Getting to the core of the problem

The Committee on the Safety of Nuclear Installations (CNSI) sponsored a workshop in November 2000 on the RASPLAV project. This project was to provide data on behaviour of molten core materials on the reactor pressure vessel lower head under severe accident conditions, and assess possible physicochemical interactions between molten corium and vessel wall. Data was also obtained to confirm heat transfer modelling for a large convective corium pool in the lower head. The project consisted of the following components:

•To provide data from large-scale integral experiments of behaviour and interactions of core-melt materials in the lower head.

•To perform small-scale corium experiments to measure the thermophysical properties (density, electrical and thermal conductivity, viscosity) to perform and interpret the integral large-scale tests.

•To determine uncertainties introduced by using non-prototypic conditions and materials in the small-scale corium experiments.

•To carry out molten-salt experiments to study heat transfer processes in the melt, to justify the choice of procedures for large-scale experiments such as the heating method, and to develop an understanding of relevant phenomena, such as crust formation, and non-eutectic materials behaviour.

•To develop computer tools and models to analyse results from large-scale integral tests and the supporting small-scale experiments.

During the first phase of RASPLAV, the large-scale experiments demonstrated that behaviour of corium melts differed from simulant materials. Under certain conditions, corium would separate into two layers enriched in zirconium or uranium.

Phase two concentrated on the physical and chemical phenomena in a convective molten pool. The effect of different corium compositions, potential for and effects of material stratification and influence of various boundary conditions were investigated. This involved a combination of integral and separate effect tests, including molten-salt tests to investigate non-eutectic mixtures and the effects of stratification, extension of material property database to allow interpretation and modelling of experimental data.

The CSNI held a seminar where the major outcome of RASPLAV could be presented and discussed in the context of other experiences on severe accidents. The objectives of the seminar were:

•To review RASPLAV experimental results.

•To exchange information on research.

•To discuss progress made on understanding severe accident progression.

•To discuss applicability to nuclear power plants and use of the results.

Overview

First priority issues for current plants:

•Ex-vessel debris/melt coolability. This is essential for timely stablisation and termination of a postulated accident and to assure the public that this is so. Current research programmes have not reached that goal. New ideas are needed to assure ex-vessel melt/debris coolability.

•Ex-vessel steam explosions. These can lead to early containment failure for some BWRs and possible leakage in containments of some PWRs. There is a connection between ex-vessel steam explosions and coolability. Lack of the former can provide the credible accident management option of establishing a pool of water under the vessel and forming a coolable particulate debris bed. Data shows that oxidic corium may be resistant to propogation of steam explosions. This may be the key to resolving the steam explosion issue. A fundamental understanding of these observations is essential, in particular with ex-vessel conditions of low pressure and high subcooling.

•Basemat failure. This is important where access of water to ex-vessel melt/debris is not available. Basemat failure may imply contamination of ground water supplies and spread of radioactivity to the environment. The technical issue is predicting basemat failure time due to long term multidimensional erosion of the concrete basemat.

As second priority issues for current plants have been selected:

•The mode of lower head failure is needed to specify initial conditions of melt discharge for containment loadings, in particular for the ex-vessel steam explosion analyses. Timing is important for the feasibility of AM measures to prevent vessel failure.

•Core quenching is accident management actions of water delivery to the vessel to flood a damaged but not relocated core and flooding the lower head when an oxidic melt pool, covered by a metallic layer, is present. The former may produce more hydrogen, latter could produce stratified steam explosions.

•Iodine chemistry. This was investigated in the Phebus FP project. Data indicates formation of organic iodine, which may need extra systems for removal. It can increase environmental release in case of leaky containments and filtered-vent releases.

•Instrumentation and diagnostics. These identify progression of the accident to the operator, and facilitate proper accident management actions.

•Steam generator tube failure. This is of concern for high pressure scenarios and aged steam generators. The accident management actions are to flood the secondary side and reduce the primary pressure.

In addition, first priority issues for new plant have been determined.

•In-Vessel melt retention (IVMR). The remaining questions on this are: (i) effects of melt stratification on vessel wall thermal loading; (ii) composition of the metal layer and its effect on the focussing of the heat flux; (iii) reliability of the gap cooling mechanism and (iv) plant maximum power level that can be reliably certified for IVMR.

•Core melt spreading and retention in an ex-vessel core catcher. This is the severe accident management scheme employed for the EPR. The main uncertainty is in the process of retention in a crucible for mixing sacrificial material in the corium melt and its subsequent failure. It is necessary to generate high flow rates to assure spreading over the whole surface area of the core retention device. The other uncertainty is in the long term cooling of the spread melt.

•Core melt retention in an external vessel. This is the concept of having core melt of a large power LWR discharge from the vessel into a larger diameter steel vessel housed in the containment below the reactor vessel.

•Innovative ex-vessel melt/debris coolability concepts. These are concepts for stablising core melt in the containment. One is adding water to the melt layer from the bottom. This helps cool and quench even relatively deep layers of melt. Another concept is using downcomers which increase the dry out heat flux in particulate debris beds.

Experimental results

There were eight papers presented at the seminar, four on results from RASPLAV, and four on results from other institutions. These papers covered construction of the facilities used in RASPLAV, including the small tungsten facility in which 1-2kg of corium can be heated to 3100°C and the large-scale facility in which experiments on 200kg of corium could be carried out; the physical property measurements for the high temperature corium melts to provide the database of physical properties for various corium compositions used in RASPLAV experiments; the data measured was for kinematic viscosity, surface tension, thermal conductivity, liquidus-solidus temperatures and density. Data was already measured for thermal conductivity of fluoride salts used in the RASPLAV-Salt test programme.

The results of material studies were presented. In addition, the results of the salt tests were also presented. These showed that the side-wall heating did not differ greatly from the direct electric heating in terms of the heat flux experienced by the vessel walls. The other tests investigated the crust formation and its effect on the heat transfer. It was found that in the thermal hydraulic steady state, the interface temperature was the liquidus temperature of the remaining liquid after the segregation.

In addition, test results on corium-steel interaction were presented. A small corium melt pot was in a cold crucible and specimin such as a steel rod was inserted into the melt pot. Post test calculations examined the parameters involved, like heat flux at interface.

Four experiments were performed to support in-vessel melt retention (IVMR). These were the Isabel-miscibility gap experiment to identify liquid/solid interface conditions and to demonstrate the miscibility gap in O-U-Zr system; the Isabel-vessel experiment to clarify physico-chemical interaction between metallic corium and vessel steel using electron beam heating at a heat flux of 1MW/m2; the Sultan experiments on external coolability of vessel under natural convection two-phase flows; and the Bali experiment to simulate corium pool convection in lower plenum and to study heat transfer characteristcs including the focusing effect.

The seminar also discussed the Simeco experiments to study the effects of melt pool stratification on vessel wall thermal loads. The experimental results indicated that high heat flux occurrred compared to non-stratified cases. The data indicated that a Richardson number of 5 as critical value for stability of the interface between layers.

Theoretical analyses

The seminar discussed development, validation, and application supporting RASPLAV, the major results achieved during the six years of RASPLAV and highlighted the continuation of work within the Masca project, the intrepetation of RASPLAV results, and discussions of the salt test analysis efforts with a CFD method. The first paper compared side-wall heating and direct heating. Comparing cases with crust and cases without crust and comparison of eutectic and non-eutectic salts were discussed in the second paper. The third paper pointed out the important result that the liquidus temperature of the residual liquid, and the absence of the mushy zone, can be applied as the boundary temperature for the oxidic pool when performing heat transfer calculations under thermal hydraulic steady state. For noneutectic materials, the final steady state temperature distribution in the melt depends on test procedure, as the chemical composition of the crust and liquid pool becomes different, and the pool liquidus temperature is different. The fourth paper pointed out the importance of fundamental heat transfer physics and the importance of the obtained RASPLAV salt test results for the code analysis validation.

Corium tests were the object of the macroscopic mass and heat transfer analyses applying the Conv codes. Conv codes are an adequate tool for planning and control during RASPLAV corium tests. Development of Conv codes includes the extension of applicability of the code to higher Rayleigh numbers, implementation of physico-chemical models and arbitrary geometry.

Corium separation in two layers, observed with C22 corium-containing carbon, is not due to a miscibility gap, as carbon-free corium with oxidation beyond 20% is outside the U-Zr-O miscibility gap. Below the liquidus temperature, there is partial separation of the metallic liquid through gaps, cracks and porosoties in carbon-free C22 or C32 corium tests. In the case of carbon-free corium, this phenomenon is not sufficient to promote transfer of the metal to the surface of the oxidic material. Two driving forces have been proposed, density differences and CO gas bubble entrainment.

Application

Corium material properties that were obtained in the RASPLAV programme have added valuable information to materials propertiies data base. This information was used in assessing IVR at Loviisa.

The Sonata program analysed formation of a gap between aluminium oxide/iron crust and vessel wall, permitting water ingression into the gap, cooling the vessel. This was demonstrated in a 1/8th scale test. The feasibility of gap cooling has not been demonstrated for reactor size lower head.

In the Forever program, data on the multiaxial creep behaviour of French reactor pressure vessel steel which maintained melt pool convection inside a 25 bar pressurised 1/10th scale vessel. Maximum temperatures of 1000°C were maintained and vessel failure was achieved.

Recommendations

The physico-chemical behaviour regarding macro stratification showed that solid zirconium oxycarbides could form at low temperatures. These do not melt when the temperature increases and should not be dissolved by the oxidic phase. Thus these oxycarbides could separate by density effects after the melting of the oxidic phase.

In the applicability of RASPLAV results to severe accident modelling, two areas of work must be followed up: completion of RASPLAV and related experiments to solve physicochemical aspects of convective corium pool behaviour; and continued refinement of the models and codes as the results of future experimental work becomes available.

In addition, issues that need further experimentation to be resolved include:

•The effect of melt stratification on thermal loading of the vessel walls must be quantified before the more complex issues of corium thermochemistry as a function of composition can be addressed.

•Thermodynamic analyses of small-scale and medium-scale corium tests are performed for real corium compositions to establish if melt stratification is caused by liquid immiscibility and if thermodynamic equilibrium was achieved.

•To establish if prototypic corium compositions might fall within the miscibility gap before launching a large programme on corium thermochemistry. This would involve examination of hypothetical melt relocation scenarios.

•Further tests should be performed with non-eutectic salt mixtures in the recently modified molten-salt test facility.

•The molten-salt test facility can be used for experiments with a stratified molten pool containing a thin metallic surface layer.

•Physical property measurements should be continued and their relevance to severe accident phenomena established. For successful modelling of reactor accidents, the following material properties for corium as a function of chemical composition are needed: heat capacity, thermal conductivity, viscosity, density, emissivity and surface tension.

•Critical properties governing prolonged in-vessel retention of molten corium are probably density and thermal conductivity. Densities of different corium compositions may influence the extent of stratification, and thus the local heat flux distribution.

•Corium surface tension may play an important role in controlling relocation behaviour, and thus the likelihood of steam explosions.

•Additional code development work is still needed to address the full range of likely accident scenarios.

•The code development recommended for verifying the in-vessel melt retention concept can be divided into melt pool thermal hydraulics, corium chemical behaviour and interactions, mechanical response of the reactor vessel, and upper crust stability.

•The applicability range of the codes needs to be extended from the current Rayleigh number limit of 1014 up to values of 1016-1017. This will require development and assessment of a turbulence model.

The ultimate goal is to be able to predict corium thermochemical behaviour, like crust formation, uranium and fissionproduct partitioning between the diffrerent layers and phases, and intermetallic exothermic reactions between zirconium and steel.

Rapid developmental work is taking place in the analytical work that is being carried out in support of and as a result of RASPLAV experimental work. This analysis effort is being done such that the code is directly applicabile to reactor cases.