Validation of the SCALE/Polaris−PARCS Code Procedure with the ENDF/B-VII.1 AMPX 56-Group Library: Pressurized Water Reactor†
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors
Very comprehensive and complete paper. Congrats!
Comments for author File: Comments.pdf
Author Response
Comments 1: L53: I understand the expensive time to produce all T16 files, but do you think that the number of energy groups (56) should be increased to reach even better computational results compare to experiment?
Response 1: I think that the 56-group is good enough for PWR analysis because the 56-group library was developed to be very consistent with the AMPX 252-group library for PWR and BWR.
Comments 2: L69-74: how many (concentric?) rings did you use in the fuel? Does it affect the calculation? I’am thinking to gadolinium rods as well. Same for moderator (I didn’t see well the details in the MoC picture –Fig1-), especially for the assembly in front of the reflector.
Response 2: We used three radial rings for fuel. We suggest 5 rings for gadolinia. However, the benchmark problems do not include any gadolinia burnable poisons. References 17-20 provide detailed models including sample Polaris inputs, which are not available for public yet, but will be available pretty soon.
Comments 3: L76: how many spatial points (for flux+depletion solving) per assembly do you have for the nodal calculation? Well, later on I understood (L354-360), one radial point per assembly.
Response 3: Yes, the PARCS developer is improving to have 2x2 node capability per assembly.
Comments 4: L105-106: this report is not available. Well, I saw that we can ask for (L388-389)
Response 4: The reference will be available for public pretty soon. Since the ORNL reports were prepared through the NRC’s support, we are waiting for NRC’s approval.
Comments 5: L121-123: I thought the contrary: in-core detector response represents local flux distribution (only 3x3 surrounded fuel pins) and not the assembly…could you please explain a bit more your sentence?
Response 5: The sentence was modified to be “Because the in-core detector response does not represent pin flux (or power) but represents more global flux (or power) including neighboring pins,”.
Comments 6: L174: “each burnup point”. How many burnup steps did you accounted for the beginning of the irradiation? I’m thinking to FP saturation.
Response 6: We considered 21 burnup points up to 60 MWD/kgU. References 17-20 provide detailed information for the reference burnups and branch cases.
Comments 7: L219: is it S^4 or S^2?
Response 7: S^4 is correct.
Comments 8: L227: “greater conservatism without considering experimental uncertainties”. Well I thought it is a supplementary uncertainty to add. I understand that you do not want to add it (no need to change in the paper) but presuming that it is peak check gamma scanning technique, it involves systematic/statistical uncertainties…
Response 8: The critical experiments do not provide the uncertainty based on peak check gamma scanning technique. However, we estimated the measurement uncertainty from the symmetric positions. Since our goal is to estimate calculation uncertainty, the measurement uncertainty can be statically subtracted. However, we did not subtract the measurement uncertainty from the difference uncertainty to have more conservative calculation uncertainty.
Comments 9: L242: very good picture. The reason why you did not distinguish C/M for fuel pins located in front and in angle of a water hole is because, for sure, there is only one mode in this frequency spectrum?
Response 9: Since we just need to estimate the pin-to-box factor uncertainty, we did not distinguish the pin locations.
Comments 10: L259: did you account for resonant up-scattering (U238 mainly) for HFP calculations? This is an important point to mention in the text.
Response 10: No. We could not consider the resonance upscattering because SCALE 6.3 Polaris with the 56-group library does not support the capability. “without considering epithermal resonance upscattering” was added to line 56.
We will provide the resonance upscattering capability for SCALE 7.0/Polaris with the 61- and 258-group libraries.
Comments 11: L268-27
- y-axis legend “Soluble boron difference (PPM)” -> “Critical soluble boron difference (ppm)”.
- For both graphs, do you mean C/M or M/C differences?
- What happened to the point “TP3 cycle 1”: it is in between uncertainty band for soluble boron and it goes off for reactivity difference. Is there a pb on the boron reactivity worth for this particular point?
Response 11:
(1) It was corrected.
(2) Always (C-M), which was indicated in Figures 4 and 5.
(3) This is a great finding. Actually we performed the PARCS calculations with two different reflector models. We were confused in the ppm graph (Figure 4) and uncertainty analysis. So we fixed the graph with Figure 4 and statistical analysis for critical ppm in Table 4 even though there is a negligible impact on uncertainty. Thank you so much.
Comments 12: L272: is there a spatial (radial) trend for power maps? I mean, overestimation/underestimation in the center and the contrary for the peripheral part of the cores?
Response 12: Yes, There is a slight power tilt in the Polaris-PARCS result due to a little poor procedure to obtain effective reflector cross sections. We have a plan to improve the procedure to generate reflector cross sections. The assembly power uncertainty already includes the power tilt.
Comments 13: L282: “SU1+TP3 graphs”: what does “simple reflector” mean, please?
Response 13: The ‘simple reflector’ was removed in graphs. As discussed above, we tested two reflector models for internal sensitivity analysis, the simple means relatively simple reflector model between two.
Reviewer 2 Report
Comments and Suggestions for Authors
I appreciate the presented results as it provides a valuable overview of the SCALE validation effort. On the other hand I had a problem to detect any uniq and original results and outcomes of the article. The SCALE validation is an ongoing process and all the presented results are available in existing ORNL reports as it is even also referenced in the article. A great deal of the results were also published in a conference paper from ANS Annual Meeting in 2023.
The presented paper does not exaplin in detail the nuclear data generation methodology even though the article title suggests focus on the AMPX library. It seems that the key part of the validation report is statistical evaluation of the measurements and calculations. I lack any references to literature in this part as I believe that it should be supported by some theory.
In summary, the article is missing original results and sufficient references to other works thus I need to suggest rejection of the article.
Author Response
Comments 1: I appreciate the presented results as it provides a valuable overview of the SCALE validation effort. On the other hand I had a problem to detect any uniqe and original results and outcomes of the article. The SCALE validation is an ongoing process and all the presented results are available in existing ORNL reports as it is even also referenced in the article. A great deal of the results were also published in a conference paper from ANS Annual Meeting in 2023.
The presented paper does not exaplain in detail the nuclear data generation methodology even though the article title suggests focus on the AMPX library. It seems that the key part of the validation report is statistical evaluation of the measurements and calculations. I lack any references to literature in this part as I believe that it should be supported by some theory.
In summary, the article is missing original results and sufficient references to other works thus I need to suggest rejection of the article.
Response 1: This investigation focuses on a validation of SCALE/Polaris-PARCS code procedure with the ENDF/B-VII.1 AMPX 56-group library which has never been performed previously. We partly published the benchmark results as a summary for WBN1 and Beavrs at the ANS conference as a summary. Typically conference papers would introduce some preliminary results during investigation. We improved models and added more benchmark results, and additionally performed statistical analysis to obtain uncertainties for key nuclear parameters to complete the validation. In addition the Polaris benchmark result for the critical experiments have never been published.
The ORNL technical reports details all the calculations including plant design and measured data, input preparations for Polaris, GenPMAXS and PARCS which are comparable to calculation notes.
We used typical and basic statistical analysis methods which may not require any reference. Similar statistical analysis to obtain uncertainties for key nuclear parameters has been used in nuclear vendors for long time. However, they have never released their statistical analysis result and procedure.
It is not clear what the ‘missing original results and sufficient references’ means. Please let us know what this means.
The following two references were added for the AMPX 56-group generation.
- Williams, M.L.; Wiarda, D.; Kim, K.S.; Jessee, M.A. Multigroup Data Processing for the Embedded Self-Shielding Method in SCALE. Proceedings of PHYSOR 2016, Sun Valley, ID, USA, 1-5 May 2016.
- Kim, K.S.; Williams, M.L.; Wiarda, D.; Mertyurek, U. Automatic Coarse Energy Group Structure Optimization by Minimizing Reaction Rate Differences for the SCALE and CASL Code Systems. Proceedings of M&C 2017, Jeju, South Korea, 16-20 April 2017.
The SCALE/Polaris-PARCS code procedure has been used for the NRC’s confirmatory analysis and many users are using the procedure for LWR analysis. However, there has been no clear validation effort for the procedure. I think that this investigation would provide a clear reference and guideline for the NRC’s confirmatory analysis and the users for the SCALE/Polaris-PARCS code procedure.
Reviewer 3 Report
Comments and Suggestions for Authors
Comments on Validation of the SCALE/Polaris−PARCS Code Procedure with the ENDF/B-VII.1 AMPX 56-Group Library: Pressurized Water Reactor
The paper presents the validation of the SCALE/Polaris modules coupled with PARCS for LWR validation, adopting ENDF/B-VII.1 libraries. The validation highlights the newer SCALE/Polaris compared to the older SCALE/TRITON for better efficiency.
Comments:
- Can you comment on the choice of using ENDF-B/VII.1 instead of the newer ENDF-B/VIII.0?
- Figure 1 shows the ¼ core geometry. However, the caption says it depicts the reactor physics with a two-step procedure. Can you extend the label to explain how the XY cross-section of the core fits into explaining the two-step procedure? Also, at lines 183-184, update Figure 1 with a different label.
- Line 179: Can you comment on the estimation of computational costs? Have you used the bare minimum configurations, or is it possible to omit some cases?
- Line 229: What do you mean by “the (-) uncertainty”?
- Table 3 compares measured data and calculated data. Due to the “case” column, the length seems excessive. Can you change the nomenclature for the BEAVRS to fit the page? Additionally, it would be helpful to depict the relative error in the table instead of the absolute difference. The previous paragraph elaborates on the error in pin factors. What about the amount of error for other quantities?
- Figure 6: The labels are very small, especially for (a) and (b). Also, why do cycles 1 and 2 start almost from the same burnup level in (a) and (b)?
- Table 4: Why is the statistical analysis not taken into account in the previous figures?
- Line 328: Can you comment on why the uncertainty for 2D is lower compared to 1D?
- Line 340: Figure 7 reports the axial 1D flux in the measured and calculated cases for two different burnup values. However, the radial region where the flux is taken is not specified. Also, can you observe the same tilt in burnup? Is it a feature that remains constant, increases, or decreases with depletion?
Some typos/minor comments:
- Line 316: “A ~10 ppm” instead of “An ~10 ppm”
Comments on the Quality of English Language
-
Author Response
Comments 1: Can you comment on the choice of using ENDF-B/VII.1 instead of the newer ENDF-B/VIII.0?
Response 1: Authors performed some sensitivity analysis using ENDF/B-VII.1 and VIII.0 and published the following manuscript.
K. S. Kim, W. A. Wieselquist, "Neutronic Characteristics of ENDF/B-VIII.0 Compared to ENDF/B-VII.1 for Light-Water Reactor Analysis," J. of Nucl. Engr., 2, 318-335 (2021).
The sensitivity analysis was performed using single fuel pins and assemblies. ENDF/B-VIII.0 significantly underestimates keffs for LWR at high burnups. It was expected that using ENDF/B-VIII.0 would result in large reactivity biases. In addition authors performed a sensitivity calculation using the Polaris-PARCS code procedure with the ENDF/B-VII.1 and VIII.0 56-group libraries for Watt Bar Unit 1 and Peach Bottom Unit 2. We obtained similar trend with the pin and assembly calculations. We submitted the investigation for ANS 2024 Winter.
K.S. Kim, W. A. Wieselquist, " Error Propagation Analysis, from LWR Assemblies to Plants, Using SCALE/Polaris-PARCS with the ENDF/B-VII.1 and VIII.0 AMPX 56-Group Libraries," ANS 2024 Winter, Orlando, FL (2024). (in review)
Therefore, we selected the ENDF/B-VII.1.
Comments 2: Figure 1 shows the ¼ core geometry. However, the caption says it depicts the reactor physics with a two-step procedure. Can you extend the label to explain how the XY cross-section of the core fits into explaining the two-step procedure? Also, at lines 183-184, update Figure 1 with a different label.
Response 2: The previous paragraph (lines 69-74) already describes the 2-step procedure which also indicates the assembly and reflector models to obtain assembly-homogenized cross sections and reflector cross sections.
Comments 3: Line 179: Can you comment on the estimation of computational costs? Have you used the bare minimum configurations, or is it possible to omit some cases?
Response 3: We used full case matrix as discussed in Lines 178-179. Since the full case matrix is standard based on previous investigation for PWR analysis, we followed the standard procedure. Computing cost is very expensive and machine dependent. For example, if it takes about 1 minutes for each state-point calculation, it will take total 15200 minutes. However, since there are independent 16 Polaris inputs, if we perform the 16 Polaris calculations at the same time, it will take about 16 hours for each assembly.
Comments 4: Line 229: What do you mean by “the (-) uncertainty”?
Response 4: Uncertainty itself should be described with ±. However, when the uncertainty is determined by non-parametric method with simple counting, the negative bound was called as the (-) uncertainty.
Comments 5: Table 3 compares measured data and calculated data. Due to the “case” column, the length seems excessive. Can you change the nomenclature for the BEAVRS to fit the page? Additionally, it would be helpful to depict the relative error in the table instead of the absolute difference. The previous paragraph elaborates on the error in pin factors. What about the amount of error for other quantities?
Response 5: We tried to change the format as review indicated, but we couldn’t. If relative errors are included, Table 3 will be too wide. We did our best with the limited funding and time. We will have better statistical analysis in the validation of future version of SCALE and PARCS as we get much more power plant measurement data.
Comments 6: Figure 6: The labels are very small, especially for (a) and (b). Also, why do cycles 1 and 2 start almost from the same burnup level in (a) and (b)?
Response 6: We drew the figures using cycle burnups (starting from 0) instead of total burnup.
Comments 7: Table 4: Why is the statistical analysis not taken into account in the previous figures?
Response 7: Question is not clear to us.
We performed the statistical analysis by collecting all the flux map comparisons together.
Comments 8: Line 328: Can you comment on why the uncertainty for 2D is lower compared to 1D?
Response 8: While the 2D flux maps include axially integrated flux maps for individual in-core detectors, the 1D flux maps include radially integrated flux maps over the core for individual axial planes.
Therefore, the 2D uncertainty can be lower than the 1D uncertainty when the axial flux prediction is poor but the radial flux prediction is good.
Comments 9: Line 340: Figure 7 reports the axial 1D flux in the measured and calculated cases for two different burnup values. However, the radial region where the flux is taken is not specified. Also, can you observe the same tilt in burnup? Is it a feature that remains constant, increases, or decreases with depletion?
Response 9: As discussed above, the 1D flux maps are radially integrated values over the core for each axial plane. So there is no radial location.
Comments 10: Line 316: “A ~10 ppm” instead of “An ~10 ppm”
Response 10: An ~10 ppm is correct because of ~.
Round 2
Reviewer 1 Report
Comments and Suggestions for Authors
Many thanx for your answers !
Hope your paper will be publised soon.
Best regards
Reviewer 2 Report
Comments and Suggestions for Authors
I am still convinced that the article has qualities of a technical report. It is a review of past and some new validation efforts without focus on the theory and methodology of the procedures. Most of the key references are to non-public ORNL reports, which further complicates potential reproducibility. In my opinion, the article is too broad to have a chance to go into detail. There is hardly any feature that could lead to improvements in future related articles.
Nuclear data generation methodology is standard and without modifications. Thus the extensive collection of calculation results is the primary outcome of the article. A technical report would be a more appropriate platform for such results.
I do not want to diminish you efforts, I am just convinced that the form of an article is not appropriate for the content.
Reviewer 3 Report
Comments and Suggestions for Authors
The paper, in this revised and corrected form by the authors, can be accepted for publication.