The 2008 ANSWERS Seminar was held at the Wessex Hotel in Bournemouth, a location near to the superb Dorset coast and in close proximity to the town centre.
Over one hundred delegates attended all or part of the three-day meeting, with the programme covering the shielding, reactor physics and criticality technical areas. Programme topics included ANSWERS code developments, code applications, related general interest issues and software demonstrations. In addition to the seminar itself, parallel meetings of the UK Shielding Forum, the UK Working Party on Criticality and the UK Nuclear Science Forum were hosted at the same venue.
The emphasis of the first day of the 2008 ANSWERS Seminar was Radiation Shielding, which was hosted by Pat Cowan, ANSWERS Applications Area Manager for Radiation Shielding. Pat introduced a variety of presentations on the recent developments in the ANSWERS Shielding codes - MCBEND and RANKERN and supporting tools - and a number of applications using the ANSWERS codes.
The morning started with a presentation on “Ten Statistical Tests in MCBEND” by Ted Shuttleworth. The talk gave a full description of the ten tests and the results for a number of examples. This new development will be available in the next release version of MCBEND. This presentation was followed by an applications talk by Andy Smith of Sellafield Ltd. Andy talked about “Dose Uptake and Shielding Assessment for Transports by Road and Sea” using MCBEND. The next presentation was carried out by Adam Bird, who spoke about the Automatic Test Tool being developed by ANSWERS and its use in the current release update of RANKERN 15A. It is expected that the Automatic Test Tool will help to reduce the effort required for code release testing in the future.
Following the morning break, Geoff Dobson gave a presentation on a development he has been working on in MCBEND – “The Detailed Thermal Point Energy Adjoint”. Geoff presented the theory behind the method, the work carried out to date and the proposed future work. This was followed by a presentation made by Joanna Alderman (Serco, Quedgeley) on MCBEND flux calculations that had been carried out for Wylfa. Jo demonstrated that new features in MCBEND 10A significantly cut down on the complexity of the whole reactor model, compared to previous work. Jo had used Visual Workshop to display impressive graphical images of her geometry model. Adam Bird ended the morning session with a presentation entitled “Tetrahedral Mesh Import into MCBEND and MONK“. This gave an update on last year’s presentation and demonstrated that we are now getting excellent comparisons between MCBEND calculations run with traditional FG and the Tetrahedral Mesh for dose-rates and energy deposition on a transport flask. A lot of work has been done to speed up the processing of the Tetrahedral mesh over the past year.
The afternoon session started with complementary talk on Importing CAD Data into MCBEND given by Keith Searson, Sellafield Ltd. This development involves tracking through the NURB surfaces directly. This was followed by a presentation given by George Wright describing the role of the ANSWERS SSQA team. George gave an example of a typical hotline query and talked through the various stages of it until its final resolution. Next Adam Bird was presenting again. This time he gave a demonstration of the latest features available in Visual Workshop and the ‘results display’ which is under development. The first release of Visual Workshop is expected shortly.
Following the afternoon break, a presentation given by Ted Shuttleworth described an MCNP Geometry to FG converter that he has developed. This is compatible with many MCNP features and has been used to convert a number of models. Next the “HSE's Safety Assessment Principles for Radiological Protection” were presented by David Simister of NII. Finally, Pat Cowan ended the day by describing the current status and future plans for the ANSWERS Shielding Area.
The second day of the seminar, The Reactor Physics day, started with a presentation by Pete Smith of Serco on the new ECCO module to be included in WIMS10. This module provides an alternative resonance self shielding module to the PRES and RES modules. Its advantage is that it uses a 1968 energy group data library with sub-groups rather than the WIMS standard 172 groups. This also gives improved accuracy at higher energy making WIMS suitable for the analysis of fast neutron systems. The main disadvantage is the increased computer time taken by the module. The second presentation was given by Pavel Mikolas from Skoda and reported the analysis of measured isotopic compositions from fuel assemblies in the Novovoronezh unit IV VVER 440 reactor. This programme of work was performed jointly between the United States DOE, Moscow ISTC and the Dimitrovgrad Research Institute of Atomic Reactors. Several different codes schemes were used in the analysis including WIMS9. The overall results were not as good as for previously analysed PWR reactors and there are thought to be some problems with measured values.
The final presentation before the mid-morning coffee break was given by Gwendoline de Hemptinne of Tractebel and discussed the WIMS9 analysis of the fuel inventory measured in an 8% by weight Gd2O3-UO2 fuel pin irradiated in the Tihange 1 reactor as part of the gadolinium demonstration programme. The measurement data included pin radial distributions for several Gd and Uranium isotopes. Overall agreement between calculation and measurement was particularly satisfactory. Sensitivity investigations of the analysis methods concluded that 12 rather than 6 energy groups were required and that the sub-group method, accounting for the resonance self shielding skin effect, was needed to give correct isotopic radial profiles.
The first presentation following the coffee break was given by Massimo Fatone of Colenco and discussed spatial meshing effects and the radial reflector treatment applied to PWR reactors. The presentation described the reactor geometry and burnable poison loading as well as the WIMS9 calculational; route used to analyse the reactor. The k-effective was found to be between 600 to 900 pcm over estimated if insufficient radial meshes were used. In addition, insufficient radial meshes resulted in a radial power profile that was too peaked. The talk finished with the presenter inviting an audience discussion on how best to apply the radial reflector treatment.
The next presentation was given by Paul Bryce of British Energy on scoping calculations to investigate a WIMS/PANTHER methodology for the treatment of MOX fuel. Paul first described some current and previous studies with pin cells identifying a stronger condensation effect for MOX rather than UO2 pins and a reactivity bias introduced by the PERSEUS module due to the assumption of a cylindrical outer boundary. It was confirmed by the chairman that this bias could be removed if required. Paul next considered assembly and supercell calculations and explained that the approximation of a zero current assembly boundary condition, currently assumed for UOX cores, was a poor approximation for cores containing a mixture of UOX and MOX assemblies and that this could lead to errors of ~ 7% for pin powers when using traditional pin power reconstruction techniques. An alternative approach was to assume a zero current approximation at the centre of assemblies and to solve a series of supercells consisting of four quarters of assemblies modelling the interfaces between the MOX and UOX assemblies. Using an appropriate selection of embedded supercells, pin power reconstruction could then be performed from the whole core smeared solution. Some simple demonstration calculations were described of this technique and showed that errors on pin powers were reduced to ~ 1%. Further work is required to develop this approach and to include a reflector treatment. The final presentation of the morning outlined further advances in nuclear data measurements and analysis. In particular the PhD programme of Tim Ware (Serco) was outlined which consists of performing cross-section measurements at the Gelina facility at Geel in Belgium using time of flight and capture/transmission yield measurements. The measurements will be analysed using the REFIT code developed by Mick Moxon to yield the cross-section resonance structure. Recent work on the REFIT code include bug fixing, updated documentation and inclusion of a standard set of test cases. The updated version of the code was released by Serco to the NEA Databank in August 2007. It was emphasised that the UK must maintain this type of measurement and analysis expertise as nuclear data has a fundamental importance in many aspects of the nuclear industry.
The first presentation after lunch was given by Dave Powney of Serco and described the neutron streaming options available in WIMS. These include the Benoist, Ariadne, Bonalumi and Behrens options each applicable to slightly different situations. Each of the methods is based on the calculation of directional collision probabilities which are used to form separate axial and radial diffusion coefficients. The application of the Benoist method was illustrated from analysis of the BICEP experiments using various enrichments of uranium metal pins in lattices with different pin sizes, pin pitches and air channel sizes. Depending on the configuration the streaming correction to the k-effective can be worth several thousand pcm. The second presentation in the session was presented by Paul Smith of Serco on behalf of Matthew Eaton of Imperial College who was not able to attend the Seminar. The presentation discussed the modelling of a postulated MOX powder system criticality accident using the FETCH code. The FETCH code uses a finite element transient criticality model including a consistent treatment of coupled spatial physics including the full transport equation and multi-phase gas-liquid or particle flows. Compressible or incompressible phase are modelled and energy, volume fraction and momentum equations are solved for each phase. The movement of delayed neutron precursors is allowed within one or more of the phases. Also modelled are bubbly free surface fissile flows, porous media, granular flows, shock waves and structural effects, heat transfer between phases, phase changes and their detailed dynamics and complex powder systems. The responses of a layered powder system of plutonium enriched MOX fuel between zinc stearate layers in a cone was considered to two different perturbations. The first perturbation was a 3$ step reactivity insertion with the second being mixing of the system by a screw mixer placed on the inner surface of the cone. Films of the calculated response showed the evolution of the system solid volume fraction and temperature, the onset of fission rate fluctuations and evolution of delayed neutron concentrations, powder phase velocities and system pressure changes. The overall observations were that in mixing the zinc stearate initially acted to increase moderation providing a positive reactivity feedback but as the stearate became fully mixed reactivities decreased. Subsequent fission peaks were also noted and could lead to temperatures in excess of 6000°C causing melting and vaporization of the powders.
The final presentation before the afternoon tea break described the application of the new WIMS10 CACTUS3D module to the generation of incremental cross-sections for Advanced Candu Reactor (ACR) reactivity control devices and was presented by Brian Turland of Serco. Brian explained the AECL, who operate the ACR reactors, currently use the DRAGON code suite for their reactor analyses but require an ”alternative physics package” to support licensing of the ACR-1000 reactor. The WIMS9/PANTHER code package is being tested for this purpose. Modelling of the reactivity control devices is a true three dimensional problem as the devices enter the core vertically while the fuel is in horizontal channels on either side. The new CACTUS3D module in WIMS10 with its fractal geometry capability was used for the modelling. This allowed the direct comparison of results with those from the MONK Monte Carlo code which uses identical geometry input as well as the use of the MONK graphics facilities to display the model for verification. The required incremental cross-sections represent changes in cross-sections due to the introduction of the empty guide tube followed by the introduction of the reactivity control device itself. A comparison of the results from WIMS with those from the DRAGON code showed excellent agreement. Brian also presented some of the difficulties that had arisen during the work when first applying CACTUS3D to problems much larger than normal cell/supercell size. Larger problems were shown to need tighter convergence criteria and sub-mesh sizes of the order of 1 cm for good spatial convergence with the flat source approximation. For the future initial trials using diffusion acceleration look promising as does the introduction of a linear source approximation. Also of interest is the effect due to P1 scattering.
The first presentation in the final session of the day was in two parts, the first described the investigation of a new once through tracking method for CACTUS and the second, special options in the WIMS BURNUP module. This presentation was made by Glynn Hosking of Serco. The once through tracking method (OTM) for CACTUS has the objective of providing uniform track coverage of large three dimensional problems using as few tracks as possible to reduce running times. The tracks cross the problem only once and the method is only applicable to problems where the inward angular boundary flux condition is known such as for a free boundary. Trials against the existing tracking method (ETM) showed that for a similar number of track segments the OTM results in a better sampling of meshes and improved accuracy in tracked volumes in addition there was an improvement in the k-effective agreement with the Monte Carlo code MONK. The method shows promise for further development. The second part of the talk explained the xenon and samarium, alpha, diff and burnable poison options, poison and bp in the BURNUP module of WIMS. The xenon and samarium options introduce equilibrium concentrations of the isotopes at the start of a burnup calculation to avoid the need for short times steps while the concentrations build up. The alpha option is applicable to nuclide with large resonances outside the resonance range treated by WIMS such as Pu240 and Pu242. The option avoids the need for short time steps due to changes in the resonance self shielding as the nuclides deplete. The poison options again try to limit the need for short timesteps, however, this time by allowing for the change in flux in the region of the poison to be re-calculated during the timestep. The poison option is used where poisons are separate from the fuel and the bp option where poisons are close to or mixed with the fuel. The diff option allows multi-definitions of a fuel composition that occurs multiple times in a model to allow each occurrence of the material to follow an individual burnup history. The diff option assumes the materials, however, remain sufficiently similar in composition to use their average composition to perform resonance self shielding.
The second presentation was given jointly by Paul Bryce of British Energy and David Powney of Serco and described progress with the development of a whole core AGR capability using WIMS/MAX/PANTHER. Paul explained that current methods for AGRs ARGOSY/PANTHER although giving acceptable results for normal operation are not validated for the graphite oxidation affecting AGR reactivity effects nor for the preparation of the safety case for steam ingress. The required route needs to generate few group cross-sections for fuel representing standard perturbations; to model interstitial absorbers as well as absorbers in empty fuel channels; to generate smeared reflector data; to apply sub-group theory to properly represent spatial self-shielding; to model AGR 3D end gap effects and to model burnable poisons wrapped azimuthally around the fuel. The MAX module in WIMS provides the ability to include all these effects. Currently routes for fuel data, rod data and reflector data have been developed for application in PANTHER. Outstanding work items include the verification/validation of each of the “building blocks” against reference WIMS and/or Monte Carlo calculations, the resolution of axial streaming effects and the comparison of the new WIMS model against plant data. In the second half of the presentation David Powney presented more detail on the calculational route employed and details of the MAX method. David explained that although the required model could be prepared using the MONK Monte Carlo code computer running times would be too long and as the CACTUS3D module was not available at the start of the project, it was decided to use the MAX module due to its 3D geometry capability. The module uses an unperturbed deterministic solution as a base case from which it builds up a series of perturbations, scoring only the change in flux and k-effective using Monte Carlo methods. The perturbations modelled include 2 to 3 dimensional modelling as well as burnup, standard core parameter changes such as temperature, the introduction of burnable poisons as well as the effect of steam ingress. The resulting calculational route becomes very complex and WIMSBUILDER is used to simplify the production of the WIMS input decks. In conclusion the mechanics of the route has been established but some aspects of validation and verification need to be completed.
Peter Smith of Serco then made a short presentation on the role of the SSQA team within ANSWERS. The SSQA team run the customer hotline, support the customer web pages and organise training courses presented by the technical experts. Peter explained how every hotline query was logged in a “Sunrise” database system with a unique reference number before being referred to an appropriate expert. Queries are then monitored to ensure an appropriate response and an agreed course of action with the customer. Between April 2007 and March 2008 in excess of 430 hotline enquiries were logged and processed. In the same period the SSQA team issued approximately 950 code licences. Peter finished the talk emphasising how important the hotline service is and that Serco would be please to here of any ideas to improve the service.
The day was completed with a short presentation from Tim Newton, the host of the Seminar reactor physics day, on the contents of WIMS10 relative to WIMS9 and including future development plans for WIMS.
The third and final day of the seminar, The Criticality Day began with the a MONK Status presentation giving an update on the code and any relevant memos since the previous seminar.The next presentation was from Liang Zhi (representing the China Nuclear Power Engineering Co Ltd) on their experience of using MONK for burnup calculations of the AFA-3G element and a results comparison with other codes.
This demonstrated the effective use of MONK in these situations and highlighted the general good agreement in number density for most nuclides, but it did raise a question regarding a few select nuclides. The future work includes experimental investigations which should answer some of the questions raised. The final talk of the first session was on CAD development and was presented by Adam Bird of Serco. He presented a talk on the TETMESH hole developed by Serco, and the NURBS processing method developed by Sellafield Ltd. Adam noted that both these methods have been successfully tested and the examples given highlighted their merits and problems such as run-time efficiency. The talk concluded by stating that with further development they will both improve in performance and be useful additions to the Fractal and Hole geometries already available.
The first presentation of the second session was on the use of MONK 8B Calculations in the design of a Furnace Load Station at Springfields Fuels Limited by Anna Booth from Nexia Solutions. This walked us through the various stages in the testing and design of the load station and the offload station as the safety review highlighted problems with the existing and initial re-designed equipment. Anna noted the extensive use of CODEMORE with MONK to speed up the calculations for the criticality safety assessments. The next presentation was on the Critical Mass Calculations for Americium from Chris Hodkinson from AWE. Chris described their findings from analysing Am241, 242m and 243 using DICE data in MONK and comparing with MCNP. He found good agreement with JEF2.2 and JENDL3.2 libraries, but poor agreement for Am242m with ENDF/B-VI due to an assumption made with the spectrum submitted to NJOY. Chris also highlighted the wide variation in minimum critical mass with library.
The comment was made that these findings support Christopher Dean’s presentation in 2003 that concluded that BINGO is the preferred route for higher actinide systems. The third presentation of this session was from Richard Neal of Sellafield Ltd on the run-time Doppler Broadening being developed for the BINGO collision processor. BINGO currently uses spot-temperature data with no interpolation. Richard described the development objectives (efficiency, accuracy, storage) and the methods used in the processing. He also described the work required to handle the unresolved resonance range thought interpolation of the sub-group data, or the use of GENEX ladders as an alternative. He identified the temperature range supported (room to 1 million Kelvin) and the current performance penalty (up to 3-4 times slower). He continued by reporting the results of comparing the method against spot data and showing excellent agreement. He concluded by listing future work that included handling spikes and discontinuities in the data, integrating the new routines in MONK and MCBEND, and validation of the resulting codes at elevated temperatures. The morning session concluded with a second presentation by Adam Bird, this time on the Visual Workshop visualisation package. He described the status of upcoming release of version 1, and gave a demonstration of the capabilities of the next version. This included the display of the birth store from MONK so the user can assess the state of settling when scoring began, and a much improved speed of response to the user input. He also showed the new surface overlay capability that could display flux profiles over a wire frame image of the model, with a range of user inputs to change how the data is shown.
The afternoon session began with a presentation by Paul Smith of Serco on the progress of the MAX into MONK task. Paul identified the problems with the current MAX implementation (such as restricted geometries), then a review of the MAX method being used, and concluded with an overview of the implementation of the method into the MONK tracking algorithm.
The next presentation was from Jim Gulliford of Nexia Solutions on the UK Working Party on Criticality (WPC). He listed the scope, objectives and a brief history of the WPC before looking at some areas of particular interest. These included Nuclear Science, Safety Methodologies, Decommissioning, Emergency Planning and Professional Skills Development. He described the workshops available and the links to external bodies, including academia. He particularly mentioned the ICNC-11 to be held in Edinburgh, where he needs support from the UK criticality community to make it a successful conference. Jim concluded by presenting the proposal to form the UK Code User Forum to allow code users to provide feedback to the code developers and vendors. The final talk of the third session was from Alwin Weiderhold of BE on the importance of bound and free hydrogen in criticality analyses. Alwin described an intercomparison he had made between MONK and HELIOS for an AGR fuel element where he identified discrepant results. He then described the analysis of a simple system comprising a fuel pin in water to identify the source of the problem. Alwin noted an increasing discrepancy with pin pitch that lead him to rerun the MONK calculations with free hydrogen, resulting in better agreement with HELIOS. His conclusions were that code users must be careful to check what hydrogen data is used when water is present because the difference in results can be significant.
The first talk of the final session was from Will Turnbull of Sellafield Ltd on Calculations on Sludge / Magnox Fuel Mixtures. He described the problem of recovering sludge from ponds and gave some detail of the model used for the calculations. Will described the initial scoping calculations to determine the most reactive configurations of fuel modelled as spheres and rods by varying their sizes and pitches. He continued by describing the analysis of the results, and concluded that Reactivity driven by the discrete fuel matrix/moderator component and that the safe parameters will have minimal impact on their planned sludge retrieval operations. The next presentation was by Peter Smith of Serco on the ANSWERS Hotline and the Verification of MONK. He described the Hotline, listed the members of Serco directly involved with it, and displayed several charts showing the number of queries by application area. He then gave an overview of the SSQA Verification of MONK for the release of version 9A_RU1, and finished with a description of the new ANSWERS Test Machine and the Auto Test Tool. One feature of this machine will be to run MONK, MCBEND and WIMS through all their Verification and Validation test sets whenever a code update is made. This will simplify checking the effect of code changes and, with the Auto Test Tool, simplify the reporting of test results.
The third presentation of this session was from Simon Warburton of the NII on the HSE's Safety Assessment Principles for Criticality Safety. Simon identified the role of SAPs, and then described in detail the two specific engineering principles and six supporting paragraphs relevant to criticality safety. These included the double contingency principle, geometrical safety, and always having a means to terminate an event should one occur. He finished by describing Technical Assessment Guides (TAGs) and how they, and SAPs, provide detailed guidance to NII assessors. The final talk of the 2008 ANSWERS Seminar was by Malcolm Armishaw of Serco on the future development plans for MONK. He described the short term plans including further BINGO related work, changes to support Visual Workshop, and the testing of the BINGO point energy burnup option. He also listed the nuclear data tasks including Doppler Broadening and a BINGO JEFF3.1 library. Malcolm also highlighted the Validation work done by David Hanlon to bring BINGO to production status. He continued by describing some of the long term plans. These included the release of MONK version 9B, further CAD enhancements, and ENDF/B-VII BINGO library, closer links to Visual Workshop, and finally continued work on investigating parallelisation options, including the use of Multi-Threading where possible (gains have been seen in Visual Workshop and BINGO using this). Malcolm concluded the session by thanking all the presenters for their efforts, all the attendees for their support, and hoped they had enjoyed the seminar. Paul Smith closed the proceedings by thanking all the presenters and attendees for making this another successful ANSWERS Seminar.