A fuzzy and Bayesian network CREAM model for human reliability analysis – The case of tanker shipping

Abstract

This paper proposes a quantitative human reliability analysis (HRA) model based on fuzzy logic theory, Bayesian network, and cognitive reliability & error analysis method (CREAM) for the tanker shipping industry. The common performance conditions (CPCs) in conventional CREAM approach are custom-modified to better capture the salient aspects of the situations and conditions for on-board tanker work. Fuzzy logic technique using triangle and trapezoidal membership functions is applied to model the uncertainty and ambiguity of the CPCs as well the control modes in CREAM. A Bayesian network reasoning model using the membership of CPCs as inputs is developed which determines the probability distribution of the control modes. Human error probability (HEP) is obtained from memberships of the control modes and the results of Bayesian network reasoning. A case study in tanker shipping industry with 18 crew members is provided, and the results show that the evaluation of HEP according to the proposed HRA model is very promising and the HRA model is consistent with the original CREAM approach. The sensitivity of the model is also checked against the inputs of the crew members. It is concluded that the enhanced HRA model is able to provide reliable human performance failure results.