A weighted CREAM model for maritime human reliability analysis

Human error can be regarded as a significant factor contributing to marine accidents. Crew onboard vessels often perform duties in circumstances where technological, environmental and social factors emerge which may contribute to the occurrence of human failures. Fuzzy Cognitive Reliability and Error Analysis Methods (CREAM) is one of the most recognized HRA methods capable of tackling such difficulties. However, shortcomings are still disclosed and weaken the applicability of such an approach. These include the lack of considering input weights, the dubitation of the logicality of adopting rule base approaches to evaluate the relations between inputs and output and the loss of useful information due to the application of min-max fuzzy inference method. A new fuzzy CREAM methodology capable of resolving the aforementioned difficulties is proposed based on a rule base approach. The framework is validated using two axioms and demonstrated by virtue of an oil tanker example. The results are consistent with the principles evolved from the axioms since the outcomes are sensitive to the minor alterations of input data and weights. It is concluded that the weighted CREAM model is able to produce reliable human performance failure results and the strengths will not be compromised even if applied in circumstances where membership function shapes of fuzzy sets are various from traditional studies. (C) 2014 Elsevier Ltd. All rights reserved.