Criteria and margins in nuclear criticality safety as applied in Sweden

Some of the criteria and margins that are applied in Sweden, differ from those applied in some other countries. Two unlikely, concurrent events must not lead to criticality. The small, sometimes negative bias corrections, established by comparison of computer calculations to experiments that are applied in other countries, are in most cases not judged as sufficient. Statistical uncertainties in the estimated multiplication factor from calculations (Monte Carlo), mechanical variations and other sources are restricted in that the standard deviation normally must be less than 0.01. Mixed arrays of dissimilar units must be evaluated in combination, not separately as this is not generally considered safe. Unrealistic (conservative) assumptions concerning some parameters in the safety analysis should not be used to cover errors and uncertainties in other parameters