APPROXIMATE TECHNIQUES FOR REPRESENTING NUCLEAR DATA UNCERTAINTIES

摘要

Computational tools are available to utilizernsensitivity and uncertainty (S/U) methods for a widernvariety of applications in reactor analysis and criticalityrnsafety. S/U analysis generally requires knowledge of thernunderlying uncertainties in evaluated nuclear data, asrnexpressed by covariance matrices; however, only a fewrnnuclides currently have covariance information availablernin ENDF/B-VII. Recently new covariance evaluationsrnhave become available for several important nuclides, butrna complete set of uncertainties for all materials needed inrnnuclear applications is unlikely to be available for severalrnyears at least. Therefore if the potential power of S/Urntechniques is to be realized for near-term projects inrnadvanced reactor design and criticality safety analysis, itrnis necessary to establish procedures for generatingrnapproximate covariance data. This paper discusses anrnapproach to create applications-oriented covariance datarnby applying integral uncertainties to differential datarnwithin the corresponding energy range.