ABSTRACT This work calculates the reliability of protective systems of industrial facilities, such as nuclear, to analyze the case of equipment subject to aging, important in the extension of the qualified life of the facilities. By means of the method of supplementary variables, a system of partial and ordinary integral-differential equations was developed for the probabilities of a protective system of an aging channel. The system of equations was solved by finite differences. The method was validated by comparison with channel results with exponential failure times. The method of supplementary variables exhibits reasonable results for values of reliability attributes typical of industrial facilities.

Neste artigo apresenta-se a aplicação de um modelo híbrido, baseado na análise envoltória de dados e em conjuntos fuzzy, para estabelecer uma priorização dos modos de falha levantados numa análise dos modos e efeitos de falha. O objetivo principal é ilustrar a aplicabilidade da modelagem para analisar sistemas complexos, como o sistema de controle químico e volumétrico de uma usina nuclear do tipo água pressurizada, onde uma grande quantidade de modos de falha deve ser avaliada. O modelo apresentado possibilita tratar variáveis linguísticas, i.e., permite que se tratem as opiniões de especialistas de modo mais efetivo num contexto de análise de riscos industriais quando há poucos dados disponíveis, como pode ocorrer em projetos com inovação tecnológica. Os resultados obtidos demonstram o potencial e a viabilidade de se utilizar uma modelagem híbrida para se estabelecer uma ordem de prioridade entre os diferentes modos de falha associados a um sistema complexo.

In this paper one presents a data envelopment analysis and fuzzy sets based hybrid model application, in order to establish a priority ranking for the failure modes evaluated in a failure mode and effect analysis. The main objective is to present the applicability of this hybrid model to analyze complex systems, as the chemical and volumetric control system of a pressurized water reactor power plant, where a great bunch of failure modes must typically be evaluated. The presented modeling makes it possible the consideration of linguistic variables, i.e., it permits that one considers most effectively expert opinion in a risk analysis context when scarce data is available, as may occur in the plant design phase. The results obtained demonstrated the powerfulness and the viability of use of the proposed hybrid modeling to establish a priority ranking among many failure modes concerning a complex system.