An efficient method for fault tree analysis to support nuclear plant operation decision-making

Fault tree analysis (FTA) is extensively used in the probabilistic risk assessment of nuclear power plants. One application of FTA is to support nuclear plant operation decision-making. In a nuclear plant, when the availability of components changes, that will affect the basic event probabilities in the fault tree of the plant. A fast and accurate evaluation of the fault tree to update the plant risk, represented by the top event probability in the fault tree, is essential to make adequate operation decisions. However, this task is challenging when the fault tree is large and complex. In this paper, a new method is introduced to efficiently update the top event probability without recalculating from scratch when basic event probabilities change. This method involves an offline phase where a set of scenarios is generated via importance sampling. Each scenario includes a sequence of basic event states and the associated weight, which accounts for importance sampling. In the online phase, only the stored weights need to be adjusted to reflect changes in basic event probabilities, allowing fast updating of the top event probability. A case study using synthetic fault trees demonstrates the proposed method’s effectiveness in achieving fast and accurate top event probability updating. Relative errors remain low and confidence intervals for given confidence levels remain narrow when the increase in the basic event probability is under one order of magnitude, though they increase or become wider when the increase in the basic event probability becomes larger, highlighting areas for future improvement. • An integrated offline–online approach to updating top event probability in fault tree analysis. • Reuse of sample scenarios from importance sampling to reduce computation time. • Capability of quantifying confidence in the updated top event probability. • Case study shows substantial computation time reduction.