1. 1 Interpretation of Melt Oxidation Observations in QUENCH-09 test M.S. Veshchunov * Nuclear Safety Institute (IBRAE) Russian Academy of Sciences * Visiting Scientist at FZK ( August-September 2005 ) 11 th International QUENCH Workshop FZK, Karlsruhe 25-27 October 2005

2. 2 Observations of melt oxidation in previous FZK tests (QUENCH, CORA and crucible tests)

3. 3 T=2200ºC Zr:O= 42:58Zr:O= 39:61Zr:O= 37:63 T=10 min.T=15 min.T=25 min. Zr melt oxidation in ZrO 2 crucible tests tests (J.Stuckert)

4. 4 Cross-section of QUENCH-03 test bundle at elevation 750 mm. “ Bulk” oxidation of melt. Zr-O melt oxidation in QUENCH tests

5. 5 Position 1Position 3 Ceramic phase (U,Zr)O 2  x VoidsMetal phase (U,Zr,O) Cross section of CORA-W2 test bundle U-Zr-O melt oxidation in CORA tests “Bulk” oxidation of melt

6. 6 Completely oxidised melt consisted of grown up bulk ceramic precipitates and the peripheral oxide crust, was formed in the AECL and FZK crucible tests on Zr melt interactions with ZrO 2 crucible walls during long-term (  1500 s) oxidation stage at high temperatures (2100- 2200  C) under a temperature gradient between the crucible walls and melt A close similarity of corium melt appearance is revealed in the bundle QUENCH and CORA tests, where bulk precipitation of ceramic particles up to complete conversion into ceramic phase was observed According to the present interpretation, ceramic structure of the central corium was formed under non-equilibrium test conditions in the course of melt oxidation by precipitation of ceramic phase in the oversaturated metallic melt along with the growth of peripheral oxide crust. Main conclusions from comparison of FZK bundle and crucible tests

7. 7 Corium melt oxidation model

8. 8. Fragment of quasi-binary phase diagramSpatial distribution of temperature and oxygen concentration SVECHA model for U-Zr-O melt oxidation under non-equilibrium conditions (1/2)

9. 9 Main findings:  Two processes of pellet dissolution and melt oxidation in the convectively stirred melt cannot be considered separately and should be modelled self-consistently.  Under non-equilibrium conditions, melt oxidation and UO 2 dissolution depend on temperature difference between solid and liquid phases and can proceed after attainment of the melt saturation, resulting in the ceramic phase precipitation in the bulk of the melt.  Depending on test conditions, the precipitation process can be accompanied with the peripheral oxide layer (crust) growth or dissolution.  The source of temperature gradients is oxidation heat at the oxide crust-melt interface and fission heat in the fuel pellet. Estimations show that these temperature drops can attain several tens K. SVECHA model for U-Zr-O melt oxidation under non-equilibrium conditions (2/2)

10. 10 Melt downward relocation in the bundle tests

11. 11. Progression of temperature front during melt relocation in CORA tests Axial mass distribution after the test and axial temperature distribution during the transient phase in CORA-W1 test Estimated velocity of relocation front:  1mm/s

12. 12. Progression of “flame” and “droplets/rivulets” fronts in CORA tests Analysis of on-line video inspections in CORA-5 test (W. Hering, thesis)

13. 13. Progression of temperature front during melt relocation in Q-09 test Temperature evolution at various elevations before quenching. Estimated velocity of relocation front:  1mm/s 1050 mm 850 mm 750 mm650 mm Time, s T, K

14. 14 Temperature front with T  2000ºC in CORA tests relocated downward with a characteristic velocity v 1  1-2 mm/s, which was extremely small in comparison with the characteristic velocities of metal rivulets and droplets (v 2  0.5 m/s) “Flamefront” in CORA tests relocated coherently either with a “droplet/rivulets front” or with Zr melting isotherm, i.e. fairly associated with the melt progression front A similar melt progression (v 1  1mm/s) apparently took place in QUENCH-09 test A new SVECHA model for melt oxidation/dissolution during relocation of a massive melt “slug” is currently under development (ISTC Project #2936) Main conclusions from CORA tests observations

15. 15 Analysis of melt oxidation in QUENCH-09 test

16. 16 extremely high temperatures during quenching Comparison of temperatures of cooling jacket by survived thermocouples in two tests Q-03 and Q-09 Special features of Q-09 test (1/3)

17. 17 extremely high hydrogen generation (~ 300 g) during quenching with increasing rate within ~ 100 s Hydrogen release (~ 16 g) during molten pool oxidation in FPT1 (~ 1000 s) Special features of Q-09 test (2/3)

18. 18 practically complete melt oxidation during quenching within ~ 100 s with formation of “foaming” ceramic structure Cross section at 590 mm bundle elevation FPT1 bundle cross-section Special features of Q-09 test (3/3)

19. 19 Microstructure of oxidised local molten pools Pool (ceramic): equiaxed fine grains Pellet : coarser grains Flow channel scale : columnar grains (from analysis of G. Schanz et al.)

20. 20 Comparison of melt oxidation in bundle tests Melt cross- section (diameter) Melt temperature Period of melt oxidation H 2 release (during melt oxidation) H 2 rate ( during melt oxidation) FPT 1  75 mm  2670 K 1000-3000 s (MP phase)  16 g decreasing Q-09  100 mm  2670 K  100 s (quenching)  300 g increasing Conclusions: The new model of molten pool (MP) oxidation based on the bulk precipitation mechanism (valid for interpretation of molten corium oxidation in FP tests) can explain post-test observations of melt microstructure in Q-09 test; However, additional mechanisms which can further enhance melt oxidation rate during quenching, should be considered for interpretation of Q-09 measurements.

21. 21 Void formation after relocation of melt Non-oxidised meltOxidised melt Melt dispersion Elevation 950 mm

22. 22 Oxidation of relocated melt

23. 23 Oxidation of dispersed melt (through open pores formed after melt relocation) Elevation 507 mm

24. 24 Elevation 590 mm Oxidation of dispersed melt (through channels formed after melt relocation)

25. 25 Conclusions from post-test observations (1/2) 1.Melt dispersion: In the course of slow relocation of molten “slug” (massive melt), parts of metallic melt relocate downward (droplets, rivulets) from the slug, leaving debris (oxide scales and precipitates ) and forming open voids and channels in the melt Compare with visual observations in the CORA tests of the “front of rivulets” relocated coherently with the “flamefront” (i.e. with the melt progression front) Typical size of voids and channels is comparable with the typical size of droplets and rivulets, that can be characterized by the capillary length of the corium melt:  Instability of melt progression front (local melting through the supporting crust and rapid downward relocation of droplets and rivulets)

26. 26 2. Oxidation of dispersed melt: a)relocated non-oxidised melt with “fresh” metallic surfaces is further attacked by steam; b)steam immediately penetrates into the formed open voids and channels and additionally attacks fresh internal surfaces of the non-relocated melt (forming oxide scale around the channels); majority of the channels and voids in the slug are open (filled with epoxy) ! Both processes provides stepwise and enhanced oxidation of dispersed melt. Conclusions from post-test observations (2/2)

27. 27 Small MP: incomplete pellet dissolution + complete precipitation Simulation of local MP oxidation (1/5)

28. 28 Larger MP: incomplete pellet dissolution + incomplete precipitation Simulation of local MP oxidation (2/5)

29. 29 Larger MP: complete pellet dissolution +  complete precipitation Simulation of local MP oxidation (3/5) High temperature scenario

30. 30 Oxidation of dispersed melt: - low temperature scenario Simulation of local MP oxidation (4/5) “effective” MP

31. 31 Oxidation of dispersed melt: - high temperature scenario Simulation of local MP oxidation (5/5)

32. 32 Conclusions The new model of molten pool oxidation based on the bulk precipitation mechanism, can explain post-test observations of melt microstructure in Q-09 test. In order to explain extremely high oxidation rates of metallic melt during high-temperature quenching observed in the Q-09 test, additional consideration of a strong dispersion (or fragmentation) of the downward relocating massive melt (slug) is proposed. Subsequent oxidation of relocated away “fresh” portions of melt (droplets and rivulets) and “internal” oxidation of remaining melt through the formed channels and voids, should be self-consistently considered. This complicated mechanism can be taken into consideration in the new slug relocation model (under development in the ISTC Project #2936). Preliminary calculations with the stand-alone model for melt oxidation in application to dispersed melt (forming small-sized local pools) allows qualitatively consistent interpretation of Q-09 test observations.