Impact of nuclear data revised from JENDL-4.0 to JENDL-5 on PWR spent fuel nuclide composition

ABSTRACT The burnup calculations for estimating the nuclide composition of the spent fuel are highly dependent on nuclear data. Many nuclides in the latest version of the Japanese Evaluated Nuclear Data Library JENDL-5 were modified from JENDL-4.0 and the modification affects the burnup calculations. This study confirmed the validity of JENDL-5 in the burnup calculations. The PIE data of Takahama-3 was used for the validation. The effect of modifications of the parameters, e.g. cross sections and fission yields, from JENDL-4.0 to JENDL-5 on the nuclide compositions was quantitatively investigated. The calculation results showed that JENDL-5 has a similar performance to JENDL-4.0. The calculation results also revealed that the modifications of the cross sections of actinide nuclides, fission yields, and thermal scattering low data of hydrogen in H2O affected the nuclide compositions of PWR spent fuels.


Introduction
The burnup calculation estimates the nuclide composition of irradiated nuclear fuels and related quantities such as radioactivity.This calculation requires various nuclear data, e.g. the nuclear reaction cross section, the fission yield, and the decay data such as half-lives and decay branching ratio.The nuclear data should be validated in burnup calculations with the post-irradiation examination (PIE) data [1]] for the safety evaluation of the nuclide composition in the spent fuel.PIE data allow the validation of nuclear data that have a large impact on the nuclide composition of nuclear fuel.PIE data are particularly effective for the validation of nuclear data that are difficult to validate with criticality experiments, such as cross sections of minor actinide and fission product (FP), fission yields, and decay data.
The latest version of the Japanese Evaluated Nuclear Data Library JENDL-5 was released at the end of December 2021 [2,3].Nuclear data of many nuclides that have a particular impact on burnup calculations were modified in JENDL-5.Cross sections for many actinides were revised by modifying resonance parameters and adopting new cross section measurement data.Fission yields and decay data were thoroughly reviewed.In addition, JENDL-5 adopts newly evaluated thermal scattering laws (TSL) data.TSL data of H in H 2 O affect the thermal neutron spectrum in light water-moderated systems.The change in the thermal neutron spectrum affects the results of burnup calculations for light water reactors (LWRs).
Although JENDL-5 has been validated by benchmark calculations and experimental analyses [4][5][6][7][8][9], validation with PIE data has not been performed yet.For example, burnup calculations for a simple LWR fuel were carried out [6].These burnup calculations mainly focused on the effect of nuclear data on the criticality through burnup.In addition, only the modification of cross-section data from JENDL-4.0 to JENDL-5 was considered, and the modification of fission yields and decay data was not considered in these burnup calculations.In other words, the impact of JENDL-5 nuclear data, including fission yields and decay data, on the nuclide composition of the LWR fuel has not been sufficiently investigated.
This paper validates JENDL-5 for the burnup calculation for LWR.The nuclide compositions of PWR spent fuel estimated with JENDL-5 were compared to those with JENDL-4.0 [10].The effect of modifying nuclear data from JENDL-4.0 to JENDL-5 was also investigated.We investigated the effect of the modifications of cross sections, TSL data, fission yields, and decay data to investigate the cause of the difference in the nuclide composition.
The PIE data of PWR Takahama-3 was used as the calculation target [11].The PIE data of Takahama-3 has been used to validate various calculation codes and nuclear data libraries [12][13][14].We chose PWR as the target LWR because the calculation conditions of PWR are simpler than those of BWR.BWR has a more complex assembly structure, many types of fuels with different enrichment, axial enrichment distribution of the fuel, void distribution in the coolant, and complexity of operating conditions such as control rod insertion.These complexities of BWR increase the uncertainty of the calculation results.Therefore, the present study focuses on PWR to more clearly investigate the effect of JENDL-5 on the nuclide compositions.The validation of JENDL-5 with PIE data of BWR is a future issue.
The remainder of this paper is organized as follows: Section 2 presents an overview of Takahama-3 PIE samples; Section 3 outlines the specific calculation conditions and methods; Section 4 shows the calculation results of C/E-1 and discusses the effects of nuclear data revised from JENDL-4.0 to JENDL-5; Section 5 summarizes concluding remarks.

Overview of Takahama-3 samples
Takahama-3 is a Japanese PWR operated by the Kansai Electric Power Company, Inc.The nuclide composition measurement data of 16 samples were obtained from three fuel rods (SF95, SF96, and SF97) irradiated in Takahama-3 [11].SF95 and SF97 are the UO 2 fuel, and SF96 is the gadolinia fuel.SF97 and SF96 were used in this study.SF95 was omitted because the fuel rod specifications are similar to SF97, while the average burnup is lower, and the number of measured nuclides is fewer than SF97.Table 1 shows an overview of SF97 and SF96, and Table 2 shows the operating histories.SF97 and SF96 were loaded in the different assemblies and irradiated through three and two operation cycles, respectively.Six and five samples were obtained from SF97 and SF96, respectively.SF97-1 and SF96-1 samples were excluded from the present calculation because the sample position of these samples is close to the top of the active core.The neutron current from the axial direction in these sample regions is larger than that in the other sample regions.For these reasons, Ref. 11 considered that these samples are unsuitable as a benchmark.
JENDL-4.0 and JENDL-5 were used as nuclear data libraries.JEDNL-5 has been updated several times since its release in December 2021 [22].In this study, JENDL-5 reflecting the updates as of June 1, 2022 was used.For JENDL-4.0, all updated files (JENDL-4.0u) [23] were adopted.Since JENDL-4.0 does not include the decay data, the JENDL Decay Data File 2015 (JENDL/DDF-2015) [24] was used as the decay data of JENDL-4.0.
Serpent 2 uses the ACE file [25] as a cross-section library.The ACE files based on JENDL-4.0 and JENDL-5 were generated using the nuclear data processing code FRENDY (ver.2.00) [26,27].The TSL data of water for hydrogen (H in H 2 O) and oxygen (O in H 2 O) were newly evaluated in JENDL-5.Both TSL data were used in this study.In JENDL-5, cross sections of the oxygen isotopes O-17 and O-18 are newly evaluated, but only O-16 is considered in this study for simplicity.
Serpent 2 can utilize the decay data and fission yields for the burnup calculation from the ENDF-6 format files [28].This feature facilitates using the decay data and fission yields of JENDL-5.In this study, the fission yields and decay data of JENDL-5 in the ENDF-6 format were partially modified because some data prevented the successful execution of Serpent 2 or produced unreasonable calculation results.This modification is explained in Table S1 in the Supplemental Online Material.
The neutron flux of each fuel pin was calculated by considering the 1/8 symmetry of the fuel assembly configuration.Figure 1 shows the fuel assembly configuration and the positions of the SF97 and SF96 fuel rods on the 1/8 symmetrical arrangement [11].Note that the assembly configuration of SF96 and SF97 is identical.The tally region of the gadolinia fuel was divided into eight regions with an equal volume along the radial direction to account for the spatial dependence of the nuclide composition [29].The reflective boundary conditions were adopted for both horizontal and axial directions.Table 3 shows the assembly geometry used in the present calculations [11].The gap between the fuel and the cladding was filled by the cladding, and the density was adjusted to maintain the mass of the cladding [30].Table 4 shows the axial position of each sample and coolant temperatures estimated according to the axial position [11,31].The temperature of the TSL data (H and O in H 2 O) was given as the temperature closest to the moderator temperature from those provided in JENDL-4.0 and JENDL-5, respectively.The fuel and cladding temperatures were set to 900 K and 600 K, respectively.Boron concentration in the coolant was assumed to be average values for each irradiation cycle (5 th cycle: 682 ppm, 6 th cycle: 680 ppm, and 7th cycle: 629 ppm) [11].Table 5 shows the densities, enrichments, and nuclide compositions of fuel and cladding used in the present calculations [11].
The number of neutron histories, active cycles, and inactive cycles in Monte Carlo calculations of Serpent 2 were 10,000 histories per cycle, 1,000 cycles, and 100 cycles, respectively.
The burnup value of each sample was calibrated based on the burnup values and the power history listed in Ref. 11.The calibration of the burnup values is explained in Section 4.1.The power history normalized by the calibrated burnup values was used for the burnup step.The burnup step was set to be less than 0.1 GWd/t and 0.5 GWd/t in the initial and subsequent steps, respectively, by dividing the irradiation periods.The predictor-corrector method [32] was applied at each burnup step to improve the calculation accuracy of nuclide compositions at each burnup step with a reasonable calculation cost.

Calibration of burnup value
We calibrated the burnup value of each sample based on the Nd-148 composition.The burnup values of SF97 and SF96 samples were evaluated by Nd-148 method [11].In the Nd-148 method, the burnup value is calculated using the measured number density of Nd-148 and nuclear data such as the fission yield of Nd-148.This means that the burnup value varies with the nuclear data library used.In this subsection, we calibrated the burnup value using JENDL-4.0 and JENDL-5, respectively.
At first, in order to calculate the number density of Nd-148 before the calibration, we performed burnup calculations for each sample using the burnup values and the power history listed in Ref. 11.The burnup values and the C/E-1 values of Nd-148 are shown in Table 6.The abbreviations J4 and J5 in Table 6 refer to JENDL-4.0 and JENDL-5, respectively, and they are also used in the following table and figures.As shown in the 'Before calibration' column in Table 6, there were slight differences in the Nd-148 number density between the   6, the calibrated burnup values of JENDL-5 are lower by about 0.1% than those of JENDL-4.0.This comes from the difference in the fission yield of Nd-148 between JENDL-4.0 and JENDL-5.The cumulative fission yield of Nd-148 at the thermal energy of U-235 and Pu-239 updated by about +0.30% and −0.05% from JENDL-4.0 to JENDL-5, respectively.This modification of the Nd-148 fission yield increases the production of Nd-148 per fission in a typical UO 2 fuel.It contributes to a decrease in the burnup value corresponding to the production of Nd-148.S2 and Table S3 in the Supplemental Online Material.

Average C/E-1 value of nuclide compositions
According to the C/E-1 values shown in Figure 2, nuclide compositions calculated with JENDL-5 are similar to those of JENDL-4.0 in the burnup calculation of PWR.As shown in Figure 2(a), the C/E-1 values of SF97 were within ±10% for both JENDL-4.0 and JENDL-5, except for Am-242 m, Cm-242, Cm-246, Cm-247, Sb125, and Ru106.Although the C/E-1 values for these nuclides significantly deviated from zero, similar results have been confirmed in previous studies [12][13][14].The reason may be that the experimental uncertainties were underestimated or there are common issues in the current  Regarding the difference in the C/E-1 values between JENDL-4.0 and JEDNL-5, there are relatively large differences exceeding ±5% for Pu-238, Am-241, Cm-245, Cm-246, and Cm-247 among actinide nuclides.The large differences exceeding ±5% were also confirmed for Cs-134, Sb-125, Sm-151, and Sm-152 among FP nuclides.The cause of these differences is investigated and discussed in Section 4.3.

Effect of the differences in cross section, fission yield, and decay data between JENDL-4.0 and JENDL-5 on nuclide compositions
To investigate the causes of the larger differences in nuclide compositions between JENDL-4.0 and JENDL-5, each of the following three data were individually replaced from JENDL-4.0 to JENDL-5: (a) ACE files of nuclear reaction cross-section data including TSL data, (b) fission yield data, and (c) decay data such as half-life and branching ratio.These investigations were carried out for the SF97-5 sample, which has the highest burnup among all the samples.Figure 3 shows the relative differences in number densities due to the change of each  nuclear data from JENDL-4.0 to JENDL-5. Figure 3 shows the following four cases: (a) only cross section data is replaced (XS), (b) only fission yield data is replaced (FY), (c) only decay data is replaced (DEC), and all the nuclear data is replaced (ALL).Note that the gray bars in Figure 3 indicate the result of ALL. Figure 3 also shows statistical uncertainties of number densities by the Monte Carlo calculations as the error bar for the ALL case.The statistical uncertainty was estimated as a standard deviation by calculating 10 times with different initial random numbers [34].As shown in Figure 3, the statistical uncertainties are sufficiently small compared to the differences due to the nuclear data change.
As shown in Figure 3, the differences in the number densities of actinide nuclides and Cs-134 of the ALL case are almost identical to those of the XS case.This means that the differences are due to the cross-section change.This section discusses the effect of the crosssection change for Cs-134.The effect of the crosssection change for actinide nuclides is investigated and discussed in Section 4.4.
Cs-134 is mainly produced by the capture reaction of Cs-133 in irradiation [35].Figure 4 shows the Cs-133 capture cross section between JENDL-4.0 and JENDL-5.The microscopic cross section in Figure 4 uses the SCALE 238-group structure [36] and is energy-averaged with a constant weight.Figure 4 shows that the Cs-133 capture cross section of JENDL-5 is higher than JENDL-4.0 in the thermal energy region.This cross-section difference in the thermal energy region contributes to increasing the number density of Cs-134.We confirmed that the total number of Cs-133 capture reactions through the burnup calculation of the XS case was about 4.5% higher than that of the case based on JENDL-4.0.This increase in Cs-133 capture reactions corresponds to the rise of about 4% in the Cs-134 number density of the XS case.
The result of the FY case in Figure 3 shows that the update of fission yields from JENDL-4.0 to JENDL-5 affects the number densities of FP nuclides, especially for Sb-125, Sm-151, and Sm-152, where the relative differences are larger than ±5%.We compared the fission yields to discuss the cause of these results.Table 7 shows the cumulative and effective yields between JENDL-4.0 and JENDL-5 for nuclides of major contributors to fission reaction (U-235, U-238, Pu-239, and Pu-241).The effective fission yield in Table 7 is a sum of the cumulative yields with the weight of the ratio of the number of fissions to the total fissions through the burnup calculation.The ratio is 0.56, 0.07, 0.31, and 0.06 for U-235, U-238, Pu-239, and Pu-241, respectively.The cumulative yields of U-238 are for fast neutron fission, and the cumulative yields of the other nuclides are for thermal neutron fission.As shown in Table 7, the effective yield for Sb-125 from JENDL-5 is lower than that from JENDL-4.0 due to a significant reduction of the cumulative yield of U-235.This decrease in the effective yield causes a reduction in the number density of Sb-125.The cumulative yields for Sm-151 of all four nuclides significantly increased in JENDL-5, which increases the number density of Sm-151.The cumulative yields for Sm-152 are almost unchanged between JENDL-4.0 and JENDL-5, although the number density of Sm-152 significantly changed by replacing the fission yield data.The main production paths of Sm-152 are from fissions, including fissions to upstream nuclides in the β-decay series, and capture reactions of [35].Since the cumulative yields of Sm-151 changed significantly from JENDL-4.0 to JENDL-5, the number density of Sm-152 also changed by replacing the fission yield data.
On the decay data, no significant differences in number density are observed in the result of the DEC case in Figure 3.The present result clarifies that the effect of the decay data modification from JENDL-4.0 to JENDL-5 on the target nuclide compositions is negligible for the burnup calculations of PWR.

Effect of cross section and TSL data modifications
This section investigates the effect of the cross section and the TSL data modification in JENDL-5 on the nuclide composition of actinide nuclides.The cross section of actinide nuclides and the TSL data of H and O in H 2 O was individually replaced.The SF97-5 sample was used for this investigation.Table 8 shows the relative differences in the number density of actinide nuclides.Each column in this table represents the nuclide whose cross section or TSL data were changed from JENDL-4.0 to JENDL-5.Each row in this table represents the relative difference in the number density of actinide nuclides for SF97-5 before and after the cross-section data replacement from JENDL-4.0 to JENDL-5.The bold italic data with gray background mean that the relative difference is larger than ±2%.The cross section was replaced for all the target actinide nuclides except for those whose capture or fission cross section does not change significantly.For the H in H 2 O and O in H 2 O cases, only the TSL data was replaced to JENDL-5, and the H-1 or O-16 cross section was not replaced.
Table 8 shows that the difference in cross section of U-235, Pu-238, Pu-242, Am-241, and Cm-244 significantly affects the actinide nuclide compositions, i.e., the resulting difference in the number density of the actinide nuclides is larger than ±2%.We compared the cross sections of these nuclides to clarify the effect on the actinide nuclide compositions.Figure 5 shows the crosssection differences of these nuclides between JENDL-4.0 and JENDL-5.
The cross-section difference of U-235 affects the number densities of U-235 and U-236.As shown in Figure 5(a-b), the fission and capture cross sections of U-235 from JENDL-5 are higher than those from JENDL-4.0 in the thermal energy region.These increases in the fission and capture cross sections reduce the U-235 number density.In addition, the U-236 number density increases due to increased U-235 capture reactions.
The cross-section change of Pu-238 almost causes the difference in the Pu-238 number between JENDL-4.0 and JENDL-5.According to Figure 5(c), the Pu-238 capture cross section in the thermal and resonance energy region of JENDL-5 is considerably higher than that of JENDL-4.0, which causes a significant decrease in the number density of Pu-238.shows that the capture cross section of Pu-242 from JENDL-5 is higher than that from JENDL-4.0 in the thermal and resonance region.This crosssection difference of Pu-242 affects the increase of the number densities for nuclides produced via capture reactions of Pu-242, such as Am-243 and Cm isotopes.
The difference in number density for Am-241 is almost caused by the cross-section change of Am-241.Figure 5(e) shows that the capture cross section of Am-241 from JENDL-5 increases around 1 eV compared to JENDL-4.0.This capture cross-section difference decreases the number density of Am-241 in the present calculation.For the other actinide nuclides, the present calculation clarified that the change of the cross section from JENDL-4.0 to JENDL-5 has no significant effect on the target nuclide compositions.
Table 8 also shows that the change in the TSL data of H in H 2 O from JENDL-4.0 to JENDL-5 significantly reduced the number density of actinide nuclides, especially Cm isotopes.Figure 6 shows the comparison of the incoherent inelastic scattering cross sections of H in H 2 O with JENDL-4.0 and JENDL-5.Figure 6 shows that the incoherent inelastic scattering cross section of JENDL-5 is lower within 0.01-1 eV and higher below 0.01 eV than JENDL-4.0.
We compared the energy spectrum ratio of the thermal neutron flux (with the energy<0.625eV) to the total neutron flux to investigate how this difference in the thermal scattering cross section affects the calculation.The initial burnup step of the SF97-5 sample was used to calculate the energy spectrum ratio.The JENDL-4.0 library was used as the calculation and the TSL data of H in H 2 O was only modified from JENDL-4.0 to JENDL-5.As a result, the energy spectrum ratio of JENDL-4.0 and JENDL-5 were 0.13135 and 0.13086, respectively, where the statistical uncertainties of the energy spectrum ratio were approximately 0.00004 for both results.Namely, the neutron spectrum becomes harder using the TSL data of H in H 2 O from JENDL-5.This spectrum change seems to complicatedly affect the nuclear reactions of nuclides that have a significant sensitivity of the cross section to the number densities of Cm isotopes (i.e., U-235, U-238, Pu-239, Pu-240, Pu-241, Pu-242, Am-243, Cm-244, and Cm-245 [35]).Detailed sensitivity analyses are necessary for further discussion on the effect of modification in the TSL data of H in H 2 O.
As shown in Table 8, the present calculation clarifies that the effect of the TSL data of O in H 2 O on the nuclide compositions of actinide nuclides is negligible.

Conclusion
To validate JENDL-5 in the burnup calculation, the PIE calculations for the PWR samples were carried out using JENDL-4.0 and JENDL-5 and the C/E-1 values of the nuclide compositions were compared.The samples of fuel rods of SF97 and SF96 of Takahama-3 were chosen as the calculation target.The result showed that the averaged C/E-1 values were within ±10% for many of the target nuclides with JENDL-5.A similar result was also obtained with JENDL-4.0.Namely, the numerical result of the present calculation using JENDL-5 was equivalent to JENDL-4.0 from the viewpoint of validation using Takahama-3 PIE data.
The difference larger than ±5% in number density between JENDL-4.0 and JENDL-5 was observed for following nuclides: Pu-238, Am-241, Cm-245, Cm-246, Cm-247, Cs-134, Sb-125, Sm-151, and Sm-152.Additional burnup calculations by replacing the nuclear data partially from JENDL-4.0 to JENDL-5 revealed the following: • The Pu-238 number density was estimated as smaller using JENDL-5 because of the increased capture cross sections of Pu-238 in the thermal and resonance energy region.• The Am-241 number density was estimated as smaller using JENDL-5 because of the increased capture cross sections of Am-241 in the resonance region.For the feedback to the nuclear data evaluation based on burnup calculations of PIE data, a further detailed evaluation of the nuclear data contribution (for example, burnup sensitivity analysis to cross sections of each neutron energy group [37]) is necessary as a future study.

Figure 1 .
Figure 1.Configuration of the fuel assembly on the 1/8 symmetrical arrangement.

Figure 2
Figure 2 shows the sample average of C/E-1 values of nuclide compositions obtained using JENDL-4.0 and JENDL-5.Error bars in Figure 2 indicate standard deviations as the variation of C/E-1 values depending on samples.The gray bars indicate the maximum uncertainties of the measured values given in the report[11].The raw data of C/E-1 values for each sample are shown in TableS2and TableS3in the Supplemental Online Material.According to the C/E-1 values shown in Figure2, nuclide compositions calculated with JENDL-5 are similar to those of JENDL-4.0 in the burnup calculation of PWR.As shown in Figure2(a), the C/E-1 values of SF97 were within ±10% for both JENDL-4.0 and JENDL-5, except for Am-242 m, Cm-242, Cm-246, Cm-247, Sb125, and Ru106.Although the C/E-1 values for these nuclides significantly deviated from zero, similar results have been confirmed in previous studies[12][13][14].The reason may be that the experimental uncertainties were underestimated or there are common issues in the current calculation condition or model.In the case of SF96 shown in Figure2(b), the C/E-1 values of some nuclides, such as Cm-242, were improved and those of some other nuclides, such as Np-237, became worse for both JENDL-4.0 and JENDL-5, compared to SF97.

Figure 2 .
Figure 2. Average of C/E-1 values over samples for SF97 and SF96.

Figure 3 .
Figure 3. Relative difference in the number density of SF97-5 due to the nuclear data replacement from JENDL-4.0 to JENDL-5.XS: only cross sections (ACE files including TSL data) are replaced.FY: only fission yields are replaced.DEC: only decay data are replaced.All: all nuclear data (cross sections, fission yields, and decay data) are replaced.

*
The result of All nuclides represents the result of XS in Figure3.

Figure 5 (
Figure 5(d)  shows that the capture cross section of Pu-242 from JENDL-5 is higher than that from JENDL-4.0 in the thermal and resonance region.This crosssection difference of Pu-242 affects the increase of the number densities for nuclides produced via capture reactions of Pu-242, such as Am-243 and Cm isotopes.The difference in number density for Am-241 is almost caused by the cross-section change of Am-241.Figure5(e) shows that the capture cross section of Am-241 from JENDL-5 increases around 1 eV compared to JENDL-4.0.This capture cross-section difference decreases the number density of Am-241 in the present calculation.

Figure 5 (
Figure 5(f) shows the capture cross-section change of Cm-244.As shown in Table 8, this cross-section difference significantly affects the number densities of Cm-245, Cm-246, and Cm-247.For the other actinide nuclides, the present calculation clarified that the change of the cross section from JENDL-4.0 to JENDL-5 has no significant effect on the target nuclide compositions.Table8also shows that the change in the TSL data of H in H 2 O from JENDL-4.0 to JENDL-5 significantly reduced the number density of actinide nuclides, especially Cm isotopes.Figure6shows the

•
The number densities of Cm-245, Cm-246, and Cm-247 were estimated smaller using JENDL-5 because of the modification of the Cm-244 capture cross section and the TSL data of H in H 2 O.• The Cs-134 number density was estimated as larger using JENDL-5 because of the increased capture cross section of Cs-133 in the thermal energy region.• The modification in the fission yield from JENDL-4.0 to JENDL-5 significantly affects the number density of Sb-125, Sm-151, and Sm-152 in the target nuclides.• The modification of the cross section of U-235 and Pu-242 has some impact on the number density of U-235 and Pu-242, respectively.• The effect of the modification of the decay data and the TSL data of O in H 2 O is negligible on the nuclide compositions of the target nuclide.

Figure 6 .
Figure 6.Comparison of scattering cross section of H in H 2 O in the thermal energy region between the TSL data of JENDL-4.0 and JENDL-5.
*Positions from bottom of active fuel length.

Table 6 .
Sample burnup values and C/E-1 values of Nd-148 before and after calibration.

Table 8 .
Relative difference in number density of SF97-5 due to the replacement of each cross section of actinide nuclide and TSL data from JENDL-4.0 to JENDL-5.