Reliability Sensitivity analysis

How to estimate the New Reliability Sensitivity analysis, based on integral domain modification in Borgonovo indices integral, proposed by Florian Schmid.

In this thesis, section 3.3.1, equation 60 and 61, the summation of denominator equals one since we are making histogram based PDF of model responses belonging to the failure domain. However, according to the definition, it should equal to Pf.

Kindly clear this doubt. I am performing importance sampling to sample from the failure domain and applying algorithm suggested in section 3.3.1. I hope I am doing this right (considering samples only from failure domain and ignoring samples which give safe according to limit state)