The paper focuses on the modeling features of ACON and the related validation work which includes the calculations of nearly 50 experiments performed in various test facilities. The validation methodology is discussed and the validation calculations are summarized as a validation matrix.
Trang 1Contents lists available atScienceDirect
Progress in Nuclear Energy journal homepage:www.elsevier.com/locate/pnucene
Containment model library of the Apros process simulation software: an
overview of development and validation work
Ari Sildea,∗, Jukka Ylijokia, Esa Ahtinenb
a VTT Technical Research Centre of Finland Ltd., VTT, P.O.Box 1000, FI-02044, Finland
b Fortum Nuclear Services Ltd., Fortum Power and Heat Oy, Keilalahdentie 2-4, 02150, Espoo, Fortum HQ Campus Keilalahti, Finland
A R T I C L E I N F O
Keywords:
Nuclear safety
Nuclear safety analyses
Containment modelling
Apros
Validation
Containment thermalhydraulics
A B S T R A C T The Apros CONtainment model library (ACON) is an add-on product of the Apros®(Advanced Process Simulation Environment) Nuclear software cooperatively developed by VTT and Fortum ACON is suitable for compre-hensive simulation of containment phenomena during nuclear reactor design basis accidents and, to some extent, severe accidents The lumped parameter approach applied enablesflexible modeling of various containment/ compartment systems ACON is a suitable tool for both safety analysis use and accurate training simulator purposes with real time calculation speed The Apros containment model can be used fully separately, or a containment simulation can be coupled with other thermal-hydraulic calculation to create a complete simulation model of a power plant, including e.g the reactor and turbine systems Modeling of relevant engineering safety features is also included
The paper focuses on the modeling features of ACON and the related validation work which includes the calculations of nearly 50 experiments performed in various test facilities The validation methodology is dis-cussed and the validation calculations are summarized as a validation matrix The paper provides a detailed presentation of selected validation cases, in which the main studied phenomena are related to general con-tainment thermal-hydraulics, spray effects, blowdown modeling, steam condensation on a structure, steam stratification in containment, and ice melting with associated natural circulation flow Finally, an example of applications is described Severe accident containment phenomena are out of the scope of this paper The results of the validation demonstrate that Apros can be used for analyses of containment thermal-hy-draulic behavior including related aspects of engineering safety systems in various containment geometries
1 Introduction
The main objective of the paper is to highlight the features and
validation process of the nuclear power plant containment modeling of
the Apros®Nuclear (Advanced Process Simulation Environment)
soft-ware developed in cooperation between VTT and Fortum Apros is a
commercial simulation software utilized in over 25 countries
world-wide (Apros, 2015; Silvennoinen et al., 1989) The Apros platform
provides an environment for configuring and running simulation
models of industrial processes, such as combustion and nuclear power
plants The Apros CONtainment library (ACON) is part of the Apros
Nuclear package (Fig 1)
The ACON library is developed mainly for analyzing containment
phenomena during nuclear reactor accidents, but the applied lumped
parameter approach also ensures aflexible modeling of various types of
containment/compartment systems outside the nuclear industry ACON
also includes the modeling capabilities for all relevant engineering safety features and accident management hardware One powerful characteristic of Apros is that the containment calculation can be cou-pled (integrated) with a complete simulation model of a power plant, including e.g the reactor, turbine and automation systems, with their interactions
The main period of development of ACON was during the end of the 1990s, but some code modifications and enhancements were also made later Basic verification of the ACON models was performed mainly by the code developers and involved different kinds of testing and code reviews The main validation process of ACON started in the early 2000s Nearly 50 calculation cases concerning various experiments in-cluding the separate effect, coupled effect and integral tests have been calculated so far (Silde, 2015) In addition, the validation and testing include several code-to-code comparison exercises/benchmarks
https://doi.org/10.1016/j.pnucene.2019.03.031
Received 8 November 2018; Received in revised form 26 February 2019; Accepted 17 March 2019
∗Corresponding author
E-mail address:ari.silde@vtt.fi(A Silde)
Available online 06 April 2019
0149-1970/ © 2019 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/BY-NC-ND/4.0/)
T
Trang 22 Overview of modeling features
2.1 General
The Apros containment code uses a so-called lumped parameter
approach The compartments/rooms of the simulated containment can
be divided into an arbitrary number of homogeneous control volumes
(nodes) connected byflow paths (branches) for steam-gas mixture and
liquid water A containment node consists of three separate phases: gas,
mist droplets and liquid water pool The mist droplets may be formed
due to volumetric condensation (fog), or from the liquid share due to
theflashing process of blowdown water Liquid droplets may also be
introduced by a boundary condition sources The mist droplets are
al-ways in a thermalequilibrium with the gas phase, whereas the water
pool may be in a thermal non-equilibrium state in which the pool mass
and enthalpy (temperature) are solved to determine the pool properties
2.2 Governing equations
The LP solution principle of ACON is a simplified form from the
approach used in the one-dimensional homogenous thermal-hydraulic
model of Apros (Hänninen, 1989) One simplification is that ACON does
not consider two-phaseflow, i.e the gas and liquid phases of the system
are solved separately and the interaction between phases takes place
only via heat and mass transfer processes through the interface inside
nodes
The simulated thermal-hydraulic system is described with the
dif-ferential equations for conservation of mass, momentum and energy
Because the gasflow is homogeneous, the equations are applied for the
mixtureflow, and therefore only three equations are used
∂
∂
Aρ
t
Aρw
(1)
∂
∂
Aρw t
A p
z S
(2)
∂
∂
Aρh
t
Aρwh
(3) where
A =flow area [m2
]
ρ = density of the mixture [kg/m3]
w = velocity of the mixture [m/s]
p = node pressure [Pa]
h = specific enthalpy of the mixture [J/kg]
t = time [s]
z = coordinate value [m]
Si, Sj, Sk= source term of mass, momentum and energy, respec-tively
The right-side terms of Eqs.(1)–(3)describe the sources of mass, momentum and energy In the mass equation, the source terms include the additional massflows into/from the system The source term of the momentum equation contains all pressure losses across theflow paths The enthalpy source term consists of all heatflows and the pressure derivative with respect to time The pressure derivative term appears in the enthalpy source term, because the enthalpy is used instead of the internal energy, i.e
∂
∂ =
∂
∂ −
∂
∂
u t
h
t v
p
where v is the specific volume [m3/kg] and u is the specific internal energy [J/kg]
Massflow rate in the junctions between the nodes is calculated from the momentum conservation equations A uniform temperature in each node is solved from the energy balances For pressure solution, also the mass balances are needed The pressure,flows, and enthalpies of the system are solve implicitly Because all the terms, such as material properties of water, steam and non-condensable gases cannot be cal-culated implicitly, the iteration procedure must be used One simplifi-cation of the LP system is that the convection term∂Aρw2/∂ztypical in the conservation equation of momentum of one-dimensional flow models is missing in Eq.(2) This simplification means in practice that the momentum offlows is not transferred across the node
In the implicit solution algorithm the pressures, flows and en-thalpies of theflow system are solved implicitly The commonly used staggered mesh discretization scheme is employed The integration method applied is the implicit Euler Because not all terms can be calculated implicitly, the iteration procedure has been used (Hänninen,
1989)
Fig 1 Main features of the Apros simulation environment
Trang 32.3 Modeled phenomena
The license of the basic ACON package includes the models
asso-ciated with the general containment thermal-hydraulics, hydrogen
be-havior and related engineering safety features (Silde and Ylijoki, 2017)
(Table 1) Particular severe accident models, such as the behavior of
aerosols and fission products, are available only in the Apros SA
module, which requires a separate license
2.3.1 Steam/non-condensable gas mixture thermodynamics
ACON calculates the thermodynamics of a gas mixture including
water vapor and six non-condensable gases (oxygen, nitrogen,
hy-drogen, helium, carbon monoxide and carbon dioxide) Air is
re-presented by a mixture of oxygen and nitrogen The gas space in each
node is perfectly mixed Steam is treated as a real gas and the
non-condensable gases comply the ideal gas law
2.3.2 Intercellflows
Theflow of gas and liquid water between adjacent nodes is
simu-lated by connecting the nodes with specific flow paths called branches
in the Apros terminology The gasflow is driven by the pressure
dif-ference and the buoyancy effect The flow loss coefficient may be a
constant value including all possible frictional and form loss terms
across theflow path, or alternatively, the loss coefficient can be
in-ternally calculated from a discharge coefficient according to the theory
of isentropic compressibleflow In the latter case, also chocked flow is
checked Gas flow may also carry fog droplets from one node to
an-other Flow of liquid water between the adjacent sumps is calculated by
the Bernoulli mechanical energy balance equation
Valve and pump components can be connected to branches in order
to drive and control both the gas and liquid waterflows The specific
model for suppression pool vent pipes calculates the vent clearing
processes including e.g the acceleration and movements of a water
plug inside the pipes
2.3.3 Heat and mass transfer The containment system includes various types of interfaces, where heat and mass transfer are of importance and should, therefore be modeled Heat transfer between containment and other thermal-hy-draulic system such reactor cooling system can also be modeled The principles of calculation for heat and mass transfer phenomena in ACON are similar at all gas–surface interfaces such as on a structure, ice, water pool, or droplets Total heat transfer rate is the sum of con-vective heat transfer, latent heat flow caused by condensation/eva-poration processes and radiative heat transfer The calculation is based
on Nusselt's theory using the heat and mass transfer analogy The Nusselt number for both natural and forced convectionflow is calcu-lated, and as a default the higher of the two values is used in the heat and mass transfer calculation However, the user is allowed to override the default assumption In the mass transfer modeling, the so-called mass diffusion theory, in which water vapor diffuses through the boundary layer and condenses on the surface, is applied Alternatively, the Uchida correlation is available
High vapor condensation rate reduces the thermal and mass layers
in size due to the suction effect of the condensation process (Corradini,
1983) The reduction in the boundary layer increases the heat and mass transfer coefficients, which are also taken into account in ACON using so-called Ackermann's approximate correction method (Ryti, 1968) In the case of a gas–liquid interface, such as on water pool and spray droplets, a separate interface temperature and its effect on steam partial pressure is iteratively calculated due to its strong effect on heat and mass transfer processes The principal numerical method used in the iteration of the interface temperature is the Secant method If the trials
of the Secant iterations are unsuccessful, subsequent trials are con-ducted using the Regula-Falsi method
At the gas–structure interface, the effect of condensate water film is taken into account as afilm resistance, or alternatively, its influence is calculated in more detail using so-called water tracking model, in which the waterfilm is allowed to flow down from one structure to another Heat transfer between aflowing water film and a structure (e.g wall condensate, waterfilm caused by external spray cooling) is calculated using the theory presented byCovelli et al (1982) The heat transfer calculation at the“stagnant” liquid–structure interface (e.g in a sump)
is based on Nusselt correlations for laminar and turbulent natural convection influid (Ryti, 1968b)
2.3.4 Sump and suppression pool
In the default approach, a water sump or a suppression pool consists
of a single-phase liquid having a uniform temperature However, a user can activate the so-called pool stratification model, in which the pool is divided into two different vertical layers which have their own mass and energy balances The heat transfer between the water layers takes place only through heat conductance Massflow between the layers takes place if the mass inventory or density of the lowest layer changes
2.3.5 Blowdown and other sources The ACON calculates the node thermodynamic conditions during blowdown release with the pressure flash model, which allows the blowdownfluid to flash into steam based on the total node pressure (Fig 2) The mass fraction offlashed steam is
′′ − ′
stm bld
(5) wherehbldis the specific enthalpy of fluid which enters the node, and ′′h
and ′h are the specific enthalpies of saturated vapor and liquid water in total pressure, respectively An input parameter defines that part of the liquid share, which is transferred directly to droplets (called the droplet fraction and marked as X inFig 2) The rest of the liquid share goes directly to the sump Because the droplets are assumed to be in thermal
Table 1
The most important phenomena modeled in ACON
Modeling phenomenon/system Remarks
Steam/non-condensable gas mixture
thermodynamics (water vapor, O 2 , N 2 , H 2 , CO 2 ,
He)
Water droplets (mist)
Intercell flow of gas, liquid sump water, droplets
Buoyancy effect in gas flow
Heat and mass transfer at different interfaces:
gas-structure, gas-ice, gas-sump, gas-droplets
Heat transfer at sump-structure interface
Thermal radiation heat transfer
Heat conduction inside heat structure
Condensate film flow on vertical structure
Ice condenser
Internal and external spray system
Water pool (sump):
homogenous or thermally stratified
Nodes with different shape For water elevation
calculation BWR suppression pool including vent pipes
Coupling of containment and other
thermalhydraulic calculation
Explicit sources and sinks: water vapor, liquid
water,
non-condensables, dry energy
Pump, valve, heat exchanger
Hydrogen combustion Discrete or continuous
Hydrogen recombiners AECL or Areva types
Hydrogen igniters Can be simulated with
discrete burning model Fission products and aerosols Available only in Apros SA
package General concentration solution for FP's and
boron
FP's available only in Apros SA
Trang 4equilibrium with the node atmosphere, their influence on node pressure
and temperature is minor However, if the node atmosphere tends to
become superheated, droplets evaporate and drive the humidity
to-wards the balance The default value of the droplet fraction in ACON is
0.2, which means that 20% of the liquid share of blowdown is
trans-ferred to droplets The default value is found to be suitable for most
typical blowdown situations A user may modify the droplet fraction
This is recommended e.g in a case when subcooled water flows
“slowly” from systems (which are in equal pressure) to the
contain-ment In this case, a droplet fraction of zero (or close to zero) may yield
the most reasonable results It is always recommended that a user
should check the result's sensitivity to the droplet fraction if there is a
source of liquid blowdown
2.3.6 Spray systems
Modeling of both internal and external spray systems is included
Operation of an internal spray system is simulated with the complete
mixing droplet model The droplet temperature is assumed to be
otherwise uniform, except that the temperature of an infinitely thin
surface interface is solved iteratively due to its strong influence on heat
and mass transfer Change of droplet size, temperature, material
properties, vertical and horizontal velocities and their effects on heat
and massflow are updated during the fall Five different classes with
different droplet sizes can be included As default, all spray water is
injected to the atmosphere, but a user-specified part of the spray water
may be injected directly onto structure surfaces Three different
in-ternal spray models with different accuracy levels are available
The external spray modeling covers a one-dimensional calculation
of energy balance of external water film on the dome outside the
structure, where the temperature, thickness, velocity and heat transfer
coefficient of the water film are determined as a function of angular
position of the semi-hemispherical dome structure (Covelli et al., 1982)
The evaporation of waterfilm is also considered
2.3.7 Ice condenser
The ice condenser (IC) modeling is based on the Westinghouse-type
system/design Ice is located in vertical cylindrical columns having a
user-specified diameter and height Ice is assumed to be at a uniform
temperature, i.e heat conduction inside the ice is not considered Both
axial and radial ice melting are modeled Different types of ice
con-denser doors are included The positive pressure difference across the
lower inlet, intermediate and top deck doors of IC pushes them open
The spring forces, gravity and inertia effect are also taken into account
in calculating the door movements
2.3.8 Node shape
If very exact calculation of water elevation inside a compartment is required, an approach in which the nodes have a constant cross section may not always be satisfactory in all geometries ACON has the cap-ability to define node geometry in three different ways: using a constant cross section, a varied cross section area as a function of node height and a varied node volume as a function of node height With these options, the water elevation in various containment geometries can be solved precisely
3 Validation process
3.1 General
Development of ACON started in the early 1990s when the principle models and the“heart” of the solution system were coded During the 2000s, the main work has proceeded by adding the modeling cap-abilities of engineering safety systems relevant to various containment types and geometries At this time, the preliminary validation process was also initiated Nearly 50 experiments performed in Finnish and international test facilities have been calculated so far In addition, about 10 code-to-code comparison cases have been carried out Most of the validation calculations are performed in the framework of the Finnish National Research Program, SAFIR, funded by the National Nuclear Waste Management Fund (VYR) and VTT
3.2 Methodology
Verification of ACON ensures that the program is coded properly and that it produces the intended results The code developers mainly perform the basic verification and preliminary code testing involving the test calculations and code reviews The aim of the validation cal-culations is to ensure and demonstrate that ACON has an appropriate capability to simulate the containment thermal-hydraulic phenomena and related effects of engineering safety systems and accident man-agement hardware during accidents/incidents
The ACON validation methodology contains two steps (Silde, 2015) Firstly, the selected experiments conducted in test facilities are calcu-lated in order to validate the code itself Secondly, the validity of changes made between the different code versions is checked by cal-culating always a certain set of experiments/transients with the new code version, and by comparing the results to those obtained by the earlier versions The comparison demonstrates how the new modifica-tions affect the code results and ensures that the changes made between the versions are valid and do not include any errors or other undesirable features The both steps of the validation process include the calcula-tions against the experimental data and the comparison calculacalcula-tions (benchmarks) against the results of well-validated other codes The ideal aim of the validation would be that all validation cases are cal-culated with all code versions Unfortunately, this is not possible due to limited resources and time Therefore, only the validation cases as-sumed to be the most representative are calculated in the version va-lidation process Some of the vava-lidation calculations are carried out as blind-calculations, i.e without knowing the experimental measure-ments beforehand, but the most calculations are carried out as open-exercises Both the code developers and pure code users have partici-pated in the validation process in two different organizations, at VTT and Fortum, in order to ensure that the validation calculations are performed independently and objectively One important aim of the validation process has been to provide essential exercise and experience
to young code users, needed in applying the code to real plant appli-cations
The choice of suitable values for some critical input parameters may have a remarkable influence on the simulation results The input values used in ACON validation calculations are mostly based on the best-es-timate approach If the best-esbest-es-timate values are not known, the default Fig 2 Principles of modeling blowdownflashing
Trang 5Table 2
Validation matrix of the Apros containment code
Experiment/benchmark Studied phenomena
CONAN experiment - Steam condensation rate on a duct wall
- Local heat fluxes to a duct wall
- Forced convection heat and mass transfer
EREC test no 1 - Pressure in bubbler condenser
containment
- Air mass concentration in SG box
- Gas temperatures in SG box, BC shaft,
BC air volume and air trap
- Water temperature in the water tray
- MELCOR comparison GEKO test series GEKO-E and
GEKO-F
- Total heat flow to the condenser IRSN CARAIDAS spray test - Droplet diameter as a function of falling
height (condensation/evaporation on droplets)
Marviken BWR experiment
MX-II (no 18)
- Drywell and wetwell pressures
- Gas temperature of drywell and wetwell
- Water pool temperature in wetwell
- Air and steam mass flow from
- drywell into wetwell MISTRA containment spray
experiments MASP-1 and
MASP-2
- Pressure
- Gas temperatures
- Steam concentrations
- Steam condensation MISTRA tests HM 2-1 and HM
3-2
- Gas temperature
- Helium concentrations (stratification)
- Effect of PARs on stratification MISTRA ISP-47 - Steam condensation on the walls
- Pressure
- Steam and helium concentration profile (stratification)
- Gas temperature profile (stratification) NUPEC experiment M-7-1
(ISP-35)
- Pressure
- Gas temperature
- Helium concentration in dome
- Spray effects PAR test calculation against the
AREVA data
- Efficiency of AREVA type recombiners PAR test calculation against the
AECL data
- Hydrogen concentration
- Gas temperature
- Recombination rate of PARs PPOOLEX test PCC-6 - Condensate flow rate in PCCS
- Flow rate through the NCG line
- Drywell and wetwell pressure
- Water temperature in the PCC pool
- Water temperature in wetwell pool
- Steam mass fraction in PCCS
- PCC pipe temperatures POOLEX tests 20 and
STB-21, PPOOLEX tests STR-9 and
STR-11
- Pool temperatures (stratification)
PPOOLEX test WLL-5-2 - Steam condensation on a wall
- Pressure
- Gas temperatures
- Wall temperature THAI experiment HM-2 - Pressure
- Gas temperatures
- Hydrogen concentrations (stratification) THAI experiment TH24 - Pressure
- Gas temperatures
- Steam concentrations (stratification)
- Dissolution of concentration
- Gas velocities PACOS Px1.2 test - Dome pressure
- Gas temperatures
- Inner wall temperatures
- Flow velocities
- Spray effects
Table 2 (continued)
Experiment/benchmark Studied phenomena PANDA test ST4.1 - Pressure
- Total cooling rate in cooler
- Total cooling power of the cooler
- Gas temperatures
- Steam and helium concentration profiles in vessel (stratification)
- Steam and helium concentrations in the cooler
- Condensate mass in the cooler
- Cooling water outlet temperature in the cooler tubes.
PANDA test T1.1 - Flow rate in PCCs feed line
- Total PCC condensate flow rate in the drain lines
- PCCS drum temperatures
- Drywell and wetwell pressures
- Drywell and wetwell gas temperatures
- Wetwell pool temperature
- Steam and helium concentrations in the drywells and wetwells.
PANDA ISP-42 - Drywell pressure
- Water mass in the wetwell
- Helium concentration in the drywell
- Pressure difference between the drywell and wetwell
- Liquid mass in the RPV
- Gas temperature distribution in the drywells and wetwells
- Wall temperature in drywell 1
- Water temperatures (stratification) in the wetwells
PPOOLEX test STR-4 - Drywell and wetwell pressure
- Drywell and wetwell gas temperature
- Suppression pool layer temperatures TOSQAN sump test T201 - Sump evaporation rate
- Pressure
- Gas temperature
- Pool temperature TOSQAN spray test 101 - Pressure
- Gas temperature
- Droplet size
- Droplet falling velocity VICTORIA no 13 - Pressure
- Gas temperatures
- Natural circulation flow
- Ice melting VICTORIA no 29 - Pressure
- Gas temperatures
- Natural circulation flow VICTORIA no 42 - Pressure
- Gas temperatures
- Natural circulation flow
- (Ice melting) VICTORIA no 50 - Pressure
- Gas temperatures
- Natural circulation flow COCOSYS benchmark - Function (movements) of ice condenser
doors COCOSYS benchmark - Licensing calculations for Loviisa IC
containment (LLOCA, SLB sequences) COPTA BWR benchmark - Pressure
- Gas temperature
- Suppression pool temperature HAMBO, GSIM benchmark - Pressure
- Gas temperature
- Suppression pool temperature
(continued on next page)
Trang 6values are used as often as possible If the input deck used yields
un-desirable calculation results, sensitivity runs are carried out in order to
determine the main reasons for the deviations
3.3 Validation matrix
The validation matrix of ACON is based on the Containment Code
Validation Matrix (CCVM) presented by the OECD/NEA/CSNI task
group (CSNI, 2014) CCVM describes a basic set of available
experi-ments and related phenomena suitable for code validation ACON's
validation matrix consists of the containment experiments categorized
as separate effect, integral, or combined effect tests according toCSNI
(2014) Certain separate phenomena are studied in the first type of
tests In the integral tests, the main aim is to investigate the integral
behavior of the system In the combined tests, the intention is to study
both separate effects and the integral behavior In containment tests,
the gap between the“separate effect” and “integral” is not always so
straightforward (CSNI, 2014)
Only those phenomena ranked “major” (the most important) in
CCVM are included in the ACON validation matrix.Table 2summarizes
the validation matrix of ACON describing the experiment under
con-sideration, the code version used for the validation case and the main
features/phenomena studied Also selected benchmark calculation
cases, in which the ACON results are compared to those of some other
codes are shown
3.4 Documentation
Extensive documentation is an important part of the validation
procedure The ACON model features, initial condition, relevant input
values used and the results of all validation calculations are
docu-mented in the research/project reports of VTT and Fortum The ACON
user's guide provides the instructions and hints needed in constructing
the input of the simulation model The code reference manual describes
the phenomenological (physical and chemical) models and related
equations implemented in ACON code (Silde and Ylijoki, 2017) A
successful code validation requires that the choice of physical model
options, default input values and used correlations including their
va-lidity ranges, are also justified and documented (Silde, 2004)
3.5 Examples of validation cases
This section presents the results of selected validation cases in-cluding calculations for tests performed in the simple one-room test facilities and in the multi-room geometries representative of BWR Mark
II, PWR large dry and PWR ice condenser containments The main studied phenomena of the described cases are related to general con-tainment thermal-hydraulics, spray effects, blowdown modeling, steam condensation on a structure, steam stratification in containment, and ice melting with associated natural circulationflow
3.5.1 Single-droplet spray tests at the IRSN CARAIDAS facility
An example of separate effect tests are the single-droplet spray tests conducted at the IRSN CARAIDAS facility (Malet and Vendel, 2009; Malet et al., 2011) The tests address the condensation and evaporation processes on mono-sized spray drops in a simple geometry Hence, the tests were focused on studying the droplet characteristics, not general thermahydraulic behavior The main aim of the Apros calculation was
to ensure that the mass transfer (condensation/evaporation) modeling
of the spray module is valid
3.5.1.1 Test arrangements The height and inner diameter of the cylindrical facility are 5 m and 0.6 m, respectively The atmospheric pressure, temperature and relative humidity, and also initial drop size have been varied test by test: 1… 5.4 bar, 20 … 141.6 °C, 3.0 … 87%,
295 … 673 μm, respectively Mono-sized spray droplets are injected into the top of the facility The test series consist of the evaporation and condensation tests In the evaporation tests, droplets are injected into
an atmosphere where the humidity is relatively low (20% or less), and droplets evaporate continuously as they fall In the condensation tests, cold droplets are injected into an atmosphere of high humidity Steam condenses on drops in the early stage of drop fall, and the drop size increases, whereas in the later stage during the fall the drops start evaporating and the droplet size decreases There are optical measurements of drop size at three different elevations downwards from the drop generator, i.e the net condensation/evaporation mass of droplets can be estimated Steady-state thermodynamic conditions and very good homogeneity along the height of the vessel were reached during the tests
3.5.1.2 Calculation model and assumptions Because the atmosphere in the tests was well mixed, use of one-node nodalization is justified A high vapor condensation rate reduces thermal and mass transfer boundary layers in size and the heat and mass transfer coefficients increase due to the suction effect of the condensation process (Corradini, 1983) By contrast, high evaporation rate decreases the coefficients These effects are considered in ACON by using the Ackermann's approximate method to correct the heat and mass transfer coefficients (Ryti, 1968) One aim of the calculations was also to check the validity of the correction method
3.5.1.3 Calculation results Comparison of calculation results to measured drop size in two evaporation tests, in which the evaporation rate was relatively low or high, is shown inFig 3 The X-axis of the Figure illustrates the distance from the nozzle.Fig 3shows calculations with and without Ackermann's correction The results indicate that when the vaporization rate was low, the droplet size was slightly overestimated and the evaporation rate was underestimated, particularly in the lower part of the facility (Z = 4.39 m) (Silde, 2011) In the high-evaporation cases (such as EVAP18), Apros predicted the droplet size extremely well at all elevations of measurements Furthermore, the disappearing of highly evaporated drops could be modeled satisfactorily If the evaporation rate was high, the best agreement was achieved when using Ackermann's correction, whereas in the low-evaporation cases the correction had only a minor influence on the simulation results
Table 2 (continued)
Experiment/benchmark Studied phenomena
SARNET 2 Generic Containment
benchmark
- Pressure
- Gas temperatures
- Relative humidity
- Hydrogen concentrations
- Pool water temperature
- Mass and heat transfer rate at pool surface
- Pool surface temperature
- Effect of radiation heat transfer
- Effect of heat structure nodalisation
- Gas flow pattern
- Hydrogen recombination rate in PARS
- Mass of steam and non-condensable gases
- Mass of fog droplets
- Effects of fog droplets on pressurisation
- Mass transfer to fog droplets due to bulk condensation
- Wall heat transfer contributions (convection, radiation, condensation)
- Heat transfer coefficient at wall and pool surfaces
- Wall temperature SUPLES benchmark - Pressure
- Gas temperature
- Blowdown modeling
Trang 7In condensation tests, a drop size increase due to condensation
oc-curred within a fall distance of about 0.5 m, after which the drops
started evaporating and the drop size decreased
Fig 4shows the calculation results in two condensation tests with a
low and a high condensation rate The general trend was that Apros
predicted the drop size very well at a short distance Z = 2.51 m from
the spray generator In the lower position, the drop size was slightly
overpredicted, because the drop evaporation rate was underpredicted
However, the simulation results were mostly within the error bar of the
measurements In the condensation tests, Ackermann's corrections had
no noticeable influence on the simulation results, because the mass
transfer rate was relatively small compared to that of the evaporation
tests Simulation results of the pure condensation phase could not be
compared extensively to the test data, since only one drop size
mea-surement was made in the part of the vessel where the condensation
occurred
In order to assess droplet behavior near the injection location with
the best possible accuracy, the use of a small system time step (of the
order of 0.1 s or less) was recommended
The results also leaded to the recommendation that the Ackermann's
correction should be always used in ACON simulations for spray cases
The overall conclusions of the calculations of spray tests at the
Caraidas facility and the large dry NUPEC test facility (Ylijoki et al.,
2018;Harti, 2005) were that ACON is able to model the basic physics of
spray droplet heat and mass transfer phenomena reasonably well, and
that the model is suitable for simulation of containment spray systems
in real plant applications
3.5.2 Liquid blowdown experiment MX-II (no 18) at Marviken facility The main aim of the validation task was to simulate overall thermal-hydraulic behavior in a large-scale containment geometry during blowdown in large break LOCA
3.5.2.1 Test arrangements The Marviken full-scale BWR test facility includes a reactor pressure vessel, a discharge pipe to the containment, drywell rooms of the containment building, a wetwell with the suppression pool and vent pipes leading the gas into the wetwell water pool (Fig 5) When the pressure in the drywell increases as a consequence of the primary coolant discharge, the steam-gas mixture flows from room 104 via four down flow channels to the vent pipe header (106) andfinally via vent pipes to the wetwell pool The total volume of the drywell is 1978 m3, the volume of the wetwell pool is
561 m3and the volume of the wetwell atmosphere is 1583 m3
3.5.2.2 Calculation model and assumptions The Marviken containment building consists of several partly separated compartments, thus forming a complex system of air-steam mixtureflow paths Therefore, the drywell in the simulation model is divided intofive separate volume nodes (Fig 6): DRY1, DRY2, DRR, DRY111 and DEAD The area of the
28 open vent pipes is 1.98 m2 The walls and other massive solid structures of the containment have been modeled with heat structures (Hänninen, 2003)
Pressure and liquid temperature in the pressure vessel were 46.6 bar and 237–259 °C, respectively The diameter of the discharge pipe was
280 mm and the duration of the blowdown was 170 s Total initial Fig 3 Droplet diameter in the low evaporation test EVAP13 (left) and in the high evaporation test EVAP18 (right)
Fig 4 Droplet diameter in the low condensation test COND1 (left) and in the high condensation test COND10 (right)
Trang 8amount of water in the wetwell suppression pool was 550 000 kg,
cor-responding to a submerged depth of 2.81 m of the vent pipes The
measured discharge mass flow and the corresponding enthalpy are
given as an input to the containment calculation (Fig 7)
3.5.2.3 Calculation results The multi-room geometry of the Marviken
plant made the simulation of the air and steamflows rather complex
However, the general time histories of the drywell and wetwell
pressures and temperatures were predicted quite well (Fig 8)
(Hänninen, 2003) During the simulation, it was found that the
pressure increase in the drywell and the wetwell was dependent on
how fast the drywell airflowed into the wetwell On the other hand, the airflow rate was dependent on the modeling accuracy of the drywell Regarding the airflow, it was particularly important how the gas flows to/from the dead-end room 124 (denoted as DEAD in Fig 6) were arranged The wetwell pressure was slightly overestimated in the calculation As long as the discharge was active, the calculated pressure difference between the drywell and the wetwell was slightly too low By increasing the pressure loss coefficient in the vent pipes, the pressure difference became larger but then the drywell pressure became too high The reason for the too low pressure difference may be the relatively simple modeling of the ventflow into the pool The complex 3-D phenomena and consequent losses at the vent pipe outlet were not taken into account in the LP containment modeling All steamflowing through the vent pipes is assumed to condense in the pool The airflow from the pool into the wetwell gas space has some steam content (humidity) corresponding to the saturation state at pool temperature
As in the case of pressures, the temperatures represent those in the volume DRY1 and in the room 111 The calculated drywell temperature was after thefirst 10 s very close to the measured data (Fig 8) The calculated wetwell gas temperature increased much faster than the measured temperatures, but later on remained below the measured data The too fast temperature increase in the wetwell at the beginning indicates the problem of the lumped parameter model The use of the averaged quantities in the control volumes causes the too-fast spreading
of the hot air-steam mixtures By using a denser nodalization, the re-sults could be improved slightly, but the basic problem remains As a parametric study, the transient was modeled so that the vent pipe header was separated from the volume DRY3 to its own control volume
In this case, the temperature of the header increased somewhat more slowly than in volume DRY1, but it had no effect on the overall con-tainment behavior The temperature in the wetwell still behaved as in Fig 8
The calculated pool temperature remained on a clearly higher level than the measured value throughout the experiment, which implies that the calculated energyflow through the vent pipes to the pool, parti-cularly in the beginning of the transient, must be higher than in an actual situation The reason for the overpredictedflow rates is assumed
to be the too-fast mixing tendency of the containment model numerical solution The LP approach used assumes full mixing properties in every calculation volume The steam released in the blowdown spreads throughout the drywell and through vent pipes into the wetwell much too fast The spreading of steam can be restricted somewhat with denser nodalization, but the basic problem of the numerical solution remains The conclusion of the validation calculation was that the complexity
of Marviken containment makes the simulation of air and steamflow rather difficult using the LP approach Accumulation and purging of air Fig 5 Outline of the Marviken containment (Marviken, 1977)
Fig 6 Simulation nodalization of the Marviken containment (Hänninen,
2003)
Trang 9from the dead-end compartment above the pressure vessel had a great
influence on the pressure behavior of the drywell and wetwell
However, the Apros-SUPLES benchmark (Hänninen et al., 2003) and
the Marviken validation calculation indicated that the blowdown
modeling of Apros is sound and the code works reliably in the
sup-pression pool applications
3.5.3 Steam condensation on the wall: PPOOLEX test WLL-5-2
The main objective of the validation task was to check that the
steam condensation and heat transfer to a wall structure is modeled
correctly in ACON Furthermore, the capability to model the general
thermal-hydraulics in a simplified suppression pool geometry was
stu-died
3.5.3.1 Test arrangements The POOLEX test facility is located at
Lappeenranta University of Technology (LUT) in Finland (Fig 9)
(Laine et al., 2008) The primary component of the test facility is a
cylindrical stainless steel vessel with a free volume of 31 m3 The
cylinder is divided into the drywell and wetwell compartments,
separated by an intermediate deck The free volume of the drywell is
13 m3 The facility also includes a suppression pool system with a vent
pipe The steam condensation was measured by collecting condensate
in two gutters, located on different vertical positions of a drywell wall
In test WLL-5-2, a relatively constant steam injection rate
(470–550 g/s) takes place for 240 s The injected steam is saturated,
having an injection pressure of 6.5 bar and a specific enthalpy around
2680… 2730 kJ/kg
The facility was dried out before the test by blowing hot dry air through the facility Because the initial humidity in the facility was not measured, the Apros calculations included some sensitivity studies with Fig 7 Measured discharge massflow (left) and specific enthalpy (right) (Hänninen, 2003)
Fig 8 Drywell and wetwell pressures (left) and gas temperatures (right) in the Marviken blowdown experiment (Hänninen, 2003)
Fig 9 The PPOOLEX facility (Laine et al., 2008)
Trang 10varied humidity.
3.5.3.2 Calculation model and assumptions A simple three-cell
nodalization is used in the simulation: one node for drywell and
wetwell, and one node representing an environment to model heat
losses there (Luukka and Silde, 2010) A suppression pool including one
vent pipe is modeled in the wetwell Three gasflow paths exist between
the drywell and wetwell: a vent pipe, a vacuum breaker and a leakage
hole in the intermediate deck door The liner of the wall, ceiling,floor,
andflange, and the lumped mass of pipe connections and valves, etc.,
are modeled as heat structures
3.5.3.3 Calculation results From the validation point of view, the most
important measured variable of this test was the amount of condensate
water collected from the lower wall segment of the drywell to the gutter
(Fig 10) Because the initial humidity was not measured in the test,
three Apros simulations were performed with a varying initial humidity
of the drywell The facility was dried out before the test, and hence, the
low humidity value was considered to represent the most realistic
value The results inFig 10show that Apros simulated the condensate
mass rather well The best agreement was obtained with very low initial
humidity (1%) Generally speaking, the initial humidity appeared to
have only a small influence on the condensate mass
The initial humidity determines the initial mass of steam and air in
the facility The mass of air did not change during the test Therefore, as
the initial humidity is higher, the initial air mass is lower and also the
partial pressure of non-condensable air remains lower This effect can
clearly be seen in Fig 11, in which the calculated drywell pressures
with varied initial humidity levels are compared to the measurements
Best agreement was obtained once again assuming 1% initial humidity
A general conclusion of this validation task is that Apros heat and
mass transfer modeling on a wall structure works well and gives reliable
results Similar conclusions were obtained also in the ACON calculation
of steam condensation test ISP-47 at the MISTRA facility (Silde, 2007)
The greatest deviation between the simulation and measurements was
observed in wetwell gas temperature, which increased too fast in the
simulation The reason for this was probably the same as in the
simu-lation of the Marviken test no 18 described above: the calculated
en-ergyflow through the vent pipe(s) to the pool, particularly at the
be-ginning of the transient, is probably higher than in the experiment
3.5.4 Steam stratification at the THAI facility (test TH24) The goal of the THAI tests of the series TH24 was to study the dissolution of steam stratification under the presence of natural con-vection (Freitag et al., 2016) The tests provided data for both CFD and
LP models in order to develop simulation capabilities in a containment atmosphere of nuclear reactor containments The benchmark exercise was of special interest, because it included both the blind and open calculations VTT participated in the blind exercise using the Apros code (ACON) The main aim of the Apros simulation was to study the capability of the LP code to model very challenging stratification and dissolution processes by utilizing the experiences gained from the re-levant previous exercises
3.5.4.1 Test arrangements The THAI test vessel has volume, height and diameter of 60 m3, 9.2 m and 3.2 m, respectively (Fig 12) (Freitag
et al., 2016) The steel vessel is thermally insulated and the walls are equipped with heating/cooling mantles in three vertical sections The vessel space also includes an open inner cylinder A sump compartment
is located at the bottom of the vessel
The test TH24 is preceded by the preheating phase, in which the vessel pressure and gas temperature increase to 1.2 bar and around
90 °C, respectively During the main steam injection phase, which takes place for 500 s saturated steam (35 g/s) is injected into the upper part of the vessel where a stratification layer is evolved Due to steam con-densation on the wall and to ensure isobaric conditions beyond the main steam injection phase, the steam injection continues at a low rate
of 3.8 g/s and the steam stratification layer was later mixed by a thermal convection induced by heating of the lower and middle heating mantles on the wall The upper mantle is simultaneously cooled The natural circulation motion is upwards near the wall and downwards inside the inner cylinder of the vessel (Fig 12)
3.5.4.2 Calculation model and assumptions A specific “pseudo-3D” nodalization concept has been developed for ACON to capture the main natural circulation flow path with associated stratification phenomena (Fig 13) The vessel is modeled by 28 vertical node levels, each of which is further divided into 5 horizontal nodes The lowest and highest parts of the vessel are modeled by own nodes Due to the certain simplifications of numerical solution of an LP code, it is impossible to model a forced convection steam jet directly However, Fig 10 Condensate mass in the lower gutter on the drywell wall in the PPOOLEX test WLL-5-2 (Luukka and Silde, 2010)