MCNP 5.1.51 ENDF/B-VII and other versions - validation for criticality safety evaluation

Nowadays, the assessment of k_eff values for criticality application is usually performed with Monte-Carlo transport codes. The accuracy of the evaluation depends on the code and on the quality of chosen nuclear data. It can be quantified by testing the method on a set of critical experiments where k_eff has been rigorously monitored. The comparison between calculated and measured k_eff allows one to determine a so-called 'calculation bias' which accounts both for systematic errors as well as spreading. This bias is to be added to any k_eff assessed value. In this paper, we will present several MCNP validations for criticality safety applications. It will include a review of results produced using different nuclear libraries (ENDF/B-V, ENDF/B-VI, ENDF/B-VII, JEFF 3.1). The combination of MCNP5.1.51 and ENDF/B-VII, leads to a substantial decrease of the code bias, providing significantly wider margins between the multiplication coefficient of the system (k_eff) and the safety criticality criteria. (authors)