Human error probability assessment for LNG Bunkering based on Fuzzy Bayesian Network-CREAM Model

Liquified natural gas (LNG) as a marine fuel has gained momentum as the maritime industry moves towards a sustainable future. Since unwanted LNG release may lead to severe consequences, performing quantitative risk assessment (QRA) for LNG bunkering operations has become mandatory according to some regulations. Human error is a main contributor to the risks, and the human error probabilities (HEPs) are essential for inclusion in a QRA. However, HEPs data are unavailable in the LNG bunkering industry so far. Therefore, this study attempts to infer HEPs through on-site safety philosophical factors (SPFs). The cognitive reliability and error analysis method (CREAM) was adopted as a basic model and modified to make it suitable for HEP assessment in LNG bunkering. Nine common performance condition (CPC) indicators were identified based on the fuzzy ranking of 23 SPF indicators (SPFIs). A Bayesian network (BN) was built to simulate the occurrence probabilities of different contextual control modes (COCOMs), and a conditional probability table (CPT) for the COCOM node with 19,683 possible combinations in the BN was developed according to the CREAMs COCOM matrix. The prior probabilities of CPCs were evaluated using the fuzzy set theory (FST) based on data acquired from an online questionnaire survey. The results showed that the prior HEP for LNG bunkering is 0.009841. This value can be updated based on the re-evaluation of on-site SPFIs for a specific LNG bunkering project to capture the dynamics of HEP. The main innovation of this work is realizing the efficient quantification of HEP for LNG bunkering operations by using the proposed fuzzy BN-CREAM model