ML20078L276

Plant Phenomenological Evaluation Summary on Direct Containment Heating in Support of Ipe

ML20078L276
Person / Time
Site:	Farley
Issue date:	06/30/1992
From:	FAUSKE & ASSOCIATES, INC.
To:
Shared Package
ML20078K997	List: ML20078K993 ML20078L002 ML20078L262 ML20078L276 ML20078L295 ML20078L329 ML20262G008 ML20262G013 ML20262G018 ML20262G022
References
FAI-91-73, NUDOCS 9411290137
Download: ML20078L276 (73)

Text

{{#Wiki_filter:- ( 0 FAI/91-73 FARIEY NUCIIAR PIANT UNITS l' AND 2 PHENOMENOIDCICAL EVAIEATION

SUMMARY

ON DIRECT CONTAINMENT HEATING IN SUPPORT OF THE INDIVIDUAL PIANT EXAMINATION Submitted To: Southern Nuclear Operating Company Birmingham, Alabama Prepared By: Fauske & Associates, Inc. 16WO70 West 83rd Street Burr Ridge, Illinois 60521 (708) 323-8750 June 1992 t O l ) 9411290137 941109 PDR ADOCK 05000348 i a9m3 0'O,l P PDR

A A _.4.u .,u a ......m A 1-i (? i ABSTRACT P Phenomenological issues on direct containment heating (DCH) have been examined in support of the Farley Individual Plant Examination Program. The I approach taken was to (1) synthesize _DCH knowledge from analyses of data obtained'from downscaled mockup DCH experiments and (2) estimate the poten- ' I tial magnitude of DCH for the Farley plant. From the latter evaluation one can assess if the results of such a postulated phenomena would approach conditions sufficient to challenge the containment integrity. Experiments performed at Sandia (SNL) and Argonne National Laboratories (ANL) include linear scale representations (up to 10%) of the the reactor cavity and instrument tunnel that can be applied to the Farley Plant. The av.silable experiments were collectively analyzed to identify, for typical high pressure melt inj ection (HPME), the effects of the fo11owin5 factors: I the potential for dispersal of debris, the effect of water in the reactor cavity and on the lower compartment floor, and most importantly, structural barriers at the tunnel exit. The instrument tunnel exit into the lower compartment has been identified as an effective separator of entrained debris which would greatly suppress the -potential for extensive debris l dispersal by de-entraining a significant fraction of debris particulate. l These experiments-clearly show the influence of containment geometry-(structures) in limiting the extent of debris dispersal in the containment atmosphere During the IDCOR Program, a set of 1% linear scale " building block" experiments were performed to examine the influence of structures such as~ the seal table and the lower containment compartment. These single purpose tests demonstrated that these structures had a dominant effect in preventing efficient debris dispersal. If used to assess what should be investigated in larger scale experiments, these results point to the reactor cavity and lower compartment geometry as being of primary importance in mitigating DCH. With this. extensive background of small scale experiments showing the importance of the containment geometry, further experiments at 5% linear ~.

11 - im Ji I '\\ / scale were performed. These included representations of the reactor coolant system, high temperature core debris, the reactor cavity, important struc-tures immediately outside of the reactor cavf.y (seal table support, steam generators, reactor coolant pumps, etc.), the lower containment compartment and the upper containment compartment. The test series also included one experiment in which the simulated primary system was pressurited with steam. All four experiments produced very similar results and none of the experi-ments demonstrated any significant direct heating. Maximum DCH efficiency found in these experiments ranged from 0% to 2.5%. Based on this result and the modeling of the de-entrainment process, the best estimate calculations for the Farley reactor system result in an estimated containment pressuritation, due to DCH, of less than one atmos-combined hydrogen burn shows the containment phere. An assessment of a atmosphere would be inerted for all sequences.

Hence, such a combustion process would not occur.

However, even if a hydrogen burn is assumed, the resulting pressure does not approach the containment ultimate capability. ( As a result of the experiments, scaling analyses, and the evaluation of perceived uncertainties in the DCH process, it is concluded that dispersal of the core debris by high pressure melt ejection would result in a contain-ment pressure far less than conditions sufficient to challenge containment integrity. 1 f 'g ~~ )

- iii - 4 7N TABLE OF CONTENTS run ABSTRACT 1 TABLE OF CONTENTS iii LIST OF FIGURES vi LIST OF TABLES vii 1.0 PURPOSE 1-1 2-1 2.0 PHENOMENA 2.1 Description 2-1 2.1.1 Controlling Physical Processes 2-1 /' - 2.1.2 Relationship to Containment Failure . 2-4 (, Mechanisms and Modes 2.1.3 Relationship to Source Term . 2-5 2.2 Experimental Results 2-6 2.2.1 Sandia HIPS Experiments 2-6 2.2.2 Argonne C'JTI Experiments 2-9 2.2.3 SNL-DCH (Surtsey) Experiments 2-13 2.2.4 BNL Simulant Fluid Test 2-18 2.2.5 IDCOR/FAI k'ood's Metal Tests 2-18 2-26 2.2.6 FAI-DCH Experiments 2-31 2.2.7 Scaling Effects 2.2.7.1 Sweepout and Extent of 2-32 Entrainment Scaling 2.2.7.2 De-Entrainment Scaling (Due to the Change in Flow Direction) 2-33 f 2.2.7.3 Time of Flight Scaling 2-34 (}/

- iv - 'Q TABIE OF CONTENTS (Continued) f.AE2 2.3 Analyses. 2 35 2.3.1 Zion Probabilistic Safety Study 2-36 2-36 2.3.2 Sandia's Preliminary Calculation 2.3.3 HARDCORE 2-37 2.3.4 PARSEC 2-38 2.3.5 Kiva-DCH 2-39 2.3.6 CONTAIN 2-40 2.3.7 Simplified Model for Representing Comparable Debris Dispersal Processes 2-40 2-42 2.3.8 Jet Breakup Medel m 1 3.0 METHODOLOGY 3-1 3.1 Mass of Debris Distributed into the Containment Atmosphere 3-1 3.1.1 Extent of Entrainment 3-2 3.1.2 De-Entrainment 3-3 3-3 3.2 DCH Pressure Rise 3.3 Likelihood of Hydrogen Combustion 3-3 3-4 3.4 The Influence of Uncertainties. 3-5 3.4.1 Core Melt Progression? 3.4.2 RPV Failure Mechanism? 3-6 3.4.3 Extent of Debris Particulation?. 3-6 3-7 3.4.4 Influence of Structures? 3.4.5 Potential for Hydrogen Combustion? 3-8 a

.v. J'(,, TABLE OF CONTENTS (Continued) Pare 41 4.0 PIRiT SPECIFIC APPLICATION 4.1 DCH Calculations. 4-1 4.2 Conclusions . 4-7 . 5-1 5.0

6.0 REFERENCES

6-1 APPENDIX A: Proceedings of the ANS/ ENS International Topical Meeting on Probability, Reliability and Safety Assessment, Pittsburgh, Pennsylvania (April, 1989) A-1 APPENDIX 3: Calculations of DCH Temperature and Pressure Rise. B-1 ( APPENDIX C: Hydrogen Combustion Limits C-1 APPENDIX D: Scaling of Entrainment and Displacement Rates. . D-1 E-1 APPENDIX E: Scoping Calculation for De-Entrainment 0

. vi e t 'N LIST OF FIGURES fJrure No. Eggg 2-1 ANL thermite apparatus for the C'JTI tests, taken from-[ Spencer, 1987) 2-10 2-2 Schematic of the Surtsey Direct Heating Test Facility, taken from [ Marx, 1989). The apparatus used in both the DCH-2 and the DCH-3 experiments 2 15 is shown. 2-3a 1% linear scale building block experiments / reactor cavity. 2-20 2-3b 1% linear scale building-block experiments / reactor cavity and seal table 2-21 4 2-3c 1% linear scale building block experiments /2D representation of the lower compartment 2-22 2 3d 1% linear scale building block experiments / = elevation view of 3D lower compartment 2-23 2-3e 1% linear scale building block experiments / plan view of 3D lower compartment 2-24 2-4 Elevation view of the vessel interconnection to simulate the containment configuration 27 2-5 Cross-section of the melt generator. reactor cavity, instrument tunnel and the simulated 2-28 lower compartment with structures 2-6 Characteristic debris length and 2-41 displacement length for debris movement 4-1 Comparison of Farley and Zion cavity configuratices 4,. i I

vLL - I LIST OF TABLES %Y Table No-EAER 2-8 2-1 Selected HIPS Experiments . 2-12 2-2 ANL CVTI DCH-Related Experiments 23 SNL-DCH Experiments 2-17 2-4 Model Comparisons for DCH Test 2-19 2-5 Zion Cavity Building Block 2-25 2-6 Ceco /FAI DCH Experiments 2-30 4-1 DCH Calculations for Farley Nuclear Plant 4-5 0 O

11 iQ l.0 PURPOSE The possibility of a localized vessel failure should severe accident conditions result in molten debris draining into the lower plenum was first addressed in the Zion [ CECO, 1981] and Indian Point (Con.Ed. and PASNY, 1982) Probabilistic Safety Studies. Along with this, the possiLility of dispersing high temperature debris from the reactor cavity due to rapid steam generation in the cavity or due to.a high pressure blowdown of the primary system was considered. Sensitivity analyses in these studies con-sidered the consequences, if the debris were to directly exchange heat with the containment atmosphere. However, these were not considered to be physi-cally possible because cf the lack of extensive debris particulation and influential structures in close proximity to the reactor cavity. In this regard it is necessary to distinguish between high pressure melt ejection (HPME) and direct containment heating (DCH). The former is the process of hydrodynamically forcing melt out of the reactor cavity due Q to the primary system blowdown. DCH is the process of directly heating the containment atmosphere by the molten core debris should HPME occur. The Zion and Indian Point studies concluded that the first coul d occur but second could not. Given the geometry of the Farley reactor cavu.y, there is no disagreement

that, given the conditions for a hign pressure melt ejec-tion, such a dynamic debris transfer could occur.

Therefore, this report will principally focus on those features which influence the potential (or lack thereof) for debris transport into, and fine scale interaction with the containment gas space. Potential failure of a Pk'R containment due to debris dispersal and consequence of HPME has been the subject of numerous direct heating as a technical exchange meetings between the Nuclear Regulatory Commission (NRC)

Staff, NRC contractors, and the nuclear industry.

Discussions have been motivated by the concern that a large fraction of molten core debris exiting a failed reactor vessel during a high-pressure blowdown could be finely fragmented and distributed into the containment atmosphere. The timing (shortly after vessel failure) of the postulated containment failure has an

r 1-2 . p. kl important ramification regarding the radiological releases, That is, 'if s containment failure due to DCH is. postulated, natural fission product l deposition mechanisms in the containment ~ would not have time to -sig-nificantly reduce the masses.of fission products that could be released. initially through the failure location. In addition, the process of high pressure melt ejection has the potential to increase the. fission product inventory released to the containment atmosphere. 4 1 In Generic Letter 88-20 (NRC,1988), the NRC states that results1from research concerning~ containment failure induced by direct containment heat-ing have not been conclusive and utilities should consider strategies to deal with this issue while awaitin5 its generic resolution. The' objective of this paper is to evaluate. with due concern for realistic uncertainties a significant likelihood for in the physical processes, whether there is I containment failure due to direct' containment heating at the Farley Nuclear. j Plant. These evaluations will provide best estimate conclusions which will form the basis for the probabilistic and deterministic assessments of plant f performance to be incorporated in the Farley IPEs. (;) 8 1 va

21 ,. 3 i (,/ 2.0 PHENOMENA 2.1 Descriotion Direct containment heating is a postulated event of rapid heat transfer between finely fragmented core debris and the containment atmosphere assum-ing (1) the occurrence of post core melt reactor pressure vessel failure at high pressure and (2) that HPME causes extensive debris dispersal. DCH a has been hypothesized as a means of early containment failure because the stored energy of the debris, including potential energy releasable through debris oxidation and hydrogen burning, is enough to cause high containment pressure if a large quantity of the core inventory participates. The extent of pressurization thus depends upon: the amount of debris which would be discharged at vessel

failure, the containment geometry which could be conducive to or an pQ impediment to dispersal beyond the reactor cavity, and-the fraction of the debris which could be finely fragmented and dispersed into the containment atmosphere..

2.1.1 Contro111ne Physical Processes The fundamental issue for DCH is whether a massive dispersion of par-ticulated core debris (tens of tonnes) into the containment ateosphere would be possible. Four steps can be identified which would be necessary for such an outcome. First, the plant stemetry must be such that debris dispersal could oc ar from the reactor cavity. Second, fine fragmentation of debris must occur and be sus-tained. Third, the debris particles must remain dispersed in the flow stream and be distributed throughout the containment gas atmos-phere.

1 l 22 j i .A k-s/ . Fourth, the particles must remain airborne long enough to trans- ~ fer crergy and reace chemically if in contact with steam or oxygen. As will be discussed, some bounding analyses have considered that hydrogen j combustion would occur and contribute a pressure rise which would be in addition to that created by the thermal energy transfer. For this to occur, the atmosphere must be such that hydrogen combustion could be sustained. Major impediments to these processes are the inherent inability to entrain and finely particulate debris, the number of directional changes which particles must endure, the presence of structure in the flow path, stagnation of the flow at corners or around obstacles, expansion of the flow path into the compartment, and the presence of water both in the cavity and in the lower compartment. Each of these is briefly characterized below. Also, the steam partial pressure in the containment would have a significant influence on the likelihood of hydrogen combustion. O t j High velocity gas flow in the reactor cavity would entrain debris but %./ the act of entrainment would also tend to decrease the gas velocity, thereby causing the reactor cavity to pressurize. (This is the subject of [ Henry, 1989], which provides the basis for the methodology discussed herein. For the readers convenience, this is included as Appendix A.) This would decrease the entrainment potential and increase the potential for displace-of the debris by pushing it along the wall of the cavity and instrument ment tunnel. Consequently, only a fraction of the melt would be particulated a large characteristic with the remainder exiting the reactor cavity with dimension. It would also be difficult to maintain an entrained flow of finely particulated, high density debris. Changes in flow direction, stagnation points, and structure all tend to provide for impaction of the particulate on walls, i.e. de.entrainment. Each change in flow direction tends to separate the heavier debris from the gas stream since the debris would move to the outside of the flow curvature as a result of the large density dit-ference. Only the smaller particles, which could respond to the gas flow G t a NJ

2-3 O (_,/ more quickly, could avoid separation from the high velocity gas stream. The a liquid layer or particulate captured on the structure walls would form film and would have to be re-entrained to form particulate. In experiments simulating the reactor cavity at Zion Nuclear Plant (which is not neces-sarily prototypic for the industry) most debris is ejected from the cavity. For these experiments, the materials dispersed in the flow have a wide range of size from 10 micron to several millimeters. However, the size range can be strongly influenced by the extent of the containment geometry modeled in the experiment. As will be discussed later, virtually all of the dispersed mass is associated with the larger particle sizes when the lower containment compartment structures are modeled. These sizes result in a very ineffi-cient heat transfer process between the hot debris and the containment gases, as demonstrated by the relevant experiments. Water would have substantial mitigating capacity in direct containment heating scenarios for those nuclear plant designs where cavity flooding is possible. In many cases, some water would be present in both the reactor f' cavity and the lower compartment and for others the cavity would be entirely e flooded. With the exception of containment bypass sequences with no flow back to the pressurizer relief tank, there would always be water on the containment floor; likely to a depth of at least 15 cm (6 in).

Also, for those sequences in which sufficient water would be injected and accumulated in containment to submerge the RPV lower head, CECO sponsored experiments performed at FAI indicate that failure of the vessel wall would not occur

[ Henry, et al., 1991). Hence, if the vessel penetrations maintain their integrity, as was clearly the case in the TMI-2 accident, the RPV lower head would not be breached and debris would not be discharged to the containment. Obviously, water would have a majer influence on these sequences. For those sequences with limited water in the reactor cavity which could progress to RPV failure, debris discharged from the vessel breach would undergo dynamic interactions and rapid heat transfer to the water upon impact. Subsequent removal of debris from the cavity would occur simul-taneously with removal of the water, and quenching would continue in this multiphase stream. Experiments involving water in the simulated cavities s and on the containment floor have consistently verified the efficacy of g

m 24 9 water as a mitigator. This will be discussed further in the context of the s. CECO /FAI 5% linear scale experiments. r Expansion (deceleration) of the flow from the instrument tunnel into the lower compartment and the presence of the extensive structure therein allows for significant de-entrainment and deposition of most particle sizes los the flow area increases, flow velocities decrease. and large particles would no longer remain airborne. Also, impaction on structures would serve to remove particles of all sizes. Only that fraction of the debris which could persist' as fine particulate could substantially contribute to direct containment heating. If containment failure is postulated, the release to the environment would be a function of the containment failure size. The failure size would in turn be a function of the containment strain, or to the first order. the extent to which the pressure would exceed that corresponding-to the yield stress of the meridinal and hoop tendons. The approach taken in *his study. [~7s is to perform representative scaled experiments (because of_ the importance \\~1) of the lower compartment structures) and couple these with scaling con-l siderations to relate the experiments and the reactor system. In addition, integral accident sequence (MAAP) analyses and phenomenological uncer-tainties will be examined to determine if there is a realistic likelihood for challenging the containment integrity. As will be discussed, there is' no realistic challenge to containment integrity,

hence, the end state evaluations in the containment eser.t tree do not include considerations for DCH.

2.1.2 Relationship to Containment Failure Mechanisms and Modes Direct containment heating is a postulated mechanism for containment failure immediately after reactor vessel failure. If_such a mechanism could l occur, the largest potential for the occurrence would be expected during I core melt accident scenarios that would maintain a high reactor vessel pressure until the time of vessel failure. The containment failure-I (:]) l

2-5 (__-) mechanism would be overpressurization due to a rapid increase in gas tem-perature as the core debris thermal and chemical energy is transferred to the gas. 2.1.3 Relationshio to Source Term Prior to the core debris transport into the lower plenum which could lead to vessel failure and high pressure melt ejection, fission products such as the noble gases, iodine, cesium and perhaps tellurium would have been released from the fuel matrix and into the primary system. For such conditions, significant quantities would likely be released to the contain-ment through openings in the RPV pressure boundary such as PORV operation, a break in the RCS, failure of the in-core instrument thimbles during the core melt progression process, etc. In addition, the fine particulation of core debris in the reactor cavity, due to high pressure melt ejection, could liberate additional fission products to the containment atmosphere (Appendix A). Therefore, any postulated process which would fail the containment f' g shortly after RPV failure would also likely occur when the airborne fission k _-) product inventory in containment would be at, or near, the maximum.

Hence, s

the potential release would also be at a maximum. After this point in time, the airborne fission products would decay with time due to natural deposi-tion mechanisms such as sedimentation, impaction, condensation, etc. For sequences with water in the cavity or containment sprays activated, the expected fission product source term would be reduced relative to a dry cavity sequence due to the enhancement of fission product aerosol capture by impaction of the aerosol particles and spray droplets or by increased Stefan flows to steam condensing surfaces. These sequences would also provide an additional heat sink (water) to quench the core debris by steam generation, which also inhibits or prevents hydrogen combustion. The use of sprays also would cool the containment atmosphere, as well as the deposited debris, and to some extent the surface of the RCS, thereby minimizing or eliminating revaporization from the reactor vessel inte the containment gas space. Should revaporization and release from the vessel occur, the sprays would scrub the fission products from the atmosynere. Functionally, the injection of water into a reactor vessel with a lower head failure would accomplish r-) EsY u l

2-6 ts k_-) the same objectives, i.,, cooling the debris, preventing or minimizing revaporization and establishing long term coolability. Injection into a vessel would essentially prevent any revaporization by cooling the inner RPV surfaces but would be somewhat less effective in scrubbing fission products from the containment atmosphere. 2.2 Ernerimental Results Several investigations have been performed and documented which provide a data base for understanding direct containment heating phenomena. These experiments have been carried out at Argonne National Laboratory, Sandia National Laboratory, Brookhaven National Laboratory and Fauske 6 Associates, Inc. Results of these experiments are briefly described here. Many of the experiments only relate to processes in the reactor cavity and the potential for removal of debris from the cavity. Commonwealth Edison documented its position that debris could be removed from the reactor cavity / instrument tunnel as a result of HPME conditions in the Zion Study [ CECO, 1981). This ~% position has not changed. Hence, those experiments which only represent the )/ Zion reactor cavity provide no additional insight to this issue for the present IPE and are only briefly summarized below. The influence of con-tainment structures on the debris dispersal was also documented in the Zion Study. CECO remains convinced that structures have a first order influence on the mitigation of DCH processes. Therefore, experimental programs re-laced to the influence of structures will be discussed in depth. 2.2.1 Sandia HIPS Experiments i l The Sandia HIPS experiments (High Pressure Melt Streaming) [Tarbell, 1984) were conducted following the SPIT series to confirm the dispersal of debris from a simulated cavity geometry, to assess the phenomena of jet geometry, gas solubility, and aerosol generation [Tarbell, 1986a). In those tests which have been reported (Tarbell, 1986b and Pilch, 1986], a 1:10 linearly scaled model of the cavity was used. This model was either open to the desert or placed within an expansion chamber with one end open to the desert. In these experiments, no attempts were made to maintain geometric 1 similarity with containment internal structures. Iron-alumina thermite i U,f-~s. i J

2-7 ) j% (,) charges with masses of about 176 lb (80 kg) were used. A table of reported tests presented in Reference [Pilch, 1986) is reproduced here as Table 2-1. In experiments (HIPS 3C, -7C, and -8C) carbon dioxide was used as the cover gas, at pressures between about 3 MPa (435 psia) and 5.6 MPa (812 psia). Pressures as high as 11.7 MPa (1697 psia) were used in two other tests (HIPS 2C and 4W). In two reported experiments (HIPS-4W and -6W), a water-filled cavity was used, and destroyed by overpressure during the blowdown. Experiments HIPS-4W and -6W are not relevant to the response of the RCS and containment because it is doubtful that the RPV could be failed if the reactor cavity were filled with water. One experiment (HIPS-8C) featured an annular gap around the thermite generator, simulating the gap around the RPV if the insulation were assumed to be removed. The reflective insulation used has substantial strength relative to the expected phenomena, particularly in rapid transient events. Therefore, tests which ignore the insulation do not represent the response of the RCS and containment and are not used in our assessment. qk,/ For all the HIPS tests, dense aerosol clouds surrounded the debris jet, s and sweepout from the cavities was nearly complete in all cases, above 95%. After tests in which the cavity was placed inside the confinement room, this debris was found either within the room or on the concrete pad just outside its open end. While not a representation of the actual plant configuration, this test series is included here because it demonstrated the substantial influence of structures to capture the debris outside of the reactor cavity / instrument tunnel. 'In the HIPS 7C tests, over 30 percent of the debris was found in the rear of the confinement chamber, much of it against the rear wall. Observations with high speed films show that debris exiting the cavity splashed off the ceiling and rained down in that location. Over 97 percent of the original melt mass was recovered from the chamber and concrete pad, and the rest either landed on the ground beyond the pad or was in t5e aerosol cloud which billowed out. Thus, the confinement chamber does not model the containment lower compartment structures. On the other hand, it does illustrate that a minimal representation of the structure results in 7-i

2-8 l'V Table 2-1 SELECTED HIPS EXPERIMENTS Thermite Initial Vessel Mass Pressure Extent Test Dispersal Water In Name Scale (kg) (1bm) (MPa) (psia) (%) Cavity SPIT-19 1:20 10.3 22.7 12.6 1827 95 No HIPS-2C 1:10 80.0 176 11.7 1697 99 No HIPS-4W 1:10 80.0 176 11.7 1697 -95* Yes O HIPS-SC 1:10 80.0 176 6.7 972 99 No HIPS-6W 1:10 80.0 176 3.8 551 -95* Yes HIPS-7C 1:10 81.5 179 5.5 798 98 No HIPS-8C 1:10 80.0 176 3.7 537 98 No

• Cavity destroyed during test.

1 1 \\ N._)

1 + 2-9 substantial removal of the debris from the gas stream. In the HIPS-8C tests, aluminum collection. pans were placed on.the ~ chamber floor to determine 'the spatial distribution of debris exiting the cavity. and deduce which fraction exited through the keyway outlet, and l which fraction exited. through the angular gap around the melt generator. About 25% of the debris was concluded to have been released through this

gap, deflected off the ceiling, and fallen into pans around'the melt gener-ator.

This corresponds to the area fraction taken up by the annular opening assuming the insulation had been removed. Again, we believe this experiment is not representative of the actual plant because the reflective insulation around the reactor. vessel was not represented in any manner. 2.2.2 Arronne C57TI Exoeriments EPRI sponsored a series of reactor material experiments at Argonne National Laboratory known as the CUTI tests (Corium/ Water Thermal Interaction, Reference (Spencer, 1987 and Spencer, 1988]). The experimental configuration represented some major features of a typical PWR large dry lower containment compartment cavity, and upper compartment at a 1:30 linear and stainless steel) were used scale. Reactor materials (principally UO2 and were created by an exothermic thermite reaction. Molten debris was injected downward into a simulated cavity and keyway, see Figure 2-1, which was connected to an expansion volume that was partitioned like the lower and upper compartments of a containment. In some

tests, water was present either in the cavity or the expansion volume or both.

The objectives of these tests were to examine heat transfer between core debris and water, sweepout of water and core debris from the cavity, steam generation, hydrogen generation, and to characterize the spatial distribution of dis-persed debris. The test apparatus included the thermite reaction vessel (which is the source of the molten core debris), an interaction vessel (which represents the reactor cavity), a pipe simulating the instrument tunnel, an expansion -vessel representing the containment, a " trap" above the pipe discharge to simulate a seal table and a baffle plate to separate the expansion vessel

2 10 .I p y g. 9, INSTRUMENTATION TREE ASSEMBLY v, S AMPLE P RT 2 NOTE: ALL DIMENSIONS IN 67.9/34 6 TC12-2 CENTIMETE R S/INCHE S 'I M e % TC12-3 15 4 76.2 1.D. > 4-- .6 4 /.2 5 TC12-4 5 148.1/58.3 CONTAINMENT CELL SEALING FLANGE TC12-5 h O -1.6 /.6 3 LOWER GAS SAMPLE PORT 57.6/22.1 CWEV LOWER INTERNALS TC12-6 A 4 $A 'Y 1' 'Y n,,__ PTS-5.10 J' h aj q ~~ y M 4 5.3/17. ~ p N NP 'P ARTICLE TR AP' 13 TC13-3 THERMITE VESSEL f$40/.3f.,, TC13-5 1r N j 188.8/88.5' W [ d m TC13-6 l -P MO " M '" INST. FEED THROUGH'S # DRAIN VALVE ri k fESSEL INSTRUMENTED SPOOL DISCHARGE LINE PTS-4 q, I / \\ .....ae CORIUM/ WATER EXPANSION CHAMBER Figure 2 1 ANL thermite apparatus for the CWTI tests, taken from (Spencer. 1987]. i

2-11 C/ into upper and lower compartments. The simulated core debris produced by the thermite reaction was composed of 604 UO, 16% Zr0, 24% stainless steel 2 2 (67% Fe, 21% Cr, 12% Ni) and had a temperature of about 3100 K (5120'F). performed with low (less than 0.5 MPa or 75 psia) and high Tests were (- 5 MPa or 725 psia) pressure blowdowns as well as with inert and oxidizing atmospheres. Several different initial water level conditions were tested in the interaction vessel ranging from completely empty to essentially full of water before the melt was released from the thermite furnace. Initial conditions and results for DCH related CWTI tests are listed in Table 2-2 (adapted from [ Spencer, 1988]). The corium masses used in the CVTI tests corresponded to a range of about 36%-60% of the corium inventory for Zion. The DCH-related tests were conducted by starting a thermite reaction at nominally atmospheric pressure. The melt was subsequently pressurized (by gas from external high pressure gas cylinders) to burst a bottom diaphragm and initiate melt ejection. The duration of the gas blowdown was suffi-p ciently long to assure sweepout of all available melt. Debris that was not remained as a 1 - 3 mm thick crust uniformly deposited on the wall swept out of the interaction vessel and the entire pipe surface (for CUTI-5 and -6). Various mass fractions of debris were observed to be dispersed from the cavity into the simulated containment volume. Debris was collected reactor at various locations in the apparatus including (1) the interaction vessel and pipe, (2) the " particle trap / reflector" at the top of the pipeway as it exited into the bottom of the expansion vessel, (3) on the floor of the simulated containment, and (4) on the top of the simulated lower compartment which represented the operating deck. Characteristic particle sizes for the debris swept out into the air atmosphere ranged from 64 to 700 microns for j tests CVII-11 and -13, which had no water in the cavity. Most particles were in the range 100 - 300 microns (Spencer, 1988]. The experiments with water in the interaction vessel or expansion volume prior to melt ejection demonstrated the effectiveness of water for removing heat from dispersed debris, including the energy released due to oxidation. These tests ex-hibited little or no direct containment heating. Results for test CWTI-12 l 1

V U Table 2-2 ASL CWT 1 DCE-RELnTED EIFERIMENTS ' I l Test No. I 5 6 11 12 13 l 1 l h l Driving Fressure, MPa/ psia 5.0/725 4.7/681.5 5.1/739.5 2.8/406 4.0/580 l I 1 l Atmodphere l At At air air air l 1 l Corium Mass injected, kg/lba 3.94/t.67 3.75/S.25 2.93/6.45 2.69/5.92 2.27/4.99 l I l corium Mass Swept out, kg/lba 2.44/5.37 1.21/2.66 0.88/1.94 1.31/2.88 0.20/.44 l l l corium Mass menained, kg/lba 1.5/3.3 2.54/5.59 2.05/4.51 1.38/3.04 2.07/4.55 l l l l water in cavity, kg/lba 5.6/12.32 dry dry 4.6/10.12 day l 1 l g l Dispersal Impediments l s/b8 s/b s/b mone none l [ l l Atmosphere Initial Temp. E/'r 419/294.2 408/274.4 411/279.8 422/299.6 298/76.4 l I I l Atmosphers Peak Temp, R/*r 417/290.6 461/369.8 435/323 407/272.6 621/657.8 l i l l Atmosphere Initial pressure, MPa/ psia 0.52/15.4 0.22/31.9 0.22/31.9 0.37/53.65 0.32/46.4 l 1 1 l DCN Efficiency', t I 0 5 1 0 62 l l 1 83.31% lineet scale model of Eion.

• s/b shroud / baffle.

= Measured Atm. Neetup

• DcH Efficaency

= a 100. Man. Equilibration Atm. Heatup . -- - -^---"- --- ..,.... ~. --

2-13 I 7-~ show no significant contribution to DCH with the codispersal of debris and a relatively large amount of water into the containment atmosphere. Test k performed without the presence of the impeding structure and CUTI-12 was clearly demonstrates the influence of codispersed water. Modest direct containment heating was observed in tests CWTI-6 and -11 which had water only in the pan of the expansion vessel (containment. floor). The largest atmospheric heating occurred in test CUTI-13 where no structures and no water were present in either the interaction vessel' (cavity) or expansion vessel (containment).. This heatup was caused by a sweepout mass of only 0.2 kg. Hydrogen generation and/or oxygen depletion that occurred during these tests were also measured to determine the extent of oxidation and its con-tribution to the overall energy input to the system. The most important observation from the ANL experiments concerns the influence of structure on the debris dispersal. The " particle reflector" in this experiment was similar to the in core instrument tube seal table con-figuration in the Zion containment building. In this case, the horizontal i baffle plate above the " particle reflector" would represent the floor O separating the containment building into upper and lower compartments. Comparison of results from CUTI Tests 6,11, and 13 reveals the effect of ] structure on direct containment heating. All three of these tests were performed with a dry interaction vessel (reactor cavity), but Tests 6 and 11 included structure in the expansion vessel (containment) while test 13 had no structure. The DCH efficiencies (whose definition is given in Table 2-2 and calculated in [ Spencer, 1988)) for Tests 6 and 11 were 5% and-1%, while the DCH efficiency for Test 13 was reported as 62%. Thus, the substantial q effect of the seal table structure was to prevent a significant fraction of the debris from entering the containment atmosphere directly, Because of the structure, a major fraction of the entrained debris was deposited on the ] lower containment floor. 2.2.3 SNL-DCH (Surtsey) Eroeriments Sandia also has conducted a series of experiments designated as Surtsey [Tarbell, 1987a;

Tarbell, 1987b;

Tarbell, 1988a;

Tarbell, 1988b].

.I Experiments DCH-1, 2, 3 and 4 used cavity configurations similar to those of )

~ .l 2-14 l the HIPS 1:10 experiments. These cavities have been completely enclosed in-3 a large expansion vessel, i.e. 3637 cubic feet (103 m ) capacity and a 145 1 psia (1.0 MPa) design pressure. No attempts were made to represent the geometry of any containment internal structures such as the seal table, the lower compartment, the operating deck, etc. At the time of this writing, four tests have been performed, and results have been presented in detailed reports for two of these tests [Tarbell, 1987 and Tarbell, 1988a), i DCH-1 involved 44 lb (20 kg) of molten iron-alumina thermite injected into a 1:10 linear scale model of a cavity only,.with a nonprototypic exit' guide box added to the instrument tunnel exit to direct debris vertically upward along the centerline of-the vessel. Figure 22 illustrates the Surtsey facility. The thermite was propelled with nitrogen gas initially at 2 55 MPa (370 psia). Peak pressures ranged from 0.09 MPa (13 psia) to 0.13' l MPa (18.9 psia) and were achieved less than one second after debris disper-sal. High speed film shows that debris shooting upward at 40 m/s (131 ft/s) expanded laterally and filled the entire chamber cross-section within a few meters of the cavity exit. About 11.6 kg (25.5 lb) was dispersed from the cavity, which includes the correction for estimated-oxidation. Melt retained within the cavity and chute was in the form of a thin crust, and a 1.2 kg (2.64 lb) mass was found at the base of the keyway inclination. Aerosol measurements indicated that much material was fragmented to a size under 10 microns, but the measure-ments may be inaccurate and a large uncertainty is present in the actual amount of such material. The calculated range of aerosolized debris was 5 to 25 percent of the dispersed mass. Mechanical sieving of debris collected in the chamber showed a log-normal size distribution with a mass mean size of 0.55 mm. Thus the bulk of the debris ejected from the cavity. was of millimeter size. This test was analyzed using the methodology described in. [ Henry, 1989) and good agreement was obtained when the specific configura-tion without structure was considered. 4 DCH-2 involved 80 kg (176 lb) of molten iron alumina thermite inj ected into the same 1:10 linear scale model of a cavity with an instrument tunnel and the nonprototypic exit guide box. The thermite was propelled with l l

2-15 b (12.2 m ELEVATIONI r 3 SURTSEY VESSEL / E SUPPORT STRUCTURE B / LEVEL 4 A T / (7.02-m ELEVATIONI p b(/ ) LEVEL 3 6 15.80 m ELEVATION) p l-CHUTE l l - CAVITY ,/ -A MELT GENERATOR ] (O ELEVAT1ONI Figure 2-2 Schematic of the Surtsey Direct Heating Test Facility, taken from [ Marx, 1989). The apparatus used in both the DCH 2 and the DCH-3 experiments is shown. O

2-16 A '( j' V nitrogen gas initially at 982 psia (6.77 MPa) and peak pressures ranged from 0.22 MPa (32 psia) to 0.31 MPa (45 psia), and were achieved less than one second af t.er debris dispersal. High speed film shows that debris leaving the guide box expanded laterally and filled the entire chamber cross-section within a fraction of a second, likely indicating that the debris-gas two-phase mixture was " choked" at the exit of the guide box. The total mass recovered from the test chamber, cavity, and melt gener-ator was about 99.5 kg (219 lb), which represents an increase in mass of 24%. This attributed to several potential sources related to the construc-tion of the apparatus and test conduct as well as oxygen uptake by oxidation of the iron in the thermite. The late.er source was considered to be the most dominant mechanism. About 91.3 kg (201 lb) was dispersed from the cavity. Five types of debris were identified, with each type being generally associated with a particular area of the apparatus. On the upper third of the vessel,in line with the cavity exit, 1-2 mm thick sheets of brittle debris were tightly bonded to the vall. Debris stuck to lower portions of the vessel wall was loosely bonded and formed sheets of 2-4 mm thickness. On horizontal surfaces, debris was found to be approximately 1-mm diameter spheres, and agglomerations of such spheres with irregularly-shaped masses of previously molten debris. Melt retained within the cavity and chute was in the form of a thin crust, and a 4.4 lb (2 kg) mass was found directly underneath the melt generator. This test was also analyzed using the methodology described in [ Henry, 1989) and found to be consistent with the proposed methodology considering the simplified geometry used in the experiment. DCH-3 and DCH-4 have not been fully reported and the extent of ex-perimental aburvation is limited.

However, the overall results are summariud w.

'ftble 2-3 [Tarbell, 1987a). Test DCH-1 as reported in [Tarbell, 1987b) had the largest efficiency of energy transfer from the debris to the atmosphere with > 90%, while DCH-4 test was lowest with nominally 35%. The latter was affected strongly by the absence of chemical energy release because of the lack of oxygen in the chamber atmosphere. For consumption of oxygen in the chamber, the degree of oxidation in the DCH-1 Oi test was the greatest. while the DCH 3 debris was the least oxidized. As

2-17 (g - (_s/ i 1 Table 2-3 '( SNL-DCH EXPERIMENTS **2 i Driving Thermite Mass (ke/lbm) Atmosehere Pressure Driving Peak Temp Peak Press f Test MPa/ psia Gas Initial Swept-Out ' Remained 'C/*F MPa/ psia l DCH-1 2.4/348 N 20/44 .-10/22 -10/22 260/500, 0.095/13.78 2 DCH-2 6.8/986 N 80/176 ~76/167.2 -4/8.8 880/1616 0.225/32.63 2 DCH-3 6.0/870 N 80/176 -75/165 ~5/11.0 860/1580 0.205/29.73 2 (::) DCH-48 6.9/1000.5 N 80/176 -74/162.8 -6/13.2 860/1580 0.19/27.55 2 t f t s 110% linear scale model of Zion cavity only, no representation of the contain- + ment structures outside of the reactor cavity / instrument tunnel. 2 Dry cavity and containment. j 1eert atmos,here. i l t k f I ? i

2 18 <s ( 1 ',/ s discussed previously, the methodology presented in Appendix A [ Henry, 1989) constitutes part of that used for the present IPE assessment. Table 2-4 compares the methodology with the SNL DCH tests. There is good agreement with the measured results when the simplified geometry of the experiments is considered. Therefore, the difference between the SNL experiments and the evaluation of the reactor system is dominated by the absence of structure and not by the experimental scale, the materials used or the driving medium (nitrogen or steam). 2.2.4 BNL Simulant Fluid Test Brookhaven National Laboratory has performed a series of simulant fluid tests to investigate the extent of entrainment in the reactor cavity and the influence of reactor cavity geometry [ Tutu, 1988). Scaled (1/42) models were constructed of the Zion, Surry and Watts Bar reactor cavity and instru-ment tunnel configurations. Of particular note for this application is that models of the Zion containment (Cinsberg, 1988] "suggest that the structures () exerts a strong influence on the flow which issues from the cavity, redirecting the flow and, very likely, changing its droplet size characteristics". These obse rvations clearly point to the inclusion of containment structures in both experimental studies and plant applications such as an IPE. l l 2.2.5 IDCOR/FAI Vood's Metal Tests During the IDCOR Program. a set of " building block" experiments were performed with the single purpose of demonstrating the extensive influence of structures in the lower compartment. These 1% linear scale models began with the reactor cavity and instrument tunnel, added the seal table, ex-panded this to a two-dimensional representation of the lower compartment and completed the test series with a three-dimensional representation of the lower compartment. The same driving pressure of nitrogen gas (2 MPa/300 psia) mass of Wood's metal (0.3 kg/10.66 lbm) was used in all tests and the j principal measurement was the fraction of the Wood's metal retained in the test section. Figure 2-3a through 2-3e illustrate the various test sections used to sequentially add the influential structure. Table 2-5 lists the [} v

o o r Table 2-4 MODEL COMPARISONS FOR DCE TEST I 1 l Test Potaseters/ l l Predictions DCH-1 DCH-2 DCH-3 DCH-4 l l Measurements { l I l 1. Geometrical Potameters l l and vessel Pressure l A te8/ft*) 0.002/.021 0.002/.021 0.002/.021 0.002/.021 A Im*/ft*3 0.5/5.37 0.5/5.37 0.5/5.37 0.5/5.37 i L in/ftl 2.8/9.18 2.3/7.54 2.8/?.18 2.8/9.18 l P l l L to/ft) 1.45/4.75 1.45/4.75 1.45/4.75 1.45/4.75 l l A (m'/ft') 0.06/.645 0.06/.645 0.86/.645 0.06/.645 c l l I P (MPe/pssa1 2.5/362.5 6.8/986 6.0/8 70 6.9/2000.5 v I 1 w l 2. Debris Size l l Predicted (an/ftl 240/.00079 65/.00021 40/.00013 30/.000098 l g3 l Measured Mass Mean tan /ft) 550/.0018 mot Available not Available mot Avettable l I I l 3. Mass Despersed as l l Pine Particulate l l Fredacted (kg/lbel 50/44 27/59.4 11/68.2 33/72.6 l l Measured (kg/lbel 11.6/25.52 33/12.6 11.4/40.48 -30/66 l l (Mass Deposited on) (Mass Deposited on) (Mass Deposited on] l l ( the Lower Mead J L the Lower Need J ( the Lower Need y l 1 l l 4. Peak Tempetature l l Predicted 681/765.8 1032/1397.6 1104/1527.2 1138/1588.4 l l (Using Measused) l l LMass Dispersedi - 1150/1610 1193/1687.4 1193/1687.4 l l l Measured (K/*P) 501/441.8 I 1 1 1

5. Peak P essuse _

0.19/27.55 0.30/43.5 0.32/46.4 0.33/47.85 l l l Fredacted IMPa/psial l Using Measured .l t l Mass Dispersed [ l Measured IMIa/pseat 0.18/26.1 0.33/47.85 0.'28/40.6 0.27/39.15 l 1 1 .I

O O O 1 i TRANSPARENT PLATE

~

m T

3 SUPPLY TUBE i f / // '/4 \\ / c, u / / / // YYYYf/A Figure 2-3a It line ar scale building block experinnents/ reactor cavity. s -w v-

O O O SEAL TABLE [/ TRANSPARENT PCATE / h A /[}< SUPPLY TUBE m-!jj;jj g/ ii;-se: a ~ ////d [x ///// WW 4 \\ Figure 2-3b 14 l i nic a r scale buildling block expeiliaciits/ reactor cavity and seal table. I - - ~ ~ e <- ~ - - -. - -. - + - - - -, -

2 p&.- 2 + a uu 2-22 o w -- w y q g iau 8 m\\3 \\\\ o --= w \\ SUPPLY i /// /// acI o F. ow c r ee pe O

'W O 04 O I I I I I I 8 I i l l 1 l i l l l I i l 1 i I I SEAL i l TABLE i l 1 ll(J f*i (J -N

I

\\ l DOOR I i ."1 i iSTEAM ~ cDOOR 6 BIOLOGICAL 8 l l l-l GENERATOR p SHIELD l i'8 l l l l 1 I i i f { SUPPLY TUBE l I Figure 2-3d 11 linear scale building block experiments / elevation view of 3D lower compartment. I i m.

O O O DOOR j I Vl_. Vl_ i i KEYWAY SCREEN IN l l [ TOP SURFACE OPENING i l +- x l -? SUPPLY

. I' L..

i- ' TUBE l E + - EI .[ DOOR I m;, L. I h-b

• 2 ZZ BIOLOGICAL SHIELD SEAL l

TABLE ~ ~~ ~ ~ ~ LIP -i -}- l- -1 -t-Ml V STEAM GENERATOR I ~ Figure 2-3e 11 Linear scale buildling block experimeints/ plan view of 3D lower compartment. -. ~ - -. - - ... ~ - -. -.. ~.

2-25 J ,e n V Table 2-5 ZION CAVITY BUIIRING BIDCK Percent of Debris Retained in the Test Section Reactor Cavity and Seal Table 6.07 2D Lower Compartment 48.0 2D Lower Compartment Within Reduction Flow Area to the Upper Compartment 76.6 ) 'w/ 3D Lower Compartment 97.4 1 l l 1 1 l 1 l sJ i i

.2-26 b Wood's metal retained in the test apparatusLand shows that, with only the seal

table, virtually no material remains;>yet with the 3-D lower compart-a ment' essentially all of the simulated core material remains in the lower compartment, most.of it close to the instrument tunnel (If we assume that the debris which is not retained in the lower compartment could cause direct

heating, these results suggest the process would have an efficiency of only about 2.5%.)

These experiments clearly-show the first order influence of structure in mitigating DCH, 2.2.6 FAI-DCH Experiments i l A jointly funded Commonwealth Edison and Fauske & Associates. Inc. ~ experimentcl program was established and carried out at Fauske 6 Associates, Inc. [FAI, 1990). The program addressed the issue of direct containment heating in support of the Zion IPE. In keeping with lessons learned from smaller scaled tests, these experiments were conceived and designed to have a 5% linear scale simulation of the Zion reactor cavity,-instrument tunnel, l lower compartment (including the steam generator, the reactor coolant pumps l and the refueling canal wall) and upper compartment of the containment building. I Figures 2-4 and 2-5 show the experimental apparatus. Iron aluminum i thermite was used to simulate the molten core material, with the 20 kg (44 lbm) thermite mass representing about twice the scaled debris mass and energy content. This was done to compensate for the greater propensity of freezing on the reactor cavity walls in small scale experiments. The ex-periments can also be viewed as a study of the effect of lower containment structures on the dispersal of iron aluminum debris when compared with SNL-i f DCH tests, which also used iron-aluminum as a core material simulant, but had no containment internal structures. This work represents the most i complete geometric similarity of cavity and containment internal structures I for scaled DCH experiments performed to date. l The experimental matrix included four tests, all with 20 kg of ther-l mite, three driven by nitrogen and one by steam. Water was available on the containment floor in all tests and was also available, to different extents. p V i t I i +

2-27 C) (./ l I l: I i. 3 V = 12.2 m 1 / ,-1=.-, a..,,. s - s

• ls l

[ / , ',.s,' v,. -, i, ) I i'* 1.33m, 3 s I { 1.4 5 m s s V = 45L l \\ \\ l i,..; # .a / 's p.3 O 2 co j i m = [] 6' PIPINO l d l n r \\,, m m t UPPER COMPARTMENT spuupe OWER g j COMPARTMENT N 1.19 m i g th(-egM i i i /////////////////////////////////// ..a,,, 1.2 4m 1.D. Figure 2-4 Elevation view of the vessel interconnection to simulate the containment configuration. 1

em rx O J L] FAI Zion DCH Model Fuel Transfer Canal - T ~a i, Divider Wall ' G< '

O Keyway Mouth

\\ Seal Table Uk -( p" 'r g7

[J;.) Floor Opening

( Position of .\\ -- s'i ' - -Steam Generators-(2) [. Top Venling Port /. I* / Reactor Coolant Pumps-(2) eht O erati D ) To Expansion Vessel g.. y Pl Sleam Generator Compartment y 7 = Melt Generator b';,,,,, 9 _f; p.= --Representative Crano Wall. Main Steam Line Room Wall, y Hx Room Wall, Seal Table Room IE...J$.... [ 8.... Floor Enclosure

Cavity, N (

U

Tunnel,

.,m w. Keyway m 1 I Figure 2-5 Cross-section of the melt generator, reactor cavity, instrument tunnel and the simulated lower compartment with structures. l l i

t 2 29 l O in the various tests; two having about I cm (- 1/2").of water on the cavity floor, one test had 8 cm (- 3 in) and the other was performed with a dry reactor cavity. Steam was used to drive the last test and was accomplished by installing a boiler next to the test apparatus to pressure the RCS with. steam after the thermite reaction was-initiated. Table 26 presents the test conditions for the experimental matrix. Table 2 6 also summarizes the test results and clearly demonstrates-that (1) the tests are very similar in their response and (2) caused very. little pressurization in the upper containment compartment. Measurements of. the containment pressurization showed very little,-if any, contribution to direct heating even though' 90% of the material was dispersed. Pressurization in the reactor cavity and instrument tunnel, as well as the lower compartment, was essentially due to debris-water thermal interactions in the reactor cavity, the blowdown of the melt generator and. steam gener-ated.in the debris quenching procees on the lower compartment floor. The measured peak temperatures of the lower compartment atmosphere were mostly the results of rapid steam generation due to quenching -followed by slow heating of the gas due to debris frozen on structures. This latter energy transfer did not contribute to the pressurization because of the cooling provided by other heat sinks. I Maximum local DCH efficiency of about 2%, based on the peak pressure, was measured in FAl DCH 1 where only a limited water mass was present in the reactor cavity before the blowdown. This is an upper-bound estimate of DCH efficiency because the steam partial pressure was conservatively estimated (underestimated). The temperature incraase in the lower compartment was about 5-6 times as high as in the upper compartment but was dominated by the rapid steam generation since the measured peak temperature was only slightly superheated. The results for' all four tests were very consistent with respect to the extent of debris dispersed, the peak temperature in the lower compartment and the pressuritation transient in the containment. The only differences are the response caused by dynamic interactions in the reactor cavity (determined by the water mass in the cavity) and the hydrogen gener- ) ated which is a function of the water in the cavity and the driving medium. Civen the similarity of the global response for all four, tests. these dif- .[ \\ J

i l ) Table 2-6 l CECO /FAI-DCS EIPERINEsTS 1 i 1 Peak l l' l Measured l Driving Thermite Mass (kg/lbe) Containment Pressure Water-(kg/lbel l-l Pressure Driving _ Compartment 'l .l' l Test (MPa/pslai Gas Initial Swept-Out Pa Psia. Cavity Floor J. i l l-1 DCH-1 3.3/478 N 20/44 ~18/'39.6 165 23.9 1.1/2.4 90/198' l s w. .t I I u: DCH-2 2.9/420 N, 20/44

• 18/~39.6 170 24.7 5.3/11.7 47/103.

o l l DCH-3 3.2/464 N 20/44

• 18/~39.6 155 22.5 Dry

-47/103 l l j l DCH-4 2.3/333 Steen 20/44

• 18/~39.6 172 25.0 1.1/2.4 47/103

.l' g l I i i l t i 9 l i s.9 'l s t l .s.,_, v v ~.... <, s ..,..-..w,__, ,,,4,---,__s._my .my. ... +,.. -.. -.,., s-s,,,,v.

2-31 D ferences have a second order influence'on the containment response. The results were consistent with CWTI-DCH tests in that significant heacup of the containment atmosphere was not observed in any runs that included the important structural barriers of a seal table and the lower containment compartment. In fact, the capacity of structures to mitigate DCH can be seen by comparing these results to SNL-DCH results which did not represent these structures. It is to be noted that the use of steam to eject the debris simulant yielded a containment temperature and pressure response virtually identical to the nitrogen blowdown tests, even though more hydrogen was generated, i.e., from 7% to 15% of the metal reacted. 2.2.7 Scaling Effects Experimental observations in scaled experiments must be related to the full scale conditions (reactor system) with similar initial conditions. For complex phenomena like DCH, the scaling effects may distort the time scale of the processes and analyses must be performed to see if such distortion occurs and, if so, whether it is important in the overall conclusion. As discussed, DCH experiments at FAI and ANL were performed, respec-and3ft linear scale mockups of the containment building.

tively, using 5%

a 10% Contrast these with the SNL-DCH experiments, which included only linear scale mockup of the cavity without modeling the geometry of the containment compartments. Consequently, only the FAI and ANL tests can be related directly to determine if there is any experimental indication of a scale dependency. Using the two different scaled c.xperiments available which incorporated all of the major influential f eatures of a reactor system, we find a maximum of 2% contribution to direct heating of the atmosphere and this is a given experimental matrix, and between reproducible for tests within different tests. We also find for experimental scales varying from 1% to 10%. that when the structures are not represented, the dispersion of debris is virtually complete and that significant heating of the gas occurs. (] Hence, the applicability of the results to the reactor system should focus V

2-32 p. i primarily on the influence of structures. Another feature of the reactor \\' system is the availability of water in the containment which can substan-tially mitigate the gas temperature rise. This is also discussed below in terms of its influence on scaling. 2.2.7.1 Sweenout and Extent of Entrainment Scalinz The phenomenological discussions in the Zion Study (CECO, 1981] described several mechanisms whereby debris could be swept out of the reac-tor cavity given the conditions of core melt, failure of the RPV lower head, debris discharge into the reactor cavity and substantial pressure in the RCS. Sweepout mechanisms included (1) roll wave formation and displacement by the follow-on gas flow, (2) wave formation by the impact of high velocity gas with subsequent breakup in the gas stream, and (3) entrainment off the debris surface by the high velocity gas stream. Subsequent experiments at different facilities (ANL, BNL, FAI and SNL) and different scales (1% to 10%) have demonstrated that sweepout would occur given the necessary condi-(~') tions. Therefore, debris sweepout is not an issue for scaling considerations.

However, the extent of debris particulation (entrainment)

'~ is an issue to be considered for scaling. As discussed in Appendix A, the issue for entrainment is not which mechanism applies for debris removal; rather the issue is how much entrain-ment could occur while other processes are also occurring. Specifically, as debris would be entrained, what would be the feedback on other mechanisms such as displacement of the debris by an imposed pressure difference. Considerations of the simultaneously occurring processes resulted in the model for the extent of debris which could be entrained and accelerated into the gas stream while the remainder was being displaced (driven) from the

cavity, as a more contiguous mass, by the imposed pressure difference.

Comparisons of this model with the Sandia Surtsey experiments shows good agreement with the measured results. Given the co-existence of these processes and the agreement with experiments, what is the influence of experimental scale? 1 0

4 A 2-33 If two processes are occurring simultaneously, the experimental scale - hould not distort one process with respect to the other. To accomplish s this, the ratio of the rates of debris transport should be preserved in scaled tests. 2.2.7.2 De-Entrainment Scaline (Due to the Chance in Flow Direction) Sandia National Laboratory [ Walker, 1987) developed a model for es-timating the likelihood of' debris particles not deflecting with the flow due to a change in flow direction, thus impacting structural boundaries of the flow path. The model was developed for the Zion seal table and instrument tunnel structure where the steam and debris escaping from the cavity makes a J 90 degree turn to escape into the lower compartment. An analogous equation can be used here to describe de-encrainment of debris particles. (For a full development of Equation (2-1) see Appendix E.) The model was benchmarked against HIPS test results mentioned in Section 2.2.2 and it is presented here in the following dimensionless form O a1 in 1+E 1-1L 2 (2-1) o ArW y f where 1.1 Y

1. s c' 4

.d. p d d '~ yw, r - (1 + y2)1/2 fraction of particles failing to make the turn [dimensionless), a-W - height of aperture (vertical dimension) particles travel through as they turn to leave the reactor cavity [m or ft), w - width of aperture (horizontal dimen.sion) particles travel through as they turn to leave the reactor cavity [m or ft), L - height of instrument tunnel (vertical dimension) in reactor cavity [m or ft), f - width of instrument tunnel (horizontal dimension) in reactor cavity [m or ft), C - kag coe W ctent [dimensionless), y 1 l

2-34 ( ) d - debris particle diameter [m or ft). The ratio W/L 7, and r in Equation (2 1) do not change with the linear scale. However, because 1 does not necessarily vary with the scale, the terms ArW and AFL will, therefore, change with the scale. How much a is affected by the scale can be examined by the following calculation. A 5% linear scale with W - 1 ft (0.3 m), u - 1 ft (0.3 m), L - 6 in (0.15 m) and, the same value of 1 would yield an a value of 99.8%. Therefore, the small scale mockup tends to overpredict the de-entrainment compared to the full scale (96.4%). However, the overprediction is only a few percent. It is noted that the 99.8% o va. sue in the 5% linear scale corresponds to - 0.2% entrainment which is close to the low DCH efficiency measured in the FAI DCH experiments. Although, a large-fraction sweepout (as mentioned in Section 2.2.6.1) be expected in full scale, most of the swept-out mass will fail to make can the turn at e entrance into the lower compartment, and will most likely be de-entrained onto the floor of the cavity. This model is combined with the (~'} entrainment fraction model to develop a methodology for relating scaled ~' experiments to a specific plant. This is discussed in Section 3. 2.2.7.3 Time of Flight Scalinz One of the issues to be addressed for scaled experiments is the in-fluence of the " time of flight" for airborne debris which could be exchanging heat with the local atmosphere or oxidizing. Fractional scaled experiments have shorter lengths, which means that the airborne debris is in flight for a shorter time in the tests than would be the case in the reactor accident sequence. This is an inherent shortcoming of scaled experiments and must be addressed when evaluating the likely plant response. As discussed previously, the issues of particular note are that all debris cannot be entrained and that all the entrained material cannot remain entrained because of the influence of structure in de-entraining particu. late. Therefore. the entrained fraction may only encounter a small fraction ['") of the containment gases as the debris particles traverse the reactor V i i 1

~ 2 35 cavity,. instrument tunnel and lower compartment. To address this for the reactor system, we will evaluate the energy transfer from the fraction of ' debris dispersed. into the atmosphere by assuming that the debris' equi.. librates with the containment atmosphere. This eliminates any consideration with respect to the time of flight. The influence of cladding oxidation and l steam j hydrogen combustion will be assessed by considering the potential for inerting as a consequence of the accident sequence. Also we will perform- 'f analyses assuming the hydrogen burns completely. Again this removes any sensitivity to details of the combustion behavior. In the scaling analysis, the instrument tunnel and seal table are the only structures considered in-removing discrete material from the high velocity gas stream. While other structures would certainly be influential, the consideration of the instrument tunnel and seal table is-sufficient to show that the resultant calculated pressurization is much less than that required to challenge the containment integrity. 2.3 Analyseg .O i The direct containment heating issue was first. discussed as sensitivity calculations in the Zion Study [ Ceco, 1981], Since then a number of-ef-forts, from simple modeling to complex computer codes, have been directed at ) gaantifying the magnitude of direct containment heating. These works in-clude Sandia's DCH preliminary calculations, FAI's DCH model and computer codes such as HARDCORE, PARSEC, KivaDCH, and CONTAIN. These will be sum-marized below. 2.3.1 Zion Probabilistic Safety Study The Zion Study [ CECO, 1981) considered a wide variety of ex-vessel interaction scenarios. of which the high pressure melt ejection was notable because of the conclusions regarding debris dispersal. Three phenomena were identified which could eject debris from the cavity should core slump and vessel failure occur. The mechanisms include (1) hydraulic jump at the j keyway slope, (2) wave formation during the steam / hydrogen blowdown, and (3) entrainment of the debris by high velocity gases. The first two phenomena

= - i 2, ./7 \\ Il provide for a debrisLconfiguration which could directly impede the progress of the gas mixture exiting the vessel, resulting in debris sweepout of the cavity. The third phenomenon recognizes the fact that the~ relatively high gas velocities could entrain, levitate and accelerate.large particles co the cavity exit. Since the adjacent lower containment compartment had substan-tial structure to capture and redirect debris and water to the containment floor, debris' ejected from the cavity was anticipated to be ~ deposited and-l quenched thereafter. Aerosol ' formation and direct heat transfer to the -i f lower compartment atmosphere was not considered, although sensitivity cal-culations included in the study did consider the effect of direct heating. The best estimate evaluation attributed the containment pressurization to be dominated by the primary-system blowdown and steam formation from debris I quenching. 2.3.2 Sandia's Pre 1Lainary Calculation t Sandia National Laboratory has performed parametric scoping calcula-tions for the effects of direct containment heating in a sample problem }/~') N-s' [Pilch, 1986]. These calculations were taken up in two parts, steady-state calculations of pressurization by simple energy balances, and rate-dependent energy release assessments. The parametric steady-state predictions are energy balance calculations in which an energy inventory is determined for assumed debris constituents at a given temperature, including contributions from oxidation of circonium,- iron, and uranium dioxide (to U 0 ). The fraction of dabris required to 38 participate in direct heating (as aerosol) to cause containment failure by overpressure was presented as a function of mass expelled from the cavity. The calculations show that at least 101,200 lb (46,000 kg) would be required for containment failure assuming 100% DCH efficiency. The rate-dependent calculations only evaluated the correlations for the Nusselt and Sherwood numbers to predict heat and mass transfer from par. ticles of given diameters. Such calculations presume the processes to generate and maintain a finely divided aerosol and neglect issues related to rate of entrainment nd de-entrainment due to structural barriers. Because 6 .m

,2-37 i i 1 t O .of theisubstantial limitations in the parametric calculations, these-can j only be used to assess. the minimum debris quantities required to challenge containment integrity. Hence, these will only be used to compare the mini-mum mass-to that which could be' realistically anticipated. l .1 2.3.3 HARDCORE I I In parallel-to the-conduct of the DCH experiments as part of the CWTI. the HARDCORE and PARSEC codes were developed at Argonne National.

program, i

Laboratory HARDCORE analyzes debris sweepout, while PARSEC determines the extent of direct containment heating by. dispersed debris. ' HARDCORE is [ formulated using one dimensional, Eulerian, three-fluid equations and two-phase flow and heat transfer correlations. The code includes mechanistic modeling of' the entrainment race-of liquid droplets from films or layers, -l mass transfer-p crust-formation, correlations for torced convective heat and for spherical particles, debris cloud thermal radiation, droplet size, and I drag coefficients. Particle gas. interactions assume the debris particulate-single particle in the. gas stream and does not consider the influence ./N - is a of a dense particulate layer separating the gas flow stream'and the debris which has not been entrained. The model also does not include the' dispersal J l nof debris out of the reactor cavity by any means except entrainment. .The o for steam oxidation of corium metallic constituents such code does account as circonium, iron, and chromium and accompanying hydrogen generation. I Oxidation rates are set by whichever of the following processes is'ilmiting-transport of steam molecules to the droplet surface or diffusion of ions chrough the oxide shell assumed to form on each core debris droplet. l Given these assumptions, HARDCORE [Sienicki, 1986] calculates the size of the dispersed co;ium droplets and the swept-out debris mass. The cal-culations require the total core debris mass ejected from the reactor vessel, the time-dependent flowrate of blowdown gas, and temperature data as i inputs. HARDCORE satisfactorily reproduced the measured values for debris j sweepout and debris particle size for Tests CUTI-13 and SNL DCH-1.

However, i

i no comparisons with other data have been reported to date. HARDCORE does not treat particle de-entrainment due to structural barriers at the tunnel exit. Given the substantial experimental evidence for' other removal 1 i ,,i

2 38 processes and de entrainment, these processes must be incorporated into the 4 calculation before the model could be used to assess the Farley plant response to severe accident conditions. 2.3.4 PARSEC PARSEC {Sienicki, 1987] is a one dimensional, coupled Eulerian-Lagrangian code for solving particle-laden fluid flow problems 'with heat transfer and oxidation reactions. PARSEC calculates the heatup of a gas atmosphere in a closed containment resulting from (1) dispersal of_ high temperature debris droplets / particles, (2) heat transfer, and (3) oxidation of reactive debris constituents with oxygen or steam to-produce hydrogen. PARSEC accounts for-the formation of aerosols by the oxidation-enhanced vaporization of metal from the surfaces of the core debris droplets. but neglects the decrease in the mass of an individual droplet. A debris cloud, thermal radiation model accounts for back scattering by debris particles into the cloud and has been shown to be significant. Othe rwis e, radiation heat loss to the structures could be overpredicted resulting in less atmos-O pheric heatup. Outputs from HARDCORE such as swept-out mass, dispersed- -debris. size distribution, entering debris temperature, entering gas pressure and-temperature are used as inputs to PARSEC and makes the resulting model limited because of the limitations to HARDCORE discussed above. PARSEC calculations reportedly showed good agreement with CVTI-13 and SNL DCH-1 tests where impeding structures were not present,

however, no other com-parisons with data have been reported to date. 'Due to its one-dimensional nature PARSEC cannot account for internal structural effects mechanisti-cally.

Until models representing other modes of debris dispersal and the influence of de-entrainment are ine '.ude d, the model cannot be used to evaluate the response of the Farley plant to severe accident conditions. 2.3.5 Kiva-DCH Kiva-DCH [ Marx, 1989), which is similar to PARSEC, calculates atmos-pheric heatup due to the release of molten debris into a container.

However, in contrast to the one-dimensional PARSEC code, Kiva-DCH simulates O

2-39 o) / (_,/ the processes in two or three dimensional geometry. Kiva-DCH is a modifica-tion of the Kiva computer code which was developed at Los Alamos National Laboratory to simulate spray transport and combustion processes in internal combustion engines. The similarity between the spray and debris droplets allows a straightforward modification for direct containment heating ap. plications. Kiva-DCH utilizes a three dimensional finite difference scheme for gas flow calculations, coupled with a Lagrangian particle transport algorithm. The debris-vall interaction is treated by assigning a probabil-ity for trapping particles on the walls, and for particles dripping off the walls. Kiva-DCH simulates the transport of the debris through the gas and evaluates radiative and convective heat transfer effects, and chemical reaction of the debris. Theoretically, due to its multidimensional fea-

tures, Kiva-DCH could be used to simulate internal structure effects on debris transport of any containment design.

However, since Kiva-DCH repre-sents an effort to simulate virtually every individual DCH process, using the most sophisticated turbulent flow model in multi-dimensions, it requires ~~ \\ extensive computational resources. Kiva-DCH simulations of SNL DCH-1 and DCH 3 tests showed modest differences between calculations and experimental obs e rvations.

However, like PARSEC, until the code includes models for processes which compete with entrainment for debris removal from the reactor cavity, and until it includes a representation for de-entrainment, the model cannot be used to evaluate the plant response.

2.3.6 CONTAIN CONTAIN [Bergeron, 1986], developed at Sandia National Laboratory, is a l lumped-parameter code for studies of the full scale reactor containment. Many of the models in CONTAIN have been imported directly from other codes. The DCH model implemented in CONTAIN assumed the mass particulated and the debris size (user specified) and assesses the transport of finely dispersed debris with the blowdown gas, heat transfer among the gas, debris, valls, removal or trapping of the debris as it is transported, and chemical reac-tions between unoxidized metals and oxygen or steam. The DCH model f-implemented in CONTAIN is less mechanistic than the one-dimensional HARDCORE k

i'; 2a40 i i -\\ and PARSEC codes and suffers from the same limitation as these codes with respect to assessing the response of the plant.to postulated severe accident conditions, f 2.3.7 Stan11fied Model for Reoresentinz [ f Comparable Debris Disnersal Processes i 1 A closed form model has been developed [ Henry,1989] for estimating the ' f character of the debris mass dispersed out of the cavity / tunnel by the f r combination of entrainment and displacement (Appendix A). The thesis for f this model is that the entrainment of the debris and its transport into the gas stream (where it would be accelerated), would cause the gas to be j decelerated due to conservation of momentum. With a decreased gas velocity and a constant delivery rate (choked flow) from the RPV, the reactor cavity i would begin to pressurize uhich would enhance debris removal due to the j pressure difference. Entrainment rates are represented by the Ricou-l Spalding correlation [Ricou-Spalding, 1961] and the entrained debris is assumed to equilibrate with the gas thereby decreasing the gas velocity. j i \\%d Debris displacement is evaluated ~by imposing the calculated pressure dif-ference on a debris layer with a characteristic length equal to that of the j reactor cavity floor. This is shown as dimension L in Figure 2-6. To f escape the reactor cavity / instrument ~ tunnel, the debris layer must be dis. placed over the distance Lp shown in Figure 2-6. The time required for this displacement determines the interval for debris entrainment and the ratio of I the entrained mass to the total mass represents the fraction of the' debris which could be finely particulated. An estimate of the debris size'for the particulated fraction is obtained through a Veber number stability criterion based upon the initial (highest) gas velocity through the reactor cavity. For the condition of interest, this results in particle sizes in the range of 100 pm. The model, which was originally formulated to evaluate fission product releases for HPME and does therefore not include a de-entrainment assess-

ment, was compared to the SNL Surtsey tests which did not represent the containment structures outside of the reactor cavity.

Table 2-4 illustrates the comparison of the predicted and measured behavior, and the predictions i 1

2-41 O Pv A[ c CORE DEBRIS A, _/ i P, f Ue 4 -- - --- O if V/H/HHH/HHHHHHHH/H/H/H/HHH/HHHb L = = L p REH.880814 A.A Figure 2-6 Characteristic debris length and displacement length for debris movement. O

= 2-42 a i in good agreement with the observations for the tests using 80 kg (DCH-are ' 2, 3 and 4). Test DCH 1 only used 20 kg and more than half of the debris froze as a thin film on the cavity walls. Freezing was not included-in the j closed form model, but if we assume that the first 11 kg freezes on the cold- [ walls and' insulates the walls for the remaining 9 kg, which could then be entrained, the model predicts that all of the 9 kg would be entrained due to the blowdown. This is consistent with' experimental observation. Another facet of the model is that substantial pressurization of-the reactor cavity could occur as a result of the debris entrainment. This could lead to a critical (choked) discharge of the debris gas mixture, which would cause the mixture to expand rapidly anc' N11 the cross section of the Surtsey vessel, While a qualitative argument, .c is consistent with the experimental observations. As mentioned above, this model focused on 'i:G-f sessing the fraction of melt which could be particulated in the reactor cavity and hence did not include a de-entrainment model. To assess the behavior of a specific plant, this model for the fraction particulated must be combined with the de-entrainment fraction for the seal table proposed in

i i

(Walker 1987). 2.3.8 Jet Breakuo Model i A model has been presented for initial jet breakup between the vessel exit and impact with the cavity floor {Frid, 1986). This model quantifies + the effect of effervescence of dissolved gases from the debris through an assumed bubble nucleation density in the jet. Bubble growth at these nucleation sites is tracked, and the jet is assumed to break up when a void fraction of 50% is attained. It is concluded that. jet breakup could occur within a few diameters of the reactor vessel, generating particles in the 30 150 micron range. However, when the debris-gas flow encounters the-cavity floor and changes direction by 90*, the debris would be expected to be redeposited as a molten layer. Therefore, this aspect was not included in the IPE. O

3-1 (m U) 3.0 ME310D011X7Y Much like design basis analyses, if reactor vessel failure were to occur at high pressure, the most important consequence would be the increase in containment pressure. Direct containment heating by core debris expelled at vessel failure has been postulated to be a significant contributor to the containment pressure increase. This section provides a method for estimat-ing the energy released to the containment by the core debris which could be discharged immediately after vessel failure. The resulting pressure in-crease should then be part of the assessment for the likelihood of containment failure by overpressurization immediately following vessel failure. The methodology described below is somewhat conservative (i.e. it overestimates the containment pressurization). If the calculated pressure loadings are far less than those which would challenge the containment capability, failure by this means should be assessed as impossible. [-] First, the extent of debris entrainment which could occur in the reac-V tor cavity / instrument tunnel is determined. Next, we assess the fraction of entrained debris which would escape the directional change at the seal table.

Then, we address the effect of hydrogen combustion that possibly accompanies DCH.

Finally, the potential for containment pressurization is assessed and compared to the containment capability to characterize the likelihood of containment failure due to overpressurization. 3.1 Mass of Debris Distributed Into the containment Atmosphere The methodology focuses on (1) the debris mass that could potentially be particulated in the reactor cavity and instrument tunnel and (2) that fraction of the entrained (particulated) debris which could escape the change in flow direction caused by the seal table. (The seal table does not represent all of the debris-structure interactions. Hence, the analysis is inherently conservative.) That fraction of the debris which is not particu-laced would have such a large characteristic dimension that the seal table would collect all of the debris and prevent it from entering the containment atmosphere. V

3-2 b A limitation exists on the extent of entrainment which could be af-fected by the RCS blowdown. The entrainment mass is not dependent upon the mass discharged from the RCS, except to the extent that the discharged mass should equal, or exceed, the entrained mass. The assessment of the particle size utilizes the conservative assessment described in Appendix A based on the oaxinum gas velocity in the reactor cavity and a single droplet Weber nuroer criterion. 3.1.1 Extent of Entrainment of debris which could be entrained, according to Equation 20 The mass of Appendix A, is proportional to the product of the square root of RCS pressure and the size of the breach in the RPV, i.e. the larger this product, the greater the mass which could be entrained. To evaluate the maximum mass which could be particulated, we may consider the pressure to be the pressurizer safety valve lowest set point of approximately 17 MPa (2475 psia) If we consider the initial vessel failure to be an instrument tube, the methodology presented in (Ceco, 1981 and IDCOR, 1983] would suggest a vessel breach of about 0.3 m (12 in) in diameter at the time of gaseous blowdown. Using the combination of this failure size and the RCS pressure, in the reactor cavity can be obtained from the calculated entrained mass Equation (20) of Appendix A. Particulation of the melt by the RCS gaseous blowdown would result in a nominal debris size determined by Equation (21) of Appendix A with a critical Weber number equal to 12. For the small LOCA sequences, the RCS pressure would be less, resulting in a small RPV failure size (for the same debris mass). Both of these would reduce the mass of debris which could be entrained. In addition, gas velocity in the cavity f would be reduced causing the average debris particulate size to be larger. Therefore, the conditions used for the station blackout like scenario bound the parameters of interest. 1 This evaluation of the mass of particulated debris is determined by the gas kinetic energy. As such, it is not a function of the debris mass dis-charged from the RPV, except for the trivial point that the entrained mass is the minimum of the calculated value and the mass discharged from the vessel. Assessing the particle size based on a single particle in the gas

3-3 (-- free stream underestimates the particle size. Hence. the following section '~ will overestimate the fraction of particulate melt which could remain in the gas stream despite the directional change at the seal table. Since the melt is assumed to equilibrate with the containment gases, this results in an overestimate of the pressurization due to DCH. 3.1.2 De-Entrainment Application of Equation (2-1) with a debris size calculated in a pre-vious step will result in a prediction that a fraction of the particulated melt would be captured by the structure. Comparing this to the total debris mass results in a melt fraction discharged to the atmosphere. The methodology conservatively calculates a particle size which is less than the observed value in the available experiments. Hence, the impact of the assumption that the droplet size is determined by the Weber number criterion based on the free stream velocity may also be evaluated for a (} larger particle size as a sensitivity study. 3.2 DCH Pressure Rise Debris of sufficiently small size to be swept past the seal table and remain airborne in the containment atmosphere is assumed to thermally equi-librate with the containment atmosphere. With this assumption, issues with respect to " time of flight" (extent of energy transfer) are included in the manner. This is an accepted methodology in a conservative (overestimated) l conservatism in the model. The method for ca ;ulating the containment pressure due to DCH is illustrated in Appendix B. 3.3 Likelihood of Hydroren Combustion The particular sequence conditions analyzed are typical of those for a l Station Blackout-like scenaric. MAAP calculations for such a sequence may /N be performed to calculate the pressure in the containment prior to RPV U

34 s i failure. Most of the hydrogen generated by cladding oxidation would still be in the RCS prior to vessel failure and would enter the containment during the gaseous blowdown; i.e. after the debris discharge has occurred. The } anticipated blowdown would be a few tens of seconds long. This is important since the initial debris discharged from the RPV would contact and vaporize i i the. sm.111 water mass accumulated due to condensation on the reactor cavity valls and the initial debris discharged from the instrument tunnel could contact water on the containment floor resulting in substantial steam gener-ation. Moreover, the RCS blowdown may involve the flashing of water in tha-RPV lower plenum. All of these would immediately displace oxy 5en from the lower compartment and add to the steam inventory in the containment atmos-t phere at a greater rate than hydrogen would be discharged from the RPV. More importantly, the steam concentration in the region of debris dispersal would be very high Consequently it is doubtful that the hydrogen-steam-air environment would be combustible for this sequence. Hydrogen combustion [ limits are discussed in Appendix C. 3.4 The Influence of Uncertainties For the containment integrity to be threatened by a DCH event, all of the following phenomena must occur. i 1. Core melt progression must occur in a manner where most of the i core is molten and slu=ps to the lower plenum such that it could be discharged immediately following RPV failure. 2. RPV failure must occur in a manner that would allow a large mass of core debris to be discharged into the reactor cavity before RPV blowdown would be completed. 3. Once the melt is discharged from the RPV to the reactor i cavity, the RCS blowdown must be sufficient to particulate and entrain moi.L. or all, of the melt. 4. If sufficien melt is finely particulated, it must remain in this beate as it flows out of the reactor cavity instrument tunnel and into the lower compartment. 5. Burning of the hydrogen in the containment following debris particulation and dispersal must occur before containment integrity can be challenged. \\

3-5 l' k Should any one of these phenomena not occur to a sufficient degree, the containment would not be threatened. Each of the major elements that must g a manner which over-be available for a DCH event are discussed below in states the potential for debris fragmentation, dispersal, and heating of the atmosphere. Realistic uncertainties would decrease each of these such that more realistic models result in lower calculated pressures than were calcu-laced by the methodology. The best estimate assessment is that none of these phenomena would occur except for the second and then only for selected l accident scenarios. 1 3.4.1 Core Melt Prorression? core material would For a large dry containment virtually all of the need to be available for fine particulation and dispersal before containment integrity could be threatened. The current assessment provided by the MAAP l codes allows only fully molten material to slump into the lower plenum and i be available for discharge following RPV failure. Should the core melt progression also-result in a substantial amount of solid material being discharged to the lower plenum the material discharge rate could be substan-tially limited as a result of " clogging" by_the solid debris. For the analyses presented in this document, this uncertainty has been set aside by assuming that the solid material would not be available for limiting the melt discharge and has used melt fractions well in excess of those that are calculated by the PWR MAAP code. Any deviation from this characterization would decrease the potential for melt discharge prior to RCS blowdown. I Another limitation caused by solidified debris would be a reduction in the melt velocity and therefore a reduction in the rate-of ablation for the vessel wall.

Hence, the failure size associated with a given melt mass would be decreased thereby reducing the potential for dispersion of the melt from the reactor cavity by the RCS blowdown.

Deviations from the core melt i progression processes represented in this analysis would decrease the poten-i tial for DCH. l

i l 36 4 (N \\~- 3.4.2 RPV Failure Mechanism? Debris drainage into the lower plenum is the specific condition of 1 concern for RFV integrity. This is of interest with respect to vessel failure immediately after debris transport into the lower plenum (i.e., failure of the in-core instrument penetrations) and also with respect to the long term behavior should the penetrations retain their integrity (as was the case in the TMI-2 accident). The former leads to the type of analyses carried out in this assessment where debris could be discharged from the RPV and then dispersed by the RCS blowdown. In the latter case, the likely failure mechanism would be at or near the top of the debris pool and would be due to creep rupture. Given the strong temperature dependence of the creep rupture phenomenon. the anticipated failure mechanism would be a localized failure (blowout of the vessel wall) followed by RCS depressuriza-tion due to gas blowdown. In this type of failure, most of the debris would remain in the RPV, and there would be no potential for a DCH type behavior. In the present analysis, uncertainties of this nature have been set aside by assuming that the failure would be localized and at the bottom of the ves-O' sel. Deviations from this behavior would decrease the potential for both a high pressure melt ejection and DCH. 3.4.3 Extent of Debris Particulation? Once debris is discharged from the reactor vessel and accumulated in the reactor cavity, the RCS blowdown represents the potential for debris particulation. The action of particulation would also impact the ability of the gas flow to sustain the entrainment process, as discussed in Appendix A. This debris-gas interaction is used in this assessment to de te rmine the response of the reactor cavity under worst case conditions, i.e. full RCS pressure vessel failure at the bottom of the RPV and substantial melt ac-cumulation within the reactor cavity. Application of this methodology to available experiments shows agreement with the extent of debris particula-tion and, if anything, tends to overestimate the mass of material which could be finely particulated. While this is considered to be one of the cwatrolling processes in the debris particulation and dispersal, other f results presented by [Cinsberg and Tutu, 1988] assumed more melt was finely i

3-7 (m(_-)' particulated, with no mechanistic assessment of the debris behavior within i the reactor cavity. Deviations from the best estimate approach, such as that obtained by.using the assumptions of [Ginsberg and Tutu, 1988), would increase the

ntainment loading.

However, large deviations from the mechanistic approach would be required before the pressure would approach a value sufficient to threaten the containment integrity.

Such large devia-tions are not considered to be realistic, particularly when considering the influence of structures as discussed below. 3.4.4 Influence of Structures? Numerous experiments have demonstrated the extensive role of structures in limiting the mass of material which could remain airborne during the debris dispersal. The approach taken uses published results for the struc-tures which require an assessment of the debris particle size. In carrying out this part of the evaluation, the particle size was determined by the Veber number stability criteria based on the maximum gas velocity in the g'"% reactor cavity. A comparison of this particle size to the data in the \\~- Sandia SURTSEY DCH tests show that such a characterization tends to underes-timate the average particle size. As a result, the influence of the seal table structure is underestimated, i.e. the mass of material " trapped" by the structure is understated which therefore overestimates the mass of finely particulated debris which could be distributed into the containment atmosphere. Hence, the major uncertainty associated with this behavior is that of the particle size and these analyses used a size which would overes-timate the mass distributed into the containment atmosphere. In addition, there are other structures in the containment compartments which would be effective in removing particulated debris from the contain-ment atmosphere. None of these were credited in the analysis. Neglecting these structures also tends to overestimate the heat transfer to the gas atmosphere and the pressurization that could result by directly heating the containment atmosphere. d

38 V,0% 3.4.5 Potential for Hydronen Combustio.D7 The potential for hydrogen combustion is greatly influenced by steam in the containment atmosphere. Calculation results that assume complete hydrogen combustion in addition to direct heating of the atmosphere by-the airborne debris mass should be performed to see whether this will result in a pressure sufficient to challenge the containment integrity. This is an additional conservatism with respect to the uncertainties of these physical processes. O 1 0 O

m._ _. 41 t .O\\s / 4.0 PLANT SPECIFIC APPLICATION This section describes the application of the methodology of Section 3 to the Farley Nuclear Plant. Note that the Farley design is not conducive to flooding the reactor cavity, so that no appreciable amount of water would be present there at the time of vessel failure. j 4.1 DCH Calculations The only si nificant flow path for core debris to exit the Farley 6 cavity is through the instrumentation tunnel. The annular region between the reactor vessel and the biological shield does not provide a significant path for debris dispersal because it is mostly occupied by the reactor ve1sel insulation. Any deformation of this insulacion, due to cavity pres-surization at vessel failure, would only plug this region further.. The and into the cavity ventilation duct passes through the lower compartment flow path from the cavity, thereby reducing cavity. This provides a vent both the gas velocity available for entrainment and the drivin8 Pressure for debris dispersal. No significant mass of debris would turn 90' to follow the vent this gas vent path; rather, the debris inertia would carry it past opening toward the much larger flow opening provided by the instrument tunnel. The magnitude of DCH can be evaluated in a straightforward manner by the methodology described in Section 3. First, the debris mass entrained from within the cavity is determined from Equation (20) of Appendix A, Cavity dimensions and RPV failure conditions are parameters required in Equation (20). For clarity, the equation and values of required parameters specific to plant Farley are shown here. The entrained mass is calculated as: O

4-2 ,m A M 'p ' - 0.19 A P L Lp "D y y p D (4 1) - 41000 kg where: 2 RPV failure area - 0.07 m, A y P - RPV pressure prior to failure - 17 x 108 Pa, y combined length of cavity and instrumentation tunnel - 27.2 m, L p L - cavity floor length - 13.2 m, debris density - 7000 kg/m3, p D cavity gas density - 0.67 kg/m (obtained from Equation (24) of 3 p E Appendix A), 2 A, - cavity / tunnel floor area - 44 m, steam molecular weight - 18 kg/kg-moi, M y .) R - universal gas constant - 8314 J/kg-mol K, 17 T - RPV steam temperature - 625 K (saturation temperature at P y MPa). The nominal debris size (d) is related to the critical Weber number of 12 by Equation (21) of Appendix A: p" U3 d We - (4 2) where: We - critical Weber number - 12, relative vel city of the droplet to the gas stream, U r o - surface tension of corium to the cavity atmosphere - 1 N/m. O l l

43 (A U is conservatively over-estimated to be the velocity of gas in the r cavity. This velocity, defined as U,, is given by Equation (22) in Appendix A: 0.6 P 'A ' g7,1/2 y y (4-3) U , c, fw, 'c ^ where 6.2 flow area A* = minimum instrumentation tunnel flow area + vent a m, P - initial cavity pressure, assumed to be 0.3 MPa. e For the Farley plant, U - 206 m/s and the mean particle diameter is calcu-c lated from Equation (4-2) to be d - 422 pm. Equation (2-1) may now be used to determine the fraction of core material (a) that will be de-entrained at the 90' turn from the instrument tunnel to the lower compartment: 1h a (4-4) a-in 1+ 1-where - 0.075 m'1 (4 5) A-f CD 7- - 2.15 (4-6) P - 31 + y 2 - 2.37 (4-7) and the appropriate geomett ical values for Farley are V - 3.9 f t (1.2 m), w 13.25 ft (4.0 m), L - 8.4 ft (2.57 m), and 1 - 13.25 ft (4.0 m). Solving Equation (4 4) with Farley-specific values yields a - 0.91. Therefore, only 9% of the entrained debris mass is expected to make it past the 90' turn from the instrument tunnel to the lower compartment. This percentage cor-3690 kg. responds to a debris mass of 0.09 mD gU

u.a Q l / Direct containment heating by this amount of debris, assumed to be 100% efficient (i.e. the debris quickly reaches thermal equilibrium with the containment atmosphere), will increase the containment temperature and The extent of this heat-up and pressurization may be calculated pressure. by Equations (B 2) and (B-3) of Appendix B. Several plant specific data and initial conditions are required in the calculation of Equation (B-2). These are listed in Table 4-1. Corium temperature is assumed to be 2500 K at the time of reactor failure. Corium compositions as indicated in Table a-1 are equivalent to a mixture of a?.1 Zircaloy cladding, all UO fuel and a lower core plate made of stainless 2 steel. Stainless steel is assumed to be composed of 67% Fe, 21% Cr and 12% Ni. Containment conditions such as air mass, steam mass, and temperature are typical values calculated by MAAP for a station blackout sequence. These values are the same as those used in the Phenomenological Evaluation Summary on hydrogen combustion (Fauske 6 Associates, Inc., 1991). 40% of to reactor failure which is Zircaloy cladding is assumed oxidized prior O consistent with typical MAAP results for a station blackout sequence.) b DCH calculation results are also shown in Table 4 1. When it is as-sumed that all hydrogen generated before and after HPME is burned the resultant containment temperature and pressure are 793*F and 81.2 psia, respectively. Without hydrogen burn, the final containment temperature and pressure would be only 286*F and 54.1 psia, respectively. These conserva-tive estimates of the containment peak temperature and pressure due to DCH are well within the containment capabilities. Furthermore, it is very unlikely that hydrogen in the blowdown stream would burn during the RCS high pressure melt ejection. It is not feasible to form a combustible mixture locally in the cavity or in the lower compart-ment since the blowdown gas would result in the displacement of air out of the cavity. During the discharge of corium into the reactor cavity, the rapid steam expansion within the cavity would increase the steam molar fraction in the reactor cavity and lower compartment beyond 53% where hydrogen burns cannot occur. Hence, the sequence would not result in a combustible mixture locally. Under the conditions postulated in Table 4-1, Ol j

.~ 4-5 l Table 4-1 DCH CAlfUIATIONS WR FARIEY NUCLEAR PIANT ' Plant-Specific Data and Initial Conditions 76% UO + 16% Zr + 5% Fe + 2% Cr + 1%.Ni Corium composition (V)) 2 Corium temperature 4040*F (2500*K) Total corium mass 242.000 lbm (110,000 Kg). DCH participating debris mass 8120 lbm (3690 Kg) ( f, m ) D Hydrogen generated in core (m ) 701 lbm (318 Kg) (40% Zr oxidation) g Initial mass of air in contain-135,866 lbm (61,629 Kg) ment (m ) Initial containment temperature 279'F (410*K) (T ) Mass of steam in containment 132860 lbm (60391 Kg) prior to RPV failure (m,) Steam mass expelled from RPV 20733 lbs (9424 Kg) ("sps) 3 Primary system volume (V,) 8.127 ft3 (230 m ) p Primary system pressure (P,)/ 2,500 psia /666'F (17 MPa/625'K) p temperature (T,) p 8 Containment free volume (V + V,) 2 x 108 ft8 (56,600 m ) c c Extent of chemical reaction with 50% steam (f ) Extent of chemical reaction with 31% oxygen (fj) Temocrature and Pressure Results 0% Hydrocen Burninz 100% Hydrocen Burninz Final containment temperature 286*F (414*K) 793*F (696*K) Final containment pressure 54.1 psia (0.4 MPa) 81.2 psia (0.56 MPa) ] Pressure rise 6.3 psi (0.04 MPa) 33.4 psi (0.23 MPa)

i 4-6 i s\\ steam colar fraction in a globally mixed atmosphere would be 57%. This mixture (according to Appendix C) is not combustible. Hence, the likelihood of simultaneous occurrence of both DCH and hydrogen burn is not feasible. We, therefore, conclude that: 1. hydrogen combustion is not considered feasible in the best estimate analysis, and 2. even if a burn la postulated, containment integrity would not be challenged. The fact that the assessment is tolerant of this conservatism adds to the robustness of the conclusion. Finally, the results of the FAl-DCH experiments are also applicable to the Farley IPE. Figure 4 1 shows the Farley and Zion cavity configurations. Core debris leaving the Farley reactor vessel will have to make two 90 degree turns before exiting the instrument tunnel to the lower compartment. Debris leaving the Zion cavity must only go up to 64 degree incline and make a 90 degree turn to enter the lower compartment. Clearly, the Farley con-figuration will trap and de-entrain more debris than in the Zion configuration. Therefore, the results of these experiments are a conserva-tive estimate of what they might be for an analogous Farley DCH experiment. 4.2 Conclusions Of principal interest for the DCH issue are the processes of debris entrainment and particulation in the reactor cavity and de-entrainment of the dense discrete phase by structures. Both of these processes were modeled and the models were found to be in agreement with available experi-ments, includin5 the St linear scaled tests performed in support of the Zion IPE. Pressure increases were assessed for the debris thermally equilibrat-ing with the containment atmosphere and also for hydrogen burning and equilibration. Neither of these conse rvative assessments were found to cause pressure increases which wculd approach containment threatening levels. i ) i

47 y Farley Cavity Configuration (V seal Table [_,. :.t Debris 6} cxit f vent opening a 1 db Reactor Vessel e -{., L-zu+:r

3. ;=.rn:..

%:p;. e. qe . >y.y L ) r .N Z - L.... Zion Cavity Configuration Seal Table g. .f ^ 33s f,$ Debris Q E xit Reactor 'h Vessel I.E"j i m.,.. C II67 '.1

k d:.

. ; " :.s;e... :.. }.?..v.,?.Q a w..f .s.eu1..c,. a e,- -n.*- '

• I'.c.4 [ '.L.

. - a =. 4.$ 0f...-$ *,M*" , f. 1d, o:w*.a. Y ** . t- .s ..s..:' oa.;p a s.g *..c. ::;q.~..,.. i; ..-.r. e. -;X3-g s 4.;y'.'!'I.1.G.;:.:.U tFt.~. #';FA. rs

.w.. e.a..

=.

w..y

.m i 1 i Q:.lt 9:c. '.i E$.. _s;-

t.. _ _ _ i

% Acove Figures are in accrossmate S:sie witn Each Otner Figure 4-1 Comparison of Farley and Zion cavity configuration.

48 f3 '\\ ') The potential for hydrogen combustion was also evaluated. It was found that the combination of steam released from the RCS as the accident progresses and the rapid steam generation caused when high temperature debris encountered water in the containment, would result in conditions sufficient to completely inert the containment against a hydrogen burn. These evaluations clearly show that DCH related behavior would not challenge the Farley containment integrity. -Therefore, the plant response trees need not include DCH as a top event. <a' s./ I \\

i 5-1 N \\ 5.0

SUMMARY

Direct containment heating is a phenomenon in which, during a severe accident, hot core debris (approximately 4040*F) is ejected from the reactor

vessel, entrained in high velocity gasses from the subsequent blowdown and, dispersed into the upper compartment of the containment building in the form of small particles. These small particles can ideally remain airborne long enough to transfer most of their energy to the containment atmosphere and cause an increase in containment pressure. This paper presents experimental data and bounding calculations which indicate that the Farley containment is not vulnerable to the direct containment heating phenomenon.

DCH experiments for a large dry Pk'R (Zion) have shown that structure in the lower compartment and the seal table region of the cavity compartment de-entrain almost all the small debris particles that could cause direct containment heating. In the Argonne Cb'TI experiments, tests performed with a dry cavity compartment and seal table structures present showed that only 1-5% of the material that left the cavity contributed its energy directly to the air of the containment building. The test performed without the seal table structure had 62% of the material that left the cavity contribute directly to the heating of the air in the simulated containment building. Other experiments explored this idea further by including, in addition to the seal table, realistic lower compartment structures and showing that they would also capture debris particles. Experiments were also performed to assess the danger of hydrogen burn-ing occurring at the same time as the direct containment heating phenomenon. Results showed that hydrogen burning cannot occur if the mole fraction of steam in the containment atmosphere is above 53%. In a TMLB with seal LOCA type accident, the only type of accident in which DCH could possibly be a

threat, we expect the mole fraction of steam in containment to be ap-proximately 674 just before primary system blowdown.

Therefore, hydrogen burning and DCH are not expected to occur together.

5-2 rx i s Calculations show that for the minimum expected particle size, 422 pm, 91% of the entrained particles will fail to make the turn into the lower compartment at Farley's seal table structure. If all of the entrained particles have the minimum expected size then only 9% of the core material could be dispersed into the containment atmosphere assuming lower compart-ment structures are ignored. Even with 100% hydrogen burn, this corresponds to a pressure increase of only 33 psi and a final containment pressure of 81 psia. Since the ultimate failure pressure of Farley's containment building is about 117 psia, the experiments and bounding calculations show that the Farley containment is under no threat due to the direct containment heating phenomenon. Therefore, the Farley plant response trees need not include a node for containment failure due to DCH. tO v O

61 1 'l l ~

6.0 REFERENCES

Benedick, W. B., et al., 1984, " Combustion of Hydrogen: Air Mixtures in the l VCES Cylindrical Tank", NUREG/CR-3273, SAND 83-1022.

Bergeron, K.,.

et al., 1986, " Development and Applications of the Interim Direct Containment Heating Model for.the CONTAIN-Computer Code", Transactions of the. Fourteenth Water Reactor -Safety.Information Meeting, FUREG/CP-0081, Gaithersburg, MD. Commonwealth Edison _. Company (CECO), 1981, Zion Probabilistic Safety Study, Chicago, Illinois. Consolidated Edison (Con.Ed.) and the Power Authority of the State' of New i York (PASNY), 1982, Indian Point Probabilistic Safety Study, New York, NY. Fauske 6 Associates. Inc.,

1990, "FAI/ CECO Direct Containment ' Heating Experiments for a Zion-like Geometry", FAI/90 60, report submitted to Commonwealth Edison Company.

Fauske & Associates, Inc., 1991, "Phenomenological Evaluation Summary on the Probability and Consequences of Deflagration and Detonation of Hydrogen", FAI/91-58, submitted to SNC. i Frid, W. E., 1986, " Behavior of a Corium Jet in High Pressure Melt Ej ection from. a Reactor Pressure Vessel", Proc. of'the International ANS/ ENS Topical Meeting on Thermal Reactor Safety, San Diego, CA.

Cinsberg, T.,

and Tutu, N. K., 1988, " Progress in Understanding of Direct Containment Heating Phenomena in pressurized Light Water Reactors", Invited Paper Presented as the Third International Topical Meg. on Nuclear Power Plant Thermal Hydraulics and Operations, South Korea.

Henry, R.

E., 1989, "An Evaluation of Fission Product Release Rates During Debris - Dispersal". Proc. cf the ANS/ ENS Intl. Topical Mtg. on Probability, Reliability and Safety Assessment. Vol. 1, pp. 375-383 Henry, R. E., et al., 1991, " Cooling of Debris Within the Reactor Vessel Lower Head", Invited paper presented at the ANS Annual Mtg., Orlando, FL. IDCOR Technical Report 15.2, 1983, " Debris Coolability, Vessel Penetration, and Debris Dispersal". i

Marshall, B.

W...

1986,

" Hydrogen: Air: Steam Flammability Limits. and Combustion Characteristics in the FITS Vessel" NUREG/CR 3468, SAND 84-0383. i O e.-,

6e2 ss/. Marx, K. D., 1989, "A Computer Model for the Transport and Chemical Reaction of Debris in Direct Containment Heating Experiments", Nuclear Sci. 6 Eng., 102, 391-407. NRC (1988) letter to All. Licensees Holding Operating ' Licenses and Construction Permits for Nuclear' Power Facilities, " Individual Plant Examination for Severe Accident ' Vulnerabilities .10 CFR 50. 54( f)", Generic Letter No. 88-20.

Pilch, M.

and Tarbell, W. W., 1986, " Preliminary ~ Calculations on Direct Heating of a Containment Atmosphere by Airborne Core Debris",. SAND 85-2439, NUREG/CR 4455, Sandia National Laboratories. l Ricou, F. B. and Spalding, D. B.,

1961,

" Measurements of Entrainment. by Axisymmetrical Turbulent Jets", Journal of Fluid Mechanics, 11, pp. 21 32. Sienicki. J. J. and Spencer, B. W., 1986, "Corium Droplet Size in Direct Containment Heating",-ANS Trans...V.53, pp. 557 558.

Sienicki, J.

J. and Spencer, B. W., 1987, "The PARSEC Computer Code for Analysis of Direct Containment Heating By Dispersed Debris",- AIChE. .Symp. Series, 257, V, 83,.pp 355-362. Spencer, B. W., et al., 1983a, "Corium/ Water Dispersal Phenomena' in Ex-i Vessel Cavity Interactions", Paper TS-15.5, proc.. Int'l. Mtg. on UVR Severe Accident Evaluation, Vol. 2, Cambridge, MA. (

Spencer, B.

W., et al., 1983b, " Overview and Recent Results of ANL/EPRI Corium/ Water Thermal Interaction Investigations", Paper TS-15.2, Proc. Int'l. Meg. on LWR Severe-Accident Evaluation, Vol. 2, Cambridge, MA. Spencer, B. W., et al., 1987, " Hydrodynamics and Heat Transfer Aspects of Corium Water Interactions",.EPRI NP-5127, Argonne National Laboratory. Spencer, B. W.,-et al., 1988, "Results of EPRI/ANL DCH Investigations and Model Development", ANS/ ENS Conference on Thermal Reactor Safety, Avignon. France.

Tarbell, W.

W., et al., 1984, "High Pressure Melt Streaming (HIPS) Program Plan", SAND 82-2477, NUREG/CR-3025, Sandia National Laboratories.

Tarbell, W.

W., et al., - 1986a, " Melt Expulsion and Direct Containment' Heating in Realistic Plant Geometrics", Proc. of.the International ANS/ ENS Topical Meeting on Thermal Reactor Safety, San Diego, California.

Tarbell, W.

W., et al.,

1986b,

" Pressurized Melt Ejection into Scaled i Reactor Cavities", SAND 86 0153, NUREC/CR 4512, Sandia National Laboratories. Tarbell, W. W., et al., 1987a, "DCH Experiments and Analyses at Sandia National Laboratories", Containment Loads and Molten Core Containment ( Expert Opinion Meeting, Albuquerque, New Mexico. m-... ..,,m.. ,.v}}