State-of-the-Art Report on Nuclear Aerosols

Unclassified NEA/CSNI/R(2009)5

Organisation de Coopération et de Développement Économiques

Organisation for Economic Co-operation and Development

17-Dec-2009

___________________________________________________________________________________________

English text only

NUCLEAR ENERGY AGENCY

COMMITTEE ON THE SAFETY OF NUCLEAR INSTALLATIONS

STATE-OF-THE-ART REPORT ON NUCLEAR AEROSOLS

JT03276300

Document complet disponible sur OLIS dans son format d'origine

Complete document available on OLIS in its original format

NEA/CSNI/R(2009)5

Unclassified

English text only

NEA/CSNI/R(2009)5

THE GROUP HAVING PREPARED THIS SOAR CONSISTED OF

Hans-Josef Allelein GRS, Germany

(Chairman)

Ari Auvinen VTT, Finland

Joanne Ball AECL, Canada

Salih Güntay PSI, Switzerland

Luis Enrique Herranz CIEMAT, Spain

Akihide Hidaka JAEA, Japan

Alain V. Jones formerly JRC Ispra, EC

Martin Kissane IRSN, France

Dana Powers SNL, USA

Gunter Weber GRS, Germany

NEA/CSNI/R(2009)5

ANNOTATION ON THE REFERENCES

The reader will find a list of references at the end of each sub-chapter. The advantage of this is close

connection between written text and the quoted reference, the disadvantage is that some references are

listed more than once in the whole report.

The authors hope the advantage will outweigh the disadvantage.

NEA/CSNI/R(2009)5

EXECUTIVE SUMMARY

Background

Nuclear aerosol investigations began in the late 1960s and early 1970s. The progress made both in

experimental studies and in code development is attested to by the publication of three CSNI-sponsored

State-of-the-Art Reports (SOAR) on nuclear aerosols since 1979. Initially, fast reactor safety was the main

concern of the work, as reflected in the content of the first aerosol SOAR. The TMI accident in 1979

motivated an interest in LWR source terms and resulted in the production of a supplement to the first

SOAR which concentrated on LWR aerosol issues. In contrast to the fast reactor situation, the large

quantity of steam present in the LWR containment atmosphere following an accident tends to make aerosol

particles more compact, reducing the modelling problem of defining shape factors. On the other hand, two

new effects had to be modelled: deposition due to steam condensation onto surfaces (diffusiophoresis) and

particle growth by steam condensation onto the particles themselves.

The second SOAR dealt with primary-system fission-product release and transport. This SOAR included a

survey of the models available in the literature for homogeneous and heterogeneous nucleation, for vapor

condensation on aerosols, and for aerosol agglomeration, and noted that these and other models had been

incorporated in circuit codes.

The last workshop on aerosols organised by the OECD was held in Cologne in June 1998. One

recommendation made at this workshop was to prepare a SOAR on aerosol behaviour in both the primary

circuit and in containment.

Objectives

Following the recommendation of the Cologne workshop, the CSNI mandated the preparation of a SOAR

with four main objectives:

• To assess the status of existing experimental data and analytical capabilities required for predicting

aerosol source terms from LWR accidents

• To address the strengths and weaknesses of the codes used to predict aerosol behaviour in the

reactor coolant circuit and containment, and provide validation results and uncertainties, in

particular for plant applications,

• To summarise findings from recent ISP exercises on nuclear aerosols and related thermal-hydraulic

behaviour in the reactor coolant circuit and containment.

• To identify safety-relevant pending issues and to propose research activities that will reduce gaps in

the understanding of relevant processes and deficiencies in modelling capabilities.

Scope

The Nuclear Aerosol Writing Group met several times over a three-year period from 2003 to 2006 to

prepare the current SOAR, which provides a survey of issues surrounding the prediction of nuclear aerosol

behaviour in the primary system and containment. It should be noted that the subject matter of the report is

NEA/CSNI/R(2009)5

limited to nuclear aerosols and does not extend to other LWR accident source term topics that do not have

a direct bearing on aerosol behaviour.

The structure of this SOAR is in line with the objectives listed above. The introduction is followed by a

background chapter discussing the fundamentals of aerosol behaviour with bibliography of the pertinent

literature. Chapter 3 is devoted to the topics of aerosol formation/growth, transport, and retention and

release. Chapter 4 provides descriptions of codes for modelling circuit and containment aerosol behaviour,

and Chapter 5 describes the experiments used for their validation. Chapters 6 and 7 provide examples of

validation results, and plant applications of source term calculations. Chapter 8 summarises the findings of

the Cologne Workshop and progress made since, and provides recommendations for future work.

Results and their significance

The theoretical, experimental and modelling studies presented in this SOAR summarise the status and

current understanding of a wide range of nuclear aerosol topics. The significance of these results is that

they allow for the identification of existing gaps in code capability and experimental data that prevent

accurate predictions of the source term. A general conclusion from this document is that there still exist a

number of items for which additional work is required. However, the status of aerosol codes and

experimental data-bases has improved substantively since the publication of previous SOARS.

At the time when the 1994 SOAR on circuit aerosols was written, plant analysis codes, (as distinct from

special-purpose research tools) were relatively crude, and their validation base was largely confined to tests

using simulant materials. Now, while these codes still use lumped-parameter descriptions of the thermal

hydraulics that can place some constraints on the accuracy of their predictions, they include a more

complete calculation of gas-phase chemistry, interactions with particles and surfaces in the primary circuit,

and a better treatment of coupling between aerosol dynamics and thermal hydraulics in the containment.

Models are also available for aerosol interactions with engineered safety features such as suppression pools

in BWRs, and complex structures such as steam generators in PWRs.

The experimental database for development and validation of aerosol codes has benefited from more

accurate and extensive instrumentation to measure a wider range of phenomena (e.g., the ability to measure

airborne water). Numerous analytical tests have been performed, and are continuing, to address specific

phenomena such as resuspension, revaporisation, and trapping in complex structures. Large-and

small-scale containment aerosol experiments have also been performed and analysed in international

programmes, notably the KAEVER and VANAM tests. Perhaps the most significant advance on the

experimental side is the availability of integral experiments in the PHÉBUS facility on fission product and

structural material release and transport. These tests demonstrate the complex inter-linkage between

different phenomena and have refined our understanding of nuclear aerosols, particularly in the circuit,

where we are now able to refer to measured particle sizes and compositions. PHÉBUS has also provided

information on specific phenomena such as revaporisation. The PHÉBUS integral experiments are now

complete but post-test analyses and interpretation of the results continues. Data from aerosol-related

experimental programmes such as ARTIST and THAI also still require analysis.

A large number of Probabilistic Safety Analyses (PSA2) plant studies have been performed around the

world, frequently involving aspects of aerosol behaviour. This report provides some examples, including

sensitivity studies that demonstrate the impact of aerosol-related processes, however few such analyses

have been published. The aerosol community is therefore not always fully briefed on the risk-relevance of

the numerous phenomena currently under study. The question of risk significance deserves a more

concerted effort, in aerosol physics as in other relevant fields. This will require additional probabilistic

analysis for new plant designs and evaluation of their sensitivity to various models and parameters.

NEA/CSNI/R(2009)5

It is expected that increasingly stringent safety standards and new power reactor designs will generate

aerosol-related safety questions that will require experimental capabilities, analytical tools, and the

expertise to use both intelligently. Aerosol-related computational and modelling capability is now being

utilised to address industrial problems, and major environmental issues such as climate change are

producing a body of knowledge that should be integrated with that developed within the reactor safety

community. The CSNI structure appears well suited to developing structures and mechanisms for

stimulating such cross-fertilisation.

Conclusions and recommendations

A number of conclusions and recommendations have been made throughout the text of this document. The

most important are summarised briefly below subdivided into three categories

Generic Issues

Code-users’ workshop on plant analysis: The CSNI workshop recommended in 1998 that a meeting on

the use of severe-accident codes in plant calculations for source term estimation, including codes used by

utilities, be held. More than eight years after formulation of this recommendation it has still not been acted

upon. There remains a need to harmonise user practices with respect to plant analyses in order to reduce

divergence in results. The objective would ideally be to produce online “best practice” guidelines for the

major codes.

Shape factors: Models describing aerosol dynamics generally assume spherical, fully dense particles but

nuclear aerosols are often neither, particularly those originating from core melt sequences, or accident

scenarios in which large parts of containment have low humidity. The importance of shape factors has been

demonstrated in uncertainty and sensitivity analyses. Although such parametric studies can be performed

for accident scenarios, the difficulty is in knowing what constitutes a “reasonable” variation of the shape

factors for representative conditions. At the minimum, it is recommended that measurements of two

diameters (e.g., aerodynamic and volume-equivalent diameter) of aerosols from prototypical integral

experiments be obtained to determine the dynamic shape factor.

Reactor Coolant System Issues

Analytical support by the use of CFD Codes: CFD treatment of coupled aerosol/flow phenomena is not

universal because of the complexity of the phenomena and geometries, and the effort required for

implementation. At present, efforts are being made to simulate particle transport and deposition in complex

reactor structures with CFD, however more work needs to be done this treatment becomes satisfactorily

accurate.

Mechanical Resuspension: Existing models are inadequate to evaluate the safety impact of aerosol

resuspension in the primary circuit, which is both design- and scenario-dependent. Experiments are

required for the development and validation of models to: a) account for the porosity of a deposit, which

affects the mass and size distribution of resuspended material and b) simulate the effect of shock and

vibrations (alone and in combination with flow increases) on resuspension. There is also limited work on

the resuspension of wet deposits or wet deposits which have dried in situ. It is recommended that scoping

calculations be performed to assess the potential impact of this behaviour on the source term. Finally, it has

to be said that the consequences for a severe accident of a light-water reactor due to resuspension from the

RCS varies depending on the scenario, i.e., bypass or non-bypass sequences. For non-bypass sequences,

the influence of this resuspension on the potential source term may well be insignificant within a few

hours. However, improvement of resuspension modelling with respect to bypass sequences is clearly very

desirable.

NEA/CSNI/R(2009)5

Deposition in singularities and complex structures: No significant advances appear to have been made

since 1998 in the experimental investigation or the modelling of deposition in singularities such as changes

in cross section or bends. Some code improvements have been made by including the best-available

models from published literature.

Significant progress has been made with respect to deposition in steam generators (SGs). Most

probabilistic risk assessments (PRAs) and severe accident codes assume that a significant fraction of

fission products flowing through an non-isolated break in a SG escapes to the environment. This may not

be the case however, as demonstrated in several recent experimental programmes. SGTR was the first

European project (2000-2002) to improve understanding in a systematic way of possible retention

mechanisms in tubes and in the complex structures of the secondary side of a SG. In particular, the

PSAERO and HORIZON experiments from Finland were conducted to study in-tube retention whereas

retention in the SG bundle has been investigated in PECA-SGTR and the ongoing ARTIST experiments

(respectively in Spain and Switzerland). In addition, certain modelling efforts to develop a correlation for

the retention in the bundle were initiated. Thorough interpretation of the data from the above experiments

is incomplete and modelling efforts continue. It is expected that the main issues regarding SGTR will be

answered after termination of these activities.

Particle break-up in highly turbulent flows: Highly turbulent flow inside a tube may induce break-up of

particles due to (i) impaction on the walls, (ii) strong shear in the flow or (iii) the vena contracta and shock

wave at the tube exit. This phenomenon was observed, e.g., in Phase I of the ARTIST tests as well as in

supplementary experiments. However, to determine whether aerosol particles may break up in severe

accidents would require more information on the structure and the nature of the bonding forces between

primary particles for severe accident aerosols as well as for the TiO

agglomerates used in ARTIST Phase

I. In Phébus tests, which produced more prototypical aerosols, SEM micrographs showed fairly compact,

sintered agglomerate structures unlike the ARTIST TiO

particles. Consequently, intra-particle forces can

be expected to be higher in reality than in these TiO

agglomerates. Nevertheless, the issue is of some

importance since the generation of submicron particles from supramicron ones leads to less-efficient

retention of the aerosols concerned. In the first instance, accident sequences other than SGTR inducing

highly-turbulent flows need to be identified; comparison of plant calculations without and with (assumed)

break-up for SGTR and the other pertinent sequences would then allow evaluation of the risk relevance of

the break-up phenomenon. If the risk impact is significant then, in the absence of relevant data for severe

accident aerosols, data for prototypical particles are needed in highly-turbulent accident-relevant

conditions.

Impact of Chemistry: The fundamental importance of chemistry to prediction of source terms arising from

potential severe accidents is well established. Nuclear safety codes model chemical reactions using the

thermodynamic equilibrium approach, which relies on having thermodynamic data for each chemical

species covered. Unfortunately, the uncertainties associated with some of these data are often large and

divergence between data series originating from different sources can be considerable. It is recommended

(as is being done for the ASTEC code) that the thermo-chemical data used by codes be thoroughly verified

and that key sources of uncertainty be identified. Moreover, while the thermodynamic approach is

pragmatic, it is of limited value when addressing accident scenarios in which slow chemical kinetics

precludes achievement of thermodynamic equilibrium (e.g. secondary-side, cold-leg and containment

conditions). In this context it is recommended to:

• Assess the value of the simplistic approach (taken in ASTEC/SOPHAEROS and VICTORIA codes)

of assigning user-defined cut-off temperatures below which chemical reactions in the RCS do not

occur. Use of PHÉBUS FP results with respect to iodine in the RCS should prove helpful;

• Follow closely experimental programmes that are investigating reaction rates for iodine species

(e.g., the French CHIP series).

NEA/CSNI/R(2009)5

• Assess whether FP species other than iodine, which may reach the containment in the vapor phase,

require kinetic modelling (especially important for hot-leg and high-pressure sequences where vapor

fractions at the RCS breach will be significant).

• Lastly, the effects of radiolysis in the RCS are unknown. Radiolysis may have little impact in the

core region (temperatures being so high that only simple atomic and radical species exist), but may

be important in cooler regions of the RCS involving significant deposits (high local dose rate) such

as in a cold-leg sequence. In terms of direct consequences for aerosols, one effect will be reduction

of the threshold super-saturation at which vapors nucleate since a high density of

electrically-charged condensation nuclei will form. In terms of consequences for the source term, the

meagre state of knowledge renders even qualitative evaluation difficult.

Revaporisation of deposits: Although significant progress has been made regarding studies on

revaporisation from several projects (Phébus FP, the EC 4th Framework Programme projects RVP and

REVAP-ASSESS and the EXSI project), experimental work on revaporisation remains scarce; further

theoretical and experimental developments are necessary to understand the revaporization process. In

addition, the safety relevance ought to be further demonstrated by assessing, conceivably, the impact of a

weak source of fission products from the reactor coolant system for some hours after the main

release-from-core phase.

Containment Issues

Charge effects: Small deposits of aerosols on the outer walls of the 10 m

containment vessel in the

PHÉBUS test FPT0 could not be explained by electrophoretic effects. However, even if charge effects on

aerosol deposition are not seen to be significant in test facilities (with or without a radiation field) there is

no firm evidence that this effect would be negligible in an accident. At present there is no consensus

among experts on whether further investigations of charging effects are necessary or not.

Mixed aerosols in condensing atmospheric conditions: Although there has been considerable progress in

modelling aerosol deposition as a function of relative humidity, a comparison of the adequacy of code

results from ISP 37 and ISP 44 indicate that there is still some work to be done to ensure satisfactory

coupling between thermal hydraulic and aerosol models so that these capture correctly aerosol behaviour in

most environments. An additional uncertainty in modelling aerosol behaviour in the containment in humid

conditions arises from determining the hygroscopicity associated with a mixture of aerosols of different

compositions. Finally, there is some uncertainty regarding the density of multi-component aerosols, and

whether this parameter is important for accident conditions with a wide variety of aerosol components.

Both the PHÉBUS tests and the KAEVER experiments suggest that an average aerosol particle

composition and size might be attained in containment. The KAEVER experiments further suggest that this

average particle would behave like the most hygroscopic of its individual components. Confirmation of

these findings, in a large- or intermediate-scale experiment performed under saturated conditions might

significantly simplify modelling aerosol behaviour in wet conditions.

Mechanical Resuspension: Compared with deposits in the primary system, deposits in containment will

have significantly lower particle loadings, and be distributed over larger areas. Resuspension of such

deposits is possible as a consequence of a breach of containment and/or a hydrogen burn. It is

recommended that experimental studies be performed on resuspension of real or simulated deposits at

containment-typical loadings, particularly during flow disturbances, as a function of the deposit

composition and history (dry, wet, wet then dried in situ etc.) and as the result of hydrogen deflagrations.

In addition, probabilistic studies should be undertaken to evaluate the safety significance of containment

resuspension.

NEA/CSNI/R(2009)5

Re-entrainment in the Containment: Re-entrainment of particulate fission products will occur at several

water and core melt pools during a severe accident. The release rates of radiological materials are relatively

small but the sources are persistent. A significant contribution to the source term by re-entrainment is

possible in the late accident phase. However, reliable analytical investigations including risk relevant

aspects do not exist yet. With a few exceptions, there are only a few codes able to simulate the FP release

from boiling or flashing sumps.

Up to now all re-entrainment tests including ThAI have been made on ideal systems with appropriate

concentrations of soluble and insoluble materials but without surfactants or impurities. Additional

experiments are required to evaluate entrainment under realistic conditions. After improvement and

validation of re-entrainment models for realistic conditions is achieved reliable accident calculations to

quantify the effect of re-entrainment from boiling pools on the source term will be possible. In the light of

these results it may be desirable to investigate measures reducing the release of fission products by

re-entrainment.

Pool scrubbing: Some BWR and PWR severe accident scenarios involve transport of radioactive aerosols

through pools of water where particles can be retained. This phenomenon, known as pool scrubbing, has

the potential to reduce the source term. Results provided by both stand-alone and integral code models

indicate satisfactory agreement with simple experiments for integral retention. However, a systematic

experimental database is required for validation purposes. Particular attention should be given to removal

of aerosols during formation and subsequent disintegration and coalescence of bubbles, and the effects of

submerged structures and contaminants (surfactants).

Removal by sprays: This issue has been extensively investigated by the French organisations CEA and

IRSN using specific apparatuses and the CARAIDAS, MISTRA and TOSQAN test facilities. The data

should be made accessible to the nuclear community, at least the OECD partners. Validated modelling

based on these experimental investigations has been implemented in the codes ASTEC and TONUS. The

ASTEC model can be found in the open literature. Further work on containment sprays is low priority for

countries that have access to this data but in other countries and for certain advanced designs it remains

important to establish effective removal by spray systems and both experimental and analytical efforts

continue.

Influence of recombiners: Phenomenological experiments have demonstrated that there is a potential for

the operation of Passive Autocatalytic Recombiners (PARs) to generate volatile molecular iodine by

thermal decomposition of metal-iodide containment aerosols. Scoping calculations indicate that this

conversion process might be a significant contribution to the molecular iodine in the containment

atmosphere. It has also to be recognised that iodine may not be the only fission product concerned, e.g.,

some formation of the highly volatile species ruthenium tetroxide might be possible in the conditions

expected within PARs though no investigation of this has been performed. Further experimental

investigations in realistic conditions (mixed-aerosol and mixed-atmosphere composition) are necessary.

Hydrogen-burn effects on suspended aerosols: Heat release by hydrogen burns may have a strong effect

on aerosol characteristics and could possibly liberate volatile forms of iodine. Experiments in the ThAI

facility demonstrate that there is a clear effect of hydrogen-combustion-induced flows on resuspension of

already deposited CsI aerosol. These findings have to be expanded to consider other types of relevant

aerosol species and other plant applications. However, chemical effects on iodine-containing aerosols have

not been studied; information from the ongoing investigation of the impact of recombiners on aerosols may

be relevant.

Release from MCCI pool: The behaviour of aerosols formed after the release from a molten corium

concrete pool was investigated in the 1990s in the ACE phase C and BETA tests and more recently in the

NEA/CSNI/R(2009)5

OECD-MCCI and EC Framework Programmes (LPP and MP projects). However, efforts to further aerosol

code development and validation from these experiments has been limited. As there is currently little

known activity in this area it is recommended that the following steps be initiated in the near future:

● Recalculation of the old test results to evaluate the thermodynamic basis of the present models by

comparing concrete erosion processes and aerosol nucleation due to vaporization from the free

upper surface and into gas bubbles in the corium concrete pool,

● Obtain information on the characterization of concrete-based aerosols; conventional

(non-nuclear) knowledge about concrete aerosols may be a good starting point;

● Evaluate information from the EC projects MP and LPP from the specific view of fission

products release due to MCCI;

Penetration of aerosols through leak paths: From a review of available data and models including recent

research in SARNET on dry aerosol transport in cracks, the recommendation can be made that additional

separate effects and integral tests are required:

● Separate effect tests should provide a detailed characterization of the scenarios and the

phenomena necessary for model development and validation of individual depletion mechanisms;

● Integral tests should focus on measuring overall process variables such as mass retained to

provide a data base for checking the overall model performance.

The boundary conditions (hydraulic diameter, curvature of the path, fluid composition pressure drop, wall

temperature, and aerosol size) for these experiments should, obviously, be as close as possible to those

postulated under accident conditions.

For wet aerosols, it is recommended that additional experiments be performed to characterise aerosol

transport through sequential expansion and contraction regions representative of the leak path from

containment to the outside atmosphere and to evaluate the extent of leak-path plugging.

Fire aerosols: Little is known about the properties or amounts of aerosols produced from fires, and how

mixing of flaky fire aerosol with nuclear aerosol particles would impact on the aerosol depletion rate in

containment. There is a need for further experimental investigation of fire aerosols especially from cable

fires. This could be done in a future step of the OECD PRISME and/or the Sandia CAROLFIRE projects.

The aerosol production rate, the particle size distribution and the shape factors are of main concern. It is

recommended that existing multi-component aerosol models be extended in order to simulate the

interaction of fire aerosols with a pre-existing nuclear aerosol.

NEA/CSNI/R(2009)5

TABLE OF CONTENTS

EXECUTIVE SUMMARY ............................................................................................................................ 5

TABLE OF CONTENTS .............................................................................................................................. 13

1. INTRODUCTION ................................................................................................................................. 19

2. GENERAL PHENOMENA................................................................................................................... 21

2.1 Aerosol Physics in Reactor Accident Analyses .............................................................................. 21

2.2 Aerosol Formation .......................................................................................................................... 23

2.3 Growth of Aerosol Particles ........................................................................................................... 25

2.3.1 Growth by coagulation ............................................................................................................. 25

2.3.2 Growth by condensation ........................................................................................................... 28

2.4 Aerosol Growth as a Result of Hygroscopicity .............................................................................. 30

2.5 Fractal Nature of Aerosols.............................................................................................................. 31

2.6 Aerosol Deposition ......................................................................................................................... 31

2.6.1 Gravitational settling and diffusion .......................................................................................... 32

2.6.2 Inertial deposition of aerosol particles ..................................................................................... 33

2.6.3 Phoretic deposition processes ................................................................................................... 35

2.7 Resuspension of Deposited Particles .............................................................................................. 36

2.8 Aerosol Removal by Engineered Safety Systems .......................................................................... 37

2.8.1 Steam suppression pools .......................................................................................................... 40

2.8.2 Fan coolers ............................................................................................................................... 41

2.9 Aerosol Removal by Filtered Vents ............................................................................................... 41

2.10 Computer Modelling of Aerosols ................................................................................................... 43

2.11 Annotated Bibliography ................................................................................................................. 43

3. ACCIDENT PHENOMENOLOGY ...................................................................................................... 49

3.1 Formation and Growth ................................................................................................................... 49

3.1.1 Phenomena considered and state of knowledge ....................................................................... 49

3.1.2 Particle formation in the primary system ................................................................................. 52

3.1.3 Containment aerosols ............................................................................................................... 55

3.2 Impact of Thermal Hydraulics........................................................................................................ 57

3.2.1 Impact of thermal hydraulics on aerosol behaviour in the primary circuit .............................. 57

3.2.2 Impact of thermal hydraulics on aerosol behaviour in the containment................................... 59

3.3 Fission-Product Transport and Deposition ..................................................................................... 64

3.3.1 Introduction .............................................................................................................................. 64

3.3.2 Synopsis of RCS phenomena and their modelling ................................................................... 64

3.3.3 Agglomeration of aerosols ....................................................................................................... 66

3.3.4 Deposition of aerosols .............................................................................................................. 67

3.3.5 Containment ............................................................................................................................. 70

3.4 Aerosol Behaviour in Complex Structures of Steam Generator Secondary Site ............................ 73

3.4.1 Issue and status ......................................................................................................................... 73

3.4.2 Theoretical background on possible aerosol removal in the steam generator .......................... 76

3.4.3 Inertial impaction and interception in the separator and dryer sections ................................... 79

3.4.4 Other processes ........................................................................................................................ 79

NEA/CSNI/R(2009)5

3.5 Resuspension .................................................................................................................................. 82

3.5.1 Resuspension in primary circuit ............................................................................................... 82

3.5.2 Resuspension in containment ................................................................................................... 85

3.6 Pool Scrubbing ............................................................................................................................... 87

3.7 Re-Entrainment .............................................................................................................................. 92

3.8 Aerosol Formation during High Pressure Melt Expulsion from the Reactor Coolant System ....... 96

3.9 Aerosols from MCCI .................................................................................................................... 100

3.10 Aerosols from Fire ........................................................................................................................ 108

3.11 Spray Systems .............................................................................................................................. 111

3.12 The Impact of Catalytic Hydrogen Recombiners ......................................................................... 121

3.13 Filters ............................................................................................................................................ 123

3.14 Two Phase Flashing Jets and Water Aerosol Behaviour .............................................................. 127

3.15 Penetration Leakages .................................................................................................................... 130

3.15.1 Introduction ............................................................................................................................ 130

3.15.2 Current modelling and experimental studies .......................................................................... 131

3.15.3 Measurement of the retention of wet aerosols ........................................................................ 132

4. MODELLING APPROACHES IN CODES (INCLUDING BASIC VALIDATION) ....................... 137

4.1 Integral Codes ............................................................................................................................... 137

4.1.1 MELCOR ............................................................................................................................... 137

4.1.2 ASTEC (SOPHAEROS and CPA) ......................................................................................... 142

4.1.3 MAAP4 .................................................................................................................................. 147

4.1.4 THALES 2.............................................................................................................................. 149

4.1.5 ECART ................................................................................................................................... 155

4.1.6 APROS SA ............................................................................................................................. 157

4.2 Circuit ........................................................................................................................................... 161

4.2.1 VICTORIA ............................................................................................................................. 161

4.2.2 ATHLET-CD ......................................................................................................................... 168

4.3 Containment ................................................................................................................................. 169

4.31 CONTAIN .............................................................................................................................. 169

4.3.2 COCOSYS ............................................................................................................................. 175

4.3.3 ART/REMOVAL ................................................................................................................... 181

4.3.4 SMART .................................................................................................................................. 183

4.3.5 GOTHIC ................................................................................................................................. 184

4.3.6 KUPOL .................................................................................................................................. 185

5. RECENT VALIDATION WORK ....................................................................................................... 187

5.1 Integral Experiments .................................................................................................................... 187

5.1.1 The PHÉBUS fission product programme ............................................................................. 187

5.1.2 ISP-46 (PHÉBUS FPT1) ........................................................................................................ 190

5.2 Circuit ........................................................................................................................................... 193

5.2.1 Revaporization in the FPT0, FPT1 and FPT2 circuits ........................................................... 193

5.2.2 FALCON (Including ISP-34) ................................................................................................. 194

5.2.3 STORM (Including ISP-40) ................................................................................................... 197

5.2.4 WIND ..................................................................................................................................... 204

5.3 Containment Experiments ............................................................................................................ 207

5.3.1 DEMONA .............................................................................................................................. 207

5.3.2 MARVIKEN-V ...................................................................................................................... 209

5.3.3 LACE ..................................................................................................................................... 211

5.3.5 VANAM (including ISP-37) .................................................................................................. 217

5.3.6 KAEVER (including ISP-44) ................................................................................................. 222

NEA/CSNI/R(2009)5

5.3.7 AHMED ................................................................................................................................. 228

5.3.8 VICTORIA ............................................................................................................................. 232

5.4 Aerosol Release in the Course of MCCI (ACE-C) ....................................................................... 236

5.5 REST, REVENT, ThAI Re-Entrainment Tests ............................................................................ 237

5.6 Pool Scrubbing Tests .................................................................................................................... 239

5.7 Resuspension Tests ....................................................................................................................... 242

5.7.1 Continuous flow experiments ................................................................................................. 242

5.7.2 Transient flow experiments .................................................................................................... 243

5.8 Spray Systems (French Tests) ...................................................................................................... 244

5.9 Containment Venting Filters ........................................................................................................ 247

5.9.1 International efforts for qualification of containment venting rilters ..................................... 248

5.9.2 CCI containment venting filter ............................................................................................... 255

5.9.3 Removal of iodine by the containment venting filters ........................................................... 258

5.9.4 International status of use of containment venting filters ...................................................... 258

5.10 Ongoing Tests ............................................................................................................................... 258

5.10.1 Aerosol behaviour in steam generators .................................................................................. 258

5.10.2 EVAN ..................................................................................................................................... 264

5.10.3 Penetration leakage tests ........................................................................................................ 265

5.10.4 ThAI aerosol tests ................................................................................................................... 266

5.10.5 Impact of catalytic hydrogen recombiners ............................................................................. 269

5.10.6 Aerosol growth under saturated conditions (Canadian tests) ................................................. 272

6. EXAMPLES FOR VALIDATION ...................................................................................................... 273

6.1 Comparison of MELCOR Predictions to Experiments ................................................................ 273

6.2 CONTAIN for PHÉBUS FPT1 .................................................................................................... 277

6.3 Example for Uncertainty and Sensitivity Analyses of Two Aerosol Calculations ....................... 283

7. SOURCE TERM CALCULATIONS IN PLANT APPLICATIONS ................................................. 287

7.1 MELCOR Calculation for a US American PWR (Station Blackout Scenario) ............................ 287

7.2 ASTEC Calculation for a German PWR (MB LOCA Scenario) ................................................. 294

7.3 Aerosol-Related Uncertainties in the Prediction of Severe Accident Source Terms ................... 304

8. IDENTIFIED OPEN ISSUES (INCLUDING RECOMMENDATIONS) .......................................... 307

8.1 Introductory Remarks ................................................................................................................... 307

8.2 Generic Issues ............................................................................................................................... 308

8.2.1 Code-users’ workshop on plant analysis ................................................................................ 308

8.2.2 Shape factors .......................................................................................................................... 308

8.2.3 Computational fluid dynamics codes ..................................................................................... 309

8.3 Reactor Coolant System Issues .................................................................................................... 309

8.3.1 Mechanical resuspension ........................................................................................................ 309

8.3.2 Deposition in singularities and complex structures ................................................................ 310

8.3.3 Particle break-up in highly turbulent flows ............................................................................ 311

8.3.4 Influence of chemistry ............................................................................................................ 311

8.3.5 Revaporisation of deposits ..................................................................................................... 312

8.4 Containment Issues ....................................................................................................................... 313

8.4.1 Charge effects ......................................................................................................................... 313

8.4.2 Mixed aerosols in condensing atmospheric conditions .......................................................... 313

8.4.3 Mechanical resuspension ........................................................................................................ 313

8.4.4 Re-entrainment from pools ..................................................................................................... 314

8.4.5 Pool scrubbing ........................................................................................................................ 314

8.4.6 Removal by sprays ................................................................................................................. 315

NEA/CSNI/R(2009)5

8.4.7 Influence of recombiners ........................................................................................................ 315

8.4.8 Hydrogen-burn effects on suspended aerosols ....................................................................... 315

8.4.9 Release from MCCI pool ....................................................................................................... 316

8.4.10 Penetration through leak paths ............................................................................................... 317

8.4.11 Fire aerosols ........................................................................................................................... 317

8.5 Concluding Remarks .................................................................................................................... 318

9. APPENDIX 1: Characteristics of Aerosols under LWR Severe Accident Conditions ................... 323

9.1 Context ......................................................................................................................................... 323

9.2 Review of Available Information ................................................................................................. 323

9.2.1 Aerosols in the RCS ............................................................................................................... 324

9.2.2 Aerosols in the containment ................................................................................................... 329

9.3 Discussion .................................................................................................................................... 331

9.3.1 Uranium contribution ............................................................................................................. 331

9.3.2 Tin contribution ...................................................................................................................... 331

9.3.3 Silver contribution .................................................................................................................. 331

9.3.4 Cadmium contribution ............................................................................................................ 331

9.3.5 Indium contribution ................................................................................................................ 331

9.3.6 Fission product contribution ................................................................................................... 332

9.3.7 Size, shape and structure ........................................................................................................ 332

9.3.8 Variability............................................................................................................................... 333

9.3.9 Other considerations ............................................................................................................... 333

9.4 Conclusion .................................................................................................................................... 334

10. APPENDIX 2: Aerosol Shape Factors ............................................................................................. 337

10.1 Fundamentals ................................................................................................................................ 337

10.2 Dynamic and Collision Shape Factors ......................................................................................... 338

11. APPENDIX 3: Spray Modelling Developed from Recent Analytical Work ................................... 345

11.1 Relaxation of Droplets .................................................................................................................. 345

11.2 Droplet Coalescence ..................................................................................................................... 346

11.3 Coupling of Droplet Relaxation and Coalescence ........................................................................ 347

11.4 Evolution of the Atmosphere Conditions ..................................................................................... 349

11.5 Aerosol Removal Modelling ........................................................................................................ 350

11.5.1 Inertial capture ........................................................................................................................ 350

11.5.2 Interception............................................................................................................................. 351

11.5.3 Brownian diffusion ................................................................................................................. 352

11.5.4 Phoretic capture ...................................................................................................................... 352

11.6 Gaseous Iodine Removal by Sprays ............................................................................................. 353

12. APPENDIX 4: Overview of Filtration Devices ............................................................................... 357

12.1 Nuclear Aerosol Filtration ............................................................................................................ 357

12.1.1 Development of the high-efficiency particulate air (HEPA) filter ......................................... 357

12.1.2 Challenges for test procedures ............................................................................................... 359

12.1.3 Demonstrated safety by the use of HEPA filters .................................................................... 359

12.1.4 Deep-bed sand and glass fibre filters ..................................................................................... 359

12.1.5 Brief history of gas adsorption ............................................................................................... 360

12.2 Type of Filters in Use ................................................................................................................... 361

12.3 Normal Operation and Design Basis Accidents ........................................................................... 361

12.4 Brief Review on Mechanisms of Filtration .................................................................................. 363

12.5 Characterisation of Filter Performance......................................................................................... 365

NEA/CSNI/R(2009)5

12.6 Cut Diameter Method .................................................................................................................. 367

12.7 Fibrous Filter ............................................................................................................................... 368

12.8 Granular Beds .............................................................................................................................. 370

12.9 Sand Beds .................................................................................................................................... 373

12.10 Gravel Beds ................................................................................................................................. 374

12.11 Venturi Scrubbers ........................................................................................................................ 376

12.12 Water Pools ................................................................................................................................. 379

13. APPENDIX 5: Point-by-Point Review of Issues Identified at the 1998 Aerosol Workshop ........... 385

NEA/CSNI/R(2009)5

1. INTRODUCTION

The prediction of nuclear aerosols behaviour associated with postulated accidents in nuclear power plants

is necessary for the assessment of the radiological consequences of these accidents and hence is an

important aspect of reactor safety evaluation. The nature and behaviour of nuclear aerosols can potentially

influence both the course and the consequences of reactor accidents. For example, nuclear aerosols can

affect the performance of engineered safety systems (e. g. air cleaning systems) as well as the magnitude,

dispersion and effects of the radioactive source term leaked to the atmosphere. As a result of the extreme

conditions for nuclear aerosols in case of an accident, they exhibit very dynamic mostly physical, but also

chemical behaviour and pose special analytical and experimental problems different from those associated

with aerosols found under industrial and ambient conditions.

Following a LWR severe accident, the overall radionuclide retention in form of aerosol, both in reactor

coolant system and containment, is the result of three processes: transport, deposition and release from dry

or wet surfaces (resuspension) or (boiling) water pools (re-entrainment).

Approximately a 30 years tradition in the preparation of state-of-the-art reports on nuclear aerosols and the

proceedings of workshops concerning the aerosol issue (summarised in chronological order at the end of

this chapter [1 - 8]) demonstrate the substantial work performed by the nuclear community and the

progress made in this period of 30 years.

Nevertheless at the last workshop in Cologne, 1998, some important issues were identified, where

knowledge did not seem to be sufficient for realible predictions. Following the recommendations of this

workshop the CSNI decided to prepare a state-of-the-art report with three main objectives:

• to summarise the present experimental and analytical knowledge on nuclear aerosols

• to give some examples for validation results and in particular for plant applications, both including

uncertainties

• to identify safety relevant still unresolved issues

Nevertheless, it should be understood that the subject matter of the report is nuclear aerosols, and the report

does not purport to cover the whole field of radioactive source terms for severe LWR accidents. For

example, there is no detailed discussion of fission product chemistry processes that do not have a direct

bearing on aerosol behaviour, such as iodine solution chemistry and organic iodine production.

Finally the authors of this report wish to acknowledge significant contributions from a large number of

colleagues, which have increased the quality of the report considerably.

References

[1] NEA Group of Experts, Nuclear Aerosols in Reactor Safety , A state-of-the-art report by a group

of experts of the NEA committee on the safety of nuclear installations, June 1979

[2] CSNI Specialist Meeting on Nuclear Aerosols in Reactor Safety in Gatlinburg, April 1980

NEA/CSNI/R(2009)5

[3] CSNI Group of Experts on Nuclear Aerosols in Reactor Safety, March 1983

[4] CSNI Specialist Meeting on Nuclear Aerosols in Reactor Safety Workshop Karlsruhe, September 1984

[5] NEA Group of Experts, Nuclear Aerosols in Reactor Safety - Supplementary Report

NEA / OCDE /OECD, Paris 1985

[6] Nuclear Engineering Applications Section Species to Aerosols - Workshop on Water-Cooled

Reactor Aerosol Code Evaluation and Uncertainty Assessment, Brussels, September 1987

[7a] OECD / NEA, Aerosol behaviour and thermal-hydraulics in the containment (2

workshop

Brussels 1987, Technical Summary), NEA/CSNI/R(92)1, Fontenay-aux-Roses, 1990

[7b] OECD / NEA Aerosol behaviour and thermal-hydraulics in the containment (2

workshop

Brussels 1987, Proceedings) NEA/CSNI/R(92)1, Fontenay-aux-Roses, 1990

[8] Third OECD Specialist Meeting on Nuclear Aerosols in Reactor Safety GRS-166,

NEA/CSNI/R(98)4, ISBN 3-931995-31-3, Cologne, June 1998

NEA/CSNI/R(2009)5

2. GENERAL PHENOMENA

This chapter provides background and introductory information and is directed primarily at those readers

with an interest in being reminded of the fundamentals of aerosol behaviour under severe reactor accident

conditions. In addition, there is an annotated bibliography that will introduce the reader to the pertinent

research literature.

2.1 Aerosol Physics in Reactor Accident Analyses

Most of the radioactive material that can escape from a nuclear power plant during a severe reactor

accident will do so in the form of aerosols. Much of reactor accident analysis is the prediction of the

behaviour of these radioactive aerosols. Aerosols are very small solid particles or liquid droplets suspended

in a gas phase. The suspended solid or liquid particles typically have a range of sizes. Particles may range

in size from 0.01 µm to 20 µm. Aerosol concentrations in reactor accident analyses are typically less than

100 g/m

and usually less than 1 g/m

. At these concentrations, the aerosol particles little affect the gas

hydrodynamics, but the gas dynamics profoundly affect the behaviour of the suspended particles. The

behaviours of the larger aerosol particles are described usually by continuum mechanics. The smallest

particles have diameters less than the mean free path of gas phase molecules and the behaviour of these

particles can often be described well by free molecular physics. The vast majority of aerosol particles

arising in reactor accident analyses have behaviours in the very complicated regime intermediate between

the continuum mechanics and free molecular limit. In this regime, aerosol behaviour must be described

using some approximate solution of the Boltzmann equation.

The mechanical regimes of aerosol behaviour are usually categorised in terms of the Knudsen number, Kn,

which is the ratio of the mean free path of

gas molecules, λ, to the particle diameter

D :

Continuum behaviour is approached for Kn < 0.1 and free molecular flow is approached for Kn > 10.

These regimes vary with temperature and pressure which affect the mean free path. At atmospheric

pressure and room temperature the mean free path of air is about 0.06 µm.

The number density of aerosol particles in a gas phase can be huge - exceeding 10

. It is quite

impossible to predict aerosol behaviour by calculating the dynamics of individual particles. Instead

aerosols must be considered in a collective sense and the aerosol is taken to have some continuous

distribution of particle sizes. Given sufficient time in a quasi-steady state environment with no continued

injection of aerosol into the gas phase, an aerosol subjected to particle growth by agglomeration and

gravitational deposition onto surfaces will develop a size distribution that is approximated well by a

log-normal distribution. The probability density for particles having sizes in the interval

D to

D + d

is:

NEA/CSNI/R(2009)5

Dn1d

σn21

µ/nD1

exp

σn12

)D(pdf

























−

where:

µ = mean particle size

σ = geometric standard deviation of the size distribution

It is common to report aerosol size data in terms of the parameters of the lognormal size distribution (µ and

σ) whether or not the aerosol satisfies the criteria for the lognormal distribution being a good

approximation of the actual size distribution. Example size distribution for the case of a very narrow size

distribution (σ = 1.8) and for the case of a broad distribution (σ = 3) are shown in Fig. 2.1-1.

When there are continuing sources of aerosol to the gas phase or when there are complicated processes

involving engineered safety features much more complicated size distributions develop. It is not

uncommon for aerosols in reactor containments to have bimodal size distributions at least for some

significant periods of time early in an accident.

Salient features of aerosol physics under reactor accident conditions that will affect the nature of the

aerosols are:

• formation of aerosol particles

• growth of aerosol particles

• shape of aerosol particles

• deposition of particles on surfaces

• resuspension of aerosol particles

Introductory or background information on these and other aspects of nuclear aerosols are provided in the

subsections of this chapter that follow. This chapter is concluded with a brief description of the numerical

methods used to model the important aerosol phenomena.

NEA/CSNI/R(2009)5

Particle Size (µm)

0.01 0.1 1 10 100

Probability Density

0.0

0.2

0.4

0.6

0.8

σ = 1.8

σ = 3.0

Fig. 2.1-1 Comparison of log normal distributions with different geometric standard deviations, s, and

the same mean, µ

2.2 Aerosol Formation

Aerosol particles can be formed by the mechanical comminution of materials or by the condensation of

vapors. Mechanical processes that produce aerosol particles in reactor accidents include:

• entrainment of solids or liquid droplets in high velocity gas flows,

• expulsion of droplets by gases bubbling through liquids,

• shock waves such as those produced in energetic interactions of molten materials with coolants, and

• high pressure melt ejection from the reactor coolant system.

Mechanical processes seldom produce very fine particles that can remain suspended in the gas phase for

protracted periods of time. Typically, mechanical processes will not produce particles smaller than about

1-2 µm. Furthermore, most mechanical processes that can produce aerosols are not of prolonged duration

and are, thus, not enduring sources of aerosols through large periods of an accident. An exception is the

production of aerosols by the bursting of bubbles of gas sparging through liquids. But, the bubble burst

source is not an especially intense source of aerosol particles. Aerosol production by mechanical processes

is largely ignored in most reactor accident analyses.

Nucleation of particles from supersaturated vapors is the more important source of aerosols in reactor

accidents and certainly the source given the most attention in reactor accident analysis computer codes.

Vapors rich in radioactive materials are formed from high temperature core debris materials. The vapors

become supersaturated as they are transported from the vicinity of the core debris to cooler regions. These

supersaturated vapors can nucleate aerosol particles.

Vapor nucleation is a very complicated physical process even when the vapor consists of a single

condensable species. Nucleation can occur homogeneously from the vapor when the supersaturation ratio

NEA/CSNI/R(2009)5

(ratio of the actual partial pressure of condensable species divided by the equilibrium partial pressure of

that species) exceeds a value of 4 - 10. Nucleation is a kinetic process and there is not a sharp onset to this

process. Nucleation is usually taken to be occurring when the rate of particle production exceeds one

particle per cubic centimeter per second. The supersaturation necessary for any particular vapor to nucleate

is a sensitive function of the surface energy of the condensed phase (surface tension in the case of a liquid

being produced from the vapor). The rate of tin vapor nucleation is shown in Fig. 2.2-1 as a function of

temperature for a vapor initially saturated at 1800 K. It can be seen from this example that nucleation

occurs whenever the gas is saturated, but nucleation is quite slow until a substantial supersaturation

develops. That is, the vapor has to cool well below saturation for detectable nucleation to be observed.

Detectable nucleation (Rate > 1 particle/cm

-s) of the tin aerosol occurs rather abruptly over a narrow

temperature range at a temperature over 100 K lower than the temperature at which the vapor becomes

saturated. The enormous nucleation rates predicted at still lower temperatures are difficult to realise in

practice because heat liberated by nucleation cannot be dissipated from the system.

Temperature (K)

1600 1620 1640 1660 1680 1700 1720

Nucleation Rate (particle/cm

-s)

Nucleation Rate

1 particle/cm

-s

Fig. 2.2-1 Nucleation kinetics of tin vapor initially saturated at 1800 K

Heterogeneous nucleation of vapor on ions in the gas phase can occur at supersaturations that are lower

than those needed to drive homogeneous nucleation. The gas phase in a nuclear reactor accident will be

intensely irradiated so there will be relatively high concentrations of ions to act as sites for nucleation.

Some heterogeneous nucleation of vapors will occur. But, supersaturation ratios that will develop when

high temperature vapors escape the regions of core debris into cooler regions can exceed 10

. Since

nucleation is kinetic in nature, it is quite likely that both homogeneous nucleation and heterogeneous

nucleation will occur simultaneously.

The theory of nucleation of complex vapors produced in reactor accidents is vastly more complicated than

the theory for single pure vapors outlined above. Indeed, a general theory of nucleation for complex vapors

has not yet been developed. The qualitative features of nucleation from complex vapor mixtures are

expected to be similar to those of pure vapor nucleation although quantitative features may differ and

calculation of these features will involve vapor and condensed phase properties that are not readily

NEA/CSNI/R(2009)5

available. Consequently, when nucleation phenomena are considered in accident analyses, approximate

models based on nucleation of single vapors are used. The upper and lower bound models of Reiss, Katz

and Cohen are quite popular.

It is more common especially in the so-called “lumped node” computer models of reactor accidents to omit

mechanistic modeling of nucleation processes. Particle formation is simply assumed to occur when the

vapor reaches saturation. The omission is justified by the rapid development of supersaturation in the gas

flowing into ever cooler environments and the rapid growth of particles that are nucleated from the vapor.

This sudden growth of the nucleated particles can obscure any effects of the details of nucleation kinetics

at the “lumped node” level of approximation. Recently, mechanistic treatments of vapor nucleation have

been included in computational fluid dynamics models of flows during reactor accidents. In some cases,

these models are showing unexpected complexity in the aerosol deposition because of the spatial variations

in the supersaturation along a flow path. These findings are, of course, subjects of continued research.

2.3 Growth of Aerosol Particles

2.3.1 Growth by coagulation

Aerosol particles grow by continued condensation of vapor on the nucleated particles and by the

coagulation of these particles. Vapor condensation on nucleated particles is limited by the rate at which

heat liberated by condensation can be removed. Interior to a cloud of nucleating vapor, only conduction

can remove this heat. Velocity differences between the gas phase and the particles are too small to promote

much convective heat transport and optical paths for radiation heat transfer are effective only near the

perimeter of the cloud. Though some continued vapor condensation will occur, coagulation is the dominant

mechanism of particle growth following nucleation. Coagulation greatly affects the size and consequently

the mobility of aerosol particles, so the aerosol particles quickly develop a size distribution. The evolution

of this size distribution is described by the intergrodifferential equation:

du)uv(n)uv,u(K

)v(n

−−

∫

∂

du)u(n)v,u(K

)v(n

∫

∞

−

)v(S)v(n)v(R +−

where:

n (v) = number density of particles of volume v to v + dv (particles/cm

)

K (u,v) = collision kernal for particles of volume v and volume u (s

-1

)

R (v) = rate constant for deposition of particles of volume v on surfaces (s

-1

)

S (v) = source rate of particles (particles/cm

-s)

The first integral on the right-hand side of this equation is the coagulation of particles of volume u and v- u

to form a particle of volume, v. The second integral is the coagulation of particles of volume v with

particles of any volume to form a particle that is no longer in the v to v + dv volume interval.

Particles coagulate because they cross stream lines of flow to come into contact. There are a variety of

mechanisms that can cause particles to cross stream lines. Each of these mechanisms is characterised by a

distinct collision kernel, K(u,v). The more common collision kernels recognised in models of aerosol

growth during reactor accidents are:

NEA/CSNI/R(2009)5

• Gravitational - larger particles sweep out smaller particles as they fall under the force of gravity

()

() ()

|uuCvvC|uvv,u

v)(u, K

3/23/2

3/13/1

3/4

−+∈













where:

= collision shape factor

χ = dynamic shape factor

g = gravitational acceleration

ρ = material density of the particle

µ = gas viscosity

()

3/13/1

3/2

=∈

β = 0.5 or 1.5

C(v) = Cunningham slip correction factor for a particle of volume v

• Brownian diffusion - fluctuations in molecular bombardment drive particles across streamlines of

flow to contact other particles.

()

(

)

(

)













3/13/1

kT2

v,uK

where:

k = Boltzmann constant = 1.38066 x 10

-16

ergs/K

T = absolute temperature (K)

• Turbulent dif fusion - turbulent eddies carry particles across streamlines to contact other particles

()

2/1

3/13/1

v,uK













ρ∈

where:

∈ = turbulent energy dissipation rate

1.29 ≤ Z ≤ 5.65

• Turbulent inertia - particles expelled from turbulent eddies impact other particles

()

()( )

4/1

3/23/2

3/13/1

|vu|uvZv,uK













ρ∈

−+

′

where:

0.188 ≤ Z

′

≤ 0.204

NEA/CSNI/R(2009)5

The Cunningham slip correction factor, C(v), that arises in these various expressions for the collision

kernels to account for deviations from continuum mechanics is empirically derived from data for spherical

particles. Many different expressions have been given for this correction factor. A commonly used one is:

()

(

)

(

)

Kn/999.0exp588.0142.1Kn1vC −++=

3/1













The collision kernels and the slip correction factor are shown in Fig. 2.3-1 and Fig. 2.3-2 as functions of

particle size for a particular set of circumstance (pressure, temperature, etc.).

In many circumstances, one of the mechanisms of coagulation is dominant. There are other circumstances

in reactor accident analyses in which several coagulation mechanisms act with similar magnitudes. This

gives rise to the issue of addition of kernels. Often kernels are simply added. In other cases more

complicated summations are used such as:

() () () () ()

v,uKv,uKv,uKv,uKv,uK

TDgBtotal

+++=

Research in aerosol physics includes efforts to develop more defensible descriptions of the combined

effects of simultaneous coagulation mechanisms.

1/3

(µm)

0.001 0.01 0.1 1 10 100

Collision Kernel

1e-13

1e-12

1e-11

1e-10

1e-9

1e-8

1e-7

1e-6

1e-5

gravitational

Brownian diffusion

turbulent diffusion

turbulent inertia

Fig. 2.3-1 Example collision kernels for various mechanisms of coagulation

NEA/CSNI/R(2009)5

Slip Correction Factor

Knudsen Number, Kn

0.001 0.01 0.1 1 10 100

Cunningham Slip Correction Factor, C

Fig. 2.3-2 Cunningham slip correction factor as a function of Knudsen number

Other coagulation mechanisms can be identified, such as laminar shear. These are not often included in

coagulation models for nuclear aerosols. One of the additional coagulation mechanisms of current research

interest is coagulation effects of electrostatic charging of aerosol particles. In most aerosol analyses, it is

assumed that the average charge on aerosol particles is zero and the distribution of charges on particles is a

Boltzmann distribution with zero mean. Certainly, this is what is expected in most laboratory tests

including those used to validate aerosol behaviour models. But, in the highly radioactive environment of a

reactor accident, there can be net charging of aerosol particles. The charging takes place because of the

intense radiation field that produces ions in the atmosphere. The irradiation of the atmosphere produces

equal amounts of positively and negatively charged ions. But, positive and negative ions have different

mobilities. In air, negative ions are more mobile than positive ions. Consequently, aerosol particles in air

are bombarded with higher fluxes of negative ions and can develop a net negative charge. The actual value

of the charge depends on the particle size. The charge is also dynamic. The net charge on any given

particle fluctuates with time. Strong, long-range electrostatic forces between particles with net charge

could have a dramatic effect on coagulation. These effects under accident conditions are elements of

current research.

2.3.2 Growth by condensation

The two principal mechanisms of aerosol growth, as opposed to aerosol formation, are condensation upon

an initially existing particle, and agglomeration/coagulation. The former process usually takes place on a

faster timescale than the latter except for the situation following homogenous nucleation, when particle

sizes are small but number densities are very high.

The aerosol growth mechanism of condensation on existing particles is sometimes referred to as gas phase

conversion. It includes both heterogeneous nucleations, where the initial seed particles are comparable in

size to but energetically favoured over the embryos of homogeneous nucleation, as well as condensation

upon larger particles, such as takes place in cooler parts of the reactor circuit and in the containment.

Chemical reaction between the gas phase and the particle surface is also possible but is not considered

NEA/CSNI/R(2009)5

here. A formula frequently used in safety codes to calculate the condensation rate on existing particles is

that due to Fuchs and Sutugin [7], which covers both the high and low-Knudsen number ranges:

2( )

11.71 1.333

FDdnn



=−





(5)

where F is the number of molecules condensing on the surface of the particle of diameter dp in unit

time, D is the diffusion coefficient, n

and n

are the molecular concentrations of the condensing species in

the bulk gas and at the particle surface respectively, and Kn is the Knudsen number. When the particle is

small the effect of the curvature upon the effective saturation pressure at the droplet surface must be taken

into account. A rearrangement of equation (2), applied to particles of size other than the critical droplet

size, yields for the difference between the vapor pressure over a curved surface and that over a plane

surface (the Kelvin effect)

liq

drop sat

kTd







(6)

where v

liq

is the liquid specific volume and σ the surface tension. This formula can be used to

calculate the bulk saturation ratio necessary to induce condensation on droplets of a given size. Smaller

droplets are seen to require higher saturation ratios in order to grow by condensation.

For the case of heterogeneous nucleation, where the initial particle radius r

is of the same order as the

critical droplet radius r

crit

it can be shown from the above equations that the rate of condensation is

proportional both to (S-1), where S is the saturation ratio, and to r

– r

crit

. Hence condensation takes place

preferentially to the larger seed particles.

Depending on the deposition process it is possible for a given class of particles to derive expressions for

the rate of particle growth. Friedlander [8], for instance, tabulates growth laws for particles of a given

diameter depending on the condensation mechanism (diffusion, surface reaction etc.). Such growth laws

may then be applied to calculate e.g. the lifetime of a droplet drying out in an unsaturated atmosphere of

saturation ratio S <1. The Langmuir equation for the lifetime t of a droplet shrinking from an initial

diameter d

to a final diameter d

8(1)

()if

sat

Mp S

Td d

−

(7)

where M is the molecular weight of the condensing species. Such lifetimes are usually short compared

with the timescales typical of containment analysis.

While particle translation is strongly damped by viscous forces for small particles, and thus has little effect

on condensation, heat transfer is an important consideration in particle growth by mass transfer.

Condensation on particles adds thermal energy, raising the particle temperature and hence the saturation

pressure. The increasing saturation pressure then acts to limit the condensation rate. Mason [9] proposed a

correction factor f

to equation (5) to account for heat transfer limitation of the condensation rate. It is

widely used in safety codes and is calculated from

1(1)

sat

pLM

fLD

kT RT

−



=+ −





(8)

where k is the gas phase thermal conductivity and L is the latent heat of condensation.

NEA/CSNI/R(2009)5

The specific case of hygroscopic particles in a humid containment atmosphere is considered in more detail

in a separate subchapter of this report.

2.4 Aerosol Growth as a Result of Hygroscopicity

Aerosol particles can also grow as a result of water adsorption. All aerosol materials exhibit some amount

of hygroscopicity. That is, they can absorb water vapor from the ambient atmosphere. For many aerosol

materials, this hygroscopicity amounts to the absorption of little more than a monomolecular layer of water

on the surfaces of primary particles that have coagulated to form a larger aerosol particle. Some materials,

those with high water solubility such as CsI and CsOH, are much more hygroscopic. They are, in fact,

'deliquescent' and will absorb and dissolve in water until the equilibrium vapor pressure of the solution is

equal to the partial pressure of water vapor in the ambient atmosphere. This water absorption causes the

particles to grow and the growth can be quite dramatic in the case of pure CsOH or CsI particles. An

example calculation of the growth of a CsOH particle initially 2.0 µm in diameter to its equilibrium size,

(eq) = 8.83 µm, is shown in Fig. 2.4-1. What is noteworthy in this figure is the time for growth of the

particle. Particle growth to the equilibrium size can occur in seconds. Because of the so-called

hygroscopic growth, such particles will settle more rapidly in a humid environment than would be

otherwise expected based on their initial sizes and growth by coagulation.

The difficulty encountered in taking hygroscopic growth of particles into account in reactor accident

analyses is that seldom are there any aerosol particles that are purely deliquescent in nature. CsOH, CsI,

and other very hygroscopic materials will be co-agglomerated with other far less hygroscopic materials.

They may, in fact, make up only a small fraction of the mass of any aerosol particle and be buried deep

within a matrix of primary particles that have little ability to absorb water. Water absorbed in these

composite particles may affect shape factors by filling voids in the composite particle, but may not amount

to sufficient growth of the particle to greatly affect deposition and settling characteristics.

Time (seconds)

0.0001 0.001 0.01 0.1 1 10

(t) / D

(eq)

0.0

0.2

0.4

0.6

0.8

1.0

Fig. 2.4-1 Hygroscopic growth of a CsOH particle initially 2 µm in diameter in an atmosphere with

relative humidity of 99 % at 363 K

NEA/CSNI/R(2009)5

2.5 Fractal Nature of Aerosols

Aerosol particles are quite small and a substantial fraction of the atoms or molecules that make up an

aerosol particle are on the particle surface. As a consequence, aerosol particles are quite reactive

chemically. When two particles encounter one another, Van der Waals forces and even chemical reactions

between particles make it quite likely that the particles will adhere. There typically are not opportunities

for multiple encounters among two particles so when the particles adhere, they do not necessarily adopt a

minimum energy configuration. In fact, agglomeration of aerosol particles leads to composites that have

fractal geometries. That is, the particle sizes are not proportional to the cube root of the volume of material

making up the composite. Under conditions of predominantly coagulation by Brownian diffusion, particle

dimensions are often found to be proportional to about the square root of the volume of the composite. As

the composites age and sinter, the proportionality evolves toward the usual cube root proportionality.

An important element of current aerosol research is the prediction of fractal characteristics of aerosol

particles. Interest in the fractal geometry arises because it can affect particle behaviour. The equations of

aerosol coagulation (and as discussed later aerosol deposition) have been derived for perfectly dense

spherical particles. Since the real particles do not have this geometry and are not perfectly dense,

correction factors have to be introduced into the equations. Two correction factors, called shape factors, are

commonly introduced:

χ = dynamic shape factor

= collision shape factor

The dynamic shape factor accounts for the higher drag of a nonspherical, porous particle. The collision

shape factor accounts for the greater spatial extent of a nonspherical particle. When the envelope of a

particle is spherical, but the particle is not perfectly dense, the dynamic shape factor and the collision shape

factor are equal but not necessarily equal to one. Aerosols in humid environments such as those expected

in reactor containment do tend to have spherical envelopes. In the reactor coolant system, such

spheroidization of the particles cannot be anticipated and the dynamic and collision shape factors can be

quite different.

It is difficult now to predict shape factors for particles. Often test data are used to 'back calculate' shape

factors. This is usually done assuming that the shape factors will be independent of particle size which is

known not to be the case. There is no assurance that the shape factors derived in this way will be applicable

to the analysis of reactor accidents.

2.6 Aerosol Deposition

The natural attenuation of radioactive material available for release from nuclear power plants during

accidents occurs because aerosol particles will deposit on surfaces in the reactor. Aerosols deposit on

surfaces because they cross stream lines of flow over the surfaces or because they extend far enough to

intercept the surface even when the particle center of mass is following a streamline. Van der Waals forces,

surface tension forces and even chemical reactions between particles and the surfaces can bind particles to

the surfaces when the velocity of impact is sufficiently low, as it usually is.

The rates of aerosol deposition on surfaces are often characterised in terms of 'deposition velocities' which

are coefficients that relate the particle flux to the particle concentration in the gas phase:

)v(n)v(V

)v(dN

NEA/CSNI/R(2009)5

where:

)v(dN

= flux of particles of volume v to a surface

)v(V

= deposition velocity of particles of volume v

)v(n = number concentration of particles of volume v in bulk gas phase

Processes that can lead to particle deposition include:

• gravitational settling

• diffusion to surfaces

• turbulent deposition

• inertial deposition

• phoretic processes

2.6.1 Gravitational settling and diffusion

Gravitational settling and diffusion nearly always affect aerosols. Gravitational settling affects the larger

particles most:

χµ













3/2

)v(Cg

)v(V

= gravitational deposition velocity (cm/s)

µ = gas viscosity (g/cm•s)

g = gravitational acceleration (cm/s

)

ρ = particle material density (g/cm

)

Of course, gravitational deposition only occurs on the upward projections of surfaces.

Diffusion affects the smaller particles the diffusion coefficient for aerosol particles is:













πµ

3/1

Tk)v(C

)v(

The diffusive deposition velocity depends, of course, on geometry and flow conditions. Van der Vate

suggests for stagnant enclosures:

(

)

s/cmD22.0V

735.0

dif

NEA/CSNI/R(2009)5

where D is in units of cm

/s. Gravitational deposition velocities and diffusive deposition velocities are

shown in Fig. 2.6-1 as functions of the equivalent spherical particle diameter. Clearly, when these two

deposition processes are summed, there will be a particle size range of minimum deposition velocity. This

is quite characteristic of aerosol deposition when two or more deposition processes are operative. The

particle size corresponding to the minimum total deposition velocity is often called the “maximum

penetration particle size”. The consequence of having such a minimum in the deposition velocity is that

particle deposition can not only reduce the total aerosol concentration, it can also affect the size

distribution of particles that remain in the gas phase. Deposition preferentially removes larger and smaller

particles, so the size distribution is narrowed around the maximum penetration particle size.

Particle Diameter (µm)

0.001 0.01 0.1 1 10 100

Deposition Velocity (cm/s)

1e-7

1e-6

1e-5

1e-4

1e-3

1e-2

1e-1

1e+0

1e+1

gravitational

diffusion

total

Fig. 2.6-1 Comparison of deposition velocities for particular circumstances

Convection and especially turbulent convection can greatly enhance particle deposition. Turbulent

deposition velocities used in reactor accident analysis models are nearly all based on empirical correlations

of data. The most notable use of empirical correlations for turbulent deposition is deposition of aerosol in

straight pipe flow.

2.6.2 Inertial deposition of aerosol particles

Aerosol-laden flows especially in the reactor coolant systems follow tortuous pathways. Whereas the gas

phase responds promptly to the accelerations caused by obstacles like bends, contractions and expansions

in the flow path, the responses of aerosol particles are delayed by inertia. Time is required for the velocity

of the particle in the direction of an obstacle to be arrested and the particle accelerated in the new direction

of gas flow. When particles are too massive to do this, they impact and presumably adhere to the obstacle

surface. This is often called 'inertial particle deposition'.

Flow patterns in the vicinity of an obstacle can be quite complicated especially under turbulent flow

conditions. It has been common, then, to use correlations of experimental data to account for inertial

particle deposition in bends, contractions and expansions of flow paths. The independent variable in the

NEA/CSNI/R(2009)5

correlations is usually the Stokes number which is the ratio of the distance required to arrest the motion of

a particle in the direction of the obstacle to some characteristic dimension of the flow path:

χµ

vD)v(C

Stk

where:

Stk = Stokes number (-)

)v(C = Cunningham slip correction factor (-)

ρ = particle material density (g/cm

)

D = particle diameter (cm)

v = gas velocity (cm/s)

µ = gas viscosity (g/cm•s)

L = characteristic dimension of flow configuration (cm)

The dependent variable in the correlations can be either the fraction of particles of a specified size that

deposit in the flow field, η(D

), or the fraction that successfully negotiate the path through the change in

the flow field. Some notable correlations are:

• Correlation of deposition efficiency in turbulent flow through a bend:

(

)

[

]

Stk963.0D1log

p10

−=η−

where the Stokes number is defined in terms of the hydraulic diameter of the pipe.

• Deposition efficiency in a contraction of the flow path from a diameter Di to a diameter Do:

(

)

x227.2x557.8721.1

+−=













−

31.0

Stkx













• Deposition efficiency from a vena contracta:

()













































++−

























−=η

−1

Stk

617.0Stk11

for 0.18 < Stk < 2.03

17.0













NEA/CSNI/R(2009)5

There has been a major dependence in reactor accident analyses on experimental determinations of

deposition velocities. Mass transfer to heat transfer analogies are very difficult to justify for aerosols

because of the very high Schmidt numbers for aerosols of interest. Recently, a major thrust of aerosol

research has been to analyse turbulent aerosol deposition in great detail using computation fluid dynamics

techniques. Often these studies are done for turbulent deposition along with other deposition mechanisms.

The studies do, no doubt, contribute greatly to the understanding of aerosol behaviour. Unfortunately, the

computational labour associated with these detailed models is incompatible with most accident analysis

codes. Indeed, even replicating published calculations would entail more labour than is usually available to

support improved aerosol modelling in reactor accident analysis codes. It would be most useful if

investigators exploring the details of aerosol deposition would correlate their findings into simplified

expressions that were similar to the correlations now used in the accident analysis models. Then, results of

these detailed analyses could be incorporated into accident analysis models promptly.

2.6.3 Phoretic deposition processes

A final, important category of aerosol deposition process is made up of the phoretic processes - usually

thermophoresis, which is particle movement in a thermal gradient, and diffusiophoresis, which is particle

movement in the flux of a condensing vapor such as steam. Both of these phoretic effects arise because the

gas impinging on one side of the particle imparts more momentum than that impinging on the other side.

Thermophoresis arises because of temperature differences between the gas phase and the surfaces exposed

to aerosol-laden gases. Such temperature differences can be quite large in the coolant system of a reactor

during a severe accident. Very hot aerosols and gases emerge from the core region of the reactor and

encounter large areas of cool surface. The thermophoretic deposition of aerosols on these surfaces was

vigorously researched thirty years ago. From this research emerged what has come to be called the Talbot

interpolation formula for the thermophoretic deposition velocity:

(

)

()

nT1

KnC2

1KnC31

KnC

kvC2

therm

∇













+++













χρ

−=

where:

C = 1.128

()

α−

()

α−

α = momentum accommodation coefficient

α = thermal accommodation coefficient

k = thermal conductivity of the gas phase

k = thermal conductivity of the aerosol particles

The thermophoretic deposition velocity is not an especially strong function of particle size. It does depend

on the thermal conductivity of the aerosol particles relative to the thermal conductivity of the gas phase.

NEA/CSNI/R(2009)5

Seldom is this particle thermal conductivity well known relative to the ability to estimate the thermal

conductivity of the gas phase. Usually bulk material thermal conductivities are used though it is

recognised that this may be quite in error. The thermophoretic deposition velocity also depends on

accommodation coefficient for gas impinging on particle surfaces. There does not appear to be much of a

useful data base on such accommodation coefficients and there are not readily accessible theories for

estimating accommodation coefficients. Fortunately, the velocity is not especially sensitive to

accommodation coefficients in the plausible range of possible values.

Recently, there has been a renewed interest in thermophoresis. Researchers have obtained detailed

solutions of approximate forms of the Boltzmann equation to assess thermophoretic deposition of particles.

Their numerical analyses do identify ranges where particles will move up rather than down the thermal

gradient. This is called 'negative thermophoresis'. Their numerical results show excellent agreement with

data taken in regimes predicted to be challenging to the theory. It is unlikely, however, that such detailed

analyses will ever find their ways into accident analysis computer codes. The Talbot interpolation formula

also shows good agreement with the available experimental data though the formula will not predict the

negative thermophoresis found possible in the more detailed analyses of thermophoresis.

Diffusionphoresis is not usually encountered in the reactor coolant system during a reactor accident since

large amounts of vapor are not condensing on surfaces in the reactor coolant system. Diffusiophoresis can

arise in the containment of a reactor under accident conditions. It is an especially important aerosol

removal process in the AP600 reactor where the steel containment is cooled by water on the outside. The

usual form of the diffusiophoretic deposition velocity is:

()

(

)

O2H

Total

12phor

OHD

1V ∇

−

























−σ

+−=

where:

phor

V = diffusiophoretic deposition velocity

σ = scattering kernel

P = partial pressure of the condensing vapor which is usually steam

()

OHD

= diffusion coefficient of the condensing vapor

A variety of expressions for the scattering kernel have been proposed. Williams has formulated a very

general expression for the kernel to account for diffuse, specular, Lambert and backward reflection of

molecules impinging on the particle surface. Others have considered both momentum and energy

accommodation in the formulation of the scattering kernel. Unfortunately, the nature of the scattering of

gases from aerosols in reactor accidents is not well known and such sophisticated models are seldom found

in reactor accident analysis codes.

2.7 Resuspension of Deposited Particles

Aerosol particles deposited on surfaces can be resuspended in the gas phase. Resuspension can occur when

there is:

• a sudden increase of gas flow over the surface

• shock or vibration of the deposition surface

NEA/CSNI/R(2009)5

Experimental data on resuspension shows that increased gas flow and shock or vibration of the deposition

surface can act synergistically to cause accentuated resuspension.

Events are expected in reactor accidents that lead to resuspension of deposited particulate. These events

include:

• sudden generation of steam when core debris relocates from the core region to the lower plenum

of the reactor vessel,

• steam generation associated with attempts to quench the core,

• rupture of the reactor pressure vessel,

• violent fuel-coolant interactions such as steam explosions, and

• hydrogen combustion events.

Experimental investigations of resuspension phenomena have focused on the effects of sudden increases in

gas flow. These studies have shown that gas flow increases produce a significant initial resuspension

followed by a longer term, much lower rate of particle resuspension. The studies also show that the aging

of aerosol particles on surfaces makes resuspension more difficult.

The first models of resuspension were based on a balance between the forces holding particles to a surface

(Van der Waals forces, surface tension, chemical bonding, etc.) and lift forces from the flowing gas. Such

models could predict the initial, intense resuspension, but these early models could not predict the longer

term, low intensity resuspension. More recent models have approached modeling resuspension in two

ways:

• turbulent eddies are predicted to penetrate the laminar boundary layer to the surface and scour

deposited particles from a surface region approximately the size of the penetrating eddy,

• particles vibrate in a potential well created by the adhesive forces. Episodic fluctuations in the

vibrations put particles in a state where lift forces can overcome the weakened adhesive forces.

Even more recently, resuspension research has focused on the potential for depositing particles to impart

sufficient momentum to already deposited particles to cause resuspension.

2.8 Aerosol Removal by Engineered Safety Systems

The natural aerosol deposition processes discussed above will eventually cleanse the gas phase of

particulate. These processes often operate over a longer time scale than the progression of reactor

accidents. Fortunately, engineered safety systems, installed in reactor containments to reduce pressure

during design-basis loss of coolant accidents, are very effective also at removing aerosol from the

containment atmosphere. Important engineered safety features are:

• containment sprays

• steam suppression pools

• fan coolers

Sprays

Sprays flood the containment atmosphere with water droplets. The size distribution of water droplets from

a type of spray nozzle commonly found in pressurised water reactor is shown in Fig. 2.8-1. The spray rates

NEA/CSNI/R(2009)5

can be quite high so the number of density of water droplets is high. After an initial transient when spray

droplets condense excess steam in the containment atmosphere, spray droplets remove aerosol particles

from the atmosphere by a combination of three processes - diffusion, interception and impaction. The

capture of particles by falling spray droplets is usually expressed in terms of capture efficiency. This

efficiency is the ratio of particles actually captured to the number of particles in the droplet path:

()

(

)

()

LdnD

dN4

∆

=∈

′

where:

(

)

′

∈ = capture efficiency for droplet of diameter D

and particles of diameter d

(

)

dN∆ = number of particles of diameter d

captured

L = droplet path length

Impaction describes the impact on droplets by particles that cannot follow the streamlines around the

droplet. Of course, it is the larger aerosol particles that cannot follow these streamlines. A commonly used

expression for the capture of these large particles is:

where:

dgT

= droplet terminal velocity

Stk = Stokes number

(

)

χµ

UdvC

δ = 0.25 to 0.75

Number-weighted Droplet Size Distribution

0 200 400 600 800 1000 1200 1400 1600

Probability Density

Droplet Diameter (µm)

Fig. 2.8-1

Number-weighted droplet size distribution for a common containment spray

NEA/CSNI/R(2009)5

Interception occurs when the center of mass of a particle is following a streamline closer to the droplet

surface than the particle radius. A common expression for the efficiency of interception is:

()

() ()

60/Re1

60/potint,Reviscint,

int

∈

+∈

=∈

()

3/1

D/d1

D/d5.1

viscint,

γ+

=∈

()

potint,

=∈

A variety of expressions are available for estimating the collection efficiency due to diffusion. An example

is:

()

(

)

8/32/1

ScRe557.02

ScRe

diffusion +=∈

where:

Sc = particle Schmidt number =

Dρ

= particle diffusion coefficient =

(

)

χπµ

kTvC

The aerosol capture mechanisms are not entirely independent. A common expression for the combined

effects of the three mechanisms is:

() ( )

[]

(

)

[

]

()

[

]

diffusion1erceptionint1impaction11total ∈−∈−∈−−=∈

Total aerosol capture efficiencies for 1000 µm and 200 µm droplets as functions of aerosol particle size are

shown in Fig. 2.8-2. The capture efficiency passes through a minimum for both droplet sizes. This, of

course, means that spray droplets both remove particles and narrow the size distribution. The minimum

varies in location with the size of the spray droplet. Accurate calculation of the effects of spray removal

requires that the size distribution of the spray droplets be considered. This is not often done because it

requires that the coagulation of the spray droplets during flight be calculated as well. Alternative

mechanistic modelling has been developed based on recent analytical work; this modelling is described in

Appendix 3.

NEA/CSNI/R(2009)5

Spray Droplet Removal of Aerosol Particles

Particle Diameter (mm)

0.001 0.01 0.1 1 10 100

Total Capture Efficiency

0.0001

0.001

0.01

0.1

1000 µm droplet

200 µm droplet

Fig. 2.8-2 Comparison of the capture efficiencies of 1000 and 200 µm droplets

2.8.1 Steam suppression pools

Calculation of the aerosol removal by steam suppression pools is quite involved. Much aerosol removal

occurs during the formation of the bubble when diffusiophoresis occurs as steam condenses and impaction

occurs at the stagnation of the flow against the growing bubble envelop. Once the bubble detaches and

begins to rise in the pool, further aerosol removal can occur by diffusion and gravitational settling within

the bubble. Diffusion within the bubble is resisted by expansion of the bubble as pressure head is lost

during bubble rise. Vapor flow into the bubble when the water pool is near the boiling point can resist

particle removal. Gases within the bubble circulate during bubble rise, so there can be aerosol removal by

inertial deposition as aerosols are unable to follow streamlines of flow in the ellipsoidal bubble. Other

removal mechanisms can arise. For instance, oscillation of the bubble can remove particles because of

delay in the response of particles to changes in bubble shape.

Aerosol removal from gas bubbles sparging through water depends on the depth of the water, but can

produce substantial decontamination. The decontamination is usually expressed in terms of the

decontamination factor, DF:

poolemerginggasinionconcentrataerosol

initiallyionconcentratmassaerosol

DF =

Decontamination factors calculated for gases sparging through the so-called 'T' quenchers in the steam

suppression pools of boiling water reactors are shown in Fig. 2.8-3. Even a pool as shallow as about one

meter will remove over 90 % of the particulate in the gas stream. It is to be emphasised that the removal

also narrows the size distribution of the remaining aerosol around the maximum penetration aerosol

particle size. This narrowing of the particle size distribution around a small particle size that is inefficiently

removed by the combination of diffusion, gravitational settling and impaction mechanisms makes

subsequent filtration processes for the residual aerosol much less efficient than would be expected based on

the size distribution of the aerosol entering the suppression pool.

NEA/CSNI/R(2009)5

Pool Depth (cm)

0 100 200 300 400 500 600 700 800

Decontamination Factor

200

400

600

800

1000

Fig. 2.8-3 Aerosol removal by sparging through a water pool

2.8.2 Fan coolers

Fan coolers can remove aerosol by thermophoresis and by the impaction of particles on rotating fans.

Thermophoretic removal decreases with time as the depositing aerosol develop an insulating layer on the

cool surfaces.

2.9 Aerosol Removal by Filtered Vents

A final protection against radioactive material released from reactors under accident conditions is to filter

the gas stream. Three types of filters are commonly used:

• fibre filters used to protect particular locations such as the control room

• venturi scrubbers

• gravel bed filters

Fiber filters remove particles by trapping them. The overall penetration of a fresh fiber filter, i. e. before it

is loaded with particles, can be expressed by:

()













−

−=

ftot

exp

απ

= fraction of aerosol penetrating fiber filter

α = volume fraction of fibers in filter, i.e. 1.0 - porosity

L = thickness of fiber bed

= fiber diameter

tot,f

= total efficiency of a single fiber at particle trapping

NEA/CSNI/R(2009)5

The efficiency of a fibre at trapping particles is calculated considering the mechanisms:

• gravitational settling

• inertial impaction

• interception

• diffusion

• thermophoresis

• diffusiophoresis

Electrostatic charging of the filter and particles can also be considered especially for radioactive aerosols.

The filter collection efficiency η and the decontamination factor DF are related to the overall penetration

η = 1 – P

respectively

The overall penetration, the collection efficiency and the DF depend strongly on the particle size. The

characteristics of the filter, like collection efficiency and pressure drop, change as mass is collected and

reduces the porosity of the filter.

Venturi scrubbers precondition the aerosol by injecting water (about 1 liter of water per cubic meter of gas)

along with gas through a constricting throat. Gas velocities can be as high as 120 m/s. The liquid water is

“atomised” into small droplets at the high velocities. Water droplets can coagulate with the aerosol or

otherwise capture the aerosol particles so that they are more efficiently removed by a subsequent filtration

system which might be a water pool or a gravel bed.

Gravel beds remove particles by deposition onto the large surface area provided by the gravel. Submerged

gravel beds use water to wash away deposited materials from the gravel.

Mechanistic models for the gravel bed filtration efficiency are based on the single grain capture efficiency,

extended to a bed of particular size. The penetration of a fresh granular bed filter is given by













−=

gtot

exp

= fraction of aerosol penetrating gravel bed

α = volume fraction of grains in filter, i.e. 1.0 - porosity

H = gravel bed depth

= grain diameter

tot,g

= total efficiency of a single grain at particle trapping

tot,g

for a single grain is comparable to the single fibre-capture efficiency used for fibrous filters. The

particles are captured mainly by a combination of particle diffusion, interception and impaction.

NEA/CSNI/R(2009)5

2.10 Computer Modelling of Aerosols

The aspect of aerosol modelling that does pose computational challenges is the solution of the dynamic

equation to describe the evolution of the aerosol size distribution as a result of particle coagulation. As

discussed near the beginning of this chapter, the evolution of the size distribution is described by a

complicated intergrodifferential equation.

In the 1970's at the time of the Reactor Safety Study, this differential equation was solved using what were

called “moment’s methods”. These methods involved hypothesizing a particle size distribution with

undetermined parameters. Typically the lognormal size distribution was hypothesized. The mean and

geometric standard deviation of the distribution were then determined through the solution of the

differential equation consistent with conservation of mass. These methods were found through an extensive

effort of experimental and analytical research to yield inadequate results during periods when aerosols

were being injected into a pre-existing aerosol with a different chemical composition. This situation gave

rise to multi-modal size distributions for the aerosol that could not be easily described by a single

distribution with limited numbers of parameters.

Today, most accident analysis models solve the aerosol dynamic equation using what are called 'sectional

methods'. Sectional methods were introduced in MAEROS and NAUA models for the first time. These

methods divide the aerosol size distribution into a set of contiguous intervals. Each interval is characterised

by a representative particle size. The differential equations are solved for these representative sizes. The

methods do suffer from numerical diffusion that can be controlled to some extent by the judicious choice

of the size bins. The sectional methods are not especially powerful in situations were vapors and steam are

condensing on the aerosol particles. Williams and Loyalka (see Bibliography) provide further critique of

these methods. For vapors and steam condensation fixed grid and moving grid methods were developed.

The MGA (Moving Grid Analysis) is successfully applied in combination with the MAEROS model

through a variety of tests. MAEROS/MGA are incorporated in codes like ASTEC, COCOSYS, CONTAIN

and MELCOR.

2.11 Annotated Bibliography

This introduction has not been intended to be comprehensive and the literature cited in this bibliography is

similarly incomplete. The works cited here are selected to provide an introduction to the field of aerosol

physics pertinent to the issues of reactor accident analyses. Annotations are provided to assist the reader in

the selection of documents of interest.

General aerosol physics

The classic reference in the field is:

− N.A. Fuchs,

The Mechanics of Aerosols, Pergmon Press, 1964

Readers may find:

− S.K. Friedlander, Smoke,

Dust and Haze: Fundamentals of Aerosol Behaviour, J. Wiley and

Sons, 1977.

more readily digested. For a work with much greater mathematical sophistication, see:

− M.M.R Williams and S.K. Loyalka,

Aerosol Science Theory and Practice with Special

Applications to the Nuclear Industry

, Pergamon Press

NEA/CSNI/R(2009)5

These texts will provide an introduction to the topics discussed in this chapter as well as other topics not

listed in this bibliography.

Nucleation

A good review of the experimental and analytical modeling of the nucleation of pure vapors is provided in:

− E. Honatan,

Theoretical and Experimental Research on Homogeneous Nucleation Phenomena,

A State of the Art Report

, Catedra de Tecnologia Nuclear E.T.S. de Ingeniero Industriales, Madrid,

1991.

The upper and lower bounds on the rates of pure vapor nucleation are modelled in:

− H. Reiss, J.I.Katz, and E.R. Cohen,

J. Chem. Phys., 48 (1968)5553.

Aerosol hygroscopicity

An account of aerosol growth by water adsorption is found in:

− J. Jokiniemi, 'The Growth of Hygroscopic Particles During Severe Core Melt Accidents',

Nuclear

Technology, 83

(1988) 16.

Details on the derivation of models of hygroscopic growth of pure particles by water absorption including

both mass and heat transfer are provided in:

− F.R. Newbold and N.R. Amundson, 'A model for Evaporation of a Multicomponent Droplet',

American Institute of Chemical Engineers, Journal, 19 (1973) 22.

Fractal nature of aerosols

A modern discussion of fractal dimensionality of aerosols that will provide an introduction to a growing

literature base is:

− M. Kostoglou and A.G. Konstandopoulos, 'Evolution of aggregate size and fractal dimension during

Brownian coagulation',

J. Aerosol Science, 32 (2001) 1399.

Cunningham slip correction factor

A recent derivation of the expression for the Cunningham slip correction factor from the Milliken oil drop

experimental data and a discussion of other expressions is presented in:

− M.D. Allen and O.G. Raabe,

Aerosol Science and Technology, 4 (1985) 269.

A more theoretical expression for the slip correction factor is provided by:

− W.F. Phillips,

Physics of Fluids, 18 (1975) 1069.

Electrostatic charging of an aerosol in a radiation field

Radioactivity of the aerosol material and the presence of an intense radiation field is what makes nuclear

aerosols different from aerosols encountered in other fields. The potential of electrostatic charging of

aerosols by the radiation field is discussed in:

NEA/CSNI/R(2009)5

− C.F. Clement and R.G. Harrison, Enhance Localised Charging of Radioactive Aerosols,

J. Aerosol

Science, 31

(2000) 363.

Diffusive deposition in enclosures

− J.F. van der Vate,

Investigations into the Dynamics of Aerosols in Enclosures Used for Air

Pollution Studies

, ECN-86, Netherlands Energy Research Foundation, Petten, The Netherlands,

July 1980.

Turbulent deposition of aerosol

The classic study of aerosol deposition in smooth pipes is:

− B.Y.H. Liu and J.K. Agarwal,

J. Aerosol Science, 5 (1974) 145.

Deposition in rough pipes is discussed in:

− L.A. Hahn, J.J. Stukel, K.H. Leong and P.K. Hopke, 'Turbulent Deposition of Submicron Particles

on Rough Walls',

J. Aerosol Science, 16 (1985) 81.

Inertial deposition of aerosols

Aerosol deposition in bends is discussed in:

− R.I. Crane and R.L. Evans,

J. Aerosol Science, 8 (1977) 161.

− D.Y. Pui, F. Romay-Novas, and B.Y. Lui, 'Experimental Study of Particle Deposition in Bends of

Circular Cross Section',

Aerosol Science and Technology, 7 (1987) 301.

− C-J. Tsai and D.Y.H. Pui, 'Numerical Study of Particle Deposition in Bends of a Circular

Cross-Section-Laminar Flow Regime',

Aerosol Science and Technology, 12 (1990) 813.

Aerosol deposition in sudden contractions and expansions of the flow path is discussed in:

− G. Ahmadi and Q. Chen, 'Dispersion and Deposition in Turbulent Pipe Flow with Sudden

Expansion',

J. Aerosol Science, 29 (1998) 1097.

− Y.Ye and D.Y. H. Pui, 'Particle Deposition in a Tube with an Abrupt Contraction',

J. Aerosol

Science, 21

(1990) 29.

Aerosol deposition in a vena contracta:

− S.P. Belyaev and L.M. Levin, 'Investigation of Aerosol Aspiration by Photographing Particle Tracks

Under Flash Illumination',

Aerosol Science, 3 (1972) 127.

Thermophoresis

The Talbot interpolation formula is described in:

− L. Talbot, R.K. Cheng, R.W. Schefer and D.R. Willis,

J. Fluid Mechanics, 101 (1980) 737.

NEA/CSNI/R(2009)5

Modern analyses that show instances of negative thermophoresis are:

− K. Yamamoto and Y. Ishihara, 'Thermophoresis of a spherical particle in a rarefied gas of a

transition regime', Physics of Fluids, 31 (1988) 1421.

− S. K. Loyalka, 'Thermophoretic force on a single particle - 1. Numerical solution of the linearized

Boltzmann equation',

J. Aerosol Science, 23 (1992) 291.

Comparison of models to experimental data on thermophoresis is presented in:

− F. Zheng and E.J. Davis, 'Thermophoretic force measurement of aggregates of microspheres',

J. Aerosol Science, 32 (2001) 1421.

Diffusiophoresis

The classic references on diffusiophoresis are:

− L. Waldmann and K. H. Schmitt, 'Thermophoresis and Diffusiophoresis' Chapter VI in

Aerosol

Science

, C.N. Davies, editor, Academic Press, 1966.

− P. Goldsmith and F.G. May, 'Diffusiophoresis and Thermophoresis in Water Vapor Systems',

Chapter VII in

Aerosol Science, C. N. Davies, editor, Academic Press, 1966.

The scattering kernel in the expression for the diffusiophoretic deposition velocity is discussed in:

− M.M.R. Williams,

Z. Naturforsch., 27 (1972) 1798 and 1804.

Resuspension

An overall review of resuspension modeling and experiments is provided in:

− G. Ziskind, M. Fichman and C. Gutfinger, 'Resuspension of Particulates from Surfaces to Turbulent

Flows - Review and Analysis',

J. Aerosol Science, 26 (1995) 613.

A recent review of the issues of resuspension is:

− L. Biasi, A. de los Reyes, R.W. Reeks, and G.F. de Santi, 'Use of a simple model for interpretation

of experimental data on particle resuspension in turbulent flows',

J. Aerosol Science, 32 (2001)

1175.

The issue of resuspension caused by the impaction of other aerosol particles is addressed in:

− W. John, D.N. Fritter and W. Winklmayr, 'Resuspension induced by impacting particles',

J. Aerosol

Science, 22

(1991) 723.

Spray removal of aerosol

− D.A. Powers and S.B. Burson,

A Simplified Model of Aerosol Removal by Containment Sprays,

NUREG/CR-5966, Sandia National Laboratories, Albuquerque, NM, June 1993.

NEA/CSNI/R(2009)5

Decontamination of aerosol by steam suppression pools

− D.A. Powers,

A Simplified Model of Decontamination by BWR Steam Suppression Pools,

NUREG/CR-6153, Sandia National Laboratories, Albuquerque, NM, May 1997.

Computer modelling of aerosol coagulation

Some of the complexities of the aerosol dynamic equation are described in:

− R.L. Drake, 'A General Mathematical Survey of the Coagulation Equation', in

Topics in Current

Aerosol Research

, Volume 3, G,M, Hidy and J.R. Brock, editors, Pergamon Press, 1972.

The sectional method for calculation of aerosol coagulation is described in:

− F. Gelbard and J.H. Seinfeld, 'Simulation of Multicomponent Aerosol Dynamics',

J. Colloid

Interface Science, 78

(1980) 485.

For a comparison of moments methods, sectional methods and other methods, see:

− C. Seigneur, A.B. Hudshwekyj, J.H. Seinfeld, K.T. Whitby, E.R. Whitby, J.R. Brock, and H.M.

Barns, 'Simulation of aerosol dynamics: a comparative review of mathematical models',

Aerosol

Science and Technology, 5

(1986) 205.

An attempt to overcome the problems of numerical diffusion is discussed in:

− T.H. Tsang and A. Rao, 'Comparison of different numerical schemes for condensational growth of

aerosols',

Aerosol Science and Technology, 9 (1988) 271

NEA/CSNI/R(2009)5

3. ACCIDENT PHENOMENOLOGY

3.1 Formation and Growth

3.1.1 Phenomena considered and state of knowledge

Successive stages in core degradation during a severe accident in a PWR or BWR result in the release of

both vapors and aerosol particles which differ in composition depending on their dominant source, be it

control rod and fuel burst events, cladding oxidation, fuel heat-up accompanied by geometric changes,

molten pool formation, or reflooding of the degraded core. Temperatures in the core region are generally

considerably higher than those in the plenum above the core, which in turn are higher than those in the

primary circuit. Concentrations are also highest in the core region, and hence the vapor/aerosol mixture

carried by the steam/hydrogen stream from the core to the assumed breach in the primary system and out

into the containment vessel or the environment undergoes a succession of cooling transients which induce

both condensation of vapors upon existing aerosols (heterogeneous nucleation) and the formation of new

particles (homogeneous nucleation). Competing with the former process is condensation upon structures,

which serves to remove material from the carrier stream, at least temporarily. The two processes mentioned

do not alter the elemental content of the carrier stream, but instead affect removal by vapor condensation

further downstream and also the mass, size distribution and chemical composition of the suspended aerosol

and thus its deposition behaviour. The size distribution of the aerosol is also strongly affected by

agglomeration, and in a simulation of fission product transport during a severe accident it is necessary to

take into account all the formation and growth processes mentioned.

3.1.1.1 The state-of-the-art report, 1994

At the time of writing of the State of the Art report on Primary System Fission Product Release and

Transport [1] physically-based models of all these processes were already available. The SOAR included a

survey of the models available in the literature for homogeneous and heterogeneous nucleation, for vapor

condensation on aerosols, and for aerosol agglomeration, and noted that these and other models had been

incorporated in circuit codes such as VICTORIA, which were being applied both to separate-effect test

series e.g. FALCON and to reactor sequences. In the absence of substantial experimental data on the

formation and growth of nuclear aerosols during transport under severe accident conditions the SOAR

could only detail the competing or reinforcing processes at work in such formation and growth without

being able to come to firm conclusions on their relative importance and combined effect under given

thermal-hydraulic conditions.

Concerning growth through

Vapor interaction with aerosols the SOAR concluded that while safety

importance (in the circuit) was high, understanding was incomplete. There is a clear need for

vapour-aerosol interaction data. Vapor-surface interaction tests may need to be done with more complex

FP species. Data on appropriate species will hopefully come from PHÉBUS FP tests. There is nothing

more about this topic in Chapter 5 of the SOAR, but in the appropriate section of Chapter 4, Section 4.6,

we read “Interactions of fission product vapors with primary circuit surfaces and aerosols can

substantially modify the magnitude and nature of the source term to the containment. However, the relative

importance of the processes is uncertain: while the extensive surface area of the aerosol surfaces indicates

that vapor-aerosol interactions should predominate, heat and mass transfer limitations can considerably

affect the balance. … few studies have addressed vapor interactions with suspended aerosols.

NEA/CSNI/R(2009)5

Consideration of timescales indicates that the latter processes will be governed by diffusion limitations

rather than chemical kinetics. Lack of data on the diffusion of fission product vapors through reactor gases

prevents accurate prediction of condensation phenomena. (Representative) data (e.g.from PHÉBUS FP)

should guide the requirement for separate-effects studies and the development of more sophisticated

models to treat these processes (vapor condensation on suspended aerosols and on deposits).”

Aerosol nucleation and growth, on the other hand was judged by the writers of the SOAR to be of

medium safety importance. Concerning the state of knowledge, existing data appeared to be adequate but

further modelling or reveiw/consolidation of data was required. The need for detailed nucleation/growth

modelling in the codes has not been clearly established. Chapter 4 provided more detail and recommended

further assessment. The importance of mechanistically modelling aerosol nucleation has not been assessed.

At issue is whether ad hoc treatments of nucleation phenomena are adequate for the purposes of reactor

safety analyses. We recommend that coordinated efforts be initiated to determine the importance of aerosol

nucleation … to RCS fission product transport analyses. These assessments would provide the basis for

determining if experiments are needed in addition to those already performed or planned.

In another section the SOAR writing group offered a more general comment on the impact of

thermal-hydraulic and core-degradation uncertainties on FP release and RCS transport: We view these

issues as sufficiently important to be considered regarding release and transport analyses and

experimental design. The following sections of this sub-chapter will attempt to review the progress made

since the appearance of [1] in the understanding and prediction for nuclear safety purposes of aerosol

formation and growth, including vapor condensation on aerosols, nucleation and agglomeration. A final

section on containment aerosols will highlight the similarities and differences between the formation and

growth of such aerosols and of those in the primary circuit, in the light of PHÉBUS FP results and those of

ISPs such as ISP-37 and ISP-44. The subchapter ends with conclusions and some proposed

recommendations.

3.1.1.2 Theory of formation and growth

This section presents some of the principal equations used to model the formation and growth aerosols for

nuclear safety purposes, together with remarks on their physical basis and applicability. A number of

references in the open literature are also included. For further details on the practical calculation of the

formation and growth processes considered in safety evaluations the theoretical manuals of the most

widely used codes may also be consulted.

Formation

Aerosol particles may be formed by a wide range of processes such as gas-phase chemical reactions,

attrition of solids, resuspension etc. Here we will consider the theoretically and practically important topic

of homogeneous nucleation, the formation of gas-borne particles by condensation of a supersaturated

vapor. Binary and ternary nucleation, in which several vapor species condense together, is important eg in

atmospheric science, but is not normally considered in nuclear safety. Vapor condensation onto existing

seed particles of another material, termed heterogeneous nucleation when the seed particles are sufficiently

small, is considered subsequently.

In homogeneous nucleation a thermally activated first-order transition takes place from the metastable

vapor phase to form local concentrations of the vapor molecules which may be considered the precursors

of fully formed liquid drops. The liquid phase has a lower chemical potential, but there is a free energy

“cost” through the formation of a phase boundary with its corresponding surface energy. Clusters of a

certain critical radius r

crit

or greater will tend to grow still larger, while those of smaller size will tend to

lose molecules and hence shrink. The classical theory of nucleation considers that the probability of

NEA/CSNI/R(2009)5

attaining a cluster of a given size is exponentially related to the energy necessary to form it [2]. This

energy is the difference between a volume term proportional to the logarithm of the saturation ratio S of

the free vapor, S being defined by

S = p

vap

sat

(1)

where p

sat

is the saturation pressure, and a surface term proportional to the surface tension σ. The free

energy reaches a maximum when the droplet radius takes the value r

crit

, leading to the following equation

for the critical radius:

crit

liq

kT S

(2)

where v

liq

is the specific volume of the liquid. Classical nucleation theory [3] proceeds to calculate the

nucleation current J, i.e. the rate of liquid formation per unit volume, assuming a population of clusters

with a size distribution determined by the free energy of formation, the single-molecule arrival rate being

calculated from kinetic theory. Its final result is the following equation for the nucleation rate per unit

volume:

()()













−













232

ln3

exp

STkn

Bliq

liq

vap

CNT

πσ

(3)

where n

vap

and n

liq

are the bulk vapor and liquid number densities respectively and M is the molecular

weight.

Some comments are in order. The numerical predictions of classical nucleation theory are usually not very

accurate (they are to within 5-10 orders of magnitude in most cases) but it successfully predicts the

isothermal variation of the nucleation rate with saturation ratio. However it overpredicts the variation of

the nucleation rate with temperature. Numerous attempts have been made to ameliorate the theory (for a

review see [4]), improving on the assumption that there is no free energy cost in forming a cluster

consisting of a single molecule or taking into account such matters as the curvature-dependence of the

surface tension or the translational invariance of the nucleation process, which contributes to the energy via

the entropy of mixing. Disappointingly, efforts to render classical nucleation theory more self-consistent in

these and other ways have not generally improved its predictions.

This may not matter greatly in nuclear safety evaluations, since for sufficient supersaturations all theories

in current use for that purpose predict the rapid formation of a very large number of very small primary

particles. The nucleation theorem [5] as extended by Ford [6] provides a formula for the number of

molecules in the critical droplet nucleus, n

crit

in terms of the nucleation rate J

nuc

crit

∂



=−



∂



(4)

from which it may be calculated that a critical nucleus contains 30 - 50 molecules. In situations of safety

interest where homogeneous nucleation occurs the resulting tiny droplets or embryos rapidly agglomerate

to form particles in the 100 - 1000 nm range, referred to in atmospheric science as the accumulation mode,

NEA/CSNI/R(2009)5

and the precise sizes, number densities and rates of formation of the original embryos which give rise to

them may not be of significance.

Agglomeration and coagulation

Agglomeration usually refers to the formation by collisions of loosely-packed assemblages of primary

particles, often highly irregular in shape, such as certain fire aerosols [10]. If the aerosol so formed

includes a liquid component its constituents may rearrange to produce roughly spherical particles, and the

process is then termed coagulation. As an example, the particles found in the hotter portions of the

PHÉBUS circuit are agglomerates of primary particles, whereas those found in the containment vessel are

more rounded and compact as a result of water adsorption. Chapter 2 of this report gives the main

mechanisms of aerosol agglomeration, as well as appropriate formulae for the individual agglomeration

kernels to be used in the General Dynamic Equation. That chapter also supplies a recipe for computing the

kernel when several agglomeration processes act simultaneously.

More detailed information for the circuit is given in chapter 3.3.

3.1.2 Particle formation in the primary system

3.1.2.1 PHÉBUS FP tests

The information subsequent to the SOAR of 1994 on particle formation comes largely from the PHÉBUS

FP tests, mainly PHÉBUS FPT1, which was the subject of ISP-46, but also test FPT0 [13] and test FPT2,

which is currently under analysis. All three tests have the great advantage of working with prototypical

materials, and in the latter two tests, used fuel rods of moderate irradiation. They also feature sections of

the test circuit operating at temperatures and concentrations characteristic of the degrading core, in-vessel

volumes, hot primary circuit elements (1000 K), steam generator tubes and the PWR containment

respectively. Both high-steam and high-hydrogen conditions have been investigated, however, tests have

been performed with and without boric acid in the coolant stream, and in test FPT3 which was executed in

2004, but is not yet finally analysed, the silver-indium-cadmium control rod of previous tests was replaced

by a control rod of boron carbide clad in steel. These tests are considered the best available for the integral

investigation of the combined processes governing aerosol growth and vapor condensation. The remarks

below are based on PHÉBUS tests FPT0, FPT1 and FPT2.

3.1.2.2 Particle composition

Concerning particle formation, it is found that the particles sampled from the simulated upper plenum and

hot leg components are largely composed of silver, together with other elements in proportions which vary

depending on the stage of the degradation scenario reached at the time of sampling. During the period of

zircaloy oxidation in FPT1, for instance, the particle composition (weight percent) is approximately

Ag: 30, Sn: 20, In: 20, Cd: 10, with fission products such as Cs and Mo making up the remainder. All the

elements cited arise from the control rod apart from Sn, released from the zircaloy cladding on its

oxidation. During the subsequent phase in a typical PHÉBUS transient, where heat-up due to decay heating

in a reactor core is simulated by increasing the driver core power to compensate for lateral heat losses from

the fuel bundle, the particles, while still predominantly silver, contain significant proportions of the fission

products Cs and Mo, whereas during the late phase where a molten pool is formed which then relocates to

the bottom of the test section the particles remain silver-rich but contain proportions of uranium and the

less volatile fission products as well as the more volatile ones seen in previous phases. In mass terms,

silver accounted for the majority of the total release, Cd, Sn and In for 25 %, Cs and Mo together for

another 9 % and the remaining fission products including Te, Rb, Tc and I amounted to less than 5 %.

“Structural materials”, mainly silver, thus formed the aerosol stream on which the radiologically interesting

NEA/CSNI/R(2009)5

fission products were carried as inclusions or condensates, and the location and fate of the latter may

largely be deduced by modelling the transport of the structural materials and the interaction of fission

products with them. The above remarks refer largely to the results of PHÉBUS tests FPT0 and FPT1. The

release of control materials was high in these tests, but in an accident it will depend on the degradation

scenario. In PHÉBUS test FPT2 it was found that there was little silver release until the control materials

relocated to lower portions of the bundle. The early fission product releases thus took place when silver

was in short supply in the primary circuit. The consequences for particle composition in this test are still

under investigation.

Chemically the state of the structural materials remains relatively simple. Silver is present as the metal, In

and tin as oxides. The chemical form of cadmium is less easy to determine.

3.1.2.3 Particle morphology

The particles from the hot portion of the PHÉBUS circuit have an AMMD of 1.5 – 2 µ and as noted above

are principally composed of metallic silver. When examined under a microscope they are seen to consist of

agglomerates of primary particles of diameter < 200 nm. The agglomerates appear fairly solid in nature,

with no sign of the hollow spheres typical of fly ash, for instance. The size range observed is too small for

the primary particles to have been created by droplet fragmentation or other mechanical means, and too

large for the particles to be the direct result of homogeneous condensation, and it is most likely that they

are formed through a multistage process in which different formation and growth mechanisms dominate at

different stages.

3.1.2.4 Formation and growth of primary particles in the core and pressure vessel

Although some seed particles will be present in the flow channels of the core and in the pressure vessel

during fission product release, it appears most probable that the first stage in the creation of the primary

particles of silver which provide the backbone and building material for particle transport arise through

homogeneous nucleation [14, 15]. The particle mass concentration in the portion of the PHÉBUS

installation where temperatures in the right range prevail (3000 to 1500 K, [1]) is most representative in

test FPT2, and takes a value of the order of 5 g.m

-3

of silver. Although considerably higher values may be

found during surges in silver release these densities are always much less than the saturation density of

silver at the higher end of this temperature range, implying that the rate of transfer from liquid to silver

vapor is subject to kinetic, heat transfer or other limitations. The unsaturated carrier stream from the core

then cools in the upper plenum region and homogeneous nucleation of silver droplets occurs. The

formation of metal aerosols at high temperature through rapid cooling is well-known in metallurgy,

volcanology etc., and it is generally accepted that rapid cooling favours homogeneous over heterogeneous

condensation, resulting in droplets with sizes in the 10 nm range [16, 17]. As mentioned above the primary

particles seen in PHÉBUS tests are ten times larger than this, implying either that the droplets must

coalesce before freezing or that rapid condensation of silver takes place on the homogeneously nucleated

silver droplets, causing their growth to the size observed. Possibly both processes occur. Further cooling

then produced the solid silver “primary particles” seen in the PHÉBUS tests.

3.1.2.5 Effect of pressure

The formation scenario described is based on PHÉBUS measurements made at pressures of the order of

2 bars. Early studies of severe accidents frequently included scenarios, e.g. station blackout, where the

system pressure remained high throughout, of the order of 100 - 150 bars, but it is now generally assumed

that during such events partial depressurisation will be applied as an accident management measure,

reducing the operating pressure during the in-vessel release phase. Such pressure reduction is limited by

the accumulator set-points and in PWRs the reduced pressure is of the order of 40 bars. At intermediate

NEA/CSNI/R(2009)5

pressures of this order the coolant stream moves an order of magnitude more slowly through the vessel and

primary circuit than at low (near-atmospheric) pressure, implying lower cooling rates, while concentrations

of released fission products are higher. In this case, for which few experimental data are available,

heterogeneous nucleation may play a greater role in the vessel, while the droplets/particles produced by

homogeneous nucleation will tend to be larger. Information on these aspects of aerosol physics may

become available as a result of Japanese elevated pressure fission product release and transport

experiments using simulants.

3.1.2.6 Formation of particles > 1

Independent of the pressure, the extremely numerous primary particles produced in the core and upper

plenum region agglomerate to a considerable extent within the vessel and continue to do so in the primary

circuit. All agglomeration mechanisms are expected to be active in different degrees depending on the

particle size and local conditions within the primary system. Note that agglomeration may take place

between chemically dissimilar particles, as occurs in PHÉBUS. The shape of the agglomerates, as shown

by SEM photographs, is by no means spherical, a fact which is known to affect both agglomeration and

removal processes. Non-spherical shape is usually modelled by the use of shape factors, but it appears

extremely difficult to predict a priori the correct shape factors to be used in circuit calculations, particularly

as particles of the same mass and composition may have different shapes. The shape factors in turn

influence thermophoretic and gravitational deposition rates, as well as agglomeration rates. Sensitivity

studies could usefully be performed to evaluate the impact of shape factor variation upon the transmission

of fission products through the primary circuit.

3.1.2.7 Vapor condensation on particles

Simultaneously with agglomeration of primary particles fission product and other vapors condense both on

the primary particles and on the larger particles resulting from their agglomeration. In PHÉBUS, despite

the over-large ratio of pipe surface to particle surface area about half the Te, for instance, is transported in

particle form through the hot portion of circuit, and in reactor scenarios one can expect fission product

transport as condensate on particles to dominate over vapor condensation on structures [18].

More detailed information for the circuit is given in chapter 3.3.

3.1.2.8 Considerations of chemistry

The similarity of the primary circuit to a tubular reactor has been noted many times [1]. A wide variety of

elements and isotopes arise from the degrading core where they react chemically with the carrier gas and

with each other, and the resulting maze of active reaction pathways represents a severe challenge to

understanding and to the modeller. The reactions continue along the circuit as temperatures drop and

elements are selectively removed by deposition processes. Availability of an element for further reaction

will depend not only on the chemical compound of which it currently forms a part, but also on the physical

phase in which it has become segregated - vapor, surface deposit, condensed layer on aerosol particle or

particle matrix. Particles of the same overall composition may have different chemical characteristics

depending on whether they are well-mixed or layered. Most important for the prediction of transport

behaviour appears to be the element’s volatility, a function of both temperature and chemical speciation.

Modern codes such as VICTORIA and SOPHAEROS [19] attempt a full equilibrium calculation to

determine the speciation of all elements point by point, without prior assumptions. This is a valid general

approach, but the calculated speciation is sensitive to uncertainties in the thermochemical properties of the

compounds involved, and it has been suggested that chemical kinetics may also play a role under low

temperature and pressure conditions, e.g. in the speciation of iodine [20]. The radiologically important

element iodine has a volatility which varies along the circuit in a manner which is hard to predict without

NEA/CSNI/R(2009)5

positive identification of the species responsible, and indeed the chemical nature of the volatile fraction

detected in PHÉBUS in the cold part of the circuit remains a matter for considerable speculation and

research. Some fixed data are now available to guide and correct the modeller, however, such as the

metallic nature of the silver, the dominance of oxides for the elements In, Sn, U, Cd and Mo, and the

probable dominance of the hydroxide for Cs. It is also known from PHÉBUS FPT2 that boric acid in the

PWR coolant has only a minor effect on circuit chemistry, aerosol growth and fission product transport.

3.1.2.9 Revaporisation in the primary circuit

Revaporisation is the subject of a separate contribution to this Report. Here it is appropriate to note that

there is evidence for the revaporisation of Cs after shutdown in PHÉBUS test FPT1 and late in the transient

of FPT2. The Cs is released either in the bundle region or in the hot portion of the circuit and on cooling

produces a fine aerosol, presumably through homogeneous nucleation, there being liittle suspended

material present at the time to act as seeds for heterogeneous nucleation. The fine aerosol behaves

differently from the rest of the aerosol population in the containment, causing Cs to display different

deposition behaviour from the other elements (see next section).

3.1.3 Containment aerosols

The removal time for containment aerosols is generally longer than the transient of aerosol release from the

degrading core, so that one may expect some averaging of aerosol composition and properties to take place

in the containment building. Such averaging is observed to be active in PHÉBUS, despite its rather short

removal time, in that the composition of the aerosol is found to be almost constant, consisting largely of

Ag metal and the oxides of Sn and In, independent of particle size. All fission products appear to be

incorporated in or deposited on this single aerosol “species”, and are removed from the containment

atmosphere at the same rate. The sole exception is found to be Cs. Possibly the Cs aerosol resulting from

revaporisation (see above) does not deposit on the much larger particles of the main aerosol but maintains

an independent existence. A further caveat regarding the “single aerosol” picture of the containment

suspended mass must be mentioned: in post-test analyses the solubility of aerosol particles sampled from

the containment was found to vary with the stage in the release transient reached at the time of sampling,

implying a different composition of the surface layer. This may in turn imply differing hygroscopic

behaviour for aerosols of differing genesis.

Particle size and morphology

The process of agglomeration continues within the containment, resulting in an AMMD in PHÉBUS which

at 3.5-4 µ is somewhat larger than the values seen in the cold part of the circuit (the geometric standard

deviation remains 2.0). The distribution is found to be log-normal, again suggesting a possible

simplification in the modelling as for the circuit. The impactor measurements refer to the wet aerosol, with

an uncertain amount of adsorbed water, some of which may have been lost in the sampling lines. However

the measured removal times are consistent with the AMMD stated. Furthermore, the size distribution is

found to remain almost constant throughout a containment transient lasting tens of hours. Compensation

appears to be taking place between the competing processes of agglomeration and preferential removal of

the larger particles by settling. This finding suggests that a drastic simplification of containment aerosol

codes might be made, but it should be recalled that the PHÉBUS containment atmosphere was rather dry,

with relative humidities normally less than 80 %. A different behaviour might be seen with higher

humidities, with differentiation of the aerosol population according to size and hygroscopicity, as occurred

in the KAEVER tests of ISP-44. The topic of aerosol behaviour in a humid containment environment is

considered further in a separate chapter of this report.

NEA/CSNI/R(2009)5

References

[1] A.L. Wright et al., Primary System Fission Production Release and Transport NEA/CSNI/R(94)2 (1994)

[2] J.Frenkel Kinetic theory of liquids, Dover, New York, Chapter 7, 1955

[3] Y.B. Zeldovich, On the theory of new phase formation: Cavitation J. Exp. Theor. Phys.12, 525, 1942

[4] Laaksonen, V. Talanquer, D.W. Oxtoby Nucleation: measurements, theory and atmospheric

applications Ann. Rev. Phys. Chem. 46, 489, 1995

[5] D. Kaschiev On the relation between nucleation work, nucleus size and nucleation rate J. Chem.

Phys. 76, 5098, 1982

[6] I. Ford Nucleation theorems, the statistical mechanics of molecular clusters, and a revision of

classical nucleation theory Phys. Rev. E., 56, 5615, 1997

[7] N.A. Fuchs, A.G. Sutugin High dispersed aerosolsin Topics in Current Aerosol Research, Vol II;

Hindy, G.M. and Brock, J.R. Eds. Pergamon, Oxford, 1971

[8] S.K. Friedlander, Smoke, Dust and Haze Fundamentals of Aerosol Behaviour. Wiley, New York, 1977

[9] B.J. Mason The Physics of Clouds Clarendon Press, Oxford, 1971

[10] I. Colbeck Introduction to aerosol sciencein Physical and Chemical Properties of Aerosols. Colbeck,

I. Blackie Academic and Professional (Chapman and Hall), London, 1998

[11] M.W. Williams, S.K. Loyalka Aerosol theory and practicePergamon Press, Oxford, 1991

[12] W. Koch, S.K. Friedlander The effect of particle surface coalescence on the surface area of a

coagulating aerosol J. Coll. Interface Sci., 140, 419-427, 1990

[13] B. Clement, N. Hanniet-Girault, G. Repetto, D. Jacquemain, A.V. Jones, M.P. Kissane, P. von der

Hardt LWR severe accident simulation: synthesis of the results and interpretation of the first

PHÉBUS FP experiment FPT0Nucl. Engrg. and Design 226 5-82, 2003

[14] C.T.R. Wilson, Condensation of water vapor in the presence of dust-free air and other gases Phil.

Trans. Roy. Soc.London 189: 265-307, 1897

[15] E. HontanonTheoretical and experimental research on homogeneous nucleation phenomena

CTN-67/91, Catedra de Tecnologia Nuclear E.T.S. de Ingenieros Nucleares 2000

[16] D.W. Oxtoby, Nonclassical nucleation theory for the gas-liquid transition J. Chem. Phys. 89(12):

7521-30, 1988

[17] D.W. Oxtoby, Homogeneous nucleation: theory and experiment J. Phys. Condensed Matter 4:

7627-7650, 1992

[18] R.D. Spence, A.L. Wright The importance of fission product/aerosol interactions in reactor accidents

Nucl. Technol. 77, 150, 1987

[19] H.-J. Allelein, K. Neu, J.P. Van Dorsselaere, K. Mueller, P. Krostka, M. Barnak, P. Matejovic, A.

Bujan, J. Slaby European validation of the integral code ASTEC (EVITA) Nucl. Eng, and Desn. 221

95-118, 2003

NEA/CSNI/R(2009)5

[20] L. Cantrel, E. Krausmann Reaction kinetics of a fission product mixture in a steam-hydrogen carrier

gas in the PHÉBUS primary circuitNuclear Technology 144(1): 1-15, 2003

3.2 Impact of Thermal Hydraulics

3.2.1 Impact of thermal hydraulics on aerosol behaviour in the primary circuit

Introduction

In nuclear safety evaluations thermal-hydraulic behaviour must be determined as a precondition for the

calculation of the transport and deposition of released materials, whether in vapor or in aerosol form, from

the degrading core through the primary circuit to the postulated breach and into the containment or into the

auxiliary building in the case of bypass sequences. In certain special situations such as pool scrubbing or

steam generator tube rupture thermal-hydraulics and material transport are intimately coupled and are

usually calculated together. More normal is the situation in which the coupling is considered to be looser

and the thermal-hydraulics are calculated as a separate issue, with feedbacks to and from the aerosol/vapor

transport calculations. Thermal-hydraulics is a general term, including the evaluation of the velocity field

of the carrier (usually a single-phase gas), the determination of the temperature field, including structural

temperatures which are required as boundary conditions, calculation of the pressure field, and evaluation of

the concentrations of the various components of the carrier gas, normally steam and hydrogen, and of the

transported materials. Each of these fields affects the aerosol behaviour, as described below, and in some

circumstances the aerosol behaviour has a substantial influence on the thermal-hydraulics.

The flow field

The flow field provides the carrier velocity at every point in the primary circuit. It therefore determines the

residence time of the aerosol-laden gas stream within each component of the circuit, and hence the

magnitudes of the fractional removal of the aerosol by the various deposition processes, e.g. settling,

impaction, thermophoresis. The flow is also important in determining the particle size spectrum. For

instance, a fast flow through the circuit will provide little time for agglomeration to occur, resulting in a

smaller average particle size. It can normally be expected that the calculated thermal-hydraulics will

conserve mass, so that the mass transfer rates from one volume to another can be considered reliable.

Details of the flow are not usually determined explicitly, although they may be important to the aerosol

physics. As an example, the turbulent energy dissipation, which is difficult to evaluate in practical

situations, is an essential parameter in the theory of turbulent agglomeration and in eddy impaction. The

streamlines of the flow also play a role in determining the concentration of fission products available at a

particular point of the circuit at a particular time. Thermal-hydraulic models used in association with

fission product transport codes normally assume a sequence of well-mixed connected volumes, in which

streaming, recirculation and other effects of the detailed flow field are ignored.

As mentioned elsewhere in the report, another important thermal-hydraulic parameter is the thickness of

the boundary layer in transition or turbulent flow, and this is supplied in the one-dimensional codes only

through the use of correlations based on fully-developed flow. When the flow is transient or is developing

e.g. downstream of a change of section, the calculated boundary layer can be incorrect with consequences

for the calculated deposition rate by thermophoresis, for instance.

The role of the flow in remobilising deposits through resuspension is treated elsewhere in this report.

Temperature effects

Thermal-hydraulic calculations determine not only the flow velocity, but also the gas temperature and

(when coupled with heat transfer calculations for structures) the local temperature difference between

NEA/CSNI/R(2009)5

surface and bulk gas. Both quantities are involved in the determination of the material properties of the gas

(thermal conductivity, diffusivity etc.), themselves employed in the calculation of aerosol deposition rates.

The gas and wall temperatures also play an essential role in determining the rates of any vapor

condensation on existing aerosols, of homogeneous nucleation, and of condensation of vapors on structural

surfaces. Sensitivity of aerosol deposition to calculated temperatures varies considerably depending on the

scenario, and needs to be determined case by case. The local temperature is important in the chemistry of

the transported materials (and that of the carrier), determining reaction rates and chemical equilibria.

Pressure effects

Pressure differences within the primary circuit are not usually large except in special situations such as

bypass sequences where critical flow can develop. Of more interest is the overall system pressure. Pressure

effects upon core degradation and fission product release are outside the scope of this article. It should be

noted however that for a given release the fission product concentrations are higher for higher system

pressures. The impact of pressure upon formation and growth was considered in Chapter 3.1. The pressure

is also an important factor in determining the Knudsen number, which features in a wide range of

processes in aerosol dynamics. Perhaps of most importance is the effect of pressure on the condensation

temperature of the carrier gas, usually steam. At high system pressures combined with moderate

temperatures condensed water can be formed within the primary circuit, affecting the thermal-hydraulics,

which becomes two-phase, and introducing the powerful aerosol deposition mechanism of

diffusiophoresis. Transport of fission products in liquid water, revaporisation by evaporation of liquid

deposits and other phenomena will then occur, not seen in experiments or scenarios with only single-phase

flow in the circuit.

Concentration and composition

The composition of the carrier stream and of the released materials carried within it depends greatly on the

core geometry, the degradation model, and assumptions concerning the supply of steam or water and the

generation of hydrogen and other gases. Hydrogen being much lighter than steam and its molecule much

smaller, carrier gas properties are highly dependent on the steam-hydrogen ratio, and hence so are the

calculated aerosol removal rates, local temperatures etc. Non-homogeneous concentrations are likely in

many situations, but are generally not considered in aerosol transport codes.

Boundary conditions

Circuit thermal-hydraulics plays a decisive role in fission product transport, deposition and remobilisation,

including the formation and dynamics of aerosols. Conversly, fission products and other released materials

can affect the thermal-hydraulics. The mass loading of the carrier gas stream by fission products is

generally insufficient to have a significant effect on its gas dynamics or on the gas thermal properties, and

the thermal energy released by fission product decay during transport is usually comparatively small

during the main flow transient. However, fission products deposited on surfaces, whether as aerosols or

through sorption, can have significant effects. Mechanical effects include changes in the geometry

available for the flow, up to and including plugging of flow paths, and increases in surface roughness.

Thermal effects become important when sufficient quantities of self-heating deposit have built up to

provide a local heat source. Such a heat source can provoke subisidiary flows, including recirculating

flows, can be a cause of revaporisation, and in extreme scenarios can lead to loss of geometry through

structural failure.

NEA/CSNI/R(2009)5

Modelling aspects

As has been mentioned, the circuit thermal-hydraulics calculated to drive the aerosol transport calculations

in severe accident codes is usually relatively unsophisticated. A pipe flow or parallel-channel geometry is

combined with one-dimensional fluid dynamics to produce the required velocities, concentrations and

temperatures. For reasons connected with the resolution of the aerosol size spectrum the number of

volumes cannot be very large, resulting in rather a coarsely discretised resolution of the flow and

temperature fields. Numerical experiments show sensitivity of the calculated aerosol processes to the

thermal-hydraulic model. The location of wall condensation is particularly sensitive. The application of

material transport models in combination with CFD is currently at an experimental stage in nuclear safety

evaluations.

Conclusion

Good thermal-hydraulic data are essential for the correct calculation of aerosol and vapor transport

processes in the primary circuit. Investigations into calculating more accurate and detailed

thermal-hydraulics than current severe accident codes can provide continue. Meanwhile, it is prudent in

safety evaluations to determine sensitivities to the thermal-hydraulic side of the calculational route

adopted.

3.2.2 Impact of thermal-hydraulics on aerosol behaviour in the containment

Introduction

In a severe accident the particulate fission product source term is not only governed by the aerosol

processes but depends also on the thermal hydraulic conditions prevailing in the containment. The main

interrelations between thermal hydraulics and aerosol physics are

• Transport processes

• Condensation processes

• Fission product decay heat processes

Fig. 3.2-1 shows schematically the most important interrelations in the containment.

Atmospheric flows

Fission product

decay heat

Steam saturation ratio

Natural

convection

Local aerosol

concentration

Agglomeration

Aerosol deposition

Droplet growth

Diffusiophoresis

Fig. 3.2-1 Interactions between thermal hydraulics and aerosol behaviour in a LWR-containment

NEA/CSNI/R(2009)5

Transport processes

The core melt aerosol entering the containment atmosphere through a leak in the coolant system or from

the cavity is distributed in the containment by atmospheric flows. The atmospheric transport is mainly

accomplished by pressure balancing flows in the early accident phase and by natural convection in the later

phase. In a LWR containment stratifications with the lighter steam at the top and the heavier colder air

below are likely. Parts of the containment may be excluded from atmospheric mixing due to such a

stratification or to a hindered atmospheric access, e. g. into a dead-ended room. An inhomogeneous aerosol

distribution is the consequence.

In the VANAM and the older CSE tests aerosol concentration differences between different compartments

in the vessel of approximately two orders of magnitude were measured. In a LWR containment the

differences are expected to be higher.

The particle growth by agglomeration as well as the aerosol deposition depend on the local concentration.

The aerosol concentration in vicinity of a source is usually higher than in the rest of the containment.

Surfaces of containment areas which are not reached by the aerosol, e. g. because of an atmospheric

stratification, are ineffective for the global aerosol depletion. In case of a small containment leak the local

aerosol concentration around the leak determines the aerosol source term.

Condensation processes

The aerosol behaviour is mainly influenced by two condensation processes: volume and wall condensation.

Steam condenses on the aerosol (volume condensation) when the relative humidity in the containment

atmosphere has the tendency to exceed saturation conditions. For aerosols of hygroscopic materials volume

condensation occurs even at conditions somewhat below saturation conditions. The particle growth by

volume condensation accelerates aerosol settling significantly. Fig. 3.2-2shows the depletion of an

insoluble SnO

aerosol under dry and condensing conditions [1]. In general volume condensation occurs

locally differently in the containment. Predominantly volume condensation takes place around steam

sources and close to cold structures. If the atmospheric mixing is poor large aerosol concentration

differences may develop between regions with and without volume condensation. An example is the

aerosol distribution in the VANAM tests [2].

NEA/CSNI/R(2009)5

Fig. 3.2-2 Aerosol depletion measured in three DEMONA tests under dry and wet conditions [1]

Fig. 3.2-3 shows the aerosol depletion in test M2. In the outer annular room the aerosol concentration is

significantly higher than in the rest of the containment.

Fig. 3.2-3

Measured and calculated aerosol depletion in the test VANAM M2 [3]

NEA/CSNI/R(2009)5

With wall condensation aerosol is deposited by diffusiophoresis onto the wall. Among others wall

condensation depends on the steam saturation ratio in the atmosphere and the temperature difference

between bulk and wall. Dry and wet walls at the same time and even in the same compartment are possible.

In general diffusiophoresis contributes less than sedimentation to the global aerosol depletion. But the AP

1000 and AP 600 reactors are exceptions to this. The used external shell cooling leads to an accentuation of

diffusiophoresis to the deposition of aerosol within the containment.

FP decay heat processes

The fission product (FP) decay heat carried by aerosols has a strong feedback on thermal hydraulics.

Airborne and deposited radioactive fission products hosted by aerosols present a considerable source of

heat. This heat is released partly into the atmosphere and partly into the structures and the sump. The

heat-producing gases are distributed in the atmosphere whereas the heat-producing aerosols will be found

predominantly on the floors and in the sump.

The decay heat released has a pronounced impact on the relative humidity, the wall and volume

condensation rates, and on the natural convection flows within the containment. The decay heat released

into the atmosphere reduces the relative humidity and the decay heat released to the sump promotes

evaporation which increases the relative humidity. At certain conditions the sump will come even to boil.

All described interaction processes have to be considered in a comprehensive and accurate aerosol source

term analysis.

Modelling aspects

Interrelation phenomena can only be examined in code systems where the individual models are

numerically coupled. In such couplings parameters of one model are provided for one or several other

models. The transfer parameters have to be updated in time. The tightness of the numerical coupling has to

be chosen according to the strength and velocity of the physical interrelation.

The modeling of aerosol transport processes requires a moderate coupling of the aerosol and thermal

hydraulic models and a multi-compartment representation of the containment. The characteristic period of

time of the aerosol processes and the atmospheric transport processes are comparable. The same applies to

the modeling of the decay heat feedback.

More complicated is the coupling for the condensation processes. Volume condensation and evaporation

are rather fast processes which require more sophisticated coupling methods, like the moving grid

technique used in several codes.

In most of the well known containment code systems (ASTEC, COCOSYS, CONTAIN, MELCOR) all

mentioned interrelation processes are modeled. Fig. 3.2-4 illustrates the interrelations treated in the

CONTAIN code [4]. Sometimes simplifying assumptions are made, e.g. the Kelvin-effect which appears

with volume condensation cannot be considered in multi-compartment calculations.

NEA/CSNI/R(2009)5

Fig. 3.2-4 Illustration of feedback mechanisms treated in the CONTAIN code [4]

As a consequence of the modeled interrelations the uncertainty of some thermal hydraulic parameters has a

very pronounced impact on the uncertainty of the calculated aerosol source term. In the conclusions of the

ISP-37 [5] the following is stated:

“However, the aerosol source term to the environment is overestimated in general. The largest

uncertainty concerning the aerosol results is caused by a limited number of thermal hydraulic

variables like relative humidity, volume condensation rate and atmospheric flow rate.”

The findings of this international study emphasise the importance of aerosol interrelation processes for

reliable source term analyses. A few years later the ISP-44 [6] concluded that there is still a need to

improve the quality of some thermal hydraulic parameters which have a strong impact on the aerosol

results.

NEA/CSNI/R(2009)5

References

[1] T. Kanzleiter DEMONA-Versuche Abschlußbericht, Final Report BIeV-R65.523-01 (1987)

[2] T. Kanzleiter Versuche zum Verhalten von Kernschmelzunfall-Aerosolen in einer

Mehrraum-Containment-Geometrie (“VANAM-Versuche”) Abschlussbericht BIeV-R67.098-01

(1995)

[3] H.-J. Allelein, K.O. Fischer, T. Kanzleiter, S. Schwarz, G. Weber The VANAM Experiments M1

and M2 – Test Results and Multi-Compartmental Analysis European Aerosol Conference AEC ’91,

Karlsruhe, 16 – 20 Sept. 1991

[4] K. K. Murata et al.Code Manual for CONTAIN 2.0: A Computer Code for Nuclear Reactor

Containment Analysis NUREG/CR-6533 (1997)

[5] M. Firnhaber, T. F. Kanzleiter, S. Schwarz, G. Weber International Standard Problem ISP-37

VANAM M3 - A Multi Compartment Aerosol Depletion Test with Hygroscopic Aerosol Material

GRS-137, OECD/GD (97) 16 (1996)

[6] M. Firnhaber, K. Fischer, S. Schwarz, G. Weber International Standard Problem ISP-44

KAEVER-Experiments on the Behaviour of Core-melt Aerosols in a LWR Containment

OECD/NEA/CSNI/R (2003) 5 (August 2002)

3.3 Fission-Product Transport and Deposition

3.3.1 Introduction

During a severe accident in a nuclear power plant (NPP), fission products (FPs), actinides and structural

materials are released as gases or vapors from the degrading core into the reactor coolant system (RCS).

These are then swept, in general, by a steam-hydrogen gas mixture towards the breach in the RCS. A

number of important physico-chemical processes occur between the point of release from the core and

release via the breach of still-suspended materials into the containment (or into the auxiliary building in the

case of a containment-bypass sequence). Not only do the phenomena occurring in the RCS reduce the

quantity of material released into the containment, they also condition its physico-chemical form; the

formation of highly-volatile chemical species leading to vapor-phase contributions to the source into the

containment is of particular concern. The importance of the different phenomena in conditioning the source

from the RCS varies in terms of the accident sequence, e.g., gravitational settling of aerosols is very minor

for a hot-leg break sequence since horizontal surfaces seen by the transported aerosols are limited, much

radioactive material remains in the vapor phase due to the relatively high temperatures between core and

breach, and transport times through the RCS are short due to the short distance travelled before release into

the containment and the high flow rates. Hence, below, a list of phenomena is presented representing the

principal phenomena arising beyond that of convection of gases and aerosols by the steam-hydrogen flow

and are briefly discussed in Chapter 3.

3.3.2 Synopsis of RCS phenomena and their modelling

The starting point here is a mixture of gases and vapors at a given temperature where this input is assumed

along with thermal-hydraulic boundary conditions for a particular RCS geometry.

NEA/CSNI/R(2009)5

Homogeneous nucleation of vapors

Even in an unsaturated vapor, molecular clusters exist but they are unstable. Once supersaturation is

established for a vapor it is ready to become aerosol or condense on a structural surface. It can be broadly

considered that homogeneous nucleation only occurs in conditions of extreme supersaturation in the

absence pre-existing aerosol. This is because the high surface energy of the small particles formed by

homogeneous nucleation constitutes a greater barrier than that due to condensation on existing surfaces

(aerosol or structural).

Nevertheless, it must be remembered that in the present context, the high radiation field reduces this

resistance by providing abundant ionic nucleation sites.

In general, therefore, if aerosols are not being produced by mechanical phenomena (e.g., bursting and

spraying of control-rod alloy or a steam explosion in the reactor vessel), moving downstream from the

point of release, the first aerosols are produced by homogeneous nucleation of the most refractory species.

This may be, for example, silver vapor or, in a later stage of the core degradation process, uranium dioxide

(or, more likely, uranium trioxide vapor non-congruently reverse-subliming to uranium dioxide). The

molecular clusters, or embryo particles, that form then quickly agglomerate while being simultaneously the

target of further condensation of the initial nucleating vapor as well as other supersaturating low-volatility

species. The picture may well be more complicated since more than one species may supersaturate and,

rather than unary nucleation, binary or even ternary nucleation will occur.

Beyond the vapor pressure itself, models require fundamental properties including chemical activities,

surface tensions and densities. In nuclear-safety computer codes the phenomenon is simplified.

Condensation of vapors on aerosols and structural surfaces

It should firstly be noted that the process of heterogeneous condensation onto existing aerosols and onto

surfaces can be considered a critical phenomenon: in the accident context it is fundamental to the quantities

of volatile species reaching the containment in aerosol form rather than being retained on RCS surfaces.

The volatile species concerned, i.e., those more volatile than the more refractory species creating the first

aerosols, comprise the majority of the most important FPs radiologically, e.g., cesium iodide.

Heterogeneous condensation occurs when a vapor supersaturates and a surface (aerosol or structure) is

available. The phenomenon is mainly governed by mass-transfer limitations where, in the RCS, such

limitations usually constitute considerably lower resistance to condensation than that of homogeneous

nucleation. It must also be remembered that while a vapor may be subsaturated in the bulk flow, it can be

supersaturated with respect to a structural surface exhibiting a cooler temperature where (as computer

codes assume) aerosols are at the same temperature as their surroundings, i.e., their minuscule heat

capacity and large surface-to-volume ratio mean that thermal equilibration with the surrounding gas is

always fast. Hence, condensation on a wall can arise without competition from aerosol formation or

condensation on aerosols. Once a vapor supersaturates in the bulk flow and aerosols are already present,

condensation onto these aerosols is efficient since a small amount of aerosol (in terms of mass) represents a

large surface area

Heterogeneous nucleation onto aerosols including the significant negative feedback from warming of the

particle due to latent-heat release by the condensing species has been correctly understood for many years.

Nevertheless, three complications exist. One arises from an aerosol particles’ considerable surface

curvature which, combined with the inherent surface tension of the condensed species, induces an increase

To consider a representative situation, given an aerosol loading of the flow of 5g.m

-3

, assuming aerosols to be spherical with a

uniform diameter of 1µm and a material density of 3000 kg.m

-3

, the aerosol surface area is 10m

per cubic metre of the flow.

NEA/CSNI/R(2009)5

in effective vapor pressure at the liquid surface formed on the aerosol. This so-called Kelvin effect can

both reduce total condensation and considerably bias it towards larger particles in an aerosol population

(such particles inducing lower surface curvature of the liquid phase). A second complication arises when

the particle size is smaller than or of a similar order to the mean free path between molecular collisions

(i.e., large Knudsen number). If this is the case consideration of the particle in a continuum regime breaks

down and the condensation rate must be corrected [1]. A third complication arises from the probable

heterogeneous nature of the aerosol surface affecting sites for condensation: this introduces considerable

greater complexity to the condensation phenomenon [2].

Chemisorption of vapors

Chemisorption of certain vapors, i.e., their chemical reaction with structural materials, is a well-known

phenomenon but for which little data is available. Some cesium and tellurium species react with metal

alloys and empirical chemisorption rates as a function of temperature have been derived for CsOH, CsI, Te

(and by extension, to SnTe) for stainless steel and Inconel surfaces.

There are also data for a handful of other species with respect to specific materials such as Zircaloy.

However, it can be said that this is a poorly investigated area where a vast amount of experimental work

would be required to provide rates for the species likely to be affected. At present the influence of vapors

chemisorption cannot be quantified.

3.3.3 Agglomeration of aerosols

Agglomeration occurs due to particle collisions arising from their differing velocities. Particle motion is

induced by Brownian diffusion, sedimentation and turbulence (shear and inertial effects) where other

influences such as electrical forces and acoustic influences are less relevant in the present context. Particles

combine due to Van der Waals forces, changes in surface free-energies and/or chemical reactions where

codes generally assume that the sticking efficiency is unity (i. e., colliding particles always stick together).

In the RCS, it is the Brownian mechanism which is of most importance since, once embryo particles have

formed, this phenomenon will rapidly lead to larger (and fewer) particles. Once the aerosols have grown to

a greater size the other agglomeration mechanisms (properly termed kinematic agglomeration) will come

into play. However, the generally short residence times of aerosols and turbulent conditions in the RCS

mean that sedimentary agglomeration is usually insignificant.

A particular point that should not be overlooked is that if agglomeration is a significant mechanism then

the numerical treatment of the aerosol population is critical in reproducing what the models intend. If a

fixed-grid scheme is used then a high number (50 or more) of size bins is usually required to avoid

significant spurious diffusion in the re-sizing scheme: a newly-formed particle will not find a size in the

discretized scheme that suits it perfectly and fractioning is needed between two adjacent size classes. The

coarser the discretization the worse the spurious diffusion.

Probably the most important uncertainty in this area is related to particle shape. As explained in Chapter 2,

it is common to associate two shape factors with an aerosol particle, one affecting its mobility (or dynamic)

properties, the other its collision properties. Spheres are the most compact particle form possible and so

any deviation from this has some impact on the resistance to movement and the probability of colliding

with another particle. Particles in the presence of high steam humidity tend to collapse to compact forms

under the influence of water surface tension. However, RCS conditions are generally highly superheated

and compaction due to steam is only likely to occur near the breach for particular sequences producing

saturated or near-saturated conditions in this region (cold-leg break or in the steam-generator tube in the

case of a steam-generator tube rupture). Nevertheless, perhaps other condensing species are abundant

NEA/CSNI/R(2009)5

enough to cause a compacting effect since there is evidence from representative experiments that particles

are, in fact, fairly compact despite superheated steam conditions. This means that high values for the shape

factors, such as for chain-like agglomerates, can probably be excluded. Nevertheless, the shape factors and

their evaluation remain a significant uncertainty. Evaluation techniques are often empirical where, in

relation to an arbitrary particle, there are no reliable analytical techniques for estimating the appropriate

values for use in an agglomeration model. Nonetheless, it is desirable that review of the most

representative experiments is undertaken with the objective of proposing more realistic values for the

shape factors, these values becoming the default values (rather than unity as is now assumed) in nuclear-

safety computer codes.

Brownian agglomeration is most significant for small particles where the free-molecular regime (Knudsen

number >> 1) and transition regime (Knudsen number of the order of 1) must be considered. The mobility

of small particles is very large but this effect is tempered by the reduced target area that they present.

Brownian agglomeration is most effective between very small and larger particles. In general, models

derive from Brownian diffusion theory with correction factors for the free molecular regime and

non-spherical particles.

Gravitational agglomeration is most clearly understood in terms of particle terminal velocities showing the

phenomenon to be proportional to the difference in the velocities of the two particles and the sum of their

projected areas (the target). Disparity arises in a factor termed the collision efficiency (see [3, 4]), where

this constitutes a correction from the ideal situation in which the larger particle sweeps and collects with

perfect efficiency all the smaller particles in its projected cylinder during free fall. The correction reduces

the efficiency due to hydrodynamic effects where smaller particles tend to flow around the larger particles

allowing some to avoid collection. In the RCS, the limited impact of gravitational agglomeration mean that

exploration of the different efficiencies is not required here [5].

Turbulent agglomeration arises due to the relative particle velocities induced by the shearing flow field and

particle drift relative to the flow arising from inertial differences. This latter contribution is zero for

particles of the same size and turbulent agglomeration reaches a minimum in this case. The present

modelling is affected by rather high uncertainties.

3.3.4 Deposition of aerosols

Aerosol deposition will occur in the RCS due to Brownian diffusion, thermophoresis, diffusiophoresis,

electrophoresis, sedimentation (gravitational settling), inertial impaction (projection onto surfaces by

flow-geometry changes and turbulent eddies) and pool scrubbing. Among these phenomena, not all are

certain to be significant since they can be considered scenario dependent. Brownian deposition will be

significant if particles, once formed, remain small. Pool scrubbing phenomena, the subject of a separate

Chapter in this report, may occur in the pressuriser (e.g., high-pressure sequence such as station black-out)

or the steam-generator secondary side (steam-generator tube rupture). Probably only thermophoresis can be

guaranteed to produce significant deposition whatever the accident sequence.

Deposition due to diffusion

Diffusional deposition is most relevant to the RCS zone where the first vapors nucleate and particles

remain small, i.e., within the reactor vessel. For laminar flows, use of the classic Brownian-diffusion

approach (Fick’s law) is only appropriate in the limit of stagnation. Hence, deposition from a laminar flow

due to diffusion is often based on empirical models where that of Gormley and Kennedy [6] is used. A key

parameter in this model is the length over which diffusion occurs: in ASTEC/Sophaeros, for example, a

fixed value of 10

-4

m is used. For turbulent flows, the phenomenology changes where the turbulence brings

particles close enough to the wall for diffusion to become the dominant transport mechanism and lead to

NEA/CSNI/R(2009)5

particle deposition [7]. Diffusional deposition, affecting above all highly-submicron particles, is usually

such a small contribution to overall deposition that the modelling approximations do not matter.

Deposition due to thermophoresis

Thermophoresis is a phenomenon which, unusually, does not depend strongly on particle size though is

larger for submicron particles. As a result of imbalanced collisions between gas molecules and aerosol

particles, particles in a temperature gradient experience a force directed towards cooler temperatures

(countered by the hydrodynamic drag force). Highly-conducting (e.g., metallic) particles are less affected

by thermophoresis than are particles of more insulating materials. For larger particles a thermal gradient

may be established in the particle introducing considerable complexity. Correction for slip due to the

greater ease of transport for particles in the free-molecular regime must also be accounted for.

A number of models exist for this phenomenon though it is common to use the Brock formula with

numerical coefficients provided by Talbot et al. [8] that allow good agreement with experimental data over

the free-molecular, transition and continuum regimes. This modelling would seem to be reliable (well

validated) though recent work shows some controversy concerning the validity of the approach in the case

of very high temperature gradients, [9, 10]. It is desirable to follow this work in case of any changes

required to nuclear-safety computer codes since the aspect of high temperature gradients is relevant.

Deposition due to diffusiophoresis

Diffusiophoresis occurs as particles are entrained by a net flow of a vapor species (steam) towards a

surface where the vapor condenses. The flow of the condensing vapor is characterised by its Stefan

velocity, a key parameter for the aerosol diffusiophoresis. In the RCS, only certain accident sequences lead

to steam condensation on surfaces, those involving lower temperatures such as cold-leg break or

steam-generator tube rupture. In fact, diffusiophoresis never occurs alone but must necessarily be

accompanied by thermophoresis so we actually observe thermo-diffusiophoresis. Once the thermal-

hydraulic problem of determining the Stefan velocity of the condensing steam is resolved, the

diffusiophoretic velocity of aerosols is proportional to this velocity but must be corrected as a function of

the gas composition (involving not just steam but incondensable gases) and the particle flow regime

(Knudsen number): this is complex and the most generally valid model is that of Loyalka [11]. The much

simpler model of Waldman [12] is valid in the free-molecular regime only. This modelling has benefited

from broad validation and cannot be considered at present a priority for further attention.

Deposition due to electrophoresis

Deposition may be enhanced due to electrophoresis, a complex process dependent on aerosol decay

activity, size, concentration and confining geometry. Radioactive aerosols self-charge electrically where,

broadly, due to the dominance of α decay (this event generally stripping out more electrons from the

particle than the two positive charges it carries away) and β

decay, a positive charge develops though very

small particles can develop a small negative charge – see the relevant theoretical [13] and experimental

[14] work. However, the theoretical work indicates that particle charging is very sensitive to two factors: in

the RCS the small size of the particles should considerably limit their capacity to self-charge while their

high number concentration should reduce their self-charging by facilitating diffusive combination with

gas-phase ions. Then again, it must be acknowledged that, as predicted in [13], confinement of the aerosols

such as in the RCS may aid charging by lowering ion concentrations. In summary, the theoretical work

implies limited self-charging in these conditions except, perhaps, for regions of particularly confined flow.

At present, nuclear-safety computer codes do not take account of this effect. If this is justified or not, can

only be shown by a proper assessment of its impact on the fission product source term to the environment.

NEA/CSNI/R(2009)5

Deposition due to sedimentation

Particles settle under the influence of gravity where the settling (or terminal) velocity is broadly

proportional to the square of the radius. In a turbulent flow particles fall across the boundary layer whereas

in a laminar flow the aerosols of the whole flow are concerned. Gravitational settling is not particularly

significant in the RCS due to the usually limited size of particles (small settling velocities) and their short

residence time before release to the containment. Corrections must be introduced to the situation of a

perfectly laminar flow around the particle (Stokes regime) for the cases of slip of small particles and the

extra drag on larger particles where flow is no-longer perfectly laminar and wake effects can be induced.

Nuclear-safety computer codes do not take this latter effect into account though, as already stated, the

settling mechanism most often has a limited impact which may well mean that modelling approximations

have little effect on predictions of the source term to the containment. Further work in this area is not a

priority.

Deposition due to inertial impaction

Particles can be projected onto surfaces by flow-geometry changes where, due to their inertia, they deviate

from the mean flow direction to come into contact with confining walls or obstructions. Such geometries

are common in the RCS but most can be generally considered as bends, contractions, expansions or

combinations of these. The impaction phenomenon clearly comes in to play as particle size increases

where, in general, submicron particles are not concerned being more influenced by other deposition

mechanisms. Models for simple geometry changes do exist in the literature but have received little

validation in reactor-relevant conditions, particularly highly turbulent flows. Furthermore, there are

empirical models for these geometries as well as models for specific complex geometries encountered in

reactors such as the components of the steam-generator secondary side.

In turbulent flow, eddies act to project particles towards surfaces where sufficient momentum is imparted

to the particles that they can cross the boundary layer and come into contact with the wall. The

phenomenon is highly sensitive to particle mass, or more strictly the relaxation time of the particle. A

number of theoretical models exist but with varying degrees of agreement with respect to data. Hence,

more empirical models are often used such as that derived from the careful experiments of Liu and

Agarwal [15]. Since few particles usually become quite large (supramicron) in RCS conditions, this

phenomenon is not a dominant mechanism.

One phenomenon which is overlooked in nuclear-safety computer codes which is associated with inertial

impaction phenomena is that of particle bounce. The sticking efficiency of particles is not perfect

especially for the higher impact velocities. It is possible for impacting particles to bounce, to disintegrate

and/or to dislodge particles already on the surface. Significant data requirements are apparent since the

models of particle sticking often involve parameters such as the Hamaker constant. What is certain is that

for high velocity flows in dry (super-heated steam) conditions with hard (salt, ceramic and metal) particles,

bounce will occur. The present understanding on particle bounce is summarised in [16].

References

[1] S.K. Loyalka, J.W. Park“Aerosol growth by condensation: a generalization of Mason’s formula” J.

Colloid Interface Sci. 125, 712-716 (1988)

[2] L.J. Willett et al.“Adsorption on heterogeneous regular surfaces” J. Colloid Interface Sci. 238,

296-309 (2001)

[3] I.H. Dunbar, J. Fermandjian“Comparison of sodium aerosol codes”CEC report EUR 9172 (1984)

NEA/CSNI/R(2009)5

[4] R.L. Buckley, S.K. Loyalka“Implementation of a new model for gravitational collision cross

sections in nuclear aerosol codes”

Nucl. Tech. 109, 346-356 (1995)

[5] H.R. Pruppacher, J.D. Klett“Microphysics of Clouds and Precipitation”Reidel, New York (1978)

[6] P.G. Gormley, M. Kennedy“Diffusion from a stream flowing through a cylindrical tube”Proc. Roy.

Irish Academy, 52, 163 (1949)

[7] C.N. Davies"Aerosol Science"Academic Press (1966)

[8] L. Talbot, R.K. Cheng, R.W. Schefer, D. R. Willis“ Thermophoresis of particles in a heated

boundary layer”J. Fluid Mech., 101, 737-758 (1980)

[9] R. Muñoz-Bueno, E. Hontañón, M.I. Rucandio “Deposition of fine aerosols in laminar tube flow at

high temperature with large gas-to-wall temperature gradients” J. Aerosol Sci. 36(4), 495-520

(2005)

[10] C. Housiadas, Y. Drossinos“ Thermophetic deposition in tube flow” Aerosol Sci. Tech. 39,

304-318 (2005)

[11] S.K. Loyalka “Velocity slip coefficient and diffusion slip velocity for a multicomponent gas

mixture” Physics of Fluids, 14 n°12, 2599-2604, (1971)

[12] L.Z. Waldmann“On the motion of spherical particles in non-homogeneous gases”Rarefied Gas

Dynamics, Academic Press, New York (1961)

[13] C.F. Clement, R.G. Harrison“Enhanced localised charging of radioactive aerosols“ J. Aerosol Sci.

31, 363-378 (2000)

[14] F. Gendarmes, D. Boulaud, A. Renoux“Electrical charging of radioactive aerosols - comparison of

the Clement-Harrison model with new experiments“ J. Aerosol Sc. 32, 1437-1458 (2001)

[15] B.Y. Liu, S.K. Agarwal“Experimental observation of aerosol in turbulent flow” J. Aerosol Sci., 5,

145-155 (1974)

[16] Special Issue of Aerosol Science and Technology Volume 22, Issue 1 (1995)

3.3.5 Containment

During a severe reactor accident nuclear aerosols composed of fission products, actinides and structural

material are released into the reactor containment at different locations and at different times. The aerosols

enter through a leak in the reactor coolant system (RCS) and after the vessel failure they are released from

the molten core in the cavity. Resuspension processes give further nuclear aerosol sources. At a

-defla-gration deposited aerosol is resuspended and aerosols are re-entrained from flashing or boiling

water pools.

The amount of aerosol released by the different sources as well as the particle size distribution varies

considerably. During core degradation several 100 kg of aerosol are expected to be swept through the RCS

leak. During this release in general the aerosol concentration in the containment (> 1 g/m

) reaches the

highest values in the accident sequence. The aerosol release with re-entrainment is comparatively small but

NEA/CSNI/R(2009)5

it represents a persistent source in the late accident phase. At this time most of the aerosol from other

sources has already been deposited.

Especially in the lower part reactor containments are subdivided by structures into compartments of

different sizes. The thermal hydraulic conditions within the compartments can vary significantly. The

parameters most relevant for the aerosol behaviour are the gas temperature, the relative humidity, the wall

and the volume condensation rates and the local atmosphere flow velocity. Locally different thermal

hydraulic conditions will be established when atmospheric mixing in the containment is weak or one or

several compartments are excluded from mixing. This happens with atmospheric stratifications when a gas

with lower density lies above gas with higher density or with dead-end rooms. Thus it is possible that in a

part of the containment wet conditions with volume condensation occur increasing aerosol depletion while

the other part stays dry.

Transport

The containment atmosphere is mainly mixed by natural convection driven by locally different gas

densities. Lowly situated steam and heat sources enhance natural convection. Only in relatively short

periods forced convection appears in the containment as during the blow-down phase or with

-deflagrations.

From the locations of release the aerosols are distributed within the containment by atmospheric flows. In

general the aerosol particles are carried with the gas velocity. Only large particles like condensate droplets

have a noticeable slip in the gas. This slip causes an aerosol transport in direction of settling. This

additional transport is relevant in stagnant atmospheres.

Deposited aerosols are transported by condensate flows. The aerosols are washed down from the walls and

transported by the condensate flows to the sump. Soluble aerosols are washed down more completely than

insoluble aerosols. The aerosol wash down process determines the fission product and the decay heat

distribution between surfaces and sump. The decay heat released to the gas and to the dry walls heats the

atmosphere and reduces the relative humidity. Decay heat released in the sump increases the steam

production and the relative humidity.

Aerosol transport and deposition in multi-compartment containment geometries were investigated in the

experimental series DEMONA (test A9 only) [1], VICTORIA [2], and VANAM [3]. The VANAM tests

were the most comprehensive ones. They were carried out in a large-scale model containment structured

into nine compartments. The aerosol concentration measured in the compartments differed by up to two

orders of magnitude. The highest concentration was measured in compartments in which the conditions

were dry throughout the entire test. In a real containment leaks in the shell are likely to occur in this area.

Details are described in chapter 5.3.5. The VANAM tests impressively demonstrated the need of

multi-compartment accident analyses. Simple single volume calculations with implicitly well-mixed

conditions do not necessarily deliver conservative source term estimations.

Condensation

Particle growth by volume condensation of steam can increase aerosol depletion considerably. In general

the particle growth is described by the Mason-equation [4]. Small particles with a strong curvature can be

excluded from the condensation process (Kelvin-effect), stay dry and deposit only slowly. In accident

calculations the Kelvin-effect is often not treated because of numerical problems. On hygroscopic aerosols

condensation already occur at slightly superheated atmospheres (relative humidity < 100 %). Hygroscopic

nuclear aerosol materials are CsOH, CsI, Sr(OH)

and others. Particles mixed from hygroscopic and

non-soluble materials and in droplets where the soluble material is dissolved in water have a reduced

NEA/CSNI/R(2009)5

hygroscopic effect. The KAEVER tests [5] demonstrate the impact of the hygroscopic effect on the

depletion of single component aerosols and mixed aerosols. The hygroscopic materials employed were CsI,

CsOH, and the insoluble materials Ag, SnO2 (see Chapter 5.3.6).

Volume condensation occurs in a saturated atmosphere, which has a tendency to supersaturation.

Equilibrium volume condensation rates are calculated by thermal hydraulic models with the assumption of

thermal equilibrium between the stem and the water droplets. Volume condensation appears in

compartments with a steam source and certain conditions like cold walls, the presence of non-condensable

gas (air), etc. [6]. Up to about 7 % of the released steam may condense in the volume. Decay heat released

to the atmosphere reduces volume condensation. The impact of decay heat is investigated analytically

since no tests with realistic decay heat releases are available. Even in the PHEBUS FP tests the decay heat

was significantly lower than estimated for a severe accident. In general volume condensation occurs locally

but the fog droplets are distributed by the atmospheric flows. In other parts of the containment the droplets

may dry and shrink again.

Agglomeration

Agglomeration causes a growth of the aerosol particles, a reduction of the particle number, and a mixing of

chemically different aerosol materials within the growing particles. The following processes are relevant in

the containment: Brownian agglomeration, gravitational agglomeration, and agglomeration by turbulent

inertia and turbulent diffusion. In general Brownian agglomeration will prevail. Gravitational

agglomeration becomes important when the particle size distribution is broad with pronounced fractions of

small and large particles. Agglomeration is only effective at relatively high aerosol concentrations

(> 1 g/m

In multi-component aerosols where the components are not well mixed in each particle the components can

deposit with different velocities. Multi-component models have some model restrictions, e.g. the material

density for all components, and are only partly validated. A multi-component treatment is necessary when

the deposition behaviour of the components is rather different and the aerosol concentration is not very

high (< 1g/m³).

Deposition

The deposition of aerosols on the containment surfaces depends on the geometry, the aerosol parameters

like particle size distribution, density of the aerosol material, shape of the particles, and thermal hydraulic

conditions etc. Aerosol deposition takes place due to the processes sedimentation, diffusion,

diffusiophoresis, thermophoresis, and electrophoresis (s. chapters 2.6 and 3.3.1). Here only containment

specific aspects on the deposition process including condensation on particles and agglomeration are

treated.

In general sedimentation is the most effective depletion mechanism in the containment. Particle growth by

agglomeration and volume condensation can increase settling considerably. If condensation is involved the

largest uncertainty in the calculated deposition rate is given by uncertain thermal hydraulic parameters like

the relative humidity and the volume condensation rate. In the dry case a main uncertainty comes from

uncertain shape factors.

Small particles have considerable deposition by diffusion. This process can be essential e. g. for

fine/disperse aerosol from re-entrainment processes which especially occur in the late accident phase. The

largest modelling uncertainty comes from the boundary layer thickness, which is an input parameter in

most models.

NEA/CSNI/R(2009)5

Diffusiophoresis is the only important phoresis process in a LWR-containment. Deposition due to

diffusiophoresis is nearly independent from the particle size. The local wall condensation rate may differ

significantly in the compartments and has to be determined by the thermal hydraulic model.

Thermophoresis is only significant with strong temperature gradients in the atmospheric boundary layer

along cold walls. This may be the case near the hot corium pool or with fires in the containment.

All important aerosol transport and natural deposition processes relevant in severe accidents are well

investigated and modelled. Most accident codes allow a multi-compartment representation of the

containment. Additionally aerosol behaviour and thermal hydraulic models are tightly coupled in order to

simulate the important interrelation phenomena.

References

[1] J.O. Liljenzin, J. Collén, W. Schöck, F.J. Rahn Report from the MARVIKEN / DEMONA / LACE

Workshop Proceedings of the OECD/NEA Workshop on Aerosol Behaviour and

Thermal-Hydraulics in the Containment, Fontenay-aux-Roses (France), 26-28 Nov. 1990, CSNI

Report No 176

[2] J.M. Mäkynen et al. Experimental and Modelling Studies on Containment Aerosol Behaviour in the

Victoria Facility 3rd OECD-Specialist Meeting on Nuclear Aerosols in Reactor Safety, Cologne,

15-18 June, 1998

[3] T. Kanzleiter Versuche zum Verhalten von Kernschmelzunfall-Aerosolen in einer

Mehrraum-Containment-Geometrie ("VANAM-Versuche") Battelle-Institut e. V., Frankfurt/Main

Abschlußbericht BleV-R67.098-01 (1995)

[4] J. B. Mason The Physics of Clouds Claredon Press, Oxford (1971)

[5] G. Poss, D. Weber Versuche zum Verhalten von Kernschmelzaerosolen im

LWR-Containment-KAEVER Fachbericht BF-R-67863, Battelle Ingenieurtechnik GmbH, Mai 1997

[6] G. Weber Calculation of Local Equilibrium Fog Formation Rates with a Multi-Compartment

Containment Code Proceedings of the OECD/NEA Workshop on Aerosol Behaviour and

Thermal-Hydraulics in the Containment, Fontenay-aux-Roses (France), 26-28 Nov. 1990, CSNI

Report No 176

3.4 Aerosol Behaviour in Complex Structures of Steam Generator Secondary Site

3.4.1 Issue and status

Steam Generator (SG) reliability and performance are serious concerns in the operation of pressurised

water reactors. In particular, SG tubing is subject to a variety of degradation processes that can lead to tube

cracking, wall thinning, and potential leakage or rupture. However, over the last decade considerable

efforts have been spent to understand these degradation processes and to develop improved modes of

operation, preventative and corrective measures. Nonetheless the SG tube leakage incidents occurred in the

past (see Table 3.4-1) proved that such occurrence can not be completely ruled out. Even if they are design

basis events and plants are designed to cope with them it cannot be excluded, that such an event escalates

into a severe accident with a significant radioactive release to the environment.

NEA/CSNI/R(2009)5

Table 3.4-1 A summary records of steam generator tube rupture accidents

Plant Date Leak Rate

(lpm)

Break Type and location Plant status and

cause of the

rupture

Point Beach

Unit 1

February 26,

1975

470 2 adjacent ruptured bulges, each

~20 mm long and wide, hot leg

side, above tube sheet, in the

sludge pile region

Full power,

Wastage

Surry Unit 2 September

15, 1976

1250 Top of U-bend, 114.3 mm long

axial crack

Full power,

PWSCC in U-bend

Doel Unit 2 June 25,

1979

510 Top of U-bend, 100 mm long

axial crack

Heated to normal

conditions,

PWSCC in U-bend

Prairie

Island 1

October 2,

1979

1270 Fish mouth, 76 mm above tube

sheet, 38 mm long, 13 mm wide

Full power,

Loose parts

Ginna Unit 1 January 25,

1982

2900 Fish mouth, 50 mm above tube

sheet, 150 mm long

Full power,

Loose parts and

tube wear

Fort

Calhoun

May 16,

1984

425 Horizontal run at the top, 32

mm long cracks, yielding small

fish mouth break of 6 mm long

Plant start-up,

ODSCC at a

crevice

North Anna

Unit 1

July 15,

1987

2410 Top of the 7

upper tube

support plate on cold leg side,

360

circumferential

crack

Full power,

High cycle fatigue

in a U-bend

McGuire

Unit 1

March 7,

1989

1900 95 mm axial crack with 9.5 mm

wide at the maximum

Full power,

ODSCC in the free

span

Mihama

Unit 2

February 9,

1991

2600 Top of the 7

upper tube

support plate on cold leg side,

360

circumferential

crack

Full power,

High cycle fatigue

Palo Verde

Unit 2

March 14,

1993

910 Freespan region between 8

and

tube support plate, 65 mm

long fish mouth in a 250 mm

long axial crack

98 % full power,

ODSCC

Indian Point

Unit 2

February 15,

2000

564 A low row U-bend Full power,

PWSCC in U-bend

Legend for degradation mechanisms identified below by stressor:Fretting, Wear – flow

induced vibration, aggressive chemicals

High cycle fatigue – high mean stress level and flow induced vibration, initiating defect

(crack, dent, pit, etc.)

ODSCC – tensile stresses, impurity concentrations, and sensitive materials

PWSCC – temperature, residual tensile stresses, sensitive materials (low mill anneal

temperature)Wastage – phosphate chemistry, chloride concentration, resin leakage

NEA/CSNI/R(2009)5

A leakage of radionuclides from the primary circuit into the secondary side negates the effectiveness of the

containment. Under certain conditions high release of radionuclides to the environment is possible during

postulated severe accidents. Risk assessment studies generally consider two types of severe accidents:

• An operational event or design basis fault that causes a SG tube rupture which then results in core

damage;

• A core damage sequence, which might impose prevailing pressure and temperature conditions,

could lead to SG tube rupture.

Failure of one or a few tubes can spontaneously happen as a result of any degradation process. However,

multiple SG tube failures can only be induced by a very large pressure difference between the primary and

the secondary side of the steam generator following occurrence of one of many foreseeable scenarios. A

steam line break, turbine trip with a stuck open secondary side safety valve, loss of feed water, ATWS

without turbine trip, station black out with a stuck open secondary side safety valve, etc., are examples of

such scenarios. Several studies have provided insight into the tube failure mechanisms and conditions

leading to tube failure. Conditions necessary to preserve core cooling were identified. These studies also

estimated the containment bypass frequencies with a core melt if the latter cannot be avoided.

Most probabilistic risk assessments (PRAs) and severe accident codes assume that a significant fraction of

fission products flowing through an unisolated break in a SG escapes to the environment. For example, in

NUREG 1150, the median estimate of the fraction of the core inventory of iodine released to the

environment was 27%, and the 95

percentile estimate was 80 %. This estimate was based on an expert

elicitation panel since none of the specific features of the retention mechanisms in the secondary side of a

SG as modelled in severe accident codes were assessed against any experimental data; nor was it possible

to quantify the uncertainty involved in the code predictions.

The SGTR project of the 5

Framework Programme of Euratom is the first European project (2000 - 2002)

started generating understanding in a systematic way for possible mechanisms for retaining aerosol

particles in tubes and in the complex structures of the secondary side of a SG. In particular, PSAERO and

HORIZON experiments from Finland were conducted for studying in-tube retention, whereas the retention

in the bundle region was investigated in the ARTIST and PECA-SGTR experiments, respectively from

Switzerland and Spain. In addition, certain modelling efforts to develop a correlation for the retention in

the bundle were initiated.

International cost share project ARTIST (AeRosol Trapping In a Steam GeneraTor), run and coordinated

by PSI, (2003-2007) studies aerosol behaviour in tubes of a western design inverted u-tube SG, in the

vicinity of the tube break where a transition from a very high to low flow velocities occurs, the far field

stages of the bundle, the separator and dryer. A wide range of operational conditions of the components

(namely, dry, wet and transition), and thermal-hydraulic and aerosol parameters are addressed.

Additionally the droplet behaviour in the separator and dryer is systematically investigated. An extensive

interpretation of the data and modelling efforts has accompanied the experimental program.

The SARNET (Network of Excellence for a Sustainable Integration of European Research on Severe

Accident Phenomenology) project of the 6

Framework Programme of Euratom (2004 - 2007) provides a

further forum where additional efforts are coordinated to understand the complex aerosol-phenomena and

to model the available data within the SARNET project.

NEA/CSNI/R(2009)5

3.4.2 Theoretical background on possible aerosol removal in the steam generator

The main components of a SG are the inlet and outlet plenums, tube bundle, droplet separator and steam

dryer. The western design SGs are vertically built and the bundle is composed of straight tubes or inverted

u-tubes connecting the inlet and outlet plenums. The Russian VVER designs utilise a horizontal bundle.

As a result of large variations of the flow velocity, and the differences in the geometry of these main

components, the important processes are introduced in the following subsections depending on their

physical location in the SG.

The following subsections focus on the aerosol removal processes in a vertical SG bundle with an inverted

U-bend. Tentative values of various magnitudes important for aerosol behaviour are provided in the

following subsections in order to give the reader a feel about their significance. The values are based on a

break flow estimated by assuming that there is only one ruptured tube and the ratio of the primary to the

secondary pressure is about 5.

3.4.2.1 Turbulent deposition inside the broken tube

For any size of the tube break, the flow out of the broken tube is choked because the pressure ratio between

the primary and the depressurised secondary side is bigger than two. This produces velocities inside the

tube, which are on the order several 100 m/s. At these velocities, turbulent deposition is the dominant

aerosol removal mechanism. Turbulent deposition is an eddy diffusion-driven mechanism that becomes

important whenever the flow is highly turbulent. Earlier studies [1] indicate that turbulent deposition at

these conditions is important and is almost solely a function of tube dimensions (L/D) for particles of

interest (> 0.3 µm). On the other hand, resuspension may also become important at these high velocities,

and somewhat cancel the retention to a degree yet to be investigated. If the break is located in the

downstream side of the bend at the cold leg side of the bundle, relatively higher aerosol deposition can be

expected in the bend. However, such deposition can cause gradual increase in the pressure drop, which

might in turn cause a rapid flushing of deposits. Other possibility for an additional deposition is the fission

product vapor condensation in the tubes if the aerosol–laden gas flowing into the inlet plenum from the hot

leg is at a temperature of >900 K, a level at which a certain fraction of volatile fission product mass is still

in vapor form. This vapor will then condense on the inner surface of tubes as well as on the airborne

particles as the gas cools off due to heat transfer. However, any increase in pressure drop in the broken

tube could easily cause flow redistribution among the other intact tubes. Such induced flow redistribution

will definitely contribute to retaining aerosols and fission product vapors in the intact tubes. The very

dynamic nature of these complex processes of coupled aerosol deposition/resuspension-flow

hydrodynamics requires very detailed and simultaneous treatment of the flow and aerosol dynamics and

thermodynamics that is not yet attempted.

3.4.2.2 Inertial and turbulent deposition on the secondary structures in the bundle

Inertial and turbulent deposition in the break stage

The break stage is the section of the bundle between two support plates (the first could be also the tube

sheet) where the aerosol-laden gas is discharged from the ruptured tube. The flow at the break location will

be sonic, and local velocities are expected to be on the order of several 100 m/s. The aerosol retention by a

single cylinder in turbulent cross flow has been investigated by Douglas [2, 3].The collection efficiency

was correlated as a function of the Stokes number, which is defined by:

NEA/CSNI/R(2009)5

UCd

Stk

spp

≡=

where ρ

is the particle density, d

the particle geometric diameter, U the gas velocity, C

the slip

factor, µ the gas viscosity, D the cylinder diameter, and τ the particle relaxation time. Douglas [2] found

that the collection efficiency varied between 2 % to about 30% when the Stokes number was varied

between 0.004 and 0.06. For full nominal flow rate (about 900 kg/h), it is expected that the velocity will

decrease from a couple of 100 m/s at the break to 10 m/s over the distance spanned by a few tube rows.

Table 3.4-2 below illustrates the expected Stokes numbers and particle relaxation time for three different

aerosol particle sizes at two representative flow velocities.

Table 3.4-2 Stokes numbers for break stage velocities

AMMD

µm

Stk

U= 300 m/s

Stk

U= 10 m/s

1 5.8 10

-6

0.09 0.003

3 4.6 10

-5

0.72 0.024

10 3.8 10

-4

6.0 0.200

In the actual case the retention is a function of many multiple and dynamic parameters:

− Break type and orientation induced velocity distribution,

− Aerosol characteristics (size distribution, stickiness, etc.),

− Gas and wall temperatures, heat-up of the structures,

− Transport and condensation of fission product vapor, if available.

− Deagglomeration of aerosol agglomerates due to high shear forces produced by the expansion, if it

occurs, which might modify the aerosol size distribution,

− Many different aerosol deposition processes, aerosol growth by vapor condensation, removal by

interception and impaction, etc.

Inertial impaction on the support plates

Away from the break, the flow will spread and move upwards towards the support plate. As a result the

velocity will decrease considerably compared to the sonic speed at the break point. The flow exits the

break stage through passages in the support plate. The passages are narrow enough that there exists a

potential for inertial impaction on the plate surface. The aerosol retention at a contraction is a function of

flow and aerosol characteristics [4] as measured by the Stokes number. For flow at contractions, the

aerosol retention has traditionally been correlated as a function of velocity in the far field and the diameter

of the contraction. The mean velocity U in the bundle can be obtained from the following relationship:

UAm

NEA/CSNI/R(2009)5

where

m is the flow rate out of the break, ρ the gas density, and A the free flow area in the bundle. At

a typical secondary temperature of 400 °C and pressure of 1 bar, the gas density is 0.32 kg/m

, and thus the

mean velocity U in the bundle space is 0.2 m/s for the nominal full flow rate. For an effective diameter of

each hole in the support plate of about 8 mm, Stokes numbers as a function of particle diameter at a mean

velocity of 0.2 m/s are shown in the Table 3.4-3below:

Table 3.4-3 Stokes numbers for far field mean velocity of 0. 2 m/s

AMMD

µm

Stk

(Stk)

1/2

1 5.8 10

-6

1.5 10

-4

0.012

2 4.6 10

-5

1.1 0

-3

0.033

10 3.8 10

-4

9.6 10

-3

0.098

In the experiments [4] investigating deposition on the upstream face of a critical orifice, it is found that

there is no aerosol retention below Stk

1/2

of 0.3. Similarly, in the experiments [4] on aerosol collection in

membrane filters, very little retention was found for values of Stk

1/2

below 0.2. Simple extrapolation of

these results to SG tube support plate might result in expectation of little aerosol deposition on the support

plate. However, since the geometry of the holes in the support plate and the flow field underneath or just

above the tube support plate are quite complex, the extent of validity of this extrapolation is in question.

Flow expansion at the top surface of the support plate can induce recirculation that could easily alter the

deposition pattern of the aerosols. Such details require an in depth analysis of local flow fields in detail and

particle tracking.

Inertial impaction on the tubes in the far field stages and on U-bends

Beyond the break stage the flow will be mainly in the vertical direction, and little retention by inertial

impaction on the tube structures is expected at a gas mean velocity of in the order of 0.2 m/s. On the

U-bend region, the velocity would remain quite small and the retention, based on the Stokes number, can

be estimated from the cross-flow data of Douglas [2]. The Stokes numbers are shown for different aerosol

sizes in Table 3.4-4 for a flow velocity U of 0.2 m/s and the tube diameter D of 19 mm. The results from

Douglas indicate very insignificant retention for Stokes number below 4.0 10

-3

., which means that for

particles under 10 µm the retention should be insignificant if the inertial impaction is the only mechanism.

Table 3.4-4 Stokes numbers for U -bend section

AMMD

µm

Stk

1 6.2 10

-5

3 4.8 10

-4

10 4.0

-3

Although the aerosol removal by inertial impaction alone on the surfaces of the straight tubes and on the

U-bend can be small, however, expected enhanced recirculation especially in the bend region might

NEA/CSNI/R(2009)5

modify the retention process considerably. An experimental and analytical database on the latter is

currently not available.

If the break location is within the U-bend section, depending on the number of tube rows that the flow

cross through until it reaches the free space below the separator inlet, the retention processes introduced

under ‘break stage’ also apply here.

3.4.3 Inertial impaction and interception in the separator and dryer sections

Separators and dryers, due to their complex geometry produce very complex and changing flow patterns

and velocities. Inertial impaction, interception, turbulence initiated particle agglomeration are the main

processes that can have a role on the particle size dependent aerosol removal. The swirl vane, cyclones and

the lid (which hinders water up flow in the normal operation) are the main components of any PWR

separators that modify the flow pattern and hence interact with the motion of the particles. The retention

efficiency of these components under normal power operation for removing large size water droplets in the

order of several 10 µm is very high. However, due to relatively small flow rate and hence gas velocity and

the micron-sub-micron aerosol size expected under severe accident conditions, the removal efficiency of

these components at full power conditions cannot be applicable to the severe accident conditions.

3.4.4 Other processes

Agglomeration

If the aerosol concentration and residence time in the secondary side are large enough, agglomeration

occurs and shifts the original size distribution. The aerosol concentration in the secondary side of a PWR

can be on the order of 10 g/m

. Assuming spherical aerosols with a density of 2 g/cm

, the number

concentration for 1 µm aerosols is in the order of 3

particles/cm

. Based on the theory of

monodispersed agglomeration [5], the number concentration as a function of time is given by

NKt

()=

where

N(t) = number concentration at time t

= number concentration at time 0

K = rate constant = 3.5 10

-10

/s (for diameters greater or equal to 1 µm)

t = time in seconds

Assuming an average residence time of about 60 s based on an estimated break flow of 900 kg/h from a

broken tube (about 45 s in the bundle, 10 s in the separator, 10 s in the space between the separator and

dryer), one has:

Nt N() .

064

Thus, the number concentration drops by about one third in the SG, assuming no other retention

mechanism enters into play. The diameter of the particles is given by

3/1

16.1)()( d

dtd ×==

NEA/CSNI/R(2009)5

Thus, in the course of the 60 seconds, agglomeration would cause an increase of 16 % in the original

diameter. If the flow rate is smaller then the residence time will be increased linearly.

The turbulence-initiated agglomeration can take place in various locations as a result of change in the

geometry, shape of the hardware causing interferences. As an example to the first is the small flow

passages around the tubes in the support plates, which cause contraction and expansion and hence creation

of certain turbulences on the top surface of the support plates, the zigzag shape and the pockets of dryer

panels for the second. The quantification of the effect of turbulence requires detailed coupled analyses of

flow hydrodynamics and the particle tracking, which is still in its research and development phase.

Gravitational settling

Settling velocities as a function of particle diameter is given in Table 3.4-5 below. Even for a large 10 µm

particle, the settling velocity is only 18 cm/min, meaning that gravity settling in the secondary side of a SG

is negligible for a residence time of the order of 1 minute, shown earlier to be the case.

Table 3.4-5 Settling velocities for different AMMD's

AMMD

µm

Settling velocity

cm/min

1 0.21

3 1.8

10 18.3

Thermophoresis

Following an accident with dry SG secondary side, the incoming gas is hot (300 °C - 900 °C) while the

tube structures have temperatures of about 300 °C, thus deposition by thermophoresis is possible. The

thermophoretic velocity depends primarily on the temperature gradient and is a mild function of particle

size [5]. Typical velocity values per unit temperature gradient are given in Table 3.4-6 below.

Table 3.4-6 Thermophoretic velocities per unit temperature gradient

AMMD

µm

Thermophoretic velocity for unit gradient (1 °C/cm)

cm/s

1 1.3 10

-4

3 1.0 10

-4

10 7.8 10

-5

For the purposes of quantifying the possible effect of thermophoretic deposition, the maximum

temperature difference between the gas and the structure can be assumed to be 500°C. Given that the

distance between the flow centreline and the tube wall is 1 cm, the temperature gradient is of the order of

500 °C/cm. Thus the maximum expected thermophoretic velocity is as given in Table 3.4-7 below:

NEA/CSNI/R(2009)5

Table 3.4-7 Thermophoretic velocities for expected temperature gradient

AMMD

µm

Thermophoretic velocity

cm/s

1 0.065

3 0.050

10 0.040

Since the residence time in a far-field stage is of the order of 6 seconds (1.1 m height and 0.2 m/s velocity),

one can see from the above table that the horizontal displacement due to thermophoresis is of the order of

0.3 cm in a stage. Given that less than 1 cm separates the flow centreline from the tube wall, it is concluded

that thermophoresis can lead to measurable deposition in dry SG tube rupture conditions.

Aerosol removal in flooded bundle

For accident management purposes, water injection in the dry secondary side may be an option in order to

re-establish heat removal and provide a pool where the incoming aerosols can be scrubbed. The

POSEIDON [6] pool scrubbing experiments conducted by PSI provided a database on the aerosol removal

efficiency of hot pools. The data indicated for example that for a high carrier gas steam fraction (70 %),

and a jet injection regime, the decontamination factor (DF) can be higher than 10 even for a shallow pool

of 0.3 m submergence. In the context of SG tube rupture, the DF can be expected to be even more

significant in view of the high velocity of the jet, the potentially deeper pool, and the presence of dense

structures, which can easily disintegrate the incoming jet into a multitude of smaller bubbles with a greater

potential for scrubbing than regular pools.

The gas-pool interactions can be analysed in three regions:

• The immediate injection zone characterised by the formation of a gas jet or globule depending on

the gas flow rate;

• The break-up zone where the jet or large globule disintegrates into smaller bubbles;

• The bubble rise zone, where individual bubbles rise through the pool stages with a terminal

velocity, and periodically squirt out from the support plate narrow constrictions.

Beyond the break-up zone (typically 10 globule diameters away from the injector), the bubbles are

oblivious of the details of the injection process, and thus the carrier gas injection rate has no effect on

decontamination in the bubble rise zone.

When a jet enters a pool, some of the incoming steam condenses almost instantaneously, with a

corresponding scrubbing of a fraction of the aerosols. The extent of condensation depends on the

temperature of the water pool. If the pool is close to boiling, which is likely in the long term because of the

hot structures and continued steam input, little condensation is to be expected. Hence, the main aerosol

removal mechanisms in the hot pool would be inertial in nature: i.e.

• Jet impaction at the injection point;

• Centrifugal impaction and gravitational settling during bubble rise.

NEA/CSNI/R(2009)5

The first mechanism depends on the injection flow rate. This supposes that the jet momentum would be

locally dissipated because of the dense structures as well as the buoyancy of the bubbles. In addition past

experience [6] indicates that the jet is fairly confined near the injection point where most of the removal

takes place.

References

[1] B.Y.H. Liu and J.K. Agarwal, “Experimental Observation of Aerosol Deposition in Turbulent

Flow“, J. Aerosol Sci., Vol. 5, pp. 145-155, 1974

[2] P. Douglas and S. Ilias “On the Deposition of Aerosol Particles on Cylinders in Turbulent Cross

Flow“, J. Aerosol Sci., Vol. 19, No.4, pp. 451-462, 1988

[3] D. Leaver, J. Li, R. Sher “New Design Applications of Natural Aerosol Deposition in Nuclear Plant

Accident Analysis“, OECD-CSNI Meeting, June 1998

[4] Y. Ye and D.Y.H Pui “Particle Deposition in a Tube with and Abrupt Contraction“, J. Aerosol Sci.,

Vol. 21, No.1, pp. 29-40, 1990

[5] W. Hinds Aerosol Technology, John Wiley & Sons, 1982.

[6] A. Dehbi, D. Suckow, S. Güntay The Effect of Liquid Temperature on Pool Scrubbing of Aerosols“,

J. of Aerosol Science, Vol. 28, Suppl. 1 pp. S707-S708, 1997

3.5 Resuspension

3.5.1 Resuspension in primary circuit

Introduction

Particles deposited on the surfaces of the primary coolant circuit may resuspend back to the gas stream.

Such phenomenon can take place, if the removal forces affecting the particles are greater than forces that

adhere them to the surface. Usually resuspension is caused by sudden increase in the gas flow rate. In a

severe accident this may happen, if the rate of steam generation increases for example because of core

quench or relocation of molten material. Also either intentional or accidental depressurization of the circuit

would significantly increase the flow rate. A very high gas flow rate causes particles to resuspend, even if

the flow rate is constant.

Resuspension decreases retention of radionuclides into the circuit. The phenomenon is especially important

in by-pass sequences, in which the radionuclides may be released directly to the environment. In these

sequences the flow rates may also be very high.

ISP-40, which was based on test SR11 of the STORM series, was the first modeling exercise to address

aerosol resuspension in LWR explicitly. The outcome of the exercise was that concerning resuspension

none of the participants succeeded in coming close to the experimental results. There was disagreement

even on which are the important parameters that affect resuspension [1].

Particle adhesion

In order to calculate particle resuspension it would be important to have an estimate of the forces that

bonds particles to the surface. It is known that the most important adhesion force for dry particles is van

der Waals force. Other electrostatic forces also bond particles to the surface, but their effect is much

NEA/CSNI/R(2009)5

weaker [2]. If there is liquid material in the deposit, surface tension may become the dominant adhesion

force. In addition to these, sintering and chemical bonding may further increase the strength of the deposit

layer.

In principle it is possible to calculate the strength of the van der Waals force for spherical smooth particles,

when Hamaker constant for the material is known and the deformation of the surfaces has been taken into

consideration. The calculated results have matched atomic force microscope (AFM) measurements closely

[3]. However, the adhesion force is very sensitive to the shape of the particles and to surface roughness.

Experimentally the adhesion force has been observed to decrease by an order of magnitude, when the

roughness of the surface has been increased from 0.2 nm to 0.7 nm [4]. For this reason direct calculation of

the adhesion force is not feasible in reactor safety applications.

In practice adhesion force is always randomly distributed between the particles. Experimentally the

adhesion force distribution of single particles on the surface has been observed to approximately follow a

log-normal distribution. Log-normal approximation has also been used, when particle resuspension both

form monolayer [5, 6] and multilayer [7, 8, 9] deposits has been modeled. It should be noted however, that

the adhesion force distribution of a thick deposit layer has not been directly measured.

In a thick deposit layer the number of contacts between particles should greatly influence the strength of

the adhesion force. The number of contacts increases as the porosity of the deposit decreases.

Experimentally it has been verified that resuspension is much easier from high porosity deposits as from

low porosity deposits formed of the same material. Also the difference between PARESS and STORM

experiments was explained by the difference in porosities between the deposits [8]. According to the study

the main deposition mechanism in PARESS experiments was settling. The resulting deposit layer was

highly porous and easily resuspended. In STORM experiments the deposit was formed by thermophoresis,

which resulted in a much denser and stronger particle layer, because polydisperse aerosol form a stronger

deposit with lower porosity than monodisperse particles. This was shown in a study by Biasi, in which the

results from several experimental series were compared [9]. When modeling resuspension, parameters

affecting the porosity should thus be included. These would include at least the size distribution and the

velocity of the particles depositing on a surface. Hardness of the deposited particles as well as the fraction

of liquid material on the deposit layer would also be parameters, which could be relatively easily included

in the integral codes.

Removal forces

Forces that cause particle resuspension can be divided to lift forces affecting normal to the surface and drag

forces, which tend to move particles parallel to the surface. When the particles are within the viscous

sublayer of turbulent flow, the lift forces should be several orders of magnitude weaker than the adhesion

forces [10]. Drag forces are not only much stronger, but in experiments done with a centrifuge it has been

found out that parallel force required for particle detachment are approximately 1/100 of the required

normal force [11]. Parallel forces need to provide energy only to overcome deformation of particles,

whereas normal forces have to move the particles out of the range of conservative forces.

Therefore, particles have been observed to move on the surface always before they are resuspended to the

gas stream [10]. Because incipient motion on the surface decreases the contact area between particles and

the surface, even a relatively weak lift force may resuspend the particles. Experimental observations also

show that particles detach from surfaces in discrete events, occurring at random intervals. These

observations relate the resuspension to the burst-sweep events taking place in the laminar sublayer of

turbulent flows. First resuspension model by Cleaver and Yates assumed that resuspension depends on the

frequency, force and area of these events [12].

NEA/CSNI/R(2009)5

In some studies it has been claimed that only long term resuspension would be related to turbulent bursts.

According to these studies, the rapid resuspension following the increase in the flow rate would be related

to the increase in the average drag force. However, in his experiments Wu showed that even short term

resuspension during the flow acceleration was significantly decreased, when only the large scale

turbulence was dampened [13]. In other studies it was found out that increasing the flow acceleration

increases the resuspension rate in short term but substantially decreases it in the long term. These studies

strongly support the conclusion that short and long term resuspension are based on the same phenomenon

and can not be treated separately.

The diameter of the deposited particles is also an important parameter in many resuspension models. As

evident from studies with monolayer deposits, the flow rate needed to cause resuspension decreases as the

particle size increases. In these experiments the drag force affecting particles increases with their size.

However, particles resuspend as agglomerates from a thick deposit layer. Therefore the primary size of the

particles can not be directly used as a parameter, when resuspension is modeled.

In addition to turbulent gas flow also impaction of particles can cause already deposited aerosol to detach

from the surface. Erosion has been observed to remove particles efficiently from the surface in jet flows

although a pure gas flow did not have the same effect. This was explained by the fact that the momentum

of a particle is approximately three orders of magnitude greater than the momentum of gas with the same

volume [14]. In her experiments Theerachaisupakij found that gas flow containing at least micron sized

particles was substantially more efficient in causing resuspension than pure gas flow. Resuspension from a

pipe surface was enhanced as the size of the impacting particles was increased. Submicron particles did not

increase the resuspension rate [15]. In PSAERO experiments only a small fraction of particles resuspended

from the pipe inlet into pure gas flow. The fraction of resuspended particles increased towards the outlet of

the pipe as the concentration of particles in the flow increased. The mass median diameter of particles in

these experiments was 3 µm and their density was 9 g/cm

[16]. Erosion is an especially efficient process,

if the gas flow rate is so high that particles impacting on the surface bounce back to the gas stream.

Whether the particles bounce depends on their size and velocity as well as their elasticity and the angle

with which they hit the surface [17]. Particle erosion is likely to be an important process in a severe

accident. Especially during core relocation the size of the aerosol particles is probably large and their mass

concentration in the gas flow is high.

Resuspension dynamics

The adhesion force distribution of deposited particles changes due to resuspension. The average adhesion

force in a deposit layer tends to increase, because loosely adhered particles are likely to be the first to

resuspend [8, 18, 19]. The increase in the average adhesion force and the discrete random nature of the

turbulent bursts also explains why the resuspension rate measured in a number of studies decreases

exponentially after the flow rate has been increased [1, 7, 16, 20].

A fact receiving less attention is that resuspension is a continuous process in a turbulent flow as particles

are depositing on the surface. Because of resuspension, the adhesion force of a deposit increases as the

flow rate during the deposition phase increases. In PSAERO experiments flow rate during the deposition

phase was a major parameter influencing the resuspension rate. If the flow rate was lower than during the

deposition phase, particles did not significantly resuspend even if the flow was pulsed [16]. Resuspended

particles may also deposit back to the surface in the turbulent flow. When they do, they adhere to the

surface on average more strongly than before resuspension.

The history of the flow rate during the deposition phase as well as the deposition mechanisms influence the

adhesion force distribution. Because of this memory effect, resuspension has to be modeled dynamically.

The state of the deposit on the surface has to be also tracked. In addition to adhesion force distribution, the

NEA/CSNI/R(2009)5

model should contain frequency, force and area of the turbulent bursts. Erosion caused by depositing

particles could be calculated using a similar frequency distribution. Most deposition mechanisms are

already adequately modeled for the purpose of predicting resuspension. However, large uncertainties

remain in the estimation of turbulent eddy impaction. Obtaining a realistic model for eddy impaction is

especially important, because the same mechanisms that carry particles to the surface also cause

resuspension.

3.5.2 Resuspension in containment

A large fraction of the fission products released from the reactor cooling circuit into the containment is

bound to aerosol particles. The aerosol concentration is expected to be high in the early phase of an

accident scenario. Later on the aerosol concentration decreases due to sedimentation of aerosols into liquid

pools or onto solid surfaces. Atmospheric air currents inside the containment are able to re-release already

sedimented aerosol material from solid surfaces into the containment atmosphere. This effect is called dry

resuspension. Possible reasons for such heavy air currents can be hydrogen deflagrations or steam

explosions. For the determination of the aerosol source term a detailed balance of the release, the

sedimentation, and the resuspension of aerosols is necessary.

The detachment of the aerosol particles from the surface is a very complicated process that depends on

many unknown parameters. The particles are bound to the surface due to different physical effects that are

gravitational forces, Van-der-Waals forces, and electrostatic forces, if the particles carry a small electrical

net charge. In the presence of humidity liquid bridges retain the particles on the surfaces. In a multilayer

particle bed agglomeration of the particles occurs and a dense particle layer is formed. The aerodynamic

forces that lift the particle up into the atmosphere are very hard to determine. It is agreed in literature that

the air current must be turbulent, if resuspension occurs. It is distinguished between different mechanisms

of resuspension. Erosion describes the steady release of aerosol particles from the surface layer by layer

over a longer period of time, while denudation describes the abrupt release of larger amounts of aerosol

material in a fraction of a second. The latter effect occurs only, when the velocity above the particle bed is

increased suddenly. Further resuspension appears under the impaction of suspended aerosol particles onto

a deposition bed. Important parameters are the wall shear stress of the airflow over the surface and the size

of the aerosol particles deposited. The influences of these parameters define different ranges of the particle

size, in which different adhesive and aerodynamic forces dominate. These influences are theoretically

described by Phillips [21].

Most models describe the resuspension under the condition of a constant airflow over a longer period in

time of at least several minutes and the particles are removed layer by layer on the basis of erosion. Such

conditions are expected during the transport of aerosol in the primary cooling system. In these models the

particle bed is completely submerged in a laminar boundary layer that is damaged by randomly occurring

bursts. Under these bursts aerodynamic forces act on the particles and lift them up into the atmosphere. A

bursting frequency n is assumed that determines how often a burst occurs. The quantity of the removed

aerosol mass under a single burst is gained from a criterion that compares the magnitude of the acting

aerodynamic and adhesive forces. Such force balance models are developed by Fromentin [18], Braaten et

al. [22] and de los Reyes et al. [23] for instance. Reeks [11] has chosen another approach and developed an

energy cumulation model. In this model the particles oscillate on the surface and perform a rolling

movement. With every burst the particle absorbs energy and oscillates with a higher frequency until it

gains enough energy to leave the surface and is entrained by the airflow current. Different versions of such

models have been built into severe accident codes and a benchmark has been performed on ISP40

consisting of a combined deposition and resuspension experiment in the STORM programme (de los Reyes

et al. [1]).

NEA/CSNI/R(2009)5

In contrast to the erosion the process of the denudation is very little modelled. It is expected to be the

dominating effect, if an air current lasts only a short period in time. These conditions are fulfilled by

hydrogen deflagrations or steam explosions in the containment. A model on the basis of the resuspension

model of Fromentin was built into the GRS containment code COCOSYS (Nowack et al. [24]). The

advantage of the Fromentin model in comparison to other erosion models is that it describes a multi-layer

particle bed. In the modified version for transient events the aerosol particles are not detached layer by

layer in different time steps, but a single event can resuspend multiple particle layers at once. The strength

of the airflow is only characterised by a maximum velocity and the duration of the current is limited to a

few seconds. These changes better fit the conditions found under transient air currents generated by

hydrogen deflagrations or steam explosions.

References

[1] A. de los Reyes, J.A. Capitão, G.F. De Santi, G.F. International Standard Problem 40 - Aerosol

deposition and resuspension Final comparison report. Joint Research Centre of the European

Communities Report

: EUR 18708 EN, NEA/CSNI/R(99)4, Ispra, February 1999

[2] D.S. Rimal, L.P. Demejo, R.C. Bowen Particle adhesion: Recent advances and current challenges

Mittal Festschrift (eds.) W. J. Van Ooij, Anderson Jr., H. R., p.

161-178, 1998

[3] S. Eichenlaub, C. Chan, S.P. Beaudoin “Hamaker Constants in Integrated Circuit Metalization”

Journal of Colloid and Interface Science 248: 389-397, 2002

[4] Y.I. Rabinovich, J.J. Adler, M.S. Esayanur, A. Ata, R.K. Singh, B.M. Moudgil "Capillary forces

between surfaces with nanoscale roughness" Advances in colloid and interface science 96: 213-230,

2002

[5] M.W. Reeks, J. Reed, D. Hall "On the resuspension of small particles by a turbulent flow” Journal of

Phys. D: Appl. Phys. 21: 574-589, 1988

[6] G. Wen, H.Y.a.K. "On the kinetics of particle reentrainment from surfaces" Journal of Aerosol

Science 20(4): 483-498, 1989A

[7] A. Fromentin "Time dependent particle resuspension from a multi-layer deposit by turbulent flow"

Journal of Aerosol Science 20(8): 911-914, 1989B

[8] H. Friess, G. Yadigaroglu Inclusion of Structural Parameters in the Modeling of Aerosol

Resuspension 3

OECD Specialist Meeting on Nuclear Aerosols in Reactor Safety, 15-18.6.1998,

Cologne, Germany, 1998

[9] L. Biasi, A. de los Reyes, M.W. Reeks, G.F. de Santi “Use of a simple model for the interpretation

of experimental data on particle resuspension in turbulent flows” Journal of Aerosol Science 32:

1175-1200, 2001

[10] A. Ibrahim, R. Brach, P. Dunn "Microparticle detachement from surfaces exposed to turbulent air

flow: microparticle motion after detachement" Journal of Aerosol Science 35: 1189-1204, 2004

[11] M.W. Reeks, D. Hall Kinetic models for particle resuspension in turbulent flows: theory and

measurement In: Journal of Aerosol Science 32, p. 1-31, 2001

[12] J.W. Cleawer, B. Yates “Mechanism of detachement of colloidal particles from a flat substrate in a

turbulent flow” Journal of Colloid Interface Science 44: 464-474, 1973

NEA/CSNI/R(2009)5

[13] Y.-L. Wu, C. Davidson, A. Russell “A Stochastic Model for Particle Deposition and Bounceoff”

Aerosol Science and Technology 17(4): 231-244, 1992B

[14] W. John, D.N. Fritter, W. Winklmayr “Resuspension induced by impacting particles” Journal of

Aerosol Science 22: 723-736, 1991

[15] W. Theerachaisupakij, S. Matsusaka, Y. Akashi, H. Masuda. “Reentrainment of deposited particles

by drag and aerosol collision” Journal of Aerosol Science 34: 261-274, 2003

[16] A. Auvinen, J. Jokiniemi, T. Renvall Aerosol Resuspension Experiments in Horizontal SG Tube. EC

Report: SAM-SGTR-D025, VTT Processes report

: PRO3/P9/03, Espoo, February 2003

[17] X. Li, P.F. Dunn, R.M. Brach “Monte Carlo Analysis of Polydisperse Microparticle Impacts with

Surfaces” Aerosol Science and Technology 33(4): 376-385, 2000

[18] A. Fromentin Particle Resuspension from a Multi-Layer Deposit by Turbulent Flow Paul Scherrer

Institut (PSI 38), 1989

[19] S. Matsusaka, H. Masuda “Adhesive strength distribution of particles deposited on a wall surface”

Journal of Aerosol Science 32 S1: S937-S938, 2001

[20] H.Y. Wen, G. Kasper, R. Udischas “Short and long term particle release from surfaces under the

influence of gas flow” Journal of Aerosol Science 20(8): 923-926, 1989B

[21] M. Phillips A force balance model for particle entrainment into a fluid stream In: Journal of Physics

D: Applied Physics 13, p. 221–233, 1980

[22] D.A. Braaten, U.K.T. Paw, R.H. Shaw Particle resuspension in a turbulent boundary layer-observed

and modelled In: Journal of Aerosol Science 21, No. 5, S. 613–628, 1990

[23] A. de los Reyes, E. Hontañón, R. Arias, J.A. Capitão The CÆSAR model for particle resuspension

in turbulent flows In: Journal of Aerosol Science 28 Suppl. 1 (1997), p. 327-328, 1997

[24] H. Nowack, M. Dapper, H.-J. Allelein, M.K. Koch Aerosol resuspension phenomena in the

containment In: Annual Meeting on Nuclear Technology, Aachen, 2006

3.6 Pool Scrubbing

Scenario description and characterisation

Pool scrubbing or wet scrubbing is the removal of aerosol particles in gas bubbles rising in a water pool.

The pool thus acts as a filter. Pool scrubbing has applications in various industrial fields where aerosol

particle laden gas cleaning is required. The following deals only with aerosols scrubbing; vapor scrubbing

is out of its scope.

Pool scrubbing is a very relevant issue in the nuclear safety since it provides a means to reduce source term

to the environment during hypothetical severe accidents. Several severe accident scenarios involve the

transport paths of fission product aerosols which include passages through stagnant pools of water where

pool scrubbing can occur. In Boiling Water Reactors (BWR’s) pressure suppression pools (“wetwell”), are

such pools although they are primarily designed to avoid over pressurization of the wetwell space. Pool

scrubbing in such pools has been given credit as an engineering safety feature capable of mitigating the

source term and hence associated risk posed by accidents. Nonetheless, the pool scrubbing scenario is not

NEA/CSNI/R(2009)5

BWR specific. Some Pressurised Water Reactor (PWR) sequences may also involve transport paths where

pool scrubbing can occur. As an example: A V-Sequence, involving release of radioactivity through safety

injection piping (either due to a break or malfunctioning check valves) directly into a tank containing

water located in the auxiliary building. Fission products can be scrubbed if the transport path involves a

full pressuriser and a stuck open safety relieve valve. During Steam Generator Tube Rupture (SGTR)

accidents concurrent with the stuck-open safety relief valve, a good potential for aerosol scrubbing is

provided if the secondary side is filled with water. Another scenario common to any type of reactor is a

severe accident with vessel breach in which the molten material falls into the reactor pit or drywell and

attacks concrete. Corium concrete interaction causes release of medium and low volatile fission products as

well as the production of large amounts of concrete aerosols. The gas generated in the interaction will

transport the aerosol particles. If the corium is overlaid by a water pool with a certain height it the latter

also acts as an aerosol scrubber which attenuates the aerosol particle release to the containment

atmosphere. Containment venting filter systems employing water pools utilise also the pool scrubbing to

reduce the activity level out of a containment below limits defined by the regulatory bodies.

Several fundamental processes take place during aerosol pool scrubbing: diffusiophoresis, thermophoresis,

inertial impaction at the nearby of gas injection, gravity settling, centrifugal deposition and diffusion

during bubbles rise, Brownian diffusion, etc. Aerosol characteristics, i.e., size, hygroscopicity, etc, are the

key factors for the effectiveness of these removal processes. Gas hydrodynamics plays an essential role

determining key variables for pool scrubbing such as bubbles size and surface/volume ratio. In addition,

other parameters like pool depth water sub-cooling, carrier gas composition and temperature and velocity,

injection mode, water composition, etc., heavily influence individual pool scrubbing processes. In

Table 3.6-1 and Table 3.6-2 the ranges of some of these key variables during relevant severe accident

sequences [1, 2] are shown.

Table 3.6-1 Pool scrubbing boundary conditions in BWRs

Sequence

Particle

AMMD (µm)

Submergence

(m)

Injection

velocity

(cm/s)

Mass %

of steam

Temperature

SBO HP-ST

< 3

(“Quencher”)

0.2 – 1.5

(downcomer)

2 – 5

(“Quencher”)

(downcomer)

25-3000 /

1-15

10-97

subcooled pool

SBO HP-LT 25-3000

SBO LP-ST 3000-9000

SBO LP-LT 100-25000 / 1-15

NEA/CSNI/R(2009)5

Table 3.6-2 Pool scrubbing boundary conditions in PWRs

Sequence

Particle

AMMD (µm)

Submergence

(m)

Injection

velocity

(cm/s)

Mass %

of steam

Temperature

SGTR --- < 12 40 – 5000 10-90 saturated

RHR 0.75 – 2 100 - 250 3 - 97 saturated

Traditionally, the scrubbing efficiency has been expressed in terms of a Decontamination Factor (DF),

which is defined by the ratio of the aerosol mass flow rate entering (m

) and the leaving (m

out

) the pool:

out

DF =

Based on the different nature of hydrodynamic- and aerosol phenomena, the path of aerosols along the pool

height is usually split into three regions [3]: injection, rise and pool surface. Consistently, the overall

Decontamination Factor is a multiplication of individual Dfs of the three regions of the pool:

surriseinj

DFDFDFDF

⋅

The injection and surface decontamination processes are considered instantaneous whereas it is assumed

that several continuous aerosol scrubbing mechanisms are active during bubble rise which can be described

by:

sur

rise

ninj

dA)r(v

expDFDF ⋅













⋅













∑

⋅

∫

⋅=

The depletion velocity of each individual mechanism is represented by v

and the time of the bubble in the

rise zone is the ratio of water height (S, submergence) and the bubble rise velocity (u

Injection zone

At the injection zone in the pool, the mechanical as well as the thermal gas-liquid interaction determine the

scrubbing process.

The gas injection velocity is a key variable in this region. The inlet gas regime is usually classified

according to the non-dimensional Weber number:

vDρ

injinj

⋅⋅

as either jet regime (We≥10

) or globule regime (We<10

). Under the jet regime a steady jet cone attached

to the injection orifice is formed. Inside the cone a transition from pure gas flow to a churn-turbulent

two-phase flow occurs. Intermediated steps in the transition are the bubbly flow and entrained droplet flow.

Aerosol impaction on liquid surfaces, either at the cone gas-liquid interface or at surfaces of entrained

NEA/CSNI/R(2009)5

droplets, can effectively remove particles from the incoming gas flow. At moderate or low gas velocities

characterising the globule regime, other depletion mechanisms such as diffusion, sedimentation and

centrifugal deposition could become important. The latter process accounts for the removal of big particles

which cannot follow internal gas streamlines and eventually strike the gas-pool surface. Primary bubble

oscillations would also enhance aerosol scrubbing. In Fig. 3.6-1 [4] the globule (bubble) and jet regimes

for various relevant accident sequences as characterised by a Weber number range are displayed.

Fig. 3.6-1 Weber range of relevant accident sequences

The thermal interaction between gas and liquid is determined by the temperature difference between both

phases and by the gas composition as well. Any temperature difference between gas and water will initiate

heat exchange. In the case where gas is hotter than water, particles would be carried to the gas-water

interface by thermophoresis. If the gas contains steam and the water temperature is at lower than the

saturation temperature the mass transfer as a result of condensation will initiate aerosol removal by

diffusiophoresis.

Bubble rise zone

As in the injection zone, the bubble hydrodynamics is closely linked to the scrubbing efficiency. The

primary bubble is usually not stable and splits to form a swarm of smaller bubbles. Gas dynamics is quite

intense during the bubble rise and a combination of many processes defining bubble size and the

distribution will occur. Bubble splitting, growth by vapor flux into the bubble (by hydrostatic pressure

decrease) and coalescence coexist. Some key hydrodynamic variables in the aerosol scrubbing are the

bubble size distribution and shape and the rising velocity. Small bubbles are usually spherical (i.e., the

largest surface-to-volume ratio), which enhances decontamination. However, larger bubbles, either

ellipsoidal or cap-shaped, can become more easily deformed, which also contributes to aerosol removal.

The bubble residence time (i.e., S/u

rise

) or gas holdup ultimately determines the final aerosol mass being

removed.

Regardless of the shape of the bubbles the major pool scrubbing mechanisms are the diffusion,

sedimentation and centrifugal deposition. The combined action of these mechanisms would lead to a

minimum of decontamination efficiency at particle aerodynamic diameters around 0.1 µm. In addition to

the main aerosol removal processes, change in the particle size directly affects the pool scrubbing.

Particles could grow, if the conditions prevail, because of agglomeration and/or steam condensation and/or

hygroscopicity. Agglomeration can be ignored due to short bubble residence time. The effect of

condensation becomes more significant if the particles are hygroscopic. Until the bubbles reach a thermal

JET BUBBLE

NEA/CSNI/R(2009)5

equilibrium and depending on the steam content, and the temperature gradient diffusiophoresis and

thermophoresis could contribute to the aerosol depletion process during bubble rise.

Pool surface zone

At the pool surface bubbles rupture causing production of many micro-droplets (from the fragmented

bubble skin). Some of these droplets can be entrained by the gas flow and the others fall back due to their

size by the gravity. These entrained droplets transport the very fine aerosol particles as well as dissolved

fission products and hence cause accumulation of the activity in the gas space. The amount of droplet

entrainment depends primarily on the bubble size distribution reaching the water surface and the gas

velocity.

In addition a minor experimental programme was intended to assess the effect of water impurities in the

pool scrubbing process [3].

Major outcome of most recent investigations

• On theoretical basis, independent investigations have shown that decontamination by bubble

formation and equilibration in a water pool can be significant, both in BWR’s and PWR’s risk

relevant sequences. Under such scenarios carrier gas composition and flow rate are the key

variables determining the decontamination efficiency. For shallow pools, the relative significance

of the bubble formation and equilibration processes in determining the decontamination can be

even larger than that by the decontamination process during the bubble rise through the pool

height.

•

A peer review of hydrodynamic models in different codes showed that they are remarkably

different in both their bases and results. Given their significance in decontamination, bubble size

and shape are particularly relevant. No less important, several hydrodynamic phenomena are

missing in codes despite they could notably influence particle retention: jet injection, primary

bubble rupture, churn-turbulent flow and bubble oscillations and deformations.

•

The DF displays an inverted Gaussian type of trend as a function of particle diameter with a

minimum at about 0.1 µm.

•

Uncertainties in the particle size distribution at the inlet can largely influence DF estimates. The

effect of pool scrubbing on the particle size distribution is narrowing towards the particle size

yielding the minimum scrubbing efficiency.

•

Experiments under different conditions have confirmed that steam condensation at the injection

point, particle size, carrier gas flow rate and composition affect significantly the DF, even under

hot conditions.

•

DF increases smoothly and exponentially with submergence. Increased gas residence time

through the pool efficiently raises the DF.

•

The decontamination factor of hot water pools is notably larger than 1, a value which was

assumed in the past safety studies evaluating the source term from BWRs. This is the case even

for small sizes of aerosols with AMMD’s on the order 0.2-0.3 µm.

NEA/CSNI/R(2009)5

References

[1] D.A. Powers, J.L. Sprung A Simplified Model of Aerosol Scrubbing by a Water Pool Overlying

Core Debris Interacting with Concrete NUREG/CR-5901, SAND92-1422, 1993

[2] R.M. Bilbao, J.M. Fernández-Salgado, J.A. Fernández-Benítez, A. Alonso-Santos, J.V. López-

Montero Severe Accident Phenomenology in BWRs and the Role of PHÉBUS-FP. EUR 15297 EN,

1993

[3] K. Fischer Modelling of Retention Processes in Water Pools Battelle Ingenieurtechnik, December 1998

[4] V. Peyrés, L.E. Herranz Programa de Investigación SAC-PS: Influencia de sustancias

Contaminantes en la Retención de Aerosoles en Lechos Acuosos. DFN/SN-06/SP-00, 2000

3.7 Re-Entrainment

This chapter describes the re-entrainment process in general, its occurrence in reactor accidents, and the

status of analytical and experimental work. Actually re-entrainment processes are modelled in accident

codes in a very incomplete way. Proposals are made to reduce this deficit.

Re-entrainment process

When gas is flowing through a pool, droplets are formed at the pool surface and transported with the rising

gas flow into the atmosphere. Together with the droplets soluble and insoluble fission products are released

in aerosol form to the atmosphere. This process is called re-entrainment or wet resuspension because the

released material comes in general from aerosols deposited earlier.

Droplet generation depends essentially on the superficial velocity of the gas flow through the pool. At low

superficial velocities the gas flow consists of single bubbles rising through the pool (bubbly flow) and

bursting at the surface. Each bursting bubble produces several hundred small so-called film droplets and a

small number of larger so-called jet droplets.

At high superficial velocities bubbles of different sizes and shapes rise in a turbulent way (churn turbulent

flow). Droplet formation results from a direct momentum exchange between gas flow and liquid. Hollow

columns of liquid, whose upper part rapidly disintegrates into droplets due to surface tension, erupt at the

pool surface.

Under the accident conditions in a boiling reactor sump the transition from bubbly to churn turbulent flow

occurs at a superficial gas velocity of some cm/s. At this transition the droplet formation process changes

fundamentally. Therefore different models are required for each regime.

The re-entrainment

E is the ratio of the liquid (droplet) mass flux

to the gas mass flux

rising

through the pool surface. It can also be defined in terms of the density

and the superficial velocity

for

the liquid and the gaseous phases:

⋅

(Eq. 1)

NEA/CSNI/R(2009)5

For bubbly and turbulent flows the re-entrainment increases with an increasing gas velocity. Additionally a

strong dependence of

E on the bubble size exists for the bubbly flow regime. The re-entrainment is much

larger with small gas bubbles than with large ones.

Re-entrainment processes in reactor accidents

In a reactor accident re-entrainment processes can occur in the circuit and the containment at different

pools and in different accident phases (Table 3.7-1).

Fission product (FP) re-entrainment takes place on boiling sumps in PWR as well as BWR core melt

scenarios. Sumps can boil in case of an extreme decay heat release, or at contact with core melt. Another

case is sump flashing during rapid containment depressurization as consequence of a containment leak.

With controlled venting the pressure transient is smaller and re-entrainment will be reduced. Venting filters

may be loaded additionally by the released aerosol and its decay heat.

Moreover, all pool scrubbing processes, e.g. in a BWR suppression system, are accompanied by

re-entrainment phenomena. On the one hand aerosol particles are efficiently scrubbed by the pool water

but on the other hand a fraction of the aerosol material retained is re-entrained. Similar re-entrainment

processes occur in the suppression systems of the Russian-type reactors WWER-440/213 and RBMK 1000

and 1500.

Re-entrainment also occurs in core melt pools agitated by rising gases from the molten core concrete

interactions (H

O, H

, CO

, CO, etc.). Together with the melt droplets non-volatile fission products are

released.

A discussed accident management strategy is to cool the spread melt by a water layer on top. However, the

decay heat of the melt can make the water boil which causes re-entrainment.

Table 3.7-1 Re-Entrainment processes in LWR accidents

Reactor type Design basis and degraded core

accidents

Core melt accidents

PWR • RPV during LOCA, core partly

dried out

• Boiling sump by FP heat release

• Pressuriser and pressuriser

relief tank at pressuriser leak

• Boiling sump at venting

• Steam generator pipe break • Flashing Sump at containment

leak

• Core melt agitated by gas

• Water on top of core melt pool

(EPR)

WWER-440 • Wet condenser during LOCA

BWR • Discharge into suppression

pool

• Boiling sump due to FP heat

release

• PPV during LOCA • Core melt agitated by gas

RBMK • Accident localisation system at

LOCA

NEA/CSNI/R(2009)5

The aerosol and fission product release rates by re-entrainment are relatively small compared to those of

other aerosol generation mechanisms, e.g. by evaporation/condensation processes during the core melt

down. Nevertheless re-entrainment can contribute significantly to the radioactive source term in the late

accident phase because the process is long lasting. At that time aerosols from other sources have already

been depleted to a large extent.

Modelling

In all accident codes re-entrainment is modelled in a very incomplete way as summarised in Table 3.7-2.

Very simple approaches which cannot describe this complex process sufficiently, e.g. constant re-

entrainment rates, are not considered in the table.

Table 3.7-2 Re-entrainment processes modelled in accident codes

Accident codes Re-entrainment model for

Boiling or

flashing sump

Suppression pool

with pool

scrubbing

Agitated core

melt pool

Containment

codes

COCOSYS no yes no

CONTAIN no no yes

Integral

codes

ASTEC no yes no

MAAP for iodine only no no

MELCOR no no yes

Re-entrainment from boiling or flashing sumps, the most important process in source term analyses, is

modelled only in MAAP and for iodine species only. In MAAP the re-entrainment of iodine droplets (CsI,

, HOI) for bubbly flows is calculated by the oscillation model of Azbel [1]. For churn turbulent flows

the semi-mechanistic correlations of Rozen and Kataoka-Ishii are applied [2]. The Rozen correlation is

used with low superficial velocities. It delivers a higher re-entrainment than the correlation of

Kataoka-Ishii [3].

In COCOSYS [4] and ASTEC [5] as well as in the circuit code ATHLET-CD re-entrainment is only

treated in connexion with pool scrubbing. Re-entrainment is calculated by the module RECOM [6] coupled

to the pool scrubbing model SPARC [7]. RECOM uses an oscillation model with critical lamina thickness

based on Azbel. For the jet droplet re-entrainment, results of the highly mechanistic code RESUS are taken

[6]. RESUS solves the Navier-Stokes equations numerically by use of a finite difference approach. The

calculated number and size of the jet droplet is used in RECOM. In the churn turbulent flow regime the

re-entrainment is calculated by Kataoka-Ishii. For film droplets and the droplets generated under churn

turbulent flow conditions the droplet size distribution has to be given in the input.

In CONTAIN re-entrainment is modelled only for core melt pools [8]. In the module CORCON/VANESA

the Azbel model is used for bubbly flow regime and the Kataoka-Ishii correlation for the churn turbulent

flow regime [9]. The same applies to MELCOR [10] which also uses CORCON.

With all accident codes the re-entrainment from water as well as core melt pools is calculated by the

oscillation model of Azbel for bubbly flows and by the correlation of Kataoka-Ishii for churn turbulent

NEA/CSNI/R(2009)5

flows. Only in MAAP the correlation of Rozen for low velocities in the churn turbulent flow regime is

applied additionally [11]. Only in the module RECOM used in COCOSYS, ASTEC and ATHLET-CD the

jet droplet formation is explicitly treated. Some of the correlations have been modified compared to the

original versions.

Especially the re-entrainment models for boiling water pools need several improvements concerning the

following items:

−

Film droplet formation and dependence on bubble size distribution

−

Enrichment of insoluble material in re-entrained droplets

−

Re-entrainment spike at onset of boiling

−

Size distribution of re-entrained droplets

Conclusions

During a severe accident in a PWR or a BWR re-entrainment of fission products can occur at several water

and core melt pools. The release rates are relatively small but the sources are long lasting. Thus a

significant contribution to the source term by re-entrainment is likely in the late accident phase. However,

thereto no recent analytical investigations are available. One reason is the very incomplete modelling of

re-entrainment processes in accident codes.

Out of five codes only one is able to simulate the FP release from boiling or flashing sumps. In general

re-entrainment is modelled in the bubbly flow regime based on the oscillation model of Azbel and in the

turbulent flow regime by the semi-mechanistic correlation of Kataoka-Ishii. The largest uncertainties come

from the film droplet generation model and especially from the uncertain assumptions for the bubble size

distribution. Only in RECOM the jet droplet generation is explicitly considered. Only in a few experiments

(REST, REVENT, and ThAI re-entrainment tests) re-entrainment has been investigated under LWR

accident typical conditions.

References

[1] D. Azbel and A.L. Liapis Mechanisms of Liquid Entrainment Handbook of Fluids in Motion Ann

Abbor Science, Michigan (1983)

[2] I. Kataoka and M. Ishii Mechanistic Modelling of Pool Entrainment Phenomenon Int. J. Heat Mass

Tranfer, Vol. 27, No. 11, pp 1999 – 2014 (1984)

[3] T. Ginsberg Aerosol Generation by Liquid Breakup Resulting from Sparging of Molten Pools of

Corium by Gases Released During Core/Concrete Interactions Nucl. Sci. and Eng.: 89, 36 – 48

(1985)

[4] W. Klein-Heßling, S. Arndt, G. Weber COCOSYS V 1.2 User Manual GRS-P-3/1 (July 2000)

[5] F. Jacq, H.-J. Allelein ASTEC V0. 2 General Overview, Rev. 0 ASTEC-V0 /DOC/99-09 (April

2000)

[6] M.K. Koch et al. Radionuclide Re-entrainment at Bubbling Water Pool Surfaces J. Aerosol Sci.

Vol. 31, No. 9, pp 1015 – 1028 (2000)

NEA/CSNI/R(2009)5

[7] P.C. Owczarski, K.W. Burk SPARC-90: A Code for Calculating Fission Product Capture in

Suppression Pools NUREG/CR – 5765 (1991)

[8] K.K. Murata et al. Code Manual for CONTAIN 2.0: A Computer Code for Nuclear Reactor

Containment Analysis Sandia National Laboratories NUREG/CR-6533 (1997)

[9] D.R. Bradley et al. CORCON-MOD3: An Integrated Computer Model for Analysis of Molten Core-

Concrete Interactions NUREG/CR-5843 (1993)

[10] R.O. Gauntt et al. MELCOR Computer Code Manuals Vol. 1: Users' Guide for Version 1.8.5

NUREG/CR-6119 Rev.2 (2000)

[11] D.L.Y. Louie and D.J. Osetek MAAP-IMPAIR Interface for Analysis of Iodine Behaviour in

Advanced Reactor Accidents (Phase II) DOE/ID-10466 (March 1944)

3.8 Aerosol Formation during High Pressure Melt Expulsion from the Reactor Coolant System

The accident at Three Mile Island demonstrated that substantial core degradation and core debris relocation

to the lower head of the pressure vessel could occur while the reactor coolant system remained at elevated

pressure. Prior reactor accident analyses had assumed that depressurization would occur early in the course

of an accident, perhaps as part of the accident initiating event. The Three Mile Island accident raised the

possibility that molten core debris could penetrate a pressurised reactor vessel and be dispersed throughout

the containment.

Safety concerns associated with the pressurised expulsion of core debris from the reactor vessel focused on

issues of containment integrity. Initial concerns were that dispersed debris could heat the atmosphere to the

point containments failed by overpressure. This is the so-called “direct containment heating” scenario.

Further investigations suggested melt expulsion and dispersal might trigger hydrogen combustion events

that threatened containment integrity. These concerns prompted several experimental investigations. A

cross section of the multinational experimental efforts is reported in a CSNI state of the art report [1].

Following preparation of this state of the art report, several scaled tests were undertaken to support

analyses of the containment threats [2]. To a significant extent the concerns over prompt containment

failure associated with pressurised expulsion of core debris have been resolved favorably for large dry and

subatmospheric containments [3].

Though investigations of high pressure melt expulsion and direct containment heating were focused on the

containment loads, there were some early efforts to examine aerosol production associated with the

expulsion and dispersion processes [4, 5]. A sequence of frames from an open air test of high pressure melt

expulsion is shown in Fig. 3.8-1.

It is apparent that a very large amount of aerosol is produced. Efforts were made to identify aerosol

formation mechanisms and to characterise the aerosol produced in tests contained within the Surtsey test

facility (a 110 m

containment vessel). As might well be expected, all of the tests were done with simulant

materials rather than real core debris. Usually the simulants were mixtures of molten iron and aluminum

oxide produced by a thermitic reaction within a vessel pressurised with nitrogen or steam. Consequently, it

is not readily apparent how applicable the aerosol characterization results are to the analysis of reactor

accidents. It is thought that the aerosol formation mechanisms are, at least, qualitatively applicable.

NEA/CSNI/R(2009)5

Fig. 3.8-1 Open air test of melt expulsion from a vessel pressurised to 41.3 bar with nitrogen

Three mechanisms of aerosol formation were identified:

•

Vaporization from surfaces of high temperature melts,

•

Droplet formation by the effervescence of gas from the melt as it emerges from the pressurised

vessel, and

•

Comminution of melt droplets as the impact surfaces within containment models.

The three mechanisms produce characteristically different aerosol particles. Examples of these particles

are shown in Fig. 3.8-2. Vapors evolving from the high temperature melt rapidly transport to cooler

environments where they nucleate extremely fine primary particles. More detailed views of these fine

aerosol particles are provided in Fig. 3.8-3 and Fig. 3.8-4. It is suspected that the cooling of the vapor is so

fast and supersaturations so high that homogeneous nucleation is a predominant path for vapor

condensation rather than condensation on either structural surfaces or surfaces of mechanically produced

particles. There also appears to be a tendency for the primary particles to coagulate in chains and these

chains either fold or agglomerate to form very low density particles.

The second mechanism of aerosol production arises because gases including steam and hydrogen are

soluble at high pressure in molten materials. When the melt emerges from the pressure vessel, the

solubility drops dramatically and bubbles nucleate in the melt. Frid demonstrated using x-rays to monitor

the melt as it emerged from the pressure vessel that gas effervescence was sufficient to completely disrupt

the stream of melt emerging from the pressure vessel. As each bubble bursts at the surface of the melt it

NEA/CSNI/R(2009)5

cases off droplets in the size range of a few microns. These droplets are small enough that they remain

airborne until they have cooled to solidify as very nearly spherical particles.

Fig. 3.8-2 Examples of the qualitatively different types of aerosol produce in a high pressure melt

expulsion event

Fig. 3.8-3 Electron micrograph of very fine aeosols produced by vapor nucleation during a high

pressure melt expulsion experiment

NEA/CSNI/R(2009)5

Fig. 3.8-4 Electron micrograph of fine aerosols produced by vapor nucleation during a high

pressure melt expulsion experiment. Note the apparent tendency to form chain

agglomerates

Larger particles (10-100 µm) are produced by a combination of aspiration and comminution. As melt is

expelled from a pressure vessel, the hole in the pressure vessel enlarges. Part way through the expulsion

process, the hole is enlarged to the point that the pressurizing gas penetrates the liquid pool and two phase

expulsion takes place. The high velocity gas entrains melt droplets. Furthermore, gas also entrains melt

previously expelled from the vessel. Droplets accelerated in the gas phase disintegrate until a critical

Weber number of about 12 is reached. Further distintegration of the melt droplets can occur as they impact

surfaces at very high velocities.

About 1 % of the melt mass expelled from a pressurised vessel is thought to be converted to aerosol. This

can produce a substantial aerosol loading of the containment atmosphere. But, if the aerosol production is

not coincident with containment failure, the aerosol rapidly agglomerates and sediments so it may not

contribute significantly to the risk associated with reactor accidents.

Following resolution of concerns about the containment loads produced by pressurised melt ejection, little

effort has been devoted to further characterizing the associated aerosol production. Interest has been further

attenuated as accident management measures have been introduced to depressurise reactor coolant systems

during accidents involving core degradation. Most accident analysis codes do not include models of

aerosol formation associated with core debris expulsion from the reactor coolant system.

References

[1] Fauske and Associates and Sandia National Laboratories, State-of-the-Art Report (SOAR) on HPME

and DCH, OECD, 1996

NEA/CSNI/R(2009)5

100

[2] T.K. Blanchat, M.M. Pilch, M.D. Allen, Experiments to Investigate Direct Containment Heating

Phenomena with Scaled Models of the Calvert Cliffs Nuclear Power Plant, NUREG/CR-6469,

Sandia National Laboratories, Albuquerque, NM, January 1997, and references therein

[3] M.M. Pilch, M.D. Allen, B.W. Spencer, K.D. Bergeron, K.S. Quick, D.L. Knudson, E.L. Tadios,

and D.W. Stamps, The Probability of Containment Failure by Direct Containment Heating in Surry,

NUREG/CR-6109, Sandia National Laobratories, Argonne National Laboratory, Idaho National

Engineering Laboratory, May 1995, and references therein

[4] J.E. Brockmann and W.W. Tarbell, Nuclear Science and Engineering, 88 (1984)342

[5] J.E. Brockmann, Progress in Nuclear Energy, 19 (1987) 478

3.9 Aerosols from MCCI

An important source of aerosol input to reactor containment during an unarrested severe reactor accident

comes from the interactions of core debris with structural concrete. Classic reactor accident sequences

envisage a point at which core debris will penetrate the reactor pressure vessel and fall into the reactor

cavity (see Fig. 3.9-1). The core debris will be a mixture of molten uranium dioxide fuel, zirconium

dioxide and zirconium metal from the fuel cladding, structural steel and fission products. Much of the core

debris will be molten when it penetrates the reactor pressure vessel. If the core debris is not quenched as it

emerges from the vessel, it will thermally attack the structural concrete in the reactor cavity. If the core

debris is quenched but the quenched debris is not coolable, core debris will reheat and eventually become

hot enough to thermally attack the concrete.

Fig. 3.9-1 Photograph of a high temperature melt interacting with calcareous concrete and the

associated aerosol production

NEA/CSNI/R(2009)5

101

Aerosol production comes about during core debris interactions with concrete because the core debris is

sufficiently hot to produce significant vapor pressures of some core debris constituents and the constituents

of ablated and melted concrete. Aerosol production can become significant (see Fig. 3.9-2.) because the

gaseous products of concrete decomposition sparge through the core debris. Three mechanisms are usually

considered in the analysis of aerosol production during core debris interactions with concrete [1]:

•

vaporization from the free, upper surface of core debris,

•

vaporization into gas bubbles sparging through molten core debris, and

•

mechanical aerosol production when bubbles burst at the surface of molten core debris.

Time after start of core debris/concrete interactions (min)

0 50 100 150 200 250 300

Aerosol Source Rate (g/s)

200

400

600

800

1000

1200

REACTOR TYPE concrete

BWR calcareous

PWR calcareous

PWR siliceous

Fig. 3.9-2 Comparison of predicted aerosol production rates during core debris interactions with

concrete at different reactors with various concretes

The first of these mechanisms is usually found to be small in comparison to the second if the core debris is

molten.

Concretes used in the structural elements of nuclear power plants are typically classed as either calcareous

concretes that use limestone (CaCO

) or dolomite (CaMg(CO

)

) as aggregate or siliceous concretes that

use basalt, granite or other silica-rich materials largely free of carbonates as aggregate. Chemical

compositions of three representative types of concrete found in nuclear power plants [2] are shown in

Table 3.9-1. Clearly, the distinction between calcareous and siliceous concretes is the large amounts of

carbon dioxide that can be liberated in the thermal decomposition of calcareous concretes. The other

gaseous product of concrete decomposition is water vapor. All concretes have a significant amount of

water vapor. About half the water (often called 'free water') is present in the concrete as water in the pore

structure of concrete and water of crystallization. The rest of the water (bound water) in the concrete is in

the form of calcium hydroxide (Ca(OH)

) which is the binding agent that holds together the aggregate. At

sufficiently high temperature, this calcium hydroxide decomposes to produce calcium oxide and water

NEA/CSNI/R(2009)5

102

vapor. Temperature ranges over which the various gaseous concrete decomposition products are released

are [3]:

•

Free water: ambient to 230 °C

•

Bound water: 300 to 450 °C

•

Carbon dioxide: 600 to 1100 °C

Table 3.9-1 Chemical compositions of some representative concretes at nuclear power plants

Chemical

Constituent

Mildly Siliceous

Concrete*

(weight %)

Calcareous

Concrete

(weight %)

Limestone/

Common Sand Concrete

(weight %)

SiO

54.7 3.6 35.7

CaO 8.8 45.4 31.2

8.3 1.6 3.6

MgO 6.2 5.67 0.5

6.2 1.2 1.4

5.4 0.7 1.2

TiO

1.0 0.15 0.2

1.8 0.08 0.8

MnO 0.4 0.01 0.03

0.2 0.004 0.014

5.0 4.1 4.8

1.5 35.7 22.0

* Siliceous aggregates are quite variable in composition and may contain much higher concentrations

of silicon dioxide than basaltic aggregate.

Siliceous concretes melt at temperatures somewhat above 1100 °C and calcareous concretes melt at

temperatures from 1200 to 1600 °C. During thermal attack on concrete, there are fronts of gas generation

propagating through the concrete ahead of the ablating interface between concrete and the core debris.

The gases liberated from the concrete are observed to pass relatively easily through substantial heads of

molten core debris. The oxidizing gases from the concrete, CO

and H

O, react with the metallic

constituents of core debris, predominantly Zr and Fe, to form CO and H

. The very exothermic reaction of

the oxidizing gases with zirconium metal in the core debris can be a significant source of heat that sustains

Note that in addition to these chemical constituents of concrete, structural concrete at nuclear power plants contains about

0.25 g/cm

reinforcing steel which is mostly iron

NEA/CSNI/R(2009)5

103

high temperatures for protracted periods in the core debris. Such high temperatures, as well as the reactive

nature of the gases mean that constituents of the melt will vaporise into the bubbles of gas sparging

through the liquid. When these vapor-ladden bubbles escape the core debris, the vapors are released into a

cooler environment and they condense to form aerosol particles.

Gases sparging through the core debris at sufficiently high velocities can entrain molten material in the

form of droplets [4] that have dimensions small enough to be considered aerosol (< 20 µm). At lower gas

generation rates, gases sparging through the core debris emerge as bubbles that burst at the surface. The

bursting of these bubbles imparts very high accelerations to the liquid films (~10.000 g). Such high

accelerations can launch small liquid droplets into the gas flow as drinkers of carbonated beverages well

know.

Aerosol production during core debris interactions with concrete, then, depends on the composition of the

core debris, especially the zirconium metal content, and the amount of gas that can be produced by the

thermal decomposition of the concrete. Comparisons of the aerosol production rates predicted to occur in

reactor accidents of different types are shown in Fig. 3.9-3. The highest aerosol production rates occur

soon after core debris accumulates in the reactor cavity. The greatest production of aerosol occurs in the

case of the boiling water reactor with calcarous concrete. Aerosol production is extremely high in this type

of accident because of the large inventory of zirconium metal in a boiling water reactor much of which is

not oxidised at the time core debris penetrates the reactor vessel.

fp fuel+structural concrete

1000

2000

3000

4000

5000

fission products

fuel & structures

concrete

Fig. 3.9-3 Sources of aerosolised materials during core debris interactions with concrete [in kg]

NEA/CSNI/R(2009)5

104

Without this large inventory of zirconium metal in the core debris, aerosol production predicted for an

accident at a pressurised water reactor with calcareous concrete is much attenuated though still significant.

Aerosol production when pressurised water reactor core debris interacts with siliceous concrete is smaller

still.

Aerosol production is persistent during core debris interactions with concrete. Even after the core debris

has solidified, temperatures are sufficiently high that concrete decompositions continues and the gaseous

products of decomposition sweep vapors of the more volatile species into cooler environments where they

condense to form aerosol particles.

Much of the aerosol produced during core debris interactions with concrete is nonradioactive material.

Most comes from constituents of concrete and structural materials as is shown by the comparisons in Fig.

3.9-3 for an accident involving boiling water reactor core debris interacting with calcareous concrete.

Nevertheless, significant releases of radionuclides can occur during core debris interactions with concrete.

Release fractions for various classes of radionuclides derived from tests and model predictions of core

debris interactions with concrete and used for regulatory analyses in the USA [6] are shown in Table 3.9-2.

But recent investigations lead to the assumption that the release of the volatile radionuclides (halogens,

alkali metals and to a less extend Te) are not coming from the MCCI. They are in-vessel release from fuel

remaining in the RPV, but being released during the ex-vessel phase of the accident.

Table 3.9-2 Radionuclide release fractions associated with core debris concrete interactions at

BWRs and PWRs

Boiling Water Reactor Pressurised Water Reactor

Radionuclide Release

Duration (hr)

3.0 2.0

Fraction of Initial Core Inventory of Radionuclide Released During Core

Debris/Concrete Interactions

Halogens (I, Br) 0.30 0.25

Alkali Metals (Cs, K) 0.35 0.35

Tellurium 0.25 0.25

Ba, Sr 0.10 0.10

Ru, Mo, Pd 0.0025 0.0025

Ce 0.005 0.005

La 0.005 0.005

Releases of nonradioactive isotopes of important fission product elements observed in the ACE tests [5]

are shown in Fig. 3.9-4. Only the relative releases can be compared to results of accident analyses since

total release is quite dependent on details of melt composition and the accident scenario. Surprising

features of the tests are the measured releases of Mo and Ru which are not predicted by the computer

models of fission product release during core debris interactions with concrete. This has been attributed to

improper modeling of the partitioning of fission products between the oxide and metal phases of core

debris interacting with concrete.

NEA/CSNI/R(2009)5

105

Relative Release Fraction

1e-5

1e-4

1e-3

1e-2

1e-1

1e+0

Fig. 3.9-4 Relative release fraction observed in the ACE L6 test

Some attempts have been made to characterise the particle size distribution of aerosols evolved during core

debris interactions with concrete [7]. Much of this effort has been directed at determining the fraction of

aerosol formed by entrainment and the fraction formed by the condensation of vapors. In general, aerosols

produced by entrainment are expected to be larger than those produced by vapor condensation.

Furthermore, the aerosol particles produced by entrainment should have compositions much like the bulk

melt whereas aerosol particles produced by vapor condensation should have compositions reflecting the

relative vapor pressures of the melt constituents. It is, however, difficult to get reproducible data

immediately above the melt where the environment is very dynamic and farther from the site of melt

interactions with concrete, coagulation of the aerosol obscures differences in aerosol produced by

mechanical and vaporization processes.

NEA/CSNI/R(2009)5

106

Aerodynamic Particle Diameter, D

, (µm)

0.1 1 10 100

Mass-weighted Probability Density

Fig. 3.9-5 Typical aerosol size distribution derived from impactor data taken during core debris

interactions with concrete

A typical mass-weighted aerodynamic particle size probability density function produced from data

collected by aerosol impactors during melt interactions with concrete is shown in Fig. 3.9-5. Aerodynamic

mass mean particle sizes are typical around 1 µm. The geometric standard deviation values are variable

exceeding 3 when interactions are very vigorous. This breadth of the aerosol particle size distribution when

melt/ concrete interactions are vigorous is attributed to multiple mechanisms of aerosol production being

operative. As interaction vigor subsides, aerosol size distributions are typically found to be narrower with

geometric standard deviations sometimes less than 2. There is some evidence of compositional variability

across the size distribution, but it is not especially pronounced.

A common accident management strategy is to provide water to cover any core debris that escapes to the

reactor cavity. Though it has not been demonstrated that the water will quench core debris in a reactor

cavity and prevent interactions with concrete, it has been shown that a water pool overlying core debris can

attenuate aerosol production rather significantly. Attenuation of the aerosol production during core debris

interactions with an overlying water pool is thought to occur by two processes. The water pool will

typically cause a crust to form over the core debris that interferes with or filters aerosol produced by the

entrainment of molten core debris in the gas flow and aerosol production by bubble bursting. Vapors can

also condense as they pass through the crust. The water body will also scrub aerosol and vapors from

bubbles emerging from the core debris and sparging through the water much as does a steam suppression

NEA/CSNI/R(2009)5

107

pool. That is, aerosols are removed from bubbles by a combination of diffusion, inertial impaction and

sedimentation. As little as 30 cm deep pools of saturated water have been observed to reduce aerosol

production during tests of core debris interactions with concrete by a factor of up to 15 [8].

Models of aerosol production during core debris interactions with concrete

Most of the important accident analysis codes have some modeling of aerosol production during core

debris interactions with concrete. The most thoroughly documented of these models is the VANESA model

[1] that is now part of the MELCOR computer code [9]. The models of aerosol production have to be

coupled with a model of core debris attack on concrete such as the CORCON model [10] in MELCOR.

The models of core debris attack on concrete provide melt temperatures, melt compositions, gas production

rates and geometrical data for the aerosol production models. All of the models estimate the vapor

pressures of important melt constituents and the mass transport of these vapors into bubbles of the gas

sparging through the molten core debris. Correlations are based largely on data obtained for bubbles

sparging through water or similarly conventional liquid [4]. Some of the models account for the

mechanical production of aerosol by entrainment or bubble bursting. Entrainment modeling is usually

based on data obtained for gas sparging through water. Typically, the detailed behaviour of vapors released

from the core debris via the sparging gases is not modeled. It is more typical for the vapor mass to be

assumed to nucleate particles that populate the smallest bin in an sectional representation of the aerosol

size distribution in the reactor containment atmosphere. The subsequent coagulation and deposition of

these aerosols along with aerosols from other sources are modeled using the aerosol physics models in the

accident analysis codes.

References

[1] D.A. Powers, J.E. Brockmann, and A.W. Shiver, VANESA: A Mechanistic Model of Radionuclide

Release and Aerosol Generation During Core Debris Interactions With Concrete, NUREG/CR-4308,

Sandia National Laboratories, Albuquerque, NM, July 1986

[2] D.A. Powers and F.E. Arellano, Large-scale Transient Tests of the Interaction of Molten Steel with

Concrete, NUREG/CR-2282, Sandia National Laboratories, Albuquerque, NM, January 1982

[3] D. A. Powers, "Empirical Models for the Thermal Decomposition of Concrete", Transactions of the

American Nuclear Society

26, (1977) 401

[4] I. Kataoka and M. Ishii, International Journal of Heat and Mass Transfer,

27 (1984) 1999

[5] J.K. Fink, D.H. Tompson, D.R. Armstrong, B.W. Spencer, and B.R. Sehgal, High Temperature and

Materials Science,

33 (1995) 51

[6] L. Soffer, S.B. Burson, C.M. Ferrell, R.Y. Lee, J.N. Ridgely, Accident Source Terms for

Light-Water Nuclear Power Plants, NUREG-1465, U.S. Nuclear Regulatory Commission,

Washington D.C., February 1995

[7] J.E. Brockmann, Progress Nuclear Energy,

19 (1987) 7

[8] R.E. Blose, J. Gronager, A. Suo-Antilla, D. Bradley, and J.E. Brockmann, SWISS: Sustain Heated

Metallic Melt Concrete Interaction with an Overlying Water Pool: Experiments and Analyses,

SAND85-1546, Sandia National Laboratories, Albuquerque, NM, 1985

NEA/CSNI/R(2009)5

108

[9] R.O. Gauntt, R.K. Cole, C.M. Erickson, R.G. Gido, R.D. Gasser, S.B. Rodriguez, and M.F. Young,

MELCOR Computer Code Manuals, NUREG/CR-6119, Rev.2, Sandia National Laboratories,

Albuquerque, NM, December 2000

[10] D.R. Bradley and D.R. Gardner, CORCON-Mod3: An Integrated Computer Model for Analysis of

Molten Core Concrete Interactions: Users Manual, NUREG/CR-5843, Sandia National Laboratories,

Albuquerque, NM, 1992

3.10 Aerosols from Fire

Fire is a hazard that can afflict all nuclear power plants. On average, a plant will have a reportable fire

about once every two years. In most cases, these fires are readily suppressed and extinguished using the

fire protection features of the plant. Occassionally fires of such magnitude that they threaten the safety of

the plant. Of course, the most famous of the threatening fires was that at the Browns Ferry nuclear station

[1] in 1975. Many enhancements to fire protection were made at nuclear plants following the incident at

Browns Ferry. Still, significant fires occur [2]. Risks of fires leading to core damage have been assessed by

all operating nuclear plants in the USA [3]. As shown in Fig. 3.10-1, the risk of fire leading to core damage

is typically commensurate with the risk that other, internal plant events more familiar in accident analyses

will lead to core damage. The risk analyses have shown that the major types of fires at nuclear power

plants include combustion of hydrogen used to cool turbines, lubricating oil fires, and cable fires. Of these,

cable fires are of the most safety concern. The cables of interest include power cables, control cables and

instrumentation cables. Combustion of the insulation on these cables can lead to electrical shorts that can

cause spurious operation of equipment, loss of safety functions and misleading control room information.

Cable fires can be the source of an enormous amount of aerosol. A photograph of smoke coming from a

cable fire at a non-nuclear power plant is shown in Fig. 3.10-2. Fire and smoke from the burning of a

transformer at a nuclear power plant substation is shown in Fig. 3.10-3. The smoke from cable fires is

known to be quite corrosive and often contains substantial hydrochloric acid. There is a concern, then, that

smoke produced in any fire, but especially smoke from cable fires, could damage equipment in plants.

Because of aerosol transport, the equipment damage by deposited, corrosive aerosol particles could occur

at locations well removed from the site of the fire. The effects of smoke from burning cable insulation on

digital circuits that might be used in nuclear power plants have been investigated [4]. Their investigations

identified types of circuits that were susceptible to smoke damage. There is, then, safety reasons to want to

understand the behaviour and transport of aerosol produced by cable fires aside from interest in the

interaction of fire aerosol with radioactive aerosol from a core degradation accident.

NEA/CSNI/R(2009)5

109

Internal Events Core Damage Frequency (yr

-1

)

1e-6 1e-5 1e-4 1e-3

Frequency of Fire-induced Core Damage (yr

-1

)

1e-6

1e-5

1e-4

1e-3

PWR

BWR

Fig. 3.10-1 Frequency of core damage caused by fire plotted against the expected frequency of core

damage caused by all other credible events during plant operations for US nuclear power

plants. Points are shown for US boiling water reactors (BWR) and US pressurised water

reactors (PWR). Points above the diagonal black line indicate plants with greater risk from

fire than from other internal initiators of plant damage. In all cases, the fire risk is

commensurate with risk from other internal events.

Fig. 3.10-2 Smoke from a cable fire at a non-nuclear power plant

NEA/CSNI/R(2009)5

110

Fig. 3.10-3

Fire and smoke produced during a substation fire

Whereas much is known about smoke and soot aerosols produced in hydrocarbon fires, relatively little is

known about the aerosol generated when cable insulation burns. The extinction coefficient of aerosol was

measured in smoke tests with digital equipment [4]. Also the relative amounts of aerosol deposition on

horizontal and vertical surfaces in a 92x92x123 cm test chamber were measured. Examples of these results

are shown in Fig. 3.10-4 and Fig. 3.10-5. Otherwise, no quantitative information has been found about the

characteristics of cable fire aerosols such as size distribution, shape factors and production rates. There are

plans in ongoing cable fire tests being sponsored by the US Nuclear Regulatory Commission to collect

some limited amounts of information about cable fire aerosol.

Optical Turbidity

Time (hour)

0.00 0.25 0.50 0.75 1.00

Extinction Coefficient (cm

-1

)

0.0

0.1

0.2

0.3

0.4

0.5

0.6

1.060 µm light

0.4579 µm light

Fig. 3.10-4 Optical extinction coeffiient of aerosol-laden gas phase during test of digital equipment

in a cable fire. Extinction coefficient was measured at two different wave lengths each

over path lengths of 1.9 and 3.2 cm to correct for the deposition of aerosols on lens

surface

NEA/CSNI/R(2009)5

111

Aerosol Deposition on Surfaces

Time (hour)

0.00 0.25 0.50 0.75 1.00

Mass Deposited (µg/cm

)

Vertical Surfaces

Horizontal Surfaces

Fig. 3.10-5 Deposition of cable fire aerosols on vertical and horizontal surfaces during test of digital

equipment

References

[1] R.L. Scott “Browns Ferry Nuclear Power Plant Fire on March 22, 1975” Nuclear News, 17 (1976) 592

[2] S.P. Nowlen, M. Dazarians, F. Wyant Risk Methods Insights Gained From Fire Incidents,

NUREG/CR-6738, Sandia National Laboratories, Albuquerque, NM, September 2001

[3] Division of Risk Analysis and Application, Office of Nuclear Regulatory Research, Perspectives

Gained From the Individual Plant Examination of External Events (IPEEE) Program, NUREG-1742,

Volumes 1 and 2, U.S. Nuclear Regulatory Commission, Washington, DC, April 2001

[4] T.J. Tanaka, S.P. Nowlen, and D.J. Anderson, Circuit Bridging of Components by Smoke,

NUREG/CR-6476, Sandia National Laboratories, Albuquerque, NM, October 1996, and references

therein

3.11 Spray Systems

The reactor containment buildings of most pressurised water reactors and the drywells of most boiling

water reactors are equipped with water spray systems. These spray systems were installed to condense

steam and reduce the pressure threat to containment or drywell integrity in the event of a design-basis,

large break in the reactor coolant system. Sprays are also very effective at removing aerosol particles from

the containment or drywell atmospheres during severe reactor accidents [1, 2].

Spray systems consist of a large number of spray nozzles oriented differently near the top of the

containment or drywell. These nozzles discharge large numbers of water droplets that fall along ballistic

trajectories through the atmosphere and sweep out aerosol particles. Some characteristics of two spray

nozzles commonly used in US reactors are shown in Fig. 3.11-1 and Fig. 3.11-2. The Whirljet spray nozzle

NEA/CSNI/R(2009)5

112

discharges water through multiple orifices over a circular pattern when directed downward so that at a

distance of 3.35 m from the nozzle, 65 % of the water flow is relative uniformly dispersed over a region

3.35 m in diameter and the remaining 35 % of the flow is over an annular region with inner diameter 3.35

m and outer diameter 5.2 m. The Sprayco Model 1713-A nozzle discharges water droplets through a single

nozzle. The diameter of the sprayed region is 2.58 m at 3.05 m from the nozzle, 3.60 m at 4.57 m from the

nozzle and 5.8 m at a distance of 30.5 m from the nozzle. The water distribution over the sprayed region is

not entirely uniform even at large distances from the nozzle as shown in Fig. 3.11-3. To achieve more

uniform droplet distribution over the sprayed volume of a containment or drywell, spray patterns are

overlapped.

Spray droplets remove aerosol particles from the containment or drywell atmospheres by several

mechanisms:

•

Diffusiophoresis: Steam condensing on the droplets sweeps aerosol particles into the droplet.

•

Impaction: Aerosol particles unable to follow streamlines of flow around a droplet collide with

the droplet.

•

Interception: Because of their finite dimensions, aerosol particles with centers of mass moving

along streamlines of flow contact and adhere to the droplets.

•

Diffusion: Brownian motion carries particles across streamlines of flow into contact with falling

droplets.

Fig. 3.11-1 Characteristics of a whirljet spray nozzle

NEA/CSNI/R(2009)5

113

Fig. 3.11-2 Characteristics of a sprayco model 1713-A spray nozzle

Fig. 3.11-3 Nonuniformity of spray droplet distribution from a sprayco model 1713-A spray

NEA/CSNI/R(2009)5

114

The diffusiophoresis mechanism is only important early in an accident when the atmosphere is steam rich

and aerosol concentrations are quite low. Consequently, this mechanism is not usually considered in the

analysis of the steady state effectiveness of aerosol removal by sprays.

Water droplets falling through a gas are approximately oblate ellipsoids. The droplet presents a

cross-section characterised by an effective diameter D

(e) as it falls through the gas. Were the aerosol

particles fixed, the water droplet would sweep out aerosols within the volume defined by the droplet

cross-sectional area times the length of it trajectory. Aerosol removal would be independent of particle

size. In actuality, the aerosols will try to follow the streamlines of flow around the falling droplets. Because

some particles will flow around the falling droplet, there is an inefficiency in the aerosol removal by falling

droplets relative to removal of fixed aerosol particles. The efficiency of aerosol removal by impaction,

interception and diffusion is expressed as the ratio of the number of particles actually removed from the

atmosphere by a particular mechanism to the number of fixed particles that would be removed by a droplet

along the same trajectory.

()

(

)

() ()

LineD

iN4

=∈

∗

where:

()

i∈ = removal efficiency of particles of size class i

N*(i) = actual number of particles of size class i removed

()

= effective droplet diameter

()

in = concentration of particles of size class i in the atmosphere

L = pathlength of droplet trajectory

Because the hydrodynamics of flow around a droplet, especially in a field of other droplets, can be

complicated, there are different descriptions of the particle collection efficiency especially for impaction

and interception mechanisms. Some commonly used expressions are [3]:

•

Impaction

()

(

)

(

)

60/Re1

60/potentialReviscous

impaction

∈

=∈

where:

()

impaction∈ = efficiency of particle removal by impaction

NEA/CSNI/R(2009)5

115

()











<<−













δ+

≥













δ+

≤

=∈

2.0Stk08334.0for08334.0Stk

Stk

57.8

2.0Stkfor

Stk

08334.0Stkfor0

potential

()























−

≤

=∈

214.1Stkfor

214.1Stk

Stk2n175.0

214.1Stkfor0

viscous

Stk = Stokes number =

()

χµ

eD9

UdC

(

)

dgT

eDU

numberynoldsRedropletsRe

= aerosol particle diameter

(e) = effective droplet diameter

= terminal velocity of the falling droplet

δ = uncertain constant 0.25 < δ < 0.75

µ = gas viscosity

ρ = gas density

χ = dynamic shape factor for aerosol particle

•

Interception

()

(

)

60/Re1

60/potentialReviscous

int

=∈

where:

()

potentialΨ = 3I

()

viscousΨ =

()













−+

5.0

5.1

1I1

I =

()

NEA/CSNI/R(2009)5

116

and γ is the collision shape factor of the aerosol particles. Gieseke and Lee [4] have discussed the

complications and approximations involved in these particle removal efficiencies for the interception

mechanism.

•

Diffusion

Many expressions have been developed for the removal of aerosol particles by water droplets by diffusion.

Among these are:

()

(

)

8/32/1

ScRe557.01

diffusion +=∈

()

3/2

18.3

diffusion =∈

()

eDPe2

diffusion

=∈

()

3.0

for

18.2

diffusion

<=∈

where:

ReScnumberPecletPe ==

TkC

numberSchmidtSc

πµ

k = Boltzmann constant

C = Cunningham slip correction factor

These expressions are based on the collection by diffusion to an isolated sphere. Of course water droplets

are neither spherical nor isolated. It is known [5] that the mass transport to a sphere is affected by the

presence of another spherical collector.

Terminal water velocities and drag coefficients on water droplets falling through air are discussed in

conventional sources [6] and include the model [7]:

[

]

DDT

Nn101912.0Nn101.1126.3Ren1 −+−=

3550Re1.010N4.2

<<<<

where:

Re = terminal Reynolds number =

(

)

gdgT

/eDU µρ

N = Best number =

(

)()

3/eDg4 µρρ

C = drag coefficient =

Re/N

NEA/CSNI/R(2009)5

117

The combined effects of impaction, interception and diffusion on the removal of aerosol are modeled either

additively:

() ( )

(

)

(

)

diffusionintimpactiontotal ∈+∈+∈=∈

or with a compound model:

()

(

)

[]

(

)

[

]

(

)

[

]

diffusion1int1impaction11total ∈−∈−∈−−=∈

The removal efficiency is highly dependent on both the particle size and the effective droplet diameter.

Diffusion is effective at the removal of very small aerosol particles (<0.1 µm) Impaction affects mostly

aerosol particles larger than about 5 µm. Interception affects particles in the size range of 0.5 to 2 µm.

Consequently, as shown in Fig. 3.11-4, there is a minimum in the total aerosol removal efficiency when

plotted against aerosol particle size. This minimum depends on the droplet diameter. For droplets on the

order of 1000 µm, this minimum is in the vicinity of 0.1 to 0.2 µm. For smaller droplets, the minimum

shifts to larger particle sizes. The effect of the minimum in the aerosol capture efficiency is that sprays not

only reduce the amount of aerosol present in the containment atmosphere, they also sharpen the size

distribution of the aerosol around the size of minimum removal efficiency. This effect is shown for an

example calculation in Fig. 3.11-5. For this example calculation, an initial aerosol with a mean diameter

near 1 µm and a broad size distribution evolves under the action of a spray to become an aerosol with a

much narrower size distribution centered round about 0.3 µm.

Fig. 3.11-4 Overall aerosol removal efficiency by a spray droplet as a function of particle size and

droplet size. Note that the aerosol particle size least efficiently removed by spray

droplets varies with the size of droplet

NEA/CSNI/R(2009)5

118

Fig. 3.11-5 Evolution of the aerosol size distribution and mass concentration during spraying.

Curves are labeled by the total decontamination achieved by the spray. Note both the

attenuation of the aerosol as well as the shift and narrowing of the size distribution

Many models of the effects of spray on aerosol in reactor accident consider only a single droplet size and

often predict a persistent residual aerosol concentration that is only very slowly removed from the

atmosphere. In fact, spray nozzles produce a distribution of droplet sizes. The number distribution of

droplets from two types of spray nozzles are shown in Fig. 3.11-6 and Fig. 3.11-7. The number

distributions are not simple, but they are peaked for reasonably small droplets. On the other hand, the mass

weighted distribution of the spray droplet sizes are peaked in both cases above 1000 µm.

Whirjet Spray Head

Droplet Diameter (µm)

0 500 1000 1500 2000

Relative Number Frequency

Fig. 3.11-6 Number distribution of spray droplet sizes for a whirljet spray head

NEA/CSNI/R(2009)5

119

To avoid an artificial prediction of a persistent aerosol in a sprayed atmosphere, it is necessary to account

for the distribution of spray droplets sizes. A complication in doing this is that spray droplets collide with

each other. A commonly used expression for the efficiency of droplet-droplet interactions is:

()

(

)

() ()

[]

() ()

jDiDfor

jDiD

dropletdroplet

=−∈

Experimental evidence suggests that this expression over estimates the interactions of droplets with other

droplets of similar size. A more empirical expression consistent with available experimental data is [8]:

()

(

)

()

(

)

()











<−

=−∈

otherwise0

for

dropletdroplet

Sprayco Model 1713-A

Droplet Diameter (µm)

0 500 1000 1500 2000

Relative Number Frequency

Fig. 3.11-7 Number distribution of spray droplet size for a sprayco model 1713-A nozzle

The evolution of the spray droplet size distribution as a function of fall distance is shown in Fig. 3.11-8.

The effect of droplet-droplet interactions is to remove from the distribution the smaller droplets which are

more effective at trapping aerosol particles.

There has not been a great incentive to improve models of spray removal of aerosol particles since sprays

are enormously effective at decontaminating a atmosphere. Reductions in the aerosol concentration by a

factor of 10 can initially be achieved in a time on the order of one hour. Further reductions in the aerosol

concentration are slower because the action of the spray alters the size distribution of the aerosol so that

particles are less efficiently removed. Nevertheless, decontamination factors in excess of 1000 are

practically achieved.

NEA/CSNI/R(2009)5

120

Watanabe

et al. [9]. have reported integral tests to validate the modeling of spray removal of aerosols in

the MELCOR accident analysis model. The tests were done in a facility 18 m tall with a volume of 12.2

. CsI particulate were used in the test. Other tests in the effort used latex particles. The investigators

were able to obtain good agreement between test results and model predictions.

Fig. 3.11-8 Evolution of droplet size distribution as a function of fall distance based on the empirical

model of droplet-droplet interactions

References

[1] R.K. Hilliard, et al Nuclear Technology, 10 (1971) 499

[2] M. Tanaka, et al Nuclear Technology, 54 (1981) 54

[3] D.A. Powers and S.B. Burson A Simplified Model of Aerosol Removal by Containment Sprays,

NUREG/CR-5966, Sandia National Laboratories, Albuquerque, NM, June 1993

[4] K.W. Lee and J.A. Gieseke J. Aerosol Sci., 11 (1980) 335

NEA/CSNI/R(2009)5

121

[5] A.R.H. Cornish Trans. Institute of Chemical Engineers, 43 (1965) 332

[6] R. Clift, J.R. Grace and M.E. Weber Bubbles, Drops and Particles, Academic Press, 1978

[7] C.S. Pemberton J. Air Pollution, 3 (1960) 168

[8] N. Arbel and Z. Levin Pure and Applied Geophysics, 115 (1977) 869

[9] A. Watanabe, H. Nagasaka, S. Yokoobori, and M. Akinaga

http://www.nupec.or.jp/database/paper/paper_11/

3.12 The Impact of Catalytic Hydrogen Recombiners

Background

Passive hydrogen re-combiners have been studied for many years as a means of preventing accumulation

of hydrogen in the containment of a water-cooled reactor following a design-basis or severe accident [1, 2].

Their installation is now an established and sometimes mandatory measure. Back-fitting of such devices on

existing nuclear power plants has been performed or is in progress in many countries while Generation III

reactor designs (AP600, EPR, etc.) are already equipped with them.

Passive autocatalytic recombiners (PAR) are usually based on palladium and/or platinum dispersed on

some substrate medium. A PAR comprises a vertical conduit with the catalytically-active part near the

bottom. If hydrogen is present in the containment, the hydrogen molecules making contact with the

noble-metal catalyst react with the oxygen of the containment atmosphere producing steam and heat. The

reaction heat provokes heating up of the catalytic elements inducing natural convection which draws more

of the containment atmosphere into the PAR from below. The effect is not only to recombine hydrogen

with oxygen as it passes through the catalytic elements but to mix the containment atmosphere eliminating

any pockets where hydrogen concentrations may be elevated. We note also that carbon monoxide is

oxidised on the catalysing surface. While the operating principle of PARs is well understood, there remains

some progress to be made before it can be said that a thorough understanding exists of the sensitivity of

PARs to the conditions expected to be produced by design-basis or severe accidents. It should be noted,

however, that while little information is in the public domain (e.g., see [2, 3]) much more probably remains

proprietary. There are a number of companies marketing PARs, e.g., Atomic Energy of Canada Limited,

AREVA/ Framatome-ANP, Nuklear Ingenieur Service mbH, Electrowatt-Ekono AG, and each design is

different having its own operating characteristics.

There is a potential problem with the use of PARs in relation to aerosol particles suspended in the

containment atmosphere. As these pass through the catalytic elements they will be heated up along with the

gases and this will inevitably lead to partial evaporation of volatile species from the aerosols. Peak

temperatures of the catalytic elements varies according to the manufacturer but is in the range 500 - 900

°C, i. e., high enough to evaporate many fission-product species. These fission-product vapors may well

interact chemically with each other and/or with the gases of the atmosphere (in general a combination of

nitrogen, oxygen, steam, carbon monoxide and carbon dioxide), especially oxygen. There is therefore the

possibility that reactions in PARs could lead to formation of more volatile forms of fission-product species

aggravating the source term by converting easily-filtered aerosol material into more troublesome vapors

and gases.

Current knowledge

Only a single series of experiments, known as RECI, has been performed to investigate the thermal

decomposition of fission-product species in conditions related to those of a severe accident [4, 5]. The

NEA/CSNI/R(2009)5

122

objective of the RECI experiments was measurement of any creation of more volatile forms of iodine, viz.

, HOI and HI, from metal-iodide species expected in severe accidents, viz. AgI, InI, CdI

and CsI. Two of

these species, cesium iodide or cadmium iodide, in an atmosphere of humidified air were injected as

aerosols into a heated tube in tests both with and without the presence of a catalysing surface. The

experiments comprised separate tests for aerosols of each of the two pure substances. The peak

temperature of the heated tube was varied between 500 and 950 °C. The gas-phase iodine species were not

measured in the heated tube but downstream where temperature had once again become ambient. It was

found that thermal decomposition producing molecular iodine could occur even for the more stable of

these species, cesium iodide, where the presence of a catalyst increased the iodide to iodine conversion by

a factor of two to three. At the highest temperatures, conversion rates of more than 10 % were measured

for cesium iodide. Conversion rates for cadmium iodide were significantly greater. Sensitivity to

parameters such as particle size and transit time through the heated zone was checked where longer transit

times and smaller particles led to higher conversion rates. A further effect of the partial evaporation of the

aerosols was to produce a population of small particles as the vapor(s) re-condensed during cooling

downstream of the heated tube. This new population resulting from homogeneous nucleation was

significantly smaller than any of the source particles injected into the heated tube.

A potential effect of the molecular iodine production would be to poison the catalyst with respect to

hydrogen conversion. However, in separate tests, this poisoning was found to be very limited [3].

Initial analysis of the RECI tests [6] supports the idea that the kinetics of the chemical reactions

downstream of the heated zone is the key to understanding the conversion rates. In other words, if a

thermodynamic approach to chemistry is assumed (i.e., equilibrium chemistry) as the flow cools then the

volatile species convert back to less volatile forms and conversion is virtually zero. Furthermore, the

conversion rates calculated at the peak temperature, i.e., in the heated tube, are close to those measured at

ambient temperature implying almost complete quenching of chemical reactions once they left the heated

tube.

It should be understood that these experimental results are simply a confirmation of the potential for iodine

species from containment aerosols to convert to more volatile species while in transit through a PAR. A

number of representativity aspects prevent these results from being extrapolated to the reactor case. In the

first instance, the aerosols investigated were pure substances: the consequences of using prototypical

mixed-substance aerosols will necessarily alter the gas-phase chemistry produced in the PARs. In addition,

the presence of mixed aerosols with refractory kernels will perhaps facilitate re-condensation of evaporated

species during cooling onto existing aerosols rather than formation of new ones, i. e., heterogeneous rather

homogeneous nucleation. The gas mixture used in these tests was humidified air where, in the

containment, at least during a severe accident, a less-oxidizing mixture containing a higher steam fraction

as well as hydrogen will be present: it is known that reducing conditions are thermodynamically less

favourable to formation of volatile iodine species. Finally, the cool-down characteristics of the flow

leaving the PARs have not been reproduced in the RECI tests: this is crucial to the propensity for vapors to

re-convert as they cool where more rapid cooling will not favour this re-conversion to a gas-vapor-aerosol

mixture closer to that at equilibrium at ambient temperature.

References

[1] F. Fineschi, G. Koroll, J. Rohde “Mitigation of hydrogen hazards in water cooled power reactors”,

IAEA-TECDOC-1196 (2001)

[2] E. Bachellerie, F. Arnould, M. Auglaire, B. de Boeck, O. Braillard, B. Eckardt, F. Ferroni, R.

Moffett "Generic approach for designing and implementing a passive autocatalytic recombiner PAR-

system in nuclear power plant containments” Nucl. Eng. Des. 221(1-3), 151-165 (2003)

NEA/CSNI/R(2009)5

123

[3] F. Morfin, J.-D. Sabroux, A. Renouprez “Catalytic combustion of hydrogen for mitigating hydrogen

risk in case of a severe accident in a nuclear power plant: study of catalysts poisoning in a

representative atmosphere”, Applied Catalysis B: Environmental 47, 47-58 (2004)

[4] F. Deschamps, J.-C. Sabroux “Etude de la production d’iode gazeux par un aerosol d’iodure de

césium soumis à des temperatures élevées dans l’air humide” Proc. 18

French Aerosol Conference,

Paris, 11-12 Dec. 2002, ASFERA (2002)

[5] F. Deschamps, J.-C. Sabroux “Particulate iodide to gaseous iodide conversion in a passive

autocatalytic hydrogen recombiner” J. Aerosol Sci. 34, S1087-S1088 (2003)

[6] N. Baltzer, M.P. Kissaneprivate communication (2005)

3.13 Filters

Type of filters in use

The requirements for the design of a filter system in removing the fission products depend on the thermal

hydraulic conditions (temperature, pressure, humidity level or steam condensation, flow rate through the

filter system) and concentration of the fission products in gaseous and aerosol form. The severity of the

conditions imposed by the normal operational conditions or accidents (design basis or severe accidents

involving core damage) produce the challenge for the design of the systems.

Normal operation and design basis accidents

During operational conditions containments are normally kept sub atmospheric in order to avoid any

accidental release of even small activity into the environment. Ventilations systems, depending on the

design and regulatory requirements, may contain parallel trains of filtration systems. A typical filter train

contains the following sequential components: (1) a moisture separator to remove entrained water droplets,

(2) a heater to control relative humidity (RH) when the RH of the air entering the carbon adsorber exceeds

nominally 70 %, (3) prefilters, (4) HEPA filters, (5) a charcoal adsorber, (6) HEPA filters downstream of

the adsorbers, and (7) a fan. Fig. 3.13-1, adopted from [1], displays a typical filtration unit. Ducts, valves,

and dampers are also included for system isolation and flow control, as well as related instrumentation.

When the moisture and dust loads are low for all credible operating modes, the prefilter and moisture

separator may not be required.

The source of activity in the containment originates from the release of dissolved activity (e.g., iodine) in

the primary coolant water as a result of a leak from the circuit. The source of the activity in the circuit

could be due to leaking fuel elements or due to iodine spiking. The maximum permissible activity level in

the circuit and in the containment is defined by regulations.

NEA/CSNI/R(2009)5

124

Fig. 3.13-1 A typical air cleaning system [1]

By definition design basis accidents (DBA) involving loss of coolant foresee rupture of cladding of certain

number of fuel rods, which then causes the release of fuel cladding gap inventory into the primary coolant

water. The inventory is mostly composed of noble gases, and volatile fission products (iodine, cesium).

The release of the primary coolant water into the containment causes the activity release into the

containment. If the accident is a containment by-pass accident then the activity is released directly either

into the auxiliary/reactor building (V-sequence) or through the broken steam generator tube(s) (SGTR) into

the steam generator secondary side and after some scrubbing in the secondary water, eventually directly

into the environment. Loss of coolant accidents due to SGTR with few tube failure normally do not cause

any fuel cladding failure, but a certain fraction of the dissolved activity in the primary circuit at the time of

the accident may be released into the environment in a combination of gas and in water dissolved form.

Release of water droplets or continuous water flow will be the transport media for the dissolved activity

out of the steam generator secondary side. The break location determines the extent of each of such release.

Containment or auxiliary (reactor) building filtration systems are designed to avoid any substantial release

of activity transported by aerosol particles and gaseous iodine. Of course the main assumption at the

background of this statement is that the containments are isolated and there are no uncontrolled leak paths.

Severe accidents

Severity of the conditions during severe accidents varies depending on the success of the accident

management measures attempted. However, anticipated conditions: high temperature, high pressure, high

NEA/CSNI/R(2009)5

125

steam concentrations, hydrogen burns or detonations, high aerosol and gaseous fission product

concentrations, and resulting high activity levels, are very much depending on the accident progression

driven by the status of the hardware as well as the accident management efforts. Contrary to the

operational or design basis accident conditions the anticipated severe accident conditions cannot be used as

the design parameters for the conventional filtration systems within the practical limitations. However, as a

result of the emerging new regulatory requirements for the severe accidents, which can not be ‘normally’

fulfilled by the conventional filtration systems new filtration concepts were developed in 1980s to backfit

the current operating reactors in some countries. The main emphasis in the new regulatory requirements is

to keep the pressure in the containment under the design limits in order to avoid catastrophic containment

failures by relieving the pressure by venting through a containment venting filter, which should at the same

time remove the aerosol particles and molecular gaseous iodine with certain efficiencies.

Some of the new developed systems by various vendors at that time were tested by an international project,

Advanced Containment Experiments (ACE) Phase A, 1988-1992 [2]. The following section summarises

the filter concepts and the results achieved from the ACE experiments. Another containment venting filter

unit developed and tested in Switzerland is also introduced.

Characterisation of filter performance

There are several parameters currently in use to describe the effectiveness of a filtration device. These are,

however, all based on the mass of material injected to a device, m

in,

and the mass of material exhausted,

out

In the nuclear industry the most commonly used parameter is the Decontamination Factor, DF, which

is defined as the ratio of the injected to exhausted mass:

out

≡

Intimately related to the DF is the penetration, Pt

which is defined as the ratio:

out

≡

This is just the inverse of the DF. The collection efficiency, η, is defined the ratio of the collected to the

injected mass. This can be written in terms of the injected and exhausted masses as:

out

1 −≡η

For most filtration devices, the efficiency of collection depends strongly on the particle size. Hence it is useful

to introduce the concept of a "fractional efficiency" where the definitions above apply for particular particle

size ranges. This definition introduces a difficulty for processes where the particles change size within the

filtration device. For example, hygroscopic particles can grow substantially [3] in wet scrubbers. In this

situation, the amount of exhausted mass in a particular size range must be based on the inlet size

distribution. Unfortunately, this is impossible to do experimentally, where only inlet and outlet size

distributions are measured, but the inlet and outlet particle concentrations in each size range cannot be

related. The only ways to determine this information are to either use aerosols, which do not grow, or less

ambiguously, inject only monodisperse aerosols.

NEA/CSNI/R(2009)5

126

The penetration can be quantified on the basis of a one dimensional filtration model as illustrated in Fig.

3.13-2

Fig. 3.13-2 Scheme of a one-dimensional filtration model

If the flow of aerosols through a control volume of length dz is considered and additionally a deposition

velocity v

dep

and a deposition area of P

dep

dz are postulated, a mass conservation equation for aerosols can

be written as:

Aerosols in













−

- Aerosols out













− =

= Aerosols deposited

dzPvn

depdep

and after the integration as:





















−=≡−≡≡

depdep

out

P expη1





















−=

depdep

exp

These expressions for overall penetration or efficiency apply to all the filters considered.

Nomenclature

Latin Symbols

Cross sectional flow area

dep

Deposition area

DF Decontamination factor

Mass of aerosol particles injected to filtration system

out

Mass of aerosol particles exhausted from filtration system

Particle mass

n Aerosol particle number

NEA/CSNI/R(2009)5

127

number of particles entering in the filter

out

number of particles leaving the filter

dep

Deposition area in the vertical surface of a filter slice along the integration direction

Penetration

Volumetric flow rate of gas

Gas velocity

dep

v Deposition velocity

z Axial distance along venturi from the water injection point

Greek Symbols

η Overall collection efficiency

References

[1] DOE handbook, nuclear air cleaning handbook, DOE Technical Standards, DOE-HDBK-1169-2003,

November 2003

[2] M. Merilo, I.B. Wall “Containment Filtration Systems Tests, Summary Report” Electric Power

Research Institute, ACE Phase A, TR-A22, February 1992

[3] M.S. Hoseyni, A.T. Wassel "Growth of Aerosol Particles in a Steam Environment and its Effect on

Removal", Nuclear Eng, & Des, Vol. 97, 103 - 109, (1986)

3.14 Two Phase Flashing Jets and Water Aerosol Behaviour

The scenario for generation of a two-phase flashing jet in a reactor accident involves pressurised coolant

upstream of a break being released into a region of lower temperature and pressure. The value of the

superheat (T

) for the pressurised water depends upon its initial temperature, and the saturation

temperature of the atmosphere to which it is released, T

sat

∞

). If the saturation temperature is less than the

initial temperature of the liquid, then the release is defined to be superheated, with the degree of superheat

equal to T

= T

- T

sat

∞

). If the liquid is released into standard atmospheric conditions, this simply

represents the amount the initial temperature exceeds the normal boiling point of that liquid (T

= T

- T

The initial state of the water is termed sub-cooled. If there is a break in the heat transport system, and the

water remains subcooled at the exit of the break, it would emerge as a single-phase homogeneous jet that

would disintegrate through mechanical processes. If downstream conditions lie below the liquid/vapor

phase line, (T

> 0), the liquid pressure in the vicinity of the break may drop below the local saturated

vapor pressure, and the process known as flashing begins. The flashing process would be expected to be

the dominant mechanism for formation of water droplet aerosols after breaks in the high-pressure

primary-heat transport system (HTS).

Most current jet dispersion codes divide the process of atomisation or flashing of superheated jets into two

possible outcomes governed by mechanisms that operate independently, and whose relative importance

depend on the degree of superheat of the liquid: the first is termed mechanical break-up, the second

flashing break-up. However, there is no consensus regarding the relative contributions of these two

processes to the droplet size distribution from a flashing jet.

The behaviour of the two-phase (water-droplet and steam) jet has important implications for reactor

accident safety analyses. High enthalpy coolant flashing can cause the containment pressure to rise above

atmospheric, thereby increasing the potential for airborne fission products to leak out of containment. The

water droplets generated in the jet can “wash” previously released airborne aerosols and gaseous fission

NEA/CSNI/R(2009)5

128

products out of the atmosphere. The jets themselves are also a primary means by which fission products

enter containment. Water droplets formed by the jet discharge can coalesce to form larger droplets, settle to

the floor of the discharge room, impinge upon containment walls and structural surfaces, or remain

suspended, where they could be carried to the atmosphere through containment leaks, or to another room in

containment. Consequently, evaluation of the radiological consequences of nuclear reactor accidents

requires models for two phase flashing jets.

Water droplet aerosols generated by two-phase flashing jets are subject to the same aerosol removal

processes that are modelled routinely for dry aerosols in atmospheric aerosol codes (see Chapter 2).

However, the large momentum of the jets from an HTS break can result in inertial impaction becoming a

uniquely dominant removal mechanism if there are any obstacles in the jet path [1, 2]. The combination of

processes such as inertial impaction and turbulent deposition (see Chapter 2) that act to remove water

droplets after jet-obstacle interaction is referred to as impingement. The solid surface acts as an aerosol

collection device, with a collection efficiency

that depends on droplet thermophysical and transport

properties (size, angle of incidence, velocity), and surface roughness.

Aerosol collection efficiencies for inertial impaction are generally correlated to the droplet Stokes number

(St

), defined as the ratio of particle stopping distance, or inertial range (τ) at a given velocity, to the nozzle

(or jet) radius (see Chapter 2). Assuming that the velocity and droplet size distribution are known at the

region where jet impaction takes place, and that correlations between collection efficiency and droplet

Stokes number hold, jet impingement could be modelled using the same correlations. However all of the

data correlating collection efficiencies to Stokes numbers was obtained at Reynolds numbers of less than

1 × 10

, [3, 4, 5] whereas Reynolds numbers for inertial impaction from 10 MPa discharges could be more

than an order of magnitude higher [Chiang, 1998]. There are no studies of impactor collection efficiencies

at high Reynolds numbers.

A large body of data on the behaviour of flashing liquid propane is available from a European Union

funded program performed by the CEA in Grenoble [7, 8]. The propane release behaviour was examined

as a function of orifice diameter and initial storage pressure. The experiments measured the void-fraction,

pressure and temperature data within the blowdown vessel. Droplet size and velocity measurements were

taken at 3 downstream axial locations, and the distributions have been described as well-behaved functions

that could be adequately represented by general size distribution functions such as the lognormal or Rosin-

Rammler

[9]. Unfortunately, there are very few experiments suitable for validating models on the

behaviour of flashing jets from high temperature, high pressure water releases, and even fewer for

validation of processes such as impingement

The WALE facility experiments [1, 2] represent the largest collection of data on flashing water jet releases,

and are the only intermediate scale studies performed on droplets generated by flashing and removed by

impingement on obstacles. The studies were performed in a 19 m

cylindrical pressure vessel, with

discharge of liquid water at 1 to 10 MPa, and various degrees of sub-cooling. The effects of nozzle

geometry (size, length to diameter ratios), impingement plate characteristics (size, orientation with respect

to the jet and distance from break) were examined. Droplet velocities and size were also measured,

although experimental difficulties were encountered with measurement of the former. In general, the data

defined as the ratio of the mass of droplets or particles deposited on the solid surface per unit time to the total mass of droplets or

particles moving with the jet per unit time.

The weight-size distribution proposed by Rosin and Rammler in 1933 is

where M(>ℓ) is the cumulative mass of all particles of mass greater than size ℓ, M

is the total mass of the distribution, Φis a

size related to the average size of the distribution, and the exponent k is a free parameter. [10]:

Small-scale studies published by [11] have been used to correlate liquid collection efficiency for a two-phase jet impinging

normally on a flat surface.

NEA/CSNI/R(2009)5

129

showed that, for the entire range of conditions studied, more than 99 % of the water droplets were removed

from the vessel by various mechanisms.

The behaviour of flashing jets, including their production of water droplets, is reasonably well understood

qualitatively. However, there is no consensus in the literature regarding the prediction of water aerosol

droplet sizes from these jets, and very little data with which to validate models for aerosol removal from

this high temperature, high pressure flashing jet releases.

References

[1] R.J. Fluke, K.R. Weaver, G.L. Ogram, L.N. Rogers, C.F. Forrest The Water Aerosol Leakage

Experiments: Programme Description and Preliminary Results In Proceedings of the Second

International Conference on Containment Design and Operation, Toronto, 1990A

[2] R.J. Fluke, G.L. Ogram, L.N. Rogers, K.R. Weaver Aerosol Behaviour in the Water Aerosol

Leakage Experiments In Proceedings of the Second International Conference on Containment

Design and Operation

, Toronto, 1990B

[3] V.A. Marple A Fundamental Study of Inertial Impactors. Ph.D. Thesis, University of Minnesota,

Minneapolis, MN, 1970

[4] V.A. Marple, B.Y.H. Liu Characteristics of Laminar Jet Impactors Environmental Science and

Technology, 8. 648, 1974

[5] V.A. Marple, K.L. Rubow Theory and Design Guidelines In Cascade Impactor Sampling and Data

Analysis, American Industrial Hygiene Association Monograph, J.P. Lodge and T.L. Chan (editors),

1986

[6] H.W. Chiang A Model for the Removal of Water Droplet Aerosols from a Flashing Jet Impinging

onto a Plate In Proceedings of an OECD/CSNI Workshop on Nuclear Aerosols in Reactor Safety,

Cologne Germany, NEA/CSNI/R(98)4,1998

[7] E. Hervieu, T. Veneau Experimental Determination of the Droplet Size and Velocity Distributions at

the Exit of the Bottom Discharge Pipe of a Liquefied Propane Storage Tank During a Sudden

Blowdown J. Loss Prev. Ind., 9, No. 6, 413-455, 1996

[8] S. Vandroux-Koenig, G. Berthoud Modelling of a Two-Phase Momentum Jet Close to the Breach, in

the Containment Vessel of a Liquefied Gas J. Loss Prev. Ind., 10, No. 1, 17-29, 1997

[9] H.W.M. Witlox, P.J. Bowen Flashing Liquid Jets and Two-Phase Dispersion a Review Health and

Safety Executive (HSE) Contact Research Report 403. St Clements House, 2-16 Colegate, Norwich,

UK, 2002

[10] W.K. Brown, H.K. Wohletz Derivation of the Weibull Distribution Based on Physical Principles and

Its Connection to the Rossin-Rammler and Lognormal Distributions Journal of Applied Physics

78,

No. 4, 2758-2763, 1995

[11] N.A. Fuchs, C.N. Davis (editors) The Mechanics of Aerosols Pergamon Press, Oxford England,

1964

NEA/CSNI/R(2009)5

130

3.15 Penetration Leakages

3.15.1 Introduction

An understanding of the aerosol-removal and transport in containment leak paths is important both from

the perspective of calculating the amount of airborne activity released into the environment and in

estimating the doses to containment for equipment qualification and habitability assessment.

In the event of a reactor accident in which the containment envelope becomes pressurised, there is a

potential for aerosol and gas leakage to the outside atmosphere through a variety of paths that can be

categorised as belonging to one of three general types:

•

valves and seals

•

concrete joints, cracks and penetration gaps (e.g., process and control wiring and piping), and

•

pores in intact concrete.

Leakage past seals and valves in containment would involve a sudden change in the flow cross-sectional

area and short path lengths relative to those of the other flow paths. Leak paths along concrete-penetration

gaps, joints and cracks are more tortuous and significantly longer. Finally, leak paths through concrete

pores are characterised by small diameter cross-sectional areas, long and complex flow channels and high

flow resistances. Depending upon the temperature of and flow through these paths, and the size of the

aerosols, significant aerosol retention might be expected. However, many safety analyses still assume that

aerosol release rates and gas release rates are identical.

Most experimental investigations on aerosol retention in leak paths [1, 2, 3, 4] have focused on small leaks

and capillaries of diameters ranging from few microns to few millimetres, submitted to pressured

differences up to several bars. However, large scale experiments were conducted in the Containment

System Experiments (CSE) programme which used a one-fifth linear scale model of a typical 1000 MWe

PWR [5]. The CSE results indicated aerosol decontamination factors from 10 to 100 under dry conditions

(15 for iodine and 100 for cesium), and complete retention under wet conditions. More recently, [6]

performed large-scale experiments on actual containment penetrations of BWR containments and they

found decontamination factors for CsI under dry conditions ranging from 10 to 1000.

Some theoretical developments have been also undertaken. [7] proposed a broadly accepted correlation

(Equation 1) that relates the total aerosol mass, m (g), leaked before reaching a complete plugging of the

crack to duct of diameter D (cm

⋅

(1)

where k (30 ± 20 g/cm

) is an empirical constant determined by fitting data obtained from

straight and smooth ducts 20 µm to 26.5 cm in diameter with circular cross sections, submitted to pressure

differences ranging from 0.3 kPa to 0.7 MPa.

The theoretical basis for aerosol penetration in cracks was reviewed by [8]. However, a mechanistic model

based upon this theory would be too complex to be implemented in current safety analyses codes.

Intermediate approaches between empirical correlations and mechanistic models look to be the most

promising. Anyway, whatever modelling nature, models used in containment codes must be validated over

a range of conditions (e.g., leak path geometries, temperatures, pressures etc.,) that may be reactor and

accident specific. Presently, existing models are not mature enough and a sound, reliable and representative

database against which to validate them is still missing.

NEA/CSNI/R(2009)5

131

The first subchapter presents the status of current modelling and experimental studies on aerosol leakage.

The first summary describes work being carried out within the Severe Accident Research NETwork

(SARNET) project of the 6

Framework Programme of the European Commission, and it is based on a

recent paper presented at the International Conference on Nuclear Energy for New Europe 2005 by [9].

The second subchapter is CANDU specific work being performed at AECL to characterise wet aerosol

leakage through containment leak path geometries.

3.15.2 Current modelling and experimental studies

The factors influencing aerosol retention in a leak path have been identified and classified as follows:

•

Thermo-hydraulics; the most important variables are the pressure inside the containment, the

gas composition, the pressure drop and the wall temperature along the crack. Factors which

influence steam condensation onto crack surfaces will be particularly relevant.

•

Crack geometry; the important parameters are primarily the crack path length and hydraulic

diameter; crack section shape (likely flat) and curvature are also very important.

•

Aerosol characteristics; the reference parameters would be in-containment average

composition, concentration and size distribution.

It is worth noting that variables considered highly uncertain at the time the leak path initiates (like

composition and morphology of aerosols), have been disregarded. Otherwise, accurate models relying on

those variables would not yield reliable estimates.

The values of these variables should encompass two bounding scenarios: the “short term scenario”, in

which energetic phenomena, such as hydrogen deflagrations and/or steam explosions might result in an

early containment impairment; and the “long term scenario” where steam and gas released from the molten

core-concrete interaction would cause containment over-pressurization. In the former, the in-containment

aerosol concentration could reach high values (1-10 g/m

) and size distribution would be shifted toward

particles of the order of some microns. In the latter, the aerosol concentration would be lower and the size

distribution would be weighed more heavily in the range of sub-micron particles. According to

experimental data available, the maximum interest from a safety point of view should be placed on

particles around 0.1 µm (since they would undergo the lowest retention).

Table 3.15-1 Some variables influencing the aerosol retention inside containment cracks

Crack path ≥ wall thickness (1.0 – 1.4 m)

Crack cross section and shape << 2-3 cm

- flat, in the case of long term scenario

Containment internal pressure 5 – 10 bar

Crack surface temperature room to containment gas temperature

Atmosphere composition depending on accident scenario and evolution

Aerosol size distribution 0.1 – 5 µm

Two different approaches are being used to model aerosol retention: Eulerian and Lagrangian. A brief

description is given next. In-crack flow dynamics and heat transfer studies [10, 11], even though highly

important for aerosol behaviour, are not reported here.

NEA/CSNI/R(2009)5

132

An experimental programme on aerosol retention within concrete cracks is being conducted by

IRSN [12, 13]. Reinforced concrete samples (128 cm x 75 cm x 10 cm) are being subjected to shear

stresses up to 645 kN. The resulting cracked wall is placed between two identical sealed boxes.

Fluorescein particles are injected in the upstream box and their mass concentration is measured in

both upstream and downstream boxes. Particles having diameters of 1

µm and 4 µm in dry

atmospheres have been tested so far, and it is planned to conduct tests with 50 nm-particles. The

maximum pressure drop across the wall has been 80 mbar

Fig 3.15-1 is a scheme of the IRSN facility.

Fig. 3.15-1 IRSN facility to investigate aerosol retention in containment cracks

Total retention of aerosol particles was observed for 4 µm particles under all the velocities tested.

However, only a fraction of the 1 µm aerosols was retained. The main retention mechanisms involved were

diffusion, sedimentation and impaction (at the crack inlet). Experiments with air-steam mixtures are

planned to study the effect of wet conditions in the crack decontamination factor.

Presently, another experimental facility, named COLIMA (CEA), is being considered as a future candidate

in which to conduct experiments since it has appealing characteristics like the ability to generate

prototypical MCCI aerosols, design pressure and temperature of 5 bar and 140 ºC, respectively, and the

capability of producing steam.

3.15.3 Measurement of the retention of wet aerosols

Wet aerosols (droplets containing suspended or dissolved fission products) are more prevalent than dry

aerosols in most postulated design basis CANDU accident scenarios. However, the published literature on

aerosol behaviour in containment leak paths is predominantly for dry aerosols, as are the empirical

correlations used to describe the behaviour. Although some empirical deposition models may be applicable

to wet aerosol behaviour, aerosol characteristics such as particle shape, and deposit characteristics (fluid

rather than solid) could cause somewhat different behaviour in potential leak paths. Over the past few

years, experiments and modelling studies have been performed at AECL to characterise wet aerosol

leakage through containment leak path geometries. The following section describes experiments on aerosol

droplet deposition and transport through paths that are representative of seals and valves in containment.

-Patm

∆P=P

-P

Pressure regulator

Pressure gauges

Upstream aerosol box Downstream aerosol box

Wall cracked by shear loading

10 cm

Cracks

Pump

Valve

HEPA sampling

Filter

Fluoresceine

Aerosol

(known

particles

massflow)

Mass

Flowmeter

downstream

Vibrating Orifice

Aerosol Generator

Filter

Excess

Neutralisation

source

(600 MBq

Kr)

upstream

-Patm

∆P=P

-P

Pressure regulator

Pressure gauges

Upstream aerosol box Downstream aerosol box

Wall cracked by shear loading

10 cm

Cracks

Pump

Valve

HEPA sampling

Filter

Fluoresceine

Aerosol

(known

particles

massflow)

Mass

Flowmeter

downstream

Vibrating Orifice

Aerosol Generator

Filter

Excess

Neutralisation

source

(600 MBq

Kr)

upstream

NEA/CSNI/R(2009)5

133

Flow paths and conditions

As described earlier, aerosol leakage through valves and airlock-door seals in containment will be

characterised by an abrupt change in the cross-sectional flow area, where gas and aerosol flows move from

a high pressure (containment) region, through small gaps resulting from seals or valves being improperly

seated, to a lower pressure region. A diagram of the generalised leak path through a single contraction is

shown in Fig. 3.15-2. Note that the leak path from containment to the environment through airlock door

seals and isolation valves generally involves more than one such abrupt contraction.

Fig. 3.15-2 Diagram of a generalised leak path

The abrupt change in flow area for a pressurised flow is expected to lead to aerosol leakage occurring

under choked-flow (or sonic) conditions at the contracted portions of the leak paths. The flow through an

opening of a given diameter increases with increasing pressure differential until a pressure wave, created

within or near the exit of the orifice, results in sonic flow. A further reduction in downstream pressure

cannot increase the velocity through the opening, and the gas velocity through the hole will be maintained

at the speed of sound

. A choked-flow condition results in a constant mass-flow rate, at the speed of sound,

through a nozzle or an opening. The criterion for choked flow is a downstream absolute pressure of less

than 53 % (for air) of the upstream absolute pressure [14].

An experimental benefit resulting from a choked-flow condition is that the leak paths are self-regulating to

ensure a constant flow rate. Choked-flow conditions can be achieved by either pressurising the upstream

side of the leak paths or by maintaining a sufficiently low pressure downstream of the leak paths to ensure

that the downstream-to-upstream pressure ratio is lower than 0.53

. The latter choice was selected because

it was expected to adequately simulate aerosol leakage from containment, but without the experimental

difficulties associated with a design in which the upstream chamber would be at an elevated pressure.

Experiments

Experiments on aerosol leakage were performed in a cubic plexi-glass chamber fitted with an aerosol

characterization port and a pair of filtration ports. The filtration ports are shown schematically in Fig.

3.15-3. A deflector was installed above each port to prevent collection of surface runoff from the vertical

surface in to the port entrance. The constricted leak path portion of an isolation-damper valve was

simulated using commercially available copper tubing, with diameters and lengths representative of

The sonic velocity U

sonic

(m·s

-1

) for a gas is: where: R = gas constant (=287 m·N·kg

-1

·K

-1

), T

= upstream temperature (K) and =

specific heat ratio (= 1.4) [15].

The atmospheric (101 kPa) to containment (225 kPa) pressure ratio following an accident is estimated to be 0.45 in CANDU

reactors.

NEA/CSNI/R(2009)5

134

damper valve leak paths possible in CANDU stations. The simulated constricted flow path of the

airlock-door seals was fabricated in-house, using cylindrical copper tubing of much smaller dimensions.

Fig. 3.15-3 Schematic of filtration sampling lines

The filtration assembly shown in Fig. 3.15-3 consisted of two separate flow lines (U1 and U2) containing

matched leak paths. The U1 line consisted of a high efficiency filter assembly to collect water droplets,

upstream of the prototypical leak path, whereas in the U2 line, the filter assembly was positioned

downstream of the leak path. Each line was connected to a rotameter through a valve to measure the flow;

the valves were used to match the flow rates through the U1 and U2 lines. A vacuum pump was connected

to the downstream port of the rotameters to maintain low pressures and ensure choked-flow conditions

through the leak path.

A Cyclone Fogger (Curtis Dyna-Fog

, Model 3000) containing a solution of an uranine dye in water was

used for aerosol generation. The uranine (sodium fluorescein) dye fluoresces at a characteristic wavelength

and concentrations of the uranine-water retained on the filter assemblies and in the leak paths were

determined using fluorescence spectrophotometry. In a typical experiment, the vacuum pump was started,

and the flow valves for each line were adjusted so that the rotameter reading on each line was the same.

The fogger was switched on to inject aerosols into the cubic reservoir, and a phase-Doppler anemometer

(PDA) focused through the aerosol characterisation port was used to measure the mean (average) diameter

of wet aerosols. The mean droplet size generated for these experiments was 9.7 µm. Droplet mass

concentration was not measured, but can be estimated from the ratio of the aerosol injection rate and the

volumetric flow rate through the filtration port to be around 5 g·m

-3

for the isolation damper valve

experiments. The experiments described here were performed at ambient temperatures.

Filters and leak path assemblies were removed after each experiment and washed thoroughly with known

quantities of distilled water, to re-dissolve the deposited uranine dye. Samples of the filter and leak path

washes were then submitted for fluorescence analysis to determine the uranine concentration. The

aerosol-penetration and aerosol-deposition efficiencies were estimated directly from the measured uranine

concentrations. The method has been adapted from one reported by [16].

Results in Table 3.15-2 show that for the simulated damper-valve leak paths, the average penetration

efficiency,, is 36 % ± 5% (1σ), and the average deposition efficiency,, is 64 % ± 6 % (1σ). There is no

significant effect of tube length on the deposition efficiency. For airlock-door seal leak paths, the average

penetration efficiency,, is 80 % ± 2 % (1σ), and the average deposition efficiency,, is 20 % ± 2 % (1σ).

NEA/CSNI/R(2009)5

135

Table 3.15-2 Retention and transmission efficiency in an idealised isolation-damper valve contraction

Test No.

(%)

(a) 9-cm long tube

D9R1

D9R2

D9R3

(b) 6-cm long tube

D6R1

D6R2

D6R3

D3R1

D3R2

D3R3

Average 36 64

Table 3.15-3 Retention and transmission efficiency in an idealised airlock-door seal leak contraction

Test No.

(%)

ηη

+ (%)

(%)

0.6-cm long tube

A6R1

A6R2

A6R3

113

146

103

Average 89 23 112 80 20

Leak path plugging

From earlier measurements of depressurisation tests performed with a simulated isolation-damper valve

leak paths, it was shown that the post-accident elevated pressure in a CANDU containment building would

level off (from 140 kPa(g) to zero) in about 15 minutes. Hence, tests were performed to monitor aerosol

flow through the U2 lines for the same duration of time to examine whether leak path plugging would

occur. The experiments indicated that there would be no leak path plugging during containment

overpressurization for an aerosol mass density of about 5 g·m

-3

NEA/CSNI/R(2009)5

136

References

[1] C.T. Nelson, R.P. Johnson Aerosol leakage tests -Status of Technology for Radiological

Consequence Assessment of postulated Accidents in LMFBR’s, ERDA-56, Springfield, USA, 1975

[2] S.L. Sutter Depleted Uranium Dioxide Powder Flow Through Very Small Openings NUREG/CR-

1099, PNL-3177, 1979

[3] J.P. Mitchell, R.T. Edwards, H.M.E. Ball The Penetration of aerosols Through Fine Capillaries

Journal of Radioactive Materials Transport (RAMTRANS), n. 3, 5-17, 1992

[4] D.A.V. Morton, J.P. Mitchell Aerosol Penetration through Capillaries and Leaks: Experimental

Studies on the Influence of Pressure Journal of Aerosol Science, n. 26, pp 353-367, 1995

[5] R.K. Hilliard, A.K. Postma Large-Scale Fission Product Containment Tests

Nuclear Technology,

Vol. 53, n. 2, pp 163-175, 1981

[6] A. Watanabe, T. Hashimoto, M. Osaki Fission Product Aerosol Trapping Effects in the Leakage

Path of Containment Penetration under Severe Accident Conditions in Proc. 3

OECD Specialist

Meeting on Nuclear Aerosols in Reactor Safety, Cologne, Germany, 1998

[7] E.U. Vaughan Simple Model for Plugging of Ducts by Aerosol Deposits Trans. Am. Nucl. Soc., 28,

507, 1978

[8] C.F. Clement Aerosol Penetration through Capillaries and Leaks: Theory Journal of Aerosol

Science, 26, 369, 1995

[9] F. Parozzi, S. Paci ECART Code Integrated Approach for Thermal-Hydraulic-Source Term

Analysis, Proc. 4

International Conference on Nuclear Option in Countries with Small and Medium

Electricity Grids, 16-20 June, 2002, Dubrovnik, Croatia, 2002

[10] H. Simon, N. Coulon, R. Tomassian, G. Nahas A model for two-phase flow through small opening

cracks 4

European Thermal Sciences Conference, 29-31 March 2004, Birmingham, UK, 2004

[11] H. Simon, N. Coulon, R. Tomassian, G. Nahas Air-steam leakage through cracks in concrete walls.

NURETH-11, 2-6 October, 2005, Avignon, France

[12] T. Gelain, F. Gensdarmes, J. VendelExperimental study on aerosol penetration through cracked

concrete wall, Congress EAC 2004 Budapest, September 6-10, 2004

[13] T. Gelain, J. VendelDétermination de modèles de dépôt d'aérosols dans un réseau de fissures,

Congrès français sur les aérosols (CFA) Paris 14-15 décembre 2005

[14] H.C. HindsAerosol Technology: Properties, Behaviour, and Measurement of Airborne Particles,

John Wiley & Sons, Toronto, 1982

[15] V.L. Streeter, E.B. Wylie Fluid Mechanics, McGraw-Hill Book Co., New York, 1979

[16] J.K. Agarwal Aerosol Sampling and Transport Ph.D. Thesis, University of Minnesota, 1975

NEA/CSNI/R(2009)5

137

4. MODELLING APPROACHES IN CODES (INCLUDING BASIC VALIDATION)

4.1 Integral Codes

4.1.1 MELCOR

4.1.1.1 Introduction

MELCOR [1] is a systems-level model for the prediction of the progression of accidents in light water

nuclear power reactors and other nuclear facilities. It was developed initially by the U.S. Nuclear

Regulatory Commission (NRC) as a model to support the quantitative estimation of risk to the public

associated with nuclear facilities and especially nuclear power plants. This initial thrust of code

development has been augmented, and MELCOR has become a tool for the mechanistic analysis of reactor

accidents. MELCOR has become the repository used by the NRC for the preservation of knowledge and

understanding of severe accidents developed in a variety of experimental and analytical research efforts.

For the purposes of reactor accident analyses, the code is intended to predict accident progression from the

initiating event, to the point of core uncovery, through vessel failure and the expulsion of core debris into

the containment, to the point of containment failure and the prolonged escape of radioactive materials into

the nuclear power plant environment. The MELCOR code provides input to a companion code, MACCS,

for the analysis of radioactive material dispersion in the environment and the consequences of this

dispersion.

The MELCOR code has a substantial, world-wide community of users. The code has a rather flexible

architecture so that it can be used to predict accident progression in many different types of nuclear

reactors. MELCOR is also applied to the prediction of accident progression in facilities for processing of

nuclear materials especially for accidents involving fires.

The MELCOR code was developed initially as a systems-level analysis code and was not intended for the

analysis of experiments. Instead, it was envisaged that the code would adopt and adapt essential features

though not all the detailed features of computer codes used for the analysis of major experiments. This has

been done in many cases as will become apparent in the discussion below of the major “packages” in the

MELCOR code. There has been, however, an increasing willingness to use the systems-level code directly

for the analysis of major experiments. It is a statement concerning the flexible architecture of the code that

such analyses are possible. It is also a statement concerning the increase in capabilities of routinely

available computational devices that the MELCOR code continues to be developed in the direction of ever

more detailed descriptions of accident phenomena. The MELCOR code has become, consequently, an

important tool in the drive to make the safety regulation of nuclear power plants based ever more on

realistic rather than overly conservative depictions of accident phenomena.

4.1.1.2 Major MELCOR “packages”

The MELCOR computer code consists of a number of “packages” that address particular aspects of reactor

accident analyses. The more important of these packages are listed in Table 4.1-1. Most of the packages in

MELCOR are associated with the analysis of reactor core and structure degradation. The RN package is

devoted to the prediction of radionuclide behaviour under accident conditions. The RN package will be the

focus of the discussions in this brief description of the MELCOR computer code.

NEA/CSNI/R(2009)5

138

Table 4.1-1 Pertinent 'packages' in the MELCOR computer code for reactor accident analysis

Symbol Package Name Description

EXEC

Executive Package Responsible for overall execution control of the calculations

BUR

Burn Package models the combustion of gases in control volumes

CAV

Cavity Package

Models the attack on the basemat concrete by hot or even

molten core materials

CND

Condenser Package

Models the effects of Isolation Condenser Systems and

Passive Containment Cooling Systems found in some boiling

water reactors

Control Function

Package

Allows users to modify the modeling in MELCOR by

defining functions of variables in the MELCOR database and

make the values of these functions available to other

MELCOR packages

COR

Core Package

Calculates the thermal response of the reactor core, the lower

plenum internal structures, core internal support structures

and the reactor vessel lower head.

CVH

Control Volume

Hydrodynamics

Package

Modeling of the thermal-hydraulic behaviour of liquid water,

water vapor and gases in control volumes

DCH

Decay Heat

Package

Models the decay heat power from fission products

FCL

Fan Cooler

Package

Models the heat and mass transfer associated with operation

of fan coolers in the reactor containment

FDI

Fuel Dispersal

Package

Models fuel expulsion from the reactor vessel to the reactor

cavity. This includes modeling high pressure melt ejection

and the dispersal of core debris over several volumes

Flow Path Package

Description of interconnection of volumes and the

condensation or evaporation of water along flow paths.

Heat Structures

Package

Models energy transfer to and within structures

Material Properties

Package

Models thermophysical properties of materials needed in the

modeling done in other packages

NCG

Noncondensible

Gas Package

Models noncondensible gases as ideal gases.

PAR

Passive

Autocatalytic

Hydrogen

Recombiner

Package

Calculates the removal of hydrogen from the containment

atmosphere caused by the operation of passive hydrogen

recombiners

Radionuclide

Package

Models release, transport and behaviour of radionuclides

SPR

Containment Spray

Package

Models heat and mass transport between spray droplets and

the containment atmosphere

NEA/CSNI/R(2009)5

139

The MELCOR code categorises radionuclides and other pertinent materials into elemental classes that

exhibit similar chemistry. The default classes are listed in Table 4.1-2. Users of the code can define

additional classes to track particular materials more exactly. The behaviours of all the materials in a class

are tracked based on the predicted behaviour of a particular representative of the class. The representatives

of the default classes are also shown in Table 4.1-2.

Table 4.1-2 Material classes in MELCOR

Class Name Representative Member Elements

1 Noble Gas Xe He, Ne, Ar, Kr, Xe, H

and N

2 Alkali Metals Cs Li, Na, K, Rb, Cs, Fr, Cu

3 Alkaline Earths Ba Be, Mg, Ca, Sr, Ba, Ra, Es, Fm

4 Halogens I F, Cl, Br, I At

5 Chalcogens Te O, S, Se, Te, Po

6 Platinoids Ru Ru, Rh, Pd, Re, Os, Ir, Pt, Au, Ni

7 Early Transition

Elements

Mo V, Cr, Fe, Co, Mn, Nb, Mo, Tc, Ta, W

8 Tetravalents Ce Ti, Zr, Hf, Ce, Th, Pa, Np, Pu, C

9 Trivalents La Al, Sc, Y La, Ac, Pr, Nd, Pm, Sm, Eu,

Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu, Am,

Cm, Bk, Cf

10 Uranium U U

11 More Volatile Main

Group

Cd Cd, Hg, Zn, As, Sb, Pb, Tl, Bi

12 Less Volatile Main

Group

Sn Ga, Ge, In, Sn, Ag

13 Boron B B, Si, P

14 Water H

O H

15 Concrete - -

16 Cesium iodide CsI CsI

The MELCOR code treatment of radionuclides includes:

•

Release of radionuclides from intact fuel and from core debris,

•

Transport and deposition of radionuclide vapors and aerosols through the reactor coolant system,

•

Behaviour of radionuclides and radioactive aerosols in the reactor containment, and

•

Effects of engineered safety systems on the amount of radioactive material that can be released

from the reactor containment.

Each of these aspects of the code is discussed in subsections that follow.

NEA/CSNI/R(2009)5

140

Radionuclide release from fuel

MELCOR considers radionuclide release from fuel both within the reactor vessel and when reactor fuel has

been expelled from the reactor coolant system into the containment. Radionuclide release from fuel within

the reactor vessel can be calculated using one of three closely related models:

•

CORSOR,

•

CORSOR-M, and

•

CORSOR-Booth

All three of these models have an empirical relationship to tests of fission product release from fuel heated

usually out of pile. Most users now prefer the CORSOR-Booth model. Diffusion coefficients in this model

have been upgraded to match well more recent tests such as those being done as part of the PHÉBUS-FP

project.

Ex-vessel release of radionuclides is done with the VANESA model [2] developed based on experimental

data explicitly for this purpose. The model considers fission product release by vaporization into bubbles

of gas sparging through core debris attacking structural concrete. It also considers the mechanical

formation of aerosols due to the bursting of bubbles at the surface of molten core debris. Radionuclide

release can be retarded substantially by the presence of a water pool over the surface of the core debris.

Modeling of this attenuation of the ex-vessel release is akin to that used in MELCOR to model

decontamination of aerosol-laden gas flows through steam suppression pools.

Aerosol nucleation from vapors

MELCOR does not explicitly model the nucleation of aerosols from vapors that cool once they escape core

debris. Extensive numerical experimentation has shown that the details of nucleation are quickly

obliterated by the rapid coagulation of fine particles as predicted by the model of aerosol dynamics

discussed below. Consequently, once vapor reaches saturation, saturation is relieved by condensation on

surfaces and nucleation of particles that are assumed to be in the smallest size bin of the aerosol dynamics

model. Water vapor is allowed by the MELCOR code to form aerosols called ‘fog’ in the code. These

water droplets can interact with other types of aerosol particles.

Aerosol dynamics

Modeling of the agglomeration and deposition of aerosol particles is done in MELCOR using the

MAEROS model. The MAEROS model was developed by Gelbard [3] and is one of the earliest models to

use the sectional method for the prediction of aerosol behaviour in a control volume. MAEROS is both

multisectional and multicomponent. A model for hygroscopicity effects is available in MELCOR. Each

size class considered in MAEROS can have a different chemical composition. Deposition mechanisms

considered in MAEROS are:

•

Gravitational settling,

•

Diffusion,

•

Thermophoresis, and

•

Diffusiophoresis

MELCOR does not currently treat inertial deposition processes.

NEA/CSNI/R(2009)5

141

Vapor deposition

MELCOR considers vapor deposition by condensation. There is also the capability to model vapor

chemisorption onto surfaces. This aspect of the modeling is undergoing re-examination in comparison to

the more comprehensive vapor deposition modeling found in the VICTORIA code which is described

elsewhere in this report.

Engineered safety features

The MELCOR code considers the effects of engineered safety features on aerosols. Specific features that

are modeled include decontamination by:

•

Steam suppression pools

•

Sprays

•

Ice beds

•

Filters

•

Fan coolers

Decontamination by steam suppression pools is done with the SPARC90 model [4] developed for the

NRC. This model calculates removal of both aerosol and iodine gas from gases sparging through

suppression pools. The spray model allows a distribution of droplet sizes but does not now mechanistically

calculate droplet-droplet interactions. Models for decontamination by filters and fan coolers are rather

simple and have not received a great deal of attention.

4.1.1.3 Code status

The MELCOR code is widely used for both regulatory and safety analysis purposes. The code is finding

increased use for the safety analysis of non-reactor nuclear facilities in addition to continued use for

analysis of reactor accidents. The code is still being actively maintained and developed. A new version of

the code was released in 2005. A FORTRAN 95 version of the code is expected in 2006. Following release

of the FORTRAN 95 code version, much of the development effort will be devoted to an extensive

comparison of code predictions to test results and documentation of the validation status of the code.

Because of the flexibility of the MELCOR design and the component nature of the package architecture, it

is felt that the MELCOR code will be adaptable to the assessment of future reactor designs.

References

[1] R.O. Gauntt, R.K. Cole, C.M. Erickson, R.G. Gido, R.D. Gasser, S.B. Rodriquez, and M.F. Young,

MELCOR Computer Code Manuals, NUREG/CR-6119, Volumes 1-3, Sandia National

Laboratories, Albuquerque, NM, December 2000

[2] D.A. Powers, J.E. Brockmann, A.W. Shiver, VANESA: A Mechanistic Model of Radionuclide

Release and Aerosol Generation during Core Debris Interactions with Concrete, NUREG/CR-4308,

Sandia National Laboratories, Albuquerque, NM, July 1986

[3] F.Gelbard MAEROS User Manual, NUREG/CR-1391, Sandia National Laboratories, Albuquerque,

NM, December 1982

[4] P.C. Owczarski and K.W. Burk, SPARC-90: A Code for Calculating Fission Product Capture in

Suppression Pools, NUREG/CR-5765, Pacific Northwest Laboratory, Richland, WA, October 1991

NEA/CSNI/R(2009)5

142

4.1.2 ASTEC (SOPHAEROS and CPA)

4.1.2.1 An overview of the ASTEC/SOPHAEROS code

Introduction

The ASTEC/SOPHAEROS code, developed by IRSN as part of the IRSN-GRS ASTEC integral code [1],

models radionuclide transport in the reactor coolant system (RCS). The main features of the code in its

current version (v1.2) are described here along with an outline of validation activities and development

work. The principal use of ASTEC/S

OPHAEROS arises from the following applications:

−

IRSN has performed a Probabilistic Safety Analysis level 2 for French 900 MWe PWRs using

ASTEC and is set to start a PSA level 2 for 1300 MWe PWRs using ASTEC v1.3;

−

SOPHAEROS-IST 2.0 (corresponding to SOPHAEROS v2.0 with some generic and CANDU-specific

improvements implemented by AECL) is the Canadian Industry Standard Toolset code for analysing

FP transport in the RCS of CANDU reactors [2];

−

ASTEC/SOPHAEROS was used and validated in a European context through the EVITA project [3],

part of the European Commission’s 5

Framework Programme;

−

The ASTEC V1 code has been distributed internationally in the context of the SARNET network of

excellence, part of the European Union’s 6

Framework Programme.

Overview of modelling

The basic element of ASTEC/SOPHAEROS modelling is the control volume where the RCS must be

described as a sequence of volumes, each comprising one or several freely-oriented truncated cones.

Arbitrary geometries can be dealt with involving, e. g., bifurcations, closed loops, etc. Within each control

volume, the modelled vapor and aerosol phenomena are summarised in

Table 4.1-3 where the associated literature source and/or a brief description for each model is provided.

A chemical element is partitioned among its possible chemical species via the use of an independent

thermochemical database. ASTEC/S

OPHAEROS v1.2 operates with a comprehensive database covering

65 elements generating 747 compounds. Recent restructuring of the numerical solver for

ASTEC/S

OPHAEROS V1.2 allows this database to be used routinely (a reduced database exists which was

once of interest in reducing calculation time by covering just over 100 chemical species but can now be

considered obsolete). Each chemical species can exist in one of five physical states: vapor; vapor

condensed on a wall; vapor sorbed on a wall; aerosol and deposited aerosol. The fraction of a species

converted into condensed or deposited states no longer participates in the chemistry where this chemical

inactivation is permanent unless revaporization occurs. Aerosol phenomena are handled by discretizing the

(arbitrary, freely evolving) size distribution over a user-determined, logarithmic grid of up to 50 size

classes. Use of a large number of size classes is particularly important in situations where homogeneous

nucleation, heterogeneous nucleation, or agglomeration is important. S

OPHAEROS is supplied with thermal-

hydraulic conditions via coupling to the ASTEC/CESAR module.

NEA/CSNI/R(2009)5

143

Table 4.1-3 Phenomena modelled in ASTEC/SOPHAEROS v1.2

Mechanism Literature source and/or brief description

V P

a h

p e

o n

u o

r m

vapor-phase chemistry Equilibrium; comprehensive database (800

species).

homogeneous nucleation [4] for condensation rate.

heterogeneous nucleation

(reversible)

Brownian-diffusion-limited mass transfer onto a

sphere, inclusion of Mason effect [5].

sorption on metal-alloy surfaces Empirical velocities as a function of temperature.

condensation on surfaces

(reversible)

Chilton-Colburn analogy [6];

laminar, Nu=3.66 (cylinder); turbulent [7]

A P

e h

r e

o n

s o

o m

l e

sedimentation Stokes’ velocity with Cunningham correction.

turbulent (eddy) impaction [8]

diffusion laminar (Brownian)

turbulent

[9] laminar;

[10] turbulent.

thermophoresis [11]

diffusiophoresis Based on Stefan velocity [17]: [14] for free-

molecular regime; option [15,16] more general.

inertial impaction in bends Hybrid model based on [18] for laminar flow, [19]

for turbulence; option of centrifugal model.

inertial impaction in geometrical

contractions

[23] or [24] depending on contraction angle

agglomeration Brownian,

gravit-ational,

turbulent

[20] continuum regime, [12] free-molecular

regime; [21] gravitational; [22] turbulent.

mechanical resuspension Semi-empirical resultant-force model devised for

the ECART code, retuned by GRS [25];

optional model based on [26].

Numerical method

The mass-balance equations resulting from the intra-volume phenomena combined with inter-volume

transport produce a non-linear system solved numerically by a Newton-Raphson method. To illustrate how

the code deals with the diverse phenomenology involved in the transport process, we take the aerosol state

(state 2 in S

OPHAEROS, hence suffix 2 below) entailing the following mass-conservation equation for a

given aerosol size class

NEA/CSNI/R(2009)5

144

where, taking each right-hand-side term successively,

s is an aerosol source rate,

are flow-dependent transport rates into and out of the volume,

is a compound term representing particle agglomeration into and out of size class

is a compound term representing heterogeneous nucleation onto smaller particles bringing mass

into size class

i and evaporation from class i particles taking mass away,

is the mechanical resuspension rate of class i deposited aerosols (state 4, hence suffix 4),

is the mass rate of formation of aerosols due to homogeneous nucleation of vapor species n where

the Dirac delta is zero except for

i=1, i.e., vapors seed into the smallest size class only.

Further details of the numeric method can be found in the theoretical manual for SOPHAEROS where,

notably, a full description of a number of matrix optimizations used to accelerate the solution scheme are

described.

Validation

A wide variety of data sources is employed for validation. Table 4.1-4 shows the currently-used experiments

where a full reassessment of the experimental database is underway (i.e., some cases may be abandoned while

others will be added).

Table 4.1-4 Current validation matrix

Test Type Project Character Tests Used Main Phenomena

aerosol

phenomena

LACE

consortium

aerosol,

large-scale,

semi-analytical

LA3B

• eddy impaction

• 90°-bend impaction

TUBA-T

IRSN

aerosol,

SGT-scale,

single-effect

TT14,22,24-31

• thermophoresis

TUBA-D

IRSN

aerosol,

SGT-scale,

analytical

TD01-TD12

• diffusiophoresis

• thermophoresis

TRANSAT

IRSN

aerosol,

large-scale,

semi-analytical

TR1, 2, 4-8

• eddy impaction

• 90°-bend impaction

• settling

DEPAT

IRSN

aerosol,

large-scale,

analytical

DEPAT01-03

DEPM01-03

• eddy impaction

vapours

iiiriidicondiaggiif

ifi

Jmmmmmms

,1,4,,2,,,,2,,2,,2

)(

Σ++−++−+=

δττττ

iagg

icond

NEA/CSNI/R(2009)5

145

Test Type Project Character Tests Used Main Phenomena

ADPFF

AEA Tech.

aerosol,

full-scale,

analytical

WT10-23, 25

• eddy impaction

• 90°-bend impaction

• settling

STORM

CEC-ENEL

aerosol,

large-scale,

semi-analytical

SD 04, ISP 40

(in progress)

• thermophoresis

• eddy impaction

•mechanical resuspension

vapor

& mixed

phenomena

DEVAP

IRSN-CEA

vapor,

small-scale,

analytical

8,13-15, 17, 18,

• chemisorption

• condensation

AERODEVAP

IRSN-CEA

aerosol/vapor

small-scale,

semi-analytical

01, 02, 04

•heterogeneous

nucleation

•condensation

•vapor-aerosol

interaction

Falcon

AEA Tech.

simulant fuel,

small-scale,

semi-analytical

Fal-17, 18, 19, 20

•vaporchemistry

• condensation

• vapor-aerosol

interaction

REVAP-ASSESS

Framework

vapor,

small-scale,

analytical

2 VTT tests,

Fal-25

• revaporization

integral

VERCORS HT

IRSN-EDF-CEA

irradiated fuel,

small-scale,

integral

HT1, 2, 3

(in progress)

• full range

HCE

COG

irradiated fuel,

small-scale,

integral

3 (in progress)

• full range

BTF

COG

irradiated fuel,

in-pile,

integral

104 (in progress)

• full range

PHÉBUS-PF

IRSN-CEC-EDF

irradiated fuel,

in-pile,

integral

FPT0, 1, 2, 4

(in progress)

• full range

Development

Future work is determined essentially on the basis of applications requirements and the feedback from

validation activities. Hence, in the shorter term, this work will include implementing models for more

complex geometries (e.g., steam generator secondary side as part of the ARTIST project, [12]) as well as

retention by water volumes that arise in the RCS, e.g., [13]. In addition, verification of all the

thermochemical data and completion of the species considered is in progress where the first stage of this

work was to compare the list of species used by S

OPHAEROS (originating mainly from the SGTE database

[14]) with species included in other database collections. The 65 elements dealt with have been prioritised

NEA/CSNI/R(2009)5

146

in terms of importance and the process of data verification is now well advanced. In the longer term,

development of physical models may address limited gas-phase chemical kinetics; this need arose from

feedback from PHÉBUS analyses leading IRSN to prepare the CHIP experiments studying gas-phase

chemistry for key iodine reactions. In the context of improved accident-analysis capabilities, the study of

cold-leg break sequences requires that steam condensation onto aerosols be examined.

References

[1] J.P. Van Dorsselaere, F. Jacq, H.-J. Allelein, B. Schwinges “ASTEC code status and applications”

US NRC CSARP meeting, Bethesda, USA, 5-7 May 2003 (2003)

[2] L.W. Dickson, R.S. DicksonFission-product transport and retention in the PHTS under accident

conditions”, 20

Annual Conference of the Canadian Nuclear Society, 30 May-2 June 1999,

Montreal, Canada, ISBN 0-919784-80-1 (1999)

[3] H.-J. Allelein, K. Neu, J.P. Van Dorsselaere, K. Müller, P. Kostka, M. Barnak, P. Matejovic, A.

Bujan, J. Slaby “European validation of the integral code ASTEC (EVITA)”

Nucl. Eng. Des. 221, 95-118 (2003)

[4] S.L. Girshick, C.P. Chiu, P.H. McMurray “Time dependent aerosol models and homogeneous

nucleation rates”, Aerosol Sci. Tech. 13, 465-477 (1990)

[5] B.J. Mason “The physics of clouds (2nd ed.)” Clarendon Press, Oxford (1971)

[6] T.H. Chilton, A.P. Colburn “Mass transfer (absorption) coefficients” Industrial and Engineering

Chemistry 26, 1183–1187 (1934)

[7] P.W. Dittus, L.M.K. Boelter “Heat transfer in automobile radiators of the tubular type” Univ.Calif. Pub.

Eng. 2 (13), 443–461 (1930) (re-published in Int. J. Comm. Heat and Mass Transfer 12, 3-22 (1985))

[8] B.Y. Liu, S.K. Agarwal “Experimental observation of aerosol in turbulent flow” J. Aerosol Sci., 5,

145-155 (1974)

[9] P.G. Gormley, M. Kennedy “Diffusion from a stream flowing through a cylindrical tube”, Proc.

Roy. Irish Academy, 52, 163 (1949)

[10] C.N. Davies “Aerosol Science” Academic Press (1966)

[11] L. Talbot, R.K. Cheng, R.W. Schefer, D.R. Willis “Thermophoresis of particles in a heated

boundary layer” J. Fluid Mech., 101, 737-758 (1980)

[12] S. Guentay, D. Suckow, A. Dehbi, R. Kapulla “ARTIST: introduction and first results”, Nucl. Eng.

Des. 231(1), 109-120 (2004)

[13] B.M. Schmitz “Pool scrubbing module SPARC-B/98 for Sophaeros v2mod0_1- model description”

GRS Technical Notice TN-SMZ-00-1 (2000)

[14] Landolt-Börnstein “Numerical data and functional relationships in Science and Technology, Group

IV: Physical Chemistry”, Volume 19: Thermodynamic properties of inorganic materials compiled by

SGTE, Subvolume A, Pure Substances’, Springer (1999)

NEA/CSNI/R(2009)5

147

4.1.2.2 ASTEC-CPA

The aerosol behaviour inside the containment is modelled in the ASTEC module CPA (Containment Part

of A

STEC), which is almost identical with the main COCOSYS modules THY and AFP. For the

description see chapter 4.3.2.

4.1.3 MAAP4

4.1.3.1 Brief description

The Modular Accident Analysis Program (MAAP) Version 4 is a computer code that simulates the

response of light water and heavy water moderated nuclear power plants, during severe accident

sequences. MAAP4 is an integrated code with capabilities to calculate the thermal-hydraulic response of

the core, the Reactor Coolant System (RCS), the containment and the auxiliary buildings, as well as the

fission product release, transport and deposition during postulated severe accident conditions [1]. MAAP4

is used in a wide range of plant evaluations including design basis evaluations, Level 1 and Level 2 success

criteria for probabilistic risk assessments, radiological source term assessments, and accident management

guidance evaluations. MAAP4 also includes a graphical interface, MAAP4-GRAAPH, enabling the user

to interactively interface with the code during execution, to modify the status of on-site power, pumps,

valves, etc., as well as, to directly observe the results [2].

4.1.3.2 Status

MAAP has been developed and maintained by Fauske & Associates Incorporated (FAI), since the

beginning of the code in 1981, under the sponsorship of the Electric Power Research Institute (EPRI) and

the MAAP Users Group (MUG). The code continues to be developed and maintained by FAI. Recent

versions include MAAP4-CANDU, that extend MAAP4 applicability to CANDU power plants. It should

be noted that the CANDU specific versions do not involve any changes to the previous fission product /

aerosols transport capabilities. Validation of MAAP4 was performed against HDR experiments, CORA

tests, TMI-2 accident, CSTF tests, FEBUS FPT0 test, ORNL VI test series, SFD tests at INEL, AP600

OSU tests, and LOFT experiments [3].

4.1.3.3 Modelling

Brief Description / Literature Source

Reactor Components

Modelled

• Reactor Core

•

Reactor Cooling System (RCS)

•

Containment

•

Reactor auxiliary buildings

Geometry Idealization

• A network of nodes inter-connected by flow paths

Inter-nodal Transport

• Transport of corium, water and gases including aerosols through

flow paths

Thermal hydraulics

• Modelled by mass and energy equations

•

Flow rates are determined from quasi-steady momentum balances

•

Pressure, temperatue and micellaneous thermodynamic properties

are determined from mass and energy of gaseous components in a

node.

Fission Product

Transport

• Fission products can exist in the solid, liquid, and vapor forms.

•

Twelve fission product compounds and three chemical compounds

NEA/CSNI/R(2009)5

148

Aerosol Transport

which affect the pH value in the water pool [4] are tracked by the

code.

•

In the primary system, fission products are not dissolved in the

water and hence are not transported by water, but only by carrier

gas.

•

In the containment, the dissolution of fission products in water pools

is modelled, and hence they are transported by both water and

carrier gas.

•

Fission product vapor - aerosol condensation/evaporation is

modelled by maintaining phase equilibria [1].

•

Vapor - surface condensation / evaporation is modelled by diffusion

[1].

•

Simultaneous aerosol growth by agglomeration and removal by a

number of aerosol deposition mechansisms are modelled by

empirical correlations specially developed for the code, based on a

principle of similarity and scaling [1, 5, 6,]. This empirical

approach differs from the traditional, time-consuming, detailed

calculations involving a solution of the Smoluchowski equation for

the evolution of particle size distribution, but results in reduced

computational time. The empirical approach considers the

Brownian and gravitational agglomeration and the aerosol removal

mechanisms listed below:

•

Sedimentation

•

Diffusiophoresis

•

Thermophoresis

•

Inertial impaction

References

[1] R.E. Henry, C.Y. Paik, M.G. Plys, “MAAP4 – Modular Accident Analysis Program for LWR Power

Plant - Code Structure and Theory, Computer Code Manual” Fauske & Associates Inc., prepared

for: Elecric Power Research Institute (EPRI), Volume 2, May 1994

[2] R.E. Henry, C.Y. Paik, M.G. Plys, “MAAP4 – Modular Accident Analysis Program for LWR Power

Plant – MAAP4-GRAAPH, Volume 4, Rev 0.0 / 0.1, May 1994 (Some Rev 0.1 updates to June

1995)

[3] R.E. Henry, C.Y. Paik, M.G. Plys, “MAAP4 – Modular Accident Analysis Program for LWR Power

Plant – Major Benchmarks” Fauske & Associates Inc., Volume 3A, Rev0.0 and 0.1, May 1994

[4] FAI Internal Memo from C.D. Wu to R .J. Hammersley, “Implementation of pH Model in MAAP4”,

October 8, 1993

[5] “Technical Support for Issue Resolution” Fauski and Associates Incorporated Report 85.27, July,

1985

[6] M. Epstein, P.G. Ellison, and R .E. Henry “Collelations of Aerosol Sedimentation” J. Colloid

Interface Sci., Vol. 113, No:2, October, 1986

NEA/CSNI/R(2009)5

149

4.1.4 THALES 2

The THALES-2 code [1] is an integrated severe accident analysis code developed at the Japan Atomic

Energy Research Institute (JAERI) in order to simulate the accident progression and transport of

radioactive materials for probabilistic safety assessment (PSA) of a nuclear power plant (NPP).

In 1982, JAERI developed, as a first step, the computer code system THALES [2] (T

hermal-Hydraulic

nalysis of Loss of Coolant Emergency Core Cooling and Severe Core Damage) for the analysis of

accident progression. In 1988, the code was combined with the ART (Analysis of Radionuclide Transport)

code [3] developed also by JAERI and the THALES/ART code system started. After that, the code system

was improved by coupling the radionuclide transports models with the thermal hydraulic ones and a

prototype of single code, namely, the THALES-2 code [1, 4] was completed in 1991. Then, the

abbreviation THALES was changed to the T

hermal Hydraulics and radionuclide behaviour Analysis of

ight water reactor to Estimate Source terms under severe accident conditions.

The code has also been validated through analyses of experiments [5] and comparison with other computer

codes [6, 7]. The THALES-2 code currently consists of BWR and PWR versions. Since the development

of BWR version [8] advances compared with the PWR version, the outlines of BWR version are described

below.

4.1.4.1 Overview of THALES-2 BWR

Accident progression

The THALES-2 BWR code [1, 7, 8]

calculates various physical phenomena such as thermal-hydraulics,

melt progression and radionuclide behaviour under the severe accident conditions. While progression of

core meltdown accidents are markedly different depending on the initiating event and the operability of

safety features, the THALES-2 BWR code postulates the accident progression such as described below.

Either the direct release of the primary coolant due to a break at system boundary or the coolant boil

off ascribed to insufficient cooling during a transient causes decrease of the primary coolant

inventory. When no cooling function such as those by Emergency Core Cooling System (ECCS) is

unavailable, core will be uncovered.

Fuel rods begin heatup when they are exposed to the steam over the mixture level. When the

temperature of a fuel rod rises high enough for activating Zr-H

O reaction, cladding oxidation occurs

and hydrogen is generated. Since this reaction is exothermic, it accelerates the core heatup. Further

heatup of the core results in core meltdown; fuel rods lose their own shape by melting or

fragmentation, and slump into lower part of the core. These slumping debris boil off the water

remaining in the lower part of the pressure vessel and heat up the support plate and the bottom bead.

Heated up support plate and bottom head lose their strength and finally the bottom head

melt-through occurs. After pressure vessel melt-through, corium-core debris together with structure

materials drops into the reactor cavity.

3. Containment pressure and temperature rise with blowdown of the primary coolant, gas generated by

concrete decomposition and hydrogen burning. After the reactor vessel failure, the molten material

falls down into the reactor cavity and core/concrete interaction is initiated. During the core/concrete

interaction, non-condensable gases are produced by the core/concrete interaction and the pressure in

the containment atmosphere begins to increase. When the sufficient cooling with the engineered

safety features such as the containment spray is unavailable, containment failure will occur and the

gases in the containment are released to the environment.

NEA/CSNI/R(2009)5

150

Environment of THALES-2 BWR

THALES-2 BWR code requires information for a configuration of a plant including geometrical data,

accident sequence (operational conditions of ECCS and operator actions), initial conditions of

thermal-hydraulics and so on. THALES-2 BWR code gives trends of thermal-hydraulic condition, mass

distribution of radionuclides, timing of events such as containment failure, and source terms.

THALES-2 is a highly modularised code system. The main part is supported by the subroutines, LOGIC1,

UCL2 and SPLPACK. The UCL2 is for unit conversion, which enables the user to prepare input data with

any unit, and the LOGIC1 for modeling the control circuit of safety-related system and for event control

which defines the accident sequence. The SPLPAC is used to support graphic functions of THALES-2.

Structure of THALES-2 BWR

In the THALES-2 BWR code, calculations of thermal-hydraulics, release of radionuclide from fuel,

radionuclide deposition and transport are performed at a given infinitesimal time increment δt. Fig. 4.1-1

shows data transfer inside the code. Firstly, thermal-hydraulic calculations are performed including the

melt-relocation. After that, release of radionuclides from fuel and molten materials are calculated. By using

these results, deposition and transport of radionuclide are calculated. Based on the results of radionuclide

distribution, decay heat in each phase in the volume are calculated and delivered to the thermal-hydraulic

calculation at the next time step t + δt.

THALES-2 uses three kinds of time increments for: 1) thermal-hydraulic in the containment system,

2) thermal-hydraulic in the reactor coolant system, and 3) radionuclide behaviour calculation. These three

time increments are automatically determined in the code.

4.1.4.2 Plant Model in THALES-2 BWR

Multiple-volume presentation of BWR plant

The multiple volumes representation is used in the THALES-2 BWR code to describe the mass/energy

transport radionuclide behaviour in LWRs plants. In this method, the primary and the containment systems

are divided into many volumes. Fig. 4.1-2 shows a typical configuration of volumes for a BWR with a

Mark-I containment

NEA/CSNI/R(2009)5

151

Fig. 4.1-1 Data transfer in THALES-2 BWR code system

NEA/CSNI/R(2009)5

152

Safety relief valve

Containment

Steam dome

Plenum

Core

Lower plenum

Loop ALoop B

Pedestal

Vent

pipe

Vent

pipe

Vacuum

breaker

Vacuum

breaker

Suppression pool

Flow path for fluid and FP

Environment

Reactor building

Safety relief valve

Containment

Steam dome

Plenum

Core

Lower plenum

Loop ALoop B

Pedestal

Vent

pipe

Vent

pipe

Vacuum

breaker

Vacuum

breaker

Suppression pool

Flow path for fluid and FP

Environment

Reactor building

Reactor coolant systems

Recirculation loop A

Recirculation loop B

Lower plenum

Core

Upper plenum

Steam dome

Downcomer

Control volumes

Containment systems

Pedestal

Drywell

Wetwell

Vent pipes

Others

Reactor building

Environment

Reactor coolant systems

Recirculation loop A

Recirculation loop B

Lower plenum

Core

Upper plenum

Steam dome

Downcomer

Control volumes

Containment systems

Pedestal

Drywell

Wetwell

Vent pipes

Others

Reactor building

Environment

Fig. 4.1-2 An example of multi-volume representation of a BWR/MARK-I plant

Concept of system, volume, junction and path

In the multiple-volume presentation, a plant is represented by a combination of 'system', 'volume', 'tank',

'path' and 'junction' which are briefly described below.

Volume: A unit of region such as recirculation loops, upper plenum and core,

System: A set of volumes, which have the same pressures are connected by junctions

Tank: A region in which coolant inventory changes such as condensate storage tank,

Junction: A model element to calculate mass/energy transport among volumes,

Path: A model element to calculate mass/energy transport among systems.

A concept of ‘system’ and ‘volume’ was introduced to realise fasten running capability without losing

accuracy. The system consists of volumes, which are combined by junctions and have the same pressure.

NEA/CSNI/R(2009)5

153

Some of 'system's and 'volume's in the THALES-2 BWR code are fixed to get fast running capability as

follows:

System

Volume

Reactor Coolant System Downcomer (DC), Lower Plenum (LWPL)

Core (CORE), Upper Plenum (UPPL)

Steam Dome (STMD),

Loop A (LOPA), Loop B (LOPB)

Drywell System Drywell (DRYW), Pedestal (CAVI)

Vent Pipe (VENT)

Wetwell System Wetwell (WETW)

User can define additional systems and volumes by input data. Each volume is further divided into a gas

region and a liquid region by a mixture level. The geometry of a volume maybe represented by several

cylindrical segments to calculate the correct mixture level.

Coolant sources outside systems such as the condensate storage tank are modeled by 'tank' which treats

only coolant inventory change. Fluid flow at a system boundary such as coolant injection by ECCS, break

flow and coolant release through safety and relief valves are modeled by 'path' which connects a volume

within a system and a tank outside a system. The path flows are used as a boundary condition of the

hydraulic calculation within a system, while the junction flow rates are given as a result of the hydraulic

calculation within the system.

As for radionuclide in volumes, six phases are defined to describe mass distribution of radionuclides as

follows:

Phase 1: Radionuclide of vapor form in gas region

Phase 2: Radionuclide of aerosol form in gas region

Phase 3: Radionuclide in liquid region or on the floor

Phase 4: Radionuclide of condensed form at structure surface

Phase 5: Radionuclide of aerosol form at structure surface

Phase 6: Radionuclide of absorbed form at structure surface

4.1.4.3 Outlines of radionuclide behaviour models

Radionuclide group

In the THALES-2 BWR, 20 radionuclides are classified into several groups (maximum 10) in terms of

their chemical characteristics. Typical elements or compounds of each group are Kr-Xe, CsI, CsOH, Te,

Sr, Ru, La and other particulates. An aerosol form is assumed in the code for Sr, Ru and La because their

vapor pressures are very low in severe accident conditions.

Total mass of radionuclides in the core are implemented by table data in the code for typical BWR plant.

The core inventory of plant concerned is given by a ratio of thermal power and inventory. In the

THALES-2 BWR code, a core region is represented by the groups of fuel assembly (maximum 5) and

vertical nodes (maximum 25). Distribution of radionuclide inventory in the core region is automatically

allocated to each core region in the code.

NEA/CSNI/R(2009)5

154

Radionuclide release from fuel

For radionuclide release from fuel before the reactor vessel failure (in-vessel), the CORSOR model and the

new model with pressure effect proposed by the VEGA program [9] are applied to calculate release rates of

radionuclides. After the vessel failure (ex-vessel release), the empirical model proposed by W.Murfin is

used to calculate generation rates of aerosols of concrete components during core/concrete interaction. In

addition, the CORSOR model is also applied to calculate release rates of radionuclides during the ex-vessel

release.

Radionuclide behaviour

The radionuclide transport models are illustrated in Fig. 4.1-3. In this code, radioactive materials can take

the form of gas, aerosol, deposit on structure walls and floors, and solution in water. The code solves the

governing equations for multi-component aerosol, taking into account the size growth by agglomeration

and condensation/evaporation of steam and volatile materials on the aerosol. Models are provided for

various transport processes, including the condensation/evaporation and chemical absorption of the gas

species at structure surfaces, deposition of aerosol to walls and floors, removal by sprays and filters,

scrubbing by water pools, and convection by liquid as well as gas flow.

Brownian

Agglomer ati on

Absorption

Turbulential

Agglomeration

Gr avitational

Agglomeration

Diffusion

Evaporation

Laminar,

Turbulent

Resuspension

Diffusion

Thermophoresis ,

Diffusiophore sis

Source

Interception

Im paction

Partition

Diff usi oph or esis

Decay

Filter

Leak

Filter

Source

Spray

Droplet

Partition

Scrubbing

Bubble

Flashing

Gas-Liquid

Partition

Gravitational

Settling

FP Transportation

by Gas Flow

FP Transportation

by Liquid Flow

Motion of Gaseous FP

Motion of Aerosol

Condensation

Diffusion

Gr avitational

Settling

Dissolution

Brownian

Agglomer ati on

Absorption

Turbulential

Agglomeration

Gr avitational

Agglomeration

Diffusion

Evaporation

Laminar,

Turbulent

Resuspension

Diffusion

Thermophoresis ,

Diffusiophore sis

Source

Interception

Im paction

Partition

Diff usi oph or esis

Decay

Filter

Leak

Filter

Source

Spray

Droplet

Partition

Scrubbing

Bubble

Flashing

Gas-Liquid

Partition

Gravitational

Settling

FP Transportation

by Gas Flow

FP Transportation

by Liquid Flow

Motion of Gaseous FP

Motion of Aerosol

Condensation

Diffusion

Gr avitational

Settling

Dissolution

Fig. 4.1-3 Radionuclide transport models in THALES-2 BWR

NEA/CSNI/R(2009)5

155

References

[1] M. Kajimoto et al."Development of THALES-2, A Computer Code for Coupled Thermal-Hydraulics

and FP Transport Analyses for Severe Accident at LWRs and Its Application to Analysis of FP

Revaporization Phenomena" Proc. Int. Topical Mtg. on Safety of Thermal Reactors, Portland, pp.

584-592 (1991)

[2] K. Abe et al."Overview of Development and Application of THALES Code System for Analyzing

Progression of Core Meltdown Accident of LWR’s" Proc. 2

Int. Symp. on Nucl. Power Plant

Thermal Hydraulics and Operations, Tokyo (1986)

[3] T. Ishigami et al."User’s Manual of ART Code for Analyzing Fission Product Transport Behaviour

during Core Meltdown Accident" JAERI-M 88-093 (1988) (in Japanese)

[4] K. Muramatsu et al."Modeling Improvement and Application of the Integrated Severe Accident

Analysis code THALES-2 for Assessment of Accident Management Strategies" Proc. Int. Conf. on

Probabilistic Safety Assessment Methodology and Applications (PSA'95), pp. 216-221, Korea,

(1995)

[5] A. Hidaka et al."Comparative Study of FP Deposition in WIND Project by ART and VICTORIA"

Proc. Int. Conf. on Probabilistic Safety Assessment Methodology and Applications (PSA'95), pp.

241-246, Korea, (1995)

[6] S. Kondo et al."Comparison of Analytical Models and Calculated Results of Source Term

Evaluation Codes" OECD/CSNI PSA Workshop, NUREG/CP-0115, pp. 273-287 (1990)

[7] A. Hidaka et al."Comparative Study of Source Terms of a BWR Severe Accident by THALES-2,

STCP and MELCOR" Proc. 1992 National Heat Transfer Conf., HTC-Vol. 6, pp. 408-416 (1992)

[8] J. Ishikawa et al."Systematic Source Term Analyses fro Level 3 PSA of a BWR with Mark-II Type

Containment with THALES-2 Code" 10

Int. Conf. on Nucl. Eng. (2002)

[9] A. Hidaka et al. Proposal of Simplified Model of Radionuclide Release from Fuel under Severe

Accident Conditions Considering Pressure Effect"J. Nucl. Sci. Technol. 41(12), pp. 1192 - 1203

(2004)

4.1.5 ECART

ECART (ENEL Code for the Analysis of Radionuclide Transport) is an integrated primary circuit and

containment code, originally created in 1989 for nuclear power plant severe accident analysis, but now also

applied to fusion plants, industrial fires etc. The utilities ENEL and EdF contributed to its intial

development, while its wider application for risk analysis is being carried forward by CESI. The code

calculates the two-phase thermal-hydraulics for both the primary system and the containment and for

material transport calculates aerosol and vapor transport together with chemical equilibrium and specific

reactions between the gaseous phase and important structural surfaces. The principal models used are listed

in Table 4.1-5.

An interesting feature of the code useful in experimental analysis is its ability to accept incomplete

thermal-hydraulic data e.g. data specified only at certain junctions or boundaries, and then to use its

internal calculation capabilities to complete the thermal-hydraulic conditions required by the aerosol and

chemistry sections of the code. Validation studies have used data from the ATT-Marviken V, LACE,

DEMONA, VANAM, STORM and PHÉBUS programmes. Recent applications of ECART have been to

NEA/CSNI/R(2009)5

156

aerosol resuspension and chemical reactions in PWR circuits, the safety analysis of IRIS, and transient

analyses of the fusion experiment ITER.

Table 4.1-5 Circuit and containment models

Mechanism Literature source and/or brief description

Thermal-hydraulics

Transport of carrier gas/liquid

mass, momentum, energy

Control volumes, each with a liquid and a gas volume

in equilibrium

1-d and 2-d connection of volumes

Bulk and wall condensation split

Exchanges with structures Included

Vapor/gas transport Secondary gases e.g. nobles, accounted for

Sprays and sprinklers

Ad-hoc Lagrangian model accounting for droplet size

distribution and injection speed

Pool scrubbing Included within control volume

Gas combustion (hydrogen

and others)

Accounted for through equilibrium chemistry

Fires and explosive aerosol

clouds

Models of pyrolysing solid surfaces, pool fires and

detection of explosible aerosol clouds

Radiative heat transfer from

flames

View factors among flames and structures calculated

with Montecarlo method

Aerosol cloud absorption accounted

Decay heat 13 most powerful elements accounted for

Vapor phenomena

Vapor-phase chemistry

Equilibrium with 126 reacting species (including

carrier gases)

Homogeneous nucleation Not modelled (source seed required)

Heterogeneous nucleation Not modelled (source seed required)

Sorption on surfaces (one-

way)

Selected species/surface combinations e.g. irreversible

sorption of I, I

, HI, CsOH, Te and Te

vapors on steel

Condensation/Evaporation

onto/from surfaces and

aerosol particles

Calculated by diffusion equations

Aerosol phenomena

Transport

Well-mixed within each volume. Corrections for

components with concentration gradients e.g. long

pipes

Discretised size distribution; simplified multi-

component description (composition accounted for

each size bin in each volume, for both airborne and

deposited particles)

Aerosol shape Aerodynamic and collision shape factors

Particle growth Includes hygroscopic behaviour, Kelvin effect

Settling Stokes and non-Stokes regimes

Turbulent impaction Liu-Agarwal data

NEA/CSNI/R(2009)5

157

Diffusion Davies, Gormley-Kennedy

Thermophoresis Brock correlation with Talbot coefficients

Diffusiophoresis Schmidt-Waldmann

Bend impaction

Stokes and non-Stokes regimes; size-dependent

trapping in narrow bends

Agglomeration Brownian (Smoluchowski)

Gravitational

Turbulent (Saffman and Turner)

Mechanical resuspension Modelled through experimental correlation [1]

Aerosol fall-back Accounted for

Scrubbing in water sumps

Lagrangian model accounting aerosol depletion within

rising bubble - [2]

References

[1] F. Parozzi A fast-running model for physical aerosol resuspension accounting for STORM

experimental results Int. Mtg. “Best-Estimate” Methods in Nuclear Installation Safety Analysis

(BE-2000), Washington Dc, USA, November 2000

[2] F. Parozzi, S. Paci ECART code integrated approach for thermal-hydraulic source term analysis 4

Int. Conf. on Nuclear Option in Countries with Small and Medium Electricity Grids, Dubrovnik,

Croatia, June 2002

4.1.6 APROS SA

4.1.6.1 Introduction

APROS (Advanced PROcess Simulator) is a multifunctional advanced process simulation software

developed by Fortum and the Technical Research Centre of Finland. It is used for process and automation

design, developing operation procedures and training at nuclear power plants, as well as for the full

spectrum of the transient and accident analyses. APROS has passed an extensive validation programme.

Most accident analyses for the new operating license of Loviisa NPP at uprated reactor power are made

using APROS. The recenct versions of APROS include components

for detailed containment analysis, as well.

APROS software is applicable also in engineering and operator

training of thermal power plants.

The versatile APROS simulation software suits as well for purposes

of building engineering or training simulators and plant simulation

models.

4.1.6.2 Severe accident extension

During the implementation of SAM for the Loviisa NPP, a tool for

simulating severe accident phenomena was seen necessary. Specific

phenomena, as in-vessel retention and hydrogen mitigation and

NEA/CSNI/R(2009)5

158

fission product behaviour in an ice condenser containment, were to be introduced into the system. The

selection of using APROS software as a platform for the severe accident was a natural choice for Fortum.

The development of APROS SA [1] aims at a severe accident simulator for the Loviisa NPP that is capable

of calculating the severe accident sequences faster than real time. The necessary phenemona are the core

relocation, molten pool formation, core slumping into the lower plenum, in-vessel retention of the molten

pool, hydrogen formation and fission product release from the overheating core, hydrogen mitigation with

passive autocatalytic recombiners in the ice condenser containment, fission product retention in the RCS

and in the containment, as well as in the long narrow pipelines in the contaiment bypass routes. In addition,

the radiation levels at the plant site are to be included into the simulator.

The visualisation of the calculation results plays an important role in training the severe accident

phenomena. In Fig. 4.1-4 can be seen an example of temperature profile in the Loviisa NPP ice condenser

containment at two situations.

Fig. 4.1-4 Temperature distribution in Loviisa NPP ice condenser containment at 0 s (left) and 3 s,

(right) after a 500 cm2 break in cold leg calculated by APROS SA. (Color gradient is from

blue to red to yellow with blue corresponding 20 °C, red 60 °C and yellow 100 °C) [1]

4.1.6.3 Aerosol modelling

The aerosol modelling in APROS SA is based on fixed, user defined aerosol size classes, where no aerosol

growth is assumed. Each fission product group acts as individual substance with their own aerosol material

densities. Currently, for each FP group there are three different aerosol size groups, one group for gaseous

species, and one for deposited mass. The deposited mass can be either on the dry surfaces of a simulation

node or dissolved in the water within the node. The FPs are transported between the nodes along with the

flowing gas or water.

NEA/CSNI/R(2009)5

159

The fission product groups in APROS SA are:

−

Noble gases (Kr, Xe)

−

Iodine (I, Br)

−

Cesium (Cs, Rb)

−

Tellurium (Te)

−

Metallic compounds (not released from the core)

−

Oxidic compounds (not released from the core)

The TH condition calculated by the other models in APROS or APROS SA are used for FP calculation,

ann the FP calculation gives the decay power distribution as input to the other models. The distribution of

metallic and oxidic compounds not released from the reactor core is used for the decay power distribution

calculation between the metallic and oxidic molten pools at the bottom of the RPV lower head.

The fission product release from the reactor core is based on the CORSOR-M model [2], and the released

FPs (other than noble gases) are instantaneously in form of aerosol after the release. This means that the

vapor phase chemistry in the RCS is not considered.

The aerosol processes in the APROS SA are those that play the most important role in the case of the

Loviisa NPP:

−

Sedimentation

−

Thermophoresis (Talbot correlation; independent of aerosol particle size)

−

Diffusiophoresis (independent of aerosol particle size)

−

Turbulent (eddy) impaction in the RCS and bypass pipelines (modified model to meet the coupling

with the resuspension model by [3])

−

Revaporisation

−

Resuspension [3]

−

Spray droplet washout (impaction, interception, diffusion)

−

Inter-volume aerosol fall to lower levels (similar to sedimentation)

The FP calculation has no other effect on the other parts of calculation than through providing the decay

power distrubution. Iodine chemistry is modelled as simple equilibrium calculation with fixed water pool

pH and temperature dependent partitioning of gaseous iodine (I

). The formation of organic iodine is not

accounted for.

The limitation of the FP modelling is that it cannot be used for detailed analyses on FP behaviour. The

essential information from the APROS SA fission product calculation is qualitative data with a reasonable

accuracy for training purposes of severe accident simulations and usage of EOPs for SAM.

The aerosol models have been validated against different experimental data. The revaporisation and

deposition in the pipe systems have been verified against the data from the RCS part in the ISP-46 from the

PHÉBUS FPT1 (see Fig. 4.1-5) [4]. The aerosol behaviour in the containment has been compared with the

results from the VICTORIA facility (a scale model of the Loviisa ice condenser containment) [5, 6]. The

NEA/CSNI/R(2009)5

160

resuspension model validation is currently under work, and the results are to be compared with the results

from the Horizon [7] and PSAERO [8] facilities.

Fig. 4.1-5 Comparison of APROS SA aerosol calculation with VICTORIA experimental results [6]

4.1.6.4 Future plans

In future, the models for aerosol size distribution in the RCS calculated from the aerosol mass

concentration are to be implemented, which probably acclerates the calculation speed significantly. There

is a very high number of calculation nodes in a typical simulation environment in a detailed plant model,

and therefore a reduction of individually calculated aerosol groups is foreseen.

The radiation level calculation from the FP distribution is one assential part of a severe accident simulator,

as it provides information for simulation of radiation measurement device used during accidents. The

essential information on the coefficients to calculate the radiation levels from the activities is to be

implemented rather in the user interface of APROS SA than directly into the FP calculation.

References

[1] E. Raiko, K. Salminen, P. Lundström, M. Harti, T. Routamo Severe accident training simulator

APROS SA. Proceedings of NURETH-10 The 10

International Topical Meeting on Nuclear

Reactor Thermal Hydraulics, October 5-11, 2003, Seoul, Korea, 2003

[2] M. Silberberg, J.A. Mitchell, R.O. Meyer, C.P. Ryder Reassessment of the Technical Bases for

Estimating Source Terms Final Report. U.S. Nuclear Regulatory Commission, Office of Nuclear

Regulatory Research, Washington, July 1986. NUREG-0956

0,00 %

14,19 %

12,35 %

6,18 %

3,12 %

4,06 % 3,89 %

0,19 %

0,60 %

0,05 %

0,03 %

0,76 %

2,39 %

2,12 %

3,55 %

0,00 %

2,47 %

0,04 % 0,02 %

0,63 %

2,04 %

0,96 %

2,12 %

0,00 %

2,11 %

3,12 %

28,74 %

0,02 %

2,58 %

1,49 %

0,16 %

VICTORIA 62

Total injected mass

in the simulation 2350 mg

in the experiment 2412 mg

0,00 %

14,19 %

12,35 %

6,18 %

3,12 %

4,06 % 3,89 %

0,19 %

0,60 %

0,05 %

0,03 %

0,76 %

2,39 %

2,12 %

3,55 %

0,00 %

2,47 %

0,04 % 0,02 %

0,63 %

2,04 %

0,96 %

2,12 %

0,00 %

2,11 %

3,12 %

28,74 %

0,02 %

2,58 %

1,49 %

0,16 %

VICTORIA 62

Total injected mass

in the simulation 2350 mg

in the experiment 2412 mg

Node 26

100

60 90 120 150 180 210 240 270 300 330 360

Time (min)

Concentratio (mg / m

)

small

medium

large

Total

TEOM, line 1

Node 27

60 90 120 150 180 210 240 270 300 330 360

Time (m in)

Concentratio (mg / m

)

small

medium

large

Total

TEOM, line 2

NEA/CSNI/R(2009)5

161

[3] A. Auvinen, J.K. Jokiniemi Modelling resuspension of particle layer in an internal tube flow

Presented in the European Aerosol Conference (EAC04), September 6-10, 2004, Budapest, Hungary

[4] V. Karttunen Validation of a fission product revaporization model Helsinki University of

Technology, Master's thesis, Espoo, January 10, 2005. 81 p. + Apps, 2005a

[5] V. Karttunen APROS SA fission product validation in containment nodes with CsOH Fortum

Nuclear Services Ltd, Report 24.1.2005. TERMO-435, 2005b

[6] V. Karttunen APROS SA fission product validation in containment nodes with Ag Fortum Nuclear

Services Ltd, Report 25.4.2005. TERMO-466, 2005c

[7] T. Routamo, J. Jokiniemi, J. Mäkynen Aerosol Deposition in Horizontal Steam Generator Tubes in

Severe Accident Conditions Proceedings of NURETH-10 - The 10

International Topical Meeting

on Nuclear Reactor Thermal Hydraulics, October 5-11, 2003, Seoul, Korea, 2003

[8] A. Auvinen

, J.K. Jokiniemi

,. A. Lähde

, T. Routamo

, P. Lundström

, H. Tuomisto

, J. Dienstbier

S. Güntay

, D. Suckow

, A. Dehbi

, M. Slootman

, L. Herranz

, V. Peyres

, J. Polo

“SG tube rupture

(SGTR) scenarios” Nuclear Engineering and Design 235 (2005) 457–472

VTT Processes, Biologinkuja 7, P.O. Box 1602, VTT Espoo 02044, Finland

Fortum Nuclear Services, Vantaa, Finland

Nuclear Research Institute Rez plc, Czech Republic

Paul Scherrer Institute, Villigen-PSI, Switzerland

Nuclear Research and Consultancy Group, Arnhem, Netherlands

Centro de Investigaciones Energeticas, Medioambientales y Tecnologicas, Madrid, Spain

4.2 Circuit

4.2.1 VICTORIA

The VICTORIA computer code [1, 2] attempts to predict the behaviour of radionuclides in the reactor

coolant system under severe accident conditions. It addresses the release of radionuclides, the chemical

forms adopted by the radionuclides, as well as the formation, growth and deposition of aerosol particles.

The computer code was not intended for the routine analysis of severe reactor accidents. The idea for the

development of a mechanistic treatment of radionuclide behaviour in the reactor coolant system arose in

the aftermath of the accident at Three Mile Island. It had become obvious that simplistic treatment of the

radionuclide releases during core degradation in terms of noble gases, gaseous iodine and general

particulate could not lead to results of adequate accuracy. At the same time, a wealth of experimental

evidence of chemical interactions of radionuclides in the reactor coolant system environment under

accident conditions was coming available from research programs in the USA, UK, and France. Simply

"patching" existing radionuclide transport models with

ad hoc chemical models became increasingly

unsatisfactory. Experimental results, though quite significant, were obtained for limited conditions and

selected species. A broader analytic approach was needed that could account for experimental findings and

extrapolate these findings to more drastic conditions expected to develop in some classes of risk-significant

accidents and even to other chemical systems that had not received individual study. The VICTORIA

model was developed to address this need. It was intended to be used to interpret experiments and to

extrapolate the results to accident conditions. The objective of the detailed mechanistic analyses was to

determine important phenomena that had to be represented in systems level accident analysis models such

as MELCOR.

NEA/CSNI/R(2009)5

162

VICTORIA is a lumped computational node computer code. Computational nodes can be quite small in

comparison to those used in systems-level reactor accident analysis codes. Key elements of the VICTORIA

computer code are models of:

•

Fission product release from degrading reactor fuel and precious metal control rods

•

Chemical speciation of radionuclides and other materials vaporized during reactor core

degradation

•

Formation and growth of aerosol particles

•

Deposition of vapors on surfaces along the flow paths through the reactor coolant system

•

Deposition of aerosols on surfaces along the flow paths through the reactor coolant system

•

Resuspension of deposited aerosol

•

Revaporisation of radionuclides on surfaces in the reactor coolant system

The VICTORIA computer code does not attempt to predict the inventories of radionuclides in reactor fuel

at the start of an accident. These inventories are usually obtained using the ORIGEN [3] computer code.

The VICTORIA computer code does not try to predict the flows through the reactor coolant system or the

temperatures of the structures along the flow paths. These are obtained from other models such as

MELCOR.

Synoptic descriptions of the models in VICTORIA are presented in the sections that follow.

Radionuclide release from the fuel

The VICTORIA model was not originally intended to address issues of radionuclide release from the fuel.

Other models like CORSOR [4] exist for making such predictions. It was found, however, convenient to

include release models into the computer code. Still this is an area that has received the least model

development attention.

The code considers two configurations of the reactor fuel for the purposes of estimating radionuclide

release:

−

intact fuel in rod geometry that may or may not have cladding still in place

−

rubble that may be molten produced as the core loses its geometry

The distinct models of radionuclide release in these configurations are discussed in subsections below.

Release from intact fuel

For the purposes of estimating release from fuel in rod-like geometry, VICTORIA originally had two

models. One was a very mechanistic model patterned after the FASTGRASS [5] model developed at

Argonne National Laboratory. The other was the more empirical model based on the well-known Booth

solution to the diffusion problem and parameterised based on results of tests of radionuclide release from

irradiated fuel done at Oak Ridge National Laboratory. It was found that the mechanistic model required

input that is not routinely available. Furthermore, the model was not especially reliable. Based on a

recommendation of a peer review group the very mechanistic release model was deleted from the

VICTORIA computer code.

NEA/CSNI/R(2009)5

163

The Booth-type modeling of fission product release has been replaced in the more recent version of

VICTORIA by a more mechanistic model that considers:

•

diffusive release of fission products from fuel grains

•

transport of fission product vapors through the pore network of the fuel

•

transport of fission product vapors through the gap between the fuel pellets and the cladding to a

break in the cladding when the cladding is present

•

mass transport of fission product vapors away from the break in the cladding or other free

surface.

Diffusion coefficients for fission products in fuel grains are taken from Matzke [6]. Gas phase diffusion

coefficients are calculated based on the well-known Chapman Enskog expression for binary diffusion

coefficients assuming a Lennard-Jones intermolecular potential. Binary diffusion coefficients are combined

to define effective diffusion coefficients for multicomponent gaseous mixtures. The assumption of

Lennard-Jones intermolecular potentials is a common one for computer codes like VICTORIA. Two issues

are associated with this assumption. First, the Lennard-Jones intermolecular potential is not precisely

applicable to the vapor species of interest since many are polyatomic and many are either radicals or have

permanent dipole moments. Most consider this to be an issue of modest importance. More difficult is that

reliable estimates of the parameters for the potential, collision diameter and energy well depth, are not

available. Parameters used in the modeling can be considered no more than educated guesses.

The advantages of the current release modeling in VICTORIA include accounting for the effects of both

ambient pressure and ambient gas composition as well as temperature on the releases of fission products

from the fuel.

Release from fuel debris

It is assumed in the VICTORIA computer code that relocation of fuel yields a rubble bed. Modeling of

fission product release from the rubble bed is based on the assumption that the rubble is composed of fuel

cylinders with lengths and diameters specified by the user to adequately reflect the surface to volume ratio

of real debris. Fission products diffuse through grains and pores of the fuel cylinders much as they do in

intact fuel. Mass transport of vapors that reach the geometric surfaces of the fuel cylinders is modeled.

The

VICTORIA code does have a model of fission product release from a molten pool. There have not been

many opportunities to evaluate and refine this model.

Release of control rod material

Vapors and aerosol are produced when silver-indium-cadmium control rods found in pressurised water

reactors fail during reactor accidents. These vapor and aerosol materials can interact with radionuclide

vapors and aerosols produced by core degradation and affect reactor accident source terms. VICTORIA

models the production of vapor and aerosols produced by the failure of silver-indium-cadmium control

rods. The production is considered to involve two phases - a burst failure when the control rod cladding

fails because the internal pressure exceeds the pressure in the reactor coolant system and a continued

vaporization release from the molten control rod alloy.

The burst release involves the venting of vapors from the control rod and the entrainment of 2 % of the

control rod alloy above the failure location. The entrained aerosol is taken to have a geometric mean

particle size of 0.5 µm with a geometric standard deviation of 1.8.

NEA/CSNI/R(2009)5

164

Vaporization of the Ag-In-Cd alloy is done by either fixing the location of the alloy or allowing the alloy to

"candle" or drain downward. Vaporization is calculated for material that is candling.

There is not now in VICTORIA a model of vapor and aerosol production from boron carbide control

blades degradation during boiling water reactor accidents.

Chemical speciation of released material

The essential model in VICTORIA is a prediction of the chemical speciation of the materials vaporized

from the reactor core. The model is applied in each of the computational nodes of the model so the

evolution of the speciation is taken into account. The model is based on the assumption of chemical

equilibrium at the local temperature and constant volume rather than the more common condition of

constant pressure. The computational method for finding the equilibrium speciation was specially

developed for the VICTORIA code, but is akin to the well-known NASA method.

The chemical equilibrium analysis considers the elements Ag, Ar, B, Ba, Cd, Cr, Cs, Eu, Fe, H, He, I, In,

Kr, Mn, Mo, Ni, O, Ru, Sb, Sn, Sr, Te, U, Xe, Zr. The equilibrium calculation considers 288 chemical

species. Chemical species considered in the equilibrium model were chosen from species with readily

available thermodynamic data. The species are primarily from binary M-O, M-H, M-Te and M-I systems

where M designates a metal or electropositve species. Very few mixed oxide vapor species are considered.

Certainly, the speciation has not been optimised for the high temperature conditions encountered in severe

reactor accidents. There has not been an aggressive attempt to include vapor species not addressed in

conventional compilations of thermodynamic data.

Multiple condensed phases can be predicted to form in the equilibrium calculation. VICTORIA considers

all condensed phases to form ideal solutions of one type or another. Three types of condensed phase

solutions are recognised: metallic, oxidic, and metal iodides. This is, at best, an approximate treatment

since many important solid species even within a given class of species recognised by VICTORIA are not

mutually soluble in the solid state.

A great deal of attention has been devoted to making the chemical equilibrium solver in VICTORIA

execute both reliably and quickly. The routine is called multiple times in each computation node and in

each time step as is discussed below in connection with vapor deposition.

Aerosol formation and growth

The VICTORIA computer code does not make a mechanistic prediction of aerosol particle nucleation from

supersaturated vapors. Rather, it assumes that aerosol is formed promptly when the vapor reaches a point

of saturation of one or more of its constituents as predicted with the chemical speciation model. Neglect of

nucleation modeling was based on calculational exercises that showed details of the aerosol size

distribution produced by mechanistic nucleation processes were promptly obscured by the coagulation of

the product aerosol particles.

The implementation of the approximate treatment of aerosol formation depends on the results obtained

from the chemical speciation model. When chemical speciation calculations indicate that a condensed

phase is formed or more mass is transferred to the condensed phase, this mass is placed in the smallest size

bin of aerosol considered in the aerosol modeling (See below).

The aerosol growth is calculated based on a conventional master equation using the CHARM [7] code.

CHARM uses a sectional method to solve the master equation. That is, the aerosol size distribution is

divided into a set of bins.

NEA/CSNI/R(2009)5

165

The solution is achieved using a finite interval collocation method. The coagulation kernel considers

coagulation of aerosol particles by:

•

gravitation settling

•

Brownian diffusion

•

turbulent inertia

•

turbulent diffusion

Aerosol growth as a result of hygroscopicity is not considered in VICTORIA since temperatures arising in

typical analyses exceed the critical temperature of water.

Vapor deposition and revaporization

As vapors pass though and out of the core region, they cool. A temperature is reached eventually where the

equilibrium chemical speciation will include some condensed phases. That is, the vapor will condense or

react with surfaces. Vapor condensation can involve deposition on surfaces, deposition on aerosol particles

and nucleation on particles. On the other hand, changes in temperature or chemical conditions along the

flow path or with time can cause previously deposited materials to vaporize.

The VICTORIA computer code has an involved method for the description of the condensation and

evaporation so that the processes are fully reversible. The equilibrium speciation of the bulk gas phase at

constant volume and temperature and including any aerosol present is calculated. Another equilibrium

speciation calculation is done for surfaces adjacent to the flowing gas which might be at quite different

temperatures than the bulk gas phase. Differences in the vapor pressures at the surface and in the gas phase

define the driving force for either vaporization or vapor deposition. The partitioning of vapor among

structural surfaces, aerosol surfaces and new aerosol is found through mass transport considerations. The

competitive mass transport to structural surfaces and to suspended aerosol is evaluated. Any additional

condensation (or evaporation) necessary to relieve disequilibrium in the system is apportioned to nucleated

aerosol that is assumed to appear in the smallest aerosol "bin".

Aerosol deposition

Aerosol deposition is considered in the VICTORIA computer code to occur as a result of:

•

gravitational settling

•

Brownian diffusion

•

turbulent deposition

•

thermophoresis

•

inertial deposition from flow in

- bends

- flow path contractions

vena contracta

A specialised model for inertial deposition in the separators and dryers of boiling water reactors is also

included in the VICTORIA modeling.

NEA/CSNI/R(2009)5

166

Diffusiophoretic deposition is not considered in VICTORIA. There is not usually condensing water in the

reactor coolant system under severe accident conditions and the condensing fluxes of vapor to structural

surfaces within the system are too small to significantly affect aerosol deposition rates.

Conventional models are used for gravitational settling and Brownian diffusion (see Chapter 2).

Turbulent deposition modeling is based on particle size. For submicron particles, turbulent deposition is

based on a model by Davies [8]:

()





































Φ+Φ

Φ+

−

1 - 2

tan

- 1

14.5

-2/3

and for larger particles, turbulent deposition is based on a model by Sehmel [9]:

02.3

1.2

01.1

Re10x69.3V













−

where:

V = particle disposition velocity (cm/s)

v = gas velocity (cm/s)

ρ = particle material density (g/cm

)

d = particle diameter (cm)

d = hydraulic diameter of flow path (cm)

Re = Reynolds number (-)

= Schmidt number

Φ =

9.2

3/1

Thermophoretic deposition is based on the Talbot [10] interpolation expression.

Inertial deposition in bends is based on the Pui [11] correlation. Deposition as a result of contractions in the

flow is based on the Ye and Pui [12] correlation. Deposition produced by a

vena contracta is calculated

from a correlation by Belyaev and Levin [13]. These correlations are discussed in Chapter 2 of this report.

4.2.1.1. Aerosol resuspension

VICTORIA uses an outdated model of resuspension based on a curve fit to experimental data [14]:

()







2300Ret/*U05.0

2300Re0

1.25

NEA/CSNI/R(2009)5

167

where:

F = fraction of deposit remaining

Re = Reynolds number

U * = friction velocity (m/s)

t = time from point at which Re became greater than 2300

This model is applicable only to aerosol that is present on surfaces as dry dust and should not apply to

deposited aerosols that have interacted chemically with the surface or with other deposited materials. It

does not account for any aging of the deposited material which will usually reduce the propensity for

resuspension. There is no consideration of surface vibration or shock waves on the ability to resuspend

deposited materials.

4.2.1.2 Status

Development and use of the VICTORIA computer code is no longer supported. Currently underway is an

effort to evaluate the findings obtained with VICTORIA in the analyses of a variety of experiments to

identify improvements needed in the fission product modeling in the MELCOR computer code.

References

[1] N.E. Bixler VICTORIA 2.0: A Mechanistic Model for Radionuclide Behaviour in a Nuclear Reactor

Coolant System under Severe Accident Conditions NUREG/CR-6131, Sandia National

Laboratories, Albuquerque, NM, 1998

[2] T.J. Heames et al.VICTORIA: A Mechanistic Model of Radionuclide Behaviour in the Reactor

Coolant System Under Severe Accident Conditions NUREG/CR-5545, Sandia National

Laboratories, Albuquerque, NM, December 1992

[3] A.G. Croff ORIGEN2: A versatile computer code for calculating the nuclide composition and

characteristics of nuclear materials Nuclear Technology, 62 (1983) 335

[4] M.R. Kuhlann, D.J. Lehmicke and R.O. Meyer CORSOR User's Manual NUREG/CR-4173,

BMI-2122, Battelle Memorial Institute, Columbus, Ohio, March 1985

[5] J. Rest and S.A. Zawadzki FASTGRASS: A Mechanistic Model for the Prediction of Xe, I, Cs, Te,

Ba and Sr Release from Nuclear Fuel Under Normal and Severe Accident Conditions: User's Guide

for Maintenance, Workstation and Personal Computer Applications NUREG/CR-5840, Argonne

National Laboratory, Argonne, IL, 1992

[6] H. Matzke A Diffusion in Ceramic Oxide Systems Advances in Ceramics, 17 (1986) 1

[7] C.J. Wheatley CHARM: A Model for Aerosol Behaviour in Time-Varying Thermal Hydraulic

Conditions NUREG/CR-5162, Sandia National Laboratories, Albuquerque, NM, 1988

[8] C.N. Davies Aerosol Science Academic Press, 1966

[9] G.A. Sehmel J. Geophys. Res., 75 (1970) 1766

[10] L. Talbot, T.K. Cheng, R.W. Schefer and D.R. Willis J. Fluid Mechanics 101 (1980) 737

[11] D.Y. Pui, R. Fomay-Novas and B.Y. Liu Aerosol Science and Technology, 7 (1987) 301

NEA/CSNI/R(2009)5

168

[12] Y.Ye and D.Y. Pui J. Aerosol Science, 21 (1990) 29

[13] S.P. Belyaev and L.M. Levin J. Aerosol Science, 3 (1972) 127

[14] A.L. Wright, W.L. Patterson and J.Y. King TRAP-MELT Validation Tets: Series 2 Aerosol

Resuspension Test Data Summary Report, Letter Report to the US Nuclear Regulatory Commission,

Oak Ridge National Laboratory, Oak Ridge, TN, 1984

4.2.2 ATHLET-CD

The system code ATHLET-CD (Analysis of Thermal-Hydraulics of LEaks and Transients with Core

egradation [1]) is designed to describe the reactor coolant system thermal-hydraulic response, core

damage progression, fission products and aerosol behaviour during severe accidents, to calculate the source

term for containment analyses, and to evaluate accident management measures [2]. It is being developed

by GRS in cooperation with the Institut für Kernenergetik und Energiesysteme (IKE), University of

Stuttgart. ATHLET-CD also includes the aerosol and fission product transport code SOPHAEROS (see

chapter 4.1.2), which is being developed by the French Institut de Radioprotection et de Sûreté Nucléaire

(IRSN).

The ATHLET-CD structure is highly modular in order to include a manifold spectrum of models and to

offer an optimum basis for further development. It contains the original ATHLET models for a

comprehensive simulation of the thermal-hydraulics in the reactor coolant system. The ATHLET code [3]

comprises: (a) a thermo-fluid-dynamic module, (b) a heat transfer and heat conduction module, (c) a

neutron kinetics module, (d) a general control simulation module, and (e) the general solver of differential

equation systems FEBE (Forward Euler - Backward Euler [4]). The thermo-fluid-dynamic module includes

two different fluid-dynamics equation systems: (a) a six-equation model, with fully separated balance

equations for liquid and vapor, complemented by mass conservation equations for up to 5 different non-

condensable gases and by a boron tracking model, and (b) a five-equation model, with a mixture

momentum equation and a full-range drift-flux formulation for the calculation of the relative velocity

between phases. Specific models for pumps, valves, steam separators, mixture level tracking, critical flow,

etc. are also available in ATHLET.

The rod module ECORE consists of models for fuel rods, absorber rods (AIC and B

C) and for the fuel

assembly including BWR canister and absorber. The module describes the mechanical rod behaviour

(ballooning), zirconium and boron carbide oxidation (Arrhenius type rate equation), Zr-UO

dissolution, as

well as melting of metallic and ceramic components. The melt relocation (candling model) is simulated by

rivulets with constant velocity and cross section, starting from the node where rod failure is predicted to

occur. The model accounts for oxidation, freezing, re-melting, re-freezing and melt accumulation due to

blockage formation. The feedback to the thermal-hydraulics considers steam starvation and blockage

formation [5]. Besides the convective heat transfer, energy can also be exchanged by radiation between

fuel rods and to surrounding core structures.

The release of fission products is modeled by rate equations or by a diffusion model within the module

FIPREM [6]. The transport and retention of fission products and aerosols in the reactor coolant system are

simulated by the code SOPHAEROS [7].

For the simulation of debris bed a specific model MESOCO [8] is under development, with its own

thermal-hydraulic equation system, coupled to the ATHLET fluid-dynamics on the outer boundaries of the

debris bed. The transition of the simulation of the core zones from ECORE to MESOCO depends on the

degree of degradation in the zone. The code development comprises also late phase models for core

slumping, melt pool behaviour and vessel failure.

NEA/CSNI/R(2009)5

169

The code system ATHLET/ATHLET-CD can be coupled to the containment code system COCOSYS

(see chapter 4.3.2), and it is the main process model within the German nuclear plant analyser ATLAS.

The ATLAS environment allows not only a graphical visualization of the calculated results but also an

interactive control of the simulation.

The code validation is based on integral tests and separate effect tests, as proposed by the CSNI validation

matrices [9], and covers thermal-hydraulics, bundle degradation as well as release and transport of fission

products and aerosols.

References

[1] K. Trambauer, C. Bals, J.-D. Schubert, H. Austregesilo ATHLET-CD Mod 2.1 - Cycle A User’s

ManualGesellschaft für Anlagen- und Reaktorsicherheit (GRS) mbH, GRS-P-4, Vol. 1., 2006

[2] V. Teschendorff, K. Trambauer

Simulation of severe accidents with detailed thermal hydraulic models Kerntechnik, 63, 18-24, 1998

[3] H. Austregesilo, C. Bals, A. Hora, G. Lerchl, P. Romstedt ATHLET Mod 2.0 Cycle A - Models and

Methods Gesellschaft für Anlagen- und Reaktorsicherheit (GRS) mbH, GRS-P-1, Vol. 4., 2003

[4] E. Hofer An A(α)-Stable Variable Order ODE-Solver Topical Meeting on Advances in Math.

Methods, Munich, April 1981

[5] K. Trambauer Coupling methods of thermal-hydraulic models with core degradation models in

ATHLET-CD In: Proceedings of the 6

International Conference on Nuclear Engineering,

ICONE-6368, San Diego, USA, May 10-15, 1998

[6] K.D. Hocke, A. Paller, A. Schatz Modelling and Code Development for the improved Description of

FP and Aerosol release during LWR Core Heat–up and Degradation CL-NA-15210–EN–C,

European Commission, 1994

[7] C. Fiche User’s manual of SOPHAEROS V2.2, IRSN N.T. SEMIC 2004/13, 2004

[8] M. Bürger, M. Buck, W. Schmidt, G. Pohlner, W. Widmann Ausbau und Verifikation der

Spätphasenmodelle und des Gesamtmodells zum Kernschmelzen in KESS und ATHLET-CD IKE

2-145, Universität Stuttgart, 2001

[9] K. Trambauer, T.J. Haste, B. Adroguer, Z. Hózer, D. Magallon, A. Zurita In-Vessel Core Degradation

Code Validation Matrix - Update 1996-1999 OECD Nuclear Energy Agency, NEA/CSNI/R(2000)21,

2000

4.3 Containment

4.31 CONTAIN

Introduction

The CONTAIN 2.0 computer code is a modular and integrated analysis tool used for predicting physical

conditions, chemical compositions, and distributions of radiological materials inside a containment

building following the release of material from the primary system in a light-water reactor accident [1]. It

can also predict the source term to the environment. Interactions among thermal-hydraulic phenomena,

aerosol behaviour and fission product behaviour are taken into account. The code includes atmospheric

NEA/CSNI/R(2009)5

170

models for steam/air thermodynamics, inter-cell flows, condensation/evaporation on structures and

aerosols, aerosol behaviour, and gas combustion. It also includes models for reactor cavity phenomena

such as core-concrete interactions and coolant pool boiling. Heat conduction in structures, fission product

decay and transport, radioactive decay heating, and the thermal-hydraulic and fission product

decontamination effects of engineered safety features are also modeled. For those phenomena affected by

significant uncertainty, CONTAIN 2.0 is provided with the capability of introducing user-specified input

parameters in order to perform sensitivity studies.

Status

CONTAIN was developed at Sandia National Laboratories under the sponsorship of the US Nuclear

Regulatory Commission (USNRC) for analyzing containment phenomena under severe accident and

design basis accident conditions. The first version was released in 1984 [2]. A detailed history of

CONTAIN code release can be found in the CONTAIN 2.0 code manual [1]. This note refers to the third

major release of the code (CONTAIN 2.0).

An extensive independent review of the CONTAIN code was done in the mid 90s [3]. At that time, the

Peer Review Committee judgement was extremely positive and stated that the code could be applied to

both selected design basis accidents and severe accidents for both current reactors and advanced light water

reactors. In this regard, the CONTAIN code was assessed against data concerning the passive containment

cooling system of the AP600 Westinghouse reactor [

4].

Even though in the revised severe accident code strategy of USNRC the role of CONTAIN for source term

analyses is de-emphasised, aerosol and fission product models are in many cases state-of-the-art and

suitable for many types of source term analyses.

Extensive validation of the CONTAIN code modeling has been carried out. A good example is the thermo-

hydraulics, inter-cell flow and hydrogen combustion models that have been compared against more than 20

tests [5]. In the particular case of aerosol modeling, recent validation exercises have been carried out in the

frame of OECD/CSNI against experiments like VANAM-M3 [6], KAEVER [7], PHÉBUS-FPT1 [8]. In

addition, further CONTAIN-to-data comparisons are set on individual bases [9, 10].

Aerosol modelling

The basic aerosol modeling in CONTAIN comes from the MAEROS code [11] together with some

enhancements. The major generic features are:

−

Size distribution can be described by 10-20 classes.

−

Chemical composition can be detailed by using 8 material components.

−

Steam condensation onto particles is tightly coupled with thermal-hydraulics.

As shown in the Table below, two methods are available in CONTAIN to model condensation/evaporation

of water onto/from particles: fixed- and moving-grid.

CONTAIN 2.0 is provided with two aerosol injections into the containment: the primary circuit and the

core-concrete interaction. The former is given through the code input deck, whereas the latter is a result of

the VANESA module in CORCON-Mod3.

Some relevant hypotheses assumed in the aerosol treatment are:

NEA/CSNI/R(2009)5

171

−

The size interval over which aerosol distribution extends is partitioned geometrically.

−

The particle volume in a class i is at least twice the one in the previous smaller class.

−

The aerosol material density is assumed to be the same for all components.

−

The aerosol shape (considered through dynamic and agglomeration shape factors) is independent of

aerosol composition.

−

The medium in which the aerosol processes are assumed to occur has fixed composition and is taken

to be air.

−

Gas properties appearing in equations for deposition rates are evaluated by-default for bulk

conditions.

−

The degree of turbulent agglomeration is fixed throughout the problem (input deck).

−

Thermal conductivity ratio between air and aerosol is not composition-dependent.

INTERACTION MECHANISMS

Agglomeration Brownian diffusion Friedlander

Differential gravitational settling MAEROS code

(Fuch’s collision effciency)

Turbulent shear Saffman and Turner

(Collision efficiency set to 1.0)

Turbulent acceleration in eddies Saffman and Turner

(Fuch’s collision effciency)

Steam condensation Moving-grid Solute effect

Kelvin effect

Mason equation

Fixed-grid

(default model)

None of above

effects modeled

NATURAL DEPOSITION MECHANISMS

Gravitational settling MAEROS code

Diffusiophoresis Waldmann & Schmidt

Thermophoresis Talbot

Diffusion MAEROS code

ESF DEPOSITION MECHANISMS

Sprays Diffusion

Diffusiophoresis

Thermophoresis

Impaction

Interception

[12]

Ice condensers Gravitational settling

Diffusion

Diffusiophoresis

Thermophoresis

Impaction

Interception

ICEDF code [13]

Fan coolers Diffusiophoresis Waldmann & Schmidt

Pools Core-concrete interactions VANESA (SCRUB) [14]

BWR suppression pools SPARC (85) [15]

NEA/CSNI/R(2009)5

172

Processes and systems with impact on aerosols

DESCRIPTION TREATMENT

Atmosphere/Pool

Thermal-Hydraulics &

Intercell Flow

Atmosphere bulk fluid:

−

Non-condensable gas

−

Coolant vapor

−

Homogeneously dispersed

liquid coolant

Pool bulk fluid

−

Coolant

−

The cell fluids are stagnant

and well-mixed

−

Flow occurs through flow

paths

−

Flow is controlled by a fluid

slug with an effective length.

−

Flow paths are not

repositories

Lower Cell and Cavity

Core-concrete

interactions(with/without

overlying coolant):

−

Basemat ablation

−

Production of gases and

steam (H

, CO,CO

, H

and ablation products

•

CORCON-MOD3 (the

VANESA code is inte-

grated):

−

Gas generation

−

Convective or boiling heat

transfer

−

Aerosol generation

Aerosols are scrubbed if there is

a coolant pool on the core debris

upper surface.

or, alternatively,

•

A one-dimensional system of

material layers

Direct Containment Heating

Molten debris ejection (under

high pressure) as a finely

dispersed collection of

particles into the containment

•

Standard containment models

(intercell flow, two-phase

atmosphere and coolant

thermodynamics, heat

transfer to structures,

radiative heat transfer,

hydrogen transport and

combustion, …) and

•

A suite of DCH-specific

models (transport and

trapping of multiple dispersed

core-debris fields, debris

droplets chemistry,

convection and radiation heat

transfer from the debris, …).

Fission Product Behaviour

−

Fission product assigned to

repositories (hosts):

atmosphere gas, suspended

aerosols, coolant pool, and

structure surface.

−

Fission product transfer

between hosts feasible.

NEA/CSNI/R(2009)5

173

DESCRIPTION TREATMENT

−

Two alternate ways of

representing fission product

decay

Combustion

− Deflagration of premixed

gas

−

Diffusion flame burning of

jets

−

Bulk spontaneous

recombination

−

Modified HECTR 1.8

−

Parametric model ( dynamics

of the diffusion flame not

modeled

−

Parametric model (user

provides recombination

threshold temperature and

time constant)

Heat & Mass Transfer

In-containment heat and mass

absorption of the thermal

energy entering containment

during a reactor accident.

−

Natural and forced

convection

−

Condensation mass and

heat transfer

−

Condensate film mass

transfer

−

Radiative heat transfer

−

Boiling heat transfer

−

Heat conduction

−

State of the art correlations

−

Heat/Mass transfer analogy

−

Fixed film thickness

(structure-pool) and film

tracking model (structure-

structure)

−

Net enclosure model and

direct radiation model

−

Adapted model from

CORCON

−

The solving method uses

finite difference techniques

Other models

− BWR suppression pool

vent discharge

−

Safety relief valve (SRV)

discharge

−

Fan cooler

−

Ice condenser

−

Sprays

−

Liquid transport system

components

Computational considerations

Overall, CONTAIN solves the aerosol equations by an explicit Runge-Kutta algorithm.

NEA/CSNI/R(2009)5

174

Steam condensation/evaporation onto/from particles is numerically handled in a different way depending

on the approach adopted (i.e., fixed- or moving-grid methods).

The fixed grid model models the condensation of water vapor onto aerosols and the evaporation of water

from them. Condensation is taken into account within the Runge-Kutta integration of the

integro-differential equation that describes aerosol evolution. Two solution methods are used for

evaporation. For high superheat, when aerosol water evaporation is insufficient to keep the atmosphere

saturated, a method of characteristics is employed. For evaporation under nearly saturated conditions, the

Runge-Kutta method is used.

The moving-grid model uses an algorithm based on the method of characteristics. After the effects of

condensation are calculated over a system timestep, the aerosol is remapped onto the fixed grid. This

remapping is performed by approximating the particle mass distribution as being constant with respect to

the logarithm of particle diameter.

The moving-grid method has the computational advantage of reducing numerical diffusion compared to the

fixed-grid one. However, this could result in an order-of-magnitude larger amount of computer time. In

addition, the fixed-grid method is more robust.

References

[1] K.K. Murata, D.C. Williams, J. Tills, R.O. Griffith, R.G. Gido, E.L. Tadios, E.J. Davis G.M.

Martínez, K.E. Washington “Code Manual for CONTAIN 2.0: A Computer Code for Nuclear

Reactor Containment Analyses” NUREG/CR-6533, SAND97-1735, June 1997

[2] K.D. Bergeron K.D. User’s Manual for CONTAIN 1.0, A Computer Code for Severe Nuclear

Reactor Accident Containment Analysis NUREG/CR-4085, SAND84-1204, Sandia National

Laboratories, Albuquerque, NM, May 1985

[3] B.E. Boyack “CONTAIN Independent Peer Review” LA-12866, Los Alamos National Laboratory,

Los Alamos, NM, 1995

[4] R. Vijaykumar, M. Khatib-Rahbar “Applicability of the CONTAIN Code Heat and Mass-Transfer

Models to Asymmetrically Heated Vertical Channels” Nuclear Technology 128,3, 313-326, 1999

[5] NEA “SOAR on Containment Thermalhydraulics and Hydrogen Distribution”

NEA/CSNI/R(1999)16

[6] M. Firnhaber, T.F. Kanzleiter, S. Schwarz, G. Weber “VANAM-M3: A Multi Compartment Aerosol

Depletion Test with Hygroscopic Aerosol Material” NEA/CSNI/R(96)26, 1997

[7] M. Firnhaber, K. Fischer, S. Schwarz, G. Weber “ISP-44 (KAEVER tests): "ISP-44 - Comparison

and Interpretation Report - Final”, NEA/CSNI/R(2003)5, 2003

[8] B. Clement, T. Haste “Comparison Report on International Standard Problem ISP-46 (FPT1)” Note

Technique SEMAR 03/021, 2003

[9] I. Kljenak, B. Mavko “Simulation of FPT1 Test Containment Phase with the CONTAIN computer

Code” IJS-DP-8614 rev. 1, Dec. 2002

[10] L.E. Herranz, L. Del Prá C., A.M. Rincón “CIEMAT Contribution to the PHEBEN-2 Project:

Interpretation of the PHÉBUS-FPT1 Experiment” CIEMAT 1025, Sept. 2003

NEA/CSNI/R(2009)5

175

[11] F. Gelbard “MAEROS User Manual” NUREG/CR-1391, SAND80-0822, Sandia National

Laboratories, Albuquerque, NM 1982

[12] N.A. Fuchs “The Mechanics of Aerosols”, Pergamon Press, 1964

[13] W.K. Winegardner, A.K. Postma, M.W. Jankowski “Studies of Fission Product Scrubbing with the

Ice Compartments” NUREG/CR-3248, PNL-4691, Richland, WA, 1983

[14] D.A. Powers, J.E. Brockmann, A.W. Shiver “VANESA: A Mechanistic Model of Radionuclide

Release and Aerosol Generation during Core Debris Interactions with Concrete” NUREG/CR-4308,

SAND85-1370, NM 1986

[15] P.C. Owczarski, R.I. Schrenk “Technical Bases and User’s Manual for the Prototype of a

Suppression Pool Aerosol Removal Code” NUREG/CR-3317, WA 1985

4.3.2 COCOSYS

Brief overview

The COntainment COde SYStem COCOSYS [1, 2] has been developed for the comprehensive simulation

of severe accidents in light-water reactor containments. The aim is to simulate all relevant phenomena,

processes and conditions that may occur inside the containment during such accidents. COCOSYS is also

able to simulate design basis accidents. One focus of COCOSYS is on the extensive consideration of

interactions between the various developing phenomena, such as thermal hydraulics, hydrogen

combustion, and aerosol and nuclide behaviour.

The structure of COCOSYS is shown in Fig. 4.3-1. The thermal hydraulic main module THY simulates

the thermodynamic of the steam/air atmosphere, atmospheric flows, hydrogen distribution, heat and mass

transfer to walls, heat conduction and storage in structures, H

deflagration, and pyrolysis processes. In the

aerosol and fission product main module AFP the transport and deposition behaviour of aerosols, the

behaviour of fission products including the decay heat release, and chemical reactions especially those of

iodine are treated. The

core concrete interaction main module CCI describes the physical and chemical

behaviour of the core melt, the concrete erosion, and the fission product release from the corium concrete

pool. Apart from the three main modules further programs are coupled to COCOSYS, namely DET3D for

the simulation of H

detonation, the commercial CFD code CFX for special fluid dynamic problems, and

the in-house-developed LAVA code for melt relocation from one room to a neighbouring one and

spreading. The data can be visualised both online and offline with the ATLAS programme. For the purpose

of complete sequence calculations COCOSYS has been coupled to the primary circuit code ATHLET-CD

which simulates core degradation.

NEA/CSNI/R(2009)5

176

THERMAL HYDRAULIC

zone models

junction models

H deflagration

pressure suppression

pyrolysis

safety systems

decay heat

AEROSOL-FISSION-

PRODUCT

aerosol behaviour

iodine behaviour

FP transport

nuclide behaviour

engineered systems

- filter

- spray

- pool scrubber

CORE-CONCRETE-

INTERACTION

corium erosion

melt chemistry

aerosol-FP-release

COCOSYS MAIN DRIVER

synchronisation, data management

ATHLET-CD

primary circuit

DET3D

H detonation

H deflagration,

distribution, jets

CFX 4

ATLAS

visualisation

LAVA

melt relocation

Fig. 4.3-1 Modular structure of COCOSYS

Aerosol modelling

The aerosol module AERIKA in the main module AFP describes the behaviour and the transport of a

multi-component polydisperse aerosol in a multi-compartment containment geometry. The aerosol may be

composed of up to eight chemically different aerosol components and the size distribution is discretised

into a number of size classes. All essential agglomeration and deposition processes are treated. The growth

of non-soluble and hygroscopic aerosol particles by volume condensation is modelled. The aerosol

retention by engineered systems like granular and fibrous filters, by water pools (pool scrubbing) and by

containment sprays is also modelled. Furthermore, all essential interactions with other containment

processes like the aerosol intercompartmental transport by gas and water flows, the

condensation/evaporation to/from aerosol particles, the depletion of aerosols carrying iodine, and the decay

heat release from airborne and deposited aerosols are treated.

The FIPHOST [3] module calculates the transport of the fission products within the containment. It

differentiates between the atmosphere, the aerosol and the sump water as fission product carriers. The

fission products can be deposited on surfaces in the atmosphere and in the sump. They are transported

together with the mobile carriers by the prevailing atmosphere and water flows. All relevant transfers

between the carriers are considered: aerosol deposition by natural processes or aided by technical systems

such as filters and spray systems, washing-off from walls, and the carrier change due to radioactive decay.

The module FIPISO [4] calculates the decay, the dynamic behaviour and the transport of fission product

inventory in the containment and the release to the environment. Generally, between 400 and 600 different

nuclides are considered in the calculation. The decay heat release in the individual zones and on the

individual carriers is taken into account in the THY main module.

NEA/CSNI/R(2009)5

177

Aerosol dynamics in a control volume

Within the module AERIKA in each control volume the aerosol dynamics are calculated by the model

MAEROS, developed by Sandia National Laboratories, USA [5], which is also used in other codes like

CONTAIN and MELCOR. The agglomeration kernel is determined by a two-dimensional integration over

particle size and the deposition coefficients by a one-dimensional integration. Owing to this accurate

method, a comparably low number of maximum 20 particle classes is sufficient. The aerosol coefficients

are calculated for various thermal hydraulic states at the beginning of a COCOSYS run and do not have to

be re-calculated anew for each time step. The first order ODE system for the aerosol concentration is

solved by FEBE (Forward Euler, Backward Euler) either implicitly or explicitly.

The way the coefficients for a multi-component system are calculated bring with it the following

restrictions: (1) aerosol material density is the same for all aerosol components (max. 8), all particles in

a certain size class have the same composition of materials, and (3) the dynamic and the agglomeration

shape factors are independent of the aerosol composition.

Particle growth by condensation is calculated by the moving-grid method, Moving Grid Analysis (MGA) [6]. In

contrast to the conventional fixed-grid method in MGA the grid of the particle size classes is apparently moved.

By this the numerical diffusion, which is an uncontrolled smearing of aerosol mass over several size classes, is

reduced. The Kelvin effect and the solubility effect for hygroscopic aerosols are included in MGA.

Table 4.3-1 gives an overview on the natural aerosol phenomena modelled in COCOSYS-AERIKA.

Table 4.3-1 Modelling of natural aerosol processes in COCOSYS

Phenomenon Mechanism

Modelling (reference and/or short

description)

Agglomeration

Brownian MAEROS [5]

Gravitational

MAEROS; optional collision efficiencies:

(1) Fuchs [7], (2) Pruppacher Klett [8], and

(3) Pruppacher Klett truncated

Turbulent shear MAEROS

Turbulent inertial

MAEROS; same collision efficiencies as for

gravitational agglomeration

Steam condensation

Condensation on non-soluble

aerosol

Mason equation including Kelvin effect;

Moving-grid-method (MGA) [6]

Condensation on hygroscopic

aerosol

Mason equation including solubility and

Kelvin effects;

Moving-grid-method (MGA)

Natural deposition

Sedimentation Stokes velocity with Cunningham correction

Diffusion Brownian (laminar) diffusion

Thermophoresis MAEROS

Diffusiophoresis Waldmann und Schmitt [9]

Resuspension

Dry resuspension by transient

gas flows

Extension of semi-empirical Fromentin

model [10]

Re-entrainment; droplet

production by bubble bursting

RECOM [11]

NEA/CSNI/R(2009)5

178

Aerosol interactions with other containment phenomena

An essential characteristic of COCOSYS is the fact that all important interactions between different

containment phenomena are taken into account. In Table 4.3-2 the modelled interactions between thermal

hydraulics, fission product behaviour and aerosol behaviour are summarised.

Table 4.3-2 Main aerosol - thermal hydraulic interactions modelled in COCOSYS

Interaction Modelled process Remarks

Thermal hydraulics

→ aerosol behaviour

Aerosol transport by gas flows

Aerosol transport by water flows

Diffusiophoresis

Particle growth by condensation

Dry resuspension by hydrogen deflagration Modelling not

completed

Re-entrainment (wet resuspension) from boiling

sump

Aerosol → thermal

hydraulics

FP aerosol decay heat releases into atmosphere,

walls and sump

Contribution of aerosol to the atmosphere

density

The modelled interaction processes take place with various strengths respectively velocities. This is

considered in COCOSYS by an adaptation of the numerical coupling of the models to the strength of the

prevailing interaction. In general tight couplings are used for fast interaction processes and loose ones for

slow processes.

The interactions were also thoroughly examined in two uncertainty and sensitivity analyses on a “dry” and

a “wet” aerosol case. In the first application the dry aerosol behaviour in the test VANAM M2 was

investigated [12]. The test was carried out in a nine-compartment geometry. 179 uncertain aerosol and

thermal hydraulic parameters were identified and analysed. The second application was on LACE test LA4

with a two-component aerosol under wet conditions, i.e. with steam condensation on the aerosol, in a

single volume geometry [13]. 58 uncertain parameters were selected.

The uncertainty of the aerosol results was found to be significantly larger in the wet case than in the dry

case. The main contributors identified were uncertain input parameters (dynamic and agglomeration shape

factors, diameter of injected particles), and the model option with/without Kelvin effect for condensation

calculation. The results revealed also the importance of uncertainties in thermal hydraulic boundary

conditions (atmosphere flows, volume condensation rate) for the calculated aerosol behaviour.

In another uncertainty and sensitivity study the modelling of the volume condensation rate was

investigated for a multi-compartment application [14]. The study shows that the uncertainty of the total

volume condensation rate in the entire containment is moderate and the local volume condensation rate in

certain compartments is large in parts, causes mainly by uncertainties in the calculated gas flows.

Engineered safety system

In AFP three engineered aerosol retention systems are modelled:

NEA/CSNI/R(2009)5

179

−

Filters (fibrous and granular)

−

Spray

−

Pool scrubber

In Table 4.3-3 the modelling of these systems is briefly described:

Table 4.3-3 Modelling of engineered aerosol retention systems in COCOSYS

System Retention mechanism Modelling (reference and/or

short description)

Fibrous filter

Diffusion

Semi-empirical model [15, 16],

use of Kuwabara-factor [17]

Interception

Impaction

Gravitation

Granular filter

Diffusion [18]

Interception

Impaction [19]

Gravitation [20]

Spray

Impaction Based on [7]

Interception

Diffusion Based on model in CONTAIN 2.0

[21]

Diffusiophoresis

Thermophoresis

Pool scrubber

Brownian SPARC-B [22], an extended

version of SPARC-90 [23]

Gravitational

Turbulent shear

Turbulent inertial

Validation

The aerosol modelling in COCOSYS has been continuously validated on experiments of the most

important containment projects within the last 20 years. Some older tests were calculated with the

COCOSYS precursor code FIPLOC. Table 4.3-4 gives an overview on the validation work performed. For

detailed descriptions of the tests mentioned see chapter 5.1.2 and 5.3.

NEA/CSNI/R(2009)5

180

Table 4.3-4 Validation work on the aerosol modelling in COCOSYS

Project Test Remarks

DEMONA

LACE

LA4 Uncertainty and sensitivity analysis

with FIPLOC

VANAM

M1, M2, M2*,

M3, M4

International Standard Problem

ISP-37 (VANAM M3)

AHMED

NaOH tests CSNI benchmark exercise

KAEVER

K90, K106, K123, K148, K151, K156,

K159, K186, K187, K188

ISP-44 (KAEVER tests),

VICTORIA

61, 62 With FIPLOC

PHÉBUS

FPT0, FPT1 ISP-46 (PHÉBUS FPT1)

Status

Aerosol modelling in COCOSYS has been completed except for the processes dry resuspension and

re-entrainment. The modelling of the Kelvin-effect is not yet satisfactory because of numerical instabilities

occurring especially with multi-compartment applications.

References

[1] W. Klein-Heßling, S. Arndt, G. Weber COCOSYS V 1.2 User Manual GRS-P-3/1, 2000

[2] H.-J. Allelein, S. Arndt, W. Klein-Heßling, S. Schwarz, C. Spengler, G. Weber COCOSYS:

Status of Development and Validation of the German Containment Code System Technical

Meeting on Severe Accident and Accident Management for NPP, Japan, March 2006

[3] S. Arndt, G. Weber FIPHOST - A Module to Calculate the Fission Product Transport in a LWR

Containment GRS-A-2553, April 1998

[4] U. Hesse FIPISO-98, ein Rechenmodell zum Nuklidverhalten in einem Raumzellensystem nach

einem Reaktorstörfall GRS-A-2750, 1998

[5] F. Gelbard MAEROS User Manual NUREG/CR-1391, 1982

[6] F. Gelbard Modeling Multicomponent Aerosol Particle Growth By Vapor Condensation Aerosol

Science and Technology 12:399-412, 1990

[7] N. A. Fuchs The Mechanics of Aerosols Pergamon Press, Oxford, 1964

[8] H. R. Pruppacher, J. D. Klett Microphysics of Clouds and Precipitation Reidel, New York, 1978

[9] C. N. Davies Aerosol Science Academic Press, 1966

NEA/CSNI/R(2009)5

181

[10] A. Fromentin Particle Resuspension from a Multi-Layer Deposit by Turbulent Flow Paul Scherrer

Institute, Würenlingen, Bericht Nr. 38, 1989

[11] A. Voßnacke Untersuchung zur Tropfenbildung an Oberflächen gasdurchströmter Flüssigkeiten bei

Störfällen in kerntechnischen Anlagen Dissertation, Ruhr-Universität Bochum, Bochum, 1999

[12] E. Hofer, B. Krzykacz, L. Langhans, G. Weber Uncertainties and Sensitivities in Aerosol

Calculations-An Analysis of a FIPLOC-M Application to the Experiment VANAM M2 PSAM

II-Conference, San Diego, 20 – 25 March 1994

[13] J. Langhans, H.G. Friederichs, E. Hofer, B. Krzykacz, B.M. Schmitz, G. Weber Unsicherheits-und

Sensitivitätsanalyse zum Aerosolmodell in FIPLOC-M, Teil 2: Nasses Aerosolverhalten im

LACE-Experiment LA4 GRS-A-2217, 1994

[14] G. Weber, E. Hofer, B. Krzykacz Uncertainty and Sensitivity Analysis of Fog Formation Rates

Calculated with the Containment Code FIPLOC-M J. Aerosol Sci., Vol.23, Suppl., pp S 269 - S 272,

1992

[15] K. W. Lee, B. Y. H. Liu Experimental Study of Aerosol Filtration by Fibrous Filters Aerosol Sci.

Technol., 1:35-46, 1982

[16] K. W. Lee, B. Y. H. Liu Theoretical Study of Aerosol Filtration by Fibrous Filters Aerosol Sci.

Technol., 1:47-62, 1982

[17] S. Kuwabara The forces experienced by randomly distributed parallel circular cylinders or spheres in

viscous flows at small Reynolds numbers J.Phys.Soc.Japan 14(4):527-532, 1959

[18] C. Gutfinger, G. I. Tardos Theoretical and Experimental Investigation on Granular Bed Dust Filters

Atm. Environment, 13:853-867, 1979

[19] T. D’Ottavio, S. L. Goren Aerosol Capture in Granular Beds in the Impaction Dominated Regime

Aerosol Sci. and Technol., 2:91-108, 1983

[20] G.I. Tardos et al. Experiments on Aerosol Filtration in Granular Sand Beds J. Colloid and Interface

Science, 71(3):616-621, 1979

[21] K.K. Murata et al. Code Manual for the CONTAIN 2.0: A Computer Code for Nuclear Reactor

Containment Analysis\NUREG/CR-6533, 1997

[22] K. Fischer Modellierung von Abscheidevorgängen in Wasservorlagen Battelle Ingenieurtechnik

GmbH, Eschborn BF-R68.411-1, 1998

[23] P.C. Owczarski, K.W. Burk SPARC-90: A Code for Calculating Fission Product Capture in

Suppression Pools NUREG/CR-5765, 1991

4.3.3 ART/REMOVAL

4.3.3.1 ART code

The ART (Analysis of Radionuclide Transport) code [1] has been developed at JAERI as a module of the

source term analysis code, THALES-2 [2] to analyse the behaviour of fission products (FP) in reactor

coolant system and containment during LWR severe accidents. The thermal-hydraulic condition needs to

NEA/CSNI/R(2009)5

182

be given as a function of time. The code can consider FP transport and removal by natural deposition and

by engineered safety features such as spray systems. As for the FP aerosol deposition, the code can treat

gravitational settling, thermophoresis, diffusiophoresis, Brownian diffusion and resuspension. For the

gaseous FP, the code can deal with condensation, chemisorption and revaporization. The code also

considers the aerosol growth by gravitational, Brownian and turbulent agglomeration and condensation or

revaporization of gaseous FP at the aerosol surface. Major physical models in ART were verified through

analyses of the international standard problems and JAERI’s WIND/WAVE experiments [3].

4.3.3.2 REMOVAL code

The REMOVAL (REMOval of Volatile fission products and Aerosol in an LWR containment) code [4]

was developed at JAERI to analyse the fission products behaviour in the containment of LWR during a

severe accident. Physical models considered in the code are

agglomeration due to Brownian, gravitational and turbulent motion,

Brownian diffusion

thermophoresis

diffusiophoresis

spray removal by gravity and collision

steam condensation onto aerosol particle

entrainment form flashing pool.

In the analysis, the containment is divided into several compartments and the behaviour of the aerosol,

iodine and noble gas is analysed for each compartment assuming the uniformity inside the compartment.

Thermal-hydraulic condition should be given for each compartment as a function of time.

4.3.3.3 ART/REMOVAL (ART combined with REMOVAL model)

In order to participate in the ISP-44 (International Standard Problem No.44)

, the hygroscopicity model

based on the Mason’s equation in REMOVAL code was tentatively combined with ART code because

ART does not treat it and the development of REMOVAL finished in early 1990’s [5]. It is noted that

present ART does not include the related REMOVAL model due to numerical problems. For calculation of

aerosol growth by steam condensation, the relative humidity at aerosol surface in Mason’s equation is

reevaluated taking into account the Kelvin’s effect and the Raoult’s law. Then, the equilibrium aerosol size

is calculated by iteration until the growth rate becomes zero. Since it usually takes much CPU time for

iteration calculation, REMOVAL calculates the equilibrium radius of aerosol by assuming the aerosol

surface temperature equal to atmospheric temperature and uses the calculated value during the iteration to

reduce computation time.

References

[1] T. Ishigami et al. "User's Manual of ART Code for Analyzing Fission Product Transport Behaviour

during Core Meltdown Accident" JAERI-M 88-093 (1988)

[2] M. Kajimoto et al. "Development of THALES-2, A Computer Code for Coupled Thermal-

Hydraulics and FP Transport Analyses for Severe Accident at LWRs and Its Application to Analysis

of FP Revaporization Phenomena" Proc. Int. Topical Meeting on Safety of Thermal Reactors,

Portland, pp.584-592 (1991)

NEA/CSNI/R(2009)5

183

[3] A. Hidaka et al. "Experimental Analyses of Iodine Behaviour under Severe Accident Conditions

with ART" J. Nucl. Mater. 248, pp. 226-232 (1997)

[4] K. Muramatsu, K. Abe, M. Ida, N. Watanabe, K. Soda, N. Yamano "Sensitivity Study on BWR

Source Terms using the THALES/ART and REMOVAL" Proceedings of the International

ENS/ANS Conference on thermal reactor safety, Vol. 6, pp.2407-2416 (1988)

[5] M. Firnhaber et al."International Standard Problem ISP44 KAEVER Experiments on the Behaviour

of Core-melt Aerosols in a LWR Containment Comparison Report TABLE OF CONTENTS"

NEA/CSNI/R(2003)5 (2002)

4.3.4 SMART

Brief theoretical description

SMART is a code that models containment aerosol and iodine behaviour. Fission products are assumed to

enter into containment in one of two physical forms, either as gases or dissolved or suspended in small

water aerosols generated by a primary heat transport system break discharge. The initial droplet size

distribution is considered to be log–normal and is characterised by two parameters: the geometric mean

diameter and the geometric standard deviation. In SMART, the geometric mean diameter is considered to

be equal to the critical diameter, where the critical diameter is defined as a diameter when the drag force on

the droplet exceeds the surface tension forces. Currently SMART does not model aerosol release during a

steam-only discharge, however this capability is being incorporated into the code.

SMART assumes that aerosol is transported within containment by bulk gas flows. Within a particular

node or a room, the liquid aerosol droplets undergo coagulation and also deposit on walls and floors due to

various natural phenomena (e.g., gravitational settling). To model the coagulation and removal processes

mathematically, the continuous aerosol size spectrum is divided into a finite number of intervals, and

within each interval, the aerosol size is assumed to be constant. The initial mass distribution of the fission

products in the aerosol size classes are provided by upstream analysis, while the changes in the fission

product masses in the size classes due to radioactive decay and build-up, coagulation and deposition in

containment are calculated in SMART.

Brief Description

Reactor

Components

Modelled

• Containment only

•

Emergency Filtered Air Discharge System (EFADS), models

aerosol removal by HEPA filters and iodine removal by charcoal

filters

•

Dousing (models water droplet aerosol formation by operation of

dousing systems)

•

Moderator (for removal of aerosols by moderator scrubbing)

Geometry

Idealization

• A network of nodes inter-connected by flow paths

•

Up to 200 nodes possible, nodalization must be consistent with the

thermalhydraulics containment model.

Inter-nodal

Transport

• Flow paths modelled as a fictitious pipe with convective flow of gas

mixtures, aqueous solutions, and water droplet aerosols (containing

dissolved or suspended fission products).

NEA/CSNI/R(2009)5

184

Thermalhydraulics

• Interfaces with a containment thermalhydraulic model (GOTHIC) to

obtain transient thermalhydraulic properties

•

Gases, fission products and aerosols in a node are assumed to be

perfectly mixed

•

Fission product and aerosol processes do not contribute to mass,

momentum or energy exchange (e.g. no feedback to thermalhydraulics

code)

Fission Product

Transport

Aerosol Transport

• Convective flow through links of fission products dissolved in water

droplet aerosols, bulk liquid or in the gaseous phase

•

Solid (dry) fission products are not modelled

•

Break or other discharge sources into a room

•

Convective flow out of holes to the external atmosphere

•

Removal by sinks (EFADS, adsorption of iodine on walls, aerosol

deposition on walls and floor)

•

Generation from precursors in radioactive decay-build up processes

•

Removal by decay

•

Iodine species alone can be transported and tracked between different

phases in each node (bulk water, condensing films, gas, suspended

water droplets, adsorbed on surfaces)

See fission product transport mechanisms above.

Aerosol processes modelled are

•

Generation of aerosols by break discharge or dousing

•

Aerosol agglomeration

•

Brownian

•

Gravitational

•

Turbulent-Inertial and Turbulent-Shear

•

Removal by Impingement (inertial impac

•

Moderator Washout

•

Gravitational Settling

•

Turbulent Deposition

•

Stefan flow

•

Thermophoresis

Reference

Private communication between J. Ball and the Chairman.

4.3.5 GOTHIC

The containment thermalhydraulics code GOTHIC has the potential to be developed into a code with

aerosol modelling capabilities, since it contains models to generate, entrain and transport small water

droplets while tracking these droplets. However, the current GOTHIC release (GOTHIC 7.2a) does not

contain aerosol modeling capabilities.

NEA/CSNI/R(2009)5

185

GOTHIC solves conservation equations for mass, momentum and energy for multi-component,

multi-phase flow and these equations are solved for up to three primary fields (Steam/gas mixtures,

continuous liquid, liquid droplet), and up to two secondary fields (mist, liquid components).

There are a few options available in GOTHIC to generate, entrain and transport small water droplets.

Droplets can be formed either by operations such as flashing (break discharge), by dousing, or by the user

specifying that droplets should be formed when the atmosphere becomes supersaturated with steam. For

example, the spray nozzle component is used to model containment sprays, ice condenser drain flows into

the lower containment and vessel spray systems. The nozzle model converts liquid flow through a flow

path, network link or 3D connector from the continuous liquid phase to the droplet phase with the fraction

of the liquid converted into droplets and the drop diameter specified as functions of time.

For modeling volume condensation, the user has the option of choosing a fog model, in which very small

drops are created when the atmosphere becomes supersaturated or the “mist” option, in which mist is

created when the atmosphere becomes supersaturated. In the fog model, the droplets can be tracked,

however, when small drops are combined with the large drops from a break or spray, the average drop size

is not representative of either population, resulting in possible excessive heat and mass transfer at the drop

surfaces. The mist phase is not combined with the drop phase and is assumed to move with the vapor

phase. When the mist concentration exceeds the user specified limit, the excess mist is either converted to

drops or is deposited in the liquid phase, depending on the user specification for the drop diameter from

excess mist. If the drop diameter from excess mist is set to zero, then the excess mist is deposited in the

liquid phase. Otherwise, new drops are created at the specified diameter (default is 200 µm).

Despite the potential for modeling some aerosol phenomena (wet aerosol) in GOTHIC, the code does not

currently allow fission product transport in any other component except the vapor phase. Incorporation of

aerosol phenomena to simulate fission product transport and deposition in containment would require

significant effort.

There is a GOTHIC development task, approved by the GOTHIC Advisory Group, to implement a drop

field distribution model into the next release of GOTHIC (7.3, not expected to be released for at least

2 years). This will allow GOTHIC to calculate a distribution of droplet diameters for each lumped control

volume and cell within a subdivided control volume. However, modeling fission product transport in these

aerosol droplets, or from dry aerosols, is not currently being considered for future development.

Reference

Private communication between J. Ball and the Chairman.

4.3.6 KUPOL

An aerosol model was implemented into the Russian thermal-hydraulic code KUPOL-M [1]. This coupled

version allows to simulate aerosol transport in a subdivided containment. For a lumped-parameter model a

complete kinetic equation of aerosol transport is the Liouville equation for a size distribution function. It

takes into account processes of condensation, sedimentation, convective exchange, diffusional, turbulent

and thermophoretic deposition as a source form and coagulation process as a collision integral form. We

will only consider the processes of condensation and evaporation onto and from aerosol particles and

aerosol coagulation. So, we write the equation in such a way:

The last term on the left-hand side of the equation represents steam condensation onto droplets with mass.

The first and the second terms on the right-hand side of the equation describe coagulation mechanism,

NEA/CSNI/R(2009)5

186

taking into account particle arrival into the given size space from the smaller ones and particle removal

from the given size space into the larger ones.

The accepted coagulation models use the coagulation kernels, simulating mechanisms of gravitational

coagulation, Brownian coagulation, turbulent and diffusional coagulation and turbulent shear one.

The differential equation of aerosol transport is solved by a fractions method [2]. The collision integrals are

expressed as a source form in each fraction.

References

[1] A.D.Efanov, A.A.Loukianov, N.N.Shangin, Yu.S.Yuriev, A.A.Zaitsev, S.L.Soloviev, L.I.Zaitchik

Development of a containment code KUPOL-М for a substantiation of the VVER's accident

localization system functions Proc. of ICONE10, #22281, Arlington, VA, April 14–18, 2002

[2] A.G. Godizov, A.D. Efanov, A.A. Lukianov, O.V. Supotnitskaya Simulation of processes of water

aerosol coagulation-condensation growth using a combination of methods of groups and fractions

Proc. of NURETH-11, Paper: 224, Popes’ Palace Conference Center, Avignon, France, October 2-6,

2005

NEA/CSNI/R(2009)5

187

5. RECENT VALIDATION WORK

5.1 Integral Experiments

5.1.1 The PHÉBUS fission product programme

Introduction

Since the TMI2 accident in 1978, an important international effort has been made on research on severe

accidents, involving integral experiments, separate-effect tests and model developments. The PHÉBUS-FP

programme has played and is still playing an important role in this landscape. A number of lessons have

been learned from the analysis of the results – see e.g. [1, 2] - and from their use for source term

assessment studies, [3]. As a result, increasingly better-qualified assessment tools have been developed but

remaining uncertainties have been evidenced. For reducing them IRSN, CEA and EDF have launched a

new International Source Term Programme comprising a series of separate-effect experiments, [4].

The international PHÉBUS FP programme, initiated in 1988 and supported by the European Commission,

is one of the major research programmes on severe accidents for light-water reactors. Regarding the

interpretation of PHÉBUS, an international co-operation has been established for more than ten years,

particularly helpful for the improvement and common understanding of severe accident phenomena. After

a short description of the facility and of the test matrix, below the main outcomes and results of the first

four integral tests are provided and analysed. Several results were unexpected and some are of importance

for safety analyses, particularly concerning fuel degradation, cladding oxidation, chemical form of some

fission products, especially iodine, effect of control rod materials on degradation and chemistry, iodine

behaviour in the containment. Prediction capabilities of calculation tools have largely been improved as a

result of this research effort.

Description of the facility and test matrix

The PHÉBUS FP programme is investigating key phenomena involved in LWR severe-accident sequences

through a series of five in-pile integral experiments. The facility provides prototypic reactor conditions

which allow the study of basic phenomena governing core degradation through to its late phase (melt pool

formation), hydrogen production, FP release and transport, circuit and containment phenomena, and iodine

chemistry. The phenomena studied take place (i) in the core region simulated by a 1-m long bundle of 20

irradiated fuel rods and a control rod, (ii) in the primary system components for which a steam generator is

simulated by a single inverted U-tube, (iii) in the containment building simulated by a 10 m

tank including

a sump, the atmosphere and painted wet and dry surfaces. All three of these components are scaled down

by roughly 1/5000 as compared with a French 900 MWe power plant. This is illustrated in Fig. 5.1-1. A

more comprehensive description can be found in, e.g., [5, 6].

NEA/CSNI/R(2009)5

188

Reactor Containment

Model

(10 m3)

Painted

Condensers

Sump

(100 l)

Sampling

furnaces

Sampling

furnace

Experimental

Cell

Phebus Reactor

Core

Test

Assembly

gamma

Spectro

Steam Generator

Model

Aqueous Sampling

FP Caisson

Phebus Reactor

Fig. 5.1-1 The PHÉBUS-FB facility

The test matrix is given in Table 5.1-1. The first two tests were performed in similar conditions except for

fuel burn-up (FPT-0 used trace-irradiated fuel whereas FPT-1 fuel had been irradiated to 24GWd/tU):

bundle including a silver-indium-cadmium control rod subjected to a steam-rich environment, hot leg

heated up to 700 °C, cold leg at 150 °C, acidic and cold sump water. For FPT-2, the bundle was subjected

to a steam-poor environment and the sump water was alkaline and evaporating at the end of the test. For

FPT-3, the silver-indium-cadmium control rod was replaced by a boron carbide one, other test conditions

being close to those of FPT-2. The FPT-4 test was of a different nature as it investigated the release of

low-volatility fission products and actinides from a debris bed made of fuel pellet fragments and oxidised

cladding shards heated up to the formation of a molten pool.

Table 5.1-1 PHÉBUS-FP test matrix

Test n° Type of fuel Fuel Degradation Primary Circuit Containment Date

FPT-0 Fresh Fuel

1 Ag-In-Cd rod

9 days pre irrad.

Melt Progression & FP

release in steam-rich

environment

FP chemistry and

deposits in non-

condensing steam

generator

Aerosol

deposition

Iodine

radiochemistry at

pH5

Dec. 2

1993

FPT-1

BR3 fuel

≈ 23

GWd/tU

1 Ag-In-Cd rod

Re irradiation

As FPT-0 with

irradiated fuel

As FPT-0 As FPT-0 July 26

1996

FPT-2 As FPT-1

BR3 fuel

≈ 32

GWd/tU

As FPT-1 under steam

poor conditions

As FPT-1 with effect

of boric acid

PH9 evaporating

sump

Oct. 12

2000

FPT-3 As FPT-1 with

B4C instead of Ag-

In-Cd

BR3 fuel

≈ 24

GWd/tU

As FPT-2 As FPT-0 PH5 evaporating

sump recombiner

coupons

Nov. 18

2004

FPT-4 EDF fuel 38

GWd/tU no

reirradiation

Low volatile FP&

actinide release from

-ZrO

debris bed

up to melting

Integral filters in test device

Post-test analyses on samples

July 22

1999

NEA/CSNI/R(2009)5

189

Main outcomes with respect to FP transport

The release of volatile fission product is generally well calculated even if some codes using the CORSOR

approach tend to overestimate the kinetics at the beginning of the transient, [7, 8]. Semi-empirical models,

though not modelling all the processes, are able to do well using a consistent set of parameters for

separate-effect tests and PHÉBUS-FP integral experiments. The situation is more contrasted for less

volatile elements for which chemistry plays a major role. A better understanding of the underlying

phenomena has been gained using mechanistic models [8].

In the particular case of FPT-0, using trace-irradiated fuel, the early release of volatile fission products can

only be explained by taking into account fuel dissolution during the cladding oxidation phase. Barium

release is much smaller in the PHÉBUS tests performed in a bundle geometry than in separate-effect

annealing tests. This difference is attributed to interactions of fuel with the cladding material and maybe

iron which greatly reduce the volatility of barium, [9]. These two observations illustrate the strong

coupling between fuel degradation and fission-product release phenomena.

The release from the silver-indium-cadmium control rod is often not well enough calculated by certain

codes, especially for silver which has an important impact on iodine chemistry. The governing phenomena

are well understood but a modelling effort is still needed mainly concerning the coupling between

degradation and release.

Fission-product transport in the RCS is characterised by two zones of significant deposition coinciding

with sections in which temperatures dropped rapidly. These were the vertical section of the hot leg above

the bundle where the gas cooled from temperatures as high as 1600 °C down to 700 °C and in the steam

generator (SG) riser where temperatures cooled from 700 °C to 150 °C. For the first two tests, FPT-0 and

FPT-1, marked differences in volatility were seen for different elements where iodine and cadmium barely

deposited at all in the hot leg of the circuit with significant deposition only in the SG. Cesium deposition,

in contrast, was significant and nearly the same in both cooling zones; completing the range of volatilities,

others elements such as molybdenum and silver deposited primarily in the vertical hot leg with relatively

little additional deposition downstream.

Recent analyses performed applying a standard version of the ASTEC/Sophaeros code show that the code

is able to reproduce some important aspects of the FPT-0 and FPT-1 tests, principally vapor/aerosol

behaviour and total retentions, [10]. As for fission-product speciation, it is worth noting that cesium was

condensed in the hot leg of the circuit, thus not being transported as CsOH as often assumed in the past.

Calculation results indicate that cesium molybdate and iodide are the main species. Though the overall

retention in the circuit is well calculated, the deposition in the vertical hot leg is underestimated. This is

largely accounted for by the effect of simultaneously-developing flow characterised by much higher

temperature gradients and mass transfer to the wall than in the case for developed flow. The deposition in

the steam generator is generally overestimated by a factor of about two. A number of hypotheses have been

explored in relation to this discrepancy [10] and 2D particle-tracking calculations performed [11], but no

conclusive explanations have been so far obtained.

In the containment, thermal-hydraulics was mainly governed by the balance between the incoming steam

and its condensation. Simple lumped-parameter codes using a coarse noding (1 or few volumes) are able to

reproduce the measured phenomena. Most of the aerosol mass was deposited on the bottom of the

containment and most of the remainder on the condensing surfaces. A small fraction was observed on the

containment walls. The first two kinds of deposits are generally well captured by lumped-parameter codes

using classical models for gravitational settling and diffusiophoretic deposition [12] although some had

difficulties to reproduce the split between deposition on the bottom and on the condensing surfaces. A

NEA/CSNI/R(2009)5

190

mechanism based on turbulence damping in the boundary layer has been identified to account for the small

deposit (2 %) on the walls in FPT-1.

Generally speaking, the simulation by calculation codes of thermal-hydraulics and aerosol behaviour in the

PHÉBUS FP containment is satisfactory.

References

[1] M. Schwarz, B. Clément, A.V. Jones “Applicability of PHÉBUS FP results to severe accident safety

evaluations and management measures”, Nucl. Eng. Des. 209, 173-181 (2001)

[2] J. Birchley, T. Haste, H. Bruchertseifer, R. Cripps, S. Güntay, B. Jäckel “PHÉBUS-FP: Results and

significance for plant safety in Switzerland” Nucl. Eng. Des. 235, 1607-1633 (2005)

[3] J.M. Evrard, C. Marchand, E. Raimond, M. Durin “Use of PHÉBUS FP experimental results for

source term assessment and level 2 PSA” Proc. 5

Technical Seminar on the PHÉBUS FP

Programme, Aix-en-Provence, France, 24-26 June 2003

[4] B. Clément “Towards reducing the uncertainties on source term evaluations: an IRSN/CEA/ EdF

R&D programme” Proc. Eurosafe Forum, Berlin, Germany, 8-9 Nov. 2004

[5] M. Schwarz, G. Hache, P. von der Hardt “PHÉBUS FP: a severe accident research programme for

current and advanced light water reactors” Nucl. Eng. Des. 187, 47-69 (1999)

[6] B. Clément, N. Hanniet-Girault, G. Repetto, D. Jacquemain, A.V. Jones, M.P. Kissane, P. von der

Hardt “LWR severe accident simulation: synthesis of the results and interpretation of the first

PHÉBUS FP experiment FPT0” Nucl. Eng. Des. 226, 5-82 (2003)

[7] B. Clément, T. Haste “ISP-46 – PHÉBUS FPT1”, NEA/CSNI/R(2004)18, August 2004

[8[ R. Dubourg, H. Faure-Geors, G. Nicaise, M. Barrachin “Fission product release in the first two

PHÉBUS tests FPT0 and FPT1” Nucl. Eng. Des., article in press

[9] R. Dubourg, P. Taylor “A qualitative comparison of barium behaviour in the PHÉBUS FPT0 test

and analytical tests” J. Nucl. Mat. 294, 32-38 (2001)

[10] M.P. Kissane, I. Drosik “Interpretation of Fission-product transport behaviour in the PHÉBUS FPT0

and FPT1 tests” Nuclear Eng. Des. 236, 1210-1223 (2006)

[11] C. Housiadas, K. Müller, J., Carlsson, Y. Drossinos “Two-dimensional effects in thermophoretic

particle deposition: the PHÉBUS-FP steam generator” J. Aerosol Sci.32, pp. S1029-S1040 (2001)

[12] I. Kljenak, B. Mavko “Simulation of containment phenomena during the PHÉBUS FPT1 test with

the CONTAIN code” Proc. Nuclear Energy for New Europe 2002, Kranjska Gora, Slovenia, 9-12

Sept. 2002

5.1.2 ISP-46 (PHÉBUS FPT1)

Introduction

The objective of this ISP is to assess the capability of computer codes to reproduce an integral simulation

of the physical processes taking place during a severe accident in a pressurised water reactor, i.e. including

NEA/CSNI/R(2009)5

191

fuel degradation and associated hydrogen production and fission product release, fission product and

structural material transport in the primary circuit, aerosol behaviour in the containment and iodine

radiochemistry.

The ISP was conducted as an open exercise, with all the relevant experimental results being available to

the participants. It was divided into four phases:

•

Fuel degradation, hydrogen production, fission product and structural material release (‘bundle’,

phase 1);

•

Fission product and aerosol transport in RCS (‘circuit’, phase 2);

•

Thermal-hydraulics and aerosol physics in containment (‘containment’, phase 3);

•

Iodine chemistry in containment (‘chemistry’, phase 4).

The emphasis was on integral calculations (all phases). The aim was not to carry out interpretation work,

but to use the codes as in plant studies, i.e. with standard models/options as far as possible, representing the

facility in a similar level of detail; this constituted the mandatory ’base case’ calculation. A more detailed

‘best-estimate’ sensitivity study could also optionally be performed.

Fifteen different codes were used: ASTEC, ATHLET-CD, COCOSYS, CONTAIN, ECART, FEAST,

IMPACT/SAMPSON, ICARE/CATHARE, IMPAIR, INSPECT, MAAP4, MELCOR, SCDAP/RELAP5,

SCDAPSIM and SOPHAEROS, of these 4 are integral codes (ASTEC, IMPACT/SAMPSON, MAAP4

and MELCOR). For the base case, 47 calculations were received, with 21 for the optional best-estimate

version. Of the base case calculations, 14 were integral (at least 3 phases calculated).

For the base case, a noding scheme was recommended in the specification report. The bundle is divided

into 11 axial nodes and typically 3-5 radial rings, with normally 1 or 2 thermal hydraulic flow channels.

The circuit is divided into 11 nodes, this being the minimum considered necessary for an adequate

calculation of deposition. The containment model is simple, with 1 node for the main volume and 1 for the

sump, taking advantage of the well-mixed conditions. For best-estimate calculations, often only noding

density was increased by typically a factor 2 or more, at the choice of the user.

The results were analysed in detail, comparing the results amongst each other and with the FPT1 data.

There was considerable scatter amongst the results obtained from each code by different users, the ‘user

effect’. To minimise this effect, representative cases were selected where necessary, taking into account the

quality of key output variables, completeness and accuracy of the technical reports, and including code

developers where possible. This analysis led to an assessment of the main models in each of the four areas

considered. Here only phases 2 and 3 dealing with aerosols are summarised below. There was on the whole

little significant difference between the base and best-estimate cases, with at most a small improvement

only in the results of the latter cases, so conclusions could be drawn on the basis of the former.

Assessment of codes and models

The following phenomena/parameters are in general well simulated by the codes:

•

Bundle – thermal response (given adjustment of input nuclear power and shroud thermal

properties within experimental uncertainties), hydrogen production (including oxidation of

relocated melt), bundle final state material distribution (given suitable reduction of the bulk fuel

relocation temperature from the ceramic value, in the longer term a more mechanistic model is

desirable), total release of volatile fission products;

NEA/CSNI/R(2009)5

192

•

Circuit – total retention of fission products and structural materials (but after cancellation of

errors);

•

Containment – thermal hydraulic behaviour (as exemplified by average gas temperature,

pressure, relative humidity and condensation rate), depletion rates;

•

Chemistry – models of the Ag/I reaction in the liquid phase are adequate for FPT1 (this cannot be

extended to other cases where the Ag is not so much in excess with respect to I; due to the large

excess of silver, in the experiment, radiolytic production of gaseous iodine and dissociation of

silver iodide did not play an important role in the overall iodine behaviour).

The following phenomena/parameters were reasonably well simulated, but some modelling improvement

is desirable:

•

Bundle – outlet coolant temperatures (overprediction), time dependence of volatile FP release

(generally too fast a release at low temperatures, e.g. for CORSOR-type approaches);

•

Circuit – distribution of deposition in the circuit (underestimation in the upper plenum where

vapor condensation and thermophoresis are the dominant mechanisms, overestimation in the

steam generator hot leg where the mechanisms are thermophoresis for all elements + vapor

condensation for I and Cd), noting that too coarse a noding leads to underestimation of

deposition;

•

Containment – relative importance of the two main depletion processes (diffusiophoresis and

gravitational settling), but it is hard to make firm conclusions owing to the variability in the

results;

•

Chemistry – no items identified.

The following phenomena/parameters were not well simulated and substantial model development is

necessary:

•

Bundle – release of medium and low volatiles (e.g. tendency to calculate low for Mo, very high

for Ba, reasonable order of magnitude for Ru and U but considerable scatter), and of structural

materials (Ag/In/Cd from the control rod where the basic process of evaporation from a molten

AIC pool is not captured, tin from the Zircaloy cladding);

•

Circuit – iodine speciation and physical form;

•

Containment – no items identified;

•

Chemistry – gas phase reactions, organic iodine reactions, including production and destruction

through radiolytic processes (definition of optimum parameters for the modelling codes such as

adsorption velocity and desorption rate on/from painted surfaces, and the facility to input the

gaseous iodine fraction at containment entrance, are recommended).

Concerning the circuit, the overestimation of bundle outlet temperature cannot fully explain the upper

plenum results; its main effect is to displace the zone where vapors nucleate. For some elements, part of

the discrepancy in the deposition pattern is due to the wrong prediction of the chemical form, and thus of

its volatility; Cs is generally calculated as a vapor at 700 °C, whereas it was condensed in the experiment.

However, this is also not enough to explain the underestimation in the upper plenum and overestimation in

the steam generator rising line. Finding explanations is presently part of the work performed in the frame

of the interpretation of PHÉBUS-FP tests.

NEA/CSNI/R(2009)5

193

Care is needed in extrapolating the rather good results for the containment directly to the reactor case, as

the PHÉBUS containment thermal hydraulics are relatively simple, and the role of gravitational settling is

overscaled, with a shorter residence time of aerosols in the atmosphere and probably less effect of

agglomeration than for real plants.

Concerning the

source term, the accuracy of containment calculations in integral treatments is sensitive,

often highly, to results of previous stages (propagation of uncertainties). Key features are the calculation of

FP release from the bundle, and of the structural materials Ag, In, Cd and Sn (the kinetics of release of

these and of FPs are as important as the final amounts); the temperatures at the entrance to the circuit,

which strongly influence the deposition pattern; while for those codes which calculate the chemistry, the

speciation is influenced by the calculated release. The release of structural materials was often

undercalculated or not calculated at all, leading to undercalculation of total mass of aerosols, but this had

only a weak impact on overall retention in the reactor coolant system (RCS) and depletion in the

containment. Iodine speciation and physical form in the circuit was poorly predicted - no code reproduced

the observed gaseous iodine fraction in the RCS.

5.2 Circuit

5.2.1 Revaporization in the FPT0, FPT1 and FPT2 circuits

The following is a review of information concerning chemical revaporization of deposits in the circuits of

the FPT0, FPT1 and FPT2 tests. The information comes from γ-spectrometry at the instrument station

known as Point C with corroborating data from γ-spectrometry of the steam generator located a little

further downstream. Some speculation on the mechanism affecting deposited cesium is included (after [1]).

The results of the FPT0 and FPT1 circuits were reviewed by Kissane [1]. Firstly, it must be said that no

evidence can be found of revaporization during FPT0 because the relevant spectrometer signals were

saturated for much of the test. Even so, if the apparent revaporization of hot-leg deposits seen in FPT1 and

2 (see below) also occurred in FPT0, it would probably not have been detectable. This is because the initial

inventory of fission products was quite low and the hot leg was, at the measuring station of Point C, a

low-deposition region of the circuit. For FPT1, evidence of revaporization is clear since about 40 % of the

Cs deposited at Point C (a zone of very light deposition, as in FPT0) was observed to relocate to the SG

riser following termination of the bundle transient - see Figure below. Indeed, this relocation appears only

to stop when isolation of the containment from the circuit occurred, i.e., steam flow stopped. No other

fission product (FP) was observed to relocate implying that relocation was not the result of mechanical

resuspension. Furthermore, the fact that the cesium species released from the deposit involved no other

(measurable) FP significantly limits the possible candidates for the revaporizing species.

For FPT2, at the 21

meeting of the CCIC Zabiego presented in some detail the current status of results

from sequential samples and on-line gamma-spectrometry measurements [2]. The findings are more

complex than for FPT1. Significant deposition was observed along the hot leg during the steam-starvation

phase. Deposits at Point C were then seen to lose some cesium (50 %), iodine (80 %), indium and rubidium

when the steam-starvation phase ended. Given that the flow regime at this location was laminar throughout

the test, mechanical resuspension would seem an unlikely explanation despite the relocation of a number of

FPs (though it would still be prudent to check that a refractory species such as silver was unaffected). The

results support the idea of partial re-vaporization of cesium, iodine, indium and rubidium from hot-leg

deposits when reducing conditions switched to oxidizing conditions. The shut-down phase also produced

loss of cesium from the Point C deposits as observed in FPT1. Coincidentally, downstream in the steam

generator, significant Cs deposition occurred after the core was shut down.

NEA/CSNI/R(2009)5

194

FPT1 - Cs

137

on-line activity measurement in the circuit hot leg (point C) and cold leg (point G)

200

400

600

800

1000

1200

1400

1600

1800

2000

0 5000 10000 15000 20000 25000 30000 3500

time from the test origin (s)

activity (MBq/cm)

zone of saturation

of point C detector

first oxidation phas

main release phas

late oxidation phas

containment isolatio

point C, 700°C

point G, 150°C

radioactive decay

origin : fission

half-life t

1/2

: 30.174 years

isotope with similar behaviour :

136

, Cs

138

Cs desorption from point C

surfaces after reactor shut-down

137

core shut-dow

Fig. 5.2-1 (Figure 22 from [1]; Figure 5.2-18 of the FPT1 Final Report): FPT1 experimental data

on Cs activity at Point C (upper line in red) showing the drop in activity following core

shutdown that continued until steam injection stopped

Concerning the remobilization of Cs, it is well known that some Cs species tend to chemisorb on alloy

surfaces such as Inconel with “trapping” of a fraction of the Cs. The mechanism involves penetration of Cs

into the substrate and reaction with silica, present as an impurity. The trapping appears accentuated by

surface corrosion, longer times and higher temperatures. While the PHÉBUS hot leg was hot enough for

the effect to be very significant, crucially it did not have a pre-corroded surface. Furthermore, there were

revaporization experiments carried out in the RVP project on pipe samples cut from the vertical section of

the FPT1 hot leg. Though not directly comparable to Point C conditions, the sample zone most closely

resembling Point C shows a large fraction being easily resuspended in pure steam at 630 °C with only a

smaller fraction trapped in a low-volatility form. Indeed, despite the mixed nature of the deposit, the

revaporization rate was analogous to that of CsOH measured in control tests. Hence, revaporization of

largely untrapped Cs as CsOH from Point C fits the FPT1 observations and is compatible with some code

analyses.

References

[1] M.P. Kissane, “Interpretation of PHÉBUS tests FPT0 and FPT1: synthesis of analyses of

fission-product behaviour in the circuit”, Note Technique SEMAR 03/96, Oct. 2003

[2] M. Zabiego, L. Pantera, “FPT2 Data Analysis” in N. Girault, “Containment Chemistry Interpretation

Circle”, IRSN Compte Rendu de Réunion DPAM/SEMIC/2004-70, April 2004

5.2.2 FALCON (Including ISP-34)

5.2.2.1 Objectives and procedure

The FALCON small-scale experiments were performed at AEA Winfrith in 1992-3 and were intended as a

preparation for the PHÉBUS FP series of in-pile tests. Their primary purpose was to provide information

NEA/CSNI/R(2009)5

195

on the influence of fission product chemistry on circuit transport and containment deposition. Also

investigated were multi-component aerosol effects and vapor-aerosol interactions. In each experiment one

to six simulant fuel samples were heated by an induction furnace in a silica vessel. The samples contained

concentrations of non-radioactive fission product elements typical of high burnup fuel in a UO

matrix.

These included CsI, CsOH, Te, SrO. BaO and MoO. The pellets were clad in Zircaloy-4. Depending on the

experiment control rod alloy materials could also be included, clad in stainless steel. The temperatures

were increased sufficiently to burst the fuel and control rod pellets. Heating was then continued up to 3000

K to produce a continuous release of materials, transported by a carrier gas through the model circuit. The

carrier gas was mainly helium admixed with low concentrations of steam and/or hydrogen. Optionally

boric acid could be injected above the heated region.

5.2.2.2 Model circuit and experimental details

The circuit consisted of a thermal gradient tube of length 480 mm followed in the primary circuit

experimental series by a straight stainless steel pipe 800 mm long. In the containment series this pipe was

replaced by a stainless steel circuit with several bends followed by a steel box modelling the containment.

Analysts have largely considered the data from the thermal gradient tube to be of most interest [1] and this

component and the experimental data obtained from it will now be described in more detail. The geometry

of the thermal gradient tube varied during the test series. In the earlier tests, including FAL-17 and FAL-18

the first section of length 100 mm had a diameter of 35 mm, while for the remaining 380 mm the diameter

was reduced to 25 mm. In subsequent tests, including FAL-19 and FAL-20 the diameter was 25 mm

throughout. A series of five independently heated zones, each of length approx. 60 mm, allowed a wall

temperature profile to be maintained. A uniform temperature gradient was generally imposed, from 1100 K

to 400 K approximately. Both wall and gas temperatures were measured at selected points along the tube.

The tube was lined with stainless steel foil, which could be removed after the test for analysis. Bulk

deposits were examined with ICPOES, atomic absorption spectrometry, and colorimetry. Iodine was

collected in Maypacks and examined with X-ray fluorescence spectroscopy. Individual particles were

measured using energy dispersive spectroscopy with a scanning electron microscope, and particle

characterisation was also applied to obtain size and morphology data. Speciation was studied through

powder X-ray diffraction and X-ray photoelectron spectroscopy.

5.2.2.3 FALCON tests and ISP 34

As part of the ongoing series of code validation problems promoted by the CSNI [2] International Standard

Problem 34 was based on four tests of the FALCON series, FAL-17 to FAL20 [3, 4, 5, 6]. The first two

were primary circuit tests while the latter two used a circuit including a containment vessel. However as far

as circuit phenomena are concerned this made little difference. We here focus attention on the thermal

gradient tube. The sources differed in the materials injected into the carrier gas stream, as follows:

FAL-17 FAL-18 FAL-19 FAL-20

He (l/min) 10 10 4 4

Steam (cm

/min at 20 °C) 0.21 0.21 3.1 0

5 % of H

O 5 % of H

O 5 % of He

Boric acid (ppm) 2000 0 2000 0

The experimental results offered for comparison with analyses were the deposition profiles for a specified

series of elements.

NEA/CSNI/R(2009)5

196

5.2.2.4 Analyses

The ISP proved popular, attracting 50 submissions from 12 countries, using the system or circuit codes in

use at the time, including MELCOR VICTORIA and RAFT. The final report was issued in December 1994

[7, 8]. It was concluded that several chemistry issues of aerosol and fission product behaviour required

further code improvements, and that there was a strong user effect. It also appeared that revaporisation

might have been important in determining the final deposition profiles. Non-representative aspects of the

experiments identified included laminar flow conditions throughout and a high surface to volume ratio.

There were also uncertainties in the mass release rates.

[1] pointed out that for constant wall temperatures and carrier gas conditions (obtaining during most of

each experiment analysed) the percentage of a given element deposited at a specified location was

independent of the instantaneous release rate. These authors were thus able to treat thermal gradient tube

deposition in FALCON as a steady flow problem. They also noted that the deposition profiles for vapors

were qualitatively different from those for aerosols, the former showing a threshold dependent on the

condensation temperature. They used this finding plus correlations between the deposition profiles of

different elements to obtain information on speciation. It was concluded that

•

The results show that speciation is important, and can vary with the carrier gas composition

•

In the presence of excess boric acid Cs is predominantly in the form of cesium borate. In the

concentrations present in FALCON the borate condenses at a temperature 250 K higher than the

hydroxide

•

The iodine data are ambiguous; possibly in FAL-18 the dominant species was CsI, but the

volatility of iodine varied between tests, suggesting differing speciation

•

Molybdenum was more volatile in the boric acid tests

•

Cadmium showed similar behaviour in all tests

•

Silver was an aerosol in the test in reducing conditions, but entered the thermal gradient tube as a

gas in the other experiments

•

The deposition velocities of aerosol particles (deposited mostly by thermophoresis) appeared to

be largely independent of the aerosol composition

5.2.2.5 Status

Currently there are no ongoing analyses of the FALCON tests. The thermal gradient tube data from this

series are however are of good quality, despite uncertainties in the timing and magnitude of the releases

from the simulant source, and might usefully be revisited to help shed light on the circuit data from

PHÉBUS FP, which are copious but not always easy to interpret.

References

[1] I.M. Shepherd, Y. Drossinos, C.G. Benson Modelling fission product vapor transport in the

FALCON facility Nuclear Technology 110,181-197, 1995

[2] CSNI: CSNI International Standard Problems (ISP) Brief description (1975-1999).

NEA/CSNI/R(2000)5. May 2000

[3] A.M. Beard, P.J. Bennett Data report 23, Integral test 17, FAL/P(92)77 Winfrith Technology Centre,

May 1992

NEA/CSNI/R(2009)5

197

[4] A.M. Beard, P.J. Bennett FALCON data report 24, Integral test 18, FAL/P(93)83 Winfrith

Technology Centre, Jan 1993

[5] A.M. Beard, P.J. Bennett FALCON data report 26, Integral test 19, FAL/P(93)90 Winfrith

Technology Centre, Feb 1993

[6] A.M. Beard, P.J. Bennett FALCON data report 27, Integral test 20, FAL/P(93)94 Winfrith

Technology Centre, May 1993

[7] D.A. Williams OECD International Standard Problem No. 34, FALCON code comparison report.

AEA RS 3394, AEA Winfrith (June 1994)

[8] D.A. Williams OECD International Standard Problem No. 34 FALCON fission product experiments

FAL-ISP-1 and FAL-ISP-2 Comparison Report. NEA/CSNI/R(94)27 December 1994

5.2.3 STORM (Including ISP-40)

5.2.3.1 Recent analyses of the STORM deposition and resuspension experiments (ISP-40)

The STORM series of experiments was performed at the European Commission’s Joint Research Centre,

Ispra (Italy) in the period 1996-97. Each experiment was divided into two phases. In the deposition phase a

SnO

simulant aerosol carried by a steam-air-nitrogen mixture through a 63 mm diameter horizontal pipe

of length 5000 mm built up a layer of deposit, mostly by thermophoresis and eddy impaction, while in the

resuspension phase a higher-velocity nitrogen flow mechanically resuspended and relocated a fraction of

the deposit. The international standard problem ISP-40 was based on the two phases of STORM test 11 [1],

and is described elsewhere in Section 5.1.5.1. Subsequent to the publication of the ISP-40 Final Report a

number of teams have continued to analyse tests of the STORM series with the aim of providing further

support and validation to resuspension models, judged to be inadequate in the conclusions of the ISP.

Most of the information available on recent STORM analyses is to be found in the Final Report of the

EVITA project [2, 3] and concerns Test SR11. There has also been some work on STORM test SR09 [4],

the essential differences of which from SR11 are as follows. The deposition phase used the same simulant

aerosol SnO

and a similar steam-nitrogen-air mix as carrier gas, but took place at the slightly lower

temperature of 340 °C. The temperature difference between the test section wall and the gas stream was

50 °C, as compared with 100 °C in SR11. The carrier velocity was higher, 24.9 m/s versus 18 m/s in the

later test. This combination was expected to result in more deposition by impaction and relatively less by

thermophoresis. The deposited mass was similar in the two tests. The resuspension phase of SR09 took

place in a single step, at a nitrogen velocity of 52.3 m/s. As in SR11, to minimise thermophoresis during

this phase the gas and wall temperatures had the same value, 340 °C.

A team at VEIKI has re-analysed test SR11 using the SOPHAEROS module of the integral code ASTEC

v0.3. The test section was divided into 11 longitudinal sections, and the aerosol size distribution was

divided into 20 logarithmically equal size bins, ranging from 2 nm to 40 µm. These analysts found that

nearly all the deposition was by thermophoresis, and succeeded in calculating the mass deposited to within

30 %. The profile of the deposit along the tube was nearly flat in the calculations, although in the test an

estimated 30 % reduction in surface density of aerosol was seen from inlet to outlet. The calculated deposit

was then used in the resuspension phase. The resuspension rate calculated by the semi-empirical model

used in the code was found to be constant during each velocity step, and with default values for the

adhesive force and resuspension rate parameters matched the total mass resuspended rather well for the

earlier steps. As carrier gas velocities exceeded 60 m/s, however, the resuspended mass became

NEA/CSNI/R(2009)5

198

underpredicted, by a factor of two at the end of the final velocity step. Significantly, redeposition of

resuspended aerosol by impaction was predicted towards the outlet of the test section.

The JRC team and collaborators have continued the analysis of experiments of the STORM series,

including the deposition tests SD04, SD05, SD07, SD08, SR09, SR11, SR12 and the resuspension tests

SR09 and SR11 as well as SR12, again using SOPHAEROS. The deposition phase of SR10 was similar to

that of SR09 but with a lower carrier velocity, and the velocity during the resuspension phase was double

that in SR09. The deposition phase of STORM SR12 was identical to that of SR11, while the resuspension

phase had three steps with carrier velocities increasing from 90 to 110 m/s. Like the VEIKI team the JRC

reseearchers were able to achieve good prediction of the deposition phase in SR11, and their results for the

resuspension phase are similar.

The SOPHAEROS model [5] decides whether particles of a given size will resuspend on the basis of a

balance of adhesive and aerodynamic forces, the rate of the subsequent resuspension being calculated by a

semi-empirical model originally used in ECART. Predominant among the adhesive forces in the STORM

tests is the cohesive force, proportional to a user-specified cohesive force coefficient. With the default

value of this coefficient the predicted resuspension rate in SR09 is too great, but the JRC team found that

good agreement with the data, including the final profile of the deposit, could be achieved with a cohesive

force coefficient ten times the default value. The JRC team has also analysed tests SR10, SR11 and SR12

using the SOPHAEROS implementation of the Reeks-Biasi rock’n’roll model (see below). It has been

found that the rock’n’roll model underestimates the final resuspended fraction in all the tests analysed. The

time-dependence of the resuspension behaviour is also different from that observed. Nearly all the

resuspension is predicted to occur within a very short time, whereas the experimentally observed

resuspension has a duration of some 25 s.

The JRC analysis also points up the importance of the deposition mode in determining the resuspension

behaviour. In test SR10 a lower temperature difference between pipe wall and carrier gas resulted in a

significant contribution from eddy impaction to the total deposition, accompanied by an increased aerosol

mass median diameter. During the resuspension phase resuspension began at a lower fluid velocity than in

tests where the deposition was predominantly by thermophoresis (80 m/s in SR09 compared to 100 m/s in

SR12).

The use of semi-empirical models with adjusted coefficients is not very satisfactory, and a less empirical

resuspension model has been developed by [6] which is claimed to account for the resuspension behaviour

observed in several STORM tests, in preceding test series which were reviewed in the SOAR on circuit

aerosol behaviour [7] and in other experiments available in the literature, without becoming too complex

for use in engineering safety codes. The model assumes a log-normal distribution of adhesive forces, and

includes correlations for the geometric mean and standard deviation of the adhesive force distribution. It

has been found that the effective adhesion is much less than for particles in smooth contact with the pipe

surface, consistent with an effect of surface roughness, and that sub-micron particles are more tightly

bound when a deposit already exists than when there is less than a mono-layer of deposit present. The Biasi

model has been implemented in SOPHAEROS.

It should be noted however that in STORM and in other tests e.g. the VTT PSAERO [8] tests, the

resuspension rate for a given gas velocity declines with time, and there is also a contribution apparently

due to gas acceleration [9] effects not currently accounted for (see chapter 3.5.1). Gas velocities favouring

resuspension also favour turbulent impaction, the models for which are not fully validated, and in

resuspension experiments the two processes are likely to occur simultaneously. Compounding of model

uncertainties may therefore occur during their analysis. A further point is that essentially all the data

considered in ISP-40 and in subsequent related developments refer to dry aerosols, and separate-effect tests

to provide data on the resuspension of wetted and semi-liquid deposits do not appear to have been

NEA/CSNI/R(2009)5

199

performed. There is thus no great incentive at present for the development of models treating resuspension

from such deposits.

References

[1] P. Dilara, A. Krasenbrink, R. Hummel STORM test SR-11: ISP-40 Quick Look Report Technical

Note, Joint Research Centre, Ispra (Italy), March 1998

[2] H.-J. Allelein, K. Neu, J.P. Van Dorsselaere, K. Mueller, P. Kostka, M. Barnak, P. Matejovic, A.

Bujan, J. Slaby European validation of the integral code ASTEC (EVITA).

Nucl. Eng, and Desn.

221 (2003) 95-118

[3] H.-J. Allelein, K. Mueller et al. European validation of the integral code ASTEC (EVITA) Final

Report, Parts 1&2. European Commission report (2004)

[4] P. Dilara, A. Krasenbrink, R. Hummel, J. Area Capitao STORM test SR-09: Deposition of SnO

Partially Insulated Pipes with Steam and Resuspension of SnO

from Partially Insulated Pipes with

Quick Look Report. Technical Note, Joint Research Centre, Ispra (Italy), February 1998

[5] M. Missirlian, N. Alpy, M.P. Kissane “ASTEC V0 SOPHAEROS code version 2.0. Fission product

analysis in PWR circuits Theoretical manual, Rev 0", ASTEC-V0/DOC/00-06. NT/SEMAR/01/98,

December 2001

[6] L. Biasi, A. de los Reyes, M. Reeks, G. de Santi Use of a simple model for the interpretation of

experimental data on particle resuspension in turbulent flows

Journal of Aerosol, Science

32,1175-1200 (2001)

[7] A.L. Wright et al. Primary System Fission Production Release and Transport NEA/CSNI/R(94)2

(1994)

[8] A. Auvinen

, J.K. Jokiniemi

,. A. Lähde

, T. Routamo

, P. Lundström

, H. Tuomisto

, J. Dienstbier

S. Güntay

, D. Suckow

, A. Dehbi

, M. Slootman

, L. Herranz

, V. Peyres

, J. Polo

“SG tube rupture

(SGTR) scenarios” Nuclear Engineering and Design 235 (2005) 457–472

VTT Processes, Biologinkuja 7, P.O. Box 1602, VTT Espoo 02044, Finland

Fortum Nuclear Services, Vantaa, Finland

Nuclear Research Institute Rez plc, Czech Republic

Paul Scherrer Institute, Villigen-PSI, Switzerland

Nuclear Research and Consultancy Group, Arnhem, Netherlands

Centro de Investigaciones Energeticas, Medioambientales y Tecnologicas, Madrid, Spain

[9] N. Ardey, F. Mayinger Aerosol resuspension by highly transient containment flow: insights by

means of laser optical methods

Kerntechnik, 68-75 (1998)

5.2.3.2 ISP-40: aerosol deposition and resuspension

Uncertainty in the modelling of aerosol resuspension had been mentioned as as a problem in the SOAR of

1994 [1], although it was found difficult to ascertain its safety import. International Standard Problem

ISP-40 was set up to address aerosol deposition and resuspension phenomena in the reactor cooling system

and was based on test SR11 of the STORM series performed at the European Commission’s Joint Research

Centre, Ispra (Italy) in 1997 [2]. The scenario was one of nuclear aerosol deposition in the relief lines of a

PWR during a steam blackout followed by resuspension of the deposits by the steam surge resulting from a

core slump. Like other STORM tests the experiment upon which this ISP was based took place in two

NEA/CSNI/R(2009)5

200

Resuspension in steps

100

150

200

250

02468

Step no.

g/s N2

g deposit

phases. In the deposition phase SnO

simulant aerosol was transported through a horizontal test section

(diameter 63 mm, length 5000.5 mm) by a carrier gas consisting of a mixture of steam, N

, air, Ar and He.

The constant gas velocity was approx. 18 m/s, and while the gas temperature was approx 370 °C the wall

temperature was 100 °C cooler so that deposition took place largely by eddy impaction and by

thermophoresis. The aerosol had a GMD of 0.43 µm and a GSD of 1.7, and the deposition phase continued

long enough to build up 162 g of deposit.

The test section and its deposit were then kept at 370 °C for a period of about 24 hours, during which the

following phase of the test was prepared. In this resuspension phase, nitrogen gas was passed through the

test section in a series of mass flow rate plateaux, and the mass resuspended during each phase and its size

distribution were measured by downstream sampling stations. Velocities up to 127 m/s were investigated.

Fig. 5.2-2 shows the mass flow rates for the various steps as well as the mass remaining in the deposit after

each step. A feature of the STORM tests is that the mass distribution of the deposit along the pipe could be

measured by a radioactive tracer, and such profile data were collected in test SR11 for comparison with

predictions.

Fig. 5.2-2 Mass flow rate and mass remaining in the deposit

The ISP attracted 18 participants, who could choose to calculate one or both of the experimental phases.

The deposition phase was run blind, while for the resuspension phase participants were supplied with the

size distribution of the resuspended mass for each step in the gas flow. Contributions were collated in

January 1998. It was subsequently found that the steam flow during deposition had been given incorrectly.

More importantly, the sizes of the resuspended aerosols had been overestated by a factor of two. Working

now with open conditions, participants were invited to recalculate their submissions. Sensitivity studies

were also produced by some participants, and the final report of the ISP [3] was based on all submissions

received, taking due account of the data available to the participants at the time each calculation was made.

The final report came to the following conclusions:

Modelling of thermophoretic deposition is adequate

There are problems with the modelling of deposition due to turbulent flows

NEA/CSNI/R(2009)5

201

Aerosol deposition and resuspension need to be treated together

Aerosol retention depends strongly on the thermal-hydraulic conditions

One-dimensional bulk parameter modelling is generally adequate to calculate aerosol retention in

fully developed flow in straight pipes

The particle tracking codes used in this ISP are still in a preliminary phase of their development and,

while having a stronger physical basis, generally require the knowledge of parameters which are not

generally known

In terms of the results obtained, there was a tendency of almost all codes to overpredict aerosol

deposition in the test pipe

Aerosol resuspension can significantly affect the source term in the case of dry aerosol deposits in

turbulent flows

Experimental data is needed for resuspension of aerosol mixtures with different liquid fractions

10.

Present aerosol resuspension models are inadequate

11.

The potential for resuspension depends strongly on the characteristics of the deposit

12.

Deposition models should give an indication of the state of the deposit, not only of the mass

deposited

13.

Separate effects tests are needed to relate the characteristics of the deposit to their chemical

composition and to the mechanisms by which the deposit was formed

14.

Mono-layer resuspension models are only a step towards the development of multi-layer models

Modelling and code effects

It is of interest for the appreciation of the impact of uncertainties in resuspension modelling to

quantify the variation in model predictions seen in the ISP-40 submissions.

Fig 5.2-3 shows the calculated masses of aerosol resulting from the deposition phase, and compares them

with the experimental value. With the exception of those from MELCOR users and the particle tracking

code MARIE predictions are similar, and may be considered acceptable. Note that the result marked Pisa-2

is produced by allowing resuspension to take place simultaneous with deposition. The predicted very low

net deposited mass is a result of too much resuspension.

NEA/CSNI/R(2009)5

202

Fig. 5.2-3 Total deposited mass

The situation is less encouraging for the resuspension calculations. Fig. 5.2-4 provides the fractions of

deposited mass remaining after each step of the resuspension part of the test, as calculated by each code but

normalised to the same total deposited mass to compensate for differences in calculations of the deposition

phase. The column corresponding to the experimental data is marked SR11. Very wide variations in

predicted resuspension behaviour with increasing gas flow can be observed, quite apart from the

differences in predicted total resuspended mass. For instance, ART predicts near-total resuspension at the

first velocity step, whereas one of the SOPHAEROS predictions has over 65 % of the deposit still present

after the final step with highest gas flow. The substantial user effect also visible (see the pairs of

calculations made using CAESAR, ECART, SOPHAEROS or VICTORIA) arises from uncertainties in the

parameters of the particular resuspension model implemented in each code. In addition, most codes

experienced difficulty in calculating the time-dependence of resuspension during each step. In the

experiment most of the resuspension took place over a relatively short time at the beginning of the step,

whereas a number of models calculated a constant resuspension rate for given thermal-hydraulic

conditions.

NEA/CSNI/R(2009)5

203

Fig. 5.2-4 Aerosol mass remaining in the deposit

Besides the resuspended masses participants were also invited to reproduce the measured size distribution

of resuspended aerosol. It was found in the test that increasing gas flow resuspended progressively smaller

particles, and this trend was broadly reproduced in the submitted calculations. However, the absolute sizes

varied very widely (see Fig. 5.2-5) and the trend mentioned was nearly absent in some calculations but

exaggerated in others.

Fig. 5.2-5 Geometric mean diameter of the particle at the outlet of the test pipe

NEA/CSNI/R(2009)5

204

References

[1] A.L. Wright et al. Primary System Fission Production Release and Transport NEA/CSNI/R(94)2

(1994)

[2] P. Dilara, A. Krasenbrink, R. Hummel STORM test SR-11: ISP-40 Quick Look Report Technical

Note, Joint Research Centre, Ispra (Italy), March 1998

[3] A. de los Reyes Castelo, J.A. Capitão, G. de Santi International Standard Problem 40 – Aerosol

Deposition and Resuspension Final Comparison Report, Joint Research Centre, EUR 18708 EN,

OECD NEA/CSNI/R(99)4 February 1999

5.2.4 WIND

WAVE experiments

Small-scale basic tests have been performed with the WAVE (Wide range Aerosol model Verification)

facility [1] at JAERI to investigate the cesium iodide (CsI) deposition onto the inner surface of a pipe

under typical severe accident conditions. In the tests, a crucible including the CsI powder was placed at

upstream part of the electrically heating test pipe and generated CsI vapor or aerosol was delivered to the

downstream pipe by nitrogen gas. Several coupons made of SUS304 were set in the test pipe and the

deposited mass on them were measured after the tests.

The tests with a horizontal straight pipe showed that a relatively large amount of CsI was deposited on the

upstream floor (bottom) of the pipe and at the downstream ceiling (top) of the pipe [1]. Analyses of the

experiments have also been conducted with the three-dimensional thermal-hydraulic code and the

radionuclide transport analysis code, ART [2, 3]. The experimental results were reproduced well with ART

by using a peripherally subdivided pipe cross section and associated representative thermal-hydraulic

information from the thermal-hydraulic code prediction. These experiments and analyses indicate that the

major deposition mechanisms for the chemical form of CsI are thermophoresis and condensation, both of

which depend on the thermal gradient in the gas.

Moreover, several tests using a quartz glass pipe with a 90 degree bend have been performed to investigate

the CsI aerosol behaviour at the bend and the effect of pipe orientation after the bend on aerosol behaviour

[4]. Three tests with horizontal, upward, and downward pipe orientation after the bend showed

enhancement of CsI deposition in the upward vertical pipe due to a transition from laminar to turbulent

flow. The flow regime in the horizontal and downward pipes was kept at laminar flow in spite of the same

flow velocity. The reason for turbulent flow only in the upward pipe is that the downward flow near the

low-temperature pipe surface is opposite to the direction of main stream. The ART calculation which

considers the thermal-hydraulics of only the main stream reproduced well the thermophoretic deposition in

the horizontal and downward pipes but underpredicted the deposition for the upward vertical pipe by one

order of magnitude. To precisely predict the deposition in the upward oriented pipe, the flow regime

should be evaluated carefully by taking into account the natural circulation in the pipe as well as the main

stream.

These studies showed that the coupling of the radionuclide behaviour and the detailed fluid-dynamic

phenomena was essential to accurately predict the CsI deposition at a bend or at a vertical pipe as well as

in a horizontal straight pipe. Little attention has been paid to such coupling in previous studies.

NEA/CSNI/R(2009)5

205

WIND experiments

The piping of reactor coolant system could be heated by high temperature gases from the reactor core

region and the decay heat of the deposited fission products (FP). Under such conditions, the integrity of the

reactor piping would be threatened by this temperature increase together with the high system pressure.

The WIND (Wide range piping Integrity Demonstration) project [5, 6] was initiated in 1993 to evaluate the

integrity and safety margin of reactor piping under severe accident conditions. The WIND project consists

of tests and analyses related to aerosol behaviour and piping structural integrity.

A schematic diagram of the WIND facility is shown in Fig. 5.2-6.

HHHHHH

H H H H H

HHHHHH

H H H H H

Argon Gas

Supply System

Buffer Tank

Induction Heating Type

Aerosol Generator

Nebulizer Type

Aerosol Generator

Upstream Test Section

Water

Mist Separator

Heat Exchanger

Dumper

Aspirator

Exhaust

Drain

Waste Water

Tank

Pump

Water

Steam

Generator

Aerosol Sampling

System

HEPA

Filter

Downstream Test Section

Water Purification

System

Aerosol

Chamber

Mixer

Fig. 5.2-6 Schematic diagram of the WIND facility

In the aerosol deposition tests, cesium iodide (CsI) aerosol was introduced into two horizontal test sections

connected in series. These sections consist of straight stainless steel pipes with an inner diameter of

approximately 10 cm and a length of 2 m. Argon or a mixture of argon and superheated steam was used as

a carrier gas to suspend and transport the aerosol into the test sections. Temperature gradients with the

maximum temperature of 1273 K and 973 K were developed in the test sections. The deposition density of

cesium onto the upstream test section in WAV1-D is shown in Fig. 5.2-7. The test results on the spatial

distribution of the aerosol deposition and the thermo-fluiddynamic analysis with the WINDFLOW code [7]

showed that the formation of a natural convective secondary flow of gases resulted in the deposition of CsI

onto the ceiling and side wall areas of the test sections and that the dominant deposition mechanisms were

the condensation of vapor and the thermophoresis of CsI aerosol [8]. No remarkable decomposition of CsI

was observed in the tests with environments of argon and superheated steam. The results obtained from the

aerosol deposition tests have been applied for the validation of the ART code, which is used for a detailed

analysis of the FP behaviour during a severe accident.

NEA/CSNI/R(2009)5

206

0.0

0.2

0.4

0.6

0.8

1.0

1.2

Floor

Wall 270°

Ceiling

Wall 90°

0.5 1.0 1.5 2.0

Deposition Surface Density (mg/cm

)

Axial Location (m)

0.0

(After deposition phase)

Fig. 5.2-7 Deposition density of cesium onto the upstream test section in WAV1-D

References

[1] A. Hidaka, et al. "Experimental and Analytical Study on the Behaviour of Cesium Iodide

Aerosol/Vapor Deposition onto Inner Surface of Pipe Wall under Severe Accident Conditions" J.

Nucl. Sci. Technol. 32 (10), pp.1047-1053 (1995)

[2] T. Ishigami et al. "User's Manual of ART Code for Analyzing Fission Product Transport Behaviour

During Core Meltdown Accident" JAERI-M 88-093 (1988).

[3] M. Kajimoto et al. "Development of THALES-2, A Computer Code for Coupled

Thermal-Hydraulics and FP Transport Analyses for Severe Accident at LWRs and Its Application to

Analysis of FP Revaporization Phenomena" Proc. Int. Topical Meeting on Safety of Thermal

Reactors, Portland, pp.584-592 (1991)

[4] A. Hidaka, et al."Deposition of Cesium Iodide Particles in Bends and Sections of Vertical Pipe under

Severe Accident Conditions" J. Aerosol Sci., 31(9), pp.1045-1059 (2000)

[5] K. Hashimoto, et al."Current Status of WIND Project" JAERI-Conf 99-005, pp.161-164 (1999)

[6] A. Hidaka, et al. "Experimental and Analytical Study on Aerosol Behaviour in WIND Project" Nucl.

Eng. and Des., Vol.200, pp.303-315 (2000)

[7] Y. Maruyama et al.Vapor Condensation and Thermophoretic Aerosol Deposition of Cesium Iodide

in Horizontal Thermal Gradient Pipes J. Nucl. Sci. Technol., 36(5), pp. 433-442 (1999)

[8] H. Shibazaki et al. Revaporization of CsI Aerosol in a Horizontal Straight Pipe in a Severe Accident

Condition Nucl. Technol., 134, pp. 62-70 (2001)

NEA/CSNI/R(2009)5

207

5.3 Containment Experiments

5.3.1 DEMONA

The DEMONA tests were performed to investigate the natural aerosol removal in a PWR containment

under core melt accident conditions. Their main objective was to demonstrate the validity of the aerosol

code NAUA. From 1983 to 1986 nine large-scale tests were carried out in the Battelle model containment

in Frankfurt/Main [1, 2]. The programme was sponsored by the German Federal Minister for Research and

Technology (BMFT) and the USNRC.

The 630 m

model containment was made of reinforced concrete and had roughly the shape of a German

PWR containment. The interior was subdivided into several compartments (Fig. 5.3-1).

Fig. 5.3-1 Battelle model containment for DEMONA tests

In each test the containment structures were heated up by steam releases first. Then the test aerosol

generated by three plasma torches was injected up to a maximum concentration of about 10 g/m

. Aerosol

material and composition as well as the thermal hydraulic conditions varied in each test (Table 5.3-1).

Except for test A9 a single room configuration of the containment was used.

Table 5.3-1 DEMONA programme test matrix

Test

Aerosol

material

Description

A1 SnO

Thermal hydraulic test without aerosol

B2 SnO

Dry test without volume condensation

B3 SnO

Base test with volume condensation

B4 SnO

Base test, repetition

B5 SnO

Test with delayed volume condensation

B6 Ag + MgO Test with low concentration and delayed volume condensation

A7 Fe

Base test, repetition with iron oxide aerosol

A8 Fe

+ SnO

Test with mixed aerosol

A9 Fe

+ SnO

Multi-compartment test with mixed aerosol and delayed volume condensation

NEA/CSNI/R(2009)5

208

The base test B3 simulated a scenario with a late overpressure failure of the containment. After the

SnO

-aerosol release the atmospheric pressure was kept constant by a steam injection near the bottom

which guaranteed a well mixing of the containment atmosphere. During the aerosol depletion phase a

moderate volume condensation took place accelerating the aerosol settling (Fig. 5.3-2).

A9 was the only DEMONA Test with a multi-compartment configuration. A four-room-geometry was

formed by closing several flow openings. The mixed aerosol injection was followed by a dry and a wet

depletion phase. The aerosol was inhomogeneously distributed with generally higher concentrations in the

lower compartments than in the upper ones (Fig. 5.3-3). Because of the limited number of aerosol sampling

stations it was not possible to measure spatial aerosol distribution in detail.

Fig. 5.3-2 Aerosol depletion in DEMONA base test B3

Fig. 5.3-3

Aerosol depletion in the DEMONA multi-compartment test A9

NEA/CSNI/R(2009)5

209

Calculations with the aerosol code NAUA showed a good agreement with the measured aerosol behaviour

in the dry DEMONA tests but had some deficits for the wet tests. The multi-compartment effects measured

in test A9 could not be considered by the single volume code NAUA. DEMONA test B3 was the subject of

a code comparison study initiated by the CEC [3].

References

[1] T. F. Kanzleiter DEMONA-Versuche Abschlussbericht, Final Report; BIeV-R65.523-01 (1987)

[2] W. Schöck, H. Bunz, R.E. Adams, M.L. Tobias, F.J. Rahn Large-Scale Experiments on Aerosol

Behaviour in Light Water Reactor Containments Nuclear Technology 81, pp 139 - 157 (1988)

[3] W. Schöck Assessment of the outputs from the various computer codes used in Member States to

study the behaviour of aerosols in LWR containments in case of severe accidents Final Report to

the, CEC, ECI-1413-B7210-85-D, KfK Karlsruhe (1987)

5.3.2 MARVIKEN-V

In the large-scale MARVIKEN-V tests (1982 to 1985) the transport and deposition of aerosols and volatile

species were measured in a simulated LWR primary circuit [1]. The reactor vessel, the pressuriser, and the

relief tank were the main components interconnected by pipes (Fig. 5.3-4). In five tests performed the

circuit geometry and the test conditions were varied. The effects of different temperature ranges (25 °C to

1200 °C), superheated steam, condensing steam, and water were studied. Non-radioactive materials were

used for fission product simulants (CsOH, CsI, Te) and core structure simulants (Ag, Mn). The materials

were vaporized by separate plasma torches forming condensation aerosols. The aerosol concentrations in

the tests lay between 36 and 132 g/m³.

Fig. 5.3-4 Sketch of the MARVIKEN-V test facility

NEA/CSNI/R(2009)5

210

Table 5.3-2 shows the test matrix. In the first three tests only the part of the circuit downstream of the

reactor vessel was used and thermal hydraulic conditions were varied. In tests 4 and 7 the whole circuit

was integrated and sequences with simultaneous core damage with fission product release and structural

aerosol release were measured. Table 5.3-3 summarises the results showing the aerosol retention in the

components and pipes measured in the five tests. The retention in the circuit ranged from 15 % in test 2A

to 74 % in test 4. Sedimentation and inertial impaction were the dominant deposition phenomena observed.

The data show also the high aerosol removal efficiency of the relief tank when filled with water. There are

some indications that the fissium and corium aerosols were not well-mixed in the reactor vessel. The

MARVIKEN –V experiments demonstrated that LWR primary circuits have some aerosol retention

capability. The generated data have been used for code validation.

Table 5.3-2 MARVIKEN-V test matrix

Test Geometry Conditions Aerosol

Pressuriser

↓

Pipe

↓

Relief tank

High temperatures

Relief tank dry

Fissium

Low temperature

Condensation in piping

Water in relief tank

Reactor vessel

PWR internals

↓

Pipe

↓

Pressuriser

↓

Pipe

↓

Relief tank

High aerosol concentration

Water in relief tank

Fissium + Corium

High aerosol concentration

Water in relief tank

Fissium

Fissium: simulated by a mixture of CsOH, CsI, Te;

Corium: simulated by a mixture of Ag, Mn

Table 5.3-3 MARVIKEN-V, Aerosol concentration and retention in different components (in %)

Test 1 2A 2B 4 7

Aerosol concentration g/m³ 35 62 51 132 52

Retention, %

Reactor vessel - - - 30 11

Piping to pressuriser - - - 8 3

Pressuriser 32 14 45 25 6

Piping to relief tank 4 1 5 11 20

Relief tank 10 85 49 26 59

Scrubber 41

Final filter 0.3 0.1 0.04 0.1 0.2

NEA/CSNI/R(2009)5

211

Reference

[1] J.O. Liljenzin, J. Collén, W. Schöck, F.J. Rahn Report from the MARVIKEN-V / DEMONA /

LACE Workshop Proceedings of the OECD/NEA Workshop on Aerosol Behaviour and

Thermal-Hydraulics in the Containment, Fontenay-aux-Roses (France), 26-28 Nov. 1990 CSNI

Report No. 176

5.3.3 LACE

The LACE tests [1-4, 6] were sponsored by an international consortium in the later half of the 1980's when

the understanding and identification of issues concerning aerosols in nuclear reactor accidents were still

developing. These tests investigated the behaviour of aerosols in high velocity gas flows and in large

models of containment buildings. Four LA tests were conducted. One of these tests, LA3, consisted of

three phases denoted by LA3A, LA3B, and LA3C. These LA tests were supplemented by three tests of

aerosol behaviour in high velocity gas flow through pipes called the 'Containment Bypass' or CB tests [5].

The facility used in the LA and CB tests is shown in Fig. 5.3-5. It consists of a piping system connected to

an 852 m

volume used variously to simulate an auxiliary building or a reactor containment building. In

tests of aerosol behaviour in high velocity gas flows, only the piping system of the facility was used.

Fig. 5.3-5 Schematic diagram of the facility used for the LACE tests

For the LA tests, the piping system was a 6.3 cm diameter, about 28 m long pipe with six 90° bends, two

vertical sections and four horizontal sections. In the CB tests, the 6.3 cm pipe connected to a 30 cm

diameter pipe and then to a scrubber.

Some of the details of the LA tests are shown in Table 5.3-. Test objectives and the qualitative results of

the tests are:

NEA/CSNI/R(2009)5

212

•

LA-1:

The LA-1 test was intended to simulate a containment bypass accident sequence with aerosol

passing through the reactor coolant piping system into a reactor auxiliary building vented to the air.

The aerosol used in this and other LA tests was a mixture of water-soluble CsOH and insoluble

MnO. More than 98 % of the aerosol was retained in the piping system. The aerosol behaviour in the

auxiliary building model is shown in Fig. 5.3-6 Revaporization of deposited aerosol was not

investigated in the test.

•

LA-2:

The LA-2 test examined the leakage from a containment volume with relatively large leaks at

elevations of 26.5 and 4.0 meters. The test was to explore the possibility of density stratification of

the aerosol-laden gas phase in the containment model much as had been suggested by results of the

DEMONA tests. About 38 % of the injected aerosol mass was allowed to leak from the containment

volume. No significant differences were observed in the mass escaping through high and low leaks

and it was concluded that no significant stratification of the gas phase in the containment volume

occurred.

•

LA-3:

The LA-3 test examined aerosol behaviour in high velocity flows through the piping system only.

The 3 phases of the test examined the effects of the gas velocity and the ratio of soluble, liquefied,

aerosol and insoluble, solid aerosol. In general, 70 % or more of the aerosol material was retained in

the piping system. Again, revaporization of the deposited aerosol was not studied in the LACE

program.

•

LA-4:

The LA-4 test examined the behaviour of aerosol in a containment. Cesium hydroxide aerosol was

injected for 50 minutes after the start of the test. MnO aerosol injection was started after 30 minutes

and continued for 50 minutes. During aerosol injection, the pressure in the containment model

increased to about 3 bar. It was held at this pressure for 200 minutes and, then, the containment

model was depressurised to simulate containment failure.

NEA/CSNI/R(2009)5

213

Fig. 5.3-6 Aerosol behaviour in the auxiliary building model during the LA-1 test

The CB tests examined aerosol behaviour in high velocity gas flows and were conducted much like the

LA-1 test though there were some differences in the piping system. Features of the tests are shown in

Table 5.3-5. The tests showed that large quantities of aerosol can be retained in the piping system. The

dispositions of aerosol following the tests are shown in Table 5.3-6. The variations in the amounts of

materials retained in the piping system in these tests strongly suggest that very hygroscopic aerosol such as

NaOH particles behave quite differently than aerosol materials like Al(OH)

that are much less

hygroscopic. The pertinence of this observation to reactor accidents depends critically on the chemical

forms of materials released as aerosol during the accidents.

References

[1.]. F.J. Rahn Summary of the LWR Aerosol Containment Experiments (LACE) Program - Interim

Report LACE TR-012, Electric Power Research Institute, Palo Alto, CA, January 15, 1987

[2] R.K. Hilliard, L.D. Muhlestein, T.J. Albiol, Final Report of Experimental Results of LACE Test

LA2 - Failure to Isolate Containment LA-TR-007, Westinghouse Hanford Company, Richland, WA,

June 1987

[3] D.R. Dickinson, R.K. Hilliard, L.D. Muhlestein D.C. Mecham, G. Carraro Aerosol Behaviour in

LWR Containment Bypass Piping - Results of LACE Test LA3 LACE TR-011, July 1987

NEA/CSNI/R(2009)5

214

[4] J.D. McCormack, R.K. Hilliard, J.M. Slgado Final Report of Experimental Results of LATE Test

LA4 - Late Containment Failure with Overlapping Aerosol Injection Periods LACE-TR-025,

October 1987

[5] G.R. Bloom, R.K. Hilliard, J.D. McCormack, L.D. Muhlestein Aerosol Behaviour Under LWR

Containment Bypass Conditions - Results of Tests CB-1, CB-2, and CB-3 LACE TR-001,

Westinghouse Hanford Company, Richland, WA, November 1986

[6] T.A. Allbiol, J.D. McCormack, L.D. Muhlestein Instrumentation and Measurement Techniques

Used in the LACE Program LACE TR-006, Hanford Engineering Development Laboratory,

Richland, WA, June 1987

Table 5.3-4 Features of the LA tests

Test CsOH

input rate

(g/s)

MnO input

rate

(g/s)

input gas

velocity*

(m/s)

aerodynamic

mass

median

diameter

(µm)

geometric

standard

deviation

Auxiliary

building

temperature

(K)

LA-1

1.8 2.3 97 1.64 1.91 388

P = 1.1 bar

LA-2

0.63 0.68 1.7 1.7 P varies 1.1

to 2.2 bar

LA-3

-3A 0.11 0.51 75 1.4 2.0 not used

-3B 0.10 0.76 24 2.4 2.0 not used

-3C 0.34 0.56 23 1.9 2.1 not used

* velocity at the start of the piping system. Because of pressure drop through the piping system, velocity

increases along the piping system.

NEA/CSNI/R(2009)5

215

Table 5.3-5 Features of the CB tests

Test Aerosol

Composition

Aerosol

input

rate

(g/s)

Aerodynamic

Mass Median

Diameter

(µm)

Geometric

Standard

Deviation

Input

Gas

Velocity

(m/s)

Gas

Temperature

(K)

Auxiliary

Building

Temperature

(K)

CB-1 100 % NaOH 3.0 3.9 2.9 100 459 358

CB-2 67 % NaOH

33 % Al(OH)

0.9 3.1 2.6 91 384 354

CB-3 100 % Al(OH)

2.0 4.3 2.5 97 433 351

Table 5.3-6 Disposition of aerosol in the CB tests and the LA-1 test

Test Retained in Pipes*

(%)

Auxiliary Building

horizontal surfaces

(%)

walls

(%)

vented

(%)

CB-1 58 26.9 7.1 8.0

CB-2 48 33.8 9.9 8.3

CB-3 14 46.4 5.2 34.4

LA-1 98 <1 - <1

5.3.4 ACE-B: large scale combined effects experiment

One large scale experiment was conducted at the Containment Systems Test Facility (CSTF) located at the

Hanford site in the USA. The CSTF, shown in Fig. 5.3-7 was consisted of an aerosol generation facility, a

test vessel (volume 852 m

) and associated equipment. The internal surface of the test vessel was painted

with the same epoxy coating as used in the RTF or laboratory scale ACE experiments. The purpose of the

large scale experiment was to evaluate the effectiveness of volatile iodine sinks (i.e., bulk aerosols, painted

surfaces, steam condensation and water reservoirs). Fig. 5.3-8 shows various phases of the experiment.

CsOH and MnO aerosol particles were fed in the test vessel in the first two phases. In addition, HI, in the

first phase, and I

, in the second phase, were injected. Steam was continuously fed in order to keep the

vessel wall at 100 °C for duration about 4.5 days. The pH value of the sump was kept at 5.6 until the I

injection, during which it was simultaneously increased to 8.5. The thermal-hydraulic parameters,

temperatures, pressure, wall condensation rate, pool water amount, were measured and the deposition

coupon samples were collected.

NEA/CSNI/R(2009)5

216

Fig. 5.3-7 Diagram of the CSTF test arrangement

The major output of the experiment was that both forms of iodine, HI and I

, were removed from the

atmosphere at the same rate as the aerosol material. The aerosol material (CsOH) had experienced

significant particle growth due to its hygroscopic nature which then enhanced the removal rate by

sedimentation. Airborne organic iodine constituted only 0.2 % of the initially introduced iodine after the

major species had been removed from the atmosphere by settling or deposition. Fig. 5.3-9 shows the total

iodine and particulate iodine concentration measured in the test vessel atmosphere during the experiment.

Fig. 5.3-8 The CSTF iodine experiment test sequence

Legend:

M: Monday,

Tu: Tuesday,

S: Saturday

NEA/CSNI/R(2009)5

217

Fig. 5.3-9 Total iodine and particulate iodine concentration measured during the CSTF experiment

5.3.5 VANAM (including ISP-37)

The VANAM tests (1988 – 1993) provided relevant experimental data for the validation of coupled thermal

hydraulic and aerosol codes for accident analyses [1]. Five large-scale tests were carried out in the 626 m³

Model Containment of the Battelle Institut in Frankfurt/Main. The main features of the VANAM tests were:

−

Multi-compartment geometry with locally different thermal hydraulic conditions

−

Local volume condensation and its effect on aerosol depletion

−

Behaviour of insoluble and hygroscopic aerosol materials in saturated and super-saturated steam-air

atmospheres

−

Impact of a hydrogen deflagration on aerosol behaviour (dry resuspension)

Table 5.3-7 gives an overview of the five VANAM tests. In test M4 the dry resuspension of settled aerosol

by a hydrogen deflagration was measured. Although the deflagration was rather mild a significant

resuspension effect was observed.

Table 5.3-7 Overview of the VANAM tests

Test

Aerosol

Material

Description

M1 SnO

Dry aerosol depletion period followed by a wet depletion period with wea

volume condensation

M2 SnO

Dry aerosol depletion period followed by a wet depletion period wit

strong volume condensation

M2* SnO

Dry aerosol depletion period followed by a further aerosol injection and

wet depletion period

M3 NaOH

Dry aerosol depletion period followed by a further hygroscopic aerosol

injection and a wet depletion period; Subject of ISP-37

SnO

NaOH

Dry aerosol depletion period followed by a hydrogen-air-detonation wit

aerosol resuspension (Scoping Test)

NEA/CSNI/R(2009)5

218

The VANAM test M3 was subject of the International Standard Problem (ISP) No. 37 which dealt with

thermal hydraulics and aerosol behaviour in a LWR containment following a severe accident with core

melt down [2]. Only this test is decribed in more detail here.

The Battelle Model Containment (BMC) was made of reinforced concrete and consisted of nine individual

compartments interconnected by vent openings (Fig. 5.3-10). In the first test phase the BMC was heated up

by steam injected into compartment R5, representing the blowdown of the primary circuit over the

pressuriser quench tank. Then in test M3 a hygroscopic NaOH aerosol generated by three plasma torches

was injected into the steam-air-atmosphere during two separate periods. During and after the first injection

the non-condensing containment atmosphere was stratified causing an inhomogeneous NaOH aerosol

distribution. In the dome and in the upper compartments the NaOH concentration was significantly higher

than in the lower compartments. The depletion of the hygroscopic NaOH aerosol was somewhat faster than

the depletion of the insoluble SnO

in test M2* under similar conditions (r.h. approx. 90 %) because of the

hygroscopic water take up.

Fig. 5.3-10

Battelle Model Containment in the configuration for VANAM M3

After the second NaOH aerosol injection representing the aerosol release from the core concrete interaction

steam was injected into the central bottom compartment. This steam production was corresponding to a

contact of core melt to sump water and induced a natural convection which mixed the atmosphere in large

parts of the containment. In most compartments the rising steam content caused volume condensation

which increased the aerosol depletion significantly. Only in the annular part of R9 no volume condensation

occurred and the measured NaOH concentration there was for many hours up to two orders of magnitude

higher than in the other compartments (Fig. 5.3-11).

NEA/CSNI/R(2009)5

219

Fig. 5.3-11 Measured NaOH aerosol concentration in test VANAM M3

The ISP-37 consisted of a thermal hydraulic and an aerosol part. Main objective of the aerosol part of the

open exercise (the participants knew all relevant test results) was to calculate the inhomogeneous

distribution and the depletion of the NaOH aerosol. Representatives of 22 organisations from 9 OECD

member countries, two non-OECD countries and the CEC submitted a total of 23 thermal-hydraulic and

22 aerosol calculations. They used the codes CONTAIN, ECART, FIPLOC (a precursor of COCOSYS),

and MELCOR for coupled thermal-hydraulic and aerosol calculations, and MACRES, MOSAIC and

REMOVAL for stand alone aerosol calculations.

Fig. 5.3-12 to Fig. 5.3-14 show the results of the predominantly used codes. In the MELCOR standard

version the solubility effect was not modelled. The user’s influences on the CONTAIN and FIPLOC

results are high.

NEA/CSNI/R(2009)5

220

Fig. 5.3-12 ISP-37; NaOH aerosol concentration in the dome (CONTAIN)

Fig. 5.3-13

ISP-37; NaOH aerosol concentration in the dome (MELCOR, part 1)

NEA/CSNI/R(2009)5

221

Fig. 5.3-14 ISP-37; NaOH aerosol concentration in the dome (FIPLOC)

With most codes the thermal-hydraulic behaviour (pressure, temperatures, etc.) was calculated with

sufficient accuracy. But with respect to the needs of aerosol behaviour analysis the accuracies, both

analytical and experimental, of specific thermal-hydraulic variables like humidity, wall condensation and

heat transfer should be improved.

All codes with a tight coupling of thermal hydraulics and aerosol physics, i.e. CONTAIN, FIPLOC, and

MELCOR, predicted the tendency of the aerosol concentration during the course of the experiment well.

But the actually calculated NaOH concentrations showed a spread of approximately two orders of

magnitude at the end of the “dry” depletion phase and about five orders of magnitude at the end of the

“wet” depletion phase. Mostly the NaOH aerosol concentrations were overpredicted due to an

underestimation of the droplet growth by steam condensation, e.g. because of a missing solubility model.

With respect to reliable source term analyses some discrepancies are clearly too large.

In [1] the ISP-37 outcomes are summarised as follows:

Although much progress has been made in the simulation of aerosol behaviour in multi-compartment

geometries the calculated local concentrations scatter widely. However, the aerosol source term to the

environment is overestimated in general. The largest uncertainty concerning the aerosol results is caused

by a limited number of thermal hydraulic variables like relative humidity, volume condensation rate and

atmospheric flow rate. In some codes a solubility model is also missing.

By choosing a physically correct nodalization

the lumped-parameter concept was found to be basically

suited to predict the inhomogeneous distributions of gases and aerosols in the containment. But the user

has great influence on the calculated results, both the thermal-hydraulic and the aerosol ones if specific

nodalization features and different code options are selected unfavourably .

physically correct nodalisation means:

The nodalisation allows the simulation of all relevant phenomen, probably occurring in the case investigated

NEA/CSNI/R(2009)5

222

References

[1] M. Firnhaber, T.F. Kanzleiter, S. Schwarz, G. Weber International Standard Problem ISP-37

VANAM M3 - A Multi Compartment Aerosol Depletion Test with Hygroscopic Aerosol Material

NEA/CSNI/ R(96)26 (Dec. 1996)

[2] T. Kanzleiter Versuche zum Verhalten von Kernschmelzunfall-Aerosolen in einer

Mehrraum-Containment-Geometrie (“VANAM-Versuche”) Battelle-Institut e.V., Frankfurt am

Main Abschlussbericht BIeV-R67.098-01 (1995)

5.3.6 KAEVER (including ISP-44)

In the KAEVER (Kernschmelz-Aerosol-Versuche) project carried out from 1993 to 1997 at the Battelle Institut

in Frankfurt/Main the depletion behaviour of aerosols typical for LWR core melt accidents under various

thermal hydraulic conditions was measured [1]. The KAEVER tests are complementary to the large scale tests

of e.g. the VANAM and the LACE series in which the number of tests was limited due to high the costs.

As test aerosols the insoluble materials Ag and SnO

as well as the hygroscopic materials CsI and CsOH

were used. Single-component aerosols as well as mixed-component aerosols were investigated. The

horizontal cylindrical test vessel made of steel had a volume of 10.6 m³. Large parts of the walls were as

well thermally insulated as equipped with a wall heating (Fig. 5.3-15).

Some tests were repeated with the same aerosol and the same thermal hydraulic conditions in order to

check the reproducibility of the results (e. g. tests K100/K102 and K125/K148). All these tests could be

reproduced well. The differences in the results were insignificantly.

Origin

Z-Axis

X-Axis

2090

2500

475

Door

Fig. 5.3-15 Top view of the KAEVER test vessel (numbers in mm)

The thermal hydraulic conditions for the tests have been varied between rather dry conditions (r.h. ≈ 0 %) and

saturated conditions with strong volume condensation, The conditions "slightly supersaturated" and "strongly

supersaturated" were achieved by two different test procedures. E. g. in the strongly supersaturated tests the

aerosol was injected into the cold test vessel fellowed by steam. The estimated condensation rate for strongly

supersaturated conditions (strong colume condensation rate) is 10

-4

kg/ (s⋅m

). This for slightly supersaturated

conditions (weak volume condensation rate) is 10

-5

kg/(s⋅m

). A total of 190 tests was performed. 33 of these

were evaluated in detail (Table 5.3-8).

NEA/CSNI/R(2009)5

223

Table 5.3-8 Evaluated KAEVER tests

Test Aerosol Thermalhydraulic conditions

(rh = relative humidity)

Remarks

K100 CsI rh = 0 %

K102 CsI rh = 0 % repetition of K100

K106 CsI rh = 85 - 90 %

K108 CsI rh = 95, goal 100 %

K110 CsI rh = 95, goal 100 % repetition of K108

K123 CsI rh = 100 %, weak volume condensation ISP-44

K125 Ag rh = 100 %, weak volume condensation

K128 Ag + CsI rh = 0 %

K130 Ag + CsI rh = 100 %, weak volume condensation

K131 Ag rh =85 - 90 %

K132 Ag + CsI rh =85 - 90 %

K134 Ag + CsI rh =85 - 90 % repetition of K132

K138 SnO

rh =85 - 90 %

K143 SnO

rh = 100 %, weak volume condensation

K148 Ag rh = 100 %, weak volume condensation repetition of K125, ISP-

K149 Ag rh =85 - 90 % repetition of K131

K150 no aerosol rh = 100 %, weak volume condensation fog formation test

K151 Ag strong volume condensation, steam released

into cold vessel

K156 no aerosol rh = 100 %, weak volume condensation fog formation test

K159 CsI strong volume condensation, steam released

into cold vessel

K167 CsOH rh = 0 %

K168 CsOH rh = 95 %

K171 Ag + CsOH rh =85 - 90 %

K174 CsOH strong volume condensation, steam released

into cold vessel

K175 Ag + CsI strong volume condensation, steam released

into cold vessel

K178 Ag + CsOH strong volume condensation, steam released

into cold vessel

K179 CsI + CsOH rh = 100 %, weak volume condensation

K181 CsOH rh =85 - 90 %

K183 CsI + CsOH rh = 85 - 90 %

K186 Ag + CsOH rh = 100 %, weak volume condensation ISP-44

K187 Ag + CsI +

CsOH

rh = 100 %, weak volume condensation ISP-44

K188 CsOH rh = 100 %, weak volume condensation ISP-44

NEA/CSNI/R(2009)5

224

Fig. 5.3-16 gives an overview of the experimental KAEVER results. It shows the normalised

aerosol

depletion rates measured for different aerosol components and different thermal hydraulic conditions. The

depletion rates in general increase with rising saturation ratio respectively rising volume condensation rate

because of the particle growth by condensation. The influence of the hygroscopic effect can be seen

clearly. CsI is weakly hygroscopic (Van't Hoff factor 1.7) and CsOH strongly (2.0). Under supersaturation

conditions, with weak as well as with strong volume condensation, the depletion rate of CsOH is always

the largest one.

Fig. 5.3-16 Comparison of depletion rates measured in the KAEVER tests

Five KAEVER tests were selected for the International Standard Problem ISP-44 which dealt, like the

earlier ISP-37, with thermal hydraulics and aerosol behaviour in a LWR containment following a severe

accident with core melt down [2]. In this exercise the main emphasis was put on the depletion behaviour of

different aerosol materials under thermal hydraulic conditions expected in a LWR severe accident scenario

over longer periods. The KAEVER tests selected were in particular:

−

K 148, test with insoluble Ag-aerosol

−

K 123, test with hygroscopic CsI-aerosol

−

K 188, test with strongly hygroscopic CsOH-aerosol

−

K 186, test with mixed aerosol of Ag and CsOH

−

K 187, test with a Ag-CsI-CsOH mixture (blind calculation)

The vessel geometry, the tests procedure and the thermal hydraulic conditions were the same in all tests.

First the vessel was preconditioned by release of steam and heat until quasi-stationary conditions were

obtained. In all selected tests the atmosphere was slightly supersaturated causing some condensation on the

insoluble Ag-aerosol and an increasing condensation on the soluble aerosols respectively. Then the

thermally generated aerosols were injected and the depletion of the aerosols in the vessel was measured

without any change of the boundary conditions.

in each case related to the maximum value for aerosol concentration measured in each test

strongly

supersat

urated

slightly

su p e r sa t

urated

slightly

superhe

ated

strongly

superhe

ated

Sn-Oxid

CsI/Ag

CsI

CsOH

CsOH/Ag

CsOH/CsI

CsOH/CsI/Ag

187

179

183

178

186

171

174

188

181

167

159

123

106

100

175

130

132

128

151

148

149

143

138

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

normalized depletion rate

thermal hydraulic conditions

aerosol component

Comparison of KAEVER depletion rates

NEA/CSNI/R(2009)5

225

Representatives of 10 organisations from 9 countries participated in the ISP-44. The codes used were

ASTEC-V0.3 (GRS-IPSN), COCOSYS V1.2 (GRS, RUB), CONTAIN 2.0 (IJS), MELCOR 1.8.3

(VEIKI), 1.8.4 (Studsvik, JRC, KAERI), 1.8.5 (NRI) and ART Mod2/REMOVAL (JAERI). Open

calculations were performed on all tests except K 187, which was offered as a blind exercise.

Several discrepancies occurred concerning the test with insoluble and slightly soluble materials. Fig.

5.3-17 and Fig. 5.3- 18 show the results of test K148 with a Ag-aerosol. It has to be said that several

calculations showing a good agreement with the measured aerosol concentration histories show other

results which are inconsistent with the measurement. E.g. four calculations show no airborne water and in

one calculation the water mass is far too high. Therefore only two out of nine calculations are in a good

agreement with the experimental results.

In MELCOR calculations on the three tests with the strongly hygroscopic CsOH the droplet growth was

either significantly overestimated or the solubility effect is simply ignored. The blind test K187 with a

Ag/CsI/CsOH-aerosol was well calculated by four out of ten participants. The other calculations showed

larger deviations from the measurement, e.g. no airborne water was calculated. Fig. 5.3-19 and Fig. 5.3-20

depict the results for test K187.

K148 Total Dry Concentration

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1.0E+00

1.0E+01

0123456

Time [h]

Concentration [g/m³]

Ex p

KAER

JRC

STUD

VEIK

Fig. 5.3-17 KAEVER test K148, Ag-aerosol concentration (MELCOR)

NEA/CSNI/R(2009)5

226

K148 Total Dry Concentration

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1.0E+00

1.0E+01

0123456

Time [h]

Concentration [g/m³]

Ex p

GRS

IJS

GRS- IPSN

RUB

JAER

Fig. 5.3-18 KAEVER test K148, Ag-aerosol concentration (miscellaneous codes)

K187 Total Dry Concentration

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1.0E+00

1.0E+01

6 7 8 9 10 11

Time [h]

Concentration [g/m³]

Ex p

KAER

JRC

STUD

NRI

VEIK

Fig. 5.3-19 KAEVER test K187, Ag/CsI/CsOH-aerosol concentration (MELCOR)

NEA/CSNI/R(2009)5

227

K187 Total Dry Concentration

1.0E-04

1.0E-03

1.0E-02

1.0E-01

1.0E+00

1.0E+01

67891011

Time [h]

Concentration [g/m³]

Ex p

GRS

IJS

GRS-IPSN

RUB

JAER

Fig. 5.3-20 KAEVER test K187, Ag/CsI/CsOH-aerosol concentration (miscellaneous codes)

In ISP-44 the agreement of the results with the measurements of these fairly simple tests was not as good

as expected. In particular the aerosol condensation modeling and the accuracy of related thermal hydraulic

parameters were insufficient with some codes.

−

The solubility effect increases the growth of hygroscopic particles by condensation and enhances the

aerosol depletion. An appropriate modeling is absolutely necessary. The solubility model in

MELCOR needs some corrections.

−

The Kelvin-effect may slow down the depletion of insoluble and slightly soluble aerosols in a

condensing atmosphere. There are indications that the Kelvin-effect is not treated sufficiently by any

codes except ASTEC and COCOSYS. The modeling of the Kelvin-effect should be checked. In

some codes the numerical stability should be improved (e.g. COCOSYS).

In calculations with a correct aerosol condensation treatment the largest discrepancies in the aerosol results

were caused by the uncertain thermal-hydraulic variables “relative humidity” and “volume condensation

rate”.

The multi-component behaviour in the KAEVER tests was fairly easy to calculate, because the

components were well mixed among all particles. Experiments with incompletely mixed aerosol

components are needed to validate the multi-component capability of codes, as in an incompletely mixed

aerosol the composition of components is not the same for all particles.

ISP-44 showed a large user influence on the thermal hydraulic and aerosol results. Possible measures to

reduce this influence are the improvement of code input assistance, user training, further benchmarks, etc.

The KAEVER experiments calculated for ISP-44 had some measurement uncertainties (relative humidity,

wet droplet size, aerosol injection rate, etc). In future tests the measurement accuracy of important

parameters should be improved.

In the ISP-37 on test VANAM M3 all codes were already used except the new ones ASTEC and

COCOSYS. The earlier exercise comprised the depletion of a strongly hygroscopic aerosol in a

NEA/CSNI/R(2009)5

228

multi-compartment geometry and was more complex than ISP-44. Among the most uncertain variables and

models analysed were already “volume condensation rate”, “relative humidity”, and the solubility model.

This demonstrates the necessity to perform several exercises on an important topic within a certain time.

References

[1] G. Poss, D. Weber Versuche zum Verhalten von Kernschmelzaerosolen im

LWR-Containment-KAEVER Fachbericht BF-R-67863, Battelle Ingenieurtechnik GmbH, Mai 1997

[2] M. Firnhaber, K. Fischer, S. Schwarz, G. Weber ISP 44 – KAEVER Experiments on the Behaviour

of Core-melt Aerosols in a LWR Containment NEA/CSNI/R (2003)5 (Aug. 2002)

5.3.7 AHMED

Introduction

The AHMED facility was constructed by VTT (Technical Research Centre of Finland) to study the effect

of T-H on hygroscopic and inert aerosol behaviour in nuclear power plant containment during a

hypothetical severe reactor accident. The objective of this work was to understand the reason for the

observed large discrepancies between the results from model calculations and measurements carried out in

large scale experiments. The discrepancies may be caused by inaccuracies in aerosol modelling,

uncertainties in T-H conditions or measurement errors.

Facility description and instrumentation

Hygroscopic NaOH, CsOH and CsI and inert Ag aerosol behaviour at different temperatures and relative

humidities (RH) was studied in a well instrumented and controlled AHMED vessel of 1.81 m

total free

volume. The relevant parameters of the vessel are presented in Table 5.3-9.

Special emphasis was placed on temperature measurements, because of homogeneous temperature field

inside the vessel was necessary for the experiments. The vessel gas temperature (13 locations), inside and

outside surface temperatures (14 locations) and surround temperatures as well as pressurised air and steam

temperatures were measured using Pt-100 resistance type detectors (RTD).The RH was measured using

three Vaisala Humicap detectors.

Table 5.3-9 System parameters

Vessel volume 1.81 m

Vessel radius 63.5 cm

Vessel height 142.5 cm

Sedimentation area 1.27E4 cm

Pressure 1 atm

Sampling rate 2.6 l/min (206 % of volume/24h)

The vessel and input line pressures and steam and air flows were also monitored continuously. Input air

flow was filtered and dried. The vessel surface temperature was controlled using computer controlled

heating cables. The input gas temperature was regulated using a heat exchanger.

NEA/CSNI/R(2009)5

229

For aerosol characterisation state-of-the-art aerosol measurement systems were used. Aerosol number and

mass concentrations were measured continuously using a condensation nuclei counter (CNC) and a tapered

element oscillating microbalance (TEOM) mass monitor. Before the measurement the aerosol was dried in

an ejector-type diluter and in diffusion drier. Aerosol sampling into the diluter could be done from three

heights through heated sampling lines.

The dry size of the particles was measured using Berner low pressure impactors (BLPI). The results were

compared to measurements done with a TSI aerodynamic particle sizer (APS). A differential electrical

mobility analyser (DMA) was used to check that the amount of small particles was negligible.

The particle size distribution and chemical composition in the test conditions (wet particle size) were

analysed by Berner low pressure impactors (BLPI). Some size distribution measurements were also made

with a PALAS PCS-2000 Optical Particle Counter.

Spherical aerosol particles were generated from water solutions of aerosol material using ultrasonic aerosol

generator. Spherical Ag particles were made from AgNO

+ H

O droplets utilising a Lindberg tube furnace

at 900 °C. In order to ensure that no AgNO

particles were produced, the solubility of Ag particles

collected on Nuclepore polycarbonate impactor foils to water and nitric acid was measured with ICP-MS.

The Aerosol was always generated in the same way and it was also checked that the size distribution was

same during different generation times.

Experiments

At first the behaviour of aerosol of different chemical species was studied separately in AHMED facility.

Finally both Ag and CsOH aerosol particles were generated and injected into the AHMED vessel

simultaneously to study the behaviour of multicomponent aerosol. The temperature in the experiments was

varied between 17 °C and 51 °C and the RH between 7.3 % and 97 %. Approximate temperatures and RHs

during the different experiments, mass and number concentrations as well as aerodynamic mass median

diameters (AMMD) and geometric standard deviations (GSD) of the dry particle size distributions at the

beginning of the experiments are presented in Table 5.3-10

Before the experiments the vessel was purged with filtered air. It was also checked that the aerosol number

concentration measured with CNC was practically zero. After the aerosol generation was stopped the

aerosol in the vessel was mixed with a propeller having a very small sedimentation area to ensure constant

homogeneous concentration in the AHMED vessel. After the mixing the aerosol particles were settled in

the calm air and the aerosol mass and number concentrations were monitored continuously over 300

minutes. From this data the decay rate of different aerosols used in the AHMED experiments can be

characterised by a decay constant in a logarithmic scale.

Table 5.3-10 Experimental parameters

Material RH [%]

T [°C]

AMMD

[µm]

GSD Mass conc.

[mg/m

]

Number conc.

[1/cm

]

NaOH 22 50 2.4 1.6 112 64 000

NaOH 34 20 2.4 1.6 200 65 000

NaOH 82 27 2.4 1.6 208 137 000

NaOH 91 17 2.4 1.6 638 400 000

NaOH 96 23 2.4 1.6 218 165 000

NEA/CSNI/R(2009)5

230

Material RH [%]

T [°C]

AMMD

[µm]

GSD Mass conc.

[mg/m

]

Number conc.

[1/cm

]

CsI 27 23 2.3 1.7 102 71 820

CsI 82 24 2.3 1.7 98 94 500

CsI 96.5 27 2.3 1.7 94 92 700

CsOH 7.3 51 2.1 1.7 89 91 000

CsOH 33 19 2.1 1.7 86 48 000

CsOH 39 19 2.1 1.7 73 80 000

CsOH 96 23 2.1 1.7 64 77 000

CsOH 97 28 2.1 1.7 94 144 000

Ag 54 22 2.7 1.7 79 120 000

Ag+CsOH 91.5 24 2.1 (CsOH) 1.7 82 103 500

Ag+CsOH 91 25 2.1 (CsOH) 1.7 112 135 000

Ag+CsOH 97 24 2.1 (CsOH) 1.7 60 79 200

Results

Homogeneous temperature and RH conditions for aerosol measurements in the AHMED facility were

achieved. The maximum gas temperature differences were measured to be smaller than 0.4 °C. Apart from

the three sampling line connections the inside surface temperatures were 0.1 - 0.5 °C lower than the gas

temperature. The slightly lower temperature is enough to cause the necessary natural convection for mixing

and maintaining homogeneous aerosol concentration during the experiments [1]. Because the sampling

lines were heated, the inside surface temperature at these locations was 0.5 °C higher than the gas

temperature. The RHs were near each other at different locations inside the vessel and maintained nearly

constant levels during experiments. During some experiments the RH changed slightly, which could also

be seen on the measured wet particle size distribution.

Mass and number concentrations measured from three different heights were the same within the limits of

the measurement accuracy. The total mass and number concentrations followed each others fairly well at

the beginning of the experiments. Later the number concentration decayed slower than the mass

concentration. This was probably due to the small particles, which settled slowly but could not be seen in

the evolution of mass concentration. Mass concentration behaviour measured with TEOM and impactors

also agreed fairly well. The calculated decay constants decreased as a function of time because large

particles had higher sedimentation velocities.

For NaOH the ratio of aerosol mass concentration half lives at low to high RH experiments was about 4.

For CsI and CsOH this ratio was about 2. This difference was due to the density effect. CsOH and CsI have

higher densities than NaOH and thus during condensation their aerodynamic sizes do not increase as much

as the AMMD of NaOH aerosol. The mass concentration half lives for CsOH and NaOH decreased

smoothly as the RH was increased. For CsI the mass concentration behaved similarly as a function of time

at RHs 27 % and 82 %. CsOH and NaOH absorb water at all RHs, but CsI starts to absorb water as the RH

exceeds 91 %. The half life of silver aerosol was less than that of the other species. This was due to the

NEA/CSNI/R(2009)5

231

larger AMMD of the high density silver particles. However, silver behaved as a dry aerosol at all

humidities. The AMMD of silver particles at high humidity was the same as in the dry conditions.

Dry size distributions measured with BLPI and APS agreed within the measurement accuracy. The wet

mass size distributions measured only with BLPI showed that the AMMD of particles decreased as a

function of time due to settling of larger particles. It could be seen that the hygroscopic particles reached

their equilibrium size immediately after being exposed to high humidity. When there was a slight decrease

in RH the AMMD of CsOH particles was also observed to decrease. The measured AMMD of CsOH

particles was compared to the corresponding equilibrium values calculated from the literature data reported

for the van´t Hoff factor i. The equilibrium AMMD was calculated from the following equation [2]:

3/1

2/1

































−













AMMDAMMD

with

AMMD equilibrium diameter of the particle

AMMD

initial diameter of the dry particle

density of the measured (wet) particle

P,O

density of the initial (dry) particle

i Van't Hofffactor '

molecular weight of water

molecular weight of the soluble aerosol material

rh relative humidity

The Van’t Hoffs factor for CsOH is reported only down to 95.61 % RH [3]. Using this data, reliable values

can be obtained by semi-empirical method down to 90 % [4, 5]. In Table 5.3-11 the measured and

calculated AMMD values are presented. It can be seen that measured and calculated values agree within

the measurement accuracy. This comparison supports both the AHMED measurements and the method to

calculate equilibrium particle size.

Table 5.3-11 Measured and calculated AMMD of CsOH aerosol at different RH

AHMED Calculated

RH [%]

AMMD [µm] AMMD [µm]

Dry 2.1 2.1

97.0 3.4 3.55

91.5 2.6 2.77

91.0 2.7 2.74

Conclusions

The AHMED experiments filled the gap on our knowledge of hygroscopic aerosol behaviour and the

results can be used for aerosol model validation in containment codes. AHMED was the first facility where

NEA/CSNI/R(2009)5

232

the aerosol mass concentration could be monitored on-line and wet aerosol size distribution was measured

directly in containment conditions. The behaviour of hygroscopic aerosol was observed to be very sensitive

to RH at high humidity conditions. At the same RH the aerosol behaviour was not affected by changing the

absolute temperature as can be expected from the available water activity data. As the steam condensation

at supersaturated conditions is also well known [6], we may conclude that the data base for steam

condensation on aerosol particles in containment conditions is sufficient for code validation.

References

[1] N. Fuchs The mechanics of Aerosols. Pergamon Press, Oxford, pp. 250-257, 1964

[2] J. Jokiniemi The effect of airborne hygroscopic matter on aerosol behaviour in severe nuclear power

plant accidents Technical Research Centre of Finland, Publications 59 (Dissertation), 1990

[3] R.A. Robinson, R.H. Stokes Electrolyte Solutions, 2

rev. ed. Butterworth, London, 1970

[4] H. Meissner Prediction of activity coefficients of strong electrolytes in aqueous systems ACS Series

133, pp. 495-511, 1980

[5] J. Jokiniemi The effect of selected binary and mixed solutions on steam condensation and aerosol

behaviour in containment Aerosol Sci. Technol. 13 (2), 1990

[6] P. Wagner Condensation processes in aerosols J. Aerosol Sci. 26 S1, pp. 203-204, 1995

5.3.8 VICTORIA

5.3.8.1 Introduction

The VICTORIA facility [1] is a scale model of the ice condenser containment of Loviisa NPP with linear

scale of 1:15 and volume scale of 1:3375. The height of the pressure vessel is 4.6 m and the diameter is

3.14 m.

The aim of aerosol experiments on the VICTORIA facility was to validate the containment aerosol models

used in the nuclear reactor accident codes. Especially important was to ensure that containment aerosol

codes were able to calculate correctly the radioactive hygroscopic and non-hygroscopic aerosol behaviour

in non-homogeneous multicompartment containments.

Two research programmes has been carried out on the VICTORIA facility. In the earlier research

programme (1990-1995) the thermal-hydraulic behaviour and hydrogen distribution were studied in severe

accident conditions [2, 3]. During 1996-1997 VICTORIA was used in a modified geometry for aerosol

experiments in a CEC Fourth Framework Programme project [4].

5.3.8.2 Facility description

Since experiments with VICTORIA facility were a part of the CEC Fourth Framework Programme project

Aerosol Physics in the Containment, details of the facility and its instrumentation is included in the

Asterism database.

VICTORIA containment

The dimensions of the pressure vessel were determined according to the chosen scale of the facility (1:15).

The cylindrical part of the liner was built on site out of galvanised steel plate (thickness = 1 mm). A dome

NEA/CSNI/R(2009)5

233

was attached to the upper part of the pressure vessel. The inner diameter of the dome of the steel liner

(thickness = 1.5 mm) was 2950 mm. Both the upper and inner surfaces were painted with alkyd paint. The

design pressure of the vessel was 3 bar, but the pressure relief valve was set to 1 bar [5].

All the structures that are of concrete in the containment of the Loviisa NPP are made out of concrete also

in the facility. All the concrete surfaces were painted with two component epox-paint, Reapox-GPL-VS.

The concrete level, at + 25.4-elevation, was replaced with plywood plate, coated on both sides with

phenolic resin. The upper level of this plate was at +25.8-elevation. There was two round holes

(∅ = 125 mm) at the edges of the plate (90° and 270°, r= 13.80 m / 0.92 m), through which the aerosols

and steam, injected into lower compartment, could flow into upper compartment of the containment. The

upper compartment free volume was 17 m

and the total free volume was 22 m

The ice condensers used in the previous VICTORIA experiments were removed before the aerosol

experiments. The ice condenser concrete structures, lower inlet doors and both the intermediate and top

deck doors were still part of the containment. A strong convective loop between the upper and lower

compartments, through the ICs, could be induced by opening the lower inlet doors of the two IC sections

non-simultaneously.

The lower compartment of the VICTORIA containment could be pre-heated. The pre-heating was done by

circulating hot water from central heating boiler in the piping below the lower compartment floor.

TH-measurements and control

The steam used in the experiments was generated by a small steam generator. The maximum capacity of

the steam generator was 25 g/s. The injection of steam flow into the containment was adjusted by a control

loop, which consists of regulator, control valve and flow meter. The injection run into the containment and

the maintenance of set values were controlled by a computer.

In the experiments the piping of both the containment outer spray and the inner spray is used as a exhaust

line of gases. The nozzles of the spray have been removed. The outer diameters of the pipes were 12 mm

and the wall thickness 2 mm. The exhaust line was needed to equalise the overpressure caused by aerosol

generation. The exhaust gases flowed through a heat exchanger and a filter in order to separate water and

particles.

The outlets of the drainage piping of the sumps installed in the containment were at the bottom of the

pressure vessel. The piping was made symmetrical from both of the sumps, in order to maintain uniform

flow out of the lower compartment.

Gas temperature was measured at 99 locations and wall temperatures at 65 locations with T-type

Thermocouples. Relative humidity (RH) was measured at 10 locations. In 8 locations the humidity was

measured with a VAISALA probe HMP 135Y, that measures relative humidity in range of 0...100 % and

temperature in range of 0 ... 160 °C. One probe in upper and one in lower compartment (VAISALA

HMP-243) could measure relative humidities in condensing condition. The temperature range of these

probes was -40 ... +180 °C. Pressure difference and water levels were measured in seven locations. The

total pressure and flow rates of steam and air into the vessel were measured as well.

5.3.8.3 Aerosol generation and instrumentation

The aerosols in the experiments were generated by two aerosol generators Water-soluble aerosol, either

NaOH or CsOH, was generated with two opposite jet atomisers. The dry particle AMMD was ~2.3 µm,

NEA/CSNI/R(2009)5

234

GSD ~1.9 and the feed rate was close to 3 mg/s. High temperature Laminar Entrainment Flow Reactor

(LEFR) was used in the generation of silver particles.

The piping from the aerosol generators (outer diameter = 23 mm, wall thickness 1.5 mm) was connected

into one injection line, which was directed into the upper or lower compartment of the containment. The

injection pipe was made out of acid-resistant stainless steel. The temperature of the injection line was kept

at slightly over 100 °C and it was adjusted by a separate controller.

Sampling lines from the upper and lower compartments of the containment, were also made out of

acid-resistant stainless steel (outer diameter 12.7 mm, wall thickness 1.65 mm). Heating cables, similar to

those in the injection line, were attached onto the surfaces of the collecting lines and the temperatures were

adjusted by controller of their own. The six sampling lines started from various locations inside the

containment. At the inlet of each line there was a conical cover installed, in order to prevent droplets and

impurities to enter the line. The samples passed through an ejector type diluter (PALAS), in which they

were conditioned for the measurement devices using dried air.

The aerosol number and mass concentration was measured continuously using a Condensation Nucleus

Counter (CNC) and the Tapered Element Oscillating Microbalance (TEOM) mass monitor. Particle mass

and chemical composition size distributions were determined with 11-stage multijet Berner type low

pressure impactors (BLPI). In order to measure the wet size distributions the impactors were placed inside

the vessel. The measurement size range of impactors was 0.03 - 15 µm [6]. The elemental concentration

collected on nuclepore filter substrates of each BLPI stage were analysed with Inductively Coupled

Plasma-Mass Spectroscopy (ICP-MS). The dry number size distribution was measured with an Electrical

Low Pressure Impactor (ELPI) and with a Differential Mobility Analyser (DMA). The size ranges of these

instruments were 0.01 - 5 µm for ELPI [7] and 0.02 - 0.8 µm for DMA. Settling of particles was studied

using deposition trays and deposition coupons. Coupons were acid resistant stainless steel and their size

was 46 x 70 mm. Samples were analysed using ICP-MS. Also on-line Optical Particle Sizer (OPS),

Aerodynamic Particle Sizer (APS) and filter samples were applied in the aerosol measurements.

5.3.8.4 VICTORIA aerosol experiments and analysis

VICTORIA aerosol experiments consisted of eight TH-experiments and aerosol generator tests as well as

four actual aerosol behaviour experiments. Detailed information on the seven preliminary tests, numbered

as 51…57, is presented in a data report [8]. Aerosol behaviour experiment number 58 and 59 belonged to

CEC Fourth Framework programme project Aerosol Physics in the Containment. In both experiments

aerosol was injected into the lower compartment of VICTORIA facility. The main difference between the

experiments was that in experiment 58 the aerosol material was CsOH, where as in exeriment 59 CsOH

aerosol was mixed with silver. Information on these experiments can be found in the final report of the

project [9]. Experiment number 60 was again a TH-test, in which steam and gas injection to upper

compartment was tested. In aerosol behaviour experiment number 61 and 62 the particles were also

injected to upper compartment of the facilty. As before the main difference between the experiments was

that in experiment 61 CsOH was applied as aerosol material, where as in experiment 62 the material was

silver. Results from experiments 60 and 61 have been published [10]. Details of experiment 62 can be

found in a data report [11].

When aerosol was fed to lower compartment particle mass and number concentrations in the upper

compartment were much smaller than in the lower compartment. Aerosol mass and number concentrations

in the lower compartment decreased fast due to high RH and sedimentation during the steam feed into the

lower compartment. Eventually they reached values that were smaller than in the upper compartment. With

hygroscopic material the average particle diameter increased during the steam feed and after that the

NEA/CSNI/R(2009)5

235

diameter decreased as a function of time because large particles were settling faster. The measured wet size

of CsOH aerosol responded to the increasing RH as was expected.

When the aerosol was fed to the dome part of the upper compartment, the dome was in a saturated

condition at a temperature of about 60 °C. A part of the aerosol particles was carried into the ice

condensers and the lower compartment by a convective flow, which was driven by asymmetrical

TH conditions between the two ICs. The lower part of the upper compartment, that is the volume between

the ICs, remained superheated due to heat transfer through the surrounding structures, and no mixing with

the saturated dome part took place. Thus, a clear difference (factor >3) could be found between the aerosol

concentration decay times in these two parts of the upper compartment.

The results from experiment number 61 were compared with calculated results using FIPLOC and

CONTAIN codes [12, 13]. Both the CONTAIN [14] and FIPLOC [15] calculations of the VICTORIA

experiment gave fairly good overall results on thermal hydraulics. Containment pressurisation as well as

temperatures and humidities in the ICs, the lower compartment and the dome part of the upper

compartment were modelled accurately. The largest deviations from the measurements were found in the

temperature and RH of the lower part of the upper compartment. It was evident that of a strong temperature

and humidity gradient between two inter-connected volumes could not be modelled with

lumped-parameter codes. Inaccuracy in modelling the convective flow pattern in the containment was

another significant source of deviation in aerosol transport and deposition. The study indicated that most of

the straits in containment aerosol modelling can be traced not to the aerosol model package itself, but to its

physical and numerical coupling to the thermal hydraulic calculation, and parameter choices [10].

References

[1] O. Hongisto, K. Lammila, H. Tuomisto Experiments for Hydrogen Distribution in an Ice Condenser

Containment USNRC 20th Water Reactor Safety Information Meeting, October 21-23, 1992,

Bethesda, Maryland, USA

[2] O. Hongisto, H. Tuomisto Experimental Verification of the Loviisa Ice Condenser Containment

Transient Operation in Reactor Accidents American Nuclear Society, Nuclear Reactor Safety

Division’s ”Safety of Thermal Reactors”, International Topical Meeting, July 21-25, 1991, Portland,

Oregon, USA, 1991

[3] P. Lundström, H. Tuomisto T. Lamberg, O. Hongisto Experimental studies of Hydrogen Behaviour

in Ice Condenser Containments In: Proceedings of the OECD/NEA/CSNI Workshop on the

Implementation of Hydrogen Mitigation Techniques. Winnipeg, Manitoba, 1996 May 13-15, pp.

159-179 AECL-11762, NEA/CSNI/R(96)8

[4] V. Saldo, E. Verloo, T. Montanelli, J. Mäkynen, J. Jokiniemi, E.I. Kauppinen, H. Tuomisto,

T. Routamo Aerosol physics in containment In: FISA-97 – EU research on severe accidents,

Mid-term review symposium on shared-cost and concerted actions in reactor safety. Ed: Van

Goethem, G., Keinhorst, G., Martin Bermejo, J. and Zurita, A. EC, Luxembourg, 17 to 19 November

1997, pp. 270-278. EUR 18258 EN

[5] E. Heikkilä Technical description of VICTORIA Facility E.C. Report ST-APC(96)-P07 (1996)

[6] R.E. Hillamo, E.I. Kauppinen On the performance of the Berner low pressure impactor Aerosol Sci.

Technol. 14, 33-47, 1991

NEA/CSNI/R(2009)5

236

[7] J. Keskinen, K. Pietarinen, M. Lehtimäki Electrical low pressure impactor J. Aerosol Sci.23, 353,

1992

[8] J.M. Mäkynen, J.K. Jokiniemi, E.I. Kauppinen, H. Tuomisto,

T. Routamo LWR Containment

Aerosol Experiments at Victoria Facility - Data Report 1/96; E.C. Report ST-APC(96)-P08 (1996)

[9] J.M. Mäkynen, J.K. Jokiniemi, E.I. Kauppinen, H. Tuomisto, T. Routamo LWR Containment

Aerosol Experiments at Victoria Facility – Final Report; E.C. Report ST-APC(98)-P19 (1998)

[10] J.M. Mäkynen, J.K. Jokiniemi, E.I. Kauppinen, A. Silde, S. Outa, T. Routamo, H. Tuomisto

Experimental and Modelling Studies on Containment Aerosol Behaviour in the VICTORIA Facility,

(1998)

[11] J.M. Mäkynen, T. Routamo LWR Containment Aerosol Experiments with the VICTORIA

Facility-Data report Experiment 62 (1999)

[12] G. Weber, S. Schwarz, F. Ewig, K. Fischer FIPLOC 3.1, Benutzerhandbuch Gesellschaft für

Anlagen- und Reaktorsicherheit (GRS) mbH, Januar 1998 GRS-P-2

[13] K.K. Murata et.al.User´s Manual for CONTAIN 1.1, A Computer Code for Severe Nuclear Reactor

Accident Containment Analysis Revised for Revision 1.11. Nov. 1989, Revised July 1990

[14] A. Silde Analyses of three containment accident sequences using the CONTAIN code: Source term

analyses for Loviisa nuclear power plant Research Report ENE4/58/96, VTT Energy, Espoo 1996

[15] S. Outa Modelling of Aerosol Behaviour in the Containment During Severe Reactor Accidents

Master's Thesis, Helsinki University of Technology, January 1998

5.4 Aerosol Release in the Course of MCCI (ACE-C)

As part of the internationally sponsored Advanced Containment Experiments (ACE) program, seven

large-scale experiments on molten core concrete interactions (MCCIs) have been performed at Argonne

National Laboratory. One of the objectives of these experiments is to collect and characterise all the

aerosols released from the MCCIs. Aerosols released from experiments using four types of concrete

(siliceous, limestone/common sand, serpentine, and limestone/limestone) and a range of metal oxidation

for both BWR and PWR reactor core material have been collected and characterised. Release fractions

were determined for UO

, Zr, the fission products: BaO, SrO, La

, CeO

, MoU

, Te, Ru, and control

materials: Ag, In and B

C. Release fractions of UO

and the fission products other than Te were small in

all tests. However, release of control materials was significant [1].

Results of aerosol release calculations by six groups from six countries were compared with the releases

from ACE MCCI test L6. The codes used for these calculations included: SOLGASMIX-PV,

SOLGASMIX Reactor 1986, CORCON.UW, VANESA 1.01, and CORCON mod2.04/VANESA 1.01.

Calculations were performed with the standard VANESA 1.01 code and with modifications to the

VANESA code such as the inclusion of various zirconium-silica chemical reactors. Comparisons of results

from these calculations were made with test L6 release fractions fro U, Zr, Si, the fission-product elements

Te, Ba, Sr, Ce, La, Mo and control materials Ag, In, and Ru. Reasonable agreement was obtained between

calculations and test L6 results for the volatile elements Ag, In and Te. Calculated releases of the low

volatility fission products ranged from within an order of magnitude to five orders of magnitude of test L6

values. Releases were over and underestimated by calculations. Poorest agreements were obtained for Mo

and Si [2].

NEA/CSNI/R(2009)5

237

References

[1] J.K. Fink, D.H. Thompson, B.W. Spencer, B.R. Sehgal Aerosols released during large-scale integral

MCCI tests in the ACE program Technical Report, January 1992

[2] J.K. Fink, M. Corradini, A. Hidaka, E. Hontanon, M.A. Mignanelli, E. Schrödl, V. Strizhov Results of

aerosol code comparisons with releases from ACE MCCI tests Technical Report, January 1992

5.5 REST, REVENT, ThAI Re-Entrainment Tests

Introduction

Only a few experiments on re-entrainment from boiling sumps have been performed under LWR accident

typical conditions. The most comprehensive work are the REST-tests. In the REVENT-tests especially the

re-entrainment during a controlled filter venting was investigated. Finally two large-scale re-entrainment

tests were carried out in the multi purpose ThAI containment facility.

REST tests

The REST tests (Resuspension Source Term) were carried out at the Kernforschungszentrum Karlsruhe

[1]. The re-entrainment of soluble and insoluble FP surrogates was measured at four different heating

levels. The superficial steam velocity lay both in the bubbly flow as well as in the low churn turbulent flow

regime. The geometric sump surface was 91.6 cm

In more than 100 separate tests the following

parameters were varied: heating power, concentration of soluble and insoluble materials, condensing and

non-condensing atmospheres above the sump, temperature and pressure.

Fig. 5.5-1 shows the measured re-entrainment of the FP surrogates in REST tests with dry atmospheric

conditions. The given superficial velocity is uncertain because the reported thermal hydraulic data for the

tests are not complete. The re-entrainment of insoluble material (Fe, Zn) is an order of magnitude higher

than for soluble material (CsI) because the insoluble material accumulates at the pool surface and thus in

the droplets. The scatter of the measured values is wider for the insoluble material than for the soluble

material.

REST-Tests: Measured Re-entrainment

Non-condensing atmosphere

1,0E-05

1,0E-04

1,0E-03

1,0E-02

0,00 0,02 0,04 0,06 0,08 0,10

Superficial steam velocity (m/s)

FP surrogate re-entrainment

soluble,

0,1% solution

bubbly flow

churn turbulent flow

insoluble,

0,1% solution

Fig. 5.5-1 Re-entrainment for insoluble and soluble material measured in the REST tests

NEA/CSNI/R(2009)5

238

The relative humidity in the atmosphere had a strong impact on the measured re-entrainment. The

re-entrainment was significantly higher in a dry, non-condensing atmosphere, where the droplets shrank by

evaporation of water and consequently sedimentation decreased. It was also observed in the REST tests

that at the beginning of boiling when there are many small bubbles the re-entrainment can increase

drastically (up to a factor 50) for a short time. Details of the mechanism are still not well understood and

the phenomenon has not been modelled.

In the REST-study at the basis of the measured values an empirical re-entrainment correlation for rough

estimations has been developed. In codes this correlation is not used because important dependencies of

the re-entrainment like the superficial velocity are not considered and the correlation is valid for a limited

gas flow range only.

REVENT tests

In the REVENT tests the re-entrainment of aerosols from boiling pools during a controlled filtered

containment venting was measured [2]. The vessel was a 1:20 linear scaled model of a PWR containment.

The pool was electrically heated (20 kW) and had a heated surface of 0.24 m². The vessel was equipped

with a depressurization system for a controlled pressure relief. In the REVENT facility re-entrainment was

measured under steady state conditions as well as under transient venting conditions. Soluble (Na

CsI, KI) and insoluble (Al

, SiC) fission product simulants were used.

For soluble material the measured re-entrainment is in the same order of magnitude as in the REST tests.

For the insoluble material the re-entrainment was found to be one order of magnitude smaller than for the

soluble one. This result is in contradiction to the REST tests where the re-entrainment for insoluble

substances was always significantly higher than for soluble ones. By the REVENT tests it was further

shown that solid airborne particles are always smaller than 3 µm. In the REVENT test a higher

re-entrainment was measured at steady state conditions than at venting conditions. The reason is a

pronounced volume condensation in the containment during venting, which causes droplet growth and

enhances aerosol settling considerably.

ThAI tests TH14 and TH15

Two tests, TH14 and TH15, in the 60 m³ ThAI containment facility of Becker Technologies, Eschborn,

have been devoted to the re-entrainment of fission products from a boiling sump [3]. The surface of the

boiling sump was 0.55 m². The sump was electrically heated. In order to increase the superficial gas

velocity additionally steam and/or air bubbles were released into the sump via a gas distributing system.

Fission products in the sump water were simulated by soluble CsCl and KI salts.

The superficial gas velocity was varied between 0.03 m/s and 0.12 m/s. The released droplets were dried

in the superheated vessel atmosphere, the residual salt particles being monitored by a Condensation Nuclei

Counter to determine the number concentration, and by a Differential Mobility Analyser to obtain the size

distribution of the salt particles. Additionally, filter samples have been taken to determine the aerosol mass

concentration. The tests performed with a boiling sump surface area of 0.55 m² and 72 kW heating power

yielded re-entrainment factors in the order of 5·10

-5

, re-entrainment being defined as the mass ratio of

released droplets and released steam. The re-entrainment values obtained confirm former REST

laboratory-scale test data of [1]. However, as a result of the more sophisticated aerosol instrumentation

applied in ThAI, the size of the released droplets and of the dried particles respectively turned out to be

significantly smaller. The mass median diameter of the dried particles was about 0.1 µm. The contribution

of re-entrainment to the overall source term should be re-evaluated in the light of the recent and coming

experimental findings.

NEA/CSNI/R(2009)5

239

References

[1] H. Bunz, M. Koyro, B. Propheter, W. Schöck, M. Wagner-Ambs Resuspension of Fission Products

from Sump Water CEC, JRC, EUR 14635 EN (Nov. 1992)

[2] M. Müller et al. REVENT Program – Aerosol Re-entrainment from Boiling Pool During Controlled

Filtered Venting after a Severe Core Melt Accident NEA/CSNI/R(98)4, Session VII (Feb. 2000)

[3] T. Kanzleiter, G. Poss, F. Funke, H.-J. Allelein ThAI Multi-Compartment Containment Test

Program 14

International Conference on Nuclear Engineering (ICONE-14) July 17-20, 2006,

Miami, Florida, USA

5.6 Pool Scrubbing Tests

Until mid 90’s, a good number of experimental programmes had addressed the pool scrubbing issue.

Table 5.6-1 and Table 5.6-2 [1] present, respectively, the main variables investigated and major specifics

of the tests conducted.

Table 5.6-1 Pool scrubbing experimental programmes conducted until mid 90’s

Programme Variables investigated

UKAEA [2]

•

Total gas flow

•

Steam mass fraction

GE [3]

•

Aerosol particle size

•

Injector diameter

•

Submergence

EPRI [4-6]

•

Total gas flow

•

Steam mass fraction

•

Submergence

•

Aerosol species

JAERI [7]

•

Total gas flow

•

Particle size

•

Aerosol concentration

•

Submergence

POSEIDON/

POSEIDON II [8-10]

•

Injection system

•

Submergence

•

Water temperature

ACE [11-13]

•

Total gas flow

•

Steam mass fraction

•

Submergence

•

Water temperature

•

Water pH

LACE-España [14]

•

Total gas flow

•

Steam mass fraction

•

Particle size

SPARTA [15]

•

Only one test conducted

CIEMAT (RCA and

GECI) [16]

•

Total gas flow

•

Steam mass fraction

•

Particle size

EPSI [17, 18]

•

Submergence

•

Pressure

NEA/CSNI/R(2009)5

240

Table 5.6-2 Boundary conditions of the pool scrubbing experimental programmes conducted until mid

90’s

Research

project

No.

of tests

We Submergence

(m)

% steam Pool conditions

UKAEA (1966) 7 6 10

- 1.8 10

0.61 4 - 75 Room

GE (1982) 18 Bubble regime 0.34 - 1.7 0 Room

EPRI (1986-91) 146 10 - 10

0.15 - 1.65 0 - 95 Room

near saturation

JAERI (1987) 1.5 10

- 2 10

1, 2 0 Room

POSEIDON

(1987)

1.5 10

1, 2, 3 100 273 ºC

1 - 6 MPa

ACE (1992) 4 5 10

- 1.5 10

1.4 - 4.6 1 - 40 25 - 80 ºC

LACE España

(1992)

11 5 10

- 1.5 10

2.5 10 - 90 110 ºC

3 bar

SPARTA (1992) 1 2 10

1.8 0 near saturation

CIEMAT (1996)

(RCA and GECI)

10 2.5 10

- 2.8 10

0.25 - 2.5 0, 50, 90 80 ºC

near saturation

POSEIDÓN-II

(1998)

17 0.3 – 4.00 0, 0.55,0.72 85 ºC

A good part of the experimental investigation were encapsulated in computer codes such as

SPARC90 [19], BUSCA [20] and SUPRA [21] which were principally developed in the 80’s. They contain

models based on the basic knowledge gained on pool scrubbing phenomena (as summarised in the

supplementary report on nuclear aerosols in reactor safety issued by NEA [22]) from the experimental

programmes carried out until late 80’s. Some limited model updates were done using the data made

available later. Nonetheless, no further major development were made from early 90’s although more

detailed information were made available by the tests carried out throughout the last decade of the XX

century.

Major analytical activities associated with the recent pool scrubbing research can be summarised as:

•

The Source Term project of the 3

FWP of Euratom [23]: a model for fission product scrubbing

under churn turbulent flow was developed. In addition, a peer review of last versions of SPARC and

BUSCA codes were assessed.

•

Simplified models for scrubbing of particles during Molten Core Concrete Interaction (MCCI) with

overlying water pool [24] and for BWR suppression pools [25]. Both of them were used to develop a

simplified model of decontamination accounting for the uncertainties associated with both scenarios.

NEA/CSNI/R(2009)5

241

References

[1] M.J. Escudero, M.J. Marcos Crespo, M. Swiderska-Kowalczyk, M. Martin Espigares, J. López

Jiménez State of the Art Review on Fission Products Aerosol Pool Scrubbing under Severe Accident

Conditions ST(93)-P61, Rev. 2, ITN/TS-22/DP-93, 1994

[2] J.J. Hillary, J.C. Taylor, F. Abbey, H.R. Diffey Iodine removal by a scale model of the SGHW

reactor vented steam suppression system UKAEA, TRG Report 1256 (W), August 1966

[3] S.A. Ramsdale, G.J. Barford, S. Fishwick, H.C. Starkie Status of research and modelling of water

pool scrubbing Final report, EUR 14566 EN, 1992

[4] J.C. Cunnane, M.R. Kuhlmann, R.N. Oehlberg The scrubbing of fission product aerosols in LWR

water pools under severe accident conditions - experimental results Proc. of Fission Product

Behaviour and Source Term Research, 1984

[5] D.D. Paul, L.J. Flanigan, J.C. Cunnane, R.A. Cudnik, R.P. Collier, R.N. Oehlberg Radionuclide

scrubbing in water pools - gas-liquid hydrodynamics EPRI-NP-41554, Vol. 1, 1985

[6] M.R. Kuhlmann, J.A. Gieseke, M. Merilo, R.N. Oehlberg Scrubbing of fission product aerosols in

LWR water pools under severe accident conditions IAEA-SM-281/47

[7] K. Hashimoto, K. Soda, S. Uno, H. Nakatani, H. Tateoka Effect of pool scrubbing of insoluble

aerosol in two-phase flow in a pipe IAEA-SM-296/60

[8] S. Güntay Experiment POSEIDON: Elemental iodine retention in water pools ANS 1990 Winter

Meeting, Washington D.C., November 11-15, 1990, Volume 62, TANSAO 062 1-722 (1990), ISSN:

0003-018X

[9] S. Güntay Experiment POSEIDON: Pool scrubbing effect on iodine decontamination pp 937-941,

ENC'90, ENS/ANS-Foratom Conference Transactions, Verlag TÜV Rheinland

[10] A. Dehbi, D. Suckow, S. Güntay Aerosol retention in low-subcooling pools under realistic accident

conditions Nuclear Engineering and Design 203, 2-3, 229-241, 2001

[11] J.D. McCormack, D.R. Dickinson, R.T. Allemann Experimental results of ACE vent filtration pool

scrubber tests AA1-AA4 and DOP1 - DOP5 ACE-TR-A1, January 1989

[12] K. Fischer, M.R. Kuhlmann, V. Kogan Fission product pool scrubbing data and modelling

assessment ACEX TR-A-01, Vol. 1, December 1996

[13] K. Fischer, M.R. Kuhlmann, V. Kogan Fission product pool scrubbing data and modelling

assessment ACEX TR-A-02, Vol. 2, June 1997

[14] M.J. Marcos, F.J. Gomez, I. Melches, M. Martin, J. López LACE-España Experimental programme

on the retention of aerosols in sugmerged beds Final report, CIEMAT, ITN/TS-08/DP-93, April

1993

[15] M. Furrer, R. Passalaqua SPARTA project: scoping test results and comparison with pool scrubbing

codes predictions ENEA-CRE Casaccia, January 1992

NEA/CSNI/R(2009)5

242

[16] V. Peyrés, M.M. Espigares, J. Polo, M.J. Escudero, L.E. Herranz, J. López Jiménez Pool scrubbing

and hydrodynamic experiments on jet injection regime CIEMAT 785

[17] K. Hashimoto, K. Soda, S. Uno High pressure pool scrubbing experiment for a PWR severe accident

ANS International Topical Meeting on Safety of Thermal Reactors, July 1991

[18] K. Hashimoto, K. Soda, S. Uno, H. Nakatani, H. Tateoka Effect of pool scrubbing of insolubile

aerosol in two-phase flow in a pipe IAEA-SM-296/60

[19] P.C. Owczarski, K.W. Burk SPARC-90: A Code for Calculating Fission Product Capture in

Suppression Pools NUREG/CR-5765, 1991

[20] S.A. Ramsdale BUSCA-JUN90 Reference Manual, SRD-R542, 1991

[21] A.T. Wassel, A.F. Mills, D.C. Bugby Analysis of Radionuclide Retention in Water Pools Nucl. Eng.

& Des. 90, 87-104, 1985

[22] NEA, Nuclear Aerosols in Reactor Safety: Supplementary Report, 1985

[23] J. López Jiménez, L.E. Herranz, M.J. Escudero, M.M. Espigares, V. Peyrés, J. Polo, Ch. Kortz, M.K.

Koch, U. Brockmeier, H. Unger, L.M.C. Dutton, Ch. Smedley, W. Trow, A.V. Jones, E. Bonanni,

M. Calvo, A. Alonso Pool Scrubbing. ST(95)-P195, ITN/TS-12/SP-95, 1995

[24] D.A. Powers, J.L. Sprung A Simplified Model of Aerosol Scrubbing by a Water Pool Overlying

Core Debris Interacting with Concrete NUREG/CR-5901, SAND92-1422, 1993

[25] D.A. Powers A Simplified Model of Decontamination by BWR Steam Suppression Pool

NUREG/CR-6153, 1997

5.7 Resuspension Tests

The experiments performed on the topic of resuspension of aerosols are divided into two classes. The first

class analyses the resuspension under a continuous airflow of at least several minutes. Such conditions are

expected in the primary cooling system or at a slow depressurisation of the containment in the course of an

accident scenario. The second class examines the influence of transient airflows after short time events like

hydrogen deflagrations or steam explosions.

5.7.1 Continuous flow experiments

PARESS

In the PARESS (Particle Resuspension Studies) experiments Fromentin [1] analyses the resuspension

under a continuous airflow in a wind channel. In a first step aerosol is injected into a sedimentation

chamber, where the aerosol settles under predefined conditions onto deposition coupons. In a second step

the deposition coupons are recovered and put into the wind channel. There a constant air current velocity is

adjusted and the removal of aerosol mass is measured in dependence of time. Different aerosol materials

are used in this study like iron oxide (Fe

), tin (Sn), and silicon (Si) or mixtures of these materials. A

resuspension flux

res

is observed that decreases inversely proportional to time. Furthermore other

influences like aging of the deposition layer, relative humidity and homogeneity of the deposition are

investigated.

NEA/CSNI/R(2009)5

243

STORM

The experimental conditions of the STORM facility at JRC are chosen to simulate the conditions in the

pressure relief lines of a PWR reactor in a station blackout scenario (de los Reyes [2]). The test section

consists of a 5.0055 m long pipe with an internal diameter of 63 mm. The test SR11 is chosen as ISP40 and

is performed in two phases. In the first phase the pipe is streamed by tin oxide (SnO

) aerosol with a carrier

gas of nitrogen and steam. During this period of 5 h a part of the aerosol is sedimented on the internal pipe

walls. In the second phase the pipe is fed with pure nitrogen. During this phase the prior sedimented

aerosol is resuspended and relocated inside the pipe or released with the carrier gas. The gas flow is

increased in six consecutive steps and the resuspension rate is determined by the aerosol concentration in

the carrier gas. A benchmark with different severe accident codes modelling resuspension has been

performed on ISP40. Further details of the STORM experiments can be found in chapter 5.2.3.

5.7.2 Transient flow experiments

VANAM-M4 add-on

Subsequent to the Vanam-M4 test in the Battelle Model Containment (BMC) an add-on experiment with

resuspension due to a hydrogen deflagration in a multi-compartment geometry is performed (Kanzleiter

[3]). In a first step hygroscopic NaOH aerosol and insoluble SnO

aerosol is injected into the inner

compartments of the BMC. The outer rooms are separated from the inner rooms by rupture foils. In a

second step hydrogen is injected into the same compartments as the aerosol. The hydrogen is ignited in the

inner rooms and the expanding atmosphere spreads out into the outer containment rooms. Although the

deflagration runs very mild, the generated air currents are able to resuspend round about 1 % of the initially

injected aerosol mass. It is observed that the resuspended aerosol material settles faster than the initially

injected aerosol. This indicates the generation of a larger aerosol component. The observation of the

particles found in filter stations with a scanning electron microscope show a wide particle spectrum with a

large number of very fine particles but also some very large particles, which carry a great fraction of the

aerosol mass.

AEREST

At technical university of Munich an experimental programme in the aerosol resuspension shock tube has

been performed (Schneider [4]). The aerosol test section is located in a 7 m long pipe with a pressure tank

on the input side. This pressure tank is discharged over a ball valve and the expanding atmosphere hits an

aerosol deposition bed that is deposited on coupons with different angles of attack, which are prepared in

an extra sedimentation vessel. The aerosol materials used in these experiments are silver particles of

different shapes and tin oxide (TiO

) particles. In these experiments a large scattering of the resuspended

aerosol mass is observed under comparable conditions. Anyway the AEREST experiments show a large

relative particle removal at high aerosol mass loads on the surface but only a minor relative particle

removal at lower particle mass loads. This can be explained by a larger adhesive force of particles located

directly on the surface in comparison to particles in higher positions of the particle bed.

ThAI experiments Aer-1, Aer-3, Aer-4

In the ThAI facility at Becker Technology three experiments on the topic of resuspension have been

performed (Kanzleiter [5]). For that purpose a horizontal deflagration tube has been installed into the

facility that was loaded with a hydrogen-air mixture. In a first phase CsI is injected into the atmosphere and

settles during a period of 18 h. Then the hydrogen in the deflagration tube is ignited and the expanding

atmosphere is released through a 2 * 50 cm² large nozzle over a deposition plate, on which aerosol material

has deposited during the sedimentation phase. In the three experiments Aer-1, Aer-3, and Aer-4 the

NEA/CSNI/R(2009)5

244

hydrogen load and thus the airflow velocities are varied. The findings are similar to that of the

VANAM-M4 test with a resuspended fraction of 1 % of the initially injected aerosol mass. Also in

accordance with VANAM-M4 is the higher settling velocity of the resuspended aerosol in comparison with

the deposition phase.

References

[1] A. Fromentin Particle Resuspension from a Multi–Layer Deposit by Turbulent Flow Paul Scherrer

Institut, 1989 (PSI 38)

[2] A. de los Reyes, J.A. Capitão, G.D. Santi International Stanard Problem 40 - Aerosol Deposition and

Resuspension OECD, 1999 (NEA/CSNI/R(99)4)

[3] T. Kanzleiter VANAM-Mehrraum-Aerosolabbau-Versuch M4 mit Mischaerosol (lösliches und

unlösliches Material) und Resuspension durch einen Wasserstoffbrand Battelle–Institut e.V., 1993

(BIeV-R67.098-305)

[4] R. Schneider Resuspension von Aerosolen durch transiente Strömungen im Containment Lehrstuhl

für Thermodynamik, TU München, 2003 (150 1102)

[5] T. Kanzleiter, G. Ahrens, K. Fischer, W. Häfner, A. Kühnel, G. Poss, F. Funke, G. Langrock, H.-J.

Allelein, G. Weber, S. Schwarz Versuchsanlage und Programm zur Untersuchung offener Fragen

zum Spaltproduktverhalten im Sicherheitsbehälter; ThAI Phase II Becker Technologies, Areva NP

und GRS, 2006 ( 150 1272 – S1)

5.8 Spray Systems (French Tests)

For model development and validation purposes concerning the effects of sprays in the containment

atmosphere, the Institut de Radioprotection et de Sûreté Nucléaire (IRSN) conducted two experimental

programmes; CARAIDAS (an extensive analytical programme) and TOSQAN (comprising

small-to-intermediate-scale tests that are still in progress). The Commissariat à l’Energie Atomique (CEA)

is carrying out some tests in the MISTRA facility where these are large-scale and complementary to those

of IRSN but in which it is not foreseen to include injection of aerosols.

The CARAIDAS programme comprised analytical tests covering three effects of sprays: droplet relaxation

(condensation and evaporation); aerosol capture; and iodine absorption, [1, 2]. The CARAIDAS apparatus

(see Fig. 5.8-1) involved a vessel of dimensions: height of 5.7 m; internal radius of 0.6 m; total volume of

nearly 1.4 m

. Single drops were generated that fell through a homogeneous air atmosphere with different

humidity levels and with or without cesium-iodide aerosols or iodine vapor, respectively. The atmospheric

temperature was controlled by the regulated wall temperature. Nearly 100 tests were performed showing

the collection efficiency to range over more than two decades and allowing development of modelling for

the ASTEC code covering droplet kinetic and thermal relaxation, gravitational coagulation, inertial capture

and interception of aerosols, and iodine mass transfer and chemical transformation in the droplets, [3]. In

particular, twenty-one of these tests were retained for evaluation of aerosol-capture modelling where the

polydisperse (GSD ≤ 1.7; 0.5 < AMMD < 5 microns) CsI aerosol interacted with pure-water droplets (see

Fig. 5.10-1). These aerosol-capture tests can in fact be grouped into five families: two analytical series

using particles with an aerodynamic volume-median diameter (AVMD) of 2-4 µm and either large

(700 µm diameter) or small (300 µm diameter) spray droplets; another two analytical series using smaller

particles (AVMD between 1 and 1.5 µm) with the same large and small droplets; and a fifth series in more

severe-accident representative conditions, i.e., higher pressure (0.2 - 0.4 MPa) and higher temperature

(120 to 140 °C). Unfortunately, this latter series was affected by large experimental uncertainties (with

NEA/CSNI/R(2009)5

245

consequent dispersion of results) arising mainly from unexplained difficulties with respect to droplet

collection at the containment base.

Fig. 5.8-1 Comparison of the ASTEC aerosol-capture model with the CARAIDAS data - after [1]

The TOSQAN experiments are performed in thermal-hydraulic conditions representative of containment

severe-accident scenarios (wall condensation, spray, sump). The principal aim has been the measurement

of steam condensation onto (or water evaporation from) spray droplets in atmospheres comprising steam

and non-condensable gases while, recently, a new phase including aerosols has started. The TOSQAN

facility is based on a vessel 7 m

in volume (4.8 m high, 1.5 m wide) with thermostatically-controlled walls

which is highly instrumented (gas temperature, steam concentration, droplet velocity, diameter and

temperature), [4]. The main thermal-hydraulic phenomena when sprays are used the mixing induced by

spray entrainment and the heat and mass transfer with respect to droplets and walls where validation of the

modelling of these processes constitutes the main objective of the programme, [5, 6]. The TOSQAN spray

test matrix consists of one reference test with an air-steam mixture and another with an air-steam-helium

mixture. Other tests were changed the spray injection rate, the spray temperature, the superheating, the

initial gas temperature, the difference between saturation temperature and injection temperature, and the

gas composition. A benchmark was initiated between SARNET partners based on TOSQAN Test 101

studying droplet and wall heat and mass transfer, [7]. In this test, a cold-water spray was injected into the

steam-air-filled vessel resulting in partial steam condensation and pressure reduction. This exercise

involved both lumped-parameter and computational fluid dynamics (CFD) codes with a first phase that

was a blind analysis and a second one (in 2005) using open results. It was seen that the calculated total

pressure was very sensitive to the modelling of droplet vaporization into the gas or onto structures.

Modelling of the spray injection required care and use of experimental data was necessary. A third phase

with open results is in progress that includes new experimental data on spray injection. In general,

calculated pressure is found to be lower than in the experiments where a possible explanation is the

NEA/CSNI/R(2009)5

246

droplet-wall interaction which is not taken into account in the calculations. A second TOSQAN spray

benchmark is being performed based on Test 113 which concerned mixing by sprays of an

initially-stratified air-helium atmosphere.

Fig. 5.8-2 Comparison of the ASTEC aerosol-capture model with the CARAIDAS data - after [3].

Tests designed for the validation of CFD codes were performed by CEA in the MISTRA facility, a much

larger-scale experiment (vessel volume of 100 m

) than TOSQAN, [8, 9]. The experiments study

axisymmetric and non-axisymmetric injection of steam into the containment with condensation on

temperature-regulated walls investigating heat and mass (steam condensation) transfer rates as well as

hydrogen distribution. Standard models for condensation used in lumped-parameter codes and multi-

dimensional codes have been validated. Some results of these tests are available to the SARNET

community with respect to tests M2 and M3, [10].

Finally, it is worth mentioning that many TOSQAN, MISTRA as well as ThAI results were used in

ISP-47 [11]. This ISP assessed lumped-parameter and CFD codes with respect to containment thermal

hydraulics involving analyses that comprised progressive modelling difficulties. The exercise highlighted

the dominant impact of user effects, especially concerning nodalization, both with lumped-parameter and

CFD codes. Furthermore, in spite of their higher numerical burden, CFD results were not generally better

than those of lumped-parameter codes. Nevertheless, for assessing hydrogen distribution in the

containment, combined use of both lumped-parameter and CFD codes was recommended it was concluded

that further code development and establishment of nodalization rules and user guidelines are necessary to

achieve more accurate predictive capabilities.

NEA/CSNI/R(2009)5

247

References

[1] D. Ducret, Y. Billarand, D. Roblot, J. Vendel Study on collection efficiency of fission products by

spray: experimental device and modelling 24

DOE/NRC Nuclear Air Cleaning and Treatment

Conference, Portland, USA, 15-18 July 1996, NUREG/CP-0153.2, 1996

[2] D. Ducret et al. Etude expérimentale et modélisation du rabattement des aérosols par des systèmes

d’aspersion 14e Congrès Français sur les Aérosols, Dec. 1998, Paris, France, 1998

[3] W. Plumecocq, V.D. Layly, A. Bentaib Modelling of the containment mitigation measures in the

ASTEC code, focusing on spray and hydrogen recombiners International Topical Meeting on

Nuclear Thermal-Hydraulics (NURETH-11), 3-6 October 2005, Avignon, France, 2005

[4] P. Lemaitre, E. Porcheron, G. Grehan, L. Bouilloux Development of a global rainbow refractometry

technique to measure the temperature of spray droplets in a large containment vessel Meas. Sci.

Technol. 17, pp 1299-1306, 2006

[5] J. Malet, E. Porcheron, J. Vendel Filmwise condensation applied to containment studies: conclusions

of the TOSQAN air-steam condensation tests International Topical Meeting on Nuclear Reactor

Thermal-Hydraulics (NURETH-11), 3-6 October 2005, Avignon, France, 2005

[6] J. Malet, W. Plumecocq, P. Lemaitre, E. Porcheron, J. Vendel Modelling of sprays in containment

applications: results of the TOSQAN spray tests International Congress on Advances in Nuclear

Power Plants (ICAPP), 4-8 June 2006, Reno, NV, USA, 2006

[7] A. Bentaïb, J. Malet, P. Lemaitre, E. Porcheron, J. Vendel, W. Plumecocq Modelling of sprays in

containment applications; Results of the TOSQAN spray benchmark European Review Meeting on

Severe Accident Research (ERMSAR), 14-16 November 2005, Aix-en-Provence, France, 2005

[8] M. Caron-Charles, J.J. Quillico, J. Brinster Steam condensation experiments by the MISTRA facility

for field containment code validation Proceedings of the International Conference on Nuclear

Engineering (ICONE), Volume 3, pp 1041-1055, 2002

[9] I. Tkatschenko, E. Studer, J.P. Magnaud, L. Blumenfeld, H. Simon, H. Paillère Status of the

MISTRA programme for the validation of containment thermal-hydraulic codes International

Topical Meeting on Nuclear Thermal-Hydraulics (NURETH-11), 3-6 October 2005, Avignon,

France, 2005

[10] A. Bentaïb, J. Vendel, H. Simon, L. Blumenfeld, I. Tkatschenko, H. Paillère Air-steam tests in the

MISTRA facility: Experimental results and validation of the lumped-parameter/CFD TONUS code

European Review Meeting on Severe Accident Research (ERMSAR), 14-16 November 2005,

Aix-en-Provence, France, 2005

[11] H.-J. Allelein, K. Fischer, J. Vendel, J. Malet, E. Studer, S. Schwarz, M. Houkema, H. Paillère, A.

Bentaib International Standard Problem ISP-47 on containment thermal hydraulics - final report.

OECD/NEA (to appear), 2006

5.9 Containment Venting Filters

Severity of the conditions during severe accidents varies depending on the successfulness of the accident

management measures attempted. However, anticipated conditions: high temperature, high pressure, high

steam concentrations, hydrogen burns or detonations, high aerosol and gaseous fission product

NEA/CSNI/R(2009)5

248

concentration, and resulting high activity levels, are very much depending on the accident progression