You are on page 1of 64

Lappeenranta University of Technology

Faculty of Technology
Department of Energy and Environmental Technology
Large Break Blowdown Test Facility Study
The subject of Masters thesis has been accepted on 27th of February 2008.
Supervisor: Riitta Kyrki-Rajamaki
Instructor: Heikki Purhonen
Lappeenranta, 26.3.2008
Arto Ylonen
Ylasatamakatu 31 B 13
80100 Joensuu
Finland
Tel. +358 50 354 4346
TIIVISTELM

A
Lappeenrannan teknillinen yliopisto
Teknillinen tiedekunta
Energia- ja ymparistotekniikan osasto
Arto Ylonen
Ison LOCA:n ulospuhallusvaiheen koelaitteistojen kartoitus
Diplomityo
2008
60 sivua, 33 kuvaa, 4 taulukkoa ja 1 liite
Tarkastajat: Riitta Kyrki-Rajamaki
Heikki Purhonen
Hakusanat: LBLOCA, ulospuhallusvaihe, koelaitteistot
Keywords: LBLOCA, blowdown, test facilities
Tyon tarkoituksena on kerata yhteen tiedot kaikista maailmalta loytyvista ison LOCA:n
ulospuhallusvaiheen tutkimiseen kaytetyista koelaitteistoista. Tyon tarkoituksena on
myos antaa pohjaa paatokselle, onko tarpeellista rakentaa uusi koelaitteisto neste-
rakenne-vuorovaikutuskoodien laskennan validoimista varten. Ennen varsinaisen koe-
laitteiston rakentamista olisi tarkoituksenmukaista myos rakentaa pienempi pilottikoe-
laitteisto, jolla voitaisiin testata kaytettavia mittausmenetelmia. Sopivaa mittausdataa
tarvitaan uusien CFD-koodien ja rakenneanalyysikoodien kytketyn laskennan validoi-
misessa. Naita koodeja voidaan kayttaa esimerkiksi arvioitaessa reaktorin sisaosien ra-
kenteellista kestavyytta ison LOCA:n ulospuhallusvaiheen aikana. Raportti keskittyy
maailmalta loytyviin koelaitteistoihin, uuden koelaitteiston suunnitteluperusteisiin se-
ka aiheeseen liittyviin yleisiin asioihin. Raportti ei korvaa olemassa olevia validointi-
matriiseja, mutta sita voi kayttaa apuna etsittaessa validointitarkoituksiin sopivaa ison
LOCA:n ulospuhallusvaiheen koelaitteistoa.
ABSTRACT
Lappeenranta University of Technology
Faculty of Technology
Department of Energy and Environmental Technology
Arto Ylonen
Large Break Blowdown Test Facility Study
Masters thesis
2008
60 pages, 33 gures, 4 tables and 1 appendix
Examiners: Riitta Kyrki-Rajamaki
Heikki Purhonen
Keywords: LBLOCA, blowdown, test facilities
The main goal of this work is to gather all suitable LBLOCA blowdown tests together
and compare them and the results obtained. By reviewing existing measurement data it
is easier to decide if a new test facility with a modern measurement system is needed for
structural analysis and computational uid dynamics (CFD) code validation purposes.
Measurement techniques and methods could be rst tested with a smaller pilot facility.
Suitable data is needed to facilitate the coupling of new CFD codes with structural
analysis codes (uid-structure interaction analysis). These codes could be used, for
example, to evaluate structural integrity of the reactor core and its supporting struc-
tures during a LOCA blowdown. The report concentrates on the existing experimental
knowledge of the blowdown, the design of the test facility and discusses some related
general issues. This work is not meant to replace any existing validation matrices, but
it can be used as an additional source of information when searching for a suitable test
facility for LBLOCA blowdown validation purposes.
Contents
Nomenclature 3
List of Acronyms 5
List of Figures 6
List of Tables 7
Acknowledgements 8
1 Introduction 9
2 Large Break Loss-of-coolant Accident 10
3 Phenomena during LBLOCA 13
3.1 Basic thermodynamic phenomena . . . . . . . . . . . . . . . . . . . . . 13
3.1.1 Evaporation due to depressurisation or heat input . . . . . . . . 13
3.1.2 Condensation due to pressurisation or heat removal . . . . . . . 13
3.1.3 Friction and pressure drop . . . . . . . . . . . . . . . . . . . . . 14
3.1.4 Pressure wave propagation . . . . . . . . . . . . . . . . . . . . . 15
3.2 Critical ow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2.1 Single-phase ow . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.2.2 Two-phase ow . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3 Phase separation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3.4 Entrainment and de-entrainment . . . . . . . . . . . . . . . . . . . . . 21
3.5 Stratied ow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.6 CCF/CCFL phenomena . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3.7 Heat transfer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.8 Pool formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.9 Quench front propagation . . . . . . . . . . . . . . . . . . . . . . . . . 24
4 Pipe break mechanics 26
4.1 Longitudinal break . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.2 Circumferential break . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
4.3 Pipe break simulations using Fluent CFD . . . . . . . . . . . . . . . . . 28
4.3.1 Eect of pipe length on depressurisation . . . . . . . . . . . . . 30
4.3.2 Eect of break opening time on depressurisation . . . . . . . . . 30
1
5 Large Break test facilities 34
5.1 Separate Eects Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
5.1.1 Edwards-OBrien pipe . . . . . . . . . . . . . . . . . . . . . . . 35
5.1.2 Bartaks pipe . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
5.1.3 CANON and Super CANON . . . . . . . . . . . . . . . . . . . . 36
5.1.4 Marviken . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
5.1.5 CSE blowdown experiments . . . . . . . . . . . . . . . . . . . . 38
5.1.6 Semiscale (Bettis Flask) . . . . . . . . . . . . . . . . . . . . . . 41
5.1.7 Piper . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5.1.8 Battelle-Frankfurt RS-16B . . . . . . . . . . . . . . . . . . . . . 42
5.1.9 Hei Dampf Reaktor (HDR) . . . . . . . . . . . . . . . . . . . . 43
5.1.10 Other tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
5.2 Integral Eects Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.2.1 LOFT and SEMISCALE . . . . . . . . . . . . . . . . . . . . . . 47
5.2.2 LOBI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
6 Test facility summary 50
7 Designing the new facility 52
7.1 Scaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
7.2 Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
7.3 Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
7.4 Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
8 Conclusions 56
Appendices
2
Nomenclature
T the degree of subcooling at inlet

M mass ow
A ow area
a sonic velocity
c specic heat
CRF carry-over fraction
D diameter
E fraction of liquid entrained as droplets
f friction factor
G mass ux, mass ow rate
g
c
gravitational conversion factor
h enthalpy
K bulk modulus
K friction factor
k slip ratio
K
l
loss coecient
L initial defect length
L length
m mass per unit area
N experimental parameter
n thermal equilibrium polytropic exponent
p pressure
Q volumetric ow
q

peak linear heat rate


s entropy
t time
V core ooding rate
V velocity
V volume
v specic volume
v velocity
w phase velocity
Z quench front level
3
Greek symbols
void fraction
quality
critical pressure ratio
isentropic expansion factor c
p
/c
v
density
stress
sector from which the liquid lm is removed
Subscripts
0 stagnation, upstream reservoir, initial
1 inlet
1 single-phase
3 branch
c critical
E entrainment
E equilibrium
e equivalent
f liquid
f water added to core
fg evaporation
g gas
i index for area change
l liquid
m mixture
p constant pressure
p fracture propagation
s constant entropy
s pipe material
sg steam generation
t throat
TP two-phase
v constant volume
y yield
4
List of Acronyms
CCF Counter Current Flow
CCFL Counter Current Flow Limitation
CFD Computational Fluid Dynamics
DBA Design Basis Accident
DEGB double-ended-guillotine break
DNB departure from nucleate boiling
ECCS Emergency Core Cooling System
FSI Fluid-structure interaction
HEM homogeneous equilibrium model
HFK Henry-Fauske model
HFM homogeneous frozen ow model
ISP International Standard Problem
LBLOCA Large Break Loss-of-coolant Accident
LWR light water reactor
NRC US Nuclear Regulatory Commission
OECD Organisation for Economic Co-operation and Development
RPV Reactor Pressure Vessel
5
List of Figures
2.1 Expansion wave propagation and growth of core barrel deformations [1] 11
2.2 Phenomena during end of blowdown phase [2] . . . . . . . . . . . . . . 12
3.1 Schematic of ashing transient phenomena during blowdown [3] . . . . 14
3.2 Critical Pressure Ratio Data as a function of Length/Diameter Ratio [4] 17
3.3 Dierences between introduced two-phase critical ow models . . . . . 18
3.4 Countercurrent ow of steam and liquid [3] . . . . . . . . . . . . . . . . 23
3.5 Pool formation in upper plenum during reood [3] . . . . . . . . . . . . 24
3.6 Quenching during reood [3] . . . . . . . . . . . . . . . . . . . . . . . . 25
4.1 Dierent rupture types considered in High Energy Line Breaks . . . . . 27
4.2 Predicted opening times of the break in three cases [5] . . . . . . . . . 27
4.3 Hoop stress in cylindrical shell . . . . . . . . . . . . . . . . . . . . . . . 28
4.4 DEGB locations, realistic break sizes and times [6] . . . . . . . . . . . . 29
4.5 Mesh/Geometry for 1.369 m pipe . . . . . . . . . . . . . . . . . . . . . 31
4.6 Arrival of the rarefaction wave to the vessel in Case 2 (0.5 ms) . . . . . 31
4.7 Pressure at outlet in Case 1 simulations . . . . . . . . . . . . . . . . . . 32
4.8 Eect of pipe length on depressurisation . . . . . . . . . . . . . . . . . 32
4.9 Pressure at outlet in Case 2 simulations . . . . . . . . . . . . . . . . . . 33
4.10 Eect of break opening time on depressurisation . . . . . . . . . . . . . 33
5.1 Schematic of Edwards-OBrien blowdown pipe [7] . . . . . . . . . . . . 35
5.2 Schematic of Bartaks horizontal experimental channel [8] . . . . . . . . 36
5.3 Schematic of CANON facility [9] . . . . . . . . . . . . . . . . . . . . . 37
5.4 Schematic of Super CANON facility [10] . . . . . . . . . . . . . . . . . 37
5.5 Schematic of Marviken critical ow test facility [11] . . . . . . . . . . . 39
5.6 Schematic of CSE reactor simulator [12] . . . . . . . . . . . . . . . . . 40
5.7 Schematic of the vessel used for SEMISCALE test 711 [13] . . . . . . . 41
5.8 Schematic of Piper vessel and blowdown nozzle [14] . . . . . . . . . . . 42
5.9 Schematic of RS-16B blowdown facility and models[15] . . . . . . . . . 43
5.10 Schematic of HDR pressure vessel and core barrel [16] . . . . . . . . . . 44
5.11 Instrumentation of the long and short discharge pipes in HDR [17] . . . 46
5.12 LOFT test facility [18] . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
5.13 LOFT reactor vessel instrumentation [18] . . . . . . . . . . . . . . . . . 48
5.14 Semiscale Mod-1 (left) and Semiscale Mod-2C (right) [18] . . . . . . . . 48
5.15 LOBI test facility [18] . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
6
List of Tables
4.1 FLUENT simulation settings . . . . . . . . . . . . . . . . . . . . . . . . 30
5.1 HDR preliminary and main phase test matrices . . . . . . . . . . . . . 45
6.1 Test facilities summarized . . . . . . . . . . . . . . . . . . . . . . . . . 51
7.1 Time-reducing scaling law [19] . . . . . . . . . . . . . . . . . . . . . . . 54
7
Acknowledgements
I would like to express my gratitude to professor Riitta Kyrki-Rajamaki and my su-
perior Heikki Purhonen for their support and guidance during the writing process of
this thesis. I would also like to thank my family and friends for their support through
the studies. Also big thanks belongs to all of my colleagues and especially to Markku
Puustinen and Vesa Tanskanen for their support and expertise.
Lappeenranta, 26.3.2008
Arto Ylonen
8
1 Introduction
Large Break Loss-of-coolant Accident (LBLOCA) is a Design Basis Accident (DBA) in
light water reactor (LWR). The design basis accident scenarios dene the operational
demands of the safety systems. No matter how unlikely these accident types may be,
all the systems must be designed so that the safety of the plant is as high as reasonably
achievable. LBLOCA was for a long time considered the worst case scenario, but after
the Three Mile Island accident (1979) it was realized that also small and intermediate
breaks could cause very severe damage to the reactor core. Therefore the researchers
moved their focus to the smaller breaks for decades. However, the development of new
computer codes for nuclear safety analysis has invoked the interest on LBLOCA issues
again.
Large break experiments done in the 60s and 70s were mainly designed to study
the coolability and the long term cooling of the core. All main phases of the LBLOCA
were studied: blowdown, rell and reood. Only a few test facilities were used to study
the rst phase of the loss-of-coolant accident, the blowdown. Pipe rupture phenomena,
depressurisation, critical ow, propagation of the pressure waves and possible defor-
mation of the core barrel are the main subjects of interest during the blowdown phase
when designing adequate emergency cooling systems and guaranteeing the safety of the
plant. [20, 21]
The main goal of this work is to gather all suitable LBLOCA blowdown tests to-
gether and compare them and the results obtained. By reviewing existing measurement
data it is easier to decide if a new test facility with a modern measurement system is
needed for structural analysis and Computational Fluid Dynamics (CFD) code val-
idation purposes. Measurement techniques and methods could be rst tested with
a smaller pilot facility. Suitable data is needed to facilitate the coupling of new CFD
codes with structural analysis codes (Fluid-structure interaction (FSI) analysis). These
codes would be used, for example, to evaluate structural integrity of the reactor core
and its supporting structures during a LOCA blowdown. To give some kind of over-
all view of the LBLOCA the most important phenomena are described. The report
concentrates on the existing experimental knowledge of the blowdown and the design
of the test facility and discusses some related general issues. This work is not meant
to replace any validation matrices, but it can be used as an additional source of infor-
mation when searching for a suitable test facility for LBLOCA blowdown validation
purposes. [20]
9
2 Large Break Loss-of-coolant Accident
The LBLOCA is a DBA in light water reactors and the double-ended-guillotine break
(DEGB) of the largest primary piping system is limiting condition for the Emergency
Core Cooling System (ECCS) requirements. Although a LBLOCA is considered very
unlikely to happen, the safety systems must be designed to secure adequate cooling of
the reactor core during the accident to prevent the meltdown of the core. Main course
of a LBLOCA is described next. [20]
In numerical analyses hypothetical loss of coolant accident is most often assumed to
take place in the cold leg of the primary circuit. Sudden rupture of a pipe in the primary
circuit leads to the rapid depressurisation of the system, this is called the blowdown
phase. During the depressurisation high-pressure coolant in the primary circuit is in
contact with the low-pressure containment atmosphere. As a result so called rarefaction
or expansion wave propagates through the primary circuit and the Reactor Pressure
Vessel (RPV) (Figure 2.1). From the rst milliseconds to approximately 50 ms of the
accident the depressurisation wave propagates in single-phase liquid ow (subcooled
blowdown), later on in two-phase ow (saturated blowdown). The coolant ow in the
reactor core is reversed in the beginning of the blowdown phase, but as the break ow
becomes saturated the ow in the core is restored. This takes only a couple of seconds
from the beginning of the accident. The velocity at which the pressure wave propagates
through the system depends on the compressibility of the uid and the elasticity of the
primary circuit piping and structures. Forces caused by the pressure waves are the
highest in the beginning of the blowdown phase. Break opening time has a signicant
inuence on the blowdown forces. When decreasing the break size and increasing the
distance from the break, forces decrease. Loads induced by the blowdown forces could
cause deformation of the core barrel and fuel bundles which could lead to ineective
cooling of certain parts of the core and end up to the partial meltdown of the reactor
core. [20, 21, 22, 23]
Critical ow rate, the highest rate at which the coolant leaves the circuit, is limited
by the sonic velocity or pressure wave propagation velocity. When the coolant is in
subcooled state, the velocity is limited by the liquid sonic velocity. When the coolant
reaches saturated state and ow at the break is in two phases, the sonic velocity is
reduced signicantly. As the break ow velocity equals the sonic velocity, the ow is
said to be choked or sonic ow. During the saturated blowdown the pressure waves
become damped and the coolant exits the primary circuit at critical ow rate in two
phases. Damping results from the equalities and the opposite directions of the blow-
down pressure wave velocity (rarefaction waves) and break ow velocity. Pressure wave
10
Figure 2.1: Expansion wave propagation and growth of core barrel deformations [1]
propagation velocity drops sharply when ashing begins and vapor is present. Velocity
is strongly dependent on the void fraction . [3, 24, 25]
As the primary circuit pressure drops, boiling front propagates through the cooling
system starting from the hottest regions of the reactor core (upper core and upper
plenum). As a result of increasing void fraction, neutron moderation weakens and
ssion process stops. Reactor scram is triggered when certain signals happen (low
pressure or low coolant level etc). Though ssion is stopped, the decay heat and
generation of void result in departure from nucleate boiling (DNB). This could lead
to high fuel cladding temperatures. The maximum allowed peak cladding tempera-
ture for Zircaloy-clad is dened by US Nuclear Regulatory Commission (NRC) in 10
CFR 50.46 regulation as 2200

F (1204

C). When cladding temperature increases above


980

C, zircaloy and steam reaction produces heat. As the cladding temperature ex-
ceeds 1100

C, the amount of heat generated in zircaloy-steam reaction is even more


signicant than the decay heat. [3, 20, 22]
After the reactor scram the primary system pressure continues decreasing. When
it has dropped below the accumulator pressure, emergency coolant is discharged into
the primary system. As a result of the blowdown, the ow in the reactor core has been
reversed again. Steam water counter-current ow blocks ECC ow in downcomer and
turns it towards the broken loop instead of the reactor core (Figure 2.2). Heat stored
in the RPV walls boils ECC water and strengthens the bypass phenomenon. In a cold
leg break more emergency core coolant is needed than in a hot leg break, partitially
because of this phenomenon. When the break is located in the hot leg similar counter-
current phenomenon doesnt happen. As the pressure dierence between the vessel
and containment decreases the amount of steam generated due to ashing is decreased.
ECC water is now able to counter the upward steam ow and enter the lower plenum
and the core, rell of the RPV starts. The second peak in fuel cladding temperature
occurs now as the steam ow through the core decreases. As the water level reaches
the bottom of the core, reood of the reactor core begins. ECC water gradually rewets
the fuel bundles and heat transfer increases. All two-phase ow regimes are present
in the core during the reood period. Heat transfer rate strongly depends on the ow
11
Figure 2.2: Phenomena during end of blowdown phase [2]
regime. After the core is fully reooded, long-term cooling must be arranged to remove
the decay heat. [3, 20, 26]
12
3 Phenomena during LBLOCA
There are many complicated phenomena present during a LBLOCA. Accurate mod-
elling of these phenomena is needed to simulate the overall behaviour of the proper
reactor system. These two-phase ow phenomena are described in this chapter. How-
ever, the main focus is on the equations and developed models, because quite complete
overall description of the phenomena already exists. This characterisation can be found
from OECD Code Validation Matrix. As the subject of this study is the blowdown test
facilities it is reasonable to emphasize those particular phenomena which are present
in the beginning of the LBLOCA. The idea is to introduce relevant phenomena as
background for the test facility study. [27, 28]
Phenomena during the LBLOCA can be basically divided in nine main categories:
basic thermodynamic phenomena, critical ow, phase separation, entrainment and de-
entrainment, stratication, CCF/CCFL phenomena, heat transfer, pool formation and
quench front propagation. [28]
3.1 Basic thermodynamic phenomena
Basic phenomena consist processes of evaporation and condensation, friction and pres-
sure drops and pressure wave propagation. One could say that they are present during
more complex phenomena. [27]
3.1.1 Evaporation due to depressurisation or heat input
As the system is depressurised and the pressure drops rapidly, saturation temperature
of the uid decreases. If the system is initially at saturated state and depressurisation
starts, uid ashes instantaneously. In the beginning of the LBLOCA depressurisation
wave propagates through the primary system and uid ashes starting from the hottest
regions of the reactor core.
During the LOCA the other form of evaporation is also observed. Fluid evaporates
as a result of the heat input. Hot walls of the reactor pressure vessel or uncovered hot
fuel bundles evaporate water and steam is generated. Evaporation phenomena during
blowdown are illustrated in Figures 2.2 and 3.1. [3, 27]
3.1.2 Condensation due to pressurisation or heat removal
As the pressure rises over saturation pressure, the steam starts to condensate. Similar
condensation follows when temperature drops below the saturation temperature due
13
Figure 3.1: Schematic of ashing transient phenomena during blowdown [3]
heat removal. Some condensation phenomena can be seen in Figure 3.1. [27]
3.1.3 Friction and pressure drop
Friction and pressure drops have great importance in nuclear safety. Interfacial fric-
tion, wall to uid friction and pressure drops at discontinuities are the main types.
Modelling of these phenomena is based on global measurements like pressure dier-
ence, pressure and temperature. Correct modelling is needed to simulate many other
two-phase phenomena.
Pressure drops and friction factors are easily obtained in case of single phase ow.
Pressure drop in a ow channel can be calculated from the equation
P
friction
=
_
fL
D
e
_
V
2
2g
c
+

i
_
K
i
V
2
i
2g
c
_
(3.1)
Friction factor f can be solved from the well-known Moody friction factor chart as a
function of Reynolds number and relative roughness of the ow channel wall. Dierent
factors K for contractions and expansions can be found from literature. Pressure
drop in dierent locations like bends and valves can be calculated in the same manner
as pressure drop by ow area change, with loss coecient K
l
or as a pipe that has
equivalent pressure drop with length L
e
P
friction
= K
l
V
2
i
2g
c
=
_
fL
e
D
e
_
V
2
2g
c
(3.2)
In case of two phase ow, calculation of the pressure drop is more complex and therefore
several correlations have been developed for dierent purposes. [22, 27]
14
3.1.4 Pressure wave propagation
The main interest in this work is to determine how to study experimentally the eects
of pressure waves on the reactor core barrel during the LBLOCA blowdown. The
pressure waves may cause deformation of the core barrel which could result to weakened
coolability of the reactor core or make the insertion of the control rods impossible.
Therefore it is important to know the eects to ensure the safety of the reactor. [27]
3.1.4.1 Single-phase ow
In a single-phase case the propagation velocity of the pressure waves is simply the
sonic velocity. Pressure pulse for the uid velocity change is derived from momentum
equation and as a result we get
p
2
p
1
V
2
V
1
= a
1
(3.3)
Sonic velocity for the single-phase uid is dened using bulk modulus K which describes
substances resistance towards compression. For example bulk modulus for water in
atmospheric pressure and 20

C is 2.19 10
9
Pa. Equation for the single-phase sonic
velocity is formed as
a
1
=

V dp
dV
=

2
_
dp
dv
_
s
(3.4)
The sonic velocity in water decreases as temperature rises. Increase in the pressure has
opposite eect on the sonic velocity. [22, 29]
3.1.4.2 Two-phase ow
In a two-phase case the calculation of pressure wave propagation velocity is more com-
plicated as it is strongly dependent on the void fraction and two-phase ow pattern.
Therefore, many dierent models have been developed during the years. One simpli-
cation is to assume two-phase ow as homogeneous mixture and then calculate average
density from equation
=
g
+ (1 )
l
(3.5)
The frozen two-phase sonic velocity is derived by substituting equation (3.5) with
equation (3.7). The propagation velocity in two-phase uid is derived from sonic ve-
locities in each phase as
a
2
TP
=
_

g
a
2
g
+
(1 )

l
a
2
l
_
1
(3.6)
The equation (3.6) can be used when void fraction is low (Grolmes and Fauske). As it
was said earlier, the propagation velocity depends on the ow pattern or to be exact
on the slip ratio. Complexity of the propagation velocity in two-phase medium makes
it dicult to model two-phase ow systems. [22, 30]
15
3.2 Critical ow
The ow is said to be critical, when the ow rate is as high as physically possible.
It means that the downstream conditions dont aect the discharge rate anymore.
Critical ow or choked ow can also be dened in single-phase system with sonic
velocity, choking velocity equals sonic velocity. This means that pressure signals cant
be transmitted to higher pressure upstream anymore, because transmission speed cant
exceed sonic velocity. In single-phase case the sonic and critical velocities are far
more easier to calculate than in two-phase case. Critical mass ow rate can be easily
obtained. When the ow is in two phases, the calculation procedure is quite dierent.
In this chapter the calculation of the critical ow rate is described and some developed
critical ow models are introduced. Critical ow is an important topic, because it
determines the severity of the LOCA. The eect of break opening time and break
location on critical ow and depressurisation will be discussed in the chapter 4. [21,
22, 27]
3.2.1 Single-phase ow
In case of the single-phase ow, sonic velocity and critical mass ow rate are simply
related as
a
2
1
=
g
c
v
2
(dv/dp)
s
, (3.7)
where a
1
is sonic velocity, g
c
gravitational conversion factor, v specic volume and s in
subscript indicates that derivate is evaluated at constant entropy. Mass ow rate can
be calculated from
G =
Q
A
, (3.8)
Now by substituting volumetric ow Q and ow area A with sonic velocity using the
denition of specic volume, we get
G
2
c
= a
2
1

2
=
g
c
(1/)
2

2
(dv/dp)
s
=
g
c
(dv/dp)
s
. (3.9)
Critical mass ow rate tells mass ow in time unit per area.[22]
3.2.2 Two-phase ow
Several two-phase critical ow models have been developed during the last 60 years.
Couple of them are introduced here. More information can be found from CSNI re-
port about two-phase critical ow models [31]. Two-phase models can be divided
into two main categories, thermodynamical equilibrium and non-equilibrium models.
Thermodynamical equilibrium means that pressure and temperature of the liquid and
vapor phases are equal. Pressure and temperature are also linked by the Mollier dia-
gram. The other way is to divide them to homogeneous and non-homogeneous models.
Homogeneity means that theres no slip between phases (phase velocities are equal,
16
Figure 3.2: Critical Pressure Ratio Data as a function of Length/Diameter Ratio [4]
k = 1). The question is which model should be applied in each case. Homogeneous
equilibrium model is not a bad way to calculate critical ow rates in long pipes where
equilibrium state can be achieved. When pipe is short more sophisticated models
for non-equilibrium ows must be applied to improve accuracy. Length to diameter
ratio is used to estimate the type of the critical ow (Figure 3.2). Mechanical and
thermodynamical dierences of later introduced critical ow models are presented in
simplied scheme (Figure 3.3). Phases are presented separately only for visual reasons.
[21, 22, 31, 32]
3.2.2.1 Homogeneous equilibrium model (HEM)
The simplest analytical model is so called homogeneous equilibrium model (HEM)
(Starkman et al. (1964)). The model assumes that velocities are equal for both phases,
thermodynamical equilibrium exists and ow is isentropic and stationary. Also, as
the name says, uid is considered as a homogeneous mixture whose thermodynamical
properties are averages between the two phases. Subscript 0 refers to reservoir con-
ditions or stagnation conditions. Flow would achieve stagnation values if it would be
brought isentropically to rest. Therefore, it can be said that the stagnation conditions
upstream are directly the initial conditions if the ow is at rest before the break. The
basic balance equations for this model are
17
Figure 3.3: Dierences between introduced two-phase critical ow models

m
wA = const. (3.10)
1
2
w
2
+ h = h
0
(3.11)
h = h
f
(p) + h
fg
(p) (3.12)
1

m
= v
m
= v
f
(p) + v
fg
(3.13)
s
0
= s
f
(p) + s
fg
(p) (3.14)
The critical mass ow rate can be solved by substituting equations (3.12), (3.13) and
(3.14) to energy equation (3.11). Now by applying the equation (3.9), we get
G
c
=
1
v
[2 (h
0
h)]
1/2
=
[2 (h
0
(p
0
) h
f
(p) (s
0
(p
0
) s
f
(p))/(s
fg
(p))h
fg
(p))]
1/2
v
f
(p) + (s
0
(p
0
) s
f
(p))/(s
fg
(p))v
fg
(p)
(3.15)
18
The model can be applied for long pipes, but it has some restrictions. It might give a
good estimate for critical ow rates, but it predicts poorly the pressure at the blowdown
pipe outlet (Moody). [22, 29, 31, 33]
3.2.2.2 Slip equilibrium model (Moody)
One well-known and widely used model was developed by Moody (1965). In his non-
homogeneous equilibrium model he assumed thermodynamic equilibrium and adiabatic,
frictionless and stationary ow. Void fraction is dened as
=
1
1 + k
1

v
f
v
g
(3.16)
For mass conservation Moody formed
G =

w
g
v
g
=
1
1
w
f
v
f
(3.17)
and for energy conservation
h
0
=
_
h
g
+
w
2
g
2
_
+ (1 )
_
h
f
+
w
2
f
2
_
(3.18)
All other equations are similar with the homogeneous equilibrium model. From the
equations it can be derived that critical ow rate is
G
c
=
_

_
2
_
h
0
h
f

h
fg
s
fg
(s
0
s
f
)
_
_
k(s
g
s
0
)
s
fg
v
f
+
s
0
s
f
s
fg
v
g
_
2
_
s
0
s
f
s
fg
+
s
g
s
0
k
2
s
fg
_
_

_
1/2
(3.19)
Slip ratio k at choking point is assumed as
k =
_
v
g
v
f
_
1/3
(3.20)
Exponent 1/3 is obtained from data comparison. Fauske assumed exponent as 1/2.
[24, 31, 34]
3.2.2.3 Homogeneous frozen ow model (HFM)
Starkman et al. (1964) developed one frequently in literature referred two-phase ow
model. In their homogeneous frozen ow model (HFM) it is assumed that average
phase velocities are equal. Also no heat and mass transfer between the phases is
assumed, therefore quality remains constant throughout the expansion. Isentropically
expanding vapor behaves like a perfect gas. Kinetic energy is due only to the vapor
expansion. Critical ow rate is dened by gas-dynamic principles. Applying all the
before mentioned assumptions, equation for the critical ow rate can be formed as
G
c
=
1
v
_
2
0
v
g0
p
0
(

1
)(1
1

)
_
1/2
(3.21)
19
The specic volume for two-phase condition is dened as
v = (1
0
)v
f0
+
0
v
g0

1/
(3.22)
In the equation (3.22) critical pressure ratio is denoted with which is
=
_
2
+ 1
_

1
(3.23)
when (1
0
)v
l0
/(
0
v
g0
) << 1. [33, 35]
3.2.2.4 Henry-Fauske model (HFK)
One of the basic non-equilibrium models was developed by Henry and Fauske. Henry-
Fauske model (HFK) is semiempirical and uses highly experimental parameter N which
describes ratio of the actual change in quality with pressure to the rate of change in
quality under equilibrium conditions
N =
d/dp
d
E
/dp
(3.24)
They determined experimentally that
N =
Et
/0.14 for
Et
0.14
N = 1.0 for
Et
> 0.14
For the critical mass ow rate they formed the equation
G
c
=
_

0
v
g
np
+ (v
g
v
l0
)
_
(1
0
)N
s
gE
s
lE
ds
lE
dp


0
c
pg
(1/n 1/)
p(s
g0
s
l0
)
__
1/2
(3.25)
However, equation (3.25) contains two unknowns, critical pressure p
c
and critical mass
ow rate G
c
. Thermal equilibrium polytropic exponent n is dened as
n =
(1 )c
l
/c
pg
+ 1
(1 )c
l
/c
pg
+ 1/
(3.26)
Therefore additional equation needs to be used in order to get solution for the problem.
Momentum equation was formed
(1 x
0
)v
l0
(p
0
p
t
) +

0

1
[p
0
v
g0
p
t
v
gt
] =
[(1
0
)v
l0
+
0
v
gt
]
2
2
G
2
c
(3.27)
Solution for the critical mass ow rate can be obtained from equations (3.25) and
(3.27). Pressure ratio can be expressed as
=
_
1
0

0
(1 ) +

1
1
2
2
t
+

1
_

1
(3.28)
20
where
=
p
t
p
0
(3.29)
=
_
1
n
+
_
1
v
l0
v
gt
__
(1
0
)Np
t

0
(s
gE
s
lE
)
t
ds
lE
dp
_
t

c
pg
(1/n 1/)
(s
g0
s
l0
)
_
(3.30)

0
=

0
v
g0
(1
0
)v
l0
+
0
v
g0
(3.31)

t
=

0
v
g0
(1
0
)v
l0
+
0
v
gt
(3.32)
v
gt
= v
g0
_

_
(3.33)
This model can be used in cases of dispersed liquid and vapor ow in nozzles. If
theres separation of phases, it cant be used. Solution requires only the knowledge of
the geometry and the upstream stagnation conditions. Henry and Fauske applied the
model also for orices and short tubes. [22, 33]
3.3 Phase separation
During the LOCA it is important to know the quality of the ow in the break location.
It determines the mass and energy removal from the system. Phase separation in
branches depends on the quality of the ow in inlet and on ow pattern. Experiments
have been conducted for dierent types of branches like Ts and Ys. Also gravity and
inertia forces cause phase separation. Phenomenon has signicant inuence on heat
transfer and quenching of the core during LBLOCA, therefore accurate modelling is
needed.
For T-branches Whalley and Azzopardi proposed a correlation for the annular ow
of high velocity in the horizontal and vertical lines.

3
/
1
=
1
2
( sin )/(

M
3
/

M
1
) (3.34)
where the angle denes the sector from which the liquid lm is removed
= 360

_

M
3
(1
3
)

M
1
(1
1
)(1 E
1
)
_
_
1.2(D
3
/D
1
)
0.4

(3.35)
and

M is mass ow, E
1
fraction of the entrained liquid in inlet and D is diameter of the
conduit. However, the correlation can only be veried with low-pressure data. [22, 27]
3.4 Entrainment and de-entrainment
Entrainment is a term for the phenomenon in which water droplets separate from the
liquid phase to the vapor phase. After entrainment droplets either fall back (horizontal
ow), deposit in the liquid phase (vertical ow) or end up being carried over by the
21
vapor. Carry-over is a term for the water droplets entrained in the vapor phase. Carry-
under is used for the opposite case in which the vapor bubbles are entrained in the
liquid phase. De-entrainment is the inverse phenomenon of entrainment, entrained
water droplets are removed from the vapor. Entrainment and de-entrainment have big
inuence on heat transfer in the core during LOCAs. Phenomena also have eect on
ECC performance during an accident. [22, 27]
These phenomena have been studied in dierent cases and some correlations have
been developed for the calculations during the years. However, the lack of detailed
information makes it possible only to form global correlations. One example of the
correlations is the one developed for the reood phase of the LOCA. During the reood
signicant amount of water is carried over as the ECC water evaporates because of
decay heat. When the average steam velocity is more than the carry-over velocity, the
following correlation can be used to estimate the entrainment

M
E
= (CRF)
_

M
f
_


M
sg
(3.36)
where CRF is carry-over fraction,

M
f
mass added to core and

M
sg
steam generated
by the decay heat. Empirical CRF factor is formed as
CRF = 0.9553 exp
_
0.0013935(p/60)
2

{1 exp [4.3127(q

/1.24)]}
{1 exp [16.62(V/6)]} exp
_
0.037161(T/140)
2

1 + sin 3.1416
_
0.5 + (Z/6)
0.025733(6/V )+0.13553(T/140)+0.0656(P/60)

2
(3.37)
where pressure is in psia, q

peak linear heat rate in kW/ft, V ooding rate in in./s,


subcooling at inlet in

F and quench front level Z in ft.
As can be seen from the multiplication factors and exponents the correlation is
highly empirical and it cant be used in any other case than ECCS evaluation during
LOCA reood. Other correlations can be found from literature, but all of them are
more or less experimental just like the one presented above. [22]
3.5 Stratied ow
Flow stratication is a term for the phenomenon in which gravitational forces have
separated the phases in the ow channel. Frictional forces and heat transfer are dierent
from homogeneous ow, because the liquid phase interacts only with a part of the ow
channel walls. Phase separation in the branches can also be signicantly dierent. [27]
3.6 CCF/CCFL phenomena
Counter Current Flow (CCF) is a phenomenon in which the subcooled emergency
core coolant uid is in contact with the hot walls of the reactor pressure vessel during
LBLOCA rell phase. Steam is generated which rises upwards and reduces penetration
22
Figure 3.4: Countercurrent ow of steam and liquid [3]
of ECC uid to the core. Counter current ow phenomenon is presented in Figure 3.4.
Counter Current Flow Limitation (CCFL) sets limits for the ECC penetration and
determines how eectively the reactor pressure vessel is relled and how much of the
ECC uid is bypassed directly to the broken loop. As the steam generation decreases
the ECC uid starts to ll the lower plenum. [3]
3.7 Heat transfer
Correct modelling of the heat transfer from fuel rods to coolant is essential when
predicting the fuel and the peak cladding temperatures during the transients and ac-
cidents. Accurate modelling of the heat transfer during the LBLOCA blowdown is a
quite complicated task due to dierent boiling modes. In the beginning of the blow-
down the mode of the heat transfer is still single-phase convection and Dittus-Boelter
and Rohsenow-Choi correlations can be used. As the nucleate boiling starts Chens
correlation can be applied. Hsu and Beckner developed correlation for critical heat
ux during rapid depressurisation. For heat transfer calculations after critical heat
ux there are couple of correlations. Tong and Young correlation for transition from
critical heat ux to lm boiling. During the lm boiling correlation by Bishop et al. or
modied Condie-Bengston can be applied. As the LBLOCA blowdown is very rapid
event, the accuracy of the heat transfer correlations is not so critical issue. Conserva-
tive and underpredicting correlations that lead to the highest temperatures of the core
are well-founded if there are no other appropriate correlations available. [22]
23
3.8 Pool formation
During the reood phase of LBLOCA water droplets may entrain in the core region
and then again de-entrain at the upper plenum support plate. This phenomenon can
cause two-phase pool formation at the upper plenum (Figure 3.5). In some reactor
designs emergency core coolant is also injected to the upper plenum. Due to the steam
generated in the reactor this water may form a pool at the upper support plate of the
core. This phenomenon is also referred as CCFL. [3]
Figure 3.5: Pool formation in upper plenum during reood [3]
3.9 Quench front propagation
Several correlations have been derived to calculate quench front propagation and rewet-
ting of the fuel pins. Some of the parameters that have eect on the quench front
propagation are wall temperature, coolant temperature and subcooling degree, pres-
sure, wall material etc. Couple of the most used parameters in correlations are Peclet
and Biot numbers, and dimensionless temperature. Eect of a single parameter on this
phenomenon is usually hard to observe. Also spacer grids have eect on quenching
(Figure 3.6). [3, 36]
24
Figure 3.6: Quenching during reood [3]
25
4 Pipe break mechanics
Break characteristics determine the consequences of the pipe rupture. Break charac-
teristics consist of fracture orientation, break opening time and break area. Normally
break is estimated to open in 1 millisecond to have certain conservatism in the safety
calculations. Amplitude of the rarefaction wave is a function of the opening time.
Therefore, when the break opening time is longer the dynamic loads are reduced, but
exist for a longer period of time. [5]
4.1 Longitudinal break
Baum summarizes in his article eects of dierent break characteristics [5]. He describes
tests with the gas-pressurised pipe to determine the opening time of the axial break. He
introduces a simple model to predict the minimum break opening time. Using the initial
crack length and breach tip velocity ratio L/v
p
(three cases) and the characteristic time
(equation (4.4)) for the rupture process, the opening time can be predicted (Figure
4.2). In the gure the opening time is presented as the function of the characteristic
time. Baum describes the charateristic time as a term which presents the eect of
the pipe diameter and how the pipe is stressed (hoop stress). For smaller and highly
stressed pipes opening times are naturally much shorter. [5]
The break opening time to full area is derived from the idealised breach. Breach
area as a function of time can be expressed as
A =
2P
0
v
p
t
3
3m
+
P
0
Lt
2
m
(4.1)
in which the rst term presents area generated by the movement of the crack tip and the
second term area generated by the horizontal movement of the free edges. When initial
defect length L is large it has more eect on the break area growth than the crack tip
propagation has until t > (3L)/(2v
p
). The ductile fracture propagation velocity v
p
can
be dened as the propagation velocity of a displacement wave in a plastic membrane
v
p
= (
y
/
s
)
1/2
(4.2)
The hoop stress before the rupture is
0
= P
0
R/h in which h is the wall thickness and
R is the radius of the pipe (Figure 4.3). Mass per unit area is m =
s
h. Substituting
these assumptions to (4.1) and using full break area (200%) as A, we get equation
2
3
=
2
3
t
3
+ (L/v
p
)t
2
(4.3)
26
Figure 4.1: Dierent rupture types considered in High Energy Line Breaks
0.0001
0.001
0.01
0.0001 0.001 0.01
O
p
e
n
i
n
g

t
i
m
e

(
s
)
alpha (s)
0
0.3 ms
2 ms
Figure 4.2: Predicted opening times of the break in three cases [5]
27
Figure 4.3: Hoop stress in cylindrical shell
in which
= (R/v
p
)(
y
/
0
)
1/3
(4.4)
The equation introduced by Baum (4.3) is simple and quite well-founded way to
approximate the minimum break opening time. The model suggests smaller break
opening times than were measured in the large scale experiments. The reason for this
is the conservative assumption that the rupture pressure P
0
remains the same. Baum
summarizes that break opening times are larger than 1 ms when exceeds 0.5 ms. [5]
4.2 Circumferential break
Circumferential break is more familiarly known as a guillotine break. The break is con-
servatively assumed to open instantaneously (1 ms) in the safety calculations. However,
there has been a discussion that the conservative assumption of 1 milliseconds is un-
realistic and instead of it the best estimate break opening time could be used. Some
assumptions for the realistic break opening time in dierent locations of the primary
circuit can be found in Figure 4.4. Instead of using the conservative assumption the
break opening time of 20 ms would be used in the case of cold leg guillotine break. [6]
4.3 Pipe break simulations using Fluent CFD
Series of simulations with Fluent CFD software were done for this report. Purpose
of these simulations is mainly to illustrate the eect of dierent break parameters
on depressurisation. Using of Fluent to LBLOCA blowdown simulations may sound
questionable. However, in the beginning of the blowdown the ow is mainly in one-
phase. Therefore it is possible to use Fluent for the sensitivity studies.
Preliminary simulations were done to test the eect of dierent solver settings and
calculation meshes. After these simulations the solver settings for the nal simulations
were decided. The pressure based solver was found out to be the fastest and the results
28
Figure 4.4: DEGB locations, realistic break sizes and times [6]
were the same as with the density based solver. Coupled method was chosen as the
pressure-velocity coupling scheme. Realizable k- turbulence model was applied to the
simulations. Courant number was calculated using the speed of sound. Due to relatively
large speed of sound small time steps had to be used (t = 3 10
6
s). Because of small
time steps the simulation time was selected as 6 milliseconds. However, the reection
of the rarefaction wave from the constant pressure inlet boundary condition happens
much earlier. Pressure recovery in the vessel is a result of this issue. Two main cases
were simulated using Fluent. The eect of pipe length on depressurisation was studied
in Case 1 and the eect of break opening time in Case 2. Three dierent geometries
were used in Case 1. In total 6 dierent Fluent cases were simulated. The center line of
the pipe was dened as the rotational axis (x-axis). One of the geometries is presented
in Figure 4.5. As an example, arrival of the rarefaction wave to the vessel is presented
in Figure 4.6. Other geometries were similar with an exception that the pipe length
was dierent. The pipe radius (100 mm) and the downcomer width (140 mm) and
length (1000 mm) were the same in all geometries. The location where pressure across
the core barrel wall was measured is marked in Figure 4.5 with a red circle (referred
in gures of curves as KP9). General and case specic settings of the simulations are
summarized in Table 4.1.
29
Table 4.1: FLUENT simulation settings
Solver settings
Solver Pressure Based, Axisymmetric
Pressure-Velocity Coupling Coupled
Formulation and Time Implicit, Unsteady
Simulated time 6 milliseconds
Time Step 3e-06 s
Iterations per Time Step 150
Discretization Pressure: 2nd Order
Density: 2nd Order Upwind
Momentum: 2nd Order Upwind
Turbulent Kinetic Energy: 2nd Order Upwind
Turbulent Dissipation Rate: 2nd Order Upwind
Velocity formulation Absolute
Unsteady formulation 1st-Order Implicit
Gradient Option Green-Gauss Cell Based
Viscous Model Realizable k-, Standard Wall Functions
Case specic settings Case 1 Case 2
Mesh size [cells, quadrilateral] 15627, 22819, 28039 22819
Initial Gauge Pressure [bar] 109 109
Inlet Gauge Pressure [bar] 109 109
Outlet Gauge Pressure [bar] 31.99, 52.39, 57.29 52.39
Pipe length [m] 0.5, 1.369, 2 1.369
Break Opening Time [ms] 1 0.5, 1, 1.5, 2
Water Temperature [K] 541 541
Water properties
Density [kg/m
3
] Pressure variable (UDF)
Viscosity [kg/ms] Constant, 9.998e-05
Speed of Sound [m/s] Pressure variable (UDF)
4.3.1 Eect of pipe length on depressurisation
Setting of the outlet boundary condition was the most dicult part in these simulations
(Figure 4.7). It was decided to set the outlet pressure as the pressure in the critical
ow situation (Fauske diagram, Figure 3.2). Other option would have been to set break
open to atmospheric pressure. In Case 1, the pressure dierence across the core barrel
wall behaved as could be expected (Figure 4.8). When the pipe is shorter, the pressure
dierence is greater and the depressurisation of the vessel begins earlier.
4.3.2 Eect of break opening time on depressurisation
The eect of break opening time was studied in Case 2. The HDR geometry (Figure
4.5) was chosen as the simulation case and the outlet boundary condition was dierent
in each simulation (Figure 4.9). The pressure dierence across the core barrel wall
30
Grid (Time=0.0000e+00)
FLUENT 6.3 (axi, pbns, rke, unsteady)
Feb 27, 2008
Figure 4.5: Mesh/Geometry for 1.369 m pipe
Contours of Absolute Pressure (pascal) (Time=1.6470e-03) Mar 06, 2008
FLUENT 6.3 (axi, pbns, rke, unsteady)
1.10e+07
5.34e+06
5.53e+06
5.72e+06
5.91e+06
6.09e+06
6.28e+06
6.47e+06
6.66e+06
6.85e+06
7.04e+06
7.23e+06
7.42e+06
7.60e+06
7.79e+06
7.98e+06
8.17e+06
8.36e+06
8.55e+06
8.74e+06
8.93e+06
9.11e+06
9.30e+06
9.49e+06
9.68e+06
9.87e+06
1.01e+07
1.02e+07
1.04e+07
1.06e+07
1.08e+07
Figure 4.6: Arrival of the rarefaction wave to the vessel in Case 2 (0.5 ms)
behaved as could be expected (Figure 4.10). When the break opening time is shorter
the shape of the rarefaction wave becomes much steeper. This results in greater pressure
dierences across the wall.
31
3e+006
4e+006
5e+006
6e+006
7e+006
8e+006
9e+006
1e+007
1.1e+007
0 0.0005 0.001 0.0015 0.002 0.0025 0.003 0.0035
[
P
a
]
TIME [s]
Outlet pressure 0.5m
Outlet pressure 1.369m
Outlet pressure 2m
Figure 4.7: Pressure at outlet in Case 1 simulations
-2e+006
-1.5e+006
-1e+006
-500000
0
500000
0 0.0005 0.001 0.0015 0.002 0.0025 0.003 0.0035
[
P
a
]
TIME [s]
KP9 0.5m
KP9 1.369m
KP9 2m
Figure 4.8: Eect of pipe length on depressurisation
32
5e+006
6e+006
7e+006
8e+006
9e+006
1e+007
1.1e+007
0 0.0005 0.001 0.0015 0.002 0.0025 0.003 0.0035
[
P
a
]
TIME [s]
Outlet pressure 0.5ms
Outlet pressure 1ms
Outlet pressure 1.5ms
Outlet pressure 2ms
Figure 4.9: Pressure at outlet in Case 2 simulations
-1.5e+006
-1e+006
-500000
0
500000
1e+006
1.5e+006
0 0.0005 0.001 0.0015 0.002 0.0025 0.003 0.0035
[
P
a
]
TIME [s]
KP9 0.5ms
KP9 1ms
KP9 1.5ms
KP9 2ms
Figure 4.10: Eect of break opening time on depressurisation
33
5 Large Break test facilities
Large Break tests have been conducted in several laboratories and with numerous
test facilities. The tests carried out before the Three Mile Island accident (1979)
were mainly designed to study phenomena and overall behaviour of the system during
LBLOCA. However, after the accident it was realized that also small breaks could
cause severe damage to the reactor core. The main focus in the studies moved to
smaller breaks and leaks and the majority of the test facilities was designed to study
these accidents.
The main goal of this work is to review LBLOCA tests and more precisely the tests
which were designed to study phenomena during the blowdown phase of the accident.
During the rapid depressurisation in the blowdown phase, pressure waves propagate
through the primary system. Forces induced by these waves could cause deformation
of the core barrel. Therefore, the interest is in the tests which were designed to study
depressurisation, critical ow or uid-structure interactions during LBLOCA.
The reviewed tests are divided in two categories, to separate eects tests and in-
tegral eects tests. Separate eects tests are designed to investigate individual or
localized phenomena. Integral eects tests are quite opposite of the separate eects
tests. They aim to simulate the overall behaviour of the reactor system and the in-
teractions between the dierent components. The relatively large scale of the most
integral test facilities makes the detailed measurement of the two-phase phenomena
quite impossible. Therefore the data acquired in the separate eects tests is needed in
the detailed modelling of the two-phase phenomena and code validation. [27]
Separate eects tests are reviewed in section 5.1 and integral eects tests in section
5.2.
5.1 Separate Eects Tests
Some separate eects test facilities have been constructed to study LBLOCA blowdown.
However, the vast majority of them have been designed for critical ow rate studies.
In this report Edwards-OBrien pipe (5.1.1), Bartaks pipe (5.1.2), CANON and Super
CANON (5.1.3) and Marviken (5.1.4) depressurisation facilities are described. As it was
said earlier deformation questions are important during LBLOCA blowdown. There-
fore couple of test facilities with core internals have been built to study the eects of
blowdown on the internal structures of the reactor. CSE (5.1.5), SEMISCALE aka
Bettis Flask (5.1.6), Piper (5.1.7), Battelle-Frankfurt RS-16B (5.1.8) and Hei Dampf
Reaktor (HDR) (5.1.9) are described here in more detail.
34
Figure 5.1: Schematic of Edwards-OBrien blowdown pipe [7]
5.1.1 Edwards-OBrien pipe
Edwards and OBrien conducted a test in 1973 which is known as International Stan-
dard Problem (ISP) 1. Their experimental test facility was designed to study critical
ow and contained a straight pipe whose length was 4.1 meters and internal diame-
ter 73 millimeters (Figure 5.1). They conducted a series of depressurisation tests in
which the pipe was initially lled with water. In the ISP test the pipe was pressur-
ized to 7 MPa and the water temperature was 240

C. The desired initial state was


obtained by rst removing the air from the pipe and then lling it with demineralized
water. The leakage test was made by pressurizing the pipe without heating to over
depressurisation state. After that the pipe was heated to get the wanted temperature
distribution. Depressurisation was initiated by using a glass rupture disc (thickness
0.5 in. and diameter 3.5 in.) which was shot with a pellet gun. Opening time of the
break is estimated as 1 ms using a thin conductive silver layer on the surface of the
rupture disk. Flow area was reduced by about 13%, because the glass disc did not fully
break. Instrumentation system of the pipe consisted of 7 pressure and 7 temperature
measurements during the transient. Also local void fraction was measured in two lo-
cations with x-ray densitometers. Axial blowdown forces were measured using a load
cell which was rmly attached to concrete. [7, 37, 38]
It was found in these tests that thermal non-equilibrium conditions are signicant
during the fast depressurisation transients. During the years these tests have been the
most popular benchmark experiments when testing code capabilities. [38]
5.1.2 Bartaks pipe
One blowdown experiment series was conducted by Bartak. Bartaks facility was a
scaled model of the soviet WWER-type pressure vessel with one inlet and one outlet
nozzle (Figure 5.2). The length of the blowdown pipe was 1.7 meters and the diameter
88 millimeters. The blowdown transient was initiated by using a rupture disc assembly.
The assembly consisted of two rupture discs. The pressure between these discs was 50%
35
Figure 5.2: Schematic of Bart aks horizontal experimental channel [8]
of the system pressure. The outer rupture disc was broken by increasing the pressure
between the discs. After that the inner rupture disc broke as well. Total 13 tests were
carried out with dierent initial conditions. Maximum initial system pressure was 12
MPa and temperature 300

C. Findings from these tests were similar as in Edwards-


OBrien experiments, thermodynamic non-equilibrium eects are signicant. [8]
5.1.3 CANON and Super CANON
Series of blowdown experiments were performed with the CANON (Figure 5.3) and
Super CANON blowdown (Figure 5.4) facilities. Blowdown experiments from high
initial temperature (320

C at maximum) and pressure (15 MPa) were carried out with


the latter facility. The length of the Super CANON blowdown pipe was 4.389 meters
and the diameter 102.3 millimeters. The length of the CANON blowdown pipe was
3 meters and the diameter 102.3 millimeters. However, the initial pressure was much
lower (3.2 MPa) in CANON and therefore the initial temperature was lower as well
(230

C at maximum). A rupture disc assembly (two discs) was used as a break device
in both facilities. Diameter of the break area ranged from 30 to 100 millimeters. In the
Super CANON experiments absolute pressure was measured in 7 locations along the
pipe and temperature in 6 locations. Also void fraction was measured in one location
using neutron scattering technique. In the CANON experiments absolute pressure and
temperature were measured from 4 locations and void fraction with neutron absorption
technique. Experimental results gathered in these experiment series have been used for
code assessment purposes (CATHARE). [10, 39, 40]
5.1.4 Marviken
The Marviken power plant or R4 reactor in Sweden was supposed to operate as a
boiling heavy water direct cycle reactor, but instead it served as a critical ow test
facility. The plant was never used as a nuclear reactor so it was an ideal large scale test
facility for critical ow experiments. The test facility was operated in the years 1972-
1985. Containment response, critical ow, jet impingement and level swell tests were
carried out in 1977-1981. The critical ow test project started in 1976 and the tests
were carried out in the years 1978-1979. The tests are well documented and reports are
36
Figure 5.3: Schematic of CANON facility [9]
Figure 5.4: Schematic of Super CANON facility [10]
37
available in the OECD Nuclear Energy Agency data bank. Also an info sheet about
Marviken experiments can be found from the Code Validation Matrix. [11, 39]
5.1.4.1 Overview of Marviken experiments
As the Marviken power plant was originally designed as a full scale reactor it provided
excellent opportunities for plant scale testing (Figure 5.5). The net volume of the
pressure vessel was 425 m
3
, design pressure 57.5 bar and temperature 272

C. Inner
height of the vessel was 24.55 meters and diameter 5.22 meters. The vessel was modied
by removing parts of the internals and installing a discharge pipe vertically to the
bottom. Test nozzle was attached to the other end of the discharge pipe. Break device
was a rupture disc assembly with two rupture discs. The volume between the discs was
pressurized to half of the system pressure. In the reports it is referred as the balancing
pressure. The volume between the discs was pressurized by using one tank and the
break was initiated by using an other tank with higher pressure. [11]
In total 27 tests were performed in the critical ow test series. The test parameters
were varied so that the main objective of the series was reached. The objective was to
determine how critical mass ow rate depends on dierent parameters like upstream
stagnation pressure and enthalpy, nozzle length and diameter and air concentration.
Steam dome pressure was slightly varied and therefore the saturation temperature
varied as well. The mass of water (saturated and subcooled), nozzle size and subcooling
degree were also varied. Oxygen content in the water was measured in part of the tests
and in test 27 it was lowered signicantly and results were compared with test 21.
[11, 41]
5.1.4.2 Instrumentation
The pressure vessel and the discharge line had vast instrumentation. The pressure ves-
sel had six absolute pressure transducers and several dierential pressure transducers.
Also temperatures were measured from the cooling lines for the water cooled pressure
transducers. Inside the vessel 28 temperature measurement points were used to obtain
the temperature prole. The discharge line had several dierential and absolute pres-
sure transducers. Temperature was measured from many points as well. For density
measurement a three-beam gamma densitometer was used. [41, 42]
5.1.5 CSE blowdown experiments
Containment Systems Experiment (CSE) reactor simulator was instrumented to mea-
sure pressures, temperatures, water level, void fraction in the exit duct, mass of uid
in vessel, forces and stresses during loss of coolant accident. The reactor simulator was
designed to operate in temperatures up to 600

F and pressures up to 2750 psig (approx-


imately 316

C and 191 bar). The volume of the vessel was 150 ft


3
(appr. 4.2 m
3
). The
blowdown pipe was attached to the lower part of the vessel and the break was initiated
using a double rupture disc assembly (Figure 5.6). The core was simulated with a quite
similar approach as in the Piper facility by using a perforated plate dummy core. The
object of the blowdown experiments was to gather data for code comparison. Special
interest was in blowdown forces which are subjected to the simulated core. [12, 43]
38
Figure 5.5: Schematic of Marviken critical ow test facility [11]
39
Figure 5.6: Schematic of CSE reactor simulator [12]
40
Figure 5.7: Schematic of the vessel used for SEMISCALE test 711 [13]
5.1.6 Semiscale (Bettis Flask)
The Water Reactor Research Test Facility (Idaho National Laboratories, Test Area
North) includes the Semiscale Facility, the Blowdown Facility, and the Full Area Steady
State Testing (FAST) Facility. The Blowdown Facility was constructed to study the
behaviour of the instruments and components during the depressurisation phase of
the LBLOCA. The facility was used from 1965 to 1968 and three test series were
conducted, named as 500, 600 and 700 series. The whole vessel was initially lled
with subcooled water and the depressurisation was initiated by using a rupture disk
assembly. Schematic of the vessel used in Semiscale test series 700 and test 711 shows
the places of the instrumentation and the scale of the facility. (Figure 5.7). [13, 18, 44]
5.1.7 Piper
Piper facility was designed to study dierent aspects of blowdown (Figure 5.8). Eval-
uation of loads on internal structures was one studied subject. The facility was mainly
used to simulate BWR blowdown as the initial content of the vessel was steam-water
mixture. However, initial void varied from 0 to 0.9 in the performed tests. The pres-
sure range was from 1 to 9 MPa and the subcooling degree from 0 to 150

C. Blowdown
was initiated using the double rupture disc assembly, probably with similar working
principle as in Bartaks experiments. Break opening time varied between 3 and 5 ms.
Internal structures of the vessel were simulated by a plate with a central hole. [14]
41
Figure 5.8: Schematic of Piper vessel and blowdown nozzle [14]
5.1.8 Battelle-Frankfurt RS-16B
A series of blowdown experiments (indicated as DWR tests) was carried out in Battelle-
Institute in Frankfurt. These RS-16B experiments were designed to study blowdown
loads on pressurised water reactor internals. The tests were carried out in 1975. The
test facility simulated a reactor pressure vessel with internals (Figure 5.9). The height
of the vessel was 11.3 meters and the diameter 0.8 meters. The outside diameter of the
core barrel was 0.62 meters and the length 7.37 meters and it was attached at the top
to a grid plate. This plate was rigidly attached to the vessel. Electrical heaters and
some passive rods simulated the core at the lower part of the core barrel. The most
signicant dierence between this facility and later introduced HDR facility was the
diameter scale. The heights were almost identical, but the diameters in this facility
were much smaller than in the HDR facility. Also the way to simulate the reactor core
diered as well. Because of the scale, thickness of the core barrel was less than in the
HDR facility, the upper one third of the core barrel had a thickness of 8 millimeters and
the lower two thirds 18 millimeters. As an example it can be said that in test DWR5
the initial pressure was 14.2 MPa and temperature 301

C at maximum. The discharge


pipe had a length of 0.35 meters and a diameter of 0.142 meters. Break was initiated
using a rupture disc and the break opening time was approximately 1 ms. The large
length-to-diameter ratio of the vessel resulted in highly one-dimensional behaviour in
the lower parts of the vessel. [15, 45]
42
Figure 5.9: Schematic of RS-16B blowdown facility and models[15]
5.1.9 Hei Dampf Reaktor (HDR)
Hei Dampf Reaktor (HDR) was originally built as a research reactor and was located
in Growelzheim near Karlstein am Main, in West Germany. However the reactor had
to be shut down in 1971 only after a short period of operation. The reason for this
was constructional faults in the fuel elements. After the shutdown the reactor got a
new research purpose as a test facility for large scale experiments when Project HDR
was established in 1973. The main goal of these experiments was to determine the
importance of the multidimensional uid-structure interaction phenomena during the
blowdown phase of a loss of coolant accident. Fluid-structural tests were performed
earlier in a couple of smaller scale test facilities. For the code validation purposes larger
scale tests were needed. In the end of the 1970s and in the beginning of the 1980s
series of large scale blowdown experiments were performed in HDR facilities.
5.1.9.1 Overview of HDR experiments
HDR had a near full scale reactor pressure vessel (Figure 5.10). For the blowdown
tests the reactor internals had to be modied. The original core barrel was too thick
for deformations to happen. It was replaced with a new one. This new core barrel
was carefully designed and uid-structure interaction codes were used in determining
the wall thickness of the barrel (23 mm). To get clear boundary conditions for the
simulations the core barrel was rigidly clamped at the top in most of the experiments
43
Figure 5.10: Schematic of HDR pressure vessel and core barrel [16]
and the core itself was simulated with a mass ring attached to the bottom of the
core barrel. The weight of the mass ring was approximately half of the core mass.
In pressurised water reactors the horizontal movement of the lower core barrel end is
limited by the core barrel bumpers. In HDR facility there were also such bumpers,
but clearances were so large that impacts were not expected to happen. The bumpers
were there as an additional safety feature to prevent overloading of the barrel clamping.
[1, 17, 46]
Issues that had to be solved during planning of HDR experiments included core
barrel clamping (rigid or non-rigid?), designing the new core barrel, air removal from
the system, the eects of initial temperature distribution, simplication of the core and
planning the tests according to the pre-calculations to avoid plastic deformations.
HDR-RPV-I blowdown test program was divided in two phases, to preliminary
test phase and main test phase. Blowdown experiments V29.2, V31 and V31.1 were
performed during the preliminary test phase. For obvious reasons testing was started
with initial conditions which would result in low blowdown loads. Blowdown loads were
increased in test V31. Test V31.1 was carried out to test repeatability. The preliminary
test phase was also needed to test instrumentation and measurement techniques and
to get experience of the facility and its behaviour. Test matrix for the main test phase
was nalized during preliminary tests. [17]
During the main test phase four blowdown tests were performed: V31.2, V32, V33
and V34. Instrumentation was increased signicantly from the preliminary test phase.
44
Table 5.1: HDR preliminary and main phase test matrices
Test Pressure Subcooling in downcomer Upper Core Downcomer
No. (bar) (

C) (

C) (

C)
V29.2 90 30 293 273
V31 110 50 308 268
V31.1 110 50 308 268
V31.2 110 50 308 268
V32 110 78 308 240
V33 110 78 308 240
V34 110 78 240 240
Test V31.2 was a replication of test V31 to examine how well experiments can be
reproduced after a long time. Subcooling degree was increased in test V32. This test
is also German Standard Problem number 5. In test V33 the eect of reduced break
area (25% ow area) was studied and the initial conditions were identical with V32.
Test V34 was done with non-rigid clamping of the core barrel. Non-rigid clamping
allowed axial movement of the core barrel and impacts between the lower end of the
core barrel and the pressure vessel. Initial conditions of all tests are summarized in
Table 5.1. Originally, more isothermal tests were supposed to be done in the main
phase, but the test matrix was changed so that test V34 was the only one. The length
of the blowdown nozzle was 4.524 meters in V29.2 and 1.369 meters in all others. [47]
More in depth analysis of the HDR test results can be found from Wolfs arti-
cles [17, 47]. From the preliminary tests Wolf concludes that the repeatability of the
experiments and the reliability of the measurements were excellent despite some di-
culties in density and displacement measurements. The results of tests indicated that
FSI reduces blowdown loads to the core barrel, three-dimensional eects exist during
blowdown, higher subcooling and shorter blowdown pipe increase the core barrel loads,
maximum loads occur during subcooled blowdown, the core barrel displacements are
quite small and that only substantial displacements of the pressure vessel have been
measured. From the main test phase Wolf summarizes that higher subcooling increases
the blowdown loads, the reduction of the break area to 25% reduces loads by 10 to 40%
and the non-rigidly clamped core barrel have signicant eect on core barrel dynamics,
but hardly any on uid dynamics. Many pre- and post-analyses have been done with
several uid-structure interaction codes. Comparisons between the measurement data
and calculations can be found from the several articles. [17, 47, 48]
5.1.9.2 Instrumentation
The instrumentation of the HDR facility was quite extensive. Instrumentation in the
discharge pipe during the preliminary test phase consisted of several fast pressure trans-
ducers, a two-beam gamma densitometer, two dragbodies and many pairs of pressure
and temperature sensors (Figure 5.11). In order to repeat experiments initial temper-
ature distribution was measured. Measurement was done rst with 122 temperature
sensors and during the blowdown 25 temperature sensors were used. By doing so it
45
Figure 5.11: Instrumentation of the long and short discharge pipes in HDR [17]
was possible to reliably interpolate the temperature distribution inside the HDR vessel.
The vessel itself had very vast instrumentation. In the core barrel there were absolute
pressure transducers, dierential pressure transducers and temperature sensors to mea-
sure uid dynamic quantities. To obtain structural quantities strains, accelerations and
displacements were measured. The most signicant problem in the use of HDR data
for CFD validation is the lack of pressure data inside the core barrel. [17]
5.1.10 Other tests
Some other critical ow and break ow studies that have been referred in literature and
technical reports are LOFT-Wyle Transient Fluid Calibration test facility (WSB03R
etc.), Carofano-McManus, Cumulus, Deich, Fincke-Collins and Neussen. Some other
critical ow tests are shortly described in the research report made by Yildiz et al.[49,
50]
46
5.2 Integral Eects Tests
Integral test facilities together with the well planned separate eects tests have been the
base for the thermal-hydraulic code validation. Integral test facilities have been built
to study dierent aspects of LOCA. Some facilities were used only to simulate small
and intermediate breaks while couple of facilities were designed to study all LOCA
breach windows. LOFT, SEMISCALE (5.2.1) and LOBI (5.2.2) facilities are briey
described next.
5.2.1 LOFT and SEMISCALE
Loss Of Fluid Test facility (LOFT) was constructed in the Test Area North at Idaho
National Engineering Laboratory. The building was done between 1965 and 1975.
LOFT facility was volumetrically scaled (coolant volume to power ratio) and it had
a real nuclear core which made it very unique in the eld of the thermal-hydraulic
research (Figure 5.12). The facility had two loops, one intact and one broken loop.
The intact loop was scaled and designed to simulate three unbroken loops of a generic
four-loop PWR during LOCA. The broken loop was not closed, hot leg and cold leg
had quick opening valves which initiated the LOCA. The coolant was discharged to
the pressure suppression system to prevent contamination and to simulate containment
pressure. In total 38 experiments ranging from large to small breaks were carried
out in the LOFT facility. Instrumentation was quite extensive, because the facility
had normal PWR instrumentation and in addition vast temperature, pressure and
mechanical measurements. The number of measurements and some main dimensions
are presented in Figure 5.13. [18, 51]
Figure 5.12: LOFT test facility [18]
47
Figure 5.13: LOFT reactor vessel instrumentation [18]
First model of SEMISCALE integral test facility became operational in September
1974. It was scaled to LOFT facility and its original purpose was to study LBLOCA. It
was a full height and volume scaled facility. The test facility had an electrically heated
core so it was easier to operate than the LOFT facility. Westinghouse 4-loop PWR
was used as a reference plant in the further modications of the facility. Those later
modications were also a full height and volume scaled (1/1705.5). [18]
Figure 5.14: Semiscale Mod-1 (left) and Semiscale Mod-2C (right) [18]
48
5.2.2 LOBI
LOop Blowdown Investigation facility (LOBI) was originally designed to study large
break loss of coolant accidents. The project began in the early 1970s when there
was a special need to research the ECCS performance in large power reactors. The
test facility was build in Joint Research Center in Ispra and the German BIBLIS
B power reactor was used as a reference plant. The LOBI-Mod1 test facility was
commissioned in December 1979 and the experimental program was concluded in June
1982 (Figure 5.15). During that period of time 28 experiments were carried out of
which 25 were LBLOCA tests. The facility was scaled using the power to volume scaling
criterion (volume ratio 1/700) except that the pressurizer had to be scaled dierently to
accommodate the internal heaters. The downcomer scaling in the original test facility
is said to be the only major exception. It was made intentionally larger to avoid ECC
bypass, but later on this was found out to be a mistake. The downcomer was replaced
with a new one which had a 12 millimeter gap instead of 50 millimeters. During the
rst phase of the LOBI project some experiments were also made to evaluate test
repeatability and the eects of the break size and geometry on the course of LBLOCA.
[18, 52]
After the rst phase of the LOBI project the facility was modied for the small
break studies. The experimental work continued in April 1984 and the last experiment
with LOBI-Mod2 was done in June 1991. [52]
Figure 5.15: LOBI test facility [18]
49
6 Test facility summary
LBLOCA blowdown phase has been studied from many angles. There have been simple
pipe facilities which have been used to study depressurisation and critical ow. Couple
of the test facilities have been more reactor like and have been used to study also
structural forces and loads. Third category is the integral test facilities which simulate
the whole primary circuit and have been used for thermal-hydraulic studies. All the
test facilities described before are summarized in table 6.1 with the estimated scale
(S=Small, M=Medium, L=Large) and studied phenomena.
The task of this study was to nd out whether there are better test facilities for
uid-structure code validation than HDR or not. After the very wide search for the
facilities it can be concluded that none of the found facilities can compete with the
results gained from the HDR experiments. The next question is: could we somehow
improve measurements from the HDR experiments to gain some additional knowledge
of the uid-structure interaction phenomena with a new test facility. General issues
and problems related to designing and scaling a new test facility are introduced in
chapter 7.
50
Table 6.1: Test facilities summarized
Test facility Scale Studied phenomena
Separate Eects Tests
Edwards-OBrien pipe S critical ow, depressurisation
Bartaks pipe S critical ow, depressurisation
Canon S critical ow, depressurisation
Super Canon S critical ow, depressurisation
Marviken L critical ow, depressurisation
CSE M critical ow, depressurisation, loads, forces
Semiscale (Bettis Flask) S critical ow, depressurisation
Piper S critical ow, depressurisation, loads, forces
Battelle-Frankfurt RS-16B L critical ow, depressurisation, loads, forces
HDR L critical ow, depressurisation, loads, forces
Integral Eects Tests
LOFT L thermal-hydraulic eects during LBLOCA
Semiscale L thermal-hydraulic eects during LBLOCA
LOBI L thermal-hydraulic eects during LBLOCA
51
7 Designing the new facility
The facilities like HDR and Battelle-Frankfurt facility were nearly full scale and built
to study uid-structure interactions during the blowdown phase of LBLOCA. Building
of such a large scale facility is expensive. The size of the existing laboratories may
also set some limitations. Therefore one solution is to build a small scale facility with
a lot of instrumentation. The initiative to design a new test facility came from the
code developers. They are searching for suitable test data to validate uid-structure
interaction calculations done with a coupling of CFD and structural analysis codes.
Some calculations have been done with the HDR geometry, but there are some aws
in the facility that restrict the use of experimental data for code validation purposes.
For obvious reasons (tests were done in the 70s and 80s) the facility was not designed
to validate CFD codes. For that purpose there must be a dense grid of pressure and
temperature measurements. In this case the pressure measurements would be enough.
When the task is to validate the coupling of CFD and structural analysis codes extensive
structural measurements are also needed. Certain important aspects have to be taken
into account when designing a new separate eects test facility; scaling, structural
questions, instrumentation and planned test program. These questions are considered
next each in their own sections.
7.1 Scaling
In order to get results that could be scaled to the full scale of the reference plant, the
test facility should be designed using appropriate scaling laws. It is essential to choose
the right scaling law to capture the nature of the phenomena correctly. If the purpose
of the experiments is to study a highly multidimensional phenomenon, dimensions of
the test facility must be well chosen. During the blowdown phase the phenomena are
three-dimensional. For example in Battelle-Frankfurt blowdown experiments it was
found out that in the lower parts of the pressure vessel the behaviour of the system
was mainly one-dimensional due to the small width to height ratio. The problem with
small scale or volumetrically scaled facilities is that phenomena are quite distorted.
One example of this is the dierence in ECC bypass phenomenon between UPTF and
SEMISCALE experiments.
There are two traditional ways to scale a test facility: time-preserving volumetric
scaling and time-reducing length scaling. Volumetric scaling is not an option for this
facility, because the essential phenomena would become distorted. Nahavandi et al.
introduced three scaling laws: time-reducing length scaling, time-preserving volumetric
scaling and time-preserving idealized model/prototype scaling. The most applicable of
52
these laws to this case would be the time-reducing length scaling. The problem with
this scaling law is that accelerations are scaled with the inverse of length scale that
means 1/. However, they concluded that
For the transient problems such as the blowdown phase of the loss-of-
coolant accident (LOCA), steam break, and water hammer phenomena, body
forces due to the acceleration of gravity play an insignicant role compared
to the pressure dierential, and distortion in the model is insignicant.
However, for transient problems involving density wave phenomenon, such
as steam generator hydrothermal instability and the reood phase of the
LOCA, the elevational head in the downcomer due to gravity is consider-
able compared to the pressure dierential. For these problems, the model
predictions using the above similarity rationale will be signicantly distorted
(i.e., dierent from the prototype behavior).
This means that the scaling law in question would be applicable to this particular case.
The development of the time-reducing scaling law begins from the basic equations:
continuity, momentum and energy equations. As a result it is obtained that model and
prototype are similar if

= 1 (7.1)
where is the time-scale and is the length scale. This means that the time-scale
should equal the length-scale (time-reducing). The time-reducing scaling law is pre-
sented in table 7.1.[19]
As the wall temperature, power input, number of rods and rod diameter are not
subjects of interest, the law adapts quite nicely to blowdown phase study. Area and
volume are simply developed from the length scale. For example hydraulic diameter
for round tube equality can be derived as
D
h
=
4A
U
=
4
2
D
2
4
D

=

2
D
2
D

=
(D

)
2
D

= D

=
D

= D
D
D

=
(7.2)
Dimensions of the prototype are marked with the star (*). A test facility scaled with
this law is a model of the prototype. The scale of the test facility is another issue and
is discussed in the next section. [19]
7.2 Structures
Structural design of the test facility is a very challenging task. Pre-calculations should
be made to evaluate forces on the core barrel. The scale of the facility should be opti-
mized using the results of these calculations and restrictions set by the instrumentation.
53
Table 7.1: Time-reducing scaling law [19]
Model/Prototype Time Reducing
Length
Time
Hydraulic diameter 1
Area
2
Volume
3
Velocity 1
Acceleration 1/
Properties (pressure, temperature, etc.) 1
Heat generation/unit volume/unit time 1/
Power
2
Rod diameter 1
Number of rods
2
Heat ux 1/
Wall temperature Distorted
Limitation Turbulent ow
Negligible buoyancy
The task is not the detailed designing of the test facility, but to give an overall view
what the test facility could be like and what structural decisions could be applied to
this case. One subject of the future studies could be the assessment of break opening
time eects. The most challenging task in this case would be the designing of such a
break assembly by which dierent break opening times could be produced and repro-
duced reliably. Code developers have expressed that the shape of the dynamic load
should be correct. Due to the time-reducing scaling law the break opening time should
be scaled with the dimension ratio (the length scale). This means smaller opening
time. However, the physical restrictions of actual break components set the value for
the smallest achievable break opening time (possibly 1 ms). This means that the con-
servative scenario of the LBLOCA may not be studied as the break opening time would
have to be much smaller than 1 ms in a small scale test facility. In the best-estimate
calculations the break opening time is assumed to be signicantly larger than 1 ms.
As far as the break opening time is concerned the facility could thus be used for best
estimate studies.
One problematic issue is the fact that the pressure wave induced by the break has
travelled already through the test facility before the break has even fully opened. This
is the case also with the best estimate break opening time in the real size plant. In the
small scale facility and with the scaled and therefore much smaller break opening time
that could be a problematic issue since the higher measurement frequency is needed
due to shorter propagation time. However, the rst test facility could be used to test
new structural ideas and measurement techniques, which could be later used in a larger
test facility.
54
7.3 Instrumentation
Some issues concerning the instrumentation must be solved if a new test facility is
decided to be constructed. Limited room for measurements, needed relatively high
measurement frequency and demanding conditions (high pressure and temperature)
are the main restrictions. The combination of instrumentation would be quite similar
compared with the HDR experiments. Pressures, temperatures, displacements, strains
and accelerations would be measured. As the object of these tests would be the gath-
ering of measurement data for uid-structure interaction code validation purposes, the
placement and type of instrumentation is a critical issue. Due to the scale of the facility,
there might not be enough space for all sensors needed. Therefore it is quite probable
that some compromises must be made in order to resolve the placement limitations
of instrumentation. Possibility of using high speed three-axis acceleration sensors to
replace displacement sensors used in the earlier test facilities would be worth of further
studying.
It must be remembered that the HDR experiments were carried out in the early
eighties. Measurement technology has developed through the years and nowadays
results probably are more accurate and measurements more reliable.
7.4 Experiments
The test facility could be used to study the eect of dierent parameters on the blow-
down forces. Investigated parameters could be the length and the diameter of the
blowdown pipe and the break opening time. Geometrical parameters can be varied
quite easily. However, the varying of the break opening time is the more challenging
task. Dierent opening times could be possibly produced by using dierent types of
rupture discs and changing the balancing pressure between the discs. Also at least in
one test series the degree of subcooling in the blowdown pipe could be varied.
55
8 Conclusions
A wide search for the blowdown phase test facilities has been carried out. It can be
concluded that the HDR experiments carried out in the early 1980s have been the most
advanced so far. There have also been separate tests to study the eects of the pipe
size and break opening time on the pressure loads. However, in the HDR test program
the eects of these parameters were not studied. Pressures inside the HDR core barrel
were measured along the central pole at dierent axial positions. It was concluded in
the articles that measured pressures represented well local pressures inside the core.
For CFD code validation purposes it would be better if there were more pressure mea-
surements and if they were also at dierent radial positions. Measurement technology
has improved since the HDR test program. Displacements and accelerations can be
measured nowadays with higher accuracy and more reliably. If the above mentioned
issues are found worth of further studying, a new test facility is then justied. This
could be the case if the conservative assumption of the break opening time is considered
to be too strict. Experiments with more realistic opening times are then needed and
phenomena could be studied even with a small scale test facility.
56
Bibliography
[1] R. Krieg, E.G. Schlechtendahl, and K.-H. Scholl. Design of the HDR experimental
program on blowdown loading and dynamic response of PWR-vessel internals.
Nuclear Engineering and Design, 43(2):419435, 1977.
[2] Reactor Safety Issues Resolved by the 2D/3D Programs. Technical Report
NUREG/IA-0127, GRS-101, MPR-1346, Joint Report, July 1993.
[3] Compendium of ECCS Research for Realistic LOCA Analysis. Technical Report
NUREG-1230, U.S. Nuclear Regulatory Commission, December 1988.
[4] H.K. Fauske. The Discharge of Saturated Water Through Tubes. Chem Engr
Prog Symp Ser, 61:210, 1965.
[5] M.R. Baum. Break opening times for axial rupture of a gas-pressurised pipe.
Nuclear Engineering and Design, 77:161167, 1984.
[6] OECD Nuclear Energy Agency. Proceedings of the Joint CSNI/CNRA Workshop
on Redening the Large Break LOCA: Technical basis and its implications, vol-
ume 2, August 2003.
[7] A.R Edwards and T.P. OBrien. Studies of the phenomena connected with the
depressurization of water reactors. Journal of the British Nuclear Society, 9:125
135, 1970.
[8] J. Bartak. A Study of the rapid depressurization of hot water and the dynam-
ics of vapour bubble generation in superheated water. International Journal of
Multiphase Flow, 16(5):789798, 1990.
[9] J.C. Rousseau. Blowdown experiments and interpretation, presented in European
two-phase ow group meeting (Haifa, Israel), June 1975.
[10] J.C. Rousseau and B. Riegel. Super CANON experiments, presented in OECD
Specialists Meeting on Transient Two-Phase Flow (Paris, France), June 1978.
[11] The Marviken Full Scale Critical Flow Tests, Description of the test facility. Tech-
nical Report MXC-101, December 1979.
[12] Nuclear Safety Quarterly Report November, December, 1968, January, 1969 for
Nuclear Safety Branch of USAEC Division of Reactor Development and Technol-
ogy. Technical Report BNWL-1009, Battelle Northwest, March 1969.
57
[13] R.N. Oehlberg. A Review of the EPRI Hydroloads Program. In Transactions
of 7th International Conference on Structural Mechanics in Reactor Technology,
1983.
[14] F. DAuria and P. Vigni. Evaluation of Fluiddynamic Loads on RPV Internals
During Blowdown. In Transactions of 10th International Conference on Structural
Mechanics in Reactor Technology, 1989.
[15] F.H. Chang, G.E. Santee, G.A. Mortensen, M.B. Gross, and T.B. Belytschko.
LOCA hydroloads calculations with multidimensional nonlinear uid-structure in-
teraction. Nuclear Engineering and Design, 70:335355, 1982.
[16] R. Gehrmann. Comparisons of Experimental Blowdown Data with the Results
of a Fluid-Structure Coupled and an Uncoupled Code. In Transactions of 7th
International Conference on Structural Mechanics in Reactor Technology, 1983.
[17] L. Wolf. Experimental results of coupled uid-structure interactions during blow-
down of the HDR-vessel and comparisons with pre- and post-test predictions.
Nuclear Engineering and Design, 70:269308, 1982.
[18] Appendix A of Compendium of ECCS Research for Realistic LOCA Analysis.
Technical Report NUREG-1230A, U.S. Nuclear Regulatory Commission, Decem-
ber 1988.
[19] A.N. Nahavandi, F.S. Castellana, and E.N. Moradkhanian. Scaling Laws for Mod-
elling Nuclear Reactor Systems. Nuclear Science and Engineering, 72:7583, 1979.
[20] W.L. Riebold. Loss of Coolant Accident. In Owen C. Jones, Jr., editor, Nuclear
Reactor Safety Heat Transfer, pages 273323. Hemisphere Publishing Corporation,
1981.
[21] W.L. Riebold, M. Reocreux, and O.C. Jones. Blowdown phase. In Owen C. Jones,
Jr., editor, Nuclear Reactor Safety Heat Transfer, pages 325378. Hemisphere Pub-
lishing Corporation, 1981.
[22] L.S. Tong and Joel Weisman. Thermal analysis of pressurized water reactors.
American Nuclear Society, 1996.
[23] Marie-France Robbe, S. Potapov, and F. Tephany. Simulation of the depressurisa-
tion occuring at the beginning of a LOCA in a 4-loop PWR. Nuclear Engineering
and Design, 224:3363, 2003.
[24] Joel Weisman. Elements of nuclear reactor design, chapter Loss-of-cooling acci-
dents, pages 371401. Elsevier Scientic Publishing Company, 1977.
[25] Fei Huang, Minoru Takahashi, and Leijing Guo. Pressure wave propagation in air-
water bubbly and slug ow. Progress in Nuclear Energy, 47(1-4):648655, 2005.
[26] R.W. Shumway. General features of Emergency Core Cooling Systems. In James H.
Rust and Lynn E. Weaver, editors, Nuclear Power Safety, Georgia Institute of
Technology Series in Nuclear Engineering, pages 281302. Pergamon Press Inc.,
1976.
58
[27] OECD Nuclear Energy Agency. OECD/NEA-CSNI Separate Eects Test
Matrix for Thermal-Hydraulic Code Validation Volume I. Technical Report
NEA/CSNI/R(93) 14, 1993.
[28] OECD Nuclear Energy Agency. Thermohydraulics of Emergency Core Cooling in
Light Water Reactors. Technical Report CSNI Report 161, 1989.
[29] Frank M. White. Fluid Mechanics. Mc-Graw-Hill, 2003.
[30] Yih-Yun Hsu. Review of critical ow rate, propagation of pressure pulse, and sonic
velocity in two-phase media. Technical Report TN D-6814, NASA, June 1972.
[31] F. DAuria and P. Vigni. Two-Phase Critical Flow Models. Technical Report
CSNI Report 49, Universit`a degli Studi di Pisa, Istituto di Impianti Nucleari, 5
1980.
[32] M.-S. Chung, S.-B. Park, and H.-K. Lee. Sound speed criterion for two-phase
critical ow. Journal of Sound and Vibration, 276:1326, 2004.
[33] R.E. Henry and H.K. Fauske. The Two-Phase Critical Flow of One-Component
Mixtures in Nozzles, Orices, and Short Tubes. Journal of Heat Transfer,
93(2):179187, May 1971.
[34] Michael Corradini. Fundamentals of Multiphase Flow. 1997.
[35] Thirasak Rirksomboon. Investigation of small scale bleve tests. PhD thesis, The
University of New Brunswick, March 1997.
[36] Seok Cho, Sang-Ki Moon, Se-Young Chun, Yeon-Sik Kim, and Won-Pil Baek.
Spacer Grid Eects during a Reood in an Annulus Flow Channel. Journal of
Nuclear Science and Technology, 44(7):967976, 2007.
[37] E.T. Tomlinson and D.L. Aumiller. An Assessment of RELAP5-3D using the
Edwards-OBrien blowdown problem. Technical report, Bechtel Bettis, Inc., 1999.
[38] OECD Nuclear Energy Agency. CSNI International Standard Problems (ISP),
Brief descriptions (1975-1999). Technical Report NEA/CSNI/R(2000) 5, 2000.
[39] OECD Nuclear Energy Agency. OECD/NEA-CSNI Separate Eects Test Ma-
trix for Thermal-Hydraulic Code Validation Volume II. Technical Report
NEA/CSNI/R(93) 14, 1993.
[40] H. Paillere, J. Garcia-Cascales, A. Kumbaro, D. Bestion, S. Mimouni, H. Stadtke,
G. Franchello, B. Worth, U. Graf, P. Romstedt, E.F. Toro, E. Romenski, H. De-
coninck, M. Ricchiuto, F. de Cachard, and B. Smith. Advanced three dimensional
two-phase ow simulation tools for application to reactor safety (ASTAR). In
FISA-2003 symposium, 2003.
[41] The Marviken Full Scale Critical Flow Tests, Summary report. Technical Report
MXC-301, December 1979.
59
[42] The Marviken Full Scale Critical Flow Tests, Measurement system. Technical
Report MXC-102, December 1979.
[43] Nuclear Safety Quarterly Report February, March, April, 1969 for Nuclear Safety
Branch of USAEC Division of Reactor Development and Technology. Technical
Report BNWL-1084, Battelle Northwest, June 1969.
[44] INEL. Resolutions of State of Idaho and Environmental Protection Agency com-
ments plus discussion of additional changes made in the work plan by DOE, March
1992.
[45] E.G. Schlechtendahl. Coupled Fluid-Structure Analysis for LWRs in the Federal
Republic of Germany. Nuclear Safety, 20:551563, 1979.
[46] Ted Belytschko and Ulrich Schumann. Fluid-structure interactions in Light Water
Reactor systems. Nuclear Engineering and Design, 60:173195, 1980.
[47] L. Wolf. Overview of the Experimental and Analytical Results for the HDR-RPV-I
Main Test Series (V31.2-V34). In Transactions of 7th International Conference
on Structural Mechanics in Reactor Technology, 1983.
[48] Special issue on uid-structure interaction and internal loading in thermal reactors.
Nuclear Engineering and Design, 70, 1982.
[49] E.D. Hughes and B.E. Boyack. TRAC-P Validation Test Matrix. Technical Report
LA-UR-97-3990, Los Alamos National Laboratory, September 1997.
[50] D. Yildiz, P. Rambaud, J. Vanbeeck, and J-M. Buchlin. A Study on the Dynamics
of a Flashing Jet. Final Contract Research Report EAR0030/2002, von Karman
Institute for Fluid Dynamics, 2002.
[51] Idaho National Engineering and Environmental Laboratory. Test Area North fact-
sheet.
[52] C. Addabbo and A. Annunziato. The LOBI Test Matrix.
60
APPENDIX I: Databases and resources
U.S. NRC, United States Nuclear Regulatory Commission
http://www.nrc.gov/
Agencywide Documents Access and Management System (ADAMS)
http://www.nrc.gov/reading-rm/adams.html
U.S. Department of Energy, Oce of Scientic & Technical Information
(OSTI)
http://www.osti.gov/
OECD Nuclear Energy Agency
http://www.nea.fr/
OECD NEA Data Bank
http://www.nea.fr/html/databank/welcome.html
International Association for Structural Mechanics In Reactor Technology
(SMiRT)
http://www.iasmirt.org/
Lappeenranta University of Technology, Nuclear Safety Research Unit
http://www.et.lut./yty/en/

You might also like