Nuclear Engineering and Design 236 (2006) 1256–1263 Hazard and operability study using approximate reasoning in light-water reactors passive systems Antonio C.F. Guimar˜ aes ∗ , Celso Marcelo Franklin Lapa Instituto de Engenharia Nuclear (CNEN)-Divis˜ ao de Reatores/Coordena¸ c˜ ao de Ensino, Ilha do Fund ˜ ao s/n, P.O. Box 68550, Rio de Janeiro 21945-970, Brazil Received 22 July 2005; received in revised form 24 November 2005; accepted 24 November 2005 Abstract In this paper, a risk evaluation approach is applied on Hazard and Operability Study (HAZOP). First, the concept of the traditional failure mode and effects analysis (FMEA) for the risk priority number (RPN) has been adapted and applied to HAZOP study. Then, a HAZOP rpn concept was created. The HAZOP rpn enables evaluation of the risk level over the system caused by an abnormality on the physical phenomena. The approach created from the fusion between the HAZOP and the traditional RPN has been named hybrid HAZOP. Following this, a fuzzy inference was established over hybrid HAZOP system defining another approach to safety assessment and risk estimation. As an example case, a passive system designed for removing the decay heat of an advanced light-water reactor, was used. The set of results demonstrated the great adherence of the hybrid HAZOP approach in this class of problems. It endorses also the advantages of using of a fuzzy inference system to model the uncertainty parameters levels in risk analysis. © 2005 Elsevier B.V. All rights reserved. 1. Introduction Failure mode and effects analysis (FMEA) is an important technique (Stamatis, 1995) used to identify potential failure modes and to estimate their consequences on complex engi- neering systems. This approach provides information to support decisions that may lead to plant risk decrease. Recently, some FMEA studies (Pillay and Wang, 2002) have generated an index, called risk priority ranking (RPN). It is produced in order to iden- tify the main failure modes. The traditional RPN is obtained by the aggregation of the some safety factors. According to many authors, the traditional RPN considers: • the probability of failure occurrence; • the failure severity to the system; • the probability of a failure remains unknown (detectability). The FMEA approach has been used successfully on many nuclear risk studies. However, the more recent nuclear power ∗ Corresponding author. Tel.: +55 21 2209 8249; fax: +55 21 2209 8259. E-mail addresses: tony@ien.gov.br (A.C.F. Guimar˜ aes), lapa@ien.gov.br (C.M.F. Lapa). plants generations contain safety systems, to which the FMEA boarding presents some restrictions. FMEA approach is per- formed at component level, but the failure modes in passive safety systems are, fundamentally, related to physical basic parameters (e.g. flow, temperature). Generally, passive sys- tems reliability; studies should be analyzed from two aspects. First, the component reliability, there are a lot of method- ologies applicable in this context, the FMEA is one among other. The other one is reliability centered on physical phe- nomena. This second aspect is concerned with the way the natural physic phenomena operate. In this case a qualitative analysis, associated with the passive system critical parame- ters, is required. A well-structured procedure commonly used in this situation is the Hazard and Operability Study (HAZOP) approach. Considering the RPN importance to risk analysis and that the HAZOP approach is a very suitable approach to passive systems reliability assessment, a hybrid HAZOP method has been developed, creating consequently a new risk ranking: the HAZOP rpn. An approach using fuzzy rule base and Grey rela- tion theory is also proposed in this paper. The fuzzy rule basis is used to aggregate the HAZOP rpn safety factor and generate a “fuzzy risk”, as in the RPN FMEA pioneer papers (Xu et al., 2002; Pillay and Wang, 2003; Guimar˜ aes and Lapa, 2004). 0029-5493/$ – see front matter © 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.nucengdes.2005.11.007