Full Text 01

Master Thesis
Application of CFD to Safety and Thermal-Hydraulic Analysis of

Lead-Cooled Systems
Marti Jeltsov
Supervisor:
Pavel Kudinov
Division of Nuclear Power Safety

Royal Institute of Technology
Stockholm, Sweden
June 2011
TRITA-FYS 2011:37
ISSN 0280-316X
ISRN KTH/FYS/11:37SE
ii
ABSTRACT
Computational Fluid Dynamics (CFD) is increasingly being used in nuclear reactor safety analysis as a tool
that enables safety related physical phenomena occurring in the reactor coolant system to be described in
more detail and accuracy. Validation is a necessary step in improving predictive capability of a computational
code or coupled computational codes. Validation refers to the assessment of model accuracy incorporating
any uncertainties (aleatory and epistemic) that may be of importance. The uncertainties must be identified,
quantified and if possible, reduced.
In the first part of this thesis, a discussion on the development of an approach and experimental facility
for the validation of coupled Computational Fluid Dynamics codes and System Thermal Hydraulics (STH)
codes is given. The validation of a coupled code requires experiments which feature significant two-way
feedbacks between the component (CFD sub-domain) and the system (STH sub-domain). Results of CFD
analysis that are used in the development of a flexible design of the TALL-3D experimental facility are
presented. The facility consists of a lead-bismuth eutectic (LBE) thermal-hydraulic loop operating in forced
and natural circulation regimes with a heated pool-type 3D test section. Transient analysis of the mixing and
stratification phenomena in the 3D test section under forced and natural circulation conditions in the loop
show that the test section outlet temperature deviates from that predicted by analytical solution (which the
1D STH solution essentially is). Also an experimental validation test matrix according to the key physical
phenomena of interest in the new experimental facility is developed.
In the second part of the thesis we consider the risk related to steam generator tube leakage or rupture
(SGTL/R) in a pool-type design of lead-cooled reactor (LFR). We demonstrate that there is a possibility
that small steam bubbles leaking from the SGT will be dragged by the turbulent coolant flow into the core
region. Voiding of the core might cause threats of reactivity insertion accident or local damage (burnout)
of fuel rod cladding. Trajectories of the bubbles are determined by the bubble size and turbulent flow field
of lead coolant. The main objective of such study is to quantify likelihood of steam bubble transport to
the core region in case of SGT leakage in the primary coolant system of the ELSY (European Lead-cooled
SYstem) design. Coolant flow field and bubble motion are simulated by CFD code Star-CCM+. First, we
discuss drag correlations for a steam bubble moving in liquid lead. Thereafter the steady state liquid lead
flow field in the primary system is modeled according to the ELSY design parameters of nominal full power
operation. Finally, the consequences of SGT leakage are modeled by injecting bubbles in the steam generator
region. An assessment of the probability that bubbles can reach the core region and also accumulate in
the primary system, is performed. The most dangerous leakage positions in the SG and bubble sizes are
identified. Possible design solutions for prevention of core voiding in case of SGTL/R are discussed.
Keywords: Coupled Codes, Verification&Validation, CFD, System Thermal-Hydraulics, Lead Cooled systems, Steam Generator Tube Rupture/Leakage, Bubble transport, Core voiding.
iii
iv
ACKNOWLEDGEMENTS
I would like to express my gratitude to my supervisor Pavel Kudinov for his priceless guidance and loading me
with exergy during this thesis project. I would like to thank Francesco Cadinu for the work and discussions
on development of methods for coupling CFD-STH codes, Walter Villanueva for the help to get acquainted
with CFD and discussions on TALL-3D and Aram Karbojian for providing the technical information and for
the development of the 3D-test section drawings.
I wish to thank also Johan Carlsson from JRC for the ELSY geometry and useful hints to get started.
Special thanks to Kaspar K
o
op who was the best Estonian buddy around and helped to settle into the
department and also for the constructive discussions on TALL-3D design.
Moreover, I would like to send a warm wishes to my program mates from the Nuclear Energy Engineering
09 masters, namely Paul, Simone, Greg, Song and HQ.
Big Thank you! to all Nuclear Power Safety, Reactor Physics and Reactor Technology guys and girls who
made my being here at KTH as cool as one could wish.
Last but not least I want to thank my family and friends in Estonia for offering me quality time during the
short visits I could do.
Suured t
anud Teile k
oigile!
This work is performed with support of the European Commissions 7th FP projects THINS and LEADER.
vi
LIST OF PAPERS AND PUBLICATIONS
I. F. Cadinu, M. Jeltsov, W. Villanueva, K. Koop, A. Karbojian and P. Kudinov, Program of work for
experimental tasks, software development and validation tasks on TALL, Technical Report, Royal
Institute of Technology (KTH), 2011.
II. M. Jeltsov, F. Cadinu, W. Villanueva, A. Karbojian, K. Koop and P. Kudinov, An approach to

validation of coupled CFD and system thermal-hydraulic codes, 14th International Topical Meeting
on Nuclear Reactor Thermalhydraulics (NURETH-14), 2011.
III. M. Jeltsov and P. Kudinov, Simulation of steam bubble transport in primary system of pool type
lead cooled fast reactors, 14th International Topical Meeting on Nuclear Reactor Thermalhydraulics
(NURETH-14), 2011.
vii
viii
CONTENTS
1 Introduction and background

1.1
1.2
1.3
Motivation
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.1.1
Validation of coupled CFD and STH codes . . . . . . . . . . . . . . . . . . . . . . . .
1.1.2
Steam generator tube leakage in a pool-type LFR design . . . . . . . . . . . . . . . . .
Theoretical background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2.1
Turbulent heat transfer in non-unity Prandtl number fluids . . . . . . . . . . . . . . .
1.2.2
Stratification and mixing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Goals and tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2 Development of the TALL-3D test section design

2.1
TALL-3D experimental facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.1
Specific requirements and description of the TALL-3D design . . . . . . . . . . . . . .
Calculations in support of the design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.1
Case I: Forced circulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.2
Case II: Natural circulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2.3
Transients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10
2.3
Identification of key physical phenomena and development of validation test matrix . . . . . .
12
2.4
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15
2.2
3 Analysis of a steam bubble transport in the primary system of a pool-type LFR design 17
3.1
3.2
Discussion of the scenarios and uncertainties . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17
3.1.1
Bubbles size distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19
3.1.2
Leak rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20
Selection of the bubble drag coefficient correlation in liquid lead . . . . . . . . . . . . . . . . .
21
3.2.1
Bubble shape and rise behavior regimes . . . . . . . . . . . . . . . . . . . . . . . . . .
21
3.2.2
Modeling bubble motion in a column of liquid lead . . . . . . . . . . . . . . . . . . . .
23
3.2.3
Drag coefficient correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
25
3.2.4
An approach to verification of modeling method . . . . . . . . . . . . . . . . . . . . .
26
3.2.5
Results of verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
27
ix
CONTENTS
3.3
Bubble transport to the ELSY core in case of SGTL . . . . . . . . . . . . . . . . . . . . . . .
28
3.3.1
ELSY reactor design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
28
3.3.2
Modeling of primary system at nominal operational conditions . . . . . . . . . . . . .
30
3.3.3
Modeling of bubble transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
33
3.3.4
Strategy for estimation of core voiding . . . . . . . . . . . . . . . . . . . . . . . . . . .
36
3.3.5
Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
38
3.4
Suggestions for mitigation of consequences of SGTL . . . . . . . . . . . . . . . . . . . . . . .
47
3.5
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
48
4 Summary
49
5 Outlook
51
Bibliography
55
LIST OF FIGURES
1.1
General view of a pool-type LFR. Circular annulus in the middle accommodates the core. . .
2.1
TALL loop configuration after the introduction of the CFD test section (7). The temperature
of the fluid at the heat exchanger (14) inlet is defined by the temperature at the CFD test
section outlet. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3D-test section geometry[18]. More than one train of thermocouples (1) will be used to measure
non-axisymmetric temperature field. There are TCs on the wall surface (2) to measure correct
temperature and determine heat losses. For CFD validation data, TCs on the disk surface
are used (3). The section is heated with a band heater (4). For the velocity measurements,
vertically and rotationally adjustable Pitot-Prandtl tube assembly is being implemented. . . .
2D temperature field (a), streamlines (b) and axial temperature distribution (c) for the steady
state forced circulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2D temperature field (a), streamlines (b) and axial temperature distribution (c) for the steady
state natural circulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10
Difference in outlet temperature between analytical and CFD solution. Transient from forced
to natural circulation, test section heater always at 5 kW. . . . . . . . . . . . . . . . . . . . .
11
Difference in outlet temperature between analytical and CFD solution. Transient from forced
to natural circulation, test section heater is switched switched to 5kW (from 0 kW) at the
beginning of the transient. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11
2.7
Different steady states and possible transients between them. . . . . . . . . . . . . . . . . . .
14
3.1
Three possible bubble behaviors in the core. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
18
3.2
Bubble diameter distribution dB [mm]. The width of the slit is 0.015 mm. Gas flow rate is
0.067 106 m3 /s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
19
3.3
Effect of slit dimensions on bubble diameter. Gas flow rate is 0.83 106 m3 /s. . . . . . . .
20
3.4
Vapor bubble size distribution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
20
3.5
Shape regimes for bubbles and drops in unhindered gravitational motion through liquids. . .
22
3.6
Column geometry used for bubble model analysis. . . . . . . . . . . . . . . . . . . . . . . . .
23
3.7
Bubble terminal rise velocity vs. bubble diameter. Analytical predictions by Stokes and
Mendelsons laws. Experimental data for velocities in Hg. . . . . . . . . . . . . . . . . . . . .
27
3.8
Bubble terminal rise velocities calcuated with different drag coefficient correlations. . . . . . .
27
3.9
Bubble terminal rise velocity vs. bubble diameter. Results obtained with and without modeling turbulent dispersion of a bubble. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
28
3.10 ELSY reactor reference configuration [44]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
29
2.2
2.3
2.4
2.5
2.6
xi
xii
List of Figures
3.11 Scheme of the primary pump - steam generator unit [45]. . . . . . . . . . . . . . . . . . . . .
29
3.12 Tubes - headers connections. Upper header is attached to the feed water line and the lower
one to the steam line [45]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
30
3.13 3D volume mesh and a vertical cross-section of it. . . . . . . . . . . . . . . . . . . . . . . . . .
32
3.14 Temperature field during normal operation. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
32
3.15 Velocity profiles in the steady state flow field . . . . . . . . . . . . . . . . . . . . . . . . . . .
33
3.16 Locations of the planes used as bubble injectors in SG. (Color bar legend shows the vertical
upward velocity of the lead at these planes). . . . . . . . . . . . . . . . . . . . . . . . . . . . .
34
3.17 Locations of the planes used as bubble injectors in core and in pump. (Color bar legend shows
the vertical upward velocity of the lead at these planes). . . . . . . . . . . . . . . . . . . . . .
35
3.18 Emergence rate as a function of number of seeds. The span of respective Point Inclusion
Probabilities is from 0.005 to 1. Injection plane at 5.2 m was used. . . . . . . . . . . . . . . .
35
3.19 Different locations where the probabilities for a bubble to escape the primary loop are estimated
(1-3). P1-P3 are defined at these locations as the probabilities for a bubble to stay in the loop. 36
3.20 Injection height 5.2 m. Bubble diameter 0.2 mm. . . . . . . . . . . . . . . . . . . . . . . . . .
38
38
39
39
39
40
40
3.27 Fraction of bubbles that reach the core inlet. Obtained with different drag correlations. . . .
41
3.28 Fraction of bubbles that are dragged to core. . . . . . . . . . . . . . . . . . . . . . . . . . . .
41
3.29 Fraction of the bubbles that are carried through the core and further into the pump and SG.
42
3.30 Fraction of bubbles that continue circulating in the primary loop. . . . . . . . . . . . . . . . .
42
3.31 Injection height 6.87 m. Bubble diameter 0.2 mm. . . . . . . . . . . . . . . . . . . . . . . . .
43
44
44
44
45
45
45
3.38 Fraction of the bubbles that reach the core inlet. . . . . . . . . . . . . . . . . . . . . . . . . .
46
3.39 Injection height 8.55 m. Bubble diameter 0.2 mm. Point inclusion probability is 1. . . . . . .
47
3.40 Injection height 8.55 m. Bubble diameter 1.0 mm. Point inclusion probability is 0.01. . . . .
47
LIST OF TABLES
2.1
Inlet boundary conditions corresponding to the two cases simulated for the design of the CFD
test section. In both cases, dinlet = 50 mm and is Q heater = 5 kW. . . . . . . . . . . . . . . .
Physical phenomena, which STH, CFD or STH-CFD coupled codes can be validated against.
The check marks in the table indicate which transient allows a particular validation. . . . . .
13
2.3
Validation test matrix. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
14
3.1
Liquid lead properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
24
3.2
Initial tentative parameters of ELSY plant [45], [46], [47]. . . . . . . . . . . . . . . . . . . . .
30
3.3
Defined source terms per region. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
31
2.2
xiii
xiv
List of Tables
CHAPTER
1
INTRODUCTION AND BACKGROUND
1.1
1.1.1
Motivation
Validation of coupled CFD and STH codes
Nuclear power plants are complex systems whose behavior is driven by the interactions between many different
physical processes at different scales. Quite naturally then, the modeling and simulation (M&S) of nuclear
power plants requires coupling between different physical models (which can span different time and length
scales).
In order to achieve high maturity of any (single or coupled) code, verification, validation and uncertainty
analysis must be performed. The validation process targets the accuracy of the code results compared to
the experimental measurements. The validation of a coupled code consists of two steps. First, every single
code has to be verified and validated (the validation is obtained by performing separate physical effect
tests). Experiments designed to validate coupled codes should feature mutual interconnections between
sub-components of the complete system resolved by each sub-code. This brings us to the second step
the algorithm which ties the codes being coupled, must be validated. Important requirement for validation
experiments is a measurement system that can provide adequate quality data for validation of both STH and
CFD. The importance, key aspects and a set of methods for validation of a code is described in a detailed
manner in Oberkampf et al. (2007)[1].
TALL facility was proposed as a platform for development of experiment for validation of coupled codes.
Pre-design analysis with CFD is necessary to satisfy requirements for such type of experiments.
1.1.2
Steam generator tube leakage in a pool-type LFR design
Lead and lead-alloy cooled Fast Reactor (LFR) systems constitute one of the six concepts of advanced reactor
design considered for research and development under the Generation IV framework. Mission and criteria
for development and operation of future fast reactors were discussed by Spencer (2000)[2], who provided a
comprehensive review of various aspects of using lead coolant technology. The use of heavy liquid metal
coolants (i.e. lead, LBE) presents an attractive potential for simpler, safer and economically efficient power
production due to the basic inherent inertness of the coolants, favorable neutronic and thermodynamic
properties. Nevertheless, in order to take advantage of mentioned features one needs to overcome problems
such as corrosion, coolant chemistry and operational issues related to hot, opaque coolant. A fine balance
between economics and safety of LFR lies on assuring the feasibility of having the steam generator (SG) in
the primary coolant circuit, thus eliminating the need for (and economic burden of) an intermediate circuit
as such in sodium-cooled fast reactors.
Currently proposed pool-type design of LFR has still safety issues that are waiting for resolution. A discussion
of different safety concerns associated with close proximity of steam generator to the core in pool type LFR
design can be found in Spencer (2000)[2], Hwang (2005)[3] and Ciampichetti (2010)[4]. Due to the high
pressure by design in the secondary side water circuit and of the large number of pipes housed in SG unit,
1
Introduction and background
the probability of a leak or rupture cannot be considered negligible. First, due to the high leak flow rates and
waterlead interaction in case of rupture of a SG tube there is a risk of over pressurization of the primary
vessel. This was shown by Ciampichetti et al. (2010)[4] when they injected water at 180 C and 185 bar
into LBE tank at 400 C. First sharp pressure peak was detected in the bulk liquid followed by subsequent
pressurization of the reaction vessel up to 2.4 MPa. Different story is when a small crack happens. Then one
needs to consider the transport of small steam bubbles, injected from the SG secondary side to the primary
system, by the turbulent coolant flow into the core region. Consecutive slow voiding of the core might then
cause threats of reactivity insertion accident or local damage (burnout) of fuel rod cladding. Trajectories of
the bubbles are determined by the bubble size and turbulent flow field of the lead coolant in the vicinity.
Moreover, it is practically difficult to detect the leak and identify its location at small flow rates. Some water
leak detection techniques in liquid metal systems have been developed in the past (e.g. [5], [6]) but they have
mainly been tested within sodium.
Leakage of SG tube is not a new safety issue per se. Significant efforts were devoted to changes in the
design, coolant chemistry, and adjustment of frequencies of SG inspections, to keep under control the SGTL
in pressurized water reactors (PWRs). Nevertheless, analysis of available statistics [7] for PWRs shows that
there were 9 cases of SG tube rupture in US during 1975-2000 and about 40 cases of SG tube leak incidents
during 9 years of 1990-1998. Comparing potential for occurrence of Steam Generator Tube Leakage/Rupture
(SGTL/R) in PWR and in LFR it is important to mention that lead, as a coolant, has higher density and
is more corrosive for the structural materials in comparison with water. These factors generally increase
frequency of SGTL/R occurrence.
In present thesis, a reactor system developed under the framework of European Lead-cooled SYstem (ELSY)
is used as a reference design for the investigation. The ELSY design aims to be a competitive and safe fast
critical reactor using simple technical engineering solutions, whilst fully complying the Generation IV goals
[8]. The ELSY is a 600 MWe pool-type reactor cooled by lead. Figure 1.1 illustrates a compact pool-type
reactor system with submerged steam generators and decay heat removal heat exchanger. In this design, the
primary coolant system is not pressurized. Yet, the secondary system has a pressure of 20 MPa. From the
PWR operational experience and from experiments performed on degraded SG tubing (e.g. [3]) it is well
known that a small leak usually precede a large rupture of the tube. The leak can last for many days. In
fact EPRI guidelines [9] recommend increased monitoring if leak exceeds 20 litres per day, and below that
value no specific actions are recommended during normal operation of a PWR.
Figure 1.1: General view of a pool-type LFR. Circular annulus in the middle accommodates the core.
Risk (probability multiplied by consequences) related to SGTL/R is one of the main criteria for licensing
of a pool-type LFR design. Given that potential direct consequences of SGTL/R are high (core damage)
and frequency is uncertain due to lack of operational experience, SGTL/R can become a show-stopper for
licensing of LFR technology. To assess the risk and provide adequate defense-in-depth for SGTL/R in LFR
both frequency and consequences need to be clarified.
Nevertheless, before one tackles to reduce the uncertainty in probability and consequences of SGTL accident,
it is necessary to look into the possibility that a bubble can be transported to the core in the first place.
Theoretical background
1.2
1.2.1
Theoretical background
Turbulent heat transfer in non-unity Prandtl number fluids
The turbulent heat and momentum transfer is of high importance in a variety of engineering applications. In
the framework of this thesis, we consider heat transfer in the flows with very low Prandtl number (order of
102 ). In lead-cooled systems, the fraction of natural convection to the forced flow is more significant than
it is in water cooled systems, for example.
Prandtl number P r is a dimensionless number which describes the ratio between kinematic (momentum)
diffusivity to thermal diffusivity. It is defined as:
Pr =
(1.1)
where and are the kinematic viscosity and thermal diffusivity, respectively.
P r number can be used to identify whether the heat transfer in the fluid takes place mainly in form of
conduction (low-P r fluids) or of convection (high-P r fluids). This is the reason why, when it comes to liquid
metals, the thickness of thermal boundary layer is bigger than velocity boundary layer. Therefore modeling
of the thermal-hydraulics of such fluids is somewhat different than modeling fluids with Prandtl number close
to unity (e.g. 0.7-0.8 for air, 7 for water).
At very low Prandtl values the nature of turbulent natural convection must be studied in detail [10], [11].
Turbulence modeling is important to obtain correct temperature and velocity fields. As an example, a DNS
(Direct Numerical Simulation) study of turbulent heat transfer in pipe flows and the effect of P r number is
done by [12].
The temperature fluctuations in the flow field increase with higher P r number. In low-P r fluids the fluctuations in velocity field are more frequent as the temperature field is smoother. This all implies the need to
introduce different turbulence modeling approaches for different fluids. Still today, there is lack of quantitatively credible experimental results on heat and momentum transfer parameters [12]. It is also very difficult
to measure many of the turbulence properties (stresses, turbulent heat flux etc), especially near the wall and
with non-intrusive methods.
1.2.2
Stratification and mixing
Mass and energy transport between interconnected enclosures is of interest, among many other industrial
applications, in nuclear reactor systems (containments, plenums, piping). The transport mechanisms can
be characterized by time and length scales which can differ by orders of magnitude, from low velocity
stratification development to mixing due to high velocity jets. Stratified conditions are induced by the
difference in densities and/or temperatures. The main sources for mixing inside an enclosure are wall jets
created by natural convection boundary layer on a heated wall and free jets generated by fluid injection or
heat sources. Peterson et al. [13] showed that convenient scaling parameters for design of scaled experiments
for large stratified volumes (e.g. in reactor containment system) can be derived for stratified conditions.
Their work results in non-dimensional parameters which govern when the onset and breakdown of ambient
stratification occurs for enclosure flows driven by wall jets and free jets. When considering a tank with fluid
injection from the bottom, a transient from thermally mixed to stratified is a rather slow process requiring
heat source in the tank. In this process the injection rate from the bottom has to be low to form higher
temperature layers the top of the tank. Essentially the injected jet (can be buoyant and/or forced) is of too
low momentum for causing the breakdown of the stratification and collection of hotter fluid to the top part
is enhanced. The opposite transient is usually faster and is due to injection of sufficiently high momentum
jet into the tank which gradually breaks the stratification down. Several useful references are available for
mixing by jets in mixed and non-mixed environments [14], [15] and [16].
1.3
Goals and tasks
The aim of the first part of the thesis is to develop a design of experimental facility for validation of multiscale STH-CFD coupling methods for reliable prediction of steady state and transient thermal-hydraulic
phenomena in liquid metal cooled reactors systems. The existing TALL facility is selected as a platform
for development of such experiments. The facility consists of a liquid lead-bismuth (low Prandtl number)
Introduction and background
thermal-hydraulic loop operating in forced and natural circulation regimes with a heated fuel rod simulator.
In order to justify the need for CFD code taking part in calculations the facility loop must feature significant
3D flow effects, for which we are adding a small pool in the loop. The pool, or 3D section, must be designed
in a way that 3D effects are strong enough to impose significant two-way feedback between the loop and
the 3D section. Star-CCM+ CFD code will be used to analyze different designs of 3D section with respect
to different loop configurations and physical phenomena taking place in the tank. The basis of the design
development study is to achieve:
Full mixing of the 3D test section when the TALL loop is in forced circulation conditions.
Significant thermal stratification development in the 3D test section when the TALL loop is in natural circulation conditions, which changes the 3D section outlet temperature and affects its transient
behavior.
After reaching these points, it is important to create surrogate model of a 3D test section within the TALL
loop STH model to confirm the feedbacks and STHs incapability in capturing the loop multi-scale behavior
(transients essentially). This surrogate model development part, however is not embraced in this thesis. After
the design has been finalized, a test matrix for the experimental program that aims at providing systematic
data for validation of different coupling approaches and codes must be developed. The test matrix must
include all key physical phenomena possibly present in the TALL-3D loop and capture important system
(1D) and local (3D) phenomena in separate effect and coupled behavior. Finally, in order to be able to
perform reliable and comprehensive validation work, the requirements for instrumentation system are to be
defined.
The goal of the second part of this work is to perform analysis of the steam bubble transport to the core
of ELSY in case of a small leakage from the SG. We start with defining and quantifying the epistemic
uncertainties playing role in the SGTL accident. Important unknowns such as uncertainty in the accident
scenario (size and morphology of the crack) and in the modeling (bubble size distribution and leak rates)
must be defined and explained. Also the uncertainty in the drag coefficient closure model have to be reduced
by selecting a drag correlation for steam bubble moving in liquid lead that fits the best with the available
experimental data and analytical solutions. Next, we model the ELSY LFR geometry and simulate its
thermal hydraulics aiming to achieve the nominal full power operation conditions of the reactor. Thereafter,
steam bubbles are injected to the system at different locations in the SG to simulate the SGTL. Here we
use the drag coefficient correlation that gives the best results in the previous point. The main objective is
to quantify the likelihood that a steam bubble is transported by the primary coolant flow to the core and
to estimate the probability that a steam bubble continues to circulate in the primary loop without escape.
With these probabilities calculated, it is also possible to assess the void accumulation rate in the primary
loop at different leak rates and bubble size distribution. If the resulting probabilities suggest that in case
of SGT leakage it is probable to have void accumulation in the reactor core (or in the circulating primary
coolant flow), then it is necessary to develop solutions to prevent SGTL in the first place or mitigate the
consequences considering changing the design of the primary system..
CHAPTER
DEVELOPMENT OF THE TALL-3D TEST SECTION DESIGN
This chapter contains the description of work done in the support of development of the TALL-3D experimental facility and development of an approach for the validation of coupled Computational Fluid Dynamics
(CFD) and System Thermal Hydraulics (STH) codes.
Firstly, a short overview of the existing TALL facility is given. Then we dig deeper into the TALL-3D specific
issues a discussion of requirements for the facility whose purpose is expressly validation of CFD, STH and
coupled codes. According to the specific needs, the development methods (criteria) for the geometry of the
3D section in the TALL loop is given.
Second section comprises calculations that confirm that the selected design of the 3D section meets the
required criteria are. Results for both circulation steady state regimes, natural and forced, are presented.
Transients, in which case the 1D STH code alone is expected not to perform correct results, are also discussed.
Two last sections are concentrating on the development of the experimental validation test matrix.
2.1
TALL-3D experimental facility
Any experimental facility that is going to be used for validation of coupled codes has to meet two basic
criteria. The first and perhaps most important one is that feedbacks between local and integral phenomena
in different sub-domains resolved by different codes are significant. Secondly, a facility should allow separate
validation of every code used in the coupled system. [17], [18]
The TALL-3D facility is a modification of KTHs TALL, a Lead-Bismuth eutectic, 7 meters tall, thermalhydraulic loop which has previously been used for the study of natural and forced circulation transients.
The TALL facility consists of a primary and secondary loop. In the current configuration, the primary loop
consists of a pump, an electrical heater, a heat exchanger and piping. The internal diameter of the main
piping is 27.8 mm. The maximum LBE velocity in the heater section is 2 m/s. Experiments performed in the
TALL facility, both, in the forced and natural circulation flow regimes, have already been used to validate
STH codes [19].
In TALL-3D, a new test section (Figure 2.1) is introduced in the existing loop-type facility, representing the
CFD sub-domain in the coupled code analysis. The CFD test section provides different feedbacks to the
system depending on experimental conditions.
5
Development of the TALL-3D test section design
Figure 2.1: TALL loop configuration after the introduction of the CFD test section (7). The temperature of the
fluid at the heat exchanger (14) inlet is defined by the temperature at the CFD test section outlet.
2.1.1
Specific requirements and description of the TALL-3D design
In TALL-3D, the goal of the CFD test section design is to obtain a strong two-way feedback between the
local thermal hydraulic phenomena inside the test section and the system dynamics of the loop. A STH
code, then, is not expected to capture the behavior of the system alone.
This goal is obtained by leveraging the dynamic interplay between the following key physical phenomena:
(i) development of stratification at small flow rate (in natural circulation flow in the loop) inside a heated
pool-like CFD test section; (ii) mixing in the test section at high flow rate (at forced loop convection); (iii)
transient natural circulation in the loop under conditions of changing 3D tests section outlet temperature
(which is in turn affected by the loop flow velocity and mixing/stratification phenomena). It is important to
note that in the fully mixed regime or in steady state loop circulation a 1D modeling (heat balance and total
pressure drop) can be applied to resolve the effect of the 3D tests section on the loop. The local 3D phenomena
(mixing/stratification) are important for the integral system behavior only in transients where the 3D test
section pool is not completely mixed and the instantaneous outlet temperature Tout can significantly deviate
from what is predicted by a simple heat balance. Tout is important because it affects the natural circulation
flow rate in the loop. This ensures the presence of multi-scale interactions between the component and the
system dynamics, mediated by the physics of natural circulation.
TALL-3D experimental facility
CFD modeling has to resolve a number of physical phenomena to capture the outlet temperature transient
behavior. Specifically important are: (i) the buoyant plumes and forced jets, (ii) the jet/plume impingement
and interactions with the obstacles and walls inside the test section, (iii) erosion of thermally stratified layer
by buoyant plumes and forced jets, (iv) development of buoyant boundary layer on the 3D test section heater
surface, and finally (v) interactions between the jet/plume and flows created in buoyant boundary layers
that define the recirculation dynamics in the test section. Consistently with the theory of buoyant jets in
pool-like geometries, a full mixing in the test section can then be achieved for inlet velocities above a certain
critical level Vcrit [14].
The goal of the pre-design CFD calculations (presented in the next section) is to select the main test section
parameters (geometry, loop mass flow rate, heater power) in such a way that the pool is completely mixed
in forced circulation regime and thermal stratification develops in natural circulation regime.
There other desirable requirements for the test section design. First, the design has to be as simple as possible,
ideally 2D axisymmetric. Second, the design should be inherently flexible with respect to the parameters
(dimensions etc.) in order to allow validating the widest range of code coupling strategies. Third, the
boundary between pure 1D and 3D flows has to be defined as clearly as possible. Finally, the quantity of
interest for the loop dynamics, which is the temperature profile inside the component, should be accurately
measured, providing data suitable for the separate effect validation of the CFD code. The instrumentation
must allow also measuring velocity profiles inside the 3D component and the integral pressure difference
over the section. Mass flow and temperature measurement instruments for the rest of the loop are already
implemented in the existing facility.
Figure 2.2: 3D-test section geometry[18]. More than one train of thermocouples (1) will be used to measure nonaxisymmetric temperature field. There are TCs on the wall surface (2) to measure correct temperature and determine
heat losses. For CFD validation data, TCs on the disk surface are used (3). The section is heated with a band
heater (4). For the velocity measurements, vertically and rotationally adjustable Pitot-Prandtl tube assembly is
being implemented.
The design that meets all above mentioned criteria is presented in Figure 2.2. Changeable test section inlet
nozzle (with different inlet diameters), vertically movable disk and band heater with adjustable power are
considered in the design for flexibility in providing different configurations of the test section. The length of
the inlet and outlet pipes is sufficient to provide fully developed flow, which enables to define realistic inlet
and outlet boundary conditions for 3D test section. The disk is introduced in the upper part of the test
section to provide an obstacle for the buoyant jet, thus enhancing the mixing in the test section. The inlet
diameter is chosen to ensure that in forced flow conditions the jet reaches the disk and creates a large scale
recirculation flow that mixes the pool even if the heater is switched on. In natural circulation flow conditions,
the momentum of the jet is not enough to penetrate the developing thermally stratified layer.
It is instructive to note that in the first version of the test section design an immersed heater along the
central vertical axis, was considered. After preliminary CFD analysis it became obvious that this scheme
is not capable to provide both stratification development in the natural circulation regime and mixing in
forced circulation. This was related to the complicated interactions between the buoyant jet at the inlet
and buoyant boundary layer on the heater wall. Both flows had the same direction promoting mixing and
inhibiting development of stratification in case of natural circulation regime in the loop. By decreasing the
inlet jet velocity to the value at which stratification can be developed in natural circulation regime, it was
found that sufficient mixing was not achievable in the forced circulation regime. Furthermore, the immersed
heater represents an obstacle for the free jet which might introduce additional undesirable complications in
the 3D phenomena of tests section behavior. Therefore, the version of the design with the band heater was
chosen.
2.2
Calculations in support of the design
The goal of the calculations in support of the design is to confirm that, for the test section geometry shown
in Figure 2.2, full mixing is obtained in steady state forced circulation in the loop and thermal stratification
develops in steady state natural circulation conditions.
The simulations have been performed with the CFD code STAR-CCM+, version 5.06 [20], by solving the
steady state RANS equations using the segregated solver and treating the gravity term using the Boussinesq
approximation. Turbulence is modeled with the realizable k model, using a two layer formulation developed
for buoyancy driven flows (Xu model [21]). It can be expected that, from the quantitative point of view,
some of these modeling hypotheses might have an adverse effect on the accuracy of the simulation results
(in particular the hypothesis of the flow being 2D axisymmetric). On the other hand, the scope of these
calculations is mainly to obtain a qualitative confirmation that stratification develops for a given set of inlet
conditions with sufficient margin. Therefore, the modeling hypotheses above were deemed to be reasonable
defaults. The calculation matrix with the corresponding inlet conditions is summarized in Table 2.1. The
characteristic values of mass flow rates in forced (4.77 kg/s) and natural (0.83 kg/s) circulation conditions are
taken from previous tests in the original configuration of the TALL facility. Performed assessments suggest
that additional pressure drop in the 3D test section is minor and characteristic values of the flow rates in the
modified TALL-3D facility are not going to change significantly.
Table 2.1: Inlet boundary conditions corresponding to the two cases simulated for the design of the CFD test section.
In both cases, dinlet = 50 mm and is Q heater = 5 kW.
Cases
Inlet conditions
Description
Case I
Inlet velocity vin =0.239 m/s

Inlet temperature Tin =609 K
Case II
Inlet velocity vin =0.042 m/s

Inlet temperature Tin =695 K
Inlet conditions corresponding to a steady state

forced circulation in the unmodified TALL loop with
mass flow rate m
=4.77 kg/s.
Inlet conditions corresponding to a steady state natural circulation in the unmodified TALL loop with
mass flow rate m
=0.83 kg/s.
2.2.1
Case I: Forced circulation
In the forced circulation case the inlet velocity and temperature are, respectively, 0.239 m/s and 609 K (the
inlet is located 35 cm below the test section). The calculated temperature distribution and the streamlines
pattern in the pool are shown in Figure 2.3.
The interaction of the high momentum jet with the disk produces a recirculation pattern characterized by
the presence of two, large scale counter-rotating vortexes (Figure 2.3.b). The vortex at the top of the test
section mixes the cold jet fluid with the hot fluid adjacent to the heater. The vortex at the bottom of the
test section drives the hot fluid adjacent to the heater towards the bottom of the test section. Therefore, the
action of both vortexes tends to homogenize the temperature field inside the test section. Predictably, a hot
spot is present in the stagnation point between the vortexes and the wall.
The resulting temperature field (Figure 2.3.a) shows that the recirculation induced by the jet-disk interaction
and buoyant boundary layer on the heater mixes effectively the fluid in the test section. Figure 2.3.c illustrates
that temperature in most of the cells in the simulation domain is uniform around 625 K except the jet region
where it is determined by the inlet jet temperature (609 K) and thin layer in the vicinity of the heated wall
where it has peak value of 642 K.
Figure 2.3: 2D temperature field (a), streamlines (b) and axial temperature distribution (c) for the steady state
forced circulation.
2.2.2
Case II: Natural circulation
In the natural circulation regime in the loop the inlet velocity and temperature for the 3D test section
are, respectively, 0.042 m/s and 695 K. The calculated streamlines and temperature profiles are shown in
Figure 2.4.
In this case, the low momentum jet is not able to penetrate thermally stratified layer and it dissipates not
reaching the disk at the top. The top bulk part of the pool is mostly stagnant. A buoyant boundary layer
flow develops along the heater surface and pushing hot liquid through the gap between the disc and the top
wall of the test section to the outlet. Figure 2.4.c shows an almost constant temperature gradient in the
top part of the test section. The difference between temperatures at the bottom and at the top in steady
state conditions is about 50 K. Although the volume of the test section is stratified and the temperature
distribution is not uniform, the outlet temperature in steady state is defined by the heat balance and can be
predicted by a STH 1D code. However, the transient development of stratification and mixing in the tests
section is a complex 3D process that is generally not resolved by a 1D code.
10
Figure 2.4: 2D temperature field (a), streamlines (b) and axial temperature distribution (c) for the steady state
natural circulation.
2.2.3
Transients
Transient calculations were performed to show the behavior of the 3D test section. The analytical solution
for the temperature at the 3D section outlet can be obtained with the assumption that the tank is always
uniformly stirred. Meaning that if the fluid with a temperature Ti enters the tank then it gets immediately
mixed and the bulk temperature, which is assumed to be also the outlet temperature can be obtained by
solving the following heat balance expression for the tank:
dT
F
Q
= (Ti T ) +
dt
V
V cp
(2.1)
where T is the bulk temperature of the 3D test section (equal to outlet temperature), F is the volumetric
flow rate, V is the volume of the tank, Ti is the inlet temperature, Q is the heater power, is density and
cp is isobaric specific heat. Tanks time constant can be defined as = V /F . After solving this equation for
different transients (different initial conditions) one obtains following expressions:
Forced to natural
(heater always on 5 kW)
T (t) = 738 122 e 259.6 t
Forced to natural
(heater 0 kW to 5 kW)
T (t) = 738 129 e 259.6 t
The comparison between the outlet temperatures of 3D section calculated with CFD code and with analytical
solution, shows that with current design of 3D test section there is a deviation in CFD calculated 3D test
section temperature from the analytical solution. This implies that the 3D effects are present. Figure 2.5
shows the difference in case of transient from forced circulation to natural (loss-of-pump) whereas the heater
is always at 5 kW.
11
14
TsimulationTheat balance
12
temperature [K]
10
200
400
600
800
1000
1200
time [s]
Figure 2.5: Difference in outlet temperature between analytical and CFD solution. Transient from forced to natural
circulation, test section heater always at 5 kW.
Figure 2.6 shows the difference in case of transient from forced circulation to natural (loss-of-pump) but the
heater was at 0 kW in the forced steady state and is switched on in the beginning of the transient.
4
TsimulationTheat balance
2
0
temperature [K]
2
4
6
8
10
12
14
16
200
400
600
800
1000
1200
time [s]
Figure 2.6: Difference in outlet temperature between analytical and CFD solution. Transient from forced to natural
circulation, test section heater is switched switched to 5kW (from 0 kW) at the beginning of the transient.
Both transient CFD results show deviation from analytical solution. It is worth to mention that in the steady
12
states the deviation is zero, which implies that the flow 3D effects are affecting only the transient results
whereas the deviation is always less than 13 K. Whether this is enough to influence the mass flow (and
temperatures) in the loop to desired extent regarding requirements for system-component feedback depends
on the temperature differences between the cold and the hot leg. They should be in the comparable range
to cause the strongest feedbacks.
2.3
Identification of key physical phenomena and development of

validation test matrix
Any attempt towards validation of coupled codes must be preceded by a separate validation of the STH and
CFD components. The validation tasks can, then, be broken down into three sub-tasks:
Separate effect validation of STH
Separate effect validation of CFD
Validation of Coupled Codes
Successful validation of separate STH, CFD and coupled codes implies that all important physical phenomena
can be resolved with sufficient accuracy by the respective codes. Important physical phenomena that define
the behavior of TALL-3D facility are presented in Table 2.2. The list of physical phenomena is divided in
three parts that correspond to validation of STH, SFD and coupled codes respectively. To provide data for
code validation against key physical phenomena we propose the validation test matrix presented in Table 2.3.
Three classes of experiments are envisioned: forced circulation steady states (SSF), natural circulation steady
states (SSN) and transients (T). The nomenclature On and Off in Table 2.2 and Table 2.3 refers to the
state of the 3D test section heater. It is instructive to note that total number of steady states (4) is defined
by 2 circulation regimes (natural or forced) in the loop and two states of the heater (On or Off). We use
following notations for defining the steady states (see also Figure 2.7):
Forced circulation in the loop, 3D test section pool heater Off

Forced circulation in the loop, 3D test section pool heater On
Natural circulation in the loop, 3D test section pool heater Off
Natural circulation in the loop, 3D test section pool heater On
SSFOff
SSFOn
SSNOff
SSNOn
Identification of key physical phenomena and development of validation test matrix
13
Table 2.2: Physical phenomena, which STH, CFD or STH-CFD coupled codes can be validated against. The check
marks in the table indicate which transient allows a particular validation.
Table 2.2 lists the physical phenomena relevant to each test. It is important to note that steady state data
only enables separate effect validation against fewer basic phenomena. That helps to implement step by step
validation with gradually increasing complexity of the task.
Given the final goal, which is validation of coupled STH to CFD codes, we prioritize tests according to the
expected significance of the feedbacks between 3D test section and system (loop) behaviors. Tests #1 to #6
belong to the high priority group and #7 to #12 belong to the low priority group. Both groups of transients
can be executed as a continuous sequence in a single experiment.
14

Table 2.3: Validation test matrix.
Steady state
Forced circulation
Pool heater ON
Steady state
Natural circulation
Pool heater OFF
Steady state
Natural circulation
Pool heater ON
mass flow
Steady state
Forced circulation
Pool heater OFF
pool heater power

Figure 2.7: Different steady states and possible transients between them.
The validation tests matrix presented in Table 2.3 can be executed for each fixed configuration of the TALL3D facility. The configuration of the CFD test section is defined by (values in bold are used as base case
configuration):
The inlet diameter d (40 mm; 50 mm).

The disc axial position zD (15 mm (upper); 150 mm (middle); 285 mm (lower)).
The test section heater power Qheater (2.5 kW; 5 kW; 10 kW).
Conclusions
15
In addition to the above, the test section heater timing can be also considered as variable parameter, however
number of possible transients in this case is beyond reachable in practical sense.
2.4
Conclusions
We have presented an approach to design an experimental facility for validation of STH, CFD and coupled
STH-CFD codes. General requirements for a code validation experiments together with specific requirements
for the proposed TALL-3D facility are described. In order to meet those criteria, we have performed a
computational analysis, which have shown qualitatively that the necessary significant two-way feedback
between the implemented 3D-test section and the system is achievable. Expected full mixing in steady state
forced flow conditions and thermal stratification in steady state natural circulation flow conditions are both
confirmed. A preliminary design of the 3D test section with certain degree of flexibility has been selected.
The list of key physical phenomena has been discussed and used for development of validation matrix and
procedures that cover both separate effect test measurements and complex transient tests which feature the
two-way feedback. A series of transient pre-tests simulations is necessary to finalize selection of the test
parameters for the transient tests.
16
CHAPTER
ANALYSIS OF A STEAM BUBBLE TRANSPORT IN THE PRIMARY

SYSTEM OF A POOL-TYPE LFR DESIGN
The first section of this analysis discusses the uncertain parameters and scenarios that are related to steam
generator tube leakage and associated possible threats to the LFR reactor core. Secondly, the epistemic
uncertainty in the drag coefficient correlation for a steam bubble in liquid lead is addressed. Validation
of the proposed approach to prediction of steam bubble trajectories in lead is presented. Using the drag
correlation which matches best the analytical and available experimental data, analysis of a steam bubble
transport in the primary coolant system in case of leakage of steam generator tube is performed. The
probabilities that bubbles can reach the core and the accumulation rates with respect to different bubble
sizes are estimated.
3.1
Discussion of the scenarios and uncertainties
There are at least three different scenarios that can emerge as consequences of a leak from a steam generator
tube:
Homogeneous voiding of the coolant Very small bubbles are leaking from the steam generator
tube. Bubble size is up to 0.5 mm and the leak rate is very low, about 0.01 l/min, in the early stage of
the crack development. This may lead to a situation where there are small steam bubbles circulating
in the primary system and passing through the core. Small bubbles are not expected to get stuck
in the core region, instead they threaten the core by being homogeneously distributed over the core
and therefore representing an effective void present in the coolant. This can cause criticality (power)
oscillations throughout the whole core. This is illustrated in Figure 3.1 (a).
Bubbles stuck in spacers Bubbles from the leak are somewhat bigger than in previous case, say
0.5 mm to 2 mm. It can happen that they do not fit through the free space and at the same time
the surface tension forces are stronger than buoyancy forces driving bubble upwards. As additional
bubbles reach the same location, they coalesce, introduce higher amount of void and also a larger dry
(voided) surface area on a fuel pin. This scenario is thought to cause problems mainly per assembly,
meaning that local neutron multiplication factor of a particular assembly may increase or flow blockage
can cause damaging of fuel. This is illustrated in Figure 3.1 (b).
Bubbles stuck at the core inlet Depending on the geometry (see 3.3.1 for detailed description),
there may be corners behind which local flow stagnation (closed vortexes) places may appear. Bubbles
can get caught in these local vortex zones one after another, accumulate and eventually form a big
bubble there. Then, either due to its too big size or a disturbance in flow field, this bubble can start
moving towards the core. Now depending on its size and shape, it can get stuck at the core inlet or be
dragged as a long slug into a core channel. This is a transient process and poses a risk of criticality,
local overheat and flow blockage. This is illustrated in Figure 3.1 (c).
17
18
Analysis of a steam bubble transport in the primary system of a pool-type LFR design
Figure 3.1: Three possible bubble behaviors in the core.
Each of those scenarios contains a number of unknown factors and uncertain parameters.
The first step towards the reduction of uncertainties in an application is to identify and thereafter quantify
them. There are big uncertainties in SG degradation probability and types, in lead-cooled systems due to
lack of experimental research and operational experience. The whole process, starting from the development
of a small crack in one of the steam generators tube and its expansion in time, ending with a different set of
consequences caused by bubbles that actually reach the core, is full of unknowns to begin with. In general
one can describe the approaches to treat the uncertainty in two principally distinct ways:
1. Epistemic or systematic uncertainties are due to the properties, sizes, conditions, factors that we
possibly could know but we do not. Also the assumptions and parts we neglect in the model rise the
fraction of this type of uncertainty. This approach is based on deterministic way of estimating the
errors in the system. Identification of this type of errors is a step towards reduction of them. By doing
more accurate measurements, taking all possible factors in the system into account helps to reduce this
uncertainty. Still, it is deemed to be potential uncertainty, meaning that the inaccuracy may or may
not exist (even if there is lack of knowledge, we sometimes model the phenomena correctly)[22].
2. Aleatory or statistical uncertainties stem from the fact that every time we measure, observe, model,
simulate some system we will have different results. The scenarios of the same accident can lead to
different results. There is no real way for an experimentalist to eliminate those entirely. What one
should do then is to quantify the uncertainty in this case. This can be done by increasing the number
of tests, using more particles, different methods/devices of measurements, increase the mesh sizes.
Common examples is applying the Monte Carlo methods in analysis or performing mesh convergence
studies. Identification of this type of errors allows to quantify them.
The uncertainty in every system can be considered in both of the aforementioned ways. In reality, an engineer
who is performing such assessment must choose and distinguish which approach to use for quantification of
uncertainty considering the complexness of the system and costs of treating different sources of uncertainties
as aleatory or epistemic.
To add the validity of decisions, choices, engineering competence of a user (human mostly) that can never be
eliminated and is impossible to quantify too, the map of uncertainty becomes an extremely important issue
to deal with. Therefore, in order to achieve most reliable M&S results, one first needs to turn as many type 1
uncertainties into type 2 and then quantify the latter one. A deeper discussion on treatment of uncertainties
with respect to safety in nuclear systems and respective computer codes can be found in papers by Theofanus
(1996)[23] and by Pourgol-Mohamad et al. (2011)[24].
Discussion of the scenarios and uncertainties
19
In the following chapters the main sources of the uncertainties such as the size distribution of the bubbles,
the leak rates from cracks, the correlation for bubble drag coefficient, are addressed.
3.1.1
Bubbles size distribution
The bubble size distribution is dependent on the gas-liquid properties, orifice dimensions and orientations,
flow directions (co-current, counter-current, stagnant), gass flow rate, gas chamber volume. Out of many
studies performed on bubble formation at single submerged orifices, Leibson et al. (1956)[25] showed that
the bubble size is relatively uniform at given Re and depends mainly on orifice diameter. Marmur and Rubin
(1976)[26] found in the analysis of slow bubble formation process that a bubble sustains in equilibrium at
maximum the radius of the orifice and this radius depends on the liquid properties. Reported observations
show that the size of the bubble formed, dB , is affected by the viscosity, however the effect diminishes at
large bubble diameters and higher flow rates. Surface tension effect increases with bigger orifice sizes and
thicknesses, this affects the detachment time and hence the diameter. Among aforementioned impacting
factors, also the concentration of surface active agents (surfactants) has a determining role since they cause
variations in surface tension forces [27].
A study of bubble formation through different opening morphology and respective bubble sizes performed
by Terasaka et al. (2007)[28] showed that in case of slit-like orifice the diameters of the bubbles can fall into
sub-millimeter scale. Since Terasaka et al. studied bubbles in water which physical properties differ from
liquid metals, one can use their results only qualitatively. If a crack appears in a SG tube, it is reasonable to
think it will be slit-like rather than a circular opening. Figure 3.2 shows the resulting bubble size distribution
from one of the slits they used. The most frequent bubbles had a diameter between 0.45 mm and 0.50 mm.
They also showed that the slit length does not affect the bubble diameter, but the width does (wider slit
creates larger bubbles). This is shown in Figure 3.3.
Figure 3.2: Bubble diameter distribution dB [mm].

0.067 106 m3 /s.
The width of the slit is 0.015 mm.
Gas flow rate is
20
Figure 3.3: Effect of slit dimensions on bubble diameter. Gas flow rate is 0.83 106 m3 /s.
Experimental data for bubbles formed into lead is very scarce. Together with uncertainties in the crack
location, properties and flow field at that location, it is impossible to provide fully validated analysis for the
bubble size distribution. Beznosov et al. (2005)[29] studied a process when water with pressure of 2224
MPa and temperature of 150250 C was pumped through a 10.0 mm diameter opening into liquid lead at
temperatures 500600 C. The minimum radius they detected was 0.5 mm which is corresponding to the
highest resolution of the measurement system. According to their results, the most probable bubble radius
is around 1 mm. The measured distribution of vapor bubble radius is shown in Figure 3.4.
Figure 3.4: Vapor bubble size distribution.
According to the results of the above mentioned studies, we have chosen bubbles with diameters of 0.2 mm,
0.4 mm, 0.5 mm, 1.0 mm, 2.0 mm, 4.0 mm and 6.0mm to be considered as possible realistic bubble sizes.
3.1.2
Leak rate
Among other factors, also the gas flow rate through the leak is determining the size of the appearing bubbles.
Leakage flow rate, in turn, depends on morphology, area and geometry of the crack and the driving pressure
difference between two sides [30], [3]. In systems where heavy liquid metal, especially lead and lead-bismuth,
is circulating, so called leak before break (LBB) is important due to possible slow corrosion and degradation
of the SG tube wall. We assume that hard-detectable failures, mainly small cracks due to corrosion and
stresses are the most dangerous. Critical chocked flow phenomena in case of a large opening (rupture of SG
Selection of the bubble drag coefficient correlation in liquid lead
21
tube) is not considered in this work.

To explain SG tube leakage and rupture phenomena further, it is useful to present studies performed by
Hwang et al. [3], [31] and [32]. These studies were initiated due to a plant outage in a Korean light water
nuclear power plant where SG tube leaks were reported. Because cracks in highly corrosive environment can
develop already at relatively low stresses (high pressure difference is not necessary), even 100% through-wall
cracks are found not to be a sufficient criterion for detectable leakage at normal operation conditions of PWR
[3]. Therefore, it is expected that cracked tubes at operating pressure of 10.8 MPa will not show any leakage.
Depending on the morphology of the opening (crack), first leaks were detected at pressures around 17 MPa
to 25 MPa (which is characteristic range of pressure difference in pool-type LFR). The corresponding flow
rates were from 0.005 l/min (at 23.4 MPa) to 0.25 l/min (at 31.7 MPa). As a conclusion of all experiments
performed by Hwang et al.(2004; 2005; 2008), including a series of burst rupture tests, the leak rates start
from almost zero (initial state of the crack), develop up to several liters per minute after some time and can
have the maximum rate of 30-50 L/min (choke of the flow becomes the limiting factor).
For lead-cooled systems it is important to assess the impacts of small leak rates to the system, since they are
difficult to detect. In case of accumulation of smaller amounts of steam void takes place in the core, critical
safety limits can be still tackled (and exceeded).
3.2
Selection of the bubble drag coefficient correlation in liquid

lead
For validation of the approach, it is crucial to demonstrate that numerical solution obtained with selected drag
coefficient can reproduce experimentally observed terminal rise velocities for different bubble diameters. This
section discusses the selection of a correlation for a steam bubble drag coefficient. For that purpose, terminal
rise velocity of a steam bubble in steady liquid lead column is calculated with different drag correlations
developed in the past. Then the predicted velocity is compared with analytically developed solutions and
also with available experimental data on a terminal rise velocity of a bubble in heavy liquid metals.
3.2.1
Bubble shape and rise behavior regimes
We begin with a theoretical description of the bubble shapes and motion in a quiescent viscous liquid.
Dynamics of a bubble motion can be affected by many factors such as temperature, viscosity, pressure,
purity of the surrounding fluid. Different regions of the bubble size, shape and the respective vertical
upwards motion behavior are generally defined. The different properties and behavior of different bubbles
makes the realistic bubble size distribution in the system a very important part of the study. An overview
of the studies investigating bubble formation, sizes and rise velocities in water and in higher viscosity fluids
(unfortunately not molten metals) is extensively presented in paper by Kulkarni and Joshi (2005) [27] and
Yang et al. (2007) [33]. Further into the theory behind the bubbles, drops and particles is explained in a
book written by Clift et al. (1978) [34].
Therefore, prediction of terminal rise velocity is not a trivial task, especially taking into account a non-linear
behavior of drag coefficient, which dependents on the size and the shape of the bubbles. Different regimes,
that govern the bubble motion in a continuous liquid with respect to different inter-facial shear (or drag) and
resulting terminal velocity, can be defined as done by Maneri and Vassallo (2000) [35]:
Spherical - dB < 0.25 mm
In this region viscous forces dominate and their shape is relatively spherical. The terminal rise velocity
is well described by Stokes law (bubble velocity is proportional to the square of the diameter). Flow
around the bubble is smooth, streamlines reattach fully after the bubble, no separation occurs.
Ellipsoidal - 0.25 mm < dB < 1.0 mm
In this region, with increasing bubble volume, the pressure on the front side increases which flattens
the bubble in the direction of motion.The rise velocity reaches the peak of about 20 cm/s (depends on
the gas and fluid of course) after which the smooth streamlines are destroyed and turbulent wake is
formed. This wake grows with the bubble diameter and results in gradual decrease in terminal velocity.
Nevertheless, this turbulent wake reattaches downstream rather steadily .
Ellipsoidal (oscillatory) - 1.0 mm < dB < 2.5 mm
When bubble diameter exceeds 1.0 mm, the vortexes behind the bubble do not reattach in a steady
22
manner any longer inwards rolling eddies produce oscillation. Bubble shape stretches out alternately
in width and in motion direction. This elongation or compression in the vertical direction can be
explained by the varying pressure field below the bubble due to varying vortex field.
Ellipsoidal (wobbly) - 2.5 mm < dB < 5.0 mm
Bubble size is large enough to be effectively affected by the surface tension forces present in the liquid
film between the bubble sides and the wall of the injecting slit/crack/opening. This causes initial
perturbation. The bubble starts to oscillate, wobble, stretch and distort about a planar elliptical
shape as rising. Now, after reaching the lowest rise velocity (happens in the mostly oscillating regime)
increasing dB makes the buoyancy forces more dominating and the velocity starts to increase, again.
Cylindrical cap - 5.0 mm < dB < 10.0 mm
Now the bubble vertical cross section resembles a cap with flat bottom and spherical top. Top surface is
relatively stable, bottom may somewhat wobble and distort and cause vortex shedding. Inertial forces
are dominating from here on.
Slug - dB < 10.0mm - The rise velocity approaches to a steady value around 18-19 cm/s and is
independent on the equivalent bubble diameter. The bubble is called slug when the horizontal crosssectional area is larger than two-thirds of the test section area.
The resulting plot is shown in Figure 3.5. This brings qualitative but very broad insight to the relations
between terminal velocities (present in Re number) and shape regimes (Eo and M o).
Figure 3.5: Shape regimes for bubbles and drops in unhindered gravitational motion through liquids.
23
The dimensionless groups, the Reynolds number (Re), the Etvs number (Eo) and the Morton number (M o)
are often used to describe the shape of a bubble and its rise behavior
3.2.2
Re =
l ub db
l
(3.1)
Eo =
gdb 2
(3.2)
Mo =
gl 4
2l 3
(3.3)
Modeling bubble motion in a column of liquid lead
In this work, I have used the Lagrangian model for bubble transport in a continuum flow field of liquid
lead. For the task of validation of bubble drag correlation, a lead column of 0.5 m x 0.5 m x 5 m, was
modeled. It was built using 3D-Cad software existing in Star-CCM+ package. The top and the bottom
surfaces of the column were defined as walls and particle interaction type was escape there. The sidewalls
of the column were defined as symmetry planes with rebound option for particles. The geometry can be
seen in Figure 3.6. Both, the surface and the volume were meshed using internal mesher of Star-CCM+.
Column geometry consists of 86 444 vertices, 23 826 cells and 116 788 faces.
Figure 3.6: Column geometry used for bubble model analysis.
Liquid lead material properties were defined in Star-CCM+ according to the recommended correlations for
main thermo-physical properties [36]. All the values are based on reference temperature of 750 K (476.85 C),
which is in the range of operating temperatures of ELSY coolant (core inlet and outlet temperatures 400 Cand
480 C, respectively). All defined material values are presented in Table 3.1.
24
Table 3.1: Liquid lead properties
Parameter
Value Parameter
Value
Tref (K)
Density (kg/m3 )
Thermal expansion coefficient (1/K)
Specific heat (J/kgK)
Mol. viscosity (Ns/m2 )
Therm. cond. (W/mK)
Tcritical (K)
Pcritical (MPa)
750.00
10 471.2
1.14104
145.5
0.00124
17.45
4870
100
2016.00
8.46104
2016.00
858 200
0.43425
207.2
1 737.97
Tboil (K)
Psat (Pa)
Tsat (K)
Qvaporization (J/kg)
Surface tension (N/m)
Mol. weight (g/mol)
Speed of sound (m/s)
Lead flow field, through which bubbles are transported, is described as a space with specified formulations in
it, called continuum. The physics continuum is a continuous phase whose governing equations are expressed
in Eulerian frame of reference. In Star-CCM+, the continuum is a collection of models that represents
the substance (fluid or solid) being simulated [37]. Bubbles were assumed to be incompressible; their size
(density) was not a function of pressure. Density and viscosity of the bubble gas was assumed equal to water
vapor density at 100 C under atmospheric pressure.
For modeling stagnant flow without turbulence, the steady state solution was imposed as initial condition.
All the solvers for flow parameters were turned off, only Lagrangian Steady State solver is now active. Lead
flow velocity was set to zero and temperature was 450 C. Models chosen for this (liquid Pb) continuum, were
following:
Constant density
Gravity
Segregated flow
Segregated Fluid Temperature
Steady
Three dimensional
Laminar
Liquid
Lead
Lagrangian Multiphase
Water vapor
In a reactor simulation, it is important to account the effect of turbulence, since it allows bubbles to jump
from one averaged streamline to another. Without the effect of turbulence, the bubbles will always follow
the same streamline, which is physically unreasonable for a turbulent flow. Set of used models for turbulence
is following:
Constant density
Gravity
K-Epsilon turbulence
Reynolds-Averaged Navier-Stokes
Segregated flow
Segregated Fluid Temperature
Steady
25
Three dimensional
Turbulent
Two-Layer All y+ Wall Treatment
Liquid
Lead
Lagrangian Multiphase
Water vapor
Turbulent dispersion
So-called random-walk technique[20] is employed in turbulent dispersion model of Lagrangian phase to simulate the fluctuating velocity field. A bubble is assumed to be affected by a sequence of eddies as it travels
through the turbulent flow field. Every eddy causes a local disturbance to the Reynolds-averaged velocity
field
v = v + v 0
(3.4)
where v is the local Reynolds-averaged velocity and v 0 , is the eddy velocity fluctuation, unique to each
particle. The magnitude of the fluctuation is random at each time instant and has a normal (Gaussian)
distribution with zero mean value and a standard deviation given by eddy velocity scale, which is described
by the following formula
r
r
2
lt 2
=
k
(3.5)
ue =
t 3
3
where lt and t are the length-scale and time-scale of the turbulence and k is turbulent kinetic energy provided
by the turbulence model (k-epsilon used here) [37]. In present case, the value for turbulent kinetic energy
was 0.1 J, which corresponds to eddy velocity fluctuation of about 0.25 m/s. No average flow was actually
modeled in this case. One can envisage a situation when a steady pool is stirred to introduce eddies with
constant energy while the averaged flow is zero through the pool.
Terminal velocity of rising bubble in laminar stagnant column was detected as the bubble velocity at the
column outlet. In case of turbulent stagnant flow the terminal rise velocity was calculated based on the
column height and particles residence time.
3.2.3
Drag coefficient correlations
In 1992, Karamanev and Nikolov [38] experimentally showed that the trajectories of rising bubbles do differ
from the ones of free falling spherical particles. Also the terminal rise velocity was shown to be smaller in
case of rising bubbles. Since, it was realized that in addition to skin drag (which is affected by internal
circulations too), there is so called form drag present in case of non-rigid bubbles (different shape regions
described in earlier chapter). This makes the whole drag formulation for bubbles more complex. Production
of the drag force that tends to slow down the relative motion of a bubble is one of the most important effects
of viscosity on the displacement of the bubble in a liquid. The drag force is essentially a balance between the
work done by the drag force and the viscous dissipation within the fluid environment and can be formulated
as
Cd vT 2 A
Fd =
(3.6)
2
where Cd is the balancing (proportionality) constant (or drag coefficient) and can have different forms depending on Re number, systems physicochemical properties of the liquid and the bubble dimensions. vT
is the terminal relative velocity between two phases and A is the cross-sectional area. One of the widest
overview for the drag coefficient and different correlations of the last, is performed by Kulkarni and Joshi
(2005) [27].
The drag coefficient, Cd , has a determining influence on the bubble terminal rise velocity. Cd in turn is a
function of Re in most cases (the effect decreases with low Re). The complexity of estimation of Cd as well as
Re is due to the time-dependent viscosity and its effect on the bubble shape. Since it is not straightforward
to obtain the rise velocities directly, one aims to develop correlations for one of its variants (viz. Cd , Re).
In this thesis, I have tackled the uncertainty reduction in bubble terminal rise velocity by choosing the drag
correlation that fits the best with both, existing data and analytical solutions.
26
There are four different drag coefficient correlations in the analysis. Star-CCM+ provides two general methods
for defining the drag coefficient:
Schiller-Naumann correlation
User-Defined Field Function (UDF)
Firstly, the only correlation implemented in Star-CCM+, Schiller-Naumann correlation, is used. This correlation is suitable for spherical solid particles and small diameter bubbles (as it is defined in [37]). It is
formulated as:
24 (1 + 0.15Re0.687 ),
B
Cd = ReB
0.44,
ReB 103
(3.7)
ReB > 103
The dispersed phase (bubble) Reynolds number is defined as:

ReB =
c |vr |dB
c
(3.8)
where c and c are the density and the dynamic viscosity of the continuous phase (lead) respectively, vr is
the relative velocity between the bubble and lead, and dB is spherical bubble diameter.
Secondly, Schiller-Naumann coefficient correlation with modification in the low Reynolds number region
according to Stokes law was used [39]. This was implemented using a User-Defined Field Function. It has a
form of:
d
24 2 + 3 c
ReB 2
ReB 3 + 3 dc
Cd =
(3.9)
24
(1 + 0.15Re0.687
), 2 < ReB 103
ReB
0.44,
ReB > 103
where d is the dynamic viscosity if the dispersed phase (steam bubble).
Thirdly, a correlation proposed by Rodrigue [40] was used:
9/4
(0.5 + 32 + 0.5 1 + 128)1/3
16
+ (0.5 + 32 0.5 1 + 128)1/3
Cd =
ReB
128 1/9
8/9
1/9
)ReB M o
+ 0.036(
3
(3.10)
where = (0.018)3 ( 32 )1/3) ReB (8/3) M o(1/3) .

Fourth correlation used in this work was developed by Tomiyama et al. [41]:

16
48
8 Eo
0.687
Cd = max min
(1 + 0.15ReB ),
,
ReB
ReB 3 Eo + 4
(3.11)
This general correlation fits well with available experimental data.
3.2.4
An approach to verification of modeling method
Since there is no available experimental data on bubble terminal rise velocity in lead, the simulation results
were compared to Stokes law that predicts velocity for small, spherical shape bubbles and to Mendelsons
equation that predicts velocities in surface tension- and inertia-dominated regimes [42]. Comparison of Stokes
law and Mendelson equation with experimental data for bubble terminal rise velocity in mercury obtained
by Mori et al. [43] is shown in Figure 3.7. As it can be seen, the bubble velocity is predicted reasonably well
by Mendelsons equation in the surface tension and buoyancy dominated regime. Small bubbles are expected
to behave as solid spheres and their terminal velocity is described by Stokes law. This agreement provides us
a reference against which we are validating the results obtained with different drag coefficient correlations.
27
0.6
Stokes law
Mendelsons eq. for Pb
Mori et al. data for Hg
0.5
velocity [m/s]
0.4
0.3
0.2
0.1
10
bubble diameter [mm]

Figure 3.7: Bubble terminal rise velocity vs. bubble diameter. Analytical predictions by Stokes and Mendelsons
laws. Experimental data for velocities in Hg.
3.2.5
Results of verification
Four different drag coefficient correlations discussed in previous system were tested. First calculations were
performed without activating turbulent dispersion model for particles. Figure 3.8 presents the terminal rise
velocities obtained using different drag correlations. It can be seen that Schiller-Naumann correlation is
applicable only for very small bubbles (which behave as they were rigid spheres). The modified SchillerNaumann correlation does not improve the prediction. Rodrigues correlation shows a better agreement than
Schiller-Naumann, but is still deficient in capturing the peak in the curve. Tomiyamas drag correlation, which
aims to match the whole drag curve, provides the best agreement with the Stokes solution and Mendelson
equation for bubble rise velocities. Therefore, Tomiyamas correlation was used in following study of bubble
transport in reactor conditions.
0.6
0.5
velocity [m/s]
0.4
0.3
0.2
Stokes law
Mendelson eq.
SchillerNaumann
Modif. SchillerNaumann
Rodrigue
Tomiyama et al.
0.1
10
Figure 3.8: Bubble terminal rise velocities calcuated with different drag coefficient correlations.
28
Without turb. dispersion

With turb. dispersion
velocity [m/s]
0.4
0.3
0.2
0.1
Figure 3.9: Bubble terminal rise velocity vs. bubble diameter. Results obtained with and without modeling
turbulent dispersion of a bubble.
A comparison between the results obtained with and without turbulent particle dispersion modeling was
done for Tomiyamas correlation. As it appears in Figure 3.9, a difference between terminal rising velocity
of a bubble in laminar and in turbulent lead column is negligible as it is expected in isotropic homogeneous
turbulent flow field.
3.3
Bubble transport to the ELSY core in case of SGTL
In this section we address the possibility of bubble transport from the steam generator to the core region
in the ELSY LFR design. First a brief overview of the preliminary configuration of ELSY design and its
nominal full power operation parameters are presented. Then the averaged steady state full power operation
flow is resolved in a realistic model of the ELSY primary system. Injector features in Star-CCM+ are used
to initiate bubbles in the system. The behavior of individual bubbles in the obtained steady state flow field
will be calculated using Lagrangian specification, taking into account dispersion of the bubbles by turbulent
vortexes. The results are containing estimation of the probability that a bubble of certain size and injected
at a certain location in the SG can reach the core region is given.
3.3.1
ELSY reactor design
Geometry
The primary system of ELSY design comprises of a circular reactor vessel that accommodates all the internals
which are used for safe heat transfer from the core via moving liquid lead to the water as working fluid in the
secondary system. The vessel is a pool that has a fixed roof (large annular steel plate) and a half-spherical
bottom head. The volume between the primary coolant free level and the reactor roof is filled by a cover gas
plenum. Reactor core is submerged in the middle part of the pool and surrounded by 8 steam generators
in direct contact with molten lead, located at the top part of the pool. To fit all required components,
the outline dimensions are: vessel outside diameter of 12.3 m by 8.65 m of height. The ELSY reference
configuration can be seen in Figure 3.10 [44].
29
Figure 3.10: ELSY reactor reference configuration [44].
SG in lead-cooled systems provides heat removal from low-pressure heavy liquid metal on the primary side to
the high-pressure water/steam flow on the secondary side. The preliminary configuration of ELSY foresees
8 Steam Generator Units (SGU) (see Figure 3.11), each of them characterized by spiral-wound tube bundle
housed in the bottom-closed, annular space with vertical inner and outer permeable shells. The inlet of
each tube is connected to the feed water header and the outlet is connected to the steam header. Each
SG has a coaxial flow pump in it which provides the head required to force the coolant radially through
perforated inner shell, past heat exchanger tube bundles, through the outer shell. Effectively this scheme
is a counter-current heat transfer since the feed water in tubes circulates from outer spiral to inner spiral
(Figure 3.12).
Figure 3.11: Scheme of the primary pump - steam generator unit [45].
30
Figure 3.12: Tubes - headers connections. Upper header is attached to the feed water line and the lower one to the
steam line [45].
Operational parameters
The pool-type ELSY fast reactor, used as a reference case here, has electrical power of 600 MW [45]. Thermal
efficiency is about 40%. The nominal operational parameters that are considered in the preliminary ELSY
configuration for the steady state flow are following:
Table 3.2: Initial tentative parameters of ELSY plant [45], [46], [47].
3.3.2
Parameter
Value
Mass flow
Core inlet temperature
Core inlet temperature
Core pressure drop
Steam generator pressure drop
Total primary pressure drop
16.05 ton/s/SG
400 C
480 C
0.9 bar,
0.3 bar
1.4 bar
Modeling of primary system at nominal operational conditions
The simulation domain represents 1/8th slice of the ELSY reactor vessel, comprising one SG including hot
leg and pump and a slice of the core region together with corresponding parts of the hot/cold plenum and
downcomer.
The parameters that were presented in the previous section were used to determine a set of boundary
conditions and uniformly distributed heat source term in the core region, heat sink term in the steam
generator region and momentum source term in the pump region (see Figure 3.13), which are presented in
Table 3.3.
31
Table 3.3: Defined source terms per region.
Region
Core
Steam generator
Kq [kg/m4 ]
Q[W/m3 ]
Porosity
Mom. source [N/m3 ]
112635
9.2 107
0.5
0.65
0.6
0.9
540 000
0.9
T 335 C
145 C
1.25 107
Upper/lower plenum
112635
Pump
74093
Shielding
2.2527 10
187.5M W
VSG
A source of momentum in the pump was adjusted together with quadratic resistance coefficients in the core,
SG, upper/lower plena and shielding regions, to obtain design mass flow rate. Heat sink expression, as
described in Table 3.3, assured that the lead temperature at the SG inlet was 480 C. If the temperature
was lower or higher than design temperature, then less or more heat was removed respectively in the steam
generator. In total, 9 regions were defined in the simulation domain. All the solid boundaries in the design
are modeled as adiabatic, no-slip walls. Left and right side of the simulation domain are defined as symmetry
planes. And free surfaces (hot and cold) are defined as adiabatic, slip walls.
Modeling flow physics
Firstly, to obtain design based operational steady state flow field, a physics continuum was defined, which
contains the collection of all models that represent the simulated substance (liquid lead and steam particles).
For mass, momentum and energy equations the coupled implicit solver was used. Density was modeled
as constant (Boussinesq model is invoked when this option is chosen together with gravity, meaning that
variation in density due to temperature is still taken account in gravity term in momentum equation by the
use of thermal expansion coefficient). Choice for the turbulence modeling was the RANS realizable k
model, using a two-layer ally + formulation [37].
Modeling and meshing ELSY geometry
The geometry, used in the simulations, was extracted as boundary surface from initial volume mesh which
was created with Pointwise Gridgen by our colleague Carlsson in JRC [48]. The new mesh was created with
Star-CCM+ internal mesher.
The mesh used in the calculation consists of 679 480 polyhedral cells. Target characteristic size of a cell in
the domain was 0.07 m. We used 5 layers of prism cells for the modeling of boundary layers on the solid
walls. The total thickness of the prismatic cell layer was 0.02 m and smallest prism cell height near the wall
was 0.0015 m and the wall y + values were mostly below 100. The 3D volume mesh (with distinct regions)
and the mesh on a vertical cross-section is shown in Figure 3.13.
32
Figure 3.13: 3D volume mesh and a vertical cross-section of it.
Steady state flow results

A satisfactory steady state was achieved at nominal mass flow rate of 16.09 ton/s and area averaged temperatures at core inlet 402.7 C and core outlet 485.3 C, which gives an average temperature rise in the core
82.6 C. The predicted pressure drop over the core was 0.85 bars and 0.29 bars in the SG. Total pressure drop
of 1.6 bars was calculated for the whole primary system. Coolant temperature distribution in the vertical
cross-section plane is shown in Figure 3.14. Analysis of the velocity vector field presented in Figure 3.15
suggests that there is a region with relatively high downward speed of the coolant flow near the vessel wall in
the downcomer. As we will show in the next section such flow is sufficient for dragging steam bubbles from
the SG to the core region.
Figure 3.14: Temperature field during normal operation.
(a) 3D velocity representation
33
(b) Velocity on 2D vertical cross-section
Figure 3.15: Velocity profiles in the steady state flow field
3.3.3
Modeling of bubble transport
To study the possibility of steam bubble transport to the core region we use Lagrangian model that resolves
individual trajectories of the bubbles in turbulent flow field of the primary coolant system. Bubble is assumed
to have constant density. The following phase specific models were used in simulation to characterize steam
bubble behavior in liquid lead:
Constant density
Drag force
Tomiyama drag coefficient (implemented as a User-Defined Field Function)
Gas
H2 O
Material particles
Residence time
Spherical particles
Track file
Turbulent dispersion
Virtual mass
Rebound conditions were implemented on the solid walls of the primary system and escape conditions
were used on the free coolant surfaces. Initial velocity of a bubble was set to 0 m/s and flow rate of a bubble
from one seed point was selected as one bubble per second. Behavior of the bubble with diameters of 0.2
mm, 0.4 mm, 0.5 mm, 1.0 mm, 2.0 mm, 4.0 mm and 6.0 mm was studied. To investigate what is the effect
of different bubble drag coefficients, both, Tomiyama and Schiller-Naumann correlations were used in the
analysis.
34
Injectors
In Star-CCM+, injectors are used to define the location, direction, rate at which the bubbles enter the fluid
continuum and the velocity of an entering bubble. Also the flow rate distribution type (rate per injection
point or the total rate), flow rate specification type (particle, mass or volume flow rate) and particle size
specification type (diameter or mass) are defined by injector. Particle diameter, in turn, can be defined
as: a) constant; b) by a specific size distribution (Log-Normal, Rosin-Rammler) or c) by User-Defined Field
Function. In the main core of the analysis particle diameters are set to a constant value and changed manually
between discrete values that are of interest in SGTL. In one case we used an UDF according to Eq. 3.12 to
see how much the results are affected by the change of particle diameter due to changing hydrostatic pressure
while traveling in the domain.
s
dB (Pdepth ) =
P0
dB0
Pdepth
(3.12)
where P0 is atmospheric pressure, Pdepth is the hydrostatic pressure at the depth of a bubble and dB0 is the
bubble diameter at atmospheric conditions.
Figure 3.16: Locations of the planes used as bubble injectors in SG. (Color bar legend shows the vertical upward
velocity of the lead at these planes).
The SG leakage was simulated by injecting bubbles at three different levels (see Figure 3.16) in the SG 5.2
m, 6.87 m and 8.55 m from the bottom of the pool (the total depth of the pool is 8.65 m). All these injectors
are by type spherical planes belonging to SG region only. These particular heights are selected to simulate
the SG tube leakage happening in the bottom, center and upper part of the SG. Each plane contains all mesh
cells at this height inside SG, whereas each of those cells can be an injection point.
There were also two other injection locations used according to the requirements set by the method we used
for estimation of the void accumulation rates (see Section 3.3.4). One of them is situated close to the exit of
the core and the other in the vicinity of the pump outlet. They are shown in Figure 3.17.
35
Figure 3.17: Locations of the planes used as bubble injectors in core and in pump. (Color bar legend shows the
vertical upward velocity of the lead at these planes).
A point inclusion probability (probability of a point being included in a set of points from which bubbles
are injected) is used in Star-CCM+ injectors to control the number of bubble seeds. The point inclusion
probability is equal to ratio of the number of seeds to the total number of cells in the seed plane. In the
SG region we used several point inclusion probabilities from 0.005 to 1.0 that correspond to the numbers of
particle seeds from 24 to 4142. The results of convergence study presented in Figure 3.18 show that changes
in the statistical results are insignificant if number of seeds is larger than 2500. In this thesis bubbles are
injected only at different axial leakage positions to estimate the probability that a bubble can reach the core.
Radially are leak points uniformly distributed over the selected plane of injection.
0.28
0.27
Emergence rate [*100%]
0.26
0.25
0.24
0.23
0.22
0.21
0.2
0.19
500
1000
1500
2000
2500
3000
3500
4000
4500
Number of seeds
Figure 3.18: Emergence rate as a function of number of seeds. The span of respective Point Inclusion Probabilities
is from 0.005 to 1. Injection plane at 5.2 m was used.
36
3.3.4
Strategy for estimation of core voiding
In general, the quantification of the probability that a bubble of certain size can reach the core region is
calculated by dividing the number of bubbles that reached the core by the total number of bubbles released
in a particular location (region of injection points) in SG. The bubbles are injected at three height levels
inside the SG.
The following three-step method is used in assessment of void accumulation rate in the core. The bubbles
injected through the crack in SGT into the primary coolant flow have generally three possibilities (depending
on their size, leak location and flow characteristics) to escape the system (see Figure 3.19):
1. They escape after injection immediately to the cold free surface above the downcomer and the SG.
2. They are dragged to the core for the first time and escape from the loop to the hot free surface above
the core.
3. They stay in the primary coolant flow until they travel through the pump and reach the SG again and
emerge to the cold free surface this time.
All the bubbles which do not escape the system at any of those steps, are considered to be carried continuously
by the coolant flow through the primary loop. Having obtained valid probabilities to reach the core for
different size bubbles and different injection locations, is the starting point for assessment of the SGTL
consequences according to different scenarios of void behavior (see the first section in this chapter for scenario
definitions).
Figure 3.19: Different locations where the probabilities for a bubble to escape the primary loop are estimated (1-3).
P1-P3 are defined at these locations as the probabilities for a bubble to stay in the loop.
Here is a more detailed explanation on how to calculate the probabilities related to aforementioned ways of
escape with respect to different injection locations.
First, all bubbles are injected at one of the three injection levels in the SG. The injection rate per all injection
point in total has set to a constant value of one bubble per second. We can estimate the probabilities that
37
a bubble will stay in the loop or will escape from the primary system by counting the number of bubbles
that reach the lower plenum and the number of those that escape through the cold free surface during the
given residence time and relating them to the total number of bubbles injected during the same time. E.g.
if 50% of the bubbles reach the lower plenum and 50% reach the cold free surface in the downcomer region,
then the measured particle flow rate at both of them is 0.5 1/s. Now, after the liquid lead steady state
flow field has been calculated, the solvers for flow will be switched off by setting the number of cycles of
algebraic multi-grid solver to 0. Only the trajectories of Lagrangian particles (with necessary parameters
for them) are calculated. This enables us to catch and count the bubbles that are entering the core (and
the reflector/dummy) region by changing the interface type at the bottom boundaries of lower plenum and
reflector regions to baffle without affecting the flow field. Baffle is the only interface type on which the
boundary condition for Lagrangian phase can be set to escape. This saves calculation time and more
importantly, allowing to count the particles entering the core only once (they do not influence the results by
going through the whole primary system and re-enter the core again) while keeping the maximum residence
time of particles sufficiently high, so the maximum number of bubbles will either escape to the core or emerge
to the free surface. Bubbles that entered the core and reflector regions are counted at the inlet boundaries of
these regions. The fraction of those to total number of bubbles is P1 measure of probability that bubbles
injected in SG can reach the core.
Now we proceed to estimation of the second probability, P2 . This step is in principal done only for the bubbles
that have already traveled through the core. The bubbles are injected at the horizontal plane slightly below
the outlet of the core. The bubbles that will escape from the simulation domain to the hot free surface (the
liquid lead surface above the core region) are counted. P2 is one minus the fraction of those bubbles to the
total bubbles injected at the top of the core, namely the ones that are moving with the coolant flow back
towards the steam generator.
Finally, to assess the fraction of bubbles that will stay in the primary loop even after the third stage, having
just passed through the pump, the bubbles are injected close to the exit of the pump and those that reach
the core inlet are counted. This is defined as probability P3 .
Having these three probabilities, one can estimate the probability that a bubble can reach the core in the
first place and also make an assessment about the the total probability for a bubble to stay in the primary
loop and therefore contribute to the accumulation of the void in the core. The accumulation rate depends
on three parameters: 1) steam injection flow rate; 2) actual bubble size distribution of injected bubbles; 3)
probability that bubbles of each size can stay in the primary loop. This can be expressed as follows:
Qprimary (t) = Qleak (t)
N
X
fd i P1d i P2d i P3d i
(3.13)
i=1
where Qprimary(t) is the rate of accumulation of void in the primary system, Qleak (t) is the volumetric leak
rate [litres/day], fd i , P1d i , P2d i , P3d i are volume fraction and respective probabilities for a bubble with size
di , and N is the number of bubbles sizes considered in the bubble size distribution.
It is also interesting to estimate the void accumulation rate in the core (assuming that all bubbles that reach
the core will stay in the core, e.g. below spacers):
Qcore (t) = Qleak (t)
N
X
fd i P1d i
(3.14)
i=1
To estimate the realistic rate of void accumulation in the core one needs to consider the probability that
bubble will pass or stuck in the core given detailed geometry of the core internals. This task is beyond the
scope of the present work.
38
3.3.5
Results and discussion
Injection level 5.2 m

Trajectories of injected bubbles are shown below (Figures 3.20, 3.21, 3.22, 3.22, 3.23, 3.24, 3.25 and 3.26). It
can be seen from these figures that small (0.2 mm and 0.4 mm in diameter) bubbles are dragged to the core,
medium size (0.5 mm and 1.0 mm) bubbles escape the primary coolant system after certain residence time
and bigger bubbles (2.0 mm, 4.0 mm and 6.0 mm) again are more affected by the coolant flow and dragged
to the core.
Figure 3.20: Injection height 5.2 m. Bubble diameter 0.2 mm.
39
40
The probability that a bubble can reach the core when injected at the 5.2 m injection level as a function of
bubble size is presented in Figure 3.27. One can see that using the drag correlation from Tomiyama (blue
curves in Figure 3.27), there is non-zero probability for bubbles with diameter less than 0.5 mm and bigger
than 1.0 mm to reach the core if the leaking tube is located at the lower part of SG.
These plots also confirm that Schiller-Naumann drag correlation (red curves in Figure 3.27) can be used only
for very small bubbles. Results obtained with Schiller-Naumann correlation for bubbles larger than 0.2 mm
can be generally misleading and therefore this correlation was not used in the further analysis.
P1 probability to reach the core after injection
41
1
Tomiyama
SchillerNaumann
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Bubble diameter [mm]

Figure 3.27: Fraction of bubbles that reach the core inlet. Obtained with different drag correlations.
Figure 3.28 shows the probabilities that a bubble can reach the core inlet for simulations with constant
bubble diameter and simulations where bubble diameter is a function of surrounding hydrostatic pressure.
Two curves are exhibiting the same tendency, but are somewhat shifted. This shift can be explained by
two facts: a) in case of pressure dependent bubble diameter we have chosen the free surface of lead as the
reference altitude for diameter dB0 which means that the initial diameter of a bubble at location plane 5.2 m
from the bottom ( 3.5 m hydrostatic head above it) is already about 1.5 times smaller and b) when a bubble
moves downward, its diameter decreases and it starts to behave as a smaller bubble. This moves the curve
rightwards.
In this range of bubble diameters, according to Figure 3.28, the shifts are about half a millimeter. This can
be considered the within error of measurements of the bubble size distributions that we have used in this
analysis since there are factors affecting bubble sizes that we have not taken into consideration in detail and
which define the bubble sizes in real application (uncertainties in the crack morphology, pressure differences
in a particular accident scenario, bubblehot lead interaction). Therefore the qualitative picture can be
captured well enough even with simulations with constant bubble diameters.
1
constant dB
0.9
dB=dB(pressure)
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0

Figure 3.28: Fraction of bubbles that are dragged to core.
42
Figure 3.29 illustrates the probability P2 as a function of bubble diameter. One can see that there is a
significant fraction of bubbles which will escape to the hot free surface after passing through the core once
(more than 50% of bubbles with diameter around 0.5 mm is removed). Nevertheless, the plot suggests that
more than half of bubbles with diameter smaller than 0.5 mm and bigger than 1 mm will continue traveling
in the system.
P2 probability to reach the pump from core
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0

Figure 3.29: Fraction of the bubbles that are carried through the core and further into the pump and SG.
P3 probability to stay in the primary loop after one cycle
Third escape possibility, however, seems to decrease the amount of void in the system significantly. Figure 3.30
shows that except for very small (0.2 mm) bubbles, the probability to stay in the primary system after one
cycle through the core is very low.
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0

Figure 3.30: Fraction of bubbles that continue circulating in the primary loop.
According to the resulting probabilities that a bubble of certain size will stay in the primary loop we have
43
estimated the void accumulation rate in the primary system. Analysis is done for two distinct cases and is
based on the procedure described in Section 3.3.4:
1. Assuming the size distribution of injected bubbles found by Sasaka and Terasada (2011) [28] and the
minimum leak rate of about 0.005 L/min (7.2 L/day) (probable in the early stages after the SGTL).
And the resulting accumulation rate of steam bubbles in the primary system is:
Q(t)small = Qleak (t)small diSasaka&T erasaka P1d i P2d i P3d i = 0.012
L
day
2. Assuming the size distribution of injected bubbles found by Beznosov et al. (2005)[29] and the leak
rate of about 0.25 L/min (360 L/day) (probable do develop upon some time after SGTL). And the
resulting accumulation rate of steam bubbles in the primary system is:
Q(t)big = Qleak (t)big diBeznosov P1d i P2d i P3d i = 0.85
L
day
Since there is uncertainty in the scenarios of how the void behaves in the primary system, we have analyzed
also a situation when bubbles are staying in the core (without passing it) due to getting stuck in vortexes
e.g. This means that we remove the probabilities which describe the bubble behavior after passing through
the core, viz. P2d i and P3d i , from the accumulation rate expression. The void build-up values in the core if
the bubbles would not leave it, are:
1. In case of smaller leakage:
Q(t)small = Qleak (t)small diSasaka&T erasaka P1d i = 0.09
L
day
2. In case of bigger leakage:

Q(t)big = Qleak (t)big diBeznosov P1d i = 96.44
L
day

Trajectories of injected bubbles are shown below (Figures 3.31, 3.32, 3.33, 3.33, 3.34, 3.35, 3.36 and 3.37).
44
45
46
If the leak is located in the middle part of SG only very small (diameter less than 0.4 mm) bubbles can be
dragged down to the core inlet (see the results in Figure 3.38). At this height level, both drag correlations
used suggest the same tendency.
1
Tomiyama
SchillerNaumann
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0

Figure 3.38: Fraction of the bubbles that reach the core inlet.
Here we used probabilities P2 and P3 that were presented in previous section. The void accumulation rates
in the system and in the core only, are:
1. In case of smaller leakage in the whole primary system:
Q(t)small = Qleak (t)small diSasaka&T erasaka P1d i P2d i P3d i = 0.01
L
day
and in case of bigger leakage:

Q(t)big = Qleak (t)big diBeznosov P1d i P2d i P3d i = 0.02
L
day
2. In case of smaller leakage in the core:

Q(t)small = Qleak (t)small diSasaka&T erasaka P1d i = 0.02
L
day
and in case of bigger leakage:

Q(t)big = Qleak (t)big diBeznosov P1d i = 0.03
L
day
It is instructive to mention that in each case the biggest contribution to the void comes from the smallest,
0.2 mm, diameter size bubbles (as it can be seen from trajectory figures and probability plots as well).
Suggestions for mitigation of consequences of SGTL
47

All bubbles injected at the topmost level (8.55 m) close to the cold free surface escape from the lead-pool.
Therefore already P1 is zero and no threat is expected. We can already say that the higher location the SG
tube leakage happens the smaller is the probability that a bubble can be dragged to the core. The trajectories
of 0.2 mm and 1.0 mm bubbles that are injected at this height are presented in Figure 3.39 in Figure 3.40,
respectively.
Figure 3.39: Injection height 8.55 m. Bubble diameter 0.2 mm. Point inclusion probability is 1.
Figure 3.40: Injection height 8.55 m. Bubble diameter 1.0 mm. Point inclusion probability is 0.01.
3.4
Suggestions for mitigation of consequences of SGTL
Significant fraction of the bubbles that can reach the core inlet in normal operation conditions suggest that
SGTL issue has to be considered in the design with adequate preventive and mitigative measures. First of
all, in case of SGTL (which seems to be likely), it is important to concentrate on the scenarios that might
pose the tendency to cause threat to the reactor. This means, to the scenarios in which the void actually
reaches the core.
Therefore, one should definitely look into the design in order to have the flow path which prevents the
appearance of so called local stagnation zones (corners) where bubbles can get stuck, accumulate and move
as bigger amount of void at a time into the core region. Also the flow path should be designed to let most
of the bubbles escape through the free surface of lead coolant.
48
3.5
Conclusions
In this paper we consider potential consequences of a steam generator tube leakage in a pool type lead-cooled
fast reactor design. Primary coolant flow in nominal operation conditions and steam bubble transport in the
primary coolant loop is simulated with CFD code Star-CCM+. Bubbles are injected at different locations
inside the steam generator volume. Tomiyamas drag model is selected for simulation of steam bubbles motion
in lead. Fraction of the bubbles that can reach the core inlet has been estimated. We found that bubbles
with diameters between 0.5 mm and 1 mm, leaking from the bottom part of the SG, are likely to escape
from the primary system before they reach the core inlet. The other bubbles can contribute to voiding of
primary system. If the leakage is located in the middle of the SG, only bubbles with diameter less than 0.4
mm can reach the core. There is a non-zero void accumulation rate in the primary system, especially due
to small bubbles. Bubbles appearing in the top part of the SG will escape from the primary loop regardless
of their diameter. Our finding about considerable fraction of the bubbles that can reach the core inlet in
normal operation conditions as a result of SGTL infers that adequate preventive and mitigative measures
are necessary in the design of pool-type LFR systems with positive void reactivity coefficient.
CHAPTER
4
SUMMARY
First, an approach to the validation of coupled codes is described. The approach is implemented in the
design of the TALL-3D experimental facility, expressly conceived for the validation of coupled CFD/STH
codes. The facility consists of a liquid lead-bismuth thermal-hydraulic loop operating in forced and natural
circulation regimes with a heated pool-type 3D test section. A parametric analysis was performed with the
CFD code Star-CCM+ to obtain the desired natural and forced circulation steady states with stratification
development and full mixing, respectively. Results of steady state calculations in support of the TALL-3D
design development showed the expected behavior of the tank flow profile (mixing, stratification, buoyant jet
penetration depth etc). Transient analysis illustrate the deviation of tank outlet temperature calculated by
CFD from the analytical 0D tank outlet temperature. In order to have more pronounced two-way feedback
effects, we would like to get a bigger difference in these temperatures than calculated so far. There is also a
test matrix provided for the experimental program that aims at providing systematic data for validation of
different coupling approaches and codes and captures important system (1D) and local (3D) phenomena in
separate effect and coupled behavior.
Second part describes a CFD analysis of the steam bubble transport to the core of ELSY in case of a small
leakage in the SG region. The main objective here is to quantify the likelihood that a steam bubble is
transported in the primary coolant system to the core in case of SG tube leakage (if at all). Important
uncertainties in the accident scenario (size and morphology of the crack) and in the modeling (bubble size
distribution, used turbulence and drag correlation models etc) are identified and addressed.
For the analysis of steam bubble transport, the theory of bubble motion in liquids is very important. Different
regimes, depending on the size of the bubbles, are discussed with respect to behavior of their terminal
rise velocity. Predicted terminal rise velocity is dependent on the selection of the proper drag coefficient
correlation. For simple bubble motion a validation test section with stagnant liquid lead was used to simulate
bubble upward movement in it. Accordingly chosen drag coefficient correlation is used in modeling bubble
transport in ELSY reactor primary system.
Finally, a method for estimation of the probability that a steam bubble can reach the core and also stay in
the primary system, is proposed. These probabilities, together with injected bubble size distributions and
leak rates, enabled us to estimate the accumulation rates of the void in the system. The main outcome of
this study is that there is a real threat that steam bubbles can be dragged to the core in significant amounts.
The extent to which this happens is dependent on the bubble size distribution, leak flow rate and injection
location. The results presented in this paper show that there may be steam void accumulation in the primary
system with a rate up to 0.85 L/day and if bubbles should get stuck in the core, then the accumulation rate
there may be even up to 96.4 L/day.
49
50
Summary
CHAPTER
5
OUTLOOK
Since the TALL-3D test section design development for validation purposes is still still in progress, there are
subjects for the analysis in the future. Here are some ideas presented on how to reach the final goals set in
this project:
In order to increase the feedbacks between 3D test section and loop behavior, it is necessary to have the
difference between the 3D section outlet temperature predicted with CFD and the outlet temperature
predicted with 0D or 1D models, in the range that is comparable to the temperature differences between
the cold and the hot leg of the TALL loop in natural circulation conditions. This will increase the 3D
flow effects on the natural circulation flow. In order to achieve that, the main core simulator heater
should be rather low, so there is no need to use powerful secondary side to remove the heat, because
heat transfer in the secondary side would dampen the 3D effects as well.
3D effects can be emphasized by changing the test section design. We have adjusted the inlet diameter
(changes velocity) and the pool volume until now. We also added a circular plate that diverts the flow
and enhances stratification (therefore 3D effects). So far we have kept the test section design axissymmetrical, which simplifies CFD calculations, but in the future one could simulate non-symmetrical
design of the tank so that flow path would change according to the velocity, for example. So, there
would be different outlet temperatures and time constant of the tank at different circulation conditions.
Possible mock-up of the 3D test section based on water could be built where it is possible to analyze the
thermal stratification and mixing behavior depending on the flow conditions. In this case the scaling
analysis must be done in order to represent the liquid lead flow with water.
The ideas for future work regarding the ELSY project are following:
Apart from voiding issue, water causes the oxidation of lead depending on the oxygen content in the
coolant. As a results, solid slags of PbO can form and deteriorate the thermal-hydraulic performance
of the circuit or cause even flow blockage. Possible water dissolution into hydrogen and oxygen and
consecutive oxygen dissolving into lead can change the bubble size which affects the drag and bubble
transport. The magnitude of this phenomena in ELSY conditions should be assessed.
Mesh convergence study for ELSY design should be done.
Use of bubble size distribution in the injectors should be studied to quantify its effects.
To build a facility for bubble terminal rise velocity analysis in liquid lead or LBE. This could be done
jointly with planned modifications to TALL facility. To decrease the epistemic uncertainty in bubble
size distribution, a specific experiment could be developed with parameters used in ELSY.
The analysis of the possible consequences depending on different void behavior scenarios in the core
should be performed. How the amounts of void presented in this paper would affect the criticality and
power, essentially, would be informative with respect to further studies.
51
52
Outlook
It would be also possible to use the knowledge gained by doing this thesis in other future projects dealing
with heavy liquid metal cooled reactor systems and experimental facilities. For example the XT-ADS system
MYRRHA, which is selected as on of the priorities in the research in Europe now, can be the subject of
similar analysis as made here since the steam generator is also located in the primary lead-bismuth pool of
the reactor.
BIBLIOGRAPHY
[1] W. L. Oberkampf, M. Pilch, and T. G. Trucano, Predictive capability maturity model for computational
modeling and simulation, tech. rep., Sandia National Laboratories, Albuquerque, New Mexico, 2007.
[2] B. W. Spencer, The rush to heavy liquid metal reactor coolant - Gimmick or reasoned, Proceedings of
8th International Conference on Nuclear Engineering (ICONE-8), April 2000.
[3] S. S. Hwang, H. P. Kim, J. S. Kim, K. E. Kasza, J. Park, and W. J. Shack, Leak behavior of SCC
degraded steam generator tubings of nuclear power plant, Nuclear Engineering and Design, vol. 235,
pp. 2477 2484, 2005.
[4] A. Ciampichetti, D. Bernardi, and N. Forgione, Analysis of the LBE-water interaction in the LIFUS 5
facility to support the investigation of an SGTR event in LFRs, tech. rep., ENEA, September 2010.
[5] R. Fuge and H. Langenbrunner, The detection of small water leaks in a LMFBR steam generator using
magnetic flowmeter bubble noise, Nuclear Engineering and Design, vol. 54, pp. 301 307, 1979.
[6] K. Hayashi, Y. Shinohara, and K. Watanabe, Acoustic detection of in-sodium water leaks using twice
squaring method, Annals of Nuclear Energy, vol. 23, no. 15, pp. 1249 1259, 1996.
[7] P. Kudinov and T. N. Dinh, Evaluation approach and case set-up for simulation of consequences of the
SGTR event, tech. rep., Royal Institute of Technology (KTH), February 2010.
[8] U. S. DOE and GIF, A technology roadmap for Generation IV nuclear energy systems, GIF-002-00,
December 2002.
[9] EPRI, PWR primary-to-secondary leak guidelines Revision 2, Tech. Rep. TR-104788-R2, EPRI, Palo
Alto, CA, 2000.
[10] D. T. J. Hurle, Temperature oscillation in molten metals and their relationship to growth stiae in
melt-grown crystals, Philosophical Magazine, vol. 13, pp. 305 310, 1966.
[11] A. Gill, A theory of thermal oscillations in liquid metalss, Journal of Fluid Mechanics, vol. 64, pp. 577
588, 1974.
[12] L. Redjem-Saad, M. Ould-Rouiss, and G. Lauriat, Direct numerical simulation of turbulent heat transfer
in pipe flows: Effect of prandtl number, International Journal of Heat and Fluid Flow, vol. 28, no. 5,
pp. 847 861, 2007.
[13] P. F. Peterson, V. E. Schrock, and R. Greif, Scaling for integral simulation of mixing in large, stratified volumes, 6th International Topical Meeting on Nuclear Reactor Thermalhydraulics (NURETH-6),
vol. 1, pp. 202211, 1993.
[14] P. F. Peterson, Scaling and analysis of mixing in large stratified volumes, International Journal of
Heat and Mass Transfer, vol. 37, no. 1, pp. 97 106, 1994.
[15] C. J. Chen and W. Rodi, Vertical turbulent buoyant jets: A review of experimental data, HMT Science and Applications of Heat and Mass Transfer, vol. 4, p. 94, 1980.
53
54
Bibliography
[16] E. J. List, Turbulent jets and plumes, Annual Review of Fluid Mechanics, vol. 14, pp. 189 212, 1982.
[17] F. Cadinu, M. Jeltsov, W. Villanueva, K. Koop, A. Karbojian, and P. Kudinov, Program of work for
experimental tasks, software development and validation tasks on TALL, tech. rep., Royal Institute of
Technology (KTH), 2011.
[18] F. C. M. Jeltsov, W. Villanueva, K. Koop, A. Karbojian, and P. Kudinov, An approach to validation
of coupled CFD and System Thermal-Tydraulics codes, 14th International Topical Meeting on Nuclear
Reactor Thermalhydraulics (NURETH-14), 2011.
[19] W. Ma, E. Bubelis, A. Karbojian, B. R. Sehgal, and P. Coddington, Transient experiments from the
thermal-hydraulic ADS lead bismuth loop (TALL) and comparative TRAC/AAA analysis, Nuclear
Engineering and Design, vol. 236, 2006.
[20] Star-CCM+ User Guide version 5.06.
[21] W. Xu, Q. Chen, and F. T. M. Nieuwstadt, A new turbulence model for near-wall natural convection,
International Journal of Heat and Mass Transfer, vol. 41, 1998.
[22] W. L. Oberkampf, S. M. DeLand, B. M. Rutherford, K. V. Diegert, and K. F. Alvin, Error and
uncertainty in modeling and simulation, Reliability Engineering & System Safety, vol. 75, no. 3, pp. 333
357, 2002.
[23] T. G. Theofanus, On the proper formulation of safety goals and assessment of safety margins for rare
and high-consequence hazards, Reliability Engineering and System Safety, vol. 54, pp. 243 257, 1996.
[24] M. Pourgol-Mohamad, A. Mosleh, and M. Modarres, Structured treatment of model uncertainty in
complex thermal-hydraulics codes: Technical challenges, prospective and characterization, Nuclear
Engineering and Design, vol. 241, pp. 285 295, 2011.
[25] I. Leibson, E. G. Holcomb, A. G. Cacoso, and J. J. Jacmic, Rate of flow and mechanics of bubble
formation from single submerged orifices, AIChE Journal, vol. 2, p. 296, 1956.
[26] A. Marmur and E. Rubin, Equilibrium shapes and quasi-static formation of bubbles at submerged
orifice, Chemical engineering Science, vol. 28, p. 1455, 1973.
[27] A. A. Kulkarni and J. B. Joshi, Bubble formation and bubble rise velocity in gas-liquid systems: A
review, Industrial & Engineering Chemistry Research, vol. 44, no. 16, pp. 58735931, 2005.
[28] K. Terasaka, Y. Sasada, D. Kobayashi, and S. Fujioka, Submilli-bubble dispersion from a slit orifice
into water, Journal of Chemical Engineering, vol. 44, no. 3, pp. 140 145, 2011.
[29] A. V. Beznosov, S. S. Pinaev, D. V. Davydov, A. A. Molodtsov, T. A. Bokova, P. N. Martynov, and V. I.
Rachkov, Experimental studies of the characteristics of contact heat exchange between lead coolant
and the working body, Atomic Energy, vol. 98, no. 3, 2005.
[30] B. Flesch and B. Cochet, Leak-Before-Break in steam generator tubes, International Journal of Pressure Vessels and Piping, vol. 43, pp. 165 179, 1990.
[31] S. S. Hwang, H. P. Kim, and J. S. Kim, Evaluation of the burst characteristics for axial notches on SG
tubings, Nuclear Engineering and Design, vol. 232, pp. 139 143, 2004.
[32] S. S. Hwang, C. Namgung, M. K. Jung, H. P. Kim, and J. Kim, Rupture pressure of wear degraded
alloy 600 steam generators tubings, Journal of Nuclear Materials, vol. 373, pp. 71 74, 2008.
[33] G. Yang, B. Du, and L. Fan, Bubble formation and dynamics in gas-liquid-solid fluidization - A review,
Chemical Engineering Science, vol. 62, no. 1-2, pp. 2 27, 2007. Fluidized Bed Applications.
[34] R. Clift, J. R. Grace, and M. E. Weber, Bubbles, drops and particles. Academic Press, Inc., 1978.
[35] C. C. Maneri and P. F. Vassallo, Dynamics of bubbles rising in finite and infinite media, Tech. Rep.
LM-00K048, Lockheed Martin Corporation, Schenectady, NY 12301 (US), 2000.
[36] OECD and NEA, Handbook on lead-bismuth eutectic alloy and lead properties, materials compatibility,
thermal-hydraulics and technologies, 2007.
[37] CD-Adapco, Star-CCM+ User Guide version 5.02.
Bibliography
55
[38] D. G. Karamanev and L. N. Nikolov, Free rising bubbles do not obey Newtons law for free settling,
AIChE Journal, vol. 38, pp. 1993 1997, 1992.
[39] S. L. Sullivan and B. W. Hardy, Formation of air bubbles at orifices submerged below liquid, AIChE
Journal, vol. 10, p. 848, 1964.
[40] D. Rodrigue, A simple correlation for gas bubbles rising in power-law fluids, Canadian Journal of
Chemical Engineering, vol. 80, no. 2, p. 289, 2002.
[41] A. Tomiyama, I. Kataoka, I. Zun, and T. Sakaguchi, Drag coefficients of single bubbles under normal
and microgravity conditions, JSME International Journal Series D, vol. 41, no. 2, p. 472, 1998.
[42] H. D. Mendelson, The prediction of bubble terminal velocities from wave theory, AIChE Journal,
vol. 13, p. 250, 1967.
[43] Y. Mori, K. Hijikata, and K. Kuriyama, Experimental study of bubble motion in mercury with and
without a magnetic field, Journal of Heat Transfer (Trans. ASME), vol. 99, pp. 404410, 1977.
[44] A. Alemberti, J. Carlsson, E. Malambu, A. Orden, L. Cinotti, D. Struwe, P. Agostini, and S. Monti,
ELSY European LFR activities, Journal of Nuclear Science and Technology, vol. 48, no. 4, pp. 479
482, 2011.
[45] L. Cinotti, Reactor assembly preliminary configuration, Tech. Rep. ELSY DOC 08 049, Del Fungo
Giera Energia S.p.A., 2008.
[46] M. B
ottcher, CFD analysis of decay heat removal scenarios of the lead cooled ELSY reactor, Benchmarking of CFD Codes for Application to Nuclear Reactor Safety (CFD4NRS-3), September 14-16 2010.
[47] A. Onea, M. B
ottcher, and D. Struwe, Lead pressure loss in the heat exchanger of the ELSY fast leadcooled reactor by CFD approach, Benchmarking of CFD Codes for Application to Nuclear Reactor
Safety (CFD4NRS-3), September 14-16 2010.
[48] J. Carlsson, Steam generator tube leakage calculations on the ELSY design, tech. rep., EC JRC,
Institute for Energy, 2009.

Full Text 01

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Full Text 01

Uploaded by

Copyright:

Available Formats

Master Thesis

Application of CFD to Safety and Thermal-Hydraulic Analysis of

Division of Nuclear Power Safety

LIST OF PAPERS AND PUBLICATIONS

II. M. Jeltsov, F. Cadinu, W. Villanueva, A. Karbojian, K. Koop and P. Kudinov, An approach to

1 Introduction and background

Validation of coupled CFD and STH codes . . . . . . . . . . . . . . . . . . . . . . . .

Steam generator tube leakage in a pool-type LFR design . . . . . . . . . . . . . . . . .

Turbulent heat transfer in non-unity Prandtl number fluids . . . . . . . . . . . . . . .

Stratification and mixing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Goals and tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 Development of the TALL-3D test section design

TALL-3D experimental facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Specific requirements and description of the TALL-3D design . . . . . . . . . . . . . .

Calculations in support of the design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Case I: Forced circulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Case II: Natural circulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Identification of key physical phenomena and development of validation test matrix . . . . . .

Discussion of the scenarios and uncertainties . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Bubbles size distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Selection of the bubble drag coefficient correlation in liquid lead . . . . . . . . . . . . . . . . .

Bubble shape and rise behavior regimes . . . . . . . . . . . . . . . . . . . . . . . . . .

Modeling bubble motion in a column of liquid lead . . . . . . . . . . . . . . . . . . . .

Drag coefficient correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

An approach to verification of modeling method . . . . . . . . . . . . . . . . . . . . .

Bubble transport to the ELSY core in case of SGTL . . . . . . . . . . . . . . . . . . . . . . .

ELSY reactor design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Modeling of primary system at nominal operational conditions . . . . . . . . . . . . .

Modeling of bubble transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Strategy for estimation of core voiding . . . . . . . . . . . . . . . . . . . . . . . . . . .

Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Suggestions for mitigation of consequences of SGTL . . . . . . . . . . . . . . . . . . . . . . .

Different steady states and possible transients between them. . . . . . . . . . . . . . . . . . .

Three possible bubble behaviors in the core. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Vapor bubble size distribution. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Column geometry used for bubble model analysis. . . . . . . . . . . . . . . . . . . . . . . . .

3.10 ELSY reactor reference configuration [44]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.13 3D volume mesh and a vertical cross-section of it. . . . . . . . . . . . . . . . . . . . . . . . . .

3.14 Temperature field during normal operation. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.15 Velocity profiles in the steady state flow field . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.21 Injection height 5.2 m. Bubble diameter 0.4 mm. . . . . . . . . . . . . . . . . . . . . . . . . .

3.22 Injection height 5.2 m. Bubble diameter 0.5 mm. . . . . . . . . . . . . . . . . . . . . . . . . .

3.23 Injection height 5.2 m. Bubble diameter 1.0 mm. . . . . . . . . . . . . . . . . . . . . . . . . .

3.24 Injection height 5.2 m. Bubble diameter 2.0 mm. . . . . . . . . . . . . . . . . . . . . . . . . .

3.25 Injection height 5.2 m. Bubble diameter 4.0 mm. . . . . . . . . . . . . . . . . . . . . . . . . .

3.26 Injection height 5.2 m. Bubble diameter 6.0 mm. . . . . . . . . . . . . . . . . . . . . . . . . .

3.28 Fraction of bubbles that are dragged to core. . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.30 Fraction of bubbles that continue circulating in the primary loop. . . . . . . . . . . . . . . . .

3.31 Injection height 6.87 m. Bubble diameter 0.2 mm. . . . . . . . . . . . . . . . . . . . . . . . .

3.32 Injection height 6.87 m. Bubble diameter 0.4 mm. . . . . . . . . . . . . . . . . . . . . . . . .

3.33 Injection height 6.87 m. Bubble diameter 0.5 mm. . . . . . . . . . . . . . . . . . . . . . . . .

3.34 Injection height 6.87 m. Bubble diameter 1.0 mm. . . . . . . . . . . . . . . . . . . . . . . . .

3.35 Injection height 6.87 m. Bubble diameter 2.0 mm. . . . . . . . . . . . . . . . . . . . . . . . .

3.36 Injection height 6.87 m. Bubble diameter 4.0 mm. . . . . . . . . . . . . . . . . . . . . . . . .

3.37 Injection height 6.87 m. Bubble diameter 6.0 mm. . . . . . . . . . . . . . . . . . . . . . . . .

3.38 Fraction of the bubbles that reach the core inlet. . . . . . . . . . . . . . . . . . . . . . . . . .

Validation test matrix. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Liquid lead properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Initial tentative parameters of ELSY plant [45], [46], [47]. . . . . . . . . . . . . . . . . . . . .

Defined source terms per region. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Steam generator tube leakage in a pool-type LFR design

Introduction and background

Stratification and mixing

Goals and tasks