WIMS-D Library Update Project (WLUP)
History and Current Status as
of 31 January, 1999
Andrej Trkov
"Jozef Stefan" Institute
Ljubljana, Slovenia
Introduction
The WIMSD family of codes is one of the few reactor lattice codes that are available on non-commercial basis. Recently, the WIMSD5B version has been released from the OECD/NEA Data Bank. Its major improvement lies in machine portability and a few other improvements. It superseeds its predecessor WIMS-D/4, which was released from Winfrith for IBM machines and has been adapted for various computer platforms in different laboratories. Unfortunately, a few errors were introduced into the code in the conversion process, therefore the use of WIMSD5B is highly recommended.
WLUP is not entirely new. It started officially in the early 90's [1,2] when dr. Ganesan promoted the idea at the Nuclear Data Centre of the International Atomic Energy Agency (IAEA). Dr Andrej Trkov from the "Jožef Stefan" Institute co-ordinated the work and was supported by the (IAEA) through a research contract. It is important to note that except for the co-ordination work under contract, the participants on a voluntary basis contributed all other activities. In 1996 dr. Viktor Dimic (now retired) of the Physics Division of the IAEA took an initiative [3] to set up WLUP as a Co-ordinated Research Project (CRP). Such an organisational scheme would speed up the work on completing the updated WIMS-D library and provide the users with a product that can improve the quality of thermal reactor lattice calculations. After the departure of dr. Dimic from the IAEA, the task was taken over with much enthusiasm by dr. Brian Dodd.
The project is divided into several stages:
Stage 1 - Benchmarks
Stage 2 -
Data
Processing Codes
Stage 3 -
Major
Reactor Materials Replacement
Stage 4 -
Actinides
and Fission Products
Stage 5 -
Structural
Materials
Stage 6 -
Other
Moderators and Missing Materials
Stage 7 -
Final
Benchmarking and Documentation
For data processing details see the section on Data Processing Procedures.
The project stages and their current status are described below in more
detail.
The objective is to define simple lattice benchmarks that can be modelled with WIMS.
Use the "1986" and/or the
"1981" version of the WIMS-D library.
Optimise WIMS input models
on the basis of a sensitivity study on the WIMS input parameters. Under
no circumstances should physically inferior models be accepted because
of a "better agreement" with the reference results.
Provide a post-processing
code so that the calculated results can readily be extracted from the WIMS
output and compared to the reference results.
Post processing of the WIMS results with a computer code allows the analysis to be automated and performed for all benchmarks in sequence. Furthermore, it eliminates the possibility of different interpretations of the results by various authors. If the processing is found inadequate, the code can be upgraded an applied to any library being tested.
History
In Stage 1 of the project extensive work was performed to define the benchmarks [1],[2]. This was a crucial step in the project in which more than 20 laboratories participated. The work was co-ordinated by dr. Matjaž Ravnik from the "Jožef Stefan" Institute. Five light water moderated Uranium metal and Uranium oxide lattices were specified. The participants were asked to provide results for:
Standard input: the purpose
of the exercise was to identify possible differences between different
versions of the WIMS-D/4 code. Note that WIMSD5B was not available at the
time.
Optimised input: the participants
were asked to prepare WIMS inputs that would model the benchmarks as accurately
as possible.
Additional benchmarks were added later. Some of them still require an independent review. Currently, the following are defined:
TRX1
- TRX-1 Light water moderated Uranium metal fuel hexagonal lattice.
RPINU - Rowlands light water
Uranium oxide pin-cell benchmark.
RPINPU1 - Rowlands light water
mixed oxide pin-cell benchmark
RPINPU2 - Rowlands light water
mixed oxide pin-cell benchmark [4].
TRX1
- TRX-1 Light water moderated Uranium metal fuel hexagonal lattice.
TRX2
- TRX-2 Light water moderated Uranium metal fuel hexagonal lattice.
BAPL1
- BAPL-1 Light water moderated Uranium oxide fuel square lattice.
BAPL2
- BAPL-2 Light water moderated Uranium oxide fuel square lattice.
BAPL3
- BAPL-3 Light water moderated Uranium oxide fuel square lattice.
DIMP1A
- DIMPLE S01 light water moderated Uranium oxide fuel square lattice.
BUCR1A
- OECD/NEA Burnup Credit Criticality Benchmark
BUCR1B - OECD/NEA Burnup Credit
Isotopic Composition Prediction Benchmark.
PURECY - OECD/NEA Plutonium
Recycling Benchmark
Current Status
The proposed benchmarks are by no means final, but if modifications are considered necessary, they must be supported by solid evidence. To propose a modification to an existing benchmark, please follow the guidelines in the "Benchmark Revision Proposal Form".
An extension of the list of benchmarks would be desirable. Light water
reactor Uranium lattice criticality and burnup benchmarks are covered reasonably
well. Verified benchmark models for criticality prediction with other moderators
and fuel types need to be specified. To add a new benchmark, please follow
the guidelines in the "New Benchmark Proposal Form".
The objective is to check the definitions of the multigroup constants in the WIMS library and to upgrade the data processing codes. Note that at the start of the project there were no validated codes, which could process the evaluated nuclear data files to prepare a WIMS-D library. In the literature one could find a few attempts at improving the library, but none of them could provide a complete and reliable solution to the problem.
History
In Stage 2 of the project the use of various data processing codes was investigated to analyse their capability to prepare multigroup constants for WIMS. Seven laboratories participated, each with its own code system.
The overall conclusion was that none of the codes was adequate and required extensive updates to meet all the requirements for the preparation of multigroup constants for the WIMS-D library. At that time NJOY-91 was released, which included the WIMSR module. The decision was taken to adopt NJOY as the data processing system and to upgrade the WIMSR module as necessary because:
NJOY is the "state of the
art" data processing system.
The extensions due to format
changes in the evaluated nuclear data files are maintained by the author.
It is a widely used system.
The feedback from the users helps the author to quickly eliminate possible
errors in the code.
Upgrades to the existing
codes were necessary anyway, therefore adopting WIMSR of NJOY-91 as a starting
point to prepare the multigroup constants for the WIMS-D library was no
disadvantage.
Unfortunately, NJOY-94 was released with major changes to the WIMSR module, so the updates based on NJOY-91 were not compatible. After a careful examination of the changes in the source code a new set of updates for WIMSR of NJOY-94 was produced. Eventually, the author of NJOY accepted these updates and they form a standard part of NJOY-97, which is the current version of the NJOY data processing system, for which some additional updates are available.
The fission product yield and decay data are not handled by NJOY. In fact, there is no generally available standard code to provide the necessary information for the WIMS-D library. The AVRFPY code was developed, which has the capability to produce the appropriate fission product yields from the independent or cumulative yields and the decay data in ENDF files. It also performs checking and data intercomparison. Also, a set of criteria can be defined to identify the most important fission products that need to be treated explicitly. Except for a few additions, the list of fission product nuclides defined in the "1986" WIMS-D library is found sufficient.
As a by-product of the data intercomparison task the WILLIE code for WIMS-D library maintenance was developed. It allows various operations on the data in the WIMS-D library such as material deletion, insertion, replacement, library conversion to binary and vice-versa, consistency checking, etc.
Current Status
The NJOY data processing system is found adequate and reliable. The WIMSR module was used by dr. S. Bhuiyan at Oak Ridge and by dr. F. Leszczynski. A few minor corrections were suggested, but no major problems were reported.Some additional changes were made to extend the capabilities of the WIMSR module to transfer the burnup data to the output file. This significantly simplifies the input data maintenance. It is recommended to use version NJOY97.62 or higher.
The AVRFPY code performs satisfactorily, but an independent review would be desirable.
The WILLIE library maintenance code is found useful and reliable.
Major Reactor Materials Replacement
The objective is to replace the main reactor materials in the existing WIMS-D library for testing purposes. These materials are Hydrogen bound in water, Oxygen, Aluminium, Uranium-235 and Uranium-238.
History
After an extensive parametric study to investigate the effects of various data processing options on the integral benchmark results and after extensive discussions with other participants and the author of the code, the NJOY inputs for the main reactor materials were finalised. Optimisation of the data processing options was based on the Numerical Benchmarks. The actual inputs can be found in the master NJOY input files (see the section on Data Processing Procedures) under labels H_H2Os, O_016s, Al_27s, U_235a and U_238a, respectively. The last two inputs are identical to those under labels U_235g and U_238g, except that in the later two the burnup and fission product yields are added.
Current Status
Numerical benchmark results show that the multiplication factor is somewhat underpredicted in WIMS with the 69-group library. It is not possible to choose data processing options, which would improve the situation in a consistent way. The main cause of this underprediction seems to be:
The 69-group limit for
the cross sections.
The resonance treatment in
WIMS (possibly the method of calculating the Bell factor).
The use of a single fission
neutron spectrum for all fissile nuclides.
It is also observed from independent studies that most of the presently available evaluated nuclear data files tend to underpredict the criticality of Uranium fuelled lattices. The WIMS results are consistent with this observation. Again, the error introduced is systematic and does not affect significantly the quality of the updated library.
The proposed inputs are by no means final, but if modifications are
considered necessary, they must be supported by solid evidence. To propose
a modification to an existing NJOY input, please follow the guidelines
in "NJOY Input Revision Proposal Form".
Actinides and Fission Products
The objective is to replace the actinides and fission-products cross sections, fission product yields, decay and burnup data.
History
The IAEA sponsored dr. D.L. Aldama to spend three months at the "Jožef Stefan" Institute. During this time the Burnup Credit Criticality Benchmark model for WIMS was developed. The multigroup constants for all the actinides and fission products were replaced. Some decay constants and burnup chains were corrected. Preliminary results showed a marked improvement in modelling the reactivity changes, compared to the original WIMS-D library [6].
The fission product yields were generated with the AVRFPY code for all major actinides. The decay and burnup data were added to the NJOY inputs.
The energy release per fission was extracted from the ENDF files for all fissile fissible materials. The energy carried away by the neutrinos was neglected, but arbitrarily 10.5 MeV was added to account for the energy released from the capture products. Another 1 MeV was added to the fissible materials with threshold fission to account for the incident neutron kinetic energy.
This allows the replacement or insertion of all the data in the WIMS-D library in a single processing sequence with the WILLY code, starting from the "1986" version of the WIMS-D library.
Current Status
The Actinides data processing methods require an independent review. To propose a modification to an existing NJOY input, please follow the guidelines in "NJOY Input Revision Proposal Form".
The problem of the lumped fission products has not been addressed yet.
The calculated yields require an independent review.
The treatment of the energy released from fission is rather crude and
needs to be refined, but it is not expected to affect the final results.
The objective is to replace the structural materials.
Current Status
The NJOY inputs are approximately defined for some of the structural
materials. Appropriate input instructions for WILLIE should be defined
to merge the isotopic cross sections for elements and/or mixtures. Careful
checking is required, particularly for structural materials with significant
resonant behaviour.
Other Moderators and Missing Materials
The objective is to process other moderators and add missing materials for different applications.
Current Status
NJOY inputs need to be defined.
Additional materials for special applications need to be identified.
Final Benchmarking and Documentation
The objective is to complete benchmark testing and prepare the library documentation.
Current Status
Underlying materials on the work completed so far is available, but
a concise final document for the end-user needs to be produced.
The objective is to simplify the data processing task and input file maintenance. This is achieved by the following
A single NJOY master input
file contains inputs for all materials to be processed. The inputs are
separated by special delimiter strings. For details see the section
on the Current Status below.
The inputs for individual
materials are extracted from the master file with the NJISPL
code.
The multigroup constants
for updating the WIMS-D library are generated by running NJOY for all materials
in sequence. A special batch procedure is available for this purpose. For
details see the section on the Current Status below.
A single set of WILLIE
input instructions is provided to insert or replaces the multigroup
constants in the "1986" WIMS-D library..
A batch procedure is available
to execute all standard benchmark test cases.
Current Status
The programs, inputs, batch procedures and script files include descriptive comments wherever possible. For most of the programs a brief description of the code and instructions for use can be retrieved quickly from the source by using a system search utility ("grep" on Unix, "find" on Dos, "search" on VMS, etc.) to list the records beginning with "C-".
References