Sensitivity and Uncertainty Analysis of Commercial Reactor Criticals for Burnup Credit

The purpose of this study is to provide insights into the neutronic similarities that may exist between a
generic cask containing typical spent nuclear fuel assemblies and commercial reactor critical (CRC) state-
points. Forty CRC state-points from five pressurized-water reactors were selected for the study and the
type of CRC state-points that may be applicable for validation of burnup credit criticality safety
calculations for spent fuel transport/storage/disposal systems are identified. The study employed cross-
section sensitivity and uncertainty analysis methods developed at Oak Ridge National Laboratory and the
TSUNAMI set of tools in the SCALE code system as a means to investigate system similarity on an
integral and nuclide-reaction specific level. The results indicate that, except for the fresh fuel core
configuration, all analyzed CRC state-points are either highly similar, similar, or marginally similar to a
generic cask containing spent nuclear fuel assemblies with burnups ranging from 10 to 60 GWd/MTU.
Based on the integral system parameter, ck , approximately 30 of the 40 CRC state-points are applicable to
validation of burnup credit in the generic cask containing typical spent fuel assemblies with burnups
ranging from 10 to 60 GWd/MTU. The state-points providing the highest similarity (ck>0.95) were
attained at or near the end of a reactor cycle. The ck values are dominated by neutron reactions with major
actinides and hydrogen, as the sensitivities of these reactions are much higher than those of the minor
actinides and fission products. On a nuclide-reaction specific level, the CRC state-points provide
significant similarity for most of the actinides and fission products relevant to burnup credit. A
comparison of energy-dependent sensitivity profiles shows a slight shift of the CRC keff sensitivity
profiles toward higher energies in the thermal region as compared to the keff sensitivity profile of the
generic cask. Parameters representing coverage of the application by the CRCs on an energy-dependent,
nuclide-reaction specific level (i.e., effectiveness of the CRCs for validating the cross sections as used in
the application) were also examined. Based on the CRCs with ck>0.8 and an assumed relative standard
deviation for uncovered covariance data of 25%, the relative standard deviation of keff due to uncovered
sensitivity data varies from 0.79% to 0.95% for cask burnups ranging from 10 to 60 GWd/MTU. As
expected, this uncertainty in keff is largely dominated by noncoverage of sensitivities from major actinides
and hydrogen. The contributions from fission products and minor actinides are very small and comparable
to statistical uncertainties in keff results. These results (again, assuming a 25% uncertainty for uncovered
covariance data) indicate that there could be approximately 1% uncertainty in the calculated application
keff due to incomplete neutronic testing (validation) of the software by the CRCs. However, this
conclusion also assumes all other uncertainties in the complex CRC configurations (e.g., isotopic
compositions of burned fuel, operation history, data) are well known. To accomplish this, an evaluation
and quantification of the uncertainties in the CRC configurations is needed prior to the use of CRCs for
code validation (i.e., quantifying code bias and bias uncertainty).