1. Analysis of the SGTR accident for LFR by SIMMER code Nicola Forgione CIRTEN Consorzio Interuniversitario per la Ricerca Tecnologica Nucleare UNIVERSITA’ DI PISA Dipartimento di Ingegneria Meccanica, Nucleare e della Produzione 3 rd LEADER International Workshop, September 2012

2. Content 3 rd LEADER International Workshop, September 2012 Introduction SIMMER generalities SIMMER history SIMMER validation phases Example of validation study for FCI phenomenon SIMMER applications for SGTR analysis at UNIPI Simulation of SGI tests Experimental campaigns on LIFUS 5 facility Simulation of LIFUS 5 tests Simplified parametric analysis of the SGTR accident for LFR Conclusions

3. Introduction 3 rd LEADER International Workshop, September 2012 The interaction between two fluids, of which one (e.g. lead) is less volatile and at higher temperature than the other one (e.g. water), results in the production of high pressure vapour Thus, this is one of the most important concerns for safety issues of: –lead and sodium cooled reactors belonging to “Generation IV” systems –ADS where both core and target are cooled by LBE In LFR, the liquid metal (primary coolant) might come into contact with the water flowing in the steam generator because of an accidental Steam Generator Tube Rupture (SGTR) CCI In SFR a loss of coolant accident can increase the core temperature up to the fuel and steel melting, leading this mixture to interact with the surrounding coolant FCI One of the crucial issue for the safety analysis is represented by the evaluation of the energy released in such interactions, in order to have indications of the potential loads and the resulting damage on reactor structures

4. SIMMER generalities SIMMER III is a 2D, three velocity-field, multi-component, multiphase, Eulerian fluid- dynamics code coupled with neutron kinetics model. It can deals with safety analysis problems in advanced fast reactors 3 rd LEADER International Workshop, September 2012

5. 1st step: SIMMER was developed in 1974 at Los Alamos National Laboratory (USA) for HCDA (Hypothetic Core Disruptive Analysis) in LMFR 2nd step: development of a small prototype fluid dynamics code AFDM, at the base for new SIMMER code development 3rd step: SIMMER-III development in 1988 in collaboration with PNC (now JNC, Japan Nuclear Cycle Development Institute) 4th step: in 1992 Version “0”. Beginning of an European-Japanese (FZK, CEA, IRSN) cooperation on LMFR R&D 5th step: in 2000 finalization of the code assessment. Beginning of the reactor integral application for safety analysis studies 6th step: two iso-model codes (SIMMER-III and SIMMER-IV) coupled to two neutronic codes (TWODANT and THREEDANT) SIMMER history 3 rd LEADER International Workshop, September 2012

6. Phase 1 (1992-1996) applied to single- and multi-phase flow benchmark problems, small-scale experiments with reactor and simulant materials, and physical problems with known solutions. It consists of 32 problems, which tested specific code models: fluid convection algorithm, interfacial area and flow regimes, momentum exchange functions, heat transfer coefficients, melting/freezing and vaporization/condensation Phase 2 (1996-2000) applied to integral, complex multiphase situations. It is intended to cover key accident phenomena, which are directly relevant to the CDA and include: boiling pool, fuel relocation and freezing, material expansion, fuel coolant interactions (FCIs), and disrupted core neutronics 3 rd LEADER International Workshop, September 2012 SIMMER validation phases

7. 3 rd LEADER International Workshop, September 2012 SIMMER validation phases

8. 3 rd LEADER International Workshop, September 2012 SIMMER validation phases Problem # Full TitleShort TitleOrg. Category 3: Fuel Coolant InteractionFZK 3.1Analysis of the behavior of thermiteTHINAJNC 3.2Analysis of high pressure corium melt quenching testFAROJNC 3.3Analysis of large scale fuel-sodium interaction in the TERMOS T1 experiment THERMOSCEA-G 3.4Analysis of FCI experiment in alumina/water systemKROTOSJNC 3.5Analysis of PREMIX experiment PM06PREMIXFZK 3.6Analysis of QUEOS experiments Q08 and Q12QUEOSFZK7JN C Category 4: Material Expansion DynamicsCEA-C 4.1Calculation of the SGI expansion phase experiementSGICEA-G 4.2Analysis of Purdue OMEGA testsOMEGAJNC 4.3Calculation of CARAVELLE 6 experimentCARAVELLECEA-C 4.4Analysis of VECDTORS experimentVECTORSJNC 4.5Analysis of developing anular flowAnnular flowJNC Category 5: Disrupted Core NeutronicsFZK 5.1Analysis of FCA-VIII fuel slumping experimentsFCAJNC 5.2Analysis of space-time neutron kinetics using the improved quasi- static method in SIMMER-III KineticsJNC 5.3Analysis of FCA-VIII fuel slumping experiments with TWODANTFCA/TWODANTFZK 5.4Neutronic validations for reactor transition phaseERANOSCEA-G Problem # Full TitleShort TitleOrg. Category 3: Fuel Coolant InteractionFZK 3.1Analysis of the behavior of thermiteTHINAJNC 3.2Analysis of high pressure corium melt quenching testFAROJNC 3.3Analysis of large scale fuel-sodium interaction in the TERMOS T1 experiment THERMOSCEA-G 3.4Analysis of FCI experiment in alumina/water systemKROTOSJNC 3.5Analysis of PREMIX experiment PM06PREMIXFZK 3.6Analysis of QUEOS experiments Q08 and Q12QUEOSFZK7JN C Category 4: Material Expansion DynamicsCEA-C 4.1Calculation of the SGI expansion phase experiementSGICEA-G 4.2Analysis of Purdue OMEGA testsOMEGAJNC 4.3Calculation of CARAVELLE 6 experimentCARAVELLECEA-C 4.4Analysis of VECDTORS experimentVECTORSJNC 4.5Analysis of developing anular flowAnnular flowJNC Category 5: Disrupted Core NeutronicsFZK 5.1Analysis of FCA-VIII fuel slumping experimentsFCAJNC 5.2Analysis of space-time neutron kinetics using the improved quasi- static method in SIMMER-III KineticsJNC 5.3Analysis of FCA-VIII fuel slumping experiments with TWODANTFCA/TWODANTFZK 5.4Neutronic validations for reactor transition phaseERANOSCEA-G

9. THINA: out-of-pile experiments, in which a thermite mixture of molten alumina and iron was injected into a sodium pool from the bottom. TH564 and TH562 tests are simulated by SIMMER-III. The aim was to investigate the phenomenology and physics of thermal-hydraulic interactions between melt and sodium. S-III well reproduces the pressure history and the experimental mechanical energy release, so the conversion of thermal into mechanical energy S-III underestimates the axial expansion of the 2Φ region (maybe non- condensable gas initially separated from the melt) S-III REASONABLY SIMULATES THERMAL INTERACTION BETWEEN SODIUM AND MELT 3 rd LEADER International Workshop, September 2012 Example of validation study for FCI phenomenon 3270 K Al 2 O 3 +Fe 780-790 K 2D r-z Geometry Mesh 6x30 cells 4 material component 3 velocity fields K.Morita et al., “SIMMER-III applications to fuel-coolant interactions”, Nucl. Eng. Des., 1999

10. 3 rd LEADER International Workshop, September 2012 SIMMER applications for SGTR analysis at UNIPI Goal: analysis of the phenomena at the basis of the interaction between water and heavy liquid metals (CCI) due to its importance for the safety aspects of SGs foreseen for the LFRs:  SIMMER assessment for CCI with the post-test analysis of: relevant experimental tests coming from literature (e.g. SGI) experimental tests that have been carried out at ENEA by LIFUS 5 facility in two different configurations  Simplified thermal-hydraulic study of the possible consequences deriving from a Steam Generator Tube Rupture (SGTR) accident, which could occur in the LFR prototype LEADERTHINS Lead-cooled European Advanced DEmonstration Reactor Thermal-Hydraulics of Innovative Nuclear Systems

11. 3 rd LEADER International Workshop, September 2012 Simulation of SGI tests Expansion phase  In case of severe accident: discharge of molten material from core and acceleration of surrounding coolant; redistribution of granulated fuel  Important for work energy potential and mechanical structure load assessment after severe accident  Upper core and vessel structures & behavior to be known (impact on mitigation)  A good evaluation of the expansion phase is crucial for assessing the work potential with a better accuracy than that obtained through the isentropic expansion, which gives a very conservative estimation without taking into account momentum and heat transfer to structures. SGI Experiment

12. An experimental campaign called SGI (Schnelle Gas Injection) was performed in 1994 in former Forschungszentrum Karlsruhe, now KIT. The experiments dealing with the injection of a high pressure gas into a stagnant liquid pool. Three tests, respectively 91, 93 and 95, have been simulated with SIMMER III and FLUENT codes. Test n. Presence of inner structures Inner structure diameter [cm] Nozzle diameter [cm] Initial pressure [bar] 91yes23911 93yes2396 95yes2393 3 rd LEADER International Workshop, September 2012 Simulation of SGI tests Scheme of the experimental facility for the SGI Campaign (units: [cm])

13. SIMMER III R-Z geometrical domain The simple geometry of the experimental facility allows to set- up a two dimensional axial-symmetric computational domain), which was divided into about 10800 cells. The time step chosen for the simulations is equal to10 -6 s. FLUENT geometrical domain An axial-symmetric domain has been set up for representing the test section. It was subdivided into 31 radial and 68 axial cells (2108 cells). The pressure vessel, the connection tube and the inner walls inside the main vessel have been shaped through ‘no calculation’ regions. In order to perform a comparison between two different kinds of codes the tests 91, 93, 95 have been chosen. These tests have been arranged with the same geometrical features, that is the presence of an inner vessel wall with a diameter of 23 cm and the nozzle diameter equal to 9 cm. The only parameter changed has been the nitrogen injection pressure which has been respectively set equal to 11 bar for test 91, 6 bar for test 93 and 3 bar for test 95. 3 rd LEADER International Workshop, September 2012 Simulation of SGI tests

14. Test 91: P inj = 11 bar Test 93: P inj = 6 bar Pressure transient in the cover gas region Experimental and numerical bubble’s shape evaluation (Test 91) 3 rd LEADER International Workshop, September 2012 Simulation of SGI tests

15. Test 91; P inj = 11 bar Test 93: P inj = 6 bar Gas bubble volume comparison As can be seen, the bubble volume time trends obtained from FLUENT and SIMMER match each other in the considered time range 3 rd LEADER International Workshop, September 2012 Simulation of SGI tests

16. Initial configuration of LIFUS 5 facility, adopted to perform the Test n.1 of the IP-EUROTRANS campaign S1 - reaction tank Volume0.1 m 3 Inside diameter0.42 m Design pressure200 bar Design temperature 500 °C MaterialAISI 316 S2 - water tank Volume0.015 m 3 Inside diameter4 in.sch.160 [in] Design pressure200 bar Design temperature 350 °C MaterialAISI 316 S3 - safety tank Volume2.0 m 3 Inside diameter1.0 m Design pressure10 bar Design temperature 400 °C MaterialAISI 316 S5 - expansion tank Volume 10.1 liters Inside diameter 6 in.sch.160 [in] Design pressure 200 bar Design temperature 500 °C Material AISI 316 3 rd LEADER International Workshop, September 2012 Experimental campaigns on LIFUS 5 facility Reaction tank Safety tank Expansion tank Water tank Reaction tank

17. Reaction tank Water tank Safety tank Scheme of the modified configuration of LIFUS 5 facility, adopted to perform the Test n. 3 and 4 of the IP-EUROTRANS and Test n. 1 and 2 of ELSY campaigns Direct connection of the reaction vessel S1 with the safety vessel S3 through a discharge line The four welded plates inside S1 have been removed 3 rd LEADER International Workshop, September 2012 Experimental campaigns on LIFUS 5 facility Reaction tank Safety tank

18. IP –EUROTRANSELSY Test n.1Test n.2Test n.3Test n.4Test n.1Test n.2 LBE temperature [°C] 350 400 Pressure on LBE free level [bar] 111111 Water injection pressure [bar] 70640 180 Water temperature [°C] 235130235 325 LBE volume [l] 1058010080 Cover gas volume [l] 5 (in S5)20 (in S1)no20 (in S1) Orifice diameter [mm] 484444 Injector penetration [m] 0.080.060.05 0.0050.25 Test duration [s] 10 3333 Presence of S5 yesno Summary of the operating conditions in IP-EUROTRANS and ELSY tests 3 rd LEADER International Workshop, September 2012 Experimental campaigns on LIFUS 5 facility

19. Detailed computational 2D model (20x32 cells) set-up in order to improve the results of the simulations respect to a previous less detailed model SIMMER III computational model for the Test n.1 of the IP-EUROTRANS campaign 3 rd LEADER International Workshop, September 2012 Simulation of LIFUS 5 tests Reaction tank Expansion tank

20. Pressure and temperature results obtained for IP-EUROTRANS Test n.1 Comparison between the experimental pressure trend and that computed by SIMMER III Comparison between the temperature measured by the middle thermocouple in S1 and that calculated by SIMMER III The experimental temperature data were found to be in a quite good agreement with the calculated trend 3 rd LEADER International Workshop, September 2012 Simulation of LIFUS 5 tests

21. SIMMER III and IV computational models for the Test n.3 and 4 of the IP-EUROTRANS campaign and the Test n. 1 and 2 of the ELSY program 3-D calculation domain employed in SIMMER IV for Tests 3 and 4 IB=20, JB=18, KB=16, cells Two-dimensional calculation domain, 23x39 cells 3 rd LEADER International Workshop, September 2012 Simulation of LIFUS 5 tests S3 Safety tank S2 Water tank Reaction tank S3 S1 S2

22. Pressure results obtained for the Test n. 3 and n. 4 of the IP-EUROTRANS campaign, with SIMMER III and IV code Test n. 3Test n. 4 Both simulations predict the peak pressure that appear at about 0.6 s, even if in Test 4 the pressure trend is still overestimated, but a better agreement between the two simulations is observed 3 rd LEADER International Workshop, September 2012 Simulation of LIFUS 5 tests

23. Temperature results obtained for the Test n.3 and n.4 of the IP-EUROTRANS campaign, with SIMMER III and IV code The predicted temperatures are affected by large high-frequency oscillations not detected in the experimental data Test n. 3 (middle thermocouple)Test n. 4 (middle thermocouple) 3 rd LEADER International Workshop, September 2012 Simulation of LIFUS 5 tests

24. SIMMER III computational models for the Test n.1 and n.2 of the ELSY campaign Detail of the computational 2D domain for the reaction vessel (S1) Test n.1Test n.2 Test n. 1 differs from Test n. 2 only for the injector device penetration, respectively, 5 mm and 250 mm 3 rd LEADER International Workshop, September 2012 Simulation of LIFUS 5 tests

25. Obtained results for the Test n. 1 and n. 2 of the ELSY campaign with SIMMER III and IV code SIMMER III succeeded in predicting the pressure peak in terms of value and timing, though slightly overestimating the first part of depressurization phase (from 0.75 to 1.5 s). On the other hand, the SIMMER IV code was found to anticipate the peak pressure SIMMER III code was found to overestimate the pressure peak (though corresponding in timing) and the first part of the depressurization phase (up to 1.5 s), whereas, SIMMER IV resulted once again to anticipate the peak pressure 3 rd LEADER International Workshop, September 2012 Simulation of LIFUS 5 tests Test n. 1Test n. 2

26. 3 rd LEADER International Workshop, September 2012 Simulation of LIFUS 5 tests

27. ParameterValue/Type Electric Power600 MW Thermal Efficiency42 % Primary CoolantPure Lead Primary SystemPool Type, Compact Primary Coolant Circulation Forced Primary System Pressure Drops 1.5 bar Primary Coolant Circulation for DHR Natural Inlet Core Temperature400°C Outlet Core Temperature 480°C FuelMOX and nitrates (with/without MA) Maximum Clad Temperature 550°C Reactor VesselInox and Austenitic Steel, H = 9 m Steam GeneratorN°8, inside the reactor vessel Primary PumpsN°8, mechanical, from hot collector InternalsRemovable Internal VesselCylindrical DHR immersion cooling container N°4, inside the cold collector ELSY reactor section Mean parameters of the ELSY reactor 3 rd LEADER International Workshop, September 2012 Simplified parametric analysis of the SGTR accident for LFR Spiral-tube SG Primary pump

28. Parameter Value Number8 LocationReactor vessel Function Heat removal and steam production TypeSpiral tubes Primary fluidLead Secondary fluidWater Heat transfer capacity187.5 MW Lead inlet temperature 480°C Lead outlet temperature 400°C Water inlet temperature 335°C Water outlet temperature 460°C Water inlet pressure190 bar Water outlet pressure180 bar Lead total flow rate130000 kg/s Lead flow velocity  2 m/s Water total flow rate 114.7 kg/s SG design requirementSteam Generator sections 3 rd LEADER International Workshop, September 2012 Simplified parametric analysis of the SGTR accident for LFR

29. Approximation of the SG region in the calculation domain The spatial discretization of the whole domain foresees 20 radial cells and 40 axial cells (SIMMER III) The domain reproduces the real SG dimensions, radius and height, and a larger expansion volume Calculation domains and simulation matrix 3 rd LEADER International Workshop, September 2012 Simplified parametric analysis of the SGTR accident for LFR

30. Series Injector Diameter [mm] Domain configurationSimulation A 18ReferenceA1A1 18Outer wall PorosityA2A2 18Venturi nozzle in the injectorA3A3 18Tube close to the injectorA4A4 18Grid in the SG upper plenumA5A5 B 24ReferenceB1B1 24Outer wall PorosityB2B2 24Venturi nozzle in the injectorB3B3 24Tube close to the injectorB4B4 24Grid in the SG upper plenumB5B5 Matrix of the simulation campaign The total orifice coefficient value inside the injector was properly taken equal to 4 RegionComponent Volume [m 3 ] (or Length) Temperature [°C ] Pressure [bar] I Water injector (1.76 m)335190 IISG lead10.74501-4 IIISG argon1.354501 IVEV argon25.24501 VEV lead2314001-6 3 rd LEADER International Workshop, September 2012 Simplified parametric analysis of the SGTR accident for LFR

31. Geometrical domains for each particular condition Domain configurationTest n. Outer wall Porosity2 Venturi-nozzle in the injector 3 Tubes close to the injector 4 Grid in the SG upper plenum 5 In the Test n. 2 a limited porosity is presented in the SG outer wall. It is due to a non perfect adherence of the two shells (main and companion) of the outer wall, when overpressure occurs during the SGTR accident In the Test n. 3 the Venturi nozzle is introduced into the water injector pipe to try to limit the water mass flow rate coming out In the Test n. 4 two non- calculation cell groups are taken into account, close to the injector opening, to simulate the SG tube bundle, which could reduce the water mass flow rate and the perturbation propagation In the Test n. 5 a grid is designed on the SG upper side, a line below the lead level, to simulate internal upper structures 3 rd LEADER International Workshop, September 2012 Simplified parametric analysis of the SGTR accident for LFR

32. Comparison among water mass flow rate calculated for different tests In the case A1, A2, A4 and A5 the water flow increases very quickly reaching the maximum value of nearly 20 kg/s around to 0.02 s. The Venturi-nozzle (A3 case) limits the mass flow rate to a much lower value of about 3 kg/s. In the Venturi orifice critical flow conditions are reached Obtained results 3 rd LEADER International Workshop, September 2012 Simplified parametric analysis of the SGTR accident for LFR

33. Pressure trend at the injector exit for all the A simulations Only a too limited margin of about 5 bar in the first peak value and 10 bar in the following evolution after 0.01 s are noted between the test A3 and the others Comparison among the pressure peaks of A simulations in the highest cell In the lead acceleration phase, liquid metal reaches the cover gas region, determining an initial argon pressurization and compression and finally hitting the upper SG wall at about 0.2 s Obtained results 3 rd LEADER International Workshop, September 2012 Simplified parametric analysis of the SGTR accident for LFR

34. Comparison between the lead kinetic energy of A and B series. Comparison among the lead kinetic energy of A simulations. The rapid effect of the SG isolation, due to the double shell wall system, leads to the upward lead motion The A3 peak is almost not present in respect to that of the other A simulations. This is the clear proof of the damping effects deriving from the Venturi nozzle introduction in water injector pipe Obtained results 3 rd LEADER International Workshop, September 2012 Simplified parametric analysis of the SGTR accident for LFR

35. Around to 0.9 s the cover gas becomes a mixture of vapor and argon and the compression work trend loses its meaning Obtained results 3 rd LEADER International Workshop, September 2012 Simplified parametric analysis of the SGTR accident for LFR Comparison between the argon compression work of A test series, due to the lead kinetic effect

36.  The main aim of this study was the analysis of the phenomena involved in CCI with particular attention to the SIMMER code qualification, in order to have the possibility of estimating the potential loads on LFR reactor structures due to SGTR  In addition to the analysis of the available LIFUS 5 experimental tests, the analysis of some relevant experimental tests coming from literature have been performed  The injection of a high pressure gas into a stagnant liquid pool, which characterizes the expansion phase of a hypothetical CDA in liquid metal cooled fast reactors, was investigated with SIMMER III and FLUENT codes through the experimental campaign SGI (KIT), obtaining good agreement with the experimental data  The simulation activity performed up to now in support to LIFUS 5 experiments has highlighted the capability of SIMMER code to reproduce quite well the thermal-fluid-dynamics phenomena involved in the interaction between water and heavy liquid metals (LBE)  As future work we will execute post-test analysis of the experimental tests that will be performed, inside THINS EU Project, in the LIFUS 5 facility with the new configuration (LIFUS5/Mod2) 3 rd LEADER International Workshop, September 2012 Conclusions

37.  A preliminary parametric analysis of the consequences for the SGTR accident in the LFR has been performed by SIMMER III code using a simplified domain to reproduce the SG  The presence of a Venturi nozzle in the injection line and the closure of a Safety Valve haven’t influences on the impulsive first pressure peak  The Venturi-nozzle inside the LFR SG pipes has, instead, a strong influence on the reduction in the mass flow rate going out from the broken pipe and consequently on the lead kinetic energy value, on the impulsive pressure peak on the top plate and on the cover gas “compression work”  The interaction between water and lead can be subdivided in three main phases: a first impulsive shock wave, a subsequent liquid metal kinetic energy increase which leads to have pressure peaks on the top plate wall and, lastly, a compression work increase in the cover gas 3 rd LEADER International Workshop, September 2012 Conclusions