A method to evaluate the effect of human factors on probabilistic safety analysis (PSA) is developed. The main features of the method are as follows:

  1. A time-dependent multibranch tree is constructed to treat time dependency of human error probability.
  2. A sensitivity analysis is done to determine uncertainty in the PSA due to branch time of human error occurrence, human error data source, extraneous act probability, and human recovery probability.
The method is applied to a large-break, loss-of-coolant accident of a boiling water reactor-5. As a result, core melt probability and risk do not depend on the number of time branches, which means that a small number of branches are sufficient. These values depend on the first branch time and the human error probability.