Garo Azarian and Bertrand de L’Epinois

CORIUM IN VESSEL RETENTION STAKES AS REGARDS THE SAFETY DEMONSTRATION

GARO AZARIAN (consultant)Email: garo.azarian92260@gmail.comBERTRAND de L’EPINOIS,AREVA

Email: bertrand.delepinois@areva.comParis, France

Abstract

Within the severe accident strategies and safety demonstration, In Vessel Retention (IVR) has emerged as a topic of major interest within the international community.

The variety of the associated phenomenon has been and is currently subject to considerable international R&D: corium formation, relocation and composition; permanent and transient phenomenon which constrain the heat flux through the vessel; critical heat flux reachable through outside vessel cooling; vessel thermochemical and mechanical behaviour; steam explosion physics; vessel pit and containment robustness under steam explosion; etc.

Given the stakes as regards the residual heat removal, the basemat protection against ablation and the potential consequences of steam explosion, IVR safety demonstration mobilizes significant discussions at global level.

These are all the more vivid than IVR considerations shape some structuring options of the containment lower part design and construction and that some of the questions involved are related to the practical elimination of situations which could lead to early and large releases.

In order to contribute to the discussion, this paper addresses:

- The phenomenon driving the critical heat fluxes, the order of magnitude of heat flux which can be reached and the associated vessel strength;

- The phenomenon concerning the corium behaviour inside the vessel and the associated heat flux (orders of magnitude and uncertainties);

- The design options to be considered.

The paper will present the main parameters to master in order to assess an IVR safety demonstration. It addresses the roles of the deterministic and probabilistic approaches and presents the practical elimination questions related to IVR.

1. INTRODUCTION

To cope with consequences of a severe accident with core melting in Light Water Reactors (PWR, VVER, BWR), In Vessel Corium Retention (IVR) strategies with external cooling of the vessel have been incorporated for the back-fitting in some GEN 2 reactors and in some new GEN3 reactors.

The corium decay heat is removed through the vessel wall, by boiling of water injected in the reactor pit. Thus doing, the vessel wall can be partially ablated. Because there is a limitation of the effectiveness of such cooling, corresponding to the Critical Heat Flux (CHF), safety margins towards the vessel failure, which could occur in few minutes at dry-out limit, are reduced for large power reactors (e.g PWR power above 600 MW).

In the light of remaining uncertainties during the corium relocation in the vessel lower head, impacting the heat flux distribution, it had been assessed that the probability of vessel failure could not be negligible even for around 1000 MW PWR [7 ]; for some new high power reactors, ex-vessel retention strategies with implementation of core-catchers have been preferred.

In case of vessel failure with IVR, the interaction of the corium jet released with the water of the reactor pit could lead to an ex-vessel steam explosion. Steam explosions are still an open R&D issue; Ex-vessel steam explosions which could threaten the containment should be practically eliminated because it could lead to large and early radiological releases.

1

IAEA-CN-251

On the other hand, a demonstration of the robustness of IVR solution and the evaluation of the associated reactor power limit providing margins against the vessel failure risk remain serious challenges. They deserve reassessing by taking into account new R&D insights, in particular those related to the uncertainties of thermochemical phenomenon within the corium.

The main phenomenon and main parameters to master in order to build and assess an IVR safety demonstration are presented below.

2. MAIN PARAMETERS

The IVR effectiveness evaluation is mainly based on the results of recent phenomenological analyses and the evolution of the state of art of knowledge over the past years related to the following three parameters:

a) In-vessel behaviour of the corium: the transient configurations of corium phases’ stratifications leading to loads on the vessel and a focusing effect higher than previously anticipated.

Because of the miscibility gap between in-vessel components of corium ( U-Zr-Fe-O), stratification

occurs between the oxide pool (U02,ZrO2, Zr) and the metal phase (Fe, U, Zr) in long term configuration, with lower density metal located above the oxidic pool surrounded by a crust limiting the heat transfer to the vessel wall.

The turbulent convective heat transfer in the oxidic pool (which contains the major part of the decay heat) impacts heat distribution sideward and upward to the top metal layer; it is an important parameter. One part of the heat in the top metal layer is removed by heat radiation upwards to the surrounding structures (around 20 to 50%), according to their emissivity which is also an important parameter The main part corresponds to the lateral heat transfer to the vessel wall, which induces a peak of local heat flux named “focusing effect”.

The sideward power of the top steel layer, which generates the focusing effect and ablates locally the vessel wall, depends on the thickness of the top metal layer: the thinner the metal layer, the stronger the focussing effect. The thickness of the top metal layer can indeed be considered as the important critical factor for the success of the in-vessel corium retention.

The decay heat is also an important parameter impacting the level of heat fluxes entering the inner wall of the vessel; it is why core melt scenarios with rapid core degradation beyond the scram, as a Large or Medium break LOCA associated with the unavailability of safety injection would be the most critical scenarios.

b) The CHF which can be credited to remove the decay heat, depending on the optimization of the water circulation in the reactor pit and on the outside vessel wall cooling. The local flux at outer vessel wall is compared to local CHF.

c) The residual thickness and associated mechanical behaviour of the vessel partially melted by the heat flux from oxide and in particular steel layers.

3. REACHABLE CHF BY OUTSIDE VESSEL COOLING

The effectiveness of external cooling of reactor pressure vessel has been examined for more than 20 years. Cooling of outer vessel wall, by submerging the vessel pit at severe accident signal, should be done before corium relocation in the lower head.

Garo Azarian and Bertrand de L’Epinois

FIG. 1. Relocation of corium

The dry-out limit is based on the local flow characteristics. This limit is higher when the vessel wall is vertical (resulting in easier steam evacuation) or when the water is sub-saturated at the inlet of the external vessel cooling system. Submerging the vessel is not sufficient to remove heat flux higher than 1 MW/m²; it is why many solutions have been investigated to prevent the film boiling regime and to ensure an efficient heat removal by nucleate boiling.

1.1. CHF enhancement by optimizing water circulation in the pit

An optimization of the shape of the vessel insulation is necessary to enhance a natural circulation, ensuring with an upward flow along the vessel and a downward flow of cooler water. This requires implementation of adequate clearances and passive water/steam flow openings to prevent steam blockage and water inlet at bottom of the insulation and steam vents near the top.

FIG. 2 Circulation of water in the pit The enhanced geometry aims at streamlining the two-phase motions, thus reducing the pressure drop and

inducing more liquid flow in the annular flow channel at a given heat flux. There is an international consensus that CHF in the range of [1.5 MW/m²-2MW/m²] can be obtained in this way.

Solutions with implementation of a forced convection during a certain time which could be combined at long term with a natural convection of water could be also efficient but with a drawback to have to rely on an active system at a very earlier phase of the severe accident.

1.2. CHF and wall/water Heat Transfer Coefficient enhancements

Some significant enhancements (factor higher than 1.5 on CHF) have been obtained by treatment of the vessel material surface; in addition, real oxidation conditions met during reactor life are favourable with comparison of some experimental tests. The oxidation behaviour of pure iron and iron-based alloy can change physical and chemical properties of material and dominantly affects boiling heat transfer capability.

Boiling architecture can be optimized by implementing a nano-particle-based coating for the vessel, which slows the phenomenon of dry-out. Modifying porosity characteristics can also increase the heat transfer [8]. To increase nucleate boiling sites can also provide higher CHF.

3

IAEA-CN-251

In addition to material surface treatment, an approach is to find solutions to be injected into the reactor pit water in order to prevent film boiling by increasing the Leiden frost temperature.The R&D research concludes that CHF significantly higher than 2 MW/m² could be obtained. Nevertheless, the limiting factor would in this case be the residual thickness of the vessel, It is reduced to 1 or 2 cm (for the material at a temperature < 650°C, ensuring the load bearing capacity) instead of the initial thickness in the range [15 cm-20cm]), when the heat flux at the outer wall reaches the range [1.5 – 2MW/m²]. There are some uncertainties in calculations of core degradation and relocation into the vessel lower head, because expected 3D effects are generally not considered. To cope with these uncertainties and to better ensure a long-term resistance of the vessel, some margins regarding its thickness must be taken account. These margins could be also necessary regarding to the risk of pressure spikes which are possible in case of reflooding in the vessel, even though the mitigation strategy has required to fully depressurize the primary circuit. It therefore seems of little interest to justify a significant enhancement of CHF, because it would be associated with a higher wall ablation: the range. [1.5-2MW/m²] seems a reasonable limit, and related R&D to justify such heat transfer can be well mastered.

1. CORIUM LAYERING AND FOCUSING EFFECT AT THEMOCHEMICHAL EQUILIBRIUM OF THE MOLTEN POOL

The maximum heat flux from the corium is the most important parameter, and the one with the highest uncertainties. It remains to be determined with a sufficient accuracy to justify some margin towards the vessel failure risk. Recent assessments, identifying the need to address transient thermochemical phenomena with mass transfers between relocated corium layers due to the changes of density of metal and oxide phases, have outlined the risk of increased heat loads with comparison to previous methodology applied at the beginning of years 2000, which used to solely address the final state when all the corium is in the lower head.

1.1. Corium layering in safety analysis

Many analyses to determine the focusing effect have considered that the bounding configuration would be the final configuration of all corium and its decay heat power relocated into the vessel lower head with a top steel layer including all the molten steel from melting of internal structures and of the vessel) in direct contact with the vessel located above the oxidic pool embedded by a crust which is a thermal resistance.

The validity of two- layers configuration has been questioned when the OECD MASCA R&D program [9 showed that the addition of steel in a sub-stoichiometric corium (UO2-ZrO2-Zr) may extract Uranium and Zirconium from the reduction of UO2 and ZrO2 at the top metal interface and form a metal alloy containing Uranium-Steel and Zirconium which becomes heavier than the oxidic pool. Such phenomena with downward migration in the bottom of the vessel of a heavy metallic layer depends on some conditions, in particular the ratio of Uranium and Zirconium in the corium as well as on the degree of the Zirconium oxidation during the core degradation, which is dependent on the core melt scenario.

The compositions of the three layers, a light metal at top, an oxidic layer at the middle, and a heavy metal layer at the bottom, can be determined, by using thermodynamic calculations assuming global thermochemical equilibrium of the corium, without addressing the transient processes of mass transfers between the different layers.

The formation of heavy metallic layer has the drawback to reduce the thickness of the light metal layer and to increase the focusing effect [1], [10].

1.2. Potential design improvements: additional in-vessel reflooding

Some PWR applications as in [3], have shown, according to their assumptions, that the vessel integrity could be challenged, because the reduced thin metal layer would lead to maximum heat flux significantly higher than 2MW/m² to overcome the related worsened focusing effect and thermochemical uncertainties,

Garo Azarian and Bertrand de L’Epinois

implementing a system of in-vessel reflooding to be activated during the severe accident has been proposed to complement the outside vessel cooling.

Because the heavy metal layer (and therefore removed from the top metal layer) is known to be strongly dependent on the mass of un-oxidized zirconium mixed into the oxide debris, the purpose of the in-vessel reflooding can be to oxidize more zirconium, reducing the amount of metallic uranium that may form in debris bed, and preventing according to experimental MASCA results, the formation of the three corium layers.

At the same time, the in-vessel reflooding would extend the time to dry out and reduce the decay heat of corium and associated heat loads through the vessel wall.

In addition to other design measures to increase steel masses during the mixing with corium to be able to increase the metal layer thickness and to reduce the heat fluxes, the implementation of an in-vessel reflooding requests to justify its reliability, because its seems difficult to guarantee a systematic availability of injection whereas the core melt started due to unavailability of sufficient in-vessel cooling.

Dedicated accumulators (pressurized tank), dedicated core make-up tanks, both sides of which are connected with the primary circuit, and elevated gravity tanks are contemplated solutions.

A question mark regarding such a solution is indeed related to the safety demonstration: how to justify the capability of fast actuation of a safety injection in the vessel in a core melt situation, created by the lack of injection in the vessel

2. TRANSIENT EVOLUTION OF CORIUM LAYERS AND WORSENED FOCUSING EFFECT RISK.

1.1. Transient layering of corium

The melting of steel internals, core support plate, partially the vessel, is a transient process, as the oxide relocation into the lower head. The conditions of corium stratification of corium and the evolution of the different layering depend on the kinetics, which is plant and scenarios dependent. By taking into account thermochemical effect and by including in the severe accident codes the transient mass transfers between layers due to the changes of density of oxide and metal phases, it could be possible to meet transitory worsened focusing effect compared to the assumed final bounding configurations addressed in many safety analyses.

To identify risk of thinner metal layer topping significant oxidic corium is one target of new on-going development of kinetics models dealing with corium phase stratification.

The previous modelling of steady state configuration, when all corium is located in the lower head, has been determined at a global thermochemical equilibrium, whereas the experimental tests with prototypic corium have shown that a downward mass transfer of steel through the horizontal crust between oxide and metal layer could be possible.

In addition to the possibility of heavy steel layer formation with downward flow, when the oxidation level of the oxide phase is low, MASCA tests also identified possibility of reverse flow of steel from the bottom heavy layer to the top: the reduction of UO2 and ZrO2 at the top metal interface produces Oxygen which can slowly oxidize the bottom metal layer until a limit of Oxygen concentration in oxide pool,, leading to transfer U and Zr from the heavy steel layer into the oxide phase and an upward flow of super- heated steel up to the horizontal upper crust, and through it to the top the pool. According to the available steel at top, such arrival of heated steel from bottom could also lead also to a worsened focusing effect. This upward metal transient will be limited by the Uranium mass transfer from the heavy layer to the oxidic layer. The knowledge of Uranium mass transfer is mandatory and an experimental validation is requested.

After the first relocation phase, different configuration of initial corium layering is possible before starting the vessel melting:- Only bottom metal layer (U, Zr, Fe) with only oxide (U, ZrO2, Zr) above, when the relocated amount of

steel is small and a high amount of heavy metal is not fully oxidized in the oxidic phase: in this case, if the top layer is only fed during a certain time, during the growth of top steel layer; high focusing effect could then occur with corium heat flux higher than 2 or 3 MW/m² [11] for a 1000MWe PWR.

5

IAEA-CN-251

- For higher steel masses at first phase of corium relocation, other initial configurations with steel above the oxide is also possible, and worsened focusing effect could occur during the thermochemical downward transfer through the crust of top steel, by reduction of its thickness.

FIG. 3 Thermochemical steel transfers through horizontal crust

Three phases of possible transient focusing effect could worsen the loads:- During the growth of a top metal layer (high flux obtained with some tons of steel, around 10 tons).- During the reduction of a top metal layer with downward flow through the horizontal crust- During an upward flow of super-heated steel from heavy bottom metal layer reaching the top of the pool

1.2. Transient models development

These transient transfers cannot be supported by the use of a global thermochemical equilibrium assumption which is only reached at long term: new models are developed to better represent the stratification kinetics. Various assumptions can be considered in new models [2], [5], [6] regarding the molten steel pathway:

(a) upward pathway of molten vessel steel which directly feed the top layer in direct contact with the vessel

FIG. 4 (a) layers (b) upward pathway of molten vessel steel It is applied in PROCOR and MAAP 5 EDF version codes development [ 2] and [ 6] in which different

permeability of the horizontal crust are considered.This pathway is assumed to be caused by the pool weighing down on the crust and pushing the steel

upward. The heavy metal, oxide and light metal layers are enclosed by the oxide crust and supposed to be close to the equilibrium state. Under certain conditions, the current in-equilibrium position of the layers depending on the current chemical composition of the pool can change, producing an inversion of stratification calculated. The out-of-equilibrium metal layer corresponding to the melted steel of the vessel is located above the crust in direct contact to the vessel and provides the focusing effect.

Fig 5 shows an evolution of lateral heat flux of the focusing effect metal layer during the decrease of top metal layer thickness with downward mass transfer through the crust (PROCOR code application)

Garo Azarian and Bertrand de L’Epinois

FIG. 5 PROCOR code application, example of lateral heat flux evolution

(b) Molten vessel steel incorporated in corium layers (oxide/metal)

It is applied in ASTEC code [5] by assuming that all the lateral crust along the inner vessel wall is not a physical barrier. In such model, adding a certain amount of steel in the whole molten pool, where a heavy bottom layer formed at corium relocation is covered by oxide corium, can lead to the inversion of stratification with an upward flow of heated steel through oxide layer reaching the top of the pool and leading to a transient focusing effect. Fig 6 is an example of maximum heat flux calculated by ASTEC code in [5] resulting from such inversion of stratification.

FIG. 6 ASTEC code application in [5] example of lateral heat flux evolution

7

IAEA-CN-251

These transient models are expected to determine worsened focusing effect with comparison to steady state configurations. Many developments are carried out in the frame of on-going European Union IVMR project.

Transient evolutions induce non-uniformities such as local variations of compositions and temperature gradients which may result in conditions far from the overall thermochemical equilibrium conditions. This leads to uncertainties in the predictions, in particular those induced by scale effects of experiments: R&D has already been identified in order to validate mass transfers between layers, depending on crust resistance permeability, oxidation progression in corium pool and Uranium diffusion.

1.1. Specific phenomena to be addressed related to focusing effect

(a) The 2 D heat transfer in the vessel wall

During the thermal ablation of the vessel wall, the 2 D heat diffusion in the wall could maintain the heat flux below CHF assumed in the range [1.5 -2 MW/m²]. In [4], analysis performed in steady state conditions outlines that a factor up to 2 could be obtained between inner and outer heat flux. Even though a gain is expected, such result is questionable in transient conditions during the local vessel ablation. The differences between maximum inner and outer wall fluxes can depend on the duration of the peak of heat flux at top layer.

(b) The natural convection in the top steel metal layer which impacts the maximum sideward heat flux of the focusing effect

When top layer is fed by molten steel at molten temperature, a period of time is necessary to develop the natural convection and the maximum radial heat flux. For some scenarios where high heat fluxes in the top layer are only reached during a spike, such grace period could limit the heat flux through the wall.

According to the steel layer thickness, the effectiveness of natural convection cells which transfer heat to the vessel wall have to be determined: for very thin steel layer, less than around 10 cm, a limitation of radial heat flux is expected; on the other hand, with larger thickness, natural convection in transient conditions could lead to higher temperatures of molten steel at the upper part of the layer in contact with the wall,.

FIG. 7 Natural convection cells in the top metal layer

2. REMARKS ON METHODOLOGY OF IVR SAFETY ANALYSIS

Recent analysis and insights from R&D results [1] have outlined that some important parameters as enhancement of CHF and load bearing capacity are well mastered and that significant uncertainties regarding transient evolutions of corium phases could challenge the standard approaches to compute the maximum heat flux and focusing effect. It would be necessary to update the models coupling thermochemical and thermal-hydraulics phenomena of corium in the vessel lower head, in order to determine, by taking into account

Garo Azarian and Bertrand de L’Epinois

worsening and favourable factors, the safety margins regarding vessel failure risk for selected core melt scenarios.

When the robustness of IVR with safety margins cannot be justified in all cases, it means that probability of vessel failure is not negligible: corium would then be released into the water filled reactor pit, through the vessel break, and could lead to an ex-vessel steam explosion. In a defence in depth approach, the steam explosion safety case is therefore to be resolved.

Ex-vessel steam explosion mechanical consequences are depending on the reactor pit design, with risk of missile formation which could hit the containment liner, or loose the leak-tightness of containment penetrations due to vessel movement. There are still uncertainties to evaluate with sufficient accuracy, at the reactor scale, the impulses of an ex-vessel steam explosion and there is no international consensus on consequences of such a scenario.

3. CONCLUSIONS

Among the parameters which drive the IVR effectiveness (for which outside vessel cooling should be limited by CHF in the range[1.5-2MW/m²]to keep a sufficient ablated vessel thickness), the most important and the one which shows the highest uncertainties is the maximum heat flux generated by the corium through a top layer of steel, leading to the focusing effect. It is related to the thickness of top steel layer of corium in contact with the vessel, which depends on the competition between the kinetics of evolution of stratified layers of corium and on the kinetics of steel addition from melted vessel and relocated internal structures.

New insights of corium R&D and new development of models let think that worsened focusing effect, compared to steady state analysis, could challenge the vessel integrity when the transient layering of corium phases is addressed during the corium relocation into the vessel lower head, by taking into account thermochemical effects. Significant progress of R&D is expected in the frame of on-going European Union IVMR project.

At the light of new analysis, it will be possible to re-assess the ability to establish an IVR safety demonstration which provides sufficient margins as regards vessel failure, steam explosion and containment integrity, which is the main safety objective under severe accident conditions.

The need to have to implement an in-vessel injection in case of a severe accident to overcome with uncertainties is a question mark concerning the safety demonstration.

When IVR reliability cannot be guaranteed, in a defence in depth approach, the ex-vessel steam explosion case is to be resolved. It is still an open R&D issue with large uncertainties. Safety margins are required in this field, as steam explosion, which could jeopardize the containment and lead to early and large radiological releases has to be practically eliminated. The ex-vessel steam explosion risk appreciation appears as a major driver as regards IVR: continued R&D and international consensus are much needed.

9

IAEA-CN-251

REFERENCES

[1] SEILER J., TOURNIAIRE B., DEFOOR F., FROMENT K., “Consequences of Material Effects on In Vessel Retention, Nuclear Engineering and Design, 2007, Volume 237, pp1752-1758.

[2] LE TELLIER R., SAAS L., “Transient stratification modelling of a corium pool in LWR vessel”, Nuclear Engineering and Design, 287, 2015.

[3] SCOBEL J-H., PLYS H.G.,” Using Accident Management to Address Phenomenological Uncertainties Related to Lower Plenum Debris Bed Chemistry and Mixing During In-Vessel Retention (IVR) of Molten Core Debris in theAP1000 Passive Plant “. ICAPP 2011, Nice, France.

[4] PARK J.W.BAE J.H., SEOL W.C. et al., “Integrated Conjugate Heat Transfer Analysis Method for In-Vessel Retention with External Reactor Vessel Cooling; ICAPP 2015, Nice, France.

[5] CARENINI L., FICHOT F., “The Impact of Transient Behaviour of Corium in the Lower Head of a Reactor Vessel for In-Vessel Melt Retention Strategies” ICONE 24, Charlotte USA 2016.

[6] BEUZET E., BAKOUTA N., BOISSAVIT M., HAURAIS F., LE BELGUET A., LOMBARD V., TORKHANI M. “Corium-related improvements in the EDF version of MAAP code in the frame of severe accident studies “NURETH 16, Chicago, USA,

[7] ESMAIILI H., KHALIB-RAHBAR M., “Analysis of in-vessel retention and ex-vessel coolant interaction for AP 1000” NUREG/CR-6849, USA, 2004.

[8] SON H.H., JEONG G., JEUN G., KIM S.J,”Effect of oxide layer thickness on the pool boiling critical heat flux of pre-oxidized RPV material “NURETH 16, Chicago, USA 2015.

[9] BECHTA S.V., GRANOVSKY V.S., KHABENSKY V.B., GUSAROV V.V., ALMIASHEV L.P., MEZENTSEVA L.P., KRUSHINOV E.V., KOTOVA S.YU., KOSAREVSKY R.A., BARRACHIN M., BOTTOMLEY D., FICHOT F. FISCHER M., “Corium phase equilibria based on MASCA, METCOR and CORPHAD results “Nuclear Engineering and Design, Vol 238, 2008.

[10] ZHANG H. et al “Analysis on effectiveness of in-vessel retention strategy based on ROAAM and 3-layer core melt configuration, ICONE 23, Chiba, Japan,2015.

[11] SEILER J.M., TOURNIAIRE B. “A phenomenological analysis of melt progression in the lower head of a pressurized water reactor” Nuclear Engineering and Design, Vol 268, 2014.