KAPL-P-OOO 14 1 (K98085)

CONF -9 8 IO03 - - OVERVIEW OF RECENT U235 NEU TRON C ROSS S ECTION EVALUATION WO RK

C. R Lubitz

October 1998

MA R

NOTICE

This report was prepared as an account of work sponsored by the United States Government. Neither the United States, nor the United States Department of Energy, nor any of their employees, nor any of their contractors, subcontracton, or their employees, makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness or usefulness of any information, apparatus, product or process disclosed, or represents that its use would not infr-inge privately owned rights.

KAPL ATOMIC POWER LABOWORY SCHENECTADY, NEW YORK 10701

Operated for the U. S . Department of Energy by KAPL, Inc. a Lockheed Martin company

DISUAIMER

This report was prepared as an account of work sponsored by an agency of the United States Government Neither the United States Government nor any agency thereof, nor any of their anployas, makes any wammty, ex- or implid, or assumes any legal Iiability or responsibility for the accuracy, completeness, or use- fulness d any infomation, apparatus, product, or process direlnvrl or repruenu that iu use wouid not infringe privatciy owned rights. R e f e r m a h a m to my spc- afic c o m m d l product, proass, or service by trade name. vldanut manufac- turer, or otherwise does not necessarily constitute or imply its endonancnt. m m - mendzrtion. or favoring by the United States Government or l a y agency thereof. The views and opinions of authors expressed h m i a do no1 n r m u n l y mtc or reflect thoK of the United States Government or any agency thamf.

DISCLAIMER

Portions of this document may be illegible in electronic image products. Images are produced from the best available original document.

,

OVERVIEW OF RECENT U235 NEUTRON CROSS SECTION EVALUATION WORK

Cecil Lubitz Lockheed Martin Corporation P. 0. Box 1072 Schenectady, New York 12301-1072 (518)395-7103

ABSTRACT

This report is an overview (through 1997) of the U235 neutron cross section evaluation work at Oak Ridge National Laboratory (ORNL), AEA Technology (Harwell) and Lockheed Martin C0rp.-Schenectady (LMS), which has influenced, or appeared in, ENDFiB-VI through Release 5. The discussion is restricted to the thermal and resolved resonance regions, qat from some questions about the unresolved region which still need investigation. The important role which benchmark testing has played will be touched on.

I. ENDF/B-VI.0

The original Release” of ENDFD-VI.0 U235 in 1990 was a collaboration among the Oak Ridge, Los Alamos, and Argonne National Laboratories, the last two providing the high-energy cross sections, and ORNL the resonance analysis. The latter culminated in an extensive Reich- Moore multilevel fit to the resolved-resonance region 0- 2250 eV.’ The work was canid out with a powerful new resonance-fitting code, SAMMY, which incorporated Bayesian methodology into the least-squares fitting, making it possible to deal sequentially, yet consistently, with the many large data sets requiring dys is .2 The evaluation techniques were a major advance over the earlier single level treatments in ENDFD-I through V, the latter of which ended the resolved region at 82 eV and treated the thermal region (0-1 ev) as tabulated data in File 3. The ENDF/B-

a Successive versions of EXDF/B-VI materials are designated as Mods. U235 has undergone modification in every Release of ENDF/B-VI, so the Mod number, which started as 1 in Release 0, is one greater than the Release number. We use the latter, as more familiar to most users. (The number of Releases, and the fact that different ones may contain quite different cross sections, makes it important for users and authors of technical articles to insure adequate identification of which ENDF/B-VI materials they used.)

VI evaluators recognized that the experimentally-observed structure in the cross sections above 100-150 eV was largely due to clumps of resonances, but it was their opinion that fitting it as if it were resolved “pseudo-resonances” would preserve the structure and permit users to derive more accurate estimates for self-shielded multigroup cross sections than could be inferred from a traditional treatment in terms of average unresolved resonance parameters. This important question has never been settleQ and remains one of the tasks for future investigation.

In the Cross Section Evaluation Working Group (CSEWG) review procedure, which was quite extensive, involving special meetings at Oak Ridge, the Thermal Benchmark Testing Subcommittee, under the chairmanship of J. R Hardy and M. L. W m , observed that there were two “differential-intew discrepancies which affected the e~aluation:~

1. The 2200 mls values for capture and fssion, and their associated Westcott g-factors, produced a Maxwellian- averaged q (v-6ssion/absorption) which was lower than the value inferred from eigenvalue calculations of thermal reactors. Quantitatively, the effect was measured by the parameter K1 = v-fssion minus absorption (all quantities Maxwellian-averaged). The integral value was 722.7i3.9 whereas the evaluation gave about 719. The 2200 m/s values were the then-recommended thermal constants that resulted from an extensive round of least-squares fitting to pointwise and reactor-averaged thermal data, following a tradition started many years earlier by G. H. Westcott. The CSEWG discussions of the least-squares results were spirited. Within the CSEWG Task Force on Thermal Constants, led by B. R. Leonard and J. R. Stehn, some believed that least-squares was the best one could do, while others believed that the results were too sensitive to the choice of input uncertainties, and that the output uncertainties were unrealistically low. It was noted that the main input data at the low-K1 end were the Chalk River reactor-average-alpha measurements, which were high and carried low uncertainties, and some low nubar data. A complete discussion of the relevant data can be found in

.

Reference 4, where an exhaustive least-squares fit to the thermal constants resulted in K 1 = 712.6. That low value epitomized the integral-differentiaI conflict for U235.

That disagreement in philosophy is still with us, and as recent experience has shown, a straightforward fit to the thermal “differential“ data still yields a K l which is below the integral value? The result of the low K1 was that eigenvalues (multiplication constants) for well-thermalized assemblies calculated around 0.5% low using Release 0.

2. The resonance integral for rawvecapture was some 10 bams below the integral value of about 142 barns. Cofiespondingly, its ratio to the fwion i n t e a (epithermal alpha) was 6-7% below the integral value of 0.51. A consequence of uli fact, and of the low K1, was that Release 0 calculated low multiplication constants for very- thermal, large, low-leakage assemblies, but the calculated values mse sharply with increasing leakage. As the leakage increased, and the spectrum hardened, the low alpha caused the calculated eigenvalues to increase. Nevertheless, the evaluation was accepted for ENDFB-VI.0. One reason was a matter of principle among many of the CSEWG members that ENDF should represent a pure view of the microscopic measurements, and not allow itself to be influenced by integral consideralions, which (then, and still today) can manifest themselves in ad hoc adjustments to cross sections, without a clear physical basis.

Another consideration, which is as valid today as it was then, is whether the integral data are accurate. Perhaps all of the thermal benchmarks are afflicted with some unknown emr, and that in fact the low K1 resulting from a least-squares fit to the microscopic measurements is correct. It would be undesirable to change the evaluation in order to fit incorrect benchmarks.

It is impossible to answer this criticism. However, one cannot simply ignore the integral measurements, so in the recent evaluation activities at ORNL and LMS a more pragmatic attitude has been adopted. Essentially, the argument is that all the experiments are “wrong”, if by “right” one means they give the true value. The microscopic measurements can only define an envelope of credibility, and it is possible to stay within that envelope and still match the integral data. In particular, “pure” least- squares fits to the microscopic data are to be preferred only in the absence of other indications. The weakest links in the microscopic measurements appear to be the normalization of capture measurements and the specification of the true resolution in measurements of the d o alpha. At best these can be done to a few percent, yet that uncertainty propagates directly into the least-squares results, where a few percent is enough to (possibly) skew

the answers.

In the most recent evaluation &om ORNL (August ‘97), which has been adopted for ENDFB-VI Release 5, some integral data have been incorporated into the basic fitting process, by adding to SAMMY the ability to fit integral values, in particular, resonance integrals and K1. This provides a mathematical bridge between the two worlds, but puts an added burden on the evaluator to insure the results are physically reasonable.

II. ENDFB-VI.1

Release 1 (199 1) introduced an energydependence in sub-thermal eta, so-called “drooping eta”. The possibility of such an energydependence had been hypothesized by Santamarina, et al., and later confumed by measurements at Geel, Harwell and ORNL.6 This modification removed most, but not all, of the discrepancy between measured and calculated temperame coefficients which had stimulared the original investigation. Since no other modifications were introduced, Release 1 continued to e t poorly in most benchmarks.

III. ENDFB-VI.:!

Release 2 (1993) did not change the cross sections.

IV. ENDF/B-VI3

Release 3 (1995) was an adjustment to Release 2, carried out at LMS by the author in cooperation with L. C. L.eal, L. W. Weston and R. Q. Wright at Oak Ridge and M. C. Moxon at Harwell? It increased the epithermal capture to agree, on average, with the measured differential capture cross sections of Perez and deSaussure,S and also increased K1 to improve the calculation of the very thermal assemblies. The technique used in the epithermal region was to increase the capture area of each resonance so that the average capture cross section over 100 eV-wide “bins” between 0 and 900 eV agreed with the values measured by Perez and daaussure. The required increase in average capture was applied to the capture area of each resonance. Because of differences in the neutron and fission widths from resonance to resonance, the capture width change-ratio was not simply a constant. Simultaneously, the neutron and fission widths of each resonance were changed to make the fission areas slightly decrease to agree with the lower of the credible fission measurements, and the Doppler-broadened total cross section peak height of each resonance was constrained to be as close as possible to that of the unadjusted resonance. The transcendental equations were solved iteratively to achieve the desired adjustment, using the NJOY code system to perform the Reich-Moore

resonance reconstruction and Doppler-broadening? The end result was an increase in the average radiation width from 35.0 meV to 38.2, over the first 100 eV. As noted above, the higher-energy structure is dominated by multiple resonances, so that the radiation width no longer has physical significance.

The technique used to achieve the desired increase in K1 was also an iterative one, in which the nine resonances at negative and positive energies below 0.5 eV were changed. At each change, the cross sections from the entire set of 3500 resonances were reconstructed and the resulting 2200 m/s values, g-factors, and K1 were noted. Successive adjustments finally produced resonance parameters which yielded the desired K1 (723) subject to the constraint that the 2200 m/s fssion cross section and nubar increased, while the 2200 m/s capture decreased, all by the same percentage of their uncertainties as given by the CSEWG Standards Subcommittee. The procedm converged at a change in the Standards values of 0.5 standard deviations. Because the changes to the resonance parameters were quite small, the droopingeta shape was maintained.

As expected, Release 3 did quite well on the thermal benchmarks, raising eigenvalues for the very thermal cores, and greatly reducing the trend with leakage. Since the cross sections above 900 eV are identical to Release 2, its calculated fast benchmark eigenvalues are also identical. A. Jonsson, at the 1995 CSEWG meeting, reported improvements in reactivity enrichment dependence." However, Release 3 did not do well on intermediate-

and UR3 benchmarks which RQ Wright had added to the set of benchmarks being used to test U235. One can speculate that the same "low-alpha syndrome" which affected Release 2 below 900 eV extends up much higher in energy.

s ~ - cores, calculating 1-2% high three HISS(HUG)

The adjustment process, which was done using bm- averaged cross sections, introduced discontinuities in the unbroadened cross sections at the resonance region boundaries. Although smoothed out by subsequent Doppler-broadening in NJOY, they remain an undesirable feature. Release 3 also had a formatting error in that the increased value of thermal nubar was put into the total- nubar fie, but not into the prompt-nubar file. That means that processing codes which reconstitute the total from prompt + delayed will get slightly lower thermal eigenvalues than they should (nubar = 2.4320 instead of 2.4338).

V. ENDFIB-VIA

ENDFD-VI Release 4 (1996) is the same as 3, but corrects the prompt nubar file.

VI. LEAL-DERRTEN-WRIGHT JANUARY 1996

In January of 1996, Oak Ridge (LC Leal, H Derrien and RQ Wright) sent out a preliminary reevaluation which put all the resonances into one region, thus eliminating the need for carefully tailored "external" resonances in each of the previous eleven regions. The new work matched the higher epithermal alpha and K1 of Release 3, but because it was a Bayesian SAMMY fit, rather than an adjusted deck like Release 3, the detailed fits to differential data were very good. In addition, it produced better (lower) eigenvalues for the three intermediate-spectrum HISS(HUG) and UR3 cores. However, MC Moxon observed that the radiation widths fluctuated more than was expected theoretically. He also questioned whether the target backings had been properly accounted for in the experimental and evaluation phases of the work, and whether the free-gas model was adequate for Doppler-broadening the low-energy resonances. His code, REiFJT, had an option to use a quantum- mechanical Einstein model to represent the target velocity distribution, and he felt that it was doing a better job at fitting the resonance shapes."

VII. LEAL-DERFUEN-WFUGHT-MOXON MARCH 1997

To help resolve these questions, Moxon spent two months at ORNL in late 1996. Moxon and N. M. W o n worked to reconcile numerical and procedural differences between REFIT and SAMMY. All numerical constants were made consistent, as was the way in which the free-gas Doppler calculation was carried out. The Doppler question still remains, because we do not yet have a defdtive test of how much difference results from accurate fitting with both models of the same data, or of the reactivity worth of such differences. In any event, Einstein-broadening is not yet available in ENDF-processing codes, so it was decided to continue the work with the &-gas model.

Some other differences between SAMMY and ETlT, besides the use or non-use of Bayesian methods, remain. Examples are the way in which the neutron-burst moderation is calculated; the way the resolution function is synthesized from its components; and the multiple-scattering correction to capture and fission yields, which is multiple in REmT but single-scatter in SAMMY. The practical impact of such differences has not been assessed, and there is currently no support for further harmonization of the two codes. It should be emphasized however, that the existence of two such sophisticated analysis tools, and the expertise which their authors and users have brought to bear on the U235 cross sections have resulted in higherquality fits than

was possible as recently as five years ago.

During Moxon's stay at Oak Ridge, Leal, Derrien and Moxon, together with J. A. Harvey and R. Spencer, were able to answer most of the questions regarding the ORNL experimental data reduction procedures. The resulting data set received limited testing, primarily Reference 11, plus unpublished work by Wright, Kahler and Weinman. The consensus was that the benchmark behavior was reasonable, but more should be done.

VIII. LEAL-DERFUEN-WRIGHT MAY 1997

Subsequently, Leal, Demen and Wright produced an interim set for discussion at the May 1997 Cadamhe meeting of the WPEC. The generation of this set utilized a new capability in SAMMY, including integral data (resonance integrals and K1) within the fitting procedure, thus putting this phase of the work on a more objective basis. Because of its preliminary nature, this set was not extensively tested. It continued the progress of the ORNL group toward a data set which would simultaneously match both integral and differential data.

IX. MOXON MAY 1997

Independently, Moxon produced a fit to the 0-100 eV region that differed from previous U235 evaluations in having a rather low capture g-factor, about 0.95 instead of the traditional .99. This may have been the result of removing the original normalization and background from the 1966 de Saussure capture and fission data" and re- fitting it with REFIT. Like the May '97 ORNL fit, this was an interim version, for discussion at the Cadarache meeting. The only action taken there was to recommend that an attempt to reach a consensus would be made before the October '97 CSEWG meeting. That turned out to not be possible, and the two evaluation efforts are st i l l separate.

Because of its relatively low Maxwellian-averaged fission, Moxon's May '97 set implied a low K1 unless it could be coupled with a nubar much higher than the FBDFD-VI Standards value. Such a higher nubar exists in the measurements of Gwin, et al., which has already been adopted by both the Joint European File (JEF) and the Japanese Evaluated Nuclear Data Library (JE!NDL).I9 At LMS, we put together a "tri-partite" data set, by inserting Moxon's fit from 0-10 eV into the January '96 ORNL evaluation, and replacing the nubar file with the JEF values. Benchmark testing showed that such a hybrid set can perform well on thermal benchmarks, but does poorly on the harder-spectrum cases. It appears that the higher Gwin nubar is good for the soft-spectrum cases, but is high for the others. However, these need to be more closely

examined. An open question is whether the different capture shape which produces the low g-factor in Moxon's work will affect temperature coefficients.

X. LEAL-DERRIEN-WRIGHT-LARSON AUGUST 1997

After the WPEC meeting at Cadamhe, Leal and Derrien continued to refine their fits, and in August, 1997 a new data set was made available for testing. It incorporated the integralquantity fitting capabilities of SAMMY to deal with the earlier problems of low K1 and epithermal alpha. Most noticeably, thermal nubar was increased from the Release 3 value (2.4338) to 2.4367 over the range from 0 to 1 eV. This value was arrived at by treating thermal nubar as a search parameter in the SAMMY runs, while using the benchmark value of K1 as a quantity to be fitted It is close to a smooth average through the fluctuating Gwin data points in this interval, and is also (fortuitously) exactly the ENDFP-V value. By cutting off the increase at 2 eV, the fit avoids the deleterious effect of a higher nubar on the harder-spectrum cores. The radiation widths in the new L-D data set have an average value over 0-100 eV of about 40 meV, with a standard deviation of lo%, an acceptable degree of fluctuation.

At the October 1997 CSEWG meeting, AC M e r and JP Weinman presented benchmark results with this new data set, for about 25 benchmarks including ORNL spheres and cylinders, ORNL L-series, Rocky Flats, and 3 HISS(HUG) and UR3 intermediate-spectrum cases. The best results were obtained using a reduced hydrogen capture cross section, 332.0 mb instead of 332.6, and a slightly-modified version of ENDF/B-VI oxygen, which had more forward scattering? The effect of the hydrogen was to raise the low-leakage end of the thermal benchmark "trend line" about 100 pcm, while the additional oxygen leakage lowered the high-leakage end about the same amount. Together they produced a flat trend l i e with an average eigenvalue very close to unity. The intermediate-spectrum cases are close together at about 1% high. The thermal cases are therefore about the same as Release 3, while the 3 harder cores are better, although not quite as close to unity as in the January 1996 evaluation of Leal-Derrien- Wright. In view of its good benchmark performance, it was accepted for inclusion in ENDFB-VI Release 5 (as was the new hydrogen capture.) A decision on the oxygen is awaiting completion of a new evaluation at Los Alamos National Laboratory by G. Hale.

In December of 1997, two papers were presented at a

E. CWO, this conference.

meeting of the JEF Working Group on Data Evaluation and Benchmark Te~ting.'~.'~ Both showed significant improvements using L-D-W-L over the JEF2.2 data set, which is similar to ENDF/B-VI Release 2. The studies covered a wide range of quantities, among them buckling measurements, k-infiities, temperaturedependent Westcott g-facton, and spent-fuel analyses. In each area the L-D-W- L set improved the calculation/experiment agreement. Of particular interest were the spent-fuel results, since they test the cross sections in a way that is relatively independent of any remaining problems with nubar. As a result of these favorable fmdings, the new data were adopted for JEF-3.0.

XI. MOXON SEPTEMBER1997

Moxon has released a version of his latest work to the Nuclear Energy Agency Data Bank at Paris?6 It extends his earlier work to higher energies, and achieves satisfactory fits to the differential data using a single value of the radiation width, close to the 38.2 mV he recommended in Reference 25. As of this date, it has not been reviewed or benchmark tested, since that will require it to be coupled with a nubar file and higher energy cross sectons. Moxon's work represents a "pure" fit to the microscopic data, so that if it is to be useful in integral calculations, it will probably require a new treatment of nubar. That is also a topic for further investigation, as is the significance of the unusual Westcott g-factor for captme, noted above in connection with his May '97 work.

XII. THE ROLE OF INTEGRAL TESTING

The degree to which successive Releases of ENDFB- VI agreed or disagreed with various integral benchmarks has from the start been an essential element in shaping its development. With the evolution of Release 3 and subsequent refinements, that aspect of the evaluation process became more integrated into the data-fiaing process. At Oak Ridge, R. Q. Wright routinely checked successive iterations of the differentialdata fitting and/or adjustment process, and provided feedback on their integral adequacy. At Bettis Atomic Power Laboratory and at LMS, A. C. Kahler and J. P. Weinman continuously monitored the progms with highly-accurate point-energy Monte Carlo calculations. The major Releases were extensively checked by the CSEWG Thermal Reactor Testing Subcommittee under M. L. Williams. R. D. Mosteller at Los Alamos National Laboratory used both CSEWG benchmarks and the newer collection embodied in the International Criticality Safety Benchmark Evaluation Project under J. Blair Briggs W L ) to assess both ENDF and other data sets. Large-scale calculation of both published and proprietary benchmarks were carried out by the UK, France, Japan, and elsewhere, for example

References 11-16. The reader is referred to the minutes of the annual CSEWG meetings at Brookhaven, to the transactions of the annual and topical meetings of the American Nuclear Society, and to the proceedings of the International Conferences on Nuclear Data for Science and Technology, for Basic and Applied Science, for Reactor Safety, and related meetings.

Because of the many different methods used to measure the benchmarks and to calculate them, it is difficult to incorporate the results directly into the evaluation process. Instead they provide broad-brush indications of where problems exist, and can focus the evaluators' attention on the appropriak differential data.

XIII. TOPICS FOR FUTURE INVESTIGATION

A number of the points discussed in the preceding paragraphs have suggested the need for a closer examination of nubar. Both theory and e x w e n t show that nubar varies from resonance to resonance, yet that variation is universdy ignored in the cross section evaluation 24 The very good fits to capture, fission and total cross sections achieved at Oak Ridge and Harwell suggest that these are now '*well-known" and are incompatible with integral reactivity measurements unless coupled to a much higher nubar than has been used in earlier Releases of ENDF/B-VL Introducing a sound theory-based description of the energy-dependence of nubar could have a beneficial effect on many calculations, since it would open up an important new degree of freedom. Any sort of correlation between nubar and alpha could affect both the thermal and intermediate-spectrum calculations. Other areas would be discrepancies between a- and q-measurements, and moderator temperature coefficients.

The question of over-absorption in U238 continues to cloud benchmark results for low-enriched U235. Modem computational techniques, most importantly accurate point- energy Monte Carlo calculations, could settle the issue of whether the basic data are responsible, or whether the problem is in the reduction to multigroup cross sections and related calculational approximations.

As mentioned above, the pseudo-resonance representation of the region from 100-150 eV up to 2250 eV which is currently in ENDF/B-VI has never been objectively tested against experiment. Moxon has derived a preliminary fit to the unresolved resonance region, using average parameters extrapolated from the low-energy region. He uses multilevel resonance ladders to fit measured total and fission cross sections, with a proper treatment of the inelastic competition, and finds an alpha higher than predicted by an equivalent SLBW approach, an effect which

. I

,

has been noted before. This holds promise for the intermediate spectrum benchmarks which calculate high using the current ENDF values.

Moxon has also raised the issues of Bragg scattering and crystalline binding, and their effects on the deduced resonance parameters. At the present time, these seem to be too difficult to incorporate into the cycle of evaluation, processing, and use, because they are not readily available in most fitting codes, or in NJOY, and might q u i r e that the user tailor his cross sections to the physical and chemical composition of the U235.

XIV. CONCLUSIONS

A large effort has gone into the evaluation and testing of the U235 resonance region since the start of ENDFD-VI in 1990. In Release 5 we have arrived at a credible data set that fits a variety of both differential and integral data. Some questions remain, such as the extension of better methods to the unresolved region, incorporating a theory and experiment-based variation of nubar, and objectively deciding whether in fact the pseudo-resonances are superior to an accurate, conventional fit using average parameters based on channel fission theory.

ACKNOWLEDGMENTS

Much of the later work was done under the auspices of the Organization for Economic Cooperation and Development Nuclear Energy Agency Nuclear Science Committee Working Party on Evaluation Cooperation (WPEC), acting through Subgroup 18 on U235 epithermal capture. This significantly enhanced the traditional oversight provided by the USDOE Cross Section Evaluation Working Group (CSEWG).

REFERENCES

1. G. de Saussure, L. C. Leal, R B. Perez, N. M. Larson, and M. S. Moore, "A New Resonance Region Evaluation of Neutron Cross Sections for p5U," Nucl. Sci. Eng. 103,109 (1989); L. C. Leal, G. de Saussure, and R B. Perez. "An R-Matrix Analysis of the ='U Neutron-Induced Cross Sections up to 500 eV", Nucl. Sci. Eng. 109. 1-17 (1991).

2. N. M. -on, Updated Users' Guide for SAMMY,

3. C. L. Dunford, Summary of the CSEWG Meeting, April 3-7, 1989, Brookhaven National Laboratory.

4. M. Divadeenam and J. R Stehn, "A Least-Squares Fit of Thermal Data for Fissile Nuclei", Annals of Nuclear Energy, u, 375404 (1984)

5. A. D. Carlson, W. Poenitz, et al.,"ENDF6 Neutron

ORNL/I'M-9179/R3, 1996

Cross Section Measurement Standards", NISTIR 5177 -ENDF 351 (1993); E. J. Axton, "Evaluation of the Thermal Constants of U233, U235, pU239 and W 1 , and the Fission Neutron yield of Cf252", GE/PW01/86 (1986).

6. A. Santamarina, et al., "Nuclear Data Qualification Through French LWR Integral Experiments", Proc. Int'l. Conf. on Nuclear Data for Basic and Applied Science, Santa Fe, pp. 509-515, Gordon and Breach, 1985; C. Wagemans, "Low-Energy Cross Sections of the Major U and Pu Isotopes", Nuclear Data Standards for Nuclear Measurements, NEANDF-311, 1992.

7. C. R. Lubitz, "A Modification to ENDF6 U235 to haease Epithermal Alpha and Kl", Proceedings of the International Conference on Nuclear Data for Science and Technology, Gatlinburg Tennessee, pp. 646-648, American Nuclear Society, 1994; C. R Lubitz, "Neutron Cross Sections for Uranium-235 (ENDFD-VI Release 3)", KAPL-4825 (DOEEIC-4500-R75), 1996.

8. R. B. Perez, G. de Saussure, et aL, "Simultaneous Measurements of the Neutron Fission and Capture Cross Sections for Uranium-235 for Neutron Energies from 8 eV to 10 keV," Nucl. Sci. Eng. 2, 46-71 (1973).

9 RSIC Code Package PSR-171, "NJOY", Oak Ridge National Laboratory, 1992.

10. C. L. Dunford, Summary of the CSEWG Meeting, Attachment 4A-1, October 17-19, 1995, Brookhaven National Laboratory.

11. C. J. Dean, D. Hanlon, and R. J. Perry, "Benchmark Calculations for U235", JEF / DOC-682, 1997.

12. C. Mounier, "Analysis, Processing, and Integral Testing of the U235 Leal-Derrien-Lanon Evaluation", JEF I

13. Proceedings of the 1992 Symposium on Nuclear Data, M. Baba and T. Nakagawa (editors), JAERI-M-93-046, 1993; and other volumes in this series.

14. J-D Kim and C-S Gil, "Validation of ENDFB-VI.2, JEF-2.2 and JENDL-3.2 Through MCNP Analysis of BWR-Simulated Critical Experime nts", Annals of Nuclear Energy, Vol. 24, No. 3, pp. 197-204, 1997.

15. M. C. Met, A. Benslimane, C. Chabert, C. Mounier, A. Santamarina, and G. WiUermoz, "Qualification of the U235 Leal Demen Larson Evaluation Using French Integral Experiments", JEF I DOC-707, 1997.

16. F. Rahnema and S. McKinley, "On the Validation of Release 3 of ENDFD-VI Nuclear Data for Light Water Reactor Criticality Analysis", Annals of Nuclear Energy, Vol. 24, No. 14, pp. 1151-1155, 1997.

17 M. C. Moxon, "An Evaluation of U235 Resonance Parameters up to 102 eV Using the Program REFIT", AEAT-0716, 1996; M.C. Moxon, "REFIT Users' Manual", April 1991.

DOC-702, 1997.

18. G. de Saussure, et al., ORNL/TM-1804, (1967).

1

19. R. Gwin, R R Spencer and R W. Ingle, "Measurements of the Energy-Dependence of Prompt Neutron Emission from U233, U235, Pu239 and Pu241 Relative to Cf252", Nuclear Science and Engineering, Vol. 87, pp. 381-404, 1984.

20. F. J. Hambsch, H. H. Knitter and C. Budtz-Jorgensen, "Fission Mode Fluctuations in the Resonances of U235(n,f)", NucZear Physics, Vol. A-491, pp. 56-90, 1989.

21. R. E. Howe, T. W. Phillips and C. D. Bowman, "Prompt Fission Neutrons from eV Resonances in U235", Physical Review C, Vol. 13, No. 1, p.195.

22. U. Bmsa, S. Grossmann and A. Muller, "Four Channels in the Fission of Cf252", Zeitschriff fur Nafu#orschung, Vol41-a, pp. 1341-1346, 1986.

23. E. Fort, J. F~haut, H. Tellier and F. Long, "Prompt Nubar for Neutron-Induced Fission in Resonance Region", Physics of Neutron Emission in Fission, p. 123, INDC(NDS)-220, H. Lemmel (ed), 1989.

24. U. Gohs and H. Marten, "Evaluation of Total Fission Characteristics for U235 in the Resonance Region", Nuclear Data for Neutron Emission in the Fission Process, pp. 241-249, INDC(NDS)-251, S. Ganesan (ed.), 1991

25. M. C. Moxon, "Comments on the ENDFB-VI Eval- uation for U235 in the Neutron Energy Region from 1 to 20 eV", ORNLEM-12304 (ENDF-355), February 1993.

26. M. C. Moxon, "An Evaluation of the U235 Resonance Parameters Using the Program REFIT", JEF / DOC- 708, 1997