NJOY 99.24 28 Sept 00 ----------------------------------------------------------------------- This file describes updates to NJOY99. See Readme0 for more general information on NJOY99, installation, test problems, and so on. *ident up1 */ leapr -- 04apr00 */ this fixes errors that destroy the calculation of s(alpha,beta) */ for cold para-hydrogen and for both cold deuteriums. the first */ was introduced in 97.53 while making the constants symbolic, */ and the second was introduced in the cleanup process for njoy99. We found these problems this March and April. LEAPR runs for the cold hydrogen and deuterium cases made with versions earlier than 97.53 should be OK. *ident up2 */ groupr -- 04apr00 */ there is an error in the indexing for the xmas 172-group structure */ that throws all the group bounds off by one. this structure is */ used in europe for thermal reactor calculations. This one is only important for people using this group structure. */ fix a problem introduced with the activation patch of 97.102. the */ nk parameter is only used when doing mf values for activation */ products, and it is not appropriate for fission nubar values. *d groupr.3824 */ don't strip off the upscatter groups for the neutron spectra */ coming from delayed neutron emission (mt=455). */ add a missing save statement and fix an unset variable in anased. */ these problems affect delayed neutron spectra. These changes are important for anybody doing multigroup calculations for fissionable materials. *ident up3 */ acer -- 04apr00 */ the acer consistency checks include an option to readjust */ eprime values that are greater than e, when appropriate. */ there are some problems with the logic, especially for */ cases that use histogram interpolation for the distribution. This change improves the appearance of the outputs, but it doesn't actually have any effect on calculations. MCNP only uses the differential probabilities for linear-linear interpolation, and it doesn't make any difference if they are incorrect at the upper limit for histogram cases. */ when using the old format (mcnp4b and earlier), some angle-energy */ distributions from file 6 are converted into the law 67 format, */ because these earlier versions of mcnp couldn't use all the */ file 6 representations. when converting from the cm to the lab, */ the methods used in subroutine fix6 are a little crude. they get */ confused when cm energies are so small that lab cosines of -1 */ are not reached. this patch tries to fix that in a rough way, */ but evaluations that use the cm frame in file 6 will work best */ if most of the cm energies are greater than e/(awr+1)**2 for */ each incident energy e. for mcnp4c and later, the code can */ sample directly from tabulated cm representations, and the */ approximations of the fix6 routine are avoided. this patch */ is needed for one evaluation from JEFF-3. */ increase the available storage to handle the very large */ mf6/mt16 tabulation in JEFF-3 Be-9. */ increase the space available for discontinuities in convr */ to allow for JENDL-3.2 si-nat These three sets of changes are only needed for very specific cases. We already know that there will be a few more like this coming up! *ident up4 */ reconr -- 05apr00 */ be sure to count subsections of file 12 before allocating */ storage for the elements of the new directory. otherwise, */ some materials with many sections of file 12 will overflow. */ this is a longstanding problem that we never noticed before. This is another problem that only showed up for one evaluation. It has been this way in NJOY for many years. *ident up5 */ purr -- 7may00 */ fix a problem introduced while installing the heating part */ of the probability tables. it shows up when doing elements */ that have unresolved data. also, increase the number of */ resonance sections allowed to handle the very large cd-nat */ evaluation from JENDL. PURR marks regions where the unresolved range for one isotope of an element overlaps with the resolved or smooth range from other isotopes by putting a negative sign on the energy. We have to be sure to recover the energy correctly by taking the absolute value! Note that this problem only shows up for elemental evaluations. There are only two of these in ENDF, but many more in JENDL. We also had to allow for more sections in the resonance range to handle JENDL's Cd-nat evaluation. *ident up6 */ acer -- 30may00 */ fix a typo in up3 (reported by bunde, anl) */ acer -- 30may00 */ fix problems with converting cm distributions to law=7 */ and problems reading law=7 into the ace file. these problems */ show up when running newfor=0 with njoy2000, especially on */ some materials from jef-2.2. This one is pretty important. NJOY99 was pretty thoroughly tested for producing MCNP4C libraries, but we neglected to do as much checking for MCNP4B cases (newfor=0). *ident up7 */ viewr -- 30may00 */ increase the allowed size of 3d plots. */ pushed by pb-nat from jef-2.2. In most cases, NJOY just truncates 3-D plots that are too big for viewr, but in the pb-nat case, it actually overwrote some memory. The storage available for 3-D plots has now been increased, and some plots that were truncated before without causing the code to fail will plot better now. *ident up8 */ acer -- 3jun00 */ subroutine ptleg2 does not need the dynamic array xat. */ this problem was first noted by Waclaw Gudowski for ENDF/B-VI */ si-nat. it shows up as "id xat not defined". This is a fairly isolated problem. It is a case of evolution leaving behind an obsolete organ. *ident up9 */ acer -- 07jun00 */ add the capability for processing anisotropic charged particle */ emission using tabulated legendre coefficients into the */ mcnp4c law61 format. this is needed for jeff-3 cr-52. */ allow for multiple interpolation ranges in file 6. this */ also occurs for jeff-3 cr-52. currently, the neutron */ energy-angle distribution only allows for combinations of */ histogram and linear linear interpolation, but the */ charged-particle sections allow for general combinations of */ all allowed interpolation laws. There are only one or two such cases in current evaluations, but more will probably appear in the future. The limitation to interpolation laws 1 and 2 for neutrons comes from the occasional need to handle multiple subsections in File 6 for neutron emission. The increased generality for charged-particle subsections comes about because multiple subsections are not handled. Someone will probably do that someday.... Although we did work fairly hard to handle the JEFF Cr-52 evaluation, the actual representation is fairly strange, and the treatment of the particle yields near the effective emission threshold should really should be changed. *ident up10 */ leapr -- 13jun00 */ fix two incorrect constants in leapr. one affects cases with */ diffusive effects, and it has been incorrect since njoy97.0 */ (oct 97). the other affects cold hydrogen calculations, and it */ has been incorrect since njoy97.53 (dec98). NJOY99 was exhaustively checked using ENDF/B evaluations and the newfor=1 option (to support MCNP4C). Unfortunately, we didn't test it with newfor=0 (for MCNP4B and earlier versions)! Several users have reported problems (especially Waclow Gudowski in Sweden. Some changes that affect running with newfor=0 were made above. The following changes (up11) are especially noticable with ENDF/B-VI Release 6. *ident up11 */ acer -- 26jun00 */ fix an error in determinining which reactions have to be */ converted into law=7 format when using newfor=0. because of */ overzealous code cleanup, acer is trying to convert sections */ with the kalbach representation in addition to sections with */ tabulated angular distributions. */ acer -- 26jun00 */ during the cleanup of the topfil routine, the logic to process */ sections of file 6 using law=2 (two-body distributions) into */ equally probable bins for newfor=0 was omitted. this shows up */ for evaluations that use mf6/mt51, etc., to represent inelastic */ levels. */ acer -- 26jun00 */ fix an error in the reaction naming. it affects mt=44 (n,n2p) and */ mt=45 (n,npa). this problem was introduced in may of 1995. */ examples of cases that use these reactions are later releases of */ endf/b-vi al-27. */ acer -- 26jun00 */ add missing external statement. reported by bokyun seo (kaeri) */ acer -- 26jun00 */ add missing line for the sequential (n,2n) reactions for be-9. */ this line was accidentally removed in njoy 94.19 (jan96). the */ error was continued through njoy 97 and 99. discovered by */ bob little (lanl). */ acer -- 26jun00 */ fix anisotropic photon production (e.g., endf c,n,o) *ident up12 */ njoy -- 27jul00 */ fix two typographical errors in the 64-bit version of the */ slatec math library. reported by piet de leege (delft). *ident up14 */ acer -- 20jul00 */ acer fails if you run it on a pendf tape that only has the */ single reaction mt=2 (elastic). this can happen for he-4 if you */ don't run heatr, thermr, or gaspr first. found by gudowski. */ acer -- 20jul00 */ acer fails for mf=6, law=2, lang>0 (angular distribution with */ tabulated cosines). the only known example is mt=51 for pb-208 from */ release 6 of endf/b-vi. found by waclaw gudowski. */ acer -- 21jul00 */ an error was included in up9, which added a capability to handle */ anisotropic charged-particle emission represented using legendre */ polynomials. the update disabled the case of isotropic */ charged-particle emission, which occurs in a number of important */ materials from release 6 of endf/b-vi. the symptom is a serious */ clobbering of the ace file, such that it cannot even be read into */ mcnp or even printed using acer. also reported by gudowski. *ident up15 */ heatr -- 31jul00 */ incorrect initial value found by m.mattes (u.stuttgart). */ increase the allowed number of legendre terms in h6ddx */ to handle the new jeff-3t fe-56 evaluation. */ watch for ill-defined vertical segments in distributions. these */ have been seen in zr90 from cendl3 and fe56 from jeff3. actually, */ the evaluations should be fixed to avoid such features, because */ we don't really know what y value to select in the vertical */ segment. we choose to just move the second energy of the double */ point up a little. we only print the diagnostic once to keep the */ output cleaner, but there could be more than one vertical segment. *ident up16 */ groupr -- 31jul00 */ watch for ill-defined vertical segments in distributions. these */ have been seen in zr90 from cendl3 and fe56 from jeff3. actually, */ the evaluations should be fixed to avoid such features, because */ we don't really know what y value to select in the vertical */ segment. we choose to just move the second energy of the double */ point up a little. we only print the diagnostic once to keep the */ output cleaner, but there could be more than one vertical segment. *ident up17 */ acer -- 31jul00 */ fix an incorrect index in the law=61 section for the primary */ particle. the effect of this error is to give an incorrect */ angular distribution for the energy points with scattering */ probability zero (which should be isotropic). this change */ is strictly cosmetic and shouldn't affect any results. */ acer -- 31jul00 */ we want to use the compact law=4 for isotropic charged particle */ distributions, and the more general law=61 for anisotropic cp */ distributions. unfortunately, we can't tell which is which */ without reading past the first few energies for some evaluations. */ the need for this patch was first noticed by jeff3 fe56. *ident up10 */ leapr -- 13jun00 */ fix two incorrect constants in leapr. one affects cases with */ diffusive effects, and it has been incorrect since njoy97.0 */ (oct 97). the other affects cold hydrogen calculations, and it */ has been incorrect since njoy97.53 (dec98). *ident up11 */ acer -- 26jun00 */ fix an error in determinining which reactions have to be */ converted into law=7 format when using newfor=0. because of */ overzealous code cleanup, acer is trying to convert sections */ with the kalbach representation in addition to sections with */ tabulated angular distributions. */ acer -- 26jun00 */ during the cleanup of the topfil routine, the logic to process */ sections of file 6 using law=2 (two-body distributions) into */ equally probable bins for newfor=0 was omitted. this shows up */ for evaluations that use mf6/mt51, etc., to represent inelastic */ levels. */ acer -- 26jun00 */ fix an error in the reaction naming. it affects mt=44 (n,n2p) and */ mt=45 (n,npa). this problem was introduced in may of 1995. */ examples of cases that use these reactions are later releases of */ endf/b-vi al-27. */ acer -- 26jun00 */ add missing external statement. reported by bokyun seo (kaeri) */ acer -- 26jun00 */ add missing line for the sequential (n,2n) reactions for be-9. */ this line was accidentally removed in njoy 94.19 (jan96). the */ error was continued through njoy 97 and 99. discovered by */ bob little (lanl). */ fix anisotropic photon production (e.g., endf c,n,o) *ident up12 */ njoy -- 27jul00 */ fix two typographical errors in the 64-bit version of the */ slatec math library. reported by piet de leege (delft). *ident up13 */ reconr -- 12jul00 */ if a reaction uses histogram interpolation, reconr tries to */ change it to linear interpolation by moving each point down by */ one in the seventh place and and adding a point higher by one in */ the seventh place. if there is already a point in the evaluator's */ grid higher by one in the seventh place, the algorithm gets */ confused. this currently occurs for carbon from release 6 of */ endf/b-vi. the symptom is an infinite loop while processing */ mf=12,mt=51. we found this at los alamos, and skipped over the */ problem by temporarily patching the evaluation. more recently, */ it was re-reported by waclaw gudowski, and now we are making a */ real fix for the problem. *ident up14 */ acer -- 20jul00 */ acer fails if you run it on a pendf tape that only has the */ single reaction mt=2 (elastic). this can happen for he-4 if you */ don't run heatr, thermr, or gaspr first. found by gudowski. */ acer -- 20jul00 */ acer fails for mf=6, law=2, lang>0 (angular distribution with */ tabulated cosines). the only known example is mt=51 for pb-208 from */ release 6 of endf/b-vi. found by waclaw gudowski. */ acer -- 21jul00 */ an error was included in up9, which added a capability to handle */ anisotropic charged-particle emission represented using legendre */ polynomials. the update disabled the case of isotropic */ charged-particle emission, which occurs in a number of important */ materials from release 6 of endf/b-vi. the symptom is a serious */ clobbering of the ace file, such that it cannot even be read into */ mcnp or even printed using acer. also reported by gudowski. *ident up15 */ heatr -- 31jul00 */ incorrect initial value found by m.mattes (u.stuttgart). */ increase the allowed number of legendre terms in h6ddx */ to handle the new jeff-3t fe-56 evaluation. */ watch for ill-defined vertical segments in distributions. these */ have been seen in zr90 from cendl3 and fe56 from jeff3. actually, */ the evaluations should be fixed to avoid such features, because */ we don't really know what y value to select in the vertical */ segment. we choose to just move the second energy of the double */ point up a little. we only print the diagnostic once to keep the */ output cleaner, but there could be more than one vertical segment. *ident up16 */ groupr -- 31jul00 */ watch for ill-defined vertical segments in distributions. these */ have been seen in zr90 from cendl3 and fe56 from jeff3. actually, */ the evaluations should be fixed to avoid such features, because */ we don't really know what y value to select in the vertical */ segment. we choose to just move the second energy of the double */ point up a little. we only print the diagnostic once to keep the */ output cleaner, but there could be more than one vertical segment. *ident up17 */ acer -- 31jul00 */ fix an incorrect index in the law=61 section for the primary */ particle. the effect of this error is to give an incorrect */ angular distribution for the energy points with scattering */ probability zero (which should be isotropic). this change */ is strictly cosmetic and shouldn't affect any results. */ acer -- 31jul00 */ we want to use the compact law=4 for isotropic charged particle */ distributions, and the more general law=61 for anisotropic cp */ distributions. unfortunately, we can't tell which is which */ without reading past the first few energies for some evaluations. */ the need for this patch was first noticed by jeff3 fe56. *ident up18 */ acer -- 03aug00 */ some jef, eff, and jeff evaluations contain a redundant reaction */ mt=10 that gives the continuum neutron production. it is */ necessary to exclude this reaction from the reconstructed total */ cross section and to omit the associated energy-angle distribution. */ otherwise, the continuum neutron production will be counted twice. */ acer -- 03aug00 */ this change can fix an infinite loop during acer plotting */ acer -- 03aug00 */ for file 6 sections with only one subsection, the mt number is */ set to zero for the messages about energy-dependent yields. */ this is a trivial cosmetic patch and doesn't affect results. */ acer -- 03aug00 */ this change is needed to handle nubar for jendl-3.2 u-235. it */ was originally made at los alamos in april, but somehow didn't */ make it to the official update file. *ident up19 */ heatr -- 03aug00 */ as noted above, some jef, eff, and jeff evaluations use the */ redundant mt=10. this value needs to be excluded from the */ heating and damage calculations. *ident up20 */ acer -- 16aug00 */ there is an error in processing angular distributions using ltt=3 */ when newfor=0 (mcnp4b compatibility). the extra tosend causes the */ code to skip over the first reaction after the elastic mf=4. */ this leads to a bad tyr=0 value in the ace file, which causes */ mcnp to issue a confusing error message about "sabcol," even */ when s(alpha,beta) is not being used. this problem occurs only */ when processing the 150-mev evalutions from endf/b-vi.6. it */ is probably best to use release 5 with mcnp4b anyway. the */ release 5 and 6 data are identical below 20 mev. It should be reiterated that Release 6 of ENDF/B-VI differs from Release 5 only in the addition of data from 20 to 150 MeV. People who only need the conventional 20-MeV data, or people processing for MCNP4B, may want to run from Release 5 and avoid the additional complexity and bulk in the files that are produced from Release 6. */ acer -- 16aug00 */ the code is finding the wrong value for the lct parameter (lab */ or cm frame) when processing file 4 angular distributions if */ the section is fully isotropic. this can result in an incorrect */ value for the ace tyr parameter, which can result in an apparent */ error from sabcol, even with no s(alpha,beta) data in the problem. */ this problem was introduced while the njoy97 coding was being */ converted to block structuring. */ acer -- 16aug00 */ the consistency check for incorrect reference frame should take */ place for isotropic distributions also. sometimes, this check is */ not a real error. users should check the evaluation to see if the */ reference frame is really as intended by the evaluator. as fixed, */ this check would have found the two problems above! */ acer-- 16aug00 */ we are not currently handling law=5 for energy distributions. */ this occurs for u-233 fission from jef-2.2. the evaluation can */ be patched by converting the lf=5 part of the distribution to */ lf=1, which is sampled much better by mcnp using cummulative */ probability distributions anyway. */ acer -- 16aug00 */ there are some additional places where skip6 should be skip6a. */ see up6 above. this shows up for endf be-9 with newfor=1. */ acer -- 16aug00 */ there is an error in the law=7 part of up11. this shows up */ when processing endf be-9 using newfor=0 */ acer -- 16aug00 */ there is an error in the new skip6a routine introduced by up6 */ that shows up when processing sections with law=7 with newfor=0. */ acer -- 16aug00 */ missing initialization in ptlegc (this could affect incident */ charged particles on some systems). NJOY99.20 has been tested rather extensively with MCNP4B and MCNP4C. It has successfully processed all the materials in ENDF/B-VI Release 5 (which has Emax=20 MeV), many important materials from JENDL-3.2, and a few problem materials from JEF and JEFF libraries into ACE formats for both 4B and 4C. All of these ACE files have been staged into MCNP and run for a few cycles to make sure that the format is reasonably correct. It has also been used recently for high-energy and charged-particle processing, but without testing in MCNP or MCNPX as yet. A few problems are known, and several materials had problems with the evaluation (e.g., carbon). *ident up21 */ acer -- 28sep00 */ the self-shielded cross sections are not being printed out */ correctly for the reactions, but total is ok. the gendf */ file is ok, so libraries made with njoy99 are ok. *ident up22 */ reconr -- 28sep00 */ add capability to handle the new extension to the reich-moore */ resonance format that uses the sign of aj to designate which */ channel spin to use for a particular resonance. based on */ coding provided by nancy larson, ornl. This is a new ENDF format approved at the 1999 CSEWG meeting. There are no official evaluations that use it as yet, but it is felt to be important for Al-27 and Si-29. We are adding it to NJOY now so as to be ready when the new evaluations appear from ORNL. *ident up23 */ gaminr -- 28sep00 */ allow for up to 400 groups (added by request) *ident up24 */ dtfr -- 28sep00 */ allow for up to 400 groups (added by request)