AVALIAÇÃO DA CONFIABILIDADE UTILIZANDO MODELO HÍBRIDO: ESTUDO DE CASO NA INDÚSTRIA SUCROALCOOLEIRA

Abstract

The main objective of this dissertation is to develop a reliability assessment applied to reparable systems in the sugar and alcohol industry. The work has a quantitative approach with a combined research method: simulation modeling with case-study. Therefore, a dynamic modeling is used to develop a case-study. For this, a Non- Homogeneous Markov Process (PNHM) model is used to represent the states transitions (operation-maintenance) of sugar centrifuges, in order to obtain reliability indicators. Integrated to the PNHM model, it will be used the Generalized Renewal Process (PRG) to describe the probabilities of failures considering imperfect repairs. The PRG is defined according to the parameters of scale, shape and rejuvenation (α, β, q), which are estimated by Bayesian Inference. In the development of the case-study, the data collection of centrifugal failures was performed and, through the simulation of the hybrid model, the case-study was conceived. The estimated availability for 5640 hours was 93.23%, whose the other 7% represents maintenances moments. For 323 hours of simulation, the reliability resulting was 50%, demonstrating high system susceptibility to failures. The results suggest the revisions of the intervals defined for preventives maintenances, which should be applied to the most observed failure mode, called malfunction of the sensing systems.