Afin de limiter les conséquences d’un hypothétique accident grave avec la fusion du coeur dans un réacteur à neutrons rapides de génération IV refroidi au sodium, la recriticité doit être évitée au sein du mélange de combustible oxyde et de structures fondus, appelé corium. Pour cela, des matériaux absorbants, tels que le carbure de bore B4C, seront utilisés dans ou près du coeur, et des matériaux diluants dans le récupérateur de corium. L’objectif de ce travail est de présélectionner des matériaux parmi ces deux types de familles et de comprendre leur comportement au contact avec le corium. Concernant le B4C, des calculs thermodynamiques et des expériences ont permis de conclure à la formation de deux phases immiscibles dans le système UO2 – B4C à haute température, une oxyde et une borure, ainsi qu’à la volatilisation d’une partie de l’élément absorbant bore. Cette séparation de phases pourra réduire l’efficacité de l’absorption neutronique au sein de la phase oxyde. Une solution à ce comportement serait d’augmenter la quantité de B4C ou d’utiliser un absorbant oxyde miscible avec le combustible. Eu2O3 ou HfO2 pourraient convenir car il a été montré qu’ils forment une solution solide avec UO2. Concernant le matériau diluant, les oxydes mixtes Al2O3 – HfO2 et Al2O3 – Eu2O3 ont été étudiés. L’interaction de ces systèmes avec UO2 étant inconnue à ce jour, les premiers points ont été recherchés sur les diagrammes ternaires correspondants. Contrairement au système Al2O3 – Eu2O3 – UO2, le mélange Al2O3 – HfO2 – UO2 présente un seul eutectique et donc un seul chemin de solidification ce qui permet de prévoir plus facilement la manière dont le corium solidifierait dans le récupérateur. / In order to limit the consequences of a hypothetical core meltdown accident in Generation IV Sodium Fast Reactors, absorber materials in or near the core, such as boron carbide B4C, and diluting materials in thecore catcher will be used to prevent recriticality within the mixture of molten oxide fuel and molten structures called corium. The aim of the PhD thesis was to select materials of both types and to understand their behaviour during their interaction with corium, from chemical and thermodynamic point of view. Concerning B4C, thermodynamic calculations and experiments agree with the formation of two immiscible phases at high temperature in the B4C – UO2 system: one oxide and one boride. This separation of phases can reduce the efficiency of the neutrons absorption inside the molten fuel contained in the oxide phase. Moreover, a volatilization of a part of the boron element can occur. According to these results, the necessary quantity of B4C to be introduced should be reconsidered for postulated severe accident sequence. Other solution could be the use of Eu2O3 or HfO2 as absorber material. These oxides form a solid solution with the oxide fuel. Concerning the diluting materials, mixed oxides Al2O3 – HfO2 and Al2O3 – Eu2O3 were preselected. These systems being completely unknown to date at high temperature in association with UO2, first points on the corresponding ternary phase diagrams were researched. Contrary to Al2O3 – Eu2O3 – UO2 system, the Al2O3 – HfO2 – UO2 mixture presents only one eutectic and thus only one solidification path which makes easier forecasting the behaviour of corium in the core catcher.

This thesis deals with the severe accident of the gas-cooled fast reactor GFR. At the beginning of the study there is a review of the gas-cooled fast reactor subject. Next part is focused on description of possible solutions for severe accidents with emphasis on the solution applied in the Generation III+ reactors. Chapters that deal with material and thermal balance with severe accident of GFR demonstration unit, along with the chapter which analyses features of the corium, create a basis for the conceptual design of core catcher of GFR demonstration unit, which forms the final part of this thesis.

There are numerous experiments for the exploration of the corium spreading behaviour, but comparable data have not been available up to now in the field of the long term behaviour of a corium expanded in a core catcher. The difficulty consists in the experimental simulation of the decay heat that can be neglected for the short-run course of events like relocation and spreading, which must, however, be considered during investigation of the long time behaviour. Therefore the German GRS, defined together with Battelle Ingenieurtechnik a benchmark problem in order to determine particular problems and differences of CFD codes simulating an expanded corium and from this, requirements for a reasonable measurement of experiments, that will be performed later. First the finite-volume-codes Comet 1.023, CFX 4.2 and CFX-TASCflow were used. To be able to make comparisons to a finite-element-code, now calculations are performed at the Institute of Safety Research at the Forschungszentrum Rossendorf with the code ANSYS/FLOTRAN.For the benchmark calculations of stage 1 a pure and liquid melt with internal heat sources was assumed uniformly distributed over the area of the planned core catcher of a EPR plant. Using the Standard-k-e-turbulence model and assuming an initial state of a motionless superheated melt several large convection rolls will establish within the melt pool. The temperatures at the surface do not sink to a solidification level due to the enhanced convection heat transfer. The temperature gradients at the surface are relatively flat while there are steep gradients at the ground where the no slip condition is applied. But even at the ground no solidification temperatures are observed. Although the problem in the ANSYS-calculations is handled two-dimensional and not three-dimensional like in the finite-volume-codes, there are no fundamental deviations to the results of the other codes.

There are numerous experiments for the exploration of the corium spreading behaviour, but comparable data have not been available up to now in the field of the long term behaviour of a corium expanded in a core catcher. The difficulty consists in the experimental simulation of the decay heat that can be neglected for the short-run course of events like relocation and spreading, which must, however, be considered during investigation of the long time behaviour. Therefore the German GRS, defined together with Battelle Ingenieurtechnik a benchmark problem in order to determine particular problems and differences of CFD codes simulating an expanded corium and from this, requirements for a reasonable measurement of experiments, that will be performed later. First the finite-volume-codes Comet 1.023, CFX 4.2 and CFX-TASCflow were used. To be able to make comparisons to a finite-element-code, now calculations are performed at the Institute of Safety Research at the Forschungszentrum Rossendorf with the code ANSYS/FLOTRAN.For the benchmark calculations of stage 1 a pure and liquid melt with internal heat sources was assumed uniformly distributed over the area of the planned core catcher of a EPR plant. Using the Standard-k-e-turbulence model and assuming an initial state of a motionless superheated melt several large convection rolls will establish within the melt pool. The temperatures at the surface do not sink to a solidification level due to the enhanced convection heat transfer. The temperature gradients at the surface are relatively flat while there are steep gradients at the ground where the no slip condition is applied. But even at the ground no solidification temperatures are observed. Although the problem in the ANSYS-calculations is handled two-dimensional and not three-dimensional like in the finite-volume-codes, there are no fundamental deviations to the results of the other codes.

My thesis deals with the design of an experimental emergency cooling device of the reactor pressure vessel (RPV). It consists of two parts, the theoretical one and practical one. Different molten corium cooling methods in terms of their efficiency and comparison are introduced in the theoretical part. The design of an experimental emergency cooling device, which incorporates a model channel past the reactor pressure vessel , is presented in the practical part. The cooling device consists of a model channel past the reactor pressure vessel, condensator, which takes away the heat generated by the reactor pressure vessel and the pump of a secondary loop. Next, thermal and hydraulic calculations are given in this section. The conclusion is devoted to the evaluation of particular cooling technologies and their comparison in terms of nuclear and technical safety.