Paper FUZZY RISK ANALYSIS OF A MODERN -RAY INDUSTRIAL IRRADIATOR F. Castiglia and M. Giardina* Abstract—Fuzzy fault tree analyses were used to investigate accident scenarios that involve radiological exposure to operators working in industrial -ray irradiation facilities. The HEART method, a first generation human reliability analysis method, was used to evaluate the probability of adverse human error in these analyses. This technique was modified on the basis of fuzzy set theory to more directly take into account the uncertainties in the error-promoting factors on which the methodology is based. Moreover, with regard to some identified accident scenarios, fuzzy radio- logical exposure risk, expressed in terms of potential annual death, was evaluated. The calculated fuzzy risks for the examined plant were determined to be well below the reference risk suggested by International Commission on Radiological Protection. Health Phys. 100(6):622– 631; 2011 Key words: accident analysis; accidents, nuclear; emergencies, radiological; nuclear power industry INTRODUCTION THE INTERNATIONAL Commission on Radiological Protec- tion (ICRP) has issued several reports on radiological protection and safety in irradiation facilities, such as those for industrial and agricultural applications, particle physics research laboratories, and radiotherapy treatment centers (ICRP 1990, 1993, 1997, 2000, 2005). Neverthe- less, even though great care is taken in planning system safety to prevent radiological exposure accidents, the system effectiveness is conditional upon the work per- formance of the operator, who may make mistakes even in a well-designed task sequence. In light of the fact that there will always remain a considerable level of risk associated with irradiation facili- ties, this work sought to evaluate the risk of radiation exposure to the system operator. Risk, as it is often defined, is the product of the associated accidental occurrence probability times the consequence; i.e., the absorbed dose. The lack of accurate quantitative human reliability data is seen as a serious limitation to risk assessment and is a major source of uncertainty. Because of this, the authors decided to employ fuzzy set models, which ultimately proved to be well-suited for this work. Fur- thermore, risk analysis was performed using fuzzy fault tree (FT) techniques to further support these findings [FT technique is a method of describing the combination of events leading to a defined system failure, called the top event (TE)]. The hypothesized accidental event was the potential radiological exposure of an operator working in a modern -ray irradiation plant. In these analyses, to account for the uncertainties related to human error, the HEART (Human Error Assessment and Reduction Technique) model (Wil- liams 1986), a first generation HRA (Human Reliabil- ity Analysis) method, was modified to include fuzzy set concepts (Zadeh 1965, 1975). With regard to some identified accident scenarios, an attempt has been made to evaluate the fuzzy potential exposure (ICRP 1993, 1997) in order to estimate the operator radio- logical risk. The obtained fuzzy risk results of the examined accidental event were well below the refer- ence risk suggested by ICRP. FUZZY APPROACH TO HEART The HEART technique, which was derived from a wide range of findings in ergonomics, assumes that any predicted reliability of a task performance may be mod- ified according to the presence of so-called error promot- ing conditions (EPCs). EPCs are the state and importance of various factors that determine and influence the performance of the task. The method identifies nine generic task types and proposes nominal human unreli- ability values, together with 17 EPCs, whose influence on the performance of the task is considered to have the maximum effect (Williams 1986). In the HEART methodology, the failure rate is estimated using an empirical expression of the form: * Department of Nuclear Engineering, University of Palermo, Viale delle Scienze, 90128 Palermo, Italy. For correspondence or reprints contact: For correspondence con- tact F. Castiglia at the above address, or email at castiglia@din.unipa.it. (Manuscript accepted 12 October 2010) 0017-9078/11/0 Copyright © 2011 Health Physics Society DOI: 10.1097/HP.0b013e31820153eb 622 www.health-physics.com