INTRODUCTION
Transcrição
INTRODUCTION
FLOW AND SOLIDIFICATION OF CORIUM IN THE VULCANO FACILITY Christophe JOURNEAU, Eric BOCCACCIO, Claude JEGOU, Pascal PILUSO, Gérard COGNET Commissariat à l’Energie Atomique, CEA / Cadarache, Dept. of Thermalhydraulics and Physics, F13108 St Paul lez Durance, France ABSTRACT In the hypothetical case of a nuclear reactor severe accident, the reactor core could melt and form a mixture, called corium, of highly refractory oxides (UO2, ZrO2), steel in a metallic or oxidised form and of the substrate decomposition products, which could eventually flow out of the vessel. The VULCANO facility is dedicated to experiments with prototypic corium (including depleted UO2). This paper is devoted to corium spreading experiments, in which the effects of flow and solidification are studied. Due to the complex behavior of corium in the solidification range, an interdisciplinary approach has been used relating thermodynamics of multicomponent mixtures, rheological models of silicic semisolid materials, high temperature heat transfer, free-surface flow with temperaturedependant properties. This is coupled with the use of specific high temperature instrumentation requiring in-situ cross calibration. Eleven high temperature spreading tests have been performed and analyzed. The main experimental results are the good spreadability of corium, the effects of cooling rates on the microstructure, the occurrence of a large thermal contact resistance at the corium-substrate interface, the presence of a rheological boundary layer at the surface. Furthermore, the experiments showed the role of the gaseous inclusions in the melt. These tests with prototypic material have been used to validate spreading models and codes. INTRODUCTION Nuclear fission energy provides about 80 % of the electricity supply in France and 33 % in the European Union. It has several important strategic benefits: independence from fossil fuels, long-term security of primary energy supply and zero emission of greenhouse-effect gases. However, despite these advantages, there is an opposition to nuclear energy from sectors of the public in the European countries. Consequently, this implies the improvement of knowledge with the view to simultaneously increasing safety and maintaining competitiveness. Since the Three Mile Island accident in 1979 [1] and especially since the Chernobyl accident in 1986 [2], it is clear that one of the key points to increase safety is that any reasonably credible accident must be controlled within the reactor containment with no off-site consequences. In this context, for some years now, the CEA has undertaken a large program [3] on severe accidents aimed at their mastering in both existing and future power plants. For the European Pressurized water Reactor (EPR) project, core melt accident is one of the severe accidents to be mastered under the new safety approach for the reactor design. Therefore, a dedicated area of about 175 m² [4] has been devoted to the spreading of the molten mixture called corium - essentially composed of uranium and zirconium dioxides and of more or less oxidized steel - that would result from the hypothetical melting down of the core, melting through the reactor vessel and mixing with a sacrificial concrete in the reactor pit. The role of spreading is to reduce the surface thermal load due to the radioactive decay heat. To study corium spreading, a series of experiments have been performed in Europe. The CORINE facility [5] has been built for low temperature analytic experiments. The Scaled Simulant Spreading Experiments (S3C) uses simulants at temperatures up to 1400 K [6]. In the KATS facility [7] alumina above 2000 K is used as a simulant. Simulant spreading experiments have also been performed for volcanological applications [8]. Spreading experiments using prototypic corium, i.e. melt containing depleted uranium dioxide with compositions prototypic of a hypothetic severe accident scenario, have been performed in the CARLA [9], FARO [10] and VULCANO [11] facilities. This paper presents the VULCANO facility and the main results from the spreading tests that have been conducted there. They have been obtained through an interdisciplinary approach linking heat transfer, fluid mechanics, physico-chemistry, materials science and thermodynamics. Rather than focusing on specific experiments or techniques, this paper synthesizes the various data and results obtained in this test series. THE VULCANO FACILITY The VULCANO facility [11] is mainly composed of a furnace [12] and a test section, which is thoroughly instrumented. Posttest analyses, which are a necessary part of prototypic corium tests, will also be described. An interdisciplinary experimental team operates this facility. The furnace To realize prototypic material experiments, with acceptable safety conditions, the furnace must be able to melt oxidic mixtures of various compositions (UO2, ZrO2, SiO2, FeOx ) with the possible addition of metals at temperatures between 2000 and 3200 K. It shall have the capacity to melt and to pour 100 kilograms continuously and at low pouring rates (0.1 to 1 L/s). A study of the candidate technologies leads to the choice of a transferred plasma-arc furnace. rise of surface thermocouples. The geometrical calibration (see Figure 2) is obtained by positioning a checkerboard on the spreading section and correlating the pixel position with the vertices co-ordinates. Figure 2: Checkerboard used to calibrate the “fish-eye” camera before test VE-U1. Figure 1 : The VULCANO furnace. Two plasma torches are ignited by an electrical short circuit. The main arc is then created and transferred between these two torches having opposite polarity (see Figure 1). The electrode tips are made of graphite. The plasmagenic gas is of argon and/or nitrogen plus, in some cases, corium fumes. The maximum available power is around 600 kW ( 1000 A – 600 V). Powders to melt are inserted in the cylindrical rotating cavity (400 mm diameter – 500 mm long). A self-crucible of zirconia is realized in the furnace by the centrifugation (between 150 and 300 rpm) and heating of the powder. The furnace external surfaces are water-cooled while the inner surface of the load is subject to the radiation from the plasma arc (temperatures between 10 000 and 20 000 K). Radiation heats the surface, then, convection and mainly conduction transfer the heat to the inner layers. The corium powders are then loaded and molten. Two successive phases of loading and melting are used in order to increase the furnace effective capacity by compensating for the powder low density. The heating process is controlled by optical pyrometry and inboard thermocouples. When a sufficient quantity of corium has been molten, the arc power is reduced and the cathode is withdrawn. The furnace is then tilted so that the melt pours out in the test section. The plasma arc is maintained during the pouring operation, in order to prevent a too large cooling. Test sections The VULCANO-E spreading test section consists of a spreading plane. Refractory bricks (zirconia or magnesia) or a steel plate have been used as substrate. Two geometries have yet been used: spreading in an open square or in a 19° angular sector materialized by refractory magnesia bricks. Instrumentation The test section is mounted on a weighing scale in order to measure the pouring flow rate. Temperature measurements are performed by thermocouples (K, N, S and C – tungsten-rhenium – types), bichromatic pyrometers and infrared thermography [13]. Spreading progression and front velocity is measured with geometrically calibrated cameras [14] and compared with the The infrared camera is in-situ cross-calibrated [13] with a bichromatic pyrometer, the aiming point of which is positioned on the thermographic images. The least square fitted correlation is then used to convert the thermographic images in temperature maps. This calibration takes into account simultaneously the intrinsic calibration of the device and the corium surface emissivity in the considered infrared bandwidth . Post-mortem analyses After the test, the cooled corium is examined. Its shape is measured and samples are taken for chemical and material postmortem analyses (see e.g. [15]). Chemical analyses are made on crushed samples using a Philips 1404 X-Ray Fluorescence (XRF) analyzer. X-Ray Diffraction (XRD) analyses are performed with a Siemens D5000 diffractometer. Resin-wrapped samples are observed using optical microscopes (Olympus PME3 and Polyvar MET) and a Stereoscan Cambridge S360 Scanning Electron Microscope (SEM). Local composition is estimated with an Oxford ISIS 300 Energy Dispersive Spectrometry (EDS) microanalysis. For the spreading experiments, the element composition is relatively constant over the spread volume, but there are important variations against height in terms of compositions and microstructures. The chemical analyses are compared with thermodynamic computations using the GEMINI2 software and the TDBCR databases[17]. The main findings of these analyses will be reported in a later section. THE SPREADING TESTS Simulant tests on 2D sections Two tests have been performed with zirconia-alumina melts that were spread on a 2D square test section. During test VR-18, which was the first spreading test at the VULCANO facility on May 23, 1996, 6.5 kg of 55%wt zirconia, 45%wt alumina have spread over a 450x525 mm spreading section. Figure 3 shows the dissymmetric final shape of the spread. Simulant tests on 19° angular sector Five tests have been performed with simulant materials on a 19° angular sector, in order to model, with limited masses axisymmetric spreading. The melts simulated corium-concrete mixtures by replacing mole-wise uranium by hafnium. The main characteristics of these tests are reported in the following table. Simulant test with metal/oxide melt Figure 3: Postmortem view of the VR-18 spread (unbounded 2D spreading) During the next test, VR-19, 23 kg of a zirconia-31%wt alumina mixture totally filled the 0.24 m² spreading area. This concluded the technological tests of the VULCANO facility. An experiment (test VE-06) has been performed in which metallic iron has been added to the oxidic melt. The spread composition was around 5% Fe, 52%w HfO2, 14%w ZrO2, 27%w Fe2SiO4, 2%w CaO, simulating the mixture of 71% of a mixed metal-oxide corium with 29% of ferrosiliceous concrete. 34 kg of melt were spread over a final length of 45 cm. The molten iron remained in droplets which formed an emulsion with the oxidic liquid and did not sediment although the temperature remained above the iron melting point for 3 minutes and these oxides were less dense than iron. Table 1: Physical and chemical characteristics of simulant spreading tests on θ=19° angular sector. Uncertainties are around ± 70 K for the temperatures and around ± 20% for pouring rates. Test Composition (in wt %) VE-01 50%HfO2, 10%ZrO2, 34%Al6Si2O13, 6% Al2CaSi2O8 58% HfO2, 10%ZrO2, 31%Al6Si2O13, 1%CaO 40% HfO2, 5%ZrO2, 39%Fe2SiO4, 16%Fe3O4 61% HfO2, 11%ZrO2, 21% Fe2SiO4, 4% Al2O3, 3% CaO 33% HfO2, 22%ZrO2, 22%SiO2, 22%FeO, 1% CaO VE-02 VE-03 VE-04 VE-07 Mass (kg) 12 21 15 12 17 Flow rate (L/s) 0.1 discontinuous 0.1 discontinuous 0.1 continuous 0.7 continuous 0.5 continuous Pouring Temp. (K) 2070 Main results Small spreading: 22 cm Thickness 40 mm Accumulation Thickness 40-100 mm Spreading length 30 cm Thickness at front 20 mm Accumulation No spreading Spreading length 55 cm Compact homogeneous structure 2270 2070 2270 2420 VE-U1 Corium Spreading VE-U1 160 140 Spread Length (cm) The first spreading experiment with UO2, VE-U1, was performed in the VULCANO facility on December 2, 1997. The corium composition (in mass percentages: 44% UO2, 23% ZrO2, 21% SiO2, 12% FeOx) was prototypic of corium discharge from the reactor pit after ablation of sacrificial ferro-siliceous concrete. The solidification range of this mixture is of more than 900 K with a liquidus temperature of 2250 K. 38.8 kg of corium, at an initial temperature of 2090±100 K, spread over 1.2 meters. Figure 5 presents a view of the spread, taken when the front was at 0.9 m from inlet. The flow length evolution (Figure 4) has been obtained from the analysis of video images and is consistent with the time of passage recorded on surface thermocouples. 120 100 80 area-averaged length Actual final length THEMA (low flow rate) THEMA (high flow rate) 60 40 20 0 0 5 10 15 20 25 30 35 40 45 Time (s) Figure 4: Progression of corium length as measured from the video images, compared to the values calculated by THEMA spreading code (see below) The measurement has been interrupted after 15 s when part of the spread was out of the video image field. VE-U5 Figure 5: Front view of the flowingVE-U1 spread (taken with the camera calibrated in Figure 2). VE-U3 The VE-U3 test used a mixture with a smaller concrete content, having the following composition, in mass percentage: 63 % UO2, 22% ZrO2, 8% SiO2, 7% FeO, i.e. being richer in UO2 than VE-U1 corium and thus 150 K more refractory. 15.6 kg of corium, poured at an initial temperature of 2400 ± 60 K with a flow rate of 0.3 L/s, have spread over 33 cm with an average height of 32 mm. In this test, two successive spreading phases were realised with a corium of the following initial composition: 46% UO2, 10 %w ZrO2, 24%w Fe O, 20% SiO2. A first part of the corium flow was poured on the spreading section while a second part was maintained temporarily in a crucible, where an exothermic reaction with metallic zirconium provided chemical heating. As the crucible door melted, a second flow occurred which mixed with the already stopped first spread (visible in bright on Figure 6). In total, 36 kg were spread at a mass flow rate of 2 kg/s with an initial temperature around 2100 K. This test showed the ability of corium to progress when a second load is poured even after the initial progression had stopped. Table 2: Physical and chemical characteristics of prototypic corium spreading tests. Uncertainties are around ± 70 K for the temperatures and around ± 20% for pouring rates. Test VE-U1 VE-U3 VE-U5 Composition (in wt %) Mass (kg) 38.8 44% UO2, 23% ZrO2, 21% SiO2, 12% FeOx 63 % UO2, 22% ZrO2, 8% SiO2, 15.6 7% FeO 46% UO2, 10 %w ZrO2, 24%w FeO, 36 20% SiO2 + zirconium metal Flow rate (L/s) 0.6 Pouring Temp. (K) 2090 0.3 2400 0.5 2100 Main results Spreading over 1.2 m Several tongues. Large voids Spreading over 33 cm Small porosity Successive spreads Analysis in progress… 1/ 8 g ⋅G 3 1/ 2 U = t 3⋅ν The characteristic time for solidification is estimated as the time necessary to remove sufficient heat as to make the corium practically immoveable (the immovability enthalpy is approximated as the average of the enthalpies at the liquidus and solidus temperature) considering heat losses by radiation (at the inlet temperature) and convection ( with Nu=0.0023 Pe). The final average height order of magnitude δs is given, for 1D spreading by the following expression: Figure 6: VE-U5 corium spread over a steel plate (The molten metallic gate of the crucible is visible as a bright spot in the upstream of the flow). MAJOR RESULTS These tests have shown that spreading of corium-concrete mixtures is quite efficient, even at low flow rates and with composition containing up to 22%wt of silica (i.e. viscous liquids). They also enabled us to validate and improve our model of the rapid cooling processes taking place during the spreading transient. Validation of spreading models and codes One of the major objectives of this series of tests was to validate spreading models and codes for severe accident management. Dinh et al. [6] have proposed a simplified model based on a scaling approach. They assume that corium flows as an isothermal viscid hydrodynamic fluid for the time needed to cool its front down to an immobilization temperature. A characteristic spreading velocity can be estimated with Huppert’s model [17] for gravity-viscous spreading : 1/ 2 V 3/ 4[qrad + qconv −qvol ] δ s =Maxδ cap, 1/16 1/ 2 7 /16 g G ρ m ⋅∆H sold 3⋅ν This approximate method gives for the above-mentioned test, the order of magnitude final corium height within a ratio of 2. Eberlé [18] has developed the THEMA 1D software that integrates the conservation of mass, energy and momentum over verticals of the flow. This code has been satisfactorily validated against the VULCANO spreading experiments. For example, Figure 4 presents the spreading-length evolution computed for VE-U1 considering two hypothesis for the evolution of inlet flow (which is indirectly known since the mass in the inlet is weighed with that on the spreading section). Currently, the main obstacle to a precise validation is due to the large uncertainties still present on the corium physical properties. These uncertainties can lead to final spreading lengths between 0.6 and 2 m for VE-U1 – for an experimental average value of 1.13 m. Therefore, a validation has been made with low temperature simulants, the properties of which are well known [5]. More refined codes modelling the 2D behaviour of the flow have also been developed. The 2-dimensional CROCO [19] and CORFLOW [20] codes have been also satisfactorily validated against the VE-U1 spreading experiment. Sensitivity analysis have been performed and showed that the spreading length, for VE-U1, was mainly controlled by the flow rate, especially during the beginning of the spreading, the viscosity law, the temperature difference between inlet and liquidus temperatures and the corium thermal conductivity. The substratum temperatures during the time of the flow are mostly correlated to the thermal contact interface and the substrate conductivity. It must be stressed that there are still large uncertainties on material data for these mixtures and on temperature measurements ( ± 70 K). Microstructure and thermodynamic modeling The chemical analyses of the spread melts indicate that there is no sign of heterogeneity within the flow, which has a somehow constant element composition. Furthermore, solidification does not occur in the form of a dendritic mush growing from the cold region, but rather as a suspension of solids in the remaining liquid. This configuration is thus similar to that of semisolid alloys [21]. Notwithstanding the chemical element uniform repartition, the microstructure and the phases in presence vary greatly depending on the vertical position. These effects [15] have been explained by the variations in cooling rate, and for very thin layers (< 100 µm), the oxidation by air. In the spread lowermost layer (less than 1 mm thick in our cases), which, as it will be developed after, was thermally insulated by a large thermal contact resistance, the solidification was close to thermodynamic equilibrium. A tracer of equilibrium below 1900 K is the formation of silicates of hafnium, zirconium, and/or uranium, including, at least in VE-U1, a compound close to chernobylite [2], the zirconium-uranium silicate observed in Chernobyl “lavas”. In the bulk of the spread, solidification is quicker (>10 K.s-1) and diffusion in the solid phase had not enough time to occur. In some cases (e.g. VE-07), the upper interface is characterized by dendritic particles. The post-mortem analyses of these tests made with mixtures of a large number of constituents ( at least 5 major chemical elements) have also contributed to the validation of the thermodynamic databases and have lead to the determination of thresholds between the different types of solidification. Physical properties As shown by the spreading-code sensitivity-analysis, physical properties can greatly affect the spreading of corium. But, there are a lot of uncertainties in estimating the properties of a mixture of a multicomponent melt and solid particles at high temperatures. For the estimation of specific mass, we assume, following Nelson & Carmichael [23], that the partial molar volume of all the liquid phase constituent have no compositional dependence. The evolution of the corium rheology during solidification can be estimated using the methodology of Ramacciotti et al. [24]. Since cooling is mainly due to radiation, there is a moving cooling front inside the height of the corium. As solidification and physical properties are linked to temperature, there is not only a moving thermal boundary layer, but also a compositional boundary layer and a rheological boundary layer [22]. Viscosity varying very sharply with solid fraction, there is an exponential increase in viscosity in the rheological boundary layer, with e.g. a characteristic length of 0.3-0.5 mm for VE-U1 test. Nevertheless, the surface remains above the solidus temperature during the spreading phase and thus, there is no solid crust on the surface, but rather a viscous skin. This rheological boundary layer has the visible effect, in these flows, of folding the surface (see e.g. Figure 7) as volcanic lavas of the ropy pahoehoe type [25]. This effect has been observed on more than 10 tests having a wide range of compositions. On the opposite, during spreads of UO2-ZrO2 corium, with a solidification range lower than 100 K, that were performed at the FARO facility [10], solid crusts has been observed floating on the melt. In both cases, there is a strong coupling between heat transfer, thermodynamics and fluid dynamics which affects the flow of corium flow. Corium-Substrate Interface During these tests, the substrate has only been locally attacked by corium (mainly near the inlet of the spreading plane). Post test analyses show mainly a penetration of corium in the brick pores with little chemical reaction. The outcome would be different in longer term experiments for which residual heat is simulated. Indeed in this type of experiment, dissolution of the crucible has been observed at VULCANO. The heat flux at the corium-substrate interface can be estimated using an inverse conduction technique [26] from temperature measured at various depths inside the substrate. For instance, during the VE-U1 flow, the downward heat flux entering the zirconia bricks was estimated around 200 kW/m², whereas the radiated flux at the free surface was around 900 kW/m². These values must be compared to the heat flux that is expected in case of perfect contact between two large conducting bodies: b ⋅ T + bsub ⋅ Tsub b − Tsub q = sub ⋅ cor cor bcor + bsub πt In the case of VE-U1, the flux would have been of 500 kW/m² after 5 s of contact. This means that a large thermal contact resistance must have been present. It is estimated around 36 mK.m²/W and is attributed to the shrinkage at solidification of the lower crust [27]. Gases One of the main differences observed in these tests between simulant and prototypic material is the presence of large voids in the VE-U1 and VE-U5 spreads. It must be noted that for simulant test VE-07, very close in composition and operating conditions to test VE-U1 (actually it was the rehearsal of VE-U1 with hafnia instead of zirconia), only a small porosity was observed. These voids cannot be only attributed to air flowing from the substrate pores, since it also occurred when the substrate was made of steel (VE-U5) so that gases cannot come from the thermal expansion of the air trapped in the substrate pores. Figure 7 shows a view of VE-U5 spreads. Part of the crust have been destroyed by a gas pressure below it. Under the portion of crust that remained intact, grains mainly made of U3O8 and SiO2 have been observed. The inner side of the crust had a structure reminding convection cells. qconv qrad qv t T U V δs δcap ∆H ν ρm Figure 7: View of the VE-U5 corium spreads showing holes at the surface. Several evidences are consistent with an hypothesis of oxygen exchange processes during the experiment leading to the local formation of volatile UO3 and SiO. Further analyses are underway to investigate this hypothesis and assess its consequence on corium physical properties and behavior. An associated difference between prototypic and simulant melts is that many of the simulant melt displayed uncracked surfaces whereas in our prototypic material tests, the corium surface was always cracked. This features would be favorable for corium cooling by top-flooding. CONCLUSIONS Eleven spreading tests have been performed at the VULCANO facility, including three with prototypic corium containing up to 63%wt of UO2 which are representative of the products of a corium concrete interaction in the reactor pit. They have shown with real materials the promises of spreading concepts for corium management in the hypothetical case of a nuclear reactor severe accident. Spreading to thin corium thickness has been obtained even with mixtures containing more than a fifth of silica, which increases substantially the liquid viscosity (but simultaneously decreases the liquidus temperature). Spreading codes and models have been validated to the extent permitted by the current large uncertainties on corium physical properties in the solidification range. Improving our physical properties database is thus one major R&D need. Important differences have been observed between prototypic and simulant materials especially for the physico-chemistry with respect to oxygen exchange processes. For the complete validation of the spreading concepts, further experiments are needed to study the effects of spreading on a concrete substrate – which liberates gases and can be thermochemically ablated -, spreading under water and to analyze the long terms effects, for which the slowly diminishing internal heat generation must be simulated. ACKNOWLEDGEMENTS The work and efforts of the whole VULCANO team are greatly acknowledged. Some of the described experiments were part of the Corium Spreading and Coolability project of the European Commission’s 4th Work Program (Contract FI4S-CT96-0041). NOMENCLATURE b effusivity J m-2s-1/2 K-1/2 g acceleration of gravity m/s² G volume flow rate m3/s Nu Nusselt number dimensionless Pe Péclet number dimensionless convected heat flow radiated heat flow surfacic power (volumic power x height) time (discharge time or contact time) temperature velocity total spread volume spread height capillary limit enthalpy to remove for immobilization kinetic viscosity melt specific mass W/m² W/m² W/m² s K m/s m3 m m J/kg m²/s kg/m3 REFERENCES 1. J.M. Broughton, P. Kuan, D.A. Petti, E.L. Tolman, A scenario of the Three Mile Island unit 2 accident, Nucl. Technol., vol. 87, pp. 147, 1989. 2. E.M. Pazukhin, Fuel-Containing Lavas of Chernobyl NPP 4th Block, Radiochemistry vol. 36, pp. 109-154, 1994. 3. G. Cognet, J.-M. Seiler, I. Szabo, J.-C., Latché, B. Spindler, J.-M. Humbert, La récupération du corium hors cuve, Rev. Gén. Nucl., vol. 1, pp. 38-43, 1997. 4. H.A. Weisshäupl Severe accident mitigation concept of EPR, Nucl. Eng. Design, vol. 187,pp. 35-45, 1999. 5. B., Spindler and J.M. Veteau, Status of the assessment of the spreading code THEMA against the CORINE experiments, Proc. SARJ Mtg. (Mtg. Severe Accident Research held in Japan) Tokyo, Japan, 4-6 Nov 1998. 6. T.N. Dinh, M. Konovalikhin D. Paladino J.A. Green A., Gubaidulin, B.R. Sehgal, Experimental Simulation of Core Melt Spreading on a LWR Containment Floor in a Severe Accident, Proc. ICONE6 (Int Conf. Nucl. Eng.), San Diego, Ca. , 1998. 7. G. Fieg, F. Huber, H. Werle, R. Wittmaack, Simulation Experiments on the spreading behaviour of molten core melts, Proc. Nat. Heat Transfer Conf., Houston, Tx., 1996. 8. J.H. Fink and R. W. Griffiths, Radial spreading of viscousgravity currents with solidifying crust, J Fluid Mech., vol.221, pp. 485-509, 1990. 9. M. Sappok and W. Steinwarz, Large scale experiments on ex-vessel core melt behavior, Nucl. Technol. , vol. 125, pp. 363-370, 1999. 10. G. Cognet, H. Alsmeyer, S. Morris, R. Wittmaack, B.R. Sehgal, M. Bürger, L. De Cecco, R. Ocelli, G. Azarian, W. Ranval, B. Spindler, H. Werle, Corium Spreading and Coolability, Proc. FISA –97 EU Research on Severe Accidents, Office Off. Pub. Eur. Comm., Luxembourg, 1998. 11. G. Cognet, G.Laffont, C. Jégou, J. Pierre, C. Journeau, F.Sudreau, A. Roubaud, Utilisation d’un four tournant à arc plasma transféré pour fondre et couler des mélanges d’oxydes autour de 2000°C., Ann. Pharm. Fr., vol. 57, pp. 31 – 136, 1999. 12. C. Jégou, , G. Cognet, A. Roubaud, J.-M.Gatt, G. Laffont, F. Kassabji, Plasma transferred arc rotary furnace for corium melting, J. High Temp Mater. Proc. vol. 1, pp. 409420, 1998. 13. G. Cognet, C. Journeau, C. Jégou, Infrared Thermography for Measuring the Surface Temperature of an Oxidic Melt, Int. Symp. on Radiative Heat Transfer, Kusadasi, Turkey, 1995. 14. C. Journeau, Y. Jung, J Pierre,. Visualization of a 2000°C melt spreading over a plane, Proc. 8th Int Symp Flow Visualization, Sorrento, Italy, 1998. 15. C. Journeau, F. Sudreau, S. Magne, G. Cognet, Physicochemical analyses and solidification path reconstruction of multi-component oxidic spread melts Mater. Sci. Eng. vol. 299A, pp. 249-266, 2001. 16. P.Y. Chevalier, E. Fischer, B. Cheynet, A.,Rivet, G. Cénérino, Global thermodynamic approach of the molten core-concrete interaction (MCCI) and selected applications in the nuclear field, J. Chim. Phys. PCB, vol. 94, pp. 849860, 1997. 17. H.E. Huppert, The propagation of 2 dimensional and axisymmetric viscous-gravity currents over a rigid horizontal surface, J. Fluid Mech, vol. 121, pp. 43-58, 1982. 18. P. Eberlé, Modélisation physique et numérique de l’étalement d’un fluide avec solidification dans le cadre des etudes de sûreté pour les réacteurs à eau sous pression, Ph.D. Thesis, Université J. Fourier, Grenoble, France, 1997. 19. B. Piar, B.D. Michel, F. Babik, J.-C. Latché, G. Guillard, J.-M. Ruggierri, CROCO: a Computer Code for Corium Spreading, Proc. NURETH 9 (Nucl. Reactors Thermal Hydraulics), San Francisco, Ca., 1999. 20. R. Wittmaack, Numerical simulation of free surface flows, Nucl. Technol., vol. 119, pp. 158-180, 1997. 21. M.C. Flemmings, Behavior of metal alloys in the semisolid state, Metall. Trans., vol. 22A, pp. 957-981, 1991. 22. C. Journeau, F. Sudreau, J.-M. Gatt, G. Cognet Thermal, physico-chemical and rheological boundary layers in multicomponent oxidic melt spreads, Int J. Therm. Sci., vol. 38, pp. 879-891, 1999. 23. S.A. Nelson and I.S.E. Carmichael, Partial molar volumes of oxide components in silicate melts, Contrib. Mineral. Petrol., vol. 71, pp. 117-124, 1979. 24. M. Ramacciotti, F. Sudreau, C. Journeau, G. Cognet, Viscosity models for Corium Melts, Nucl. Eng. Design. vol. 204, pp. 377-389, 2001. 25. R.A.F. Cas and J.V. Wright, Volcanic Successions, Modern and Ancient, Chapman & Hall, London, p. 67,1988. 26. M. Raynaud and J. Bransier, A new finite-difference method for the nonlinear inverse heat conduction problem, Num. Heat Transfer, vol. 9, pp. 27-42, 1986. 27. T. Loulou, E.A. Artyukhin, J.P. Bardon, Estimation of thermal contact resistance during the first stages of metal solidification, Int. J. Heat Mass Transfer, vol. 42, pp. 21192142, 1999.