ML20064F644
| ML20064F644 | |
| Person / Time | |
|---|---|
| Site: | 05200004 |
| Issue date: | 02/28/1994 |
| From: | Leaver D, Li J, Sher R POLESTAR APPLIED TECHNOLOGY, INC., RUDOLPH SHER ASSOCIATES, TENERA, L.P. (FORMERLY TENERA CORP.) |
| To: | |
| Shared Package | |
| ML20064F626 | List: |
| References | |
| NUDOCS 9403150380 | |
| Download: ML20064F644 (45) | |
Text
{{#Wiki_filter:. l DRAFT SBWR. CONTAINMENT NATURAL AEROSOL REMOVAL Prepared by: David E. Leaver Polestar Applied Technology,Inc. 4 Main St. Los Altos, CA 94022 Jun Li TENERA, L.P. 1340 Saratoga-Sunnyvale Rd. San Jose, CA 95129 Rudolph Sher Rudolph Sher Associates 740 Mayfield Ave. Stanford. CA 94305 Y Prepared for the U.S. Department of Energy under Contract No. DE-AC07-931D132 DRAFT February 1994 A l
DRAFT SBWR' CONTAINMENT NATURAL AEROSOL REMOVAL TABLE OFCONTENTS ' Section Page 1 INTRODUCTION . . .. .. ... .. . . .................................................I 2 HASIC AEROSOL THEORY AND THE NAUAHYGROS CODE., ... 3 2.1 A e r o sol Fu n da m e n ta l s . . . . . . . . .. . . . . . . . . .. . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 3 2.1.1 Deposition Vel oci ti es . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 - 2.1.2 S edi m e n tati o n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . 4 , 2.1.2.1 Gra vitational S e ttli n g . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . 4 ~ 2.1.2.2 A g gl o m era ti on . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . 4 2.1.23 Condensation on Particles.. ........ ............. 2.1 3 Di ffu s i o n . . . . . . . . . . . . . . . . . . . . . ..........6 . . . . . . . . . .. . . . . . . . . 2.13.1 Di ffu siophores is. . . . . . . . . . . . . . .. . . . . . . . . . . . . .. . . . . . . . . . . . . . 6 2.13.2 Wall Condensation . . . . . . . . . . . . . . . . . . . . . . . ................ 7 2.13 3 The rm ophoresi s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . 7 l 2.13.4 Other Diffusion Processes .... . ..... .... ... ...... ... . ... . 7 2.1.4 Vali dati on of Aerosol Models . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . 8 i 2.2 Descrip tion of NA UA H YG R OS . .. . . . . . . . . . . .. . .. . .. . . . . . .. . . . .. . . . . . . . . . . 8 2.2.1 Validation of NA UA HYG ROS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . 10 ' 3 METHODOLOGY AND MODELING FOR SBWR ! C A L C U L A T I O N S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . .. . . . . . . . . 3.1 R e fe r e n c e S e q u e n e e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . , . . . . . . . . . . . . 14 3.2 M A A P T h e r mal Hyd ra ulics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . 17 ' 3.3 A erosol Sou rce Defin ition . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . 17 3.4 E ffe c t o f t h e P C CS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 3.4.1 Uniform Upward Flow Effects on Aerosol Sedimentation......... 22 3.4.2 Aerosol Removal in the PCCS and Wetwell ......................... 24 ! 4 RESULTS OF SBWR CA LCULATIONS .... ... ... . ........ .. .. . ..... . ... .. .... 25 5 4.1 B a s e C a s e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .- i. 4.2 S e n si t i vi ty S t u die s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 4.2.1 S eq uence Ti mi n g . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . 26 ' 4.2.2 A erosol Sou rce Si ze Di stributi on . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . 27 4.23 Effects of Hygroscopici ty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 4.2.4 Non-Fission Product Aerosol Source Rate........................... 28 ' 4.2.5 Effect of Dry Particle Densi ty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . 29 4.2.6 Conclusions from Sensitivity Studies....................... ..... ..... 29 DRAFT
.I r
- ,... , - r, ...-+.
DRAFT TABLEOF CONTENTS (cont'd) 5 UNCERTAINTIES... . . ... . . .. ... ........................30 5.1 Aerosol Source Characterization.... .... . ... . . ... .. . .....30 5.2 Thermal Ilydraulic Uncertainties. . . ..... ... .... . ... ..30 5.3 Code Limitations.... ...................................30 6
SUMMARY
AND CONCLUSIONS.. ... . .. .. ... .... .............33 References.. . . .. .... ... .. . ....................................34 APPENDIX A - AEROSOL REMOVAL IN flIE PCCS. ... .. . ... . .... . 36 APPENDIX B - CALCULATION OF REMOVAL COEFFICIENTS... .. 39 LIST OF FIGURES
- 1. Suspended Aerosol Concentrations in the LACETests.. .... . .........11
- 2. Base Case Thermal Hydraulic Conditions.. .. .. .... ... .. . . . . . . . . . . 18
- 3. Containment Scheraatic. .... ........ . . ......... ......... . . . . . . . . . . 23 B - 1. A vs ti m e. . . . . . . . . . . . . . . . . ...........................................41 LIST OF TABLES
- 1. Integrated Settled, Plated, and Leaked Aerosol, Tests LA-2 and LA-4 .. .12
- 2. SBWR Accident Sequences Considered for Containment Aerosol Calculations... . ... ... . . ... . ..... .. . ........ .......... . . .. . . . . . . . 1 6
- 3. GRIGEN Core Inventories. ..... . ... .......... .......................20
- 4. Release Fractions... . .........................................20 1 input Assumptions ....... ..............................................21
- 6. Ae rosol Sou rce Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . 22
- 7. Best Val ue SBWR Resul ts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
- 8. Aerosol Removal Rates ( A ). . . . .. . .. .. .. . . . . .. . .. .........................26
- 9. Alternative Release Timing Comparison. ... ..... . ... ..... ..... . . . . . . . 27
- 10. Effect of Source Size Distri bution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . 2.3 1 1. Effect of Hygnscopi ci ty. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . 28
- 12. Effect of Different Non-Fission Product Aerosol Source.......... .... ..... 29 DRAFT a
i I l
DRAFT Acknowledgment This report presents work sponsored by the Advanced Reactor Severe Ac the U.S. Depanment of Energy through Contract No DE
.- - AC07 rogram of 931D13200 wish to thank contributions Bill Usry and Doug Mcdonald of General to this work. Electric for their h l f
. The authors epu DRAFT iii
DRAFT Section 1 INTRODUCTION The Advanced Light Water Reactor (ALWR) Program, under the leadership of U.S. and foreign utilities and the management of the Electric Power Research Institute (EPRI), has developed an updated Gssion product source term for application in the ALWR. The Department of Energy Advanced Reactor Severe Accident Program (ARSAP) has provided technical support for this effort. The source term work was undertaken for three reasons: (1) to update TID-14844 and associated regulatory guidance that goes back 30 years, thus providing a more stable licensing basis,(2) to improve plant safety by basing mitigation system design on realistic source term phenomena, and (3) to provide a realistic, technically correct source term for addressing emergency planning. Initial technical work on the source term update for Evolutionary ALWRs (Leaver et al, 1990) and Passive ALWRs (Leaver et al,1991) has been completed in the past few years. Since that time, the NRC has initiated its own source term studies, including NUREG-1465 (Soffer et al,1992) which defined release fractions, timing, and chemical form. The ALWR designs have progressed, including completion of the passive plant safety analysis reports for design cenification. The ALWR Program has also continued work on the updated source term to strengthen the technical basis in areas which depart from traditional licensing practicc. One such area is natural aerosol removal for fission product control in containment, which is the subject of this report. Traditional practice in PWRs has been to utilize sprays (whose primary purpose is pressure suppression in the containment) for fission product removal in containment. In BWRs, containment spray systems are also provided. riowever, these spray systems are not credited in licensing analyses. The primary fission product removal mechanism for licensing is the suppression pool, A goal of the ALWR program is to demonstrate that natural aerosol removal processes would be effective enough to make reliance on engineered aerosol removal systems (such as containment sprays) unnecessary in licensing analyses. Furthermore, it has been specified in the Passive ALWR Utility Requirements Document (URD)(EPRI,1993a) that natural aerosol removal inside containment be credited in meeting the 10CFR100 dose limits and in meeting the protective action guide (PAG) dose limits. No spray systems are specified in the URD. The passive plant designers are relying on natural aerosol removal both for 10CFR100 and for PAG dose calculations.This report supplements and updates natural aerosol removal work documented in the earlier ALWR reports, extending it to the SBWR. DRAFT Page1
DRAFT It is a natural outgrowth of the design philosophy of the SBWR to utilize passive systems and natural processes for control of fission products in the event of core damage. This applies both t fission products inside the containment and to Gssion products that have leaked from the' containment to the secondary building. The important natural aerosol processes inside the SBWR containment include agglomeration, steam condensation on particles (including the effect of hygroscopicity), gravitational settling, thermal gradient driven diffusion (thermophoresis), and steam condensation driven diffusion (diffusiophoresis). These processes are well understood. having been demonstrated experimentally, and benchmarked calculational methods exist for evalnating them. An earlier study (Leaver et al,1993) developed a refined methodology for evaluating natural removal of aerosols from the containment atmosphere,and applied this methodology to the AP600 plant.The objective of the current study is to apply it to the SBWR plant. It uses SBWR thermal hydraulic results from MAAP 3.0B obtained from General Electric (Mcdonald,1993), applies these results in a consistent manner, and establishes a clear rationale for choosing the accident sequence type to be analyzed. The integrated aerosol leakages over 2- and 24-hours after the initiation of the accident have been calculated, and these leakages are used to derive an aerosol
- removal coefGeient that can be used for the 2-hour as well as longer term dose evaluations.
As in the earlier AP600 study the aerosol code that was chosen is a modined version of the EPRI- ' sponsored code NAUAHYGROS (Sher and Jokiaiemi,1993), a detailed mechanistic treatment of aerosol behavior in containments. This code is well benchmarked and includes almost all significant processes which contribute to natural aerosol removal. The omission of a few of these processes is of no significance and, in fact, results in some conservatism in the calculated leakages. The remainder of this report presents an introductory description of aerosol mechanics, a short discussion of the modified NAUAHYGROS aemsol code, a description of the methodology, and the results obtained in the SBWR study along with the uncertainties. DRAFT Page 2
\
DRAFT Section 2 BASIC AEROSOL THEORY AND TIIE NAUAHYGROS CODE 2.1 Aerosol Fundamentals The natural mechanisms that act to remove aerosol particles from the containment atmosphere are sedimentation (settling), diffusion (plateout), and leakage. Aerosol agglomeration and steam condensation on the particles are r.!so important mechanisms, since they result in growth of the particles, thur Sancing the sedimentation rate. The basic aerosol equations used in most aerosol codes des. N total number (and mass) of the suspended aerosol and its size distribution evolve in timt., . count of the introduction of aerosol from sources and its removal through the various removai . ;hanisms. These equations are coupled partial differential equations and in order to solve them numerically it is necessary to approximate the particle size distribution by dividingit into a set of N discrete " size bins" with each bin allotted a fraction of the total number (or mass) of particles. Doing so enables the differential equations to be converted to a set of first order ODES. In general the rate equations are of the form dn j = s, - K,9 (1) for diffusion, sedimentation, and leakage, and are of the form
- -} K qn, g + } K)n,n, (2)
J JA for agglomeration processes. In equations (1) and (2), n, is the number of particles (or the number per unit volume)in size bin i, K, is a rate constant characterizing thepth removal process, s, is the aerosol source rate into bin i, Kyis an agglomeration kemel characterizing the rate of agglomeration between particles in bins i I andJ. Kj characterizes the rate of agglomerations between those binsjand k (J k s i) which result in a particle in bin i. I 2.1.1 Deposition Velocities l For diffusion and sedimentation processes,it is convenient to replace the rate constants K, by their ; corresponding deposition velocities v, - K, , where V A is the surface area on which the
~'
DRAFT
DRAFT ; deposition is taking place (e.g., upward. facing horizontal surfaces for sedimentation), and V is the containment volume. Standard expressions for the deposition velocities (e.g. the Stokes formula l for settling velocity) are well established in the literature. It should be m>ted that throughout this discussion it will be assumed that the particles are spherical. Most of the applications of the theory in post-accident light water reactor containments are expected to involve hygroscopic aerosols and/or saturated steam atmospheres. In either case, condensation of steam on the particles will be expected to occur (see below). It has been experimentally demonstrated (Sen6ck et al,1981) that under such conditions the particles will be at least approximately spherical because of the surface tension of the condensed water. Provision for user-specified shape factors to account for non-sphericity is available in the codes. 2.1.2 Sedimentation Under normal conditions. there is only one sedimentational removal mechanism that causes aerosol panicles to settle on the floor: gravittiional settling.The mte of sedimentational removal of aerosols depends on the settling velocities of these aerosols. 2.1.2.1 GravitationalSettling The Stokes formula for the settling velocity of an aerosol particle (u,)is given by Equation (3): u, - # 9p [Cn(Kn)] (3) where p, is the particle material density, g is the gmvitational acceleration, a is the particle radius, is the viscosity, Kn is the Knudsen number and Cn(Kn)is the Cunningham slip correction factor. Since the settling velocity of an aerosol particle is proportional to the square of its aerodynamic radius ([a), growth of the particle size is imponant to particle sedimentation. In general, two aerosol growth mechanisms are considered: agglomerational and condensational growth of aerosols. 2.1.2.2 Agglomeration The imponance of agglomeration between particles is that the resulting panicles, being larger, will settle out faster. As a result of agglomeration processes, particles are both removed from and added DRAFT
I DRAFT to size bins, since if two particles of radii r, and e jcoagulate, one new particle is fonned whose radius is r, -Grl + rj . Note that there is a net loss of one particle per agglomeration: the total mass, however, is conserved. There are three principal mechanisnis that lead to particle agglomeration, namely, Brownian ! diffusion, turbulent diffusion, and gravitational "sweepout. The Drst two of these result in particle-paniclecollisions due to the random Brownian motion or random turbulence-induced motions of the particles. Gravitational coagulation results from the fact that larger or heavier particles fall with higher terminal (Stokes) velocities than smaller or lighter ones. The faster particles can thus overtake and collide with the slower ones in the course of their fall. Several other uglomeration mechanisms exist, but are of little signiDeance for reactor containment aerosols in a bat follows turbulent agglomeration is also neglected; as mentioned previously this results in conservative leakages. 2.1.2.3 Condensation on Particles In steam / air atmospheres, conditions may exist that enable steam to condense on some or all of the aerosol panicles. As with agglomeration the importance of condensation on the particles is that the aerodynamic radius becomes larger, and themfore the particles settle out faster. The model used for condensational growth on particles (or shrinkage,if evaporation occurs)is based on the Mason equation (Mason,1971), which is widely used in atmospheric aerosol and cloud physics. For hygmscopic particles the Mason equation must be modiDed to account for the additional decrease in the equilibrium pressure at the surface of the droplet due to the solute effect. The correction involves the activity coefDcient of the droplet solution, which depends on the van't Hoff factors r and the concentrations of the hygroscopic materials dissolved in the droplet. It should be noted that condensation can occur on hygroscopic particles even if the relative humidity is less than one,in contrast to the non-hygroscopic case, which requires a supersaturated steam atmosphere. Furthermore, for hygroscopic particles, at any relative humidity (RH) greater than the deliquescence point there is an equilibrium size to which particles of a given initial size will grow. If the RH is not very close to one (< 0.995) the equilibrium size will be reached rapidly enough that it can be assumed that it is attained instantaneously. This enables the condensation calculation to be speeded up considerably. In severe accidents some of the most important fission product aerosols are hygroscopic, including CsOH,Cs2 CO3, and Cs!. Agglomerated particles that contain a hygroscopic component will also be hygroscopic. -
l l 2.1.3 Diffusion - d There are various diffusion mechanisms that result in particles moving towards and depositing on surfaces. The principal ones in containments are diffusiophoresis, thermophoresis, Brownian diffusion, and turbulent diffusion. 2.1.3.1 Diffusiophoresis Diffusiophoresis is the diffusion of particles to interior heat sink surfaces, such as the containment walls, on which steam is condensing. As steam condenses, bulk steam moves towards the surface, while air or other noncondensible gases diffuse away from the surface in order to maintain constant total pressure.This msults in a net mass flow (Stefan flow) of the gas / vapor phase to the surface that entrains the aerosol particles,which can thus be thought of as being caught up in the flow of the condensing vapor molecules to the wall. The diffusiophoretic velocity is given by the following equation (Waldmann and Schmitt,1966): p1* W v"""- (h + X i,5 ) P, (4) where x,, = ratio of the mole fraction of air to that of steam M, ( M. ) = molecular weight of steam (air) p, = density of steam W = steam condensation rate per unit area on the wall (W/p, is the vapor velocity at the wall; if there is no air (x, ,. = 0), this is also the Stefan flow velocity). Note that the diffusiophoretic deposition rate is proponional to the rate of steam condensation on the wall, and is independent of the particle size. DRAFT I r,s
2.1.3.2 Wall Condensation DRAFT l It is necessary to know the wall condensation rate (or condensation heat transfer rate) for two reasons. First, it determines the diffusiophoresis deposition rate according to equation (4), and ; second,it influences the relative humidity, which drives the particle condensation process, as mentioned above in the discussion of the Mason equation. Steam condensation on the wall will be significant when there is significant steam injection into the containment and there are ample heat sinks, such as a cold wall. For the SBWR. the steam condensation rate on the " wall" is actually the MAAP-calculated condensation rate in the PCCS. ' 2.1.3.3 Thermophoresis Thermophoresis causes a diffusive effect due to a temperature gradient in the steam / air atmosphere near the surface. Here the gas molecules on the hotter side of the panicle collide with it more frequently than those on the cooler side, leading to a net momentum transfer to the panicle, which drives it towards the cooler side, i.e., the wall. The thermophoretic velocity is proponional to the temperature gradient at the wall (or, equivalently, to the sensible heat transfer rate to the wall), and is dependent on particle size, except for very small particles (less than 0.1 pm radius), although the dependence is not very strong even for particle radii as large as 1.0 pm. The MAAP calculations for the SBWR indicate that there is little or no sensible heat transfer to the containment structure, so thermophoresis has not been included in the aerosol removal calculations. 2.1.3.4 Other Diffusion Processes Brownian diffusion is the net diffusion of panicles to surfaces due to their random motions. As particles stick to the surface a concentration gradient is set up in the gas, resulting in a net diffusion of particles to the surface. Brownian diffusion is usually rather unimportant as a removal mechanism. In large containment volumes, although turbulent diffusion and impaction mechanisms also usually play a role in the transfer of panicles to the wall, these mechanisms are ignored, because of the difficulties in predicting the gas flow fields that are present. Doing so will lead to ' under prediction of aerosol deposition and therefore to conservative aerosol leakages. i p rz y p.,~ N* L4 % bge 7
l l 2.1.4 Validation of Aerosol Models "DRAI"d 1 The aerosol processes discussed above are well established and have been confirmed m many ; separate effects experiments, which are discussed in standard references (Fuchs,1989; Pruppacher and Klett,1978).The agglomeration formulas for Brownian coagulation have been validated to within about 20% The gravitational agglomeration equation has not been as well confirmed, except when the two particle sizes are very different. When the particle sizes are roughly equal, the kernel is not as well established, but on the other hand, the coaguletion rate is small because the velocities of the two particles are similar. The Stokes formula for sedimentation velocity has been well confirmed for particles whose diameters are less than about 50 p m. These usually make up the bulk of the aerosol. For larger particles the Stokes formula overestimates the settling velocity (by up to a factor of about 10 for 1 mm pellets);for the relatively few particles large enough to be beyond the range of the Stokes formula other expressions exist for calculating their terminal velocities. However, in the present calculations the maximum size of the particles is limited to 50 y m. There are some separate effects validations of the diffusiophoretic effect, but the best confirmation comes from integral experiments such as the LACE tests (Rahn.1988). Calculations of these and other integral tests accurately predict the integrated mass of plated aerosol material only if diffusiophoresis is taken into account. Ifit is neglected, the predicted plated mass is about two orders of magnitude too small, compared to the observed plated mass. , Confirmation of condensation phenomena on particles has come both in cloud physics studies and simulated containment experiments.The principal problem in the context of containment aerosols has been the lack of activity coefficient data at containment atmosphere temperatures for the important hygroscopic fission product materials such as CsOH and CsI, These data are now available, and are incorporated in NAUAHYGROS. f 2.2 Dese.iption of NAUAHYGROS The containment aerosol code used in the ALWR work reperted here is a modified version of NAUAHYGROS. NAUAHYGROS (Sher and Jokiniemi,19y3)is a revision of the German code NAUA (Bunz et al,1983), and wasjointly developed at EPRI and VIT Technical Research Center of Fmland. NAUAHYGROS uses discrete size bins, as discussed above. The number of bins, N, can be set up to 110, although N = 30 usually gives sufficient accuracy with manageabic r.s. . DRAFT
A
-{
DRAFT ! computing times.The containment geometry is specified by the user. NAUAHYGROS assumes that the containment is a single compartment in which the atmosphere and the aerosol are well mixed; there are no spatial dependencies in the calculation. The user provides geometric data such - as the containment volume and the surface areas for settling and diffusive depositions. In addition I certain themial hydraulic data must be input as a function of time. These include the gas temperature, wall temperature, steam injection rate, and volumetric leak rate. The modified NAUAHYGROS has calculational options in which additional or alternative input thermal hydraulic data such as heat transfer rates and relative humidity may be used. The aerosol source is defined by the number of aerosol species, the chemical composition and source size distribution of each specie. its injection rate into the containment, and the time interval during which it is injected. Each aerosol specie can be injected continually over one or more time intervals, or can be introduced as an instantaneous (" puff") release at arbitrary times. The particle size distribution function at any time is approximated by a set of N quantities, representing the number of particles at discrete paniele radius values r, (bins), thus forming a size histogram.The first bin corresponds to a chosen minimum radius, while the Nth bin corresponds ' to a chosen maximum radius. The bins are initially spaced logarithmically equidistant,i.e., the ratio r, / r, , is a constant. (This is done because aerosol source size distributions are often log normal or approximately so). Clearly the larger N is, the more accurate the use of discrete size bins will be: on the other hand, computing time increases faster than N (because agglomeration processes usually involve two sizes, i.e., two bins). Codes that use this scheme are called sectional codes. Some aerosol codes have attempted to treat the size distribution as a continuous function characterized by a small numberof parameters whose behavior is calculated as a function of time Although computationally much faster than sectional codes, these have not been as successfi.I i' predicting containment aerosol behavior. The log normal source size distribution for each specie is characterized by a geometric mean radius,
. r,, and a geometric standard deviation, o,, However, as discussed previously, the suspended aerosol size distribution is allowed to assume whatever numerical form evolves from the calculation at each time step.
To determine the best values to be used for r, and o, , data from failed fuel experiments (STEP l-
- 3. PBF-4 LOFT FP-2, and others) were reviewed (Sher,1993). This review yielded best values of r, = 0.25 p m and o, = 1,69. These values were used for the base case calculation. To determine the sensitivity of the results to these parameters, sume runs were done with other values. l The differences in the results are minor, and are shown in Section 4.
q q p-h\hfh
DRAFT in order to calculate the relative humidity at each time step, the steam mass balance is updated, ! l based on the gain in steam mass due to steam injection and its loss due to condensation on particles and/or the wall and leakage. Alternatively, the relative humidity can be input by the user, and this was done in the present calculations, since the relative humidity is provided directly by the MAAP calculations. The condensation rate on particles is determined by the Mason equation, including the effects of hygroscopicity of the panicles. The various removal velocities or rates are calculated using the approaches discussed earlier. These include Brownian and gravitational agglomeration, settling, Brownian diffusion, diffusiophoresis, thermophoresis, and leakage. As noted previously, turbulent agglomeration and diffusion, and certain other removal processes are neglected. The growth of particles due to condensation on them is accounted for in the rate calculations. The code containsinternal main time step control based on the most rapidly varying removal rate. Separate time step control within the main time step is also used in the particle condensation routine, since condensation is often much more rapid than the other processes, and may require smaller time steps within the main time step. 2.2.1 Validation of NAUAHYGROS Validation of NAUAHYGROS has been against the LACE experiments LA-2, LA-4, and LA-6 (Li,1992), which were sponsored by an intemational consortium led by EPRI (Rahn,1988). The 3 tests were performed in a large (850 m ) tank. Hygroscopic (CsOH) and non-hygroscopic (MnO) aerosols were introduced over an approximately one hour period at a rate that resulted in maximum aerosol concentrations of 1-10 g/m 3, similar to expected accident concentrations. Steam was also injected so as to provide saturated or near-saturated conditions. (The steam and aerosol injection protocols and leakage conditions were somewhat different in the three tests). The tests were well instrumented to measure aerosol parameters such as suspended aerosol mass as a function of time, integrated settled, plated, and leaked mass, and size distribution parameters. In addition, extensive thermal hydraulic measurements were made, which provided benchmark and input data for the code calculations. Figure I shows the measured and calculated suspended mass for the three tests. It should be noted - that at late times, when the suspended aerosol concentrations are of the order of two orders of magnitude below their peak values, the calculated and measured concentrations can disagree considerably. However, the impact of this disagreement on integrated aerosol quantities is r.s. - DRAFT
6 wm.Feffetf5 tre.C1Af!CN u.Cn. ua bb# / i'i r jIY 3.J[jg W u ,E88'N8W
.c=. uaCPt. m f!Gt :
1 1 I
.n 1 3b
- E , 's '
4 E 4s
- 1
." !\
g& F gj e6 s g, -
/
- F ,
\ . :
\ * ~
) ee l. s * -
2 1- 1
- a. ** * : -) ,
.l 2 *e
.1
. l a "
K
-se .
3' 9 N . = 1
/
t 3 O" 3 .
..: . 3 .a ..a .
Ti 4 a60 ' ?i 4 s60 mswervT5 t*. Oft.!Lf1CN ad.C'" L..J. 2 LL
, 1
. 1 1
5 i l s a
$n-:. ./
e **
/ ,
n. T - /s
/
3 -f 8
- n. '
7 P . *
*^
a .; i 6.a J yg sie
- g~ p5 A g~ 9?*
l.)g K/dI $ x Test data. Standard deviations on the data points rangefrom 1 5% to 1 60%. Figure 1. Suspended Aerosol Concentrations in the LACE Tests. Page 11
DRAFT negligible. Table I shows the integral data for tests LA-2 and LA-4 (integral data for LA-6 were not taken). It should be emphasized again that calculations that do not properly account for diffusiophoresis and steam condensation on the particles fait quite badly in predicting the acrosol behavior in these experiments. (Thermophoresis was not significant in these tests). On the other hand these effects might not be as important in other situations, depending on the thermal hydmulic conditions. There is a slight inconsistency between the measured suspended mass concentrations in LA-2 shown in Figure 1 and the measured integrated leakage for this test. The leakage derived from the measured suspended mass data agrees better with the NAUAHYGROS results than the measured leakage does. l Table 1 i INTEGRATED SETTLED, PLATED, AND LEAKTO AEROSOL, ! TESTS LA-2 AND LA-4 ) Settled (g) Plated (g) Leaked (g) LA-2. test data 1973(2 10 %) 449( 20%) 1515(* 15 %) l IA-2. calculated 2366 334 1197 LA-4. test data 4490(e 10%) 532(= 20%) 108(e 30%) LA-4. calculated 4437 609 % LA-4. calc. nv/o sicam condensanon s 4811 10 32I w~p $ [g y[ - Page 12
- . ~
prwprw L)K/kf l Further validation comes from the FALCON tests (Benson et al,1990). These tests were performed in a facility that included a containment '.ank whose volume was 0.3 m'. Three tests were proposed for an international code comparison exercise. The results of the code comparison for test ISP-2 have been reported (Williams,1993).This test had high relative humidity ( > 90 7c) and low particle concentration. Its purpose was to test hygroscopic models in the absence of agglomeration. Fourteen aerosol species were injected, including Cs. I Te, non-volatile fission product species such as Mo, Ba, and Sr, and various structural and control rod materials. Aerosol source specifications and thermal hydraulic input data were supplied to the participants. The total measured Cs sedimented mass was about 3600 pg: the NAUAHYGROS calculated value was approximately 4800 pg. The measured wall deposition of Cs was 2600 pg, while the NAUAHYGROS value was 2300 pg. For control rod materials (Ag, In, Cd) the total sedimented , value was about 10000 pg; the NAUAHYGROS value was 14000 pg. The corresponding values for control rod material wall deposition were 6000 g measured and about 5700 pg calculated by NAUAHYGROS. The agreement of the order of 5 - 40 percent in the deposited masses was considerably better than most of the other codes, many of which showed disagreements of factors of two or more. i~ LJi%s%i ~ g l P*6E 13
DRAFT Section 3 METHODOLOGY AND MODELING FOR SBWR CALCULATIONS The goal of this study is to derive an estimate of the S3WR containment aerosol concentration and leakage as functions of time for appropriate accident sequences, and to convert this information to an aerosol removal coefficient, A (hr"). The methodology includes the following elements: Select a reference accident sequence to be considered for release timing and containment thermal hydraulics. Evaluate the timing and thermal hydraulics for the reference sequence, using the General Electric Co. SBWR version of the MAAP eode. Define aerosol source characteristics on the basis of the updated NRC source term, the ALWR Utility Requirements Document, and the reference sequence timing. Calculate the containment aerosol removal and leakage, using the modified NAUAHYGROS code. Perform sensitivity studies on those parameters whose uncertainties are large enough to possibly have significant effects on the results. 3.1 Reference Sequence The parameters that influence natural aerosol removal processes include the containment geometry, the thermal hydraulic conditions in the containment, the characteristics (species, materials properties, size distribution, etc.). of the injected aerosol, and the release rates and release times of the aerosol source. Thermal hydraulic conditions in the containment strongly influence aerosol processes such as diffusiophoresis and condensational growth of the aerosol particles, as described previously. These conditions are plant design and accident sequence dependent. It would be a formidable task to evaluate aerosol behavior in every conceivable accident scenario, so one or a few representative sequences must be chosen. The basis on which these sequences have been chosen is that they represent classes of accidents which are the major contributors to core damage frequency and have significant fission product releases to containment that tend to bound the releases from other sequence types. The SBWR PRA classifies severe accident sequences into several classes whose contributions to Page 14 hgg4. g "
DRAFT core damage frequency are evaluated. From an examination of these categories it can be concluded j that only a few accident sequences need to be considered, since they represent the classes that l comprise more than 90% of the total core damage frequency and tend to produce the most fission i product aerosols in the shortest time. These dominant sequences are listed in Table 2, and consist of the low pressure core melt sequences with either loss of short term coolant makeup or loss of long term coolant makeup. In addition, the bottom drain line break sequences are included in Table 2 because of their relatively fast core damage progression. In accordance with the URD provision for sequence selection, sequences with a core damage frequency greater than approximately 10" per year must be considered in containment performance and source term evaluations. Fmm Table 2 only LPE and MPL are within a factor of two of 10" per year, and of these two sequences, LPE has by far the most rapid core damage progression (and thus the fastest aerosol release). MPE and BDE have slightly faster core damage progression, but their frequencies are more than an order of magnitude below the 10" per year cutoff. On the basis of these considerations, LPE was selected as the reference sequence. For purposes of the licensing source term analysis, the reference sequence was considered to be an in-vessel sequence, i.e., core damage temiinated with the reactor vessel still intact. There are several reasons why this was done: Most core damage events are expected to involve only partial core damage with accident management and recovery (e.g.,TMl-2) and are expected to remain in-vessel. The results of the OECD-NEA TMI-2 Vessel Investigation Project (OECD,1993) which indicate that ALWR design features such as RCS depressurization result in reductions in global vessel temperatures, and consequently tend to lead to in-vessel retention of the core during an accident. Regulatory precedent for degraded core accidents is based on partial core damage Ex-vessel releases are to be considered in the emergency planning dose calculation. DRAFT I l l l Page 15 j 4
DRAFT Table 2 SBWH ACCIDENT SEQUENCES CONSIDERED FOR CONTAINMENT AEROSOL CALCULATIONS Sequence Frequency Time of Core Onset 'of Core Reflood time, (yr) Uncovery, hr Damage, hr hr LPL(low ~2x 10-" 7 8.3 12 pressure, DPVs open, loss of long term makeup) LPE(low ~8x 10 ^" 0.9 1.6 4.5 pressure, DPVs open, loss of short term makeup) MPL(low 4x 10-" 8 9.3 13.5 pressure, DPVs not open, loss of long term makeup) MPE(low ~5x10" 0.8 1.6 4.4 pressure, DPVs not open, loss of short term makeup) BDL(bottom ~1 x 10 ~" 3 4.5 5.5 drain line break, panialinjection) BDE(bottom ~ 1x 10" 0.6 1.4 3.1 drain line break. noinjection) p en rz y ~ r-LAk g,[ Page 16
I DRAFT 3.2 MA AP Thermal Hydraulics The LPE sequence was calculated by MAAP to obtain the required thermal hydraulic data for input to the modified NAUAHYGROS. The required thermal hydraulic data include the containment pressure, relative humidity, and steam condensation and heat removal rates to the containment structure. Figure 2a shows the relative humidity, while Figure 2b shows the heat removal rate. The time variable in the figures represents time after core uncovery. In the SBWR MAAP calculation. the condensation heat transfer rate is essentially the same as the total heat transfer rate since there is little or no sensible heat transfer to the wall. Since MAAP nodalizes the containment into several compartments, but NAUAHYGROS treats it as a single volume, the MAAP outputs had to be suitably chosen. For this purpose, the volume and floor area of the lower drywell were added to those of the upper drywell, so that the lower and upper drywells were combined into one compartment, with uniform mixing of the aerosol assumed to take place. (There are a number of ducts that pmvide communication between the two drywell compartments, but in the accident scenario calculated, the aerosol is released only in the upper drywell. In bottom drain line break sequences, the aerosol would be released into both the upper and lower drywell.) However, the output thermal hydraulic data for only the upper drywell were used. As mentioned previously, the condensation rate calculated by MAAP for the PCCS was used as the condensation rate in the upper drywell for the purpose of determining the diffusiophoretic deposition rate of the aerosol. 3.3 Aerosol Source Definition The aerosol source into the containment results from various processes that can take place during different timeintervals depending upon the type of accident sequence. First. there is some release of fission products to the gap between the fuel and cladding even during normal operation. As the cladding fails during the core degradation phase of an accident, the accumulated fission products in the gap will be released. This is a small contributor to the overall release. The second phase of l release occurs during fuel degradation and melting, before reactor vessel failure. This is referred to : as early in-vessel release. In the event that molten core debris results in reactor vessel lower head failure, there is a third phase that is associated with ex-vessel releases from molten core material on the lower drywell. A fourth phase is late in-vessel release from core material that remains in the vessel after failure, including revolatilization of aerosol deposited earlier on primary cooling system surfaces. v - Page 17 m&rn l4" } I
-u - a
( f DRAFT LPE ' saturation ratio 1 -- 0.9 -- 0.8 -- [
, 0.7 -- / '
- 0.6 --
E 0. 5 - i" 0.4 -- 0.3 -- 0.2 -- 0.1 -- 0 : ; 0 50000 100000 time after core uncovery, sec (a) LPE heat flux 6E+11 --
) SE+11 --
s y 4E+11 -- 5 3E+11 - - 5 c:: 2E+11 1E+11 0 , ; O 50000 100000 i time after core uncovery, sec (b) Figure 2. Base Case Thermal Hydraulic Conditions (a) Relative humidity (b) Heat flux
,..,. DRAFT
DRAFT For the purposes of the SBWR licensing source term analysis, the reference accident sequence is considered to be an in-vessel sequence as discussed in Section 3.1 above. Thus, the aerosol source for the licensing calculation consists of the gap and early in-vessel releases. The source term calculation for the emergency planning dose evaluation will consider the ex-vessel and late in-vessel releases. . To evaluate the gap and early in-vessel release rates for SBWR calculations, the draft NUREG-1465 (Soffer,1992) release durations were used. NUREG-1465 release fractions were used for the volatile fission products. For the nonvolatile fission products, release fractions were taken from Leaver et al (Leaver.1991). Estimated core inventories (end of life after three 2-year cycles) based on ORIGEN calc ulations (Hobbins.1989) are shown in Table 3. The NUREG-1465 release duration for the gap release is one hour, and for the early in-vessel release it is 1.5 hours. The beginning of the gap release corresponds to the time of core uncovery, which, for the SBWR reference sequence, LPE,is 0.9 hours after accident initiation. The non-Dssion product source rates were based on observed ratios of non-fission product to fission product releases in failed fuel experiments and the Three Mile Island accident (EPRl(b),1993). (The sensitivity of the results to the non-fission product release rate is discussed in the following section). Table 4 shows the , release fractions that were used. Table 5 summarizes the input assumptions used in the calculations. p n g e. w . k. l I i Page 19 I
}'50 /( [~T hJ h r~t t a Table 3 ORIGEN CORE INVENTORIES ELEMENT CORE INVENTORY, GRAMS I 1.8 E4 Cs 2.2 E5 Te 3.8 E4 Ba 1.2 E5 Sr 7.1 E4 Ce 2.1 E5 12 9.7 E4 Ru 1.9 E5 '
Sb 23 E3 U (PWR) 7.0 E7 U (BWR) 1.4 E8 Table 4 REL, EASE FRACTIONS GAP RELEASE EARLY IN-VESSEL RELEASE Noble gases 0.05 0.95 Iodine 0.05 0.22 Cesium 0.05 0.25 Tellurium 0.00 0.15 Strontium 0.00 0.01 Badum 0.00 0.01 Ruthenium 0.00 0.0001 l l Lanthanum 0.00 0.0001 l Cedum 0.00 0.0001 Other 0.00 0.0001
-- DRAFT l 1
DRAFT Table 5 INPUT ASSUMPTIONS Gap release begins at 0.9 hours after accident initiation (i.e., at time of core uncovery for sequence LPE) NUREG-1465 gap and early in-vessel release fractions for volatile fission products EPRI early in-vessel release fractions for low- and non-volatile fission products and inert (structural ) materials ReDood assumed prior to the time of vessel failure, as predicted by MAAP Inert-to-fission product ratio =' 1/1 Aerosol well mixed in DW with diffusiophoretic removal based on PCCS condensation rate and sedimentation removal as calculated by NAUAHYGROS (i.e., upward DW gas flow ignored) Containment leak rate = 0.5 vol %/ day In accordance with Table 4, the gap release included CsOH and Csi. The early in-vessel release included these as well as Te, BaO, SrO, CeO 2 , La20, 3 Ru, and Sb as the non-volatile fission-products: and SnO2 , UO2 , and Zr as non-fission product species arising from structural and control rod materials. In the early in-vessel release stage the mass ratio of the inert to fission product species was taken to be unity. The aerosol source rates used in the base case calculation are shown below in Table 6. As mentioned previously, the geometric mean radius and geometric standard deviation of all aerosol sources were the best experimental values of 0.25 y m and 1.69, respectively. It should be noted that aerosol removal and " aging" due to agglomeration in the flow paths between the reactor core and the containment were neglected. These processes will lead to smaller leaked masses than ! calculated.
'~~~~~
E'" [
]
Page 21
i' DPiAFT Table 6 l AEROSOL SOURCE RATES, g/s - l 0-3600 3600-9000 , seconds seconds CsOH 3.i52 6.027 Csl 0.512 1.502 , Te 0.00 0.773 Ba0 0.00 0.0745 Sr0 0.00 0.047 CeO2 0.00 0.00147 b023 0.00 6.32E-4 Ru 0.00 1.%E ' Sb 0.00 1.27E-5 Fe,0, 0.00 8.43 i UO- 0.00- 0.88 Cd 0.00 0.00 Ag 0.00 0.00 Zr 0.00 1.50 i 3.4 Effect of the PCCS ' 3.4.1 !)niform Upward Row Effects on Aerosol Sedimentation The aerosol sedimentation modelin NAUAHYGROS is based on the assumptions that the aerosol is uniformly distributed throughout the volume under consideration (e.g., the upper drywell) and that the gas medium in which the aerosol is suspended does not have a net flow in any particular direction. The uniform distribution assumption is appropriate in severe accident scenarios since L turbulent circulation and/or vortex flow of the gas will tend to promote good mixing. The - assumption that the SBWR containment atmosphere does not have a net flow in any particular direction during a severe accident is also appropriate since. even with steam flow to the PCCS, the upward gas flow is localized and does not significantly affect bulk gas conditions. Furthermore, the turbulent circulation and/or vortex flow of the gas, which promotes mixing, will not result in a net gas flow in any particular direction. DRAR i i
4 DRAFT This can be better understood by dividing the containment volume into three subvolumes, as shown in Figure 3: the dome volume above the reactor vessel (volume 1), the annular volume between the reactor vessel shell and the containment wall (volume 2), and the lateral volume beneath the PCCS (volume 3). There are three PCCS suction inlets that are located 120 degrees apart at the ceiling of volume 3. In volume I the aerosol sedimentation surface is the projected area of the upper head of the reactor vessel;in volume 2. it is the floor surface of the lower drywell. In volume 3,it is the floor surface of the upper drywell. PCCS 1 3 3 reactor vessel 2 2 fewer drywell Figure 3. Containment Schematic. The elevation of the PCCS inlets makes it impossible to have a uniform upward flow pattern in the dome volume (volume 1). In volume 3, a uniform upward gas flow is highly unlikely since the l PCCS suction should create a flow only in the vicinity of the PCCS inlets and in the gas between the inlets and the steam injection points, as streamlines become established between thern. Thus' ! most of the region in the vicinity of the sedimentation surface in volume 3 should not be affected. In volume 2,if the ADS ir activated or there is a main steam line break, steam will be injected into : the containment near the top of the reactor vessel, so the gas flow will not be affected except possibly for a small amount of entrainment. Only if tht accident scenario involves a bottom drain i r.s. ,, DRAFT ,
DRAFT line break in conjunction with ADS failure, so that steam is injected into the containment from the bottom of the reactor vessel and a suction takes place in the PCCS, will a uniform upward gas How be possible in volume 2. However, as discussed above in Section 3.1, bottom drain line accidents are very low frequency. Even in this case only the sedimentation in volurne 2 would be affected in this type of scenario, it is therefore concluded that the effects of an overall upward gas flow induced by PCCS suction are likely to be very small and for purposes of aerosol removal are consistent with the assumption of no net gas flow in any particular direction. 3.4.2 Aerosol Removal in the PCCS and Wetwell Aerosols that are entrained in the gas flow into the PCCS will experience many types of removal processes as they travel through the PCCS to the wetwell. The conditions and systems that are favorable for aerosol removal are the following,in sequential order: the bends in the suction pipe line to each PCCS the diffuser at the upper plenum of each heat exchanger that is used to evenly distribute gas flow rates among all the tubes steam condensation in the PCCS a long discharge pipe for each heat exchanger that discharges uncondensed steam and noncondensible gas into the wetwell throagh spargers at the end of the pipe the suppression pool in the wetwell into which any remaining aerosol is discharged a the wetweilitself The aemsol removal mechanisms in the PCCS and wetwell include diffusiophoresis in the PCCS as a result of the significant steam condensation in the heat exchangers, inertial impaction deposition in the pipe bends and diffusers, turbulent deposition in the pipes, pool scrubbing in the suppression pool, and diffusion and sedimentation in the wetwell. A simple analysis provided in I Appendix A indicates that nearly all of the entrained aerosols will be removed in the PCCS by diffusiophoresis if the ratio of the outflow rate to the inflow rate is small (as is the case heret Page 24
DRAFT Section 4 RESULTS OF SBWR CALCULATIONS 4.I Base Case The accident scenario, timing of key events, and thermal hydraulics of the SBWR base case have been described above. In the NAUAHYGROS calculations, the quantities of primary interest were the integrated 2- and 24-hour (i.e., after accident initiation) leakages from the containment, but many other characteristics of the suspended and deposited aerosols were also determined, most as functions of time up to 24 hours after accident initiation.The containment leak rate was assumed to have its design value, namely 0.5 vol%/ day. The results that follow include the 2-hour and 24 hour total leakages (fission product and non-fission product aerosol species), and the 2-hour and 24-hour leakages of the principal fission products CsOH, Csi, and Te. The 2-hour fission product leakage is calculated in order to support the 2-hour 10CFR100 dose evaluation. The 24-hour leakage is appropriate for the determination of emergency planning dose i and is also a suitable approximation to the 30-day aerosol leakage, since the containment aerosol concentration at 24 hours is so low that aerosol fission product leakage beyond 24 hours is negligible. The organic iodide leakage beyond 24 hours will need to be considered in the 30-day 10CFR 100 dose evaluation since organic iodide is assumed to be not removed in the containment. It is useful to have for reference the leakages in the absence of all natural aerosol removal mechanisms (except leakage).These can easily be hand-calculated given the source and leakage - rates, and are shown as the first entries in Table 7. The second entries are "best value" leakages, which were calculated by NAUAHYGROS taking account of all the applicable removal. mechanisms. It is seen that the natural mechanisms for aerosol removal reduce the 24-hour leaked masses by .a factorof about 20, and reduce the 2-hour leaked masses by a factor of 1.7. The aerosol removal coefficients (i.e., A s) have been calculated based on the integrated leaked and deposited masses, and are shown in Table 8. Average total lambdas and lambdas for sedimentation and plateout (i.e., diffusiophoresis ) were obtained over three time periods: up to two hours after accident initiation, from two hours to 3.4 hours (end of the early in-vessel release), and long term (from 3.4 to 24-hours). As can be seen in Table 8, the total A s are of the order of 0.6 1.8 hr" . As has been mentioned, thermophoretic deposition is not calculated for the SBWR, since all of the heat removal
,g, pp ;
1 r7 {,,,,j h j g { j be
DRAFT from the containment is accounted for by steam condensation, i.e., diffusiophoresis and sedimentation are the only aerosol removal mechanisms present. The method of calculating the average tambdas is given in Appendix B. i Table 7 f BEST VALUE SBWR RESULTS : 4 Total CsOII CsI (g) Te (g) (g) (g) 24-hr. no removal 523- 198 45 19-24-hr. best value 25.65 9.43 2.16 0.93 2- hr. no removal 1.72 1.44 0.24 0.0029 2-hr. best value 1.02 0.84 0.14 0.0028 Table 8 ! AEROSOL REMOVAL RATES (A', hr-1) ' i Time (hr) A(total) A (sed) A (plateout) 0.9 - 2.0 1.83 0.24 1.59 2.0 - 3.4 0.57 0.44 0.13 t
> 3.4 -1.36 1.04 032
-)
4.2 Sensitivity Studies A number of runs were made to determine the sensitivity of the results to various changes in input or modeling parameters and to the timing of events in the accident scenario. 4.2.I ScauenceTimina A calculation was done in which the timing of the aerosol releases was changed to coincide more # DRAFT '
DRAFT elosely with the accident sequence timing calculated by MAAP. In this run the gap release was from 0 to 2520 see (0.7 hr) and the early in-vessel rele.se was from 2520 to 12960 see (3.6 hr) (times are after core uncovery). The release rates were adjusted so that the total release for each aerosol specie in each of the release periods was the same as in the base case. The results, which compare the Table 7 results(NRC) with the results using the MAAP timing,are shown in Table 9, and indicate very little change in the 24-hr leakages for the MAAP release timings. The 2-hr. leakages increase somewht , especially Te, since the early in-vessel release begins at an earlier time. 5 Table 9 ALTERNA7 RELEASE TIMING COMPARISON Total (g) CsOII (g) Csl (g) Te (g) NRC. 2-hr 1.02 0.84 0.14 0.0028 MAAP. 2-hr 1.64 1.11 0.20 0.022 NRC. 24-hr 25.65 9.43 2.16 0.93 MAAP. 24-hr 26.20 ' 47 2.19 0.96 4.2.2 Aerosol Source Size Distribution A series of runs was made in which the aerosol source size distribution parameters r, (geomtric mean particle radius)and o, (geometric standard deviation) were varied. The results are shown in Table 10. The largest effect is due to o, . : l
.i 4.2.3. Effects of Hveroscopicity i The effect of hygroscopicity is indicated in Table 11. In the run labeled "non-hygroscopic" the hygroscopicity model in NAUAHYGROS was turned off.,i.e., all species were assumed to be non-hygroscopic. The differences in leaked masses are quite small. This can be understood by referring to Figure 2a, which shows that the relative 1 umidity is low for most of the time when the aerosol concentration is high, i. e., during and immediately after the source release period.
Hygroscopicity is normally important only when the containment atmosphere relative humidity is high. However, even at later times when the relative humidity does approach saturation, hygroscopicity does not play a significant role, since the aerosol release has terminated and the suspended concentration is very small. Page 27 DRAF,s -
DRAFT Table 10 EFFECT OF SOURCE SIZE DISTRIBUTION (Normalized to Run I) R u r, geom. geom. Total (g) CsOll Csl (g) Te (g) mean standard (g) radius, d e v., r, ( m) o, 1 0.25 1.69 1.00 1.00 1.00 1.00 2 0.25 2.0 0.83 0.83 0.83 0.83 3 0.10 2.0 1.01 1,.01 1.01 1.01 4 0.20 1.69 1.02 1.02 1.02 1.02 Table 11 EFFECT OF IIYGROSCOPICITY , Total (g) CsOII (g) Csl (g) Te (g) base case 25.65 9.43 2.16 0.93 non-hveruscopic 26.71 9.78 2.24 0.97 4.2.4 Non-Fission Product Aerosol Source Rate A set of calculations were performed in which the non-fission product aerosol source rate was increased by factors of Ove and ten. The results are shown in Table 12. In another run, the non-Gssion product source was assumed to be zero: the results are also shown in Table 12. It is seen that increasing the amount ofinert aerosols increases the total leakage, but decreases the fission product leakages, and vice versa. These results are explained by the augmented particle agglomeration rates when there are more inert aerosols, which then enhances the sedimentation ; rates. In any case, the increase in Gssion product leakage when there is no inert aerosol assumed is l not significant. DRAFT 1 i Page 28
DRAFT Table 12 EFFECT OF DIFFERENT NON-FISSION PRODUCT AEROSOL SOURCE TOTAL (g) CsO H (g) Cs! (g) Te (g) Best value (1:1) 25.65 9.43 2.16 0.93 0x non-f.p. source 15.00 11.14 2.56 1.11 5x non-f.p. source 53.66 6.74 1.53 0.64 10x non-f.p. source 77.95 5.50 1.23 l 0.51 4.2.5 Effect of Dry Particle Density A run was done in which the dry particle densities of the non-hygroscopic aerosol species were. reduced to one half of their values. This is often recommended in NAUA calculations in which steam condensation on the particles is not expected to be significant, and is an attempt to take account of the porous stiucture of the particles.The effect was small, resulting in an increase in the leaked masses of about two percent. 4.2.6 Conclusions from Sensitivity Studies The results presented above indicate that both the 2-hr and 24 hour leakages of the principal fission products (CsOH. Csi, and Te) were not strongly affected by assumptions on aerosol source size distribution parameters or accident sequence timing (except for the 2-hr result for Te). Increasing the non-fission product aerosol releases decreased the 24-hour fission product leakages significantly, whereas omitting non-fission product aerosols increased the Gssion product leakages ; slightly, as expected. Assuming that the aerosols were non-hygroscopic led to very minor increases in the leaked masses.This result is also expected in view of the low relative humidity in the containment during and shonly after the aerosol release period, and the low suspended aerosol concentration after reflood, when the relative humidity was high. Reducing the panicle densities also had very little effect on the leaked masses. Other sources of uncertainties are discussed in the following section. Ir is" Page 29
yMW j A r* l Section 5 UNCERTAINTIES l 5.1 Aerosol Source Characterization Uncenainties in the release rates arise from uncertainties in each of the quantities in Tables 3-6. The end results - the 2-hour and 24-hour integrated leakages of each specie - will be roughly , proponional to its assumed release rate. The core inventories shown in Table 3 are close to being upper bounds from the point of view of fuel bumup, and are probably not in error by more than 10 percent.The uncertainties in the Cs, I, and Te release rates are also believed to be of the order of 10-20 percent. For the non-volatile fission products the uncertainties may be considerably larger, but the rates themselves are quite small, as is the impact on offsite dose. 5.2 Thermal Hydraulic Uncertainties in the modified version of NAUAHYGROS that was used for the final calculations in these studies, the thermal hydraulic inputs were containment pressure and the condensation and sensible heat transfer rates from the containment atmosphere to the containment structure, which were supplied by MAAP. The total heat removal from the ecntainment is fixed by the decay energy , released to the containment in the accident and the temperature boundary conditions at the structure. Thus a certain insensitivity of the results to the thermal hydraulic conditions arises from the fact that diffusiophoresis and thermopharesis are complementary mechanisms in the sense that when the thermal hydraulic conditions favor condensation heat transfer to the containment structure rather than sensible heat transfer, diffusiophoretic deposition will dominate, and vice versa. The total phoretic deposition will be fairly invariant. As has been mentioned, in the SBWR MAAP calculations used in this work, basically all of the heat transfer is due to condensation, with very little sensible heat transfer, and thermophoresis was therefore not expected to be important. 5.3 Code Limitations It has already been noted that NAUAHYGROS has some inherent uncertainties due to various approximations made, e.g. the representation of the aerosol source size distribution as a histogram of a limited number of size bins. The approximation that the particles are spherical has also been alluded to earlier. Some other approximations are now discussed. ! The various removal processes are assumed in the code to be algebraically additive: however, in fact they are coupled, and would be additive only if all changes in the number of particles in a given size bin during a time step wem very small compared with the total number of particles in the
- DRAFT
DRAFT bin.The internal time step control in NAUAHYGROS assures that the time steps are small enough that the various source and removal amounts within a time step can be considered additive. Other sources of uncertainty arise from the use of a Gnite number of bins to approximate the aerosol size distribution. An indication of how the results depend on the number of bins, N, can be obtained by comparing calculations with different N's. Calculations of this type in the past have indicated that using N = 30 (as was done in these calculations) gives results that are within a few percent of those using N = 100. It should be noted that individual aerosol species are not kept track ofin an exact manner. To do so would require inordinately large computing times, since in each size bin there would be particles. both primary and agglomerated, of a very large number of compositions. NAUAHYGROS makes the approximation that all particles in a given size bin have the same composition, but the composition varies fmm bin to bin and in time. This insures that if one specie is characterized by small particles and another by large particles, the composition of the smaller size bins will be richer in the Grst specie and that of the larger bins richer in the second. Again, comparison with the LACE tests indicates that this is not a large source of error; the individual specie behaviors are well predicted. Radioactive decay processes in aerosol particles could result in the particles acquiring a net residual electric charge, which could inhibit particle agglomeration. This effect is not modeled in the code. An analysis of this phenomenon (Clement and Harrison.1990) indicates that it is of importance only for large (> 5 m) particles or at low particle concentrations (< 1010/m3 ), in the present calculations the particle concentration is always larger than 1011/m3 , and the average particle radius is about 0.2 to 0.25 p m. Under these circumstances charge effects should be negligible. If, as in the present SBWR calculations, there is no late aerosol source, charge effects might be present, but on the other hand there is little contribution to the 24-hour leakage arising from the period after the cessation of the in-vessel aerosol release.. The uncertainty due to the approximation of a well-mixed containment volume is not believed to be significant, mainly on the basis of the LACE test results. For example, in one of the LACE experiments, two leakage paths were provided at different elevations in the tank, one close to the l point of injection of the aerosol and the other at an ele.aaon considerably higher. The leak rates from both were essentially the same, indicating tt m .he aerosol concentrations were the same at these two heights. However,it should be noted that in the LACE experiments there was minor 4 stratification of the atmosphere, amounting 10 5-10% of the tank volume in a post-accident reactor containment stratification effects are expected to be small, since many mechanisms for mixing
~.m DRAFT
DRAFT exist, including diffusion, circulation, turbulence induced by steam and/or aerosol jets or hydrogen combustion, etc. This is especially true for a passive plant where containment heat removal is by natural convection. Other large scale experiments have also shown that there is good mixing in large open containment volumes (Haschke and Schuck,1988). Other modeling errors are believed to be even less important than the above. Numerical errors and instabilities do not seem to be significant, except when inconsistent thermal hydraulic data are input. Use cf a consistent set of MAAP data in these calculations avoids these sources of error.
' NAUAHYGROS is an aerosol code with externally supplied thermal hydraulic data, not a fully..
integrated thermal hydraulic-aerosol code in which the coupling and feedback effects between the aerosol and the containment atmosphere are accounted for rigorously. In MAAP, the thermal hydraulic calculations do take account of the aerosol, however.The validation of NAUAHYGROS ' against the LACE tests indicates that it can calculate aerosol behavior in post-accident containment conditions satisfactorily if reliable thermal hvdraulic data are available.
,5 [k aos e ( f %[( j7 E
T Page 32
A& rLA g q% j W k Section 6
SUMMARY
AND CONCLUSIONS A methodology for evaluating natural removal of containment aerosols has been developed and has been applied to calculate the integrated 2-hour and 24 hour leaked masses from the containment to the environment for a core melt accident in the SBWR. An evaluation of core melt accident sequences from the SBWR PRA was performed and for purposes of this aerosol removal and leakage calculation, a reference sequence type was selected which is a major contributor to core damage frequency, and which has significant fission product releases to the containment that tend to bound the releases from other sequence types. Aerosol source rates were based on draft NUREG-1465 values for the volatile fission products. and on EPRI reports for other species. Timing of key events, especially core uncovery and reactor pressure vessel failure, were obtained from MAAP-SBWR. a version of the MAAP 3.0B code modified to model the SBWR. The MAAP ealculations also provided the necessary thermal hydraulic data to be input into the aerosol calculation. A modified version of the NAUAHYGROS code was used for the aerosol calculation. The principal natural aerosol removal mechanisms in the containment, including sedimentation and diffusiophoresis, were accounted for in detail. Steam condensation on the particles, including the effect of hygroscopicity was also taken into account.The containment leak rate was taken to be the SBWR design value,0.5 vol%/ day. Total 24 hour leakage (i.e., all aerosol species including fission products and non-fission product materials) and leakages of the volatile fission products of main interest (CsOH. Csi, and Te) were obtained from the NAUAHYGROS calculation. The values are: total,25.7 g: CsOH,9.4 g: Csl, 2.2 g: and Te,0.9 g. These values (and those for the low volatility fission products) are about a factor of 20 smaller than those obtained without crediting natural aerosol removal processes. Two-hour leakages were also obtained, and for these the reduction due to natural removal processes is on the order of a factor of two. The values of the 2-hour leakages are: total,1.0 g; CsOH,0.8 g: Csi, O. I g: Te,3x10~'g. Aerosol removal coefficients were derived for three periods in the accident corresponding to the period from core uncovery to two hours after accident initiation, the interval from two hours to the end of the early in-vessel release (3.4 hours), and the period from 3.4 hours to 24 hours after accident initiation.The totsi removal coefficients averaged over each of these intervals ranged from about 0.6 to 1.8 hr". In the first 2 hour period diffusiophoresis was the dominant removal mechanism; after that sedimentation was more important. Hygroscopicity was of little or no significance. DRbt'>ag Page 33 i
i L%4F M D A I" "3 Sensitivity calculations were performed to assess the effects of varying the aerosol release timing, the aerosol source size distribution, the magnitude of the non-fission product source, and the l densities of the non-hygroscopic aerosol species. The effect of hygroscopicity was also assessed. l None of these variations changed the 24-hr fission product leakages by more than 20 percent. The negligible effect associated with hygroscopicity in this SBWR accident scenario is due to the low relative humidity in the containment during those portions of the accident sequence in which aerosol is being released or in which the suspended aerosol concentration is high. It is concluded that the values of the fission product leakages obtained in these calculations are reasonably robust, for a variety of aerosol release assumptions. It is believed that the results described here tend to be conservative and provide suitable input for 1 10CFR100 dose calculations. The conservatism results from the neglect of several aerosol phenomena, which, ifincluded, would increase aerosol removal and thus reduce leakage. Among the neglected phenomena are turbulent agglomeration and diffusion, aerosol transport and removal effects in the suppression pool and wetwell , and aging of the source aerosol in the primary cooling system prior to its release into the containment. Another source of conservatism in the results arises from the inclusion of the lower drywell volume as part of the single compartment required by NAUAHYGROS. In reality one might expect that the aerosol would be mostly in the upper drywell. The aerosol concentration would thus be higher, leading to increased agglomeration and removal by sedimentation. Calculations using just the upper drywell volume bear this out: the leaked masses are about 10 percent lower than when both volumes are included. I
References:
Benson, C.G., B.R. Bowsher, and M.S.Newland, Report AEEW-M 2612, UKAEA,1990. 1 Bunz, H., M. Koyro, and W. Sch6ck, NA UA Afod4: A Codefor Calculating AerosolBehaviour i in LWR Core Afelt Accidents - Code Description and Users Afanual. Report KfK 355%, Kernforschungszentrum Karlsruhe,1981. i Clement, C.F. and R.G. Harrison, " Electric Charge Effects on Aerosol Behavior" Proc. of l Workshop on Aerosol Behavior and Thermal Hydraulics in the Containment, Fontenay-aux- .\ Roses, Nov. 26-28, 1990, CSNI Report No.176, pp.449-461, OECD,1991. l i EPRl(a), Advanced Light Water Reactor Utility Requirements Document, Rev.6, Vol.lll, Electric Power Research Institute, Palo Alto. December,1993. EPRl(b), letter to N.R.C., J.C. DeVine to L. Soffer, July 30,1993. Fuchs, N.A., The Afechanics ofAerosols Dover Publications,1989. Haschke, D. and W. Sch6ck, "Results of the DEMONA Aerosol Removal Demonstration reu p,rm r-~ L)h lpkfh 1 1
D fi r" ~c
- g#4 g j Experiments", in IVater-Cooled Reactor Aerosol Code Evaluation and Uncertainty Assessment, Proc. of a Workshop held in Brussels, Sept. 9-11,1987, p.88. OECD,1988.
Hobbins, R.R., letters to D. Leaver, Nov.14 and 21,1989. Leaver, D.E., et al, Licensing Design Basis Source Term Updatefor the Evolutionary Advanced Light IVater Reactor, DOE /ID-10298, Sept.1990. Leaver, D.E., et al, Passive ALIVR Source Term, DOEilD-10321, Feb.1991. Li, J., Steam Condensation Effects on Containment Aerosol Behavior in Severe Nuclear Power Plant Accidents, Ph.D Dissertation. Dep't. of Mech. Eng., Stanford University,1992. Mason, B.J., The Physics of Clouds, Clarendon Press,1971. Mcdonald. D., personal communication,1993. OECD-NEA TMi-2 Vessel Inyestigation Project. Calculations to Estimate the Atargin to Failure in the TAfl-2 Vessel. E.G.&G. Idaho, Inc., August,1993. ' Pruppacher, H.R. and J.T.Klett, Aficrophysics ofClouds and Precipitation. D. Reidel Publishing Co., 1978. Rahn, F.J., The LIVR Aerosol Containment Erperiments (LACE) Project. Report EPRI-NP 6094-D, Electric Power Research Institute,1988. Sch6ck, W. et al, Report KfK 3153. Kemforschungszentrum Karlsruhe,1981. Sher, R., unpublished,1993. Sher, R. and J. Jokiniemi, NAUAHYGROS 1.0: A Code for Calculating the Behavior of Hygroscopic and Nonhygroscopic Aerosch in Nuclear Power Plant Contaimnents Following a Severe Accident, EPRI TR-102775, Electric Power Research Institute,1993. Soffer. L. et al, U.S. Nuclear Regulatory Commission (NRC) Draft Report for Comment, NUREG-1465. Manuscript completed and published: June,1992. Waldmann, L. and K.H. Schmitt, "Thermophoresis and Diffusiophoresis of Aerosols",in Aerosol Science. C.N. Davies, ed., Academic Press,1966. Williams, D.A., OECD International Standard Problem Number 34 - Falcon Code Comparison q Report. Report AEA RS 3394, UKAEA,1993. l l 1 a.#it.,/kg f\ "( j l l j Page 35
DRAFT APPENDIX A - AEROSOL REMOVAL IN THE PCCS. We make the following assumptions: uniform condensation rate in the PCCS tube a uniform temperature along the tube constant vapor density uniform velocity at any cross section of the tube The total condensation rate is p G, where p = vapor density. The inlet volumetric flow rate is then fG. where f 21.
- 1. Consider mass conservation analysis for the vapor-fpG > ( Q s p G(f-1) p UA -- -
R +p ( U+dU)A
- p V.,Ddt V = Gl DL p UA = p(U+dU)A + pWrDdx + d/dt(p A dx) (A '\
The last term in equation (A-1)is assumed to be zero, thus dU = - nDIA Vdx = - Gdr/AL (A-2) or U = - Gx/AL + . const. The boundary condition that U(0) = Gf/A leads to U = (GIA)(f-x/L) (A-3) Note that if f = 1. U(L) = 0 (all the steam is condensed).
- 2. Now consider the aerosol mass balance (n = aemsol concentration). We consider a control mass of the aerosol that is bounded by a volume element of the tube of thickness 6 x.This element moves with the flow velocity U. Since U is a function of x, 6 x is also a function of x. The mass balance equation forthe aerosolis:
DRAFT n.n s
~)r y pe.4 er g
.- % j"gid ;s dm V*6A
---e *m (A-4) dt 6V where m is the aerosol mass in the control mass and 6A and 6V are respectively the surface area in the volume element on which the steam is condensing and the volume of the element. ( 6A is equal to :tD6 x; 6V equals n D' dr,where D is the tube diameter). e is the factor within the 4
parentheses in eq. (4) of the main text. Fora pure steam atmosphi.re, t = 1. Equation ( A-4) can be rewritten as: dm V :r
- D *
--e* m (A-5) di A 2
where A = n D is the cross section area of the tube. 4 Under the assumption that the condensation rate is uniform along the tube, integration of eq. (A-5) i yields eG m(t) - m(0)* exp[- ALt ] (A-6) where m(t)is the aerosol mass in the volume at any time t and m(0)is the mass at zero time,i.e., at the inlet .The total time r for the control mass to travel from the inlet to the outlet is obtained from the following equation: G r di dr - UdtA - f L so that dr A
- L
- In 'f\
1 (A-7)
-f G (j _ Q G r f - 1) .
o A_\ LI Thus the mass of aerosol at the outlet is: m( r) - m(0) * 'f - i)" (A-8) r f) The volume that encloses the control mass changes along the tube as a result of the change in 6 x that results from the change in Gow velocity along the tube. We have: ( D3 l,\ ["\ fs) k li
% CX Page 37
4 {.y$$5'Whn p tw d(6x) JU G
- - ar- *6x dt dr AL so that 62f r) [ Gar -
(f - 1) (A-9) 6x(0) - exp\ A
- L f l
The aerosol concentration at the inlet is: m(0) 1 n(0) - i A
- dr(0) while at the outlet it is:
m(0)
- f f-l) '
m(r) ff
~
nlL) - - ( f/ - n(0) *
- 1)(,.o A*6x(r)
A *6x(0) I-l ' I r f) If e - 1, n(L) = n(0), i.e., the aerosol concentrations at the inlet and outlet are equal. However, if f=1, U(L) =0, and there is no flow of the steam or aerosol out of the tube. This corresponds to all of the steam being condensed in the tube, in addition, effects such as a decreasing temperature gmdient along the tube and other aerosol removal mechanisms such as turbulent and inertial - deposition and deposition in bends will tend to insure that little or no aerosol ernerges from the PCCS. DRAFT I
'I i
l Page 38
DD /t CT JA Nf APPENDIX B - CALCULATION OF REMOVAL COEFFICIENTS. We desire to calculate the average removal coefficient, A(hr~'), which will yield the observed leaked mass over a given time interval.The equation used to approximate the suspended mass as a function of time is: dm
- S - Am (B-1) di The solution of eq. (B-1)is:
m(t) - mo e " +A E(1 - e-") (B-2) The leakage over some time interval r is given by : L( r ) = A t m(t)dt - A (m e'" + (1 - e'")dt t o (B-3)
~
where At is the containment leak rate (sec '). In general S and A will be functions of time. We wish to obtain an average A that will give the same integrated leakage from eq. (B-3) as is calculated from the actual behavior of the aerosol,i.e., from the NAUAHYGROS calculation. v Assuming that eqs. (B-1) through (B-3) are valid, the integration of eq.(B-3) is straightforward and yields: L ma S At
-(1 A
- e.u) +ASt -(1 A*
-e u) (B-4)
Knowing the leakage L and source rate S over the time interval r, eq.(B-4) can be solved for the average removal coefficient A in that time. n
!Mn $eg "5I~r !v PaSe 39
{ms u E k h* S $g r =r We are interested in the time intervals from 0(accident initiation) to 2 hours. 2 hours to 3.4 hours (the end of the early in-vessel release), and from then to 24 hours. During the first time interval the actual aerosol source consists of the gap release from 0.9 to 1.9 hours and the in-vessel release from 1.9 to 2 hours. Thus the integrals in eq. (B-4) are split into one period of duration r, = 3600 seconds with a total source rate Si = 3.664 g/s (see Table 4) and a second period of duration r, = 360 seconds with a total source rate S, = 19.2462 g/s. The total leakages in these intervals as obtained from the NAUAHYGROS calculation are 0.804 and 0.22 g, respectively. The values of m ,o the suspended aerosol mass at the beginning of each interval obtained from the NAUAHYGROS calculation are respectively 0 and 3113 g. Using these values in eq. (B-4) gives the result for the total A shown in Table 8. The partial A's for sedimentation and plateout are obtained by multiplying the total A by the fraction of total aerosol removed in the time interval by each of these processes. The A's for the other periods are obtained similarly. It is ofinterest to compare the average total A's with the actualinstantaneous A's obtained fmm 1 dm NAUAHYGROS using thedefinition A = S . Figure B-1 shows a plot of A vstime A m dt ^ ,
\
l 1 (it should be noted that the figure actually plot s - A .)It is apparent that the calculated average A's are consistent with the actual time-dependent A's. 1 DRAFT Page 40
1 DRAFT LPE removal coefficient 0 50000 100000 0 ; ; 2-4
/ F
*- t
, .4_ .
3 -s-- e
* ~ 6 --
time after core uncovery, rec DRAFT i l I
.J Figure B-1. A u time. ,
Page 41
-)
- .-}}