- Email: [email protected]
High pressure corium melt quenching tests in FARO
ltldear
".-~'~~ ' A l l ELSEVIER
Nuclear Engineering and Design 155 (:995) 253-270
High pressure cerium melt quenching tests in FARO D. Magallon, H. Holunann Commission of the European Communities, Joint Research Centre, Safety Technology Institute, 21020 lepta (Va). Italy
A~t
First-of-a-kind experimental data on the quenching of large masses of cerium melt of reafistic composition when poured mto pressunseu water at re.actor scale depths are presented and di~-'--tis.7~d.M e ~ , ~ ,uv,oav,,~: . . . . h...~ Io~vm,u--'~44 kg of a molten mixture 80 w% UO2-20 w% ZrO2, which were delivered by gravity through a nozzle of diameter 0.1 m to 1 m depth nearly saturated water a't 5.0 MPa. The objective was to gain early information on the melt/water quench process previous to tests that will involve larger masses of melt (150 kg of mixtures UO2-ZrO2-Zr). Particularly, pressures and temperatures were measured both in the gas phase and in the water. The results show that significant quenching occurred during the melt fall stage with 300/0 to 42% of the melt energy transferred to the water. About two-thirds of the melt broke up into particles of mean size of the order of 4.0 nun. The remaining one-third collected still molten in the debris catcher but did not produce any damage to the bottom plate. The maximum downward heat flux was 0.8 MW m 2. The maximum vessel overpressufisation, i.e. 1.8 MPa, was recorded with 44 kg of melt poured into 255 kg of water and a gas phase volume of 0.875 m 3. No steam explosions occurred.
1. Introduction
The general background o f the present F A R O L W R test series has been previously reported by Corradini and H o h m a n n (1993). It is summarised here. The reference situation is that o f a postulated core melt-down accident when jets o f molten cerium penetrate into the lower plenum water pool. This issue suffers a lack of d a t a on the water quenching potential that determines whether the thermal loading on the b o t t o m head structures is mitigated. In past analyses it had been assumed that this sequence o f events would have result in either settling o f most o f the fuel unquenched on the lower head with eventual RPV wall failure, or steam explosion. During the TMI-2 accident, al-
though 20 tons o I cerium did p o u r into the wa*.er lower plenum, neither o f these two possible scenarios actually occurred. Consequently, safety aspects o f the F C I issue needed to be reconsidered and it has been found o f fundamental importance to carry out tests involving large amounts o f prototypical curium poured into water at reactor scale depth, in order to characterise the melt/water mixing and quenching process, and melt/structure interaction. The J R C - ! s p r a F_ARO plant ( H o h m a n n . 1986) is used for such a purpose. It is a multi-purpose test facility in which severe accidents can be simulated out-of-pile under a variety o f conditions. Basically, a maximum quantity o f the o r d e r o f 150 kg o f UO2-ZrO2 fuel type melts (up to 3000 °C)
0029-5493[95/$09.S0 © 1995 Elsevier ScienceS.A. All rights reserved SSDI 0029=5493(94)00876=Z
254
D. Magalion, H. Hohmann{ Nuclear Engineeringand Design 155 (1995) 253-270
produced in the FARO furnace, possibly mixed with metallic components (e.g. Zr), and delivered to a test section of interest. The plant has been used previously for LMFBR safety problems such as melt relocation and molten fuel/ sodium interaction. The 1.5 m 3 test vessel TERMOS used for MFCI experiments can withstand 10 MPa at 300 °C, which makes it Imrtieularly suitable for simulating hil~h pressure accident scquenees. It can contain up to approximately 1 m 3 of coolant over a height of 2.5 m and a diameter of 0.71 m. The objective of the test series is to determine: (1) the melt quenching rate associated with the melt/water penetration, (2) the hydrogen production associated with the zirconium oxidation, (3) the thermal load on the bottom structures, and (4) to characterise the debris structure. It must be noted that these tests are not steam explosion experiments and, consequently, are not intended to produce data for this issue. The first part of the test matrix, which deals with high pressure and nearly saturation conditions, further limit the risk of a steam explosion and allow to concentrate on the quenching phenomena. Nevertheless, the possibility of occurrence of a steam explosion has been taken into account and the instrumentation chosen accordingly (fast pressure transducers in the water). In the absence of previous work under the conditions mentioned above, it had been decided to proceed by step towards the target corium quantity of 150 kg. Two tests has been performed at constant volume with a pure oxide mixture 80 w% UO2-20 w% ZrO2, which was delivered nearly by gravity through a nozzle of diameter 0.1 m to a 1 m depth saturated water pool at 5.0 MPa. The melt quantity, the water mass, the vessel diameter were ! 8 kg, 120 kg, 0.47 m and 44 kg, 255 kg, 0.71 m for the first and the second t,~st, respectively. The paper summarises and discusses the findings from these experiments. A simple assessment of the energy transferred to the water during the melt fall stage is made. The heat fluxes through the debris catcher bottom plate are calculated from the thermocouple data. The most relevant and recent experimental study on the break-up and quenching of cerium melts in water can be
is the CCM series (Wang, 1989). Particularly, the CCM-5 and CCM-6 tests dealt with approximately 13 kg of a 60 w% UO2-16 w% ZRO2-24 w% steel mixture poured into 510 kg of 45 °Csubcooled and saturated water at 0.1 MPa through a nozzle of diameter 0.051 m. Results from these tests are briefly compared with ours.
2. Experiment description 2.1. Test arrangement and procedure
The experimental arrangement is shown in Fig. 1. The interaction vessel TERMOS (10 MPa, 300 °C, main dimensions in Fig. 1) is connected to the FARO furnace via the release tube, the intersection/isolation valve unit and the release vessel. The cerium is melted at low pressure (i.e. 0.1-0.2 MPa), while the pressure in TERMOS is as required by the test (e.g. 5 MPa). The release vessel closing flap is the barrier between the low pressure and the high pressure regions. Nevertheless, the main isolation valve SO2 remains closed during cerium melting for the sake of safety. The temperature in TERMOS up to the SO2 valve corresponds to the saturation conditions of TERMOS (e.g. 263 °C). At the end of the cerium melting phase, the SO2 is opened, the melt released from the furnace to i~" catcher, and the SO2 closed again. The release vessel volume is used as a lock-chamber for pressure equalisation. This is obtained by bursting a double disc mounted on the communication line, which acts as a quick opening valve. Upon pressure balancing, the release vessel hinged-flap automatically opens, and the melt is delivered by gravity to the water. 2.2. T E R M O S test section 2.2.1. Scoping test ( S T )
An internal water stainless steel container of diameter 0.47 m was included, separated from the pressure vessel by A1203 bricks for thermal insulation. From the upper flange of the pressure vessel to the lower flange of the SO2 valve, a thermal insulator was installed externally. The heating of
D. Magallon, H. Hokmann / Nuclear Engiaeering and Design 155 (1995) 253-270
FARO fumoce
255
1 Lower electrode Release tube Closing Disc (W) 71,72 detectom Release tube
-I-NT~ -
Mirror system drive 7 W-Disc | VideocamSurveillanceJ
Protection valve S01
---'-t
Main intersection valve S02 (¢=120 mm) Communication I~ne (~--40 mm) Double rapture disc .~,o_fe~ valve (IOMPa) Release vessel (volume up to 502 = 0.028 m 3) Hinged-flap
(ergo=,= 100 ram) TERMOSve~l 710 mm Water Funnel Bottom plate (debris catcher) =
/
Supports
Fig. 1. Quenchingtest-2 arrangement. the space inside the water container was provided by means of four heater rods of total power 90 kW. The space above the upper flange of the
pressure vessel up to the lower flange of the SO2 was heated by means of trace-heaters (21 kW) put on the outer surface of the components. The
256
D. Magallon, H. Hohmann / Nuclear Engineering and Design 155 0995) 253-270
debris catcher was supported by three adjustable legs put on the bottom of the water container. The debris catcher bottom plate had a thickness of 0.04 m. Its upper face was distant 0.135 m from the bottom of the container. The catcher was provided with trace-heaters (3.5 kW) attaebed on its lower face. 2 . 2 2 Quenching test 2 (QT2) For this test the internal vessel of diameter 0.47 m and the A1203 bricks were removed, thus increasing the diameter of the water pool to the diameter of the TERMOS vessel, i.e. 0.71 m, as indicated in Fig. 1. To account for the increase of steel mass to be heated with respect to the scoping test (the pressure vessel itself 0.045 m thick, i.e. about 7 tons), additional trace-heaters for a total power of 17.5 kW were fixed on the lower part of the TERMOS outer wall. The vessel was thermally insulated from outside. The same debris catcher as for the scoping test was used with a funnel to account for the change in diameter of the test section. 2.3. Test instrumentation The principal quantities measured during the corium quenching were pressures and temperatures both in the gas region and in the water, temperatures in the debris catcher bottom plate and water level swell. The distribution and types of the probes reported in Figs. 2 and 3 correspond to the quenching test-2 configuration. They were essentially the same for the scoping test, except for level swell measurement devices, which were not present. Four KELLER pressure transducers (piezoresistive, 5 kHz frequency response) measured the vessel pressurisation. They were installed at the end of 0.012 m diameter, 0,4 m long straight tubes emerging in the gas phase region at the levels reported in Fig. 2, and water cooled. Prior to the test, the uppermost ones were used to monitor the pressure in the release vessel and in the chamber between the double disc, respectively. Four VIBRO-METERs (piezoelectric, 15 kHz frequency response) were located in the water as indicated in
Fig. 2 for rapid transient records in case of an FCI. They were protected by stainless steel grids. The eight gas phase K-thermocouples were installed in such a way that they could not 'see' a centred melt jet. Fourteen thermoeouples were fixed on the structures by means of clamps. They were fixed to the internal vessel for ST and to the pressure vessel for QT2 at the same axial positions. The 25 water Kqhermocouples were essentially sacrificial thermocouples used to determine the downward progression and radial extension of the melt jet. Those not destroyed during melt penetration recorded the long time water temperature history. The water centreline thermocouples were attached on thin (0.2 mm) stainless steel wires crossing the test section. The opening of the release vessel was indicated by the rupture of a 0.5 mm K-thermocouple (OD1) mounted opposite to the hinge, and fixed both to the release vessel lower flange and to the flap. Another 0.5 mm K-thermocouple (OD2) was placed on the centreline of the vessel, 0.25 m below the lower face of the flap, for detecting the passage of the melt. The four resistance probes for measuring the level swell in QT2 were installed 0.25 m, 0.50 m, 0.75 m and 1.00 m above the initial water level, respectively. An experimental probe for a continuous measurement of the level swell using a time domain reflectrometry method was also installed at the same radial position as the resistance probe, but 150° apart.
3. Experimental conditions The conditions are summarised in Table 1 for both tests. The interacting melt mass corresponds to the quantity of debris found in the debris catcher. The melt, water and gas volumes were roughly twice higher in QT2 than in ST (actually, 2.4, 2.1 and 1.9 higher, respectively), The melt temperature is inferred from ultrasonic temperature sensor measurements performed in previous similar melting and release tests in FARO. An attempt for measuring the temperature of the melt in the release vessel was made in QT2, but was unsuccessful. The delivery times have been deduced from the OD1 and OD2 signals as follows.
D. Magallon, H. Hohma~ I Nuclear Engietceringand Design 155 0995) 253-270
25?
4
3150 - -
2000
--
Continuous LevelIno~cotor 2100
2135
- -
4 C0nt0ctProbes
- - *
Thermocouples • ~ 1.6 turn "~ 0.6 mm 1600--
-
-
1635
!135 - -
~ =
1050--
-
-
980
-
-
7~0
4 YtBR0-g~.Rs
550-350-135--
Fig. 2. Quenching test-2 instrum©ntation(all dimensionsare in mm).
The ODI rupture was taken as the melt release start (time 0). The O1:)2 thermocouple was not destroyed by the melt (in both tests) even though it indicated temperature .jumps above 1400 °C. One reason for that was the off-centring of the melt jet induced by the hinged flap. The time corresponding to the start of temperature decrease after the maximum has been taken as the end of
the release vessel discharge, thus giving the delivery time. As noted in Table 1, the durations (280 ms and 370 ms, respectively, for ST and QT2) are higher than the theoretical values (110 ms and 235 ms, respectively, for ST and QT2) due to crust formation in the delivery tube and start-end of release disturbances. The relative Iarger difference noted for ST is explained by the formation of an
D. Magallon, H. Hohmann 1 Nuclear Engineering and Design 155 (i£195) 253-270
258
!
?
I I
< ................................................................................. ..4.L~. ................................................................................ •
15
1
...175
/
i
)
•
i
) l
•":L-..................... \
t l ~ :' e r m o c o u p l e s
3x2
^ i
at
1200 i,m
3
;14 18
~
~ . |
~
~ ==: a x 5,~ m
*
..........i ...... o° ........... ~,--
o
15
/
I
.135 ....9 5
supports
~hermocouples lhermocouples
$ 0.6 mm $ t , 0 mm
Fig. 3. Debris catcher instrumentation (all dimensions are in mm). Table 1 Summary of experimental conditions Seeping Test
Quenching Test 2
Melt Composition (w%) Mass (kg) Temperature (oC) Delivery nozzle (ram) Delivery time (s) Flow rate (kg s -t) Free fall in gas (m)
80 UO2 + 20 ZrOz 18 2650 100 0.280 64 1.66
80 UO2 + 20 ZrO2 44 2750 100 0.370 119 !.53
Waler Mass (kg) Depth (m) Temperature (°C) Subcooling at melt contact (°C) Fuel to coolant mass ratio
120 0.87 266 (230 bottom plate) 2 (38 bottom plate) 0.15
255 i.00 263 (255 bottom plate) 12 (20 bottom plate) 0.17
83 steam + 17 Ar 0.464 270
70 steam + 30 Ar 0.875 263
0.470 0,~,0 5.0 5,4
0.710 1.3 5.g 6.1
Gas phase Composition (w%) Volume (m3) Temperature at ODI rupture (release start) (°C) Test vessel Diameter (m) Overall volume (m 3) Pressure at ODI rupture (MPa) Pressure at melt/water contact (MPa)
D. Magallon, H. Holnnann I Nuclear Engineering and Design 155 (1995) 253-270
abnormally thick crust in the nozzle (5 ram) and above the melt (about 10 mm), probably favoured by the small quantity and the poor overheating. Melt streams rather than a coherent mass release is likely for this test. The vessel pressure, and water temperature and level indicated in Table ! depended on the TERMOS heating and pressure equalisation procedures, which were different for each test and are detailed in the next two sub-sections.
259
5.0 MPa in the same way as for the scoping test. To suppress boiling of the water during the pressure equalisation between TERMOS and the release vessel, argon was blown into TERMOS just before bursting the communication line. f u t u r e disc. Due to this argon addition, the pressure at the time the melt started flowing down was 5.8 MPa. It increased to 6.1 MPa as the melt crossed the gas phase. Consequently, the suheoolJng of the water at melt/water contact was 12 °C in the bulk against 20 °C at the bottom plate level.
3.1. Scoping test
The pressure in TERMOS at 80 °C was 0.4 MPa (argon). Heating was pursued up to reaching a water temperature of 266 °C and a vessel pressure of 5.3 MPa without further addition of argon (pressurisation due to steam generation only in boiling regime). Then, the heating power was reduced just to maintain these conditions. This residual power was switched off a few seeonds before the melt to water release. The water height counted from the debris catcher bottom plate upper face was 0.87 m. Upon bursting the communication line rupture disc, the pressures in both TERMOS and the release vessel equalised to 5.0 MPa. This little blow-down of TERMOS enhanced water boiling and induced a level swell. The temperature of the water did not exhibit significant changes. While the melt crossed the gas space, the pressure increased up to reaching approximately 5.4 MPa when the melt leading edge arrived near the water surface (contact time deduced from thermocouples), As a consequence, boiling should have completely ceased at that time, and the water resumed its original depth of 0.87 m. Note that water was about 38 °C subcooled near to the debris catcher (referr,~ to 5.4 MPa) because of an unaware switch-off of the power of the catcher heater. The decrease from 266 °C to 230 °C was localised within 0.25 m above the debris catcher bottom plate. 3.2. Quenching test 2
The pressure in TERMOS was 0.2 MPa at 70 °C (essentially argon). The heating of the test section was pursued up to reaching 263 °C and
4. Experimental remits and
The experimental curves presented in this section are also used as a basis for discussing the timings of the melt fall stages and for suggesting some interpretations of the facts that occun~cl. In order to avoid any ambiguity, each sub-~:tion starts with a short introduction presenting just the related experimental curves. The different times indicated in the figures and in the data summary in Table 2 were deduced from the thermocouple signals. This method does lead to some uncertainties. In particular, one does not know exactly whether the bulk leading edge of the melt or some preceding lumps contacted the probe first. For this reason, the data deduced from the thermocouples form part of the discussions. For all the data reported, time zero corresponds to the OD! rupture (release vessel hinged-flap opening, i.e. start of melt release). 4.1. Melt fall stage 4.1.1. Experimental data
The gas phase pressure traces corresponding to the first 1.4 s of both tests are reported in Fig. 4. The values are normalised to the value at time zero, i.e. 5.0 MPa for ST and 5.8 MPa for QT2. The water pressure transducers (Rot reported) gave exactly the same signals as the gas phase transducers, except the classical late drift due to heating. In particular, they did not ~.ndicate fast dynamic pressurisations typical of energetic events (steam explosions).
260
D. Magallon, H. Hohmann I Nuclear Engineering and Design 155 (1995) 253-270
Table 2 Summary of experimental results (time t = 0 is release start)
Melt Mean velocity in g~-. phase (m s-z) Mean velocity in water ( m s -I) Broken up (kg) Conglomerate1 on bottom plate (kg) Mean size of fragments (ram) Melt/debris rejection Bottom plate Maximum temperature increase (°C) Maximum downward heat flux (MW m -2) State Pressure increase Short-term maximum from release start (MPa) Short-term maximum from melt/water contact (MPa) Long-term maximum (MPa) Steam explosion Temperature increase Steam (maximum measured) (°C) Steam (mean value at t = I0 s) (°C) Water (°C) Maximum levd swell Continuous probe (mm) Resistance probes (150 ° apart from continuous probe) (ram) hfferred from thermocouples (mm)
Scoping test
Quenching test 2
4 2.3 12 6 4.5 No
5 3.7 3O 14 3.8 No
Intact
275 (contact face) 0,8 Intact
I.I (at t = 1,2 s) 0.7
!.8 (at t = !.2 s) 1.5
1.6 (at t = 12 s) No
1.8 (at t = 20 s) No
86 ~43 15 (max. at t = 12 s)
83 ~30 23 Cat t = 25 s)" ~450 Cat t = 1.15 s) >250, <500 (at t = 1.45 s)
> 130
~Maximum not reached at that time.
2.0 1.751,5-
~ 1.25-
g. t.uI
0.0
0,2
0.4
0.6 0.8 ti~ [=]
1.0
1.2
1.4
Fig. 4. Early vessel pressure increase (ST, scoping test; QT2, quenching lest 2, t = 0: release start; MWC, melt water contact; MBC, melt bottom contact; AMIW, all melt in water; E O B . end of break-up).
Fig. 5 shows the signal from the experimental continuous level-metre and the analogical signal from the resistance probe located 0.25 m above the initial water free surface (labelled cp1385) corresponding to QT2 (no data are available for ST, except that inferred from thermocouples and reported in Table 2). The other contact probes (0.5 m and above) did not move from rest. The resistance probe indicates that the level swell definitely ~eached 0.25 m at time 1.35 s (and remained above this value up to time 1.6-1.7 s), while the continuous probe indicates that the value of 0.25 m was already reached at time 0.85 s. According to the different angular positions of the two probes (they were installed 150° apart), the differences observed may indicate that the level swell was not uniform, probably influenced
D. Magallon, H. Hohmann [ Nuclear Engineering and Design 155 0995) 253-270
261
[mm] ~0-
_A/k_
~400-
oLMCP 1365
~_ J
~i''r"
_!
o 300--I, u.I
,,-I, ~ o -
100-
'
,,~,~,_~~f ~1~
/_..
O-
0.0
0.5
1.0 TIME [ o ]
1.5
2.0
Fig. 5. Level swell in QT2 (CLM, continuous level indicator; CPI385, contact probe located 250 ram above the initial water level).
by the off-centring of the melt jet. The continuous level-metre indicates that a maximum of around 0.45 m was reached at time 1.15 s. The two signals returned to be in phase while the water level moved dowlt after 1.15 s. The early gas phase temperature increases are presented in Fig. 6 for QT2. They were very similar for ST (see Fig. 8). A significant non-homogenous heating of the gas phase volume is noticed with maximum temperature increases around 80 °C. 4.1.2. Discussion
200-
0.0
~ ,~ o%=~:~-:,=
0.2
0.4.
0,6
0,8
1.0
1.2
1.4-
[s]
From the water thermocouple signals, only a broad picture of the melt penetration history could be established. Mainly the times of melt/water impact and melt arrival on the bottom plate could be determined as reported in Figs. 4 and 6. For ST, the water thermocouples appeared contacted by the melt in a rather stochastic way as the result of a non-coherent release. For QT2 it
THE Fig. 6, QT2earlystearatemperatureincrease(for legend,see Fig. 4).
could be established that only an off-centring effect from the hinged-flap was present. The melt flowed down laterally but coherently, probably impacted first the debris catcher funnel and then
262
D. Magallon, H. Hohmann I Nuclear Engineering and Design 155 (19,35)253-270
spread on the bottom plate from one side. In both eases, only few water thermocouples were damaged. The curves of Fig. 4 indicate that the break-up process was essentially the same for both tests. Before penetrating into the water, the melt crossed the gas space (height 1.66 m for ST and 1.53 m for QT2). Pressure increases of 0.4 MPa and 0.33 MPa are observed before the melt/water contact (MWC in Fig. 4) for ST and QT2, respectively. If only gas heating by the falling melt were the cause of these pressure increases, the temperature of the gas phase should have increased by 41 °C for ST and 36 °C for QT2, respectively (considering that superheated steam behaves as a perfeet gas). Clearly, the gas phase temperature traces reported in Fig. 6 do not indicate such an increase for the time period of interest. One reason could be that classical thermocouples can hardly measure transient steam superheating. Another reason could be that steam heating was more concentrated around the melt path. Steam generated by radiative heat transfer from the melt may be also invoked. For ST, steam generation induced by the 'disequilibrium' created at the time of pressure equalisation between TERMOS and the release vessel nevertheless occurred (see equalisation procedure descriptions in the previous section), and enhanced the phenomena. The pressure increase after melt/water contact (MWC) was 0.7 MPa for ST and 1.5 MPa for QT2, but only beyond the trailing melt penetration time (AMIW time in the figures) are the increases resulting from melt break-up alone. The end of break-up time (EOB) reported in the figures means that the unbroken part of the trailing edge should have reached the bottom plate at that time. It has been calculated by adding the delivery time to the time at which the leading edge touched the bottom (MBC), thus assuming that the break-up process continued for all the time the melt descended through the water. The rate of pressure increase curves are plotted in Fig. 7. These curves have been obtained by differentiating the spline smoothed pressure histolies. It is seen that the slope became negative very soon after melt/water contact (taken as the reference time in that figure), which seems not to be
[~,/s]
3.5
°
3.0-
~
2.5-
o
2.0-
'¢
I~
•
ZI ~
ST
1
, 0.4
~ n,6
t,00.50.0 -0.5. -0.4 -0.2
0.0
~ 0.2
~ 0.8
1.0
1.2
t - t¢
Fig. 7. Pressure increaserates during the melt fall stage (to, contact time). consistent with the start of quenching. One explanation, consistent with a possible previous heating of the gas phase by the melt, is that steam produced at the beginning of the melt/water interaction was colder than steam already present in the gas phase, which partially condensed. For ST, the rate of pressure increase never came up again to the value it had just before the melt/water contact. The fluctuations noticed within the break-up window can be attributed to an irregular melt penetration. 4.1.3. Estimates o f the energy released by the melt to the water
An estimate of the amount of steam added to the system as a result of the melt falling through the water has been made, assuming that the thermocouples of the gas phase measured the actual temperature of the gas whatever the steam superheating might have been. The quantities of steam present in the free-board volume at the time of melt/water contact and at the time of the maximum pressurisation have been calculated by using the specific volumes of the steam at the partial pressures of the steam and averaged temperatures of the gas phase: i.e. 4.8 MPa, 277 °C and 5.5 MPa, 303 °C for ST; 4.9 MPa, 265 °C and 6.4 MPa,
D. Maffallon, H. Holumnn I Nuclear Engineeringand Design 155 (1995)253-270 298 °C for QT2). The difference between the two quantities has been taken as the quantity of steam produced during the melt/water interaction. One finds 0.8 kg for ST and 5 kg for QT2, which corresponds to an energy of 3 MJ and 16 MJ, respectively. The time difference between the maximum pressurisation and the melt/water contact is very similar for both tests (see Fig. 7) and is approximately 0.9 s. Using the above quantifies of steam produced, mean generation rates of 0.9 kg s - t and 5.6 kg s - t are found for ST and QT2, respectively. A maximum value of the steam generation rate can be calculated for QT2 by multiplying the peak v~lue ofdp/d: hi Fig. 7 by the quantity bfV/R,~T~, where M is the steam molect,lar weight, V ~tte gas phase volume, Ts the gas phase temperature, and Ru the universal gas constant. One finds a peak steam generation rate of 11 kg s-J for T8 = 300°C. Varying Tg from 263 °C (initial gas phase temperature) to 345 °C (maximum measured) does not change very much the above value: one goes from 10.2 kg s - ' for 263 °C to
[ deg. c ]
263
11.8 kg s- ~ for 345 °C. One should further introduce a compressibifity factor for taking into account that steam does not behave as a perfect gas at the pressures and temperatures of concern here. Taking a value of 0.8 for this factor, the above estimates increa.~ by 20%. The energy required to increase the temperature of the liquid water during the melt fall from 264 °C to 270 °C in ST and from 263 °C to 277 °C in QT2 (see Figs. 10 and 11) is 5 MJ for ST and 12 MI for QT2, respectively. Comequenfly, the total energy released to the wat~ daring the .melt through water fail stage is estimated to be 3 MI for ST and 28 MI for QT2, respectively. 4.2. Longer term stage 4.2.1. Experimental data The gas phase temperature distribution over the first 10 s are plotted in Fig. 8 for ST. A trend to temperature equalisation is noticed as time increases. The pressure histories over a period of
5
1 320-
270
o
I
25O 0
1
2
3
4
5 [~]
6
7
8
9
T~AE
Fig. 8. Scopingtest steam temperaturehistoriesfor the first 10 s after releasestart.
10
264
D. Magallon, H. Hohmann / Nuclear Engineering and Design 155 (1995) 253-270 ff,4Pa]
tures are plotted together with the saturation temperature of the water corresponding to the total pressure. The averages have been calculated using all the thermocouples of the gas phase (8) and only those in water which did not exhibit strong disturbances during the passage of the melt (14 for ST and 13 for QT2). For QT2, two gas phase thermocouples, probably touched by water droplets due to condensation, showed a sharp decrease between 4 and 8 s. This explains the average temperature perturbations noti~d in Fig. 11 during this period of time.
- PT 03 150 3150 -
8.0
,~P--- I
7.5
"
7.0
II
o..
SCOPING
TEST
6.0:: 5.5 5,0 O
5
10
15
20
4.2.2. Discussion
25
[,,l
As can be seen in Fig. 8, the gas phase temperature is largely not uniform, especially soon after the end of the jet fall. It is believed that this map represents the reality of the steam generation and mixing with the cover gas. When quiet steaming from the debris bed dominates, temperatures tend to equalise. Figs. 9-11 indicate that similar vaporisation/ condensation scenarios developed at two different
TI~E
Fig. 9. Long-term pressure histories.
time of 25 s are reported in Fig. 9. The long-term maximum pressure was reached after approximately 12 s for ST and 22 s for QT2, and was 6.6 MPa and 7.6 MPa, ,espectively. In Figs. 10 and 1l, the averaged gas phase and water tempera320
E
310-
o "0
tll r1"
290Q: W
~ 280W p. 270
-
260 0
5
10
15
20
25
TIME [ s ]
Fig. i0. Scoping test steam and water temperatures compared with the saturation temperature.
D. Magallon, 1t. Hohmann [ Nuclear EngMeering and Design 155 (1995) 253-270
26~
310
,q
A 0
GAS PHASE / ~
~
+-~
| LU
.--2 • T.ttP)
w Lu n
!
J
LU I-
270-
0
5
t 10
i
+
15
20
25
TIME [ S ] Fig. 1 I. Quenching test 2 steam and water temperatures compared with the saturation temperature.
time scales for both tests: the melt break-up time scale and the long-term debris cooling time scale. During melt break-up, the production of steam was very fast and intense, inducing a fast pressurisation of the vessel up to a level that could be higher than the saturation pressure corresponding to the gas phase temperature (see Fig. I1 for QT2). As a consequence, at the end of the breakup phase, condcnsation dominated steam production fi'om the debris cooling (mainly steam to steam condensation). The higher the steam generation during the melt fall quenching stage, the higher the pressure decrease at the transition phase. Then, the contrary occurred up to the second maximum. Later on, condensation definitively dominated. Previous to the second pressure maximum, steam generation due to debris cooling is evidenced by the fact that the gas phase temperature decreased as the pressure increased. Boiling in the debris region is further evidenced by the wiggles on the signals of the water thermocouples located just above the debris (s¢¢ Fig. 12). When comparing Figs. 10 and 11 one remarks that the temperatures are of the same order for
both t¢sts. This is not so surprising considering that the melt to coolant mass ratio was very similar for both tests and that homogenisation rapidly occurred after the melt fall stages (within the first 5 s in the water). Proportionally to the melt quantity, more steam was produced in QT2 during the melt fall stage than in ST (see Section 4.1.2), inducing a higher pressurisation and a subcooling of the water. As a consequence, part of the superheated steam produced in the debris catcher region condensed while rising through the water, increasing the water temperature. 4.3. Debris catcher data The debris structure was very similar for both tests. Fig. 13 shows a view of the ST debris surface as found in the catcher. Fig. 14 shows a view of a part of the conglomerate in con*act with the bottom plate. This part was certainly still molten when it contacted the plate. As can be seen, the plate did not suffer any damage. This is not surprising using a pure oxide melt becausv we knew already, from the BLOKKER series per-
D. Magallon, H. Hohmann / Nuclear Engineerir,g and Design 155 (1995) 253-270
266
290
280-
270 -
ILl
260-
::3 t-.<
11: 2 5 0 tu O.
240-
230-
220 (~
I
I
I
I
5
10
15
20
25
TIME [ S ]
Fig. 12. Seeping test water temperature just above the debris (at level 175 mm).
Fig. 13. View of the ST debris bed from above.
Fig. 14. View of the melt spread on the bottom plate in ST.
formed in F A R O in the frame of the L M F B R programme (Hohmann, 1989; Magallon, 1990), that jets of 100 kg of pure UO2 around 3000 °C interacting in dry conditions with 0.04-m thick plates preheated to 400 °C did not induce any erosion even though the plates reached 1000 °C. Some samples of the ST debris are shown in Fig. 15. They were very irregular in shape and aspect. Only a few spheres such as that in the
figure were present. Contrarily to pure UO:, these particles were very brittle and great care had to be taken in manipulating them in order not to alter the analysis. In general, particles up to a dimension o f the order of 10 mm presented a fine granular internal structure, normally a sign ot rapid quenching. Larger particles (a few particles of some centimctres were part of the debris) presented an internal structure such as that shown in
D. Magallon, H. Hohmann I Nuclear Engineering and Design 155 (1995) 253-270
Fig. 15. ST debris samples.
Fig. 16. View of the internal part of a 4 em size particle of the ST debris.
Fig. 16. The central grain growth structure is typical of slow quenching. The particle size distributions are given in Fig. 17 and compared with that of CCM-5 and CCM6. "lhe difference in the mean particle size between ST and QT2 can be explained by the higher penetration velocity observed in QT2. The mean particle sizes of both FARO tests lie in between those of CCM, which agrees with the fact that the degree of subeooling of the water in FARO was in between that of CCM tests. It should be noted that in neither of the two CCM tests of concern, were signs of a re-agglomerated molten cerium evidenced. This happened only in CCM-2 where the water depth was 0.64 m instead of 1 m. Proportionally to the quantity of melt which experienced break-up (about 12 kg in ST and 30 kg in QT2), the energy corresponding to water
267
heating during the melt fall phase, as calculated in Section 4.1.3 (i.e. 5 MJ for ST and 12 M,I for QT2), is comparable with the CCM values (Wang, 1989), where 10 ,'eLI in subcooled water (CCM-5) and 3.5 MJ in ~::e:ated wamr (CCM-6) for 13 kg of melt were calculated. Also, the total energy transferred to the steam/water system during melt fall stage, which represents 44% of the energy content of the broken part of the debris for ST and 62% for QT2, globally corresponds to the CCM-5 (55%) and CCM-6 (36%) estimates. Most of the thermocouples located just above the surface of the bottom plate were destroyed during melt impact. 'Contact temperatures" (in fact the thermocouples were welded 0.5 mm below the contact face) and temperatures 5 mm below the bottom plate surface are shown in Fig. 18 for different radial locations (QT2 test). They reflect the fact that the melt did not spread uniformly on the plate. Downward heat fluxes calculated from these temperatures are reported in Fig. 19. A maximum value of 0.8 MW m-2 is obtained in the central zone of the plate. The time integrals of curves of Fig. 19 give the energy released to the plate (Fig. 20), which reached a maximum around 9 MJ m-2 at time 20 s in the centre of the plate. Maximum values around 2 MJ m -2 and 4 MJ m -2 are reached at 9 s (r = 80ram) and 12 s (r = 160ram), respectively. Negative downward heat flux and, accordingly, decrease in energy transfer are calculated for times beyond 9 s (r = 80ram) and 12 s (r = 160mm) as a consequence of the inversion of the axial temperature gradient noted in Fig. 18. This indicates that radial heat transfer through the plate in view of temperature homogenisation became dominant with respect to heat transfer from the debris.
5. Concluding remarks The essential objectives of the two first tests of the FARO quenching series have been achieved. Estimates of the melt quer,,,hing rates and of the thermal loads on the bottom structures, and characterisation of the debris have been made. It has been found that the melt lost more than 50% of its energy during the fall through water stage. De-
268
D. Magallon, H. Hohmann / Nuclear Engineering and Design 155 (1995) 253-270 10.0
I
I
I
I
I
I
I
I
i
I
I
i
I
I
I
I
i
IrlOI IO in A
@0
E E
[]
lZl
,jj N LLI ,_1
Io
1.0
[]
o_
@o
•
El rl
mean particlesize
Io
a.
iD I
[]
[] Q
@
SCOPINGTEST
4.5 mm
O
QUENCHINGTEST2
3.8
i
CCM-5(REF.3) CCM*6(REF.3)
5.0 3.5
[] 0.1 0.01
0.1
1-2
5 10
20 3'o'40506070 ''' 80
9'o ;s
9899 ''
99.9 '
)9.99
PERCENT MASS LESS THAN INDICATED SIZE Fig. 1% Particle size distributions (total debris for CCM, particulate debris, i.e. two-thirds of total, for ST and QT2).
5,°1
5°°1
-°"''
1
g'°° tl \ 300
s/'z" " ~
0
5
1
15
t~
20
,25
[s]
Fig. 18. Melt/bottomplate contacttemperaturesand temperatures in the plate 5 mm belowthe contact surface. spite the fact that about 30% of the cerium collected still molten on the bottom plate where the water was up to 38 °C subcooled, no steam explosions occurred and the plate remained intact.
Maximum downward heat fluxes of the order of 0.8 MW m -2 have been calculated from the there mocouple data. Some problems arose when trying to assess the steam generation rate from the pressure and temperature data. They were essentially due to the heating of the gas phase by the melt itself previous to the melt/water interaction, and to the difficulty in measuring transient steam overheating by using classical thermocouples. On the other hand, one can hardly eliminate these problems working with high pressure and necessarily limited volumes. Consequently, it is important for the interpretation of the experiments by the mixing codes and for the validation of these codes, that they include good models for the melt fall through and heat transfer to the gas phase previous to and during the melt/water interaction. With respect to past quenching experiments using smaller quantities of thermite generated melt at ambient pressure, no fundamental differences were found, but some correlation between
D. Magailon, H. Hohmann / Nuclear Engineering and Design 155 (1995) 253-270
269
1.10 =-
,g, X I< Ill "r
[
~
o
5
-2.10 =
J
~
J
10
15
20
.,
j 25
t~me Is]
Fig. 19. Downward heat fluxesat different radiat locations.
1,10 ~-
centre 8 , 1 0 =-
6
6,106-
=~ 4 , 1 0 = -
2,10 =-
0,100
160 mm
r
=
r
= 80ram
I
I
I
I
I
5
10
15
20
25
tim Is) Fig, 20. E n e r g y t r a n s f e r r e d
test pressure and steam generation on the one hand, and water depth and molten mass that reached the bottom plate on the other hand, could be further evidenced. This justifies pursuing the study using larger amounts of melt and larger water pool depths. Also, the influence of the
to
the b o t t o m plate.
presence in the melt ~f a percentage of a metalfic compound has to be evaluated. Tests involving 150 kg of UO2-ZrO2-Zr melts is in progress at JRC-lspra. The boundary conditions for these tests will be as close as possible to that of the preliminary tests reported in this paper
270
D. Magallon, H. Hohmaan / Nuclear Engineering and Design 155 (1995) 253-270
(closed volume), in o r d e r to facilitate comparisons and c o d e validation.
Acknowledgements "~,ais w o r k was p e r f o r m e d in collaboration with the U S N u c l e a r Regulatory C o m m i s s i o n in the frame o f a Technical Exchange Agreement. The authors greatly acknowledge the w o r k and efforts o f the whole F A R O experimental team.
References M,L. Corradini and H. Hohmann, Multi-phase flow aspects of fuel-coolant interactions in reactor safety research, OECD/ CSNI, Nucl. Eng. Des. 145 (1993) 207-215.
H. Hohmann, D. Magallon, H. $chins, R. Zeyen, H. Laval and A Benuzzi, Contribution to FBR accident analysis" the JI~C- !spra FARO Programme, Prec. of the Int. Conf. on Fast Reactor Safety held in Guernsey on 12-16 May 1986, BNES, Vet. 2, pp 139-144. H. Hohmann, D. Magallon, A. Renuzzi, A.V. Jones and A. Yerkess, Results of the FARO Programme, Prec. of the Seminar on the Commission Contribution to Reactor Safety Research held in Varese (Italy) on 2024 November 1989, Elsevier Applied Science, Oxford, p. 837. D. Magallon, R. Zeyen and H. Hohmann, 100 kg-scale molten UO~ out-of-p'.'l: ir,:eractions with LMFBR structures: plate erosion and fuel freezing in channels, Int. Conf. on Fast Reactor Core and Fuel Structural Behaviour held in Inverness on 4-6 June 1990, BNES. S.K. Wang, C.A. Blomquist, B.W. Spencer, Lid. Mac Umber and J.P. Schneider, Experimental study of the fragmentation and quench behavior of cerium melts in water, Nat. Heat Transfer Conf., ANS, t989.
".-~'~~ ' A l l ELSEVIER
Nuclear Engineering and Design 155 (:995) 253-270
High pressure cerium melt quenching tests in FARO D. Magallon, H. Holunann Commission of the European Communities, Joint Research Centre, Safety Technology Institute, 21020 lepta (Va). Italy
A~t
First-of-a-kind experimental data on the quenching of large masses of cerium melt of reafistic composition when poured mto pressunseu water at re.actor scale depths are presented and di~-'--tis.7~d.M e ~ , ~ ,uv,oav,,~: . . . . h...~ Io~vm,u--'~44 kg of a molten mixture 80 w% UO2-20 w% ZrO2, which were delivered by gravity through a nozzle of diameter 0.1 m to 1 m depth nearly saturated water a't 5.0 MPa. The objective was to gain early information on the melt/water quench process previous to tests that will involve larger masses of melt (150 kg of mixtures UO2-ZrO2-Zr). Particularly, pressures and temperatures were measured both in the gas phase and in the water. The results show that significant quenching occurred during the melt fall stage with 300/0 to 42% of the melt energy transferred to the water. About two-thirds of the melt broke up into particles of mean size of the order of 4.0 nun. The remaining one-third collected still molten in the debris catcher but did not produce any damage to the bottom plate. The maximum downward heat flux was 0.8 MW m 2. The maximum vessel overpressufisation, i.e. 1.8 MPa, was recorded with 44 kg of melt poured into 255 kg of water and a gas phase volume of 0.875 m 3. No steam explosions occurred.
1. Introduction
The general background o f the present F A R O L W R test series has been previously reported by Corradini and H o h m a n n (1993). It is summarised here. The reference situation is that o f a postulated core melt-down accident when jets o f molten cerium penetrate into the lower plenum water pool. This issue suffers a lack of d a t a on the water quenching potential that determines whether the thermal loading on the b o t t o m head structures is mitigated. In past analyses it had been assumed that this sequence o f events would have result in either settling o f most o f the fuel unquenched on the lower head with eventual RPV wall failure, or steam explosion. During the TMI-2 accident, al-
though 20 tons o I cerium did p o u r into the wa*.er lower plenum, neither o f these two possible scenarios actually occurred. Consequently, safety aspects o f the F C I issue needed to be reconsidered and it has been found o f fundamental importance to carry out tests involving large amounts o f prototypical curium poured into water at reactor scale depth, in order to characterise the melt/water mixing and quenching process, and melt/structure interaction. The J R C - ! s p r a F_ARO plant ( H o h m a n n . 1986) is used for such a purpose. It is a multi-purpose test facility in which severe accidents can be simulated out-of-pile under a variety o f conditions. Basically, a maximum quantity o f the o r d e r o f 150 kg o f UO2-ZrO2 fuel type melts (up to 3000 °C)
0029-5493[95/$09.S0 © 1995 Elsevier ScienceS.A. All rights reserved SSDI 0029=5493(94)00876=Z
254
D. Magalion, H. Hohmann{ Nuclear Engineeringand Design 155 (1995) 253-270
produced in the FARO furnace, possibly mixed with metallic components (e.g. Zr), and delivered to a test section of interest. The plant has been used previously for LMFBR safety problems such as melt relocation and molten fuel/ sodium interaction. The 1.5 m 3 test vessel TERMOS used for MFCI experiments can withstand 10 MPa at 300 °C, which makes it Imrtieularly suitable for simulating hil~h pressure accident scquenees. It can contain up to approximately 1 m 3 of coolant over a height of 2.5 m and a diameter of 0.71 m. The objective of the test series is to determine: (1) the melt quenching rate associated with the melt/water penetration, (2) the hydrogen production associated with the zirconium oxidation, (3) the thermal load on the bottom structures, and (4) to characterise the debris structure. It must be noted that these tests are not steam explosion experiments and, consequently, are not intended to produce data for this issue. The first part of the test matrix, which deals with high pressure and nearly saturation conditions, further limit the risk of a steam explosion and allow to concentrate on the quenching phenomena. Nevertheless, the possibility of occurrence of a steam explosion has been taken into account and the instrumentation chosen accordingly (fast pressure transducers in the water). In the absence of previous work under the conditions mentioned above, it had been decided to proceed by step towards the target corium quantity of 150 kg. Two tests has been performed at constant volume with a pure oxide mixture 80 w% UO2-20 w% ZrO2, which was delivered nearly by gravity through a nozzle of diameter 0.1 m to a 1 m depth saturated water pool at 5.0 MPa. The melt quantity, the water mass, the vessel diameter were ! 8 kg, 120 kg, 0.47 m and 44 kg, 255 kg, 0.71 m for the first and the second t,~st, respectively. The paper summarises and discusses the findings from these experiments. A simple assessment of the energy transferred to the water during the melt fall stage is made. The heat fluxes through the debris catcher bottom plate are calculated from the thermocouple data. The most relevant and recent experimental study on the break-up and quenching of cerium melts in water can be
is the CCM series (Wang, 1989). Particularly, the CCM-5 and CCM-6 tests dealt with approximately 13 kg of a 60 w% UO2-16 w% ZRO2-24 w% steel mixture poured into 510 kg of 45 °Csubcooled and saturated water at 0.1 MPa through a nozzle of diameter 0.051 m. Results from these tests are briefly compared with ours.
2. Experiment description 2.1. Test arrangement and procedure
The experimental arrangement is shown in Fig. 1. The interaction vessel TERMOS (10 MPa, 300 °C, main dimensions in Fig. 1) is connected to the FARO furnace via the release tube, the intersection/isolation valve unit and the release vessel. The cerium is melted at low pressure (i.e. 0.1-0.2 MPa), while the pressure in TERMOS is as required by the test (e.g. 5 MPa). The release vessel closing flap is the barrier between the low pressure and the high pressure regions. Nevertheless, the main isolation valve SO2 remains closed during cerium melting for the sake of safety. The temperature in TERMOS up to the SO2 valve corresponds to the saturation conditions of TERMOS (e.g. 263 °C). At the end of the cerium melting phase, the SO2 is opened, the melt released from the furnace to i~" catcher, and the SO2 closed again. The release vessel volume is used as a lock-chamber for pressure equalisation. This is obtained by bursting a double disc mounted on the communication line, which acts as a quick opening valve. Upon pressure balancing, the release vessel hinged-flap automatically opens, and the melt is delivered by gravity to the water. 2.2. T E R M O S test section 2.2.1. Scoping test ( S T )
An internal water stainless steel container of diameter 0.47 m was included, separated from the pressure vessel by A1203 bricks for thermal insulation. From the upper flange of the pressure vessel to the lower flange of the SO2 valve, a thermal insulator was installed externally. The heating of
D. Magallon, H. Hokmann / Nuclear Engiaeering and Design 155 (1995) 253-270
FARO fumoce
255
1 Lower electrode Release tube Closing Disc (W) 71,72 detectom Release tube
-I-NT~ -
Mirror system drive 7 W-Disc | VideocamSurveillanceJ
Protection valve S01
---'-t
Main intersection valve S02 (¢=120 mm) Communication I~ne (~--40 mm) Double rapture disc .~,o_fe~ valve (IOMPa) Release vessel (volume up to 502 = 0.028 m 3) Hinged-flap
(ergo=,= 100 ram) TERMOSve~l 710 mm Water Funnel Bottom plate (debris catcher) =
/
Supports
Fig. 1. Quenchingtest-2 arrangement. the space inside the water container was provided by means of four heater rods of total power 90 kW. The space above the upper flange of the
pressure vessel up to the lower flange of the SO2 was heated by means of trace-heaters (21 kW) put on the outer surface of the components. The
256
D. Magallon, H. Hohmann / Nuclear Engineering and Design 155 0995) 253-270
debris catcher was supported by three adjustable legs put on the bottom of the water container. The debris catcher bottom plate had a thickness of 0.04 m. Its upper face was distant 0.135 m from the bottom of the container. The catcher was provided with trace-heaters (3.5 kW) attaebed on its lower face. 2 . 2 2 Quenching test 2 (QT2) For this test the internal vessel of diameter 0.47 m and the A1203 bricks were removed, thus increasing the diameter of the water pool to the diameter of the TERMOS vessel, i.e. 0.71 m, as indicated in Fig. 1. To account for the increase of steel mass to be heated with respect to the scoping test (the pressure vessel itself 0.045 m thick, i.e. about 7 tons), additional trace-heaters for a total power of 17.5 kW were fixed on the lower part of the TERMOS outer wall. The vessel was thermally insulated from outside. The same debris catcher as for the scoping test was used with a funnel to account for the change in diameter of the test section. 2.3. Test instrumentation The principal quantities measured during the corium quenching were pressures and temperatures both in the gas region and in the water, temperatures in the debris catcher bottom plate and water level swell. The distribution and types of the probes reported in Figs. 2 and 3 correspond to the quenching test-2 configuration. They were essentially the same for the scoping test, except for level swell measurement devices, which were not present. Four KELLER pressure transducers (piezoresistive, 5 kHz frequency response) measured the vessel pressurisation. They were installed at the end of 0.012 m diameter, 0,4 m long straight tubes emerging in the gas phase region at the levels reported in Fig. 2, and water cooled. Prior to the test, the uppermost ones were used to monitor the pressure in the release vessel and in the chamber between the double disc, respectively. Four VIBRO-METERs (piezoelectric, 15 kHz frequency response) were located in the water as indicated in
Fig. 2 for rapid transient records in case of an FCI. They were protected by stainless steel grids. The eight gas phase K-thermocouples were installed in such a way that they could not 'see' a centred melt jet. Fourteen thermoeouples were fixed on the structures by means of clamps. They were fixed to the internal vessel for ST and to the pressure vessel for QT2 at the same axial positions. The 25 water Kqhermocouples were essentially sacrificial thermocouples used to determine the downward progression and radial extension of the melt jet. Those not destroyed during melt penetration recorded the long time water temperature history. The water centreline thermocouples were attached on thin (0.2 mm) stainless steel wires crossing the test section. The opening of the release vessel was indicated by the rupture of a 0.5 mm K-thermocouple (OD1) mounted opposite to the hinge, and fixed both to the release vessel lower flange and to the flap. Another 0.5 mm K-thermocouple (OD2) was placed on the centreline of the vessel, 0.25 m below the lower face of the flap, for detecting the passage of the melt. The four resistance probes for measuring the level swell in QT2 were installed 0.25 m, 0.50 m, 0.75 m and 1.00 m above the initial water level, respectively. An experimental probe for a continuous measurement of the level swell using a time domain reflectrometry method was also installed at the same radial position as the resistance probe, but 150° apart.
3. Experimental conditions The conditions are summarised in Table 1 for both tests. The interacting melt mass corresponds to the quantity of debris found in the debris catcher. The melt, water and gas volumes were roughly twice higher in QT2 than in ST (actually, 2.4, 2.1 and 1.9 higher, respectively), The melt temperature is inferred from ultrasonic temperature sensor measurements performed in previous similar melting and release tests in FARO. An attempt for measuring the temperature of the melt in the release vessel was made in QT2, but was unsuccessful. The delivery times have been deduced from the OD1 and OD2 signals as follows.
D. Magallon, H. Hohma~ I Nuclear Engietceringand Design 155 0995) 253-270
25?
4
3150 - -
2000
--
Continuous LevelIno~cotor 2100
2135
- -
4 C0nt0ctProbes
- - *
Thermocouples • ~ 1.6 turn "~ 0.6 mm 1600--
-
-
1635
!135 - -
~ =
1050--
-
-
980
-
-
7~0
4 YtBR0-g~.Rs
550-350-135--
Fig. 2. Quenching test-2 instrum©ntation(all dimensionsare in mm).
The ODI rupture was taken as the melt release start (time 0). The O1:)2 thermocouple was not destroyed by the melt (in both tests) even though it indicated temperature .jumps above 1400 °C. One reason for that was the off-centring of the melt jet induced by the hinged flap. The time corresponding to the start of temperature decrease after the maximum has been taken as the end of
the release vessel discharge, thus giving the delivery time. As noted in Table 1, the durations (280 ms and 370 ms, respectively, for ST and QT2) are higher than the theoretical values (110 ms and 235 ms, respectively, for ST and QT2) due to crust formation in the delivery tube and start-end of release disturbances. The relative Iarger difference noted for ST is explained by the formation of an
D. Magallon, H. Hohmann 1 Nuclear Engineering and Design 155 (i£195) 253-270
258
!
?
I I
< ................................................................................. ..4.L~. ................................................................................ •
15
1
...175
/
i
)
•
i
) l
•":L-..................... \
t l ~ :' e r m o c o u p l e s
3x2
^ i
at
1200 i,m
3
;14 18
~
~ . |
~
~ ==: a x 5,~ m
*
..........i ...... o° ........... ~,--
o
15
/
I
.135 ....9 5
supports
~hermocouples lhermocouples
$ 0.6 mm $ t , 0 mm
Fig. 3. Debris catcher instrumentation (all dimensions are in mm). Table 1 Summary of experimental conditions Seeping Test
Quenching Test 2
Melt Composition (w%) Mass (kg) Temperature (oC) Delivery nozzle (ram) Delivery time (s) Flow rate (kg s -t) Free fall in gas (m)
80 UO2 + 20 ZrOz 18 2650 100 0.280 64 1.66
80 UO2 + 20 ZrO2 44 2750 100 0.370 119 !.53
Waler Mass (kg) Depth (m) Temperature (°C) Subcooling at melt contact (°C) Fuel to coolant mass ratio
120 0.87 266 (230 bottom plate) 2 (38 bottom plate) 0.15
255 i.00 263 (255 bottom plate) 12 (20 bottom plate) 0.17
83 steam + 17 Ar 0.464 270
70 steam + 30 Ar 0.875 263
0.470 0,~,0 5.0 5,4
0.710 1.3 5.g 6.1
Gas phase Composition (w%) Volume (m3) Temperature at ODI rupture (release start) (°C) Test vessel Diameter (m) Overall volume (m 3) Pressure at ODI rupture (MPa) Pressure at melt/water contact (MPa)
D. Magallon, H. Holnnann I Nuclear Engineering and Design 155 (1995) 253-270
abnormally thick crust in the nozzle (5 ram) and above the melt (about 10 mm), probably favoured by the small quantity and the poor overheating. Melt streams rather than a coherent mass release is likely for this test. The vessel pressure, and water temperature and level indicated in Table ! depended on the TERMOS heating and pressure equalisation procedures, which were different for each test and are detailed in the next two sub-sections.
259
5.0 MPa in the same way as for the scoping test. To suppress boiling of the water during the pressure equalisation between TERMOS and the release vessel, argon was blown into TERMOS just before bursting the communication line. f u t u r e disc. Due to this argon addition, the pressure at the time the melt started flowing down was 5.8 MPa. It increased to 6.1 MPa as the melt crossed the gas phase. Consequently, the suheoolJng of the water at melt/water contact was 12 °C in the bulk against 20 °C at the bottom plate level.
3.1. Scoping test
The pressure in TERMOS at 80 °C was 0.4 MPa (argon). Heating was pursued up to reaching a water temperature of 266 °C and a vessel pressure of 5.3 MPa without further addition of argon (pressurisation due to steam generation only in boiling regime). Then, the heating power was reduced just to maintain these conditions. This residual power was switched off a few seeonds before the melt to water release. The water height counted from the debris catcher bottom plate upper face was 0.87 m. Upon bursting the communication line rupture disc, the pressures in both TERMOS and the release vessel equalised to 5.0 MPa. This little blow-down of TERMOS enhanced water boiling and induced a level swell. The temperature of the water did not exhibit significant changes. While the melt crossed the gas space, the pressure increased up to reaching approximately 5.4 MPa when the melt leading edge arrived near the water surface (contact time deduced from thermocouples), As a consequence, boiling should have completely ceased at that time, and the water resumed its original depth of 0.87 m. Note that water was about 38 °C subcooled near to the debris catcher (referr,~ to 5.4 MPa) because of an unaware switch-off of the power of the catcher heater. The decrease from 266 °C to 230 °C was localised within 0.25 m above the debris catcher bottom plate. 3.2. Quenching test 2
The pressure in TERMOS was 0.2 MPa at 70 °C (essentially argon). The heating of the test section was pursued up to reaching 263 °C and
4. Experimental remits and
The experimental curves presented in this section are also used as a basis for discussing the timings of the melt fall stages and for suggesting some interpretations of the facts that occun~cl. In order to avoid any ambiguity, each sub-~:tion starts with a short introduction presenting just the related experimental curves. The different times indicated in the figures and in the data summary in Table 2 were deduced from the thermocouple signals. This method does lead to some uncertainties. In particular, one does not know exactly whether the bulk leading edge of the melt or some preceding lumps contacted the probe first. For this reason, the data deduced from the thermocouples form part of the discussions. For all the data reported, time zero corresponds to the OD! rupture (release vessel hinged-flap opening, i.e. start of melt release). 4.1. Melt fall stage 4.1.1. Experimental data
The gas phase pressure traces corresponding to the first 1.4 s of both tests are reported in Fig. 4. The values are normalised to the value at time zero, i.e. 5.0 MPa for ST and 5.8 MPa for QT2. The water pressure transducers (Rot reported) gave exactly the same signals as the gas phase transducers, except the classical late drift due to heating. In particular, they did not ~.ndicate fast dynamic pressurisations typical of energetic events (steam explosions).
260
D. Magallon, H. Hohmann I Nuclear Engineering and Design 155 (1995) 253-270
Table 2 Summary of experimental results (time t = 0 is release start)
Melt Mean velocity in g~-. phase (m s-z) Mean velocity in water ( m s -I) Broken up (kg) Conglomerate1 on bottom plate (kg) Mean size of fragments (ram) Melt/debris rejection Bottom plate Maximum temperature increase (°C) Maximum downward heat flux (MW m -2) State Pressure increase Short-term maximum from release start (MPa) Short-term maximum from melt/water contact (MPa) Long-term maximum (MPa) Steam explosion Temperature increase Steam (maximum measured) (°C) Steam (mean value at t = I0 s) (°C) Water (°C) Maximum levd swell Continuous probe (mm) Resistance probes (150 ° apart from continuous probe) (ram) hfferred from thermocouples (mm)
Scoping test
Quenching test 2
4 2.3 12 6 4.5 No
5 3.7 3O 14 3.8 No
Intact
275 (contact face) 0,8 Intact
I.I (at t = 1,2 s) 0.7
!.8 (at t = !.2 s) 1.5
1.6 (at t = 12 s) No
1.8 (at t = 20 s) No
86 ~43 15 (max. at t = 12 s)
83 ~30 23 Cat t = 25 s)" ~450 Cat t = 1.15 s) >250, <500 (at t = 1.45 s)
> 130
~Maximum not reached at that time.
2.0 1.751,5-
~ 1.25-
g. t.uI
0.0
0,2
0.4
0.6 0.8 ti~ [=]
1.0
1.2
1.4
Fig. 4. Early vessel pressure increase (ST, scoping test; QT2, quenching lest 2, t = 0: release start; MWC, melt water contact; MBC, melt bottom contact; AMIW, all melt in water; E O B . end of break-up).
Fig. 5 shows the signal from the experimental continuous level-metre and the analogical signal from the resistance probe located 0.25 m above the initial water free surface (labelled cp1385) corresponding to QT2 (no data are available for ST, except that inferred from thermocouples and reported in Table 2). The other contact probes (0.5 m and above) did not move from rest. The resistance probe indicates that the level swell definitely ~eached 0.25 m at time 1.35 s (and remained above this value up to time 1.6-1.7 s), while the continuous probe indicates that the value of 0.25 m was already reached at time 0.85 s. According to the different angular positions of the two probes (they were installed 150° apart), the differences observed may indicate that the level swell was not uniform, probably influenced
D. Magallon, H. Hohmann [ Nuclear Engineering and Design 155 0995) 253-270
261
[mm] ~0-
_A/k_
~400-
oLMCP 1365
~_ J
~i''r"
_!
o 300--I, u.I
,,-I, ~ o -
100-
'
,,~,~,_~~f ~1~
/_..
O-
0.0
0.5
1.0 TIME [ o ]
1.5
2.0
Fig. 5. Level swell in QT2 (CLM, continuous level indicator; CPI385, contact probe located 250 ram above the initial water level).
by the off-centring of the melt jet. The continuous level-metre indicates that a maximum of around 0.45 m was reached at time 1.15 s. The two signals returned to be in phase while the water level moved dowlt after 1.15 s. The early gas phase temperature increases are presented in Fig. 6 for QT2. They were very similar for ST (see Fig. 8). A significant non-homogenous heating of the gas phase volume is noticed with maximum temperature increases around 80 °C. 4.1.2. Discussion
200-
0.0
~ ,~ o%=~:~-:,=
0.2
0.4.
0,6
0,8
1.0
1.2
1.4-
[s]
From the water thermocouple signals, only a broad picture of the melt penetration history could be established. Mainly the times of melt/water impact and melt arrival on the bottom plate could be determined as reported in Figs. 4 and 6. For ST, the water thermocouples appeared contacted by the melt in a rather stochastic way as the result of a non-coherent release. For QT2 it
THE Fig. 6, QT2earlystearatemperatureincrease(for legend,see Fig. 4).
could be established that only an off-centring effect from the hinged-flap was present. The melt flowed down laterally but coherently, probably impacted first the debris catcher funnel and then
262
D. Magallon, H. Hohmann I Nuclear Engineering and Design 155 (19,35)253-270
spread on the bottom plate from one side. In both eases, only few water thermocouples were damaged. The curves of Fig. 4 indicate that the break-up process was essentially the same for both tests. Before penetrating into the water, the melt crossed the gas space (height 1.66 m for ST and 1.53 m for QT2). Pressure increases of 0.4 MPa and 0.33 MPa are observed before the melt/water contact (MWC in Fig. 4) for ST and QT2, respectively. If only gas heating by the falling melt were the cause of these pressure increases, the temperature of the gas phase should have increased by 41 °C for ST and 36 °C for QT2, respectively (considering that superheated steam behaves as a perfeet gas). Clearly, the gas phase temperature traces reported in Fig. 6 do not indicate such an increase for the time period of interest. One reason could be that classical thermocouples can hardly measure transient steam superheating. Another reason could be that steam heating was more concentrated around the melt path. Steam generated by radiative heat transfer from the melt may be also invoked. For ST, steam generation induced by the 'disequilibrium' created at the time of pressure equalisation between TERMOS and the release vessel nevertheless occurred (see equalisation procedure descriptions in the previous section), and enhanced the phenomena. The pressure increase after melt/water contact (MWC) was 0.7 MPa for ST and 1.5 MPa for QT2, but only beyond the trailing melt penetration time (AMIW time in the figures) are the increases resulting from melt break-up alone. The end of break-up time (EOB) reported in the figures means that the unbroken part of the trailing edge should have reached the bottom plate at that time. It has been calculated by adding the delivery time to the time at which the leading edge touched the bottom (MBC), thus assuming that the break-up process continued for all the time the melt descended through the water. The rate of pressure increase curves are plotted in Fig. 7. These curves have been obtained by differentiating the spline smoothed pressure histolies. It is seen that the slope became negative very soon after melt/water contact (taken as the reference time in that figure), which seems not to be
[~,/s]
3.5
°
3.0-
~
2.5-
o
2.0-
'¢
I~
•
ZI ~
ST
1
, 0.4
~ n,6
t,00.50.0 -0.5. -0.4 -0.2
0.0
~ 0.2
~ 0.8
1.0
1.2
t - t¢
Fig. 7. Pressure increaserates during the melt fall stage (to, contact time). consistent with the start of quenching. One explanation, consistent with a possible previous heating of the gas phase by the melt, is that steam produced at the beginning of the melt/water interaction was colder than steam already present in the gas phase, which partially condensed. For ST, the rate of pressure increase never came up again to the value it had just before the melt/water contact. The fluctuations noticed within the break-up window can be attributed to an irregular melt penetration. 4.1.3. Estimates o f the energy released by the melt to the water
An estimate of the amount of steam added to the system as a result of the melt falling through the water has been made, assuming that the thermocouples of the gas phase measured the actual temperature of the gas whatever the steam superheating might have been. The quantities of steam present in the free-board volume at the time of melt/water contact and at the time of the maximum pressurisation have been calculated by using the specific volumes of the steam at the partial pressures of the steam and averaged temperatures of the gas phase: i.e. 4.8 MPa, 277 °C and 5.5 MPa, 303 °C for ST; 4.9 MPa, 265 °C and 6.4 MPa,
D. Maffallon, H. Holumnn I Nuclear Engineeringand Design 155 (1995)253-270 298 °C for QT2). The difference between the two quantities has been taken as the quantity of steam produced during the melt/water interaction. One finds 0.8 kg for ST and 5 kg for QT2, which corresponds to an energy of 3 MJ and 16 MJ, respectively. The time difference between the maximum pressurisation and the melt/water contact is very similar for both tests (see Fig. 7) and is approximately 0.9 s. Using the above quantifies of steam produced, mean generation rates of 0.9 kg s - t and 5.6 kg s - t are found for ST and QT2, respectively. A maximum value of the steam generation rate can be calculated for QT2 by multiplying the peak v~lue ofdp/d: hi Fig. 7 by the quantity bfV/R,~T~, where M is the steam molect,lar weight, V ~tte gas phase volume, Ts the gas phase temperature, and Ru the universal gas constant. One finds a peak steam generation rate of 11 kg s-J for T8 = 300°C. Varying Tg from 263 °C (initial gas phase temperature) to 345 °C (maximum measured) does not change very much the above value: one goes from 10.2 kg s - ' for 263 °C to
[ deg. c ]
263
11.8 kg s- ~ for 345 °C. One should further introduce a compressibifity factor for taking into account that steam does not behave as a perfect gas at the pressures and temperatures of concern here. Taking a value of 0.8 for this factor, the above estimates increa.~ by 20%. The energy required to increase the temperature of the liquid water during the melt fall from 264 °C to 270 °C in ST and from 263 °C to 277 °C in QT2 (see Figs. 10 and 11) is 5 MJ for ST and 12 MI for QT2, respectively. Comequenfly, the total energy released to the wat~ daring the .melt through water fail stage is estimated to be 3 MI for ST and 28 MI for QT2, respectively. 4.2. Longer term stage 4.2.1. Experimental data The gas phase temperature distribution over the first 10 s are plotted in Fig. 8 for ST. A trend to temperature equalisation is noticed as time increases. The pressure histories over a period of
5
1 320-
270
o
I
25O 0
1
2
3
4
5 [~]
6
7
8
9
T~AE
Fig. 8. Scopingtest steam temperaturehistoriesfor the first 10 s after releasestart.
10
264
D. Magallon, H. Hohmann / Nuclear Engineering and Design 155 (1995) 253-270 ff,4Pa]
tures are plotted together with the saturation temperature of the water corresponding to the total pressure. The averages have been calculated using all the thermocouples of the gas phase (8) and only those in water which did not exhibit strong disturbances during the passage of the melt (14 for ST and 13 for QT2). For QT2, two gas phase thermocouples, probably touched by water droplets due to condensation, showed a sharp decrease between 4 and 8 s. This explains the average temperature perturbations noti~d in Fig. 11 during this period of time.
- PT 03 150 3150 -
8.0
,~P--- I
7.5
"
7.0
II
o..
SCOPING
TEST
6.0:: 5.5 5,0 O
5
10
15
20
4.2.2. Discussion
25
[,,l
As can be seen in Fig. 8, the gas phase temperature is largely not uniform, especially soon after the end of the jet fall. It is believed that this map represents the reality of the steam generation and mixing with the cover gas. When quiet steaming from the debris bed dominates, temperatures tend to equalise. Figs. 9-11 indicate that similar vaporisation/ condensation scenarios developed at two different
TI~E
Fig. 9. Long-term pressure histories.
time of 25 s are reported in Fig. 9. The long-term maximum pressure was reached after approximately 12 s for ST and 22 s for QT2, and was 6.6 MPa and 7.6 MPa, ,espectively. In Figs. 10 and 1l, the averaged gas phase and water tempera320
E
310-
o "0
tll r1"
290Q: W
~ 280W p. 270
-
260 0
5
10
15
20
25
TIME [ s ]
Fig. i0. Scoping test steam and water temperatures compared with the saturation temperature.
D. Magallon, 1t. Hohmann [ Nuclear EngMeering and Design 155 (1995) 253-270
26~
310
,q
A 0
GAS PHASE / ~
~
+-~
| LU
.--2 • T.ttP)
w Lu n
!
J
LU I-
270-
0
5
t 10
i
+
15
20
25
TIME [ S ] Fig. 1 I. Quenching test 2 steam and water temperatures compared with the saturation temperature.
time scales for both tests: the melt break-up time scale and the long-term debris cooling time scale. During melt break-up, the production of steam was very fast and intense, inducing a fast pressurisation of the vessel up to a level that could be higher than the saturation pressure corresponding to the gas phase temperature (see Fig. I1 for QT2). As a consequence, at the end of the breakup phase, condcnsation dominated steam production fi'om the debris cooling (mainly steam to steam condensation). The higher the steam generation during the melt fall quenching stage, the higher the pressure decrease at the transition phase. Then, the contrary occurred up to the second maximum. Later on, condensation definitively dominated. Previous to the second pressure maximum, steam generation due to debris cooling is evidenced by the fact that the gas phase temperature decreased as the pressure increased. Boiling in the debris region is further evidenced by the wiggles on the signals of the water thermocouples located just above the debris (s¢¢ Fig. 12). When comparing Figs. 10 and 11 one remarks that the temperatures are of the same order for
both t¢sts. This is not so surprising considering that the melt to coolant mass ratio was very similar for both tests and that homogenisation rapidly occurred after the melt fall stages (within the first 5 s in the water). Proportionally to the melt quantity, more steam was produced in QT2 during the melt fall stage than in ST (see Section 4.1.2), inducing a higher pressurisation and a subcooling of the water. As a consequence, part of the superheated steam produced in the debris catcher region condensed while rising through the water, increasing the water temperature. 4.3. Debris catcher data The debris structure was very similar for both tests. Fig. 13 shows a view of the ST debris surface as found in the catcher. Fig. 14 shows a view of a part of the conglomerate in con*act with the bottom plate. This part was certainly still molten when it contacted the plate. As can be seen, the plate did not suffer any damage. This is not surprising using a pure oxide melt becausv we knew already, from the BLOKKER series per-
D. Magallon, H. Hohmann / Nuclear Engineerir,g and Design 155 (1995) 253-270
266
290
280-
270 -
ILl
260-
::3 t-.<
11: 2 5 0 tu O.
240-
230-
220 (~
I
I
I
I
5
10
15
20
25
TIME [ S ]
Fig. 12. Seeping test water temperature just above the debris (at level 175 mm).
Fig. 13. View of the ST debris bed from above.
Fig. 14. View of the melt spread on the bottom plate in ST.
formed in F A R O in the frame of the L M F B R programme (Hohmann, 1989; Magallon, 1990), that jets of 100 kg of pure UO2 around 3000 °C interacting in dry conditions with 0.04-m thick plates preheated to 400 °C did not induce any erosion even though the plates reached 1000 °C. Some samples of the ST debris are shown in Fig. 15. They were very irregular in shape and aspect. Only a few spheres such as that in the
figure were present. Contrarily to pure UO:, these particles were very brittle and great care had to be taken in manipulating them in order not to alter the analysis. In general, particles up to a dimension o f the order of 10 mm presented a fine granular internal structure, normally a sign ot rapid quenching. Larger particles (a few particles of some centimctres were part of the debris) presented an internal structure such as that shown in
D. Magallon, H. Hohmann I Nuclear Engineering and Design 155 (1995) 253-270
Fig. 15. ST debris samples.
Fig. 16. View of the internal part of a 4 em size particle of the ST debris.
Fig. 16. The central grain growth structure is typical of slow quenching. The particle size distributions are given in Fig. 17 and compared with that of CCM-5 and CCM6. "lhe difference in the mean particle size between ST and QT2 can be explained by the higher penetration velocity observed in QT2. The mean particle sizes of both FARO tests lie in between those of CCM, which agrees with the fact that the degree of subeooling of the water in FARO was in between that of CCM tests. It should be noted that in neither of the two CCM tests of concern, were signs of a re-agglomerated molten cerium evidenced. This happened only in CCM-2 where the water depth was 0.64 m instead of 1 m. Proportionally to the quantity of melt which experienced break-up (about 12 kg in ST and 30 kg in QT2), the energy corresponding to water
267
heating during the melt fall phase, as calculated in Section 4.1.3 (i.e. 5 MJ for ST and 12 M,I for QT2), is comparable with the CCM values (Wang, 1989), where 10 ,'eLI in subcooled water (CCM-5) and 3.5 MJ in ~::e:ated wamr (CCM-6) for 13 kg of melt were calculated. Also, the total energy transferred to the steam/water system during melt fall stage, which represents 44% of the energy content of the broken part of the debris for ST and 62% for QT2, globally corresponds to the CCM-5 (55%) and CCM-6 (36%) estimates. Most of the thermocouples located just above the surface of the bottom plate were destroyed during melt impact. 'Contact temperatures" (in fact the thermocouples were welded 0.5 mm below the contact face) and temperatures 5 mm below the bottom plate surface are shown in Fig. 18 for different radial locations (QT2 test). They reflect the fact that the melt did not spread uniformly on the plate. Downward heat fluxes calculated from these temperatures are reported in Fig. 19. A maximum value of 0.8 MW m-2 is obtained in the central zone of the plate. The time integrals of curves of Fig. 19 give the energy released to the plate (Fig. 20), which reached a maximum around 9 MJ m-2 at time 20 s in the centre of the plate. Maximum values around 2 MJ m -2 and 4 MJ m -2 are reached at 9 s (r = 80ram) and 12 s (r = 160ram), respectively. Negative downward heat flux and, accordingly, decrease in energy transfer are calculated for times beyond 9 s (r = 80ram) and 12 s (r = 160mm) as a consequence of the inversion of the axial temperature gradient noted in Fig. 18. This indicates that radial heat transfer through the plate in view of temperature homogenisation became dominant with respect to heat transfer from the debris.
5. Concluding remarks The essential objectives of the two first tests of the FARO quenching series have been achieved. Estimates of the melt quer,,,hing rates and of the thermal loads on the bottom structures, and characterisation of the debris have been made. It has been found that the melt lost more than 50% of its energy during the fall through water stage. De-
268
D. Magallon, H. Hohmann / Nuclear Engineering and Design 155 (1995) 253-270 10.0
I
I
I
I
I
I
I
I
i
I
I
i
I
I
I
I
i
IrlOI IO in A
@0
E E
[]
lZl
,jj N LLI ,_1
Io
1.0
[]
o_
@o
•
El rl
mean particlesize
Io
a.
iD I
[]
[] Q
@
SCOPINGTEST
4.5 mm
O
QUENCHINGTEST2
3.8
i
CCM-5(REF.3) CCM*6(REF.3)
5.0 3.5
[] 0.1 0.01
0.1
1-2
5 10
20 3'o'40506070 ''' 80
9'o ;s
9899 ''
99.9 '
)9.99
PERCENT MASS LESS THAN INDICATED SIZE Fig. 1% Particle size distributions (total debris for CCM, particulate debris, i.e. two-thirds of total, for ST and QT2).
5,°1
5°°1
-°"''
1
g'°° tl \ 300
s/'z" " ~
0
5
1
15
t~
20
,25
[s]
Fig. 18. Melt/bottomplate contacttemperaturesand temperatures in the plate 5 mm belowthe contact surface. spite the fact that about 30% of the cerium collected still molten on the bottom plate where the water was up to 38 °C subcooled, no steam explosions occurred and the plate remained intact.
Maximum downward heat fluxes of the order of 0.8 MW m -2 have been calculated from the there mocouple data. Some problems arose when trying to assess the steam generation rate from the pressure and temperature data. They were essentially due to the heating of the gas phase by the melt itself previous to the melt/water interaction, and to the difficulty in measuring transient steam overheating by using classical thermocouples. On the other hand, one can hardly eliminate these problems working with high pressure and necessarily limited volumes. Consequently, it is important for the interpretation of the experiments by the mixing codes and for the validation of these codes, that they include good models for the melt fall through and heat transfer to the gas phase previous to and during the melt/water interaction. With respect to past quenching experiments using smaller quantities of thermite generated melt at ambient pressure, no fundamental differences were found, but some correlation between
D. Magailon, H. Hohmann / Nuclear Engineering and Design 155 (1995) 253-270
269
1.10 =-
,g, X I< Ill "r
[
~
o
5
-2.10 =
J
~
J
10
15
20
.,
j 25
t~me Is]
Fig. 19. Downward heat fluxesat different radiat locations.
1,10 ~-
centre 8 , 1 0 =-
6
6,106-
=~ 4 , 1 0 = -
2,10 =-
0,100
160 mm
r
=
r
= 80ram
I
I
I
I
I
5
10
15
20
25
tim Is) Fig, 20. E n e r g y t r a n s f e r r e d
test pressure and steam generation on the one hand, and water depth and molten mass that reached the bottom plate on the other hand, could be further evidenced. This justifies pursuing the study using larger amounts of melt and larger water pool depths. Also, the influence of the
to
the b o t t o m plate.
presence in the melt ~f a percentage of a metalfic compound has to be evaluated. Tests involving 150 kg of UO2-ZrO2-Zr melts is in progress at JRC-lspra. The boundary conditions for these tests will be as close as possible to that of the preliminary tests reported in this paper
270
D. Magallon, H. Hohmaan / Nuclear Engineering and Design 155 (1995) 253-270
(closed volume), in o r d e r to facilitate comparisons and c o d e validation.
Acknowledgements "~,ais w o r k was p e r f o r m e d in collaboration with the U S N u c l e a r Regulatory C o m m i s s i o n in the frame o f a Technical Exchange Agreement. The authors greatly acknowledge the w o r k and efforts o f the whole F A R O experimental team.
References M,L. Corradini and H. Hohmann, Multi-phase flow aspects of fuel-coolant interactions in reactor safety research, OECD/ CSNI, Nucl. Eng. Des. 145 (1993) 207-215.
H. Hohmann, D. Magallon, H. $chins, R. Zeyen, H. Laval and A Benuzzi, Contribution to FBR accident analysis" the JI~C- !spra FARO Programme, Prec. of the Int. Conf. on Fast Reactor Safety held in Guernsey on 12-16 May 1986, BNES, Vet. 2, pp 139-144. H. Hohmann, D. Magallon, A. Renuzzi, A.V. Jones and A. Yerkess, Results of the FARO Programme, Prec. of the Seminar on the Commission Contribution to Reactor Safety Research held in Varese (Italy) on 2024 November 1989, Elsevier Applied Science, Oxford, p. 837. D. Magallon, R. Zeyen and H. Hohmann, 100 kg-scale molten UO~ out-of-p'.'l: ir,:eractions with LMFBR structures: plate erosion and fuel freezing in channels, Int. Conf. on Fast Reactor Core and Fuel Structural Behaviour held in Inverness on 4-6 June 1990, BNES. S.K. Wang, C.A. Blomquist, B.W. Spencer, Lid. Mac Umber and J.P. Schneider, Experimental study of the fragmentation and quench behavior of cerium melts in water, Nat. Heat Transfer Conf., ANS, t989.