ML20206C066

Draft MELPROG-PWR/MOD1 Analysis of a Tmlb' Accident Sequence

ML20206C066
Person / Time
Site:	Surry, 05000000
Issue date:	06/24/1986
From:	Dearing J, Henniger R, Joseph Kelly SANDIA NATIONAL LABORATORIES
To:
Shared Package
ML20204G644	List: ML20206C066 ML20206E142 ML20206F071 ML20206F086 ML20206F182 ML20206F313 ML20206G821 ML20206G928 ML20206H081 ML20206H113 ML20206H144 ML20206J252 ML20206J270 ML20206J368 ML20206J403 ML20206J600
References
RTR-NUREG-1150 SAND86-2175, SAND86-2175-DRFT, NUDOCS 8704100148
Download: ML20206C066 (69)

Text

, sAo str1u7t e

O ROUGE DRAFT sh/d MELPROG-PWR/ MOD 1 ANALYSIS OF A TMLB' ACCIDENT SEQUENCE J. E. Kelly, R. J. Henninger, J. F. Dearing i~ Sandia National Laboratories Albuquerque, NM 87185 l June 1986 EXECUTIVE

SUMMARY

The first complete, coupled, and mechanistic analysis of a reactor core meltdown sequence has been made with MELPROG-PWR/ MOD 1. The sequence analyzed was a station blackout accident sequence (TMLB') for the Surry plant. The MELPROG calculation was initiated at the point where the primary coolant saturated (estimated from a TRAC-PF1 calculation) and was run through the point that the reactor vessel failed. Between the beginning and the end, all important aspects of the meltdown sequence where calculated with MELPROG. While this calculation is the first one performed with the new version of MELPROG and must be viewed as preliminary at this point, the current analysis does demonstrate the advanced capabilities that this version of MELPROG possesses for core meltdown accident analyzes.

This version of MELPROG permits a full two-dimensional treatment of the in-vessel phenomena. As such, the important effects of in-vessel natural circulation can be accurately modeled. To assess the importance of natural circulation and other two-dimensional effects, the current calculation has been compared with one-dimensional MELPROG and MARCH calculations of the same accident scenario. This comparison has shown that natural-circulation reduces,the rate of core heat-up, but increases the rate of heat-up of upper plenum structures. This implies that a significant amount of the core energy is deposited in the plenum and primary piping. This increased 'g heating can inhibit fission product deposition and may lead to an early failure of the primary system. Hence, natural circulation alone can completely change the course of a meltdown sequence.

Since this calculation was preliminary and uncertainty existed in a number of key models, limited sensitivity studies where performed to assess the relative importance of various modeling assumptions. One of the key models assessed was the modeling of the initial fuel rod celting and relocation.

Variations in the modeling assumptions were found to strongly affect hydrogen production and the shbsequent course of the accident. The magnitude of hydrogen sosrce could be varied by a factor of 2 through variations in fuel ' rod modeling. This result implies that accurate and mechan'istic modeling is important for seiere accident sequence analysis.

8704100148 B70408 PDR NUREG PDR 1150 C . , ,

r l

l l L l

l

1.0 INTRODUCTION

MELPROG-PWR/ MOD 1 was used to perform an analysis of an unprotected station blackout (TMLB') accident scenario in a large pressurized water reactor (PWR). MELPROG has been developed to analyze in a mechanistic manner the wide range of phenomena which i occur in the vessel during a severe accident sequence. These accident sequences can be analyzed with MELPROG from accident initiation through vessel failure and melt ejection. This means that MELPROG treats a wide range of processes including coolant boiling and flashing, core heatup, Zr oxidation, fuel and control rod melting, candling, debris formation, thermal attack on in-i vessel structures, core slumping, melt / water interactions, vessel failure, and melt ejection. The approach used in MELPROG is to simultaneously treat all thermal, chemical and material transport processes in the vessel in as mechanistic a manner as possible.

This, then, allows the important coupling between phenomena to be accurately predicted. It should be added that while MELPROG is strictly concerned with in-vessel phenomena, the code has been designed to allow implicit linking with the TRAC-PF1 code. This feature allows the entire reactor coolant system to be modeled with the MELPROG/ TRAC linked code. Currently, there are two versions of MELPROG. The first, MELPROG-PWR/ MODO [1] , uses a 1-D hydrodynamic treatment of the vessel with some 2-D enhancements.

The second version, MELPROG-PWR/ MOD 1 uses a complete 2-D treatment of the vessel. Both of these versions are applicable to PWR analysis. The most distinguishing feature of MELPROG-PWR/ MOD 1 is that it allows full two-dimensional modeling of the reactor vessel so that important two-dimensional effects, such as natural circulation, can be accurately modeled.

This calculation represents the first complete and coupled accident sequence performed with this new version of MELPROG.

The sequence was run from the point where the primary system coolant saturates to the point where the reactor _ vessel failed.

The calculation treated the coolant voiding, core heat-up, clad-ding oxidation and melting, fuel relocation, core slump and fuel-coolant interactions, and the vessel heatup and failure. As such, this calculation demonstrates the significant advancement that MELPROG provides for severe accident analyses.

For this initial calculation, three objectives were established before the calculation was begun. The first was that the calculation should attempt to be as accurate as possible.

That is, it should attempt to be a "best-estimate" calculation.

In this regard, a detailed geometric model was used as were realistic initial and boundary conditions. However, since this was the first test with the complete code, it was realized that some physical phenomena would probably be neglected or improperly modeled. Hence, the final results would not necessarily be a "best-estimate," but more likely a "best-effort."

a.

.s c

A second and more important objective of this cciculation

- was to run a complete accident sequence which involved all mod-l ules in the code. The reason for giving priority to this goal i was that.at this stage of the code development it is more impor-i tant to test and debug.the. code as completely as possible rather I

than to " fine-tune" the results. The practical implication of this objective was that the calculation was completed even with-

) inaccurate physical results. It should be noted that the model-a ing deficiencies were duly noted and improvements are currently l being made.

i The disadvantage of this approach is that the predicted results do not necessarily represent a "best-estimate." However, the advantage of this approach is that a large fraction of the problems in the code were identified with a single calculation.

Hence, this approach is cost effective. The alternative would be to continually restart the code every time an error was found.

In view of the< fact that this was the first complete calculation, the approach taken here was considered acceptable.

j The third and most important goal of this calculation was to gain insights into the relative importance of the phenomena

! which affect the source term and the meltdown calculation. The l intent here was to assess how sensitive predicted quantities l (such as core temperature, mass flow, etc.,) were to particular i modeling assumptions. In particular, the importance of natural j circulation within the vessel was one of.the key items to be

< investigated. As will be shown later, this phenomenon has a j significant effect on the meltdown calculation.

4 The approach taken to gain these insights was to complete what we refer to as the base-case calculation. This base-case calculation was then compared to calculations made with the 1-D version of MELPROG[1,2] and with MARCH [3] . Limited auxiliary 2-D calculations were also made to investigate specific items of f interest. These auxiliary calculations were compared to the-base-case calculation to determine the sensitivity of particular modeling assumptions. In this way, preliminary insights into the important meltdown phenomena were attained.

The discussion of the MELPROG analysis-has been broken into

! a number of sections. First, the MELPROG model used in.this L calculation is discussed in Section 2. In this discussion, the  !

geometrical model and the initial and boundary conditions'for the  :

calculation are discussed. In Section 3, some of the major.

modeling uncertainties in the current calculation are addressed.

A description of the base-case calculation is provided in Sec--

. tion 4. In Section 5, the limited sensitivity studies and.com- ,

parative analyses are presented. :The important results of this i preliminary analysis are summarized in Section 6. 1 l

s

'c i 2.0 MELPROG MODEL To model a severe accident sequence in a realistic manner i requires a great deal of data and thorough analysis of the acci-dent event sequence. In the case of MELPROG, an accurate repre-sentation of the vessel and its internals is needed to properly i represent the flow paths and hydraulic resistances so that the correct flow rates and temperature distributions can be calcu- l lated. Many times, the data obtained from FSAR and other sources j is incomplete and, hence, assumptions and approximations must be l made. In the calculation discussed in this report, MELPROG was  !

run in stand-alone mode and, hence, initial and boundary condi-tions representative of the TMLB' sequence were required. If a )

coupled MELPROG/ TRAC calculation were made, then event sequence l characteristics such as relief value flow rates, pump character-istics, etc., would be required. Hence, the. development of a MELPROG model.for a given accident sequence in a specific plant is not a trivial exercise.

a l In this section, the MELPROG modeling of the TMLB' sequence is described. The sequence of-events for this sequence are reviewed first. Then, the geometric model of the vessel is described. Finally, the initial and boundary conditions used to simulate the TMLB' sequence are discussed.

2 2.1 Overview of TMLB' Sequence In the TMLB' scenario, the primary system heat rejection path through the steam generators is unavailable due to a com-plete loss of feedwater; also the emergency core cooling systems and the containment safety features are unavailable due to the-loss of all electric power. Decay heating following reactor shutdown results in complete boiloff of the water in the second-ary side of the steam geherators. After steam generator dryout, i the primary system pressure rises to the relief valve setpoint R and the primary coolant temperature rises to the saturation tem- i perature for that pressure. At this point in-time, over 6000 seconds after neutronic shutdown, the vessel and core are near the coolant saturation temperature and have very low thermal gradients. It is at this point that we chose to begin the MELPROG analysis.

In keeping with previous analyses of this accident [2,3] , the l l Surry Unit 1 power plant [4] was used as the reference PWK design. I case. By studying this particular. plant, we can directly compare to these other calculations. Figure 2.1 illustrates a typical Westinghouse PWR design which is similar to the Surry vessel and internals. The reactor coolant flows from the inlet downward between the vessel wall.and core barrel (downcomer) into a plenum between the vessel bottom and the bottom support. It'then flows upward through the various support plates and core to a zone at the elevation of the outlet nozzle where it exits. The plenum 4

h

s-i between the upper support plate and the vessel top also contains water and this will drain during core boiling.

Since MELPROG is intended to examine the core melting, relocation phenomena,and vessel failure a detailed description of the reactor core and vessel is needed. Because this calculation focused on examining the core meltdown phenomena, MELPROG was run in a stand-alone mode. This lack of direct coupling with the reactor coolant system allows for a faster calculation at the expense of some loss of accuracy. In the low flow, constant pressure TMLB' calculation, this effect should be small.

In the course of the transient, one of the first major events is the complete voiding of liquid above the upper core plate. At this point, both the upper plenum and exit plenum have emptied of water and the primary loop is no longer directly cou-pled to the reactor vessel. As voiding progresses, the liquid in the core and downcomer is boiled away until the core is complete-ly uncovered. Up to this point, an overall reactor coolant sys-tem model (e.g., TRAC-PFl[5] is desirable and this is the reason for developing the MELPROG link with TRAC-PF1. However, the phenomena (e.g., core mciting and relocation), which this calcu-lation primarily addresses, occur after core uncovering.

As the sequence proceeds,the phenomena likely to be encoun-tered are easy to identify. First, the water inventory in the vessel continues to be boiled away. As the pressure in the vessel reaches the relief valve set-point, the valve opens and relieves the system of mass to reduce the pressure. The pressure then drops to the lower set-point and the valve closes. This cycling would continue throughout the sequence.

Eventually, the fuel rods begin to overheat and oxidize.

The Zr oxidation releases large quantities of energy and this leads to rapid heating of the rods. Clad melting follows and ultimately the fuel rods fail and relocation begins. Most of the core becomes involved in the melting and relocation of the fuel rods. Blockages form in the lower sections of the core which holds up the debris and allows it to heat and melt. These block-ages will fail at some point and this, then, allows the core debris to slump into the lower plenum.

The hot core debris will contact the water remaining in the lower plenum leading to a melt / water interaction. This event J

will be violent if not explosive. Suffice to say that the inter-action will at least generate large quantities of steam and a pressure spike.

After vaporizing the water in the lower plenum, the core debris will begin to reheat and melt. The debris will heat the vessel, weakening it and eventually leading to its failure.

Failure may occur due to melt through, rupture, or penetration failure. Once the vessel fails, the_ core debris will exit the l

o e

vessel into the containment. This event marks the end of acci-dent sequence from the MELPROG point of. view.

. The above list is nothing more than a sequence of antici-

, pated events. In an actual calculation, the sequence may deviate or include other effects. For example, the primary system may fail before the vessel head fails; there is speculation that the hot leg may fail early in the accident. Events such as this will be automatically predicted by MELPROG, because it does not assume a sequence of events. Rather, it treats the thermal, chemical and mass transport processes which control the sequence of events. The coupling between phenomena prevents one from defining a generic TMLB' sequence (or any sequence for that matter). It is for exactly this reason that'the MELPROG was developed: to investigate accident sequences in a mechanistic and fully coupled manner.

2.2 Vessel Model The model used for this analysis is shown in Figure 2.2.

Five radial rings and thirteen axial cells are used in a cylin-l drical grid to represent the reactor vessel. (A total of 65 nodes is used.) The calculation is bounded on the lower surface by the lower head, on the upper surface by the upper head, and on the outer surface by the vessel wall. The first three radial rings are used to model the core region, the fourth ring repre-4 sents the core bypass region, and the fifth ring represents the downcomer. The three radial rings in the core region subdivide the fuel assemblies equally by volume. This equal volume separa-tion is assumed to be adequate to describe the radial heat trans-fer and failure incoherence. The thirteen axial cells include O " six in the fuel rod region to enable computation of the axial.

, gradients necessary to follow melt progression. These'six cells include the fuel rods, the control rods, and poison rods. All of the major vessel structures are modeled, as shown in Figure 2.2. i

.Dne cell is located near each major plate to give an accurate structural thermal calculation. In the axial zones below the core, additional structures associated with instrumentation and core support have been added as heat sinks with the appropriate volume and surface area. In developing the lower plenum noding, the volume of the plenum was made consistent with the actual dimensions.

An important, but not necessarily obvious part of the modeling, is that a realistic simulation of the upper vessel requires more than a two-dimensional model, i.e., there are three-dimensional features in the vessel. Flow paths exist between the downcomer -and the upper head (via the cooling spray nozzles) and between the upper plenum and the upper head (via the centrol rod drive covers). These flow paths need to be repro-sented, in addition to the flow paths simulated by the two-dimensional mesh. The FLUIDS module has been modified to allow the user to specify a limited number of additional interfaces l

l

. _ . - . ~ .. - - = -- - .- - - - . _ _ - - __

s

'k l

between cells. Three of these " embedded" interfaces:are used in this .model tx> represent the flow paths inside the control rod

-drive covers, and one is used to represent the head spray cooling 4

nozzles (Figure 2.2). .The upper plenum flow areas, hydraulic diameters, and structure volume fractions,.which are normally calculated by the STRUCTURES module, were set internally in_the

~ ~

j- code to properly treat the special (heat-transfer only) struc-tures that were used to model the control rod guide tube struc-tures.

All the geometric data for the core, barrel, and vessel are j readily available from either the Surry FSAR[4] or the BMI-2104 [3] documents . The geometric data for the core, support, and diffuser plates as well as the structural support columns can be inferred from other Westinghouse plants.

Steady-state values of pressure drops and flows were used i to calibrate the model. Specifically, 1% of the total vessel flow cools the upper head via the spray cooling nozzles. . The amount of flow penetrating the core barrel and flowing down the core bypass was adjusted to 0.52%. Flow resistances were ad-justed to achieve the required pressure drops between the inlet nozzles and diffuser plate, and between the diffuser plate and outlet nozzles. Axial lengths of cells representing the upper i and lower heads were adjusted to achieve the proper fluid volume of these hemispherical regions. Finally, FSAR. values of total l vessel flow and power produced the required temperature-rise and 1 core average velocity.

I 2.3 Initial and Transient Boundary Conditions j A TRAC-PFl[5] calculation [6] for Zion-1 PWR was used to

provide initial conditions for-MELPROG 6500 s into the transient, l when boiling began in the core. The cold-leg flow calculated by TRAC was equal to 3.5% of nominal full power flow at 8500 s.

This flow is due to a natural circulation loop within the primary system, and was calculated by TRAC to decrease to zero beginning

about 300 s after boiling began in the core. For simplicity, the

flow through the cold leg was set to zero at the beginning of the MELPROG calculation. This condition was chosen to allow compari-son with the results of a previous calculation with similar

assumptions [2] . The pressure boundary condition representing the i hot legs was set to the PORV setroint of 16.3 MPA. .(Later in-

!- this report, limited sensitivity studies related to variations in the outlet pressure will be discussed.)

i Boundary conditions for the convective terms'in'the' event i of inflow at the hot legs were determined assuming a zero gra-dient (i'. e . , what comes in is identical to what borders the pressure boundary condition). These boundary conditions are necessary for this. type of transient (in which the system is heating up and expanding, causing outflow at the hot legs) only during boiling oscillations.

J i i 1

. - ~ . , , , . , , ,-,.-.a -.

i e

x

?

I 3.0 MODELING UNCERTAINTIES

[ Since the calculation presented in chis report was the first complete calculation performed with MELPROG-PWR/ MOD 1, it is f

! appropriate to discuss some of the important modeling uncertain-

~

i ties in the calculation. At this point in time, this version of

MELPROG is not a mature code. Rather, it.has just attained

operational status. Development work on the code continues in a number of key-areas. This means that the current calculation was-performed.with significant simplifications in certain models.

The impact of improved modeling on the overall calculation is difficult to assess at'this time, but is expected to be impor-tant. Hence,.the current calculation must be viewed with some skepticism.

1 The modeling-uncertainties in this calculation can be divi-ded into two groups. The_first group consists of all modeling

, not related to'the code itself. This group, which we refer to as i external modeling uncertainties, includes initial and boundary c.onditions, geometrical detail of the vessel, and other accident j scenario assumptions. The second group consists of all modeling .

within the code. This group, which we refer to as internal i modeling uncertainties, includes such items as the fuel rod model, fission product model, etc. The major uncertainties in each of these groups is discussed below.

3.1 External Modeling Uncertainties 4

In performing this calculation, a number of assumptions have been made concerning the accident scenario. 'These-assump-tions deal primarily with the initial and boundary conditions for 1

! MEL'ROG which are needed to run a stand-alone calculation. While i

these assumptions are reasonable, they do introduce some uncer-tainty in the calculation and need to be identified (if not j addressed).

The first area of uncertainty deals with the initial condi-tions for the MELPROG calculation. As indicated in Section 2.3, a TRAC TMLB' calculation for the Zion reactor was used to esti-mate the timing of primary system saturation for the Surry reac--

tor case. The primary source of uncertainty is the timing of this event, but this is not expected tc significantly-alter the subsequent calculation. To reduce this uncertainty, a coupled TRAC /MELPROG calculation should be performed. However, the 2-D~

TRAC-LINK is under development and is not yet operational.

The next area of uncertainty involves the modeling of the .

! PORV. In-the base-case-calculation,-the PORV is not actually '

modeled at all. Instead,.a constant pressure at the hot leg is assumed. This assumption neglects the cycling of the PORV.

, However, an auxiliary calculation was run to assess the effect of pressure cycling at the hot leg and this is discussed in Sec-l l tion 5.3.

_ __ . . . .__ __ .~. . _ _ _ _ . _._.____.__.___t _.-. _ .__,_

s The pressure cycling effect is only one aspect of the PORV modeling. In the modeling, it is also assumed that (1) the PORV continues to operate throughout the accident, and (2) the pres-sure at the hot leg is the same as that at the PORV. The second assumption here is reasonable and of little consequence, but the first deserves further attention. The PORV may likely fail after a number of cycles. If it fails, then the primary system will begin to depressurize like a small-break LOCA. This could sig-nificantly change the meltdown progression.

Also, when the core slumps into the lower plenum, the rapid vaporization can lead to significant pressures which may fail the PORV-(or associated piping). Hence, future calculations should assess the effect of PORV failure.

The next area of uncertainty is related.to the fact that MELPROG was run in stand-alone mode. This means that the vessel was decoupled Trom the remainder of the primary system. For this particular accident sequence, this assumption is not unreason-able. However, certain effects could not be accounted for. For example, three-dimensional effects related.to the positions of the hot leg nozzles could not.be modeled. Furthermore, any influence which the steam generators may have on the heat par-titioning could not be modeled. Currently, 60MVIX calcula-tions [7] are' being performed to address this latter issue. The results of these should be factored into the MELPROG results.

The final area of uncertainty involves the actual modeling of the geometrical detail in the vessel. Uncertainty exists due to insufficient data for the dimensions of.all structural compo-nents and due to simplifications in the modeling of complex structures. The modeling of flow areas, volumes, and the thermal capacity of the vessel structures is probably adequate. However, the frictional loss coefficients for these components are largely uncertain. These loss coefficients will certainly affect natural circulation flow patterns. The degree to which the flow patterns are affected is unknown and needs to be assessed before firm-conclusions can be drawn concerning natural circulation. This same argument can be made for both intact and degraded core states.

3.2 Internal Modeling Uncertainties As indicated above, MELPROG-PWR/ MODI has just become opera-tional and development continues in a number of key areas. In the current calculation, the modeling of certain phenomena was simplified or neglected, reflecting the current state of the code. In this section, the major areas of uncertainty due to the phenomenological modeling are identified.

The first area of uncertainty involves the modeling of fuel rod failure and relocation. The current modeling does not. treat the candling phenomenon in a mechanistic manner. In determining l

_ _ _ _ _ _ _ _ _ . __ _ ~

l l' s l 1

L \

l

+

l b the failure of a fuel rod section, the entire section is treated i as a single entity. (Note.that each fuel rod modeled is divided t

into a number of axial sections.) This means that when a fuel rod section is predicted to fail, then the entire rod-(both fuel ,

and cladding masses) is transferred to the corium field. .If the

rod is shattered by quenching or if it is completely molten, then ,

this approach is not unreasonable. However, the more-likely situation in an accident sequence is that the cladding will: melt

first and relocate leaving behind bare pellet stacks. This situ-ation (referred.to as. candling) is not properly treated in the current-fuel' rod model. However, the new fuel rod model, CORE,

.which is under development will treat this situation mechanistic-  ;

I ally.

In the current'model, the' candling is approximated by allowing the fuel rod section to fail when its average; tempera- [

i ture exceeds a user input value. . Typically, this.value would be set between.th.e metallic Zr melting point (~2200 K) and the. Zr02

melting point (~3000 K). If set to the lower value, it allows relocation to occur at the earliest time and' lowest temperature.

4 If set to the higher temperature, it would allow relocation at later time and higher temperature. The timing and temperature at 4 relocation will strongly affect fission product release, hydrogen production, and the degree of relocation, and vessel failure timing. As such, it is an extremely important phenomenon. As 4 will be discussed, the base case allowed relocation at 2200 K, l i.e., the lower limit. To assess the importance of this phenome-

! non, an auxiliary calculation was run in which the failure tem-1 perature was set to 2500 K. The results of this calculation are l discussed in Section 5.2.

The second area of uncertainty.is the effect and behavior.

j of fission products. The current calculation did not treat fis-sion product behavior. Currently, the fission product module, VICTORIA, is being added to MELPROG,'but.is not yet operational.

~

Consequently, the effect of fission product deposition.and heat-e ing could not be studied. It is speculated that this heating, if 4 it occurs in the upper plenum, could alter natural circulation ,

flows. l i .

I L The third area of uncertainty is in the'modeling-of the l core slump. In particular, the fuel-coolant interactions are treated in a rather simple manner. Only1the heat transfer from i the fuel to coolant is treated and no, fragmentation is modeled.

The corium is assumed to have a constant, user specified surface ~

to volume ratio - (ef f ective diameter) . in the calculation. Cur-rently, the fuel-coolant interaction module, IFCI, is under development to provide MELPROG with a more mechanistic treatment.

The fuel-coolant interaction model in the current code is strongly dependent on the value-chosen for'the effective corium diameter. This parameter affects both the steam and hydrogen generation rates and the degree of core quenching. Further

! 1-

+

--, . _ . . - . . . ,.m.. , ,

1 6

assessment of the uncertainties in this modeling are currently in i progress. l The fourth area of uncertainty is related to the inter-facial exchange models used in the FLUIDS module. The uncer- i tainty lies in the choice of flow regime maps for the vapor to liquid exchange models. These maps and their associated correla-tions for interfacial mass, momentum, and energy transfer have a significant impact on the predicted thermal-hydraulic state in the vessel. There were two problems discovered during the cur-rent calculation. The first was related to the actual correla-tions used in the various flow regimes. During the testing of the 1-D version of MELPROG, a number of improvements were made to TRAC-PF1 models which had originally been used in the code.

These improvements primarily affect the heat transfer in non-equilibrium regimes. Due'to the fact that the testing of the two MELPROG versions (i.e., MODO and MOD 1) was occurring simulta-neously, not ill of these improvements were made to the new code (MOD 1). Hence, some of the correlations in the 2-D code were not properly updated.

The second and more important problem was that there was no horizontal flow regime map when this calculation was initiated.

The original flow regime map was strictly for vertical flow and was taken directly from TRAC. In most cases, this map is suffi-cient. However, when trying to treat the boiloff of the water in the lower plenum, the map would predict excessive heat transfer rates. This is because it would predict a bubbly flow condition while the actual state should be closer to a stratified flow state (i.e., vapor over a liquid pool). This deficiency was recognized early in the calculation, and an improved model was incorporated into the code. However, this new model has not been fully assessed, but it is very important for calculating the steaming rate in the vessel. Hence, the full impact of this modeling needs further assessment.

The final two areas of uncertainty reflect the lack of specific models in the code. The first deficiency is that the i oxidation of steel is currently not modeled. The second is that there is no instrumentation tube failure mechanism for the vessel head. Development work is currently addressing both of these modeling deficiencies.

In summary, there is sufficient unresolved uncertainty in the current modeling that firm conclusions drawn from the calcu-lation should be avoided.

l l

l

4 4.0 BASE-CASE RESULTS i

In performing this analysis, we have defined a base-case L calculation. This calculation is simply the complete calculation

! that has been made. It is not necessarily the most accurate-4 calculation possible. Some of the reasons for this have been given in Section 3. Nevertheless, it is this base-case calcula-j tion which will serve as a reference point in the comparative j analysis discussed,in Section 5.

l The sequence of important events for this base-case calcu-lation is given in Table 4.1. -This list does not give great-

detail'concerning the various events, but serves to place the events in the proper sequence. As the calculation is described,-

further detail will be given.

i l In describing this calculation, we have divided the acci-dent sequence -into five sections. .These sections chronologically l cover the entire sequence (with some overlap) and lump together l related phenomena and similar. events. The.first section is~the

boiloff and core heatup to the start of oxidation phase. This

! section discusses the period from the beginning of the calcula- ,

! tion until the maximum cladding temperature exceeds 1273 K.

j Relative to Table 4.1~, this is from 6500 s to 9300 s.

The second section is the cladding oxidation and fuel rod failure phase. This section discusses the period during which the Zr cladding is oxidizing and generating heat. The core rapidly heats during this period leading to fuel and control rod

failures. Relative to Table 4.1,.this phase includes the period i from 9300 s to 10400 s.

i The third section is the debris region formation and behav-i ior phase. This section discusses the initial formation of debris regions, their heating and eventual relocation. Relative l

to Table 4.1, this phase includes the period from.10180 s to 14877 s.

! The fourth section is the core slump-phase. This section dis-cusses the slumping of the core debris _into the lower plenum. '

] The core slump is a very fast event, but is very significant. As noted in Table 4.1, this event occurs at 14877 s.

l The fifth and final section is the vessel heatup and fail- l 4

ure section. This section discusses the core debris heatup in -i i the lower plenum and the eventual failure of_the vessel head. I l Particular attention is given to the statelof the core debris at the time of vessel failure. Relative to_ Table 4.1, this section covers the period from 14877 s to 15928 s.

- . 4

---

.---._w-, , - a v- y w , ., n_ ,-, - , _ , _ , , , _ . , , , , , _ , _ _ , _ , ,

I s 1 i '4 .~ 1 Boilof f and Core Heatup to Oxidation

.The " front end" of TMLB' transients for other Westinghouse plants has been studied in detai1[6] . The front end of a tran-sient is defined as the. period between transient initiation and 4 the onset of hydrogen generation where the.use of a degraded core

analysis tool would be required. The sequence of events during f the early phase of the TMLB' transient are-not complex; following i; the loss of feedwater, the. steam generators boil dry. With the loss of secondary. heat sink, core. decay energy can no longer be transferred from the primary to the secondary. The temperature f of the primary coolant begins to increase resulting.in an expan-

, sion of the liquid volume. This in turn causes the primary pres-sure to increase; the pressure setpoint of the power-operated

relief valve (PORV) is exceeded and the loss of. primary system.

inventory begins. Coolant heating continues until the saturation temperature is reached at the top of the core and boiling begins, j Temperatures in the primary system at the onset.of boiling are j uniform (a few degrees K across the core) because of. flow and j mixing that is induced by natural circulation flow around the j primary loops. For this analysis, we have assumed that the tim-3 ing of events to the onset of boiling could be taken directly from a scaled TMLB' calculation for the Zion-plant. This assump-tion needs to be checked by runn.ng a linked TRAC /MELPROG calcu-lation from the beginning of the transient.

i ~

The sequence of events for the_" front end" of the accident- l is given in Table 4.2. Included in the table are the results of a TRAC-PF1 calculation for this phase of the accident. The l TRAC-PF1 and MELPROG (MOD 1) models are geometrically similar and the initial conditions are approximately the same. The models

for both codes were set up so.that-boiling would begin at'about l- the same time (6500 s). For both calculations, the flow into the

vessel was terminated when boiling began. Figure 4.1 gives the

core average liquid volume fraction for the two calculations. In i TRAC the liquid fraction drops to 0.84 very

quickly, then that t level is maintained until approximately 7080 s when the_ level

] once again drops. The level-is maintained because of an equilib-

! rium that is established between vapor formation in the core and draining of water into the core. Once the core is " uncovered" at l-7080 s, there is no more draining into the core and the level drops as the water in the core is boiled away. A similar plateau-at a lower liquid fraction (~O.78) is obtained in the MELPROG calculation. Core uncovery in the MELPROG calculation occurs at l I 7070 s, 10 s earlier than TRAC.

! The core voiding rate to a liquid fraction of 0.2 was very, similar. The boiloff of the last 20% of-the water in the core was slower in the MELPROG calculation leading to a 250 s delay in I

emptying the core as compared to the' TRAC result. A possible i explanation of.this difference is. that the 150M] and MELPROG power

shapes are not identical. TRAC uses axial cell-edged values whereas MELPROG uses axial cell-centered values. This could i

o ,

b change the voiding rate near the top and the bottom of the core where the axial power distribution has a strong gradient. When the calculations are re-run, we will attempt to minimize such differences to determine their effect upon the results. At approximately 7200 s in both calculations, the maximum cladding temperature exceeds the saturation temperature and increases at a rate of 0.3 K/s. Figure 4.2 compares the maximum cladding i

temperature for the two calculations up to the initiation of i cladding oxidation (1273 K). It can be seen in the figure that the results agree to within 40 K up to a temperature of 1050 K.

At temperatures above 1050 K, radiation from the rods, which is not modeled in TRAC, results in slower heating in MELPROG as is indicated'in the figure. We find that the agreement between .

MELPROG and TRAC results is quite good and we are confident that

! remaining differences will be resolved as the models are refined and the calculations are re-run.

Figures'4.3-4.5 graphically show the state of the vessel during this phase of the accident. In interpreting these fig-ures, the top number in each cell represents the temperature (K) of the fluid with the largest volume fraction, while the bottom number represents the ensdece temperature of the structure or fuel pins (in cells that contain both fuel pins and structure, the pin surface temperature is shown). Dotted-line density is  ;

proportional to liquid water volume fraction, while.the angle of the dotted lines from the vertical is proportional to hydrogen partial pressure. The fuel pin volume fraction is represented by the vertical solid lines, while impenetrable (in the two-dimensional sense walls are represented by closely spaced paral-

, lel lines). The velocity vectors show a four-point average of the axial and radial interface velocities for the fluid with the largest volume fraction in that cell. Additional embedded flow paths that represent the control rod drive covers and the upper-

head spray nozzles are not shown for clarity. Figure 4.3 shows

the state of the vessel when boiling has just begun at 6500 s.

At this time in the transient, there is a flow of approximately 450 kg/s through the vessel. The temperature difference across i

the core is seen to be 4 K axially.and uniform radially. The

! structure and rod surface temperatures are within 1-2 K of the liquid temperature. Figure 4.4 shows the vessel at 7500 s when

the core is approximately 60% steam filled. By this time, a multi-dimensional circulation pattern has developed in the upper 4

plenum that reaches down into the core. Steam exits the center of the core at 688 K, flows upward, is cooled in the upper plenum  ;

to 665 K, and re-enters the core at the outer radial edge. There l it is heated and it flows back to the center of the core. Note i that the flow through the center of the upper plenum is larger i than the flow out of the vessel and that the steam exiting the vessel is 16 K cooler than the steam exiting the core. i Figure 4.5 shows the vessel at 8522 s, about 170 s after the core has filled with steam. The circulation that developed earlier persists, but now extends one axial cell further downward into-the core. An 86 K radial temperature gradient has developed

4 I

l i

i. !. 838 i. i. 892 i. i. B43 . .

i: !:776 i: i :824 i: ! :7.78  : :

i.: :1.: 785i i i i 1676.

! 630 . i. i i 701

:q:

:  : >  :  :  : :  : 1! :: :

: q ::

r -- _

944 i. i. !. 682

i. !. 373 ,

i. !. 959 i. i. 1. 954 i!..... 11 752

!. !. !. 6. 50 i... i. 5 i.. ..!. 9.17 /:. 879

h:

: : : 4 :....

1 - e

.a, . . . ..... ..

i 1942 i i i i 855 ii 380~

i : 940 i i 1938 )

i.....i. 3.31 i.. i.909 i... i. 8.73

i. i.s10

i. i. i. ?. 7.50

i. '. 936  :  : ocu  ;

i!

92 .i . 943 i i. 1.934 r761.

!. i. ,,.: 662

!. !. 21 i. i. 902 i. !. . 865 .

: p )

l 7 l' l l lli % 'll l' f i li :

l- : :: (. 1 ll l'  : :

l 1 l ll:  : 1

. . .._m_. ..  ;

i.: :i. w i: i.: g:822 i: 22 i

. 1.9.16 .

6  :.  :. E c, . . . . ... . 1

!. . 2 1.546 i. 1673 .

ii:619

..g

7

: 6  :. -: 1  :

.:  : : : :  : . : : : i i 77 l!963 i 48 i i673 i i i 618 l

i. 2  :. i. i: i.: : .:

. i.: i.: 's: : !:

. 37 i 1669 i : : 617 i 818  :

! 794 i: 609 i: 530 I: 46: :

i: i867 : : :

i: !'i61d: : : :

! 722 i 708 i 00 i 1 657 i ! ':. 617 i: 750  !?17 i:  : :

t i: 1835: : :

i l:i :613 i:.i:ii620 ii::::iS20  :::::ii620 i 50  :: ::W. Eil7 M  :  :  :. 613

622

..... . ......::::::4. 522.  ::::::

...... 2.1. . . . -

!!!!: 520  !!!!!$20  !!!!!)20  !!!!!!!619 iii!!!!619

.iiiiii.85 i i i i. l. i.1.614,

!!!!!5.1.9. . .

. . . . 1. .. 7..::::::::::

. !!!!!5.1.9... . iiiii.E.1.3. .. .

.:::::.:: 1.8. .

.::::::617 ::: ::.::617 . . . . . . . . . 612

. . . . . . 617

t.i.15.::  ::::::::iS1

609:

W::

:::4:::: : :::::::::615:::::::4:::::

61.5.:  :::::::  :-  :..

........::::::::::g .::1.:  :::::::

:::.:::+: .::::::::

50 CM/S .

Figure 4.5: Flow and Tet:1perature Distributions at 8522s l

(

i l

o O l l

I across the top of the core as a result of heat transfer to cooler structures at the outer radial boundary. Note that the I circulation has moved super-heated steam to the bottom of the core where it can contact the water below the core. The steam flowing from the vessel is 167 K cooler than that exiting the .

I core and the flow rate remains low compared to that in the upper plenum. At 9280 s, the rods in radial ring 1 at the top of the core (in level 7) begin to oxidize.

4.2 Cladding Oxidation and Fuel Rod Failure When the cladding temperatures exceed 1273 K, the cladding oxidation calculation begins. This process is highly exothermic and accelerates the fuel-rod heating process. As oxidation pro-ceeds, cladding becomes embrittled. Embrittled cladding will ,

shatter if the cooling rate by water or steam re-entering the 1 core region exceeds 100 K/s. The rapidly increasing fluid and structural temperatures in the core region will cause failures; the first to fail will be portions of the control rods whose '

stainless steel cladding will melt at 1700 K. The liquefied control materials drain from the failed rods into the intact core and proceed downward to a level where the temperatures are low enough to cause freezing. Throughout this time, the cladding temperatures are increasing as core decay energy and energy from cladding oxidation are deposited in the rods, fluids and strue-tures. When cladding temperatures reach 1850 K , the oxidation kinetics change and the cladding oxidation rate increases marked-i ly. Fuel-rod cladding begins to melt when cladding temperatures exceed 2180 K. The fuel rods at a given location are assumed to fail when the cladding is completely molten and above 2200 K.

The current model assumes that failed fuel rods are frag-mented and the debris which is formed is of a particulate nature.

Recent evidence indicates that the fuel rods may not simply frag-ment at such a low temperature and that more refined models such as the MELPROG CORE module are needed. The debris formed by fuel-rod failure will move downward and freeze or lodge in the lower sections of the core. The dynamics of the debris regions R

- are discussed in Section 4.3. The exothermic metal-water reac-

. tion becomes a significant heat source when it begins in the ring 1 rods at approximately 9300 s in levels 6-9. The heating rate l in these levels changes from 0.3 K/s togo.8 K/s as is indicated in Figure 4.6. Figures 4.6-4.8 give the,; cladding surface temperatures for the upper five levels for each of the three I radial rings in the core region. Ring 2 cladding begins to oxidize approximately 100 s later in levels 6-8. Ring 3 cladding oxidation begins approximately 300 s after ring 1 at 9600 s.

I A closer examination of the figures reveals that for this I part of the accident that the temperature increases upwards through the core in ring 1 (the axial temperature gradient is positive), whereas in ring 3 it increases downward through the I core (the axial temperature gradient is negative). Ring 2 l

=.. - - - . .- . - - - . - . . - . - . . .- -

L l 1

l increases upwards except for the uppermost level (level 9). The temperature distribution can be explained by the flow patterns present in the vessel. The flow and temperature distribution at 9525 s is given in Figure 4.9 Flow is seen to be upward in ring 1 and most of ring 2 (except for level 9) and downward in ring 3.

s A pattern similar to that in Figure 4.9 persists until 9970 s a when the gas in the hottest node of the core reaches the stain-less steel melting. point (1700 K). At_1700.K, it is assumed that 1 the stainless steel control rod cladding fails releasing the i molten silver-indium-cadmium absorber alloy that it contains. In

! the next_175 s, all of the control-rods in the top four levels of j rings 1 and 2 fail. The absorber material, whose freezing point i is 1070 K, flows downward through the core. The minimum rod temperature in.the core when absonby'y aftgrial begins to. move is

! approximately 1200 K in level 4 #at tke Bottom of the core i (Figure 4.10 gives the level 4 rod temperatures for the three

{ radial rings in the core). Therefore, the absorber material does not freeze until it contacts water in the lower plenum. Heat I transfer from the absorber material to the water causes steaming

!- in the lower plenum which, in turn, cools level 4 rods in all l three rings and level 5 rods in rings 2 and 3 (see Figures 4.7,

_ 4.8 and 4.10). Continued heating results in the hottest rods .

j reaching 1850 K at 10115 s at which point a change in the Zr02 lattice structure causes an increase in the oxidation rate. The 1 increased oxidation is manifested in a change in the rod heating

- rate from 0.8-K/s to 2-5 K/s which can be seen in the rod temper-ature plots (Figures 4.6-4.8). With increased heating the rods '

i rapidly reach the cladding melting point (2180 K) and then the

failure temperature (2200 K). In ring 1 at level 5 (the hottest i location in the core) these events occur at 10156 s and 10181 s, t i respectively. Rod failure is indicated by the point at which the j line ends in the rod temperature plots.

! The next event occurs at the top of the core in ring 3

where the control rods begin to fail at 10216 s. In the time i between 10241 s and 10303 K, cladding melts and the fuel rods I fail in the top four levels of the core in rings 1 and 2. Lag-

! ging by approximately 20 s, the same-levels in ring 3 fail in the time frame from 10319 s to 10377 s. Fuel rod material from all

of these failures moves downward and freezes in leve1~5, forming .

l, a debris region beginning at that level. Contact with the debris in conjunction with additional oxidation heating fails all of-the l j rods in level 5 in rings 1 and 2 by 10403 s and by.10808_s in l l ring 3. This leaves only the level 4 rods intact. The R 4 conditions and flow patterns in the reactor vessel at_11000 s are depicted in Figure 4.11. This figure shows that the convection j cell between the upper plenum and the top of the core, present l

earlier in the transient _still persists. The flow within the l debris region (indicated by the symbol X), is extremely low.

j Additionally, it can be seen that " thin" structures (control rod housings) in the center two rings _above the core are either melting or close to their 1700 K melting temperature.

)  :

3

l l

I The "two-phase" water level in the core region and the downcomer are given in Figure 4.12. The level begins at the top ,

of the downcomer (7.4 m). The two-phase level is obtained by i searching, starting from the top, for the first cell that has a liquid volume fraction greater than 1%; the level is then determined within that cell by assuming that the two phase height is proportional to the liquid volume fraction. The bottom of the core is located at 2.46 m; the core is, therefore, essentially empty at 8350 s. The gas above the water in the downcomer is cooler and, therefore, heavier than the gas in the core region.

The resulting two-phase level in the downcomer is therefore lower than in the core to obtain a static balance. The level continues to decrease below the core due to the flow of superheated vapor over the water which heats and vaporizing the water. At 9970 s, relocated core material begins to vaporize additional water below the core. The two distinct events occurring at 11522 s and 12305 s are a result of control material from level 4 causing i rapid vaporization in the lower p1 gum. The final event occurs when the core slumps at 14877 s.*i Tnteraction of the core materials with the water boils away, entrains, and displaces the remaining lower plenum water, leaving the vessel empty. The increase in level in the downcomer is a result of a spray of water that is ejected from the lower plenum.

The total hydrogen mass produced by oxidation of cladding is given by Figure 4.13. Most of the hydrogen is produced when the rods are intact and above 1850 K (between 10115 s and 10403 s). It is during this time period that all of the rods except those at level 4 and those in ring 3, level 5 are oxidiz-ing rapidly and fail. The small addition of hydrogen at approxi-mately 10800 s is due to the oxidation of the level 5 rods in ring 3. Figure 4.14, which gives the total pressure and hydrogen i partial pressure at the top of the core, shows that' steam starved i conditions were not reached in the base-case calculation. The j abrupt changes in hydrogen partial pressure are caused by the '

quenching of control rod material and induced steaming in the lower plenum. The resulting steam sweeps the gas, including the ,

hydrogen, from the vessel. The steam flow from the lower plenum i also lowers the temperature in the bottom of the core region.

This can be seen in the level 4 rod temperature plot (Figure 4.10). The low rod failure temperature (22OO K) used in l the base case limited the amount of cladding oxidation; the- j average oxidation fraction at failure was only 0.4. An increased 1 I

failure temperature of 2500 K was used in a parzmetric calculation and this case will be discussed in Section 5.2.

l 4.3 Debris Region Formation and Behavior As a fuel rod section fails, the mass and energy associated with the section is transferred to the FLUIDS corium field for subsequent analysis. While the new 2-D-hydrodynamics model l allows for the treatment of four fields, only one corium field  ;

was used in this calculation. That is, the corium field was not j

)

1

,1

i

\

I i

divided int'o solid and liquid fields. This simplified' approach was taken to allow direct comparison with previous calculation j performed with the 1-D, 3-field model used in the MODO version of the code. Work-is currently underway to make the fourth field (i.e., liquid corium field) fully operational.

i The main disadvantage of using one corium field is that it

restricts the mechanistic treatment of relocating the corium i mass. Consequently, candling of the cladding, which involves i

both solid and liquid phases simultaneously, was not properly j treated in this calculation. However, the fuel and control rod  ;

. module, CORE, will contain the proper models to treat candling. J

! Meanwhile, the. compositional information of the corium field is j

! used to treat the average corium behavior.

The term corium, as used in the code and this report, denotes the combination of all fuel rod, control rod, and structural mat.erials. Currently, the masses of UO 2, Zr, ZrO 2, j U-Zr-0 mixtures, steel, and control rod material are accounted for. This detailed information allows the code to approximate the average state (either liquid or solid) of the corium in each cell. Based on this state, the governing equations for reloca-  :

tion can be determined.

i In order to understand the formation of debris regions, it j is useful to trace through the process in a particular ring.

l (The behavior in the other rings is usually similar.) The first step in the process is the failure of a particular rod section.

In the current calculation, a fuel rod section is assumed to fail when the metallic phase of the cladding is completely molten '

i (T > 2200 K). Practically, this means'that instead of performing i a one-dimensional heat transfer calculation through the rod, a lumped average heat transfer calculation for.the rod debris is

used. More importantly, the failure triggers the transfer of the j fuel and cladding masses to the FLUIDS corium field so that.relo-cation can begin. The relocation of the corium depends on the i state of the corium. That is, the composition and temperature of

the corium will determine its motion. If the corium is entirely

solid or entirely liquid, then the question of relocation

is trivial. In the former case, the corium flows as a particulate field unless it is within an intact core region, in which case it

is assumed to be a frozen crust. In the latter case, the corium

flows as a liquid field. The difficulty. in the current treat-' l I

ment arises when the corium is a mixture of liquid and solid

, phases. In the current model, it is assumed that'if the corium temperature is greater than 2200 K, then the corium is treated as i

.a liquid (or slurry). If the corium temperature is less than l 2200 K, then.the corium is assumed to be solid. It-is important to note, however, that the corium field will also include control ,

i rod material and steel at times. The logic in the code treats

-the difference in melting points,-but decides-based on the component with:the largest mass.

i 1

- . ~ _ . _ . . - - _, _ _ _.____ ___,--. _ _ .. . _ _ . . _ -. _ _-

l l

In the current case, the material is calculated to be molten and begins to flow downward. As it moves, it loses energy, cools and eventually freezes. Depending on the degree of superheat, some material may flow into the plenum. However, the usual case is that the material freezes on intact rods in the lower section of the core. As material accumulates on intact rods, a debris region will eventually form. In the current .

model, a debris region forms when a cell is at least half full of corium at the minimum packing fraction (37%). This means that the corium occupies a minimum of 18.5% of the free volume of a cell before a debris region forms.

Once it is determined'that a debris region exists, the DEBRIS module begins to perform a detailed calculation of the heatup and melting of the region. As time-progresses the intact rods upon which the debris region formed fail and their mass is added to that in the debris region. As other rod sections fail above the region, their mass is relocated downward until reaching the top of_the region. At this point, the mass is added to the region. This process continues and leads to increases in the size of the region.

The relocation of the entire debris region is controlled by a number of factors. Recall that the debris region'was initially formed by corium freezing and stopping on intact. rods. These rods eventually fail, but this does not necessarily mean that_the region will relocate. Rather, the model assumes that the lowest part of the debris region is crusted and that thic crust supports the debris region provided that the crust, in turn, is supported and is solid. Two means of crust support are currently

considered. The first is by a plate below the crust and the second is by intact rods below the crust. Hence, if the crust is solid and supported, then the region will remain stationary.

When the support for a region fails,-it is assumed that the debris can relocate. At this point, FLUIDS resumes control of

, the debris behavior calculation. This means that a detailed i

debris calculation is not performed until a debris region reforms. Hence, if~the debris region is supported on the lower plate and the plate fails, then the debris will relocate into the lower plenum. The debris will usually encounter water in the plenum and the resulting quench will generate copious quantities I i of steam and hydrogen. Eventually, a debris region will form on l the vessel bottom and begin to reheat.

i In this calculation, the first fuel rod section fails at l 10181 s in ring 1, node 8. A debris region forms shortly there-after in ring 1, node 7. The rod section in ring 1, node 9, fails at 10197 s followed by the rod section in ring 1, node'7, at 10199 s. By 10206 s, the debris region in ring 1 is 1.25 m high and has an average temperature of 2140 K. The region is fairly porous and has steam flowing through it. Consequently, the unoxidized Zr in the region continues to oxidize.

l l

l

)

At 10273 s, the rod section.in ring 1, node 6, fails. This failure causes the debris region to lose its support and conse-quently the debris relocates downward. However, the debris'does not have much superheat and a new debris region forms in ring 1, cell 5. By 10303 s, sufficient remelting and relocation of mate-rial (mainly Zr) in the debris region has occurred that a crust (5 cm thick) has formed at the bottom of the region. At this point, the steam flow through the region ceases and this termi-nates the Zr. oxidation in this debris region.

This region continues to melt and relocate internally, but

- remains stable because the bottom of the region is cooled by steam from below. In the current model, the crust at the bottom of the region is assumed to be stable until it attains the Zr melting temperature (2200 K) or loses its support. Due_to the poor heat transfer in the region and the cool steam below the i region, the time scale to reach a failure limit is long. In I fact, not unti1 14877 s does the crust fail by melting. The long time to failure seems too large and is currently believed to be due to a heat transfer coupling error between the debris and the_ ,

gas. A development effort is currently under way to correct this problem.

The pattern of rod section failure and debris region forma-I tion is similar in the other rings and occurs during the same time that the sections in ring 1 are failing. In ring 2, the debris region ultimately forms in node 5, while in ring 3 it ultimately forms in node 4. While the actual locations of the debris regions are not exactly the same, the general meltdown behavior is the same in each ring.

It should be noted that treatment of the debris region failure and relocation does depend on the mesh size. That-is, it is always assumed that the bottom of a debris region coincides with the bottom of a FLUIDS node. While this approach is practi-cal, it does introduce some mesh size sensitivity into the results. However, until more research into crust behavior is performed a more detailed model is probably not warranted.

This concludes the discussion of the debris region forma-tion and subsequent behavior while the debris is in the core region. The following sections discuss the core slump phase-and the vessel heatup and failure.

4.4 Core Slump As the sequence proceeded, the debris regions in all-rings continued to heat, melt, and relocate internally. The region in ring 1 began in node 5 and was 2.2 m high. In ring 2, the region also began in node 5 and was 2.4 m high. The region in ring 3 4

began in node 4 and was 3.1 m high. The state of the vessel at this point is shown in Figure 4.15. At 14877 s, the crusts in ring 1 and 2 failed which marked the beginning of the core slump

I l

phase. At this time, it was arbitrarily assumed that the region 4 in ring 3 would also release.

At this point in time, the average temperature of the core  ;

debris was 2840 K. This high temperature is directly related to j the long time required to fail the crust. An earlier crust fail- j ure time would have resulted in a lower temperature. Of the i debris, approximately 72% was molten. Due to the large degree of I superheat and large molten fraction, a significant amount of the core debris entered the lower plenum after the debris regions in ,

the core released. At the time just prior to core slump, there j wae approximately 3000 kg of saturated water in the lower plenum.

The code calculates that within 5 seconds after the initia-tion of core slump approximately 75000 kg of debris relocated into the plenum. This implies an average corium fraction of 76%.

This high packing fraction is due to the large molten fraction in i

the debris prior to slump. While the current version of the code does not contain a fuel-coolant interaction model, it does model i the heat transfer between the corium and the water and steam. -

When the corium and water mix, the corium partially quenches l while the water rapidly boils via film boiling. Within 5 s, 95% l of the plenum has been voided. The voiding occurs due to vapori-zation and to sweep-out of the water (by entrainment and dis- I placement). Most of the remaining water is in the downcomer region. This water enters the plenum and vaporizes as it does.

Within 15 s after, the plenum is devoid of water and the average l corium temperature is 2350 K. l l

The steam produced during the rapid vaporization raises the i the pressure in the interaction zone (the bottom of the lower ,

plenum) to 17.9 MPa. During the interaction, a 0.6 MPa pressure I drop develops across the downward falling corium above the inter- l action zone. This pressure drop is not large enough to levitate i the corium. This pressure drop is enough, however, to move steam upward through the falling corium with a velocity in excess of 2 m/s. The steam flows through the core, the upper plenum, and out of the vessel. The vessel exit area corresponds to the flow area of one hot leg. Even with this relatively large flow area (0.426 m2 ), exit velocities in excess of 200 m/s were seen during the core-slump interaction.

In this model, there is, of course, no further limitation to the flow modeled. In reality, .the flow must enter the i smaller-area surge line, the pressurizer, and then the even smaller flow-area relief valves. These flow area reductions, which are impediments to venting of the steam, imply.that the primary system pressure must be higher than that seen in the MELPROG calculation. On the other hand, the steam production rate predicted by MELPROG is probably also too high. In order to gain more insight into the interaction phenomenon and its resultant effect on the primary system pressure, two bounding cases were considered. First, the maximum pressure can be l

estimated by assuming that no steam exits (i. e., the steaming rate is so high that the steam cannot leave the primary system in time to lower the pressure). The amount of steam initially present can be calculated, if it is assumed that one third of the system (the vessel) is initially at 1400 K and the remainder is at 622 K (the saturation temperature). Knowing the initial steam mass, the amount of steam produced (3000 kg) and the volume of the primary system, the final steam specific volume can be calculated to be 10.6 m 3/kg. The pressure is then uniquely determined by the steam temperature. If the steam generated by

the fuel interaction is completely mixed with steam present in the primary system, the temperature would be 640 K, yielding a pressure of 17.0 MPa. Since the two thirds of the system outside of the vessel is probably above the saturation temperature, a more likely resultant temperature would be 660 K, yielding a final pressure of 19.0 MPa. Higher temperature steam initially present in the system will, of course, lead to an even higher pressure. At the other extreme, the pressure would remain rovatant if all of the steam produced could be vented through the reiief valves. The Surry pressurizer has two PORVs and three safeby valves for a total steam relief capacity of 165 kg/s.

The~r ebJe, if the 3000 kg of lower plenum water were vaporized over a rariod of 18 s (or longer), then the resulting steam could be vented with no increase in system pressure. These considerations point to the need for both a more accurate fuel-coolant interaction model to accurately estimate the steam production rate and a complete primary system model to have the system condition and pressure drops outside of the vessel codeled correctly.

4.5 Vessel Head Failure The corium which enters the lower plenum after core slump is only partially quenched by the water in the plenum. This is due to the low inventory of water and the high debris tempera-ture. The debris quickly accumulates into debris regions in the first 3 rings throughout the lower plenum. In fact, the debris regions extend from the bottom of the vessel up to node 4 in the core ( >2 m high). After boiling the water away the debris has an average temperature of 2350 K.

As time progresses, the debris regions heatup, remelt, and compact. The heatup is due primarily to decay heat since there is no steam flow through the regions which could oxidize the unoxidized Zr. The heat generated in the regions is transferred upward by radiation and downward to the vessel by conduction.

As indicated above, the core slump occurred at 14877 s. At this point the average vessel temperature is 610 K. Even though the heat transfer between the debris and the vessel is poor, the high debris temperature results in a rapid heatup of the vessel.

The average temperature of the vessel bottom in the central ring

l l

1 1

heats at a rate of 1.5 K/s. This' fast heating rate results in a rapid failure of the vessel bottom.

In the current calculation, the vessel was predicted to fail in ring 1 at 15928 s. The mode of failure was a creep- l rupture type and not a complete melt-through. At the time of failure, the average temperature of the vessel bottom was 1428 K, and ablation of the vessel had begun. It should be noted that no instrumentation tube weld had been modeled in this calculation and, hence, no failure of this type could be predicted. After the vessel bottom failed, the calculation was terminated.

At this time, the debris in the vessel was 30% molten on average. The debris in the lower' plenum consisted of both the core material and structural steel. Table 4 3 gives a brief summary of the state of debris regions at the time of' vessel failure. This information is broken down on a compositional basis to indicate the distribution of the debris. Included here

~

is the amount of steel in the debris,.most of which is added to the debris after core slump. Also included here is the amount of unoxidized Zr. This amount (9575 kg) represents 58% of the orig-inal inventory. The average temperature of the debris was 2460 K.

In the present calculation, vessel failure would be local-ized to the central ring. In this case one would expect that.all of the debris in ring 1 would exit the vessel. In addition, all of the molten material in the other rings would : exit. Based on these estimates, one would expect at least 70650 kg of the debris to exit the vessel. This value represents 52% of the core debris. Since the debris remaining in the vessel is hot and is being heated, it will continue to melt. We estimate that this debris will melt at a rate of approximately 38 kg/s. Hence, that this remaining debris would be entirely molten in 1750 s. Of course, these post-vessel failure results are strictly extrapola-tions of the calculated results.

~ _ . ._ _ .. .. _ .

5.0 PRELIMINARY INSIGHTS AND ASSESSMENT The base-case calculation described in the previous sec-tion, while preliminary in nature, represents.the "best-effort" MELPROG calculation of this accident sequence. Due to the prelim-inary nature of this calculation, we feel that it is premature at

, this time to draw firm conclusions from this calculation. On the other hand, we recognize that this calculation, and MELPROG it-i self, can be used to gain important. insights into severe accident phenomena.

These insights can be gained in a number of different ways.

j The approach we have used in this study is to perform-limited sensitivity studies, auxiliary calculations, and comparative analyses to investigate specific aspects of the overall calcula-tion. The aspects chosen to be investigated were selected based on their perceived importance. Obviously, there are many impor- ,

tant aspects of such a calculation and we by.noLmeans claim.to '

have studied all of these. This whole area of study is continu-ing and, with time, more aspects will be investigated. At this point, we will simply summarize our current findings.

, A total of five aspects of the calculation have been studied in detail. These are:

1. Effect of natural circulation 2

2. Effect of fuel rod failure temperature

3. Effect of pressurizer cycling, and 4 4. Thermal history of hot leg nozzle, and

5. Core debris state at vessel failure.

l Each of these is discussed in a separate section below.

I j 5.1 Effect of Natural Circulation i

I One of the main reasons for developing the 2-D version of MELPROG was to allow the code to predict natural circulation in

,' the vessel. Previous studies had indicated that convection : cells between the upper plenum and core region could exist. If pre-sent; the resulting circulation could remove a significant amount of energy from the core. Hence, this phenomenon was expected to play a significant role in the meltdown progression. However, i the magnitude of_its effect was unknown and we felt it important to try to assess the effect of the phenomenon.

In performing this study, we have compared the results of the 2-D base case to a 1-D MELPROG calculation of the/same sequence [2] . Both calculations use the same geometrical model

! for the core, upper plenum,'and lower plenum. The main

- - - . .- -. - . . . . . _ . - - - . . . . ~ . . . _ . - - ,,, _ . ,

\

~

l l

difference between the two modelsois that the 1-D case did not model the downcomer. Nevertheless,'the overall modeling of the two. cases is similar enough that the.effect of 2-D natural p-circulation can be directly assessed.

At the initiation of the WELPROG transient in both cases, it was assumed that all primary-coolant flow had ceased and the fluid temperature was slightly subcooled. The core's nuclear decay heat quickly brings the stagnant primary fluid to satura-i tion and boiling begins within the core region. The entire length of the heated core contributes to vaporization'until'the upper plenum is completely voided, i.e., steam filled. 'During this period, a constant. steam generation rate occurs (which'is easily calculated) and this rate has been found to be in agree-

! ment with that predicted by.both versions of MELPROG. This <

j agreement indicates that mass and energy balances are correct.

During the core uncovery phase of-the accident, the heat removed from the fuel rods is deposited entirely into steam-production and eventually into the steam itself. In the one- -

t dimensional case, the steam simply exits the core region and eventually the vessel. However, in an actual PWR, multi-t dimensional flows establish a natural circulation flow path i between the core and the upper plenum and loops. This calcula-

tion is, in fact, predicted by the two-dimensional version of.the-code. The circulation transports energy more rapidly from the  !

4 core to the available heat sinks, thereby delaying the time to I

core oxidation and melting. It also' heats the vessel components l (especially in the upper plenum) much more rapidly than is the

! case for the one-dimensional model. In-terms of fission product release, the natural circulation may be very important because of its effect on mass transport processes. Also, the circulation

. between the vessel and the hot legs could lead .tx> a massive fail-

! ure in the hot leg region which would depressurize the system.

The circulation also affects aerosol and fission product deposi-tion on structures. Finally, the natural circulation makes the core meltdown more uniform which may affect the ultimate H2 and fission product-generation.

. Many of the effects of natural circulation are obvious in

! the two-dimensional calculation. For example, it is easy to see t the effect circulation has on core heatup. Figure 5.1 compares i the calculated core heat-up rates for the 1-D and 2-D cases. It l is seen that the circulation, as' expected, significantly delays-the core heat-up and beginning of oxidation. Relative to the 1-D calculation, these events are delayed'approximately 900 s.

Simply put, the circulation leads to' greater heat removal from the core and, hence, less of the energy generated goes into rais--

I ing the temperature.of the core. However,'this has-the less obvious effect of heating the plenum structures at a faster rate

~

than in the 1-D~ case. This difference can be-seen in Figure 5.~2 where the surface temperatures of upper plenum structures-are compared. The structures in the 2-D case heat much more rapidly

and eventually melt, whereas tne structures never melt in the 1-D case. This coupled behavior has significance later in the sequence in regards to availability of surfaces for fission prod-uct deposition.

Natural circulation also changes the spatial temperature distribution in the core. In the 2-D case, the radial tempera-ture distribution is non-uniform while the axial temperature distribution is nearly uniform. In the 1-D case, the converse is true. This difference can be understood by considering the cir-culating flows. The steam flows up the central regions of the core, cools in the upper plenum and flows down the outer regions of the core. Due to the higher power in the central regions of the core and the lack of heat sinks (such as the core baffle),

the temperatures are highest in the central region of the core.

However, since the magnitude of the flow is relatively high, the t'emperature gradient in the axial direction is low (relative to the 1-D case)., This means that oxidation and failure occur more or less uniformly in the central region of the core. In fact, all rods in the top half of the core in the central ring fail within 90 s of each other. However, the radial gradients are relatively large. When the first rod fails in the central ring (T = 2200 K), the maximum rod temperature in the outer ring is 1670 K. The first rod failure in the outer ring does not occur until 150 s after the first failure in the central ring. In the 1-D case, the temperatures and failures were much more uniform radially.

The change in the core heating pattern will also affect the hydrogen production. However, it is not clear as to whether this is a significant effect. The comparison of the 1-D and 2-D cases shows that the 2-D case produced 30% more hydrogen than the 1-D case (see Figure 5.3). While this difference is important, it is small compared to the difference found when the relocation temperature is changed (see next section). Hence, further assessment of the effect of natural circulation on hydrogen production is needed before definitive conclusions can be drawn.

After fuel rod relocation begins, the circulation patterns in the core change dramatically. The relocated material forms partial, if not complete, flow blockages in the lower core regions. These blockages represent a large impedance to flow and disrupt the circulating flow pattern. The flow through these ,

blockage regions is very low. The flow tends to bypass these  !

regions and new convective flow patterns are established. These exist between the upper plenum and the region above the blocked core. This flow will significantly affect fission product release and transport.

The effect of the relocating material on the core flow can be seen in Figure 5.4. This figure shows the mass flow rate of steam at the core mid-plane in the central ring. Before reloca-tion, the flow is seen to be significant at approximately l

r .

4 l 10 kg/s. The sharp drop at ~10180 s is due to-the relocation of

. fuel-rod material after the initial failure. One finds that the flow at the mid-plane is' essentially zero after relocation begins. On'the other hand, the flow rate at the top of the core -

4 in the central ring, while dropping during relocation, eventually reestablishes itself (see Figure 5.5) . The flow at the top of the core is higher than at the mid-plane before relocation,

< indicative of the natural circulation. During relocation, the flow.is temporarily. disrupted, but soon reestablishes at a level-of ~15 kg/s. It stays constant at this level'until core slump

occurs. It should also be noted that throughout this time the.

flow at the bottom of the core was essentially zero (see Figure 5.6) i lit is seen then that natural circulation.is an extremely l significant phenomenon. Its main.effect is to modify.the energy partitioning of the core. The " extra" energy extracted from the core delays the core heatup, but increases the upper plenum temperatures. The convective flows change the' axial and radial

. temperature gradients in the core. This,.in turn, modifies the hydrogen generation rate. Finally, the flows are found to be important primarily before relocation. After relocation, the 4 flow patterns, while still existing, undergo radical changes in the core region.

! 5.2 Effect of Fuel Rod-Failure Temperature j

In the current calculation, the original fuel rod model was used. This model uses a rather simplistic ~ approach to fuel rod failure. Instead of performing a mechanistic failure calcula-tion, the model relies on parametric failure criteria which are a function of oxide thickness and temperature. The model is based on the assumption that as the cladding melts it could be.

maintained (i.e., held in place) by an oxide layer if the layer 1 is both thick enough and strong enough. The strength factor was felt to be strictly a function of temperature. Since fuel rod i failure marks the beginning of relocation, the modeling was felt to be rather important in the meltdown progression.

J In the base case, the failure temperature.was set to 2200 K. This value is the oxygen stabilized Zr melting tempera-ture. Practically, this temperature means that when the metallic part of the cladding is completely molten, then relocation could ]

begin. That is, no effect of the oxide layer was~ considered in  ;

this case. For testing purposes, this lower limit would expedite I the calculation. q

l It should also be noted that the corium flow regime map l

' assumed that if the corium temperature exceeded 2200 K, then~the i H

i corium behaved like a liquid (or slurry), while if the tempera--

ture was below 2200 K, then the corium was a solid. The key item here is that with corium'being " created" at 2200 K by the fuel-4

, l

- y y g y & -

ep- ---g- -

3 m yy--- -

y.r. y7--yfe--W-= "-R-em-m 3- way t emmy ,'= g' g 9'"*--7

rod model, a small heat loss would lead to freezing of the corium. Hence, relocation was very limited. )

In order to investigate the sensitivity of the overall )

predicted results to this failure temperature, the calculation ,

was rerun through the rod failure point with the failure point increased to 2500 K. The higher value was based on observations from the PBF-SFD experiments. Again, while this modeling is still not mechanistic, it does allow the sensitivity of this one area to be assessed.

The results of this rerun calculation indicate three impor-tant areas which are very sensitive to the failure temperature.

The first is the amount of hydrogen produced. In both cases, the beginning of relocation marks a rapid decrease in the hydrogen production rate. This is due to the reduced area for oxidation and reduced steam flow. Figure 5.7 shows this effect. What this means is that ,the longer the rods remain in an intact state, the longer hydrogen can be produced (provided steam starvation does not occur). Furthermore, since the rod temperatures are high just prior to relocation, the rate of hydrogen production is large. Hence, the change in relocation temperature results in a 50% - 100% increase in the amount of hydrogen produced.

The second important area affected is the degree of reloca-tion. In the rerun calculation, the corium does not begin to relocate until it is 2500 K. Since the corium flow regime map was not changed, the corium, when formed, has significant super-heat. Hence, a large amount of energy must be removed before the corium would freeze. This means that the corium stays liquid longer and relocates further. In fact, large quantities of corium relocate into the lower plenum. The corium quenches and vaporizes much of the water remaining in the plenum. The steam generated in this manner momentarily cools the core and sweeps out the hydrogen.

The third important area affected is the amount of energy produced. The energy produced during the oxidation is propor-tional to the amount of hydrogen produced. As discussed above, the rerun case resulted in a significantly greater amount of hydrogen produced. Hence, the amount of energy generated is also much larger. Due to the relatively low decay power, the energy produced during oxidation has a dominant effect. The practical ,

result of the increased energy is that the sensible heat of the j corium is much higher. This causes an acceleration of the melt-down. It only takes an additional 35 s to increase the rod tem-peratures from 2200 K to 2500 K when they are rapidly oxidizing.

Decay heating would require approximately 600 s to raise the temperature the same amount. Hence, the corium created is much I hotter and this will result in a reduction in the time to vessel I failure. l l

1 l

l l

- - . n. -. .. . . . -. .. -- -- - - . - . ._. .

I g ,

)

o 4-In summary, the temperature of fuel rod failure,has signif- )

icant effects on the overall' calculation. The amount of hydrogen R produced is increased by nearly a factor of-2 by increasing.the i failure temperature from 2200 K to 2500 K. The corium1 created '

has more energy-(i.e., higher temperature) and can relocate much further. Also, the timing of the meltdown progression can be greatly accelerated. In view of these effects, it is vital that MELPROG have a mechanistic model-for this failure. Current

> efforts in the CORE module development are_ addressing this impor-tant-question.

1-5.3 Effect of PORV Cycle

. In Sections 4.1 and 4.2 that describe the base-case calcu-lation, it was seen that a~ natural convection' cell in the upper

plenum and top of the core region developed and was maintained.

The vessel exit boundary cogdition for the base case was a con-stant pressure,of 16.3 MPa9 'In a "real" TMLB' sequence, th QRV.

will open and close to maintain'the pressure near 16.3 MPp

the actual pressure varying'between 16.0 MP&P2Wd 16.6 MPa.

check the stability of the natural convection cell.through.a PORV -'

l cycle, an auxiliary MELPROG calculation was run in which the PORV cycle was simulated. For_this calculation, the pressuye at the i vessel outlet was varied by 9.3 MPa over a period ef d <s as is ,

! shown at the top of Figures 5.8a-c. The initial condition-for ,

j this calculation is shown in Figure 5.8a. Comparing Figure ~5.8a  :!

and 5.8b we see that the major effects of changing the vessel

t. outlet pressure are a disruption of the part of the cell that reaches downward into the core region and a cooling of the gas j throughout the vessel. Both of these effects are a result of ,

increased steam flow from the lower plenum which, in turn, occurs l when reduced pressure induces flashing of some of the water j there. Note that while the gas temperature in the level 4 cells

' is considerably cooler (~3OO K) , the rod surf ace temperatures are i not changed much(<70 K) , especially at the ' top of the core. Also note that both the flow out of the vessel and the exit temperature have' increased. Figure 5.8c shows the vessel at the

end of the PORV cycle. The gas at the bottom of the core remains cooler than it was at the start of the cycle, but the initial I flow pattern is re-establishing itself. The preliminary-l conclusion, based upon this auxiliary calculation, is that a cycling PORV can temporarily effect the-flow pattern and temperatures, but'the initial state is very stable and is rapidly re-established.

5.4 Hot Leg Nozzle Temperature f An important assumption in this and all other TMLB' se-

! quence calculations ic that the primary system remains at or near y l

the pressure corresponding:to the set' point of one of~the relief I 4 valves. The possibility of a failure somewhere in the primary  !

l system, followed by depressurization needs to be examined. In H

this section, we provide'an estimate of'the thermal condition of j

] t t

___ . . .._____u. - . _ . .-_ _ _ - ,_ _ _ .._ .-.... _,_, ___,,.._ _ _ _ ._ ._ ,.-_,m,._,.

s '!

i one of the primary = system weak points, the connection between the

-vessel outlet nozzle and the hot leg. The temperature determined-here can then be used in a structural analysis to determine.if and when a failure may occur. In the1 absence of an implicitly linked TRAC-MELPROG calculation, we can estimate the temperature.

l of.the vessel-hot leg-connection by using the MELPROG outflow L conditions as boundary conditions for a TRAC model of the' nozzle and hot leg assembly.

T

-The boundary conditions needed for the TRAC calculation are the mass flow and temperature from the vessel, as calculated it by MELPROG. These-quantities are given in Figures 5.9 and 5.10, respectively. The TRAC model includes.the portion of the nozzle-that is external'to the vessel and the hot leg itself. To:obtain the temperature distribution through the nozzle and hot leg wall,

four radial nodes were used. The boundary condition for the external surface of the nozzle and hot leg is assumed to be _

convective heht transf er to a vapor at 400 K with- a heat transfer coefficient of 10 W/m2 -K. This heat transfer coefficient is i representative of natural convection from a horizontal pipe that is surrounded by degraded insulation. The temperature distribu-tion through the hot leg wall at the nozzle connection is given i in Figure 5.11. The temperature can be seen-to be above 1000 K i for more than one hour before core slump occurs at 14877 s.

! Initial indications are that the connection.will fail rapidly if j above 1000 K[8] . If this is the case, then system depressuriza-4 tion by this means is likely by about 11000 s into the transient.

As is indicated in Figure 5.11, this is well before either core i slump at 14877 s or vessel failure at'15928 s.

5.5 Core Melt Insights i An accurate description of the core debris and. melt in the

vessel at the time of vessel failure is an important source term consideration. The state of debris needs to~be

known to provide

! appropriate sources to containment codes. The state of the i debris also affects the nr.gnitude of the in-vessel fission prod-

uct release.

! To analyze the state of the debris, one must at least determine both the composition of the debris and the average temperature of the debris. The debris will be-composed of a j

number of constituents, the major ones being UO 2, Zr, U-Zr-0  !

mixtures, Zr0 2 , steel, and control rod materials. The masses of ,

each of these components is important. Both the. steel and Zr '

masses are especially important since these materials can oxidize  ;

and release energy in the containment.and these~can also affect l i fission product release during core concrete interactions. The l

! masses of liquefied materials (or molten fraction) also must be i' calculated. The temperature of the debris is important as this

determines the~ initial thermal-loading on the containment.

?

l i

l

The tools available for calculating the state of the core debris are limited. The MARCH code is probably the most widely used and results of this code can be found in the BMI-2104 report ,

and in the QUEST study (SAND 84-0410) [9] . The approach in MARCH I is parametric and relies heavily on engineering judgement. Both i key phenomena and important coupling between phenomena are i neglected. These factors are reflected in the large uncertain-i ties in the MARCH results.

By far the most mechanistic code is MELPROG. In MELPROG, 3

the meltdown progression is treated in a continuous and mechanis-

< tic manner. That is, melting and relocation of materials is-

treated in a physical manner without user interaction. There is

no separate treatment of the core debris as it exits the core

region into the plenum. This integrated analysis, while fairly complex, does allow all the relevant phenomena to be treated in j an appropriate manner.

I While comparisons between MARCH and MELPROG are possible, a couple of points need to be considered. First, the two codes are

completely different in approach and modeling. This can lead to substantial differences in the results. Second, MELPROG is not 4

yet assessed and, hence, caution must be exercised in interpret-i ing its results. Also, only a very limited number of MELPROG

> calculations have been made and little has been done in the way

of sensitivity studies.

{

Nevertheless, it is useful to discuss the current under-standing in the area of core melt. The results of three MELPROG runs and one MARCH calculation for a TMLB' accident sequence in the Surry reactor are presented in Table 5.1. While all cases differ in one way or another, this table illustrates the spectrum of possible results. The MARCH results have been taken directly j from BMI-2104, Volume V. The first two MELPROG runs are 1-D i

calculations while the third run is a 2-D calculation. There were two differences between these two 1-D cases. Case 1 used a t relocation temperature of 2200 K and did not model the downcomer l water inventory. Case 2 used a relocation temperature of 2500 K i and modeled the downcomer water inventory. The effect of these

. modeling variations is discussed below. Also, it will become i evident, that the state of the core debris at vessel failure is a strong function of the in-core modeling.

In fact, the main p igh that needs to be made is that the state of-debris at vesse1gdepends on the assumptions and modeling 3

, used during the meltdown progression in the core region. Certain j quantities such as hydrogen generated and Zr reacted are depend-i ent on assumptions made before any fuel rod-failures occur. The

importance of these assumptions and models is discussed-below.

1

! Without getting into a great deal of detail, there are four

, models which lead to the differences in the results. The first, and most important, is the modeling of cladding relocation (in-i

, _ . , - - - , . . - _ _ _ _ _ - . , _ , - _ , , - .,, ,. _ ,,,- . , _ .,,-.n- - -

l

-i these calculations this means the temperature at which the clad-ding begins to relocate). This model affects the amount of Zr oxidation and the amount of energy generated in the core. The interesting result which-is consistent in all calculations is that the majority of oxidation (> 80%) occurs while the rods are in intact geometry. Once relocation-begins, the flow channels become blocked, which leads to a reduction in steam flow. Also, the surface-to-volume ratio of the Zr decreases, which inhibits oxidation. Both of these factors lead to marked reduction in the ,

oxidation rate following relocation. Hence, the relocation modeling is extremely important. ,

Note that the MARCH calculation and the MELPROG-ID, case 2,  ;

calculation used similar values for the cladding relocation tem-perature. Both of these calculations predict high oxidation which means that there is less Zr metal available for release to the containment. The other two cases used the lower relocation temperature an_d have significantly less oxidation. '

In addition to differences in oxidation, the relocation model has two other effects. First, the high:Trelo ca8*8 attain vessel failure earlier. ThisisduetotheSatearge increase '

in energy deposited by the Zr oxidation. It takes only 35 seconds for the temperature of an intact rod section to rise from 2200 K to 2500 K when the rod is rapidly oxidizing. This time is negligible when compared to the overall timing. The huge increase in energy (equivalent of a 300 K temperature increase) is very important and this leads to early vessel failure. The other effect is that the average temperature is lower as is the molten fraction of the debris. This is because the overall timing is reduced and the debris does not have the time to get hotter.

To see the big effect of the increased relocation tempera-ture, compare the two 1-D MELPROG cal'culations. The first case did not model the downcomer and used a lower relocation tempera-ture. The second case modeled the downcomer and used a higher relocation temperature. The effect of the downcomer is to increase the time to initial rod failure by about 10 minutes. -'

Hence, if everything were the same, then the' timing of the second case should be 10 minutes longer. However, it is 10 minutes shorter. This means that the increased relocation temperature has shortened the accident by 20 minute.s. It has.also doubled the Zr reacted (and H2 produced). Ho+ cr, the molten fraction of the debris is reduced by a facter of J. This comparison clearly demonstrates the intimate coupling between the meltdown l phenomena.

The second area of modeling which affects the core melt is y natural circulation. To see the effect, _one can compare the 2-D calculation with the 1-D, case 1, calculation. The biggest effect of the circulation is to delay the time to failure. In this case, taking into account the downcomer effect, the circula- l tion would delay the time to failure by about 30 minutes. This delay, however, is due to transport of energy from the core debris to in-vessel structures. Hence, the debris itself is not much different at the time of failure. However, by transporting heat to in-vessel structures, more structures melt relative to the 1-D case. Comparing the 1-D case, case 2, to the 2-D case, one finds about 10,000 kg more steel in the debris. This is due to the melting of the upper plenum structures. Hence, the nat-ural circulation, by transporting energy from the debris, leads to longer vessel failure times, more steel in the debris, and lower average debris temperatures at vessel failure.

The third model which affects the debris state is the core slump model. This model determines how the core debris leaves the core region and enters the lower plenum. The core slump model mainly affects the timing of the vessel failure. However, the timing of vessel failure determines the average temperature and molten fraction of the debris. It should also be noted that large uncertainty still exists in the modeling of the core slump.

If core slump occurs with the core debris at low molten fraction, then it will take a long time to attain vessel failure due to the poor heat transfer through the debris. This allows the interior of the debris region to become significantly molten and attain high temperatures before vessel failure occurs. On the other hand, if core slump occurs with the core debris at a high molten fraction, then vessel will heat faster and fail earlier. This earlier failure prevents the interior of the debris region from becoming significantly molten. However, this current finding is probably scenario dependent and should not be generalized.

The fourth model affecting the core debris state is the actual vessel failure model. Obviously, the vessel failure model will affect the timing of the failure. Again, the timing of the failure, which, in turn, affects the molten fraction and tempera-ture of the debris. Failure can occur due to melt through, mechanical rupture, or instrumentation tube failure. The current MELPROG calculations have only considered the first two modes of failure. In each case, mechanical rupture occurs before complete melting of the vessel. However, instrumentation tube failure may occur earlier. Modeling of this is currently being studied.

4

. l

-6.0

SUMMARY

, CONCLUSIONS AND RECOMMENDATIONS

[

The first complete, coupled, and mechanistic analysis of a ,

i .TMLB' (station blackout) core meltdown accident has been made

) with MELPROG-PWR/ MOD 1. The. calculation was initiated at the point. boiling began in the core region and ended with failure of l '. the reactor vessel. Most of the important phenomena occurring in i the accident sequence were modeled during this accident sequence.  ;

The important exceptions are a treatment of cladding motion prior i to. major disruption of the fuel rods . (candling) and a treatment

, of the fission product release,-transport, and deposition (as

[ treated by the VICTORIA module that is being implemented). While i this calculation should be viewed as preliminary, it does demonstrate _the advanced capabilities of this version of MELPROG.

4 The " base case" calculation has provided the timing of the j major events occurring in the accident, the amount and timing of hydrogen produ'ced by' oxidation of Zircaloy cladding, and the-I condition and composition of the disrupted material at the-time

! of vessel failure. Because of the preliminary nature of,the of j this first calculation, we have performed a limited number of l auxiliary MELPROG calculations. The base case, the auxiliary 1 calculations, and a comparison of these results with previous

! calculations have provided further insights into this accident

} sequence. These insights have allowed us to draw conclusions in j regards to the important phenomenological modeling required for

' such accident sequences.

T In particular, we have shown that natural circulation

! reduces the rate of core heat-up, but increases the rate of heat-j up of upper plenum structures. This implies that a significant amount of core energy is deposited in the plenum and primary

< piping. This increased heating can inhibit fission product j deposition and increase the amount of molten structural' steel in the melt at vessel failure. We have also shown that_the coupling between vessel flow and primary system flow may lead to early ,

heatup and failure of the primary system. Natural convection 4

cooling of the top of a debris region, such as in the lower head,

also lengthens the time to vessel failure. Hence,~ natural i circulation within the vessel with coupling to the primary system.

I can completely change the course and timing of a meltdown sequence. This underlines the importance of a multi-dimensional 4 vessel flow capability coupled to a complete treatment of the

primary system such as will be-provided by MELPROG/ MOD 1-TRAC.

In addition, we have studied the effect of the modeling.of 3 the initial fuel rod ~ melting and relocation. Variations in the

. assumptions were found to strongly affect hydrogen production and l the subsequent course and timing of the accident (total hydrogen l l

l production was doubled and vessel failure occurred 20 minutes

] earlier for a higher failure and relocation temperature). Thus, ,

we have shown the need for the more accurate models provided by l l the MELPROG CORE module that is currently being implemented.

a

I _ , , _ . _ . _----____.._,____..,____.a ._a__-_

Finally, we have alluded to the need for additional experi-mental and analytical information in areas such as debris crust failure related to core slump and vessel penetration failure prior to gross vessel wall failure. We recommend that such information be obtained and incorporated in MELPROG along with the implementation of the CORE and VICTORIA modules and the TRAC

link. With these capabilities in place, we recommend that the calculation be re-run to fully assess the effects of this improved modeling capability.

e' 1

2

7.0 REFERENCES

1. W. J. Camp, et al., MELPROG-PWR/ MODO: A Mechanistic Code for Analysis of Reactor Core Melt Progression and Vessel Attack Under Severe Accident Conditions," FIN NO. A-1342, Sandia National Laboratories, Albuquerque, NM (1985) .

2. K. A. Williams, T. J. Heames, and J. E. Kelly, " Mechanistic Calculation of a Station Blackout Accident (TMLB') in a

'Large PWR Using MELPROG-PWR/ MODO," Internal Memo, Sandia National Laboratories, Albuquerque, NM (Nov. 1985).

3. J. A. Gieseke, P. Cybulskis, R. S. Denning, M. R. Kuhlman, and K. W. Lee, Radionuclide Release under Specific LWR Accident Conditions--Volumes I-VII (Drafts), BMI-2104, Battelle Columbus Laboratories, Colombus, OH (1983).

4. "Surry Power Station Final Safety Analysis Report (FSAR) , "

, Vol. 1, 2, and 3, Virginia Electric and Power Company (VEPCO) (Dec. 1969).

5. " TRAC-PF1: An Advanced Best-Estimate Computer Program for Pressurized Water Reactor Analysis," LA-9944-MS, NUREG/CR-3567, Los Alamos National Laboratory, NM (1984) .

6. B. E. Boyack, " Loss-of-Offsite Power Transient for the Zion-1 PWR," LA-UR-83-1714, Los Alamos National Laboratory, NM (1983).

7. V. L. Shah, " COMMIX Application in TMLB' Accident Scenario Simulation," presented at USNRC, March 17, 1986.

8. V. Shah, " Structural Failure Studies of the RCS," presented at the USNRC, April 23, 1986.

9. R. J. Lipinski, P. K. Mast, D. A. Powers, J. V. Walker,

" Uncertainty in Radionuclide Release Under Specific LWR Accident Conditions," SAND 84-0410, Vol. 1, Sandia National Laboratories, Albuquerque, NM (May 1985) .

l l

ad$ *

?d [h. : -

i! r:

ENSTRuedENTATION t# pea SUPPOnT PLATE

%I l t' 4 9h]

conEsAnnEL 6

\$ Mr d % ' /

-) its 1 I 1 l i:

sueronT cotm

/

/

p d

',,i e uerEn cone cowinot noo t

I.  ;, ,,

],

OUTLET Noutt '. 3- 9

,l! l '_V -

I m -

lg~~y\ , !l t

l mLET NoHLE SAFFLE f CONE SUPronT

^-

~f i I

coLtanus \ s .

,f

[i!

ql l l

VN '

M r p s

- g nEACTon VESSEL p -  %

d l LOWEn CORE PLATE soTTw surconT ,

4

x $R s 1

,,,,,,,a .R-

./

Figure 2.1: Schematic of Westiaghouse Reactor Vessel

~39-

1.2 1-i 0.8 - \ e-i 4

% 's O '. TRAC U 0.6 - '.

g MELPROC 's a b.

p s 4

o 0.4 - 'ss c

H o

o 4, ,

3 0.2 - -

m a:

O

' o. , , ' s J

U s, \

's I 0-  ? '

l l

-0.2 . . . . .

6000 6500 7000 7500 8000 8500 9000 TIME (S)

Figure 4.1: Comparison of Core Liquid Volume Fractions

1300

• I 1200 -

d

/

/

1100 - /

o'

/

a' o

^

1000 -

6 ,

9 5 ,l' 900 -

/

@ /

N ,8

/

800 - ,

,/ TRAC MELPROG ,/

/

o' 700 - g 9

l

600 , , , , ,

! 6000 6500 7000 7500 8000 8500 9000 l TIME (S) 1 Figure 4.2: Comparison of Maximum Cladding Temperatures

~42-

l l

e e 1 S

9 l

I I

e . e..e eeeeeeeeee oeee4e e9eoeeee oe. eoee.oeoeee eeeoe e

eeea.e e e.e.

eeeeee oeoeee eeeeee eeeeeee eee e eeee eoeeee eeeeee eeeeoe eeeeeeo eee eeeoee eeeeee eeeoee oeeeoe ee eee e4eeeee eeee ee eeeeee eoeeee eeeee4 eeeeee eeeeeoe e o o 'eeeeeee eeeeee eoeeee e $eeeee e ea9eee eoeeoe eeeeeeo e ekeeeeeee eeeeeeeeeoee eeeeoeeeeeee eeoeeeaeeee oe eeeeeee e e eGe e89eeee ee eeee eee9ee eeeeoeeee e oeeoeeoeeeo 9eeeeeeeeeee 00 *eeoeeeeeeesee e eoeeeeeGeeee eeeeoeeeeeee eeo e eeeeee 800eeeeeeeee toeeeeeeeeeo eeee4eeeeee eee9eeeeeee eeeeeeeeG94e eeee eeeoeeee49eeee eeeeeaeeeeecee eeeee eseoe oe49e eeeee eaeoee eeeee eeeee eeeeeeeeeoee eeeee e eeeeeeGeeeeG$4 eeeeee eeeeeeeee seose seeee osee eseee eeese ee e eesesese9ee seseoeeeeeee eseeeoeoseeo eseeo e**ee oesesos 8mseeseene eGeeeeeoeeo eoeeee44eeee eeeeeeeeeeee eseoeseosese eeeeeeeeeeee eseceseoesesseeme 1 eeeeeeG$eeo eeeoeeeeeee eeoeeeeoeeeo eD eseeetoe eeeeeoeeesee e4eeeeeeeeeea4eee eeeoeeaeeee eeeeeeeeeoe eeeee eeeeee eoeeeeeeeee e4eeeeeeee eetceeteeeceteese eeeeeeee4eee eeeeeeeeeeee eeeeeeeeeee eeeeeeoeeeee e#89eeeeeeeeeeeJe eeeeeeeeoeoeeee4 eeeeeeeeeeee eeee4eeeeeee eeeeseeoeee oeeeeeeeeeeo eeoeeeeeseee eteeeseeeeeesecee i eeeoeeeeeee4 4eeeee4eeeoe eeeeaeeeeeee eeeeeeee4eeO eeeeeeoeeeee eeeoeeoeeGeo e4eoeeeeeeee4eeeo e4eeeeeeeeee 98eeeeeeeeee eoeeeeeeeeee eoeee49eeeeeeeeee

• Oeeeeoeeeee eeeeeeeeeee e#9eeeee4ee eeeoeeeeeeee 980eoeoeeeetseete '

eeeoe#*eeeeeoeeee eoeeeee44eeoeeee e ee ee eee ee ee eeeeoeeeeeeeeee e e e ee s

e e e 9 .e .e e. .e .e .e eeee$eeeee eeeeeeeeee ee e ee eeeeee eeeeees eeeeee. $eee.oe4eeee e ... eee4eeeeee oeeeeeo eeeeeee eeeeee eeeeeeeeee I eee estee e e e. eeeeese eeeeeee seteee eeeeeeeoee i eeeeee esettes 6000 eeeeeo seesese seeeees 00 et..e. s e e .e .e .e eee eee.e e e e i e eeeeoee oeeeeee seesto seseeeeeee eeeeeeeeeeeo eeeeee eeoeeeeeeeeea e eeeeeeeoeoee-ee eeoeee eeeeeeeeoeee ee eeeeeeoeo e I eeeee4 e4ee 0040 eeeeeeee 4eeeOe oeee eeee eeeeeeeeeeeee eeeeeeeeeeeo toeeeeeeaeeeeeee ee99Seeeeesceoge 4

l ee44ee eeee eeeeee eeeeeGe eeee eeeee4eeeeeo eeeeee eeeeeeseseee ee o eeeoee eee eeoeee eeoee seeeeo eeeeeat a I 9eeeeeeeeeSe e996eeeeeeee e$4eeee eee* 9996eo eeeeese 1 eeoeeeeeeeee eoeeeeoeeeee 4toeee eeeee eGeee seesteee*6940ete eeeeee.

e. e. s...e e .e .eee .e e e e o eeteeeeoeeee e eeeee.eee eseeoeeeeeee eeeeee4ee490404e e ee ee444eseeeee oeeeeeto4eee e .. ...eeee s t e e s e e .e e..eee s eeeeeeeeeeee oete..eSe.ee o.se .

9000eseteoee

..eee eto s e e e e .e p..e.

.eeee eee49eeeeGee

.e e e s e o s e.e 3e s e a e e o g e e t o t e .e eGe. e e eeeeee e. e e e e s .e. 1 eoee4ececeee e4644eoeeeeei eeeoeee4496e eeteeeeeeeeo seeseeeeeetteese oee#9eeoeeee eeeeeeeeeoeei 00eeeeteeeee eee40eeeseee eceSeeeeeeseeee 9eee4eeeeeeeeeee eeeee8eoeeee e4eeeeeeeeeoeeeo eeeeeeeeeeee eeeeeeeeeeoe seSGGeeeeetseene

• • eeeeeeeeeeeee oeeeeeeeee eeeee eeeeeeeeeeeeeeee eeoeeeeeeeeoeoe eee9eeeeeeeeoee eeee00eeeeeeoee ee6eeeeeeeeeee eeea eeeoe eee a e eeeeee eeeeee eeeoeee e e e e eae o e4e9e*

See4ee 9etee 4esee eeseeee eeeD9e oeeeee eeeeeeoeeeee eto9944eee4 toesee eeteee eeeeeee eeee9ee eeeeoe teetto b ette4* e s e o e t e e e e, ee 9..

ee.eGe eseeee.

t e.ee oee e .e9e .e 4 .

44eeee Seeeccee 4etece l oee eeeeee96eeoe eeeeeeoe eee

. .e.ee e4ee$eeeeeeee e e .4 eeeeeeee4eoee

.e. AD.e .e.

e e e .eeo eeeeoeeeeeee, se e e

e .e .e 6 t ee.e t eS .

e.eeOe

.. e i

eeeee9eeeeeee eeeo eeeeee eeee4$eeeeee 4

e4e9eeee4ee4eeeeee eeeoeee eee eeoe eeeeeeoeee  % 4e eea eaeee ee eo9eee eeee4ee eeeee eee eeoeeee ees.4d) ee44ee egeeee ,

e4ee eOGG.ee eeso$

eeee G e e t .e .e e.

eeeeee eoeeoe Oesesee ee e seeeGee e e . e .ee e ee o e D A D-9eeeee

% eeeoee e.e..g eoe.g. ee eeeee0 seeses u5 oseeeee e4eoee, eseeeee eeeeee eeeee9 ee e seeeeee 4eeeeo.. eeoeceeeee eoee4eeeeoee oeeoeeoeeoeeee seeeece.e eoee oeoeoee e ee -

eeseeeeeee We, e oeeeeeeoeoeeeeoe ce to Se S .e ee et g .e g e.ee.d s e t e er see..e..

ee ee so ee .sete e . c o .e.e.

e... t s . eS. e e ee eeOee eeeeeeeee ea oe ee ee G4 e e44e e eeee e o ee ea eeee e e e( eeee4 eee4 ee eeee eoee e eeee e4 eeeeeeeeee e ee eee ee4e eece 999e eeeeG$e0ee eeee 4eeee 'g 9ee eeee eeeee eteese9eee

1. 34 e*8e 80ee eee9 9ee$ eeeeoeeee eee e eeee oese seee eeGe99 etetoe 4e eG eeeeoeeeee e eeo i #4eeeeeee eeeeeeeee 4eeeeeeeee oeeeeeee eeGeeeeeee eo 0eeee eeeeeeeeeed eeeeeeeeeeeeq ea eeeeee 9eeee ee eeeeeeee

)e o te e eo eeeeee eeeeeeee eee oeee e ete ee e e ee eee eeee e4ee eeee eee 8eee eeee ee$

eeeee$eoee eeeeee$eee e

e"e eeee e9ee e4ee g eee eeee ee4 6eoe eeeeee4eee ee9 ee$e eoe Geee 9eee40ee46 eee e eeee ee eeee eee0 eeee4e eeeeee 9e eeee ee 44eee##$e eeeoeee e eeeeeeeeee eee98eeeee eeeee*$e69 eeeeeeeee ee eeeee 6eeeoeeeee eeeeeeeoee e eeoee9 eeeeeG ea

$ EGG $e94

}4 w

eoe ee69e eeeee4e eoe e 4ee eee $ee9 8e*e oeeoeeee* **eeeeee eee oeoe eGe $e oe eeeS$e4ee0

• eeS eeoe eee eeeS eeoee9eeee eee 9eoe oeee eeee eeeeeeeeee y%

eee eeee eeee eee$ eeee$eeoe$

eee e eeee 4eee eoee eeeee9 ee e eeeeaeeee eeee ee eee ea e4ee eeeeeo ee eeeeeoeee oeeeeee=ee eeeeeeeeea eeeeo eeeeeeeeeeq e eeeeee )e e e49eeeeo eeeG$eee eeoee .e eeeeeee e

e eeeeeO9eeee1 9eeee eeeeeeee eee oe e eee e e G .e eeee ee eee 44eee oe eo9 eeee eeee eeee eeee40eeee 44e 49e4 eeee 04e eee$

4eeo eeeeeeeeea eeeeeeeeoe eee oeoe eeee eee eeee e499 eee eeee eoeeeeeeee eeeeee ee o e4e e eeee eeee oeeeeeeee eO9e ** eG$eeee9*e e 8 e eee e oeee eeeoeeeee ee eeee.

4eee eeeeeeoe=e e

eeeeee eeeeee oe ee e o ee eoe4e eeeeeeee eo eeeeeeoe ee ee ee oeeeee4 eee eee q e e eee oeee eee eoee eeo e oeoe o eee eeee eee .oOeeeeo eeeoeeoeo$ **

e ee eeee eeee400ee$

eee eeoO eee eeee ee4eeeeoe4 eee eeee e$eee eee eeee O90e $ee9e4eeGe oeeeee e oee 4 eee8 e99e eeee eeee eeeeee ee

• • eeeeoeoee eeee99e$ee e eeee ea eeeee e eeeeee ee

• eeeee9e oee eeeo e eeee9eeee e49eeee4941 0eeoe8 ee oe

_ ". e eeeeee eeeeeeeeeo ee ee eeeeeee9 eee e os een eeeeee e i s eeeeeese etet****

o e. ee e e o Seeoeee esse see eeeeee 4eeoee oeeeeee e8eeeee oee994 eeoee.ee Geeeoe seeeeeeeese eeeoeeeeee eee.ee eeeeeee eeeeee eeeeee eeeeeaee eoeeeeee eoeeee9$ ee9eeee eoeeee eeee$ ee tee 99eoe se**9ee$ 0 4eee eeeeeoe e... e e e o e e e e .e e $-

u eee449 eeoe4e ee e e e e e o e e. .

e eeeeoeee. e. oe ee e

is e e e e e. e. e ee e w e eeeel 39ee9 y9eee# .. .

eoee  %

eee8ee i eee i..

eeee eee94e eeeo.l

.i eeee

$$ee 4eeee eeeeg Geeeee eeeeeo e6eeeee$ee eeeeeoeee et 0eee eeee eeeeee e

eeee e e eliF ee4Se eeeee eeeee 9989ee eeoeee e$

• e e e o e' e4eeee oeee4e 4e e e 4 e e e. . . . . . .e e e90e.e.....e.

e e eoeee....ea e .

e e . $ e e e e e. . . . . . .e oeeeee eeeeee e

oe ee...

e e .e.

e e. e e e e e e . e...e. Se eee e.e...e e ee e..e.. .e ...e.e.e eeeee e e e..eee e e o e e 61 .e e D.eeee..e .e .e 15.e .e eeeo e  : .e e e...,. o. eeee61 . Oeoeeee.ee e.Oe.e.

.....eeet.e eoeeeoeeeee eee.. O e e e . 1e e e e e e e e e e e 61

......e..

.e . .

.Oe.40009 e..e.

e..Oe.oeee

e. eee e

e e e .e .s e .e e

O.

e e.e.ee....

9e 6.

e

..e.

e.

e

.e ee.p e e

O

.e .e .eeee o e ..e e .e e e e e 00

.e.e..e

.60 O e

.e ..e.. .

.see e

. e e

.e e e O e t .. O e..e.ee

. e .e .9 eo

.. . .o e.e .ee.e.e.e.e.e.s

e. ..e

.e.

.e.O o..e 4.e ee...-

..ee... . 40

.e . .e

.ette.4..ee.6 ee e. 0. 0 o se

.e .ee.e...ee..e..

e eet ese.

e t . e .t .s .

.e9 000 00 eeee...

96e etes. eet.e

..e. .e ee te e.g.

ee.. .

...e e S e 0 e.eee....

S e.ee..ee e.6 . 90 000s.0 ee.e..e . .e es ee .t e eese

.t. .e940e ts .e....

.eep O e...e..eeece.e

..e 49.es..e tete e.9... ee oesteet eg

..ese...ee.eee

.eeet . ee 0

t e..e

. e e e s . e.e e .e e e .c .e eet

.o.eee .. .ee e.os .ct .ee ..c. ..eS.e e . 9 0 0 0

. 4. . .000e. 6

. e e ..ee..ee .e0

e. 00 09e .

...... .e e .0....

G 00 e

. . O... e ..e..eeee...

e e e...e.es ee e

.e.. 4.ee.eeee e

0e96.

e.

. 0e.eee e. . . oe .e.et .. .n e . . . e e .t e G. e o . . .

9 0 et e . .eet .o e . .eee.

..e ee

.e e....e. e.4 0.000.00ee00 00 0.

9....O m

V 50 CM/S Figure 403: Flow and Temperature Distributions at 6500s

I i

I

. i i

t t

l

i. .i 642 i. 656 .: i643 i. i 627

i. i. .: 620 629 638  : :

6  : : :

:  : :  : ..  : :: 6:17:  : : : 610 i

5 l . : :  :  : :  : : : :

• l  :.  : : : . .: : - . ....

_ a _ __ _

( i  ! 67 I 670 669 i ! 667 . ! . 620 i

! i. b. 4 658  ! i..!.

I. 645 . .

i 1633

. : 615

!i i i :M iii: . i i  !  !  ! ii i !

i.  :

i  :  :-  :  : : : :  :  : : : :  : :  :  :<  :  :  :  :  :

-a,:: ::  :

:  : :  : :s  :-

!. i, .

576 i i

.i 667 i i ! 664

.. i !. 667 i !. !. ! 662 i ! 656 i ! 560 i! 644 i 1635. i ! ! i p.wr t

i 80 i ~ 669 ii 5 665 i in 66 . . ,- om i < 57

. . i.  !. c653 i !. . 643 .

1633. i. !. !. 615

\

l ,

-  : N i  : : 4 i: :,

\

..t  :  : t  : :  : : :

.. ..:W.

!: :  !: : i  : ii: i i i !: :

i 77 9 i 76 i i 21 ii: 620

:  :  : :  : : W.

1  :

1  :

i  :: ::: :61.4  : :

!. 43  !. 3  !. 647 i! 21

.. i: !: ,:, 620

.5

.. !622

!. .! .!622

!!622

.i..! 621 .i .i : 620

!!622  ::! 622  !!!B22  !!!!!619  ! ! ! ! ! B19

!::::i :ii:: : i lii:::::1  !!::: : :i :i!.!:::i

iS18

. ::.::  :::::::iS19

::i:::i!619

i:::i' 17  !::::::"!W ::iS18

4

4

.  ::::::: . S. .

" : R17

!!!!.!615  !!::! 15

.....  !!.!!! 15

.. .  !!!:!:.:615

~

i.::.!:m

..  !.!.616 iiii

. . . ih 5 : tilii 15. iiiii .15; . ....

ist:iii 615. i:!!!!!iiisti.

615 - :. . :615 : ..:::::615 ...:...::

. :.: . ::615 . . . . . :: ::614

!!!!:!!!!61.3  :!:  :!!!!613 i:!::!!!!613 :!: !:!::612 :::!:i!!!!611

.::::: g::::::::  :: ..::::y::::::

. : 4.::::::-  :: ..::k ::: :: .::: .:.*:: ::::.

V 50 CM/S I

Figure 4.4

Flow and Te::iperature Distributions at *1500s I

I

I j

i l

I 2500 , , , , , , , , ,

2250- -

1

' f,:

.i 2000- Level 5 ,'

p I

. . . . . . . . . . Le v el 6 l Level 7 l e 1750 - / -

g - Level 8 //

- 1 2 Level 9

9

$ 1500 -

I' -

E R O

3 1250 -  ! -

1000- -

750- -

500 , , , , , , , , ,

6500 7000 7500 8000 8500 9000 9500 10000 10500 11000 11500

TIME (s)

Figure 4.6: Cladding Wall Temperatures in Ring 1

~46-

i l

\

l l

2500 , , , , , , , , ,

a 2250 - -

f!

j

, 2000- Level 5 -

I

.......... Level 6 (

Level 7 I

, o UM- . -

h5 - Level 8 .

il ~ f1

' O LeveI9 '

/

1500 - .

..Nf l.

h 1250- ,f }

,*,.~-

~

1000- _

r.

750 - -

500 , , , , , , , , ,

6500 7000 7500 8000 8500 9000 9500 10000 10500 11000 11500 l TIME (s) l Figure 4.7: Cladding Wall Temperatures in Ring 2

l l

l l

2500 , , , , . . . i i 2250-f ~I 2000- Level 5

.......... Level 6 be f750 -

Level 7 -

h5 - Level 8 ,

j Level 9 , ',-i f2. 1500- ,- -

E .. .

@ s/

3

" 1250 - ,/ -

Y. / l 1000- -

..?

750- -

J l l

500 i , , , i . . . .

6500 7000 7500 8000 8500 9000 9500 10000 10500 11000 11500 TIME (s)

Figure 4.8: Cladding Wall Temperatures in Ring 3

.: l 1032  ! ! 1111 i !!O26 i i921 i i i j774 i ! ,9 4.4 i l }0,14 l :!938, i: i770  : : :

i i i j 680.

:::s.:  : :

: m: : -2 fr:: :: :: .: ---

ll l29

! ! 1271  :

!: l1260 l l1240 i i i 782

!! l .8 l l 1199 42 l l962 i i i 749 i :!

! ! W:

: :NN  :::

! !:: :! !::  ! i si !!

i

! !o  :  :

• b.:

i 1313  ! : '!1227 l l 1234 ! ! ! ! 1061 1232 i!lN1231 ! ;117E i ! .112S  !! !1Q31 ll!?.S97

l4 .

!li 335 ! ! ! !217 !1  !! 1219  : : - m l! 226:

.&  ! t. T : 1.15.7 l* l

.  :.l1216 1116  !!J960 i. i. 'i. 7. 64

: : 4 s:t )

ll .l"l .::

: a, J: l: \ l 3l9l l/ i  : :

i ! ! 834 i 1789

! 410  !' !1.a38

!0424 i M.381  :

i 1p$.2 i:i :B.14.  : :

i: i :4:632  : :

• ]396:  ! 359  ! 1146 i i798 i i i 629

!!ils  ! 8,

! J:.5.

: 3: i:i :5.18.  : :

i: i:4:630  : :

4- :  : : : :

! 353  ! $ 329  ! !) 253 i !806 i i ! 628

! 75 i ! :1 51:  ! !O262 i i B23.

i: i 4:630  : :

!1283

!t286 !p258 i iB11 i i ! 623

! l'307 i 31.0 :i y26  :.9 i: i :824.  : :

!: !:4: 625  : :

i1167  : 1135  : 1107 i !797 i i ! 624 M83 D151

! i $139 i: !?79 i i :4:626  : -

!!!!!i622 i:!!!E22 !!!!! 623 i!/59 i i y 623 iiiii45ES. iiiiljs25. iiiiiipa4 __

ifi7R j j j $26

!!!!i  !!!!!620  !!!!!  !!!!!!!620 iiiii!620

. ...!.!.ilb.521 21. . . . . . iiii.!.T.i.20. . .. . .!!!!.!.520 62. 0. . .. . iiii. i. i.B. 2. 0. i i i 16! l61.5

618  :::::::::619 ::::::::: . . . . . . . . 19 . . . . . . . 619 . ::::::::::

.::::::::1 . . . . . . . . . 610

$.1.5.::  ::::::::::61.7

617::

$:: .:::  ::::::::.616:

::: : .:::::: W  ::::::::::::::5.:07

.::::::!?::::::

:.::::::q:::::::

M:::  ::  ::::::::::::::::.

w 50 CM/S Figure 4.9: Flow and Temperature Distributions at 9525s

1 I

f l

l 1

l l

2000 ,

i i i e i i i i j

1800- s -

6 .gN ..

\

. ..y ,

1600- l '

l. -

Ring 1 ,

-f e 400- . . . . . . . . . . Rin9 2 i:

w r,* . ,

3 g

Ring 3 ij

, e

o. 1200 -

E L

,e ...-

, *. - ,)

y 0 1000- I m y .

e 3

800- _

600 - -

l 400

• i ,

i 6000 7000 8000 9000 10$00 11dOO 12$00 13$00 udOO 15d00 16000 l

TIME (s) e d

Figure 4.10: Cisdding Wall Temperatures in Level 4

~50-

./ / .1430 .

e i .1356

. /./1247

1. ./ .1106' .f'., ' ,./928 . . 74g

:  :  : :  : : 9.18:  :

1.24.4  :  ::1741 .:  : .-

:  : ...i.< .-

.- -.- .- l

p::

: .6.

p ::  :

Sl* :: .  :

l l.Al: . ~: .

-: :: l.*

1c ::

/ :: :: ::  :: :: :: :: - : : : ,

- -  ; i

< i g747  :

./ :/: :1683 / :-./ ./1007

:. 180  :. :  : : 1 1 g37

:: 1.33,'

:  : : .1650  : 665 r  :

I.

.-l s..

.- w :  :

.:/s ./

./ - 1841 1690 ././1646 ./ ./ 1668 .i ./ ./

/ /

: 1700 . .:

/://./ :'\'1646 :i :/.:15p4

• :/ .'1488'

: : .:/: /: ?./1387 12.12.

~

. / l874 ' / /1688 / '/ 1644 // 1641 f

/ '.1700 . ./ .l 16.98 .

/././16.00 . .

/(137) './ ?' uu.): 100 .5

/ .fy* l l 'd % ' ./ .y'

.i K_. \/ . .

~/ i 19$9 '. / *

//< / 16 ' /12$1  !!

jif;: ;;

/ / (^:

.. ll: .. . ., ! l/ ;

.48 / /T201

.- <  !:isl!778 744

/

l  ! .' . i .

h; .

'lHil 1H11 lH11  :::iin l!m 111111111111Hill.

/1346

' / 1396 /

'@ !!iM

' /1180 i i ': 708

' .7353

/ Pip 5

/ 1038 -

i:. !:: :7.06

/ / 992 / / S 38

/ / 884 / /./J008 ij !4 686

/ fi463. //78.49 / / I4 97 , IR D j j $87

! ! 619 ' ! ! $42  ! ! !$43 i ! !4 68 i ! ! 632

!! $25  ! ! 6. 2,4.  ! ! !6.24.  ! ! ! 631 620 i l '.!

...... 818 ........

...::::: 618 .......

.: 618 . . . . . .: .

.: 618  :.::::: ::513 -

:::::::::617 ::::

. :: 0  :::::::::617 :o  :::

- ::::::5: . .:::

. :t:: ::::::.15

510

61.8

:: ::: ::.: ::g:::::::::: :::::::

..: : ::  :: :::.. -  :::::::g ..:.::::  :  :.:::::::::::: ..:::::+:::::::-

w 50 CM/S Figure 4.11: Flew and Temperature Distributions at 11000s

l

)

l l

i i

i

\

i

' i 4 . 1 i i .

\

\

7- ,

. \

M..

t .

Cor8 5- .

. ~ \

' ,,,,. . i 1

e i.

A- '

4 5-

• 5 .

\.' s

~T o

....-~~~'*% ..

1- s, .

0 . . . .

9000 m 3000 3000 3D000 1I000 9000 9000 14000 9000 Thrie @ i I

l l

l Figure 4.12: Water Level in Core and Downcomer

. --- - - , 7, - , , , . - , _ , - - , - . - . - . _ _ _ _ , -- ,, , - , - . . , - - . - _ . . . . , _ , ,- , . ., - -- -,-_

t l

1 l

400 , , ,

350-300-2 6 250-Q 0

2 _

e 200-e CD y 15 0 -

10 0 - I J

50-0 , , ,

9000 10000 11000 12000 13000 14000 15000 Time (s)

' Figure 4.13: Total Hydrogen Mass Produced l

- _ - _ - , . - _ , - - . .1

20 . , , , i . . . .

17.5- -

15 - -

Total 9n. 12.5 - -

---

Hydrogen i b .k.

Ei 10 - I -

n m

o .

ch l:l 7.5 - ll -

,l '

ll l .

, l t~

5-  ; : l'. . .'. .

l>: l;.;.

2.5 -

,s

'l l ,.

f f

'j I ' l.

l f  ;,.........-...,., ....... ,

. e 0 , , , , , , , , ,

6000 7000 8000 9000 10000 11000 12000 13000 14000 15000 16000 Time (s)

Figure 4.14: Total Pressure and Hydrogen Partial Pressure at Top of Core

~54-

1 1

l l

l

!!151 . :1419  !!269 i1131

i 1219

ll.1361

'1;457 -

!!426 1354 !963 i* i .  : : 9:

!*l . i w. ..

..: i. -

. . .i :.: -.:

IN . . !!510  : : !1505 . 1492 i ll50 IS ..' : .'l.507,,,llllp76 .# ;1395 1104 fi a-

/

: i

: . :w

. : ni.

.i k:i -

\

it , . . : \

. lbz3 . . 1494 i! 1481 . i 1488l. : ? .1354

...!* )..

llfl.. .l.!1W !: .%

i

* .0 '

1 1528 !11495 ~; ! id475 - m.a

! .*! 2/I 63-

. 0. .  :/. 1. 11383  : I:1 1,211 i

~

s  : i! A * :5.  :: T l  : .* -: .k .

~ -

i=.N!1542 i  : . il053 l?:152$ .. :l ;!/ J' t . ,l l: l,!311 m  ; !'l m  !

l j ii i ii s i i i ,

. 1388 . 1032

' i c !l Iijo[l l'..l  ! l l 1425 Y .988

>v .i ~ ~ ,

o -

II . l l 3 titilit , ~ i l *'

11, I t 111111 m , .'m

!!6/u

. Bos [lM r=a  : :T308

40 f ! 949

. l'i SOS

'.997 . l #52 . Ib'IS  : l,1105 i 891

1100  : 1091.  !!36

• WE iS46

!822 jQ305 805 l 714 !770 M. ' !%1 . *790 l - 7 29 'g!705 '

1 732

~

715 683 695 717 l i

.g623 23 ,613 ,!604 l621

~

I I

50 Ct1/S '

l 1

l Figure 4.15: Flow and Temperature Distributions Prior to Core Slump 1

l

_-.-.,_._.____s

i l

. l l

. i 3000 , , , ,

m y 2500 - -

v ..

Cz.1 -

a: . .. -

p 2000 - - - -

E-

< ~

c:

w 1500 - -

CL  :

2 -

N -

2D MEtPROG E

1000 - -

. . . iD uEtpRoc -

500 '

5000 7000 9000 11000 13000 15000  ;

TIME (s) l l

l l

l i

l l

Figure 5.1: comparison of Maximum core Temperatures

I l

2200 , , ,

2000 -

m k.

v 1800 -

00

% 1600 - -

D E-

< 1400 -

c: . ....

W 1200 -

CL .......,

2 ,'

w 1000 -

2D MELPROG

~

F . ID MELPROC 800 -

y ,....-

._e_

i 600 '

7000 9000 11000 13000 15000

";- TIME (s) i Figure 5.2: Comparison'of Surface Temperature in Upper Plenum

400 , , , ,

350 - -

l 300 - -

1 m of) 250 M ""

w y) 200 - -

U2 4 150 -

2 100 -

2D MELPROG

~

1D MELPROG 50 -

l -

0 5000 7000 9000 11000 13000 15000 TIME (s)

I Figure 5.3: Comparison of Hydrogen Produced

. - - - . - . . _ - - - - .. _ - - _ - - . .- ..~ - _ _ - - - . _-

o.

R - .

o* . .

S. ....................'....................j....................l............

m l s . l l c,o E  ! l RELOCATION l c

g g. .. .. . . . . . . . ..... ..y ..l .. . .. . .. . . . .. ..t ...B . ................y..........

..y EGINS l g . .

m

.! .! l

. CORE

. .' SLUMP ct

o. ....j.... l

......... ............  !,......____._ l o.

t 9 1 t '

s.o e.s I

I it.o 13.s 1s.0 TIME (S) # 10' I

Figure 5.4: Steam Mass Flow Rate at Core Mid-Plane in Ring 1 1

I

O.

I a

o

• E- .......,............;.................. ,;....................'..........

- . . RELOCATION E  !. BEGINS l

+

cn

.x. .

g p. ....... . ... ...... y....................

..................y..........

d .

m 1 . l l M ,

e ,

g 7 ' b i.

~

o .

l l

r- r

f. ....... ,...........>.............

\

. . CORE

. SLUMP o . .

F y s.o e.s it.o T

is.s to.o TIME IS) a10' i

i i

1 i

l l

l l Figure 5.5 Hass Flow at Top of Core in Ring 1 l

! i t

\

, i I

i r

i l

i l

i

)

R o

g. ....................k....................)...................k........... .........

Q l l RELOCATION CD w . l.

. BEGINS o l g9 ....................,...................yl..  !

...............y.........., .........

d  :

m .

I l o .

l j t ._ 1^

__,21 _,t_ _ ___'r ........

. . . i l l l CORE I I I SLUMP

o. .

y '

5 i

y 8.0 J.5 l1.0 I t 13.5 18.0 TINC (S) =10' Figure 5.6: Steam Mass Flow Rate at Core Bottom in Ring 1 450 i i i

• 400 -

20 MEtenoc - 2200K CASE .

2D MELPROG - 2500K CASE -

350 -

^ot>300 -

M v 250 -

m M 200 - .

l 4 150 -

l -

100 -

l I

50 - -

0 .

7000 8000 9000 10000 11000 12000 '

TIME (s)

Figure 5.7: Comparison of Hydrogen Mass Generated in Relocation Temperature Study i

. I i

16.4 -

Outlet Pressure in

_ PORY Cyces E

m 5 16.3.

• Current time f

EL 16.0 I I I 9425 9426 9427 9428 9420 Time (s)

! 1020 4 i1095 i1018 i 111 . 775 927  ! 922

9 91 li762 lr l! 675

.i l

1l, . e T

le t  !

{*i l  !  !

o /lIi <

i l

f I i 1213 1233 1215 776 l

1 3165 M

2 A 740 l i, ,

w_

<jj-

, , l

.;L l

l279 1 \llte 1100 i 1038 lfjll34 i

!~

1185 II4? I'M'Iloo i 1= ii1lm i "i 1195 l293 l, 1152 ill__ < as lilli I!P i

kiI88 YII3  ! IN '

M !l N 11l11111! e IWillii 7;* 1i-i'8 3 dr314 il lDe 798; i i 83$

f ,8 0 @3 E 3

Y. i

• 63,3 ji 351 312  !!  !!O 750 j i 529 l f ,3 3 ; %I Y l ) 03 315 - 135 217 757 i 628

. , l ' ,:232 W, *i 527 ig3I lii250

Trass i

'J 22s ig.,

i [i !i *i

, 523 s24 J232 ,

Ql46 ' ell 22  ! ' 113 - : 786 524 W

!1154 ftl26 '

l!?65 '

• 624 t 623 !tji!.623 623 *

i:} l 4626' ' i I 4 626 i t ;ll1626 .Ir t 251 13

, t 623 624

! l!  !

6fgf

618 ,619 .i  ! 619 619 611

,616 .r. , ,j617 ' l,: j 617, ,,

g 616 607 4

.i.

50 CM/S Figure 5.8a: Flow and Temperature Distributions During PORV Cycle

se.s -

outlet pressure en

, posty cyces a.

5.

e g is.: ,

E a.

cweens um.

se.e t 8 e 9425 9426 9427 9428 6429 Time N

. unos inoro es3 me rse s2s ps2 s2o 7se. s7s d -. .

, i l I  ! l 1223 ji'$6 IED 754 3 l 3165p # 132

$2 737 ii mi lisz ii44

] 17e r aiu i lies i)

sia l' h.60s l !! an$i6.n  ! tai := . i :%:- igik . h  ! ! W- ! i 'yi!  ;!': i  !!'l 'l ;8  :!;8:' ,; I 'll '*ih ! 50 CN/S Pigure 5.0b: Flow snd Temperature Distributions During PORY Cycle 16.4 - l Outlet Presome 1st l PORY Cycle j 9 ' o. E c.,-i e l g 14.3 E a. 14.0 8 i 0428 9434 9427 9433 343g hM 1010 il000 i1535 795 w y us mi o a n . ,

• I o , ,,i

! ana . szsa me sang no i y y/l .--Tla m o o;  %. 3 i! l1 i , I l291 lago ' '8100 tim 18144 , D ... . . 'M 1 814 s ' l" YP n m Nsssa . ita g .<n ! L8S 18PD . A 884 I l1752 i . 111111li i Ill ll lP lli II 11, ~P i i= ! = 3m m en '1 5 WlLsza  ? W  % 888 sas' taan sar l5 gan 7ee 1 l 54 P!' ~1214 H 1Mi 732 ' 635 0@45 .,I Taja an, sas .b hh h

• h 22

!% M.:  % *f $ liliya 113 = p:!:  ! ! <n i !I= Ililie! lili= I!llR 1lll1Cl illH !!! si!!a1ll P' ,igP ; U in n '"!lliTr!!r  ;,jud;!

il Milllik,jh@yl!ll n 3l=k!!!

50 CN/S Figure 5.8c Flow and Temperature Distributions During PORV Cycle 12 0 , , , , , , , , 30 0 . 6 1 80- - 6 60- ' - 40 . 20- - O- ~^ ~ ~ " ~ - - ^ -~ "+M - - I -20 , , , , , , , , 6000 7000 8000 9000 10000 11000 12000 13000 MOOO 15000 TIME (s) i I l i i 1 Figure 5.9 Outlet Flow Rate I . I k i i i 1800 , , , , , , , , , i t l 1600- - i ! I 8 I 1200- - 1000- - 800- - l 600 , , , , , , , , , '. 4000 7000 8000 9000 10000 11000 12000 13000 14000 15000 4000 i Tim. (s) 1 1 I Figure 5.1 Os vessel outlet Vapor Temperature i ~67-l ROyay DRAFT 1300 , , , , , , , , , , . u.g:.-y = 5 1, 6 -T (' 2.q.t; ~ ,,**, .f 110 0 - *Y ./ q . .' 9 v Possible Philure of * / Core Slumps e 1000- *z u i .r - j Hot Leg Connection //0 t Q. 900-ll // ~ /. Vessel Fhils W h l,1l ] a00- ,f 70o . ,, inner Node ,. [. . - Outer Node 600- . 500- , , i i i . . . . 6000 7000 8000 9000 10000 11000 12000 13000 M000 15000 16000 TIME (s) l Figure 5.11: Temperatures in Hot Leg i i I .-. _ _ . - - _ _ . . _ , _ .. _ _ . _ . - . , . . - - . _ _ ~ . ._ ,, - - - o J TABLE 4.2 I COMPARISON OF MELPROG AND TRAC INITIAL TMLB' EVENT SEQUENCES Time (s)

MELPROG TRAC ,

Boiling Begins 6500 6500 Core Uncovered 7070 7080

First Dryout 7240 7180 l

! Core Steam Filled 8350 8100 H2 Generation Begins 9280 8900 i I l \ I i t 1 ) TABLE 4.3 i STATE OF DEBRIS AT VESSEL FAILURE % MOLTEN / MASS (kg) LIQUEFIED t j UO2 96000 14 , ! Zr 9600 100  ! i Zr02 9250 0 Steel 19300 78 Control Rod 2850 100  : TOTAL 137000 30 1 Tavo - 2460 K 3 4 i i

l i ,

9P'

• TABLE 5.1 DEBRIS STATE AT VESSEL FAILURE MARCH WELPROG-ID WELPROG-2D CASE 1 CASE 2 Time from Primary System Saturated to .

Vessel Failure (min)

• 90 115 105 157 Tave (K) 1870 2600 2120 2460 Total Mass (kg)  ? 117500 128160 137000 Molten Mass (%)  ? 34% 16% 30%

Zr Mass (kg) 6770 11400 6500 9600

• si. \ +* 2 9 -

Steel Mass (kg)  ? 900 10300 19300 Zr Reacted (%) 59% 31% 60% 40% Hydrogen Wass (kg) 430 230 440 300 Trelocate (K) 2550 2200 2500 2200 O t o TABLE 4.1 TMLB' EVENT SEQUENCE TIME (s) EVENT 0 Loss of offsite power, loss of feedwater 4170 Steam Generators Dry 6500 Incipient boiling, begin MELPROG calculation 7070 Core " uncovered" 8350 Core empty 9280 Hydrogen generation begins at top of core 9970-10145 Control rods fail in top 3 levels in Rings 1 & 2; ' Steam temperature > 1700 K 10156 Cladding begins to melt in Ring 1, Level 5; Cladding temperature > 2100 K 10181 Fuel rods disintegrate in Ring 1, Level 5; Cladding molten and temperature > 2200 K 10216-10221 Control rods fail in Ring 3, at top of core 10241-10303 Cladding melts and fuel disintegrates in center 2 rings, top 4 levels of core 10319-10377 Cladding melts and fuel disintegrates in top 4 levels of ring 3 10377-10403 Fuel rods in Rings 1 & 2 disintegrate in Level 2 10387 Upper cote plate melts in Ring 1 10808 Fuel rods in Ring 3, Level 2 disintegrate

11345-10260 " Thin" metal in upper plenum melts 11522 Control rods fail in Ring 1, Level 1 11680 Core baffle fails mechanically 11824 Core baffic begins to melt 12305 Control rods fail in Ring 2, Level 2 i

14877 Debris region crust fails, core slumps 14878 Level 1 fuel rods disintegrate 15371-15874 Lower support structures melt i 15928 Lower head fails, end MELPROG calculation l wru e --- II 3I4 ] ] no jfg*gg 3 3 11PPf f SUPPol:7 ' LATE 9,$ j4 1 2 3 12 )I 7.915

> 4 11 PBC contact too carvt covtes

(- : 10 VBC ePrie cost PL ATE 6'486 .527. e 9 6 117 F7 5 3a6 e 4 655 3.924 W###/#/ - Loel PiATf / *462 cUIUlWFTKTi t l 1.842 2 telI2MOWRif7ttre 1 1220 - l towirkrXB I 2 3 4 5 Figure 2.2: Vessel Model Used in MELPROG Calculation