Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Even if some of CFD software developer and its user think that a state of the art CFD software can be used to solve reasonably at least the single-phase
nuclear reactor safety problems, there is still the limitations and the uncertainties in the calculation result.
From a regulatory perspective, Korea Institute of Nuclear Safety (KINS) is presently conducting the performance assessment of the commercial CFD
software for the nuclear reactor safety problems.
In this study, in order to examine the prediction performance of the commercial CFD software with the porous model in the analysis of the scale-down
APR+ (Advanced Power Reactor Plus) internal flow, simulation was conducted with the on-board numerical models in ANSYS CFX R.14 and FLUENT R.14.
Performance Assessment of the Commercial CFD Software for the Prediction of the PWR Internal Flow
Gong Hee Lee1*, Young Seok Bang1, Sweng Woong Woo1, Ae Ju Cheong1, Do Hyeong Kim2, Min Ku Kang2
1 Korea Institute of Nuclear Safety, Daejon, Republic of Korea
2 ANFLUX Inc., Seoul, Republic of Korea
RESULT & DISCUSSION
Streamline – lower plenum
Mass flow rate at core inlet plane
Frequency distribution of the mass flow rate at core inlet plane
Standard deviation of mass flow rate () in CFX was smaller than that in FLUENT. This means that CFX may predict the more uniform distribution of mass flow rate at core inlet plane.
Meanwhile FLUENT predicted the wider minimum/ maximum distribution of the mass flow rate at core inlet plane.
The measured core inlet flow rate was in the range of 86% to 126% of the averaged fuel assembly flow rate.
Both CFX and FLUENT predicted core inlet flow rate in the range of 41% to 145% and 26% to 154% respectively. Although there was a difference between the measurement and the prediction, CFX showed more agreement with the measurement in comparison with FLUENT.
SUMMARY
1. Depending on the CFD software the internal flow distribution of the scale-down APR+ was locally somewhat different.
2. Although there was a limitation in estimating the prediction performance of the commercial CFD software due to the limited number of the measured data, CFX V.14 showed the more reasonable predicted results in comparison with FLUENT V.14
3. CFD software suitable to the available computational resource should be selected for the massive parallel computation.
INTRODUCTION
Background
Objectives Test conditions
Either flow distribution tests for the scale-down reactor model or Computational Fluid Dynamics (CFD) simulations have been conducted to understand the complex thermal hydraulic characteristics inside reactor.
Although recently licensing applications supported by using the commercial CFD software are increasing, there is no commercial CFD software which obtains a licensing from the domestic regulatory body until now. In addition, there is no guideline for the comprehensive evaluation of CFD software.
From a regulatory perspective it is necessary to perform the systematic assessment and prepare the guideline for the prediction performance of the commercial CFD software.
In this study, in order to examine the prediction performance of the commercial CFD software with the porous model in the analysis of the scale-down APR+ (Advanced Power Reactor Plus) internal flow, simulation was conducted with the on-board numerical models in ANSYS CFX R.14 and FLUENT R.14.
The test matrix consists of the symmetric/asymmetric flow conditions for four-pump operation and the flow conditions for three-pump operation.
In this study, CFD simulation with the symmetric flow conditions for four-pump operation was conducted. In this condition, the Reynolds number was about 8.6105 in the downcomer.
NUMERICAL MODELING
Numerical method
Real geometry of flow skirt and beams in the lower support structure were considered.
Fuel assembly and some internal structures, for examples instrument nozzle support, fuel alignment plate, and upper plenum were considered as each simple bulky volume (porous domain) due to the limitation of computation resource.
Grid system
Boundary conditions
A hybrid mesh, made up of tetrahedron and prism, was generated to improve mesh resolution in the near wall region.
Total numbers of grid : 6.3107
Max. y+ (at downcomer) : 305
Flow passing through the flow skirt mixed in the reactor lower plenum and the flow velocities were relatively low in this region.
CFX predicted more mixed flow pattern in comparison with FLUENT near the reactor lower plenum.
CFX
1/5 scale-down model of APR+
APR+ Core Flow & Pressure Test Facility (ACOP) is a 1/5 scale-down model of APR+ and consists of a reactor pressure vessel with four inlet (cold leg) and two outlet nozzles (hot leg).
257 core simulators were installed in the model reactor to measure the hydraulic characteristics at the inlet and outlet of the fuel assemblies.
A number of the differential pressure transmitters were used to measure the core inlet flow rate and core outlet pressure distribution.
TEST FACILITY
Table Summary of scaling parameters
Turbulence model
Standard k- model was used.
Prediction performance of this model was equivalent to Shear Stress Transport (SST) model in the analysis of the turbulent flow inside the scale-down APR+ model.
Wall treatment : Scalable(CFX) or standard(FLUENT) wall function
Working fluid : Light water of 60 ℃
Inlet : Symmetric flow conditions for four-pump operation
• Flow rate : 135 kg/s at each cold leg
• Turbulence intensity : 5 %
Outlet : Average pressure over whole outlet option with the relative pressure of 0 Pa at each hot leg
Wall : No-slip condition Computational domain Grid system
Geometry modeling
Software : ANSYS CFX R.14 & FLUENT R.14
Steady, incompressible and turbulent flow
Convergence criteria : Residuals of variables were below 4×10-4 and the variations of target variables were small
Discretization scheme for the convection terms
• Momentum eqns. : High resolution scheme(CFX), 2nd order upwind scheme(FLUENT)
• Turbulence eqns. : 1st order upwind scheme
In order to reflect the velocity field and pressure drop occurring in the original flow region, porosity and an isotropic loss model were applied to porous domain.
Computational memory
Because of the difference of discretization methodology, FLUENT R.14 (node based discretization scheme) required more the computational memory (about 196%) than CFX R.14 (cell based discretization scheme) for the same grid system.
Therefore the CFD software suitable to the available computational resource should be selected for the massive parallel computation.
FLUENT