A fuzzy modeling application of CREAM methodology for human reliability analysis

This paper is using a fuzzy classification system for human reliability analysis in order to calculate the probability of erroneous actions according to CREAM in specific contexts e.g. maintenance tasks, in-field actions or control room operations in the running of a chemical plant. The complexities of such a system as well as the appropriate actions that have to be taken into consideration by the developers are analysed in detail. CREAM methodology has been selected among the most known and used methods for Human Reliability Analysis not only for being well-structured and precise, but also because it fits better in the general structure of the fuzzy logic. The first results from the application of the model are very promising and in accordance with CREAM. The numerical values produced by the model can be further used in the 'classical' risk assessment methods, such as event trees and fault trees, in order to calculate the overall occurrence frequency of a particular hazardous event. This is a pilot application that demonstrates the successful 'translation' of CREAM into a fuzzy logic model. However, the same model could be extended to cover also other industrial fields, like aviation technology and maritime transports. (c) 2005 Published by Elsevier Ltd.