Analysis and interpretation of the LIVE-L6 experiment

Transcription

1 Analysis and interpretation of the LIVE-L6 experiment A. Palagin, A. Miassoedov, X. Gaus-Liu (KIT), M. Buck (IKE), C.T. Tran, P. Kudinov (KTH), L. Carenini (IRSN), C. Koellein, W. Luther (GRS) V. Chudanov (IBRAE) Presented by A.Palagin

2 Outline Introduction The experimental facility LIVE-L6 test conduct Main results of LIVE-L6 test Application of different code systems to the LIVE-L6 test Simulation of the LIVE-L6 test by the CONV code (KIT) Post-test calculations of the LIVE-L6 experiment (USTUTT-IKE) Calculations with MEWA module of system code ATHLET-CD Calculations with CFD code ANSYS-CFX Calculations with ICARE module of ASTEC code (IRSN) Calculations with AIDA module (GRS) Calculations with the PECM model (KTH) Summary and conclusions

3 Introduction The thermophysical behaviour of a corium pool in reactor pressure vessel of a pressurised water reactor is of principal importance for the prediction of core melt down accident development. This concerns, in particular, the influence of major critical phases and timing on the accident progression in terms of assessing the possibility to remove the released heat by external vessel cooling. The general objective of the LIVE program at KIT is to study phenomena resulting from core melting experimentally in large-scale 3D geometry with emphasis on the transient behaviour. The presented work considers the analysis and interpretation of the LIVE-L6 experiment, in which the molten pool (non-eutectic melt KNO 3 -NaNO 3 ) was separated by horizontal copper plate in order to study the layering effect. Different codes and models were used for post-test calculations: This includes CFD code CONV, fast running models implemented in the severe accident codes ASTEC (ICARE module) and ATHLET-CD (MEWA module), CFD code ANSYS-CFX, AIDA module (GRS), PECM models implemented in Fluent.

4 The experimental facility Core of the LIVE test facility is a 1:5 scaled hemispherical bottom of a typical pressurized water reactor. The inner diameter of the test vessel is 1 m and the wall thickness is 25 mm. The material of the test vessel is stainless steel. The heating system consists of 6 heating planes at different elevations with a distance of about 45 mm. Each heating plane consists of a spirally formed heating element with a distance of ~40 mm between each winding. All heating planes together can provide a power of about 18 kw. The vessel wall is equipped with 17 instrumented plugs at different positions along 4 axes. Each plug has 5 thermocouples (0, 5, 10, 15, 20 mm from the vessel wall). It is possible to place up to 80 thermocouples in the melt to measure its temperature at different positions. The LIVE test vessel Scheme of live test facility LIVE volumetric heating system

5 Power, W LIVE-L6 test conduct In the L-6 test the non-eutectic 80 mole% KNO 3 20 mole% NaNO 3 melt composition was used. The total volume of melt was 210 l (68 kg of KNO 3 plus 324 kg of NaNO 3 ) which corresponds to 43.5 cm of the pool height. In the L-6 test horizontal copper plate of 2 mm thickness located at the level of 33.3 cm separated melt in two parts. That was done in order to develop the approach to the analysis of layering and focusing effects that may take place during severe accident in Reactor Pressure Vessel Time, s Evolution of heating power

6 Temperature, ºC Main results of LIVE-L6 test MT1, r = 7.4 cm, h = 7 cm MT3, r = 7.4 cm, h = 17 cm MT4, r = 17.4 cm, h = 17 cm MT5, r = 27.4 cm, h = 17 cm MT6, r = 7.4 cm, h = 27 cm MT7, r = 17.4 cm, h = 27 cm MT8, r = 27.4 cm, h = 27 cm MT9, r = 37.4 cm, h = 27 cm Time, s Melt temperature evolution averaged over azimuth angle Melt temperature vertical profiles Noticeable is the decreasing of temperature in the upper part of the melt when approaching the cooper plate from above. Such temperature gradient means that there is certain heat flux from the upper part of melt (which has no heat sources) to the very vicinity of the copper plate. This effect may be explained by redistribution of heat fluxes coming from the lower (heated) melt part in the upper part due to convection.

7 Simulation of the LIVE-L6 test by the CONV code (KIT) CONV is 2D/3D thermohydraulic CFD code for the simulation of heat transfer due to conduction and convection in complex geometry, crust formation, etc. It was developed at IBRAE (Nuclear Safety Institute of Russian Academy of Sciences, Moscow) within the framework of the International RASPLAV project and additionally improved within the ISTC #2936 and #3876 Projects. For the modelling of heat generating viscous liquid in gravity field with consideration of the buoyancy force in a Boussinesq approximation the efficient difference scheme is applied to solve unsteady 3D Navier-Stocks equations in natural "velocitypressure" variables on fully staggered orthogonal grids for Cartesian coordinates. The Large Eddy Simulation (LES) scheme with no SGS closure (i.e. with implicit filtering) was realized in the code. A frozen version of the CONV code has been transferred from IBRAE to KIT within the framework of bilateral information exchange agreement in order to simulate the LIVE experiments

8 Simulation of the LIVE-L6 test by the CONV code (KIT) In 3D calculation the cubic meshing nodalization was used with 256 nodes in the vertical direction. The vertical nodes were condensed in the vicinity of the cooper plate in order to describe in more details the heat exchange between the upper and lower melt volumes. Materials distribution at 18 kw phase of the test Measured and calculated temperature evolution during first two phases of the test. TC location: horizontal coordinate 37.4 cm, vertical coordinate 25 cm

9 Temperature,C Vertical coordinate, cm Application of different code systems to the LIVE-L6 test Simulation of the LIVE-L6 test by the CONV code (KIT) copper plate Experiment Calculation ffff Experiment Calculation, 18 kw Calculation, 10 kw copper plate Vessel height, cm Vertical temperature distribution at the distance of 36 cm from the vessel symmetry axis vessel inner surface Horizontal coordinate, cm Measured and calculated crust thickness

10 Post-test calculations of the LIVE-L6 experiment (USTUTT- IKE) Calculations with MEWA module of system code ATHLET-CD The MEWA module is being developed and integrated in German system code ATHLET-CD for simulation of late phase core melting. MEWA describes the processes of late phase core degradation, the behavior of corium in the lower head, including debris formation, coolability, debris re-melting and molten pool behavior up to failure of the RPV. For the description of molten pools an approach based on a representative model is applied. The underlying conceptual picture divides the pool into a boundary layer along the cooled wall, where the melt flows down, a stratified region in the central lower part of the pool and a turbulent, isothermal region in the upper part. For the heat transfer in the boundary layer, the model of Chawla & Chan is used. The temperature distribution in the central part of the pool is determined from a one-dimensional energy conservation equation in axial direction, assuming that mass flowing down in the boundary layer is balanced by a corresponding upward mass flow in the central pool part. Heat transfer to the surface of the upper mixed layer is described by empirical correlations. For a possible overlying metallic layer a point model is used, i.e. only an average temperature of the whole layer is calculated. The heat transfer through the layer is described by empirical correlations.

11 Post-test calculations of the LIVE-L6 experiment (USTUTT- IKE) Calculations with MEWA module of system code ATHLET-CD Measured and calculated melt temperatures at 5 cm (MT1), 15 cm (MT9) and 25 cm (MT21) elevations Measured and calculated heat flux distributions for steady state at different heating levels.

12 Post-test calculations of the LIVE-L6 experiment (USTUTT- IKE) Calculations with MEWA module of system code ATHLET-CD Comparison of crust thickness distribution at the end of the test LIVE L-6: experiment (left) and MEWA calculation (right)

13 Post-test calculations of the LIVE-L6 experiment (USTUTT- IKE) Calculations with CFD code ANSYS-CFX In order to support the analysis and interpretation of the experiments and through this the further development of simplified models to be used in severe accident codes, complementary analyses have been started at IKE using the commercial CFD code ANSYS-CFX. CFD calculations offer a more detailed analysis that can provide better insight into details of the physical processes, which can be used for checking of model assumptions (e.g. stratification, thin boundary layer, upper layer with strong mixing) and the improvement of empirical laws in more simplified approaches. A structured mesh (100k on the finest one, refine factor of 3) was used, with enhanced local treatment on the crust side. The calculations were carried out in 2D cylindrical symmetry (one cell on azimuthal coordinate). The SST (shear stress transport model) turbulence model was applied, which is a blend of the k-ε and the k-ω models.

14 Post-test calculations of the LIVE-L6 experiment (USTUTT- IKE) Calculations with CFD code ANSYS-CFX Comparison of melt temperature (left) and heat flux (right) profiles calculated by CFX for different heating powers with experimental measurements

15 Calculations with ICARE module of ASTEC code (IRSN) The ASTEC code has a modular structure, each of its modules simulating a reactor zone or a set of physical phenomena. The ICARE module is used to describe in-vessel core degradation, core thermal hydraulics and molten pool behaviour in the lower head during severe accident. The ASTEC lower plenum model, implemented in the V2.0 version, can simulate up to 3 dense corium layers, depending on the metal-oxide phases separation and their stratification. Each layer is considered having a homogeneous composition. The molten pool in the LIVE-L6 test was therefore treated as two layers in the ICARE calculation, one for the melt positioned below the separating plate and another to simulate the melt above the separating plate.

16 Calculations with ICARE module of ASTEC code (IRSN) Temperature evolution of the lower melt layer Temperature evolution of the upper melt layer

17 Calculations with ICARE module of ASTEC code (IRSN) Vertical profiles of non-dimensional temperature difference Non-dimensional heat flux distribution at vessel wall

18 Calculations with AIDA module (GRS) The module AIDA (Analysis of the Interaction between Core Debris and the RPV during Severe Accident) is a coupled integral simulation model for: the thermal behavior of molten core material in the lower head the crust formation the cooling of the melt by a transient two-phase flow through a gap between crust and (RPV) as well as for the structural response of the RPV wall. The scope of AIDA modeling is restricted to the lower head and the RPV wall. Within the model all physical phenomena with short relaxation periods are modeled by correlations. For processes with long relaxation periods, such as the heat conduction through the RPV wall or the crust formation in the meltdown, differential equations are used. For the case when a metallic layer is formed above the oxidic melt, a point model is used which determines the average temperature of this layer. The heat conduction in the crust is considered one-dimensional and stationary. The melt crust system is thermally connected to vessel wall, with the latest being modeled by a two dimensional differential equation.

19 Calculations with AIDA module (GRS) Experimental (averaged over 36 TC) and computed mean temperatures of the melt Computed crust thickness during constant heat power phases

20 Calculations with the PECM model (KTH) The PECM (Phase-change Effective Convectivity Model) has been developed for simulation of melt pool natural convection heat transfer. Using a similar approach employed in the PECM for simulation of turbulent natural convection heat transfer, the metal layer PECM has been also developed for simulation of mixed convection heat transfer in the metallic layer atop of oxidic melt pool. Both mentioned PECM models were applied for simulation of the LIVE-L6 experiment in this work. The instantaneous liquid melt velocities are not employed in PECM, therefore solving Navier-Stokes equations is not required. This allows the PECM to be computationally efficient. Only energy conservation equation is solved using the Fluent code solver. The turbulent heat fluxes in different directions are transferred to the cooled boundaries by the means of directional characteristic velocities which represent energy splitting in the melt pool.

21 Calculations with the PECM model (KTH) Transient temperatures of PECM simulation and LIVE-L6 experiment Vertical temperature profiles of the PECM simulation and experiment

22 Calculations with the PECM model (KTH) PECM and experimental crust thickness The PECM simulation and experimental heat fluxes at the end states of 18 kw and 10 kw heating

23 Summary and conclusions The thermophysical behaviour of a corium pool in pressure vessel of a Pressurised Water Reactor (PWR) in the course of core melt down accident is of principal importance for the prediction of its development. The general objective of the LIVE program at KIT is to study the phenomena resulting from core melting experimentally in large-scale 3D geometry with emphasis on the transient behaviour. In the presented work the analysis and interpretation of the LIVE-L6 experiment were described. A range of different codes and models is used for post-test calculations and comparative analyses. This includes fast running models implemented in the severe accident codes ASTEC (ICARE module) and ATHLET-CD (MEWA module), CFD code CONV, CFD code ANSYS-CFX, AIDA module (GRS) and PECM models implemented in Fluent. Generally, all the codes show satisfactory agreement with the experimental data. However, certain discrepancies (underestimation of the upper pool temperature, overestimation of the crust thickness over the separating plate, etc.) were revealed. The cross-comparison of different codes calculation results, analysis of discrepancies and possible recommendations for the models improvement are foreseen within a benchmark calculation study.