ML20070H832

Assessment of Unprotected Loss of Flow Accident at End of Cycle-4 in Crbr Heterogeneous Core Design

ML20070H832
Person / Time
Site:	Clinch River
Issue date:	12/31/1982
From:	Cahalan J, Dunn F, Gruber E ARGONNE NATIONAL LABORATORY
To:
Shared Package
ML20070H826	List: ML20070H821 ML20070H832 ML20070H852 ML20070H869
References
NUDOCS 8212270250
Download: ML20070H832 (87)

Text

{{#Wiki_filter:. i i l An Assessment of the Unprotected LOF Accident at EOC-4 in the CRBRP Heterogeneous Core Design 1 by J. E. Cahalan F. E. Dunn E. E. Gruber, J. M. Kramer E. E. Morri s D. P. Weber H. U. Wider December 1982 Reactor Analysis and Safety Division Argonne National Laboratory 9700 South Cass Avenue Argonne, Illinois 60439 I 'E212270250' 821223 ~ ~ PDR ADOCK 05000537 _A P D.7

I 11 Ar. Assessment of the Unprotected LCF Accident in the CRBRP Heterogeneous Core Design by J. E. Cahalan F. E. Dunn E. E. Gruber J. M. Kramer E. E. Morris D. P. Aeber ' H. U. Wider ABSTRACT The hypothetical unprotected loss-of-flow (LCF) accident for the Clinch River Breeder Reactor (CRBR) with a heterogeneous core design has been investigated with the SAS30 whole core accident analysis computer code. The representation of critical phenomenology with experimentally validated models has played an essential role in this best estimate analysis of the LCF scenario. Fuel motion has been modeled consistently witn the TREAT in-pile experiments L6 and L7, which were designed and executed to examine fuel disruption and dispersal under loss-of-flow conditions at elevated power. l ~ Molten cladding motion has been modeled consistently with TREAT experiments R4 i l

iis and.R5 and SLSF experiment P3A, which were designed and executed to examine coolant boiling, vapor dynamics, and cladding relocation. The effects of plenum fission gas were modeled on the basis of TREAT experiment R8, the only in-pile experiment for LOF simulation with significantly pressurized fission gas plena simulating end-of-life conditions. Fission gas distributions within the fuel matrix were examined with the fission gas migration model, FRAS3, validated against HEDL fission gas release (FGR) expe..nents.

Finally, irradiated cladding failure under plenum gas pressurization was modeled on the basis of the HEDL FCTT experiments.

The whole core best estimate analyses show,with such experimentally validated models, a mild power burst with near zero energetics. This conclusion is valid even in the unlikely event that the plenum fission gas can act to compress the disrupting fuel. Parametric variation on clad failure and plenum gas release, and molten cladding relocation show very small sensitivities in initiating phase energetics. The potential for significant energetics appears to require pessimistic phenomenological nodeling that is not supported by the present experimental database, and is therefore beyond that appropriate for a realistic assessment of the accident energetics. The likelihood of energetics approaching the Structural Margin i Beyond the Design Base (SMBDB) value is very remote. l l F

iv TABLE OF CONTENTS Page I. I NT R O D UC T I O N..................................................... 1 II. PHENOMEN0 LOGICAL CONSIDERATIONS.................................. 8 II.1 MCDELING OF PLENUM GAS BLOWDOWN IN R8....................... 10 II.2 EXPERIMENTAL RESULTS ON CLAD RELOCATION DYNAMICS............ 24 I I.3 CLADDING FA IL URE CRITERI A................................... 30 I I I. ECC -4 LO F S LMMAR Y................................................ 34 !Y. C O NC L US I O NS...................................................... 50 Appendix A: Modifications to the SAS30 Boiling Module to Account for Release of Plenum Gas into a Boiling Region.......... 51 Appendix 3: Two-Fluid Model Analyses of Plenum F i s si on G a s Rel e a se...................................... 57 Appendix C: SAS3D Modification to TREAT R-Series Cool ant Hy drau l i cs....................................... 70 Appendix D: Calculation of Plenum Blowdown Ceu;1ed wi th P re ssure-Dri ven Fuel Mo ti on......................... 72 Appendix E: Modified Treatment of Partial Clad Blockages i n the S AS3 0 B o i l i ng Model.............................. 77 REFERENCES...,......................................................... 79 t i I ( i 1

t y l LIST OF FIGURES Figure Title Page 1 R8 Ga s Pl e n um Bl owdown.................................... 12 2 R8 Coolant Flow Rates, Flooded Friction Factor in CLAZAS........................................ 14 3 R8 Coolant Flow Rates, Nominal Single Phase Friction Factor in CLAZAS............................... 15 4 Coolant Pressure Profile at tne Onset of Clad Motion.......................................... 18 5 Vapor Mass Flux at the Onset of Clad Motion.......................................... 19 ~ 6 Coolant Pressure Profile After a R e - e n t ry E v e n t........................................... 20 7 R8 Clad Motion, Flooded Friction Factor in CLAZAS........................................ 21 8 RS Clad Motion, Nominal Single Phase Friction Factor in CLAZAS............................... 23 9 R5 Inl et Fl ow-r a te C omp a ri s on.............................. 26 10 Sketch of tne Voiding Pattern in tne Reactor C o re a t the e nd o f C a s e 1.................................. 44 B1 Rupture Site Pressure History C omp a ri s o n.............................................. 59 32 TWCFLU Predicted Channel Pressure Di stri b u ti on Hi s tc ry.................................... 61 B3 TWCFLU Predicted Fission Gas Di s tri bu ti o n Hi s to ry.................................... 62 34 Lower Sodium Slug Interface Comparison............................................... 65 B5 PLUT02 Predicted Channel Pressure Di s tri bu ti o n Hi s to ry.................................... 66 36 PLUT02 Predicted Fission Gas and So di um Va por Ma s s Hi s to ry............................ 68 01 Assumed Simplified Geometry of the Pin Stub and Plenum Region.............................. 73

vi LIST OF TABLES No. Title Page 1 Active Core Regicn (36 inch) Material 'dorth at EOC-4, Dc11ars..... 3 2 Timi ng o f Ev ents i n the R8 Test................................... 13 3 Timi ng of Cl addi ng Events in P3A Experiment....................... 29 4 Heating Rates for SAS3D Best Estimate LCF Case.................... 31 5 Event sequence for Case 1...............l.......................... 36 6 Event sequence fc.r Case 2......................................... 41 7 Comparison of times between initiation of gas release and the initi ation of fuel motion in Cases 1 and 2.................... 43 3 Event sequence for Case 3......................................... 46 9 Comparison of times between initiation of gas release and the i ni ti ation of fuel motion in Cases 1 and 3........................ 48 10 Work-energies based on adiabatic expansions of super-saturated fuel to a fi nal pres sure of 1 a tm................................. 49

1. Introduction In an assessment of energetics potential for an unprotected loss-of-flow (LOF) accident, several factors may be identified as playing critical roles in determining maximum reactivity and power levels. Included in this set are the l facto,rs which add positive reactivity, such a, the sodium void contribution and relocation of cladding away from the active core region, and negative reactivity factors including axial expansion, Doppler feedback, and fuel disruption and dispersal dnder overpower conditions. An additional factor that has been postulated is the potential for adding positive reactivity due to compaction of disrupting fuel by the gas in the pressurized fission gas plenum. Although there are factors, which are sumarized below, which appear to mitigate concern for this latter scenario, the potential for its effect motivated a reassessment of the expected scenario in the unprotected loss-of-flow scenario. This reassessment, provided in rasponse to NRC Question 1 CS760.178A3, demonstrated the importance of representing important phenomer.. ology with exiierimentally consistent models. Specifically, it was shown that fuel disryption.a,nd dispersal under overpower conditions was the dominant phenomenological consideration governing the potential for initiating phase energatics for the neutronic and thermal-hydraulic model of the hiiterogeneous CRBRP core described in the Project's assessment of HCDA energetics 2, Available experimental evidence on fission gas release from the HEDL FGR tests 3 was used to validate the FRAS3 phenomenological code,4,5, which was. then used to establish fuel pin conditions in the whole core analysis code, SAS3D. Similarly, data from TREAT in-pile LCF tests under overpower conditions, Tests L6 and L7,7,8, were used to calibrate the fuel dispersal modeling in the 6 SLtHPY9 fuel motion model of SAS30. When such experimentally based modeling was used in the energetics assessment of the low sodium void worth core ,1

( s described in reference 2, it was shown that very mild excursions and essentially zero energetics would be expected. It was also shown tha' sufficient time existed in the accident sequence to ppture the._ cladding expel the stored gas, and eliminate the potential for compaction. It Sta.s noted, in addition, that because of the relatively low positive reactivit,v additions from sodium voiding and clad relocation ($1.23 for void reactivity and 284 for clad reactivity at the time of fuel motion initiation in the lead channel) that the~ system was sufficiently far from prompt critical (a net reactivity of 59% at fuel motion initiation) that the conclusions were quite insensitive to a range of modeling assumptions. A further investigation of accident sequences in the unprotected LOF area was subsequently requested by the NRC Staff after review of the Project's reassessment of sodium void worth uncertainties and their implication on the potential for the loss-of-flow driven transient overpower (LOF'd' TOP) event. 'The uncertainty analysis was provided in response to question CS760.178A21 and was based on a large experimental data base. This assessment included anslysis of ovet 100 critical experiments in LMFBR-type assemblies of CRBRP size er larger. The assessment de,monstrated that the uncertainty in sodium void worth is not as large as is commonly perceived and resulted in a net uncertainty of 7.9% in the central core (positive reactivity) region,11.3% in the external core (negative reactivity) region, and 20.7% in the axial and internal blanket regions. An additional fact, however, that came from this investigation was a more accurate assess' ment of the nominal worths for all mat' rials, but, most importantly, the sodium void and cladding worths. Table e 1 contains these material worths as used in the SAS3D analysis and a compari-son of comparable information contained in reference 2 and used in the previously mentioned LOF assessment. Particularly important aspects of this

TABLE *1. Active Core Region (36 inch) Haterial Worth at EOC-4, Dollars Steel (Clad and Flowing Sodium Wire Wrap) Material Worth Material Worth j h Assenhly Number of GEFRa Best g,,g i SAS Channel Number Type Assemhites 00523 Estimate GEFR-523 Estimate 1 B 7 .100 .142 .173 .247 2 F 21 .386 .454 .986 -1.311 3 8 21 .330 .463 .607 .807 4 F 9 .160 .189 .414 .544 5 B 36 .559 .735 -1.029 -1.267 6 F 6 .035 .303 .265 .329 7 F 12 .165 .198 .51 .607 8 8 12 .125 .158 .242 .214 9 F 6 .027 .042 .157 .174 En 10 F 12 .113 .141 .417 .471 11 F 24 .366 .425 -1.027 -1.230 j 12 F 12 .038 ,.011 .120 .123 13 F 18 .116 .141 .466 .501 14 F 18 .200 .186 +.152 +.201 15 F 24 .002 .059 .101 .068 ' Driver 162 1.098 1.438 -4.31 -5.16 Internal Blankets 76 1.114 1.498 -2.05 -2.59 Total 238 2.212 2.936 -6.36 -7.75 a - p = 0.00340 b - p = 0.00323 d

reassessment are the increase in sodium void reactivity in the driver assemblies from $1.10 to 31.44 and an increase in the driver assembly steel worths (clad and wirewrap) from 34.31 to $5.16. Such increases th the ele- .ments which typically add positive reactivity to the system have several - iap11 cations. It would be expected that the introduction of larger sodium void reactivity would increase the rate of increase of the reactor power and shorten the time scale for the initiation of fuel disruption. If stored plenum fission gas can, as hypothesized, act to compress disrupting fuel pins, the potentially shortened time scale would limit the time available for gas blowdown and increase the potential for fuel compaction. On the other hand, the decreased blowdown time also raises the po'ssibility that released fission gas may be a significant force in affecting sodium vapor dynamics and may significantly mitigate the potential for clad relocation &e to sodium vapor streaming. Also, shorter time scales would generally imply a higher retention of fission gas still within the fuel pin matrix. This gas is the main force which drives fuel disruption and dispersal under mild overpower conditions. Hence, the dispersive potential for fuel material may, in fact, increase. It is clear that several competing effects are present in this hypothetical accident sequence, so an assessment of the integrated effects has been perfomed using the whole core analysis code SAS3D. In the assessment of accident energetics with the higher void worth values, it is expected that an increased sensitivity to modeling assumptions will be present. Relating modeling to available experimental information is essential. Such a detailed approach in the area of fuel disruption and l dispersal modeling was undertaken in the previous assessment, but in other areas including fission gas plena rupture, fission gas effects on sodium vapor dynamics, and molten cladding relocation, experimentally inconsistent, and yet

o. conservative, assumptions were employed. Having established a sound basis for fuel motion modeling, this present reassesse:nt of the LOF scenario allows an opportunity to develop a similar experimentally based description of the phenomena mentioned above. in the second section of this report, the three important phenomeno-logical areas -- modeling of fission gas blowdown, molten cladding relocation, and clad failure due to plenum fission gas -- are reviewed. In each of these areas, specific experimental evidence is available to guide phenomenological and integrated analysis modeling. The TREAT R-series 10,11 provides information on sodium vapor dynamics, clad relocation and plenum fission gas release and the SLSF P-seriesI2.13 provides further information in the first two of these areas. Modeling of these experiments with the SAS3D integrated analysis code and comparison of the results with data'is discussed. In the area of clad failure, the HEDL FCTT14-18 and FCTT/TUCOP19 tests are used to establish appropriate criteria. In the khird section of this report, this experimentally consistent modeling' c,apabil,ity is used in the whole core analysis of.the CRBRP LOF HCDA. Important phenomenological issues within the whole core analysis context are highlighted and the expected power and reactivity conditions are given. The role of the plenum fission gas is also discussed. Al so, reugnizing that there is some uncertainty in this modelfng, an indication of the sensitivity of the whole core analysis results to modeling phenomena such as clad failure and cladding relocation is provided. In the appendices, we have provided an independent justification and a phenomenologically based discussion of several elements of modeling in the SAS3D code. In particular, we describe the SAS3D treatment of fission gas / sodium vapor mixtures and compare it to independent two-fluid models in ,~., - - - - - - -, e----- ne -

the PLllT0220,21 and TRANSIT-HYDR 022,23 codes. Also described are pressure and f'1ciw distributions from these refined analyses and an interpretation of their implications on the whole core scenario. - Finally., a few coments should ba made about the phenomenology of plenum f.ission gas release, its potential for fuel compaction, and the conservative m'odeling of this effect employed in the SAS3D analysis to be described in section III. As discussed in previous meetings with the NRC staff and its consultants, the model us'ed in SAS3D for the compaction is simply an acceler-ation based on the time dependent pressure difference between the fission gas plenum and the point in the disrupting channel where the non-disrupted pin exists. The mass and length of this acceleratins segment decrease as the' power burst disrupts additional axial segments and the plenum pressure decreases as gas is ejected into the-coolant channel and the gas plenum lengthens because of the downward motion of the accelerating segment. This compactive motion is extremely conservatively modeled by assuming that all fuel pini refocate coherently and all assemblies (typically 12 to 24 asemblies per SAS3D channe,1{ in a given SAS3D channel also respond coherently. Due to i significant, radially incoherent, thermal profiles in steady-state and the expected 1 to 2 second time delay in radial void propagation 25, the assumed intra-subassembly coherence must be recognized as a simplifying, conservative assumption. In the TREAT R8 testll discussed below, which was designed and ( executed to explore plenum fission gas effects, such incoherencies, required pressurization of only 3 of the 7 pins used. A second mitigating factor not incl.uded in the SAS3D analysis is the upward ejection of cladding segments during the expulsion process. In the aforementioned R8 test, it was found .that the three pressurized pins upper cladding segments had moved upward from their original locations by 6.4,10.2, and 74.3 cm, respectively. The smaller 1

~ 7 two displacements were for pins which were restricted by the integral It would be instrument sheaths at the tops of these instrumented pins. expected that larger relocations would be more typical of the CR8RP case. Such an effect alters the calculation, and reduces the driving pressure, in First, the plenum volume increases, thereby reducing the two ways. overpressure and, second, the gap length used in the blowdown calculation Calculations decreases allowing the plenum pressure to decrease more rapidly. carried out with no restriction on the upward motion of the plenum gave the result that the plenum moved up far enough (14 in) to reduce the gap length to zero in only 27 as (See Appendix D]. Clearly, a strong mitigating potential for depressurization is available that has not been included in the present In addition, the fuel pin motion calculation does not include any assessment. Although such friction or mechanical interference between fuel and cladding. restrictive forces are expected to be present, quantification of their effect Consequently, the additional without experimental guidance is difficult. I conservatism of ignoring these mitigating forces has been employed in the analysis.. In summary, several factors can be identified that mitigate, if no '1 eliminate, the potential for plenum fission gas compaction and the results discussed in the whole core analysis section should be viewed as conservati if such compactive effects play a significant phenomenological role. i l r l i I 1 i I f

II. Phenomenological Considerations In the previous assessment of the hypothetical unprotected loss-of-flow accident, it was concluded that the positive reactivity that coul.d be i , introduced into the system was sufficiently limited that power levels remained - relatively low. The scenario time scale was thus extended and release of the plenum fission gas prior to pin disruption in all SAS30 channels was predicted. It was observed, however, that in the simulation of TREAT LOF tests L6 and L7, and from the test data itself, a slight positive contribution 'I to reactivity from the initial fuel motion could be inferred in relatively low power (5 to 10 times nominal) excursions 8 This effect was accounted for in the SAS30/SLLNPY analysis but because the system was sufficiently far from prompt critical, the initial positive fuel effect was of little significance. The maximum reactivity was approximately 604. Fuel dispersal in the lead channel mitigated concern for an accelerating sequence in which l compactive fuel motion in several more channels made reinforcing positive contributions. In this calculationl, it is noted, though, that the clad relocation module of SAS30, CLAZAS26, was predicting several tens of cents of positive reactivity during this portion of the scenario. As will be demon-strated below, it is believed that CLAZAS overpredicts both tne rate and the -amount of clad relocation. If similar CLAZAS modeling were used in higher l void worth cores, it would predict higher than expected clad reactivities and l introduce the potential for nearing prompt criticality at the time of fuel disruption. It should also be noted that the whole core calculations in the previous assessmenti did not explicitly account for the presence of ejected fission gas in the coolant channel and its effect on the sodium vapor { dynamics. The expected local pressurization at the ejection site would reduce the sodium vapor flow in the active core region where the molten cladding is l I L I I o

.g. .present and partially remove the shear coupling between vapor and clad, thus mitigating the extent of vpward relocation. Such a iconsequence has been deduced from the TREAT R8 experimentll. Although many in-pile experiments - have demonstrated the existence of upper cladding blockage, the R8 TREAT test, the only test with substantial pressurized plenum gas release, did not show such an upper blockage. In this section, we review the important phenomenological areas of the ' nfluence of plenum fission gas release on sodium vapor dynamics and clad i relocation, dynamic clad relocation under experimental loss-of-flow conditions without fission gas effects, and the failure of irradiated cladding under transient loading by the plenum fission gas. ~!n the first area, we focus on the TREAT R8 experiment and a recent analysis of this experiment with version 1.0 of SAS3D with the modifications and improvements used in the whole core analysis of the CRBRP heterogeneous core. Specific details of flow patterns and comparisons with the experimental data are provided. A parametric study of vapo'r-cla'dding frictional drag is presented and compared to experimental information to provide a qualitative basis for the modeling in the whole core analysis. More detailed dynamic information from TREAT and, SLSF experiments and their analysis with SAS30 is then reviewed to provide quantitative foundations for the whole core analysis. Finally, experimental information and analytical results are summarized to establish the quantitative clad failure criteria used in the analysis in the third section. 6

II.1 Modelling of Plenum Gas Blowdown in R8 TREAT experiment R811 addressed issues related to voiding dynamics and clad relocation with the presence of released plenum fission gas. In this 7-pin test, 3 of the 7 pins were initially pressurized (4.14 MPa at 560*C) using xenon gas to account for intrasubassembly incoherence. This was a constant, nominal power test subjected to a simulated FFTF flow coastdown. Important " observations included the upward ejection of the upper cladding segments. driven by plenum gas expansion, which would have the tendency to mitigate any disrupted fuel compaction. Also observed was that when the cladding subsequently melted, little or no molten cladding was driven upward into the upper reflector region since the channel pressurization had removed sodium and effectively precluded upward sodium vapor streaming at that time. The complete planar blockage at the top of the core found in previous tests was absent in R8, replaced by an inhomogeneous pattern of debris and complete unblocked regions which* were the result of the previous cladding ejections. In RS, the remaining cladding melted downward faster than in previous test:, due to the early, complete channel voiding. The channel pressurization due to plenum gas release'resulted in the predicted expulsion of sodium from the entire core region; this early voiding and ensuing film dryout altered the subsequent heatup, melting, and relocation of cladding relative to previous tests. A complete inlet blockage ' formed about 2 s carlier in R8 than in previous tests, attributable to the much hastened downward melting progression in the absence of s.cdium " chugging". The blockage lower extremity was 8 cm into the lower reflector region; by the end of the test the steel had accumulated to 21 cm thickness. The R8 test was re-analyzed using the new SAS3D treatment of fission

O, gas / sodium vapor mixtures as well as a minor code modification, described in Appendix C, te account for the system hydraulics. The main purpose of this re-analysis was to detennine whether the CLAZAS model, using coolant velocities and pressures calculated by the new gas / sodium vapor treatment, could predict the clad relocation results observed in this test. It was found that CLAZAS could predict the observed clad relocation results Lut only if the friction factor used to calculate the shear stress between the sodium vapor and. molten clad were reduced to a nominal single phase friction factor, rather than the flooded tno-phase friction factor normally used in CLAZAS. For this re-analysis of RB, a number of SAS3D input parameters were different from those used in the SAS analysis ' reported in ANL/ RAS 78-3911 The pin failure was assumed to occur in an axial node centered 12.8 em below the top of the active ' fuel, since SAS3D predicted the highest clad temperatures at this node at the time of pin failure. At the time of clad failure, SAS3D predicts that the gap between fuel and clad at the top of the active ' core is smaller than the gap between the upper Inconel reflector and the cla'd. Ther,efore, the flow area and hydraulic diameter of the ficw pata 1 between the gas plenum and the rupture were based on the cal:ulated gap si:e between fuel and clad in the upper active fuel nodes at the time of pin failure. The length used for this flow path was 18 cm, which corresponds to the length of active core above the assumed failure point.plus a small addition for the pressure drop past the upper Inconel reflector. As shown in Figure 1, with the use of these parameters, SAS3D predictions for the plenum gas blowdown agree quite well with the experimentally observed results. Table 2 gives the timing of many significant events in this test. For the SAS3D calculations, a cladding failure temperature of 1400*C was used. Near the time of clad failure, the clad temperature at the failure i

80-T oa Legend N 60-x }A EXPERIMENT Z D X SAS3D CD CD NN 40-g 2 o \\ 'X. n Na 20-A U3 N N <4 O w 0-l -0.5 0 0.5 1 1.5 TIME SINCE FAILURE (Sec) l Figure 1. R8 Gas Plenum Blowdown l l

13-Table 2. Timing of Events in the R8 Test Experiment SAS3D Event Time (sec) Time (sec) Reactor power up 3.5 3.5 Start of flow coastdown 7.97 7.97 Local boiling 15.17 Net voiding 17.5 17.52 Inlet fitw refersal-18.18 18.17 Onset of cladding failures 18.89 18.93 Flowtube failure 18.96 19.42 Cladding motion starts Reactor power down 26.0 26.0 t node was increasing at a rate of about 700'C/second, so the SAS3D failure time would match the experimental value better if a clad failure temperature of about 1370*C were used. Since unirradiated clad was used in this test, the clad faf1ure~ temperatures in the test were probably higher than they would be for end-6f-life irradiated clad. l Figure 2 gives a comparison of the measured and computed inlet flow rates for R8. The agreement is reasonably good from the beginning of the run through the initial boiling, the clad failure, and the initial expulsion and re-entry after gas release starts. Later, when SAS3D predicts the re-entry of liquid sodium over very hot clad, the code predicts some vigorous, expulsion and re-entry events with higher frequencies and higher amplitudes than those j observed in the experiment. The SAS3D calculation shown in Figure 2 used the normal CLAZAS two-phase flooding friction factor for the shear stress between j sodium vapor and molten clad. Figure 3 shows the results obtained when the ( ' same case was re-run with a nominal single phase friction factor in CLAZAS

I i _o m in< N< t d i V -~-.......:.:..... - -==================.=.=....,.

==,m,,,____ y,,;.;z---- 3,, u c ' =_s%_ _ _ _=_ =- c' Y.. 5 D U' O i u e m B .c -m 2_, -m s -~ I z = a es". _Q y m M es M g o.. n = e / @ r< 11,: li w mi u ,-a 1 ,a e i i e. o o. I

s O

O I m l l AOld 13'INI G3ZI'IVMEION e i l i I s

1- "$o I .A N 0.5 - l s ,e >1Z )> ll f', - - - - - - = 0= {,v.> l :,/ p r4 .ta, l f l M .. f. l,CbTelld f.l,.' l l/ 4 -0.5 - EXPMillM ENT f, M l l .O 3_3_3_3 9_ _ _ _ _ _. ,E L,, 5 10 15 20 25 TIME (Sec) 1 Figure 3. R8 Coolant flow Rates Nomin51 Single Phase Fricti'on Factor in CLAZAS e 8

instead of a two-phase friction factor. Until the start of clad motion, both cases are the same, but after clad motion starts the nominal single phase case quickly predicts a clad blockage in the lower part of the active core, and this blockage reduces the amplitude of the liquid sodium re-entry and hxpul sion. Flowtube failure which occurred soon after the start of gas release from ruptured pins was not modelled in SAS3D, but it would probably have some impact on coolant flow rates. Another aspect that was not included in the SAS30 analysis was the upward ejection of the upper clad segments of the pressurized pins after pin failure. The upper parts of two pins restricted by instrument sheaths went upward 6.4 and 10.2 cm, whereas the upper clad from the third pin went upward 74.3 cm. If the upward motion of the clad from the third pin occurred soon after pin failure, it would have led to very rapid gas release from that pin, although the ejection of the upper clad segment would have reduced the impedance to upward flow for gas from the pin, and thereby would have reduced the impact of the rapid gas release on the inlet f1bw shown in Figures 2 and 3. The measured gas pressure shown in Figure 1 is for the pin that moved upward 10.2 cm. An upward motion of 10.2 cm would reduce, but not eliminate, the impedance to gas flow between the gas plenum and the rupture point. It is possible that the impedance in this pin was initially higher than that modelled in SAS3D, and that the impedance dropped as the upper clad segment moved upward. The expulsion of the inlet liquid after pin failure was somewhat faster in the SAS3D results than the experimental measurements indicate, and SAS30 l predicts re-entry after the expulsion sooner than the experiment. In the SAS30 analysis, all three pressurized pins were assumed to fail simul taneously. Staggering the pin failures would reduce the speed of the initial expulsion and delay the re-entry. l

At 19.42 seconds, when the motion of molten clad starts, asch of the gas has been released from the gas plenum; but the plenum gas pressure is still 19 atmospheres at this time; and gas release still has a large influence on the pressures and flow rates in the coolant channel. Figures 4 and 5 show the coolant pressures and mass fluxes near the time when clad motion starts. The gas is being released at 98 cm. The pressure peaks at this location. The mass flux

• is upward above this location, and downward below it.

If the gas ~ were not being released, then by the time that clad motion starts the coolant pressure would tend to peak near the bottom of the active core, where the vapor source would be, and the vapor velocities would tend to be upward above that point. After the start of clad motion, gravity and downward gas flow tend to send the clad downward, but periodic re-enty of liquid sodium into the bottom of the fuelled region provides an intennittent vapor source that exceeds the gas source and sends vapor and clad upward part of the time. Figure 6 shows the coolant ' pressure profile soon after a re-entry. Re-wetting of het clad provide} a high vapor pressure near the bottom of the fuel. Molten clad causes partial blockages at several axial locations. Most of the pressure drop in the test section is concentrated across these molten clad regions, and the combination of pressure gradients plus shear stress from upward streaming vapor sends clad upward when the hot clad is wetted. Figure 7 shows the clad behavior for the case with a flooded friction factor between clad and vapor. In this figure, shaded areas represent molten or re-frozen clad, and the density of the shading is an indication of the thickness of the clad. The clad oscillates up and 'down, but eventually a substantial clad blockage is formed above the core and the rest of the clad drains downward. I.ven with a substantial gas source near the top of the fuel,

2-m& l m l 3 n.a-l m mD 1.6 - (n cn m m l a. 1.4 - h. bZ< 0 g.g. 9o 1-i i i i i i o so 100 150 200 250 300 ,i AXI AL. POSITION (cm) i Figure 4. Coolant Pressure Profile at the Onset of Clad Motion 1

19 o -oN - 8 ^E l s z 5 o O o-M. - o cn s o e P-a = x c -8< a 2 g !it o E i i i i ~ e o e e m a 6 6 8 a i t i I (OSS/8-*Wo/m8) Xn'Id SSVli 1NV1000 i

O -QN a r w e i _$ m 5 A e I = s Z 5 / o x s ~ o~ S caoA .u A a s 2 -A ~8< 2a 4 a C ,o i i i n q m q N (mge) 3HflSSSHd 1NTIOOD .m

140 120 l lll ,.7-1, .---.$,k .j. o 8 80

3 y

I 0 60 l ll[j f I'll h 3* ~ ___---_--_.g._ 20 -f bottom of fuel I I-- I I I 0 l 19.4 20.0 20.6 21.2 21.8 22.4 23.0 } Time (see) figure 7. R8 Clad Motion, Flooded Friction Factor in CLAZA5 i 4 1 e

much of the clad goes upward. A post-test examination of the test section showed no clad blockage above the core, and no indication that molte_n clad had ever gone upward into the reflector region. Instead, all of the clad from the fuelled region was found in a massive blockage in the lower reflector. Figure 8 shows the results of the same case re-run with a nominal single phase 4 friction factor between the sodium vapor and molten clad. In this case, all of the clad ends up in the lower reflector. In either case, the coolant pressure gradients tend to concentrate across the molten clad regions, but the difference in shear stress is enough to make the difference between net upward motion and net downward motion. In suninary, the use of a flooded friction' factor in CLAZAS over-predicts the upward motion of molten clad in this test, whereas the results calculated with a nominal single phase friction factor are consistent with the post-test examination. Also, the nominal single phase results provide better agreement with the measured inlet flow after the onset of clad m6 tion. 9 t 9 l l l O 9 6

l ~23-l i o l l l 1 1 M i n 3 i m 5 = i ~ T d l a s l N u Pah I CO n %g i Ue a =-= m= s en 2 ~ f; s i n e i E e i9 =-- i r ic e c; E i l lo

• "b"E

._s-*._r;= - ; - N ,a e o 8o E- 'D l"

"u

= e o = m- , O.-- e=- i i o u rz l ,p l8 s ,a id b i l..O t i I I I I o g cn .e o o o o o o o o-w v N o. CO O v N (wo) Uo!1! sod le!x-/ as a -~ .,.,n.- ..-,,-4

?, II.2 Experimental Results on Clad Relocation Dynamics To assess further expected clad relocation in the CRBRP LOF scenario, several additional experiments and their analyses with the SAS3D codei, were reviewed to establish a reasonable treatment within the context of the SAS3D code. The experiments considered were the TREAT R4 and R510 tests, and SLSF tests P313 and P3Al2 SAS3D analyses of these tests have been performed as well as analyses with the one dimensional cladding relocation model, CLAP 27, for the R5 test and the nJ1ti-dimensional cladding relocation model, MULCLA028,29, for the R4, RS, P3 and P3A tests. A brief sumary of the tests, their results, and the analyses is provided below. TREAT tests R4 and R5 were seven-pin, los's-of-flow tests with full-length i unirradiated FFTF-type fuel pins. In LOF test R4, the sequence was run at l, constant, nominal power well beyond the inception of molten fuel motion. h RS, the sequence was terminated prior to fuel melting to preserve evidence of early molten cladding motion. Up to the point of fuel melting and motion, the l tests were consistent _with each other. It was noted, however, that the thermocouple data for the R4 test were of good quality, but the R5 data showed l numerous ambiguities attributable to erratic thermocouple performance. Co'nsequently, the interpretation of temperature data was based principally on R4 data. It was noted in reference 10 that: ..., at a time in the test sequence when SAS calculates the onset of cladding motion (about 1 see after the cladding solidus temperature is reached), thermocouple TCTS-7, located 3 inches below the top of the fuel column, indicates a rapid transient heating event. This is interpreted as being caused by the motion of molten cladding material which accumulates and bridges to the flowtube wall. Additionally, thermocouple TCTS-5, at the top plane of the fuel column, shows a similar heating event about 0.1 see after the lower TC, suggesting a net upward motion

.g - of the molten material. However, therwocouple TCTS-4, located one inch above the heated zone in the tolder insulator pellet region, does not show such an event, suggescing that the molten material stopped its upward motion between these two measurement locations". This upward relocation of 3 to 4 inches in 0.1 sec implies an average upward cladding velocity of less than 100 cm/sec. The presence of the upper cladding blockage was confirmed during post test examination and were found to be about 0.3 cm in thickness. A post-test analysis of the R5 test was performed with SAS3A code and summari:ed in reference 10. To examine the effect of the frictional coupling between streaming sodium vapor and molten cladding, this same experiment moderl was examined with version 1.0 of SAS3030 with the modifications described in the appendices. The most important modification was 'the consistent coupling l of frictional effects in the " flooded" cladding region 26 with the implication 1 of reduced vapor flow and reduced shear forces on the molten cladding. Figure 9 shows a comparison of inlet flow rate between SAS3D and the experimental measurements' and the results are seen to be quite good. SAS3D predicts the initiati,on of clad motion in node 15, whose mid-point is at 97.5 cm frem the bottom of the pin (11 cm below the top of the active fuel). As mentioned earlier, it was reported that a themocouple 3 inches below the top of the active fuel sensed molten clad motion initially and a second themocouple at the top of the fuel sensed molten clad motion approximately 0.1 sec later, implying a clad velocity of approximately 75 cm/sec. In SAS30 calculations with the normal flooded two-phase friction factor, initial velocities were calculated in excess of 200 cm/sec and in the time it took the clad to move 10 cm, the average velocity was approximately 150 cm/s'ec. By reducing the l frictional coupling by employing a nominal single phase friction factor 26, the initial velocities were calculs.ted in the range of 50 cm/sec and in the time

l1 P i l i 4 er i 0.95-0.00-N Legend 4 0.65-a R5 FM-4 Z S: 0.50-x SAS3D o A o Ex 0.35-O ~ - - - N 0.20-0.05- \\] N ((i U / A \\ g / ['i 7 1 1 x / s M -0.10 - l i o b< z -0.25 - i -0.4 0 - l 13 14 15 16 17 TIME, see s \\ l Figure 9. R5 Inlet Flow-rate Comparison l 1 a e

e it took the clad to move 10 cm, the average velocity was approximately 75 ci/sec. Recognizing that there is uncertainty in the experimental measurements, it is still clear that the modification of frictional coupling to using a nominal single phase friction factor has resulted in a reduction in the rate of upward clad relocation and provides better agreement with the experimental data. In the review of a similar event sequence and calculation from the P3A experiment and the qualitative results for clad motion in the R8 experiment, it will be seen that such a reduction in upward driving forces is consistently required to reach reasonable agreement between SAS30/CLAIAS \\ calculations and experimental observations. Such apparent deficleicies of CLAZAS have long been recognized, however, and a new cladding relocation model, CLAP 27, was, in fact, developed several years ago and incorporated in the SAS3A code. In the analysis of the TREAT R5 test with CLAP, it was noted "In comparison to CLAZAS calculations of the typt of experiment, the CLAP model allows calculation of a more realistic smaller upper blockage....".27 Analyses o' TREAT experiments illustrate that clad relocation predictions with SAS30/CLAZAS should be viewed with caution. Sinflar conclusions were drawn from tne SLSF P3A and P3 experiments and their analyses. The P3A experiment contained a fuel bundle comprising 37 fresh, full-length, prototypic FTR pins. The fuel was irradiated in the SLSF under prototypic thermal corditions to an equivalent of 25 full-power days, at maximum average linear power of 36.7 kW/m, yielding a maximum burnup of -0.6 atom *.. Follow-ing 48 hours of continuous full-power operation, the fuel was subjected to a simulated loss-of-flow accident. The test train orificing and bypass flow had been chosen such that the boiling and voiding dyna'nics that resulted from the flow reduction closely approximated those expected in a loop-type fast

reactor. Thus, the test produced data on voiding and cladding motion, as well as data on fuel disruption characteristics. This experiment and the SLSF experiment P3, which used a near.ly identical test. vehicle.and test conditions, fonn a complementary set of experiments addressing the behavior of low-burnup fuel bundle during an LOF. The reactor scram in the P3A LOF simulation was chosen to tenninate the experiment just before gross fuel melting, and thus to yield data on initial fuel disruption. The P3 LOF simulation was continued long enough to ensure wide-spread fuel :nelting and to provide an opportunity for early fuel motion. The predicticn of coolant voiding by SAS3D was observed to be reasonably good, but the clad relocation sequence, as deduced from thernoccuple response, developed more slowly than that predicted by the code. Specifically, a thermocouple positioned 76 m'below the top of the fuel detected molten steel at 12.2 sec and a second thermocouple positioned at the top of the fuel, rose to the stainless steel melting temperature between 12.5 s and 13.0 s in the P3A test. The timing of cladding events and comparisons with calculaticns are given in Table.3 [taken from Reference 12 ]. In the SAS30/CLAZAS analysis of this experiment reported in reference 12, the initial clad velocities were calculated to be great than 200 cm/sec and over the first 0.1 sec, they averaged sightly less than 200 cm/sec. In this analysis, the normal flooded two-phase friction factor was used. Measured clad motion is seen to be less than that predicted by SAS30/CLAZAS, with average velocities in the.20 to 30 cm/sec range. Similar results were obtained in comparison of experimental results and analyses for the F3 test.

29-Table 312 Timing of Cladding Events in P3A Experiment (times in seconds after beginning of coastdown) SAS3D P3A DATA l Initial Cladding 10.3 11.8 i Melting Initial Cladding Meltthrough 11.0 11.8-11.9 and Motion . Molten Cladding Reaches 11.1 12.5 Top of Fuel 11.1 -13.0 Top Blockage Reaches Final Configuration Configuration of Top Complete Partial Blockage lib NA Bottom Blockage Complete The tests discussed above have illustrated ' hat the CLAZAS module of SAS3D can qualitatively predict the behavior of molten clad relocation includ-ing the development of upper blockages and the subsequent draining to form a lower blockage. ' There are limitations, bewever, in the quantitative pre-dictions of rates of relocation and the extent of the upper blockage. From the review of these experiments and the R8 analysis in the previous section, it is clear that modeling clad relocation with the one-dimensional CLAZAS mocule of SAS3D using the high frictional coupling appropriate for " flooded" conditions can produce conclusions that are both qualitatively as well as quantitatively incorrect. Consequently,in the whole core cases to be described in section III, using the fact that CLAZAS clad relocation predictions were more consistent with experiments with significantly weaker vapor-clad frictional coupling, the choice of a nominal single phase friction f actor was employed.

t - s - II.3 Cladding Failure Criteria In our' previous SAS3D assessment of the potential for autocatalysis due to plenum pressure driven fuel compactionl, we made the simplifying. assumption that blowdown of the plenum would not begin until the cladding at the top of the active fuel reached 1400*C; essentially the melting point. Although the cladding at the fuel-blanket interface will certainly fail by the time it . reaches melting, earlier hechanical failure must occur at some temperature less than melting under EOC-4 plenum pressure loadings of the order of 50 atmospheres. Earlier failure would allow more time for depressurization. For 100-500*C/s, a reduction in the typical cladding heating rates near melting of, failure temperature of 100*C would increase the blowdown time by 0.2-1.0 seconds. This additional time is substantial compared to the time constant for blowdown of about 0.25 s (See Appendix D]. We have therefore looked more carefully here at the question of cladding failure under plenum pressure

loading, Cladding failure temperatures' depend on steady state irradiation history, mechanical loading, and transient thermal history. Table 4 provides the most important paramete. s as determined for the uppermost active fuel column clad-r ding node from the new best estimate case SAS3D calculations of the CRBR E00-4 loss-of-flow accident scenario (Case 1 in section III). The channel numbers The at the top of the table refer to SAS channels as given in Reference 2.

1 cladding hoop stress a is calculated from the maximum plenum pressure P and the thin shell formula o = Pr/h, where r is the inner cladding radius and h is i l the' cladding wall thickness. Also shown are the cladding midwall heating rates at temperatures near the melting point. Our reassessment of cladding failure under the conditions given in Table 4 consisted of a review of relevant experiments plus calculations using both e -r-- --,,r-- -es we

i i Table 4 Heating Rates for SAS3D Best Estimate LOF Case CH 2 CH 4 CH 6 CH 7 CH 11 Fluence at Top Node 6.34 6.37 3.44 6.98 7.56 1022 n/cm2 i Irradiation Temperature, *C 572 569 587 564 555 Maximum Plenum Pressure, MPa 44 44 23 44 44 Cladding Hoop Stress at Maximum 29 29 15 29 29 Plenum Pressure, MPa T 9 1200*C, 'C/S 225 219 154 208 526 T 9 1300*C, 'C/S 216 246 141 365 443 T 9 1400*C, 'C/S 347 432 166 433 320 data correlations and theoretical models. The experiments which we judged to be most directly applicable to the question of cladding failure under plenum gas loading conditions were the FCTT tests performed at HEDLl4-17 In the i FCTT tests, both unirradiated and irradiated cladding tube's were internally l l loaded with gas pressure and heated uniformly at a constant rate until failure l l occurred. Failure temperature and failure ductility were measured as a func-I ' tion of heating rate and initial hoop stress. These tests very closely simu-late the thermal and mechanical loading conditions of interest here. Further-more, multiple FCTT testslB have shown that the most important part of the i

cladding thermal history is that part near the failure temperature. For low pressures, where failure is expected to occur near the melting point, Table 4 shows that the SAS30 calculated cladding heating rates at the fuel bianket , interface are nearly constant and of the order of several hundred 'C/s. Such heating rates are close to the highest heating rate (111 *C/s) FCTT data. Some additional data does exist from recent FCTT TUCOP19 tests at higher heating rates of 550*C/s. However, in the TUCOP tests the cladding diametral strain-rate was controlled by decreasing the driving pressure as the test proceeded. Since plastic instability is an important aspect of high-temperature failure under constant pressure loading, these tests tend to give somewhat higher failure temperatures than woul'd be expected under constant pressure conditions. Although a considerable amount of FCTT data exists, almost all of the data are for conditions where the gas pressure loading was greater than 50 atmospheres. We have to make use of the full data base here to calculate failure under low-pressure conditions by extrapolating existing correlations 16,33-of the data. In our calculations, the Dorn parameter correlation 33 produced the most reasonable results when compared with the limited high-temperature, low-pressure data. Application of the Dorn parameter correlation to the conditions given in Table 4 produced cladding failure temperatures between 1250 and 1300*C. Additional calculations of cladding failure were perfonned using theoret-ical models developed by ANL/ RAS 34,35,36 These models describe the funda-mental phenomena which govern cladding failure, including high-and low-temperature matrix defonnation, annealing and recovery, grain growth, liquid i metal and irradiation embrittlement, and intergranular fracture. Previous comparisons between the theoretical models and the FCTT data mentioned above I

33- ,have shorn good agreement. Extension of these calculations to the conditions j given in Table 4 produced failure temperatures for irradiated cladding in the range of 1250-1350*C. It was also found that for unirradiated cladding under similar conditions, the failure temperatures are within 25'C of melting. This i result is consistent with the SAS3D interpretation of the R8 TREAT test given elsewhere in this response. Based on the above review of the rel'evant data and on calculations using l two independent approaches, we conclude that 1300*C is an appropriate average cladding failure temperature to be used to initiate plenum blowdown in the SAS3D assessment. of plenum pressure driven fuel compaction in CRBR EOC-4 loss-of-flow accident scenario. This temperature is 100*C less than the previous f conservative assumption of cladding failure at melting. l i j f i, i f i b I l i i t i I e i

r III. EOC-4 LDF Summary A best estimate LOF scenario for the CRBRP EOC-4 heterogeneous core, i incorporating the phenomenological considerations discussed above will now be described.' The basic reactor model is the same 15 channel model used in . Reference 2. The neutronics data are the new values which resulted from the reassessment of the rodium void worth described in the introduction. Thus, all the neutronics data tre different from those used in Reference 2. Most taportantly, in the driver subassembites, the void worth is somewhat more than 34 cents larger and the steel worths are increased by about 85 cents. Other modeling assumptions, with a few exceptions to be described below, are the same as were used in the response to question IS760.178A3. 1 Unlike previous whole core calculations using SAS3D,2, the current calculation explicitly accounts for the release of stored fission gas from.the fission' gas plenum into the SAS channel. The manner in which this is accomplished is similar to that used in the analysis 'of the R8 experiment described above. Coding changes were also introduced to allow (at the user's discretion) the. pressure in the fission gas plenum to be applied to the top of the upper pin stub in the SLLMPY calculation. When this option is used, the mass of the upper axial blanket fuel pellets is added to the pin stub mass in determining the downward acceleration of the stub. In addition, coding changes were made to allow the use of smooth-tube friftion coupling between clad motion and the sodium vapor streaming. This appears to allow the one-dimensional modeling in SAS30/CLAZAS to better approximate experimental results. The analysis of the R8 test supports this approach. Clad motion was allowed to begin when the clad melt-fraction reached unity. The boiling model was modified so that after the onset of clad motion the friction factor used to calculate the shear stress between the vapor and the clad in the boiling e

6 model was the same as tnat used in CLAZAS. Previously, the friction factor used in the boiling model did not account for any flooding that CLAZAS might be using. Finally, code changes were introduced to prevent moving cladding within the SLLMPY compressible zone from causing a zero-velocity boundary condition to be set at one or both edges of the compressible region. An event sequence for the current calculation is Itsted in Table 5. This case is designated as Case 1. It is of particular interest to note the times when gas release begins and ends in a channel. The gas release is stopped when the pressure in the fission gas plenum drops below 2.5 atm. I This value I was typical of the pressures predicted by SLtMPY at the point of fuel disruption in the previous assessment. The time required to achieve this 1 value is seen to vary from as little as about 0.7 s to more than 1.5 s. In addition, it is noted that of all the channels to initiate fuel motion during the transient, only channel 11 does so before the gas release has stopped. In this channel, the pressure in the fission gas plenum is about 4.3 atm when fuel motion initiates, while coolant channel pressures are nearly as high at the axia,1 location where fuel motion begins. Thus, compactive fuel motion is minimal and the potential for autocatz1ysis is quite small. It should be noted that in the present calculation, the pressure in the fission gas plenum is held artificially high because of a peculiarity in the gas release model. The pressure :ased by the gas release model at the clad rupture point is not the coolant channel pressure in the axial node containing the rupture location, but is the pressure at the lower bubble interface for the vapor bubble adjacent to the rupture. Nonnally, the difference between these two pressures is small enough to have an unimportant effect on the rate of gas release; however, in the present case, the lower bubble interface for channel 11 is located below a molten clad blockage. Because the lower sodium 7

Table 5 Event sequence for Case 1 TIME EVENT CllN* P/P0 Ril0 RHOP Ril00 RIIGE Ril0V RHOF-RHOC 11.9251 COOLANT BOILING 6 0.863 -0.074 0.0 -0.151 -0.058 0.134 0.0 0.0 13.4879 COOLANT POLLING 2 0.902 -0.013 0.0 -0.172 -0.068 0.227 0.0 0.0 13.8538 COOLANT BOILING 4 0.866 -0.059 0.0 -0.180 -0.073 0.193 0.0 0.0 14.3099 COOLANT BOILING 7 0.880, -0.038 0.0 -0.187 -0.079 0.228 0.0 0.0 15.0174 RELEASE GAS 6 0.981 0.062 0.0 -0.211 0.098 0.371 0.0 0.0 15.6666 CLA0 MOTION 6 1.674 0.394 0.0 -0.270 -0.150 0.814 0.0 0.0 15.7166 STOP RELEASE 6 1.625 0.367 0.0 -0.277 -0.156 0.797 0.0 0.004 15.8066 COOLANT BOILING 10 1.652 0.367 0.0 -0.289 -0.166 0.830 0.0 -0.008 15.8266 COOLANT BOILING 11 1.695 0.381 0.0 -0.291 -0.169 0.853 0.0 -0.012 16.0265 COOLANT BOLLING 9 1.706 0.356 0.0 -0.319 -0.194 0.919 0.0 -0.050 l 16.3609 C00LANT BOILING 13 1.779 0.365 0.0 -0.351 -0.225 0.969 0.0 -0.029 16.3659 RELEASE GAS 2 1.782 0.365 0.0 -0.351 -0.226 0.970 0.0 -0.028 16.5409 RELEASE GAS 4 2.672 0.561 0.0 0.377 -0.248 1.157 0.0 0.030 16.7656 COOLANT BOILING 12 5.124 0.714 0.0 -0.474 -0.317 1.283 0.0 0.223 L, i 16.8109 RELEASE GAS 7 7.089 0.782 0.0 -0.498 -0.336 1.393 0.0 0.222 i' i 16.8784 FUEL MOTION 6 8.374 0.790 0.0 -0.549 -0.376 1.489 0.0 0.225 16.9034 CLA0 HOTION 2 8.196 0.776 0.0 -0.566 -0.389 1.497 -0.000 0.235 4 16.9370 COOLANT BOILING 15 8.104 0.761 0.0 -0.592 -0.408 1.493 -0.003 0.271 16.9997 COOLANT BOILING 14 5.983 0.658 0.0 -0.62) -0.430 1.466 -0.013 0.258 17.0284 CLAD MOTION 4 5.902 0.647 0.0 -0.632 -0.4 38 1.500 -0.038 0.254 17.0897 CLAD MOTION 7 4.600 0.534 0.0 -0.645 -0.450 1.564 -0.203 0.268 17.1284 COOLANT BOILING 5 3.260 0.338 0.0 -0.650 -0.453 1.602 -0.435 0.273 7 17.1697 RELEASE GAS 11 2.179 0.010 0.0 -0.650 -0.453 1.625 -0.763 0.251 17.2297 RELEASE GAS 10 1.403 -0.519 0.0 -0.642 -0.451 1.568 -1.182 0.189 17.4309 COOLANT BOILING 3 0.889 -1.209 0.0 -0.626 -0.447 1.491 -1.677 0.050 17.4384 STOP RELEASE 4 0.871 -1.252 0.0 -0.625 -0.447 1.490 -1.714 0.044 l 17.4747 COOLANT BOILING 1 0.809 -1.391 0.0 -0.621 -0.446 1.493 .l.827 0.010 17.4797 CLAD MOTION 11 0.808 -1.393 0.0 -0.621 -0.446 1.493 -1.826 0.008 17.5622 CLA0 MOTION 10 0.774 -1.406 0.0 -0.613 -0.445 1.519 -1.854 -0.013 17.5797 RELEASE GAS 9 0.749 -1.474 0.0 -0.611 -0.445 1.531 -1.919 -0.030 17.6897 COOLANT BOILING 8 0.663 -1.691 0.0 _0.604 -0.444 1.613 -2.127 -0.129 17.7034 STOP RELEASE 7 0.679 -1.603 0.0 -0.603 -0.444 1.633 -2.095 -0.095 17.7634 RELEASE GAS 13 0.811 -1.085 0.0 -0.593 -0.443 1.795 -1.899 0.054

Table 5 (cont'd) Event sequence for Case 1 TIME EVENT CllN* P/P0 Rl10 Ril0P Hil00 R110E Ril0V Ril0F Ril0C 17.8447 CLA0 MOTION 9 1.060 ' 0.519 0.0 -0.589 -0.442 1.867 -1.538. 0.183 17.8934 STOP RELEASE 2 1.354 -0.159 0.0 -0.591 -0.442 1.918 -1.287 0.242 18.0072 RELEASE GAS 12 7.570. 0.795 0.0 -0.620 -0.446 1.864 -0.684 0.681 18.0384 PEAK REACTIVITY 0 42.984' O.960 0.0 -0.674 -0.454 1.819 -0.566 0.835 18.0434 PEAK POWER 0 46.536 0.956 0.0 -0.693 -0.456 1.810 -0.565 0.860 18.0522 FUEL MOTION 2 36.776 0.933 0.0 -0.723 -0.461 1.790 -0.572 0.898 18.0559 FUEL MOTION 4 30.773 0.923 0.0 -0.731 -0.461 1.784 -0.584 0.915 18.0597 FUEL MOTION 7 26.979 0.911 0.0 -0.737 -0.462 1.774 -0.600 0.936 18.0609 STOP RELEASE 10 25.953 0.906 0.0 -0.739 -0.463 1.773 -0.608 0.943 18.1134 CLA0 NOTION 13 2.461 _0.039 0.0 -0.757 -0.464 1.771 -1.902 1.312 18.1559 RELEASE GAS 15 1.487 -0.709 0.0 -0.744 -0.463 1.915 -3.238 1.821 18.1659 FUEL MOTION 10 1.375 -0.844 0.0 -0.741 -0.463 1.953 -3.542 1.948 18.2359 CLA0 MOTION 12 0.659 -2.926 0.0 -0.722 -0.463 2.093 -6.406 2.572 18.2596 FUEL MOTION 11 0.558 -3.672 0.0 -0.719 -0.463 2.065 -7.091 2.536 S. 18.2634 RELEASE GAS 14 0.544 -3.799 0.0 -0.719 -0.463 2.058 -7.195 2.521 I* 18.2894 TERMINATION 0 0.468 -4.626 0.0 -0.718 .-0.464 1.962 -7.647 2.241 Terminology: CHN stands for the SAS30 channel number; P/PO stands for the normalized power; Rit0 stands for the net reactivity; and Ril0X stands for reactivity component X where X = P means programed reactivity, X = 0 means Doppler, X = E means axial expansion, X = V means voidin9, X = F means fuel motion, and X = C means clad motion. Reactivities are in dollars. I i I

1 38-s

• slug re-enters the channel and rewets some very hot cladding below the molten blockage, the pressure at the lower bubble interface increases to a value near 4.5 atm and causes the gas re'1 ease model to force gas and vapor back~into the plenum, thus, causing the pressure in the plenum to increase. The coolant pressure in the axial node adjacent to the rupture site remains near or below 2.0 atm, and it is likely that had this pressure been used in the gas release calculation, the gas release would have been stopped before fuel motion started.

In the calculation shown in Table 5, it is assumed that gas release occurs at the middle of the top fuel node in the active core when the clad temperature is near 1300*C. The assumption.t6at clad failure occurs at the top of the active core may be conservative, since, depending on the condition of the fuel-cladding gap, it is likely that initial clad failure might occur somewhat! earlier at a point farther down in the core. The failure zone is likely to propas 'e upward and reach the top of the core, but by this time, the pressure in the fission gas plenum would have already been reduced somewhat. Such,behaviour was observed in the R8 test for pin number 6 which had been ejected upward out of the core region. As noted in Reference 11, the top of the long axial rip was 5.0 cm below the top of the active fuel and extended downward to about 9.4 cm where the cladding effectively severed. Thus, the assumption of initial failure at the top of the core probably prolongs the t'me required to remove the plenum gas by some undetermined time interval. On:a the release begins, the pressure in the channel rapidly increases to values as large as 5 to 6 atm. This high pressure temporarily chokes off vaporization of any sodium film that may remain on hot cladding near the bottom of the active core. In the meantime, the cladding temperature continues to incre'ase. The high channel pressure ejects any sodium slug that e -+.--.y_ y ,~,-- .g--r,yveg.,w..e,, gg

O -3 9-may remain in the top of the subassembly and often ejects the lower sodium slug from the bottom of the subassembly. While these events are taking place, the mass flow rate of gas from the plenum is decreasing, and the pressure in the coolant channel begins to drop nonsonotonically. As a result of the nonmonotonically dropping pressure, vaporization may resume intemittently in the lower part of the channel, and through much of the time required for the gas to be completely exhausted from the plenum, gas and vapor flow in the C active core may be alternately upward and downward. The implications of the gas release on the motion of molten cladding depend on the ccupling between the clad and the streaming vapor. While the motion of gas and vapor alternates between upward and downward, when upward motion does occur, vapor velocities may be very high. As a result, with the normal flooded two-phase friction factor used in SAS30, initial clad motion tends to be upward, sometimes leading to significant clad motion reactivity insertion rates. As noted earlier, this kind cf motion is also predicted in the SAS3D analysis of the R8 TREAT test, and leads to a calculated upper claddinfblockage that was not observed in the experiment. This result provided motivation for modifying the code so that the user could specify the use of nominal single phase friction coupling between clad and vaper. As already noted, when the nominal single phase friction factor is used in the R8 analysis, clad motion tends to be predominately downward and an upper cladding blockage is not predicted, a result more consistent with experimental observation. When the nominal single phase friction factor is used in the whole core analysis, the initial clad motion in most channels tends to be downward. At the time gas release stops or shortly thereaf ter, cladding frequently fills the coolant channel near an axial location about one third of the way up in the active core and begins to move upward, driven primarily by l

the coolant vapor pressure drop. Clad does not move up coherently in all, channels; the time when upward motion occurs is delayed depending on the time of gas release in the channel. This upward clad motion subsequently leads to positive reactivity insertion rates, but these rates occur at a time when the reactor is suberitical because of fuel motion in channel 6. At about the time t upward clad motion is established in chcnnel 2, fuel in channel 6, which has been initially dispersed by' fission gas, begins to fall back into tne core. The fuel fallback, together with the upward clad relocation, is responsible for the power increase that leads to the initiation of fuel motion in channels 2, 4, and 7. The transient described above differs in 3everal ways from the transient predicted in the response to question CS760.178A3. The event sequence for this latter case, Case 2, is reproduced in Table 6 for ease of reference. The first. noticable effect of the larger sodium void worth in Ute present calculation is that initial boiling occurs about 0.8 s earlier in Case 1 than. t in Case 2. At the time of initial boiling, the net reactivity is 2 cents i higher in Case.1 than in Case 2, but the void reactivity is nearly 4 cents l higher. The lower increase in the net reactivity i: caused by an increased f fuel temperature resulting in a combined Doppler and axial expansion feedback with magnitude nearly 2 cents higher in Case 1 than in Case 2. The increased fuel temperature, in turn, leads to the earlier boiling time. A second difference between the two cases is the fact that the time between initial boiling and the final shutdown in the initiating phase is more than 2.5 s longer in Case 2 than in Case 1. At least four factors may contribute to the shorter time span in the present case. The first of these is the higher sodium void worth. Based on a comparison between the present case and a case in which clad motion was not permitted, one can estimate that

.. x. w, - +, . ; ~:.. e'* - u., ....y ,.u ,c.v,C P d. m. .,j., + ',- .t' <;..o ;s-a=.- v.. .s.e..-- +.?. ,n Table 6 Event sequence for Case 2 TlHE EVENT CllN

• P/PO Ril0 Ril0P RIt00 Ril0E Ril0V RHOF RHOC 12.7655 COOLANT BOILING 6

0.821 -0.094 0.0 -0.140 -0.050 0.096 0.0 0.0 14.6697 COOLANT BOLLING 2 0.819 -0.068 0.0 -0.156 -0.057 0.145 0.0 0.0 s 15.0561 COOLANT BOILING 4 0.817. -0.069 0.0 -0.161 -0.061 0.152 0.0 0.0 15.7772 COOLANT BOILING 7 0.851.-0.019 0.0 -0.170 0.068 0.219 0.0 0.0 17.1048 COOLANT BOILING 10 1.226 0.234 0.0 -0.235 -0.125 0.594 0.0 0.0 17.1998 COOLANT BOILING 11 1.269 0.253 0.0 -0.242 -0.132 0.627 0.0 0.0 17.5298 COOLANT BOILING 9 1.321 0.252 0.0 -0.270 -0.159 0.681 0.0 0.0 11.7792 COOLANT BOILING 13 1.241 0.190 0.0 -0.287 -0.175 0.653 0.0 0.0 17.9242 CLAD HOTION 6 1.320 0.233 0.0 -0.297 -0.186 0.715 0.0 0.0 18.2117 COOLANT BOILING 12 2.225 0.514 0.0 -0.333 -0.223 0.999 0.0 0.071 18.6442 CO0lANT BOILING 15 2.570 0.531 0.0 -0.397 -0.279 1.130 0.0 0.076 18.8732 COOLANT BOILING 14 3.002 0.561 0.0 -0.440 -0.315 1.241 0.0 0.075 i 19.1867 CLAD HOTION 2 2.889 0.498 0.0 0.495 -0.366 1.282 0.0 0.076 l 19.3417 f, LAD HOTION 4 3.770 0.594 0.0 -0.521 -0.390 1.238 0.0 0.273 19.3517 FUEL HOTION 6 3.784 0.590 0.0 -0.532 -0.393 1.233 0.0 0.283 19.4129 COOLANT BOILING S 3.695 0.570 0.0 -0.544 -0.400 1.200 0.005 0.308 19.4930 PEAK REACTIVITY 0 4.654 0.644 0.0 -0.563 0.411 1.207 0.003 0.400 33 19.5017 PE AK POWER 0 4.670 0.643 0.0 -0.566 -0.413-1.201 0.002 0.419 19.6092 CLAD HOTION 7 3.453 0.492 0.0 -0.587 -0.426 1.186 -0.110 0.428 19.6163 COOLANT HalLING 3 3.437 0.488 0.0 -0.588 -0.427 1.186 -0.123 0.441 19.7705 COOLANT BOILING 1 1.744 -0.022 0.0 -0.594 -0.434 1.145 -0.701 0.560 19.7730 COOLANT BOILING 8 1.722 -0.036 0.0 -0.594 -0.434 1.145 -0.712 0.559 70.1267 CLAD HOTION 10 0.984 -0.651 0.0 -0.579 -0.432 1.365 -1.523 0.518 20.1555 CLAD HOTION 11 1.004 -0.601 0.0 -0.578 -0.432 1.389 -1.523 0.543 20.5000 CLAD HOTION 9 1.514 0.007 0.0 -0.588 -0.433 1.565 -1.523 0.987 20.7005 CLAD HOTION 13 1.506 0.019 0.0 -0.597 -0.433 1.586 -1.523 0.986 20.9430 FUEL HOTION 2 1.989 0.259 0.0 -0.610 -0.431 1.731 -1.523 1.092 21.1105 FUEL HOTION 4 2.863 0.470 0.0 -0.629 -0.428 !.773 -1.402 1.157 21.1342 CLAD HOTION 12 2.665 0.425 0.0 0.631 -0.428 1.760 -1.399 1.123 21.5380 FUEL NOTION 7 0.681' -1.324 0.0 -0.6 38 -0.425 1.848 -3.711 1.602 21.8117 CLAD HOTION 15 0.363 -3.377 0.0 -0.634 -0.426 1.864 -5.636 1.4 56 21.8830 TERMINATION 0 0.334 -3.742 0.0 -0.633 -0.426 1.837 -5.904 1.385

• Temf r9109y: CllN stands for the SAS30 channel number; P/PO stands for the normalized power; RHO stands for the net reactivity; ani RH0X stands for reactivity component X where X = P means programed reactivity X = 0 means Doppler, X = E means axial expansion, X = Y means vof ding, X = F means fuel motion, and X = C means clart motion. Reactivities are in dollars.

42-the increased sodium void worth alone shortens the time span by somewhat more than 1 s. A second factor is the higher clad worth. It is difficult to separate this fr.ctor from the third factor which is the actual clad motion. It appears that clad motion along with the increased clad worth also shortens the time span between first bofitng and reactor shutdown by somewhat more than I a second. The contribution of the fourth factor, fuel fallback in channel 6, i appears to be small because the fallback occurs simultaneously with a rapid increase in clad motion feedback. As a result, the power burst that occurs l just after 18 s in Case 1 would have occurred even without the fuel fallback. A preliminary calculation, similar to case 1, indicates that the power excursion resulting from clad motion alone, while somewhat milder than j the present excursion, is sufficient to initiate fuel motion in channels 2, 4, 1 7,10, and 11 and lead to reactor shutdown on about the same time scale as in the present case. The influence of the new neutronics data on the potential for f6f ssion-gas-driven compaction of fuel can be shown with reference to Table 7. The table shows the times between the initiation of gas release and the initiation of fuel motion for each of the driver channels. In interpreting the results, bear in mind that a clad failure temperature of 1400*C was assumed for i estimating the times listed for Case 2 while a temperature of 1300*C was used I in the present case (Case 1). In spite of the lower clad failure temperature in the present calculation, the times are considerably shorter than in the earlier case. While the margin is not as great as it was previously, there is ample time for gas release in the present calculation. The fission gas parameters used in SUJ4PY are the same for both Cases 1 and 2, and are based on a FRAS33,4,5 analysis of the best estimate case in Reference 2. They correspond to a fraction of steady-state fission gas 2.

43 Table 7 Comparison of times between initiation of gas release and the initiation of fuel motion in Cases 1 and 2. Note that the clad failure temperature was 1400*C in Case 2. SAS3D Channel Number Case 1 Case 2 2 1.69 s 2.36 s 4 1.52 s 2.38 s 6 1.86 s 2.58 s 7 1.25 s 2.51 s 9 0.71 s* 1.96 s* 10 0.94 s 2.31 s* 11 1.09 s 2.39 s* 12 0.28 s* 1.54 s* 13 0.53 s* 1.89 s* 14 0.03 s' O.66 s' 15 0.86 s*

• The time.to the end of the calculation since fuel motion did not initiate in this channel.

retained in grains of 54% and a fraction of steady-state gas on the grain boundaries of 4.7%. FR'AS3 calculations were redone for channel 6 using the thermal history obtained in Case 1. The gas fractions based on the new transient'were found to be 70% and 2.7% respectively. Case 1 has not been rerun using the new gas fractions, but based on previous experience using the SLUMPY model, the fuel dispersal computed for channel 6 would not be expected to change significantly. In concluding the discussion of Case 1, we note that the potential for fuel failure into liquid sodium is effectively absent. Figure 10 shows the voiding pattern in the reactor by channel at the end of the transient. It can be seen that' voiding is in progress in all channels and that sodium has been completely removed from the active fuel region in all driver subassemblies except for the lower third of channels 10 and 14. Fuel motion is in progress l in channel 10, but the fuel melt fraction is still below 0.1 in channel 14. ~ i I

i l ~ Q SODIUM ] VOID w (M N E-N 150 - yp 8 Z A N U 1 3 E4 ACTIVE lI: CORE o 5 s pq m r. 4 ^ ^' 0- ~ ^"'i r i i i ^ 0 50 100 150 200 250 CUMULATIVE NUMBER OF, SUBASSEMBLIES Figure 10. Sketch of the Voiding Pattern in the Reactor Core at the End of i Case 1, including the Upper and I.ower Axial Blankets. (Each l Histogram Bar Represents an SAS Channel, and a Few of These Are Numbered to Assist in Identification. Channels 1, 3, 5, and 8 Are Internal Blanket Subassemblies.)

Sensitivity studies have been carried out y n which the E clad failure temperature was changed from 1300*C to 1400*C. The event sequence for'this case, Case 3. is shown in Table 8. As expected, gas release in all channels starts later than in Case 1. This, in turn, causes events subsequent to gas release in channel 6 to be delayed compared to the times of their occurrance in Case 1. These delays are sufficient for gas release to end prior to initiation of fuel motion in five of the first six channels. The times between the start of gas release and the initiation of fuel motion is shewn for these six channels in Table 9. The results show that the reduction in the time between the start of gas release and the initiation of fuel motion is not generally as large as the time delay irt the start of gas release In fact, if the reactivity had not gone slightly above' prompt critical in Case 3 it is likely that all channels initiating fuel motion would have previously stopped releasing gas or have had sufficiently low pressure in the fission gas plenum so that fission-gas-driven compaction would not be a concern. As'can le seen in the event sequence for Case 3 in Table 8. channels 9 10,12,,and 13 begin fuel motion before ses. release has stopped. In the case of channel 10, the pressure in the fission gas plenum is only 3.3 atm when fuel motion starts and is not likely to play a significant role in the remainder of the transient. In the case of channels 9,12, and 13, the pressures are respectively 11.3, 33.2, and 15.2 atm. Thes.e pressures we'uld be high enough to influence the remainder of the transient were it not for the fact that these three channels begin fuel motion after the net reactivity has begun to decrease because of strong Doppler feedback and dispersive fuel motion in channels 2, 4, 6, and 7. To see the effects of fission-gas-driven c:mpaction of fuel in channels 9, 10, 12, 13, and possibly 14 and 15, the compaction model introduced into St.tMPY for this purpose was utilized. The

,g* k 657728503898908038667296008 . C - 000477076990983207657122308 . 0 - 000000000001 02223222220000115775 l 9 - R - 000000000000000000000000000O0000 i C 01 971 361 241 5853 - F - 000928385641325 . 0 - 000000000000000000000801 15552779 i i - R _ 0000000000000000000000111111 1000 . 4738586867411 1562993828393551372 - V. 32925576643709134585803324671372 2127777890223444445565555556779 1 l i - R - 000000000O111111 1111111111 111111 . 883923186'37334237965496664443679 . E - 56775566922684699024555555555555 . 0 - 000011 1 1 122223333444444444444444 l i . R. 00000000000000000000000000000000 2073531194971661660166540854064 1 5708778924592137891 4554443222445 - 0 _ 0. 1 111 2222333345555566666666666666 3 l i eR. 00000000000000000000000000000000 s a C r._ 0li P. o - 00000000000000000000000000000000 8f ee.R. 00000000000000000000000000000000 l c bn_ ae Tu q. .4198057113541 8571465609762345994 D.715333300070691 67521 036397975244 e 0000333333355787777761 56611 ._ I 1664 1 s n.R._ 000000000000000000000000011 t 1 0000 e l" i v E _ 32605032803997391506201385500299 - O _ 60602460621 81605961 81 2261 0800952 - P _ 09385555568497080733593229085603 - /. P _ 0000111111122797877751111 0001453 . N _ 624755019632224465747S31 10104989 11 11 1 1 1 111 l l e - C - . GGGG GGG GG G G GG 'G G - NNNN NNN NN N N NN N N - IIII III II I I I I I I - LLLL LlLELL L L LL L .E L . !lI I NSI l ISIl SNSNNI SI NI I SI SNNSNI S - T 00OOOAOoOAOoAOAOOOAOOOOAOAOOAOOA - N - BBDDIGDDDEDDGI GI IDGDI DDGDGII EI DG - E _ T L T TT T TTLT - V _ TTTTOETTTETTEOEOOTETOTTETEOOEOTE NNNNMSNNNRNNSMSMMNSNMNNSNSMMRMl S E _ l _ AAAA AAAA AAA A AAA AAAAA AA . LLLLDELLLPLLEDEDLLELDLLELED0PDLE . OOOOALOOOOOOLALAEOLOAOOLOLAAOA0L . OOOOLEOOOTOOELELUOEOLOOEOELLTL0E . CCCCCRCCCSCCRCRCFCRCCCCRCRC.CSC0R - 19895550509700505020570000722027 ._ 573961 6631 726636367134661 64271 24 28506842982605834606913458149461 _ E - M _ 9403667803468990001 11 34445667990 I 3345555666666677777777777777770 T. 1 1 1111 1 1 11 1 1 1 11 1 1 11111 1 11 1111 11 1 1 e l

Table 8 (cont'd) Event sequence for Case 3 [ l -i TIME EVENT CHN* P/PO Ril0 Ril0P --_____________.____________________________________________RH00 RHOE Ril0V RHOF RHOC 18.0272 STOP RELEASE 2 2.801 0.335 0.0 -0.654 -0.459 1.959 -1.077 0.566 18.1085 RELEASE GAS 13 1.666. -0.126 0.0 -0.648 -0.459 2.171 -1.816 0.627 18.1347 CLAD MOTION 13 1.569 -0.190 0.0 -0.648 -0.459 2.198 -1.984 0.703 I 18.2222 STOP RELEASE 7 6.503 0.723 0.0 -0.665 -0.461 2.156 -1.492 1.184 i 18.2360 FUEL MOTION 2 13.645 0.873 0.0 -0.677 -0.462 2.141 -1.424 1.294 i 18.2397 STOP RELEASE 11 19.285 0.914 0.0 -0.682 -0.462 2.139 -1.406 1.325 18.2470 FUEL MOTION 4 46.656 0.977 0.0 -0.700 -0.465 2.126 -1.372 1.388 18.2501 FUEL MOTION 7 88.822 1.003 0.0 -0.716 -0.465 2.124 -1.357 1.418 i 18.2523 PEAK REACTIY1TY 0 154.934 1.013 0.0 -0.733 -0.465 2.124 -1.350 1.437 18.2549 FUEL MOTION 10 258.308 1.003 0.0 -0.770 -0.464 2.125 -1.343 1.456 18.2555 FUEL MOTION 11 274.136 0.997 0.0 -0.781 0.464 2.125 -1.342 1.460 18.2561 PEAK POWER 0 280.618 0.992 0.0 -0.791 0.464 2.125 -1.342 1.464 e 18.2567 FUEL MOTION 9 278.637 0.985 0.0 -0.803 -0.463 2.126 -1.343 1.468 0 18.2586 FUEL MOTION 13 216.685 0.962 0.0 -0.834 0.461 2.127 -1.351 1.480 18.2593 FUEL MOTION 12 183.196 0.952 0.0 -0.843 -0.460 2.128 -1.357 1.485 18.2717 RELEASE GAS 12 3.673 0.040 0.0 -0.866 -0.455 2.145 -2.385 1.601 18.3480 CLA0 MOTION 12 0.416 -8.591 0.0 -0.808 0.456 2.339 -11.583 1.917 i 18.3730 STOP RELEASE 10 0.366 -10.027 0.0 -0.798 0.456 2.375 -13.283 2.135 18.4168 RELEASE GAS 15 0.346 -10.359 0.0 -0.791 -0.457 2.404 -13.696 2.180 18.4893 TERMINATION 0 0.311 -11.238 0.0 -0.799 -0.457 2.103 -14 ___.-_-__-___-..._....._-_....._.---___._--_-___.--____--___-..-__-__.....,__..__..._____.183 2.097

• Teminology:

CHN stands for the SAS3D channel number; P/PO stands for the nomaltzed power; RHO stands for the net reactivity; and Ril0X stands for reactivity component X where X = P means programed reactivity,'X = D means Doppler, X = E means axial expansion, X = Y means voiding. X = F means fuel motion, and X = C means clad motion. Reactivities are in dollars. I 1 l Table 9 Comparison of times between initiation of gas release and the initiation of fuel motion in Cases 1 and 3. SAS3D Channe.1 Number Case 1 Case 3 2 1.69 s 1.43 s .~ 4 1.52 s 1.26 s 6 1.86 s 1.36 s 7 1.25 s 1.14 s 10 0.94 s 0.67 s 11 1.09 s 0.81 s event sequence for the resulting case, Case 4, is almost identical to that for Case 3. The dispersive fuel motion from channels 2, 4, 6, 7, 10, and 11 (representing 84 subassemblies) was sufficient to overcome the reactivity-insertion rates produced by the compacting fuel in channeli 9,12, and 13 (representing 36 subassemblies). Channels 14 and 15 did not init,iate fuel motion and had peak. fuel melt fractions between 0.35 and 0.4 when the transient ' ended. Table 10. shows the work-energy obtained when super-saturated fuel is e,xpanded adiabatically to a final pressure of 1 atm for Cases 1, 3, and 4. These results show that the LOF transient using the modeling assumptions of Case'1 is not sensitive to the choice of the clad failure temperature used in the fission gas release calculations. e e l 9 e w me=ww & me W e ww w -s * ~ ~ * *-~ ~' '

-49 i I i l Table 10 Work-energies based on adiabatic expansions of super-saturated fuel to a final pressure of 1 atm. Case Work-Energy, K1 1 0.6 3 4.3 4 5.6 e

• S O

e 1 1 ,e 0 e O e l 8 --7---. -y

e s IV. Conclusions , The assessment of sodium void coefficient uncertainties has resulted in an increase ih the nominal values of s' odium void and clad material worths in the CRBRP heterogeneous core. These changes have increased the sensitivity of wiole core analysis results to the modeling of important phenomenology. The t importance of representing fuel disruption and dispersal consistently with the experimental database has been previously established. It has been d6monstrated in'this report that similar experimentally-based models can be developed in the areas of molten cladding relocation, the effect of release of plenum fission gas on sodium vapor dynamics and clad motion, and failure of irradiated cladding under the fission gas plenum pressures. The whole core best estimate analyses have shown that with such experimentally validated models, a mild power burst with near zero energetics is expected. This conclusion is valid even in the unlikely event that.the plenum fission gas can act to compress the disrupting fuel. Parametric variations on clad failure and plenum gas release, and molten cladding relocation show very mild sensitivities in initiating phase energetics. The potential for significant energetics appears to require pessimistic phenomenological modeling that is not supported by the present experimental database, and is therefore beyond that appropriate for a, realistic assessment of the accident energetics. The likelihood of energetics approaching the SMBDB value is very remote. O O O e _e O

-51 Appendix A Modifications to the SAS3D Boiling Module to Account for Release of Plenum Gas into a Boiling Region The gas voiding model in Version 1.0 of SAS3D is mainly applicable to voiding due to pin failure and gas release before the onset of boiling. If a pin fails in a boiling region, this model will calculate the flow of gas from the gas plenum to the failure point, but the only effect on the voiding is a reduction in the condensation coefficient of the vapor. This model has been modified to provide a better treatment of the impact of gas release into a boiling region. In the modified *model, gas released into a vapor bubble is treatec as an additional equivalent vapor source in the boiling model. This additional I vapor spurce is added at the one axial node where the pin ructure occurs. Since the molecular weight of fission ' product gas is different from that of sodium vapor, the mass of the gas leaving the rupture aust be converted into an equivalent vapor mass for use in the boiling calculation. In the SAS3D boiling model, a vapor bubble is treated as either a small bubble, with unifom vapor pressure, or a larger bubble containing pressure gradients due to streaming vapor. For gas released into a small bubble, the product, pV, of pressure times volume is conserved when the gas is converte't to vapor. For a perfect gas pV = mRT (1)

4, so aRTg g g = myRyTy (2) where g = mass of gas leaving the rupture, m my = equivalent mass of the vapor source, Rg = gas constant for the gas, Ry = equivalent gas constant for sodium vapor Tg = temperature of the gas leaving the rupture, assumed to be equal to the fuel surface temperature at the rupture point, and Ty = vapor temperature at the rupture point. Note that Tg and Ty must be absolute temperatures (K). Equation 2 gives ART 999 (3) m = y RT vv l The vapor gas constant, R, is calculated as y

P y .R (4) = y aT vv where av is the saturation vapor density and pv is the saturation pressure at temperature Ty. 'For gas released into a large vapor bubble, the friction pressure drop due to streaming gas or vapor is conserved. The friction pressure drop is G2 L APf*fg3 (5) where f = friction factor, G = mass flux p = density, L = length, and D = hydraulic diameter.

= .o 8 1 1 . Therefore, g2 - g2 y L g L 'ff-fyy* yfg g (6) y g i Differences between the vapor and gas friction factors are neglected. Al so, ' G is assumed to be proportional to the mass released at the rupture, so i 5 i ) m2 m2g ) Tii-

• Fo -

(7) y g 4 1 or IRT p y gg m =m =m (8) y g a g RT a g a yy For a small bubble, the heat flow to the bubble during a time step has a term AE o added to the tern E of Eq. 131 in ANL-813837 This term is given t to by O e

AIto

• I"gp1 ~ "gp2I IN R,T, i

where agp1 = mass of gas in the plenum at the beginning of the step agp2 = plenum gas mass at the end of the step and A = sodium heat of vaporization. y The FORTRAN variable name for E o is DQT(1), and the term is added in t subroutine TSC43A. For gas release into a larger bubble, a ters aQ, is added to the heat flux from the clad to the coolant (see Eq.153 of ANL-8138) at the rupture node in subroutine TSC4A. This term is RT E 40 = (m -m e gp1 gp.,) (10) RT atA az s where at = time step size, Ace = coolant flow area, and az = node size The boiling module in version 1.0 of SAS3D will stop the code if a vapor bubble extends out the bottom of the subassembly. This is because some FORTRAN subscripts for arrays used in the vapor pressure gradient calculation

would be equal to 0 if the bubble interface is below the lowest channel node l (node 1), and a subs ript of 0 is not allowed in FORTRAW. Ges release into a boiling region can often lead to voiding out the bottom of the channel, so the l _ code was modified slightly to handle this case. Now if a bubble extends below -the bottom of the channel, the lowest clad and structure node are ignored in the coolant calculation, and the inlet coolant temperature is used for the clad and structure temperatures at the liquid-vapor interface. The lowest vapor node then extends from the liquid-vapor interface up to node 2. By ignoring node 1 in this case, the subscripting problem is bypassed, and the calculation can continue. 1

. } i Appendix B Two-Fluid Model Analyses of Plenum Fission Gas Release The SAS3D code is ifmited in its ability to treat the release of plenum fission gas into a sodium vapor filled channel. The theoretical basis for De current treatment has been provided in Appendix A. To confirm the modeling l and results from SAS3J and to insure the model is conservative, an independent l analysis of the plenum gas ejection scenario was undertaken. No single ( i analysis capability was available that treated all aspects of this problem y and, consequently, two methodologies from the PLUT02 code and the TRANSIT-HYDRO code were employed which treated several factors not included in SAS30. The :;ualitative agreement between these three methodologies provides confidence in the SAS3D treatment and the quantitative comparisons indicate l tha the SAS3D treatment is, indeed, conservati' 2 B.1. TRANSIT-HYCR0 Results f To, evaluate the impact of approximations made in modeling the plenum fission gas injection process with the SA530 coolant dynamics module, the I TWOFLU module of the TRANSIT-HYDRO computer code has been used to simulate i plenum fission gas injection into a partially voided subassembly. The TWCFLU fomulation is based on a two field (liquid and vapor), three componant (fuel, clad, coolant) structure in which each field has an independent veiocity, and t each component within a field has an independent internal energy. In i addition, mass conservation is maintained for each component in each field. t For the fission gas injection simuistion, this permits independent tracking of the fission gas and coolant vapor components, and eliminates the need for l energy and mass mixing. The particular issue addressed here is the timing and i h

extent of downward fission gas penetration, and the validity of the assumption of the loss of condensation potential following cladding rupture in the SAS3D coolant dynamics todel. }, Initial conditions for the 1MOFLU simulation were taken as those predicted by SAS3D in channel 2 at the time of cladding rupture for the best estimate EOC-4 loss-of-flow analysis described in Section III. These conditions included geometry (channel length, flow area, liquid slug location, coolant film thickness and location), as well as thermal (axial temperature j distributions in cladding, liquid slug, and coolant vapor), and hydrodynamic i conditions (liquid slug and coolant vapor velocities). Thermodynamic and transport properties for all materials were taken from those employed in SAS30. Momentum frictiera1 modeling for the liquid and vapor were also taken from SAS30, including the two-phase m.eltiplier employed to represent the effect of liquid coolant films. To describe the gas injection process, the plenum gas prtssure and temperature fannulation, as well as the gas flow [ formulation (subroutine PIPFLO) used in SAS3D were implemented intact in j t TWOFLU. The time history for the pressure boundary condition at the l subassembly inlet and outlet were taken from the best-estimate SAS3D LOF analysis described in Section III. I ~ Given these initial and boundary conditions, TWOFLU predicts channel j i pressurization and liquid slug reversal similar to SAS30. Figure B1 compares the channel pressure history following clad rupture at the rupture site as f predicted by SAS30,TWOFLU and PLtff02. The comparison shows that TWOFLU j predicts a somewhat higher and more sustained pressure pulse than that pre-t dicted by SAS30. Close examination of analysis results shows that this is due j i to much lower gas temperatures predicted by SAS30. Inclusion of gas / structure heat ' transfer effects would tend to lower the TWOFLU-predicted temperatures, f I l t f

RUPTURE SITt PRESSURES , % s.O i i 7.0 - + i- -{ i 6.0 - v s- ~- .i j ta 5.O ~ i-M D I) gi (n 4.0 r{,g-l m ' f J. - I-- Ei-

s

.o

Qd.......i.5%;...

Le9end a i ...g % 3.0 t.. ......3.. wY: ' tr i M 0 a SAS3D -- Q i i 2.0 u K TWOFLU O Pt.UTO2 1.0-i i i i i i 0 100 200 300 400 500 600 700 800 TIME, MSEC Figure 81. Rupture Site Pressure illstory Comparison

but not to the levels predicted by SAS30. The TWOFLU pressure distribution history shown in Fig. 82 causes a more rapid expulsion of the lower liquid slug than is predicted by SAS30. In ddition, the-slug motion is more extensive (it is expelled from the bottom of .a 't'he subassembly), and slug re-entry is delayed compared to SAS30. This effectively lengthens the time frame for negative vapor velocities. Sensitivity studies have shown that augmentation of the subassembly inlet pressure by approximately 0.1 MPa (1 atmosphere) in the TWOFLU calculation results in slug re-entry times close to the SAS30-predicted value. The TWOFLU analysis shows that downward penetration of the plenum fission gas is rapid and extensive. Figure B3 shows tihe TWOFLU-predicted mass fraction distribution for fission gas following clad rupture. During this time, the liquid slug has reversed and is being expelled due to the fission gas pressurization. As the figure shows, even at early times fission gas has effectively penetrated down to the liquid slug interface, compressing a small amount of coolant vapor trapped below the rupture site. In addition, the fission gas has swept all of the coolant vapor formerly above the rupture site out through the top of the subassembly. Based on the TWOFLU analysis summarized here, it can be concluded that in the event of ciad rupture and plenum fission gas blowdown in channel 2 of the best-estimate LOF analysis of Section III, the released fission gas rapidly pressurizes and fills the channel. This analysis indicates that the assumptions made concerning coolant vapor condensation reduction due to. fission gas blanketing of liquid coolant films in the SAS3D coolant dynamics model are reasonable and justified. In addition, this analysis indicates that channel pressurization may be somewhat higher than would be predicted by SAS30. This would tend to enhance the potential for downward molten cladding 0 mm a

TWOFLU CHANNEL PRESSURES 3.5 3- -. ..j.... ..).... . j.. ... j.. 2.5 - ^5-N l-M ' ' nN - i 2 2- %. g..:,..... g.%

a.

-.y-~~. H .N .i x i. O_ f.5 - +n. ..a.. 4 N.. Le9gM W I n 3 o

t:

a 1_ ...n.....g. ...g........ A _80 MSEC e u 5 m t i d: m a, i' W-s'- x NO MSEC, X 0.5 - -.. ~. n y d 0 200 MSEC 0-p: n 2 Fi P ja ,i f a 280 MSEC

n d:

p ---h425h- - -l

• 4.90..MS.EC

-0.5 - g[-

er

.i ! M _y. M 720 MSEC i i r i i i 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 PRESSURE,PA

• 10F Figure B2. TWOfLU Predicted Channel Pressure Distribution llistory i

I e

37~,,... - - ye ,4 .c y,..... , D [_ r V., <-.;J ' % ~ ?,f f  ; "',.

• a

' l,.. p i ~ ;, , J~ e ;,.. ; ,5.,,_.S. ' A l _

p.,.

1 '.+...m.: n c. 4 f g.# va s.7 e t TWOFLU FISSION GAS MASS RATIO 3.5 .f... . N......!........ 1 1 Legend 3 i A 25 MSEC ~ i-i- 2.5 - i-50 M.SE.C. n x i 2 ......i............ ; ........i.......... O 75 MSEC i 2_ g p I i a 100 MSEC i---!-- -i U O 1.5 - E 12. 5..M..S.E.C.. Gi i i. n y M 15_0_M_S_E_C._

g...........................g...........................g.........

i b .l -*= A- - NyA-I .i ..- E- + 175 MSEC X 0*5 - /-~~ w: ,V ~y,. B f .. N'. y -,.W jf e e 200 MSEC 4 i i o 225 MSEC /p.y 0 7 7 7.

p xi.

n: + 250 MSEC W,- --l I 5

^*: -l n7 sleif /

.e -0.5^- o 300 uSEC .x w-p. n i i i i 0 0.2 0.4 0.6 0.8 1 Mfg / Mfg + Mno Figure 83. TWOFl.U Predicted Fission Gas Distribution llistory

f motion by lengthening the time frame for negative gas velocities. This observation is consistent with the experimental results seen in the R8 test (no upper blockage, downward clad motion) and supports the assumption of nominal single phase frictional coupling in SAS30/CLAZAS. B.2. PLUT02 Results An investigation into the effect of plenum fission-gas release on the channel flow behavior was also made with a special version of the SAS4A/PLUT02 code. This was done in order to verify the new SAS3D modeling of plenum gas release into a boiling channel. The SAS3D base case calculation, which uses the new fission-gas / boiling model, calculates 'a downward motion of the vaper-gas mixture in the active core region for about a hundred milliseconds follow-ing the onset of plenum gas release. In the model the plenum gas injected into the channel is replaced by an appropriate amount of Na vapor and the sodium vapor condensation for the entire channel is set to a small value. The validity of these assumptions was investigated in this study with SAS4A/PLUT02 which which has the capability of treating sodium and fission gas separately. Although the better known features of the PLUT02 module are the calculation of in-pin and channel fuel motion, it also has a fairly detailec treatment of two-phase sodium and fission-gas flow in the coolant channel. A stagnant liquid sodium film which can evaporate or be entrained by vapor flow is also modeled. In the current application the in-pin fuel motion and fuel ejection from the fuel pins was turned off. Plenum gas was injected into the coolant channel at a rate similar to that calculated by the SAS3D base case for channel 2. All geometrical and thermo-hydraulic data used in the single channel SAS4A/PLUT02 calculation were the same as the data for channel 2 in the SAS30 whole core base case. The power and inlet pressure history

s l necessary for the single channel calculations were also from the SAS3D whole-core calculation. Extensive sodium boiling and voiding took place in the curre'nt calcula- - tions before cladding failure was assumed to occur at the top of the active ~ core when the clad midwall temperature at this location had reached 1300*C. The PLUT02 calculation with the plenum fission gas injection was then initi-ated. The PLUT02 calculated pressure history at the rupture site is shown in Fig. B1 together with the SAS3D and the TWOFLU calculations. Both PLUT02 and TWOFLU predict a longer lasting and higher pressure peak during the first'300 msec because both models can account for heat transfer from hot, dried-out clad to the gas in the coolant channel, whereas the SAS30 fission gas / boiling model does not account for any heat transfer from the clad at any node where the film has dried out. This causes the gas or superheated vapor temperature ~ fn PLUT02 and TWOFLU' to be several tiundred degrees Kelvin hotter than in SAS3D and also makes the pressures in PLUT02 and TWOFLU higher. At the later times both PLUT02 and SAS3D shew pressure increases which are caused by the sodium film vaporization at the lower end of the active core. In Fig. B4 the lower sodium slug interfaces calculated by SAS30 and PLUT02 are compared. The PLUT02 predicted slug ejection is more rapid and lasts longer than in SAS3D because of the higher pressure calculated by PLUT02. The PLUT02 calculated pressure distributions in the coolant channel at different times are shown in Fig. 85. The bottom of the active core is at about 0.34 m and the top of the active core, where the plenum gas injection takes place, is at about 1.25 m. l The pressure distribution at 80 msec shows a peak at the plenum gas ejection site which corresponds to the maximum pressure achieved in this run. This overpressure caused the vapor / gas flow below the failure site to move

SLUG INTERFACE LOCATION 1 .i 0.75- -s 3 '2E i i i 0.50 ,? +:i+!-- Z i i i. O__ Q 0.25-i j T - l. t - j o O i 1 _J 0-y 4Jt e i i L) .-+ <C -0.25 - t-12. oc i i La.) i i i

5.......... :...... ;.... y w

_o,5 o. ..... ;.......... 5.. i Legend 5 i i i -0.75 - - -i - ?-

--i=-

i-......- A SAS3D i MMXU N Hh.MMMA. 5 i x PLUTO 2 i i N i i i i i i i O 10 0 200 300 400 500 600 700 800 I TIME, MSEC 4 i figure 04-Lower Sodium s1u9 Interface Comparison r } J 4

PLUTO 2 CHANNEL PRESSURES l... 3.5 3_ s ,' N....................:........... s h N i -iN.;& %; '----{~--i-2.5 - i \\\\ s.i k..h: i i 2 2_...........j... ...r.,.

y....K.. :.

..........i........... I ...}. .g........ :. .....,.........g... ....j........... O._. j,$ _ ........... 3.. w s i( i Legend I I .........i........... ).......... g .t.g:... *......_:......L..:............ a g a. a 80 MSEC j _. 38i N x 160 MSEC - ; e' "o **:" h ".- "::-" 2'"! [ X 0.5 - C a- - $ n" " - ~ ~ " - 4-9 o

c O 200 MSEC Y u.

.t 0 i a 280 MSEC h i E I,3 : x i i A

? -

t : -n -9 :i k i u 420..M..S..E.C. -0.5 - s-- u x 720 MSEC i k; a n., -1 i i i i i i 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 \\ PRESSURE, PA

• Kf Figure B5. PLUT02 P.redicted Channel Pressure Distribution History I

8 a

downwards and also led to rapid flow reversal of the lower liquid slug which can be seen in Fig. 84. The pressure distribution at 100 usec was decreased due to the upward streaming of the gas and the rapid downward motion of the lower sodium slug which is uncovering cold clad and structure in the inlet region. By 420 msee the pressure gradient below an axial location of 1 m is still downward preventing the gas vapor flow velocity below this region from becoming positive. By 720 msec the fission gas injection pressure has dropped below the inlet pressure causing a slight pressure tilt towards the outlet in the active core region. Liquid sodium film evaporation and sodium vapor condensation do not play a dominant role during the first 720 msec. The sodium film vaporization occurring at around 0.6 m keeps the pressure level at that location somewhat higher than predicted by TWCFLU at 280 and 420 msec. In Fig. B6 the ratio of fission-gas mass over fission-gas mass + sodium vapor mass is shown at different locations and times. The two symbols shown for each curve depict the axial extent of the fission-gas region which corres-ponds to the boiling region at times greater than 30 msec. At later times the fission, gas completely dominates the sodium vapor in the dried-out regior, between 0.6 and 1.25 m and also in the lower blanket and in~1et region where relative cold, low density sodum vapor exists. Between 0.25 and 0.6 m as well as above 1.25 m if quid sodium film vaporization leads to the higher density sodium vapor which is noticeable on this mass ratio plot. Because the fission gas is spreading quite rapidly in this PLUT02 calculation, it can be concluded that the SAS3D assumption of using a very small condensation co-efficient is reasonable because most of the pressure in this calculation is actually due to the pressure of the noncondensible fission gas. The SAS30 prediction of a sodium vapor flow reversal lasting only a hundred milliseconds is significantly shorter than the PLUT02 prediction of

t I PLUT02 FISSION GAS MASS RATIO Mc 3.5 ......!...............::.............. i................::...... j.... 3_ --i:: A-2.5 -- ':E -<li 2-g T. i ? - !k - . O 1.5 - -s iZi i 3: ...............i.................i................;...........i....... -_ j_ m <( i

w. '-

v- - Legend r X 0.5 - s -~ + 4 i a 20 MSEC 0-x i00 MSEC g) I O 200 M.S.E.C.. -0.5 - + i +:- as i i e 300 MSEC. i: --I i i i s' 0 0.2 0.4 0.6 0.8 1 Mfg Mfg + Mno Figure B6. PLUT02 Predicted Fission Gas and Sodium Vapor Mass History O

. about 400 msac. However, the SAS3D prediction is conservative because it leads to a more limited downward clad motion than a calculation with a more extensive vapor flow reversal. The main reason for the shorter lasting flow reversal is probably the lack of heat flow from dried-out clad in the SAS30 model. Moreover, it may be that the small condensation coefficient applied to the converted gas in SAS3D is still causing significant condensation and loss of gas mass over longer times. An attempt was also made to investigate the effect of intra-subassembly incoherencies in the clad failure. This was done by injecting the plenum gas into the channel at about one quarter of the initial injection rate of the previous case. This injection lasted for 1200' msec compared to 550 msec in the previous case but led to the same total gas injection. The same inlet pressure history was used as in the previous case. The time period of negative or icw vapor flows in the active core region lasted for more than i see which is more than twice the value of the previous case. Apparently keeping the pressure in the gas injection node above the inlet pressure for a longer t,ime has more impact than having a higher initial pressure which drops below the inlet pressure more rapidly. This indicates that the assumption of releasing the plenum gas from all pins simultaneously is also conservative i with regard to the potential upward molten of the cladding. r e M

' Appendix C SAS3D Modifications Required to Analyze l TREAT R-Series Coolant Hydraulics ~ On the R8 test, a flow orifice was put in the coolant inlet pipe upstream from the test section to simulate the pressure drop of the inlet orifice in FFTF subassemblies. This orifice is nomally modelled with an inlet orifice coefficient in SAS3D. In SAS3D calculations for R8, the gas release following pin rupture leads to voiding of the whole test section and expulsion of the lower liquid slug from the bottom of the subassembly. When the lower liquid i slug is below the subassembly inlet, SAS3D doe's not account for orifice or-friction pressure drops in computing the motion of the ifquid. The motion is based only on inertia, as driven by the difference between the inlet plenum pressure and the bubble pressure above the liquid slug. In the R8 test, the l inlet orifice was located far enough upstrear; that the gas will never void through the orifice, so the orifice pressure drop should always be accounted for, even if SAS3D predicts expulsion of the lower liquid slug from the bottom of the subassembly. For use in the R8 analysis, a special version of subroutine TSC2 was produced. In this routine, the inlet orifice pressure drop is accounted for in the equation for the motion of the lower liquid slug, even after the liquid slug has blown out the bottom of the subassembly. The load module for this modified routine is stored in data set C112.822404.SAS30 MIS. L9AD(TSC2R8) on the ANL computer system. The modifications, in UPDAT format, used to produce this routine from the SAS3D version 1.0 source are listed in table C1.

i . 1 Table C1 TSC2 Modification for R8 Calculations 100100

• IBM
• NOLIST
• 0RIGIN 2
• REWIND 2
• SUBS TSC2.192 TSC2.194 C

SLUG BLOWN OUT BOTTOM, INCLUDE ORIFICE FOR R8 XIOR1(K)=XK01 XIOR2(K)=XK02 SGN=1.0D0 IF (G1(K).LT.0.000) SGN-SGN AA0 ( K ) = SGN* 0. 5 DO* ( X IOR1 ( K )+X IOR2 ( K ) )

• G1 ( K ) ** 2+ P DCM* ( P TP 1 ( K ) -P BT 1 ( K 1))

BB0(K)=5GN*XIOR2(K)*G1(K) DG0 ( K ) =- DELT* AA0 ( K ) / ( XLL ( K ) + BB0 ( K )

• 0ELT )
• FINI
• END

-e

' Appendix D Calculation of Plenum Gas Blowdown Coupled with Pressure-Driven Fuel Motion A simple finite-difference code was written to calculate plenum blowdown, coupled with downward motion of the fuel and (optional) upward motion of the plenum. The escape of gas from the plenum region is calculated from the relation given by Chawla et. al. for isothermal flow 38 It can be shown from eqs. 25 and A.6 of that reference that the rate of pressure change is given by A 2 ~ 1/2 RT /2 g t_3 P = -P 7 F ,P - I ns, where P is the plenum pressure, R is the gas constant, T is the Kelvin is the flow area of the gap, Y is the plenum volume, S is the temperature, Ag ratio of the channel pressure to the plenum pressure, and F is given by F = 2Af /D (2) g g Here 1 is the length of the flow path, f is the friction factor (taken to be g constant, at 0.01, after Chawla), and Og is the hydraulic diameter (Dg = .0284). The geometry is depicted in Fig. 01, which shows the plenum region overlapping the blanket fu,el a distance 1, (initially, t = 14 inches, or 35.56 3 cm). The initial volume of the plenum Vp1 is taken to be 21.09 cm, and the mass of the plenum structure mp1 is estimated at 85.5g. The fuel mass mr is assumed to be half the mass of the active fuel in the. pin, added to the mass j of the blanket fuel, for a total of 155 g. l

. / //// 6 dM r a P 'I Blenkt L /4 f \\s \\ F e.1 s A Pa l l Figure 31. Assumed Simplified Geometry of the Pin Stub and Plenum Region I l l

Fuel action is calculated as a result of the forces due to gravity, , (Y) ~ F =m.g, g f and due to the excess of the plenum pressure P over the channel pressure Pch. 2 = vr (P - P F p ch where r is the fuel-pellet radius, taken te be- 0.254 cm. The acceleration of the fuel is calculated from the total force, F9+F9, (5) a = f and integrated over small time steps at to obtain the downward velocity i + apt (6) v Tv f f where vf is the velocity at the beginning of the time step. The downward displacement is obtained by calculating the displacement increment in the time step, using the mean velocity for the time step; afat ~ df= dt + (vi + 7-) at. ~ (7) f f

. A parallel calculation is carried out for the upper pin structure, with the difference that the gas pressure and gravity act in opposite directions. Upper movement of the pin structure is limited to an arbitrary (input) value, so that the effect of restraint can be considered. Ttt ;3 anum pressure is recalculated for each time step, to reflect the reduction due to the escape of plenum gas, and the reduction due to the increase in plenum volume due to the relative motion of the pin structure and the fuel. Initially, the calculated value of a may be smaller than the critical value for choked flow, calculated to be 0.13516 for this probles. When this situation does occur, the critical value is used for S. The extent to which the upper pin structure can move upward is not clear, but subassembly schematics from CRBR PSAR39 suggest that a large fraction of the pins can move a considerable distance. Given the fact that the wire wraps can unravel, and the fact that any motion would likely start from the subassembly center and progress outward, it would appear that significant upward motion of the pin structure could take place. If this motion is unimpeded,' the finite-difference calculations indicate that the plenum pressure will be released in 0.027s. It should be noted that the function multiplying P on the right-hand side of equation 1 varies slowly during the blowdown; s is small, while the parameter F is about 25. As a result, the pressure decay is about exponential, of the form P=P exp (-t/ ).

. ~ The time " constant" is defined from equation 1 as 3,,2 31/2 ' -1 k (/ 1/2 RT 3 ) F - An$ This blowdown time is the time required for the pressure to fall to 1/e of its initial value. Even though t is not constant, its value is useful in estimating the blowdown rate. The initial value of t for the present calculations, at 1200K, is about 0.25s. This value is based on a constant gap width of 0.0143 cm and a flow length of 35.56 cm. l l O y v.

. J Appendix E Modified Treatment of Partial Clad Blockages in the SAS3D Boiling Model Af ter the formation of partial blockages in the coolant channel, due to the motion of molten clad, the coolant boiling model sometimes tends to calculate negative pressures in the middle of a blockage. These negative pressures, in addition to being unphysical, cause the program to become numerically unstable and terminate. The problem is due to an acceleration term in the momentum equation for the vapor. For the current CRSR and R8 cases, this acceleration term was modified to 'give a solution that is physically more meaningful and numerically more stable. In SAS3D the momentum equation for sodium vapor contains an acceleration term of the form E (G d )' o If the flow area changes due to the motion of molten clad, then the code uses a term of the form X E (AG ) 1d o where G is the mass flux in the vapor, o is the vapor density, A is the coolant flow area, and 2 is the axial position. In case of a large vapor flow through a local partial blockage involving a large area change, this term contributes little or nothing to the over-all pressure drop across the blockage, since the pressure loss on one side of the blockage is largely

. cancelled by pressure recovery on the other side; but in the middle of the blockage this term tends to drive the pressure negative. In such cases, a rapid drop in the pressure in the blockage leads the code to cut back the _ coolant time step to very small values in an attempt to obtain an accurate and s' table solution; but the tendency toward negative pressures often causes the code to go unstable and tenninate. Therefore, the code was modified so that the acceleration term is eliminated at any node interval where the coolant flow area is less than 56% of the nominal value, or at any time when the coolant time step is cut to less than 3x10-5 seconds. Pressure drop through the blockage is always accounted for by the friction tenn, which can get large if the flow area is small. REFERENCES 'l !. Letter HQ:S:82:110 John Longenecker,.to Paul Check, Amend' No* 72 to the PSAR fof CRBRP, dated Oct 29, 1982 2. S. K. Rhow, et. al., "An Assessment of HCDA Energetics in the CRBRP Heterogeneous Reactor Core," CRBRP-GEFR-00523 (December,1981). 3. T. E. Kraft, "An Evaluation of Recent Transient Fuel Behavior Models Based on Selected Experimental Results " ANL/ RAS 80-29 (November, 1980). 4. J. M. Kramer, et. al., "An Analysis of Recent Fuel Disruption Experiments," Inti. Top Mtg. on LMFBR Safety, Lyon, France (July, 1982). 5. E. E. Gruber and E. H. Randkley, " Comparison of Fission Gas Effects in a Transient Overpower Test (HUT 5-7A) to FRAS3 Code Predictions," Intl. Top. Mtg. on Fast Reactor Safety Technology, Seattle, Washington ( August,1979). 6. R. Simms, et. al., " TREAT Experimental Data Base Regarding Fuel Dispersals in LMFBR Loss-of-Flow Accidents," Proc. of Top. Mtg. on the Reactor Safety Aspects of Fuel Behavior, Sun Valley, Idaho (August, 1981). 7. R. Simms, et. al., " Loss-of-Flow TREAT Tests L6 and L7 with Irradiated LFEBR-Type Fuel", Nucl. Tech., R,331(March,1981). 8. R. Simms, "An Evaluation of Fuel Motion in Recent TREAT Experiments with L'iquid-Metal Fast Breeder Reactor Fuel," Nucl. Tech., 50,, 257 (October, 1980). .~ 9. W.'R. Bohl, "SLlNPY: The SAS3A Fuel Motion Model for Loss-of-Flow," ANL/ RAS 74-18 (August, 1974). 10. B. W. Spencer, et.al., " Final Report on TREAT Tests R4 and R5; Seven-Pin, Loss-of Flow Tests with Full-Length Unirradiated FFTF-Type Fuel Pins," ANL-79-106 (December,1979). 11. B. W. Spencer, et. al., " Interim Report on TREAT Test R8, a Seven-pin Loss-of Flow Test with Pressurized Pins", Al*L/ RAS 78-39 (September, 1978). 12. T. E. Kraf t, et. al., " Final Report on the SLSF In-Pile Experiment l P3A", ANL/ RAS 81-20 (June,1981). 13. T. E. Kraft, et. al., "SLSF In-Reactor Experiment P3 Interim Posttest Report", ANL/ RAS 78-53 (December, 1978). ' 14 C. W. Hunter, R. L. Fish, and J. J. Holmes, " Mechanical Properties of Unirradiated Fast Reactor Cladding During Simulated Overpower Transients,"Nucl. Tech.,E.(1975). I I

e l 15. C. W. Hunter, G. D. Johnson, and R. L. Fish, " Mechanical Properties During Simulated Overpower Transients of Fast Reactor Cladding Irradiated from 700-1000 F," HEDL-TME 75-28, Hanford Engineering Development Laboratory, Richland, WA (June,1975). 16. G. D. Johnson and C. W. Hunter, " Mechanical Behavior of Fas't Reactor Fuel Pin Cladding Subjected to Simulated Overpower Transt'ents," HEDL

THE 78-13 (June,1978).

17 '. Letter, M. L. hmilton, HEDL, to A. Biancheria, W-ARD, D. B. Atcheson, GE, and G. D. Johnson, " Transmittal of FCTT Data Base for Cladding Failure Criteria," March 9,1982. 18. D. R. Duncan, G. D. Johnson, and C. W. Hunter, " Effects of Multiple Transients'on Fast Reactor Fuel Pin Cladding Mechanical Frcperties," Trans.,Am. Nucl. Soc., 30_, p. 196, (Nov., 1978). 1 19. N. S. Cannon and C. W. Hunter, " Transient Undercooled Overpower (TUCOP) Simulation on Fast Reactor Cladding," HEDL TC-2260, (Sept.,1982). 20. H. U. Wider, et. al., "The PLUTO-2 Overpower Excursion Code and a Comparison with EPIC," Proc. Int'l Mtg. on Fast Reactor Safety, i Seattle, Washington, (August 1979). t l 21. H. U. Wider and L. A. Semenza, " Analysis of TREAT Transient Overpower Experiments Using the PLUTO Cedes," Proc. Specialists Workshop on , Predictive Analysis in LMFBR Safety, Los Alamos, New Mexico LA-7938-C (March, 1979). 22. D. L. Graff, '"Results of Testing and Validation Efforts on the Two-Fluid (TWOFLU) Hydrodynamics Module of TRANSIT-HYDRO," ANL/ RAS 82-32 l (October, 1982). 23. D. L. Graff and J. E. Cahalan, " Modified ICE Technique for Two-Phase j Systems with Strong Mass Sources," Trans. Am. Nuc. Soc., 39, 505 i (1981). ' 24. D. P. Neber, "The VENUS-III HMT Algorithm: A Non-linear Implicit Eulerian Hydrodynamics Algorithm for Two-fluid Models without S1.ip," ANL/ RAS 79-5 (April,1979). 1 25. T. J. Scale, et. al., "0PERA Fifteen-Pin Sodium Expulsion Test," Trans. Am. Nuc. Soc., 43 (1982). 26. W. R. Bohl and T. J. Heames, "CLAZAS: The SAS3A Clad Motion Model," ANL/ RAS 74-15 (August, 1974). 27. W. R. Bohl, " CLAP: A Cladding Action Program for LMFBR HCDA LOF Analysis," Trans. Am. Nucl. Soc., 23, p. 348 (1976).

'. e 28. M. Ishii, W. L. Chen, and M. A. Grolmes, " Multichannel Model for Relocation of Molten Fuel Cladding in Unprotected Loss-of-Flow Accidents in Liquid Metal Fast Breeoer Reactors," Nuc. Sci. Eng., 69 (1979). ) 29. W. L. Chen and M. Ishii, "A Para:netric Study of Multichannel Molten Cladding Motion Under Unprotected Loss-of-Flow Accident Conditions in LMFBR's," Nuclear Engineering and Design, R, No.3 (December,1980). 30. J. E. Cahalan, et. al., "A Preliminary User's Guide to Version 1.0 of the SAS3D LMFBR Accident Analysis Computer Code," Reactor Analysis and Safety Division Internal Report, Argonne National Laboratory (July, 1977). 31. T. G. Theofanous, M. DiMonte, and P. D. Patel, "Incoherency Effects in Clad Relocation Dynamics for LMFBR CDA Analyses," Nuc. Eng. Design, 36 (1976). 32. G. B. Wallis, One-Dimensional Two-Phase Flow, McGraw-Hill, New York (1969). 33. G. D. Johnson and C. W. Hunter, " Mechanical Properties of Transient-Tested Irradiated Fast Reactor Cladding," Trans. Am. Nucl. Soc., 30, p. 195, (Nov., 1978). 34. J. M. Kramer and R. J. DiMelfi, "An Analysis of the Rupture Behavior of Fast Reactor Fuel Cladding Subjected to Thermal Transients," J. Eng. Mat'1s. and Tech., 101., 293-298 (1979). 35. J,. M..Kramer and R. J. DiMelfi, "Modeling Deformation and Failure of Fist R'eactor Cladding During Simulated Accident Transients," Nucl. Eng. and Des., 63, 47-54 (1981). 36. R. J. Diin'elfi and J. M. Kramer, "Modeling the Transient Failure Behavior of Irradiated Fast Reactor Cladding Tubes Including Fuel-Adjacency Effects," Trans. of the 6th Int. Conf. on Structural Mechanics in Reactor Technology, Paris, France, August 17-21, 1981 paper C3/3. 37. F. E. Dunn, et. al., "The SAS2A LMFBR Accident-Analysis Computer. Code," ANL-8138 (October, 1974). 38. T. C. Chawla, G. M. Hauser, M. A. Grolmes, and H. K. Fauske, Nucl. Sci. j Eng. _58, 21-32 (1975). 39. Chapter: 4.2 of the' CRBRP Prelimina l Corporation, Docket No. 50-537,197b Safety Analysis Pe' port, Project Mana'gemec l l This enclosure contains the response to item 8 of enclosure 1. __.}}