PAUL SCHERRER INSTITUT
ISSN 1423-7334 March 2003 u
PSI • Scientific Report 2002 / Volume IV
Nuclear Energy and Safety PAUL SCHERRER INSTITUT ISSN 1423-7334 _ I "I March 2003 H— 1
Scientific Report 2002 Volume IV Nuclear Energy and Safety
ed. by: Brian Smith and Beatrice Gschwend
CH-5232 Villigen PSI Switzerland Phone: 056/310 21 11 Telefax: 056/310 21 99 http://nes.web.psi.ch Table of Contents
Introduction 3 W. Kroger
Overview of Main Activities 7 K. Foskolos, W. Kroger
Risk Benefits of Climate-Friendly Energy Supply Options 15 S. Hirschberg, P. Burgherr
Development of Dynamic Probabilistic Safety Assessment: The Accident Dynamic Simulator (ADS) Tool 21 Y.H. Chang, A. Mosleh, V.N. Dang
Level-Swell Prediciton with RETRAN-3D and its Application to a BWR Steam-Line-Break Analysis... 29 Y. Aounallah, K. Hofer
RETRAN-3D Analysis of the OECD/NRC Peach Bottom 2 Turbine Trip Benchmark 41 W. Barten, P. Coddington
Reactivity Measurements on Burnt and Reference Fuel Samples in LWR-PROTEUS Phase II 55 M. Murphy, F. Jatuff, P. Grimm, R. Seiler, A. Luthi, R. van Geemert, R. Brogli, R. Chawla, G. Meier, H.-D. Berger
Experimental Investigation in the PANDA Facility of the Impact of a Hydrogen Release on Passive Containment Cooling Systems 61 O. Auban, D. Paladino, P. Candreia, M. Huggenberger, H.J. Strassberger
CFD Simulation of Steady-State Conditions in a 1/5th-Scale Model of a Typical 3-Loop PWR in the Context of Boron Dilution Events 69 T.V. Dury
Assessment of the MELCOR Code against PHEBUS Experiment FPT-1 Performed in the Frame of ISP-46 75 J. Birchley
The FUJI Project: Fuel Test for JNC and PSI — A JNC-PSI-NRG Collaboration 85 Ch. Hellwig, P. Heimgartner, F. Ingold
Fracture Toughness of Zircaloy Claddings 91 J. Bertsch, W. Hoffelner
Anisotropic Diffusion in Layered Argillaceous Rocks: A Case Study with Opalinus Clay 97 L. R. Van Loon, J. Soler, W. Muller, M.H. Bradbury
Uptake of Trivalent Actinides (Cm(lll)) and Lanthanides (Eu(lll)) by Cement-Type Minerals: A Wet Chemistry and Time-Resolved Laser Fluorescence Spectroscopy (TRLFS) Study 105 J. Tits, T. Stumpf, E. Wieland, T. Fanghanel
Publications List 111 3
INTRODUCTION
1/1/. Kroger
PSI's Nuclear Energy and Safety Research In parallel, and to assure continuation of the Department (NES) looks back on 2002 in terms of an successful collaboration with the European research impressive record, in regard to both its research and programmes, NES has participated in 27 "Expressions its strategic decisions. Besides the excellent results of Interest" for Integrated Projects and Networks of produced in the various research areas, both federal Excellence within the 6th EU Framework Programme. and external funds have been consolidated. Cuts in personnel resources are finally stabilising and, Common position adopted by PSI and the FIT with a current staffing quota of about 165 person- Board in regard to Swiss participation in GIF years per year, and roughly 7 MCHF for O&M costs, a position of equilibrium has now been reached. This • PSI considers that nuclear energy in has been confirmed both in PSI's mid-term planning, Switzerland cannot be abandoned, either for and in the joint position of PSI and the Board of the economic or ecological reasons. However, it Federal Institutes of Technology (FIT) regarding Swiss can only play a part in the future if it is further participation in the Generation IV International Forum developed (in an international framework). (GIF); see box. • PSI accepts the goals of GIF—more sustainable fuel cycle, increased safety, The strong collaboration between NES and its economics and resistance to proliferation—and traditional partners—the Swiss utilities (UAK), Nagra, and the regulatory authority (HSK)—has continued. focuses its contributions on the reduction of NES participation in the European Framework radioactive waste, and further increase in plant Programmes, and its collaboration with leading safety. nuclear supply industrial partners, also remains active. • PSI is committed to follow actively current As a result, both currently, and in the mid-term, a trends in nuclear energy production, its further substantial share (-45%) of the total costs for nuclear development at an international level, and the research at PSI have been assured. fostering of young specialists. • PSI does not intend to further reduce funding in Highlights in research and operation the field of nuclear energy. The NPP operators support PSI's involvement in research work Projects established in previous years have yielded within the GIF framework without, however, relevant and first-of-a-kind results, which have gained broad attention, both nationally and internationally, committing themselves in regard to the and which are presented in detail in this report. A few possibility of constructing new plant, even less outstanding examples are cited below: to a particular reactor type. • Along with its proven skills in nuclear • Successful first measurements with highly active samples in LWR-PROTEUS Phase II (high burn-up technology, and its first-class nuclear facilities fuel) have shown significant dependency of reacti- (i.e. the Hot Lab and PROTEUS), NES can vity on burn-up, and increasing discrepancies contribute to other related projects (e.g. between calculated and measured reactivity values MEGAPIE), and PSI's large facilities (SLS, with burn-up. As a consequence of these findings, SINQ). the Swiss utilities wish to extend this phase. • PSI's participation in GIF takes into account its • On-call calculations in the framework of the (financial) possibilities, the use of existing STARS project have been used to modify the facilities, the compatibility of running research feedwater system of the Leibstadt NPP. The programmes alongside its obligations, and the modification has been subsequently confirmed scientific challenges within its research during a turbine trip. priorities. • An international consortium has been established for the ARTIST project (aerosol behaviour in the case of steam generator tube rupture). First tests showed higher aerosol retention than expected. On the operational level, the year 2002 was marked by a series of significant events: • The MEGAPIE project remains ongoing, and plans • The Federal Institutes of Technology, to which PSI for post-irradiation examination (PIE) have now belongs, have drawn up strategic plans for the been established. Investigation of the leak which years 2004-2007. The proposed PSI contributions occurred in the LISOR loop in the Hot Lab has have been accepted by the FIT Board. The positively identified the cause of failure. planning document is currently in circulation for • PSI's contribution to the China Energy Technology comment. Programme has been completed and documented. • PSI has updated its "Guiding Principles" to include Among other results, the programme provided its own research capabilities, in particular in regard evidence for lower total costs (including to sustainable energy technologies. externalities) by using "clean coal" technologies. 4
• At the end of an intensive "bottom-up approach", dealing with generic research issues, strengthening an R&D planning document for the year 2003 has the networking within PSI, and implementing a new, been issued for the first time by NES; this exercise major activity in the framework of an international should be formalised in a more compact form in collaboration. The former featured in PSI's general the future. discussions concerning the future use of its large • Backfitting of the Hot Lab has been completed in 4 facilities for its own research purposes, the conditions years, at a cost of 14 MCHF (compared with the for the latter being established through the admission initial estimates of 3 years and 9.8 MCHF); the of Switzerland in the Generation IV International NPP operators have agreed to contribute to the Forum (GIF) Hot Lab infrastructure costs to the extent of 3 Basic assumptions MCHF per annum. It is assumed that the Hot Lab is itself one of PSI's With regard to education and training, in view of the large facilities, and thereby warrants what may be necessity to maintain a continuous education called "adequate care". Though NES increasingly programme in nuclear technology at university level, uses the spallation source SINQ and the synchrotron the Swiss utilities have conditionally agreed, following light source SLS, its research and its raison d'etre are the retirement of Prof. Yadigaroglu, to fund an ETHZ not exclusively defined in terms of these facilities. Chair in Nuclear (Systems) Engineering. The Rather, NES continues to follow its guiding principles, appointee will become a central part of the new which include contributions to the safe operation of Master of Nuclear Engineering programme, to be the Swiss NPPs, the safe disposal of radioactive offered by ETHZ and EPFL, and will be given the waste, and keeping the nuclear option open for the opportunity by NES to perform large-scale research at future. Essential elements in this line are maintaining PSI. The new Masters programme will become an its own assessment capabilities, and fostering the integral part of the emerging European Network on next generation of nuclear specialists, both through Nuclear Energy (ENEN). participation in future-oriented research, and At the technical teaching level, PSI's own Technical appropriate use of its own "large" facilities. In doing School for reactor operators has, after some years of so, the success and quality criteria adopted at PSI reduced participation, again registered an increase in (among other things, providing world-class the number of students: in 2002, courses had to be contributions in "niche" areas), and the Swiss-specific offered on an annual basis, instead of every 2 years, co-operation and financing approach (i.e. close in response to the increased demand. collaboration between utilities, regulators and the research community: the so-called "tripod") remains valid. Strategic options for the future
As expressed in the joint position of PSI and the FIT These basic assumptions and boundary conditions Board (see box), the value of nuclear energy research lead to the following conclusions: is being recognised both within PSI and the FIT • The backfitting of the Hot Lab must be completed domain, and its scope should, therefore, remain and consolidated. The satisfaction of clients must stable. This is not, however, the opinion of the Federal increase, the analytical tools must be kept state-of- Commission for Energy Research (CORE), which the-art, and mechanical testing of active materials questions, in principle, the necessity for future- must be implemented. A decision on a more oriented nuclear research, and recommends a focus fundamental backfitting, or the building of a new on safety issues relating to existing NPPs and waste Hot Lab, must be taken around 2010. management, and further recommends a reduction in public funding. An extensive presentation of PSI's • With regard to PSI's other large facilities, NES nuclear research programme, and discussions with participation in the MEGAPIE project (materials, CORE members, has already taken place. Further thermal-hydraulics, PIE) will continue. The SLS clarification will be forthcoming on the occasion of the MicroXAS Beamline will be commissioned, and be Energy Conference, in November 2003. operated both for our own research (LES, LWV), and for non-NES users. The climate for nuclear energy remains stormy and • With regard to NES's own large facilities, the uncertain. Following a vote at Cantonal level, the PROTEUS strategy for the next few years focuses licence for an exploration tunnel at the Wellenberg on LWR and Next-Generation Water Reactors; a site, with a view to a future repository for low and backfitting of the PROTEUS l&C system will take intermediate level waste, was denied, resulting in a place in 2005. For PANDA, it is necessary to political dead end. The new Atomic Law is still under develop a strategy for the time following parliamentary discussion, and will probably be subject completion of the OECD-SETH project (beyond to a referendum at the beginning of 2004. Two anti- 2006). The DRAGON/ARTIST facility is soundly nuclear initiatives, one requiring a phase-out of embedded in international projects until 2008. nuclear energy in Switzerland, and the other an • International Networking will be realised primarily extension (under more stringent conditions) of the in the framework of new initiatives: i.e. within GIF existing moratorium, will be presented for popular vote (new) and the 6th EU FWP (continuation). in May 2003.
In these, still difficult, times, greater flexibility and robustness are necessary, and sought after, through 5
the sense of "more reactor", and resized so that no Swiss participation in GIF redistribution of resources becomes necessary. The following reactor systems were selected as • IMF activities should be "rounded-up" with an promising candidates for implementation around irradiation at the Beznau NPP, provided the 2030: partners are willing to finance a large part of the • Gas-Cooled Fast Reactor (GFR) project in the sense of technology transfer. • Molten Salt Reactor (MSR) • Long-term field experiments in excavation- • Sodium Liquid-Metal Cooled Reactor (Na LMR) damaged rock zones are premature in the light of the current state of the waste disposal scene in • Lead or Lead-Bismuth Cooled Reactor (LFR) Switzerland. • Supercritical Water-Cooled Reactor (SCWR) • An XAS and XRF beamline on a high-field bending • Very High Temperature Reactor (VHTR) magnet at the SLS facility is relevant for future As selection criteria, the following goals have been research. However, being a "workhorse" is not identified by consensus: unique in Europe: such a beamline should be financed by its users. • Sustainability of fuel cycle (resources, waste) • Safety and reliability (no need for external Following the recommendations of the NES FoKo, emergency planning) NES decided to concentrate its efforts on two issues: • Economics (cost advantage, low financial risk) • High-temperature materials The main issue will be • Robustness against proliferation and terrorism damage evolution in HT-Materials (metals, • Missions (in regard to power generation, hydrogen ceramics and composite structures) under production, high-temperature process and heat, irradiation, temperature and mechanical loading. actinide management) In-situ mechanical testing under temperature and • Unit size (small, medium, large) irradiation is foreseen, followed by micro-structural analyses in the Hot Lab, and at SINQ and SLS. • Expected R&D costs. From today's perspective, this activity is estimated With regard to potential contributions, the NES at about 5 MCHF over 5 years, out of which about Scientific Committee (NES FoKo) recommended that 1.5 MCHF is needed for new investments and PSI should not focus on a specific GIF system, but irradiation. As "seed money" for this activity, part of continue to devote a considerable part of its resources NES's reserve of external funds will be used. to LWR technology. Further, though gas-cooled • Liquid-metal technology The issue here is the concepts seem to be the most interesting, PSI should interaction of liquid metals with structural materials mainly address cross-cutting issues; e.g. materials. under SINQ conditions, as well as in Pb/Bi-cooled After the scope of the individual GIF R&D ADS and critical reactors. The aim is to select and programmes was defined in more detail in November characterise a series of materials, in particular in 2002, possible Swiss contributions were identified and regard to determining their properties, and their proposed. These focussed on ceramics and super- deterioration with time and as a result of alloys for the VHTR (helium outlet temperature 950°C irradiation. Further studies of heat transfer in liquid to 1150°C). Participation in R&D projects for cladding metals, measurement techniques, and validation of and other structural materials, at high temperatures in CFD codes are also foreseen. Finally, there is a an irradiation field, and with different (cooling) media, possibility for in-situ fatigue and creep tests in a is also possible. Further contributions are expected liquid-metal environment under high temperature conditions. This work is embedded within the 5th from NES's research on validation of CFD codes, and th safety analyses for the SCWR (thermal). and 6 EU-FWPs and the MEGAPIE consortium. Conclusions New focal point "NES 2005" Between 70% and 80% of nuclear energy research at Parallel to the considerations regarding GIF PSI will remain focused on its primary mission: the participation, and departing from the aforementioned safety of existing NPPs, and the safe disposal of "bottom-up" process, discussions have also taken radioactive waste. Future-oriented research should be place on the possibility of a new focal point for strengthened, and increasingly oriented towards research in NES, to be fully established around the (generic) materials research (HT-materials, LM- year 2005. By mid-September 2002, sketches of the technology). proposals had been formulated, and subsequently discussed by the NES-FoKo. The assessment criteria As a general remark, it should be noted that for this new focal point were scientific quality and contradictory and unforeseeable times require high innovation, the potential for networking within PSI, the vigilance and higher robustness of nuclear energy potential for embedding our work in external research with regard to external conditions. Swiss frameworks (in particular GIF and EU FWPs), and the nuclear energy research remains, however, in good attractiveness to young specialists. shape. It makes leading contributions in many key areas, thereby enabling Swiss participation in major The NES-FoKo recommendations are: international endeavours. It is necessary to do • The proposed restart of activities to fast spectra everything possible to maintain this position, and to (FAST) emphasise too strongly on ADS and Pb/Bi- cultivate new perspectives. cooled systems. The proposal should be revised in 7
OVERVIEW OF MAIN ACTIVITIES
K. Foskolos, W. Kroger
Highlights of the main activities during the year covered by this report are given below in the framework of the corresponding projects. They include tasks which were completed during the year, as well as certain projects still in their early or start-up phases. While NES is encouraging a networking of its activities, both within the Department and throughout PSI (e.g. through its involvement in the MEGAPIE project), most of the projects are the prime responsibilities of distinct organisational units (Laboratories) within NES. This basis forms a well-defined scientific focus for the projects, each requiring particular expertise and specialised equipment or facilities (see the organisation chart at the end of this overview).
COMPREHENSIVE ASSESSMENT OF ENERGY presented in a hierarchical manner, allowing users to SYSTEMS choose the level of detail most appropriate to their backgrounds and interests. Activities in the framework of the China Energy Technology Programme (CETP) were finalized in Some task representations also have special features, 2002. During this year, the GaBE team, supported by including extended analytical capabilities. For its research partners, developed an integrated example, within the Environmental Impact software tool for high-level and medium-level Assessment task, a tool enables, for a variety of decision-makers and other users (engineers, power plant sites, impacts to be simulated environmental experts and academics). This tool interactively, and the corresponding external damage provides ready access to CETP results and insights, costs to be estimated. For each site, the user can and includes advanced analytical options for exploring select from several generation technologies, establish specific issues. basic plant characteristics, and choose emission abatement levels for major air pollutants. Based on a The software comes in the form of a DVD, and parametric model, the estimation of damage due to includes a movie about CETP, background pollution is made within seconds, compared to the information on CETP and China, core information on hours needed for the underlying full-scale simulation. individual tasks, methods, data and results, and interactive tools for their exploration and decision- The multi-criteria-based, decision-support module support functions (see Fig. below). The information is generates and presents ranking results in an
Overview
—* I CETP tfVfovi e Background
Core Research Presentation a: >1 Technical tasks: Methodology, Data and Results Guided EMpIoration of Project and Cone lus ions
Interactive Exploration
Shandong Electricity I Options Ranking i ©
Examples of assessments Above: Simulation of impacts of • • • airborne pollutants released from w y y y hi. power plants in Shandong; reduction of mortality by abatement systems fcfc T i * : i Left Power technology ranking based on multi-criteria decision analysis and example preference profile. SJLtl oo 8
interactive, graphical form. Aggregated CETP results a mixture of light and heavy water (H20-D20) to for a range of criteria are combined with the user's simulate moderating conditions at full power in a input on criteria weights to produce a ranking of either PWR. The burnt-fuel samples were transported from single technologies or combinations of technologies. the Hot Lab to PROTEUS by means of a specially The user can establish the best possible options designed and constructed combined container/sample against different stakeholders' preferences, based on changer, which allowed the simultaneous transport of their economic, ecologic and social perspectives. a number of samples and their subsequent insertion, one at a time, into the central PWR mock-up zone. LWR-PROTEUS A typical reduction in multiplication factor due to burn- up of an idealised system consisting of an infinite Phase II of the LWR-PROTEUS project is related to 235 array of U02 fuel rods with 5wt% initial U high-burnup LWR fuel discharged from the Swiss enrichment is depicted in Fig. 2. The same Figure NPPs, and includes experimental reactor physics as shows a preliminary comparison of calculated well as detailed chemical analysis. One of the most (BOXER) and measured reactivity differences important reactor physics quantities investigated in between burnt U02 and MOX samples at different Phase II is the reactivity of burnt fuel, as an integral burn-up levels, and that of a reference fresh sample. measure of the ability of the fuel to maintain a neutron The results suggest that the reactivity loss is more chain reaction in the reactor. The reactivity decreases difficult to predict for MOX than for U02 fuel, and more monotonically with burn-up because the fissile difficult for higher than for lower burn-up levels. material is consumed, and because neutrons are increasingly captured in newly created species In addition to the reactivity measurements, chemical (actinides, fission products). The extent to which burnt analysis of complementary fuel segments has begun, fuel can continue to reside in the core depends on the and neutron radiography of the special fresh pellets core reactivity; accurate knowledge of the reactivity has been carried out. In 2003, new experimental level thus largely determines the cycle length and fuel windows are planned, including a new moderation performance. condition (borated water), higher burnup U02 and MOX PWR samples, and two BWR samples.
0.93 1 1 1 1 1 1 1- 0.6 20 30 40 50 60 70 80 90 Burnup [GWd/t]
Fig. 2: Typical reduction in multiplication factor due to burn-up, and preliminary comparisons of calculation/experiment (C/E) reactivity worths. Fig. 1: PWR mock-up used for reactivity measure- ments (samples placed at its centre). STARS The recent investigation of the behaviour of the feed- In 2002, two experiments were carried out with special water system following a turbine trip event in a BWR fresh-fuel samples, and two more with burnt-fuel very well illustrates the capability of the STARS samples from the Gosgen NPP. PROTEUS has been project in providing high-quality scientific services to reconfigured by replacing the SVEA-96+ fuel the Swiss nuclear industry. assemblies with OPTIMA assemblies from the Leibstadt NPP. The central test mock-up (an array of The following transient scenario was analysed. The 11x11 PWR fuel rods, see Fig. 1) was investigated in feed-water system undergoes a depressurisation two different states of moderation: one with full- transient following a turbine trip because the steam density light water (demineralised H20), and one with flow used to heat up the condensate reduces to zero. 9
Eventually, if the depressurisation rate is too large, With this perspective, cooperation with the University feed-water pump-head degradation may develop due of Maryland was established in order to work jointly on to cavitation, with the potential risk of a low-level the analysis of the dynamic aspects of Pressurised SCRAM. Hence, the goal of the planned plant Thermal Shock (PTS) scenarios, and the development modifications, to which the STARS project contributed of appropriate tools and frameworks. Key parameters analytical support, is to avoid cavitation of the feed- in the PTS event are the excessive rate of cool-down water pump by reducing the depressurisation of the reactor, re-pressurisation of the vessel, and the transient. duration of the adverse conditions. In order to calibrate the very detailed model of the First successful runs of a dynamic PSA (dynamic feed-water system, a previous turbine-trip event was event tree) model for a Main Steam Line Break analysed for which the degradation of the pump head scenario have recently been obtained. The model of one of the feed-water pumps was identified as the uses RELAP to calculate the response of the plant key phenomenon. Modelling requires accurate and its automatic safety systems, and incorporates a prediction of the pressure in the feed-water tank. model of the operating crew, and a Probabilistic Given that measured data to define the boundary Fracture Mechanics (PFM) model. The PFM model conditions to the model are scarce, significant effort calculates the conditional probability of vessel was exercised in explaining the measured cracking resulting from the plant-operator response in depressurisation. Onset of pump-head degradation each scenario evolution. could be well captured compared with measured data by assuming the presence of a layer of slightly super- The dynamic PSA platform can treat more closely the heated water at the bottom of the (very large) tank. inter-dependence between the operators' and the plant's responses. This translates into better Possible modifications of the plant control system, and predictions of the sequence and timing of the operator sensitivity to key parameters, have been explored actions, which directly impact the threat posed by the using the calibrated model, and recommendations PTS event. To achieve this, the dynamic event tree have been made. A new turbine trip test conducted model requires models of key operator decisions, as last fall confirmed the success of the proposed well as the likely courses of actions, and their modifications to the plant control system in that, as alternatives; this characterisation and modelling is a planned, no pump-head degradation was observed. A central focus for current research. detailed analysis of this new test is currently underway. Despite its costs in terms of analytical and modelling complexity, dynamic safety assessment may, in the It should be noted that, for the first time, the project future, provide capabilities for a more realistic provided results that have been directly used to treatment of the interaction between operator action develop and support plant modifications. The positive and plant response. outcome of the last plant test underlines the high quality of the analytical work. This has only been achieved because the theoretical limits of the applied ALPHA analytical tools are now very well understood, so that Various experimental and analytical investigations of the effects of known code deficiencies could be passive decay heat removal systems, and their adequately factored into the analysis of the components, have been performed in the past in the complicated phenomenology. The study also helped framework of the ALPHA Project. In particular, the in itself to identify certain code deficiencies; these are feasibility of several passive cooling system concepts currently being addressed. have been demonstrated at large-scale in the PANDA facility. Most investigations were related to postulated Human Reliability Analysis (HRA) design-basis accidents, and were studied with regard to the particular physical phenomena observed. The The potential scenarios with human "errors of numerous accompanying analytical investigations, commission" identified in the recent work of the which were carried out both within PSI and in the project have highlighted the importance of dynamics context of international projects, identified certain in the evolution of accident scenarios. Possible human code deficiencies in predicting the system behaviour errors, and their estimated probabilities (the products for situations where non-homogenous gas distribution of the HRA that are used within probabilistic safety plays a significant role. This is particularly true in the assessment to calculate the overall safety of the case of the presence of light, non-condensable gases: plant), can be sharply sensitive to such dynamics. e.g. hydrogen, which is released as a consequence of The development of simulation-based tools, with a (postulated) severe accident involving core which the coupled responses of the plant, the degradation. For such situations, a more detailed, automatic systems and the operators can be jointly three-dimensional treatment of various phenomena modelled, derives from these considerations. For the (e.g. jet and plume behaviour) is required to scenarios most problematic in terms of dynamics, a accurately predict the resultant system behaviour. dynamic safety assessment, based on an extended, As part of the TEMPEST project (5th EU FWP), the discrete dynamic event-tree framework, represents a performance of the ESBWR Passive Containment more ambitious and long-term undertaking. Cooling System (PCCS) in the presence of non- 10 condensable gases (in particular helium as a simulant continuously sample gas from various locations, and a for hydrogen) has been experimentally investigated in mass spectrometer is used to provide time histories of PANDA. In addition, the impact of a new accident the gas molar fractions (air, He and steam) at each mitigating design feature — the Drywell Gas Re- measurement location. circulation System (DGRS) — on the long-term Figure 3 shows the containment response in three containment behaviour has been examined. different tests with the same PANDA configuration. The function of the DGRS is to mitigate system Test T2.2 is a simulation of the base-case, design- pressure build-up by the operation of a vent fan, which basis-accident transient (Main Steam Line Break), feeds non-condensable gases from the PCCS back to including the effect of the DGRS (started at about the Drywell instead of discharge to allowing them to 22 000 seconds). In addition, Tests T1.1 and T2.1 the Wetwell gas space. The ultimate goal of the four simulate core overheat, starting at 10 000 seconds, by tests with the injection of a large amount of He was to the injection of a large amount of He (to simulate H2 investigate in detail the relevant phenomena, and to release through zirconium/water interaction) over a provide a database to assess the capability of period of two hours. The graph shows the initial, rapid advanced containment and CFD codes to predict the pressure increase due to the He injection, and its containment behaviour under such conditions. subsequent stabilisation, about 3000 seconds after Accordingly, the already extensive PANDA the end of the injection phase for both tests. There is instrumentation was improved in regard to the spatial only a moderate influence of the DGRS on the system resolution of the temperature measurements, mainly behaviour, in that some of the He remained in the in the Drywell. Part of the instrumentation upgrade Drywell during Test T2.1 resulting in a system was the installation of a gas-concentration pressure reduction of about 0.3 bar. measurement system, in which capillary tubes
Time (s)
Fig. 3: Containment system pressure for Tests T2.2, T1.1 and T2. 1.
With the improved instrumentation, a clear picture of particles of 5 |jm diameter). Consequently, it is, the gas distribution, mixing and stratification anticipated that airborne fission-product particles phenomena in the Drywell vessels (including the would circulate around with the gas in the containment propagation of the stratification front and related volumes rather than settle under the influence of phenomena) could be obtained. The experimental gravity, as is usually assumed in lumped-parameter results are currently being used for code-assessment containment codes such as CONTAIN. activities. Consequently, there is, a need to assess the controlling phenomena for aerosol deposition using a Severe Accident Research more rigorous treatment of the forces acting on the Typical of actual reactor containment conditions, the aerosol particles than used hitherto. In particular, gas circulation velocity in the PHEBUS containment (a analyses are required to determine the relative 1:5000 scale test facility at CEA Cadarache, France) magnitudes of inertial impaction and gravity settling. is of the order of several tens of centimetres per An open question is whether there is a fluid-dynamics- second, and hence much in excess of the settling controlled mechanism for aerosol deposition on velocities of the largest expected aerosols (1 mm/s for vertical dry walls. To simulate particle motion in the 11
PHEBUS containment, it was necessary to know the Advanced Fuel Cycles (AFC) gas velocity field, and the CFD code CFX-4 was used One of the main activities within the AFC project is the to determine it. The model was a 1/6th sector of the fabrication of various geometrical forms of MOX fuel containment vessel, and consisted of 46 000 with 20% Pu, which is required for the FUJI project. computational cells (Fig. 4). Particle tracking was then This collaboration between PSI, the Japan Nuclear performed in a post-processing mode, based on the Cycle Development Institute, JNC, and the Dutch steady-state flow field. Nuclear Research and Consultancy Group, NRG, is A total of 117 particles, representing a reasonable directed towards the testing and comparing of statistical sample, were assigned random starting different types of fuel, with a view towards later positions in the flow domain, and then tracked. A side utilisation in the Japanese fast reactor programme. view of a typical particle track is shown in Fig. 4. The PSI provides its experience in fuel fabrication by the particle follows the flow streamlines until it reaches a Sol-Gel-Method, and NRG its irradiation facilities at location (in this case, the sump) where its inertia is too the High-Flux-Reactor in Petten. The experimental large to allow it to move further, and it impacts on the laboratory work started at the beginning of 2002, after vessel wall. the refurbishment of the PSI Hot Lab. As a first priority, fuel fabrication tests were performed with The times required for such particles to deposit vary different fuel shapes, namely pellets, microspheres, considerably, depending on their initial positions. The and vipac particles. calculations reveal that impaction times range from a few seconds to 2.7 hours, with a mean value of about The pellet fabrication tests were successfully 40 minutes, which is of the same order as the completed, including the adjustment of the sintering experimentally determined value of 60 minutes. The conditions, to achieve the required low oxygen-to- computation also indicates that 75% of the particles metal ratio of 1.97, typical for fast reactor fuel. In deposit on the lower surface of the containment (i.e. addition, pellet fuel for the irradiation tests could be the sump and the floor), which is also consistent with fabricated up to the end of 2002. The fabrication tests experimental data. for the vipac particles were also successfully completed. This was the first time PSI had produced Contrary to lumped-parameter containment codes, the such fuel using a fabrication process which involved use of CFD techniques enables both the physical and pellet pressing, crushing the green pellets, and quantitative features of the deposition process to be sintering the granules. Figure 5 confirms the captured. These encouraging results have provided satisfactory shape of the granules obtained by this incentives for performing further detailed studies. method. Series fabrication of the vipac fuel for Planned future efforts include simulation of the full irradiation will take place in January 2003. PHEBUS containment geometry (instead of the 1/6th sector), and the investigation of particle behaviour in a time-dependent flow field.
Starting §«t
CXftdeLjer
7PM-V601-S 0 106-0 250 mm SB-0X51S 14 07 08 2002 HP43 Fig. 5: MOX vipac particles of different size fractions produced at PSI.
EDEN The properties of irradiated austenitic weldings are important for lifetime assessments of reactor internals Dqpo t t in Light Water Reactors (LWRs). Typical weldings between the standard austenitic steels 304 and 347 are being studied as part of the project INTERWELD Fig. 4: The computational mesh used for the 1/6th (5th EU FWP). The mechanical properties, the model of the PHEBUS containment, and a microstructure, and the residual stress distributions in side view of a typical particle track. the welds (base metal, heat-affected zone) are being 12 investigated at PSI in regard to stress/strain response, the transformation. It was found that, in the TEM characterization, and neutron diffraction. temperature range 30-300°C, the temperature dependence of the deformation-induced martensite Figure 6 shows a comparison of the microstructures could be described by an exponential decay function. both near the fusion line, where heavily deformed Since the initial material state plays an important role areas and re-crystallised grains can be seen, and for the martensite formation rate, the dependence of away from it. The structural differences lead one to the martensite content on the cycle number was expect an increase of yield strength in this area determined for two different material states. compared with the unaffected base metal, a X-ray diffraction experiments using synchrotron light circumstance which could be experimentally verified. sources performed at ESRF Grenoble have enabled First stress measurements with neutron diffraction the martensite distribution at the surface of the fatigue reveal the presence of high residual stresses, which specimens to be analysed. Figure 7 shows the might be a direct consequence of the very martensite distribution in the longitudinal cutting plane pronounced differences in the actual microstructure. The deformation behaviour has also been 4.667 - 5.000 4.333 - 4.667 investigated. Twinning was identified as an important 4.000 - 4.333 3.667 - 4.000 deformation mechanism at room temperature. At 3.333 - 3.667 3.000 - 3.333 elevated temperatures (i.e. 300°C), on the other hand, 2.667 - 3.000 mainly dislocation glide was found. No strain-rate 2.333 - 2.667 2.000 - 2.333 sensitivity could be detected. During the next stage of 1.667 - 2.000 1.333 - 1.667 the study, irradiated material will be investigated. 1.000 - 1.333 0.6667 - 1.000 0.3333 - 0.6667 0 - 0.3333
-18 -16 -14 -12 -10 -8 -6 -4 axial position in mm
Fig. 7: Martensite distribution in the longitudinal cutting plane of a fatigue specimen, as measured by X-ray diffraction.
of a specimen fatigued to crack initiation. It was observed that martensite was formed mainly in the Fig. 6: Microstructure close to (left) and away from centres of the specimens, and in the crack areas. (right) the fusion line. The volume fraction of martensite was determined by neutron diffraction and advanced magnetic methods. Structural Integrity Neutron diffraction appeared to be the appropriate tool to serve as a calibration method for the martensite Degradation of nuclear materials can decisively affect content in the bulk of the fatigue specimens. In order the performance and operational lifetime of NPPs. to establish the correlation between martensite Monitoring of material degradation is a new trend, and content and magnetic properties, the magnetic aims at setting up a methodology for a physics-based susceptibility was measured. For given loading and lifetime assessment of nuclear plant components. The material conditions, a linear relationship was found technology includes the discovery of indications for between the magnetic susceptibility and the number ageing in the microstructure of materials, and of cycles. In addition, a highly sensitive Giant Magneto measurement of the influence on the mechanical and Resistant (GMR) sensor was used to visualise the physical properties. The technical goal comprises the martensite. Figure 8 gives an example of the development of a lifetime monitoring system, martensite distribution at the surface of a fatigue applicable to a certain class of materials. Austenitic specimen showing a technical crack. A non- stainless steels subjected to cyclic loading exhibit homogeneous distribution of the martensite at the microstructural changes before macroscopic crack surface may be observed, whereas in the bulk the initiation begins. For certain loading and temperature martensite is concentrated in longitudinal streaks. conditions, a deformation-induced martensitic phase transformation during fatigue has been observed. Waste Management In the investigations carried out at PSI, it was found that, for the given loading and material conditions, the The waste management R&D at PSI plays an volume fraction of martensite depends linearly on the important national role by supporting the Swiss cycle number (lifetime) of the applied load. Federal Government and Nagra in their tasks to safely Consequently, the deformation-induced martensite dispose of nuclear wastes from medical, industrial and was used to determine the usage factor of fatigue research applications, as well as those from the Swiss degradation. Furthermore, it was shown that the test NPPs. The activities are in fundamental repository temperature had a strong influence on the course of chemistry, chemistry and physics of radionuclides at 13 solid/liquid interfaces, and radionuclide transport and The evaluation of the solubilities and sorption retardation in geological media and man-made distribution ratios of relevant radio elements repository barriers. The work is a balanced constitutes a milestone in efforts over many years, combination of experimental activities in dedicated and is a good example on how progress in radioactive laboratories, at large facilities, such as understanding of basic mechanisms has an impact on synchrotrons, in the field, and in theoretical modelling. performance assessment. The in-built conservatism of Current efforts are directed towards repository previous assessments has been reduced projects, and the results find ready application in considerably, and the overall picture has moved comprehensive performance assessments carried towards greater realism. through by Nagra. Figure 9 presents the distribution ratios for sorption on
Fatigue specimen P16, D=1 (Martensite 8.58 SINQ) bentonite backfill (MX-80), which consists predominantly of montmorillonite. For most elements, the distribution ratios are now predicted to be considerably higher, and hence the mobility Martensite [%] (GMR) correspondingly lower, than had been assumed in
2,000 previous performance assessments — in this 3,250 comparison, for the Kristallin-I Study of 1995. In 4,500 addition, it has been possible to define an expectation 5,750 7,000 value together with uncertainty bars, whereas 8,250 previously the range of the distribution ratio was 9,500 10,75 given, somewhat subjectively, by so-called realistic 12,00 and pessimistic values. 5 10 15 20 25 The reduction of predicted mobility results from the longitudinal direction [mm] combined effects of better understanding of sorption Fig. 8: Martensite distribution at the surface of a mechanisms, new results from both in-house fatigue specimen measured by a magnetic measurements and the literature, and a novel method GMR sensor. to derive sorption in compacted clay material from measurements in diluted systems. A few of the In 2002, emphasis was placed by Nagra on sorption distribution ratios could be validated by finalisation of the contributions to the Demonstration independent, dynamic diffusion experiments. It is the of Disposal Feasibility Study in the Opalinus clay of task of future work to strengthen this validation basis, the Zurcher Weinland. This study by Nagra deals with and to reduce uncertainties by developing better the long-term safety of a repository for spent fuel, understanding of compacted clay systems. vitrified high-level and long-lived intermediate-level waste, and recommendations were submitted to the Swiss regulatory authorities in December 2002.
MX-80 BENTONITE SORPTION VALUES
Np
100 1000 10* 10° DISTRIBUTION RATIO, R (L kg"1)
"Sorption data base values 2002 * Sorption data base values 1995 Fig. 9: Sorption values in Bentonite MX-80. Use of Large PSI Facilities synchrotron-based, micro-beam X-ray absorption spectroscopy (XAS). The optical concept is optimised In 2002, the work in the microXAS beamline project towards micro-focusing (-1x1 |jm2), with the possibility focused on the finalisation of an optical layout for sub-micron spatial resolution. Within this project, a designed as an analytical facility dedicated to new technique to produce hard X-ray pulses of 14 approximately 100 femtoseconds duration will also be where the X-ray beam passes a dynamic 'roller-blade' implemented. As a result, time-resolved studies with a slit system. resolution of -100 femtoseconds will become feasible. After finalization of the optical layout, the work has The microXAS beamline will be located in a long, focused on the elaboration of CAD construction plans, straight section (X05L) equipped with a radiator and the evaluation of beamline components critical with modulator. A mini-gap, in-vacuum undulator (U19) respect to their technical requirements, and the costs serves as the radiation source, and will provide high and time schedule for the realisation of the project. In brightness X-rays in the energy range 5 to -20 keV. this context, the call for tenders (WTO) for the The front end includes SLS standard radiation safety monochromator, the mirror, the insertion device, the equipment, as well as an aperture and slit system, to front-end and the Pb-shielded hutches were all reduce bremsstrahlung and total radiation power. worked out. The offers were then evaluated, and Since X-ray satellites (created by the laser-electron contracts with manufacturers were subsequently beam interaction in the modulator) in the FEMTO placed. The first front-end components are currently mode have an angular offset of 0.5 mrad with respect being delivered, and will be installed at the beginning to the undisturbed beam, the proposed set-up will of 2003. It is planned that the monochromator be allow the main X-ray pulse to be blocked. installed in early fall 2003. Due to the long delivery After a travel distance of -21 m from the mid-point of time of the mirror and the insertion device (up to the straight section, the X-rays pass a toroidal mirror 18 months), installation of these components will take in the shielded optics hutch. This water-cooled device place at the end of 2003 or the beginning of 2004. is operated in grazing-incidence mode, dynamically bent, and produces a horizontal 1.4:1 focusing and collimation of the beam in the vertical direction. The National and International Collaborations current design foresees an additional water-cooled All NES activities are intensively and specifically mirror at -22.5 m, which will be evaluated for future embedded in national and international co-operations. FEMTO applications. At least one of the two Swiss Federal Institutes of Technology (ETHZ, EPFL) is always a partner (often Since windowless operation is required for the contractually), and all NES projects are contributing to FEMTO project, a vacuum scheme, including one or more EU projects. All important European differential pumping, is being considered. The beam research organisations (CEA, FZJ, FZK, FZR, NRG, will be energy-filtered by a Double Crystal SCK, VTT, and the JRCs), as well as the most Monochromator at -27 m. The fixed-exit important ones in the USA and Asia (EPRI, CRIEPI, monochromator will be equipped with two sets of JAERI, KAERI), are direct partners in several projects. monochromator crystals, namely Si(111) and Si(220), NES is a highly regarded and substantial partner of and can also be operated for "pink beam" mode. A international activities within the IAEA and the water-cooled beam and bremsstrahlungs-s\op is OECD/NEA. On the industrial side, and along with the necessary to protect the downstream equipment. domestic utilities (UAK), all important reactor and fuel Before entering a Kirkpatrick-Baez (KB) mirror manufacturers (GE, Framatome-ANP, BNFL- system, used for the final focusing of the X-ray beam, Westinghouse, Areva-Cogema, BN, JNC) collaborate there is an intermediate focal point (virtual source) with NES as paying partners.
Nuclear Energy and Safety Research Department W. Kroger
GaBE Si. Hirschberg
STARS Waste Management M Zimniermann Programme J. Hadermann LWR-PROTEl'S ALPHA F. JatulT J. Dreier MicroXAS A. Scheidegger Severe Accidents S. Giintay
MEGAPIE
PROTEUS HOX PANDA, LINX,DRAGON+ HOT LABORATORY LABORATORY (SLS) 15
RISK BENEFITS OF CLIMATE-FRIENDLY ENERGY SUPPLY OPTIONS
S. Hirschberg, P. Burgherr
One of the central goals of sustainable development Is the reduction of Greenhouse Gas (GHG) emissions. This Is needed In order to prevent the anticipated climate change, and the potentially serious consequences for human beings and the environment Energy supply systems constitute the dominant contributors to GHG emissions. This paper examines three Illustrative emission scenarios for world-wide energy supply In the 21st Century. These scenarios, Including the associated GHG and major pollutant emissions, were chosen from a set established by the Intergovernmental Panel on Climate Change (IPCC). Using the emissions as a starting point, and based on recent findings concerning the impact on the environment and the financial costs resulting from global climate change on the one hand, and regional air pollution on the other hand, the present work provides estimates of the scenario-dependent, world-wide cumulative damage. The fossil-intensive reference scenario leads to overall damages which correspond to very substantial losses in Gross Domestic Product (GDP), and which widely exceed the damages caused by the scenarios reflecting climate-friendly policies. Generally, the somewhat speculative estimates of the GHG-specific damages are much less significant than damages to human health and the environment caused by the major air pollutants. This means that the secondary benefits of climate- friendly, energy-supply options, i.e. those which avoid the impacts due to air pollution, alone justify strategies protecting the climate.
1 INTRODUCTION reduction of the expected accident risks, another controversial issue in the energy policy debate. The Intergovernmental Panel on Climate Change (IPCC) concludes in its latest assessment report [1] The present work addresses the costs of impacts that the Earth's climate system has changed, globally caused by climate change and air pollution. More and regionally, with some of these changes being specifically, the benefits of averting the 'climate attributable to human activities. Carbon dioxide, catastrophe', and simultaneously avoiding serious surface temperatures, precipitation and sea level are health and environmental effects, are estimated in all projected to increase globally during the 21st monetary terms. The analysis covers damages world- Century because of human activities. Models also wide, and its time horizon is set at the end of the 21st project an increase in extreme weather events. Century. Fossil energy systems are the dominant contributors to the emissions of major Greenhouse Gases (GHGs). 2 ENERGY SCENARIOS At the same time, most currently used fossil systems emit large quantities of air pollutants, such as sulphur The most comprehensive state-of-the-art assessment dioxide, nitrogen dioxide and particulate matter. The of GHG emission scenarios, reported in [2], was used impacts associated with these emissions include as the starting point in the present work. The health effects, manifested by millions of years of scenarios provide the basis for future assessments of human life lost, and by degradation of the climate change and possible response strategies. The environment. Particularly in developing countries, the report shows that the choices of preferred energy damage caused by air pollution is very extensive, and technologies, along with energy conservation policies, backfires on economic growth. As a result, the three have a decisive influence on the magnitude of future pillars of sustainability — environmental, economic emissions. The characterization of scenarios supplied and social — are each strongly affected. in [2] stops at the level of emissions; thus, no estimates of impacts or financial damage have been provided, though such estimates are necessary for The extended use of climate-friendly technologies is cost-benefit considerations. associated at least initially with increased costs. One additional benefit of such policies may be that not only Four scenario families, including in total 40 scenarios, are the climate-related risks being reduced, but there have been developed. For the purpose of the present is also a simultaneous potential for mitigating the work, two scenario families (A and B1), and three health and environmental impacts of air pollution. This associated "illustrative" scenarios, have been may then constitute the main secondary benefit of the selected. deployment of climate-friendly technological options. In fact, the reduction of air pollution, whose effects are "Family A scenarios" are characterized by rapid here and now, is considered by many developing economic growth with substantial reduction of regional countries to be the primary benefit. Of main interest differences in per capita income, global population are strategies which lead to much improved air quality peaking in mid-century, but declining thereafter, and at a reasonably low additional cost, and which also by rapid introduction of new and more efficient result in reduced GHG emissions. Such scenarios technologies. The two considered "illustrative" may also exhibit one additional secondary benefit: i.e. scenarios are: the fossil-intensive A1F1, and non- fossil ("sustainable") A1T. 16
The "Family B1 scenario" is characterized by Figures 1 and 2 show the characteristics of the significantly lower growth than for "Family A", selected scenarios in terms of primary energy shares, practically the same population pattern as A, rapid C02 and other major pollutant emissions, respectively. changes in economic structures towards a service and Some additional, implicit characteristics of the information economy with reductions in material scenarios are provided in Table 1. intensity, and by forceful introduction of clean and Table 1: Additional characteristics of selected IPCC resource-efficient technologies. scenarios (following [2]). Among all "illustrative" scenarios, A1F1 has the highest cumulative C02 emissions in the period 1990- Feature/Scenario A1F1 A1T B1 2100, while B1 has the lowest, closely followed by A1T. It needs to be emphasized that the fossil- Share of coal in 2020: 29 2020: 23 2020: 22 intensive scenario A1F1 by no means represents a primary energy (%) 2050: 33 2050: 10 2050: 21 "worst case", since substantial credit is taken for 2100: 29 2100: 1 2100: 8 expansion of clean and efficient fossil technologies. Share of zero carbon 2020: 15 2020: 21 2020: 21 in primary energy 2050: 19 2050: 43 2050: 30 (%) 2100:31 2100: 85 2100: 52 C0 emissions 2 2 189 1 068 983 (GtC), 1990-2100
3 DAMAGE COSTS 3.1 Global warming
The C02 damage costs are highly uncertain, given the A1F1 A1T B1 A1F1 A1T B1 A1F1 A1T B1 A1F1 A1T B1 complexity and limited state of knowledge of the 1990 2020 2050 2100 underlying phenomena. The estimates are sensitive to assumptions that are not only of methodological Fig. 1: Shares of primary energy for the selected character but also reflect ethical positions. Use of IPCC scenarios (following [2]). mitigation costs could be an alternative approach, but is based on two arbitrary assumptions: (a) the marginal costs of abating emissions equal the marginal damages; and (b) environmental regulation is economically efficient. The recent damage estimates reported in [3] are based on the FUND 2.0 model. The considered impacts include: health, agriculture, water supply, rise in sea level, ecosystems and biodiversity, and extreme weather events. The numerical results tend to be lower than earlier estimates, since it has been taken into account that, for rich countries with temperate climates, the climate change may bring some benefits. Toll and Downing's estimates [3] were obtained for the time horizon 2100, though the emissions take place in the period 2000-2009. The costs of emissions taking place later could be higher. All cost estimates are discounted to the year 2000, using a 1% discount rate. The world average is as follows (the reference gives the estimates in EUR0/tC02, but, taking the year 2000 as the reference year, we obtain roughly the same numbers expressed in US$):
Minimum: 0.1 US$/tC02
Central estimate: 2.4 US$/tC02
Maximum: 16.4 US$/tC02 The later review paper by Toll et al. [4] confirms that the marginal costs of C02 emissions are unlikely to exceed 50 US$ per ton of carbon; a result fully • A1F1 E1A1T DB11 consistent with the maximum value given above.
Fig. 2: C02, S02 and NOx emission characteristics for selected IPCC scenarios (following [2]). 17
3.2 Air pollution USA and Oceania have the same average damage costs as EU-15, the Economies in Transition the same The damage costs due to air pollutants are very as Asia, and the Middle East and African countries the dependent on where the emissions actually take same as South America. place. In Western Europe, the resulting damages for specific pollutants may differ by a factor of ten, Weihai • PM10 depending in which country the pollutants are emitted. UNOx • 5378 • SOx Here, three main pollutants are considered: S02, NOx Jinan 13975" and PM10 (primary aerosols). For S02 and NOx, the • 6364 principal damages caused by the formation of Heze secondary fine particles (sulphate and nitrate Power Sector • 5032 4579 aerosols) are included. Shandong 4592 The methodology used for the estimation of external Shandong 4881 costs is the "impact pathway" approach, established Power Sector • within the ExternE Project of the European Union [5]. China The main features of this approach are summarized China below.
2000 4000 6000 8000 10000 12000 - The analysis steps are: Technology and Site US$ per Ton Characterization; Prioritisation of Impacts; Quantification of Burdens (e.g. emissions); Fig. 3: Monetary damages per ton of pollutant Description of the Receiving Environment; emitted at various locations in China [7]. Quantification of Impacts (using, whenever applicable, dispersion models for atmospheric Table 2: Regional pollutant-specific damage estimates. pollutants and dose-response functions); and Economic Valuation. The pathways of pollutants are followed from the point of release to the point Region/Pollutant S02 NOx PM10
where damage take place. Quantification of health (US$/tS02) (US$/tSOx) (US$/t PM10) and environmental impacts is achieved through the EU-15 [6] 3 600 3 500 7 000 damage function. Switzerland3 [8] 7 500 9 400 11 000 - All relevant stages in the various energy chains are Asia [7] 3 000 1 800 2 400 covered (Extraction, Fuel Processing, Transport, Power Generation, Waste Management and South America [6] 150 200 450 Storage). a These estimates are based on the use of a single-source - The impacts covered in a full-scope analysis model, while the other data sets were generated by running include: Occupational and Public Health Effects a multi-source model. The results from the single-source case are representative for the Swiss power plants and may (Mortality, Morbidity); Impacts on Agriculture and differ from the average country-specific damage costs. The Forests; Biodiversity; Impacts on Water (Ground Swiss values were not used in the current analysis, but are and Surface); Impacts on Materials (Buildings and included in the Table for comparison purposes. Cultural Objects); and Global Impacts (Greenhouse Effect). Both normal operation and 3.3 Total damages severe accidents are considered. Combining scenario characteristics (emissions) with Based on our research within the China Energy damage cost per emitted ton leads to the results Technology Program (CETP), the ExternE results for summarized in Table 3, and illustrated in Figs. 4 and Western Europe, and on similar analyses for South 5. For air pollution, this was done separately for each America, the average damage costs in US$ per ton of of the six world regions. It should be noted that the pollutant were established. All these studies use the damage effect due to Greenhouse Gases other than respective regional implementations of the multi- C02 has been ignored, along with the cooling effect of source version of the EcoSense software [6, 7]. aerosols. In addition, the estimated damages have Figure 3 shows the variation of damage costs by air been corrected for GDP growth. The monetization of pollutants for different plant locations in the Chinese impacts due to air pollution is based on contingent Province of Shandong, for the power sectors in evaluation, employing the Willingness to Pay (WTP) Shandong and the whole of China, and for all method. Since health effects dominate the damages, emission sources, both in Shandong and for China as and the value of life lost is dependent on GDP per a whole. The results available for the regions capita, the corrected numbers were obtained by analysed are summarized in Table 2. It should be scaling up the results based on the current monetary noted that the variation between the various sites, values using the scenario-specific ratios between the sectors and geographical regions is very large. Thus, regional average GDP per capita in the year in the estimates provided for whole continents should be question and the year 2000. Given the large increase regarded as aggregates of a rather wide spectrum of of GDP per capita over the last hundred years, this contributing components. In order to estimate the approach may be debatable. damages for the whole world, it was assumed that 18
Combining scenario-specific air pollutant emissions with damage costs per emitted ton is based on the 100000 assumption of linear exposure response functions | H QECD 90 M REF • Asia • ALM | 28748 without thresholds. The impacts are, in most cases, 8667 dominated by sulphates formed from S02 and NH3 emissions through chemical reactions in the atmosphere. The formation of nitrates from NOx depends in a more complicated way on the conditions in the atmosphere. Thus, the extrapolation of damages into the future is more uncertain for NOx. The contribution from primary PM10 is neglected, since the corresponding scenario emission data are lacking. A1F1 A1T B1 A1F1 A1T B1 A1F1 A1T B1 A1F1 A1T B1 The cumulative estimates of the global warming 1990 2020 2050 2100 damages are based here on the maximum values, since the coverage in terms of damages and time is Fig. 4: Estimated world-wide damages associated limited. The large uncertainty interval reflects the with selected IPCC scenarios; the shares for relatively low confidence in the estimates of the the OECD90, REF, Asia and ALM regions are damages due to global warming. However, the total also indicated. estimates are dominated by the damages due to air 1400 pollution, whose assessment is much more robust. IHCO2 MA\r Pollution | The "sustainable" scenarios A1T and B1 exhibit much 1200 lower cumulative damage costs than the fossil- 1000 intensive scenario A1F1. 800 Table 3: Estimated world-wide damages associated with selected scenarios (all numbers are 600 rounded). 400
Cost A1F1 A1T B1 200 Type/Scenario 0 1990: 3-400 1990: 3-400 A1F1 A1T B1 1990: 3-400 C02 Damage 2020: 4-600 2020: 4-600 2020: 5-70 in billions of 2050:5 -700 2050: 4-700 2050: 8-1400 Fig. 5: Total cumulative damage for the period 1990- (2000)US$ 2100: 2-300 2100: 2-300 2100:11-1800 2100, based on interpolation. Air Pollution 2020: 1800 2020: 1700 2020: 1100 Damage 2050: 8700 2050: 620 2050: 3000 4 CONCLUSIONS in billions of 2100: 28700 2100: 11500 2100: 3400 (2000)US$ The total damage estimate for fossil-intensive power Cumulative generation is dominated by the impact of air pollution. 0.8-130 0.4-65 0.4-60 Carbon Dioxide In the case of scenario A1F1, this impact would Damage in the constitute about 5.5% of the world GDP in the year period 1990- 2100, while the corresponding losses for more 2100 sustainable scenarios A1T and B1 would be 2.2% and in trillions of 1.0%, respectively. Thus, it can be claimed that, on (2000)US$ the basis of the current level of knowledge, the Cumulative Air secondary benefits from mitigating the air pollution 1140 600 250 Pollution risks alone make a very strong case for the promotion Damage3, 1990 of climate-friendly systems. Based on a comparison of -2100 the damage costs assessed in this paper with the in trillions of increased system costs associated with climate- (2000)US$ friendly scenarios analysed in [9], and with the year Total 2050 used as the time horizon, there are strong 1140-1270 600-670 250-310 Cumulative indications that the avoided damage costs associated Damage3 in the with the environmentally friendly scenarios widely period 1990- exceed the increased investment costs. Furthermore, 2100 using the findings of the comparative assessment of in trillions of severe accidents in [10], it can be shown that the (2000)US$ more sustainable scenarios exhibit lower expectation 1 Based on interpolation values for severe accident risks. However, the expected damages due to such severe accidents, in view of their frequencies, are much lower than those caused by air pollution. 19
REFERENCES [6] W. Krewitt, A. Trukenmuller, T. Bachmann, T. Heck, "Country-Specific Damage Factors for Air [1] IPCC, Reports of Working Groups I, II and III of Pollutants; a Step Towards Site-Dependent Life- the Intergovernmental Panel on Climate Change, Cycle Impact Assessment", Int. J. LCA 6:4, 199- Summaries for Policy Makers, Intergovernmental 210, 2000. Panel on Climate Change, Cambridge University Press, Cambridge, UK, 2001. [7] S. Hirschberg, T. Heck, U. Gantner, Y. Lu, J. V. Spadaro, W. Krewitt, A. Trukenmuller, Y. Zhao, [2] IPCC, Special Report on Emission Scenarios, A " Environmental Impact and External Cost Special Report of Working Group III of the Assessment in the China Energy Technology Intergovernmental Panel on Climate Change, Program", PSI Report (to be published), 2003. Cambridge University Press, Cambridge, UK, 2000. [8] S. Hirschberg, R. Dones, U. Gantner, "Use of External Cost Assessment and Multi-Criteria [3] R. S. J. Toll, T. E. Downing, "The Marginal Costs Decision Analysis for Comparative Evaluation of of Climate Changing Emissions", Institute for Options for Electricity Supply", NES Scientific Environmental Studies, Amsterdam, Netherlands, Report IV, 2000. 2000. [9] L. Barreto, "Technological Learning in Energy [4] R. S. J. Toll, T. E. Downing, S. Fankhauser, R. G. Optimisation Models and Deployment of Richels, J. B. Smith, "Progress in Estimating the Emerging Technologies", PhD Thesis No. 14151, Marginal Costs of Greenhouse Gas Emissions", ETH Zurich, Switzerland, 2001. Working Paper SCG-4, Centre for Marine and Climate Research, Hamburg University, [10] S. Hirschberg, G. Spiekerman, R. Dones, Germany, 2001. "Severe Accidents in the Energy Sector", PSI Report No. 98-16, 1998. [5] EC, ExternE - Externalities of Energy, ExternE Final Report, European Commission, Luxembourg, 1999. 21
DEVELOPMENT OF DYNAMIC PROBABILISTIC SAFETY ASSESSMENT: THE ACCIDENT DYNAMIC SIMULATOR (ADS) TOOL
Y.H. Chang1, A. Mosleh1, V.N. Dang 1 University of Maryland
The development of a dynamic methodology for Probabilistic Safety Assessment (PSA) addresses the complex interactions between the behaviour of technical systems and personnel response in the evolution of accident scenarios. This paper introduces the discrete dynamic event tree, a framework for dynamic PSA, and its implementation in the Accident Dynamic Simulator (ADS) tool. Dynamic event tree tools generate and quantify accident scenarios through coupled simulation models of the plant physical processes, its automatic systems, the equipment reliability, and the human response. The current research on the framework, the ADS tool, and on Human Reliability Analysis issues within dynamic PSA, is discussed.
1 INTRODUCTION Reliability Analysis (HRA) has focused on the development of methods for systematically identifying In a traditional probabilistic safety assessment (PSA), potential EOC situations. Within the HRA project at fault trees and event trees are used to create a quasi- PSI, the Commission Errors Search and Assessment static model of accident scenarios. This model (CESA) method was developed, and a pilot essentially consists of a series of snapshots of the application has been performed [1]. The dynamics of scenario evolution. At these snapshots, the hardware the plant processes and their interactions with human systems and operators may perform successfully or response, were shown to be a contributing factor in a fail, resulting in a tree of scenarios (the event tree) in number of the potential EOC scenarios identified in which the snapshots correspond to the branching the pilot. points of the tree. Comprehensive modelling of EOC scenarios within an The development of simulation-based or dynamic event-tree/fault-tree PSA model leads to two practical PSA methodology is motivated by the need to address problems. First, some of the success/fail branches the limitations of this quasi-static, binary model. These need to be converted to non-binary branches to include difficulties represent alternative courses of action. An equivalent - with the handling of some forms of dynamics related binary representation could also be used, although it to the physical and automatic processes, and would be cumbersome. Second, information about the - with the addition of the diverse scenarios introduced causes of an EOC must be retained, because these by errors of commission (EOCs). can affect the probabilities of the subsequent human In traditional PSA, many success criteria are defined actions. This consideration reflects the trend in the in terms of how long a system must perform emerging HRA methods to consider decision successfully. An example is that a particular system performance in more detail, and be based on a must perform at 50% of its capacity for at least 30 broader set of contextual factors. In summary, an minutes. In all cases in which the criterion is not extension of PSA to include EOCs leads to many fulfilled, the function is considered failed. For the additional scenarios, as well as to the need to "carry" purpose of defining the sequence, it is often more information concerning the history of each of the conservatively assumed that the system has not scenarios individually [2]. performed at all. This assumption, while in general A range of modelling frameworks are available to conservative, affects both the sequence model and address these difficulties. These include the Markov the human performance model. It affects the chain (continuous event tree) [3], GO-FLOW [4], sequence model in that partial functioning of the Event Sequence Diagram (ESD) [5], dynamic FT [6], system may reduce the demands on other systems. It and the discrete dynamic event tree (DDET) [e.g. affects the human performance model, in that this 7,8,9]. The ADS tool is based on the DDET, because partial functioning may extend the time available for it combines all of the following advantages: response, or may make the diagnosis cues less evident. By neglecting these effects, the assumption • The physical processes, hardware states, and makes the PSA manageable and practical. human performance are directly modelled, avoiding the need to manually define a state vector or a state In general, PSA defines the failure of human transition diagram (both of which would be very interventions in the scenario as the failure to carry out large) required actions. In contrast to the omission of required actions, EOCs refer to the performance of • The coupling of these models allows all their inappropriate actions that aggravate a scenario. interactions to be explicitly and transparently treated These actions have not been comprehensively • The scenario space is systematically examined included in the PSA due to the lack of an appropriate through the DDET algorithm, avoiding Monte Carlo methodology. More recently, research in Human sampling and its associated difficulties. 22
The partial tree at the top of Fig. 1 shows the DDET after the first history (Sequence 1); the system state has been saved at three branching points. Next, the system state is reloaded at the last branching point (the point labelled 'a') of the first history, and the simulation is restarted, resulting in Sequence 2. The tree-building algorithm continues by returning to the nearest open node, now 'b', and restarting the simulation. The lower branch at node b is simulated and the branching points 'd' and then 'c' occur in Sequence 3. As was the case with the first partial tree, node 'c' is first expanded to result in Sequence 4, then node 'd'. After Sequence 5, the simulation would continue with node 'e'. (The tree building algorithm for Fig. 1: Evolution of the dynamic event tree. the DDET may follow alternative schemes.) The DDET framework is introduced in Section 2 while In contrast to a 'static' event tree, the order of the Section 3 presents an overview of the ADS tool for branching points (which correspond to event headers) dynamic PSA. The treatment of the human response is not explicitly specified a priori. Instead, the model within a dynamic PSA is addressed in Section 4. includes rules for determining what stochastic events Finally, Section 5 presents the on-going efforts within may occur next, based on the current system state the project, and the outstanding issues for further represented at a node. In addition, the timing of research. events, represented schematically in Fig. 1 by the spacing between the nodes, is variable, and depends 2 DISCRETE DYNAMIC EVENT TREES on the sequence. For example, node 'c' may refer to the hardware state as node 'a', but earlier in time, and In a traditional PSA, the analyst builds the event tree with somewhat different physical process parameter to represent the possible evolutions of various values. scenarios. This 'accident sequence modelling' process is based on separate off-line analyses of the The DDET scheduler thermal-hydraulic response of the plant. Multiple The internal structure of a DDET simulation consists analyses are performed to consider variations of the of three parts: timing of automatic and operator actions, in order to define success criteria and the associated time • a "scheduler" that manages the branching process, windows, and, in general, to consider the interactions including the saving of the system state at the of the plant, the automatic systems, and the nodes and the restart of the simulation; operators. In contrast, the discrete dynamic event tree • a model of the hardware and human performance, (DDET) is a simulation-based framework that which describes how the equipment and personnel integrates these models to generate the event tree responds in the scenario; and "dynamically" and automatically. The role of the • a model of the plant physical and control processes. analyst is then to specify these models: i.e. how these responses develop and interact with each other. The core of the DDET is the scheduler, which controls the simulation models. It introduces a probabilistic DDETs were first implemented in the Dynamic Logical branching point, for instance, when the plant process Analysis Methodology (DYLAM) software tool, [e.g. evolution leads to a demand on a hardware 7,10]. The subsequent Dynamic Event Tree Analysis component. For example, when a vessel pressure Method (DETAM) [8] and the Accident Dynamic exceeds the high-pressure protection setpoint, a Simulator (ADS) [9,11,12] share the overall approach. demand is placed on the pressure relief valve; the The differences lie in the branch control algorithms relief valve may open successfully, or it may fail to and in the models of the response, in particular, the open. Conversely, the scheduler modifies the human response. boundary conditions for the plant process simulation when the personnel performs an action. Dynamic generation of the event tree In a dynamic event tree, the basic idea is to save the As shown in Fig. 2, the generated dynamic event tree state of the system (a snapshot) at each stochastic may include branching points for the system state (probabilistic) branching point, continuing the (equipment success/failure), the process variables simulation for one of the possible outcomes. At the (when key values such as setpoints are reached), for end of the scenario, which may be based on time or software (for the control systems), and for human on an absorbing state, the analysis returns to a action alternatives [12]. branching point, but now continues the simulation Probability accounting considering each of the alternative outcomes in turn. In this way, a tree of scenarios is built up, as shown in At each branching point, the probabilities of the Fig. 1 [13]. branches sum to unity. 23
Fig. 2: Schematic of the generated dynamic event tree.
The conditional probability of the end state (ES) is the With appropriate branch generation rules and a product of the probabilities of the branches included in carefully selected minimum step size, the the scenario. The sequence probability is the product computational limits on the size of the tree can be of this conditional probability with the initiating event satisfied while retaining sufficient resolution to frequency. The notion of accounting for the represent the possible event sequences. probabilities associated with stochastic events, instead of sampling the events in Monte Carlo fashion, 3 OVERVIEW OF THE ADS SOFTWARE TOOL allows stochastic, dynamic processes to be analysed in the DDET framework while avoiding the Development of the ADS tool started at the University disadvantages of Monte Carlo methods. of Maryland (UMD) more than a decade ago. The first generation code included four computing nodes to Although branching can theoretically occur at any time represent the behaviour of the reactor, steam point, the sequence evolution is discretised into a generator and pressurizer in a PWR plant, and hard- finite number of (variable) time steps in order to coded rules to represent the operators' actions [9]. manage the size of the tree. The system state Subsequent versions of the tool have followed this transitions and operator cognitive decisions, and successful demonstration of the ADS approach. resulting actions, are then modelled as occurring at these discrete points. The available computing power limited many early DDET implementations, including the earlier ADS In practice, the combination of events with stochastic versions, to problems with reduced scope, or with outcomes, processes with stochastic durations and simplified models (in particular, of the computation- interdependence of stochastic and continuous-time intensive physical processes). With the increase in dynamics leads to an overwhelmingly large number of computing power in the last decade, however, potential sequences. Clearly, in spite of the applications of the ADS tool to practical problems with discretization of time, an exhaustive simulation that a more realistic scope can now be pursued. expands all possible branches of the dynamic event tree is generally not feasible for a realistically-sized The most recently released ADS program [14] system due to this "combinatorial explosion". In fact, it contains six modules. Its architecture is shown in Fig. may not even be desirable because the exhaustive 3. The User Interface Module enables the user to edit tree includes branches whose probability is negligible. the inputs and initial conditions, and to control the Besides discretization in time and probability analysis parameters. The Scheduler Module accounting, avoiding the combinatorial explosion is implements the DDET algorithms discussed above. one of the functions of the scheduler. The personnel, plant processes and equipment states are represented by the Crew Module, the Indicator One of the mechanisms to 'prune' the tree is to Module (the plant human-machine interface and implement a probability cut-off. At each stochastic safety logic), the System Module, and the Component event, the scheduler calculates the probability of each Reliability Module. The Scheduler controls all of these branch, and determines whether to expand this models (lines 1-4 in Fig. 3) and their interactions (lines branch further. The scheduler does not resume the 5-8). simulation of branches with probabilities below the cut-off, or of branches that have reached an absorbing state. 24
point is generated to represent the (hardware) system state branching point. Here, the scenario branches according to whether the component fails or performs successfully. After all system-related information has been updated, the Crew Module calculates the crew response. As with the hardware, branching points are generated if action alternatives are selected. The Scheduler Modules summarizes all information generated and applies the cut-off algorithms to determine whether the branches are retained; in addition, it calculates the branch probabilities and performs the probability accounting. If the end-state criteria (i.e. the criteria for a safe, stable state, or a scenario failure state) have not been reached, the process repeats again to the next time step. Once a sequence end state is reached, the Scheduler recourses to an open Fig. 3: Architecture of the ADS dynamic PSA tool. (undeveloped) branch. The DDET is complete when no open branches remain. The Crew Module, with the use of the IDAC model [14], simulates the operating crew response. This 4 THE HUMAN RESPONSE MODEL module represents the crew with three types of Scope and modelling approach operators: the Shift Supervisor (decision maker), a reactor operator (who implements actions), and a In an event-tree/fault-tree PSA, the human response technical advisor (with a consultative role). These is analysed by considering the scenario context at models interact with the plant via the Indicator Module each snapshot where an operator action is required. (line 5) and with each other (not shown). At this point, the task of the HRA analyst is to determine whether salience of the available The Indicator Module models the human-machine indications that cue this action, and the degree to interface, which includes the display of plant process which procedures and training support the personnel parameters generated by the System Module (the in assessing the situation and selecting the plant model). In addition, the Indicator Module appropriate response, is justified. The relevant processes the system information to generate alarms ergonomics are also examined for both the decision and automatic responses, according to the control and and execution elements of the action. safety logic. Finally, the automatic responses and actions performed by the operators on the plant are transferred through this module to the Component Reliability Module. The Component Reliability Module models both failures on demand of equipment and running failures. Examples include the failure of a valve to open and the failure of an operating pump, respectively. In this way, this module represents the actual state of the equipment in response to the demands of the control and safety logic, and of the operators. The System Module represents the plant physical processes. The boundary conditions for this physical process simulation are determined by the equipment states. Figure 4 shows the ADS simulation flow diagram. After the initialisation of the system, controls, and crews' initial state, the System Module calculates the system state at the next time step. The Indicator Module updates the alarms state, physical parameter values and indications of the component state based on the new system state.
After the updates of the System Module and Indicator Fig. 4: ADS simulation flow diagram. Module, the Component Reliability Module calculates the equipment failure probabilities. If the failure probability is larger than pre-set criteria, a branching 25
Table 1: The operators' activities generated by ADS for one event sequence (excerpt).
Time Decision Maker Reactor Operator Technical Advisor System (s)* (Shift Supervisor) Response Response Response Response 0 Pipe breaks, alarms triggered. 1 Changes Goal to Changes Goal to TS Changes Goal to TS 'troubleshooting' (TS) 2 Changes Strategy to Changes Strategy to LE Changes Strategy to LE 'logic expansion' (LE) 3 Selects Diagnosis Selects Diagnosis Selects Diagnosis PI1- PI1-LBRK (training- PI1-LBRK -LBRK based procedure) 5 Asks Reactor Operator Checks SG1 Level to check SG1 Level 6 Checks SG1 Level Checks SG1 Level
23 Concludes and communicates Diagnosis PI1-LBRK 24 Does not accept Advisor's conclusion 73 Changes Goal to MGSM Changes Goal to MGSM Safety injection pumps due to EOPs entrance due to EOPs entrance start point reached point reached 74 Changes Strategy to Changes Strategy to Fl procedure-following (PF) 75 Enters Procedure E-0 Step 0
301 Asks Reactor Operator to control SG2, SG3, and SG4 water levels 302 Performs SG2, SG3 and SG4 water levels control 372 System is stabilized
* The time values result from task durations assumed for demonstration runs.
In a dynamic PSA, the human response model must 4. A model for the execution of actions, to represent, represent the activities of the personnel throughout for instance, the probability of unintentional errors and the scenario. In this way, the human model has the omissions. possibility to make decisions, and to implement 5. The strategies for decision-making, and the factors actions as the cueing indications arise during the that influence the selection of these strategies, such simulation. The workload due to these activities as stress, workload, and time pressure. influence their capacity to detect, interpret, decide and act. These human model components, which contain deterministic and probabilistic elements, allow the The models for each of the operators consist of five human performance to be modelled within the DDET, major components. These are: dynamically generating the branching points, i.e. the 1. The time required for different types of tasks alternative branches and their probabilities. This (communication, processing, and the execution of basic modelling approach is shared with other DDET procedure steps and tasks) implementations, as in [8,15], as well as other human- machine system modelling frameworks, such as task 2. The bases for situation assessment: i.e. how the network simulation [16] and discrete event simulation personnel determines the state of the plant in light of [17,18,19]. procedures and training The model in the DDET and post-simulation 3. The bases for response selection (decisions concerning action alternatives), again grounded in In a conventional event tree, the scenario histories procedures and training were defined in the accident sequence modelling process by the analyst; the event tree tool calculates the probabilities of the scenarios. In contrast, the 26
DDET histories are generated directly by the SGI Level falls simulation, and arise from the interactions of the plant below safety margin and human models. Thus, the generated dynamic Goal Strategy Diagnoses Provide Accept ... Goal Strategy Action Selection Selection Formation Advice Advice \ Change Change Performed event tree yields not only the probabilities for the end- Decision PipelBrk N/A Not Accepted I MGSM FP By Action Taker Maker states, but also the scenario histories underlying the | MGSM | pp Pipe2 Brk [Accepted | LE By Self tree. : Action . Pipel Brk N/A N/A In post-simulation analyses, the generated histories Taker Pipe2 Brk [Not Perfor can be examined to identify the contributing factors. Table 1 shows the activities of the personnel during a Pipel Brk Provided N/A Pipe2 Brk Not Provided single sequence [14]. (The indicated times result from the durations assumed for code testing and demonstration.) Key plant events and alarms, "cognitive" events related to the use of procedures and situation assessments, communications among Fig. 5: Highlights of the operators' processing the operators, and the actions of the plant personnel, activities and alternatives during a sequence. are all included in the Table. Following the pipe break and alarm at time=0, the In addition to using the plant parameters at the operators perceive the alarms. Their goals switch from decision points to determine the alternative responses routine monitoring to troubleshooting (TS) and the and their probabilities, the model also tracks the strategy changes to 'logic expansion' (LE) — a form of operators' internal responses to the situation. In the rule-based reasoning. At 5s, the decision-maker asks example shown in Fig. 6, the model predicts that the the reactor operator to check the level in the steam stress level increases when the water level in the generator SG1. At 74s, the safety injection pump steam generators (SG) first drops below the desired starts. The goal switches to maintaining a global safety limit (accompanied by alarms annunciating this safety margin (MGSM), while the strategy is condition), and continues to increase until the procedure-following (PF). The decision-maker enters operators are guided by the procedures to take the the post-trip emergency procedure, E-0, at 75s. necessary action in response to the situation. In this way, the influence of the evolving stress level on the Figure 5 represents this sequence as a time line, and operators' response can be dynamically treated. illustrates the different branches in the scenario. For instance, an alternative strategy at t=1s would be procedure-following (PF), i.e. selecting a procedure immediately instead of performing rule-based reasoning based on training.
Time (s)
Fig. 6: Predicted stress level of the decision-maker plotted against the steam generator water levels. 27
5 CURRENT EFFORTS AND OUTLOOK approach to the dynamic PSA is taken in the PSI- UMD work. Although the human response is treated With the ADS tool as a framework for dynamic PSA, probabilistically, the main use of the ADS tool at the current work within the PSI-UMD joint effort has present is to identify key scenarios involving dynamic two main components: effects and human errors. Validation of the human - further development of the ADS tool in the area of response model would be required to support the computational efficiency and the treatment of quantitative results. hardware reliability; and With regard to HRA, the DDET is thus used as a tool - HRA issues in the dynamic PSA framework, which to assist in the identification of significant error includes the model of the human response in ADS, as opportunities, while the scenario histories provide well as methods for analysing the human response. specific information for additional post-simulation, and quantitative analyses of human performance. In the As noted above, the integration of a model to identification process, the human response generate the response of the plant in accident probabilities essentially represent a mechanism for scenarios in a dynamic PSA enables the screening, since low-probability scenarios are cut off interdependence between the behaviour of the plant in the branching algorithm. As in a traditional PSA, the and its systems, and the responses of the personnel, dynamic PSA model can be revised in an iterative to be more closely examined. The ADS development process to represent the scenarios of concern in more work, led by UMD, seeks to address the associated detail. The information concerning plant response, computational demands through the use of multiple plant parameters, the personnel workload, and processors and refinement of the DDET branching strategies predicted by the model, are then used to algorithm to reduce unnecessary branching. quantify the key scenarios. Accordingly, The hardware reliability issue concerns the treatment improvements in the support within ADS for post- of the support systems that allow the plant's frontline simulation analysis of the runs, e.g. for examining systems to respond to a scenario. Specifically, when a specific scenario histories, are also planned. frontline system fails to operate, due to a support A case-application study is the focal point for this system problem, the impact on other systems that rely work; it examines the risk of a through-wall crack of on that support system needs to be taken into the reactor pressure vessel in overcooling scenarios, account. In a classical PSA, these so-called support also referred to as pressurised thermal shock (PTS) system dependencies are treated by means of fault scenarios. The dynamic PSA framework is promising trees. A solution that would not require the inclusion of in this case, because the probability of RPV cracking the complete fault-tree configuration is currently being depends on the reactor coolant system pressure and examined. temperature, in terms of both magnitude and In the emerging dynamic PSA framework, the analysis gradients. Thus, the timing of automatic responses of human performance begins with the application of and operator actions are significant in the task analyses and HRA methods to characterize the determination of the risk. It is worth noting here that procedure-based and training-based responses of the the dynamic PSA is intended to complement the personnel. Through these analyses, the inputs to a conventional PSA rather than to replace it [5]. The human response model within DDET tools like ADS dynamic PSA's analytical and modelling complexity, are generated. The dynamic event tree is then and the significant effort required restricts its use to produced by the tool, which also quantifies the scenarios where dynamic issues need to be scenario evolutions and accident sequences. examined. The issue of identifying when dynamic issues are likely to be important in scenarios is The HRA-related efforts, led by PSI, address both of therefore also being examined in the application these aspects. To develop the inputs to the human study. response model, the characterisation of the human response extends the approach used to identify EOC ACKNOWLEDGMENTS opportunities in earlier work [1]. The work on the The PTS application of ADS code has been partially human model itself constitutes a major issue for the supported by the U.S. Nuclear Regulatory dynamic PSA. The next steps with the model include Commission, Office of Research. Their support is duly the revision of the human model to enable the key acknowledged. The HRA project at PSI is supported aspects of the operators' main tasks to be by the Swiss Federal Nuclear Inspectorate, and by the represented in the dynamic PSA. This will involve Swiss utilities. operationalizing and simulating the effects and interactions of the risks, together with the mitigating REFERENCES factors. [1] V.N. Dang, B. Reer, S. Hirschberg, "Analyzing Clearly, the dynamic PSA in the long term requires an Errors of Commission: Identification and a First HRA model that would be capable, without post- Assessment for a Swiss Plant', in Proc. of the simulation analyses, of both simulating the most likely OECD/NEA Workshop, "Building the New personnel responses, and quantifying these within the HRA: Errors of Commission - from Research to DDET. The quantification especially poses a Application", Rockville, MD, USA, 7-9 May 2001, challenge, as can be expected. A step-by-step OECD Nuclear Energy Agency, Paris. 28
[2] N. Siu, "Risk Assessment for Dynamic Systems: [12] Y.H. Chang, A. Mosleh. "ADS: A Computer An Overview', Reliability Engineering and Program for Dynamic Probabilistic Risk System Safety, 43, 43-73 (1994). Assessment', Proc. of the ANS International Topical Meeting on Probabilistic Safety [3] J. Devooght, C. Smidts, "Probabilistic Reactor Assessment, Detroit, Michigan, USA, 6-9 October Dynamic-I: The Theory of Continuous Event 2002, 206-211, American Nuclear Society, La Tree", Nuclear Science & Engineering, 111, 229- Grange Park, Illinois (2002). 240 (1992). [13] V.N. Dang, "Frameworks for Dynamic Risk [4] T. Matsuoka, M. Kobayashi, "GO-FLOW: A New Assessment and Their Implications for Operator Reliability Analysis Methodology, Nuclear Modeling, in Fifth International Workshop on Science & Engineering, 98, 64-78 (1988). Dynamic Reliability: Future Directions, 1998, [5] S. Swaminathan, C. Smidts, "The Event Greenbelt, Maryland, Center for Technology Risk Sequence Diagram Framework for Dynamic Studies, University of Maryland, College Park, Probabilistic Risk Assessment', Reliability Maryland (1998). Engineering and System Safety, 63(1), 73-90 [14] Y.H. Chang, A. Mosleh, "Cognitive Modeling and (1999). Dynamic Probabilistic Simulation of Operating [6] M. Cepin, B. Mavko, "A Dynamic Fault Tree", Crew Response to Complex System Accidents Reliability Engineering and System Safety, 75, (ADS-IDACrew)", CTRS-B6-06, Center for 83-91 (2002). Technology Risk Studies, University of Maryland, College Park, Maryland (1999). [7] A. Amendola, "Accident Sequence Dynamic Simulation versus Event Trees," Reliability [15] P.C. Cacciabue, G. Cojazzi, "A Human Factors Engineering and System Safety, 22, 3-25 (1988). Methodology for Safety Assessment Based on the DYLAM Approach", Reliability Engineering [8] C.G. Acosta, N. Siu, "Dynamic Event Tree and System Safety, 45, 127-138 (1994). Analysis Method (DETAM) for Accident Sequence Analysis", MITNE-295, Massachusetts [16] J.C. Schryver, "Operator Model-Based Design Institute of Technology, Cambridge, MA (1991). and Evaluation of Advanced Systems: Computational Models", Proc. of the 1988 IEEE [9] K.-S. Hsueh, "An Integrated Simulation Model for Fourth Conference on Human Factors and Power Plant/Operator Behavior in Accident Conditions", Plants, Monterey, CA, 5-9 June, 1988, 121-127, PhD Dissertation, Department of Materials and Institute of Electrical and Electronic Engineers, Nuclear Engineering, University of Maryland, New York (1988). College Park, Maryland (1992). [17] R. Laughery, "Using Discrete-Event Simulation to [10] G. Cojazzi, P.C. Cacciabue, P. Parisi, "DYLAM-3, Model Human Performance in Complex A Dynamic Methodology for Reliability Analysis Systems", Proc. of the 1999 Winter Simulation and Consequences Evaluation in Industrial Conference, Phoenix, AZ, USA, 5-8 December Plants, Theory and How to Use", ISEI/SER 1999, Vol. 1, 815-820, ACM Press, New York, 2192/92 (EUR 15265 EN), EURATOM (1993). NY, USA (1999). [11] K.-S. Hsueh, M. Ali, "The Development and [18] Y. Huang, "Modeling Control Room Crews for Application of the Accident Dynamic Simulator for Accident Sequence Analysis," Ph.D. Thesis, , Dynamic Probabilistic Risk Assessment of Massachusetts Institute of Technology, Nuclear Power Plants", Reliability Engineering Cambridge, MA, September 1991. and System Safety, 52, 297-314 (1996). [19] Y. Huang, N. Siu, D. Lanning, J. Carroll, V. Dang, "Modeling Control Room Crews for Accident Sequence Analysis", MITNE-296, December 1991, Massachusetts Institute of Technology, Cambridge, MA (1991). 29
LEVEL-SWELL PREDICTION WITH RETRAN-3D AND ITS APPLICATION TO A BWR STEAM-LINE-BREAK ANALYSIS
Y. Aounallah, K. Hofer
Level-swell experiments have often been simulated using system codes, such as TRAC and RELAP, but only cursory assessments have been performed with the operational-transient code RETRAN-3D, the main system code used within the STARS project The present study, initiated in the framework of a BWR Steam-Line-Break (SLB) accident scenario, addresses this lacuna by performing RETRAN simulations of the General Electric Level-Swell experiments, and by investigating their implications on power plant accident analyses. Parameters to which the predicted level swell is sensitive have been identified, and recommendations on code options are made. The SLB analysis objective was to determine the amount of steam and liquid discharged through the break under specified boundary conditions, and to gauge the results against reference values. The impact of the nodalization of the upper part of the reactor pressure vessel was investigated and found to play an important role, whereas the level swell induced from flashing was found not to be the predominant factor for these simulations.
1 INTRODUCTION e.g. break size, MSIV closure time, and variations in initial conditions. Three cases were identified for the The prediction of the two-phase mixture-level analysis, two of them with a guillotine break and one behaviour during blowdown conditions is of interest to with a small break size equivalent to 10% of the cross- both the chemical and power industries. It is sectional area of one main steam line. The analyses particularly salient for Light Water Reactors, for presented here were all performed at "hot-standby" example for the prediction under accidental situations conditions. of core uncovery and steam generator dryout. Also, an a priori conservative assumption may not always 2 GENERAL ELECTRIC LEVEL-SWELL lead to a conservative result. For example, a EXPERIMENTS conservative increase of reactor power may lead to overprediction of the level swell of an uncovered core, 2.1 Description of experiments and models or a steam generator, resulting in an overprediction of the cooling; hence the need for complementary best- The General Electric (GE) Level-Swell experiments estimate calculations with accurate mixture-level were designed to investigate separate-effect predictions. phenomena, such as critical flow, mixture level swell, and axial void distributions, during blowdown The capability of the operational-transient code conditions. The experiments consisted of RETRAN-3D [1], the main system code in the STARS depressurizing a quiescent pool of water, initially in (Safety Research related to Transient Analysis of the thermodynamic equilibrium with its vapour phase. The Reactors in Switzerland) project, to accurately predict fast rate of depressurization (up to 10 bar/s at the the mixture level behaviour, has been the object of beginning of the transient) resulted in a sudden rise, only cursory effort, based on reported studies [2], in or swell, of the liquid free surface due to flashing. The contrast to other system codes, such as TRAC and initial level swell was followed by a level shrink as the RELAP [3,4,5]. The present simulation of the General steam production in the mixture decreased as a Electric blowdown experiments [6], which was initiated consequence of the reduction in depressurization rate. in the framework of a steam line break analysis, aims at addressing this shortcoming. A series of experimental tests were performed with Small Blowdown and Large Blowdown vessels, both One objective was to assess the code models as 4.27 m long, vertically-oriented cylinders, with "stand-alone" components of more complex systems, respective internal diameters of 0.305 m and 1.219 m. such as the one for the present postulated main The depressurization rate was controlled by mounting steam line break in a BWR plant; the break is different size venturi nozzles on the discharge line. assumed to occur inside the turbine building. The Experimental uncertainty was reported to be typically main interest of the study is to investigate the basis for +/- 0.12 m for the mixture level, and up to +/- 0.05 for calculations of the radioactive isotopes contained in the void fraction. No information was available in the the primary coolant, and therefore entrained in the open literature regarding the uncertainty of the system steam and liquid mass discharged through the break. pressure measurements. The break outflow is interrupted by a complete closure of the main steam isolation valves (MSIVs). A rupture Two modeling schemes were used, both aimed at of a main steam line in the turbine building is most maximizing consistency between the component critical from the viewpoint of release of activity to the assessment model and (i) the BWR plant model, and environment, since this building is not included in the (ii) the experimental discretization scheme for the barrier formed by the containment. The present study measured variable. This was reflected in the consists of calculating the mass of steam and liquid nodalization schemes in both cases, as well as in user discharged from the break under different conditions: options. Thus, while a two-node model was used in the framework of the plant study, a seven-node model 30 was used for the detailed simulation of the GE seems to be no supportive evidence in the open experiments. The impact of the radial nodalization in literature for selecting a particular value within the the phase separation region was investigated for the range considered (0.3 m/s to 1.5 m/s). Very low or BWR system model, as described below (Section very large values of Vb can be selected for limiting 3.5). cases, such as the determination of the maximum level swell or the collapsed liquid level, respectively. 2.2 Base-case calculations and sensitivity tests The nodalization for the base-case calculations (GE Test 5801-15) and sensitivity tests is depicted in 1 1 1 1 1 1 1 1 1— Fig. 1.
Nodalization of RPV upper-down comer region /*/ . / / ^ '7/ • 'P / Nodalization of the GE Vessel . • r [ • Experiment i • • - • • Vb=0.3 [m/s] Vb=0.6 [m/s] - - Vb=0.9 [m/s] Vb=1.3 [m/s] (*) -- Vb=1.5 [m/s] (*) Base-case value.
0 2 4 6 8 10 12 14 16 18 20 Time [sec] Fig. 2: Effect of the bubble-rise velocity (Vb) on the mixture level (with Co=0.0 [-]).
Time-De pendent Volume The effect of varying the bubble-gradient parameter is shown in Fig. 3, where the value Co=0 corresponds to a uniform void profile, and Co=1.0 corresponds to the Fig. 1: Nodalization for the base-case model. maximum void gradient. The base-case value (Co=0.0) maximizes the level swell, since the void The RETRAN 3-D bubble-rise model (BRM) was used fraction at the free interface is then at a minimum, and for the Control Volume 2 (Fig. 1) and, in the context of the vapour flow rate leaving the mixture is then the this model, the salient parameters identified for these smallest. Conversely, the maximum void gradient tests were the bubble-rise velocity (Vb), the bubble- leads to the lowest mixture level, since it maximizes gradient parameter (Co), and the break-valve the vapour flow rate leaving the mixture free interface. contraction coefficient (Cn), as listed in Table 1 below. Unless otherwise stated, these base-case • Experiment parameters, used in the preliminary calculations for Co=0.0 [-] (*) steam line break analysis (see Section 3), were - - Co=0.2 [-] • - • • Co=0.5 [-] implemented in the calculations. Co=0.8 [-] - Co=1.0 [-] Table 1: Base-case parameters.
Parameters Values Remarks
Bubble-rise velocity 1.3 [m/s] Bubble-gradient parameter 0.0 [-] Homogeneous Break valve contraction coeff. 1.0 [-] No correction
The result shown in Fig. 2 for the Large Vessel tests, and subsequently confirmed within the range of the (*) Base-case value. sensitivity tests performed, is that, for the base case 0 2 4 6 8 10 12 14 16 18 20 (i.e. with Vb=1.3 m/s), the rate of level swell and level Time [sec] shrink is lower than the experimental values. Also, the Fig. 3: Effect of the bubble-gradient parameter (Co) peak elevation of the two-phase mixture level is on the mixture level (with Vb=1.3 m/s). underpredicted, and the timing of this peak is delayed. Sensitivity calculations, stemming from the base-case Figure 4 shows the combined effect of variations in Vb simulation, were performed to assess the impact of and Co, and thus indicates the domain swept by the varying the bubble-rise velocity, the bubble-gradient mixture level velocity during depressurization, and parameter, and the break valve contraction coefficient. identifies the envelope cases. It can be seen that Figure 2 shows the effect that the bubble-rise velocity drastically different results can be obtained for a (Vb) has on the mixture level, particularly when there relatively narrow range of Vb. The bubble-gradient 31 parameter becomes increasingly important as rates of level swell decrease (Co would be inconsequential for Vb~0 m/s since no vapor would then cross the mixture free interface).
Fig. 5: Effect of the valve contraction coefficient (Cn) on the mixture level.
Fig. 4: Effect of combined bubble-rise velocity (Vb) and bubble-gradient (Co) on the mixture level: envelop cases. With the base-case parameters (Vb=1.3 m/s, Co=0.0), the system pressure is also underpredicted during vessel depressurization, as illustrated in Fig. 5. The initial dip in the measured system pressure, followed by a recovery, is caused by the time delay in flashing (not included in the code) as steam is vented from the steam dome, with no significant void formation takes place in the superheated liquid. After this brief period, typically less than one second, significant vapor bubble formation takes place, causing the two-phase mixture level to swell and compress the steam, as manifested in the recorded pressure recovery. Significant correction of the break-valve contraction coefficient (Cn=0.8) was required to obtain the correct depressurization rate, as can be seen in Fig. 5. The Fig. 6: Effect of the bubble-rise velocity (Vb) on the impact of the bubble-rise model parameters, system pressure. particularly Vb, on the depressurization rate is also shown in Fig. 6. The changes in slope of the pressure- 2.3 Prediction of the axial void profile decay curves correspond to conditions where the mixture level reaches the break elevation (as can be A model was used to assess the code's capability to seen in Fig. 2). The initial single-phase steam flow predict the axial void profile during blowdown vented out of the pressure vessel is then replaced by situations. The experimental Test 1004-3 has been a two-phase mixture which reduces the vapour mass the most frequently investigated one, including by the vented, and thereby slows down the pressure decay. RETRAN code developers [3], and was therefore also selected here. Since the process of interest is the loss of liquid inventory in the pressure vessel, the main outcome A seven-node model (Fig. 7) was used to match the from these calculations is that the base-case axial distribution of the void fraction measured parameters would give conservative results experimentally. In other words, the node length (overestimation of liquid losses) only if the mixture corresponds to the axial distance between the water level reaches the break-line elevation. This is successive pressure tappings. Thus, there is because the experimentally observed level-shrink consistency in the control volumes used in the rate, which follows the initial level swell, is experiment and those in the code simulations. underpredicted, and therefore the predicted mixture The void fractions are very well predicted by the code, level remains high for a longer period of time. as can be seen in Fig. 8, as long as neither the experimental nor the calculated mixture level reaches the BRM lower boundary (Node 6 in Fig. 7). This 32 actually occurs at 100 s for Test 1004-03, but was predicted to occur at -150 s, as shown in Fig. 9. As a • Experiment Vb=1.3 [m/s] (*) consequence of this delay, the void fraction increase •-•• Vb=Wilson in Node 5 is also correspondingly delayed.
TDV
Saturated Liquid
(*) Time-dependent volume Fig. 9: Effect of the bubble-rise velocity (Vb) on the Fig. 7: Nodalization for GE Test 1004-3. mixture level (Test 1004-3). For comparison of the code predictions with the Among the RETRAN-3D void model options, the experimental database, the nodalization shown in algebraic slip model (isflag=3) and the dynamic slip Fig. 2 was used, together with the BRM base-case model (isflag=5), both based on the Chexal-Lellouche parameters and the pressure-dependent, Wilson drift-flux model, gave very similar results. bubble-rise velocity model [1]. Good predictions of the depressurization rates were obtained by adjusting the valve contraction coefficients, found to be Cn~0.9 and • Exp. Node 2 Calc. Node 2 Cn~0.7 for the Large-Vessel and Small-Vessel tests, 0.6- o Exp. Node 3 respectively. These adjustments resulted in lowering Calc. Node 3 + Exp. Node 4 of the maximum elevation reached by the two-phase — Calc. Node 4 0.5- A Exp. Node 5 mixture, and a reduction in the discrepancy for the Calc. Node 5 timing of this maximum.
(dP/dt)i = -0.8 [bar/s] - - (dP/dt)i = -2.1 [bar/s] 3.5 (dP/dt)i = -5.2 [bar/s] . •-•• (dP/dt)i =-10.3 [bar/s]
£ TEST 5702-16
82.5 / / C>D / TEST CD I 2 / TEST 5801-l^r^t^y ' 0 20 40 60 80 100 120 140 160 180 Time [sec] Vb=1.3 [m/s] ' Fig. 8: Comparisons of axial void profiles (GE Test 1004-3). TEST 1004-3
0.5 2.4 Comparison against the GE Level-Swell database: use of the Wilson BRM In addition to the base-case and sensitivity Time [s] calculations, the eleven experimental tests Fig. 10: Wilson bubble-rise velocity for different constituting the GE Level-Swell database were also experimental tests: Vb=1.3 m/s being the used to identify possible inconsistencies in the base-case value and (dP/dt)j the initial identified trends. The Large-Vessel experiments depressurization rate. consisted of four tests with different break sizes, the initial water level being the same for all tests. The The main finding is that the base-case parameters Small-Vessel experiments included cases with four lead to an underestimation of the mixture level peak break sizes (some of which included a vessel elevation for cases where the steam line was not restriction plate), and two initial water levels. All tests reached, and an overestimation otherwise. The were initiated at a system pressure of about 70 bar. Wilson model (Fig. 10), on the other hand, is considered to have performed well. It must be stated 33 that this option was used together with the building following a break in a main steam line. The homogeneous model option (Co=0) for consistency two most relevant components for mitigating this with the Wilson's methodology (which is based on an event are the flow limiters and the main steam axially-averaged value of the bubble-rise velocity). In isolation valves (MSIVs). The design of the integrated Fig. 10 is shown the bubble-rise velocity for four tests venturi-type flow limiters defines the maximum steam (1004-3, 1004-2, 5801-15, 5702-16) as calculated by blowdown rate, while the release of primary coolant is RETRAN-3D. Each of these tests is characterized by limited by the closure of the MSIVs (Fig. 12). its initial depressurization rate (dP/dt)j. The Figure Therefore, one objective of the steam line break shows how the bubble-rise velocity increases over the analysis is to assess the impact of an early reactor first 5 to 20 seconds of the transient, from a low value isolation as a consequence of a reduced MSIV to a quasi steady-state value, and how both the rate of closure time. The following cases were investigated: increase and final level depend upon the • 200% break, 10 second MSIV closure time, depressurization rate. The base-case value of Vb=1.3 m/s is also shown. • 200% break, 5 second MSIV closure time, • 10% break, 5 second MSIV closure time. Among the Small-Vessel tests, two tests (8-28-1 and 1004-2) clearly indicate venting of a two-phase The first case is used for benchmarking the RETRAN- mixture, and from the analysis of these tests comes 3D analysis against the results of a BWR safety the observation that the use of the base-case analysis report (SAR). parameters delays significantly the calculated level- shrink rate, as illustrated in Fig. 11, while the use of 3.2 Sequence of events the Wilson model provides much closer agreement. The first response to a break in one of the main steam lines inside the turbine building is a rapid increase in 4.51 1 1 1 1 1 steam flow as the reactor depressurizes. A steam line high-flow signal detected in the two flow limiters 4- involved initiates MSIV closure (for all valves) shortly after the break, which is followed by a reactor scram as soon as the MSIVs reach the 90% open position. The flow limiters restrict the steam flow rate out of the reactor pressure vessel (RPV) to the maximum allowed under critical flow conditions. Referring to one flow limiter, this value is approximately twice as large as the rated flow for 100% power operation. As a consequence of the reactor blowdown, and the subsequent decrease in reactor pressure, the reactor water level will swell, due to flash evaporation, with 1.5 - Base-case BRM • •-•• Wilson BRM the result that the two-phase level will rise up to the steam nozzles and liquid will flow out into the steam 1 1 1 1 1 1 lines. Finally, this event is terminated by the complete 0 20 40 60 80 100 120 Time [s] closure of the MSIVs. Fig. 11: Comparison of the mixture levels obtained Table 2: Values used for initial conditions at hot- with the base-case and the Wilson bubble- standby in the RETRAN-3D BWR steamline- rise velocities (GE Test 1004-2). break analysis. It must also be added that the conclusions reached here are not sufficient to fully characterize the Parameter Value Used in Analysis behaviour of the mixture level in more complex Core power 5% of nominal situation such as the mixture level in a BWR pressure Reactor water level 100% of nominal vessel during blowdown conditions, since, in addition Core mass flow 35% of nominal to flashing, the net mass and enthalpy balance over rate the nodes under consideration can play an important System pressure 93% of nominal role as well, as will be further shown in Section 3.5. Pressure in turbine 1 bar building 3 APPLICATION TO BWR STEAM-LINE- MSIVs MSIVs related to turbogroup A open; BREAK SCENARIO MSIVs related to turbogroup B closed 3.1 Problem description Position of turbine Closed bypass valves The objective of the main steam line break analysis Position of turbine Open applied to a BWR plant consists of calculating the stop valves release of steam and liquid transported to the turbine 34
Flow Inboard Outboard Limiter MSIV MSIV \ I / -X- Turbine B -X- ++-A t, "Junction "Splitc -H- >4- -! 1 A Turbine A -H- >4- < Breaks ^Section of Figure 13 "(see below) —h-
Drywell Reactor I Turbine | Building | Building
Primary/mar / SeconcSecondary Containmentainment ContainContainment i
Fig. 12: Main steam piping system with break in turbine building.
3.3 Initial Conditions For the RETRAN-3D analysis, the postulated break is Because the water inventory for a given reactor water modelled for the 20-inch diameter main steam line A level is larger at hot-standby than at full power, as a at the entrance to the turbine building. The break is consequence of the lower void fraction in the core, the located at about 5 m downstream of the location upper plenum and the separators, the amount of called "Junction" in Fig. 12, and about 9 m from the primary coolant discharged by the steam line break corresponding outboard MSIV location. The details of will be larger than under normal operating conditions. the nodalization for the ruptured steam line in the The RETRAN-3D analyses were therefore performed vicinity of the turbine building are given in Fig. 13. for hot-standby conditions, for which the main Node 700 is a time-dependent volume — a special parameters are summarized in Table 2. RETRAN control volume for which the constant thermodynamic conditions of the turbine building 3.4 RETRAN-3D model and analyses atmosphere are imposed. The activation of valves 3.4.1 Nodalization V18, V19, and V20 actually simulates the break (Section 3.4.3). The main steam line break analysis with RETRAN-3D was carried out based on a comprehensive system Turbine Building model including the reactor pressure vessel and its internals, the reactor core, the two recirculation loops and pumps, the complete main steam piping system, the turbines, the turbine bypass lines, the condensers, and the blowdown lines. A total of about 200 nodes was utilized. A general overview of the main steam piping system and the postulated break in one of the MSIV two steam lines in the turbine building are given in Fig. Fig. 13: RETRAN-3D model of the main steam line 12. Four main steam lines (denoted by A, B, C and D) break in the turbine building. are connected to the RPV. Each of these steam lines consists of the piping system, an integrated, venturi- During the model development stage, it was observed type, steam-flow restrictor (here referred to as flow that the calculated coolant mass discharged in the llmltei'), and two MSIVs. In addition, a dual-function turbine building is sensitive to the nodalization of the safety relief valve, and either a safety valve or a RPV components that involve phase separation motor-operated pressure relief valve, are flanged to during the depressurization phase. This mainly each main steam line upstream of the flow limiters. concerns the nodalization of the steam separators, the Inboard and outboard MSIVs are located on each side steam dryer, and the upper downcomer. Several of the primary containment, which provides the nodalization schemes were examined, and separation of the drywell and pressure suppression reasonable results were obtained with the one shown pool from the reactor building. in Fig. 14.
The 16-inch diameter main steam lines coming from The behaviour of the reactor water level is determined the RPV join together in pairs, and the resulting two in RETRAN-3D by the bubble-rise model (BRM). The 20-inch diameter lines (denoted by A and B in Fig. 12) parameters that can be specified in this model, as penetrate the reactor building, and further already described in Section 2, are: (i) the bubble-rise velocity Vb, which can be either a user-defined downstream enter the turbine building. Both lines constant (or a function defined within the code divide again into two 16-inch diameter lines that are "control" system) or a code-calculated value based on finally connected to the turbines valves. the Wilson correlation; (ii) the initial two-phase mixture 35 level; and (iii) the bubble-gradient Co. It should be During the blowdown period of the main steam line noted that the input values of the BRM, in particular break, the reactor water level swells for all three the bubble-rise velocity, can be adjusted by the code paths. When the level in Path 3 reaches the bottom of during steady-state initialization, if this option is the main steam line nozzles, liquid is entrained. The employed. two-phase mixture level in Path 1 can rise up to the top of the steam separator, and from there merge with The obvious reason for using a bubble-rise model is that from Path 2, and cause swelling in the steam that liquid flows out of a phase-separated volume only dryer. When the steam dryer is completely full, liquid if the mixture level reaches the outlet elevation. This is can also flow into the main steam lines. different for standard RETRAN-3D control volumes in which the liquid and steam phases are It is expected that when fluid enters the steam lines, homogeneously mixed. The RETRAN bubble-rise most of the liquid would flow through Path 3, and most model is therefore the most appropriate option for of the steam through Paths 1 and 2. tracking the reactor water level, and the associated The expected flow behaviour as described above was primary coolant discharge out of the RPV. correctly simulated with the nodalization scheme Special attention was given to the development of the illustrated in Fig. 14. The radial nodalization of the model so that it conforms to the flow behaviour RETRAN-3D plant model in the RPV upper region expected to occur in the RPV upper region of the includes three separated volumes: one for the steam nuclear power plant during a steam line break event. separators (here denoted as Separator Interioi), and In contrast to the GE Level-Swell facility, there are two for the upper downcomer (here denoted as three flow paths that need to be considered (Fig. 14a): Separator Exterior and Dryer Skirt). The separator exterior node comprises the volumes delineated by Path 1: steam separators steam dryer steam the steam dryer assembly seal skirt and the steam dome main steam lines; separators. The dryer skirt node represents the Path 2: upper downcomer inside of the steam dryer volume between the steam dryer assembly seal skirt assembly seal skirt steam dryer steam and the RPV, and extends vertically up to bottom of dome main steam lines; the steam lines. In order to obtain the correct flow behaviour, the height of the steam separator interior Path 3: annulus of the upper downcomer between the and steam separator exterior were extended to RPV and the steam dryer assembly seal skirt include the steam dryer volume. -> main steam lines.
(a) Schematic of upper RPV (b) RETRAN model of upper RPV
Fig. 14: RETRAN-3D nodalization for the upper RPV. 36
3.4.2 Code options selected •200% Break: V19 and V20 open completely and V18 closes at break activation time The major code options used for the main steam line within 0.1 ms. break analysis are summarized in Table 3. The analysis was carried out without coupling between • 10% Break: V19 opens by 10% at break activation thermo-hydraulic and neutronics. Instead, the core time within 0.1 ms, V18 remains fully power was kept constant at 5% of full power open, and V20 is not used (i.e. throughout the transient. remains closed). The rise of the reactor water level during the In the RETRAN-3D model, the turbines are simulated depressurization period depends on the separation of with time-dependent volumes, to provide the thermal- liquid and steam, which is mainly expressed by the hydraulic boundary conditions. In order to limit the bubble-rise velocity. It is obvious that smaller values steam flow from the turbine out of the break, Turbine result in a more rapid and larger rise of the level. In A is isolated shortly after the break by the turbine stop the context of a main steam line break analysis, the valves. The closure of the outboard MSIVs connected liquid phase would then reach the main steam line to the ruptured main steam line is not triggered by a nozzles earlier, and more liquid would be entrained in steam line high-flow signal. In accordance with the the steam lines, and finally through the break into the results of the BWR SAR analysis, they are closed at turbine building. An analysis with the bubble-rise 0.5 s after the activation of the break for both the velocity close to zero is therefore considered as a 200% and the 10% break analysis. Finally, the MSIV conservative approach. Based on the results obtained closure characteristic is based on a linear, time- from the separate-effect study (Section 2), the Wilson dependent table for the valve flow area. correlation for the bubble-rise velocity was selected 3.5 Results for this analysis. The correlation is used together with the bubble-gradient parameter Co=0.85 for the The break mass flow rates calculated by RETRAN-3D separator interior, and Co=0 for the separator exterior for the three cases analyzed are shown in Figs. 15a to and dryer skirt nodes. 15c. The results are normalized to the nominal main steam mass flow rate. The origin of the time scale is Table 3: Major code options of the RETRAN-3D main related to the break activation time. In addition, it steam line break analysis. should be noted that the break mass flow rate is defined here as that discharged at the steam line Code Option Details break; that is, referring to the model of Fig. 13, the Code version RETRAN-3D MOD003 sum of the flow rates from the steam line Nodes 561 Numerical scheme Four-equation option, thermodynamic and 562 to the turbine building Node 700. equilibrium In Fig. 15a, the results obtained for the 200% break Slip Algebraic slip equation based on drift flux model of Chexal-Lellouche with 10 s MSIV closure time are compared against the (isflag=3) data of the BWR SAR analysis. One of the most Power generation Power table, important differences is the time delay for the liquid to no reactor kinetics model used start to flow out of the break. It takes 2.4 s in the BWR Critical flow model Isoenthalpic expansion model SAR analysis, and 3.4 s according to the RETRAN-3D simulation. Furthermore, the slope is much steeper for Wilson bubble rise The correlation is used for all the velocity correlation bubble-rise volumes the BWR SAR analysis. The time delay and slope are mainly determined by the speed of the downcomer level swell, which itself is influenced by the timing of 3.4.3 Modelling of the sequence of events the void formation, by the bubble-rise model parameters, and by the initial reactor water level. The The core power and recirculation pump speed remain large peak (up to 390% of nominal main steam mass unchanged after the initiation of the postulated break. flow rate) at the beginning of the transient for the After the break, the feedwater mass flow rate is kept RETRAN-3D plot in Figs. 15a and 15b is the result of constant at the value equivalent to 5% of steam the rapid steam discharge out of both sides of the production at full power. The undamaged 20-inch guillotine break in the steam line. Such a peak is not main steam line (i.e. line B in Fig. 12), is isolated prior evident from the BWR SAR analysis. to the break by closing its outboard MSIVs. The maximum initial steam vessel blowdown rate is set by As expected, the RETRAN-3D analyses for the two adjusting the flow areas of the junctions associated cases with MSIV closure time of 5 s result in with the flow limiters. considerably smaller values of the total mass released compared with that for a closure time of 10 s. The main steam line break is simulated by the three Moreover, there is almost no liquid mass release in valves V18, V19 and V20 shown in Fig. 13. The the former case. valves V19 and V20 control the flow from the steam line to the turbine building, and valve V18 is used to The influence of the small break size of 10% is shown simulate the piping misalignment in the case of a in Fig. 15c. After the closure of the MSIVs at 5.5 s, the 200% break. For the break sizes investigated here, ruptured main steam line is isolated at both ends, and the valve operation sequences are as follows: 37 it takes another 15 s for complete depressurization to The numerical values of the integrated break mass the ambient pressure of the turbine building. flow rates are summarized in Table 4. The numbers are normalized to the total mass release of the BWR SAR analysis. Table 4: Relative integrated break mass flow rates for different steam line break scenarios.
Total Mass Liquid Mass Steam Mass Analysis [%] [%] [%]
BWR SAR • 200% break, 100.0 75.5 24.5 10 s MSIV closure time RETRAN-3D • 200% break, 90.8 37.5 53.3 10 s MSIV closure time • 200% break, 37.3 3.0 34.3 Time [s] 5 s MSIV closure time Fig. 15a: RETRAN-3D versus BWR SAR analysis • 10% break, 16.5 0.3 16.2 results for 200% main steam line break and 5 s MSIV 10 s MSIV closure time. closure time
As shown in Fig. 2 for the GE experiments, the RETRAN: Total Flow swelling of the reactor water level during RETRAN: Liquid Flow • • • • RETRAN: Steam Flow depressurization of the RPV depends on the bubble- rise model. A sensitivity study was carried out with the RETRAN-3D BWR plant model for the bubble-rise velocity of the separator exterior and dryer skirt nodes (Fig. 14). The goal was to find out by how much the time delay td, for the onset of the liquid mass flow rate determined at the RPV main steam line nozzles, can be modified in order to eventually match the results of the BWR SAR analysis (Fig. 16). The time delay is defined as the time needed by the downcomer two- phase mixture level to reach the RPV main steam line nozzles. The investigation was performed using the sum of the relative mass flow rates of the two 16-inch diameter main steam lines C and D at the exit of the Fig. 15b: RETRAN-3D result for 200% main steam RPV (Fig. 12). It should be noted that the RETRAN- line break and 5 s MSIV closure time. 3D results shown in Fig. 16 and Fig. 15a are different, since they were calculated at different locations: i.e. the RPV exit and the break location, respectively.
RETRAN: Total Flow RETRAN: Liquid Flow RETRAN: Steam Flow
10 11 12 13 14 15 16 17 18 19 20 Time [s] 10 11 12 Fig. 15c: RETRAN-3D result for 10% main steam line break and 5 s MSIV closure time. Fig. 16: Time delay td for the onset of liquid mass flow for a 200% main steam line break and 10 s MSIV closure time. 38
The bubble-gradient parameter was set at Co=0 parameter, and the valve contraction (or discharge) (uniform void profile) for the separator exterior and coefficient. While a fixed value for bubble-rise velocity dryer skirt nodes, and at Co=0.85 for the separator is useful for calculating bounding limits, it remains of interior node. For the entire study, the value limited applicability when accurate prediction is determined by RETRAN-3D for the bubble-rise required; for example, when the mixture level is a velocity of separator interior was Vb = 4.41 m/s. As triggering signal in the course of the simulation. The mentioned above, this value is calculated at the Wilson correlation for the bubble-rise velocity was beginning of the transient. Results are summarized in found to yield consistently good results. Significant Table 5. adjustments needed to be made to the break-valve contraction coefficient in order to obtain the correct Table 5: Effect of the bubble-rise velocity Vb on the vessel pressure decay rate. time delay td for the onset of liquid flow for a 200% main steam line break and 10 s MSIV In the case of more complex modelling, such as a closure time. reactor pressure vessel for the present BWR steam line break analysis, the axial and radial nodalization of Analysis Time delay td [s] the upper region was also found to play an important BWR SAR 2.4 role in predicting the reactor water level during RETRAN-3D depressurization, and therefore particular attention Vb based on Wilson correlation a 3.4 was given to the nodalization scheme. The accident Vb = 1.29 m/s b 3.7 scenarios included different boundary conditions for b Vb = 0.61 m/s 3.5 the break sizes and MSIV closure times. The bubble- Vb = 0.15 m/s b 3.3 Vb = 0.03 m/s b 3.3 rise model parameters, which are relevant to the calculated reactor water level swelling during a for separator interior, separator depressurization, were selected based on simulations exterior and dryer skirt b for separator exterior and dryer skirt of the General Electric Level-Swell experiments. Agreement between the RETRAN-3D results and The results of this study show a weak dependency of those of the reference analysis of the BWR SAR the time delay td on the bubble-rise velocity. In this steam line break analysis was reasonable. The context, an important difference is that the RETRAN- differences concerned primarily the delay for the 3D model used for this sensitivity analysis is a mixture level to reach the steam-line nozzle, and also complete plant model. Therefore, the results are also the quality of the two-phase mixture vented through functions of the response of the nodes representing the nozzle. The delay could not be explained with the the RPV lower region, especially the lower bubble-rise velocity alone since flashing in the upper downcomer. During an RPV depressurization as a downcomer was found not to be the predominant result of a main steam line break, void is formed in the factor in determining the level rise, due to the lower downcomer region. This effect is enhanced by relatively low depressurization rate (compared to the the low subcooling for the downcomer for an RPV at General Electric experiments). A conservative, as hot-standby conditions. As a consequence of this void opposed to the present best-estimate, methodology formation, a two-phase mixture flows into the upper could explain the observed discrepancy, along with downcomer, in particular in the dryer skirt, and other factors, such as the initial water level. therefore the downcomer mixture level is displaced upwards. It was found from this analysis that the ACKNOWLEDGMENTS contribution of mixture level rise due to inflow from the lower downcomer is the predominant factor compared This work was partly funded by the Swiss Federal to flashing that occurs in the upper downcomer. This Nuclear Safety Inspectorate (Hauptabteilung fur die is due to the relatively low depressurization rate (-1.5 Sicherheit der Kernaniagen, HSK), and the Swiss bar/s over the initial 3.5 seconds of the transient) Federal Office of Energy (Bundesamt fur Energie, when compared against the analysis of General BFE). Electric experiments (e.g. ~4 bar/s for the base case), in Section 2 above. The results from the Vb study REFERENCES therefore show that the RETRAN-3D results cannot [1] M.P.Paulsen, J.H. McFaden, G.C. Gose, yield a better match with the BWR SAR analysis data C.E. Peterson, P.J. Jensen, J.G. Shatford, solely by selecting a conservative value of the bubble- J.A. McClure, J.L. Westacott, "RETRAN-3D - A rise velocity. Program for Transient Thermal-Hydraulic Analysis of Complex Fluid Flow System; Volume 4 CONCLUDING REMARKS 1: Theory and Numerics", EPRI NP-7450, Revision 4, July 1999. Simulations of the General Electric Level-Swell experiments with the operational-transient code RETRAN-3D have identified important model parameters for accurate predictions of the level swell during depressurization under blowdown conditions: namely, the bubble-rise velocity, the bubble-gradient 39
[2] M.P. Paulsen, J.H. McFaden, P.J. Jensen, [4] Y.A. Hassan, "Transient Two-Phase Blowdown D.L. Johnson, C.E. Peterson, G.C. Goose, Predictions of an Initially Stagnant Saturated J.G. Shatford, and J.L. Westacott, "RETRAN-3D Liquid Steam in a Vessel Using TRAC-PF1", - A Program for Transient Thermal-Hydraulic Nuclear Technology, 69, 388 (1985). Analysis of Complex Fluid Flow System; Volume 4 Assessment Manual", EPRI NP-7450, Revision [5] C.E. Hendrix, "Analysis of the General Electric 5, July 1999. Company Swell Tests with RELAP4/MOD7", Transactions of the American Nuclear Society, [3] P. Saha, J.H. Jo, L. Neymotin, U.S. Rohatgi, Atlanta, Georgia, USA, June 3-7, 1979. G.C. Slovik, C. Yuelys-Miksis, "Independent [6] J.A. Findlay, G.L. Sozzi, "BWR Refill-Reflood Assessment of TRAC-PF1 (Version 7.0), Program - Model Qualification Task Plan", RELAP5/MOD1 (Cycle 14), and TRAC-BD1 (Version 12.0) Codes using Separate-Effects GEAP-24898, NUREG/CR-1899, October 1981. Experiments", NUREG/CR-4359, BNL-NUREG- 51919, August 1985. 41
RETRAN-3D ANALYSIS OF THE OECD/NRC PEACH BOTTOM 2 TURBINE TRIP BENCHMARK
l/l/. Barten, P. Coddington
This paper presents the PSI results on the different Phases of the Peach Bottom BWR Turbine Trip Benchmark using the RETRAN-3D code. In the first part of the paper, the analysis of Phase 1 is presented, in which the system pressure is predicted based on a pre-defined core power distribution. These calculations demonstrate the importance of accurate modelling of the non-equilibrium effects within the steam separator region. In the second part, a selection of the RETRAN-3D results for Phase 2 are given, where the power is predicted using a 3-D core with pre-defined core flow and pressure boundary conditions. A comparison of calculations using the different (Benchmark-specified) boundary conditions illustrates the sensitivity of the power maximum on the various resultant system parameters. In the third part of the paper, the results of the Phase 3 calculation are presented. This phase, which is a combination of the analytical work of Phases 1 and 2, gives good agreement with the measured data. The coupling of the pressure and flow oscillations in the steam line, the mass balance in the core, the (void) reactivity and the core power are all discussed. It is shown that the reactivity effects resulting from the change in the core void can explain the overall behaviour of the transient prior to the reactor scram. The time- dependent, normalized power for different thermal-hydraulic channels in the core is discussed in some detail. Up to the time of reactor scram, the power change was similar in all channels, with differences of the order of only a few percent. The axial shape of the channel powers at the time of maximum (overall) power increased in the core centre (compared with the shape at time zero). These changes occur as a consequence of the relative change in the channel void, which is largest in the region of the onset of boiling, and the influence on the different fuel assemblies of the complex ring pattern of the control rods.
1 INTRODUCTION presented, in which a boundary condition model between the lower and upper plena is used with a 3D- Within the STARS project at PSI, the code core. The results from CORETRAN for Phase 2, environment for the coupled 3-D reactor-kinetics/ which laid the foundations for the 3D core model in thermal-hydraulics transient analyses of the Swiss RETRAN-3D, are presented in [5]. The third part of LWRs is based principally on RETRAN-3D and the paper summarizes the RETRAN-3D results for CORETRAN. (It should be noted that other codes are Phase 3, with the plant system fully coupled to the used for specific applications: e.g. RAMONA for BWR core 3D-neutronics and power calculation. More stability.) Both codes play distinct roles in this details are given in [6]. environment: RETRAN-3D is used for the analysis of coupled 3-D core/plant system transients, while CORETRAN is used for core-only dynamic analysis. 2 PHASE 1: FULL PLANT SYSTEM WITH PRE- An important aspect is that both codes are based on DEFINED POWER an identical neutronics algorithm, allowing the use of 2.1 The RETRAN-3D Base Model CORETRAN as an interface code to help prepare the 3-D core model for RETRAN-3D. This approach forms The RETRAN-3D input model for the Phase 1 the basis of the PSI 3-D transient analysis calculation was developed from an early RETRAN-02 methodology. model for Turbine Trip 1 [7]. From this model, we used the nodalization for the reactor vessel, including the Participation in the OECD/NRC Peach Bottom 2 (PB2) steam separator and downcomer region, the re- Turbine Trip (TT) Benchmark [1] was prompted by the circulation loop and the steam lines. The reactor core following considerations. Firstly, the PSI methodology region was re-nodalized (including the accompanying has so far only been assessed for neutroncially driven heat structures), to include 24 axial control volumes, transients [2], and for a PWR system transient [3]. in combination with a core exit volume and core inlet Since the Benchmark addresses a BWR transient volume connected to the core bypass volume. driven by system thermal-hydraulic perturbations, it extends the range of the codes' assessment. From the Benchmark specifications [1], we used the Secondly, the Benchmark incorporates three different time-dependence and axial profile of the pre-defined phases, which are, from the PSI point of view, well power, together with the feedwater flow, the turbine suited to a comprehensive assessment of all the inlet flow rate, and the turbine bypass area. In participating codes. Consequently, PSI participated in addition, the value of the enthalpy in the steam dome, all three Phases of the Benchmark. the dome pressure, and the flow rates at t = 0 were used. The first part of the paper presents a summary of the Finally, for the base case, the RETRAN-3D code RETRAN-3D results [4] for Phase 1, in which the plant option of the 4-equation model (thermal equilibrium) system is modelled with the reactor power taken from was employed. The Zolotar-Lellouche correlation was the experimental data. In the second part of the paper, used to calculate the slip ratio for the two-phase flow, a selection of the RETRAN-3D results for Phase 2 are and, although the Chexal-Lellouche correlation is that 42 preferred in RETRAN-3D, it is sometimes difficult to A comparison between calculated (base-case) and obtain convergence for plant applications, while for measured steam dome pressure, Fig. 2, shows the pressures considered here (>6.0MPa) the generally, good agreement. The calculation Zolotar-Lellouche correlation has been shown to give reproduces the initial increase in the pressure as the equally good agreement with steady-state void- pressure wave propagates along the steam line from fraction experimental data [8]. the turbine, following the turbine trip and closure of the turbine stop valves. This is followed, between 0.5 s In order to determine the dynamical response of the and 0.9 s by a pressure increase as the power reactor system, it is important to understand the increases, and later by the oscillations due to the general flow of water in the reactor vessel, and how pressure wave in the steam line, with a periodicity of a this is approximated using the RETRAN-3D little below 1 s. The flattening of the curve after about nodalization (Fig. 1). Water is pumped from the jet 3 s, and the beginning of the decrease as a pumps into the lower plenum. From there, the flow is consequence of the power reduction, is also directed upwards into the core region, where the reproduced by the calculation. water is heated and steam is produced. From the core region, the steam/water mixture flows into the upper The increase in the measured pressure at about 1 s, plenum and through the standpipes into the which is due to the closure of the turbine stop valve, steam/water separator. From there, the liquid part of and the power increase is larger than the calculated the mixture flows into the "steam-separator-external" one by about 0.1 MPa, while after 1 s both curves are volume, while the gaseous fraction (steam) flows almost parallel. In order to investigate the possible through the lower dryer into the steam dome, and reasons for the discrepancy in the initial pressure rise, from there through the steam line to the turbine. we looked at two features of the plant representation and physical modelling. This was based on the From the steam-separator-external volume, the liquid knowledge that the system pressure is dependant on flows down into the upper downcomer, while a small the energy (steam) production/conservation in the amount of steam flows up into the lower dryer volume. vessel and the pressure loss along the steam line. From the upper downcomer, where the liquid from the steam-separator-external volume mixes with the When the pressure loss coefficients in the turbine feedwater, subcooled water flows down into the jet bypass line are increased, the steam dome differential pumps. The separation of steam and water in the pressure at times later than 3 s is also increased separator region is important for the system (Fig. 2). Up to about 2 s, however, there is no visible behaviour, and will be discussed in more detail below. effect on the steam dome pressure, and in particular the increased pressure rise at about 1 s is not To steam captured. Consequently, the reason for this increased line Steam dome dryer rise is not in the turbine bypass line. This is because changes in the turbine bypass pressure losses only influence the steam dome pressure after the turbine bypass is almost full open and "steady-state" flow Lower dryer along the steam line has been established. 0.50 .ii i i i i i M i I 11 i i i i i i i I i i i i i i i i i I i i i i i i i i ** 0.40 - Steam sep. ext. < Steam separator
r Feed water p Upper downcome r Standpipes measurement base case re-nod. separator j Lower downc. Upper plenum turbine bypass —i * Recirculation 0.00 lines i 11 i 11 i 11 i i 11 i 11 i 11 i 0 1 2 3 4 5 Jet pumps Core time [sec] region Fig. 2: Steam dome differential pressure versus time; variations of the base case.
Lower plenum Since the pressure increase after 0.5 s is driven primarily by the additional steam production, which results from the power increase, it is important to Fig. 1: Reactor nodalization (not to scale). determine the distribution, including any condensation 43 of this steam as it flows through the steam separators re-nodalization of the separator/downcomer region into the steam dome and upper downcomer regions. was performed, and this is shown to give an increase in the steam dome pressure equal to the measured 2.2 Investigation of Impact of Non-Equilibrium one at about 1 s (Fig. 2). In order to achieve this, the Effects volumes of the steam separator and the steam separator external control volumes had to be reduced, 2.2.1 Nodalization of Steam Separator Region while the volume of the upper downcomer control In order to determine the distribution of the steam volume was increased to preserve the total volume of flow, it is important to follow the flow of steam through water. the separators into the steam dome [9]. In the reactor, The main effect of the re-nodalization was a reduction two-phase flow from the core region passes through in the energy transfer from the gas phase to the liquid the standpipes and enters the steam separators. In phase, because the volume within which thermal each separator, the steam-water mixture passes equilibrium can be established is then smaller. turning vanes, which impart a spin to establish a Consequently, there is less condensation of the vortex, thus separating the water from the steam. The vapour as the pressure increases. To obtain the denser liquid is thrown radially outwards by centrifugal measured pressure increase, it was necessary to force, forming a continuous film on the inside wall of approximately halve the volume of the steam the inner pipe. The separator water exits from under separator and steam separator external control the separator cap and flows out between the volumes. standpipes, draining into the downcomer. Steam, with a quality of at least 90%, exits from the top of the separator and rises to the dryers. The dryers force the wet steam to be directed horizontally through the dryer panels. The steam is forced to make a series of rapid changes in direction while traversing these panels, during which the heavier drops of entrained moisture are forced to the outer walls, where moisture-collection hooks catch and drain the liquid to collection troughs, then through tubes into the vessel downcomer. By this process, the steam dryer assembly increases the quality of the steam to more than 99.9%, and this dry steam then flows into the steam dome and subsequently the steam lines.
This behaviour is approximated in the RETRAN-3D model by control volumes (CV 1 - CV 5 in Fig. 3). A two-phase mixture from the core region flows into the steam separator volume (CV 1). In this volume, the to Lower to Jet RETRAN-3D bubble rise model option is used in order Downcomer Pumps to define a liquid level below which two-phase Fig. 3: Nodalization diagram of the steam conditions are established, whereas above this level separator/downcomer region. there is only steam. The steam flows upwards to the lower dryer volume (CV 2), and continues to the steam dome volume (CV 3) and into the steam line. 2.2.2 Two-Region Non-Equilibrium Model Two-phase flow, consisting of water with some vapour carryunder, flows from the steam separator volume to Another RETRAN-3D code option that can be used to the steam separator external volume (CV 4). The obtain a reduced energy transfer between the liquid junction between these two volumes is below the and the vapour region is the "two-region, non- liquid level in CV 1. In the steam separator external equilibrium model". In this model, RETRAN-3D divides (CV 4), a liquid level is defined, again using the the fluid in a bubble-rise control volume at the liquid RETRAN-3D bubble rise model, with a steam phase level interface into two regions, the two regions being above the liquid level and a two-phase region below. internally in thermal equilibrium, but not necessarily in At t = 0 there is a very small flow of steam from the equilibrium with each other. The "liquid" region below steam separator external to the lower dryer volume the interface, and the "vapour" region above the (CV 2), while the main flow is of water draining into interface, have the same pressure, but in general the upper downcomer (CV 5). have different temperatures; for example, because of superheated steam in the vapour region. The Since the RETRAN-3D code option used in this application of this model to the steam separator analysis is the 4-equation model (i.e. full thermal volume and the steam separator external volume, equilibrium), the volumes of water in the steam CV 1 and CV 4 in Fig. 3, also produced an increased separator (CV 1) and steam separator external (CV 4) pressure rise at about 1 sec (Fig. 4). A sensitivity control volumes are in equilibrium with the steam, and study showed that the effect in the steam separator this will strongly influence the pressure increase per volume is negligible, and that the increased pressure unit of steam generated in the core. For this reason, a rise at 1 s comes mostly from the steam separator 44 external volume. The reason for this is that, in the 2.3 New Version of RETRAN-3D steam separator external volume, the pressure The calculations presented above were performed increase after 0.3 s leads to an increase in the steam using RETRAN-3D MOD003.0 [10]. Recently, temperature in the vapour region but, because of the RETRAN-3D has been updated to MOD003.1 [11], non-equilibrium model, the energy transfer between thereby correcting the area-change pressure drop for the vapour region above the interface and the liquid the case of multiple junctions. The influence of this up- region below the interface is reduced, thus date, for example for the calculation with the two- maintaining the superheated steam in the vapour and region, non-equilibrium model, is shown in Fig. 5, dryer regions. In the steam separator region, however, which confirms that the calculations using MOD003.0 there is a transfer of vapour from the liquid region at and MOD003.1 are nearly identical. the level interface, because of the rising bubbles through the liquid region, and, relative to this, It should be noted that the combination of the two- reduction in the energy transfer across the interface region, non-equilibrium model with the 5-equation due to the non-equilibrium model has only a small model in the same control volume is currently not effect. available in either code version, and the 4-equation 0.50 .i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i model is always used in the code in combination with the two-region non-equilibrium model. Therefore, calculations using the 4-equation and 5-equation 0.40 E- options in the steam separator and steam separator external volumes, together with the two-region non- equilibrium model, always produce the same results (Fig. 5). Such a combination, allowing non-equilibrium between the liquid and the vapour regions in combination with non-equilibrium between the liquid and the vapour phases in the liquid region, would usefully extend the modelling capabilities of the code.
0.50 F
2 3 time [sec] Fig. 4: Steam dome differential pressure versus time: two-region, non-equilibrium, 5-equation model. non-equilibrium+V3.1+4EQ
2.2.3 "5-Equation" Model non-equilibrium+V3.1+5EQ Using the RETRAN-3D "constrained non-equilibrium" non-equilibrium+V3.0+5EQ = or "5-equation" model in the steam separator and steam separator external control volumes also 111111111111111111111111111111111111111111111111 reduces the heat transfer between the vapour and the 0 1 2 3 4 5 liquid phase. Thus, the pressure at about 1 s in the time [sec] steam dome is increased (Fig. 4). This increase, Fig. 5: Steam dome differential pressure versus time: however, is much smaller than for the two-region, a comparison of the 4-equation and 5- non-equilibrium case. The reason for this is that the 5- equation models, and different RETRAN-3D equation model, when used in conjunction with the versions. bubble-rise model, influences only the heat transfer within the regions above and below the liquid level, and does not consider the heat transfer across the 2.4 The Non-Equilibrium Model and Parameter liquid-level interface. The smaller pressure rise of the Variations 5-equation model is also in line with the properties of In the previous Sections, we saw that the two-region the different models. While the homogeneous non-equilibrium model gave improved agreement with equilibrium model for a two-phase mixture constrains the measured pressure in the steam dome. We have all phasic temperatures to the saturation temperature also seen that the agreement at later times can be at the respective pressure, the 5-equation model improved by increasing the pressure losses in the constrains the vapour phase to the saturation turbine bypass line (Fig. 2). That is, when the change temperature only and permits the liquid temperature to in the bypass line is combined with the two-region, float freely. It is clear that, since there is no non-equilibrium model in the steam separator regions, mechanism for sub cooling the liquid in CV 1 and CV the calculated pressure after 3 s is in excellent 4 (Fig. 3), a major impact on the pressure increase agreement with measured data, while there are still through the use of this model is not to be expected. some differences between 1 s and 3 s 45
i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i_ (Fig. 6). However, as described above, the change in the bypass line has almost no effect on the differential pressure in the steam dome for times up to 2 s. j•ys Finally, we investigated a further feature of the steam separator region: the steam separator fluid inertia. Because of the complex flow behaviour in the steam separator volume, i.e. flow in a vortex, there is no unique value for the inertia appropriate for all conditions and transients. To analyse the sensitivity to fluid inertia, we approximately doubled the value in the steam separator inlet, and more than doubled the value in the steam separator exit to the lower dryer (see Fig. 3). These values were chosen based on an early analysis of this transient [7], together with recent discussions [April 2002] in the e-mail forum of the 2 3 Benchmark. The effects of these changes on the time [sec] steam dome differential pressure are relatively small Fig. 7: Steam dome differential pressure versus time: (Fig. 6) for the case considered here, i.e. with a pre- sensitivity to non-equilibrium conditions. defined power.
0.50 F1-1 3 PHASE 2: PRE-DEFINED CORE BOUNDARY CONDITION MODEL r-. 0.40 £ The principal aim of Phase 2 of the Benchmark was to define the 3D core for the Phase 3 calculation. 8 0.30 E- Consequently, flow and/or pressure boundary conditions at the lower and upper plenum were 0.20 r measurement provided in the Benchmark specifications for the I generation of a "core-only" model. Since, in Phase 2, "d non-equilibrium these (calculated) boundary conditions were I 0.10 non-equilibrium + turbine bypass J stipulated by the Benchmark co-ordinators, no non-equilibrium + inertia comparison with measured data was possible. 0.00 base case 3.1 The RETRAN-3D Model
0 2 3 In the PSI methodology, the input required for time [sec] RETRAN-3D to perform a 3D core calculation consists of three separate input files, two of which are Fig. 6: Steam dome differential pressure versus time: prepared by CORETRAN, and the remaining one the two-region, non-equilibrium models and forms part of the normal RETRAN-3D input structure. variations. The two files prepared by CORETRAN consist of a transient cross-section (tcs) file, and a file containing 2.5 Summary of RETRAN-3D Analysis of PB2TT geometric information for all the individual fuel Phase 1 assemblies (cdi or CORETRAN Data Interface file). The additional information contained within the In summary, we saw in the above Sections that the standard RETRAN input includes a "map" which manner in which the heat transfer between the liquid allocates each reactor fuel assembly to a given and vapour phases is treated in the two-phase regions RETRAN core hydraulic channel (a total of 34 such (steam separator and steam-separator external channels were used in the present analysis, see volumes in Fig. 3) had the most influential impact on below). Thus, prior to the RETRAN-3D calculation, a the transient pressure prediction. Two sensitivity CORETRAN analysis [5] of Phase 2 was performed to calculations were subsequently performed: firstly, to provide well-founded parameters for the 3D core. The understand the importance of the volume of liquid with Phase 2 calculation is described briefly below, while which the vapour is brought into equilibrium as the more information is given in the Section describing pressure increases (re-nod separator sensitivity); and Phase 3 of the Benchmark. secondly, the importance of heat transfer across the free surface in the steam separator external volume. If we now consider the flow in the core region, we see Since these two effects relate to two different physical that the flow from the lower plenum is mainly to the phenomena, a final calculation was performed (Fig. 7) core inlet volume, with a small fraction of the total core showing their combined effects. The result is very flow flowing through the core bypass volume (Fig. 8). close agreement between the calculation and From the core inlet volume, most of the coolant flows measurement up to about 2 s, after which (as into the core, while again a small amount enters the discussed above), the modelling of the pressure core bypass volume. The core region is represented losses along turbine bypass line become important. by 34 thermal-hydraulic channels, each with 24 axial 46 nodes, and the flow through each channel (Fig. 9a, dashed line), which completes the corresponds to the combined flow through a certain information needed to calculate the power (Fig. 9c). number of fuel assemblies. The steam/water mixture We can compare this with a second option in which flowing out of the top of the core channels flows into a the pressure in the lower plenum is used as the inlet single core exit volume, and from there into the upper boundary condition. With this option, a slightly smaller plenum, where it mixes with the core bypass flow. inlet flow rate (Fig. 9a, solid line) for the core region (relative to the flow boundary condition case) is calculated. In addition, the core exit flow rate is also To Standpipes reduced. Together, this results in a slightly reduced net flow of water into the core region before the time of the power maximum (Fig. 9b), which in turn means a smaller increase in the (void) reactivity. These slight differences in the flow behaviour lead to a power peak for the pressure boundary condition case about 20% lower than that for the flow boundary condition case.
flow be From Jet pumps pressure be
Fig. 8: Core region nodalization (not to scale).
In the current (Phases 2 and 3) calculations, the neutronics parameters of each of the 764 fuel assemblies have been entered at 24 axial levels. The combining, or "lumping", of the flow through these assemblies into the 34 thermal-hydraulic channels is that suggested in the Benchmark specification (Ref. [1], Fig. 3.2.2), except that for numerical reasons Channel 26 was subdivided into two different channels, with numbers 26 and 34; see Section 4 for more information. The neutronics parameters for the fuel assemblies and control rods were generated in a manner similar to the PSI CORETRAN analyses of Phase 2 of the Benchmark. 3.2 Selected Results In the Benchmark specifications, different options were provided for boundary conditions for the Phase 2 0.6 0.8 1.0 calculations. In all of these options, the upper plenum time [sec] pressure is used as (outlet) boundary condition, and Fig. 9: Flow rates and normalized power of Phase 2 the same lower plenum fluid enthalpy is used. In one calculations around the power peak for two option, the total flow rate through the lower plenum different sets of boundary conditions. into the core region, i.e. the sum of the core and bypass flows, is used as the inlet boundary condition 47
This illustrates the sensitivity of the power peak to maximum, while those relating to pressure losses in small changes in the system parameters; namely, in the turbine bypass line, becoming effective only after the lower plenum flow rate and pressure. The dynamic t = 2 s, were not included. response of the system with close interaction of the The Phase 3 calculation, which is a culmination of the thermal-hydraulic properties with the core power is analytical work of Phases 1 and 2, shows that it is discussed in more detail in the following Sections for possible, using RETRAN-3D, to perform an accurate Phase 3. analysis of a system-driven transient with a 3D core on the basis of the PSI analytical methodology of 4 PHASE 3: FULL PLANT SYSTEM COUPLED preparing the transient neutronics cross-sections and WITH THREE-DIMENSIONAL CORE core geometry within the CORETRAN code. 4.1 The RETRAN-3D Model for Phase 3 The RETRAN-3D input model of the reactor system for Phase 3 of the Benchmark exercise is based on the model for the Phase 1 calculation [4], described above in Section 2. In the Phase 1 calculation, however, the core was represented by a single thermal-hydraulic channel with 24 axial control volumes, and the reactor power was taken from the experimental data, but in Phase 3 a 3D core model is used instead. The region between the lower and upper plena is the same as that in the Phase 2 calculation. 4.2 Comparison with Measurements Q £0 III I I I I I II | I I I I I I I I I | I I I I I I I I I | I I I I I I I I I | I I I I I I I I l_ The results of the RETRAN-3D Phase 3 calculation are presented in this Section. The calculation (Fig. 10, solid lines) was performed using the same model and code options as for Phase 1, including the two-phase, two-region, non-equilibrium model (see Section 2). Results are generally in good agreement with the measured data (Fig. 10, dotted lines). The calculated maximum of the normalized power, Q(t), occurs only slightly earlier, and is slightly smaller, than measurement (Fig. 10a). The thermal-hydraulic behaviour of the reactor, as represented by the steam dome (Fig. 10b) and the core exit relative (Fig. 10c) pressures, are well predicted. The calculation shows, between t = 0.5 s Q JQ II I I I I I I I I | I I I I I I I I I | I I I I I I I I I | I I I I I I I I I | I I I I I I I I l_ and t = 0.9 s, the initial increase in the pressure as the pressure wave propagates along the steam line from the turbine, followed by the pressure rise as the power increases. The calculation also shows the pressure oscillations due to pressure wave propagation back and forth along the steam line, with a periodicity of about 1 s. The flattening of the pressure curve after about 3 s, and the beginning of the pressure decrease as a consequence of the power reduction, is also reproduced by the calculation. After about 1 s the calculated pressures are about 0.05 MPa below the measured ones, and are in very close agreement with the Phase 1 predictions with the pre-defined core power (dashed lines). 0 1 2 3 4 5 time [sec] In the analysis of Phase 1 (Section 2 above and [4]), Fig. 10: Comparison of calculated and measured sensitivity of the model to the original base case was investigated, in particular with respect to non- powers and pressures. equilibrium effects in the steam separator region and to pressure losses in the turbine bypass line. The 4.3 Dynamic Response of the System improvements with respect to the non-equilibrium effects were included in the 3D model, since they In this Section, we analyze the dynamics of the influence the magnitude and timing of the power reactor system during the transient, in particular around the time of maximum power. 48
4.3.1 Before Scram We first discuss the behaviour before t = 0.75 s, i.e. prior to control rod insertion. The transient is initiated at t = 0 when the turbine is tripped and the turbine stop valve starts to close. The flow rate from the steam line to the turbine reduces quickly from the steady-state value of -1000 kg/s to zero at t = 0.096 s when the turbine stop valve is completely closed. Simultaneously, at t = 0.06 s, the turbine bypass valve begins to open, and is full open at t = 0.852 s. The flow through the turbine bypass is choked, and has a 100% capacity of -600 kg/s. The steam line exit flow (Fig. 11a, dashed line), which is the sum of the (rapidly decreasing) flow through the turbine and the (slowly increasing) flow through the turbine bypass, is determined up to t = 0.1 s by the flow through the turbine, and later by that through the turbine bypass. Thus, at t = 0.1 s, there is almost no flow at the steam line exit, but at the same time there is still the full flow rate of -1000 kg/s from the steam dome into the steam line (Fig. 11a, solid line). These two effects generate a net mass flow into the steam line. Consequently, the pressure in the steam line increases, at first close to the exit to the turbine. The pressure rise then propagates as a pressure wave back along the steam line to the steam dome, leading -0.25 s later to a reduction of the steam flow at the steam line inlet. After t = 0.3 s, the flow at the steam line inlet begins to decrease, and eventually falls below zero. Between t = 0.4 s and 0.65 s, there is a reverse flow of steam from the steam line into the steam dome (Fig. 11a, solid line), with a maximum of - 600 kg/s at t = 0.46 s.
The reduced mass flow between the steam dome and the steam line after t = 0.3 s causes the pressure to rise in the steam dome (Fig. 10b). A pressure wave then propagates through the steam separator external volume to the lower plenum (Fig. 1). The pressure rise at the core inlet (Fig. 8), then leads to an increase in the mass flow into the core (Fig. 11b, solid line), with a maximum at t = 0.5 s. In addition, the pressure rise in the steam dome produces a wave, which propagates down through the steam separator to the upper plenum (Fig. 1). The pressure rise at the core exit (Fig. 8) results in a reduced mass flow from the core exit (Fig. 11b, dashed line), with minimum flow occurring at about t = 0.55 s. The response of the flow at the core exit is slower than that at the core inlet because the pressure wave propagates faster through the liquid in the path from the steam separator external to the lower plenum than through the (more compressible) steam/water mixture in the path from the steam separator to the upper plenum. The increased core inlet flow and the decreased core exit flow together produce a net positive mass flow into the core (Fig. 11b, dash-dotted line). The maximum of this net mass flow is about 4000 kg/s at t = 0.55 s, which is more than twice the maximum flow rate reduction at Fig. 11: Flow rates, reactivities and power around the the steam line inlet. time of maximum power. 49
The "squeezing" of the core by the pressure wave below zero. With its further decline, the power is also propagation through the lower and upper plena results reduced. This means that less vapour is being both in an increase in the liquid (mass) content of the produced in the core, which leads to a net inflow of core and an increase in the core pressure. The net water. A continuous net mass flow into the core, mass flow into the core increases the void (or density) however, begins only after t = 1.3s (i.e. beyond the reactivity (Fig. 11c, dashed line). time scale given in Fig. 11). The (mostly) negative values of the core flow balance between t = 0.7 s and The increase in reactivity then leads to a power rise t = 1.3 s occur as a consequence of the power (Fig. 11d, solid line). The enhanced heat generation, increase, which leads to an increase in the heat and consequently the heat transfer to the coolant, transfer into the coolant, and thereby enhanced promotes vaporization of water in the core, which in vaporization and core pressurization. Superimposed turn causes a further increase in the core pressure. on the general behaviour of the core mass balance This balances the pressure at the core inlet and exit (Fig. 11b) are the oscillations which occur as a and reverses the mass flow into the core, so that, after consequence of the pressure wave propagations 0.65 s, there is, for a period of 0.15 s, a net mass flow (back and forth) along the steam line. In addition, the out of the core. The result of this is that the void oscillations of the steam line inlet flow relative to the reactivity peaks at - 0.65 s, and then slowly exit flow also continue, in line with the pressure decreases. The Doppler reactivity effect is small (Fig. oscillations in the steam line after the time frame 11c, dash-dotted line), and acts only as a second- presented in Fig. 11, but with decreasing amplitude. order correction. Thus, the total reactivity p(t) seen in Fig. 11c (solid line) is dominated by the void reactivity, 4.4 Analysis of the Three-Dimensional Core and also peaks at about t = 0.65 s. The maximum in the normalized power, Q(t), follows closely at In the previous Section, we analyzed the reactivity t = 0.68 s (Fig. 11d, solid line). For comparison, the and total power of the core. In this Section, we determine the 3-dimensional power shape by measured power is also included (Fig. 11d, dotted examining the power in different thermal-hydraulic line). Note that the reduction in power begins well channels, concentrating on the time before scram. before the insertion of the control rods at 0.75 s. The power behaviour can be understood in qualitative 4.4.1 Thermal-Hydraulic Channel Radial Map and terms for this fast transient by a "prompt jump Control Rod Map approximation" [12], which predicts the normalized In the Phase 3 calculation of the core behaviour, the power to be combining, or "lumping", of the flow through the different fuel assemblies into the 34 thermal-hydraulic fi(0= p (1) channels is that suggested in Ref. [1], except that fi-p(t) since RETRAN-3D will not combine assembly types The total delayed neutron fraction at the start of the with different numbers of fuel rods within the same transient is p = 0.005526 [1, Table 2.3.1]. The power hydraulic channel, Channel 26 in [1, Fig. 3.2.2] was calculated from the prompt jump approximation subdivided into two separate channels: Number 26 (Fig. 11d, dashed line) rises slightly earlier than the (for the 4 outermost fuel assemblies with Assembly power calculated by RETRAN-3D because it assumes Design 5 [1, Fig. 2.4.2]), and Number 34 (for the 4 innermost fuel assemblies with Assembly Design 4). an immediate response of the power with respect to Figure 12 is a radial map of the thermal-hydraulic the reactivity. The power maximum is also slightly channels; those mentioned above have been earlier than that calculated by RETRAN-3D, and highlighted. The neutronics parameters of the fuel coincides with the time to the reactivity maximum. It is assemblies and control rods are generated in a also smaller, because the increased level of delayed manner similar to the PSI RETRAN-3D and neutrons due to the power rise has not been taken CORETRAN analysis of Phase 2 of the Benchmark, into account. The sensitivity of the power maximum to already described in Section 3. the core inlet flow rate, observed in the Phase 2 calculations, can be seen in the prompt jump The reactor core consisted of a complex pattern of approximation of the power ~([5-p(t)y considering control rods prior to the initiation of the transient (Fig. that the total reactivity approaches the delayed 12), with a significant number of partially inserted neutron fraction. rods. The details of the control rod pattern are given in [1, Fig. 5.2.1]. Some important features of this pattern 4.3.2 After Scram are a bank of 4 control rods inserted to 1/3rd of their The reactor scram is activated at t = 0.63 s, but since length surrounds the inner part of the core. This the control rods have a response time of 0.12 s, region is surrounded by another bank of 8 control control rod insertion begins at t = 0.75 s. (It should be rods, which are almost fully (11/12) inserted. Outside noted that the scram activation at 0.63 s was part of of this is a ring of 24 control rods, which are between the Benchmark specification, and is not connected to 1/4 and 1/2 inserted. Finally, on the outside, there is a any plant signal.) The total reactivity is then further ring of 24 control rods consisting of 8 fully inserted reduced by the control reactivity (Fig. 11c, dash-dot- control rods: 12 control rods with -1/4 insertion, and 4 dot-dash line). At t = 0.85 s, the total reactivity falls with 1/6 insertion. 50
0 0 0 0 0 0 0 0 0 0 0 0 0 0 18 17 17 17 17 17 17 18 0 0 0 0 0 33 14 15 14 15 14 15 15 14 33 0 0 0 0 33 18 16 15 14 15 14 15 14 14 15 16 18 33 0 0 0 33 14 15 13 13 13 11 3 11 13 I' 13 13 15 14 33 0 0 0 0 18 14 15 13 12 sI 13 Li1 M 11 12 11 13iji 12 13 15 14 18 0 0 0 33 33 29 M 11 13 ill 1T3 11 13 iTj fTi 11 11 1 '3 13 11 lTf fTl 13 11 2£ 29 33 0 0 0 18 29 30 27 Mf I11 12 11 7 6 7 6 2 11 12 3 4 27 30 18 0 0 0 0 33 29 30 13 13 27) ff5 7 7 7 7 8 7 7 7 7 2^1ri v 13 13 29 33 0 0 17 29 30 13 2 8 ^I 27 28 22 3H | 7 31 7 6 1 7 6 7 31 7 aI 31 22 28 27s ^ 2 8 30 29 17 0 17 30 29 13 27]ri 3 27 24 2T$hS 3 7 8 3 3 '3 8 7 '2 24 27 1 7 13 29 30 17 0 17 29 30| 27 28 27 28| ; 22 31 22 4 3 3 3 3 4 3 4 22 31 22 28 27 28 27 30 29 17 I s 0 17 30 2?1P ?7 13 13 2l) ^2 22 24 5 3 3 3 3 5 21 24 22 2 2 13 13 2' 9 30 17 0 17 29 30 27 28s §27 32 22 32 s$21 21 19 2 19 4 21 4 21 32 22 32 27 28 27 30 29 17 24 22 2' 0 20 20 20 20 2 2 22 s 7 27 0 17 30 29 27 2 J 9 J J 1 J 24 2 29 30 17 0 17 30 2 9^27 1& 22 22 32 20 20 20 2 20 20 20 32 22 22 9 M 27 29 30 17 0 17 30 2 7 M 2 32 20 20 20 2 20 20 20 32 2s 2 9 2A 2' 9 30 17 0 17 30 29 27 27 27 24 22 22 20 20 20 19 19 J J 19 19 20 20 20 22 22 24 27 27 27 29 30 17 0 17 29 30 27 2' 7 32 22 3f 4 21 2 2 1 % 21 4 2' 2 22 32 2' 27 30 29 17 0 17 30 29s 27 13 13 22 22 22 24 21 3 3 3 21 24 22 22 22 13 13 27 29 30 17 0 17 29 3 7 2 £ 27 25 2 31 22 4 3 3 3 4 22 31 25 27 28 2' 0 29 17 17 30 29 13 27 13 27 24 22 23 3 3 3 7 |22 24 27 13 29 30 17 23j 13i! | 27 0 17 29 30 13 2 7 28 22 3 6 6 7 31 y| fTl 22 28 2/ fl'8 13 30 29 17 0 0 33 29 30 13 13 27 25 7 7 7 7 7 7 25 27 13 13 30 29 33 0 0 0 18 29 30 27 1 11 6 11 12 11 27 30 29 18 0 0 33 33 29 11 11 13 13 11 13 13 11 11 11 11 13 13 1 135I 11 13 11 M 29 33 33 0 0 0 0 18 14 15 13 3 12 11 1! M M l '2 11 12 iTj fl2 13 15 14 18 0 0 0 0 0 33 14 15 13 13 13 13 13 11 13|13 11 13 13 13 13 15 14 33 0 0 0 33 18 16 15 14 I' 4 15 14 lffT4 14 15 I' 14 15 16 18 33 0 0 0 0 33 14 15 14 15 14 15 15 15 15 14 15 15 14 33 0 0 0 0 0 18 17 17 17 17 17 17 17 17 17 17 17 18 0 0 0 0 0 0 0 0 0 0 0
Fig. 12: Thermal-hydraulic channel radial map, with C 1,10, 26, 34 highlighted. The + symbols represent differently inserted control rods.
4.4.2 Channel-Averaged Power The change in the power production as a function of time in the different fuel assemblies is almost homogenous. To illustrate this, the normalized power distributions of the fuel assemblies in Channels 1,10 and 26 are shown in Fig. 13. Channel 1 is representative of those inside the ring of 8 nearly fully- inserted control rods. Channel 10 is typical of the channels outside this ring, and at the time of maximum power it has the highest power density of all the channels in the core. Channel 26 is close to the periphery of the core, and is the furthest away from the region with control rods.
Figure 13a gives the power t = 0 to t = 1.2 s, while Fig. 13b shows the detailed behaviour close to the maximum. The time-dependency of the power in the individual channels is almost identical. The rise of the normalized power in Channel 10 (Fig. 13b, dotted line) is slightly larger than that in Channel 26 (dashed line), which again is larger than that in the centre of the core: Channel 1 (dash-dotted line). However, the differences are only a few percent. time [sec] Fig. 13: Time-dependent channel power. 51
4.4.3 Axial Profiles of Channel Power density are induced by changes in the void reactivity, and the influence of the complex ring pattern of Figure 14a shows the axial distribution of the control rods. In fact, the highest change in local void normalized power density, Qchno(t = 0), at time zero for occurs immediately above the onset of boiling (Fig. Channels 1, 10 and 26. As can be immediately 14d). noticed, the axial profiles are different, both in shape and magnitude. Channels 1 and 10 have their power maxima at about mid-height, with asymmetry around the maxima such that the power increase below the maximum is steeper than the power decrease above it. This is typical behaviour for most of the channels in the core. Channel 26, and some others at the periphery of the core have their power maxima close to the core inlet. The power profile for Channel 26 is typical of those far away from the control rod banks, and in particular from those banks with partially inserted control rods. The power profile in the fuel assemblies in the centre of the core, surrounded by the second, nearly fully inserted, ring of control rods, is of similar shape to Channel 1, with a plateau at the lower end of the core. This reflects the influence of the innermost bank of control rods, which are 1/3 inserted, and surround the centre of the core. Channel 10 shows the typical axial power profile of the channels outside the second ring of control rods, where the power in the lower half of the channel is depressed, but where there is no identifiable signature for the location of the tip of an individual control rod. On the other hand, Channel 10 is surrounded by control rods which are inserted by different amounts: -1/4, 1/3, 1/2.
The overall observation, therefore, is that, because of the complex control rod pattern present in the core prior to the transient initiation, there is a large range of different channel/fuel assembly axial power profiles. At the time of the power maximum, the axial shapes of the channel power densities are similar to those at time zero, though with some small variations (Fig. 14b). On the average, the normalized power density 0, (t ) in Channel 1 is decreased by about 4%, en,no \ max / J and in Channel 26 by about 1.5%, while in Channel 10 it is increased by about 1.5%. Channel 10 has the highest axially-averaged power density, and also the highest node-based power density at the time of the power maximum. The relative decrease in the centre (Channel 1) is due to the higher control rod density. The axial redistribution of the normalized power densities is similar for all channels, where, for example, the relative power density is increased at mid-height, but reduced at the lower and upper ends. axial node This can best be seen in Fig. 14c, where the change Fig. 14: Axial profiles of channel power and void. in normalized power density, QcKnXtmJ-QcKno{t = Q), is shown. The power maximum of Channel 10 is increased by about 9%, Channel 1 by about 3%, while This is produced as a result of the combination of the the axial location of the power maximum of Channel following effects. Firstly, prior to the increase in the 26 is displaced towards mid-height. core power, the increased flow rate of water at the core inlet, and the increase in subcooling as the As noted before, there is no movement of the control pressure increases, moves the boiling boundary rods before the scram. Thus, all these changes, and upwards, while, as a consequence of the in particular the redistribution of the axial power compressibility of the two-phase mixture, the relative 52 change in the void fraction increases as the density into the boundary condition model of Phase 2. increases. As the power increases, additional steam is Comparison of the results obtained with two (pre- produced as a result of the enhanced heating, which defined) sets of pressure and flow boundary of course approximately follows the axial power conditions illustrate the sensitivity of the power profile. This point is discussed in greater detail in [6]. maximum to changes in the system parameters influencing the core inlet and exit flow rates. 5 CONCLUSIONS Finally, in the third part of the paper, the results for The paper describes the PSI RETRAN-3D analyses of Phase 3, using RETRAN-3D with the system model the Peach Bottom BWR Turbine Trip Benchmark, developed in Phase 1 and the 3D core from Phase 2, Phases 1, 2 and 3. The CORETRAN analysis of the are presented. The RETRAN-3D core model used the Benchmark Phase 2 is presented elsewhere [5]. The Benchmark-specified, thermal-hydraulic channel main goal of the Benchmark participation was to "lumping" scheme, for which the flow through the 764 assess the capability of RETRAN-3D for application to fuel assemblies was represented by just 34 channels. the coupled 3-D core/plant system transient, defined The calculated reactor system behaviour for Phase 3 in Phase 3 of the exercise. In preparation for this, is in excellent agreement with that for Phase 1 and analyses of Phase 1, in which the RETRAN-3D plant with the measured data, while there is a small under- system was initialised, and Phase 2, where the 3-D prediction of the core power response. core model was developed, were also performed. The Phase 3 calculation, therefore, which is a In the first part of the paper, the RETRAN-3D analysis culmination of the analytical work of Phases 1 and 2, of Phase 1 is presented. In this Phase, a RETRAN-3D shows that it is possible to perform an accurate plant system model was developed to predict the analysis of a BWR system-driven transient using steam dome pressure using the pre-defined transient RETRAN-3D with a 3-D core using the PSI analytical core power distribution. The results of the calculations methodology of preparing the transient neutronics are in good agreement with the measured data, in cross-sections and the core geometry within the particular the steam dome pressure. In the paper, we CORETRAN code. have demonstrated that the modelling of the steam The results for Phase 3 of the Benchmark are in good separator/downcomer region is crucial if the pressure agreement with the measured data, in particular the increase during the first 2 s is to be correctly power, the steam dome pressure and the core exit predicted. Changes in the turbine bypass line pressure. The following sequence of events in the pressure loss distribution also influence the system transient are all well predicted by the code. The pressure, but only after the first 2 s. closure of the turbine stop valve, causing the In order to demonstrate the importance of the interruption of the steam flow, leads to a pressure modelling of the steam separator region, and to wave in the steam line which propagates back to the highlight the important physical processes we reactor vessel. The pressure wave then propagates analysed and compared the impact of: re-nodalizing down into the reactor vessel, leading to an increase in this region; using the 4-equation and 5-equation the flow from the lower plenum into the core, and a models; using a two-region, non-equilibrium model; reduction in the flow out of the core into the upper and finally of changing the fluid inertia within the plenum. These two effects combine to produce a net steam separator. For this Phase, in which the mass flow into the core, the amplitude of which is experimental power is used as a boundary condition, more than double the reduction in the flow into the the calculation in which the steam steam line. The net mass flow into the core increases separator/downcomer region is re-nodalized, and if a the void reactivity, which subsequently generates a two-region, non-equilibrium model is used, provide the power excursion. The power increase enhances the closest predictions to the measured system heat transfer into the coolant, which increases the parameters (expressed in terms of the steam dome core pressure, and so leads to a mass flow out of the pressure). These results help to show that one of the core. This then reduces the void reactivity, and most important mechanisms controlling the magnitude consequently also the power. After the power peak, of the initial pressure rise, and therefore the power the plant is shut down with a reactor scram. The prediction in Phase 3, is the thermal dis-equilibrium in reactivity effects resulting from the change in the core the separator and dryer region. Using the 5-equation void can therefore explain the overall behaviour of the model, or changing the inertia in the steam separator transient prior to the time of the reactor scram. The region, has a smaller effect. We also show that the Doppler reactivity provides a second-order most recent version of the code, RETRAN-3D contribution to the termination of the power increase. MOD003.1, gives results very close to those obtained using the previous version. Finally, the 3D nature of the core power distribution was investigated by analyzing the power density of The second part of the paper presents a selection of the different thermal-hydraulic channels. It was found the RETRAN-3D results for Phase 2. The analyses of that the time-dependence of the channel-averaged the CORETRAN calculations for this Phase [5,13] power was similar for all the channels, while the provided confidence in the 3D core model, which was differences in the relative channel powers, at the time then transferred to RETRAN-3D. The RETRAN-3D of the power maximum, were within a few percent of calculations show that the 3D core was well integrated each other. The axial distribution of the channel 53 powers at the time of the maximum power, compared [6] W. Barten, P. Coddington, "Peach Bottom BWR to that at time zero, show a relative increase in the Turbine Trip Benchmark Phase 3: Analysis of Full power at the core centre, but by a different amount for Plant System with 3D Neutronics using RETRAN- each channel. The changes in the power shape occur 3D", for Nuclear Mathematical and Computational as a consequence of both the change in the void Sciences: A Century Anew, ANS Topical Meeting profile in each of the different channels, and the in Mathematics and Computations (M&C 2003 influence on the different fuel assemblies of the Conference, held in Gatlinburg, TN, USA, April 6- complex ring-pattern of the partially and fully inserted 11, 2003), accepted. control rods. [7] K. Hornyik, J.A. Naser, "RETRAN Analysis of the Turbine Trip Tests at Peach Bottom Atomic ACKNOWLEDGMENTS Power Station Unit 2 at the End of Cycle 2", EPRI The introduction of the 3D neutronics parameters into Special Report, NP-1076-SR, EPRI, Palo Alto, the RETRAN-3D model by our PSI colleague, Hakim California, USA (1979). Ferroukhi, is gratefully acknowledged. We also wish to [8] D. Meier, P. Coddington, "Evaluation of the Slip thank both him and Garry Gose (CSA/USA) for their Options in RETRAN-3D", Nuclear Technology, implementation of the effective core bypass density 128, 153-168 (1999). correction in the RETRAN-3D code. We have also benefited from discussions with Garry Gose and Andy [9] S.R. Greene, "Realistic Simulation of Severe Olson (Exelon/USA) on the RETRAN-3D model and Accidents in BWRs — Computer Modeling aspects of the experimental data. This work was partly Requirements", Oak Ridge National Laboratory, funded by the Swiss Federal Nuclear Safety NUREG/CR-2940, ORNL/TM-8517, Oak Ridge, Inspectorate (Hauptabteilung fur die Sicherheit der Tennessee, USA (1984). Kernaniagen, HSK) and the Swiss Federal Office of Energy (Bundesamt fur Energie, BFE). [10] C.E. Peterson, J.H. McFadden, M.P. Paulsen, G.C. Gose, J.G. Shatford, "RETRAN-3D — A Program for Transient Thermal-Hydraulic REFERENCES Analysis of Complex Fluid Flow Systems [1] J. Solis, K.N. Ivanov, B. Sarikaya, A.M. Olsson, (Revision 3), Volume 1: Theory and Numerics", K.W. Hunt, "Boiling Water Reactor Turbine Trip and "Volume 3: User's Manual", Computer (TT) Benchmark, Volume I: Final Specifications", Simulation & Analysis, Inc., Idaho Falls, Idaho, NEA/NSC/DOC(2001)1, June 2001. USA (1998). [2] H. Ferroukhi, P. Coddington, "The Analysis of [11] M.P. Paulsen, C.E. Peterson, G.C. Gose, PWR and BWR Reactivity Transients with J.H. McFadden, J.G. Shatford, J.L. Westacott, CORETRAN and RETRAN-3D", Proc. 10th Int. P.P. Cebull, J.Y. Wu, P.J. Jensen, "RETRAN-3D RETRAN Meeting, Jackson, Wyoming, USA — A Program for Transient Thermal-Hydraulic (2001). Analysis of Complex Fluid Flow Systems, Volume 1: Theory and Numerics", and " Volume 3: User's [3] O. Zerkak, "A PWR Main-Steam-Line-Break Manual', Computer Simulation & Analysis, Inc., Accident Analysis with RETRAN-3D", Paul Idaho Falls, Idaho, USA (2001). Scherrer Institut Scientific Report, Volume IV, March 2001. [12] D.L. Hetrick, "Dynamics of Nuclear Reactors", The University of Chicago Press, Chicago, USA [4] W. Barten, P. Coddington, H. Ferroukhi, "Peach (1971). Bottom BWR Turbine Trip Benchmark: PSI Analysis of Exercise 1 using RETRAN-3D", Proc. [13] W. Barten, H. Ferroukhi, P. Coddington, "Peach PHYSOR 2002 Int. Conf. on the New Frontiers of Bottom BWR Turbine Trip Benchmark Analyses Nuclear Technology: Reactor Physics, Safety with RETRAN-3D and CORETRAN", Nuclear and High-Performance Computing, October 7-10, Science and Engineering, submitted for 2002, Seoul, Korea, on CD-ROM (2002). publication (2003). [5] H. Ferroukhi, W. Barten, P. Coddington, "Transient 3-D Neutron Kinetic Analysis with CORETRAN of a Core Thermal-Hydraulic Boundary Condition Model - Peach Bottom 2 Turbine Trip Benchmark Phase 2", Proc. PHYSOR 2002 Int. Conf. on the New Frontiers of Nuclear Technology: Reactor Physics, Safety and High-Performance Computing, October 7-10, 2002, Seoul, Korea, on CD-ROM (2002). 55
REACTIVITY MEASUREMENTS ON BURNT AND REFERENCE FUEL SAMPLES IN LWR-PROTEUS PHASE II
M. Murphy, F. Jatuff, P. Grimm, R. Seiler, A. Luthi, R. van Geemert, R. Brogli, R. Chawla, G. Meier (Kernkraftwerk Gosgen), H.-D. Berger (Framatome ANP)
During the year 2002, the PROTEUS research reactor was used to make a series of reactivity measurements on Pressurised Water Reactor (PWR) burnt fuel samples, and on a series of specially prepared standards. These investigations have been made in two different neutron spectra. In addition, the intrinsic neutron emissions of the burnt fuel samples have been determined.
1 INTRODUCTION Of particular interest is the change in the reactivity of the fuel as a function of burnup and the capability of Present-day and future challenges for the design and modern codes to predict this change. Figure 1 shows operation of nuclear power plants in Switzerland are a typical variation of k-infinity (i.e. the ratio of total being driven by the deregulation of the electricity neutron production to total absorption) with burnup. market and the consequent fall in prices. Advances are being made in the optimisation of fuel elements, resulting in ever more complex and heterogeneous designs. One such advanced Boiling Water Reactor (BWR) element (SVEA96+) was investigated in the LWR-PROTEUS Phase I programme [1]. Another approach to optimising the return on investment is to lengthen the time the fuel remains in the reactor, i.e. to increase burnup. In order to obtain higher burnup, the initial 235U- enrichment of the fuel is increased. In fact, the current 0 10 20 30 40 50 60 70 80 90 100 tendency of fuel management and core design is to Burnup [GWd/t] increase the initial 235U enrichment from about 4% to more than 4.5%, and the batch discharge burnup level Fig. 1: Typical Variation of k-infinity with Burnup. is now typically about 50 GWd/t (instead of 30-40 GWd/t). 2 LWR-PROTEUS PHASE II Running fuel to higher burnup requires accurate techniques, both for fuel assembly inspection, and for The ongoing LWR-PROTEUS Phase II programme the analytical prediction of the operational [2], a continuing collaboration between the Swiss characteristics of LWR cores containing the highly nuclear power plants and PSI, is the framework for the burnt fuel. From the physics viewpoint, this implies an study of six uranium oxide (U02) and three mixed accurate knowledge of isotopic compositions. Fuel oxide (MOX) burnt fuel samples. These were all rods with average burnup values beyond 50 GWd/t fabricated by the same fuel vendor (Framatome ANP) are characterised by relatively large amounts of and discharged from the same Swiss PWR fission products, and by a high abundance of major (Kernkraftwerk Gosgen). and minor actinides in a degraded U02 matrix. Moreover, the composition of actinides and fission A list of the fuel rods under investigation is given in products becomes ever more complex at increasing Table 1. The wide range of burnup levels under burnup levels. The most direct consequence of the investigation (up to 85 GWd/t in the case of U02 and increased complexity at high burnup is the increase in 60 GWd/t in the case of MOX) will be extended in the discrepancies between predictions from different 2003 to 100 GWd/t and 70 GWd/t, respectively, and codes. two BWR samples will also be added to the list. A key task in seeking to improve calculational Table 1: Selected Burnt Fuel Rods in Phase II. methods is the acquisition of precise experimental Fuel Sample Discharged knowledge of the physical and chemical behaviour of Type Local Burnup high-burnup fuel in relation to a wide spectrum of uo2 39.1 GWd/ton June 1996 needs. Short and medium term interests cover issues uo2 53.5 GWd/ton June 1995 such as core-follow and core performance, shutdown uo2 74.4 GWd/ton June 1995 worth and reactivity coefficients, and criticality safety U02 74.5 GWd/ton June 1995 of pool storage and transport casks. Longer-term U02 83.5 GWd/ton June 1995 considerations include topics such as resource U0 85.3 GWd/ton June 1995 utilisation, radiotoxicity reduction, waste disposal, and 2 MOX 22.3 GWd/ton June 1998 non-proliferation, which will all play key roles in a MOX 41.4 GWd/ton more sustainable utilisation of fission nuclear energy. July 1999 MOX 60.0 GWd/ton July 2000 56
The experimental investigations cover the following A guide tube in the centre of the PWR region three independent, complementary approaches: facilitates the insertion of test samples into the core centre. The driven nature of the reactor configuration (i) Post-irradiation examination in the PSI Hot Lab enables a range of representative neutron-spectrum (Labor fur Werkstoff Verhalten) conditions to be investigated (standard spectrum with for the characterisation of fuel rod elongation and full-density H 0, and hardened spectra with mixtures deformation, metallographic analyses, fission gas 2 of H 0 and D 0, or boric acid and H 0). release measurement, cladding inspection, oxide 2 2 2 thickness determination and axial gamma scans; 3.1 Transfer of Overcanned Burnt Samples (ii) Chemical assays (in the PSI Hot Lab) Each of the burnt fuel samples has been overcanned to quantify the isotopic composition of the nine in the Hot Lab with a special Zircaloy cladding, which most abundant major actinides, eight minor was welded using a certified procedure to guarantee actinides and 33 fission products — the most leak tightness and freedom from contamination. A important from neutronics, toxicity and radioactive combined transport flask and sample changer (Fig. 3) source-term viewpoints; and was used for the simultaneous transfer of up to four burnt fuel samples from the Hot Lab for their (iii) Reactivity measurements (in PROTEUS) subsequent measurement in PROTEUS. The sample to compare the reactivity effects of the burnt fuel changer has a control unit that also acts as a data samples with well-characterised fresh fuel samples logger during the measurements. (some for calibration and others specially prepared with individual additives). The transport flask remains in place above the reactor throughout the reactivity measurements, and serves The present paper reports principally on the first to move the samples in and out of the reactor during series of reactivity effect investigations carried out the experiments. with the full range of samples.
3 INVESTIGATIONS OF REACTIVITY EFFECTS The PROTEUS facility was modified during 2001 to include a 160 mm x 160 mm central PWR test zone of fresh, full-length fuel rods of 4.3w% 235U-enrichment. This fuel was supplied by Kernkraftwerk Gosgen, with support from their current fuel supplier Framatome ANP. Analogously, Kernkraftwerk Leibstadt and Westinghouse Atom have provided the BWR fuel assemblies placed in the outer 450 mm x 450 mm region of the test tank. In the spring of 2002, the fuel in the outer region was changed from SVEA96+ to SVEA96 OPTIMA2. The fuel configuration is illustrated in Fig. 2.
Fig. 3: The Transport Flask/Sample Changer.
3.2 Burnt Sample Reactivity Measurements Fig. 2: A view of the Fuel Configuration for LWR- PROTEUS Phase II showing the Central PWR The reactivity difference between burnt and fresh fuel, Region Surrounded by OPTIMA2 BWR and its variation with burnup, is deduced from the Assemblies. effect on the neutron balance of the PROTEUS 57
reactor by introducing 40 cm long test samples, one at In the stable period method, the control rods remain a time, into the core centre. In neutronically driven fixed whilst a sample is inserted, the resulting change systems such as PROTEUS, it is useful to quote in reactivity causing the reactor power to diverge from relative reactivity worths, i.e. the differences in its steady value. After a short time, the divergence has reactivities between the burnt samples and a a simple exponential form. The time for the reactor reference sample, normalised by the difference power to increase, or decrease, by a factor of 'e' is between the same reference and a further well known known as the reactor period. The higher the worth of reference. In the current measurements, the two the inserted reactivity, the shorter is the stable period, references are 3.5w% U02 and natural U02. For small and the more rapid is the change in power. Using the reactivities (a few cents), such ratios in an data logger, count-rates are recorded from six neutron appropriately designed driven configuration are the detectors located at different positions in the reactor. same as they would be in the corresponding, hypothetical, single-zone system. (The reactivity unit Figure 5 shows the change in power recorded during of a cent is one hundredth of a dollar, and a dollar is a typical set of stable period measurements. By equivalent to the delayed neutron fraction, i.e. about analysing the recorded exponential flux evolution, the 0.007). reactivity of a sample in the PROTEUS reactor can be determined on an absolute basis. The results of such Two methods of reactivity measurement are used: an analysis are shown in Fig. 6. Stable period compensation and stable period. Each derives a measurements are usually less precise than the reactivity value by analysing the change in the reactor compensation method, but they are completely state when a sample is inserted, but each in a independent and have the advantage of providing different way. For the compensation method, the absolute values of reactivity. reactor control rods are used to keep the reactor power constant by compensating for the reactivity change caused by the introduction of a sample. In particular, an automatic fine control rod is used (known as the Autorod). This rod is controlled by a servo feedback system, which compensates automatically for small reactivity changes. The positions of all of the control rods are recorded every second by a data logger that is part of the sample changer control system.
Figure 4 shows the movements of the automatic Time (seconds) control rod during a typical measurement sequence. A reference sample (usually 3.5w% U02) is inserted Fig. 5: Flux Variation During a Period Measurement. first, followed by the first burnt sample to be measured, and then the reference sample again, and so on; the sequence is usually repeated three times. 0.02 After corrections for slight non-linearities, the 0.015 movements of the control rods are indicative of the 0.01 -IW reactivities of the test samples. The difference in § 0.005 reactivity between the burnt sample and the reference 5> o -^p^pPSfl1 ! —$ 1 is found by time-wise interpolation. This method of s o -0.005 » a> \ ill analysis has the advantage of cancelling any long- CC i i term drift in the state of the reactor. To put the -0.01 measured reactivities on an absolute basis, the -0.015 compensation method has to be calibrated using the 0 1000 2000 3000 4000 5000 6000 stable period method. Time (seconds) Fig. 6: Reactivity Values Derived from a Period 1000 Measurement. 900 800 1 700 3.3 Comparisons with Calculations 600 1 500 Figure 7 shows an example of a comparison of 400 measured and calculated reactivity worths (reactivity 300 ri r "1 H effect in PROTEUS of replacing a fresh, reference 200 w 1 100 1 sample at its centre by a burnt fuel sample), 0 expressed in each case relative to a well-known 1000 2000 3000 4000 5000 6000 reactivity difference (between the fresh 3.5% Time (seconds) reference and a fresh natural U02 sample). The Fig. 4: Autorod Movement During a Compensation Measurement. 58 calculations were performed using a simplified whole- So far, detailed reactivity measurements of the special reactor BOXER [3] model for PROTEUS. It is seen samples have been made with the water/heavy water that discrepancies are of the order of 6% higher for mixture as moderator. In 2003, this will be extended to MOX than for U02, and in both cases diverging further the other moderation conditions. The measurements from unity with increasing burnup. indicate that, in most cases, there is a linear relationship with increasing amount of additive, but 1.00 that the absolute values do indeed need to be ~~1 calibrated. Disappointingly, the caesium and boron 0.98 additives appear to have almost completely 1 0.96 • disappeared during the sintering process. A chemical LU 1 analysis of representative samples will be made in the 0.94 Hot Lab as part of the ongoing analysis of the burnt 0.92 + 1 samples. In addition, investigations have been made r-. by neutron radiography in the SINQ facility [4] at PSI. 0.90 MB. • U02 J r^f -PMOX) 0.88 4 NEUTRON EMISSIONS 20 40 60 80 100 Burnup GWd/t Determination of the intrinsic neutron source of each of the burnt samples has been achieved by inserting Fig. 7: Relative Reactivity Worths of Sample them into PROTEUS, one-by-one, with the reactor in a slightly sub-critical state. In this condition, the Replacement in LWR-PROTEUS D20/H20 mixture: Calculational Results Using a neutrons emitted from the samples into the multiplying BOXER Whole-Reactor Model, Divided by reactor environment result in a suitably significant Experimental (E) Values. count rate on the reactor instrumentation neutron channels. Figure 8 shows these preliminary determinations of neutron source strengths, which 3.4 Special Samples Reactivity Measurements increase dramatically with burnup. The immediate Special samples were manufactured by Westinghouse application of such measurements is the accurate Atom in order to provide reference reactivity worths estimation of neutron sources for the calculation of and thus added value to the series of measurements radiation shielding. In fact, the total dose rate at the with burnt samples. Additives corresponding to certain outer surface of spent fuel casks is usually governed important fission products were blended with uranium by the neutron dose rate, and thus a certified cask oxide and then sintered to produce pellets. A list of the design for <50 GWd/t fuel may very well become samples is given in Table 2. unusable if burnup is extended. The neutron source is mainly related to the spontaneous fission of 244Cm Table 2: Special Samples. (especially for fuel with a relatively long cooling time). However, other spontaneous fission and (a,n) sources ID Nuclide Additive Nominal No. of are also present in the composite sample. Mass of pellets additive g 10.00 Nd-1 Nd-143 Nd203 5.957 45
Rh-1 Rh-103 Rh203 2 55 Rh-2 Rh-103 Rh 0 5 56 2 3 ,2 1-00
Gd-1 Gd-155 Gd203 0.048 67
Gd-2 Gd-155 Gd203 0.0903 65 0.10 Sm-1 Sm-149 Sm203 0.0151 68 3 <1) Sm-2 Sm-149 Sm203 0.0248 64 -•-MOX
Sm-3 Sm-149 Sm203 0.0495 67 0.01 —•—U02
B-1 B-10 B4C 0.04 87 20 40 60 80 100
B-2 B-10 B4C 0.26 91 Burnup GWd/t
Cs-1 Cs-133 Cs02 8.7 66 Fig. 8: Relative Neutron Source Strengths of Different LWR-PROTEUS Phase II Samples. Cs-2 Cs-133 Cs02 17.5 71
Cs-3 Cs-133 Cs02 35 66 5 FUTURE DEVELOPMENTS FP- Nd-143 Nd203 1.05 65 50 Rh-103 Rh203 0.7 In 2003, the scope of LWR-PROTEUS Phase II will be Sm-149 Sm203 0.0035 expanded in the following directions. Two additional Cs-133 Cs0 1.3 2 KKG samples will be prepared with new, record 59
burnup values (9-cycle U02 with >100 GWd/t, and 4- We are particularly grateful to H. Achermann (EGL), cycle MOX with >70 GWd/t). Two KKL samples (45 H. Fuchs (ATEL), G. Straub (KKM) and W. Kroger and 70 GWd/t) will also be prepared for analysis. A (PSI) for their strong support of the experiments. Also new moderation condition, using borated water to give thanked are G. Bart, Z. Kopajtic, B. Wernli, an additional neutron spectrum environment, will be I. Guenther-Leopold and S.-G. Jahn for their important implemented in the central PWR test region of contributions to the planning work, as well as PROTEUS. D. Kuster, R. Schwarz and H. Schweikert for the preparation of samples. Thanks are also due to 6 CONCLUSIONS P. Bourquin, M. Fassbind, M. Zimmer-mann and M. Berweger for excellent reactor operation and In the LWR-PROTEUS Phase II programme during maintenance during 2002, to A. Meister for important 2002, the reactivity worths of nine PWR burnt fuel rod analytical work, as well as to J. Ulrich, J. Kohout and segments were measured, in two different neutron K. Kohlik of PSI's Logistics and Marketing Department spectra. The burnups range from 39 GWd/t to (LOG) for their valuable engineering support. 85 GWd/t for the U02 samples, and from 22 GWd/t to 60 GWd/t for the MOX samples. On completion of the REFERENCES LWR-PROTEUS Phase II measurements, a unique set of experimental data on the physics properties of [1] M. F. Murphy et al., "Neutronics Investigations for high-burnup fuel will have been obtained on samples the Lower Part of a Westinghouse SVEA-96+ with record burnups, accurately determined reactivity Assembly", Nuc. Sci. Eng., 141, 32-45 (2002). worths and chemical compositions, and accurate [2] A. Meister, P. Grimm, R. Chawla, "Neutronics irradiation histories. This will serve to support future Design of LWR-PROTEUS Phase II High-Burnup burnup increases, on the one hand, and provide the Fuel Reactivity Measurements", Trans. Am. Nucl. necessary justification to reduce the current level of Soc., 85, 268-269 (2001). conservatism in fuel element and core designs as well as in fuel storage and transport planning, on the other [3] J.-M. Paratte, P. Grimm, J.-M. Hollard, "ELCOS: hand. The PSI Code System for LWR Core Analysis, Part II: User's Manual for the Fuel Assembly ACKNOWLEDGMENTS Code BOXER", PSI Report 96-08 (1996). The LWR-PROTEUS programme is being conducted [4] S. Baechler et al., "The New Cold Neutron jointly by PSI and the Swiss Nuclear Power Plants, Tomography Set-Up at SINQ", Nucl. Instr. Meth. with specific contributions from Elektrizitats- Phys. Res. A481, 397 (2002). Gesellschaft Laufenburg (EGL) and Aare-Tessin AG fur Elektrizitat (ATEL). 61
EXPERIMENTAL INVESTIGATION IN THE PANDA FACILITY OF THE IMPACT OF A HYDROGEN RELEASE ON PASSIVE CONTAINMENT COOLING SYSTEMS
O. Auban, D. Paladino, P. Candreia, M. Huggenberger, H.J. Strassberger
The large-scale, thermal-hydraulics PANDA facility has been used in the past for investigating passive decay-heat removal systems and related containment phenomena for a number of next-generation Light Water Reactors. Passive Containment Condenser (PCC) systems operate by transferring heat via steam condensation from inside the containment to outside, and serve to mitigate pressure build-up in the event of steam discharge from the primary circuit. Five new integral tests have recently been performed in the context of the 5th European Framework Program project TEMPEST. One main objective was to assess the influence of a light gas (hydrogen) on the performance of the Passive Containment Cooling System (PCCS). Hydrogen release in the case of a severe accident is simulated in PANDA by injecting helium with steam into the Drywell. In addition, the impact of a new accident-mitigating design feature, the Drywell Gas Re-Circulation System (DGRS), on the long-term containment behaviour was tested. Another important objective was also to provide relevant data for the validation of modern — mainly Computational Fluid Dynamics (CFD) — codes. The paper reports main observations from two of these new integral tests in which standard PANDA instrumentation has provided important data concerning system response when helium is released in the course of the transient. The results show that some important stratification phenomena have occurred, as a result of the buoyant flow generated by the helium injection. The resulting temperature and concentration measurements show how helium is being distributed in the Drywell (DW) volume during the helium injection phase, and how helium is retained or vented out of these vessels once the injection stops. The paper includes some discussion concerning the influence of gas mixing and stratification and the effect of the DGRS on PCC performance and system pressure build-up.
1 INTRODUCTION AND BACKGROUND eventually non-condensibles are vented (the ends of the PCC vent lines are submerged in the suppression A key feature of next-generation Advanced Light pool) to the Suppression Chamber (SC). Peak system Water Reactors (ALWRs) is their utilization of simple pressures and temperatures are key parameters passive cooling systems [1]. These systems, which characterizing containment integrity during the long- operate following a (postulated) breach of the primary term decay heat removal phase: the overall system circuit, remove decay heat from the core, and mitigate pressure is principally governed by the performance of the effects of pressure build-up in the containment. the PCCs, and by the amount of non-condensible The Passive Containment Cooling System (PCCS) gases initially filling the containment compartments, or designed for the European Simplified Boiling Water released later in the course of the transient. Reactor (ESBWR) is one such system, transferring heat from inside the containment to external water From the case in which hydrogen is released from the pools. reactor core during the accident scenario, previous studies have demonstrated that its contribution to Component tests and integral system tests in scaled system pressure build-up is significant, due to facilities are required for improving the design of these degradation of PCC performance [2]. If the hydrogen safety systems, and for their certification. is quickly mixed with steam, conveyed to the PCC, Experimental programmes are also needed to gain and finally vented to the SC, a significant increase in better understanding of the significant physical pressure is observed right after hydrogen release has phenomena which occur (natural circulation, fluid started. Eventually, like in any Boiling Water Reactor mixing, stratification, etc.) on the basis of which these containment, if part of the hydrogen remains passive systems function. The experimental results somewhere in the DW, e.g. as a result of stable gas will also provide data for the validation of codes or stratification, it does not contribute to the overall models that could be used with reasonable confidence system pressure increase. to analyze real nuclear installations, ranging from normal operating conditions up to (postulated) severe- Within the context of the EU 5th Framework Project accident scenarios. TEMPEST, the performance of the PCCS in the presence of non-condensible gases (i.e. air or The purpose of the PCCS is to remove from the hydrogen) has been investigated experimentally in the containment the energy released in the reactor core large-scale test facility PANDA [3]. In these tests, under accidents conditions in order to guarantee hydrogen is simulated by helium. In addition, the containment integrity. Passive Containment effect of a new accident-mitigating design feature — Condensers (PCCs) condense the steam out of the the Drywell Gas Re-circulation System (DGRS) — on gas mixture fed to them from the containment long-term containment behaviour was also assessed. atmosphere, and the condensate is drained back to The system is based on a vent fan that allows the the Reactor Pressure Vessel (RPV), thereby non-condensible gases coming from the PCCs to be preserving the water inventory. Insufficient energy fed back to the DW, (instead of discharging them to removal leads to Drywell (DW) pressure increase and 62 the Wetwell (WW) gas space [4]), in order to mitigate indicated in Fig. 2. In these new tests, it was possible the build-up of system pressure. to measure temperatures with an accuracy of ±1°C.
The main objectives of these tests were: The main upgrade to the PANDA instrumentation performed for the TEMPEST project has been the to assess PCCS performance in the presence of addition of a Mass Spectrometer (MS) that permitted hydrogen; air, helium and steam concentrations to be measured to assess the combined PCCS and DGRS [11]. Gas was continuously sampled (through capillary performance in the presence of hydrogen; tubes) at different locations in the facility, enabling gas molar fractions to be monitored continuously. At each to investigate the ability of the DGRS to retain location, one new measurement (consisting of three some significant amount of non-condensible gas molar fraction values for the three gas species: air, in the DW, and thereby to reduce long-term steam and helium) is taken every 540s, with an containment pressure build-up; and accuracy of ±5% (corresponding to ±0.05 absolute to provide a database to assess the capability of error in the molar fraction). In each DW vessel, advanced containment and CFD codes to predict concentration measurements are performed at nine containment behaviour under such accident different locations, at three levels (at Levels 2, 6, 10 conditions. and radial locations A, B, C in Fig. 2). Concentrations were also measured at three elevations in the centre In the following, the PANDA test facility and the of the DW Interconnecting Pipe (IP). It should be different test scenarios will first be outlined, but then noted that all the gas sampling locations for the paper will focus on the two tests (among the five concentration measurement are situated quite close carried out in the frame of the TEMPEST programme) (typically a few millimetres) to the corresponding for which a large amount of non-condensible gas was thermocouple locations. released in the course of the transient. Observations concerning gas mixing and stratification, particularly in the DW volume, will be reported. A later section will deal with PCC performance in more detail, and finally the influence of DGRS activation on system response PCC Pool will be discussed by comparing results from the different tests.
2 PANDA FACILITY AND TEST SCENARIOS 2.1 PANDA Test Facility PANDA is a large-scale, thermal-hydraulic test facility consisting of large inter-connected vessels and open water pools (Fig. 1), and is used for investigating containment system behaviour and phenomena for different ALWR designs [5, 6]. Heights are scaled approximately 1:1, and decay power and volumes 1:40, with respect to the ESBWR. Detailed information concerning the PANDA arrangement, and its scaling with respect to an ESBWR containment, have already Main Vent2 been published [7, 8]. Wetwell 2 press. The installation of the DGRS was the main system M |Eq.-Line , modification to the PANDA facility for this new test series (compared with earlier the ESBWR Suppression • Suppression I configuration); full details have been given in a Pool 1 Pool 2 Electn previous paper [4]. In short, the DGRS consists of a Heaters vent fan and system of pipelines that interconnects the three PCC vent lines with the Interconnecting Pipe (IP) between the Drywells. Fig. 1: Schematic of the PANDA facility configuration PANDA instrumentation consists of numerous sensors used for the TEMPEST test series (No DGRS for the measurement of fluid and wall temperatures, activation means that valves V0, V1s V2 and V3 total and air partial pressures, flowrates, phase are all closed). indications, valve states and heater power [9, 10]. To enable a detailed analysis of the phenomena occurring in the DW vessels to be carried out, twelve temperature measurements levels have been defined. The actual locations of the fluid temperature sensors (thermocouples) installed in the two Drywells are 63
ii) Helium was supplied (12.9 g.s") to the two Main DW1 DW2 Steam Lines from approximately 10000s after test
IMTG.D1A.1 h initiation for a period of two hours; |MTG.D1*.2 h iii) All three Passive Containment Condensers (PCCs) |M I G.U1 A.3 \- - IMIG.U1A.4 |- - were connected to the DW vessels, and the PCC IMIG.D1A.5 h - pools (Fig. 1) were initially filled with water at |M I G.U1 ".6 |- - saturation temperature; IMIG.D1A.7 h |M I G.D1 A.6 |- iv) In Test T2.1, the DGRS was activated exactly at IMTG.D1A.9 j- the time helium injection begins, and continued at a |M I G.U1 MO |- - 1 |MT(5.D1 A.11|— well-controlled, constant flow rate of 30 l.s" until the |MTG.D1A.12|- end of the 12 hour transient.
Plan view
HELIUM, NO DGRS (T1.1)
HELIUM, DGRS (T2.1)
NO HELIUM, DGRS (T2.2)
|MTG.D1A.1..TT IMTG.D2A.1...121
• TC location in DW: MTG.Dx*.n x: DW number (1 or 2) TC Tolerances in: *: radial location (A, B or C) Elevation: +/- 30 mm n: measurement level (1 ..12) Plan view: +/- 50 mm
O TC location in connecting line: MTG.TDD.i i: level (1 ...5) 0 5000 10000 15000 20000 25000 30000 35000 40000 45000 Time (s) Fig. 2: Gas temperature measurement locations in Fig. 3: DW and WW pressures for Tests T1.1 ,T2.1 and the Drywells. T2.2.
2.2 Test Scenarios The tests presented in this report simulate the long- term cooling phase following a postulated Loss Of Coolant Accident (LOCA) caused by a Main Steam Line Break (MSLB) in the ESBWR. The main objectives were to demonstrate the performance of MTG.D2A.1 MTG.D2A.2 the PCCS, and the ability of the DGRS to retain non- MTG.D2A.3 MTG.D2A.4 condensable gas in the DW, thereby reducing long- MTG.D2A.5 MTG.D2A.6 MTG.D2A.7 term containment pressure build up. The tests start MTG.D2A.8 MTG.D2A.9 with conditions predicted for the ESBWR in the RPV MTG.D2A.10 MTG.D2A.11 and the containment one hour after MSLB scram. MTG.D2A.12 The following two tests for investigating the effect of hydrogen (simulated by helium in PANDA) on PCC 5000 10000 15000 20000 25000 30000 35000 40000 45000 performance and long-term containment response will Time (s) be discussed and compared: (a) Test T 1.1 Test T1.1: Helium release but no DGRS activation; TestT2.1: Helium release with DGRS activation; In addition, a third test will be outlined, serving as a reference test: til -
t jf 1/ MTG.D2A.1 Test T2.2: No Helium release, and DGRS activation « """ MTG.D2A.2 C3 MTG.D2A.3 late in the transient. Q MTG.D2A.4 1 c MTG.D2A.5 J t $ MTG.D2A.6 1 In this test series, PANDA was configured as follows: MTG.D2A.7 W - i ! if -o MTG.D2A.8 MTG.D2A.9 i 7 = i) One half of the break flow from the RPV was JF MTG.D2A.10 f MTG.D2A.11 directed to DW1, which feeds one PCC (PCC1), and r MTG.D2A.12 - the other half was directed to DW2, feeding two X condensers (PCC2 and PCC3). This was achieved 0 5000 10000 15000 20000 25000 30000 35000 40000 45000 Time (s) with two separate blowdown lines (Main Steam Lines (b); Test T2.1 with essentially equal flow resistances) from the RPV to each of the two DW vessels; Fig. 4: DW2 temperatures measured along a vertical axis (radial location A). 64
The nominal initial conditions and decay-heat curves (Fig. 7). The differences in the recorded DW are similar for all these tests, and correspond to those temperatures for the two tests are an important issue, predicted for the ESBWR one hour after the scram and will be discussed in detail in the following triggered by the LOCA. The total initial DW pressure is sections. approximately 2.5 bar, which includes some residual Third Phase: following helium injection air (0.05 bar air partial pressure). It should also be noted that the DW to SC pressure difference is As can be seen in Fig. 3, the system pressure allowed to vary over a small range only: on the lower continues to increase for some time (approximately side, the pressure difference is limited by the VB 3000s) after the end of helium injection. In Test T2.1, opening settings, while the higher value is essentially the effect of the DGRS is to mitigate pressure build- determined by the PCC feed-line pressure drops and up, the highest measured pressure in the DW is the PCC vent line submergence depth. approximately 5.7 bar, compared with 6.0 bar for Test The initial conditions, the decay-heat power curve T1.1. Around t=20000s, stabilization of the system pressure occurs in both tests. In Test T1.1, the during the test, and the overall PANDA configuration transition between the pressure-increasing phase and are similar in all these tests, which makes a direct the moment at which pressure starts (slowly) comparison between the measurements meaningful decreasing appears to be much more abrupt than for and straightforward; detailed information on the test Test T 1.2, for which the DGRS was active. In Test matrix and initial conditions has already been T2.1, the pressure keeps increasing, but at a slower published [4]. rate, until t=23500s, and then a continuous decrease in pressure ensues until Vacuum Breaker opening 3 OVERVIEW OF THE TESTS occurs around t=30000s. Three main phases are identified to characterize the Around t=20000s, PCC performance is quickly and course of events during the tests. significantly improving in both tests (as noted by the fact that the condensate flow rate has recovered to a First Phase: before helium injection level comparable with that measured before helium At the beginning of this phase, a steam/air mixture is injection), and a new kind of "equilibrium" has now vented from the DW to the WW during the first 20 been established. Nevertheless, this event does not minutes, because the PCCs were not able to correspond to the same conditions in the two tests. condense all the steam supplied to them Later, i.e. The observations detailed in the following section will within the first hour of the test, all the residual air in actually demonstrate that in Test T1.1 all the helium the DW was vented to the condensers and the WW had been vented out of the DW, but in Test T2.1 a gas space through the PCC vent lines. This venting significant amount of helium had been permanently process explains the increase of WW pressure at the retained in the DW. very beginning of the test (Fig. 3). Since the WW pressure governs the entire system pressure, the 4 OBSERVATIONS CONCERNING MIXING pressures in the DW and RPV increase at the same AND STRATIFICATION time. As soon as this venting process is completed, The containment pressure is principally governed by however, the WW pressure remains nearly constant the performance of the PCCs, and by the amount (and until the beginning of the helium injection phase — the transient distribution) of the non-condensible gases RPV and DW pressures then start decreasing slowly. which initially filled the containment compartments The variations in the three PCC pool levels shows that (air), or were released later in the transient from the all three PCC units have equally shared the heat load reactor core. generated in the RPV. Of course, during this phase, Containment temperatures are ultimately determined the three tests (T1.1, T2.1 and T2.2) exhibit very by the gas distribution. These new tests, for which similar system behaviour. additional instrumentation has been installed in the Second Phase: during helium injection PANDA facility, provide more refined information concerning gas mixing and stratification phenomena As soon as helium is injected, the PCC and system than was possible previously. For example, in Test behaviour changes significantly, the helium being T1.1, the helium mixes immediately with the steam in responsible for a continuous increase in RPV and DW the DW from the start of the injection. pressures. A similar increase in WW pressure is also All the concentration measurements in the DW observed, because of the almost continuous venting indicate that no vertical stratification has occurred from DW to WW through the PCC vent lines. Figures during the injection phase: a 17% helium-rich mixture 4a and 4b show increases in DW temperature in is seen from 12000s after the beginning of the accord with the increases in saturation temperatures. injection until the end of the phase (Fig. 5a). The During this phase, PCC performance is degraded: corresponding vertical temperature profiles at 11000s, condensate flow returning to the RPV is clearly and 14000s and 17000s (Fig. 6a) also show that there is a suddenly reduced (from approximately 400 g.s"1 to very uniform temperature distribution in the DW. It 150 g.s") immediately after helium injection begins. should also be noted that no significant radial PCC performance can also be monitored by distribution of concentration or temperature was examining the decrease in the pool water inventories observed in this case. 65
As soon as helium injection stops, the measurements indicate the presence of a vertical stratification (Fig. 5a, from t=17500s to t=19500s). Helium is then being vented out of the DW through the PCC feed lines. No helium can be found at Level 10 from t=23000s onwards. Figure 6a shows clearly the evolution of this front: moving upwards till the moment all the helium has left the DW volume (t=20500s). Of course, the time needed to reach a uniform spatial distribution was shorter in DW2 than in DW1, since only one PCC (PCC1) was connected to DW1 instead of two (PCC2 and PCC3) for DW2.
In Test T2.1, from the beginning of the helium Fig. 6a: Vertical temperature profiles in DW2 for Test injection phase, the gas distribution in the DW T1.1. appeared to be quite different. Vertical stratification occurred around 2500s after the start of the injection. In this test, the DGRS was active, circulating gas from the PCC vent lines to the IP. This gas is lighter (and colder) than the ambient medium: it flows to both DW vessels, and feeds the DW volumes above the level of the IP. The gas concentration measurements below the IP level in both DW vessels (Level 10) are qualitatively and quantitatively similar to those obtained in Test T1.1 (with no DGRS activation). However, above the level of the IP, the helium concentration continues to increase, and at the top measurement location (Level 2) reaches 33% in DW1 and 26% in DW2 (Fig. 5b). The corresponding vertical Temperature (C) temperature profiles show that the resulting Fig. 6b: Vertical temperature profiles in DW1 for Test stratification is stronger in DW2 (above the level of the 2.1. IP) than in DW1 (Figs. 6b and 6c). Up to now, no specific reason has been found to explain this non- symmetric gas distribution in the DW volumes.
- MCG.D2B.2 - MCG.D2A.6 - MCG.D2A.10
I Temperature (C) 20000 25000 35000 40000 Fig. 6c: Vertical temperature profiles in DW2 for Test Time (s) T2.1. Fig. 5a: Helium concentrations in DW2 for Test T1.1. There is insufficient information concerning the flow patterns in the DW vessels and IP, and no conclusions can be drawn from the fact that DW1 is connected to one PCC and DW2 to two PCCs. On the one hand, a symmetric discharge to the two DW volumes would generate a flow from DW1 to DW2. A consequence of this could be that more recirculated gas would enter DW2 than DW1. On the other hand, if we assume the DGRS flow in the IP to be distributed equally between DW1 and DW2, a greater dilution in DW2 would explain the lower helium content measured there. In fact, a difference in the two main 5000 10000 15000 20000 25000 30000 35000 40000 45000 flows (more flow to DW1) was observed, and perhaps Fig. 5b: Helium concentrations in DW1 and DW2 for is itself responsible for the measured distribution. Test T2.1. 66
5 PCC PERFORMANCE close to the DW temperature, whereas in the lower tube region the gas temperature approaches that of 5.1 Variations in PCC Performance the pool (water saturation temperature at All observations made during the tests indicated that approximately one bar). The gas mixture in the lower PCC performance changed considerably. Before region blankets the condenser surface. From the helium was injected, total PCC performance was measurements in the tubes, the boundary between following exactly the heat-decay curve. But initiation of the two zones appears to lie somewhere between helium injection led to a significant degradation in Sensor 7 (MTG.P3.7) and Sensor 8 (MTG.P3.8); see PCC performance, and then, some time after the end Fig. 8. of injection, PCC efficiency improved again. These measurements are in agreement with those The decrease in PCC pool water levels is directly already described in previous PANDA tests [2]. There, related to the amount of energy removed from the it was demonstrated that the PCCs adjust their containment by the PCCs. Figure 7 shows the PCC3 performance to the decreasing power level by pool collapsed levels measured in Tests T1.1 and progressively accumulating air in the lower part of the T2.1; the measurements in the two other pools were tubes, thereby reducing the effective condensing always very similar. During the early part of the tests, length. Obviously, in the present tests, there was the PCC levels decreased in the same way. However, insufficient air available for blanketing the heat as soon as the (non-condensible) helium is added, transfer surfaces, and the DW pressure began to PCC performance is significantly degraded, and a decrease (see Fig. 3). change in the rate of decrease of the level is recorded 5.2.2 After Initiation of Helium Injection (Fig. 7). During this phase, pool levels decreased faster for Test T2.1, because of DGRS activation. Long term, the PCC operation mode changes Around t=20000s in both tests, a better efficiency is significantly. In the presence of helium, mechanisms established, and the pool levels decrease faster. occurring in the tubes are more complex. From a Nevertheless, this time corresponds to very different general point of view, one should realize that conditions in the two tests: in Test T1.1, all the helium buoyancy forces arising from steam condensation out has been vented out of the DW, while in Test T2.1 a of a steam/helium mixture strongly influence the flow new kind of "equilibrium" had been reached in which a patterns. roughly constant amount of helium was retained in the upper part of the DW for a long time. The clear separation between the two zones described above almost immediately disappears, and Variations in PCC performance can also be related to a different vertical temperature gradient is established the condensate flow returning to the RPV. Early in the along the tube length. In both Tests T1.1 and T2.1, tests, it was observed that the amount of condensate two different modes of operation could be identified. decreased proportionally to the decay-heat curve. The first refers to the situation in which a vertical When helium injection was initiated however, the temperature gradient (with a higher temperature condensate flow rate dropped, due to the sudden at the top, and a lower temperature at the decrease in PCC performance. In both tests, at bottom) is observed. This gradient indicates that around t=20000s, this flow rate increased significantly the steam/helium mixture is flowing downwards in and then decreased more slowly (as the decay-heat the tube, and is becoming more and more helium decreased) till the end of the test. rich as condensation proceeds. This first mode 5.2 Operation Modes was, for example, observed in PCC3 in Test T1.1 from the beginning of helium injection till The observations presented in this section are based approximately t=23500s (Fig. 9). on gas temperature measurements taken in the PCC3 condenser during Test T1.1; a schematic of the The second mode refers to the situation in which condenser showing the locations of the the temperature levels in the upper (sensor thermocouples is given in Fig. 8. Some temperature MTG.P3.1) and lower (sensor MTG.P3.2) drums traces are given in Fig. 9: these are representative of are maintained (in comparison with the first what was observed in the two tests with helium. mode), while a reverse temperature gradient (the gas mixture being hotter at the bottom) is 5.2.1 Before Helium Injection established. The gas in the tube was then colder than in the lower drum, which would indicate the The steam/non-consensible mixture supplied to the flow was recirculating upwards, and again PCC upper header passes downwards through the becoming colder due to steam condensation. In vertical condenser tubes. While steam is condensing, Test T1.1, this mode is established quite the density of the mixture increases (air fraction suddenly in the PCC3 tubes from about t=23500s becomes higher), and the temperature decreases. onwards (Fig. 9). The temperature measurements indicate two separate zones: in the upper region of the tubes, steam is One should anyway realize that only one tube in each condensing out of a mixture with a rather low air PCC was well-instrumented for gas temperature content, whereas in the lower region air, in a mixture measurements, and that not all of the phenomena with a relatively low steam content, is accumulating. In occurring in the 20 tubes could be monitored. In both the upper tube region, the gas temperature remains tests, sudden changes in PCC operation mode were 67 observed at very different times during the course of the light gas would have been injected at the bottom the tests. It seems that some adjustment of PCC of the DW (and not in the IP, as in the current PANDA performance occurred. Nevertheless, this process is configuration), more volume would be available to quite different from what happens when only air and store non-condensibles in the DW, leading to an steam were present in the tubes — different tubes overall lower system pressure. With respect to an now sharing the load unequally. ESBWR containment, the design of such a DGRS system should, of course, be optimized accordingly. The available temperature measurements in the tubes have shown qualitatively similar results in both tests, the same operation modes being observed. In Test T1.1, after all the helium is vented out of the DW, one could say the helium supply to the PCCs had ended. Nevertheless, it is obvious that, in order to achieve the kind of self-adjustment of the PCC heat removal efficiency observed, some helium had to have remained in the tubes.
6 EFFECT OF THE DGRS ON THE SYSTEM RESPONSE The DGRS is able to retain non-condensible gases in the DW. In Test T2.2 (DGRS active, no helium injection), this has been observed to be approximately Fig. 8: Thermocouple locations for the gas 2.4% and 3.8% respectively for DW1 and DW2, as temperature measurements in condenser measured by the air content after DGRS activation PCC3. [12]. The gas distribution obtained in Test T2.1 has been detailed in the previous sections. As expected, the activation of the DGRS led to mitigation of system pressure build-up due to the addition of non- condensibles in the course of the transient. A 0.3 bar difference in peak and final DW pressures could be observed between Tests T1.1 and T1.2 (Fig. 3), though the course of the transients was very similar in both cases (i.e. with and without the DGRS). The re- establishment of a better PCC efficiency leads to pressure stabilization. It seems that the system had reached a new "equilibrium", though related to very different conditions in the DW in the two tests. Nevertheless, PCC operation modes appear to be 5000 10000 15000 20000 25000 30000 35000 40000 45000 very similar for the two tests in which helium is Time (s) retained in the tubes. Fig. 9: PCC3 gas temperatures for Test T1.1.
The end of Test T2.1 corresponds to the end of the DGRS activation period. During the test extension, pressure quickly increased, reaching a level comparable with that seen in Test T1.1 with no DGRS activation at all. At the same time, Fig. 5b shows that the helium retained in the DW is vented, as was observed at the end of the helium injection phase in Test T1.1. A consequence of an accidental interruption of this (non-passive) DGRS system could be an interesting feature of future PANDA tests.
1 5000 10000 15000 20000 25000 30000 35000 40000 45000 A 30 l.s" flow rate was imposed for these tests. More Time (s) tests would be needed to assess the importance of Fig. 7: PCC3 pool collapsed levels for Tests T1.1 this flowrate as a parameter influencing the amount of and T1.2. gas retained in the DW and the system pressure.
It appears that the amount of non-condensibles 7 SUMMARY retained in the DW, and thus the system response to Two integral system tests have been performed in a release of light gas, is dependant of the PANDA in which the long-term PCCS cooling phase configuration adopted for the DGRS pipe lines; i.e. following a LOCA caused by a MSLB has been how gas is reintroduced into the DW. For example, if 68 assessed considering the presence of a large amount [5] G. Yadigaroglu and J. Dreier, "Passive Advanced of light gas (helium). The tests have provided relevant Light Water Reactor Design and the ALPHA data with respect to PCCS performance and overall Programme at the Paul Scherrer Institute", system behaviour. Kerntechnik, 63, 39-46 (1998). The two tests have shown similarities with respect to [6] G. Yadigaroglu, "Derivation of General Scaling system response. In both tests, some time after the Criteria for BWR Containment Tests", Proc. 4th end of helium injection, the system appears to have Int. Conf. on Nuclear Engineering, New Orleans, established a new "equilibrium". When the DGRS is USA, 1996. activated, some helium is permanently retained in the [7] J. Hart, W.J.M. Siegers, S.L. de Boer, M. upper part of the DW, and does not contribute to the Huggenberger, J. Lopez Jimenez, J. L. Munoz- overall system pressure. This is of considerable Cobo Gonzalez, F. Reventos Puiganer, "TEPPS- interest. As the phenomena leading to this situation Technology Enhancement for Passive Safety need to be better understood, post-test analyses Systems", Nuclear Engineering and Design, 209, using CFD-like codes (as foreseen as part of the 243-252 (2001). TEMPEST project) are recommended. [8] H. W. Schenk, and A. M. Hooft Van Huysduynen, ACKNOWLEDGMENTS "Scaling Analysis of Passive Containment Cooling Tests", INNOTEPSS (96)-D004, ECN, The work reported in this paper was supported by the Petten, 1997. European Commission (European Commission 4th and 5th Euratom Framework Programme on Nuclear [9] M. Huggenberger, J. Dreier, S. Lomperski, C. Fission Safety) and the Swiss Federal Office for Aubert, "PANDA Facility, Test Programme and Education and Science. Data Base: General Description", PSI Internal Report, ALPHA-606/TM-42-96-08, 1996. REFERENCES [10] S. Lomperski, J. Dreier, C. Wilkins, "Error [1] R. J. Mccandless and J. R. Redding, "Simplicity: Analysis for PANDA Instrumentation", PSI the key to improved safety, performance and Internal Report, ALPHA-503-1/TM-42-95-03, economics", J. Nuclear Engineering International, 1996. 34, 20-24(1989). [11] O. Auban, D. Paladino, "Description of PANDA [2] T. Bandurski, M. Huggenberger, J. Dreier, C. Gas Concentration Measurement System using Aubert, F. Putz, R.E. Gamble, G. Yadigaroglu, Mass Spectrometry", PSI Internal Report, "Influence of the Distribution of Non- ALPHA-03_01/TM-42-02-01, 2002. Condensibles on Passive Containment [12] D. Paladino, O. Auban, P. Candreia, M. Condenser Performance in PANDA", Nuclear Huggenberger, H. J. Strassberger, "Hydrogen Engineering and Design, 204, 285-298 (2001). Impact on Passive Containment Cooling System [3] V. A. Wichers, J. Y. Malo, J. Starflinger, M. and Accident Mitigating Design Features: Part F", Heitsch, G. Preusser, M. Tuomainen, M. 5th Euratom Framework Programme, EVOL Huggenberger, "Testing and Enhanced Modelling TEMPEST D06F Report, 2002. of Passive Evolutionary Systems Technology for Containment Cooling (TEMPEST)", FISA, Luxemburg, 2001. [4] D. Paladino, O. Auban, P. Candreia, M. Huggenberger, H. J. Strassberger, "New PANDA Tests to Investigate Effects of Light Gases on Passive Safety Systems", Proc. 2002 Int. Congress on Advanced Nuclear Power Plants, Hollywood, Florida, USA, 2002. 69
CFD SIMULATION OF STEADY-STATE CONDITIONS IN A 1/5™-SCALE MODEL OF A TYPICAL 3-LOOP PWR IN THE CONTEXT OF BORON DILUTION EVENTS
T.V. Dury
As part of the STARS Project at PSI, the Computational Fluid Dynamics (CFD) code CFX-4.3 was used to examine boron mixing within the pressure vessel of a 1/5-scale model of a typical 3-loop PWR. The object of the first phase of the study, which is reported here, was to examine steady-state conditions with constant flows in all three cold legs. Boron concentration was simulated by a temperature difference of 10°C between one of the inlets and the other two. Particular care had to be exercised in implementing the additional resistance factors into the code for various zones simulated as porous regions, such as perforated tube plates and volumes containing support columns. Results showed that circumferential swirling occurred from the level of the inlet as far as the core inlet plate, and that the swirl appears to slightly enhance the mixing of the hotter and colder fluid streams. In general, there is a significant degree of separation between the individual flows from each cold leg, even at the inlet to the core, and even at the highest inlet flow simulated. Mixing of the cold and hot streams does not change significantly as overall flowrate increases. At the lowest flowrate, the flow area at the core inlet over which the temperature was still at the level of that of the hotter of the cold-leg inlets is 2.4%, dropping to 1.0% at 8 times this flowrate.
1 INTRODUCTION Constant material properties were used, with the Boussinesq approximation used to represent There exists a potential risk of a reactivity accident buoyancy effects. occurring in a Pressurised Water Reactor (PWR) if a slug of water with a low concentration of boron (which is introduced into the coolant as a reactivity inhibitor) 3 MODEL passes into the vessel and core under certain 3.1 Structure and Mesh conditions of Reactor Cooling Pump start-up. This could happen if the reactor is at decay-heat level and Figure 1 is a horizontal cross-section through the natural circulation has been stopped. Depending on vessel, showing the general vessel layout and the the degree of dilution due to mixing during its (arbitrary rectangular) CFD model block structure in passage, the slug could cause a local increase in core the central region. Each inlet pipe was modelled to a reactivity, leading to a power excursion and possible distance of 100 hydraulic diameters upstream of the core melt-down. bend before the nozzle. The cold leg pipe walls and the outer vessel wall were not simulated. In experiments, borated water can be simulated by injecting water through one cold leg at a higher Core barrel temperature than the water supplied through the other cold legs (as water containing boron has a lower density than pure water at the same temperature). This technique has been used, for example, in the Bora-Bora facility, tested by Electricite de France in the early 1990s [1-2]. As a contribution to the STARS Project, the CFD code CFX-4.3 [3] has been used to examine a 1/5-scale model of the pressure vessel and internal components of a typical 3-loop PWR. The first phase of the study was to simulate steady-state conditions (with constant flows in all cold legs) at three different levels of flow Cold leg 1 (to examine Reynolds Number effects), and to examine the degree of mixing by the time the inlet fluid streams have reached the core inlet plane. Fig. 1: Plan view of the vessel.
2 CONDITIONS SIMULATED Figure 2 shows the circumferential positions of the thermal shields and exposure baskets in the Three different cases were simulated, with water downcomer; the final model mesh is shown in Fig. 3. flowrates applied to the cold leg corresponding to inlet From parameter-studies, the subdivision at the cold pipe velocities of 1.63 m/s, 3.27 m/s or 13.07 m/s. leg nozzle penetrations was optimised to be 6 cells System reference temperature and the temperatures wide by 4 cells high. at the inlets to Legs 1 and 3 were (arbitrarily) set to 20°C, and that at the inlet to Cold Leg 2 set to 30°C. 70
Centring blocks at 90° spacing Thermal shield procedure prevented chequerboard oscillations of Exposure basket pressure. Details of the models are given in [3]. For each different flowrate examined, the mesh was adjusted so that the thickness of fluid cells adjacent to Cold legs at 120° spacing walls satisfied the requirements of the wall function treatment used for the standard high Reynolds Thermal shield Number k-e turbulence model (30< y+ < 100), which Block subdivision was applied for all calculations. Solution convergence was assumed once the total mass residual dropped Exposure basket by a factor of 107 from its starting value.
4 RESULTS Fig. 2: Sectional view of CFD model showing thermal 4.1 Low Flowrate Case shields and exposure baskets. 4.1.1 Velocities Local mean Reynolds Numbers in the vessel, based 3.2 Porous Patches on the overall average component of velocity in the All regions containing attachment plates or rods were horizontal and vertical planes, are given in Table 1; modelled as 3D porous regions. Because of the Reynolds Number in each cold leg is calculated restrictions in the CFX-4.3 model (see [4]), the along the pipe axis, upstream of the bend. porosity was set to unity and an individual resistance coefficient applied for each region, additionally Region Transverse Vertical factored to incorporate the resistance which would 5 have been induced by the higher fluid velocity had the Inlet pipe 2.28x10 — s porosity been correctly applied. Empirical resistance Downcomer (below thermal shields) — 6.36 x10 coefficients were applied, estimated either from the 4 4 data of Idel'chik [5] or from Incropera and DeWitt [6]. Below lower attachment plate 1.03 x10 1.10 x 10 Between attachment plates 4.48 x104 8.58 x10s Hot leg sleeve / Below Core Support Plate 8.25 x10s 1.42 x10s
4 IIST Within Core Support Plate — 2.24 x10
2 4 ir ii" Core Support Plate Core Inlet Plate 10 1.65 x10
[fllllllK 4 Within Core Inlet Plate — 1.09 x10
Table 1: Local Reynolds Numbers (1.6 m/s case).
f-llPlMs^ssssss• SSSggS^fSSSSSS;^ : Core barrier Core barrel >4 " Circumferential flow on a horizontal cross-section at the level of the centre of the cold legs can be seen in Fig. 4. Core plate
Core support plate
Attachment plates Higher velocity here
Fig. 3: Sectional view of the CFX-4.3 mesh layout. than here 1.93 m/s
3.3 Physical Models and Numerics A Dirichlet boundary condition was applied at the inlet to each cold leg pipe, and a turbulence intensity and dissipation length-scale specified; a pressure boundary condition was defined at the outlet plane. The SIMPLEC velocity-pressure coupling algorithm was used for outer solver iterations, with the Algebraic Multi-Grid model used for inner iterations. Hybrid differencing was used to model the convective terms Fig. 4: Velocity vectors in the horizontal plane of all transport equations, and central differencing through the cold legs (low flowrate case). used for the pressure; the Rhie-Chow interpolation 71
The influence of each bend is clearly visible in the direction of flow is upwards). The maximum velocity in small asymmetry of the flow field as it enters the the downcomer is about 1.0 m/s away from the downcomer. This asymmetry results in a small restriction caused by the thermal shields and anticlockwise component of velocity (as seen in this exposure baskets. view from above the vessel). In this plane, the hot-leg Flow leaving the downcomer turns around and passes sleeves inhibit the circulation, though it continues to up through the centre of the vessel, with a strongly be present lower down in the vessel. However, the non-uniform distribution present across the vessel as two hot-leg sleeves adjacent to each cold leg inlet fluid enters the core support plate. This non-uniformity nozzle are positioned asymmetrically relative to the disappears by the time the fluid reaches the core inlet nozzle (e.g. L1>L2 for Cold Leg 2), which encourages plate. Thus, the effectiveness of the design would asymmetry in the circumferential flow field within the appear to have been confirmed, in that the velocity annulus. Azimuthal expansion is restricted by the solid profile has become essentially flat by the time the hot leg sleeve blockages, which prevent flows from coolant has reached the entrance to the core. the different cold legs interacting with each other. However, although the flow distribution has become Velocity vectors in the vertical plane through the uniform, this is not true of the temperature distribution. centre of the inlet penetration of Cold Leg 2 and the centre of the reactor vessel are shown in Fig. 5. The +0.5 m/s incoming fluid expands as it enters the downcomer, with some slight penetration into the region above the inlet pipe. At the bottom of the downcomer, the downward momentum of the fluid is sufficiently high to limit expansion into the central region of the vessel above the upper attachment plate. However, the resistance of the lower plate is high enough to cause the fluid to expand into the centre of the vessel between the plates, with very little fluid penetrating below the lower plate. -0.5 m/s
+0.59 m/s +0.5 m/s
Fig. 6: Vertical velocity component in the vertical plane through Cold Leg 2 (low flowrate case).
4.1.2 Temperatures Temperatures on the horizontal cross-section through the centre of the cold-leg pipes are shown in Fig. 7.
Bottom of downcomer 0.4 m/s Upper attachment plate
Lower attachment plate'
Fig. 5: Vectors of total velocity in the vertical plane through Cold Leg 2 (low flowrate case).
It should be noted that the (locally) higher velocities which ought to occur within zones where the porosity is less than unity are not present, because the porosity is always set to unity in the model, and compensated for by increasing the applied resistance coefficients (to avoid the code restriction mentioned in Sec. 3.2.). The vertical flow pattern can be seen more clearly Fig. 7: Temperatures on the horizontal section from the shaded contour plots of the vertical velocity through the inlet plate (low flowrate case). component of flow (Fig. 6, in which the positive 72
The hot-leg sleeves prevent fluid from the different thermal shields below the hotter inlet, which tend to cold legs from interacting in this plane. A horizontal deflect fluid in the same anticlockwise direction. plot of temperature contours at the mid-point of the Parameter studies with this model could identify the thermal shields (Fig. 8) shows that the hotter fluid is relative magnitudes of each of these causes, but are quite well confined within a 120° segment of the not warranted, since the effect of swirl at this level on downcomer and has not yet mixed, in spite of the the mixing of the hot and cold fluid streams does not partial blockage of the annulus and the disturbance to appear to be significant. In addition, in spite of the the flow caused by the thermal shields. Each length of inlet ducting modelled for each cold leg, the exposure basket, due to its large surface area relative assumption of a uniform flow field boundary condition to its volume, has a temperature close to that of the at the duct inlets is unlikely to be strictly correct, and adjacent fluid, while the thermal shields have strong non-uniformity of the inlet flow distribution could still thermal gradients through their thicknesses. have a small effect on the flow distribution at the entry to the downcomer. 30°C Between the attachment plates (Fig. 10), the effect of swirl is apparent from the shape of the interface between the hotter and colder fluid zones. The anticlockwise movement has strongly distorted the interface, although it still remains relatively smooth. This is possibly due to the influence of the higher velocity of the core of the jet at this level dominating the injection pattern. At the level midway between the upper attachment 20°C plate and the core support plate (Fig. 11), the hotter stream of fluid has spread strongly towards the centre of the vessel, though the hot-cold interface is smoother than between the attachment plates at a Fig. 8: Temperatures on the section through the lower level. However, the azimuthal position of the hot centre of the thermal shields (low flowrate region shows a clockwise rotation relative to that case). between the attachment plates, though its position Comparing temperatures at the level of the core inlet has been corrected again by the time the fluid reaches plate (Fig. 9) with those in Fig. 8 shows that there is a the level of the core inlet plate (Fig. 9). This shows small anticlockwise azimuthal swirl in the flow (as that there are strong 3D effects in the fluid motion in viewed from the top of the vessel). The peak the region between the downcomer and the core temperature in the hotter fluid has moved 23° support plate, though not strong enough to circumferentially by the time it has reached the core significantly influence mixing. inlet plate. Ultimately, it is important to know how well the fluid from the hotter inlet leg (the 'boron-diluted' leg) mixes 30°C with the colder fluid from the 'boron-rich' legs by the time it reaches the core inlet plate. From Fig. 9, it can be seen that mixing has taken place, as evidenced by the fact that the fluid temperature is above that of the cold inlets over more than half of the cross-section of the core inlet plate.
30°C
20°C
Fig. 9: Temperatures through the core inlet plate (low flowrate case).
There are three possible reasons why this should occur: the first is as a consequence of the bend in the cold legs; the second, the asymmetrical distance 20°C between the hot-leg sleeves and the stagnation points at the inlet nozzles; and the third is a consequence of Fig. 10: Temperatures midway between the attach- the asymmetrical gaps which exist between the ment plates (low flowrate case). 73
The temperature is still at that of the hotter leg over At the level of the core inlet plate (Fig. 12), swirling in 2.4% of the core inlet plate area (i.e. it is unmixed at the downcomer has moved the peak temperature this level), while the contour representing 50% mixing azimuthally by slightly over 28° from the position of divides the area into approximately two-thirds (cooler) Cold Leg 2 — an increase of 5° over the low flowrate and one-third (hotter) sections. case. Otherwise, the thermal profiles are very similar in both cases, with no significant difference in the average mixing of the streams (cf. Fig. 9). Also, there is the same degree of mixing in the region above the plate (cf. Figs. 8 and 13). However, there has been some improvement in mixing at the level of the core inlet plate, for which the proportion of unmixed hot fluid has dropped from 2.4% to 1.4% of the total.
Fig. 11: Temperatures midway between upper attachment and core support plates (low flowrate case).
4.2 Intermediate Flowrate 4.2.1 Velocities The velocity characteristics of this case show only minor deviations from a straightforward doubling of Fig. 13: Temperatures through the centre of the the velocities of the low flowrate case at any given thermal shields (intermediate flowrate case). point in the model, and these are caused only by small variations in the imposed resistance factors. 4.2.2 Ternperatures 4.3 High Flowrate Case There is very close similarity in the temperature 4.3.1 Velocities contours in the vertical-plane through the hotter inlet A plot of the vertical component of velocity in the and the centre of the vessel with those of the low vertical plane through the centres of Cold Leg 2 and flowrate case, with only a slight increase in the reactor vessel (Fig. 14) shows that there is now penetration of fluid from the downcomer into the better mixing below the core support plate, with a bottom region of the vessel. This causes marginally broader core of higher vertical velocity material having better mixing of the hot and cold fluid streams below been created by an even stronger penetration from the core support plate. the downcomer into the lower region of the drum. All other aspects of the flow are similar to the previous case, with the effect of circumferential swirl again stronger than for the lower inlet flowrate cases, though the effect is not strictly proportional to the ratio of these flowrates.
4.3.2 Temperatures The effect of swirl can be seen from the distribution of temperature over the plane of the core inlet plate, seen in Fig. 15. The peak temperature has now moved around by 30° relative to the cold leg through which it entered the vessel — an increase of only 2° over the intermediate flowrate case. The general degree of mixing of the fluid remains similar to the two previous cases, and the only clear difference is that the relative area of the plate covered by unmixed hot Fig. 12: Temperatures through the core inlet plate fluid has again been reduced: to 1.0% compared with (intermediate flowrate case). the 1.4% for the intermediate-flow case. 74
1 \ ^ +4.4 m/s
30°C TZ) r«j i. ) h) ^ ^ Igift^^——^^ -10.2 m/s
-8 m/s
-4 m/s
+3 m/s 20°C
+4.4 m/s +4.4 m/s Fig. 15: Temperatures through the core inlet plate (high flowrate case).
Fig. 14: Vertical velocity component in a plane REFERENCES through the hotter inlet (high flowrate case). [1] A. Martin, D. Alvarez, F. Cases, S. Stelletta, "Three-dimensional calculations of the primary 5 CONCLUSIONS coolant flow in a 900 MW PWR vessel. Numerical simulation of the accurate RCP start-up flow rate", Steady-state calculations have been performed in Proc. ICONE-5 Meeting, Track 1, ICONE5-2426 simulation of the injection of boron-diluted water (CD-ROM), Nice, 1997. through one leg of a 3-loop PWR. The flowrates in the three cold legs to the reactor vessel were set at [2] A. Martin, D. Alvarez, J.P. Schneider, G. Petit, S. 1.63 l/s, 3.27 l/s or 13.07 l/s for the low, intermediate Stelleta, "Boron dilution reactivity transient and high flowrate cases, respectively, with the computation in a 900 MW PWR. Thermal- temperature of incoming fluid to Cold Leg 2 set at hydraulic plug mixing in the vessel and thermal- 10°C above that to the other two cold legs, to simulate hydraulic/neutronic response of the core", Proc. the presence of de-borated water. Conclusions are as ICONE-6 Meeting, Track 5, ICONE-6111 (CD- follows: ROM), San Diego, California, USA, 1998. 1 Circumferential swirl exists in the downcomer and [3] CFX-4.3: SOLVER, AEA Technology, Harwell, vessel, which increases with increasing flow, but Oxford, UK, March 2000. not at a rate proportional to the increase. [4] C.J. Fry, J.N. Lillington, "Validation of the CFX-4 2 The presence of swirl appears to have only a small Code for PWR Fault Analysis", Proc. OECD/CSNI enhancement effect on the mixing of the hotter and Workshop on Advanced Thermal-Hydraulic and colder fluid streams. Neutronic Codes: Current and Future 3 At the lowest flowrate, the area of the core inlet Applications, NEA/CSNI/R(2001)2, Vol. 2, pp. plate over which the temperature has not dropped 133-144, Barcelona, Spain, 10-13 April 2000. from that of the hotter of the three cold-leg inlets (i.e. representing the unmixed-fluid) is 2.4% of the [5] I.E. Idel'chik, Handbook of Hydraulic Resistance total, dropping to 1.4% at twice the flowrate, and — Coefficients of Local Resistance and of 1.0% at 8 times the flowrate. Friction, Gosudarstvennoe Energeticheskoe Izdatel'stvo, Moskva-Leningrad, 1960. ACKNOWLEDGMENTS [6] F.P. Incropera, D.P. DeWitt, Introduction to Heat nd The author acknowledges the assistance given to him Transfer, 2 Edition, pp. 376-379, John Wiley, by AEA Technology, Harwell, in setting up the CFD New York, 1990. model, and to Dr Brian Smith (LTH, PSI) in support of the implementation of resistance factors for porous regions in the CFX-4.3 code. 75
ASSESSMENT OF THE MELCOR CODE AGAINST PHEBUS EXPERIMENT FPT-1 PERFORMED IN THE FRAME OF ISP-46
J. Birchley
Calculations of PHEBUS FPT-1 are being performed in the frame of CSNI International Standard Problem ISP-46. The objective of ISP-46 is to assess the capability of computer codes to provide an integral simulation of a severe accident in a Pressurised Water Reactor (PWR), from the initial stages of core heat-up to the behaviour of released fission products in the containment. The present calculations are performed using MELCOR, chosen as the main tool for assessment of Swiss nuclear plants by virtue of its whole-plant-simulation capability, using modelling practices as similar as possible to those used in plant analyses. The calculations cover the bundle heat-up, degradation, the release, transport and retention of fission products and other materials, and the thermal-hydraulic and aerosol behaviour in the containment. Comparison between a best-estimate case and experiment demonstrates the code's ability to capture most aspects of the sequence with fair to good accuracy. Uncertainties remain, particularly in regard to core degradation, and the chemistry and transport of fission products. Weaknesses of code models in these areas largely reflect limitations in current knowledge.
1 INTRODUCTION The adoption of FPT-1 as the subject of CSNI International Standard Problem ISP-46 [5] enables The PHEBUS Fission Product (FP) project [1] is an direct and rigorous code-code and user-user international, severe-accident research programme comparison, and hence demonstrates the comparative conducted by the French Institute de Radioprotection strength, weakness and susceptibility to user effects of et de Surete Nucleaire (IRSN). The key feature of the various codes, thereby providing an overall PHEBUS is a "cradle-to-grave" simulation of the assessment of current tools and know-how. accident using prototypic materials, capturing the entire portfolio of processes —material, chemical and Experiment FPT-1 has been the subject of numerous transport — and their causal interaction. The PHEBUS previous analyses, for example [6], using detailed experiments thus provide a rich and unique source of code packages to examine separately the various data on fission-product release, transport and chemical aspects of the sequence, and to interpret the complex behaviour under prototypic conditions with which to phenomena exhibited. The understanding thus gained assess, improve and validate methods for source-term makes it appropriate to calculate the sequence in an evaluation, and hence place best-estimate predictions integral manner with the same tools as used for plant of source terms on a firmer basis. application. The present calculations were performed using MELCOR 1.8.5 [7], chosen as the main tool for The PHEBUS facility comprises an in-pile, 1 m long, assessment of Swiss nuclear plants by virtue of its 5x5 rod bundle, an upper volume, pipework to simulate capability to simulate all phases and aspects of a plant the hot leg, the steam generator primary side, the cold accident sequence in an integral manner, and its leg, and a vessel with a sump to simulate the comparatively economical run-time performance. In containment. Power is delivered to the fuel rods via a accordance with the aims of the ISP, the approach driver core which surrounds the bundle and is used in the calculations was to use modelling controlled by the operators. The components are practices, in regards to noding detail and model scaled approximately 1:5000 with respect to a 900 options which are as similar as possible to those to be MWe commercial PWR. The containment surfaces are used in plant analyses. The scope of the calculations slightly superheated, except for a limited area which is covers the bundle heat-up, degradation, and the subcooled and simulates the scaled surface for release of fission products and other material from the condensation. A general description of the PHEBUS bundle (Phase 1), the transport and retention in the facility as configured for experiment FPT-1 is given in circuit (Phase 2), and the thermal-hydraulic and [2]. The experiment [3] was performed using fuel aerosol behaviour in the containment (Phase 3). The irradiated to near end-of-cycle condition. The transient iodine chemical behaviour in the containment (Phase was representative of a PWR depressurised sequence 4), though part of ISP-46, was not calculated because with a dried-out core and a flow of steam sufficient to the status of the iodine chemistry models in MELCOR maintain oxidising conditions. The power transient is not yet established, and their use at PSI remains comprises a series of ramps and plateaux, with under review. shutdown according to the bundle state deduced from the temperatures in the shroud. Fission products are The following sections present an outline of MELCOR, released from the bundle and are partially transported especially the features and models relevant to FPT-1, to the containment vessel. Following reactor shutdown, a summary of the input model, a description of the the containment is isolated and the aerosol and best-estimate a calculation, discussion of sensitivities, chemical behaviour is monitored. The results are and an overall assessment of the code modelling and reported in detail in [4]. practicalities. 76
2 CODE DESCRIPTION grouped, plus a separate group for U02. In each class, a single representative material is identified (e.g. Cs MELCOR is designed to simulate all phases of a plant for alkali metals, I for halogens), so that the class severe accident, the relevant plant phenomena and inventories therefore comprise the sums of the components. The objective is to concentrate on the individual nuclides. Release from the fuel is modelled major characteristics and parameters important for using CORSOR [8], and the associated decay-heat is plant safety rather than to capture the processes in transported with the fission products. Release of detail. MELCOR comprises, typically, simple empirical absorber and cladding materials can also be modelled correlations or parametric statements, and is normally by defining additional classes. Material can be used conjunction with coarse-mesh input models. transported as vapour or aerosols. Vapour may The different packages are internally decoupled from condense on cool surfaces, or may form aerosols each other, but are linked through the executive within the bulk volume. The latter are subject to module, which synchronises the time-step processes of agglomeration, inertial and turbulent advancement, manages the data transfer, and impaction, thermophoresis, etc., and as a result are maintains conservation and consistency of physical partially retained in the course of transport. On quantities. reaching the containment, they are removed from the gas space, mainly by settling and diffusiophoresis onto The thermal-hydraulic module is pivotal within the condensing surfaces. The treatment of chemistry is code, and is the most intimately linked to the other very limited in that fission products are released from modules. In particular, it furnishes the thermal and fluid the fuel in elemental form. However, some of them conditions for all of the process and component form compounds after release, notably CsOH, TeO, models — degradation, fission-product transport, and Csl. The only area where fission product chemistry corium-concrete interaction, etc. In the spirit of the is treated in detail is the containment, via an iodine coarse-noding approach, fairly large regions are often model based on the INSPECT code, as described in represented by just one hydraulic cell. Each cell the MELCOR Reference Manual [7]. normally comprises two regions: a pool, which may be a single-phase liquid or a two-phase bubble-rise 3 MODEL DESCRIPTION OF THE PHEBUS region, and an atmosphere which may contain liquid FACILITY water in the form of fog. The mass of each non- condensable gas species is also tracked. The cells are The approach adopted was to use a level of detail connected by flow paths where the flows of the pool similar to that of a plant model, and mainly default or and atmosphere are separately calculated by means of neutral model options, taking into account that certain a simplified, one-dimensional treatment of momentum facility characteristics which are atypical of a plant balance. Closure is provided via correlations for the might need to be treated in a special way. mass, energy and momentum exchanges. Heat conducting structures provide the physical and thermal The hydraulic system was modelled using rather a boundaries for the hydraulic system; these can ablate coarse nodalisation, with a total of 14 fluid volumes. or undergo failure through thermo-mechanical loads. Just one cell was used for the heated bundle, three for the hot side of the steam generator (SG) tube, one for The core models follow the heat-up, oxidation, fuel the SG cold side, and a single volume for the dissolution, bulk melting and relocation. The core containment tank. Reflecting the layout of the components include fuel rods, absorber rods, cladding, PHEBUS test section and circuit, the volumes were non-supporting structures such as guide tubes, and connected in simple linear fashion with no parallel flow supporting structures. In order to resolve the spatial paths or volumes. The noding scheme, with the variations within the core, it is customary to subdivide it respective volume and junction elevations, is shown in into a number of radial and axial cells within a single Fig. 1. fluid volume. Heat generation in the core is by a The PHEBUS core components extend over three fluid combination of fission, decay and oxidation. Melting volumes — the inlet plenum, the active bundle and the occurs at the melting point of the pure material, or due upper plenum. The 1 m active bundle is subdivided to eutectic interactions between metallics (e.g. steel into 11 axial and 2 radial cells. The axial nodes are of and Zr) and between metallic and oxide (e.g. Zr and length 5 cm, 9*10 cm, 5 cm, in order to ensure that U0 , Zr and Zr0 ). The fuel rods are considered to be 2 2 the cell centres are located at the thermocouple locally reduced to debris, once intact geometry is lost; elevations. this according to a temperature criterion, or when melting occurs. Both liquid and solid debris can Heat structures were used to represent the bundle relocate downwards or radially, provided there is shroud, the pipework of the rising line, hot leg, SG space for them to occupy. Material relocates to the tube, cold leg, containment walls, sump, upper and lower head when the supporting structures fail, lower vaults, and the wet and dry parts of the according to an empirical correlation or parametric condensers. The material properties and dimensions criterion. of the heat structures were as described in the specification document. The models and input The large number of fission products is reduced to 11 parameters in the base and best-estimate cases are radionuclide classes into which the nuclides are summarised in Table 1. 77
Fig. 1: MELCOR noding for PHEBUS circuit.
The hydraulic conditions are determined, essentially, experiment, with the notable exception that release of by the initial conditions, the bundle power, the inlet absorber and structural material from the bundle was flow, and the circuit and containment boundary not calculated, due to the absence of an appropriate conditions. The circuit boundary temperatures were model in MELCOR. maintained at prescribed values (i.e. 700°C and 150°C for the hot and cold sections of the circuit), while the bundle shroud was cooled externally by the flow of water in a cooling jacket with inlet temperature set at 165°C. The initial conditions in the containment and steam inlet flow were defined according to the specification document. The containment external temperatures were set to their experimental values. The bundle power and inlet steam flow are shown in Fig. 2.
The starting point for the calculations was a case in which all of the facility and experimental parameters were set to their nominal values, and the code default or neutral model options and values were used throughout. The aim of the nominal case was to provide a first assessment of the ability of MELCOR to calculate the sequence without introducing any special models or "tuning" to experiment. Despite this, the Fig. 2: Bundle power and steam flow. model correctly reproduced most aspects of the 78
Table 1: Summary of MELCOR process models.
Process BASE CASE Best estimate
Bundle nuclear power Nominal, modified for movement fuel and absorber As base case * 0.9 Shroud As stated in [1]. Inter-layer gaps contain steam, As base case; thermal conductivity increased at high temperature material conductivity to simulate gap closure *1.2 Zircaloy oxidation model Urbanic/Heidrick; parabolic kinetics As base case Diffusive limitation on Zr reaction Heat/mass transfer analogy for rate of steam As base case in gas phase diffusion Oxidation of Zr-bearing mixtures As metallic Zr, adjusted for surface area of refrozen As base case debris U02/Zr dissolution model Hofmann As base case
Oxidised cladding breach Tdad > 2400 K or oxide shell < 0.01 mm As base case Bulk fuel movement T > 2500 K and not supported by intact component As base case U02-Zr02 melting T > 2800 K As base case
FP release from fuel (rods, melt, CORSOR-M for FP, U02. Low volatile FP; U02, melt pool) Same for intact, debris or molten coefficients increased Structural material release Defined as pseudo-FP with tuned release As base case; (control rods, cladding,...) parameters coefficients retuned Aerosol characteristics: mass rho=2500 kg/m3 rho = 5000 kg/m3 median diameter (D), density, Dmin / Dmax = 1 / 50 mu Dmin/Dmax = 0.1/10 mu size range and distribution log-normal, size classes = 5 Deposition and resuspension MAEROS models (from Hinds [9]): As base case processes modelled settling, diffusio-/thermo-phoresis, Brownian diffusion, inertial impaction. No resuspension model Agglomeration processes MAEROS models (from Hinds [9]): As base case modelled gravitational, turbulent and Brownian coagulation; hygroscopic effects not modelled Aerosol nucleation models Aerosol forms with mean size geometric of smallest As base case class. Composition by weighted sum of condensed species
Wall effects Wall condensation if Twan < T* < Tgas; As base case T* = species saturation temperature Revolatilisation if Twan > T* Adsorption not modelled Treatment of chemistry I as Csl; remaining Cs as CsOH; Te as TeO As base case Containment chemistry not represented Steam condensation model Heat-mass transfer analogy to convert Nusselt to As base case Sherwood number; reduction when non- condensables present
This omission made it impossible to assess the in Table 2, together with the best-estimate calculation modelling of fission-product transport and aerosol described in the following section. behaviour in the containment. To overcome this Several further calculations were performed in order deficiency, a set of parameters was constructed to to examine the following areas of sensitivity: bundle simulate the releases. The resulting model was used power, radiation exchange factors in the bundle, for the base-case submission to ISP-46, which aimed shroud thermal conductivity, oxidation model, therefore to provide a calculation corresponding to the degradation parameters, release parameters, noding data in the ISP specification document avoiding, as far of steam generator tube, containment initial boundary as possible, use of any experimental results. The conditions, aerosol density and size range. Thus, the calculated sequence is compared against experiment variation in input parameters was intended to cover the major sources of experimental and modelling 79 uncertainties. The best-estimate calculation included degradation is underestimated, and melt-pool modifications to bundle power and shroud conductivity formation is not calculated. (within their experimental uncertainty bands), and Comparison of the bundle fuel temperatures, shown in further changes to the release parameters. The best Figs. 3a, 3b, demonstrates remarkably good estimate attempted to reproduce the observed thermal agreement during the heat-up through the early response during the early part of the sequence up to plateau and main oxidation excursion. the main oxidation peak, and the bulk releases from the bundle.
4 DESCRIPTION OF BEST-ESTIMATE CALCULATION The calculated base case and best-estimate event sequence and major characteristics are compared with experiment in Table 2. The overall character of both calculations agree well with the experiment data, which suggests that MELCOR is able to capture, at least globally, the major phenomena - cladding oxidation, fuel liquefaction, debris formation, fuel movement, oxidic melting, fission product release, aerosol behaviour, and containment thermodynamics.
Table 2: FPT-1 Experimental and calculated event sequence. Fig. 3a: Fuel rod temperatures (upper region). Base case Event (units) Experiment BE Cladding rupture (s) 5700 4770 5024 Start of oxidation (s) 8580 5320 (>1% total H2 generation) 6160 Start of FP release (s) 11185 9950 (>1% Xe release) 10900 Control rod failure (s) 9690 9250 10420 First relocation (s) 11000 10660 11250 First oxidation peak (s) 11260 10630 11300 First oxidation peak 2490 2400 temperature (K) 2470 Melt-pool formation (s) ? 16925 Fig. 3b: Fuel rod temperatures (lower region).
Some discrepancies become apparent during the later Total mass of H2 (g) 96 101 period, but the comparison should be regarded as 99 qualitative only, since the degradation affects the Displaced fuel (kg) « 4.6 2.15 functioning of cladding thermocouples as well as the 2.00 thermal response itself. The oxidation history is Dissolved or melted fuel = 1.4 1.22 correctly reproduced, although there is indication, (kg) 0.36 shown in Fig. 4, of premature onset of oxidation. Maximum temperature (K) >2500 2800 The calculated final bundle state, shown in Fig. 5, 2792 exhibits degradation throughout most of the bundle, However, some of the event timings during the initial transformation to debris, and voiding in the middle heat-up are earlier than observed, particularly for the section as material relocates downwards. Almost all of base case. The bundle power was reduced by 10% the absorber has melted, and dropped below the and the shroud conductivity increased by 20% for the active region. The experimental end-state, shown in best-estimate case, values which are close to the Fig. 6, indicates the areas of quasi-intact geometry, extremes of the experimental uncertainty range. This debris, voiding, and melt-pool formation. gave better agreement for timings, but the extent of 80
0.12 MELCOR • SDHY700 expt • SDHD701 ..
0.08
w 0.06 CO E
0.04
0.02
5000 10000 15000 20000 time (s) Fig. 4: Hydrogen generation in bundle.
Thus, the degradation and voiding have been correctly reproduced, but the extent of downwards movement of fuel has been underestimated, and the molten pool, which formed near the bottom of the bundle, was not calculated. Downward movement of absorber material was overestimated. Oxide melting appears to have enhanced strongly the bulk fuel movement which occurred towards the end of this phase, but this was not calculated as the melting point (2800 K) assumed in MELCOR was not reached. Post-test examinations of the PHEBUS bundle indicate fuel melting at lower temperatures (2750K), which may be a reason for the discrepancy.
absorber rod inner fuel rods outer fuel rods debris intact debris intact debris intact ssx ss sic sic ss ssx :rx zr uo2 uo2 zr zrx < zr uo2 uo2 zr zrx I
Fig. 6: Bundle end-state (experimental).
It can also be seen that the iodine release to the I containment is underestimated. Many of the fission products exhibited chemical behaviour that cannot be Fig: 5: Bundle end-state (calculated). modelled with MELCOR, and which affected the transport and retention. In particular, some of the The total releases of noble gases and volatile fission iodine was transported as gaseous species, while products from the bundle were in generally good some of the caesium is thought to have been in the agreement although the rates of release were form of molybdate, which was deposited in the rising overestimated, as exemplified by xenon and iodine; line. Retention of aerosols in the SG tube was see Fig. 7. overestimated. 81
• COR-DMH2-TOT showed sizeable discrepancies. There are several
- Xe Expt reasons to expect quantitative differences, notably the simplified and often very approximate treatment of the physical processes, the coarse noding, the experimental uncertainty, and perhaps also the difficulty
60 in identifying an observed event or quantity with its MELCOR counterpart. Agreement was generally at least fair and within acceptable bounds, and the 40 reasons for the discrepancies were mostly traceable to known limitations in the modelling. Thus, the character of the calculated sequence was not sensitive to the 20 input model choices and parameters.
8000 10000 12000 14000 16000 18000 20000 time (s)
Fig. 7: Fractional release of xenon and iodine.
The containment pressure was mainly determined by the steam inlet flow and condensation rates. Figure 8 shows the good agreement achieved, despite the simplicity of the models. Agreement for the aerosol deposition, Fig. 9, was also quite good, bearing in mind that the total mass of aerosols transported to the containment is underestimated by about a factor two. 10000 20000 30000 40000 Time (s)
2.5 Fig. 9: Aerosol mass distribution in containment.
2.0 The calculations generally conserved mass and energy in the thermal-hydraulics, material movement a) except for an incompatibility between the core and the Ql radionuclide packages. The released structural and absorber material is ignored in the accounting of core • Total MELCOR » Noncondensable material inventory, but this is only significant for 0.5 ••• Steam • Total Expt materials undergoing large fractional release (e.g. » Noncondensable • Steam cadmium). 0.0 —I— 5000 10000 15000 20000 25000 30000 The premature calculation in the base case of many of Time (s) the events times, and most clearly the onset of oxidation, suggests that the nominal power might be Fig. 8: Total and partial pressures in containment. too large, or the heat transfer from the fuel might be too small. The best-estimate simulation gave This is due partly to the overestimate of the retention in improved agreement for the timings, and for the fuel the circuit, and partly because certain released rod and shroud temperatures, by virtue of reduced net materials, notably rhenium from thermocouples, which heating, but at the cost of worse agreement for the are specific to the PHEBUS facility, were not included in extent of degradation in the later period. As an the calculation. The lower concentration led to less alternative, increasing the radial radiation exchange agglomeration, and hence a slower settling. There was factors in the bundle from the default of 0.25 to 0.75 good agreement for deposition on the wet condenser, (values which probably represent realistic bounds) as this was not affected by the agglomeration. caused more heat to be transferred from the inner rods to the shroud. This gave results similar to the best-estimate case, with slightly better agreement for 5 DISCUSSION OF UNCERTAINTIES temperatures of the inner fuel rods but worse for the All of the calculations gave good qualitative agreement shroud temperatures. The uncertainty band for the with FPT-1 data, correctly reproduced the integral modelling of heat balance appears to roughly coincide character of the sequence, the causal chain of events, with the experimental uncertainty. and most of the main phenomena. Agreement for the magnitudes and timings was mixed — some The onset of oxidation was also too early in all of the parameters were calculated accurately, while others calculations, including the best-estimate case, in 82 which the time of the peak was accurately calculated. overestimate is broadly similar to that calculated by A possible reason might be that the default Urbanic- other codes, for example VICTORIA [12], and appears Heidrick (U-H) oxidation model possibly overestimates to be generic to 1-D models, perhaps due to the the rate in the lower temperature range, even though neglect of 2-D effects. MELCOR did not calculate the the total generation is correct. A case was run using observed retention of caesium in the hot leg, as it did the recently recommended Leistikow [10] and Prater- not calculate the correct chemical species. However, Courtright correlations [11] in the low (<1800 K) and the overall impact of chemical speciation of fission high (>1900 K) temperature ranges, respectively. The product retention appears to be minor when compared Leistikow correlation (slower kinetics than U-H) gave with uncertainties in release from the fuel and aerosol improved agreement for the time at the onset of deposition. Speciation is likely to be more important in oxidation, but the Prater-Courtright correlation (faster predictions of the chemical behaviour in the than U-H) overestimated the oxidation rate, leading to containment. an even earlier estimate of the time of the peak, and premature end of the main oxidation excursion; this The single-node containment model gave good implied that the total hydrogen generation was agreement for the pressure and temperature histories, underestimated. The U-H oxidation model is but it was necessary to specify the initial and questionable in the lower temperature range, but boundary conditions accurately, as the calculated appears to give good results overall. Modelling of values proved to be rather sensitive to the model zircaloy oxidation at the time of metallic melting and inputs. The base case overestimated the pressure, movement is problematic, and beyond the scope of but the discrepancy was small due to compensating MELCOR. errors in the input; these were corrected in the best- estimate model. Condensation and associated aerosol Uncertainties in the heat balance might also have deposition agreed well with experiment, and the small contributed to the tendency to underestimate the deposition on the vertical walls was in qualitative degradation, but this cannot be definitely concluded agreement with experiment. The mass transported to from the present analyses. PHEBUS, among other the containment was, in most cases, lower than programmes, shows that the processes that control experiment, mainly due to the neglect of released debris formation, bulk relocation and melting have not material from instruments in the bundle, and yet been clearly identified, and so must be considered overestimate of the retention in the circuit. This low a continuing uncertainty. However, it should be aerosol concentration in the containment possibly possible to vary the relevant parameters in MELCOR. contributed to a slight underestimate of the agglomeration rate, the mean size and the settling The fission-product releases calculated by the rate, but additional reasons might have been CORSOR-M model depended, of course, on the limitations in modelling the agglomeration processes. release parameters, but were fairly insensitive to other The number of size classes was probably too small to aspects of the model. Despite good agreement for properly resolve the size distribution. The single-node the total release of volatile fission products, the model appears to be adequate for the open volume of release rates were overestimated by a factor of about the PHEBUS containment. three. Releases of the U02 and the less volatile fission products were underestimated if the default 6 CONCLUSIONS model was used, and the timing of release remained in poor agreement against measurement, even if the MELCOR successfully calculated Phases 1,2 and 3 of total release was acceptable. The release of several the PHEBUS FPT-1 experiment. The character of the of the fission products depends on chemical sequence, and all the main features, were reproduced processes which are not well understood. At present, using mainly default or neutral model options, and the release parameters must be treated as uncertain with noding typical of a plant model. Good agreement until additional data are available. MELCOR provides was obtained for the bundle thermal response, no model for the release of structural and absorber hydrogen generation, and the containment thermal- material, and this should be remedied. These releases hydraulic behaviour, while most other aspects of the could only be calculated in an artificial and sequence were at least in fair agreement with unsatisfactory manner, and were highly sensitive to all experimental data. The simplicity of the physical aspects of the calculation. models, the coarse noding, together with experimental uncertainties, all resulted in some quantitative and Aerosol retention in the circuit was overestimated by timing discrepancies, which could be ameliorated, all the calculations, and was not significantly alleviated though not completely eliminated, by tuning of input by changing the assumed aerosol characteristics. parameters. There was some overestimate in the gas temperatures in the vertical line from the bundle and The code ran in a robust manner with no run-time, the hot-leg, which might have caused an overestimate numerical or instability problems. The coarse noding of the aerosol deposition in those locations, but this meant that calculations could be performed with does not appear to be a major factor. Most of the modest computing resources, so that additional trial calculated retention occurred in the first metre or so of and sensitivity cases did not impose a major burden. the steam generator tube, where the calculated The calculation was generally robust to changes of temperatures agreed well with the data. The compiler and machine, although some of the more 83 sensitive variables were slightly affected. Run-time REFERENCES performance and robustness assume importance in [1] P. Von der Hardt et al., "Nuclear Safety plant applications, for which sensitivity studies are Research: The PHEBUS FP Severe Accident necessary to address uncertainties Experimental Programme", Nucl. Safety, 35, 2 Knowledge of the results was used only to a limited (1995). extent in defining the input, notably in modelling [2] H. Scheurer, B. Clement, UPHEBUS FP Data release from the bundle to provide a more-or-less Book FPT-1", Document PH-PF IS/92/49 (1997). correct source for transport through the circuit. However, prior knowledge of the experimental [3] S. Bourdon et al., "Final Test Protocol for behaviour did help to identify certain input errors, and PHEBUS FP Test FPT-1", Document PH-PF highlighted dependences that were not previously IS/95/259 (1995). appreciated. [4] D. Jacquemain et al., "PHEBUS PF FPT-1 Final It should be realised that PHEBUS is a simplified and Report", Document PH-PF IP/00/479 (2000). highly idealised representation of a reactor accident, [5] T.J. Haste, "Specification of International and the adequacy of the simple noding employed in Standard Problem ISP-46 (PHEBUS FPT-1)", the present application might not extrapolate to a real plant, for example because of two and three- Note Technique SEMAR 2002/5 (2002). dimensional flows in the vessel or circuit. Other [6] J.C. Birchley, R.C. Cripps, "Analysis of PHEBUS phenomena not addressed include reflooding, FPT-1 Bundle Degradation, Release and external cooling, air-ingress, hydrogen-burning, and Transport of Fission Products, and Aerosol late-phase, ex-vessel corium behaviour. The Dynamics and Iodine Chemical Behaviour in the benchmarking of MELCOR (and other tools) against Containment", PHEBUS: SW/99/002 (1999). PHEBUS FPT-1 needs to be complemented by analyses of additional experiments such as those [7] R.O. Gauntt et al., "MELCOR Computer Code being carried out in the QUENCH programme [13], Manuals: Version 1.8.5", NUREG/CR-6119 and future PHEBUS experiments. (2000). The most serious model weaknesses are in the [8] M.R. Kuhlman et al., "CORSOR Users Manual", release of fission products and other materials from NUREG/CR-4173 (1985). the core, their chemical speciation, and their retention [9] W.C. Hinds, "Aerosol Technology: Properties, in the circuit. These weaknesses are unavoidable Behaviour and Measurement of Airborne without a much more detailed treatment of the Particles", John Wiley & Sons Inc. (1982). controlling processes, but in any case are shared with other codes, and reflect limitations in current [10] S. Leistikow, G. Schanz, H. v. Berg, "Kinetik und knowledge. The resulting uncertainties have a impact Morphologie der isothermen Dampf-Oxidation on source-term assessment and at present must be von Zircaloy 4 bei 700-1300 °C\ KfK2587 (1978). quantified using data from experiments and/or [11] J.T. Prater and E.L Courtright, "Properties of detailed analyses, or via sensitivity studies. A Reactor Fuel Rod Materials at High particular limitation is the absence of a suitable model Temperatures", NUREG/CR-4891 (1987). for the release of absorber and structural material. The presence of silver in the containment has been [12] T.J. Heames et al., "VICTORIA: A Mechanistic demonstrated to strongly limit the extent of iodine Model of Radionuclide Behaviour in the Reactor volatility. Limitations also exist in the models for core Coolant System under Severe Accident degradation and melting, although their impact is less Conditions", NUREG/CR-5545 (1992). severe; additional flexibility in the code input would [13] M. Steinbruck, M. Granosa (Eds.), "Eighth help to quantify the related uncertainties. International QUENCH Workshop", Karlsruhe, Despite the above reservations, the analysis of Germany, October 2002. PHEBUS FPT-1 has incresed confidence in the adequacy of the MELCOR models, their integration in the code, the specification of input, and the practicalities of actually running the code. 85 THE FUJI PROJECT: FUEL TEST FOR JNC AND PSI - A JNC-PSI-NRG COLLABORATION Ch. Hellwig, P. Heimgartner, F. Ingold The FUJI project Is especially dedicated to the early-in-life restructuring of two types of particle fuel — sphere-pac and vipac — in comparison to pellet fuel. Three types of sphere-pac fuel segments prepared by PSI will be irradiated in the High Flux Reactor (HFR) in Petten (NL), together with pellet-type fuel segments and vipac fuel segments (non-spherical, fuel-shard-packed segments), prepared in accordance with JNC specifications at PSI under the same irradiation conditions. MOX fuel, with 20% Pu and an oxygen-to-metal ratio of 1.97 (typical for fast reactor fuel), is currently being produced in different fuel shapes (pellets, microspheres and vipac particles). For the pellet and vipac fuel, the dry attrition mill method was used to prepare the MOX powder prior to pressing. Microspheres were produced following the internal gelation process, with nitrate solutions as the starting material. The pellet and the vipac fabrication have now been completed, although some analysis results are still missing. The special production route of vipac has led to particles of beneficial shape; the microsphere fabrication is underway. The preparation for the filling and welding of the segments have been completed, and production will take place before the middle of 2003. 1 INTRODUCTION • Compare fuel performance for sphere-pac, pellet and vipac type fuels under the same The use of low-decontaminated Pu for the fabrication irradiation conditions of MOX for fast reactors offers significant cost benefits. Due to the high radiation, remote-handling • Study the restructuring of sphere-pac fuel procedures are required for the fuel and pin from start-up conditions to melting fabrication. The same is true if minor actinides (to be temperatures burned in the fast spectrum) are added to the fuel. • Study the behaviour of MA and plutonium Particle fuels are therefore candidates for achieving a burning under reactor irradiation conditions simple and remote procedure for manufacturing of with a radial temperature profile of the fuel pin advanced fuel for the burning in a fast reactor of minor close to that expected in a fast neutron actinides (MA) and low-decontaminated plutonium. spectrum Additionally, particles can be produced via a wet- • Study plutonium and MA migration and chemical route, and so dust production can be greatly restructuring behaviour at high linear heat reduced in comparison to dry pellet fabrication. Two ratings in a fuel irradiation experiment with a types of particle fuel are used today: sphere-pac and radial temperature profile close to that vipac. expected in a fast neutron flux • Exchange information to keep open the option The irradiation performance of these fuel types must of a later transfer of the sphere fabrication be carefully investigated and compared against pellet process and pin-filling technology. fuel. Today, there exists only sparse technical The content and goals of the project, and the present knowledge on the fabrication and irradiation behaviour status and results of the work performed, are of sphere-pac fuels containing MA and high summarized in this report. concentrations of plutonium in the form of uranium and plutonium oxides (MOX). The same is true for vipac fuel. 2 EXPERIMENTAL The irradiation matrix In general, and contrary to pellet fuel, pins comprised of particle fuel do not have gaps. This is of advantage Three types of sphere-pac fuel segments, each thermally, but is more than compensated by the lower prepared by PSI, will be irradiated in the High Flux thermal conductivity resulting from the small area of Reactor (HFR) at Petten (NL), together with pellet- solid contact between adjacent particles. However, type fuel segments and vipac fuel segments (non- this area increases rapidly during irradiation due to spherical fuel, shard-packed), prepared in accordance different sintering mechanisms, leading to a steadily with JNC specifications at PSI under the same increasing thermal conductivity of the fuel, and irradiation conditions. therefore lower central fuel temperatures. It is obvious that this sintering behaviour, which is distinct at the A total of 16 fuel segments (four series of four beginning of irradiation, is crucial for the safety segments each) will be irradiated in the pool-side evaluation of the fuel. The FUJI project focuses on the facility of the HFR. In this facility, fuel segments (two early-in-life sintering and restructuring of the two types parallel rods) can be moved towards the reactor core of particle fuel: sphere-pac and vipac [1]. The on a trolley. As a result of this movement, any power objectives of the FUJI project are as follows: history can be applied to the test rods independently of the reactor operation. To increase the fuel testing possibilities, PSI will produce segmented rods with 86 two segments forming one test rod. The connection • The "power-to-melt test" (maximum power between the two segments will be placed in the centre 900 W/cm, duration 85 hours) imitates first the plane of the reactor. Thus, four segments (two start ramp and a subsequent holding time of segmented test rods) can be irradiated in parallel in 48 hours (the same as the "first restructuring one operation. The irradiations focus on the following test"), but then the power is increased features: stepwise until fuel-centre melting is observed. • The "initial sintering test" (maximum power The fuel types to be irradiated in the different tests are 540 W/cm, duration 36 hours) imitates the listed in Table 1. In addition to MOX fuel with 20% Pu, start ramp of the fast reactor JOYO; i.e. the MOX with 20% Pu and 5% Np will be produced and power increases linearly from zero to full irradiated. Sphere-pac segments will be produced with power in 36 hours two different filling techniques, leading to different • The "first restructuring test" (maximum power smear densities (the above-mentioned maximum 540 W/cm, duration 84 hours) again imitates power is given for the maximum axial power and fuel the start ramp, but with a subsequent holding with the highest density). time of 48 hours • The "second restructuring test" (maximum After irradiation, PIE is foreseen in the NRG Hotcells power 540 W/cm, duration 132 hours) imitates at Petten; some samples will be transported to PSI for the start ramp with a subsequent holding time burnup analysis. of 96 hours Table 1: Overview of the different fuel types to be irradiated in the FUJI tests. Initial Sintering 1st Restructuring 2nd Restructuring PTM Fuel type MOX MOX MOX MOX MOX MOX MOX MOX Fuel form Pellet Sphere Pellet Sphere Pellet Sphere Pellet Sphere pac pac pac pac Smear density 89.4% 78-81% 89.4% 78-81% 89.4% 78-81% 89.4% 78-81% Upper segments Lower segments Fuel type MOX MOX MOX Np-MOX MOX MOX MOX Np-MOX Fuel form Sphere Vipac Sphere Sphere Sphere Vipac Pellet Sphere pac pac pac pac pac Smear density 78-81% 75-78% 78-81% 78-81% 72-73% 75-78% 89.4% 78-81% 3 FUEL FABRICATION wall, agglomeration of powder at the edge of the lower part of the mill jar, and rapid fall-down of the powder. MOX fuel with 20% Pu and an oxygen-to-metal ratio The mill jar has now been separated by a steel grid of 1.97 (typical for fast reactor fuel) is currently being into upper and lower parts. The shaft has various produced at PSI: Pu0 of one single batch P-18 (PSI- 2 kinds of paddles, with specially-designed shapes for designation, Pu =88.3%), and depleted U0 from fiss 2 each part. The shapes and arrangement of the BNFL, have both been used. paddles are optimised so that the powder stays for a For pellet and vipac fabrication, the weighed amounts longer time within the milling media (balls made of of each oxide powder (Pu02 and U02) were mixed in Ystab-Zr02). Once the powder has passed through the a Turbula™ mixer for 4 hours, and then milled with an upper stage, it is continuously milled in the lower improved two-stage attrition mill developed by stage, where the milling operation is refined due to the KAERI1. The mill is an improvement over the smaller milling media and a different shape of paddle. conventional, batch-type dry attrition mill [2] in regard The attrition mill turns at about 150 rpm, which gives to its efficiency, and to problems often encountered the total length of time for one passage of powder during operation, such as powder sticking at the inner through the mill of about 2-5 minutes, depending on its properties. Several passages are applied to one powder batch to ensure homogeneity and optimal 1 Korean Atomic Energy Research Institute powder properties. 87 For pellet fabrication, the milled powder is before transfer to the welding box. There, the end plug subsequently pre-compacted and granulated to welding will be performed, after the internals (plenum achieve good fluidity. The granules are pressed to spring, particle seals, etc.) have been added to the pellets, and then sintered. In order to achieve the segment, and the whole filled with He. Finally, the desired O/M ratio, the gas flow of the reducing gas segments will be decontaminated, and then examined (N2+8%H2) can be changed, and vapour can be using He-leak and X-ray testing. The particle-fuel added. segments will additionally be scanned to ensure a For vipac fabrication, the milled powder is also pre- homogeneous density distribution using gamma rays. compacted and granulated. The green vipac particles The filling procedures for each different fuel type and are sieved in different fractions, and then sintered in a welding procedure had to be developed from new, but reducing atmosphere. are now ready to be applied. The microsphere fabrication follows the internal gelation process, with nitrate solutions as starting material [3]. All the solutions are prepared by mixing a nitrate solution of each element (U, Pu, or Np) composing the final product. The starting materials were: Pu(N03)4+x, prepared by dissolution of Pu02 (from PSI-Batch P-18) in nitric acid, with the addition of some HF, and subsequent concentration up to nearly pure nitrate in a vacuum distillation. U02(N03)2-y, prepared by dissolution of U02 (depleted U02 from BNFL) in nitric acid, and subsequent concentration up to the substoichio- metric nitrate in a vacuum distillation Np(N03)4+x, prepared by dissolution of Np02 (from 2 mm 7PN-T006-19 Obninsk, Russia) in nitric acid, with the addition of 21.01.2003 HP43 some HF, and subsequent concentration up to nearly pure nitrate in a vacuum distillation. A mixed-feed solution, with 0.80 mol/kg metal content, Fig. 1: Unetched overview of a MOX pellet; the pore a NO^Me ratio of 1.90 mol/mol, an HMTA/Me ratio of distribution is almost homogeneous. 1.30 mol/mol, and a urea/Me ratio of 1.30 mol/mol, was used for the subsequent gelation. The solution was mixed at a temperature of 0°C, and then driven though a nozzle of 700 jum diameter; the throughput was roughly 4.5 g/min. The drops from the feed * solution fell into silicon oil at about 105°C, and then gelled to spheres. The spheres were then washed in a vibro-bed column in an NH4OH solution. No broken •„•* . , - i-i^r ivijf-.** * * *>• . spheres could be observed after the washing stage. The spheres were dried in flowing air in two stages, at V • * >*. 'V, •# • y.,• vA ,< > >• -- 75°C and 110°C, converted into oxide in a calcination stage in an Ar/7%H2 gas stream at 600°C, and finally sintered at 1400°C in a reducing atmosphere A ft-" > j « » 4 (N2/8%H2) to reach the targeted O/M ratio of 1.97. The 0.02 mm ,";.. P f, - V f- • ^ 7pn-t006-1 ,,, " * - ~ , ••, batch sizes were of the order of 100 g. pm43719.12.2002 ?*, <; * ^ — & Characterisation of the sintered spheres focused on the O/M ratio, the density measurement, X-ray Fig. 2: Thermally etched MOX-pellet microspheres. diffraction, and chemical composition. Dense oxide particles were obtained utilising a thermal treatment in For the filling of the sphere-pac segments, fabrication a controlled atmosphere. parameters from former experiments were adjusted to the present case using U0 -microsphere filling tests. 4 SEGMENT FABRICATION 2 Two types of filling operation have to be distinguished: For the fabrication of the segments, the dedicated parallel filling and infiltration filling. For parallel filling, equipment in PSI's Hot Lab will be used. The segment size fractions (800 jum spheres and 190 jum spheres) to be filled will be connected to the bottom of a special are filled into the segment in parallel, under moderate "filling box". The fuel will then be filled into the vibration. Then, the fuel column is fixed, and stronger segment from above. If required (for vipac and vibrations are applied. For infiltration filling, first the sphere-pac fuels), the segment can be vibrated from coarse fraction (800 jum) is filled into the pin under below. The segment will then be temporarily closed moderate vibration, then the fuel column is fixed with 88 a sieve that allows the fine fraction (70 |nm) to the MOX granules was necessary. Results are in line penetrate. The fine fraction is added on top of the fuel with an independent comparison made of the segment, and the infiltration is forced by vibration. measured tap density (filling density of the granules in a glass cylinder after tapping in a defined manner). For vipac filling, well-defined amounts of six different size fractions have to be poured homogeneously into the segment; then the column is vibrated. 5 RESULTS Pellet fabrication All pellets for the irradiation have been produced in three batches. A pre-compaction pressure of 15 MPa, and a compaction pressure of 400 MPa, were used for the fabrication of the pellets. Sintering took place over 20 hours at 1600°C in a reducing atmosphere. It was found necessary to increase the gas flow of the reducing sintering gas to 180 l/min to achieve the desired low O/M ratio. The diameter of the pellets after sintering was 6.85 mm, their height was 9.9 mm, and the diameter after centreless grinding was 6.504 mm. The dimensions are therefore well within the specifications. The quality of the majority of the pellets was excellent, showing no cracks or flaws on the surface. The determined O/M ratio was 1.967 (3 pellets), and the geometrical density was determined at 95.3±0.2%. Ceramography shows a homogeneous distribution of the porosity (see Fig. 1). Thermal etching was performed, which revealed a grain size of roughly 2 |iim (see Fig. 2). (Though this grain size is rather small, it could be enhanced by a modified sintering procedure. However, for this project, no modification was performed, due the irrelevance of small grain size for the foreseen, short-time irradiation tests). X-ray diffraction and alpha-autoradiography, as well as chemical analysis, are currently underway. Vipac fabrication After evaluating the proposed fabrication route by sintering green U02-granules, tests with MOX fuel were successfully performed; the pre-compaction pressure was 550 MPa. The granulator sieve had to be adjusted: the size of the holes was increased from 1.0 mm to 1.2 mm in order to fabricate large granules. In order to increase the grain size, vapour was added 0 5 mm to the atmosphere during the heat-up phase and the ! 7PN-V001-6 ' 0.045-0.105 mm first five hours of sintering. To reduce the O/M ratio, 58 -02015-19 , 07.06.2002 HP43 the fuel was then sintered in a reducing atmosphere without vapour for 15 hours. The shape, i.e. the Fig. 3: MOX vipac particles of different size fractions circularity, of the particles produced was compared produced at PSI. with Zr02 granules used for the inactive segment filling tests (see also Fig. 3). This is an important To reach a high smear density, a well-defined parameter, since the circularity of the particles will combination of different particle sizes is required [4,5]. affect the achievable smear density. Three types of The optimal range and distribution of vipac particle Zr0 granules, with different circularity, were used for 2 sizes for segment filling is listed in Table 2; the the inactive segment filling tests: untreated granules particle-size distribution of the fabrication test is also produced from sintered material, granules ball-milled given in Table 2. To make better use of the material for 24 h, and granules ball-milled for 40 h. The produced, granules were sieved after the first circularity of the MOX particles produced was fabrication step, and the fraction with sizes between comparable to the second type, i.e. those granules 600 |nm and 250 jum was compacted, and again ball-milled for 24h, and which performed well in the granulated. The results from the fabrication have been inactive filling tests, and thus no further treatment of added to Table 2. The medium-size granules will now 89 be manually crushed in a mortar to increase the even closer adjustment of the derived particle-size amount of the fine-size fraction. This will lead to an distribution to that required Table 2: Range and distribution of vipac particle size for segment filling. F1 F2 F3 F4 F5 F6 Particle range (|jm) 850-600 600-425 425-250 250-106 106-45 45-25 for optimal segment filling 50.0 9.9 4.6 4.6 18.0 13.0 Weight ratio (%) Derived from MOX fabrication test 28.2 25.3 10.0 11.6 13.8 11.0 Derived from MOX vipac fabrication 37.6 21.4 8.0 18.2 9.2 5.5 Microsphere fabrication were made slightly smaller: i.e. about 750 jum instead Five types of spheres have to be fabricated for the of 800 |nm diameter. project: 800 jum diameter MOX spheres, with and For vipac filling, the homogeneous mixing and filling of without Np, 190 |nm diameter MOX spheres without the six different size fractions is nearly impossible. Np, and 70 jum diameter MOX spheres with and Good improvement can be achieved if the two without Np. At the time of writing, 800 jum size MOX- coarser-sized fractions, and the four finer-sized spheres without Np have been produced. fractions, are first mixed, and then fed into the segment in parallel, as is done for parallel-sphere filling. For industrial fabrication, a feeder with a continuously working balance for each size fraction would be the optimum for homogeneous filling. Nevertheless, the PSI method does produce filled segments of acceptable homogeneity. For the vibration procedure, a varying frequency was found to be advantageous, provided the resonance frequency of the system was avoided. 6 DISCUSSION AND OUTLOOK After the delay in the refurbishment of the Hot Lab at PSI, and the subsequent revision of time schedules, the work is now proceeding rapidly. The delivery of the segments is planned to take place before the end Fig. 4: Sintered spheres placed on a flat glass plate, of June 2003. The irradiation will then be performed in seen from below. two series: in autumn 2003, and during 2004. Post Irradiation Examination (PIE) will be performed Figure 4 shows sintered spheres placed on a flat thereafter, and anyway before the end of 2004. The glass, seen from below. The sphere size was project will be completed with the final report delivered determined at 750 jum. This is lower than the original by mid-2005. target value of 800 jum, but still within the specifications. For reasons mentioned below, it was Good coordination and smooth collaboration between decided to use the 750 jum spheres for the segment the different working groups involved in the project — filling. The O/M ratio was determined at 1.93. This is the analytical group, the actinide chemistry group, the clearly too low, and must be adjusted by further fuel performance group of the Laboratory for Materials sintering runs with a less reducing atmosphere: i.e. by Behaviour, and the Technical Laboratory (for the X- adding some water vapour to the N2/8%H2 gas. ray work) — is necessary to achieve the envisaged delivery date. Questions relating to transport are being Segment fabrication addressed, offers from different transport companies have been received, and tenders will be given out For parallel filling, the fabrication tests showed that soon. long filling times would be of advantage in better controlling the simultaneous completion of the feeding for both size fractions. Contrary to this, the homogeneous flow out of the two feeders, for both fractions, requires a minimum of throughput. The optimum was found to be 25 seconds. For the subsequent vibration, a spring-like fixing device was used to keep the fuel column down. For infiltration filling, the fabrication tests indicated that a higher smear density could be reached if the larger spheres 90 REFERENCES [4] H. Steinkopff, R. Krompass, O.V. Skiba, [1] K. Bakker, H. Thesingh, T. Ozawa, Y.Shigetome, "Automatic refabrication and quality control of fuel S. Kono, H. Endo, Ch. Hellwig, P.Heimgartner, elements for the BOR-6O reactor in the USSR", J. F. Ingold, H, Wallin "Innovative MOX fuel for fast Nucl. Mater., 178, 163 (1991). reactor applications", Proceedings of the ARWIF 2001 Conference, Chester, UK, 22-24 October [5] R. Herbig, K. Rudolph, B. Lindau, O.V. Skiba, 2001, Paper 3.2 (CD-ROM). A.A. Maershin, "Vibrocompacted fuel for the liquid metal reactor BOR-6O", J. Nucl. Mater., [2] H.M. MacLeod and G. Yates, "Development of 204, 93 (1993). mixed-oxide fuel manufacture in the United Kingdom and the influence of fuel characteristics on irradiation performance", Nucl. Technol., 102, 3(1993). [3] G. Ledergerber, F. Ingold, R.W. Stratton, H.P. Alder, C. Prunier; D. Warin, M. Bauer, "Preparation of transuranium fuel and target materials for the transmutation of actinides by gel coconversion", Nucl. Technol., 114, 194 (1996). 91 FRACTURE TOUGHNESS OF ZIRCALOY CLADDINGS J. Bertsch, W. Hoffelner Zirconium-based alloys (Zircaloy) have been used as cladding material in Light Water Reactors for many years. During fabrication, or in in-reactor service, crack-type defects can be formed, posing questions regarding mechanical integrity. As claddings change their mechanical properties (mainly toughness) during service as a result of irradiation-induced degradation, oxidation and hydride formation, it is essential for integrity considerations to provide parameters for the assessment of the influence of flaws on rupture behaviour. Usually, fracture-mechanics parameters are employed such as the fracture toughness, K,c, or, for high plastic strains, the J-integral, JiC. The applicability of these parameters is, however, limited by the dimensions of the samples (e.g. thickness). In claddings with a wall thickness of below 1 mm, determination of toughness necessitates an extension of the J-integral concept. A method based on the traditional J-approach, but applicable to thin-walled structures, is presented in this paper. 1 INTRODUCTION B > 2.5(KlC/ay)2 Zircaloy components are fabricated to be "defect free". in which ay is the material yield strength at the In practice, however, these components, in common temperature and loading rate of the test. For a roughly with all engineered components, can have defects in expected Kic = 55 MPa m1/2, and a yield stress of 600 the form of small cracks. Such cracks can originate MPa for irradiated Zircaloy at 300°C, the component either during fabrication or during in-reactor service. or specimen thickness would need to exceed about The following questions arise: "How easily will the 21 mm. This is, of course, many times the thickness of crack propagate?", and "Will the crack compromise the cladding, grid, channel and other components. For the safe operation of the component?" Answers to unirradiated Zircaloy, the required thickness is even such questions falls within the realm of fracture larger. mechanics. To accommodate soft material, like Zircaloy, another Extensive employment of fracture mechanics to the approach, called the J-integral method, is frequently issue of cracking in steel and other high-strength used. Physically, the J-integral is defined as the alloys has resulted in increased safety in pressure energy release rate (the dissipation of energy) that is vessels, aircraft engines, building structures, and in calculated by taking a line integral around a path many other commercial and military applications. enclosing a crack tip. It is strictly valid only for the However, in the past, use of fracture-mechanics case in which the crack grows in an elastic material. technology to reactor Zircaloy issues has been rather Its use, however, has been extended [6,7] to include limited. This is partly due to a lack of regulatory elasto-plastic materials, such as Zircaloy, in which emphasis on cladding failure as a safety issue [1], but considerable plastic deformation can occur. As also to the fact that, as explained below, much of the commonly used, "J" is related to the amount of work standard fracture-mechanics methodology does not (dissipative energy, both elastic and plastic) per unit apply to standard PWR- and BWR-bundle component crack surface area required to extend the crack. geometries. Nonetheless, since higher burnup increases the probability of the reactor being operated But, even the less stringent size requirements for the with defects present in the materials, and as new applicability of the J-lntegral are not met theoretically zirconium alloys are introduced, the use of fracture by thin-walled claddings. In the framework of an mechanics to predict the behaviour of cracks or international collaboration within the EPRI NFIR defects is increasing. Indeed, papers at recent project [8], different laboratories are attempting to find international conferences have illustrated fracture a J-type approach to Zircaloy claddings with the aim mechanics techniques for analyzing crack propagation of defining mechanical quantities which allow a in failed Zircaloy tubing [2, 3, 4], and, for many years, prediction of the fracture behaviour of Zircaloy leak-before-break criteria and critical crack lengths in claddings with different toughness. In analogy to the CANDU-type pressure tubes have been analyzed with conventional fracture mechanics approach, such considerable success using fracture-toughness quantities should be, as far as possible, independent methodology [5]. of specimen geometry, in order to allow their application to components and realistic crack The problem for LWR claddings is that the strict geometries to be viable. Consequently, different methodology of Linear-Elastic Fracture Mechanics specimen geometries have been analyzed in the (LEFM) usually does not apply to geometries relevant different laboratories. to reactor components (particularly in regard to the thickness). In essence, the plastic-zone sizes of 2 EXPERIMENTAL zirconium alloys are too large; in other words, the materials are too soft. In order to satisfy the criteria of The experiments were carried out with samples ASTM Standard E399 for valid K|C determination, the fabricated from the aluminum alloy AI-7050 and cold- thickness of the tested member, B (see Fig.1), must worked (CW) Zircaloy 4. Both materials were supplied satisfy the criterion: 92 by one source (EPRI) in order to produce well-defined The two segments on the top fit the inner diameter of comparisons. the sample, and can be slipped into it. The grips are machined as counterparts to the segments. This Our approach was based on the following criteria: assembly allows the application of tangentially • In order to have a well-defined starting point, oriented stresses to the cladding sample. we chose a specimen geometry which is well characterized under valid K and J conditions The testing procedure is similar to standard fracture- mechanics procedures. The notched sample is • With respect to later testing of service- fatigue-loaded until a fatigue crack starts to grow from exposed cladding material, the geometry the root of the notch. This procedure is necessary in should also allow easy manipulator handling order to have a well-defined starting point for further • All testing parameters reported for established measurements. The sample is uni-directionally K and C testing should be measurable. stressed thereafter, and the propagation of the crack is monitored as a function of load and displacement in Fracture mechanics samples can be of bending-type the load line. or of tension-type. Typical bending-type samples are the CT-specimen and the SEN-bending sample [9]. Well-known, tension-type samples are the single- or double-edge notched tension (SENT, DENT) samples, or the centre-notched panel (CN). Because of the reasons mentioned previously, and because of the fact that, in the other laboratories, bending-type samples were employed, we have chosen a pipe-type version of the DENT sample; this is shown in Fig. 1. Two machined notches act as starting points for fatigue cracks created by cyclic loading. ^ • a/2 Fig. 3: Stage of stable crack growth (Zircaloy 4, cold- // \\ \ worked) at room temperature. The dark parts •.^""•crc of the cracks represent the starter notch and the fatigue pre-crack, respectively. The white contours belong to the cracks growing under uni-directional loading. w The different areas become evident from Fig. 3. The Width W = 12 mm starter notches and the fatigue cracks both appear Wall Thickness B = 0.6 mm dark, whereas the cracks propagating under uni- Crack Length a = 4 mm at beginning of directional load appear white. The experiments were experiment carried out in an electrically driven machine; crack length was measured optically. For thin-walled Fig. 1: Geometry and typical dimensions of the samples, such as those used here, this is a valid Double-Edge Notched Tension Sample approach, since the crack front propagates rather (DENT). straight, and so the crack length measured at the surface correlates well with the crack length in the interior of the sample. The displacement in the load line is determined optically from the movement of the two edges of the starter notch. 3 EVALUATION OF DATA The sequence of Figures 4a-4d illustrates the different stages in the development of a crack during stable crack-growth conditions. Typical load-displacement curves for aluminum and Zircaloy are shown in Fig. 5. The scatter of the data still requires further investigation. Reasons might be the under-estimation of bending moments for the Fig. 2: Gripping section. stress calculations, and the fact that decreasing The gripping sections, which are used to apply a specimen size usually increases the scatter bands mechanical load to the sample, are shown in Fig. 2. because of the relative importance of the behaviour of single grains. 93 Je=K2/E' in which E' is the Young's modulus for plane stress; E' = E(1-v2) for plane strain. The stress intensity factor K is related to the stress intensity function f(a/W) according to / a \ K = p / yfltW VWy Fig. 4a: Appearance just before starting of stable where W is the specimen width (see Fig. 1), t is the crack growth; only fatigue cracks are visible. specimen wall thickness, f(aAA/) is the dimensionless stress intensity function, and P is the applied load. The plastic J-integral for the DENT specimen is evaluated from Ejl J» = 'If 2 tb in which Up is the area under the plastic part of the load deflection curve, b = W-a, and rip is a dimensionless correction factor; JD is calculated incrementally taking into account the continuous crack growth during loading. Fig. 4b: Appearance of cracks just after stable crack growth started (white crack contrast). Load - Displacement Curves Comparison Aluminum /Zircaloy 4'000 Displacement a [mm] Fig. 4c: Appearance of cracks just before final rupture. Fig. 5: Load-displacement curves for Zircaloy claddings; values for aluminum samples with the same geometry are shown for comparison. We have adapted this procedure to our modified DENT sample. For different crack lengths, a, the corresponding J-values are plotted in Figs. 6 and 7, showing an evaluation of tests performed with aluminum and Zircaloy, respectively. These so-called crack-resistance curves, or J-R curves, can be fitted by a power law. The designation Jx refers to the J- Fig. 4d: Appearance of the cladding just after final integral determined at a crack propagation of x mm; rupture. Jo.2 is a well-accepted reference value for comparison of different experiments. For reference, J0.i5 and J15 are also usually indicated; valid, stable crack-growth The J-integral is calculated as a super-position of conditions are assumed between these two values. elastic and plastic components: The slope of the J-R-curve (often referred to as the J = Je + Jp tearing modulus) provides another measure for the toughness behaviour of the material. Low tearing The elastic component J is computed from the elastic modulus corresponds to brittle materials, whereas a stress intensity Kaccording to high tearing modulus usually represents a tough material. 94 J-R-curve (Aluminum) This is the J-value corresponding to the maximum load in the load-displacement curve. In our opinion, this value depends on the sample geometry, because the measured load-displacement response can be different for different geometries. Kjmax" This is a K-value calculated according to the relation K = VJ -E' , where E'=E for plane-stress conditions, and E' = E(1-v2) for plane strain-conditions, valid 0.5 1.0 1.5 2.0 exactly only for linear-elastic conditions. In highly crack extension a a [mm] plastic situations, any formal link to an elastic parameter can give misleading results. J 0.2- Fig. 6: J-resistance curve determined with modified This value represents the J-integral at 0.2 mm crack DENT samples at room temperature for extension. It can be regarded as a rather sensitive aluminium. measure for the onset of crack growth, and is therefore considered as one possible measure for the toughness of thin-walled samples. J-R-curve (Zircaloy) dJ/da: 300 The slope of the J-R curve also characterizes the toughness of the material, and could be used as an additional measure of it. J - R curves Comparison Aluminum/Zircaloy 0.5 1.0 1.5 E 300 crack extension a a [mm] Fig. 7: J-resistance curve determined with modified DENT samples at room temperature for Zircaloy. 1.0 1.5 2.0 2.5 crack extension a a [mm] 4.4 DISCUSSION OF RESULTS One of the most interesting questions for any fracture- Fig. 8: J-R-curves for Zircaloy and aluminum at room mechanics-based approach is the comparability of the temperature. data gathered with different specimen geometries, and under different loading conditions. Only if (within a A comparison of Zircaloy and aluminum in terms of J- usual scatter band) comparable results are obtained R curves is presented in Fig. 8. The behaviour of the for different specimen geometries, is it considered that materials is as expected. The tougher Zircaloy has a the approach can be applied to actual reactor higher J 0.2 value, as well as a higher tearing modulus. components. Under the conditions of linear elastic A summary of the average values determined is given fracture mechanics, this is relatively straightforward in Table 1. because the fracture toughness K|C, if determined under valid conditions, gives such a geometry- Table 1: Summary of the results (average values) for independent quantity. For conditions which are Zircaloy and aluminium. theoretically outside the validity criteria, it is not a simple matter to find such a quantity. Aluminum Zircaloy With respect to thin-walled claddings, the following J 0.2 34.6 kN/m 75.3 kN/m values for the characterization of the toughness are Tearing 55 MPa 225 MPa used. modulus at J0.2 95 These values compare well with measurements on [3] V. Grigoriev, B. Josefsson, B. Rosborg, "Fracture the same materials performed with different Toughness of Zircaloy Cladding Tubes", techniques in other laboratories [8], verifying that it is Zirconium in the Nuclear Industry: Eleventh indeed possible to perform fracture mechanics International Symposium, ASTM STP 1295, E.R. analyses of thin-walled cladding alloys. Bradley and G.P. Sabol, Eds., American Society for Testing and Materials, 431-447, 1996. 5 SUMMARY AND OUTLOOK [4] Y.R. Rashid, C. Lemaignon, A. Strasser, A fracture-mechanics approach for the determination "Evaluation of Fracture Initiation and Extension in of toughness of thin-walled claddings which uses a Fuel Cladding", Proc. ANS Int. Topical Meeting tension-loaded modified DENT sample has been on LWR Fuel Performance, Park City, UT, USA, developed. Applying well-accepted, J-evaluation April 10-13, 2000. methods, the determination of J0.2 as a measure for [5] P.H. Davies, R.S. W. Shewfelt, "Size, Geometry, the stage of early crack growth has been established. and Material in Fracture Toughness Testing of Monitoring crack growth and applied load allows the Irradiated Zr-2.5Nb Pressure Tube Material", determination of J-R curves to be utilized as another Zirconium in the Nuclear Industry: Twelfth measure for the toughness of the material. The International Symposium, ASTM STP 1354, G.P. approach is not limited to cladding materials, and can Sabol and G.D. Moan, Eds., American Society for clearly also be employed for other piping applications, providing a quantitative relationship between a flaw of Testing and Materials, West Conshohocken, PA, given size (fretting defect, production defect, hydride 356-376, 2000. lens, etc.) and tolerable loads, and can therefore be [6] N.E. Dowling, "Mechanical Behaviour of Ma- linked to safety considerations. Future work is aimed terials", 2nd Edition, Prentice-Hall, Inc., ISBN-0-13- at improving the method, and extending its application 905720-X, 1999. to higher temperatures, other environments and other loading conditions (creep, fatigue, etc.). Another [7] G.C. Sih, "Fracture Mechanics in Two Decades", experimental challenge for the future will be to apply (Twentieth Symposium) ASTM STP 1020, R. P. the method to service exposed cladding materials Wei and R. P. Gangloff, Eds., Am. Soc. For originating from nuclear power plants, and to Testing and Materials, Philadelphia, 9-28, 1989. determine toughness values for differently hydrided [8] EPRI Nuclear Fuel Industry Research (NFIR) materials. Program, the 42nd Steering Committee Meeting, Baden, October 2002, PSI (still unpublished ACKNOWLEDGMENT results). This work was financed by the Swiss Nuclear Power [9] T. Anderson, "Fracture Mechanics-Fundamentals Stations in the Framwork of their KFEB project, and and Applications", CRC Press, 1994. was performed within EPRI NFIR IV guidelines. REFERENCES [1] U.S. Nuclear Regulatory Commission Standard Review Plan, NUREG-0800, Rev. 2, July 1981. [2] K. Edsinger, J.H. Davies, R.B. Adamson, "De- graded Fuel Cladding Fractography and Fracture Behavior", Zirconium in the Nuclear Industry: Twelfth International Symposium, ASTM STP 1354, G.P. Sabol and G.D. Moan, Eds., American Symposium for Testing and Materials, West Conshohocken, PA, 316-339, 2000. 97 ANISOTROPIC DIFFUSION IN LAYERED ARGILLACEOUS ROCKS: A CASE STUDY WITH OPALINUS CLAY L.R. Van Loon, J. Soier, W. Muller, M.H. Bradbury Anisotropic diffusion was studied in Opalinus Clay, a potential host rock for the disposal of spent fuel, vitrified high-level waste and intermediate-level waste. Diffusion parallel to the bedding was measured using a radial through-diffusion technique, while diffusion perpendicular to the bedding was measured using the classical (planar) through-diffusion method. The materials used were samples from Mont Terri (MT) and Benken (BE), respectively. Diffusion of Tritiated Water (HTO), parallel and perpendicular to the bedding, was studied under confining pressures of 7 MPa (MT) and 14 MPa (BE), respectively. The effective diffusion coefficient for diffusion parallel to the bedding, De, was found to be 3.20(±0.26)x10'11 m2s1 and for the Benken 5.39(±0.43)x10'11 m2s1 for the Mont Terri samples. The diffusion accessible porosity was ~ 0.15(±0.02) in both cases. For diffusion perpendicular to the bedding, the effective diffusion coefficient was 5.44(±0.35)x10'12 m2s1 and 1.37(±0.08)x10'11 m2s1 for the Benken and Mont Terri samples, respectively. The diffusion accessible porosity was also ~0.15(±0.02) in both cases. These first results indicate that diffusion parallel to bedding is larger than that perpendicular to the bedding by a factor of 4 to 6. This might be explained in terms of smaller path lengths (tortuosity) for species diffusing parallel to the fabric. 1 INTRODUCTION anisotropic, and consequently so is the effective diffusion in such rocks (Fig. 2). The diffusion of charged and neutral species in rocks depends strongly on their geometric parameters, such \ sedimentation of clay platelets resulting as tortuosity and constrictivity. Tortuosity takes into in a house-of-cards structure account path lengthening, because of the particulate nature of the porous medium (Shackelford, 1991), while constrictivity is concerned with pore narrowing. For aqueous-phase diffusion, the relationship between the effective diffusion coefficient, De, the geometric parameters, and the diffusion coefficient in water, Dw, I is given by: IHItlHIIH 8-e De = a 0) compaction by overburden load leading to preferential orientation of clay platelets where S represents the constrictivity, r is the 3: clay platelet tortuosity, and s the diffusion accessible porosity. Fig. 1: Structure of a mud rock during sedimentation Argillaceous rocks are mainly composed of clay (house-of-cards structure) and after platelets that have settled in an aqueous (marine) compaction (preferential orientation of clay environment to form a mud deposit. Bennett et al. platelets); the orientation of the clay platelets (1980) have reported that the clay platelets in marine after compaction is perpendicular to the deposits of low porosity are oriented preferentially direction of sedimentation. perpendicular to the direction of sedimentation. During Anisotropic diffusion has been observed in many sedimentation, the grains form "house-of-card" type different media; e.g. in crystals, textile fibres and structures, because of the electrostatic repulsion by polymers, in which the molecules have a preferential like-charged basal planes, and the attraction by direction of orientation (Crank, 1975; Seo et al., 1999). oppositely charged edge and basal planes (Fig. 1). When a species diffuses parallel to the fabric, The solution composition and ionic strength also have tortuosity is expected to be smaller than in the former an effect on the structure (Berner, 1980). After case in which diffusion, is perpendicular to the deposition, the layers of clay mud are compacted by bedding (Fig 2). Consequently, the effective diffusion the weight of younger, superimposed sediments rate will be faster. (overburden pressure), resulting in water being So far, in geochemistry, little attention has been paid squeezed out. During this process, the mutually- to these potentially anisotropic diffusion properties. repellant particles are forced closer together, and the Systematic studies comparing diffusion perpendicular porosity of the mud decreases. and parallel to the fabric for the same material and for Compaction has been reported to increase the different species are, we believe, not available, with preferential orientation of clay platelets (von the exception of one study by De Windt & Palut, 1999, Engelhardt & Gaida, 1963; Meade, 1964). Owing to who studied the diffusion of HTO and I" through a this preferentially layered structure of argillaceous piece of Opalinus Clay perpendicular and parallel to rocks, tortuosity (path lengthening) is expected to be the fabric. 98 composition, chemistry and effective porosities for ; •• groundwater in aquitards, and by Novakowski & van •• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• • •• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• ••'••'••'••'• •'_• •'_• •'_• •'_• •'_• •'_• •'_• •'_• •'_• •'_• •'_• •'••'••'••'••'• •' der Kamp (1996) for determining effective diffusion — —•—•—•—•—•—•—•—•— Pressure (overburden) coefficients, porosity and adsorption from in-diffusion TrmfufufufuJ experiments in porous geological materials. A concentration gradient is set up across the sample in such a way that diffusion takes place from the centre OPA" repository of the sample (high concentration region) to the outer Diffusion boundary (low concentration region). Because the •• •••••• ••• > •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• .• •• layering of the sample is parallel to the concentration • ••• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •'• •• gradient, diffusion will take place along the layering. A more detailed description of the cell is given in Van Loon etal. (2002). Theory of radial diffusion Diffusion parallel to bedding Consider a hollow cylindrical sample of height h, with an internal radius rint, and an external radius rext. The concept behind the experiments is the same as in one-dimensional diffusion experiments (Cho et al., 1993), except for the fact that the high-concentration reservoir is the central cavity of the cylinder (from r=0 to r=rint), and that breakthrough of the tracer occurs across the entire external surface (r=rext)- As a Diffusion perpendicular to bedding consequence, the diffusion process has cylindrical symmetry. Fig. 2: Schematic view of the anisotropy of a layered rock (Opalinus Clay, Benken, Switzerland) with respect to diffusion. Although they observed a larger value for the diffusion parallel to the bedding, they could not exclude experimental artefacts, so that their results remain somewhat in question. Most studies of natural argillaceous materials published in the open literature are on diffusion properties perpendicular to the bedding (Bourke etal., 1993; Put et al., 1998; Melkior, 2000). One of the main reasons for this is a lack of appropriate techniques for studying diffusion parallel to the bedding. In this work, diffusion of HTO was studied perpendicular and parallel to the bedding of Opalinus Clay (OPA): a layered, argillaceous rock. For diffusion parallel to the bedding, we used a radial through- diffusion technique recently developed by Van Loon et al. (2002), and for diffusion perpendicular to the bedding, a one-dimensional, through-diffusion method, as described in Van Loon et al. (2003), is 10 cm used. The experimental set-up is so arranged that the pressure applied to the samples is always Fig. 3: Cross-section of view of the radial through- perpendicular to the bedding, and the diffusion either diffusion cell. parallel or perpendicular to it, which is relevant for performance assessment, and represents the most The diffusion equation is given by: common situation in natural systems. ac dC 1_d_ n (2) 2 MATERIALS AND METHODS a dt ~~r dr v dry 2.1 Radial diffusion where Da is the apparent diffusion coefficient, and C is Equipment the concentration of a given tracer in the rock. The initial and boundary conditions are as follows: A view of the radial through-diffusion cell is given in Fig. 3. A similar concept was used earlier by van der Kamp et al. (1996) for determining isotopic 99 C(rint -2nhDc (3) dr mmmu c = c The analytical solution of the cylindrical through- diffusion problem is given by (Crank, 1975, p.84; Eqn. 5.64): Q(rext,t)_ 2(Dat-L) nC0ah ln(rext/rint) r a r a Jo ( int n )J() ( ext n 2 'I )exp(-Daa nt) (4) n=1 an (4 ( rintan )~(Jo( rextan )) where Q(rext,t) is the total amount [in mol] or activity [in Bq] of tracer which has diffused through the sample at time t. The parameter L [m2] in Eqn. 4 is c = o given by: r r + r + r n r r . M ~ ext ( ht ext) ^ ( ext ? int) (5) 4ln(rext/rint) Fig. 4: Concentration distribution in a radial diffusion The J0(x) terms are Bessel functions of the first kind of experiment in the steady-state phase. order zero, and the an terms are the positive roots of: 2.2 Planar (one-dimensional) diffusion J r a Y r a J r a Y r a m^n) = o( int n ) 0 ( ext n )~ o( ext n ) 0 ( int n ) Equipment = 0. A schematic view of the planar through-diffusion cell is (6) given in Fig. 5, and a more detailed description is given in Van Loon et al. (2003). A concentration and the Y0(x) terms are Bessel functions of the second kind of order zero. The first five positive roots gradient is set up across the sample in such a way that diffusion takes place from the high concentration rintan of Eqn. 6 for different ratios reyt/rint, are tabulated by Crank (Table 5.3, p. 380, 1975). side to the opposite boundary (low concentration side). Because the layering of the sample is normal to At steady-state, the solution (4) reduces to: the concentration gradient, diffusion will take place perpendicular to the layering. 0(rext,t)=l (Det-aL). (7) Theory of planar diffusion For a one-dimensional diffusion process through a Hence the total amount of tracer diffused through the 2 1 sample becomes a linear function of time: plane of thickness D, the flux, J [mol-m" s" ], is given by Fick's first law: Q(rext,t)=a-t-b, (8) ac with: (10) 2nC hD 2nC haL 0 e and b •• 0 ln(rext/rint) ' ln(rext/rint)' For the temporal evolution of the concentration, C, Fick's second law is applied: The concentration profile of the tracer in the rock at steady-state is logarithmic: 3£ d2C (11) InlinL r Cr—Cn I- (9) where Da is the apparent diffusion coefficient: InlinL rext J 100 d the sample thickness [m] (12) a Co the concentration of the radionuclide in the reservoir cell [Bq-m"3] and a represents the capacity factor, defined as: t time [s] oc=e+p Kd (13) a the rock capacity factor 3 _1 where Kd is the distribution coefficient [m -kg ], and p Equation 14 gives the cumulative mass or activity of a tracer in the low-concentration reservoir, Q(d,t), as a the bulk dry density of the rock [kg-m"3]. function of time. As time increases, the exponential For non-sorbing tracers (Kd=0), the rock capacity term asymptotes to zero and the Equation reduces to: factor is just the porosity, which can be deduced directly from the intercept. For K >0, the porosity can d Det a be deduced from Eqn. (13). Q(d,t) = SdC0 (15) Thus, when the diffusion process is in the steady-state ± • Tubing phase, the total diffused mass is a linear function of • Swagelok fitting time: Disc spring package h n Q(d,t) = mt-p (16) ii n End plate ii n ii ii ii ii • O-ring JLmhJL • Screw S C20 De . S d CQ 2 a • Sample where m = —- and p = —, • Stainless steel filter II Sample holder Sample origin and sample preparation d) £8= The samples used for the measurements were of Load sensor Opalinus clay, taken from the Mont Terri Underground Rock Laboratory (Canton Jura), and from the deep borehole in Benken (Canton Zurich) in the northern part of Switzerland. The OPA was deposited about 180 million years ago as a marine sediment consisting Fig. 5: Cross-sectional view of the PSI high-pressure of fine mud particles. It has a thickness of -100 m, diffusion cell for diffusion perpendicular to the and contains between 40% and 80% clay minerals, bedding (planar diffusion). 10% of which are capable of swelling (Gautschi, As defined earlier, De is the effective diffusion 1997). A bore core, either from Mont Terri or from coefficient, and is related to the diffusion coefficient in Benken, was placed in a Plexiglas™ cylinder, and free bulk water by virtue of Eqn. 1. embedded in an epoxy resin (Epofix™, Struers GmbH). After the hardening of the resin, the cylinder The boundary and initial conditions of the through- was placed on a lathe, and a cylindrical sample diffusion problem are given by: (parallel: rext=25.4mm, rint=6.35mm, h=52mm; C(x,t)=0, xe [0,c/], perpendicular: rext=12.7mm, d=11mm) was prepared. C(x=0,t>0) = C0 C(x =d,t>0) = C0. Experimental set-up An analytical solution to the through-diffusion problem The experimental set-up of the diffusion experiments may be obtained by solving Eqn. 11, with the initial is shown schematically in Fig. 6, and comprises a and boundary conditions given above (Crank, 1975; diffusion cell, as described above, an 8-channel Jakob eta!., 1999): peristaltic pump (IPC, Ismatec, Idex corporation, USA), and 250 cm3 and 25 cm3 containers. The Q(d,t)=SdC0 • sample was placed in the sample holder, together with two stainless-steel filters (discs or cylinders), and the 2 2 De-n -n -t\ (14) cells were then closed with two end plates. The bolts 2 2 2 P 2 d * n h n \ d a I were first greased with a MoS2 screw paste, and then tightened until the desired load was obtained (MT: 7 MPa; BE: 14 MPa). The confining pressure where: simulates the in-situ overburden. The large container was filled with 200 cm3 of artificial OPA pore water, 2 S = the cross-sectional area of the sample [m ] and the small one with 20 cm3. 101 High concentration reservoir (250 ml) Fig. 6: Schematic view of the experimental set-up of the radial through-diffusion experiment The composition of the artificial pore water is given in Table 1: Composition of the synthetic OPA pore Table 1. The samples were re-saturated by circulating waters used in the diffusion experiments. the OPA pore water against both surfaces of the sample for 2 weeks; this time was found to be Element 1Mont Terri 2Benken sufficient to reach re-saturation. Subsequently, the (mol-dm"3) (mol-dm"3) solutions were replaced by fresh ones, the solution in the large container was labelled HTO, and through- Na 2.40-10"1 1.50-10"1 diffusion was started (Series 1). The activity of HTO K 1.61-10"3 4.31-10"3 was 1.0-106 Bq-dm"3. Mg 1.69-10"2 5.21-10"3 The solutions in the small containers were replaced 2 3 after a given time interval At to keep the activity of the Ca 2.58-10" 7.24-10" tracer in the compartment as low as possible: i.e. <1% Sr 5.05-10"4 3.97-10"4 of the concentration in the high concentration 1 1 compartment. The activity of HTO in solution was CI 3.00-10" 1.60-10" 2 2 measured by Liquid Scintillation Counting, as S04 1.41-10" 1.00-10" described in Van Loon etal. (2003). CO3/HCO3 4.76-10"4 3.11-10"4 After completion of a through-diffusion experiment, the 7.6 7.9 solutions in both containers were replaced by artificial PH pore water without tracer, and out-diffusion was Sum cations 3.28-10"1 eq/l 1.80-10"1 eq/l started. At selected time intervals, the activity in the Sum anions 3.28-10"1 eq/l 1.80-10"1 eq/l solutions was measured, and then the solutions replaced by fresh ones. This procedure was repeated Ionic strength 0.39 M 0.20 M until all the activity in the sample had diffused out. 1 After out-diffusion, the solution in the large container Pearson Type A1 (Pearson, 1998) 2 Pearson (2000) was replaced by an OPA solution containing HTO 6 3 Unlike the cumulative plot, the flux clearly shows (1.0-10 Bq-dm" ), and through-diffusion was repeated when the system is in steady-state. The latter phase on the same samples (Series 2). was used to deduce a value for the effective diffusion coefficient for the tracer, and the porosity accessible 3 RESULTS AND DISCUSSION to the tracer. The slope of the curve (Qt vs time) gives Figure 7 (A-D) shows the flux and total diffused the effective diffusion coefficient, by applying Eqn. 7 activity for the through-diffusion of HTO parallel and for diffusion parallel to the bedding, and Eqn. 15 for perpendicular to the bedding in the two samples diffusion perpendicular to the bedding. The porosity (Series 1). The curves clearly show two stages: a can be calculated from the intercept of an asymptote transient phase, in which the flux increases, and a extrapolated from the steady-state region of the curve steady-state phase, in which the flux is constant, and to the Y-axis. The concentration of HTO at the high- where the total diffused activity increases linearly with concentration boundary decreased slightly (<10%) time. during the experiment, so that the chosen up-stream 102 boundary, as specified, can be applied in the analysis It can be seen from the data from Series 1 and of the breakthrough curves. The quality of the Series 2 that the diffusion measurements are parameters (De and a) was tested by using them as reproducible. Compared with values for diffusion input parameters for the analytical solutions (Eqn. 4 perpendicular to the bedding, the effective diffusion for parallel diffusion, or Eqn. 14 for perpendicular coefficients for diffusion parallel to the bedding are diffusion) and calculating the entire diffusion curve, larger by a factor between 4, for the Mont Terri J(rext,t) or J(d,t), as a function of time. In all cases, it samples, and 6, for Benken samples. The larger value was possible to reproduce the increased flux data for the Benken samples indicates a greater degree of (solid flux-curves in Fig. 7). An overview of the preferential orientation of the platelets, which might be preliminary results for HTO is given in Table 2. due to the higher overburden pressure (Meade, 1964). Table 2: Values for the effective diffusion coefficients and diffusion accessible porosities for the diffusion of HTO into Opalinus Clay parallel and perpendicular to the bedding. Sample De (parallel) a (parallel) De (perpend.) a (perpend.) (m2-s1) (m2-s1) Benken No. 1 3.10(±0.25)x10"11 0.15±0.02 5.44(±0.35)x10 ,-12 0.14±0.02 Benken No. 2 3.20(±0.23)x10 -11 0.13±0.02 5.41(±0.35)x10" 0.13±0.02 Mont Terri No. 1 5.39(±0.43)x10"11 0.15±0.03 1.41 (±0.09)x10,-1 1 0.17±0.02 -11 Mont Terri No. 2 5.42(±0.38)x10 0.17±0.02 1.37(±0.08)x10,-i i 0.14±0.02 1500 1.6 10" 1.2 104 1000 E O o o~ 0G DO _Q do _Q - 500 10 15 10 20 30 40 Time [days] Time [days] 3 10 2.5 10* --I21 o o E o cr 1.510 co -o • Flux -— best fit uncertainty •• 5 10 o— Q(v r ext,t' )' j I i i i I i i i I i 0 10 5 10 15 4 8 12 16 Time [days] Time [days] Fig. 7: Fluxes and total diffused mass of HTO through Opalinus Clay (Series 1) for diffusion parallel to the bedding (A: Benken, C: Mont Terri), and perpendicular to the bedding (B: Benken, D; Mont Terri). The solid curves for the fluxes were calculated using the analytical solutions (Eqn. 4 or Eqn. 14, for diffusion parallel or perpendicular to the bedding, respectively), together with the parameters specified in Table 2. The shaded areas represent the uncertainties to the calculated curves. 103 Because HTO is a non-sorbing tracer (Kd=0) the rock anionic (36CI", 125l") and cationic (22Na+) tracers are capacity factor, a, equals the diffusion accessible planned. porosity, 8. The values for the porosity, e, are similar in both samples. Because Dm e and 8 are independent ACKNOWLEDGMENTS of the direction of diffusion, the different D values e This work was partially funded by the Swiss National must be due to differences in the tortuosity. The path Cooperative for the Disposal of Radioactive Waste lengthening (tortuosity) is smaller in the direction (NAGRA). Dr. J. Hadermann (PSI) and Dr. A. Jakob parallel to the bedding, so that De is expected to be (PSI) are acknowledged for their comments on the larger for diffusion parallel to the bedding (Eqn. 1), as manuscript. has been found experimentally. A more detailed explanation for the factor 4 to 6 REFERENCES cannot yet be given. However, for porous media with [1] P.J. Bourke, N.L. Jefferies, D.A. Lever, T.R. pores smaller than the dimension of the clay particles, Lineham, "Mass transfer mechanisms in the ratio of the particle length to the particle width compacted clays", in Manning, D.A.C., Hall, P.L., seems to play a central role, as shown in a very Hughes, C.R. (Eds.), Geochemistry of Clay-Pore simplified sketch of the microstructure of a layered Fluid Interactions. Chapman & Hall, London argillaceous rock (Fig. 8). (1993). n layers of thickness r and length s [2] R.H. Bennett, W.R. Bryant, G.H. Keller, "Clay s fabric of selected submarine sediments: fundamental properties and models", J. I II ill 1 1 Ir 1 Sediment. Petrol., 51, 217-231 (1981). r ill 1 1 1 1 1 i IE ill 1 1 1 [3] R.A. Berner, "Early Diagenesis: a Theoretical Approach", Princeton University Press, Princeton, i ill II i 1 1 NJ, USA (1980). i 1 1 ill 11 i [4] W.J. Cho, D.W. Oscarson, P.S. Hahn, "The i 1 1 ill 11 i measurement of apparent diffusion coefficients in i 1 1 ijl 11 i compacted Clays: an assessment of methods", • Applied Clay Science, 8, 283-294 (1993). _ : pathway is d [5] J. Crank "The Mathematics of Diffusion", 2nd — : pathway is d + (n-1) x s/2 edition, Oxford University Press, Oxford, United Fig. 8: Simplified representation of the microstructure Kingdom (1975). of a layered argillaceous rock. [6] L. De Windt, J.M. Palut, "Tracer feasibility experiment (FM-C, Dl)", in Thury, M., Bossart, P. To test this hypothesis, the length/width ratios of the (Eds.), Mont Terri Rock Laboratory: Results of grains have to be measured for different rocks, and the Hydrogeological, Geochemical and these data related to effective diffusion data (i.e. De Geotechnical Experiments Performed in 1996 parallel and perpendicular to the fabric). and 1997. Geologische Berichte Nr. 23, Landeshydrologie und -geologie, Bern, 4 SUMMARY AND CONCLUSIONS Switzerland (1999). One-dimensional radial and planar through-diffusion [7] A. Gautschi, "Hydrogeology of the Opalinus Clay, techniques have been used for measuring the implications for radionuclide transport", Nagra effective diffusion coefficients parallel and Bulletin 31, 24-32 (1997). perpendicular to the fabric of a layered rock under a confining pressure applied perpendicular to the [8] R.H. Meade, "Removal of water and rearrange- bedding. The methods could be successfully applied ment of particles during the compaction of clayey to measure the effective diffusion coefficients of HTO sediments - revieW\ U.S. Geol. Survey Prof. in an argillaceous rock (OPA). Preliminary results Paper 497-B, B1-B23 (1964). clearly demonstrate that diffusion in Opalinus Clay is [9] Th. Melkior, "Etude methodologique de la anisotropic. This is due to differences in the values for diffusion de cations interagissants dans les the tortuosity parallel and perpendicular to the argiles. Application: mise en oeuvre bedding. Diffusion parallel to the bedding is larger by experimental et modelisation du couplage a factor of 4 to 6 than diffusion perpendicular to the chimie-diffusion d'alcalins dans une bentonite bedding. The values for the diffusion accessible synthetique", Ph.D. Thesis, Ecole Centrale des porosities, however, are comparable. Arts et Manufactures, Ecole Centrale Paris, The evaluation of the out-diffusion results of HTO is Paris, 2000. CEA report CEA-R-5890, SaClay currently ongoing, and will be presented in a 91191 Gif-sur-Yvette, Cedex, France (2000). forthcoming publication. Similar experiments with 104 [10] K.S. Novakowski, G. van der Kamp "The radial [16] G. Van der Kamp, D.R. Van Stempvoort, diffusion method. 2. A semianalytical model for L.I. Wassenaar, "The radial diffusion method. 1. the determination of effective diffusion using intact cores to determine isotopic coefficients, porosity, and adsorption", Water composition, chemistry, and effective porosities Resources Research ,32, 1823-1830 (1996). for groundwater in aquitards", Water Resources Research, 32, 1815-1822 (1996). [11] F.J. Pearson, "Opalinus Clay experimental water: A1 Type", Version 980318. Internal Technical [17] L.R. Van Loon, J.M. Soler, W. Muller, M.H. Report TM-44-98-07, Paul Scherrer Institut, Bradbury, "A radial through-diffusion technique Villigen PSI, Switzerland (1998). for measuring diffusion rated parallel to the fabric of layered rocks: measurement of effective [12] F.J. Pearson, Nagra unpublished internal report, diffusion coefficients and diffusion accessible Nagra, Wettingen, Switzerland (2000). porosities", submitted for publication, in Journal [13] M.J. Put, A. Dierckx, M. Aertsens, P. De of Contaminant Hydrology (2002). Canniere, "Mobility of the dissolved organic [18] L.R. Van Loon, J.M. Soler, M.H. Bradbury, matter through intact Boom Clay cores", "Diffusion of HTO, 36CF and 1251' in Opalinus Clay Radiochimica Acta, 82, 375-378 (1998). samples from Mont Terri: effect of confining [14] Y. Seo, H. Shinar, Y. Morita, G.A. Navon, pressure", Journal of Contaminant Hydrology, 61, "Anisotropic and restricted diffusion of water in 73-83 (2003). the sciatic nerve: A H-2 double-quantum-filtered [19] W. von Engelhardt, K.H. Gaida, "Concentration NMR study", Magnetic Resonance in Medicine, changes of pore solutions during the compaction 42, 461-466 (1999). of Clay sediments", J. Sediment. Petrol., 23, 919- [15] C.D. Shackelford, D.E. Daniel, "Diffusion in 930 (1963). saturated soil. I: Background", Am. Soc. Civ. Eng., Geotech. Eng., 117, 467-484 (1991). 105 UPTAKE OF TRIVALENT ACTINIDES (Cm(lll)) AND LANTHANIDES (Eu(lll)) BY CEMENT-TYPE MINERALS: A WET CHEMISTRY AND TIME-RESOLVED LASER FLUORESCENCE SPECTROSCOPY (TRLFS) STUDY J. Tits, T. Stump?, E. Wieland, T. Fanghanel1 1 Forschungszentrum Karlsruhe, Institut fur Nukleare Entsorgung, Karlsruhe, Germany The interaction of the two chemical homologues Cm (III) and Eu(lll) with calcium silicate hydrates at pH 13.3 has been investigated in batch-type sorption studies using Eu(lll), and complemented with time- resolved laser fluorescence spectroscopy using Cm(lll). The sorption data for Eu(lll) reveal fast sorption kinetics, and a strong uptake by CSH phases, with distribution ratios of 6(±3)-10 L kg1. Three different types of sorbed Cm(lll) species have been identified: a non-fiuorescing species, which was identified as Cm cluster present either as surface precipitate or as Cm(lll) colloid in solution, and two sorbed fluorescing species. The sorbed fluorescing species have characteristic emission spectra (main peak maxima at 618.9 nm and 620.9 nm) and fluorescence emission lifetimes (289 ± 11 jus and 1482 ±200 jus). From the fluorescence lifetimes, it appears that the two fluorescing Cm(lll) species have, respectively, one to two or no water molecules left in their first coordination sphere, suggesting that these species are incorporated into the CSH structure. A structural model for Cm(lll) and Eu(lll) incorporation into CSH phases is proposed based on the substitution of Ca at two different types of sites in the CSH structure. 1 INTRODUCTION spectroscopic information using Time-Resolved Laser Fluorescence Spectroscopy. TRLFS has proved to be Cementitious materials are commonly used world- a versatile tool for Cm(lll) speciation studies [5,6], and wide for the immobilisation of radioactive waste. In a for sorption studies on various solids [7]. It is capable nuclear waste repository, the immobilisation of of identifying different species as well as determining radionuclides by cementitious materials takes place their hydration status, thus allowing outer-sphere and both in the waste containers and in the construction inner-sphere surface complexes and species materials, such as backfill, liners, etc. A mechanistic incorporated in the crystal lattice of the solid to be understanding of the chemical processes by which distinguished. waste elements become immobilised in cement-based matrices is important for long-term predictions of the 2 EXPERIMENTAL performance of disposal sites for hazardous waste, and repositories for radioactive waste. 2.1 Materials Hardened cement paste (HCP) is a complex mixture All experiments were performed in an artificial cement of different phases [1]. The immobilisation potential of pore water (ACW) simulating the chemical conditions HCP originates from the selective binding properties during the first phase of the HCP degradation. The of these phases for metal cation and anion ACW contained 0.18 M KOH and 0.114 M NaOH, and compounds. Calcium aluminates and calcium silicate had a pH of 13.3. The Ca and Si concentrations in this hydrates (CSH phases) are considered to be among pore water were fixed at levels such that the solution the most important phases governing immobilisation was in equilibrium with the CSH phase. processes, because of their abundance and the presence of appropriate structures for cation and CSH phases were prepared using the so-called "direct anion binding (e.g. [2,3]). At the present time, reaction" method [8] in a glove box under N2 however, the understanding of uptake processes on a atmosphere. AEROSIL 300 (Si02) was mixed with microscopic level, and the thermodynamic description CaO in polyethylene centrifuge tubes to give a target of these processes, is still insufficient. Ca/Si weight ratio (C:S) of 1.0. To this, ACW was added to obtain a solid to liquid (S:L) ratio of 24 g L"1. The chemical conditions in a cementitious After an ageing period of 4 weeks, the slurries were environment are controlled by the interaction of the filtered through Whatman ashless 541-grade filter HCP with the groundwater. The first phase of the HCP paper using a Buchner funnel with vacuum pump. The degradation is characterised by the dissolution of the supernatant solution was analysed to obtain the Na, alkali oxides in the cement, resulting in a pH of K, Ca and Si concentrations. The solid material was approximately 13.3, and Na and K concentrations of then re-suspended in fresh ACW, and equilibrated for approximately 0.11 M and 0.18 M, respectively [4]. a further week. The filtration and re-suspension procedures were repeated until the Na and K The objective of the present study is to investigate the concentrations in the filtrate were identical to their interaction of Cm(lll), a representative of the trivalent respective concentrations in fresh ACW; one cycle actinides, and Eu(lll), a chemical homologue for the was usually sufficient. In preparation of the drying trivalent actinides, with CSH phases under conditions step, the slurries were then washed with Milli-Q water relevant to a cementitious repository for radioactive to remove the remaining NaOH and KOH. The contact waste by combining macroscopic information from time of the slurries with Milli-Q water was kept as batch sorption experiments and molecular level short as possible, i.e. less than 1 minute, to minimize 106 changes in composition. The CSH phases were dried With TRLFS, spectroscopic information is obtained to constant weight in a dessicator over a solution of from excitation spectra, emission spectra and lifetime saturated CaCI2, which maintains a relative humidity measurements [10]. Excitation spectra are obtained of 30%. Moreover, the procedure also prevented the from scanning the excitation wavelength, and CSH phases from becoming statically charged, which recording the emission within a fixed wavelength made handling of the material easier. range. Emission spectra result from scanning the emission wavelength at a fixed excitation wavelength. 2.2 Methods The great advantage of TRLFS is the possibility of 2.2.1 Batch sorption experiments measuring the lifetime of the exited states. This information can be used to discern energy transfer Sorption kinetic data and sorption isotherms for Eu(lll) mechanisms during de-excitation. De-excitation were obtained using the batch technique. All occurs through radiative and non-radiative processes. experiments were carried out at 25±1°C in glove In aqueous media, non-radiative processes contribute boxes under N2 atmosphere. CSH suspensions were significantly to the decay rate, resulting in shorter prepared by mixing a quantity of fresh CSH from the lifetimes of the excited state and lower fluorescence dessicator with 1 L of synthetic CSH solution. (The intensity. The non-radiative decay is mainly caused by composition of this solution is based upon the energy transfer from an excited f-state to highly chemical analysis of the last filtrate obtained during excited vibronic states of the H20 molecules in the the synthesis of the CSH phase.) Then, 40 ml_ of the first coordination sphere around the lanthanide or pre-equilibrated suspensions were transferred to 152 actinide ion. The significance of this non-radiative centrifuge tubes, and an aliquot of an acidic Eu decay depends on the ratio, R, of the energy gap tracer solution was added. The suspensions were between the emitting state and the next lower, non- then equilibrated on an end-over-end shaker for a fluorescent state, and on the vibrational quantum given period of time, depending on the type of energy. If this ratio is high, then the probability for experiment. Experiments were performed in duplicate radiationless decay will be low, and the fluorescence or triplicate. Phase separation was carried out by intensity and fluorescence lifetime will be high. centrifugation (1 hour at 95000g). The concentrations of the radionuclides in both the suspension and the This ratio, R, determines to a large extent the equilibrium solution were assayed with a Canberra applicability of TRLFS to different lanthanides and Packard Tri-Carb 2250CA liquid scintillation counter actinides in aqueous media. In the case of Cm(lll) and using an energy window between 6 keV and 70 keV. Eu(lll), the energy gap amounts to approximately 17 000 cm"1 and 12 000 cm"1, respectively, resulting in The sorption data are presented in terms of a fluorescence lifetimes in the jus range. In the case of distribution ratio, R (L kg"1), which relates the quantity d Am(lll), however, the energy gap only amounts to of radionuclide sorbed (mol kg"1) to the equilibrium 5000 cm"1, giving rise to a lifetime below 40 ns, and a radionuclide concentration in solution (M). The much lower quantum yield [9]. difference in concentration between the total activity in the suspension and the activity in the supernatant fluorescence free fluorescence solution after centrifugation was taken as a measure spectra ion process levels of the tracer activity on the solid. In this manner, 30- excitation artefacts due to wall sorption effects were avoided. spectrum H ^ 2.2.2 Time-Resolved Laser Fluorescence t nonradiative decay Spectroscopy (TRLFS) 20- emission *r spectrum - (lifetime - lO^s) Lanthanides and actinides with partially filled f-shells •7/2 absorb radiation at ultraviolet or visible (UV/VIS) excitation wavelengths of the electromagnetic spectrum, due to I 10 the excitation of electrons within their f-shells [9]. These parity-forbidden transitions are rather weak in intensity (molar absorptivity < 10 L mol"1 cm"1 for • 7/2 lanthanides, and < 500 L mol"1 cm"1 for actinides). Thus, for studies under high pH conditions, in which 7 Fig. 1: Simplified energy levels of Cm(lll) and metal concentrations are usually below 10" M, the spectroscopic characteristics, modified from a sensitivity of classical UV/VIS absorption diagram of Beitz [11]. spectroscopy is not sufficient. Fluorescence spectroscopy, with excitation in the UV/VIS region, provides information similar to UV/VIS absorption A simplified schematic of Cm(lll) energy levels, spectroscopy. The application of the monochromatic including the most important excitation and emission light of high power lasers for excitation, combined with states [11], is shown in Fig. 1. The excitation of the 5f7 sensitive photon-counting detectors, allows studies to configuration occurs via the three most intensive be made of the speciation of lanthanides and absorption bands: at 396.0 nm (absorption coefficient, actinides present at the sub-micro-molar 8 = 55.3 L mol"1 cm"1), 381.3 nm (e = 32.6 L mol"1 cm"1) concentration range. or 375.5 nm (e = 29.3 L mol"1 cm"1) in the F, G and H states. In aqueous solution at room temperature, 107 fluorescence emission is only observed from the A- centrifugation by accidental shaking, may have a state to the ground state, Z, and results in a simple significant effect on the measured concentration in the emission band with a maximum at 593.8 nm supernatant. (FWHM=8.3 nm). The decay from the excited A-state to the ground state follows a mono-exponential law. The fluorescence lifetime for aqueous Cm(lll) is found to be 65 jus. In this study, TRLFS measurements were performed using a flash-lamp-pumped Ti:sapphire laser (Elight, Titania). The laser pulse energy, at most 4 mJ, was controlled by a photodiode. The fluorescence emission was detected by an optical multi-channel analyser, which allows simultaneous detection of the whole emission spectrum. The system consists of a monochromator and spectrograph (Oriel; MS 257) with a 300 or 1200 lines/mm grating, and a ICCD camera (Andor). The excitation occurred at 395 nm, and the emission spectra of Cm(lll) were recorded in the 520-680 nm (300 lines/mm grating; lifetime measurements) and 580-620 nm (1200 lines/mm Fig. 2: Eu(llh sorption isotherm; the S:L ratio was grating; high resolution spectra for peak 5-10" kg L"1, and the equilibration time was deconvolution) ranges using a constant time window one week. of 1 ms. For measuring the time dependent emission decay, the delay time between laser pulse and camera gating was scanned with time intervals 3.2 Spectroscopic characterisation of Cm(lll) in between 10 and 50 jus. CSH suspensions in ACW at pH 13.3 For the TRLFS investigations of the Cm(lll) sorption Selected fluorescence emission spectra of Cm(lll) on CSH phases, CSH suspensions were prepared as (10"7 M) in a CSH suspension (5-10" kg L"1) measured described above. An aliquot of an acidic 48Cm tracer at different contact times are shown in solution was added to the CSH suspensions to obtain a total 248Cm concentration of 10" M. The Fig. 3. Two different peak maxima at -606.0 nm and suspensions were then equilibrated for a given period -620 nm appear in the spectra. Similar spectra were observed for 10"7 M Cm(lll) with a different S:L ratio of of time, depending on the type of experiment; the 4 1 samples were shaken periodically during equilibration. 5-10" kg L" CSH suspension. The intensity of the At chosen time intervals, Cm(lll) emission spectra Cm(lll) fluorescence emission spectra increased with were recorded. increasing contact time between the Cm(lll) and CSH phases. 3 RESULTS 3.1 Batch sorption experiments Eu(lll) sorption kinetics were studied at S:L ratios of 3.8-10"4 kg L"1 and 5.4-10"3 kg L"1, and a total Eu(lll) concentration of 2-10"9 M. The sorption kinetics were fast: sorption equilibrium was reached within one day, and the Rd values were found to remain constant over a time period of 90 days. Moreover, very high 5 -1 distribution ratios (Rd = 6(±3)-10 Lkg ) were determined. An Eu(lll) sorption isotherm was determined at an S:L ratio of 5-10"4 kg L"1, and an equilibration time of one week. The total Eu(lll) concentration in the Wavelength (nm) experiments was varied between 10"10 M and 7-10"7 Fig. 3: Fluorescence emission spectra of 10"7M M. The data are shown in Fig. 2. The isotherm was 1 found to be linear over the entire concentration range Cm(lll) in CSH suspension (0.05 g L" ) of interest (slope = 1.05±0.06). measured at different Cm(lll)/CSH contact times. The high uncertainty in the measured Rd values may originate from the extremely strong sorption of Eu(lll) In all spectra, a peak with a maximum at -606.0 nm on the CSH phases, and therefore incomplete phase was observed. Its position and intensity ratio with separation: e.g. colloidal material remaining in respect to the main peak did not change with solution, or which are re-suspended after increasing Cm(lll)/CSH contact time. Note that this 108 peak was not observed in spectra of the same system the individual species from the composite obtained at 20 K. Therefore, it was concluded that this fluorescence emission spectra. Again, two differently peak is the result of visible transitions of thermally fluorescing Cm(lll) species sorbed on CSH were occupied states at room temperature, and has no found, with maxima at 618.9 nm (F1) and 620.9 nm chemical relevance. (F2), respectively. 1 1 1 1 1 1 1 1 1 1 1 Cm(III) sorbed species F2 oa 620.9 nm oa O O O Cm(III) sorbed species F1 1J 618.9 nm M os o o <3 N a> N O £ lL i 580 590 600 610 620 630 640 I 600 610 620 640 Wavelength (nm) Wavelength (nm) Fig. 4: Fluorescence emission spectra of the Fig. 5: Fluorescence emission spectra of 10"7M fluorescing Cm(lll) sorbed species F1 and F2 -i1\ as derived by peak deconvolution; the spectra Cm(lll) in a CSH suspension (0.5 g L" ) after a were scaled to the same peak area. Cm(lll)/CSH contact time of 58 days at different delay times; the spectra were scaled to the same peak area. In order to determine the number of Cm(lll) species, and quantify their concentrations, the set of spectra was submitted to a factor analysis based on an In Fig. 6, the Cm(lll) fluorescence emission intensity eigenvector analysis using the procedure described (on a logarithmic scale) is shown as a function of the delay time after a Cm(lll)/CSH contact time of 60 days by Rossberg [12]. Two different fluorescing Cm(lll) 5 1 species were identified (Fig. 4), with peak maxima at (10 M Cm(lll) and 5-10" kg L" CSH). The decay 618.9 nm (F1) and 620.9 nm (F2). The batch sorption behaviour follows a bi-exponential decay law. The experiments with Eu(lll) indicate that under the given lifetimes of F1 and F2 were determined to be t = 289 experimental conditions more than 99% of the Cm(lll) ±11 |as and t = 1482 ± 200 |as, respectively. is sorbed on the CSH phases. Hence the two Cm(lll) According to Eq. 1, a lifetime of 298 + 11 |js species identified with TRLFS must be sorbed corresponds to 1.4 water molecules in the first species. coordination shell of the Cm(lll) F1 complex. A lifetime of 1482 + 200 \is indicates the total loss of the The fluorescence lifetime provides information on the hydration sphere of the Cm(lll) F2 complex. composition of the first coordination sphere of the fluorescing species. According to the method developed by Horrocks Jr. & Sudnick [13] for Eu(lll), 1 7 and later applied to Cm(lll) by Kimura & Choppin [14], 0.05 g L" CSH; 10" M Cm(lll) - Bi-exponential Fit the following linear correlation between the decay rate and the number of H20 molecules in the first coordination sphere applies: E CD 0) e1€ n(H2Q) = 0.65kobs - 0.88 (1) t1=289 +/- 11 |is/1-2 H20 t =1482+/- 109 ns/OHO in which n(H20) is the number of coordinated water molecules, and kobs is the reciprocal lifetime of the 1 excited state [ms" j. contact time 60 days 3+ Cm(H20)9 has a lifetime of 68 ±3 \is [15]. When no water molecules are coordinated, the lifetime 500 1000 1500 2000 2500 3000 increases to 1250 ± 80 jus. Decay time (|is) Figure 5 shows selected normalized fluorescence emission spectra of a Cm(lll)/CSH sample (10" M Fig. 6: Time dependency of emission decay of 10"7 M Cm(lll) and 5-10"4 kg L"1 CSH), measured after a Cm(lll) in a CSH suspension (0.05 g L"1) after contact time of 58 days at different delay times. With a Cm(lll)/CSH contact time of 60 days; the bi- increasing delay time, the peak, with a maximum at exponential decay behaviour should be noted. -620 nm, exhibits a red shift from 619.6 nm to 620.9 nm. A peak deconvolution was carried out to separate 109 4 DISCUSSION Fig. 7), Ca is coordinated with 7 oxygen atoms originating from Si tetrahedral (stable Ca). The 4.1 Time-dependent fluorescence intensity of fluorescing Cm(lll) species, F1, corresponds to Cm(lll) Cm(lll) in CSH suspensions substituting for Ca in the interlayer (type 1 Ca), or Ca At the beginning of the sorption kinetic experiments, bound in the type 2 sites of the Ca octahedral layers, no fluorescence signal could be detected. With time, whereas the fluorescing Cm(lll) species, F2, can be an increasing fluorescence intensity was observed. interpreted as being a Cm(lll) substituting for Ca in the The absence of a fluorescence signal is attributed to type 3 sites of the Ca octahedral layers. concentration quenching due to clustering of the Cm(lll) ions. This clustering can occur via the formation of separate Cm(OH)3 colloids, or via the formation of Cm(lll) clusters on the CSH surfaces * s * # x (surface precipitation). 4.2 Identification of the different sorption sites Interlayer space with "labile" Ca The sorption of Eu(lll) and Cm(lll) on CSH phases under high pH conditions is characterised by chain of silicate tetrahedra extremely high sorption values. This was already indicated in an earlier study [16]. Layer of "stable" Ca The fast sorption kinetics and the linear sorption behaviour suggest an adsorption-like process for Eu(lll), involving only one sorbed species and a single sorption site. The information obtained from TRLFS, on the other hand, is interpreted in terms of three Fig. 7: A three-dimensional view of the tobermorite different Cm(lll) species: non-fluorescing Cm(lll) structure according to Hamid [18]. Note that clusters, which form very quickly, and two fluorescing oxygens of H20 molecules are enlarged sorbed species with, respectively, 1 or 2 and no H20 compared to the oxygens from the silica molecules in their first coordination sphere. The tetraedra. Cm(lll) and Eu(lll) substitute for apparent discrepancy between these two three types of Ca with different coordination, observations is most probably due to the high labelled 1, 2 and 3. uncertainties of the batch sorption data. Note that the overall uncertainties on the sorption kinetics data and the sorption isotherm presented in Fig. 2 are almost ACKNOWLEDGMENTS one order of magnitude. Thus, small changes in the affinity for the sorbing cation between the different We thank Andre Rossberg of the Forschungszentrum sorption sites may then no longer be detectable. Rossendorf for providing the factor analysis program designed to analyse EXAFS, UV-vis and TRLFS The low number of H20 molecules in the first spectra. Our colleagues E. Curti and M. H. Bradbury coordination spheres of the sorbed species F1 and F2 are acknowledged for their many valuable indicate that both species must be incorporated into discussions, and their careful reviews of the the CSH structure. It is noted that the crystal ionic manuscript. Partial financial support was provided by radius of Eu(lll) (0.95A) and Cm(lll) (0.97A) are very the National Cooperative for the Disposal of close to the crystal ionic radius of Ca(ll) (0.99A) [17]. Radioactive Waste (Nagra), by the European Hence, it may be postulated that Eu(IN) and Cm(lll) Commission (contract number FIS5-1999-00157), and substitute for Ca(ll) in the CSH structure. by the Swiss Federal Office for Education and A structural model for Cm(lll) and Eu(lll) incorporation Science (on behalf of the European Commission, by CSH phases has been developed based on a contract number FIKW-CT-2000-00028). defect 11A Tobermorite structure; the crystal structure is reported by Hamid [18], and is reproduced in Fig. 7. REFERENCES The structure consists of layers of Ca octahedra [1] H.F.W. Taylor, "Cement Chemistry", Thomas ("stable Ca") linked through non-bridging oxygens to Telford Services Ltd, London, UK, 1997. chains of silicate tetrahedra on either side, forming Ca-silicate layers. In the interlayers between the Ca- [2] D.L. Cocke, M.Y.A. Mollah, "The Chemistry and silicate sheets, variable amounts of Ca atoms may be Leaching Mechanism of Hazardous Substances located ("labile" Ca). Thus, the Tobermorite structure in Cementitious Solidification/Stabilization contains Ca atoms at three different types of sites. Systems", Lewis Publishers, 1993. The "labile" Ca in the interlayers (labelled 1 in Fig. 7) [3] M.L.D. Gougar, B.E. Scheetz, D.M. Roy, is coordinated with 7 oxigen atoms, of which 2 "Ettringite and CSH Portland Cement Phases for originate from H 0 molecules. In the second type 2 Waste Ion Immobilization: A Review", Waste (labelled 2 in Fig. 7), Ca is coordinated with 6 oxygen Management, 16(4), 295-303 (1996). atoms originating from Si tetrahedra, and one oxygen from H20 (stable Ca). In the third type (labelled 3 in 110 [4] U. Berner, "A Thermodynamic Description of the [11] J.V. Beitz, "Laser-Induced Fluorescence Studies Evolution of Pore Water Chemistry and Uranium of Cm3+ Complexes in Solution", Radiochim. Speciation during the Degradation of Cement", Acta, 52/53, 35-39 (1991). PSI-Report 90-62, Paul Scherrer Institut, Villigen, [12] A. Rossberg, "Anwendung der Factorenanalyse Switzerland, 1990. auf die Rontgenabsorptionsspektroskopie zur [5] V. Moulin, J. Tits, C. Moulin, P. Decambox, P. Bestimmung der Speziation von Uran in Mauchien, O. De Ruty, "Compiexation Behaviour losungen", Ph.D. Thesis, Technische Universitat of Humic Substances towards Actinides and Dresden, Dresden, Germany, 2001. Lanthanides Studied by Time-Resolved Laser- [13] W.D. Horrocks Jr., D.R. Sudnick, "Lanthanide Ion Induced Spectrofluorometry", Radiochim. Acta Probe of Structure in Biology. Laser-Induced 58/59, 121-128 (1992). Luminescence Decay Constants Provide a Direct [6] T. Fanghanel, J. I. Kim, P. Paviet, R. Klenze, W. Measure of the Number of Metal-Coordinated Hauser, "Thermodynamics of Radioactive Trace Water Molecules", J. Am. Chem. Soc., 101, 334- Elements in Concentrated Electrolyte Solutions", 340 (1979). Radiochim. Acta, 66/67, 81-87 (1994). [14] T. Kimura, G.R. Choppin, "Luminescence Study [7] T. Stumpf, T. Fanghanel, "A Time-Resolved on Determination of the Hydration Number of Laser Fluorescence Spectroscopy (TRLFS) Cm(lll)", J. Alloys Comp., 213/214, 313-317 Study of the Interaction of Trivalent Actinides (1994). (Cm(lll)) with Calcite", J. Coll. Int. Sci., 249(1), [15] J.V. Beitz, D.L. Bowers, M.M. Doxtader, V.A. 119-122 (2002). Maroni, D.T. Reed, "Detection and Speciation of [8] M. Atkins, F.P. Glasser, L.P. Moroni, J.J. Jack, Transuranium Elements in Synthetic "Thermodynamic Modeling of Blended Cements Groundwater via Pulsed-Laser", Radiochim. at Elevated Temperatures (50°C to 90°C)", Acta, 44/45, 87-93 (1988). DoE/HMIP/RR/94.011, University of Aberdeen, [16] I. Pointeau, B. Piriou, M. Fedoroff, M.-G. Barthes, U.K., Aberdeen, 1993. N. Marmier, F. Fromage, "Sorption Mechanisms [9] J.V. Beitz, "Similarities and differences in trivalent of EU3+ on CSH Phases of Hydrated Cements", J. lanthanide- and actinides-ion solution absorption Coll. Int. Sci., 236, 252-259 (2001). spectra and luminescence studies", in K. A. [17] F.A. Cotton, G. Wilkinson, "Advanced Inorganic Gschneidner Jr., L. Eyring, G.R. Choppin, G.H. Chemistry", Interscience Publishers, New York, Lander (eds.): Handbook on the Physics and USA, 1972. Chemistry of Rare Earths, Vol. 18, pp. 159-195. North-Holland, Amsterdam, 1994. [18] S.A. Hamid, "The Crystal Structure of the 11A Natural Tobermorite Ca [S\ 0 . (0H)^ ]- 1H 0", [10] J.I. Kim, R. Klenze, H. Wimmer, "Fluorescence 223 3 7 5 5 2 Zeitschrift fur Kristallographie, 154, 189-198 spectroscopy of curium (III) and application", Eur. (1981). J. Solid State Inorg. Chem., 28, 347-356 (1991). 111 PUBLICATIONS LIST University Level and Other Teaching JACKEL B. "Aufbau und Nutzung eines W-188/Re-188 BART G. Generators", Seminar Vortrag Kernchemie und "Strahlenbelastung durch Kritikalitatsunfalle und Nuklear-Medizin, Philipps-Universitat, Marburg, verlorene Quellen", Proceedings SVA- Deutschland, 15 November 2002 Vertiefungskurs: Wirkung ionisierender Strahlung, Winterthur, Switzerland, 5-6 December 2002 NIFFENEGGERM. "Warmeubergang: Stationare und instationare DANG V.N. Temperaturverteilungen, lineare und nichtlineare "Overview of HRA and its Integration in Probabilistic Finite Elemente Analysen", Kurs Qualitatssicherung Safety Assessments", "HRA of the Post-Initiator bei FE-Analysen, Fachgruppe fur Mechanik und Response", Human Reliability Analysis Training Industrie, Institut fur Mechanik, ETH Zurich, Course, KKL Leibstadt, Switzerland, 10-11 April 2002 Switzerland, September/October 2002 DANG V.N. REER B. "Humanfaktoren, Human Reliability Analysis (HCR, "Analyse menschlicher Zuverlassigkeit fur THERP, SLIM)", im Rahmen der Vorlesung Instandhaltungsfehler", Human Reliability Analysis "Methoden der Risikoanalyse und des Training Course, KKL Leibstadt, Switzerland, 10-11 Risikomanagements", Dozenten Prof. W. Kroger, Dr. April 2002 A.V. Gheorghe, ETH Zurich, Switzerland, 26 November 2002 SCHENLER, W. "What's Optimal Anyway?", Seminar presented at DEGUELDRE C. FHS Brugg-Windisch, Brugg, Switzerland, 25 June "Impact d'un reacteur dans I'environnement", 2002 University of Geneva, Switzerland, Winter Semester, 2002-2003 SCHEIDEGGERA.M. "Molecular environmental science: Licht ins Dunkle DEGUELDRE C. dank Synchrotronstrahlung", Inaugural Lecture, ETH "Comportement des radionuclides dans les Zurich, Switzerland, 27 June 2002 environnements souterrains", University of Geneva, Switzerland, Winter Semester, 2002-2003 SCHEIDEGGERA.M. "Application of synchrotron techniques for structural GIMMI TH. analysis in chemistry, environmental sciences, "Porositat und Porenstruktur von Opalinuston - Wo physics and biology", Course at ETH Zurich, gibts denn da Probleme?" within the framework of the Switzerland, WS 2002/03 Forum Mineralogicum at the University of Bern, Bern, Switzerland, 27 May 2002 Publications in Scientific and Technical Journals GUNTHER-LEOPOLD I. ANALYTIS G.TH., CODDINGTON P. Strategische Ubung: "Kernbrennstoffe", in der "Analysis and sensitivity studies of postulated SB- Vorlesungsreihe Analytische Chemie V, ETH Zurich, LOCAs in the Muhleberg (KKM) BWR/4 by TRAC- Switzerland, November 2002 BF1", Annals of Nuclear Energy, 29, 1525-1549 HIRSCHBERG S. (2002) "Sustainable Energy Supply for Switzerland", Seminar BEHRINGER K., HENNIG D. presented at FHS Brugg-Windisch, Brugg, "A novel auto-correlation function method for the Switzerland, 11 June 2002 determination of the decay ratio in BWR stability HIRSCHBERG S. analysis", Annals of Nuclear Energy, 29, 1483-1504 "Comparative Analysis of Energy Systems", Cours (2002) "Physique des systemes energetiques II", EPFL BONHOURE I., SCHEIDEGGER A., WlELAND E., DAHN R. Lausanne, Switzerland, 19 June 2002 "Iodine species uptake by cement and CSH studied by HIRSCHBERG S. I K-edge X-ray absorption spectroscopy", "Nachhaltigkeitsindikatoren: Bewertung der Optionen", Radiochimica Acta, 90, 647-651 (2002) Vertiefungskurs "Nachhaltige Energienutzung", ETH BRADBURY M.H., BAEYENS B. Zurich, Switzerland, 28 November 2002 "Sorption of Eu on Na- and Ca-montmorillonites: HIRSCHBERG S., HECKT. Experimental investigations and modelling with cation "Health Risk of Major Energy Chains", SVA- exchange and surface complexation", Geochimica et Vertiefungskurs "Wirkung ionisierender Strahlung", Cosmochimica Acta, 66, 2325-2334 (2002) Winterthur, Switzerland, 5-6 December 2002 112 BRESSLERCH1., SAESM.1, CHERGUIM.1, GERAEDTS K.1, BRUGGEMAN C.1, MAES A.1, GROLIMUND D., ABELA R. VAN LOON L.R., ROSSBERG A.2, REICH T.2 "Towards structural dynamics research with ultrashort "Evidence for the existence of Tc(IV)-humic pulse lasers and synchrotron radiation", in substances species by X-ray absorption near-edge Femtochemistry and Femtobiology, A. Douhal and J. spectroscopy", Radiochimica Acta, 90, 879-884 Santamaria (eds.), World Scientific, Singapore, 449- (2002) 458 (2002) 1 University of Leuven, Heverlee, Belgium 2 1 University of Lausanne, Switzerland FZR Rossendorf, Germany 1 BRESSLER CH.1 SAES M.1, CHERGUI M.1' HELLWIG CH., KASEMEYER U., LEDERGERBERG. , 2 2 GROLIMUND D., ABELA R., PATTISON PH.1'2 BYUNG-HO LEE , YOUNG-WOO LEE , CHAWLA R. "Towards structural dynamics in condensed chemical "Interpretation of Experimental Results from systems exploiting ultrafast time-resolved X-ray Moderate-Power In-Pile Testing of a Pu-Er-Zr-Oxide absorption spectroscopy", J. Chem. Phys., 116, 2955- Inert Matrix Fuel", Annals of Nuclear Energy, 30, 287- 2966 (2002) 299 (2002) 1 1 University of Lausanne, Switzerland KKL Leibstadt, Switzerland 2 2 ESRF, Grenoble, France KAERI, Taejon, Korea CHAWLA R., JONEJA O.P., ROSSELET M., WILLIAMS T.1 HENNIGC.1, REICH T.2, DAHN R., SCHEIDEGGERA.M. "Definition and Analysis of an Experimental "Structure of uranium sorption complexes at Benchmark on Shutdown Rod Worths in LEU-HTR montmorillonite edge sites", Radiochimica Acta, 90, Configurations", Nuclear Technology, 139, 50-60 653-657 (2002) (2002) 1 FZR Rossendorf, Germany 2 University of Mainz, Germany 1 EGL, Zurich, Switzerland HERMANN A. CODDINGTON P., MACIAN R. "Thermal behavior of hydrogen in Zircaloy corrosion "A study of the performance of void fraction layers", Journal of Nuclear Materials, 302, 217-219 correlations used in the context of drift-flux two-phase (2002) flow models", Nuclear Engineering and Design, 215, HLAWATSCH S.1, GARBE-SCHONBERG C.D.2, 199-216(2002) LECHTENBERG F.3, MANCEAU A.4, TAMURA N.5, CODDINGTON P., MACIAN R., ZlMMERMANN M.A., KULIK D.A., KERSTEN M.6 CHAWLA R. "Trace metal fluxes to ferromanganes nodules from "Application and In-Depth Assessment of RETRAN- the Western Baltic Sea as a record for long-term 3D for Best Estimate Analysis of Nuclear Power Plant environmental changes", Chemical Geology, 182, Transients", Journal of Nuclear Science and 697-709 (2002) Technology, 39(9), 972-985 (2002) 1 GEOMAR, Kiel, Germany 2 CURTI E., DEGUELDRE C. University of Kiel, Germany 3 "Solubility and hydrolysis of Zr oxides: a review and Roentgen-Analytik-Service, Itzehoe, Germany 4 LGIT, Grenoble, France supplemental data", Radiochimica Acta, 90, 801-804 5ALS, LBNL, Berkeley, USA (2002) 6 Gutenberg University, Mainz, Germany DAHN R., SCHEIDEGGERA.M., MANCEAU A.1, CURTI E., HUMMEL W., BERNERU., CURTI E., PEARSON F.J.1, BAEYENS B., BRADBURY M.H., CHATEIGNER D.2 THOENEN T. "Th uptake on montmorillonite: a powder and "Nagra/PSI Chemical Thermodynamic Data Base polarized extended X-ray absorption fine structure 01/01", Radiochimica Acta, 90, 805-813 (2002) (EXAFS) study", Journal of Colloid and Interface 1 Ground Water Geochemistry, New Berne, USA Science, 249, 8-21 (2002) 1 1 LGIT, Grenoble, France HUMMELW., BERNERU., CURTI E., PEARSON F.J. , 2ISMRA, Caen, France THOENEN T. "Nagra/PSI Chemical Thermodynamic Data Base DAHN R., SCHEIDEGGERA.M., MANCEAU A.1, 01/01", Nagra Technical Report NTB 02-16, Nagra, SCHLEGELM.L.1, BAEYENS B., BRADBURY M.H., Wettingen, Switzerland, and Universal MORALES M.2 Publishers/uPublish.com, Parkland, Florida, USA, "Neoformation of Ni phyllosilicate upon Ni uptake on ISBN 158 112 620 4(2002) montmorillonite: a kinetics study by powder and 1 Ground Water Geochemistry, New Berne, USA polarized extended X-ray absorption fine structure spectroscopy", Geochimica et Cosmochimica Acta, INGOLD F., WICHSER J., ZUBLER R., STREITM., KOPAJTICZ., KONOS.1 66, 2335-2347 (2002) 1 LGIT, Grenoble, France "Fabrication of neptunium containing MOX fuel 2 University of Maine, Le Mans, France microspheres", Journal of Nuclear Science and Technology, Supplement 3, 717-720 (2002) 1 FLURYM. , GIMMI TH. 1JNC, Tokai, Japan "Solute diffusion", in: Methods of Soil Analysis: Part 4- Physical methods", J.H. Dane and G.C. Topp (eds.), 1323-1348 (2002) 1 Washington State University, Pullman, Washington, USA 113 JATUFF F., JONEJA O.P., MURPHY M., PRALONG C., NA B.C.1, COMETTO M., WYDLER P., TAKANO H.2, SEILER R., CHAWLA R. KELLETT M1. "Methods/Data Testing for Advanced LWR Fuel "OECD/NEA Comparison Calculations for an Designs and Operational Modes in the PROTEUS Accelerator-Driven System Using Different Nuclear Facility", Journal of Nuclear Science and Technology, Data Libraries", Journal of Nuclear Science and Supplement 2, 642-645 (2002) Technology, Supplement 2, 835-840 (2002) 1 OECD/NEA, Issy-les-Moulineaux, France KROMPHOLZ K., KALKHOF D. 2 JAERI, Ibaraki-ken, Japan "Size effect studies of the creep behaviour of a pressure vessel steel at temperatures from 700 to NAKAMURAM.1, NAKAJIMAY1, POUCHONM.A., 900°C", Journal of Nuclear Materials, 305 112-123 SEKINE N.1, HELLWIG CH. (2002) "Development of Sphere-Pac Nuclear Fuel Behavior Analysis Code", Journal of Nuclear Science and KARPOVI.K.1, CHUDNENKO K.V.1, KULIKD.A., Technology, Supplement 3, 701-704 (2002) 1 BYCHINSKII V.A. 1 JNC, Tokai, Japan "The convex programming minimization of five 1 1 1 2 thermodynamic potentials other than Gibbs energy in OZAWAT. , ABET. , KONOS. , BAKKERK. , 2 Geochemical Modeling", American Journal of Science, THESINGH H. , HELLWIG CH., INGOLD F., WALLIN H. 302, 281-311 (2002) "Design for irradiation tests in the HFR with innovative fuel", Trans. Am. Nucl. Soc., 86, 279-280 (2002) institute of Geochemistry SB RAS, Irkutsk, Russia 1 JNC, Tokai, Japan 2 NRG, Petten, The Netherlands KULIK D.A. "Sorption modelling by Gibbs energy minimisation: PELLONI S., CHAWLA R. towards a uniform thermodynamic database for "On the Feasibility of ADS-Representative Integral surface complexes of radionuclides", Radiochimica Experiments Using Limited Amounts of Low-Enriched Acta, 90, 815-832 (2002) U-Fuel", Journal of Nuclear Science and Technology, KULIK D.A. Supplement 2, 1224-1227 (2002) "Gibbs energy minimization approach to modeling PELLONI S., HAGER H., JONEJA O.P., CHAWLA R. sorption equilibria at the mineral-water interface: "Data Sensitivity of Design Calculations for High thermodynamic relations for multi-site-surface Conversion D20 Reactors", Journal of Nuclear complexation", American Journal of Science, 302, Science and Technology, Supplement 2, 880-883 227-279 (2002) (2002) KULIK D.A., KERSTEN M.1 PFINGSTEN, W. "Aqueous solubility diagrams for cementitious waste "Experimental and modeling indications for self- stabilization systems: 4. A carbonation model for Zn- sealing of a cementitious L&ILW repository by calcite doped calcium silicate hydrate by Gibbs Energy precipitation", Nuclear Technology, 140(1), 63-82 Minimization", Environmental Science & Technology, (2002) 36, 2926-2931 (2002) 1 University of Mainz, Germany PFINGSTEN W., KISLITSINS.1 1 "Radionuclide migration from underground nuclear LAKEHAL D. , SMITH B. L., MILELLI M. "Large-eddy simulation of bubbly turbulent shear tests performed at the Lira facility, Kazakhstan — flows", IOP Journal of Turbulence, 3(25), 1-21 (2002) estimations by source term and transport modelling", ModelCARE 2002: Calibration and reliability in 1 ETH, Zurich, Switzerland groundwater modelling, K. Kovar and Z. Hrkal (eds.), MACIAN R., CODDINGTON P. Prague, Czech Republic, June 16-20, 2002. "Simulation and Analysis of Experimental Blowdown Univerzita Karlova V Praze Nakladatelstvi Karolinum, and Small-Break Loss-of-Coolant Scenarios with Acta Universitatis Carolinae-Geologica 46(2/3), 461- RETRAN-3D", Nuclear Technology, 139, 185-204 466 (2002) 1 Atomic Energy Committee of the Republic of Kazakhstan, Almaty, (2002) Kazakhstan MEIER M.1, YADIGAROGLU G.1, SMITH B. L. "A Novel Technique for Including Surface Tension in POUCHONM.A., NAKAMURAM.1, HELLWIG CH. PLIC-VOF Methods", Eur. J. Mech./Fluids, 21, 61-73 "A Mixed Ceramic-Metal Sphere-Pac Concept", (2002) Journal of Nuclear Science and Technology, ETH, Zurich, Switzerland Supplement 3, 830-833 (2002) 1 JNC, Tokai, Japan MURPHY M., LUTHI A. SEILER R., GRIMM P., 1 1 1 JONEJA O.P., MEISTER A., VAN GEEMERT R., JATUFF F., SCHUESSLERW. , METZV. , LUTZENKIRCHEN J. , 1 BROGLI R., JACOT-GUILLARMOD R.1, WILLIAMS T.1, KIENZLERB. , PFINGSTEN W. HELMERSSON S.2, CHAWLA R. "Modelling of coupled transport reaction processes — "Neutronics Investigations for the Lower Part of a summary and conclusions of the TRePro 2002 Westinghouse SVEA-96+ Assembly", Nuclear Science Workshop held at Karlsruhe", atw - Internationale Zeitschrift fur Kernenergie, 47(8/9), 668-669 (2002) & Engineering, 141, 32-45 (2002) 1 1 EGL, Zurich, Switzerland FZK, Karlsruhe, Germany 2 Westinghouse Atom AG, Vasteras, Sweden 114 SHEPEL S.V. NIFFENEGGERM., REICHLIN K., KALKHOF D. "Numerical Simulation of Filling and Solidification of "The Change of the Seebeck Coefficient due to Permanent Mold Castings", PhD Dissertation, Neutron Irradiation and Thermal Fatigue of Nuclear University of Notre Dame, USA, August, (2002), also, Reactor Pressure Vessel Steel and its Application to together with Paolucci S.1, in Applied Thermal the Monitoring of Material Degradation", PSI Bericht Engineering, 22, 229-248 (2002) Nr. 02-11, 1 University of Notre Dame, Indiana, USA RITTERS., SEIFERTH.P. 1 2 STREIT M., INGOLD F., GAUCKLER L.J. , OTTAVIANI J.-P. "Characterisation of the Lower Shell and Weld "(Pu,Zr)N annular fuel pellets shaped by direct Material of the Biblis C Reactor Pressure Vessel", PSI coagulation casting", Journal of Nuclear Science and Bericht Nr. 02-01 Technology, Supplement 3, 741-744 (2002) 1 ETH, Zurich, Switzerland RITTERS., SEIFERTH.P. 2 CEA Cadarache, St. Paul-lez Durance, France "Effect of a Sulphate Transient on the EAC Crack Growth Behaviour of Low-Alloy RPV Steels under THORENZC.1'2, KOSAKOWSKI G., KOLDITZO.3, 2 Simulated BWR Operating Conditions" (CASTOC WP BERKOWITZ B. 3, PSI Test 1), PSI Bericht Nr. 02-09 "An experimental and numerical investigation of saltwater movement in coupled saturated/partially- SEIFERTH.P. saturated systems", Water Resources Research, 38 "Literature Survey on the Stress Corrosion Cracking of (6), doi: 10.1029/2001WR000364 (2002) Low-Alloy Steels in High-Temperature Water", PSI 1 University of Hannover, Germany Bericht Nr. 02-06 2Weizmann Institute of Science, Rehovot, Israel 3 University of Tubingen, Germany SEIFERTH.P., RITTERS. "Environmentally-Assisted Cracking of Low-Alloy VAN GEEMERT R., JATUFF F., GRIMM P., CHAWLA R. Reactor Pressure Vessel Steels under Boiling Water "Assessment of Reactivity Effects due to Localized Reactor Conditions", PSI Bericht Nr. 02-05 Perturbations in BWR Lattices", Nuclear Science and Engineering, 142, 96-106 (2002) SEIFERTH.P., RITTERS. "Effect of a Chloride Transient on the EAC Crack VULAVAV.M.1, KRETZSCHMAR R.2, BARMETTLER K.2, Growth Behaviour of Low-Alloy RPV Steels under VOEGELIN A.2, GROLIMUND D., BORKOVEC M.3 Simulated BWR Operating Conditions" (CASTOC "Cation Competition in a Natural Subsurface Material: WP3, PSI Test 2), PSI Bericht Nr. 02-23 Prediction of Transport Behavior", Water Resources Research 38, 7-1-7-7 (2002) TITS J., WIELAND E., BRADBURY M.H., ECKERT P., 1 University of Tennessee, Knoxville, USA SCHAIBLE A. 2 ETH, Zurich, Switzerland 3 "The uptake of Eu(lll) and Th(IV) by calcite under University of Geneva, Switzerland hyperalkaline conditions", PSI-Bericht Nr. 02-03, WIELANDE., TITS J., DOBLERJ.P., SPIELER P. Nagra NTB 02-08 "The effect of a-isosaccharinic acid on the stability of and Th(IV) uptake by hardened cement paste", Conference Proceedings Radiochimica Acta, 90, 683-688 (2002) ABOLHASSANI S., DADRAS M.1, LEBOEUF M.1 "In-situ Observation of Oxidation of Zircaloy-4", PSI Reports Proceedings of the 15th International Corrosion BERNER U. Congress, 22-27 September 2002, Granada, Spain, p. 349-357 "Project opalinus clay: radionuclide concentration 1 limits in the near-field of a repository for spent fuel and University of Neuchatel, Switzerland vitrified high-level waste", PSI Bericht Nr. 02-22, ADDABBO C.1, ANNUNZIATO A.1, AKSAN N., D'AURIA F2, Nagra NTB 02-10 GALASSI G.2, DUMONT D.3, NILSSON L.4, RIGAMONTI M.5, RIIKONEN V.6, STEINHOFF F.7, TOTH I.8, UMMINGER K.9 BRADBURY M.H., BAEYENS B. "European Network for the Consolidation of the "Porewater chemistry in compacted re-saturated MX- Integral System Test Experimental Data Bases for 80 bentonite: physico-chemical characterisation and Reactor Thermal-Hydraulic Safety Analysis (CERTA)", geochemical modelling", PSI Bericht Nr. 02-10, Nagra FISA-2001-EU Research in Reactor Safety, 12-14 NTB 01-08 November 2001, Luxembourg, EUR 20281, 581-590 BROGLI R., KRAKOWSKI R.A.1 1 JRC, Ispra, Italy 2 "Degree of Sustainability of Various Nuclear Fuel University of Pisa, Italy 3 Cycles", PSI Bericht Nr. 02-14 CEA, Grenoble, France 4 1 LANL, Los Alamos, USA Studsvik Ecosafe, Nykoping, KUREPIN V.A., KULIK D.A., HILTPOLD A., NICOLET M. 5SIET, Piacenza, Italy "Thermodynamic modelling of Fe-Cr-Ni spinel VTT, Espoo, Finland formation at Light Water Reactor conditions", PSI GRS, Garching, Germany Bericht Nr. 02-04 g KFKI, Budapest, Hungary Framatome ANP, Erlangen, Germany 115 ADROGUER S.1, CHATELARD P.1, BARTEN W., CODDINGTON P., FERROUKHI H. VAN DORSSELAERE J.P.1, DURIEZ C.1, COCUAUDN.1, "Peach Bottom BWR Turbine Trip Benchmark: PSI BELLENFANT L.2, BOTTOMLEY D.3, VRTILKOVA V.4, Analysis of Exercise 1 using RETRAN-3D", Physor MUELLER K.5, HERING W6, HOMANNC.6, KRAUSSW6, 2002, 7-10 October 2002, Seoul, Korea, CD-ROM MIASSOEDOVA.6, STEINBRÜCK M.6, HOZERZ.7, o Q 1 n BERNER U. BANDINIG. , BIRCHLEYJ., BERLEPSCH T. v , BUCKM. , 11 12 13 "Solubility calculations and their interpretation in PA", BENITEZ J.A. F. ,VIRTANEN E. , MARGUETS. , in "The use of thermodynamic databases in AZARIAN G.14, PLANK H.15, VESHCHUNOV M.16, 17 performance assessment", Workshop Proceedings, ZVONAREVY. Barcelona, Spain, 29-30 May 2001, OECD/NEA, "CORE LOSS During A Severe Accident (COLOSS)", Paris, France ISBN 92-64-198486-6, 139-149 FISA-2001-EU Research in Reactor Safety, 12-14 November 2001, Luxembourg, EUR 20281, 247-260 BERNER U., KULIKD.A. th 1 CEA-IPSN-Cadarache, St. Paul-lez-Durance, France "Ca-AI-hydrates: solid solutions?" 12 Annual 2 CEA-IPSN, Fontenay-aux-Roses, France 3 Goldschmidt Conference, 18-23 August 2002, Davos, JRC, Karlsruhe, Germany Switzerland, Geochimica et Cosmochimica Acta, 66, 4 SKODA-ÜJP, Zbraslav-Praha, Czech Republic A73 (2002) 5 JRC, Petten, The Netherlands BESTION D.1, PAILLÈREH.1, LATROBEA.1, LAPORTAA.2, FZK, Karlsruhe, Germany TESCHENDORFF V.3, STÄDTKE H.4, AKSAN N., 7 AEKI, Budapest, Hungary D'AURIA F.5, VIHAVAINEN J.6, MELONI P.7, HEWITT G.8, 8 ENEA, Bologna, Italy g LILINGTON J.9, PROSEKA.10, MAVKOB.10, MACEKJ.11, RUB, Bochum, Germany 11 12 12 10 MALACKAM. , CAMOUS F. , FICHOTF. , IKE, Stuttgart, Germany MONHARDTD.13 11 UPM, Madrid Spain 12 "European Project for Future Advances in Science LTKK, Lappeenrenta, Finland 13 and Technology for Nuclear Engineering Thermal EDF, Clamart, France 14 Hydraulics (EUROFASTNET)", FISA-2001-EU Framatome-ANP, Paris, France 15 Research in Reactor Safety, 12-14 November 2001, Framatome-ANP, Erlangen, Germany Luxembourg, EUR 20281, 591-599 16 NSI-IBRAE, Moscow, Russia 17 CEA, Grenoble, France NSI-KI, Moscow, Russia 2 EDF, Chatou, France 1 2 AKSAN N., BITTERMANN D. , DUMAZ P. , 3 GRS, Köln, Germany 3 4 5 KYRKI-RAJAMAKI R. , MARACZYC. , OKAY. , 4 JRC, Ispra, Italy 6 6 7 SCHULENBERG T. , SQUARER D. , SOUYRI A. 5 University of Pisa, Italy "A High Performance Light Water Reactor Concept", 6 Lappeenranta University of Technology, Finland Proc. Jahrestagung Kerntechnik 2002, 14-16 May 7 ENEA, Bologna, Italy 2002, Stuttgart, Germany ISSN 0720-9207, 577-580 8 Imperial College, London, U.K. 1 Framatome ANP, Erlangen, Germany 9 AEA Technology, Didcot, U.K. 2 10 CEA-Cadarache, St. Paul-lez-Durance, France Institut Josef Stefan, Ljubljana, Slovenia 3 11 VTT, Espoo, Finland NRI, Rez, Czech Republic 12 KFKI, Budapest, Hungary CEA-IPSN-Cadarache, St. Paul-lez-Durance, France 13 University of Tokyo, Japan Framatome ANP, Paris, France FZK Karlsruhe, Germany BONHOURE I., WIELAND E., SCHEIDEGGER A.M., TlTS J. EDF, Chatou, France "Radionuclide immobilization by cement and cement ANALYTIS G.TH. phases studied by EXAFS", 12th Annual Goldschmidt "Implementation and Assessment of the Conference, 18-23 August 2002, Davos, Switzerland, Renormalization Group (RNG), Quadratic and Cubic Geochimica et Cosmochimica Acta, 66, A90 (2002) Non-Linear Eddy Viscosity k - s Models in GOTHIC", BRUCHERTSEIFERH., CRIPPS R., GUENTAYS., 9th International Conference on Nuclear Engineering JÄCKEL B. (ICONE-9), 8-12 April 2001, Nice, France, CD-ROM "Fission product iodine release and retention in ANALYTIS G.TH., CODDINGTON P. nuclear reactor accidents - experimental programme th "Analysis of Different Hypothetical Recirculation Line at PSI", 14 Radiochemical Conference, 14-19 April SB-LOCAs in a BWR/4 by TRAC-BF1 A/98.1", 9th 2002, Marianske Lazne, Czech Republic, 320 International Conference on Nuclear Engineering BRUCHERTSEIFERH., CRIPPS R., GUENTAYS., (ICONE-9), 8-12 April 2001, Nice, France, CD-ROM JÄCKEL B. th AOUNALLAH Y. "Analysis of iodine species in aqueous solutions", 12 "CORETRANA/IPRE Assembly Critical Power Euroanalysis, Dortmund, Germany, 8-13 September, Assessment", 9th International Conference on Nuclear 2002,617 Engineering (ICONE-9), 8-12 April 2001, Nice, France, CD-ROM 116 CHANG Y.H.1, MOSLEH A.2 DOKHANE A., HENNIG D., RIZWAN-UDDIN1, CHAWLA R. "ADS: A Computer Program for Dynamic Probabilistic "A Parametric Study of Heated Channels with Two- Risk Assessment", Proceedings of the ANS Phase Flow Using Bifurcation Analysis", American International Topical Meeting on Probabilistic Safety Nuclear Society Annual Meeting, 9-13 June 2002, Assessment, 6-9 October 2002, Detroit, Michigan, Florida, USA, Trans. ANS, Vol. 86, 131-133 (2002) USA, 206-211 1 University of Illinois, Urbana-Champaign, USA 1 PSI, and University of Maryland, College Park, USA DOKHANE A., HENNIG D., RIZWAN-UDDIN1, CHAWLA R. 2 University of Maryland, College Park, USA "Nonlinear Stability Analysis with a Novel BWR CURTI E., BERNERU. Reduced-Order Model", Physor 2002, 7-10 October "Solubility of Ra in a radioactive repository 2002, Seoul, Korea, CD-ROM th environment", 12 Annual Goldschmidt Conference, 1 University of Illinois, Urbana-Champaign, USA 18-23 August 2002, Davos, Switzerland, Geochimica DURYT. V. et Cosmochimica Acta, 66, A161 (2002) "CFD Design Support at PSI for the MEGAPIE Liquid- DAHN R., SCHEIDEGGERA.M. , MANCEAU A.1, Metal Spallation Target", European CFX Conference BAEYENS B., BRADBURY M.H. 2002, 16-18 September 2002, Strasbourg, France, "Surface reactivity of clay nanoparticles as studied by CD-ROM th polarized EXAFS", 12 Annual Goldschmidt DURYT. V. Conference, 18-23 August 2002, Davos, Switzerland, "Simulation with CFX-4.3 of Steady-State Conditions Geochimica et Cosmochimica Acta, 66, A163 (2002) in a 1/5th-Scale Model of a Typical 3-Loop PWR in the 1 LGIT, Grenoble, France Context of Boron-Dilution Events", IAEA, OECD/NEA DANG V. N. Technical Meeting on the Use of Computational Fluid "Applications of an Operator-Plant Model for Dynamic Dynamics (CFD) Codes for Safety Analysis of Reactor Risk Assessment", in E. J. Bonano et al. (eds.), Systems, Including Containment, 11-14 November Proceedings of the 6th International Conference on 2002, Pisa, Italy, CD-ROM Probabilistic Safety Assessment and Management FERROUKHI H., CODDINGTON P. (PSAM 6), 23-28 June 2002, San Juan, Puerto Rico, "Analysis of Sensitivity Studies with CORETRAN and USA, Vol. I, 183-188 RETRAN-3D of the NEACRP Rod Ejection th DANG V.N., REERB., Benchmark", 9 International Conference on Nuclear "Decision and Commission Errors - From Engineering (ICONE-9), 8-12 April 2001, Nice, Identification to Quantification", in: J. J. Persensky et France, CD-ROM (2002) al. (eds), New Century, New Trends, Proceedings of FERROUKHI H., BARTEN W., CODDINGTON P. the 2002 IEEE 7th Conference on Human Factors and "Transient 3-D Neutron Kinetic Analysis with Power Plants, 15-19 September 2002, Scottsdale, CORETRAN of a Core Thermal-Hydraulic Boundary Arizona, Institute for Electrical and Electronical Condition Model - Peach Bottom 2 Turbine Trip Engineers, New York, 2002, 3.26 - 3.33 Benchmark Phase 2", Physor 2002, 7-10 October DANG V.N., REERB., HIRSCHBERG S., 2002, Seoul, Korea, CD-ROM "Analyzing Errors of Commission: Identification and a GAVILLET D., MARTIN M., BLANC J.Y.1, THOMAZET J.2, First Assessment for a Swiss Plant", in Proceedings of MIQUET A.3 the OECD NEA Workshop, "Building the New HRA: "Investigation of the Li and B distribution on in-pile Errors of Commission - from Research to irradiated Zircaloy-4 and Zr-1%Nb-0 oxide layers", Application", 7-9 May 2001, Rockville, USA, CD-ROM Proceedings Chimie 2002, International Conference 1 2 on Water Chemistry in Nuclear Reactors Systems, 22- DICKINSON S. , SIMS H.E. , FUNKE F., GUENTAY S., 3 4 26 April 2002, Avigon, France, CD-ROM BRUCHERTSEIFERH., LILJENZIN J.-O. , LIGER K. , 1 4 5 6 CEA, Gif-sur-Yvette, France MONTANELLI T. , KRAUSMANN E. , RYDLA. 2 Framatome ANP, Paris, France "Iodine chemistry and mitigation mechanisms 3 EDF, Villeurbanne, France (ICHEMM)", FISA-2001-EU Research in Reactor GLAUS M.A., BAEYENS B., LAUBERM., RABUNGT.1, Safety, 12-14 November, 2001, Luxembourg, EUR VAN LOON, L.R. 20281, 369-381 1 "The effect of organic matter extracted from Mont Terri AEA Technology, Winfrith, U.K. 2 Framatome ANP, Erlangen, Germany and Benken, Opalinus Clay on the speciation of th 3 Chalmers University of Technology, Sweden radionuclides", 8 International Conference on the 4 CEA/IPSN, Grenoble, France chemistry and migration behaviour of actinides and 5 JRC, Petten, The Netherlands fission products in the geosphere, Abstracts PB4-04, 6 NRI, Rez, Czech Republic 91-92, Bregenz, Austria, 16-21 September 2001 DOKHANE A., HENNIG D., RIZWAN-UDDIN1, CHAWLA R. 1 FZK Karlsruhe, Germany "Stability and Bifurcation Analyses of Two-Phase Flow GRIMM P., JATUFF F., LUTHI A., MURPHY M., Using a Drift Flux Model and Bifurcation Code 1 th SEILER R., VAN GEEMERT R., WILLIAMS T. , CHAWLA R. BIFDD", 10 International Conference on Nuclear "Channel Bowing Effects on Pin Power Distributions in Engineering (ICONE-IO), 14-18 April 2002, Arlington, a Westinghouse SVEA-96+ Assembly", Physor 2002, Virginia, USA, CD-ROM 7-10 October 2002, Seoul, Korea, CD-ROM 1 University of Illinois, Urbana-Champaign, USA 1 EGL, Zurich, Switzerland 117 GROLIMUND D., BARMETTLER K.1, BORKOVEC M.2 HOGENBIRK A.1, MEULEKAMP R. K.1, MANSANI L.2, "Colloid-facilitated transport of pollutants: phenomena GLINATSIS G.3, CETNAR J.4, PELLONI S., VICENTE C.5 and modelling", 12th Annual Goldschmidt Conference, "Uncertainty Evaluation of the Nuclear Design of PDS- 18-23 August 2002, Davos, Switzerland, Geochimica XADS", ENC 2002, 7-9 October 2002, Lille, France, et Cosmochimica Acta, 66, A293 (2002) CD-ROM 1 ETH Zurich, Switzerland 1 NRG, Petten, The Netherlands 2 University of Geneva, Switzerland 2 Ansaldo Nucleare, Genoa, Italy 3 GROLIMUND D., WARNER J.A.1, CARRIERX.X.2, ENEA, Bologna, Italy 4 Faculty of Physics, Krakow, Poland BROWN JR. G.E.3 5 CIEMAT, Madrid, Spain "Spreading of reactive chemicals in natural porous media: from the atomic to the field scale", TRePro 2002 HUMMEL W. (International Conference on Modelling of Coupled "Consistent evaluation of acid base equilibria in NaCI Transport Reaction Processes), 20-21 March 2002, media to high temperatures", 8th International Karlsruhe, Germany, Wissenschaftliche Berichte, FZKA Conference on the chemistry and migration behaviour 6712, 23-24 (2002) of actinides and fission products in the geosphere, 16- 1 Lawrence Berkeley National Laboratory, Berkeley, USA 21 September 2001, Bregenz, Austria, Abstracts PC1- 2 Universite Pierre et Marie Curie, Paris, France 3 08, 108-109 SLAC, Stanford, CA, USA GROLIMUND D., BORKOVEC M.1 HUMMEL W. "Quantitative modelling of colloid facilitated transport in "Can we estimate which organic complexes are natural porous media", TRePro 2002 (International important in PA", in "The use of thermodynamic Conference on Modelling of Coupled Transport databases in performance assessment", Workshop Reaction Processes), 20-21 March 2002, Karlsruhe, Proceedings, 29-30 May 2001, Barcelona, Spain, Germany, Wissenschaftliche Berichte, FZKA 6712, 25- OECD/NEA, Paris, France, 83-92 (2002) 28 (2002) HUMMEL W. 1 University of Geneva, Switzerland "The Influence of cyanide complexation on the GUNTAYS., DEHBI A., SUCKOWD., BIRCHLEYJ. speciation of radionuclides", 12th Annual Goldschmidt "ARTIST: An International Project Investigating Conference, 18-23 August 2002, Davos, Switzerland, Aerosol Retention in a Ruptured Steam Generator", Geochimica Cosmochimica Acta, 66, A348 (2002) International Congress on Advanced Nuclear Power JOKINIEMI J.1, LAHDEA.1, TUOMISTOH.2, ROUTAMO T.2, Plants (ICAPP), 9-13 June 2002, Hollywood, Florida, LUNDSTROMP.2, DIENSTBIERJ.3, GUNTAYS., USA, CD-ROM SUCKOWD., BAKKERP.4, HERRANZL.5, PEYRES V.5 GUNTAYS., DEHBI A., SUCKOWD., BIRCHLEYJ. "Steam Generator Tube Rupture Scenarios (SGTR)", "The PSI Artist Project: Aerosol Retention and FISA-2001-EU Research in Reactor Safety, 12-14 Accident Management Issues Following a Steam November 2001, Luxembourg, EUR 20281, 433-444 th Generator Tube Rupture", 10 International 1 VTT, Espoo, Finland Conference on Nuclear Engineering (ICONE-IO), 14- 2 Fortum Nuclear Services Ltd, Keilaniemi, Finland 18 April 2002, Arlington, Virginia, USA, CD-ROM 3 NRI, Rez, Czech Republic GUNTAYS., DEHBI A., SUCKOWD., BIRCHLEYJ. 4 NRG, Petten, The Netherlands "ARTIST: A Multinational Experimental Project to 5 CIEMAT, Madrid, Spain Investigate Fission Product Retention in a Ruptured JONES A. V.1, DICKINSON S.2, DE PASCALE C.3, Steam Generator", Jahrestagung Kerntechnik 2002, HANNIET N.3, HERRANZ L.4, DE ROSA F.5, 14-16 May 2002, Stuttgart, Germany, 143-148 HENNEGES G.6, LANGHANS J.7, HOUSIADAS C.8, 9 10 GUNTHER-LEOPOLD I., WERNLI B., KOPAJTICZ., WICHERSV. , BIRCHLEYJ., PACI S. , 11 "Isotope ratio measurements on transient signals by MARTIN-FUERTES F. an online coupled HPLC-MC-ICP-MS system", 12* "Validation of Severe Accident Codes Against Annual V.M. Goldschmidt Conference, 18-23 August PHEBUS FP for Plant Application (PHEBEN 2)", 2002, Davos, Switzerland, 18-23 August 2002, FISA-2001-EU Research in Reactor Safety, 12-14 Geochimica et Cosmochimica Acta, 66, Number 15A, November 2001, Luxembourg, EUR 20281, 456-467 A297 1 JRC, Petten, The Netherlands 2 AEA Technology, Winfrith, U K HELLWIG CH., WALLIN H., LASSMANN K.1 3 CEA-Cadarache, St. Paul-lez-Durance, France "RIA modeling-a comparison between TRANS- 4 URANUS and FREY", JAERI-Conf. 2002-009, Proc. CIEMAT, Madrid, Spain 5 Fuel Safety Research Specialists' Meeting, 4-5 March ENEA, Bologna, Italy 2002, JAERI, Tokai, Japan, 183-191 FZK, Karlsruhe, Germany 1 JRC, Karlsruhe, Germany GRS, Garching, Germany NCSR "Demokritos", Athens, Greece HERMANN A., BUHNER R., SCHLEUNIGER P. 9 NRG, Petten, The Netherlands 10 "Fuel rod puncturing and fission gas analysis: Politechnic University of Madrid, Spain 11 improvements in equipment and methodology", University of Pisa, Italy Jahrestagung Kerntechnik 2002, 14-16 May 2002, Stuttgart, Germany, 331-334 118 KALKHOF D., GROSSE M. KULIK D.A. "Influence of PbBi Environment on the LCF Behaviour "Minimising uncertainty induced by temperature of SNS Target Container Materials", 5th International extrapolations of thermodynamic data: a pragmatic Workshop on Spallation Materials Technology, 19-24 view on the integration of thermodynamic databases May 2002, Charleston, USA, CD-ROM into geochemical computer codes", in "The use of Thermodynamic Databases in Performance KALKHOF D., GROSSE M., NIFFENEGGERM. Assessment". Workshop Proceedings, Barcelona, "Monitoring of Fatigue Degradation in Austenitic Spain, 29-30 May 2001, OECD/NEA, Paris, France Stainless Steels", 2nd International Conference on 125-137 (2002) Fatigue of Reactor Components, 29-31 July 2002, Snowbird, USA, CD-ROM KULIK D.A. "Density parameters in thermodynamic surface KALKHOFD., NIFFENEGGERM., GROSSEM., BARTG., complexation models", Abstracts of the 12th Annual "Influence of Cycle Number, Temperature and Goldschmidt Conference, 18-23 August 2002, Davos, Manufacturing Process on Deformation-Induced Switzerland, Geochimica et Cosmochimica Acta, 66, Martensite in Meta-Stable Austenitic Stainless Steels", A423 (2002) 5th International Symposium on Contribution of Materials Investigation to the Resolution of Problems KULIK D.A., CURTI E., BERNER U. Encountered in Pressurized Water Reactors, 23-27 "Thermodynamic modelling of radionuclide retention in September 2002, Fontevraud, France, CD-ROM clays: surface complexation or solid solution equilibria?" in International Conference on Clays in KASEMEYER U., HELLWIG CH., CHAWLA R., DEAN D.W.1, Natural and Engineered Barriers for Radioactive MEIER G.2 WILLIAMS T.3 Waste Confinement, 9-12 December 2002, Reims, "Feasibility of Partial LWR Core Loadings with Inert France, 63-64 Matrix Fuel", ARWIF 2001, 22-24 October 2001, Chester, United Kingdom, CD-ROM MACIAN R., NECHVATAL L. 1 Studsvik Scandpower Inc., Beacon St., USA "Best Estimate Analysis of the Thermal Expansion 2 KKG, Daniken, Switzerland Scenario during Shutdown in a PWR", 9th International 3 EGL, Zurich, Switzerland Conference on Nuclear Engineering (ICONE-9), 8-12 KETELAAR K.C.J.1, DE KRUIJF W.J.M.2, AVAKIAN G.3, April 2001, Nice, France, CD-ROM (2002) 2 2 3 GUBERNEATIS P. , CARUGE D. , MANERA A. , MAZUREKM.1, JAKOB A. 3 4 VAN DER HAGEN T.H.J.J. , YADIGAROGLU G. , "Evidence for matrix diffusion in the TRUE-1 block at 5 6 6 DOMINICUS G. , ROHDEU. , PRASSERH.-M. , Aspo based on fracture characterisation and 7 CASTRILLO F. , HUGGENBERGER M., HENNIG D., modelling of tracer tests", in Radionuclide Retention in MUNOZ-COBO J.L.8, AGUIRRE C.9 Geologic Media, NEA/OECD, Paris, France, ISBN 92- "Natural Circulation and Stability Performance of 64-19695-1, 137-154 (2002) BWRs (NACUSP)", FISA-2001-EU Research in 1 University of Bern, Switzerland Reactor Safety, 12-14 November 2001, Luxembourg, EUR 20281, 535-558, MEISTER A., MURPHY M., LUTHI A., SEILER R., 1 GRIMM P., VAN GEEMERT R., JATUFF F., CHAWLA R., NRG, Petten, The Netherlands 1 2 JACOT-GUILLARMOD R. University of Technology, Delft, The Netherlands 3 "Experimental Investigation of Pin Removal Reactivity CEA-Cadarache, St. Paul-lez-Durance, France 4 Worths for a Westinghouse SVEA-96+ Assembly in ETH, Zurich, Switzerland 5 the PROTEUS Research Reactor", Physor 2002, 7-10 Forsmark Kraftgrupp AB, Stockholm, Sweden October 2002, Seoul, Korea, CD-ROM 6 FZR, Rossendorf, Germany 1 EGL, Zurich, Switzerland 7 Iberdrola S.A., Madrid, Spain Politechnic University of Valencia, Spain NORDSTROM L. A. 9 KKL Leibstadt, Switzerland "Comparative Irradiation and Transient Test of Pellet and Sphere-Pac Fuel (IFA-550.11)", Proc. Enlarged KROGER W., HIRSCHBERG S., FOSKOLOS K. HPG Meeting on Man-Machine Systems Research "The Attributes of Sustainability of Energy Options", and High Burn-Up Fuel Performance, Safety and MIT-CANES Symposium, 19-20 April 2001, Boston, Reliability and Degradation of In-Core Materials and USA, CD-ROM Water Chemistry Effects (HPR-359), 8-13 September KROGERW. 2002, Storefjell, Norway, Vol. 1, Paper 23 "ILK Recommendations on the Use of PSA in Nuclear PALADINO D., AUBAN O., CANDREIA P., Licensing and Supervision Process", 6th International HUGGENBERGER M., STRASSBERGERH. J. Conference on Probabilistic Safety Assessment and "New PANDA Tests to Investigate Effects of Light Management (PSAM6), San Juan, Puerto Rico, 27 Gases on Passive Safety Systems", Int. Congress on June 2002 Advanced Nuclear Power Plants (ICAPP), 9-13 June Workshop Proceedings, Barcelona, Spain, 29-30 May 2002, Hollywood, Florida, USA, CD-ROM 2001, OECD/NEA, Paris, France 125-137 (2002) 119 PAILLÈRE H.1, KUMBARO A.1, BESTION D.1, MIMOUNI S.2, SCHEIDEGGERA.M., GROLIMUNDD., CHEESEMAN C.R.1 LAPORTA A.2, STÄDTKE H.3, FRANCHELLO G.3, GRAF U.4, "A microspectroscopic study on the influence of the RÖMSTEDT P.4, TORO E.F.5, ROMENSKI E.5, inherent heterogeneity of waste repository materials th DECONINCK H.6, VALERO E.6, DE CACHARD F., SMITH B. on contaminant uptake", 12 Annual Goldschmidt "Advanced 3D Two-Phase Flow Simulation Tools for Conference, 18-23 August 2002, Davos, Switzerland, Application to Reactor Safety (ASTAR)", FISA-2001- Geochimica et Cosmochimica Acta, 66, A676 (2002) EU Research in Reactor Safety, 12-14 November imperial College of Science, London, UK 2001, Luxembourg, EUR 20281, 491-501 (2002) SEIFERTH.P., RITTERS. CEA-Cadarache, St. Paul-Iez-Durance, France "Effect of a Sulphate and Chloride Transient on the 2 EDF, Chatou, Fance 3 EAC Crack Growth Behaviour of two Low-Alloy RPV JRC, Ispra, Italy 4 GRS, Garching, Germany Steels in Oxygenated High-Temperature Water under 5 Manchester Metropolitan University, U.K Periodical Partial Unloading Conditions", Minutes of 6 VKI, Rhode-Saint-Genése, Belgium the 2002 Annual Meeting of the International PFINGSTEN W. Cooperative Group on Environmentally Assisted "Modelling cement-rock-water interactions - The Cracking of Light Water Reactor Materials, 14-19 influence of boundary conditions", Proc. Int. Workshop April 2002, Lyon, France, Ed.: J. Hickling, Low-Alloy on "Modelling of Coupled Transport Reaction Steel Session, Paper L6, CD-ROM Processes" - TRePro2002, 20-21 March 2002 (V. SEIFERTH.P., RITTERS. Metz, W. Pfingsten, J. Luetzenkirchen and W. "SICC and Low-frequency Corrosion Fatigue of Low- Schuessler, eds.), Karlsruhe, Germany, FZKA 6721: Alloy RPV Steels, of a Weld Filler and Weld HAZ 69-72 (2002) Material under BWR Conditions - Effect of PFINGSTEN W. Temperature and Loading Rate/Frequency", Minutes "The attempt to characterise the hydrogeological of the 2002 Annual Meeting of the International properties of a fractured shear zone by modelling of Cooperative Group on Environmentally Assisted several tracer dipole tests", 2002 IAHR International Cracking of Light Water Reactor Materials, 14-19 Groundwater Symposium on "Bridging the Gap April 2002, Lyon, France, Ed.: J. Hickling, Low-Alloy between Measurement and Modeling in Steel Session, Paper L4, CD-ROM Heterogeneous Media", Berkeley, CA, 24-29 March SMITH B. L., DURYT. V., Ni. L., ZUCCHINI A.1 2002, IAHR, Madrid, Spain, 240-245 (2002) "A Simple Coupling Strategy between CFX-4 and PLASCHY M., CHAWLA R., DESTOUCHES C.1, AN S YS/AB AQ US", European CFX Conference 2002, DOMERGUE C.1, SERVIÈRE H.1, CHAUSSONNET P.1, 16-18 September 2002, Strasbourg, France, CD-ROM 1 LAURENS J.M.1, SOULE R.1, RIMPAULT G.1 ENEA, Bologna, Italy "Foil activation studies of spectral variations in the SMITH B. L., MILELLI M., SHEPELS. MUSE4 critical configuration", Physor 2002, 7-10 "Aspects of Nuclear Reactor Simulation Requiring the October 2002, Seoul, Korea, CD-ROM 1 use of Advanced CFD Models", IAEA, OECD/NEA CEA-Cadarache, St. Paul-lez-Durance, France Technical Meeting on Use of Computational Fluid REERB., DANG V.N., HIRSCHBERG S. Dynamics (CFD) Codes for Safety Analysis of Reactor "Identifying and Assessing Errors of Commission - Systems, Including Containment, 11-14 November Results of Applying the CESA Method", in W.J. 2002, Pisa, Italy, CD-ROM Bonano et al. (Eds.), Proc. 6th International 1 2 3 SQUARER D. , OKA Y. , BITTERMANN D. , AKSAN N., Conference on Probabilistic Safety Assessment and 4 5 6 MARACZYC. , KYRKI-RAJAMAKI R. , SOURYA. , Management (PSAM 6), 23-28 June 2002, San Juan, 7 DUMAZ P. Puerto Rico, USA, Vol. I, 177-182 "High Performance Light Water Reactor (HPLWR)", RITTERS., SEIFERT H.P. FISA-2001-EU Research in Reactor Safety, 12-14 "Strain-Induced Corrosion Cracking of Low-Alloy RPV- November 2001, Luxembourg, EUR 20281, 620-630 FZK, Karlsruhe, Germany Steels under BWR Conditions", National Association 2 University of Tokyo, Japan of Corrosion Engineering (NACE) Corrosion 2002, 8- 3 12 April 2002, Denver, USA, CD-ROM Framatome ANP, Erlangen, Germany 4 KFKI, Budapest, Hungary RITTERS., SEIFERTH.P. 5 VTT, Espoo, Finland "Stress Corrosion Cracking of Low-Alloy Reactor 6 EdF, Paris, France Pressure Vessel Steels and of a Weld Filler Material 7 under BWR Conditions", 15th International Corrosion CEA-Cadarache, St. Paul-lez-Durance, France Congress ICC, 22-27 September 2002, Granada, Spain, CD-ROM 120 STREIT M., INGOLD F., GAUCKLER L.J.1, OTTAVIANI J-P.2 AKSAN N. "Annular Plutonium Zirconium Nitride Fuel Pellets", "Report on the Final Comparison Workshop of the Advanced Reactors with Innovative Fuels, OECD ISP-42 Exercise (PANDA Tests) ", OECD/NEA, 3rd Paris 2002, 22-24 October 2001, Chester, UK, CD- Meeting of the CSNI Group on the Analysis and ROM Management of Accidents (GAMA), Paris, France, 22- 1 ETH, Zurich, Switzerland 23 January 2002 2 CEA-Cadarache, St. Paul-lez-Durance, France AKSAN N. THOENENT., BERNERU., CURTI E., HUMMELW., "The ISP-42 Exercise (PANDA Tests) Final Blind and PEARSON F.J.1 th Open Phases Comparisons Reports: Executive "The Nagra/PSI Thermochemical Data Base", 12 Summary, Conclusions and Recommendations", Annual Goldschmidt Conference, 18-23 August 2002, th OECD/NEA, 4 Meeting of the CSNI Group on the Davos, Switzerland, Geochimica et Cosmochimica Analysis and Management of Accidents (GAMA), Acta, 66, A771 (2002) London, U.K., 24-25 September 2002 1 Ground-Water Geochemistry, New Berne, USA AKSAN N. THOMAS G.M.1, SOULE R.1, ASSAL W.1, CHAUSSONNET P.1, DESTOUCHES C.1, JAMMES C.1, "Developmental Needs for RELAP5 Code to Super LAURENS J.M.1, PLASCHY M. Critical Pressure Conditions", US.NRC Code "Steady-State Calculations in Support of the MUSE-4 Application and Maintenance Program (CAMP) Experimental Programme", Physor 2002, 7-10 Meeting, Alexandria, VI, USA, October 31- November October 2002, Seoul, Korea, CD-ROM 1,2002 1 CEA Cadarache, St. Paul-lez-Durance, France ALONSO U.1, DEGUELDRE C. TITS J., WIELAND E., DOBLER J.P., KUNZ D. "Modelling americium sorption onto colloids: effect of "The uptake of Sr(ll) by calcium silicate hydrates: redox potential", Poster Presentation at the Adsorption versus co-precipitation", 12th Annual symposium Colloid 2002: symposium C, EMRS, Strasbourg, France, 18-20 June 2002 Goldschmidt Conference, 18-23 August 2002, Davos, 1 Switzerland, Geochimica et Cosmochimica Acta, 66, CIEMAT, Madrid, Spain A776 (2002) BARTEN W., FERROUKHI H., CODDINGTON P. "OECD/NEA & USNRC BWR Turbine Trip VAN GEEMERT R., JATUFF F., CHAWLA R. Benchmark: PSI Final Results of Phase 1, and "A Decomposition Methodology for Quantitative Parameter Variations, Phase 2 with Flow and Interpretation of Reactivity Effect Discrepancies in Pressure Boundary Conditions, and Final Results LWR-PROTEUS", Physor 2002, 7-10 October 2002, Phase 3 Using RETRAN-3D", OECD/NRC Boiling Seoul, Korea, CD-ROM Water Reactor Turbine Trip Benchmark - Third WICHERSV.A.1, MALOJ.Y.2, STARFLINGER J.3, Workshop (BWR-TT3), FZR, Rossendorf, Germany, HEITSCH M.4, PREUSSERG.5, TUOMAINEN M.6, 28-30 May 2002 HUGGENBERGERM. BERTSCH J., HOFFELNERW. "Testing and Enhanced Modelling of Passive "J-R Curves Determination of Tube Cladding Evolution Systems Technology for Containment Material", 42nd NFIR Meeting, Baden, Switzerland, Cooling (TEMPEST)", FISA-2001-EU Research in 21-25 October 2002 Reactor Safety, 12-14 November 2001, Luxembourg, EUR 20281, 514-526 (2002) BORKOVEC M.1, GROLIMUND D. 1 NRG, Petten, The Netherlands "Colloid facilitated transport in natural porous media: 2 th CEA-Cadarache, St. Paul-lez-Durance, France Science or Fiction?", 224 American Chemical Society 3 FZK, Karlsruhe, Germany (ACS) National Meeting, Boston, USA, 18-22 August GRS, Garching, Germany 2002 1 University of Geneva, Switzerland 5 Framatome ANP, Offenbach, Germany 6 VTT, Espoo, Finland BORKOVEC M.1, GROLIMUND D. YADIGAROGLU G.1, ANDREANI M., DREIER J., "Colloids in Subsurface Environments", E-MRS 2002 CODDINGTON P. Spring Meeting, Strasbourg, France, 18-21 June, 2002 1 University of Geneva, Switzerland "Trends and Needs in Experimentation and Numerical Simulation for LWR Safety", FISA-2001-EU Research BRADBURY M.H, BAEYENS B. in Reactor Safety, 12-14 November 2001, "Test case 5: Np(V) sorption on montmorillonite", NEA Luxembourg, 559-574 Sorption Project, Technical Workshop, El Excorial, 1 ETH, Zurich, Switzerland Spain, 28-30 October 2002 BRUCHERTSEIFERH. Talks delivered at Conferences, Workshops and "lodspeziesanalyse in wassrigen Losungen mittels Specialist Meetings (without Proceedings) lonenchromatographie", 4. DIONEX Tagung ABOLHASSANI S., DADRAS M., LEBOEUF M. lonenchromatographie, Olten, Switzerland, 15 May "In-situ Observation of oxidation of Zircaloy-4", 15th 2002 International Corrosion Congress, Granada, Spain, 22-27 September 2002 121 CHAWLA R. DEGUELDRE C. "Nuclear Education in Switzerland", Frederic Joliot- "M|jSR in Monoclinic zirconia", M|jSR User Meeting, Otto Hahn Summer School Meeting, Paris, France, PSI Viligen, Switzerland, 16-17 January 2002 1 July 2002 DEGUELDRE C. DAHN R., SCHEIDEGGERA.M., MANCEAU A.1 "Read out Redox?", INE Seminar, FZK, Karlsruhe, "The use of polarized EXAFS to determine uptake Germany, 22 January 2002 processes of metal ions at the mineral-water 1 2 interface", Invited Talk, International Workshop on DEGUELDREC., ARIMAT. , LEEY.W. "Thermal conductivity of zirconia based inert matrix Mineral Surface and Colloid Chemistry in Soil and th Aquatic Environments, University of Karlsruhe, fuel: from a formal model to experimental data", 8 Germany, 8-9 October 2002 IMF Workshop, JAERI, Tokai, Japan, 16-18 October 1 LGIT, Grenoble, France 2002 1 University Kyushu, Fukuoka, Japan 2 DAHN R., SCHEIDEGGERA.M., MANCEAU A.1, KAERI, Tajeon, South Korea BAEYENS B., BRADBURY M.H. DEGUELDRE C., HELLWIG CH. "Uptake mechanisms of Ni onto montmorillonite: A th "Study of a zirconia-based inert matrix fuel under polarized EXAFS study", 39 Annual Meeting of the irradiation", EMRT II Meeting, Karlsruhe, Germany, Clay Minerals Society, Boulder, USA, 8-13 June 2002 25-27 September 2002 1 LGIT, Grenoble, France DOKHANE A. DANG V.N. "Stability and Bifurcation Analyses of Two-Phase Flow "Human Reliability Analysis (HRA) - Human Factors Using a Drift Flux Model", Seminar, University of in Probabilistic Safety Assessments", KSA Seminar Illinois at Urbana-Champaign, USA, 23 April 2002 uber Probabilistische Sicherheits-Analysen (PSA) Stufe 1, HSK Wurenlingen, Switzerland, 14-15 May DONES R. 2002 "Aspetti ecologici di vari sistemi di produzione di calore", "Giornata di Studio: Le pompe di calore", DANG V.N., GUBLER R.1, HIRSCHBERGS. Associazione delle Aziende Elettriche della Svizzera "Demonstration anhand der PSA KKM", KSA Seminar Italiana (ESI), Losone, Italy, 10 December 2002 uber Probabilistische Sicherheits-Analysen (PSA) Stufe 1, HSK Wurenlingen, Switzerland, 14-15 May GAWELDAW.1, SAESM.1, BRESSLERCH.1, ABELA R., 2002 GROLIMUND D., JOHNSON S.L.2, HEIMANN PH.3, 1 KSA consultant CHERGUI M.1 DANG V.N. "Picosecond time-resolved X-ray absorption "Quantifying Human Factors - The State of HRA and spectroscopy of aqueous organometallic complexes", Current Information Needs", invited presentation at SLS User Meeting, PSI Villigen, Switzerland, 25-26 the OECD Halden Workshop, "Review of Methods September 2002 1 University of Lausanne, Switzerland and R&D Needs to Address Issues Related to HRA", 2 University of California, Berkeley, USA Halden, Norway, 5-6 November 2002 3 Advanced Light Source, Berkeley, USA DEGUELDREC., LAUBEA. GIMMI TH. "Colloid generation in a marl aquifer", Poster "Transport von Pestiziden in Feldboden", Invited Talk, Presentation at 8th International Conference on the EAWAG, Dubendorf, Switzerland, 5 April 2002 Chemistry and Migration Behaviour of Actinides and 1 2 Fission Products in the Geosphere, Bregenz, Austria, GIMMI TH., URSINON. , FLUHLERH. 16-21 September 2001 "Analysing heterogeneous water flow in a field soil in terms of the dilution index", Poster Presentation at DEGUELDREC., FAVARGERP.-Y.1 Annual Meeting of the European Geophysical Society, "Colloid analysis by single particle Inductively Nice, France, 21-26 April 2002 Coupled Plasma - Mass Spectroscopy: a feasibility 1 University of Padua, Italy 2 study", Oral Presentation at the Symposium Colloid ETH, Zurich, Switzerland 2002: Symposium C, EMRS Strasbourg, France, 18- GIMMI TH., RUBEL A.1, WABER H.N.2 20 June 2002. "Fractionation of stable water isotopes in pore water of 1 University of Geneva, Switzerland an argillaceous rock", Poster Presentation at Annual DEGUELDRE C., LAUBE A., GECKEIS H.1, HAUSER W.1, Meeting of the European Geophysical Society, Nice, FIERZTH.2, MISSANAT.3 France, 21-26 April 2002 1 GRS, Braunschweig, Germany "Experimental results from the Grimsel colloid 2 migration", Poster Presentation at the Symposium University of Bern, Switzerland Colloid 2002: symposium C, EMRS Strasbourg, France, 18-20 June 2002 1 FZK, Karlsruhe, Germany 2 Solexperts AG, Schwerzenbach, Switzerland 3 CIEMAT, Madrid, Spain 122 GIMMITH., WABERH.N.1, RUBELA.2, GAUTSCHI A.3 HERMANN A. "Large scale-transport behaviour of a Jurassic "Effect of localized hydride accumulations on Zircaloy argillaceous rock as inferred from profiles of stable cladding mechanical properties — Completing Task 1 water isotopes and chloride", International Conference and starting new burst tests on irradiated fuel cladding on Clays in Natural and Engineered Barriers for specimens", 42nd NFIR Meeting, Baden, Switzerland, Radioactive Waste Confinement, Reims, France, 9-12 21-25 October 2002 [Handout NFIR-X104-02/H04 December 2002 (2002)] 1 University of Bern, Switzerland 2GRS, Braunschweig, Germany HIRSCHBERG S. 3 NAGRA, Wettingen, Switzerland "Air Pollution and Greenhouse Gas Damages: Issues 1 for Switzerland, China and the World", Seminar GROLIMUND D., BARMETTLER K. "Mobilization of colloidal particles: a key obstacle to in- presented at Yokosuka Research Laboratory, Central situ remediation of contaminated sub-surface systems", Research Institute of Electric Power Industry, E-MRS 2002 Spring Meeting, Strasbourg, France, 18- Yokosuka, Japan, 7 March 2002 21 June 2002 HIRSCHBERG S. 1 ETH, Zurich, Switzerland "Overview of Methodology for PSA Level 1", KSA GROLIMUND D. Seminar uber Probabilistische Sicherheits-Analysen "Colloid mobilization and transport", The Danish (PSA) Stufe 1, HSK Wurenlingen, Switzerland, 14-15 Institute of Agricultural Sciences, International May 2002 Workshop on "Colloids and Colloid-Facilitated HIRSCHBERG S. Transport of Contaminants in Soils", Foulum, Denmark, "PSA Development Status and Applications", KSA 19-20 September 2002 Seminar uber Probabilistische Sicherheits-Analysen GROLIMUND D., SCHEIDEGGER A. M., JANOUSCH M., (PSA) Stufe 1, HSK Wurenlingen, Switzerland, 14-15 DAEHN R., BONHOURE I., CHEESEMAN C.R.1 May 2002 "Microbial influenced degradation of solidified HIRSCHBERG S. radioactive waste materials: a microspectroscopic "Risk Assessment in Sustainable Development study", SLS User Meeting, PSI, Villigen, Switzerland, Projects", Keynote speech, 6th International 25-26 September 2002 Conference on Probabilistic Safety Assessment and 1 Imperial College of Science, London, UK Management (PSAM 6), San Juan, Puerto Rico, USA, GROLIMUND D. 23-28 June 2002 "Reaktiver Transport in Boden und HIRSCHBERG S. Grundwasserleitern: Die Bedeutung von mobilen "Approaches to Evaluating Sustainability of Energy Kolloiden", Institut fur Bodenkunde und Supply Options", 11th Solarpaces International Standortslehre, Universitat Hohenheim, Stuttgart, Symposium on Concentrated Solar Power and Germany, 16 December 2002 Chemical Energy Technologies, Zurich, Switzerland, HADERMANN J. 4-6 September 2002 "The safety case and chemistry — a subjective view HIRSCHBERG S. of an amateur", Workshop in honor of Prof. J.I. Kim "Swiss Methodology for Assessment of Technologies", "Nuclear chemistry-future", Invited Talk, Karlsruhe, INPRO Technical Meeting, IAEA, Vienna, Austria, 21- Germany, 4 July 2002 23 October 2002 HEIMGARTNERP., HELLWIG CH. HIRSCHBERG S. "Infrastructure at PSI for the fabrication of Pu-Pellets "Review of INPRO Methodology for Assessment of (MOX and IMF) and segments", Plenary Meeting of Innovative Nuclear Technologies", INPRO Technical the European Working Group "Hot Laboratories and Meeting, IAEA, Vienna, Austria, 21-23 October 2002 Remote Handling", SCK-CEN, Mol, Belgium, 25-27 September 2002, INGOLD G., ABELA R., BEAUD P., GROLIMUND D., MUNOZ M., SCHMIDT T., SCHLOTTV., SCHULZ L., STREUNA., HELLWIG CH., KASEMEYER U. VAN DER VEEN F., CHUBARO.1, KHAN S.2, TARNOVSKYA.3 "IFA-651.1: IMF performance during the first three "Undulator X-ray source for time-resolved ps and sub- cycles; comparison with modelling results", Enlarged ps pump/probe experiments", SLS User Meeting, Halden Programme Group Meeting on High Burn-Up Villigen PSI, Switzerland, 25-26 September 2002 Fuel Performance, Safety and Reliability, 8-13 1 CEA Saclay, Gif-sur-Yvette, France September 2002, Gol, Norway 2 BESSY, Berlin, Germany 3 University of Lausanne, Switzerland HERMANN A. "Neutron radiography of NFIR burst specimens and KALKHOF D., SEIFERT H.P. the effect of hydride lenses on local strain", 41st NFIR "Current Research Activities in the Field of Structural Meeting, San Francisco, California, 16-19 April 2002, Integrity", 25th Meeting of the OECD/NEA CSNI USA [Handout NFIR-X104-02/H03 (2002)] Working Group on Integrity of Components and Structures, NEA, Paris, France, 15 May 2002 123 KALKHOF D. LEEY.W.1, LEES.C.1, KIMH.S.1, JOUNGC.Y.1, "Status of Follow-up Actions following SESAR/FAP DEGUELDRE C. Recommendations", 7th Meeting of the OECD/NEA "Study on the mechanical properties and thermal CSNI Working Sub-Group on the Integrity of Metal conductivity of simulated Zr02 and Mg spinel-based Components and Structures, NEA, Paris, France, 13- inert matrix nuclear fuel materials after cyclic thermal 14 May 2002 shock", 8th IMF Workshop, JAERI, Tokai, Japan, 16- 18 October 2002 KALKHOF D. 1 KAERI, Tajeon, South Korea "Materials Ageing Data available for Structural Integrity Assessment", 7th Meeting of the OECD/EA MACIAN, R. CSNI Working Sub-Group of the Integrity of Metal "Best Estimate Analysis Methodology in the STARS Components and Structures, NEA, Paris, France, 13- Project at PSI", Meeting of Experts on Best Estimate 14 May 2002 Calculations and Uncertainty Analysis, Aix-en- Provence, France, 13-14 May 2002 KALKHOF D. "Moderne Diagnostik fur Alterungsphanomene", MACIAN R. Industrieseminar Oberflachentechnologie, PSI "Uncertainty in System Codes. Current Status and Villigen, Switzerland, 28 October 2002 Future Work in STARS", Meeting of Experts on Best Estimate Calculations and Uncertainty Analysis, Aix- KASEMEYER U., LEBENHAFT J. en-Provence, France, 13-14 May 2002 "Preliminary Validation of Pin Power Prediction in Partial LWR Core Loadings with Inert Matrix and U02 MACIAN R. Fuel", CMS European Users Group Meeting, PSI "Comparison of Isotopic Compositions in Highly Burnt Villigen, Switzerland, 30 September - 2 October 2002 MOX and U02 Fuel Calculated with CASMO-4 against Experimental Measurements from the KASEMEYER U., LEBENHAFT J.R., HELLWIG C., ARIANE Program", CMS European Users Group CHAWLA R. Meeting, Villigen PSI, Switzerland, 30 September - 2 "Comparison of Various Partial LWR Core Loadings October 2002 with Inert Matrix and MOX Fuel", Enlarged Halden Program Group Meeting, Gol, Norway, September 8- MORI A.1, ALEXANDER W.R.2, GECKEIS H.3, GEYER F.3, 13; and 8th Meeting of the Inert Matrix Fuel EIKENBERGJ., FIERZTH.4, DEGUELDRE C., MISSANAT.5 Workshop, JAERI Tokai, Ibraraki, Japan, 16-18 "The Colloid and Radionuclide Retardation October 2002 Experiment at the Grimsel Test Site: Influence of bentonite colloids on the radionuclide migration in a KOPAJTIC Z. granitic fracture", Symposium Colloid 2002: "Aerosol characterisation of the PHEBUS FPT4/PTA symposium C, EMRS Strasbourg, France, 17-21 June samples", PHEBUS PTA Meeting, Cadarache, 2002 France, 1 February 2002 1 Geotechnical Institute Ltd, Bern, Switzerland 2 NAGRA, Wettingen, Switzerland KOPAJTIC Z. 3 FZK, Karlsruhe, Germany "PSI investigations in the frame of the PHEBUS 4 Solexperts AG, Schwerzenbach, Switzerland 5 FPT2/PTA and FPT4/PTA programme", Oral CIEMAT, Madrid, Spain Presentation at the PHEBUS PTA Meeting, NECHVATAL L., ZIMMERMANN M.A. Cadarache, France, 11 October 2002 "Thermal-Hydraulic Scoping Analysis of the Planned KROGERW. Halden LOCA-Experiment IFA-650 Using TRAC-BF1", "Wiederaufarbereitung - Ressourcenschonung und Special Meeting on LOCA-Experiment IFA-650, Gol, Umweltbelastung", SVA Sessionsveranstaltung, Bern, Norway, 10 September 2002 Switzerland, 21 March 2002 NEGREANU C., STEPANEK J., JONEJA O.P., CHAWLA R. KROGERW. "Validation of New Electron and Positron Data "Kernenergie - Auslaufmodell Oder Hoffnungstrager?", Libraries", IRRMA-V, 5th International Topical Meeting Ringvorlesung PHS, St. Gallen, Switzerland, 24 April on Industrial Radiation and Radioisotope 2002 Measurement, Bologna, Italy, 9-14 June 2002 LEBENHAFT J.R., KASEMEYER U., CHAWLA R. NIFFENEGGERM. "Preliminary 3-D MCNP Calculations for the CROCUS "Solitonen und Mikromagnetismus", Seminar, Institut Kinetics Benchmark", 13th Meeting of the Nuclear fur Theoretische Physik, University of Fribourg, Science Committee Working Party on the Physics of Switzerland, 18 June 2002 Plutonium Fuels and Innovative Fuel Cycles, PELLONI S. OECD/NEA, Paris, France, 26 July 2002 "Data Sensitivity of Design Calculations for High Conversion D20 Reactors", JEFF Meeting, Aix-en- Provence, France, 25 April 2002 124 PELLONI S. SCHEIDEGGER A. M., GROLIMUND D., JANOUSCH M., "Detailed Sensitivity Study for a Pb-Bi Cooled DAEHN R., BONHOURE I., CHEESEMAN C.R.1, CUI D.2 Accelerator Driven System with a Pb-Bi Target", "Degradation and corrosion of solidified waste th Invited Talk, Technical Meeting, ENEA, Bologna, Italy, materials: a microscopic study", 7 International 24 September 2002 Conference on X-ray Microscopy (XRM2002), Grenoble, France, 29 July-2 August 2002 PFINGSTEN W. 11mperial College of Science, London, UK "Modelling cement-rock-water interactions — the 2Studsvik Nuclear AB, Nykoping, Sweden influence of boundary conditions", Int. Workshop on SCHEIDEGGER, A.M., GROLIMUND, D., TrePro2002, FZK, Karlsruhe, Germany, 20-21 March 1 2002 CHEESEMAN, C. R. "Microbial influenced degradation of solidified PFINGSTEN W. radioactive waste: a microscopic study", 7th "Modelling of dipole tracer tests", 8. Hyperalkaline International Conference on X-ray Microscopy, Plume in Fractured Rocks (HPF) Project Meeting, Grenoble, France, 31 July 2002 Tokai, Japan, 23-25 April 2002 imperial College of Science, London, UK PFINGSTEN W. SCHEIDEGGER, A.M. "Radionuclide migration from underground nuclear "Molecular-level information on reactions at the solid- tests performed at the Lira facility, Kazakhstan — liquid interface as determined by X-ray absorption estimations by source term and transport modelling", spectroscopy and microscopic techniques", Invited ModelCARE2002, Prague, Czech Republic, 16-20 Talk at the symposium "Long-term storage of nuclear June 2002 waste", CEA, Saclay, France, 15 March 2002 PFINGSTEN W. SMITH B. L. "Modelling of dipole tracer tests using additional "CFD Applications to Containment Flows", constraints coming up by ongoing measurements", 9. OECD/NEA Exploratory Meeting of Experts on the HPF-Project Meeting, Bern, Switzerland, 14-15 Application of Computational Fluid Dynamics to November 2002 Nuclear Safety Problems, Aix-en-Provence, France, 15-16 May 2002, CD-ROM POUCHON M. A., NAKAMURA M.1, HELLWIG CH., INGOLD F., DEGUELDRE C. STOENESCU R. "Cer-Met Sphere-Pac concept for Inert Matrix Fuel", "Mechanical behaviour of the heat-affected zone of IMF8 Workshop, Tokai-Mura, Japan, 16-18 Oktober austenitic stainless steel welds", Junior Euromat 2002 2002 Conference, 2-5 September 2002, Lausanne, 1 JNC, Tokai, Japan Switzerland 1 2 RONCHI C. , OTTAVIANI J.P. , DEGUELDRE C., TITS J., WIELAND E., DOBLER J.P. CALABRESE R.3 "Uptake of radionuclides by calcium silicate hydrates: "Thermophysical properties of inert matrix fuels", 2nd an experimental approach to distinguish adsorption Seminar on European Research on Materials for from co-precipitation processes", Poster Presentation, Transmutation. Karlsruhe, Germany, 26-27 14th Radiochemical Conference, Marianske Lazne, September 2002 Czech Republic, 14-19 April 2002 1 JRC, Karlsruhe, Germany 2 CEA-Cadarache, St. Paul-lez-Durance, France VAN LOON L.R., MULLER W., GIMMI TH., IIJIMA K. 3 ENEA, Bologna, Italy "Activation energy of the self-diffusion of water in SAESM.1, GAWELDAW.1, BRESSLERCH.1, CHERGUI M.1, compacted clay systems: a case study with Opalinus ABELA R., GROLIMUND D., JOHNSON S.L.2, HERTLEIN M.2, clay", International Conference on Clays in Natural HEIMANN PH.2 and Engineered Barriers for Radioactive Waste "Picosecond time-resolved X-ray absorption Confinement, Reims, France, 9-12 December 2002 spectroscopy of laser-excited organometallics in VAN LOON, L.R., MULLER W., BRADBURY M.H. solution", ALS User Meeting, Berkeley, CA, USA, 10-12 "Anisotropic diffusion in argillaceous rocks: a case October 2002 study with Opalinus clay", International Workshop: 1 University of Lausanne, Switzerland Clay Microstructure and its Importance to Soil 2 Advanced Light Source, Berkeley, USA Behaviour, Lund, Sweden, 15-17 October 2002 SCHAFFNERB., HOFFELNERW., MEIER A., WALLIN H. WUILLEMIN D., STEINFELD A. "Solar Thermal Recycling of Electric Arc Furnace "Using FALCON and the NFIR Cladding Burst Tests Dust", 11th Solarpaces International Symposium on to Validate Modifications to the MATPRO Zr-4 Concentrated Solar Power, Zurich, Switzerland, 4-6 Mechanical Properties Models", NFIR-Meeting, September 2002 Baden, Switzerland, 23-25 October 2002 WERNLI B. "Contribution of PSI to the MALIBU programme", Oral Presentation at the MALIBU Introductory Meeting, Brussels, Belgium, 25 April 2002 125 WlELAND E., BONHOURE I., TLTS J., SCHEIDEGGERA.M., KROGER W., WILLIAMS T.1 BRADBURY M.H. "Gen IV Nuclear Energy Systems Initiative - Gen IV "Macroscopic and spectroscopic investigations on the International Forum (GIF) , Stand der Bemuhungen immobilization of radionuclides by hardened cement und mogliche Beteiligungen der Schweiz als paste", Invited Talk, 14th Radiochemical Conference, Vollmitglied, NES Colloquium, PSI Villigen, Marianske Lazne, Czech Republic, 14-19 April 2002 Switzerland, 8 March 2002 1 EGL, Zurich, Switzerland NES and ENE Colloquia Miscellaneous Reports ABOLHASSANI S. "Investigation of Zircaloy Oxidation by TEM and other ADDABBO C.1, ANNUNZIATO A.1, AKSAN N., D'AURIA F.2, Techniques", NES Colloquium, PSI Villigen, DUMONT D.3, GALASSI G.2, HOFMAN D.4, NILSSON L.4, Switzerland, 16 May 2002 RIGAMONTI M.5, PURHONEN H6, RIIKONEN V6, STEINHOFF F.7, TOTH I.8, UMMINGER K.9 BRUCHERTSEIFER H., CRIPPS R. "Requirements for Storage and Retrieval of "Radiolytische Zersetzung von Silberiodid— Experimental Data for the Assessment of LWR Safety Experimente & Modellierung", NES Colloquium, PSI Codes", EC-JRC Report (EUR 20413), June 2002 Villigen, Switzerland, 31 January 2002 1 JRC, Ispra, Italy 2 University of Pisa, Italy CODDINGTON P. 3 CEA, Grenoble, France "Application of RETRAN 3D to Operational Transients 4 of Swiss NPPs", NES Colloquium, PSI Villigen, Studsvik Ecosafe, Nykoping, Sweden Switzerland, 14 February 2002 SIET, Piacenza, Italy VTT, Espoo, Finland CODDINGTON P., PELLONI S. GRS, Garching, Germany "Physics and Safety of Nuclear Waste Transmutation", g KFKI, Budapest, Hungary NES Colloquium, PSI Villigen, Switzerland, 24 Framatome ANP, Erlangen, Germany October 2002 ADDABBO C.1, ANNUNZIATO A.1, AKSAN N., D'AURIA F.2, DAHN R. GALASSI G.2, DUMONT D.3, NILSSON L.4, RIGAMONTI M.5, "Sorption mechanisms of radionuclides on clay RIIKONEN V6, STEINHOFF F.7, TOTH I.8, GAUL H. P.9, mineral surfaces as determined by X-ray absorption UMMINGER K.9 spectroscopy" NES Colloquium, PSI Villigen, "Maintenance of European LWR Integral System Test Switzerland, 14 March 2002 Thermal-Hydraulic Databases", EC-JRC Report (EUR DEHBI A. 19937), October 2001 1 JRC, Ispra, Italy "ARTIST: An International Project Investigating 2 University of Pisa, Italy Aerosol Retention in a Ruptured Steam Generator", 3 CEA, Grenoble, France NES Colloquium, PSI, Villigen, Switzerland, 19 4 September 2002 Studsvik Ecosafe, Nykoping, Sweden 5 SIET, Piacenza, Italy HAAG B., WILLIAMS T.1 6 VTT, Espoo, Finland "F&E Bedurfnisse der Kernenergie in der Schweiz", 7 GRS, Garching, Germany NES Colloquium, PSI Villigen, Switzerland, 20 June 8 g KFKI, Budapest, Hungary 2002 Framatome ANP, Erlangen, Germany 1 EGL, Zurich, Switzerland 1 1 2 ANNUNZIATO A. , ADDABBO C. , AKSAN N., D'AURIA F. , 2 3 4 4 HlRSCHBERG S. GALASSI G. , DUMONT D. , NILSSON L. , HOFMAN D. , RIGAMONTI M.5, RIIKONEN V.6, STEINHOFF F.7, GUBAA.8, "Greenhouse Gas Emission and Air Pollution Damage 8 9 9 Scenarios: Highlights from Recent PSI Analyses for TOTH I. , GAUL H.P. , UMMINGER K. Switzerland, China and the World, ENE Colloquium, "Development and Establishment of the CERTA PSI Villigen, Switzerland, 18 April 2002 Network Database", EC-JRC Report (EUR 20421), HOFFELNER W., HELLWIG CH., VAN LOON L., DAI Y., September 2002 1 1 JRC, Ispra, Italy BALUCN. 2 University of Pisa, Italy Hot Labor User Colloquium, NES Colloquium, PSI 3 Villigen, Switzerland, 29 August 2002 CEA, Grenoble, France 1 4 EPFL, Lausanne, Switzerland Studsvik Ecosafe, Nykoping, Sweden 5 KALKHOF D., GROSSE M., GROTH E. SIET, Piacenza, Italy "Einfluss der PbBi-Umgebung auf die 6 VTT, Espoo, Finland Ermudungsfestigkeit der Stahle 316L SS und 10.5Cr- GRS, Garching, Germany Manet-N", NES Colloquium, PSI Villigen, Switzerland, g KFKI, Budapest, Hungary 14 November 2002 Framatome ANP, Erlangen, Germany 126 BIDINGER G.H.1, CACCIAPOUTI R.J.2, CONDE J.M.3, GROLIMUND D., BARMETTLER K.1, BORKOVEC M.2 COUSINOU P.4, DOERING T.5, GRIMM P., HWANG H-R.6, "Natural colloidal particles in the subsurface LEE W.J.7, MARLOYE D.8, MURRAY R.L.9, environment: mobilization, transport and persistence", NEUBER J.-C.10, BRADY RAAP M.11, SAPYTA J.12, Danish Institute of Agricultural Sciences DIAS Report, THOMAS D.A.13, TURNER S.14, WILSON R.E.15, Plant Production No. 80, 59-67 (2002) 1 BOYACK B.16, DIAMOND D.J.17 ETH, Zurich, Switzerland 2 "Burnup Credit PIRT Report", NUREG/CR-6764, BNL- University of Geneva, Switzerland NUREG-52654, May 2002 HADERMANN J. 1 Private Consultant, Rockville, USA "Mit Sicherheit entsorgt", ENET-NEWS, Nr. 53, 36-37, 2 Duke Engineering & Services, USA December 2002 3 CSN, Madrid, Spain 4IPSN, Fontenay-aux-Roses, France HERMANN A., KÜSTER D., WIESE H., LEHMANN E., 5EPRI, Las Vegas, USA SCHWEICKERTH. 6 KOPEC, Taejon, South Korea "Effect of localized hydride accumulations on Zircaloy 7 NAC International, Norcross, USA cladding mechanical properties — Investigation of 8 Belgonucleaire, Brussels, Belgium 9 NFIR burst samples by neutron radiography", Report North Carolina State University, Raleigh, USA EPRI/NFIR-IV X104-02-04, EPRI, Palo Alto, USA 10 Siemens AG, Offenbach, Germany (2002) 11 PNNL, Richland, USA 12 Framatome ANP, Lynchburg, USA HIRSCHBERG S. 13 Framatome ANP, USA 14 Korreferat zur Studie "Marginale Zahlungsbereitschaft Holtec International, Marlton, USA für eine erhöhte Internalisierung des Risikos von 15 DOE, Washington, USA 16 Kernkraftwerken" von Prof. P. Zweifel & Yves LANL, Los Alamos, USA 17 BNL, Upton, USA Schneider, BFE, May 2002 HIRSCHBERG S., HECKT. BIOLLAZS., BÜCHI F., DIETRICH PH., GASSMANN F., KÖTZR., LIENINS., RÖDERA., SCHERER G. "Schweiz und China als Beispiel", in Bulletin, Magazin "Energie-Spiegel", "Verkehr", Facts für die der ETHZ, Nr. 287, 12-15 November 2002 Energiepolitik von Morgen, Newsletter des Projektes HOSEMANN P. GaBE, Ganzheitliche Betrachtung von 1 "Vorsorge gegen das Unwahrscheinliche", EN ET Energiesystemen, PSI /ETHZ, Nr. 6, March 2002 News, 30-31 April 2002 1 ENE/NES HUMMEL W., BERNER U. 1 DÄHN R., SCHEIDEGGERA.M., MANCEAU A. , "Application of the Nagra/PSI TDB 01/01: Solubility of BAEYENS B., BRADBURY M.H. Th, U, Np and Pu", Nagra NTB 02-12 (2002) "Local structure of Th complexes on montmorillonite clay mineral determined by extended X-ray absorption JATUFF F. fine structure (EXAFS) spectroscopy", Speciation, "Realitätsnahe Forschung im PROTEUS", EN ET Techniques and Facilities for Radioactive Materials at News, 34-35, July 2002 Synchrotron Light Sources, NEA Pubn., 75-81 (2002) METZV.1, PFINGSTEN W., LÜTZENKIRCHEN J.2, 1LGIT, Grenoble, France SCHUESSLERW.1 DANGV. N., REER, B. "Proceedings of the TRePro2002 Workshop" on "Review of the Human Reliability Analysis in the "Modelling of Coupled Transport Reaction Processes", Mühleberg Shutdown PSA," HSK 11/785, prepared for FZK, Wissenschaftliche Berichte 6721, 20-21 March HSK, October 2002 2002, FZK, Karlsruhe, Germany, p.119 (2002) 1 FZK, Karlsruhe, Germany 1 2 3 EHRNSTÉN U. , ERNESTOVAM. , FOEHL J. , 2 JRC, Karlsruhe, Germany GOMEZ-BRICENO D.4, HÜTTNER F.3, LAPENAJ.4, ROTH A.5, RITTER S., SEIFERT H.-P., ZAMBOCH M.2 POUCHON M. A. "Crack Growth of Pressure Vessel Steel in Simulated "Java Meshing Tool for Sphere Arrangements — BWR Environment, a Summary Report on Inter- Manual", JNC TN8520 2002-001, open JNC Report, Laboratory Investigations", Technical Report WP1, 2002 CASTOC R003, 4 June 2002 POUCHON M. A. 1 VTT, Espoo, Finland "CrossNeckl.O - Manual", JNC TN8520 2002-002, 2 NRI, Rêz, Czech Republic 3 MPA Stuttgart, Germany open JNC Report, 2002 4 CIEMAT, Madrid, Spain 5 Framatome ANP, Erlangen, Germany REERB., DANG V.N. "Review of the Human Reliability Analysis in the FOSKOLOS K., KRÖGER W., HIRSCHBERG S. Mühleberg Full Power PSA (MUSA)," HSK 10/780, "Energie-Spiegel", "Nuklearenergie: prepared for HSK, October 2002 Wiederaufarbeitung", Facts für die Energiepolitik von Morgen, Newsletter des Projektes GaBE, Ganzheitliche Betrachtung von Energiesystemen, PSI1/ETHZ, Nr. 7, July 2002 1 ENE/NES