Professional Documents
Culture Documents
020
A new computer code, named CORONA (Core Reliable Optimization and thermo-fluid Network Analysis), was
developed for the core thermo-fluid analysis of a prismatic gas cooled reactor. The CORONA code is targeted for
whole-core thermo-fluid analysis of a prismatic gas cooled reactor, with fast computation and reasonable accuracy. In
order to achieve this target, the development of CORONA focused on (1) an efficient numerical method, (2) efficient
grid generation, and (3) parallel computation. The key idea for the efficient numerical method of CORONA is to solve
a three-dimensional solid heat conduction equation combined with one-dimensional fluid flow network equations. The
typical difficulties in generating computational grids for a whole core analysis were overcome by using a basic unit
cell concept. A fast calculation was finally achieved by a block-wise parallel computation method. The objective of the
present paper is to summarize the motivation and strategy, numerical approaches, verification and validation, parallel
computation, and perspective of the CORONA code.
KEYWORDS : Prismatic Core, Fuel Temperature, Fluid Flow Network, Gas Cooled Reactor, CORONA, VHTR
plication areas of CFD technology have been growing. of system codes such as MARS-GCR [16] and GAM-
Commercial CFD codes are user-friendly and well vali- MA+ [17]. It should be noted, however, that the system
dated against most industrial problems which have a single codes are targeted for system transients. Their meshes
phase fluid. In spite of the remarkable development of are too coarse for a detailed thermo-fluid analysis of a
CFD technology, however, a whole prismatic core CFD prismatic fuel block. The accuracy of the system code
analysis with geometric details is still a challenging task. calculations is not satisfactory for the core thermal design
Since fine meshes are required for accurate CFD calcula- of a prismatic reactor (e.g., hot spot analysis).
tions, most CFD applications are limited to local behaviours Therefore, a new computer code named CORONA
(e.g., single fuel column [12,13], and seven fuel columns has been developed to overcome the difficulties in the
[14]) in the design of a prismatic reactor. In 2009, Pointer available tools at KAERI, i.e., CFD codes and system codes.
and Thomas [15] published the STAR-CD results for a The CORONA code is intended for whole-core thermo-
whole core of a prismatic reactor. It is clear that such a fluid analysis of a prismatic gas cooled reactor with fast
CFD analysis requires tremendous computational efforts. computation and reasonable accuracy. Fig. 3 summariz-
Additional significant efforts have to be devoted to gen- es the thermo-fluid analysis codes and their application
erate computational grids of high quality for a good con- areas for the design of prismatic gas-cooled reactors at
vergence. Such efforts are valuable for a final calculation KAERI. The position of CORONA lies in between CFD
but are too taxing for a designer who wants a large number and system codes. Its position is similar to that of a sub-
of calculations with various design options. channel analysis code such as COBRA [18] used in a light
The other available option at KAERI could be the use water cooled reactor design.
Fig. 3. Thermo-fluid Analysis Codes and Their Areas for the Design of Prismatic Gas-cooled Reactors at KAERI
In order to achieve the target of the CORONA code Major applications of the CORONA code are (1) a
(i.e., fast computation of whole core thermo-fluid analysis steady-state hot spot analysis, (2) steady-state analysis for
with reasonable accuracy), three fundamental strategies off-design conditions such as a coolant channel blockage
were established and applied to the development of CO- accident, (3) some transient scenarios (e.g., rod ejection),
RONA: (1) an efficient numerical method, (2) efficient (4) a coupled analysis (e.g., with neutronics) for high
grid generation, and (3) parallel computation. fidelity, (5) design optimization calculations. The CO-
Efficient Numerical Method: RONA code is written in C++ and can be run on either
The use of an efficient numerical method is the most Windows or Linux operating system.
important requirement for a new code to achieve
its goal. The CORONA code adopts the existing
idea of the efficient numerical method used for the 3. NUMERICAL APPROACHES
design of MHTGR and HTTR (e.g., DEMISE and
FLOWNET/TRAMP). This efficient method solves 3.1 Grid Generation Using Basic Unit Cells
a three-dimensional heat conduction equation for a The first step in the numerical calculation of thermo-
solid like a CFD code and one-dimensional conser- fluid phenomena is to set up a grid. The unique feature of
vation equations for a fluid flow like a system code. the CORONA code is the use of the basic unit cells for
In 2012, Travis and El-Genk [19] verified the ef- the generation of computational grids. The concept of basic
ficiency of this combined method using STAR-CD unit cells provides an efficient generation of unstructured
calculations. computational grids to model an entire fuel block. The
Efficient Grid Generation: present authors noticed that the prismatic fuel blocks
For a whole prismatic core thermo-fluid simulation, have regular arrangement of hexagonal or pentagonal unit
grid generation can also be a time-consuming pro- cells as shown in Fig. 4. Prismatic fuel blocks can be repre-
cess. Therefore, an efficient grid generation method sented by several types of basic unit cells and are shown
using the concept of basic unit cells was developed in Fig. 5. The unstructured computational grids are defined
by the present authors [9,20]. This concept of the within the basic unit cells. Using the combination of the
basic unit cells adopted in the CORONA code ena- basic unit cells, unstructured grids can be efficiently gen-
bles the effective generation of unstructured grids erated to simulate the thermo-fluid behavior of the fuel
for prismatic fuel blocks without a special mesh blocks. Fig. 6 is provided in order to improve the under-
generator. The revision of the generated grids is standing of the relation between the basic unit cells and
also simple and convenient. the computational grids. It shows the assignment of the
Parallel Computation: basic unit cells to generate the computational grids of the
Reliance on a parallel computing technique is inevi- standard fuel block. The change in generated grids is also
table for a whole prismatic core simulation since sig- simple and convenient. Fig. 7 illustrates an example of
nificant numbers of meshes (e.g., in the order of ten computational grids generated using the basic unit cells.
million) are required. Parallel computation leads not Fig. 8 shows the modeling procedure from the basic
only to a speed up of the calculations but also allows unit cells to a whole prismatic core. Initially, computa-
the use of a normal personal computer (PC). tional grids are defined within each type of basic unit cell.
Fig. 6. Basic Unit Cell Assignment to Generate Computational Fig. 7. Example of Computational Grids Generated Using the
Grids of Standard Fuel Block Basic Unit Cells for the Standard Fuel Block
Then, using an axial combination of the basic unit cells, the CORONA code has a similar level as the CFD analysis.
a hexagonal (or pentagonal) body called a pin is made. However, it should be noted that the computational expense
Three-dimensional computational grids for a hexagonal of a solid heat conduction equation (i.e., Eq. (1)) is much
fuel (or reflector) block are generated using a two-dimen- less than that of fluid conservation equations in a whole
sional combination of pins. Modeling of a whole core core CFD analysis of a prismatic core.
is finally finished using a combination of hexagonal fuel
and reflector blocks. 3.2 Numerical Model for Solid
The number of computational grids for a solid used in In order to simulate the heat conduction through the
Fig. 8. Modeling Procedure from Basic Unit Cell to Whole Prismatic Core
complex geometry of the fuel block, the CORONA code orthogonal. At a fluid boundary, the following equation
solves the following three-dimensional heat conduction is governed:
equation:
(6)
(1)
Equations (3) and (6) can be rearranged into matrix form as
A fluid flow network model implemented into the CO- The first step to solve Eqs. (11) and (12) is to guess the
RONA code is based on the pressure correction method pressures at all nodes. These values are treated as prelimi-
[22~24]. One-dimensional conservation equation for a fluid nary values and donated by P. The correction P is de-
flow can be written as: fined as the difference between the correct pressure field
and the guessed pressure field P.
(8) (13)
Similarly, the volumetric flow rate correction Q and density
correction are defined as:
(14)
(9)
(15)
The density of the gas can be obtained using the ideal gas law.
(10) (16)
For a junction (j) having upstream fluid node (iuj) and down- Finally, the substitution of Eqs. (14), (15) and (17) into
stream fluid node (idj) shown in Fig. 9(b), the simplified the continuity equation (i.e., Eq. (11)) yields the following
momentum equation can be applied as equation for the pressure correction.
(12) (20)
(21)
(22)
Fig. 11. Result of Verification of CORONA Using Single Unit Fig. 13. Result of Verification of CORONA Using Seven Unit
Cell Problem Cell Problem
Fig. 12. Layout of Seven Unit Cell Problem and Its Boundary
Conditions
Fig. 14. Schematic of Kaburaki and Takizuka Experimental
Apparatus for Bypass Flow and Crossflow [25]
and the temperature of the outmost boundary was fixed.
Since it is believed that a commercial CFD code can solve
a multi-dimensional heat conduction problem accurately,
a commercial CFD code, CFX, is used for the verifica- 4.2 Bypass Flow and Crossflow in Fuel Column
tion of the CORONA code for the multi-dimensional heat An air flow experiment performed by Kaburaki and
conduction problem. Fig. 11 compares the predicted tem- Takizuka [26] was selected to validate the one-dimensional
perature profile by CORONA with that by CFX. It shows fluid flow network model implemented in CORONA [27].
that CORONA accurately solves the multi-dimensional Fig. 14 shows a schematic of their experimental appara-
heat conduction equation. It is obvious that a finer mesh tus. Four graphite blocks were stacked up in a column
provides a more accurate result. and surrounded by a steel shroud to simulate the bypass
The second case considers seven unit cells composed gap (~1.2mm) between fuel columns. The hexagonal fuel
of one coolant cell (designated as C) surrounded by six block has 12 coolant holes 20 mm in diameter. One arti-
fuel cells (designated as F) as shown in Fig. 12. A uniform ficial cross flow gap (~1 mm) was created in the middle
heat generation in the fuel compacts was assumed and of the column to simulate a crossflow gap between fuel
the temperature at the coolant boundary was fixed. In addi- blocks. The air at atmospheric pressure was drawn from
tion, the adiabatic boundary condition was imposed on the the top of the test section and flowed down the channels
outmost boundary of the unit cells. and the bypass gap. The bottom of the bypass gap was
Fig. 13 shows the predicted temperature profiles along completely sealed and the coolant leakage flow, which
Line A in Fig. 12. It was observed that when a fine mesh is entered the bypass gap, was merged into the main coolant
adopted, the accuracy of CORONA is close to that of CFX. channels through the crossflow gap. The mass flow rate
Fig. 17. Comparison of Predicted Axial Temperature Fig. 19. Comparison of Predicted Coolant Velocities at the
Distributions at the Center of the Hottest Fuel Compact for the Control Hole for the RSC Fuel Column Problem
Standard Fuel Column Problem
5. PARALLEL COMPUTATION
NOMENCLATURE
Fig. 23. Predicted Maximum Temperature within Each Fuel ai = Influence coefficient at node i, defined in Eq. (5)
and Reflector Column of PMR 600 ai* = Rearranged influence coefficient of ai, used in Eq. (7)
aj = Coefficient for junction j, used in Eq. (20)
a0p = Influence coefficient at node P, defined in Eq. (5)
ap* = Rearranged influence coefficient of ap, used in Eq. (7)
As = Solid surface area [m2]
bj = Coefficient for junction j, used in Eq. (20)
bp = Source term for node P at temperature matrix,
used in Eq. (7)
cj = Coefficient for junction j, used in Eq. (20)
Cj = Resistance coefficient at fluid node j, defined in
Eq. (18)
Cs = Specific heat of solid [J/kg K]
dj = Coefficient for junction j, used in Eq. (20)
D = Hydraulic diameter [m]
ei = Source term for node i at pressure matrix, used in
Eq. (20)
f = Friction factor [-]
g = Gravitational acceleration constant [m/s2]
Fig. 24. Result of Parallel Computation Performance Test G = Parameter defined in Eq. (19)
Using PMR600 Problem h = Fluid enthalpy [J/kg]
Ii = Initiating junction of node i
kf = Thermal conductivity of fluid [W/m K]
6. CONCLUSIONS AND OUTLOOK ks = Thermal conductivity of solid [W/m K]
K = Form loss factor [-]
The development of the CORONA code was started L = Fluid channel length [m]
in KAERI for the establishment of the key technologies N = Number of meshes
for the design of a prismatic VHTR. The CORONA code Nu = Nusselt number [-]
is targeted for a whole core thermo-fluid analysis of a P = Pressure [Pa]
prismatic gas cooled reactor with fast computation and Pr = Prandtl number [-]
reasonable accuracy. In order to achieve the target, the qs = Power density generated in solid [W/m3]
development of CORONA focused on (1) an efficient nu- qfconv = Convection heat transfer at fluid [W]
merical method using the combination of a three-dimen- qsconv = Convection heat transfer at solid [W]
sional heat conduction and one-dimensional fluid flow net- Q = Volumetric flow rate [m3/s]
work, (2) efficient grid generation using basic unit cells, and R = Gas constant of helium [J/kg K]
(3) parallel computing capability. Re = Reynolds number [-]
The results of the verification and validation studies sinon = Cross-diffusion term
show that the CORONA code can provide a reasonably t = Time [s]
accurate result for the thermo-fluid analysis of prismatic fuel Tf = Fluid temperature [K]
Gas-Cooled Reactor, Transactions of the Korean Nuclear [ 32 ] Intel Corporation, Intel MPI Library for Windows OS,
Society Autumn Meeting, Gyeongju, Korea, Oct. 27-28, Reference Manual, Rev. 4.0 Update 1, http://www.intel.
2011. com (2010).
[ 31 ]
C. K. Jo, Preliminary Result File of MASTER-GCR [ 33 ] J. Y. Cho et al., DeCART v1.2 Users Manual, KAERI/
Calculation for PMR600, 12BH_B4C_080_R48_MAS_ TR-3438/2007, Korea Atomic Energy Research Institute
AS_PF3_Case27_summary.dat (2013). (2007).