51 organizations network in SARNET (S__-evere A__-ccident R__-esearch ___-work of Excellence) their capacities of research in order to resolve the most important remaining uncertainties for enhancing, in regard of Severe Accidents (SA), the safety of existing and future Nuclear Power Plants (NPPs). This project, co-funded by the European Commission (EC), has been defined in order to optimise the use of the available means and to constitute sustainable research groups in the European Union. SARNET tackles the fragmentation that exists between the different R&D national programmes, in defining common research programmes and developing common computer tools and methodologies for safety assessment. SARNET comprises most of the actors involved in SA research in Europe (plus Canada). To reach these objectives, all the organizations networked in SARNET contribute to a so-called Joint Programme of Activities (JPA), which consists in: ・Implementing an advanced communication tool for accessing all project information, fostering exchange of information, and managing documents; ・Harmonizing and re-orienting the research programmes; ・Jointly analysing the experimental results provided by research programmes in order to elaborate a common understanding of relevant phenomena; ・Developing the ASTEC code (integral computer code used to predict the NPP behaviour during a postulated SA), which capitalizes in terms of physical models the knowledge produced within SARNET; ・Developing Scientific Databases, in which all the results of research programmes are stored in a common format (DATANET); ・Developing a common methodology for Probabilistic Safety Assessment (PSA) of NPPs; ・Developing courses and writing a text book on SA for students and researchers; ・Promoting personnel mobility between various European organizations. After the first period (2004-2008), co-funded by the EC, the network will progressively evolve toward self-sustainability. The bases for such an evolution, still under discussion, are presented in the last part of the paper.

Quarante-neuf organismes européens rassemblent dans le réseau SARNET (Severe Accident Research and management NETwork) leurs moyens de recherche sur les accidents graves de réacteur, afin de réduire les problèmes importants en suspens dans ce domaine, et ainsi contribuer à l'amélioration de la sûreté des centrales nucléaires actuelles et futures. Le projet a été défini en prenant en compte la nécessité d'optimiser l'utilisation des moyens disponibles en Europe, et d'associer de manière pérenne les complémentarités des laboratoires de recherche. SARNET s'attaque à la fragmentation existant entre les différents programmes nationaux de R&D, notamment en élaborant en commun des programmes de recherche ainsi que des outils de calcul et des méthodologies pour les évaluations de sûreté. Coordonné par l'IRSN le réseau SARNET rassemble la plupart des acteurs impliqués dans la recherche sur les accidents graves en Europe.Pour atteindre ces objectifs, toutes les membres du réseau SARNET contribuent à un programme d'activités commun (JPA), constitué de plusieurs éléments :• Mise en oeuvre d'un outil de communication avancée pour favoriser l'échange d'informations ;• Harmonisation et réorientation de programmes de recherche et définition commune de nouveaux programmes ;• Analyse des résultats expérimentaux fournis par les programmes de recherche afin d'aboutir à une compréhension commune des phénomènes concernés ;• Développement du logiciel ASTEC (outil de calcul permettant de simuler le comportement d'un réacteur nucléaire lors d'un accident grave) qui capitalise en termes de modèles physiques les connaissances élaborées par SARNET ;• Développement de bases de données scientifiques dans lesquelles tous les résultats des programmes de recherche sont stockés ;• Développement d'une méthodologie commune pour les évaluations probabilistes de sûreté des réacteurs ;• Enseignement, formation et rédaction d'ouvrages ;• Mise en oeuvre d'un programme de mobilité de personnels entre les divers organismes membres du réseau.Le réseau a atteint en terme de compétence, dans le domaine des accidents graves et pour l'ensemble des réacteurs nucléaires en Europe, la masse critique nécessaire à la réalisation et l'interprétation de programmes expérimentaux ainsi qu'au développement de modèles et à leur intégration dans le logiciel ASTEC. Quelques organismes couvrent un large éventail de compétences, qui sont compétées par l'apport de contributions spécialisées dans des domaines très spécifiques.Afin de préserver les intérêts des différents membres du réseau, une politique claire en termes de protection de la propriété intellectuelle a été définie. Les documents contenant des données "protégées" ne sont remis qu'aux membres qui, de par l'activité qu'ils proposent, augmentent de manière importante la valeur des données (production d'analyses, développement de modèles et évaluation). Néanmoins, les méthodologies d'évaluation de sûreté et le logiciel ASTEC, produits intégrateurs des connaissances et modèles élaborés au sein de SARNET, seront disponibles pour tous les membres du réseau et les autres organismes européens désireux de les utiliser pour des évaluations de sûreté ou pour l'amélioration de leurs centrales nucléaires.

The treatment of zirconium oxidation kinetics in severe accident (SA) codes has been the subject of many discussions and controversies in recent years. The main problem was the existence of several correlations which could lead to large differences in the calculated results. It appeared clearly that there was a need to converge towards a common understanding of the physical processes that must be modeled (oxygen diffusion, blanketing effect, etc.) and an agreed database among code developers and users. It would help reducing an important source of uncertainties in SA calculations.The kinetic correlation database, obtained as a result of examination of complementary experimental data in Parts I and II, is applied here to analyze a few high-temperature separate-effects tests and bundle experiments where Zry oxidation reaction played a dominant role. The ICARE/CATHARE computer code developed by IRSN is used to check the validity of the high-temperature correlations derived in Parts I and II. The physical modeling provided by the code includes detailed account of specific features of chemical interactions between fuel rod cladding and steam. In particular, high reaction rates at T > 2000 K are moderated by two effects, examined in Part I: steam blanketing during thin oxide layers growth and transition to oxidation of alpha-Zr(O) phase after total consumption of primary beta-Zr in cladding metallic part.When applied to separate-effects tests, the evaluated parabolic correlations have shown their applicability to different types of temperature transients taking into account Zry oxidation specifics in rod geometry. The bundle integral experiments QUENCH-06 and PREBUS B9+ did not lead to extremely large temperature excursions. Calculated temperatures, hydrogen production and oxide thickness, as well a's parameters of melt relocation were found to agree well with experimentally measured values. As a result of this study, we believe that the new best-fitted correlations, obtained in agreement with available experimental data, can be used in further studies and can improve predictive power of the codes. The continuation of the current work will be the application of ICARE/CATHARE code with best-fitted Zry oxidation correlations to NPP accident scenarios. (C) 2004 Elsevier B.V. All rights reserved.

Safety research for nuclear power plants has been performed for more than 25 years at the French "Institut de Radioprotection et de Surete Nucleaire" (IRSN). A wide base of knowledge has been developed through various national and international research programmes characterized by a number of co-operations with national and international actors. This knowledge is the main basis of the technical support provided by IRSN to the French Safety Authority. Presently, there is till a need to fill remaining gaps in knowledge and it is important to preserve research capabilities and infrastructures to support future research programmes for operating and future reactors. This paper gives an overview of the main IRSN R&D efforts in the fields of design basis accidents, severe accidents and fire risk in nuclear installations. International co-operation is also illustrated by contributions to the EURATOM and OECD programmes. Medium term plan of IRSN on safety research conclude the paper.

The core loss during a severe accident (COLOSS) project is a 3-year shared-cost action which started in February 2000. The project is concerned with the consequences that core degradation, occurring under severe accident conditions, may have on H(2) production, melt generation and the source term. Unresolved in-vessel risk-relevant issues are studied, through a large number of experiments such as (a) UO(2) and MOX dissolution by molten zircaloy and burn-up effects, (b) simultaneous dissolution of UO(2) and ZrO(2) in rod geometry, (c) oxidation of U-O-Zr mixtures, (d) oxidation of pure B(4)C material and (e) degradation and oxidation of B(4)C control rods. A parallel effort is devoted to model developments for severe accident (SA) computer codes. These codes are finally used for plant calculations to assess SA code capabilities and to apply results produced in this project to evaluate their consequences on key SA sequences occurring in different plants such as PWR-1300, BWR, VVER-1000, EPR and in the TMI-2 accident. Following significant results have been produced at the mid-term of the project:Several B(4)C oxidation experiments have improved the understanding of oxidation mechanisms. Preliminary models have been developed and implemented in SA codes.Separate-effects tests (SET) on simultaneous UO(2) and ZrO(2) dissolution and on U-O-Zr oxidation by steam enabled progress to be made on the understanding and modelling of these interactions. There is experimental evidence that the oxidation of mixtures can contribute significantly to the large H(2) production occurring during the reflood of a reactor core under severe accident conditions.Two large-scale tests CODEX-B(4)C and QUENCH-07 have been carried out with a central B(4)C control rod. The B(4)C effects on VVER and PWR core degradation and on the related gas production have been evaluated.Twelve plant calculations of key SA sequences illustrate the current status of SA codes to predict core degradation, in particular B(4)C effects. (C) 2002 Elsevier Science B.V. All rights reserved.

The COBE project started in February 1996 and finished at the end of January 1999. The main objective was to improve understanding of core degradation behaviour during severe accidents through the development of computer codes, the carrying out of experiments and the assessment of the computer codes’ ability to reproduce experimental behaviour. A major effort was devoted to quenching behaviour and a substantial achievement of the project was the design and commissioning of a new facility for the simulation of quenching of intact fuel rods. Two tests, carefully scaled to represent realistic reactor conditions, were carried out in this facility and the hydrogen generated during the quenching process was measured using two independent measuring systems. The codes were able to reproduce the results in the first test, where little hydrogen was generated but not the second test, where the extra steam produced during quenching caused an invigorated Zircaloy oxidation and a substantial hydrogen generation. A number of smaller parametric experiments allowed detailed models to be developed for the absorption of hydrogen and the cracking of cladding during quenching. COBE also investigated other areas concerned with late-phase phenomena. There was no experimental activity – the work included code development and the analysis of experimental data available to the project partners – either from open literature or from other projects such as Phebus-FP. Substantial improvement was made in the codes’ ability to simulate heat transfer in debris beds and molten pools and increased understanding was reached of control rod material interactions, the swelling of irradiated fuel and the movement of molten material to the lower head.

The objective of the current Validation Matrix is to define a basic set of experiments, for which comparison of the measured and calculated parameters forms a basis for establishing the accuracy of test predictions, covering the full range of in-vessel core degradation phenomena expected in light water reactor severe accident transients. The scope of the review covers PWR and BWR designs of Western origin: the coverage of phenomena extends from the initial heat-up through to the introduction of melt into the lower plenum. Concerning fission product behaviour, the effect of core degradation on fission product release is considered. The report provides brief overviews of the main LWR severe accident sequences and of the dominant phenomena involved. The experimental database is summarised. These data are cross-referenced against a condensed set of the phenomena and test condition headings presented earlier, judging the results against a set of selection criteria and identifying key tests of particular value. The main conclusions and recommendations are listed. (K.A.).

One can theoretically represent the diffusion of radon or any other atmospheric pollutant in the lower troposphere by the differential diffusion equation of the K theory. We have solved this equation numerically. In this model it is possible to choose any form of the diffusivity profile and of the vertical wind profile. The boundary conditions and the initial conditions can be modified according to the problem under consideration (geometry and intensity of the source, deposition at the earth's surface in the case of aerosols). In particular, we have used this model to study the diffusion of radon from a plane infinite source, a half-plane source, and a line source. We have measured the concentrations of radon in the troposphere. The results have been interpreted by a meteorological analysis; for certain data, the numerical model has been utilized.