Ex-Vessel Core Melt Stabilization Research (ECOSTAR)

Transcription

1 Ex-Vessel Core Melt Stabilization Research (ECOSTAR) H. Alsmeyer 1, W. Häfner 2, C. Journeau 3, M. Fischer 4, M. Eddi 5, H-J. Allelein 6, M. Bürger 7, B.R Sehgal 8, M.K Koch 9, Z Alkan 10, J.B Petrov 11, M Gaune-Escard 12, F.P Weiss 13, G Bandini 14 1) Forschungszentrum Karlsruhe (DE) 8) KTH, Nuclear Power Saftey, Stockholm (SE) 2) Becker Technologies, Eschborn (DE) 9) LEE, Ruhr-University Bochum (DE) 3) CEA DEN/DTP, Cadarache (FR) 10) LRST, RWTH, Aachen (DE) 4) Framatome ANP, Erlangen (DE) 11) NRI, Rež (CR) 5) EDF, Chatou (FR) 12) Université de Provence, Marseille (FR) 6) GRS, Köln (DE) 13) Forschungszentrum Rossendorf (DE) 7) IKE, University of Stuttgart (DE) 14) ENEA, Bologna (IT) SUMMARY The Project on Ex-Vessel Core Melt Stabilization Research (ECOSTAR) is oriented to the analysis and mitigation of severe accident sequences that could occur in the ex-vessel phase of a postulated core melt accident. The issues are: (1) the release of melt from the pressure vessel, (2) the transfer and spreading of the melt on the basement, (3) the analysis of the physical-chemical processes that are important for corium behavior especially during concrete erosion with onset of solidification, and (4) stabilization of the melt by cooling through direct water contact. The project involves a large number of experiments with low and high temperature simulant melts and real corium at different scales. Model development and scaling analysis allows application of the research results to reactor design and accident mitigation. The results achieved so far, resolve a number of important issues: The amount of melt that could be transferred at RPV failure under reduced system pressure into the containment is quantified by scaled experiments. It is found that melt dispersion is also strongly depending on the location of the RPV failure, and that lateral failure results in substantially less melt dispersion. During melt release, the impinging melt jet could erode parts of the upper basement surface. Jet experiments and a derived heat transfer relation allow estimation of its contribution to concrete erosion. Spreading of the corium melt on the available basement surface is an important process, which defines the initial conditions for concrete attack and for the efficiency of cooling in case of water contact. Validation of the spreading codes based on a large-scale benchmark experiment, is underway and will allow determination of those initial conditions, for which a corium melt can be assumed to spread homogeneously. Experiments with UO 2 -based corium melts highlight the role of phase segregation during onset of melt solidification and during concrete erosion. To cool the spread corium melt, the efficacy of top flooding and bottom flooding is investigated in small scale and in large-scale experiments, supported by model developments. Project assessment is continuing to apply the results to present and future reactors. A. INTRODUCTION The ECOSTAR Project, as indicated by its title Ex-Vessel COre Melt STAbilization Research, investigates severe accident sequences that play a key role during the ex-vessel phase of a postulated core-melt accident. In addition to the improved understanding of these processes, investigations are directed towards stabilization and controlling the corium melt in order to terminate the accident and to minimize its consequences. Terminating the accident in the ex-vessel phase requires solution of the coolability issue, which means the extraction of the internal heat and of the nuclear decay heat from the corium

2 melt in order to stop basement attack and to exclude containment failure by downward penetration or over-pressurization. This also requires knowledge about the location of the melt, its geometrical configuration, and the important physical-chemical processes during corium solidification. The ECOSTAR project therefore studies the following sequence of events in the exvessel phase: Release of melt form the pressure vessel Transfer and spreading of the melt on the basement Analysis of the physical-chemical processes that are important for corium behavior especially during concrete erosion with onset of solidification, and Stabilization of the melt by cooling through direct water contact. Investigation, description and evaluation of the listed processes involve cooperation of different complex, scientific branches. The ECOSTAR project therefore brings together 14 institutions of 5 countries, which cooperate with high efficiency. The project was originally planned for a period of 3 years. However, after about 2 years, an important partner had to withdraw from the project. Therefore, project coordination was taken over by Forschungszentrum Karlsruhe and the work program had to be modified. Unfortunately, no alternative solution could be found for a large-scale experiment on long-term corium concrete interaction, while for other important experiments, an adequate solution was realized. To fulfill the revised work plan under the new conditions, the duration of the ECOSTAR contract was extended by one year until end 2003, so that at the time of the FISA 2003 meeting, some tasks are still underway. B. WORK PROGRAM The work program consists of experiments, model development, and analysis activities, which are grouped under the following work packages: (1) Melt Release from RPV: Dispersion of melt and concrete erosion by melt jet The core melt can be collected (and cooled) in the lower cavity or on the basement only if the steam pressure of the primary circuit does not disperse the melt into the containment or sub-compartments. The DISCO experiments investigate the influence of reduced primary pressure and breach location on corium release at RPV failure. Another important aspect is formation of a pressure driven melt jet at RPV failure, which may lead to local erosion of the basement, a process which needs quantification by the KAJET experiments with high temperature melt jets and by model development. (2) Ex-vessel melt transport: Spreading experiment and code validation Spreading of large melt masses on the concrete surface is a process that is important for concrete attack and melt cooling as well. Large-scale experiments are performed and are used to complete validation of existing spreading codes. (3) Data and phenomena: Properties of multi-component corium melts, solidification of melts, and concrete erosion by oxide corium melts The thermo-physical properties of corium melts are an area of major uncertainties. Also concrete attack requires continued interest, especially in the long term, when partial solidification of the melt becomes important. This may influence the direction and efficiency of heat transfer, and brings up a series of important questions, which require a coupled analysis of thermalhydraulic and physicochemical processes. One of the related questions is e.g. the existence and stability of crusts at the concrete interface. (4) Coolability: Crust and cracks, top flooding, bottom flooding The dominant question for melt stabilization is cooling of the spread corium melt by water addition. Top flooding and bottom flooding are the two modes, which are investigated in small-scale and large-scale experiments and by theoretical models. Its feasibility under accident situation has to be evaluated. (5) Assessment for reactor situations and other applications

3 The various aspects of the investigated processes are evaluated for their relevance during the course of the accident and for accident management measurements both in design and in operator actions. Other industrial applications are also considered. C. MAIN ACHIEVEMENTS C.1 Melt release from RPV After successful depressurization of the primary circuit the residual pressure at RPV failure is below 10 bar. At failure of the RPV, dispersion of the melt and the erosive action of a melt jet are investigated. Dispersion of melt DISCO experiments The DISCO experiments at FZK quantify the corium fraction that would be transported from the RPV into the containment and thus would not be retained in the reactor pit. The EPR specific geometry was investigated, characterized by a relatively narrow reactor pit. A series of experiments was performed in a 1:18 scaled geometry using water or high density liquid metal at low temperatures as melt. For the first time, lateral breaches, circular holes, horizontal slots, and unzipping of the lower head (Figure 1) were systematically investigated. The final experiment within the program was done in the DISCO-H facility with high temperature alumina-iron melt and steam, and a central hole in the lower head. The cold experiments show the strong influence of (1) breach location, e.g. central or lateral hole, (2) RPV pressure, (3) cavity geometry and (4) model fluid. 75% of the melt are dispersed out of the cavity for primary pressures in the range of bar in case of a central bottom hole. Depressurization to less than 5 bar above containment pressure is needed to get dispersion < 10 %. Generally, lateral breaches lead to much smaller dispersion. For central holes, the hot test shows smaller dispersion than for the cold simulants. For EPR conditions, further hot tests are required to specify the pressure level below which melt dispersion is acceptable. Pressure increase in the containment, which is caused by heat transfer from the dispersed melt droplets and by hydrogen burn, is in the range 2 4 bar. Development and application of a computer code, performed outside of the ECOSTAR project, is important to understand the experiments and to apply them to the reactor case. Jet erosion KAJET experiments and modeling In case of local RPV failure, expulsion of the melt into the reactor cavity may be as a compact jet for a short period, followed by a dispersed release after gas break-through. The KAJET experiments at FZK are related to the short initial phase of a compact jet. They studied concrete erosion by oxide and metal melt jets driven by gas pressure between 3 and 8 bar. The molten corium was simulated by an alumina-iron thermite melt of up to 160 kg, initial temperature about 2000 C. The highest erosion velocity measured in the tests is about 10 mm/s. With the tests, a data base for model development and validation is now available. The melting of the structure material by the impinging hot jet has been identified by LEE/RUB as a decisive mechanism, while chemical and mechanical effects have minor influence. In the impingement region, a thin melt layer is formed and continuously removed by the deflected jet. The heat transfer from the jet to the concrete can be described by a relation of the form Nu = f (Re, Pr) within a factor 2. Application to the reactor accident scenario yields an eventual erosion depth up to 130 mm. The limiting criterion is the onset of gas break-through. C.2 Ex-vessel melt transport Spreading of a simulated corium melt on concrete is investigated in experiments at FZK under two aspects: (i) In a 2D-demonstration test (ECOKATS-2), a large mass (3200 kg, 800 l) of a mixed oxide/metal melt is poured with 20 l/s on a 2 m x 2 m concrete area. This

4 corresponds to flow conditions, which are expected under accident situation. The experiment shows, that spreading was complete and very fast (Figure 2a). The melt covers the concrete surface homogeneously after less than 60 s, which is a consequence of the high release rate of the melt. The 20 cm deep melt initiated strong gas release from concrete with intense H 2 flames. (ii) In a 2D-benchmark test (ECOKATS-1), 600 kg ( 220 liters) oxide melt are poured with a low flow rate of about 2 l/s only on a 3 m by 4 m concrete surface. This surface is sufficiently large, that the melt would stop before wall contact, and is used as benchmark for the spreading codes CORFLOW, LAVA, and THEMA in order to test, if codes are able to identify those critical conditions of flow rate and melt overheat, for which spreading is incomplete. Blind post-test calculations are presently underway and will be published after completion of the benchmark. The spreading codes will be used for design and licensing calculations of plants. C.3 Data and phenomena A review of the physicochemical database has been performed, which is used to predict the phase diagrams of the multi-component UO 2 -ZrO 2 -concrete system, serving also as contribution to the European thermodynamic database NUCLEA of the ENTHALPY project [1]. A report deals with the actual status of describing the density and viscosity of the relevant oxide mixtures. Slow solidification of UO 2 -based prototypic oxide melt mixtures has been studied at NRI Rež and CEA. It was confirmed that segregation does occur during crystallization, that means that crystals form with different concentrations of low melting (e.g. CaO-FeO) and high melting (UO2-ZrO2) compounds, depending on the crystallisation rate. Dendritic, cellular, and plane-front solidification were observed as crystallisation velocities were reduced. A mass diffusion coefficient for the components is estimated, which is an important parameter for the transition to plane front solidification. For certain concentrations, a miscibility gap for the oxide melts was found. Interaction of prototypic, pure oxidic corium with different types of concrete and under different decay heating rates was investigated in small scale experiments with sustained heating at FRAMATOME ANP. The objectives are characterization of the erosion velocities and quantification of the role of phase segregation during long-term concrete erosion. Downward heat fluxes and corresponding erosion rates were determined and can be used for MCCI model improvement. The temperature of the melt was found to stabilise near the liquidus temperature of the melt. For the experiments performed so far, no evidence of a stable interface crust was found. A further test with reduced decay power is under preparation. C.4 Coolability by direct water contact Crust and cracks Spreading experiments were analysed with respect to the existence of cracks, which could contribute to cooling. In the VULCANO experiments VE-U7 and -U8 of CEA, in- and ex-vessel corium with UO 2 and ZrO 2 was spread under dry conditions on ceramic or concrete substrate. The gas bubbles in the bulk of the solidified melt are from 10 µm to several mm and may improve coolant water ingression. However, the surface of the solidified melt is mostly dense, which may be due to a silica rich upper layer. For reactor application, it has to be considered that in the reactor case the height of the melt is increased by a factor 5 to 10. Top flooding If the core debris exists in form of a particulate bed, its coolability can be substantially improved by the use of downcomers, which bring the coolant water from the top of the bed to its bottom. The POMECO experiments at KTH showed an increase of the dryout heat flux from 50 to 600 %, thus accelerating and improving the coolability. An analytical model for the improved dry-out heat flux was developed.

5 In the large-scale ECOKATS-2 experiment at FZK (see Section C.2), the melt, after complete spreading over the 2 m x 2 m concrete surface, was top-flooded by water (Figure 2b). In spite of ongoing MCCI processes with substantial gas release and an agitated melt surface, the flooding process was mild and did not cause intense interactions of melt and water. The 20 cm deep melt formed a surface crust, from which some small volcanoes developed, but no ejection of particles was observed. Cool-down of the melt was slow, indicating the limited ingression of water. The surface crust was strictly anchored to the concrete sidewalls of the large test area already shortly after flooding and the hypothesized floating crust of the melt was not observed. This experiment shows only limited melt coolabilty by top flooding. The ANSYS code for top-flooding and related cooling process was developed at FZR, which includes water progression, crust growth and crust mechanics during ongoing MCCI, based on a 2-d CFD-model. After successful pre-calculations the model is presently applied to the new ECOKATS-2 test. Bottom flooding Adding water to the bottom of the melt pool results in improved coolabilty by fragmentation of the bulk of the melt. This is confirmed by three large scale CometPC-experiments at FZK with decay heat simulation. The initial melt mass of 800 kg was arrested and quenched in a period of 30 to 60 minutes only. This confirms that the CometPC concept is attractive providing fast arrest and cooling of the melt. The porosity, generated in the melt, is however not as homogeneous as observed in the original COMET concept with prescribed flow channels. Modification of the water injection, combining dedicated flow channels and the porous concrete layer, are proposed to further improve the bottom flooding concept. The DECOBI tests at KTH studied details of the bottom flooding process, in intermediate scale with high temperate simulant melts of different viscosities. For low viscosity melts, an intensive mixing of melt and coolant occurs, which creates 40 to 50 % porosity of the solidified melt and results in fast quenching. For high viscosity melts, the porosity is concentrated above the locations of water injection, and cooling of the less porous zones is delayed. A detailed theoretical model is developed at IKE, which describes porosity formation by the expanding steam with melt break-up and build-up of porous channels, stabilized by concurrent freezing due to rapid cooling by evaporation. Application of this model shows the influence of the water supply pressure and the viscosity of the melt on porosity formation. Application of the WABE code for porous melts shows a much higher cooling potential for bottom flooding, as steam and water flow are co-current, whereas in top flooding, countercurrent steam flow may limit water ingression. C.5 Assessment for reactor situations and other applications As an example for reactor relevance, spreading calculations have been performed for an EPR scenario and have confirmed that homogeneous spreading can be expected, if realistic initial conditions are fulfilled. This information was used to set up the experimental conditions for the spreading experiments. An assessment report is under preparation, which will condense all ECOSTAR results under the aspects of application. D. DISSEMINATION AND EXPLOITATION OF THE RESULTS The results of the project will be published, as far as possible, in the open literature and are presented at international conferences. The participation of industrial and licensing institutions in the project guarantees a direct transfer to the end-user in nuclear application, both for existing and for future plants. Especially the results and models on melt dispersion, melt spreading, and melt cooling have high relevance for actual planning and safety evaluation.

6 E. CONCLUSIONS The ECOSTAR project successfully investigates a series of phenomena, which are important to control and stabilize the ex-vessel core melt, and which are evaluated for accident mitigation: (1) Consequences of melt release from the RPV and possible mitigation were identified. (2) Complete spreading of a simulated corium melt on the basement under realistic conditions was demonstrated. Computer codes will be available for spreading calculations after completion of the present code validation process. (3) A better understanding of the physicochemical processes in the multi- component oxide melt during concrete erosion and partial solidification of the melt has been achieved. (4) The efficiency of cooling by direct water contact in the modes top flooding and bottom flooding has been evaluated by experiments and models. Furthermore, models were developed and validated to allow detailed application to plant accident analysis and planning of countermeasures. This results in substantial improvements of ex-vessel corium stabilization. Specific ex-vessel mitigation aspects could be strengthened further by EU research groups, using the EU sponsored Transnational Access to Research Infrastructure, especially PLINIUS [2] and LACOMERA [3]. References [1] A. De Bremaecker et al., European Nuclear Thermodynamic Database Validated and Applicable in Severe Accident Codes, FISA-2003, Luxembourg, November 10-13, 2003 [2] C. Journeau et al., Transnational Access to the PLINIUS Prototypic Corium Experimental Platform, FISA-2003, Luxembourg, November 10-13, 2003 [3] A. Miassoedov et al., Large Scale Experiments on Core Degradation, Melt Retention and Coolability (LACOMERA), FISA-2003, Luxembourg, November 10-13, 2003 Figure 1: Melt dispersal at RPV failure with circumferential rip (a) (b) Figure. 2: Melt during spreading on 2 m 2 m concrete surface(a) and after top flooding (b)