ML20206H144
| ML20206H144 | |
| Person / Time | |
|---|---|
| Site: | Sequoyah, 05000000 |
| Issue date: | 09/02/1986 |
| From: | Bergeron K, Carroll D, Tills J JACK TILLS & ASSOCIATES, INC., SANDIA NATIONAL LABORATORIES |
| To: | Silberberg M NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| Shared Package | |
| ML20204G644 | List:
|
| References | |
| CON-FIN-A-1146, CON-FIN-A-1198, CON-FIN-A-1412, RTR-NUREG-1150 NUDOCS 8704150271 | |
| Download: ML20206H144 (22) | |
Text
.. .. . - -_ _
i < l i
) * [* 2 6 Sandia National Laboratories !
i Albuqueteue, New Mexico 87185 September 2, 1986 I
i l
Mr. M. Silberberg, Chief Accident Evaluation Branch !
Office of Nuclear Reactor Research '
U. S. Nuclear Regulatory Commission
- Washington, D. C. 20555 i
Dear Mel,
l Tim Lee requested a short time ago that we in the CONTAIN code project cpply the newly-developed Direct Containment Heating (DCH) models in the CONTAIN code to the sequoyah ice condenser plant in a sensitivity study
< for Appendix J of NUREG-1150. In view in of the high priority assigned by view of the rapidly disappearing NRC to these issue papers, and j
window of time before the deadline for NUREG-1150, we have devoted an intensive effort.since then to producing the letter report which we are i
now transmitting to you.
The principal results We of the think study these areresults givenarein Table 2 and extremely Figure 3and important of the letter report.
also quite interesting. It should be kept in mind, of course, that this study is different from the simpler analyses of DCH in Surry performed for the CLWG and reviewed in the current draft of Section J.5 of NUREG-1150; in particular, the calculations are not intended to be bounding.
.! In fact, all parameters in the calculation except the two sensitivity parameters and the amount of pre-existing hydrogen in containment (see to have values which were our best-estimates; the i below) were chosen guide for'many of these BMI-2104 Sequoyah TMLB' was our principal l
! choices. Thus, while the pressures Surry shown CLWG in Figure 3 are lower than for calculations, it should not be the worst cases in the concluded that the DCH problem is worse for Surry than for Sequoyah.
l Naturally, the details of these results are dependent on the choices uncertain parameters and assumptions about accident
- made for a number of qualitative conclusions emerge from progression. However, have rather broad applicability.
this study which we believe will This r First, it is clear that the dominant metal reaction is with steam.
is a consequence of the lack of oxygen in the environment of the debris l
particles early in their histories, and the high reaction rates.
Second, one cannot isolate used the DCH phenomenon from hydrogen combustion.
default combustion criteria, but thresholds, it is highly In this study, we likely that they are inappropriate. Lower combustion would probably not dramatically change the likelihood of however, However, one feature which will be containment failure for Sequoyah.
I l
8704150271 870408 i
1150 C PDR 2f _
s
. s M. Silberberg September 2, 1966 c
the location of the sharp censitive to the combustion threshold is i transition seen around 20% melt ejection in Figure 3.? This transition i is due to the onset of hydrogen combustion (i.e., no burns occur at melt j fractions below this point). The location of the transition is also '
offected by the presence of hydrogen in the containment just before vessel failure. In these calculations, no pre-existing hydrogen was assumed (the hypothesis being that independently-powered igniters or recombiners removed the hydrogen and corresponding amount of oxygen prior to vessel failure). We feel that our treatment of pre-existing Non-zero pre-existing hydrogen would shift the transition to the left.
hydrogen. here is somewhat non-conservative, but we made this choice in order to have a cleaner sensitivity study,(i.e., we are not interested at present in the problem of combined steamcspike by and hydrogen SARRP to beburns in a major the absence of DCH--a scenario demonstrated concern by itself.) /
A third conclusion is that the results are' auch less sensitive to assumptions about de-entrainmentdebris (or trapping) than has been previously de-entrainment time, 0.3 and 10 thought. The two choices of ceconds, are, in my mind, reasonable estimates of the upper and lower bounds of the plausible range for this parameter. But, as Figure 3 shows, the differences in peak pressure are not great an (though peak extremely temperatures show a greater sensitivity.) This is important result, and it has important implications for model development priorities, for experimental matrix designs, and for assessment of plant geometry effects. For example, it would suggest that the heuristic plant geometry categorization proposed by IDCOR may be less relevant than they suggest.
The study of DCH phenomena with CONTAIN will continue into the next fiscal year, of course, supported in part by the SASA and QUECLA projects. In fact, we have already developed and exercised an improved treatment of mass and heat transfer (to be described in a forthcoming paper at the ACS conference in Anaheim However,in October), for the Sequoyah and are developing calculations a 4 multi-size droplet field model.
presented here, we applied the version of the model which was presented The at the April 1986 meetings on DCH in Silver Spring and Bethesda.
reason for choosing this version is that the opportunity for peer review has taken place, and the code has been extensively exercised over the past four months. Also, it is the version of the model which was so successful in semi-blind post-test predictions of the DCH-1 experiments inclusion in a high-(cf. letter to T. Lee, June 4, 1986). For visibility document such as NUREG-1150, therefore, it seemed prudent to i use the older version, especially since our experience would not give significantly with the improved ;
to believe that they models leads us different results for the cases considered here.
In summary, the sensitivity study we have done in support of the NUREG-1150 issue paper on DCH must be considered only part of the story of the progress this rapidly evolving area, but it can serve an ongoing in j
, , ---- - . - - . . _ _ . , - _ . , - - - . . , , _ _ . ~ -
. . , - - _ _ _ _ . _ . .,.-_..-_.,..__.,._-.-_,,.__-_--__.n-...--.,-
l ~ (
l
/ Ssptemb3r 2, 1986 M. Silberberg
(
(1-sg important purpose in conveying the basics of our current understanding of the phenomena. We have a relatively high degree of confidence in these predictions (certainly more so than in the ultra-conservative CLwo calculation) and we feel that this work is suitable for inclusion in the issue paper.
Please let me know if you have any questions on this material.
Sincerely, J+
Kenneth D. Bergeron, Supervisor Containment Modelling Divsion j ..
" Direct Containment Heating Analysis with the Encl.: Letter report CONTAIN Computer Code" V
cc: w/ encl: h J. Mitchell, NRC/RES R. Meyer, NRC/RES P. Wood, NRC/RES T. Walker, NRC/RES F. Eltawila, NRC/NRR 11 6422 D. A. Powers 6422 W. W. Tarbell 6422 M. Pilch 6440 D. A. Dahlgren ,
6449 D. C. Williams j 6449 D. E., Carroll 6449 J. L. Tills 6449 K. E. Washington 6449 File 3.7 l
-s
^
~
.R ,
r g. :
- ("
't e*
s Attachment to Latter from K. D. Bergeron to M. Silberberg, September '2,1986 Direct Containment Beating Analysis with the CONTAIN Computer _ Code
- y_
K. D. Bergeron, D. E. Carroll, J. L. Tills
- K. E. Washington, and D. C. Williams i Containment Modeling Division 6449
[. Sandia National Laboratories
' ' Albuquerque, NM 87185 i
- 1. Introduction The phenomena of melt ejection, debris dispersal, c'nemical reactions and heat transfer between debris, water and gases which collectively contribute to the
! Direct Containment Heating (DCH) phenomenon are extremely difficult 'to predict with confidence using existing calculational tools. An ongoing NRC research program is underway to study these processes experimentally, and it is expected that a significantly improved understanding of DCH will' result from it is useful to analyze the these experiments. However, in the interim, problem with the best calculational _ tools available in order to assess the important uncertainties and to be able to interpret the' results of the experi-There are a ments as efficiently as possible when the data become available.
i number of phenomena involved in direct heating (e.g. , debris transport through complicated pathways) for which there are virtually no verified or verifiable models; tother ~ phenomena are better understood (e.g. heat and mass transfer I from a suspended droplet) . A model which treats the highly uncertain class of phenomena parametrically, and the better understood phenomena with best-estimate models is therefore a reasonable goal for an interim calculational tool. This report describes an Interim Direct F*oting Model (IDRM) which has been developed as an module of the CONTAIN corrWtt code, and which is j *This work supported by the United States Nuclear Regulatory Commission under FINS 1146,1198, and 1412 and performed at Sandia National Laboratories which is operated for the U.S. Department of Energy under Contract Number DE-AC04-
- 76DP00789. "
I
- J. L. Tills and Associates, Albuquerque, NM
, l
! l
_. . . _ ~ - - - . . .. . ._ ._ ___ .-_
f 2
intended to provide an improved understanding of the uncertainties in the analysis of accident sequences involving direct heating, and which can be of use in interpreting and guiding the DCH experiments. This model is then applied in a sensitivity study to the TM13' sequence at the Sequoyah ice condenser plant.
- 2. Backaroundt, Previous DCH Calculations The uncertain aspects of DCH limited early modeling efforts to relatively simple paremeter studies in which the mass and composition of the debris participatingarbitrary in the direct heating was simply assumed by the analyst.
Similarly, assumptions were made concerning the degree of complete.
ness of the oxidation of the metallic constituents of the debris with steam or oxygen in the atmosphere. In October, 1983, the direct heating question was raised at a meeting of the Containment Loads Working Group (CLWG) in reference to tHe Standard Problem No. 1 (SP-1), which was a THLB' sequence at a Zion-like plant. In the next CLWG meeting at Palo Alto in February, 1984, CONTAIN calculations of DCH for SP-1 and SP-2 were presented which Thewere based energies were on assuming varying amounts of energy transmitted to the gas.on the specified composition of the calculated in side calculations based debris and varying degrees of debris participation in the process.
One deficiency of this calculational approach was that the equilibration of heat between the debris and the gas was not explicitly taken into account, so that the sensitivity parameter was not the fraction of debris mass participating, but rather the fraction of the debris enthalpy transmitted to code, designated DHEAT, was developed which was the gas. For this a new simpler than CONTAIN in many respects (e.g., heat transfer to and conduction l
in heat sinks are neglected) but which explicitly equilibrated heat between debris and gas, making it impossible for example to calculate a situation in This code-which the gas was hotter than the debris at the end of the event.
was used extensively for the DCH parameter studies used in the CLVG final report, NUREG-1079, which was published in draft-for-review form in the summer of 1985. However, DHEAT was it limited in a number of ways; it was restricted to neglected heat transfer to sinks, it did not a , single control volume,the atmosphere during the chemical reactions, and it add gas to remove or the user to specify the fractions of participation of the debris, in required transfer and also in terms of oxidation. However, it had the terms of heatbeing extremely fast and well-suited to parameter studies.
The advantage of DHEAT code has been informally distributed to a number of participants in the Severe Accident Risk Reduction Program (SARRP) containment event A tree paperexpertswhich l
review panel to assist in assessing the DCH problem for SARRP.CONTAIN a includes both and Design (K. D. Bergeron and D. C. Williams, EED. 22, 153, 1985.)
Another model development effort which contributed to our understanding of DCH was due to M. Pilch, in support This of the SPIT and HIPS experiment series which model followed the trajectory of a single were fielded in 1984 and 1985. through the atmosphere under gravity. Unlike the debris droplet falling this calculation tracked the debris temperature and above, models described function of time, taking account of heat and mass transfer composition as a I
l l
1 i
limitations in the gas boundary layer surrounding the droplet. Atmosphere temperature and pressure was simultaneously tracked for a self-consistent calculation based on the simple picture of a field of droplets falling through ;
the atmosphere of a single volume containment. This model and a number of sensitivity studies using it are presented in NUREG/CR-4051, SAND 85-2435.
In February,1986, L. Baker of Argonne National Laboratories presented Thisawas paper an at the San Diego ANS/ ENS meeting on direct heating modeling.
extension of the Baker-Just model for metal droplet oxidation which had been developed for LMFBR applications in the 60's. Unlike other models discussed here, Baker's treatment explicitly considered liquid-side as well as gas-side diffusion limitation on chemical reaction rate of a droplet moving through a gas environment.
In another effort, Corradini and his colleagues and students at University of Wisconsin have developed computational models for heat transfer and chemicalA reactions of debris suspended in the atmosphere in a code called DIRHEAT.
variety of parameter studies with this code have been performed, and in addition UW personnel have used it in conjunction with a suite of containment analysis codes (HECTR, KEDICI-M1, and CORCON) which were developed by Sandia for the NRC. The direct heating models and calculations were documented in two UW reports, designated UWRSR-34 and UWRSR-35.
In November, 1985, the CONTAIN code project at Sandia National Laboratories was requested by the NRC to develop improved models to serve the purpose of interim issue resolution, and also to assist in guiding or interpreting the experimental data expected in early 1986 from the Surtsey facility. It was decided to improve the existing models by including mechanistic heat transfer and chemical reactions, and by allowing multiple volumes, but with debris transport from volume to volume controlled by relatively simple parametric models. In the following sections the resulting model is described. The features of the new model reflect the influence of the earlier models described above.
A number of model development efforts have taken place concerning melt ejection from the vessel and the cavity. These include work done at Sandia, Argonne and University of Wisconsin. However, since the focus of the IDRM work which is the subject of this report is on mass and heat transfer, these models will not be described here.
! 3. Eescriction of CONTAIN and the Interim Direct Heatine Model There are two aspects of the new calculational capabilities underThese discussion:
will be standard CONTAIN models and the new features of the IDHM. Also to be briefly discussed below, with no attempt at completeness.
discussed are verification calculations done to compare the new models against earlier calculational tools. Finally, we will flag a number of modeling uncertainties of which the reader should be aware in order to put calcula-tional results in the proper perspective.
l i
l l
l
4-3.1 Relevant CONTAIN Models CONTAIN is a system-level best-estimate containment analysis code specifically accident containment phenomena. Figure 1 designed for analysis of severe the phenomena modeled for a typical LVR problem.
illustrates schematically CONTAIN models can be found in the CONTAIN User's Manual, so Details of the only a brief description of the standard features which will be utilized in typical direct heating calculations will so if be the presented appropriate keywc*;d here. Alldoes of the not CONTAIN models are keyword-enabled, appear in the input deck, the model is not activated. Thus, for many direct heating calculations, it would be unnecessary to activate the CONTAIN models fission product decay and transport, debris-concrete for aerosol physics, The features which would interactions, and a number of other features.
ordinarily be used include the following:
- a. Intercell flow. CONTAIN uses a control volume approach to gas transport, treating each specified volume or cell as a well-mixed repository of the gases. Flow between cells occurs via an orifice flow correlation when a flow path of a given cross-sectional area and friction coefficient is specified between the cells. Arbitrary interconnections between cells are allowed. An arbitrary number of computational cells is allowed.
- b. Two-phase gas-steam-water thermodynamics. A realistic equation of state for two-phase water and a variety of non-condensible gases is solved at every time step to give the pressure and temperature of each computational cell based on the internal energy and masses of the constituent gases.
- c. Heat transfer to structures. Each cell can have an arbitrary Heat transfer occurs n==her of heat transfer structures inside the volume.
via convection, condensation (including evaporation), and radiation betweer.
the gas and the structure surfaces. Gas-structure radiation heat transfer sophisticated model for the emissivity of steam and utilizes a reasonably condensation model is applicable to both saturated and carbon dioxide. The Each structure can be represented as a superheated atmospheric conditions.
planar slab, a half-cylinder, or a half-sphere, and it can be composed of an arbitrary number of layers of materials (e.g. steel, concrete, gas). Each the user through input, and the one-dimensional heat by layer is nodalized is solved to obtain the temperature at each point in the conduction equation 4
material. A condensate film is allowed to collect at the structure surface reaches a user-controlled depth, at which point the excess runs off until it and is added to the water pool, if one is specified for that cell.
- d. Hydrogen combustion. The hydrogen burn model is taken from the EECTR code, which was developed at Sandia for the analysis Unless burns of containment have been problems involving hydrogen transport and combustion.it is assumed that an ignition source is f '
explicitly inactivated through input,the concentrations of hydrogen, oxygen, and always present, and that whena burn occurs. Propagation from cell to cell steam are in a certain envelope, will take place depending on whether certain other concentration criteria are satisfied. All burns are treated as deflagrations occuring over a time period
5-determined by the characteristic len5th of the cell sed a flame speed which is calculated from correlations.
- e. Pool boiling. If a water pool is specified for a given cell, and a debris layer is present below the pool, heat transfer between the debris and the water will occur, and if the pressure-dependent boiling temperature is reached, the pool will boil.
- f. Ice condenser. The ice condenser model involves both thermal-
' hydraulic and aerosol decontamination modeling, but since fission products are not of primary interest in a typical direct heating calculation, we will not describe the scrubbing model. The ice is modelled as a surface held at the ice temperature which changes in area as the ice melts. Condensation heat transfer is modeled between the atmosphere and the ice with a thin water film separating the two. Radiation heat transfer to the ice from the gas is also modeled. The melted ice exits the cell and is added to the pool of a user-at a user-specified temperature. Doors between the lower specified cell compartment and the ice bed and between the ice bed and the upper containment the can be modeled as being be either one-way or two-way; in the latter case, different depending on which direction the flow effective flow area can occurs, Containment Sprays. Like the ice condenser, sprays have g.
important effects on thermal-hydraulics as well as on radioisotope inventories, but the latter will not be discussed here. It is assumed in the spray model that all droplets exiting the spray nozzle are the same size, but that the size can change through evaporation or condensation as the drop falls. The fall velocity is the terminal velocity. Heat transfer from the gas to the droplet t. 'ces place via the same condensation model as is used for structures and for t..a ice condenser, except the Nusselt number used is that When the appropriate to a sphere moving in a gas at the terminal velocity.
droplets reach the floor, they are added to that cell's water pool (or another cell's pool, if the user so specifies in input.)
3.2 Direct Heatine Models )
l The principal modification is that a new field has been added to the code.
The debris field is like the gas in that each cell is a well-mixed repository of the debris mass and its energy. However, the debris mass in each cell is assumed to be composed of a large number of spherical droplets having ,
identical composition and temperature. A realistic debris equation of state is solved at each time step to give the debris temperature in each cell. There are five debris constituents allowed: Zr, Zr0 ,2Fe , Feo, and UO2 '
The debris flows with the gas in a dispersed droplet mode; that is to say, debris is transported from one cell to the next with the gas in proportion Suchto a the mass of debris and gas present in the donor cell at each timestep. In the cavity model is sometimes referred to as a homogeneous flow model.
area, this approach can be justified by the fact that gas velocities are so high that droplets which impinge on surfaces are quickly resuspended in the i
i
i gas stream and fragment down to about the maximum Weber-stable radius.
However, in the model, the droplet diameter, Dd . 18 8pecified in input and This picture of debris transport does not change throughout the calculation.of the critical Kutateladze velocity is
- at gas velocities wellthe in HIPSexcess experiments, and simulant fluid experiments
_ justified by theory, conducted at Brookhaven.
However, as gas velocities drop, it is to be expected that This some is analogous de-entrainment to the will occur that is not followed by re-entrainment.
transition from dispersed droplet flow to annular flow in two-phase flow in pipes. However, so little is known about the flow patterns under these conditions in the complex geometries and large scales of reactor containments that it is not possible to develop reasonable mechanistic models of two-phase flow for the situation under consideration, especially when the material is an unknown mixture of eutectic, metal and oxides at an unknown temperature. a user-Therefore, the process of de-entrainment is treated parametrically:
specified removal rate, f, is assigned to each cell, and in each second, that fraction of the cell's debris content is assumed to be removed from the atmosphere and deposited in a debris layer at the bottom of the The cell remaining(if the debris layer has been enabled for that cell in the input deck).
suspended debris isof the transported, without slip, with the gas to downstream cells. The mass debris contributes to the inertia of the gas if the acceleration term in the flow equation is important. The mass and energy of debris entering a cell is added to that cell's debris field, resulting in a new debris temperature and a new composition.
Heat transfer occurs between the debris and the gas via convection and radiation. A convection heat transfer coefficient is calculated based on a Reynolds correlation for cell for turbulent i relative to the flowgas over specified a sphere, by with the user. a droplet The velocity, v,g Nusselt number is the same as used by Pilch in NUREG/CR-4053, and is given by:
l (1)
! Nu - 2.0 + 0.6 (Re /2)(Pr /3) )
The use of a I where non-zero is the Reynolds Re relative number and Pr is the Prandt1 number.
velocity of the droplets is not necessarily inconsistent ~
with the zero-slip assumption when it is realized that the debris field consists of a collection of particles moving in random directions superimposed on an overall drift equal to the gas velocity.
the I Two models for radiation heat transfer from the droplets are available: In the gray gas model, radiation heat l gray gas model and the clear gas model.between the debris and a gray, non-transmitting transfer occurs A multiplier, a, is provided to reduce the heat transmission assumed. g transfer to to the gas from the black body value. This multiplier can be
' considered be the product of the debris and gas emissivities and any other reduction or enhancement factors which might come into play in gas-debris l l
l i l l
l
_ , . - _ ,. . _ _ _ . . _ - _ _ _ _ _ - _ , ~ _ _ _ _ _ _ _ _ _ , , _ _ . - _ _ _ - _ _ . . , - _ , , , , - . _ . . , , ,
radiation heat transfer. With the multiplier equal to 1, this model will probably over-estimate heat transfer to the gas.
The clear gas model assumes the gas is transparent to debris radiation and that the debris radiates to heat sink surfaces without attenuation. The maximum interfacial area for this heat transfer is the structure surface area, but a multiplier, o ,, is provided to reduce this area in order to account for the possibility that not all of the structure area can "see" the debris plume, or to account for non-unity emissivity, etc. With this multiplier equal to one, the clear gas model will underestimate heat transfer to the gas from the debris. 4 Chemical reactions can take place at the surface of the droplets if they contain oxidizable metal. The metal oxidation reactions allowed are performed in a hierarchical fashion in the following order:
(1.) Zr + O 2 -------> Zr0 2 (2.) 2 Fe + O 2 -------> 2 Fe0 (3.) Zr + 2 H 2O ------->
Zr02+2H2 (4.) Fe + HO -------> Fe0 + H 2 2
(5.) 2H+O -------> 2HO2 2 2 The first four reactions are limited to (1.) the mass of metal in the droplet and/or (2.) the mass of oxygen or steam which can diffuse throught the boundary layer to the droplet surface from the bulk gas. The heat from these reactions is added to the droplet field energy. All mass inventories (debris and gas fields) are appropriately updated in accordance with the extent of each reaction.
The fifth reaction is marked with an asterisk because it is differentThe only from hydrogen the normal hydrogen combustion event discussed in Section 3.1.
involved in this reaction is the by-product of reactions 3 and 4, and the oxygen mass, if any, is taken from the bulk gas, rather than the quantity which can diffuse to the droplet surface. This reaction represents the result of diffusion of the hydrogen byproduct back to the bulk gas. It is assumed that the near-drop environment is so hot that hydrogen-oxygen recombination occurs without need of a flame or spark source. It is also assumed that the back-flow of hydrogen does not impede the diffusion of oxygen, a reasonable assumption given the high diffusivity of hydrogen, and the uncertainty in other aspects of the mass transport model. The heat from this reaction is j
added to the bulk gas, not to the debris.
1 i
i - _ _ _
'~ _ _ _ . _ , . _ _ _ , _ _
--
- e *7-_.,_-___,m_
l 8-Oxidation of UO 2 by oxygen is not treated in this model since there is evidence that it is not favored at the temperatures of interest, and since it The chemical reactions discussed above has little significance are rate-limited in onlyenergetically.
one way: diffusion of the oxidizing gases through the It is also possible that the gas boundary layer around each droplet.
reactions are limited by diffusion on the droplet side, either in the liquid phase or in a solid crust. Baker's ANS/ ENS paper was based on a particular model for the droplet-side diffusion limitation, for example. However, For example,there the is a great deal of uncertainty in how to model this process.
solubility of the oxides in the metal phases must be considered. Also, one must consider the possibility of mixing inside the droplet due to internal circulation loops. For the present model, therefore, will droplet side limitations rely on two parametric are not modeled mechanistically; instead, we features. First, all reactions are shut off at a user-specified droplet temperature, T,. Second, multipliers on the diffusivity of the gases is available; these are designated a,x and a st fr xygen and steam diffusi-vities, respectively. In a gross sense, reducing the diffusivity will limit the reaction in a way similar to the liquid side limit, though the dependence on droplet composition will not be the same.
Mass transfer to the droplet is calculated with a mass transfer coefficient In other words the dimension-based on a heat transfer / mass transfer analogy.
less Sherwood number, Sh, is calculated from a correlation quite like that of the Nusselt correlation in Eq. 1, except the Frandel number is replaced by the Schmidt number:
(2)
Sh - 2.0 + 0.6 (Re / )(Sc ! )
The area used for diffusion of oxygen and steam to the droplets is equal to the area of a single droplet times the total number of droplets in the cell.
Since this treatment does not properly account for the fact that at any given time there will actually be a distribution of particle compositions, some with unoxidized metal left and some without, this treatment may overestimate the reaction rates. Therefore, in a model variation, a multiplier on the area for diffusion is used which is based on an estimate of the fraction of debris particles which still have some metal left.
The Sch=idt nu=ber in Eq. (2) is given by E (4)
Sc - pD where p is the gas viscosity, p is the gas density, and(Equ. D is 16.3-1, the binary gas '
- p. 505, diffusivity in air, given by Bird, Stewart and Lightfoot i
1960 edition). All gas properties are evaluated at conditions intermediate between the droplet and the bulk gas.
exercise, a simple problem was defined which could be run As a verification both on the Pilch model from NUREG/CR-4053, and on an improved version of DHEAT (which accounts for gas inventory changes due to chemical reactions).
The problem involved injecting debris consisting of four materials into an air environment at 10 m/s, and observing the pressure and temperature rises. No After thermal haat sinks were modeled, and no trapping the CONTAINwas allowed.
and DHEAT pressures and equilibration had been achieved, The Pilch model agreed within about 74 in temperatures agreed to within 0.34. the total number of moles of gas due to temperature, butMore it does not adjust important, since the Pilch model calculates droplet the reactions. function of time, it is possible to compare the time required behavior as a for the droplet to reach its peak temperature. On this quantity, the Pilch model and CONTAIN agreed to within 10-154. Given the differences in the models, this was considered adequate verification.
3.3 Modeline Uncertainties There are many uncertainties in modeling a process as complex and poorly It is important that understood as the melt-ejection / direct heating problem.
the results of any calculations using this model be interpreted in the light of these uncertainties. (This is true of all direct heating calculations performed to date, though it is too often true that the uncertainties are paid little attention.) As indicated in the model descriptions, our basic strategy for dealing with these uncertainties is to provide adjustable parameters (defaulted to our best estimates) so that sensitivity studies can be performed through input. Many of the limitations of the present model have been the course of the model descriptions, but it is worth while to identified in In the list below, we identify a nuiber of the re-iterate some of them here.
modeling or input uncertainties, and indicate what control the user has in varying parameters to account for each uncertainty. Ongoing model development will alleviate many of the problems identified below.
Fraction of debris eieeted. This amount depends on the extent of core melting at the time of vessel failure, on the location Very of thecertainty little break, andis possibly onon thethese way the hole in the vessel enlarges.Since the mass and composition of the ejec subjects.
possible debris is specified as tabular input, this uncertainty is fully controlled by the user.
Droelet size. The droplet size may be estimated on the basis of a Weber stability criterion, but smaller droplets are possible depending on the if nature of the fragmentation processes, and larger droplets are possible local gas-debris relative velocities are smaller than assumed in the Weber Therefore, the diameter itself is specified by the user.
number calculation.
Debris transoort. This uncertainty is possibly the most intractable in the problem. One of the principal purposes of the Surtsey experiment series and the simulant fluid experiments at Brookhaven is to improve our l
l
\
g..... -.-
i .
i i 1 i
understanding of the way debris-gas mixtures can be transported through
]
< complicated pathways and around obstacles. De-entrainment and isolation from the high velocity flow region is an important_ potential ILuitation on debris transport. So is freezing on metal structures, although conduction limita-tions may prevent this process from being very efficient if the debris expulsion time is short. Freezing on concrete is a little more difficult to credit, since outgassing and/or spalling of the concrete will prevent the-debris from " sticking". The principal control the user has on this process is l
the trapping fraction, which removes debris at a rate which is proportional to i
the, amount of debris in the cell. The removed debris is deposited in the debris layer, and does not participate in continued direct heating.
j Heat transfer. There is some uncertainty concerning how effective the mixing between the gas and the debris plume might be. .The Nusselt l i
correlation in Eq. 1 assumes a mean droplet velocity relative to the flowing i
gas which is quite arbitrary. Similarly, in the clear gas option radiation l heat transfer is based on good optical contact between the walls and the
- debris plume, and that the plume exterior is at essentially the same
! temperature as the bulk debris. To study that uncertainty, the user can l'
sdjust the interfacial area. The gray gas model has no such problem, but it probably overestimates the heating of the gas and underestimates the heating On the other hand, the presence of copious quantities of l
of the walls.
l aerosols and steam probably make the gray gas model more reasonable than the - !
clear gas model.
i chemical reactions. The chemistry model is quite simple, and it is ,
the actual processes going on would be far more complex. One
{
likely that important assumption is the neglect of a limitation on diffusion on the liquid side of the droplet / atmosphere interface. A solid crust could inhibit I
oxidizing gas diffusion even more. To accomodate this uncertainty, the user has control of an overall multiplier on the diffusivity, which can be used as l
i a surrogate for the liquid side limit. The temperature cutoff serves the l purpose of simulating the effects of the crust formation and freezing of a l 1
droplet, and the value of the cutoff temperature is available throught input
]
) to the user.
Flownath and control volume confiruration. As is always the case
' with computational simulations, part of the model is the nodalization. Of particular concern is the possibility that the flow paths specified are i
incorrect after the melt ejection begins. For example, dynamic loading of the j
boundaries' of the cavity region in Sequoyah could result in a failure which would create a new flow path to different parts of the containment. The j
analyst can study such possibilities with alternate nodalizations and 2
flowpaths between cells which can only open when (This theis pressure option a standarddifference CONTAIN exceeds a number specified in the input.
I feature.)
i Effect of distribution of droelet nrocerties. The well-mixed l assumption applied to the droplet field requires that material entering a cell be mixed with the existing cell material, resulting in a new effective single l
i drop properties (composition, temperature, etc.) In reality, the history of 1
i.
I
each droplet is different, and there will be a continuous distribution of properties. This difference may not be important for some processes, such as heat transfer, since there may be good radiant heat exchange among the droplets, and the overall heat transfer rate may not be very sensitive to the width of~ the various distributions. It may be more important for chemical reactions, since all of the unreacted metal may reside in a relatively small number of droplets. As discussed above, a model variation is available to take the latter effect into account in a simple way, but there is still a good deal of residual uncertainty associated with this problem.
Hydromen combustion. In the chemistry model used, the hydrogen liberated by metal steam reactions on the surface of the suspended drops is assumed to recombine with the oxygen in the bulk. However, it is also possible that the pre-existing hydrogen may be ignited by the melt dispersal event. Conventional steam inerting criteria are probably irrelevant for this situation. A modification of CONTAIN has been developed to allow the user to modify the ignition criteria for the gas concentrations, but what is probably needed is an ignition criterion based on cell gas temperature or debris temperature, or some kind of average. Such a capability for CONTAIN is under development.
- 4. IDEM Calculations of the Beauovah Ice Condenser Plant Containment Resnonse to the TMLB' Station Blackout DCH Scenario.
In this section, we will present direct Containment Heating calculations performed with the IDHM for the Sequoyah ice condenser plant ~. Before ,
discussing these calculations, functional differences between the Sequoyah containment and conventional large dry containment designs which may affect containment response to a DCH event are worth noting. Sequoyah is a small containment with a failure pressure slightly above 4 bars. In comparison, large dry containments such as Surry typically have design pressures in the vicinity of 9 bars. The relatively low failure pressure of Sequoyah is-primarily a consequence of the ability of the ice condenser to remove large ,
quantities of blowdown steam, thereby dramatically reducing peak containment pressures in the design basis accident scenario (double-ended severance LOCA). !
l In a DCH event, hydrogen burns fueled by copious amounts of ex-vessel hydrogen and other relevant DCH phenomena which are not greatly mitigated by the i I
presence of the ice condenser may give rise to a previously unforseen threat.
CONTAIN IDHM predictions of Sequoyah peak pressures in a TMLB' DCH scenario are therefore warranted.
4.1 Problem Descrintion The main objective of this study is to estimate peak containment pressure as a function of corium ejection fraction for the Sequoyah ice-condenser plant with the IDRM. The calculations were performed using a-3 cell nodalization as shown in Figure 2. The standard 3 cell CONTAIN input deck for Sequoyah was supplemented by the required IDHM input parameters and corium source tables.
Relevant IDRM options, corium source masses, vessel blowdown characteristics,
4 and other user-selectable inputs chosen for the calculations are discussed below. This discussion will focus on those parameters most closely related to the modeling uncertainties outlined in section 3.3. l The results are presented as two separate sets of f
- s. De-entrainment.
calculations. The first set of calculations were performed with a debris de-entrainment time (td ) for all three cells of 0.3 seconds. The debris de-entrainment tbne is defined as the inverse of the fractional trapping The rate, f, g
which governs the rate of debris removal from the atmosphere.
second set of calculations were done with a e of d 10.0 seconds. These two values of t d are believed to provide practical upper and lower limits with regard to the effects of debris removal from the atmosphere via interaction with cell structures (a process which is not modeled mechanistically in the IDHN).
- b. Corium Composition. The debris content was taken to be that of the l
TMLB' melt at the time of RPV failure for the Sequoyah plant as provided in BMI-2104 vol 4, Table 6.8. In each case the corium was assumed to enter the lower cavity at a steady rate over a 5 second period. The mass source rates:
(in kg/sec) for the corium constituents in the 100% case were therefore j
- 20198 (UO ), 2360 (Zr), 3052 (Zr0 2 ), and 9968 (Fe). All cases other than 1004 2
ejection consisted of uniformly scaled down corium masses with all other i
parameters held constant.
- c. In-Vessel Hydrogen. Based on the debris composition described above, the in-vessel zirconium-steam reaction was calculated to have liberated 496 kg
- l 2 of hydrogen. One half of this hydrogen was assumed to recombine with atmospheric oxygen in the containment prior to RPV failure (This assumption could be justified by hypothesizing igniters or recombiners with independent power supplies). The remaining hydrogen was assumed to enter the containment with the blowdown steam. This treatment of the in-vessel hydrogen may be non-i
- conservative; however, it separates the DCH problem from the more conventional
- hydrogen burn problem which by itself is known to be a problem for ice condenser plants. Hydrogen burns fueled by hydrogen produced after vessel ,
- l failure by the Zr-steam and Fe-steam chemical reactions were considered and i
are shown below to be the major contributing factor to the DCH induced l
{ l
- pressure rise in the cases that predict containment failure. ,
The vessel steam / hydrogen blowdown was i d. Blowdown Characteristics. 30 seconds following RPV. failure. The blowdown was
- assumed to last forassumption of constant mass flow rate over.this 30 second modeled under the made to account for a ramp at the beginning of the period. No attempt was blowdown nor a tail at the end of the blowdown. The total steam 3 mass of 39,652 kg was determined from the total primary system volume (373 m ) and the specific volume of saturated steam at the failure conditions (T-578 K. P-15.6 MPa). Therefore, the steam blowdown mass rate was 1322 kg per second.
l i
i i
?
,.,.7.- ,____.y. . , ---.--.---,..-.__.m,- _ . . - - ,
,, ,.,,,____._.,_,,.,_,__my.
, - , -, , , , , , , ~ - , , , - . , _ - ~ , , - - , , , , , , - _ _ - , - - , , _ . - , , _ _ , , , _ , , . ,
- e. User Selectable IDHM Input Parameters. As discussed in Section 3.3, several user selectable input parameters are provided to account for various i
direct containment heating modeling uncertainties. The parameters of primary l
- 1. The values chosen for the present '
interest are those listed in Table calculations are now given. All calculations were performed assuming 0.5 mm i
drops (D d), c nsistent with the Surtsey DCH-1 experimental results of debris mass median diameters. The multiplier on oxygen diffusivity, o,g, and the were both assumed to be unity. The multiplier on steam diffusivity, o,g, chemical reaction cutoff temperature, T,, of the bulk debris field was assumed to be 1000 K. Radiative exchange between drops and the' surroundings was treated under the gray-body model with a chosen effective drop emissivity of 0.8. An opaque gas model was assumed in which all of the radiated energy was The deposited in the gas and none in cell structures (o g - 0.8, a ,- 0.0).
debris velocity relative to the gas, v , (used g in the evaluation of Re) was assumed to be 6 m/sec in each of the three cells. This velocity roughly In all corresponds to the terminal fall velocity of a 0.5 mm debris In drop.
this option calculations the hydrogen recombination option was used.
hydrogen produced in a cell is assumed to immediately combine with bulk oxygen in that cell with the reaction energy going to the gas. Finally, the fraction of ice left at vessel failure, b, was assumed to be 0.8.
TABLE 1. Ilser Specified Parameters for IDHM r Fraction of nominal core mass of debris ejected em a Multiplier on black body radiation from debris to gas a, Multiplier on radiation from debris to wall a,x Multiplier on oxygen diffusivity a,e Multiplier on steam diffusivity T, Cutoff temperature for chemical reactions (K) vg Cas-debris relative velocity in cell i (m/s) fg Trapping parameter; fraction removed from cell i per second (s'1)
D Debris droplet diameter (m) d b Fraction of initial ice left in ice condenser at vessel failure l
l
'h.
14 I
I l
l 4.2 Discussion of Results ,
CONTAIN predictions of peak pressure in Sequoyah as a function of corium ejection percent , r,,, are shown in Figure 3 for the two de-entrainment times discussed in Section 4.la. For purposes of discussion, these curves will be divided into 3 separate regimes. The first regime is between 0 and 15 core percent for the t -10 d case and 0 and 20 percent for the t d-0.3 case. This regime is characterized by the sbsence of a hydrogen burn in the upper containment (cell 3). The resulting predicted peak pressures are high enough to threaten but not exceed the failure pressure. Note that at r,g-204, occurrence of a hydrogen burn in the upper compartment depends on the de-entrainment time. While the location of this threshold depends on td' I' I' seen that td has little impact on peak pressures before the burn threshold is
- crossed. In the second region the peak pressure is primarily driven by the 1 hydrogen burn process in the upper dome. Near the end of the second region, an interesting yet physically realistic behavior is predicted when tg-0.3.
i That is, the peak pressure for a 404 corium ejection was predicted to be
, higher than for a 50% corium ejection. Study of the detailed code output indicates that this behavior can be attributed to the fact that the hydrogen burn in the 50% case began at an earlier time than it did in the 40% case.
The timing of the burn has this effect on the peak pressure because at early
- times (immediately after debris ejection) the lower-cavity temperatures are considerably higher than at later times. Due to subsequent cooling, the 4
l lower-cavity can therefore serve more efficiently as a " pressure-sink" for
- early burns than it can for late burns. When t -10 d this behavior is somewhat t
overshadowed by the severity of the early burn. Note that this burn timing phenomenon does not have the same effect on the peak containment temperatures l which are monotonically increasing with core percent as shown in Table 2.
l Beyond r,g-40% is the third regime of the curve. The peak pressures in this
]
l regime for case 2 (td-10) diverge significantly from those of case 1 (tg-0.3). l This divergence is attributed to differing availabilities of hydrogen in the i upper dome at late times (near the end of and following the vessel blowdown).
i During debris ejection, the gas reaches the temperature of the debris for core ej ection fractions above 40 percent in both cases. Consequently, further ,
increase in the ejection debris mass cannot result in further heating of the l
]
atmosphere during the debris ejection phase. Therefore, predicted peak )
i pressures for cases above 40t core ejection will primarily depend upon the l amount of debris in the atmosphere after the debris ejection. Following the ejection, the gas temperature in the lower cavity drops in case 1 due to the rapid fall out of the debris. On the other hand, in case 2, considerable amounts of debris remain in the atmosphere following the ejection as a consequence of the slow fractional trapping rate. Since this debris is a reservoir of thermal energy incoming steam continues to be heated, which thereby increases the driving force that pushes hydrogen through the ice 1
l i
i m ...-..-__ , _ , . , . _ _ _ _ , _ - . _. c _ _ _ - _ ..._.__._,_-..___,_.._,_,__._.~.._,m,. , . , _ _ _ _ _ , . _ . _ , _ .
i' i 5
condenser and into the upper done. In case 1 the peak pressure reaches a j plateau beyond 404 core ejection as a result of a saturation of this driving r force. That is, the amount of debris in the atmosphere following the debris l ejection. phase is insufficient to further heat the vapor in the lower cavity. l
\
'l i
Table 2 CONTAIN Predictions of Peak Pressure and Temperature
- in Sequoyah Containment for a TM13' DCH Scenario, t =0.3 e -10.0 I
d d Percent Peak Peak Temperature Peak Peak Temperature j Core Pressure T(1) T(2) T(3) Pressure T(1) T(2) T(3)
(K) (K) (K) (bars) (K) (K) (K)
, (t) (bars) 0 2.74 430 438 394 2.74 430 438 394 l 10 3.23 1254 915 477 3.32 1456 1035 504 15 - - - - 3.58 1647 1154 548 20 3.61 1666 1107 528 5.81 1816 1235 1182 i 25 5.60 1803 1172 1128 - - - -
30 6.21 1919 1226 1177 .7.26 2070 1342 1317 35 6.86 2010 1257 1247 - - - -
40 7.18 2088 1298 1280 7.36 2246 1417 1338 50 6.90 2217 1351 1320 7.78 2283 1489 1415 l
i 75 6.72 2387 1478 1339 9.03 252: 1560 1663
! 100 6.95 2436 1557 1408 10.55 2865 2066 2076 4,3 Conclusions i
There is no question that there are many complexities to the DCH j problem, and 'that in many respects the simplifications inherent in the 4 IDHM modeling (and in other parts of the CONTAIN code) are inadequate
) representations of the phenomena. Ultimately, we must depend on an improved experimental data base for improvements in some areas of our understanding. However, in the meantime, the system-level analysis l
which the CONTAIN code affords makes it possible to make the connection between the existing knowledge base and the operating reactor situation. ;
l The uncertainties in our knowledge can, in many cases, be represented by t '
The understanding
! ranges of the user-specified input parameters.
thereby gained can help guide the experimental program and future model l development activities, as well as provide decision makers with interim l
assessments of the implications of DCH on reactor safety.
Three important conclusions emerge from the sensitivity studies which have been described here. First, the most important metal oxidation l ,
l I i
t l
l l
l l l
l
l 1
reaction is the steam metal reaction, a consequence of local oxygen starvation in the vicinity of the debris when it is first ejected, and of the rapid chemical reaction rates. A somewhat unexpected aspect of this issue is the importance of the iron / steam reaction, which in itself is not particulary exothermic, and is usually limited under more normal accident conditions by oxide crust formation, but which proceeds in these calculations because of high ambient gas temperatures.
The second conclusion concerns the importance of hydrogen combustion in DCH scenarios. The quantities of hydrogen generated by these processes are extremely large, and the generation rates are unprecedented in rasctor safety analysis. In ice-condenser plant, the pressure rise due to debris-gas heat transfer is often not as large as the pressure rises from the subsequent hydrogen burns. Uncertainty about the mode of hydrogen combustion, and the various criteria for combustion, therefore, has become a major concern for DCH analysis.
Finally, the third important conclusion is that, for Sequoyah, the details of intermediate debris transport--de-entrainment, re-entrain-ment, sticking, bouncing, and so on--which have tended to dominate DCH discussions in the past, may be less important than we have previously believed. Over a rather vide range of our debris de-entrainment time parameter, the sensitivity of peak pressure was not very large. This result is, of course, due to the importance of the hydrogen generation and combustion processes discussed in the previous two paragraphs.
While one cannot conclude that intermediate debris transport is unimpor-cant for this or any other plant, the reduced sensitivity to this aspect of the problem can have important implications for future research I activities and current assessments for regulatory applications.
4 I
f 1
i l
l
.-._,_._._,,_.,._m.__-,_._,, _
m._,, _ . _ . _ . . . , _ , , _ - . . . . , . ~ _ _ . _ , . . - , , ,
17-
/////H l
ENGINEERED / g ..
- . ?*: .:' .
SAFETY / j ,*.,
i FEATURES :
[ CONDENSATION EVAPORATION TWO PHASE THERMODYNAMICS Jh i
AEROSOL EVOLUTION AND FLOW MYDROGEN SURN RADIOISOTOPE TRANSPORT AND DECAY a TERCRL FLOW i
STRUCTURE HEAT TRANSFER
/ a - _-
l 1
i i
l
' h i l
D
.' N .:+. c._,_ _ ..-
i REACTOR CAYlTY .
' PHENOMENA i \
FIGURE 1.
i SCHEMATIC DIAGRAM OF \
PHENOMENA ANALYZED IN THE CONTAIN . CODEA INDICATING THE . CONT i
I
,--%y- pe ,,-----w9-,,---v-,+ -__%-._w%,_,---,,,-,,__ym,.c - , ,-,y,.y.-. .,,-,w-,w--..,w.w=vy-e-ww,,ve
l
+
3
+ l 2 2
~
1 I 1 Compartment Descriptors
- 1. Lower compartment
- 2. Ice compartment
- 3. Upper compartment (Dome)
Figure 2. SCHEMATIC DIACRAM OF THE CONTAIN 3 0 ELL NODALIZATION OF THE SEQUOYAH ICE CONDENSER CONTAINMENT.
I j * .
1
- 11.0 ,
' s i
l 10.0 -
1 i 9.0-
! 8.0-1
! v 7.0- ,
l @
l: $ 6.0-en in
! o 5.0-1 L.
Q '
- 4.0 - ..
.M a ta = 10.0 g
l g 3.0 ; 4 = 0.30 1
2.0-i l 1.0-i 0.0 .
i .
i 0.0 20.0 40.0 60.0 80.0 100.0
! Percent Core Melt-Ejection 1
l Figure 3. CONTAIN Prediction of Peak Pressure in Sequoyah Contaisument 4 for a TMt.B' DCH Scenario. Results for 2 Values of Debris De-Entrainment Time, e *** 8 d I'**)'
i . _ _ _ _ _ _ _ _ _ . - __
\ . 1 t).f,2l l
~) \
Sandia National Laboratories j Albuquerque, New Mexico 87185 E September 26, 1986 I
Dr. Thomas J. Walker NRC/RES ;
i 7915 Eastern Avenue '
Dear Dr. Walker,
't As you and I have Direct discussed on the phone, we have just completed Containment Heating (DCH) calculations with a number of the CONTAIN code as partwith of our support for SASA, and this some of the highlights. Details letter will provide you How-will appear, as you know, in a forthcoming NUREG report.
l
- - ever, because of the interest which Tim Lee and Farouk Eltawila have shown in this work in regard to their NUREG-ll50 issue we are providing this summary immediately, in the hopes l
i paper,
! that it will be of use right away.
t The new work is an extension of the sensitivity study which Dave Williams reported in his American Chemical Society conference l
s paper, of whichparameter you have a copy. What we have done is select a number of his variations for Surry In and performed addition, a few the new l
i same variations on the sequoyah plant.
! parameter variations were performed for both Surry and Sequoyah.
In this way we can directly compare the plant specificity of the I
modeling sensitivities. The resulting matrix of calculations provides some very interesting insights on the DCH problem and l
adds to our growing understanding of the importance of different d
l j phenomena and plant characteristics. 1 A representative result is shown in the attached Figure 1, which is the pressure-time history around the time of vessel breach for the Base Case. For this study, all the assumptions and parameter choices described in Williams paper were also applied, so you can
- get a complete description Only of the conditions of the calculation plant-specific aspects changed; e.g.,
by referring to it.
the mass of corium ejected was scaled to the core size (but was 4
still chosen to be 75% of the core mass in the hase case).
Plant-specific parameter choices are described in my recent letter to Mel Silberberg (dated September 2, 1986).
Two options for hydrogen combustion modeling were exercised for These are burn model "a", which is the r all parameter default model variations.
for all ignition, completeness, and propagation l
criteria in CONTAIN (as taken from HECTR) . The alternative model paper as " prescribed burn". In this "b" is described in the starting two seconds afthr vessel i model all hydrogen is burned,there is oxygen, regardless'of concentrations l failure, whenever As explained in the l of steam, hydrogen, and oxygen in the cell.
l l
=
- 57enfomerg l
1
'PP j i
d Dr. Walker September 26, 1986 paper, this model may be more reasonable than the default model because of the presence of hot gas and hot debris. In any case, we feel the two choices "a" and "b" are bounding on this impor-tant uncertainty.
The parameter variations are summarized in Table I. A total of 8 variations are reported there for each plant and each burn op-tion. Thus, with the base case, the total number of calculations o is 36. We present only the peak pressure (P,,x) and the ratio P,,g/Pf, where Pf is the nominal failure pressure for each plant.
We arbitrarily selected the failure pressures to be 135 psia for Surry and 60 psia for Sequoyah, consistent with SARRP studies (though there is a lot of uncertainty about these numbers).
Details about other parameters (e.g. temperatures in different compartments) will be provided in the NUREG report, but this summary provides an at-a-glance picture of what the key uncer-tainties are. The extent of each parameter variation was l
selected somewhat arbitrarily as being within the range of 1 uncertainty. These are not bounding ranges, in general. If one had uncertainty ranges on the parameters, it would be straight-forward to estimate resulting ranges in the peak pressure by linear extrapolation from the results in Table I.
~
In general, the results in the table are self-explanatory, but a few key issues should be noted. The debris de-entrainment times d used were t based on the scaling relation that Dave Williams used, but he described it in terms of a fractional trapping rate.
This is just the inverse of the de-entrainment time used in Table I. Our approach here is in contrast with the Sepoyah study reported in the above-mentioned letter to Mel, which used the
. same t d in each cell. Another important difference was that in i
the earlier Sequoyah study, we did not include drop side limita- i i tions on mass transfer. However, this was shown by Dave not to i be of crucial importance in most cases. Finally, the earlier l Sequoyah study had a different debris composition, being based on BMI-2104 results, rather than Containment Loads Working Group j Standard Problem 2, which is the basis for the results reported l
in Table I.
l Please let me know if you have any questions on this material. l 1
Sincerely, i K~kv~~
Kenneth D. Bergeron, Supervisor Containment Modeling ,
Division 6449 KDB:6449:mm l
l l
l j
J Dr. Walker September 26, 1986.
Copy to:
NRC/RES T. Lee NRC/RES P. Wood NRC/RES J. Mitchell NRC/RES R. Meyer NRC/NRR F. Eltavila 6440 D. A. Dahlgren 6449 K. E. Washington l 6449 D. C. Williams 6449 File 3.7 i
S i
l
Sequoyah Pressure History Predictions for TMLB' DCH Scenario with CONTAIN - IDHM (base case) 120.0 Default H-burn
Continuous H-burn 100.0 -
- . ~~~~~.
3m
,m -
p., 80.0 - f l
0 /
U :
En : _
l @ 60.0 -
u ./
O-. .
40.0 -
20.0 .
i .
i .
i .
9400.0 9410.0 9420.0 9430.0 9440.0 Time (sec)
f _
P 0 8 2 0 2 2 8 5 2 _
6 7 3 4 1 1 5 8 0
" /g . . . . . .
b , 1 1 1 1 2 1 1 1 1 P
n r
t u 1 1 n B x 6 7 9 4 7 7 6
5 9 1 6 a a 9 0 7 8 2 1 1 l m 1 P P h
a y .
o g 5 2 3 7 7 u P 3 8 5 4 0 8 8 9 8 q / 8 7 5 5
- x . . . . . .
1 0
S "a a 1 1 1 1 2 0 1
" a P
n r
u , 3 9 0 8 2 B 0 7 3 3 1 1 5
, 1 0 9 9 2 4
, 1 1 1 1 1 P
f P 0 6 0 7 3 2 5 6 2
/ 0 0 9 9 3 8 0 2 0
_ " x .
1 0 1 1
1 b a 1 1 0 0
" a P
n r
u .
B x 5 2 3 0 9 1 1
2 9 6
7 3
t a 3 4 2 3 7 4 1 1 n a 1 1 1 1 1 1 1 a P l
P y I r f r P 8 5 9 4 3 4 6 5 4 E 0 6 7 9 7 L
_ u / 7 8 6 7 B S x - .
A "a a 0 0 0 0 1 0 0 0 0 T
" a P
n
_ r u
B x 5 5 3 0 0
8 3
6 8
3 0
7 2
0 0
a 0 1 9 1 1 1 a 1 1 1 1 P
n o )
i m t o P f
m c r r e
_. o n a f t
(
_ r w r a s
) f o. f n r e
e dc 5 o e w s . we os t2 c
i t a d r
u a n t a e s aa b a l e0 w s s ii
- ( im b0 j d n g dm 1 eo i d r e r ppss o nd m t x0 ge e p
i g it p0 ao e o3 ks P p o. r5 r s 05 t n t et o7 r0 4e 5i e63 a e i p
p1 r1
- c. Z Ep Ed r 1 i s a. d s0 o 0- u r a p r0 o 3 0 0s
.o a c a t et e e et .i l :
r g gm gm g 2d 1c i h V
e s
t t=
sd n5 a
no ar no ar n5
- a. do dg a a F y:
oy a o at h0 hf hf h0 dc dk B N F C C C C A A d ur e qr m eu e u Ss s 1 2 3 4 5 6 7 8 9 s .
a s C A .
J < ! l
1 I
Surry DHEAT2 Calculations Without Blowdown Steam and no Stearn Spike 2400.0 16.0 -
l n
- Pressure ,... ~.. ,.
14.0 _
= Temperature l
- .. ~ .. m--
2000.0 q
g v L 12.0 -
d a)
L CQ v ,.. -
g e 10.0 -
1600.0 a L .- d L
U .- q)
$ 8.0 -
' a O
, C L
L CL 6.0 - -
~
1200.0 $
d -
~
g
- C
.6 ~
E E 4.0 -
800.0 2.0 -
0.0 / .
i i
i i
. 400.0 0.0 20.0 40.0 60.0 80.0 100.0 Percent Core Melt-Ejection
Surry DHEAT2 Calculations With Blowdown Steam and no Steam Spike
, 2400.0 16.0 - .
- Pressure 14.0 -
= Temperature -
2000.0
'W n 2 v
L 12.0 -
'.. q) d v
CO
', .. .. s *...1600.0
- a g
q) 10.0 -
M D c)
.=~ a 8.0 -
d q) C u
cL
,~..= -
1200.0 $
- 6.0 - -
d *,- d C '
- C C 4.0 ,
C 800.0 2.0 -
,s' 0.0 :f' .
i i
i
. 400.0 0.0 20.0 40.0 60.0 80.0 100.0 Percent Core Melt-Ejection
_ ----- -