1 Transport and Activation files processing with NJOY-99 J-Ch. Sublet & C Jouanne 1 CEA Cadarache, DEN/DER/SPRC,13108 St Paul Lez Durance, France 1 CEA Saclay, DEN/DM2S/SERMA, Gif-sur-Yvette, France A Tale of the unexpected JEFFDOC-1160

2 The main libraries; neutron and gamma JENDL-3.3/GP337 isotopes (2 elements) JEFF-3.1/GP 381 isotopes (7 elements) JEFF-3.1/Th 9 compounds JEFF-3.1/A 774 isotopes ENDF/B-VII/GP393 isotopes (3 elements) ENDF/B-VII/Th20 compounds ENDF/B-VII/Ph163 isotopes + CENDL, BROND, … + special purpose files, proton, deuteron,.. + “preliminary” evaluations

3 The codes: Monte Carlo and deterministic The word “processing” encompasses a large set of processes not all equivalent nor identical to one another Not all Monte Carlo codes read in and/or need the same set of data or files, the processing level may differ  COG, MERCURY, MONK, MCBEND, MCNP5, MCNPX, TART, TRIPOLI, VIM, etc… Not one cell or lattice codes use the same input libraries  APOLLO, CASMO, DRAGON, ERANOS, PANTHER, SCALE, TORT, DORT, WIMS, etc… Different modules, sequences, parameters are used every time

4 Monte Carlo processing  TRIPOLI-4.4.1: Monte Carlo, pointwise or groupwise BROADR PENDF or GROUPR GENDF CALENDF probability tables in the URR 293.6, 300, 473.6, 600, 673.6, 873.6, 900, , 273.6, , , K Modified THERMR “TRIPOLI-4.4 JEFF-3.1 Based Libraries” CEA-R-6125  MCNP4c3&5: Monte Carlo, pointwise + Photonuclear (ENDF/B-VII) ACER PENDF PURR probability tables in the URR 293.6, 300, 473.6, 600, 673.6, 873.6, 900, , 273.6, , , K THERMR (up113, continuous secondary-energy distribution not yet plugged in) “MCNP4c JEFF-3.1 Based Libraries” CEA-R-6110

5 Deterministic processing  ERANOS2: 3D deterministic code, fast NJOY GENDF* (+ MF-50) (  MERGE & GECCO) ecco 1968, Xmas 172, VitJ 175, ecco 33 groups structures 293.6, 573.6, 973.6, , K CALENDF probability tables (MF-50 in all energy ranges) P0-P1 to P5 “ECCOLIB-JEFF-3.1 Libraries” CEA-R-6100  APOLLO2: 2D deterministic cell code NJOY GENDF* (+ as/afend,) up to 50 dilutions (  LIBNJ90 & N2A2) Universal 11276, Xmas 172, SHEM 281 groups structures 273.6, 293.6, 573.6, 773.6, 973.6, , , K CALENDF probability tables (self shielding & mutual shielding) P0-P1 to P5  DARWIN: activation/transmutation code NJOY GENDF 300 & 923K 3, ecco33, sailor47, GAM-II 100, Xmas 172, 315 groups structures

6 Tripoli-4.4.1, specific processing  Treatment of thermal neutron scattering data, scattering kernels in THERMR  Modification: results in terms of equi-probable cosines instead of equi-probable angles, this is not an update  With 32 cosines, done in up124  MT’s 221 and 222 hard coded  CALENDF-2005 probability table in, and only, the URR  addition of probability tables in the URR of each isotopes that contained one –140 in JEFF-3.1 –252 in ENDF/B-VII –209 in JENDL-3.3  PURR and CALENDF PT’s differs (order, steps, production methods, scope, …

7 Pointwise cross section comparison: total CALENDF pts NJOY pts PT’s Gprs A Cubic interpolation requires less points than a linear one But many more points exists in the CALENDF pointwise file in the URR, # smooth NJOY data

8 PT’s impact on the ICSBEP benchmarks Excellent way to test the influence of the URR PT’s Gprs groups in The URR

9 Tripoli Libraries, internal processing TRIPOLI generates XDR portable binary and angular distribution files the first time it reads in a PENDF and and an ENDF file Seven File Types are used : 1.Evaluation (ascii) 2.PENDF (binary) 3.Dictionary of cross sections (ascii)* 4.Binary XDR pointwise file* 5.Anisotropy file (ascii)* 6.Probability tables (ascii) 7.Thermal file (ascii) Tests for evaluations : ENDF-102 format Normalization (angular and energetic distribution) *: generated by TRIPOLI the first time it access the data file. 140 URR JEFF URR ENDF/B-VII

10 TRIPOLI-4.4.1, angular distribution  The angular distribution is computed by TRIPOLI the first time a pendf file is accessed and the code require to be linked to the original endf file  The computed angular distribution is variable, file dependant and can contain up to 256 equally probable cosine bins  For the JEFF-3.1 U238 the angular distribution have been produced by the ECIS optical model code and tabulated in 91 bins from 0 to 180 degree. It is used as such by TRIPOLI  Elastic and inelastic channels structures are preserved  It is a much better representation than in Legendre coefficients, but not strictly identical to a 64 bins uniform processing (MCNP)

11 QA: MT=2 Angular distribution

12 NJOY modules processing scheme ModulesMain parametersOutputs moder reconr err = (0.1%) broadr errthn = Thermal quantities unresr moder pendf files thermr+ nbin = 32 tol = emax = 4.95 moder NJJOY Evaluations Tripoli library One single C-shell scripts

13 NJOY modules processing scheme ModulesMain parametersOutputs reconr err = (0.1%) broadr errthn = Thermal quantities unresr moder pendf files thermr nbin = 16 tol = emax = 4.0 heatr mtk = purr nbin = 20 nladr = 64 gasprmt = viewrQA graphs moder acer newfor = 1 iopp = 1 Ace files acer iopt = 7 viewrQA graphs NJJOY MCNP4c3&5 library

14 QA Graphs and thermal quantities Inter 7.0 or NJOY-99 Broadr NJOY-99

15 QA Graphs

16 QA graphs, Photonuclear gamma files: pendf

17 QA Benchmarking (independent) Regardless of the evaluations quality the same information is extracted from the data file with two different Monte Carlo code on very different benchmarks

18 Reaction types and MT’s in EAF-2005/A ENDF-102 MT: 1 …117 new MT: 152…200 declared unassigned

19 MT Values - Grid of reactions including all 36 MT numbers defined in ENDF and some defined in EAF-2005

20 NJOY-99 Activation file processing In NJOY reconr has been modified to handle the new MT’s allowing pendf files generation NJOY has an automatic loop (10/) that will process all the nuclide production sections found in File 8 Results can be extracted from the output listing or read from the GENDF output MATXSR knows how to add the multiple production sections together and generate a single production cross section for each product Specialized output routines could be written for EAF + all fissile isotopes format changed to be handled by /10 in JEFF-3.1/A and EAF-2005 U230 pendf file Changed to

21 NJOY-99 processing Note that there may be multiple sections generating the same ZAm, and they have to be added together for total nuclide production ZAm of productFe54 to Mn52m(n,t)   cross section An update for groupr, matxsr, etc.. is needed for those modules to be able to handle properly the new MT’s

22 Activation libraries: the way forward Elastic scattering and inelastic channels to be added, it allows to reconstruct and compare the total with experimental data, a must.. Positive Q, (n,p) and (n,a) branching ratio in MF-9 Upper energy limit; 20 Mev, 60 Mev, … 150 MeV MT-5/MF-6 yields –above 60 Mev –or for all, but the recognized ENDF-6 MT’s channels  pure ENDF-6 format MT-5 and activation yields in a separate file, for each isotopes Deuteron, Proton, Gamma activation files

23 Activation libraries: the way forward Uncertainties –In a single for EAF-2005 –MF-33 like + one comment line + isomeric MAT  Provide a measure of the accuracies without correlations  simple MF-33, groups variances without correlations between the cross sections and/or the adjacent groups  Variable group structure It provide a complete set of “exploitable” data for activation, transmutation calculation uncertainties related to and uniquely the cross sections, not the reaction rates nor the neutron flux

24 Processing pits  If the processing fails this may be due to:  the inputted data file (evaluation format, numeric,…  the process code input parameter set (P9, 128 bins,…  the processing code itself  If it succeed it may well still fail to be interpreted properly or adequately (i.e. negative PT’s, wrong sum, non positive coefficient, …) but can this be spotted, checked, and finally stamped by external verification  But if it succeed and lead to different results (processed file) on different platforms (computers) when using a same input deck and evaluated file who will see it but a sturdy, exemplified, thorough QA and issue tracker as it has been developed and openly shared at:

25 Conclusions  NJOY is a complex code that may be dragged (less than before) into the “grey” areas of the different OS/compilers handling of number  Its installation requires care and knowledge with a high level of QA, Verification and follows up  Its input parameter sets may influence the end results, but this is a requirement  There are other processing codes that can be used to ascertain and verify some aspect of the processing  It evolved significantly over the past years, but still requires care and attention from its “knowledgeable” users

26 A spell, expressed during Bob’s Symposium It has been, and still is, a rare pleasure to follow the footsteps of such exceptional Wizard, as like to call him the man from the Dorset shore, he used in fact the word Magician but we differ on those delicate language manners. I would prefer the word Enchante(u)r that suits more my mind, although who am I to spell it with a u? any how it encompasses in many better ways all I know he have given us. Other did say Virtuoso, Maestro … The Old Continent still needs you Bob, and may be your last spell of the 99 series, a 1XX should do fine. A rune that will make happy the many few able to fell its power, and use it wisely in accordance with your custodian masterminding. Take pleasure in knowing that your legacy, did and will for many years to come entice the four corners of the world, at least the ones I know …