ML20128P101
| ML20128P101 | |
| Person / Time | |
|---|---|
| Issue date: | 09/24/1984 |
| From: | Ryder C NRC OFFICE OF NUCLEAR REGULATORY RESEARCH (RES) |
| To: | Araj K HARVARD UNIV., CAMBRIDGE, MA |
| Shared Package | |
| ML20127A894 | List:
|
| References | |
| FOIA-85-110 NUDOCS 8507130202 | |
| Download: ML20128P101 (1) | |
Text
,/
UN!TED STATES
- 't NUCLEAR REGULATORY COMMISSION g
wasamoros.o. c.nosss g, v.... /
SEP 2 41984 Dr. Kamal Araj Department of Physics Harvard University Cambridge, MA 02138
Dear Dr. Araj:
Dr. Simon Goren asked me for sumaries of the aerosol projects related to the source term analyses. He gave me a list of projects.
I added several other projects to his list.
Enclosed are the sumaries for your files.
5 Sincerely, Vii6 tic! c181160 dy i
Christopher P. Ryder Accident Source Term Program Office Office of Nuclear Regulatory Research
Enclosure:
As stated 8507130202 850415 phj85-110 PDR P
pol
~
g e
THE EFFECTS OF COMBUSTION AND STEAM EXPLOSIONS ON FISSION PRODUCT BERAVIOUR Interim Status Report to the OECD/CSNI Group of Experts on the Source Term (GREST) on TASK 7 by D.A. Powers i
Sandia National Laboratories Albuquerque, New Mexico 87185 U.S.A.
and J.F. van der Vate ECN Petten The Netherlands and D.J. Wren Atomic Energy of Canada Limited Whiteshell Nuclear Research Establishment Pinawa, Manitoba ROE ILO Canada October 1984 d d A c11 h >
m TJUllJlt09 on
I
)
l INTRODUCTION The behaviour of fission products within containment is a complex function of the physical and chemical processes which occur subsequent to an accident. It is recognized that highly energetic processes such as hydrogen combustion and steam explosions may have a significant impact on fission products and consequently the source term. In this paper, we present the status of our knowledge in these areas and a summary of some particular problems where fu~ther work is recommended.
i The organization of this paper is divided into four sections:
I.
The Chemical Ef fects of Hydrogen Combustion II.
The Physical Effects of C,ombustion on Aerosol Particles III. The Formation of Aerosols by Steam Explosions and Malt Ejections IV.
Conclusions and Recommendations.
The first three sections outline the problems and potential impset of combustion and steam explosions on the source term. The last section is a summary of the major areas where the need for further work has been identified.
I.
CHEMICAL EFFECTS Steam explosions and hydrogen flames are two high temperature processes which may lead to chemical changes in the airborne radionuclide inventory in containment. In the case of steam explosions, the hot core debris will be propelled into the containment atmosphere where the high
' particle temperature, large surface ares and plentiful gas reactants will promote reaction. The major consequences of this will be oxidation reactions and a rapid approach to chemical thermodynamic equilibrium. The most important radionuclide affected by this process is ruthenium and the source term implications of ruthenium reactions will be discussed in detail in Section III.l.
~_ - _.-_.-.. -.. _ _ _
i
\\
_2_
Hydrogen combustion, in addition to generating very high tempera-tures at the moving flame front (1000-2200 C), also generates large transient concentrations of reactive radical intermediates:
OH, H and O.
These species may react with airborne radionuclides either in molecular or particulate form, leading to different fission product species. The following sections discuss the projected impact of hydrogen combustion on the source term as a function of the radionuclide chemistry.
j I.1 HALOGENS (I, Br)
The halogens, principally iodine, may exist in the gas phase as molecular species (I, HI or organic iodides) or as part of an aerosol 2
particle. In the latter case, the iodine will be present as an ionic salt, CsI. All of these species are capable of reacting with intermediates generated during hydrogen combustion.
The reactions of I, HI and small organic iodides (such as CH 1) 2 3
with each other and the species present in H /02 combustion are relatively 2
well documented in the chemical literature. Kinetic calculations ( ) have shown that, over a wide range of conditions, organic iodide will rapidly react in flames to release iodine as I, HI or I. The reverse process, which has 2
been studied for methyl iodide (2)
CH4+I2 3I + HI does not occur to any significant extent at the H /CH ratios required to 2
4 support hydrogen combustion in containment due to the competing process CH I + H CH4+I
+
3 The relative quantities of I, L,, and HI in the gas phase will depend on the flame temperature, the fraction of H in the air and the degree to 2
which equilibrium is achieved during passage of the flame front. At high temperatures the I will be completely dissociated but will recombine as the 2
temperature decreases.
The behaviour of iodine on aerosol particles is less clear. High flame temperature will lead to vapourization of some of the cesium iodide salt. Once airborne, the Cs1 may react via CaI + OR CaOH + I leading to formation of airborne I, I and HI.
Work at Sandia has demonstrated 2
that CsI will react, but has not quantified the process (3)
This will be a very difficult problem to solve owing to uncertainties in important variables such as the temperature at the aerosol surface and the rate of CsI vapourization.
Reactions of iodine species adsorbed on surfaces will be the same as the reactions of airborne species. However, the extent of the reaction will be much less easy to predict because of uncertainty in flame temperatures and chemistry near surfaces. Experimental work is required to determine the
~
extent of reaction and vapourization, particularly for adsorbed salts.
The above discussion on aerosol particles assumed dry particles.
For wet aerosol particles, the CaI salt will be dissociated in aqueous solution.
Prior to reaction and release of I r HI, the wet aerosol must be dehydrated,.
2 It is not clear what conditions of aerosol size, flame temperature and flame progation rate are required before.a significant fraction of the CsI will be vapourized and react.
A hydrogen flame in containment will also lead to secondary combus-tion of other material, such as paint and insulation. The products of such combustion will include airborne organic species. These may react with airborne HI and. I to produce organic iodides after the hydrogen deflagration 2
itself is finished. The degree to which this may occur is not known.
I.2 CHALCOCENS (Te, Se)
Tellurium may be present in containment atmospheres as H Te, Te0(OH)2 2
or as part of an aerosol particle (as Te or Te0 ).
Cas phase reactions of the 2
first two species have not been extensively studied and their behaviour in
. flames cannot be predicted.
If thermodynamic equilibrium is achieved during the passage of a flame front, the products of Te combustion will be H Te for f r atm spheres with excess 0 ( }2 atmospheres with excess H and Te0 2
2 2
Tellurium adsorbed on particles and surfaces will undergo the same processes as CsI, vapourization followed by reaction, with the products being again either H Te or TeC.
Since atmospheres with excess H are unlikely, H 2
2 2
2 combustion will have the effect of converting all Te to Te0, an inv latile 2
species which will be subject to condensation and plateout.
Small organic tellurides are also volatile species. However, these will not be formed during H e abustion for the same reasons as for organic 2
iodides. The tellurides will also be unstable with respect to attack by the combustion radicals H and OH.
There is no experimental evidence available to quantify the degree of Te reaction in H combustion and to verify the predictions of 2
thermodynamics.
I.3 RUTHENIUM AND TECHNETIUM Prior to a hydrogen deflagration, these species will be present in containment as part of aerosol particles or adsorbed onto surfaces. Their chemical state will either be as metals or involatile metal oxides Tc0, Ru0 '
2 2
Ru0. While both of these elements may exist as volatile oxides, Tc 02 7 and Ruo4 or Ru0, these species will not be formed during a hydrogen deflagration.
3 All of the higher oxide species are thermodynamically unstable at flame temperatures, particularly in the presence of even a trace amount of hydrogen.
As a consequence, hydrogen combustion will not lead to an increased source term for either of these species.
l The existence of volatile oxides presents a problem if the oxidation occurs at lower temperatures under conditions where hydrogen is not present.
These conditions may be met during steam explosions and the possible conse-quences on ruthenium volatility are discussed further in Section III.l.l.
?
. I.4 ALL OTHER FISSION PRODUCTS The remaining fission products, including the alkaline earths (Ba, Sr), the alkali metals (Cs, Rb), the lanthanides (La, Ce...),
the actinides (U, Pu) and various metals (Mo, Zr), may be considered together. None of these elements form chemical species which will be airborne in containment, except as particulate aerosols. In all cases, the most thermodynamically stable species are oxides and hydroxides. H combustion will only drive 2
reactions to produce these compounds from metals or alloys presect in aerosol particles.
Reactions and high flame temperatures may result in the temporary volatilization of metal oxides or hydroxides from particle surfaces, but these must recondense rapidly on all available surfaces af ter the flame front passes.
II. PHYSICAL EFFECTS A steam explosion or hydrogen combustion will produce a pressure pulse propagated by a shock wave. The high velocity of the particles at the shock front may have an impact on aerosol behaviour via a number of mechanisms.
These are discussed briefly below. In addition, high temperatures at a 4
hydrogen flame front may cause particle vapourization which will also affect I
aerosol behaviour.
II.1 AGGLOMERATION AND DEAGGLOMERIZATION 1
1 Airborne particles will be subject to the competing processes of agglomeration and daagglomerization at the shock front. Enhanced agglomeration will be promoted by the fast movement of particles at the shock front which will increase their collision frequency. Deagglomerization will occur if i
enough energy is transmitted by collisions with fast particles to break the bonds between primary particles and the body of the aggregate.
The two collision processes will not take place simultaneously and usually exclude each other. Desgglomerization is possible only when the
i
! particles absorb more energy than required, e.g. turbulent agglomeration (assuming reasonable sticking coefficients). Agglomeration is the dominant process under normal circumstances and is enhanced for moderately increased energy densities.
Assuming that surface tension forces govern interparticle bonding in aggregates, the transition from coagulation to deagglomeration should occur in the absorbed energy range of 0.1 fJ to 10 pJ per particle, for particles of 0.1 um to 10 Wu, respectively. In practice, however, significant deagglomera-tion during shock wave passage will be followed by fast agglomeration during the later passage of the low energy tail of the' shock wave. By and large, explosions are believed to enhance agglomeration. The dominant agglomeration process in explosions will be turbulent inertial agglomeration. Diffusional coagulation will be of minor importance because all important parameters (particle size, particle size dispersity and degree of turbulence) are sufficiently large to favour turbulent inertial agglomeration ( }.
II.2 DEPOSITION AND RESUSPENSION Particles may be deposited or resuspended from surfaces as a result of a shock, but both will not occur simultaneously except in an intermediate energy regime. Deposition may be promoted by drag of particles by the travelling shock to walls or obstacles. However, deposition due to explosions will be a very inefficient process. Only a small boundary layer adjacent to walls will be affected because of the short-lasting acceleration induced by the explosion wave. The resulting maximum range of the layer is only a few microns. This agrees with experience gained from generating aerosols via l
exploding wires (6) l Resuspension of particle deposits due to kinetic energy absorption by a dust-laden vall can go via two routes:
by direct interaction between shock wave and particles (and the reflected shock wave in the esse of a hard wall);
by elastic absorption of the shock wave energy by the wall and j
partial transfer of this energy to the dust layer.
t
a
, Resuspension by elastic wall absorption is expected to be much more effective than the direct interaction of particles and shock waves; whether this elastic absorption is important, depends on the material and dimensions of the wall.
Resuspension may also result from the blast of an explosion (e.g. the propagation of a steam explosion through the primary heat transport system).
This will be governed by various mechanisms, each of which dominate in three particle diameter regimes with the transition points at approximately 30 um and 200 um(
For source term considerations, interest is mainly in the small particle regime (diameter < 30 um). Larger particles, though possibly having a
~
higher probability for resuspension, are of little interest due to their extremely short airborne lifetimes. Resuspension in the small particle regime is dominated by the forces of adhesion and turbulent lift. The boundary condition for resuspension is given by:
3 log t,
> 4 log d + 22 where t, = wall shear stress (Pa) d = particle diameter (m).
The above equation is for dry particles. In the humid environment of the reactor containment and heat transport system, the particles will be expected to be wet.
The water on the particles will contribute to the surface tension, which will dominate the forces of particle / particle and particle / wall interactions. As a result, agglomeration will be favoured compared to deagglomeration and resuspension will be considerably hampered.
II.3 PARTICLE CHARACTERISTIC CHANGES Explosions may affect the physical characteristics of aerosol particles via two mechanisms. The first nechanism is mechanical impaction.
Fluffy aggregate particles may be compacted due to increased turbulence and more energetic turbulence. The impact of this process would depend largely on the degree of water condensation on the aerosol particles and the shape of the
~
particles prior to an explosion.
l
. The second mechanism is direct heating of the particles via energy absorption from shock waves or a hydrogen combustion.
This will lead to substantial vapourization of (at least) water from the particles, but may also vapourize chemical species containing Cs, I, Te and Ru The vapourization, if sufficiently great, may alter the size distribution of the aerosol or a particular component of the aerosols (e.g. particles comprised mainly of CsI).
t Following evaporation, the chemical species may react as discussed in Section I or simply recondense. The recondensation occurs onto the particle size fraction having the main surface area and onto walls. The process of evapora-tion is a complicated function of aerosol / shock wave-combustion interaction, vapour material properties, aerosol size distribution and concentration, and the overall spacial dimensions of the system.
III. AEROSOL GENERATION A steam explosion is expected to occur when molten debris from a reactor core streams into a water pool beneath the reactor. One consequence of the resulting energetic event is the production of large quantities of aerosol particles. There is also a second mechanism for aerosol production via an " explosive" event. This is ejection of molten core debris from a pressurized reactor vessel. Both of these processes may have a substantial impact on the source term.
III.1 STEAM EXPLOSIONS The process of a steam explosion begins with coarse fragmentation of the core melt as it enters the water pool. The melt initially fragments into 1-2 cm droplets which are surrounded by a film of boiling water (and hydrogen if the material is reactive). There is some debate on the degree of initial fragmentation with estimates varying by 2-3 orders of magnitude (').
At some point the gas envelopes around a droplet may collapse permitting intimate melt / water contact. The result is a sudden burst of steam which creates a shock wave. The conditions and forces required to cause the gas envelope to collapse are not well established. The initial shock wave may
collapse other gas envelopes leading to reinforcement as the shock wave propa-gates. The coherent propagation of the shock wave is also not well understood.
The sudden cooling of a coarse fragment upon gas envelope collapse will cause it to fragment into finer particles, further accelerating the cooling process.. The explosion concludes with the expansion of the steam, which lofts finely divided particles and water into the atmosphere.
Molten material adjacent to the explosion site but not in the water
~
may also be propelled into the atmosphere by the force of the explosion. This melt will fragment as a result of the sudden acceleration (10),
Experiments have been conducted at Sandia to determine the nature of the aerosols produced during steam explosions. The data show that, in comparison to the dimensions typical of aerosols, the particles formed by steam explosions are very coarse. Mean sizes are greater than 100 um.
Seldom is more than 1%, by weight, of the material of a size less than 1 um.
These results seem independent of melt composition. The above results have been obtained from tests with molten iron, iron oxide, alumina and uranium dioxide.
The aerosol data have been obtained from experiments with relatively small melt masses (< 25 kg) and relatively inefficient explosions. The efficiencies of thermal to mechanical energy conversion were typically < 1-2%,
though efficiencies as high as 10% were observed for single Fe0 droplets.
There is a question as to whether coarse fragmentation would still occur if the explosions approached the thermodynamic limit of e 30% efficiency.
Careful examination of particle size data suggests that the material has a multi-modal distribution. As the efficiency of the explosion increases, the population of a mode at 150-200 Um increases, but the mean size of this mode does not decrease. This suggests that even for very efficient explosions, the extent of aerosol production (i.e. particles with sizes < 20 um) will still be small.
A second source of aerosol production is water droplets formed from excess water ejected along with the core debris into the atmosphere. The size distribution of water droplets produced during stesa explosions has not been i
]
reported. There are two possible consequences of such droplet production. If very fine droplets are produced, they will participate in agglomeration with solid particles accelerating the sedimentation process. If coarse droplets are produced, these should behave like spray droplets in sweeping particles from the atmosphere. In either case, water droplets produced during steam explosions should attenuate the inventory of suspended radioactivity in the containment atmosphere.
i f
III.l.1 Chemical Processes During Steam Explosions
~
r It has been recognized in the Reactor Safety Study
, that during f
the course of a steam explosion, volatile fission products may be released from the melt fragments via vapourization and reactions with the containment atmosphere. It has. been believed that the latter mechanism would control-release of volatile molecular species. Experiments at ORNL have demonstrated quantitative release of noble gases, halogens and alkali metals, extensive release of ruthenium and tellurium and very limited release of barium and O2) strontium from irradiated fuel fragments heated in air at 500-1200 C Based on these data, release predictions were made in.the Reactor Safety Study. Despite intensive research into steam explosions, there have been no attempts to validate the release estimates made in the Reactor Safety Study.
i The predicted release of noble gas, halogens and alkali metals is inconsequential as these volatile species will have all completely escaped the fuel by the time it becomes molten and hence prior to any steam explosion. The.
release of tellurium is possibly significant, but other phases of a severe accident will result in a near quantitative tellurium release whether or not a steam explosion occurs. Hence, a steam explosion may have only a minor impact on the timing of tellurium release. However, the prediction of extensive release of ruthenium is a significant addition to the source term.
Since the addition of significint quantities of ruchenium to the source term is the only important addition to the source term, the following discussion focuses on its behaviour. There are three points at which chemical processes may cause radionuclide relesse:
l 11 -
during coarse mixing of the debris when the melt is surrounded by steam; during gas film collapse when water comes into intimate contact with the core debris; during the flight of fine debris particles through the atmosphere j
following an explosion.
Ruthenium is present in irradiated fuel as a metal inclusion alloyed withPd,Mo,Th(
Ruthenium metal has a low vapour pressure, and extensive release of ruthenium will occur only if the element forms volatile ruthenium oxides. At moderate temperatures ruthenium may be quite volatile due to the formation of the oxides rug 4 and rug.
The stability of these oxides relative 3
to the much less volatile lower oxides, rug and RuO, will be controlled by the 2
steam / hydrogen gas ratio and is reduced substantially when even minor amounts of hydrogen are present in the gas. This point has already been made in the previous discussion on hydrogen combustion.
The depression is the volatility of ruthenium in steam / hydrogen mixtures relative to the volatility in air suggests two things:
ruthenium release as a result of an in-vessel steam explosion that does not cause massive rupture of the vessel ought to be less than predicted in the Reactor Safety Study.
ruthenium release ought to be small during the coarse mixing and gas film collapse stages of a steam explosion. It appears that high release of ruthenium is possible only during the debris flight through air.
In the air environment, the release of ruthenium will be controlled,
9 by kinetics rather than thermodynamics. The kinetics observed in the ORNL l
experiments may not be applicable to particles in the air. These tests dealt j
with somewhat porous fuel containing ruthenium probably present as metallic inclusions (
)
The porosity of the fuel was probably increased as a result of oxidation of the UO2 matrix t U02 7 and U 0, increasing the rate of oxygen 38 access to the ruthenium alloy.
/
a e-~,m,-
.-,a w
-~~--~,--nnw,,----wr.r
-,w--
w---
,_,-m,,-,g-----------r-r--,-,v y
-~--+,m--
r--
~
w w
l,
Fuel participating in an explosion will have melted and porosity present in the fuel pellets will have been lost. Melting also allows the ruthenium to partition into a metallic phase which will have low porosity Thus, release mechanisms available in the ORNL experiments would not be available for ruthenium release following a steam explosion.
The chemical kinetics observed in the ORNL experiments when extra-palated to higher temperatures are very fast and ruthenium release from very hot debris might be controlled by the availability of reactants rather than the reaction rates when reactants are present. The availabil'ity.of oxygen to form the volatile ruthenium oxides might be the race-limiting process.
Correlations for the mass transport of oxygen to the surface of a single aerosol particle are available(15) m t
rate h the fond frw dN 2
g=45 r K P g 0 2 where N
moles of Ru released
=
particle radius r
=
P02 Xygen partial pressure in the atmosphere K
mass transport coefficient for oxygen.
=
g Integration of this equation requires knowledge of the particle size distribu-tion and airborne residence time. As discussed above, the particles are coarse and it is not immediately obvious that the residence times of these particles (typically a few seconds) would be sufficient to achieve nearly quantitative ruchenium release. Furthermore, the possible presence of other reactive materials such as Zr and steel may substantially reduce the local oxygen partial pressure via competitive oxidation processes.
Oxygen transport to debris expelled into the atmosphere by an explo-i sion may not be as efficient as would be estimated for single particles. Upon initial expulsion, debris and water droplets form a cloud. Oxygen transport to particles within the cloud is controlled by the entrainment of atmospheric gases into the cloud, a process that is most inefficient
13 -
III.2 HICH PRESSURE MELT EJECTION It has been recognized in reactor analyses that there is a possibil-ity of a core meltdown in a pressurized reactor vessel (
In such an accident sequence, the molten core debris is predicted to be forcibly ejected from the vessel. Early analyses of this event predicted that the core debris would disperse across the reactor containment floor where it could be quenched and permanently cooled. This was thought to reduce both gas and aerosol generation by melt / concrete interactions, and hence to significantly reduce both the inventory of airborne radionuclides and the probability of containment failure via either a hydrogen deflagration or long-term over-pressurization.
While experiments have confirmed that debris is dispersed nearly quantitatively from scale models of reactor cavities, two additional phenomena have been observed. The dispersed debris were found to directly heat and pressurize the atmosphere and hence raise the possibility of early over-pressure failure of the containment. Secondly, intense aerosol generation was observed to accompany the pressurized melt ejection (
This is an aerosol source term which has not been considered in previous, supposedly bounding, reactor accident risk studies.
The behaviour of molten core debris expelled from a pressurized vessel is poorly understood and an area of active investigation. The following is a brief discussion of some of the very recent experimental results obtained to date and the questions and uncertainties which they raise.
Within the United States, two groups have investigated the problem.
A summary of their work is pre'sented in Table 1.
The SPIT and HIPS tests have been instrumented for aerosol detection and characterization. Results of SPIT tests with N2 pressurization yielded intense production of trimodal aerosols with modes at 0.5, 5 and e 50 um(19)
The last mode is probably due to a tail of the size distribution of macroscopic droplets. The finest aerosol material consists of sgglomerates of yet finer particles (0.1 - 0.05 um diameter). The
'l mode at 5-8 ma consists of spherical particles as would be expected if mechani-cal processes caused the expelled jet of molten material to disintegrate into liquid droplets.
. Tests with CO2 pressurization yielded similar results except for an
~
absence of the 5-8 um mode for the aerosol. The current hypothesis is that this difference arises from a high solubility for N versus a low solubility 2
for C0 in the melt and evidence from X-ray analysis of the melt jets appears to support this.
Analysis of the bulk debris on the cavity floor has found it to be dispersed as macroscopic droplets rather than as a coherent liquid layer..The mean sizes of the droplets were 0.43 and 0.76 mm.
The aerosol generation could occur at three points in the ejection process:
during emergence of melt from the reactor vessel, from the melt jet or as a result of melt dispersal from the
- cavity, from debris expelled from the cavity interacting with the containment atmosphere.
Modelt for aerosol emergence during melt emergence include a coherent melt 9*.rcash followed by pressurized gas, and a melt stream penetrated by and mixed with pressurized gas.- Study of the latter model has shown that a process called pneumatic atomization can lead to production of very fine par:icles( ).
Models can be formulated to predict when the pressurized gas will penetrate the melt and the droplet sizes (
However, while criteria for entraining melt in a gas exist, there ia currently no model for the rate of entrainment which is required to predict the magnitude of aerosol generation.
The multimodal character of aerosols observed in melt ejection tests suggests that several mechanisms may be responsible for their formation. The finest agglomerated aerosols may be formed by vapourization of the melt followed by subsequent condensation. The limited composition data available for the aerosols are consistent with this hypothesis. The fine aerosol would then be enriched in the most volatile species.
Scaling of test results for
~
this process is difficult because of uncertainty in melt geometry.
I J
2
. The possibility of mechanical aerosol generation by gas effervescing has been mentioned above..
Powers has shown that H and H O would be quite 2
soluble in core melts with the pressurized reactor vessel. The volumetric expansions after ejection are predicted to be large enough to cause substantial disruption of metallic melts. Pilch and Fried (
have formulated a model of melt jet disintegration by gas' effervescence. The particle size is controlled by bubble nucleation and growth with the stability criterion y,, h
-2 12e d
=
2 pV y
ag 3
where d
maximum particle diameter
=
max Vd liquid particle velocity
=
V gas velocity
=
liquid surface tension a
=
gas density.
o
=
g Gas effervescence drastically increases the melt surface area and presumably the rate of vapourization of volatiles from the melt.
Entrained melt droplets will also be subject to disintegration during
" blowdown" of the pressurized gas. Quantitative analyses of the disintegration process shows that maximum stable droplet sizes are much larger than sizes
~
i typically associated with aerosols.
The dispersed molten core debris will be subject to chemical reaction 4
with the ~ containment atmosphere in a similar fashion to aerosol material generated by a steam explosion. As described in Section III.l.2, the most important factor is oxidation and the rate controlling mass transport of oxygen to the particle surfaces. In addition to chemical release of volatile species such as RuO or h0, oxidation of large metal droplets can lead to disintegra-4 3
tion into fine aerosol particles. Comprehensive models or definitive data on chemical reactions of debris droplets lofted from the cavity do not exist.
In conclusion, we would note that there is still uncertainty over whether in-vessel core degradation will lead to the initial conditions
l
]
-.16 -
4 I
necessary for high pressure melt ejection. These conditions are a high internal pressure and a large quantity of molten debris. Modelling of core degradation is an evolving area where more scrutiny is required. It is clear that if a pressurized melt ejection does occur, it could lead to very high radioactive aerosol concentrations and a large source term if accompanied by an over pressure containment failure.
IV.
CDNCLUSIONS AND RECOMMENDATIONS Our current level of knowledge and ability to predict the conse-quences of explosions on the source term is clearly limited. Work has just begun in some areas which have only recently been recognized as important.
The following is a brief summary of some of the conclusions which can be made to date and also some recommendations of work in areas where knowledge is particularly lacking.
IV.1 CHEMICAL EFFECTS 1.
Hydrogen combustion vill only have a significant chemical impact on the halogens and chalcogens.
2.
Organic iodides will be destroyed by a hydrogen flame, but may be regenerated by secondary reactions involving the combustion of containment materials. Experiments are required to determine the importance of the latter process.
3.
Hydrogen combustion will convert Cs1 in aerosol particles to airborne HI and I. Experiments a,re required to quantify this process.
2 4.
Experiments are required to assess the impact of hydrogen combustion on wet CsI aerosols.
5.
Experiments are required to verify that no volatile tellurides will survive a hydrogen combustion.
. IV.2 PHYSICAL EFFECTS 1.
Explvsions will promote agglomeration of aerosol particles via turbulent inertial agglomeration.
2.
Particle deposition caused by explosions will be a minor effect.
I 3.
Resuspension of particle deposits is partially understood for dry particles, but is more complicated for wet particles. More work in this area is required.
i 4.
Particle vapourization may lead to changes in aerosol size distribution and also the chemical form of volatile species. This phenomenon is very complex and requires further work to assess its impact.
IV.3 AEROSOL CENERATION l.
Steam explosions are not likely to generate large quantities of aerosol material.
2.
The expulsion of even large masses of debris into the atmosphere may not drastically affect aerosol inventories since the ability of coarse particles to remain airborne and to scavenge particles of aerosol proportions is limited.
3.
Water droplets produced by steam explosions will attenuate the inventory of suspended radioactivity in the containment atmosphere. Work is required to measure the size distribution of the water aerosol to quantify this effect.
4.
Steam explosions may lead to an increase in the quantity of airborne ruthenium. While this is expected to be low, experiments are required for confirmation.
5.
Pressurized melt ejection is a potential source of large quantities of aerosol material. Models of core slump behaviour must be improved to
O assess the probability that conditions may exist which can lead to in-vessel high pressure failure.
6.
Experiments and improved models are required to predict the magnitude of aerosol generation during pressurized melt ejection as a function of vessel pressure, salt temperature and melt composition.
7.
More work is requuired to understand the effects of reactor cavity geometry and containment geometry on aerosol production during pressurized melt ejection.
~
8.
Models of the chemical reactions of airborne molten debris with the ambient containment atmosphere leading to aerosol formation and release of volatile. species are inadequate.
REFERENCES 1.
D.J. Wren, "The Behaviour of CH I in H,/0, Combustion", review paper 3
submitted to OECD /NEA-CNI-GREST, SIND0C (83) 240 (1983).
'S.P, Pardini and D.S. Martin, Int. J. Chem. Kinetics 15, 1031 (1983).
b.
D.M. Golden, R. Walsh and S.W. Benson, J. Am. Chem. SE. 8_7, 4053 (1965).
3.
L.L. Nelson,.".The Impact of Hydrogen Burns on Fission Product Release from LWR Containments During Severe Accidents", Sandia National Laboratories.
4.
F. Garisto, " Thermodynamics of Iodine, Cesium and Tellurium in the Primary Best Transport Systen Under Accident Conditions", Atomic Energy of Canada Limited _ Report, AECL-7782 (1982).
5.
H.P. Pruppacher and J.D. 'Klect, Microphysics of Clouds and Precipitation, D. Reidel, Dordrecht, 1978.
6.
J.F. van der Vate, unpublished results.
7.
M. Phillips, J. Phys. D. Appl. Phys. 13,, 221 (1980).
8a.
L.S. Nelson and M. Berman, " Mitigation of Damaging Effects of Hydrogen Combustion in Nuclose Power Plants", Proc. U.S. NRC lith Water Reactor Safety Research Information Heating, Caithersburg, NUREG/CP-0048 (1983).
b.
C.D. Andriesse and J.F. Van der Vate, "The Ef fects of Hydrogen Combustion and Steam Explosions on the Physico-Chemical Form of Fission Products and Their Redistribution", OECD/NEA-CNI-GREST, SINDOC (83) 240 (1983).
v
, 9a.
M.L. Corradini and G. A. Moses, "A Dynamic Model for Fuel-Coolant Mixing",
Proc. Int. Mtg on LWR Severe Accident Evaluation, paper TS 6.3, Cambridge, Mass. (1983).
b.
H.K. Fauske, R.E. Henry and M.N. Hutcherson, " Interpretation of Large Scale Vapour Explosions with Application to Light Water Accidents", ibid, paper TS 6.5.
D.H. Cho et al., " Mixing Considerations for Large-Mass Energetic Fuel-c.
Coolant Interactions", Proc. ANS/ ENS Fast Reactor Safety Mts, Chicago, l
Ill., CONF-761001 (1976).
10.
M. Pilch et al., " Acceleration Induced Fragmentation of Liquid Drops",
NUREG/CR-224 (1981).
11.
USNRC Reactor Safety Study: An Assessment of Accident Risks in U.S.
^l Commercial Nuclear Power Plants, WASH 1400, NUREG 75/04 (1975).
t 12a. G.W. Parker et al., Oak Ridge National Laboratory Report, ORNL-3176 i
(1961).
- b. G.W. Parker et al., Oak Ridge National Laboratory Report, ORNL-3319 (1961).
13.
D.R. Olander, Fundamental Aspects of Nuclear Reactor Fuel Elements, UC 796, TID 2671, L975.
14a. G.W. Parker, G.E. Creek and A.L. Sutton, Jr., " Influence of Variable Physical Process Assumptions on Core Melt Aerosol Release", Proc. Int.
Meg on Thermal Reactor Saf ety, Chicago, Ill., NUREG/CP-0027, Vol. 2, p.1078 (1983).
- b. J. Fischer, J.D. Schilb and M.G. Chasanov, J. Nucl. Material 48, 233 (1973).
15.
R. Clif t, J.R. Grace and M.E. Weber, Bubbles. Drops and Particles, Academic Press, New York,1978.
~
16.
M.S. Plessel and C.G. Whipple, Physic of Fluids E,1 (1974).
- 17. Commonwealth Edison Co., Zion Nuclear Power Plant Units 1 and 2 Probabilistic Safety Study, Chicago, Ill.,1981.
18.
J.E. Brockmann and W.W. Tarbell, " Aerosol Source Ters in High Pressure Melt Ejection", Proc. Nucl. Thermal Hydraulics Mts, San Francisco,1983.
19.
W.W. Tarbell, J.E. Brockmann and M. Pilch, "High Pressure Melt Stressing Program Plan", Sandia National Laboratories Report, SAND 82-2477 (1982).
20.
M. Pilch, " Advanced Reector Safety Quarterly Report", Mar.1982, Sandia National Laboratories, NUREC/CR-2679,' SAND 82-0904, August 1983.
21.
D.F. Gluck, J.P. Gille, E.E. Zukoski and D.J. Simkin, J. Spacecraf t 3, 1691 (1966).
22.
D. A. Powers, "The Solubility of Cases in Reactor Core Melts". Sandia National Laboratories.
M. Pilch and W. Tri[d, " Disruption of a Melt Jet by Gas Ef fervescence",
23.
unpublished.
e TABLE 1 SUleERY OF PRESSdRIZED ELT EJECTION EXPERIENTS TEST SCALE MELT PgESSURE REMARKS CWT 1 1:35 U02 and stainless 40 Atas.
apparatus connected to a steel about 2 kg of Argon containment model purged with ejected into a argon refractory lined steel cavity SPIT 1-14 1:20 55"/o Fe 13-170 Atas.
study of aerosol generation and 45"/o A1 0 of CO or N heat fluxes from expelled uelt 2.3toIb.3kg 2
2 SPIT 18-19 1:20 55"/o Fe; 45"/o At 0 105 Atas.
dispersal og melt from cavity 8
10.2kgintoscalek3 of Nitrogen into a 45 m, air-filled El cavity of A1 0 and building 23 e
concrete HIPS 2C 1:10 55"/o Fe; 45"/o A1 0 105 Atas.
dispersal of melt from a 23 80 kg into a concrete of Nitrogen cavity cavity HIPS 4W 1:10 55"/o Fe; 45"/o A1 0 100 Atas.
effect of water in cavity on 23 80 kg into a concrete of Nitrogen dispersal cavity filled with water
,i
,