Some models using Boolean algebra like nuclear PSA models are very detailed and, when quantified, result in an enormous amount of Minimal Cut Sets (MCS). The negligible contributions (non-significative MCS) have to be suppressed by a cut off process to respect computing limitations and to have a reasonable quantification time. When an acceptable level of cutoff uncertainty for the baseline risk is obtained with common cutoff thresholds, the final cut set equation is only useable to study small changes in the occurrence probability of basic events. So this set of MCS may be inadequate to estimate importance measures like the Birnbaum indicator or RAW. The truncation process has to be adapted to obtain a baseline risk set of MCS adapted to any importance measure calculation.

This paper summaries and comments existing truncation processes. Then we propose a new truncation process that enables us to generate a set of MCS as small as possible and adapted to estimate all importance measures with a good level of accuracy for any elementary event in the model.