology for addressing those challenges by using Natural Language Processing (NLP), NESTOR. Historical maintenance records (2012–2017) from the University of North Dakota (UND) Aviation Academy were analyzed for, three critical engine components: intake gaskets, rocker cover gaskets and baffles in single and twin-engine fleet. Failure Mode and Effect Analysis (FMEA) was conducted for the identification and prioritization of intervention options for failure modes based on likelihood and consequences. The results demonstrate the potential of NLP techniques to support reliability analysis for RUL required for risk informed LE decisions of capital-intensive long life complex assets.