Fuzzy probability based fault tree analysis to propagate and quantify epistemic uncertainty

ASTRACK
A number of fuzzy fault tree analysis approaches, which integrate fuzzy concepts into the quantitative
phase of conventional fault tree analysis, have been proposed to study reliabilities of engineering sys-
tems. Those new approaches apply expert judgments to overcome the limitation of the conventional fault
tree analysis when basic events do not have probability distributions. Since expert judgments might
come with epistemic uncertainty, it is important to quantify the overall uncertainties of the fuzzy fault
tree analysis. Monte Carlo simulation is commonly used to quantify the overall uncertainties of conven-
tional fault tree analysis. However, since Monte Carlo simulation is based on probability distribution, this
technique is not appropriate for fuzzy fault tree analysis, which is based on fuzzy probabilities. The objec-
tive of this study is to develop a fuzzy probability based fault tree analysis to overcome the limitation of
fuzzy fault tree analysis. To demonstrate the applicability of the proposed approach, a case study is per-
formed and its results are then compared to the results analyzed by a conventional fault tree analysis. The
results confirm that the proposed fuzzy probability based fault tree analysis is feasible to propagate and
quantify epistemic uncertainties in fault tree analysis