Vibration-Induced Fatigue Failure Monitoring in Industrial Pipes Using Paraconsistent Annotated Logic with Annotation of Two Values Jose Jefferson M. de Oliveira, Carlos Eduardo A. L. Rodrigues, João Inácio da Silva Filho

Abstract

This paper presents some concepts of vibration in piping and monitoring techniques to measure the risk of fatigue failure, according to methodologies proposed by the Energy Institute (EI) and International Organization for Standardization (ISO) guidelines. Based on these two sources of information, a way to quantify these risks is investigated using the basic concepts of Paraconsistent Annotated Logic with Annotation of Two Values (PAL2v). Subsequently, from real vibration measurement data in piping of a compression unit (for different operating conditions), algorithm configuration was implemented composing a Paraconsistent Analysis Network (PANet), and the Real Certainty Degree Value was determined. The results obtained confirmed the possibility of applying this type of logic in neural networks aimed at predicting failures in piping due to vibration-induced fatigue (VIF).