This abstract contains information of PC-Cluster systems that Samsung Advanced Institute of Technology (SAIT) have developed and researched since 1998. This paper introduce not only the reasons that why SAIT has been interested and focused in such system, but also where to SAIT has applied. There are many other applicable areas but this paper emphasis only in the fields of high performance computing and storage system. With the result of SAIT's PC-Cluster research, we hope we can create many new applicable areas with collaborations.

Flow in the blood sac of the Korean artificial heart is numerically simulated by finite element method. Fluid-structure interaction algorithm is employed to compute the 3D blood flow interacting with the sac material. The motion of the actuator is simplified by a time-varying pressure boundary condition imposed on the outer surface of the sac. Numerical solutions show that there are a strong flow into the outlet and a stagnation flow near the inlet during systole. Shear stress distribution is also delineated to assess the possibility of thrombus formation.

For the large scale computation of turbulent flows around an arbitrarily shaped body, a parallel LES (large eddy simulation) code has been recently developed in which domain decomposition method is adopted. METIS and MPI (message Passing interface) libraries are used for domain partitioning and data communication between processors, respectively. For unsteady computation of the incompressible Wavier-Stokes equation, 4-step splitting finite element algorithm [1] is adopted and Smagorinsky or dynamic LES model can be chosen fur the modeling of small eddies in turbulent flows. For the validation and performance-estimation of the parallel code, a three-dimensional laminar flow generated by natural convection inside a cube has been solved. Then, we have solved the turbulent flow around MIRA (Motor Industry Research Association) model at $Re = 2.6\times10^6$, which is based on the model height and inlet free stream velocity, using 32 processors on IBM SMP cluster and compared with the existing experiment.

This paper describes numerical research on transverse jet behind rearward-facing step in turbulent supersonic flowfields without chemical reaction. The purpose of transverse jet behind rearward-facing step is to improve mixing of the fuel in the combustor. Two-dimensional unsteady flowfields generated by slot injection into supersonic flow are numerically simulated by integration of Navier-Stokes equation. Final-scale turbulence effects are modeled with two-equation $\kappa-\epsilon$ model. Numerical methods are modeled high-order upwind TVDschemes. A total of 4 cases are computed, comprising slot momentum flux ratios at four step heights downstream of the step. These numerical results are represented periodic phenomenon in unsteady flowfields.

TNumerical simulations of the supersonic impinging jet flows are carried out using the 3D Navier-Stokes code. This paper is focuses on the unsteady flow features associated with stagnation bubbles and other oscillatory behavior. The 3D code was validated by reproducing the results of Lamont's experiments. Computation is carried out for the cases in which the unsteadiness of the plate shock has been observed experimentally. The computational results confirm the oscillatory feature in several kHz. Unsteady calculation with algebraic turbulence model is also performed. It is found that the laminar and turbulent results have some discrepancy in the transient period. However, both of them reveal the oscillatory behavior with similar frequency.

TIn this study, the counter-jet flows which designed for improvement of aerodynamic performance of the blunt body vehicle have been analyzed. The variations of the drag force and jet penetration depth due to changes in the stagnation properties of counter jet new such as total pressure, mach number, and total temperature. The counter jet flow, which is injected toward incoming supersonic freestream at stagnation region of blunt cone-cylinder vehicle, have been studied by using upwind flux difference splitting navier-stokes method. The changes in the stagnation pressure and Mach number resulted in large effects on the wall pressure and drag force, on the other hand tile total temperature changes did not.

The dynamic behaviors of the single vortex in a reacting and non-reacting methane-air jet flow were investigated numerically. The numerical method was based on a predictor-corrector for low Mach number flow A two-step global reaction mechanism was adopted as a combustion model. After fuel and air were developed entirely in computational domain, the single vortex was generated by an axisymmetric jet that was impulsed to emit a cold fuel. Through comparisons of single vortex in reacting and non-reacting jet flow, it was found that global dynamic behaviors and the mechanisms leading to the formation, transport processes of vortex ring were influenced significantly by heat release from reaction. In addition, the interaction between a single vortex and flame bulge generated by buoyance effect in a reacting jet flow was found.

Computational study has been undertaken to investigate the aerodynamic influence of side jet on a supersonic missile and to find a similarity condition between the flight condition and the wind tunnel testing. Tasks were peformed to validate the existing Raytheon test body with side jet, to simulate the flow inside the supersonic wind tunnel, and to compare the flow fields between the missile in free flight and that in the wind tunnel. Then sub-scale model of body-tail configuration was analyzed to estimate the influence of the side jet on the missile components. It is found that the influence of side jet is not as significant on the tail region as on the body surface and a simple algebraic formula for aerodynamic coefficients accounting for the side jet as a point force may be cautiously utilized in setting up control logic.

The supersonic flow around multi-stage rocket system is analyzed using 3-D compressible unsteady flow solver. A Chimera overset grid technique is used for the calculation of present configuration and grid around the core rocket is composed of 3 zones to represent fins in the core rocket. Flow solver is parallelized to reduce the computation time, and an efficient parallelization algorithm for Chimera grid technique is proposed. AUSMPW+ scheme is used for the spatial discretization and LU-SGS for the time integration. The flow field around multi-stage rocket was analyzed using this developed solver, and the results were compared with that of a sequential solver The speed-up ratio and the efficiency were measured in several processors. As a result, the computing speed with 12 processors was about 10 times faster than that of a sequential solver. Developed flow solver is used to predict the trajectory of booster in separation stage. From the analyses, booster collides against core rocket in free separation case. So, additional jettisoning forces and moments needed for a safe separation are examined.

A three-dimensional incompressible Navier-Stokes solver is developed for the flow analysis around Micro Air Vehicle(MAV) designed by MACDL(Micro Aerodynamic Control and Design Lab), Seoul National Univ., Validations of this solver are presented for two cases, first flow over the circular cylinder with infinite length, second flow over infinite wing with wing section, E387 airfoil. Simultaneously, Wind Tunnel test is performed with Flatform Wire type sir-component balance and model designed by MACDL. The numerical results are also examined through comparison with experimental data.

In this paper, the preconditioned multistage time stepping methods which are popular multigrid smoothers is implemented for the compressible Navier-Stokes calculation with full-coarsening multigrid method. The convergence characteristic of the point-Jacobi and Alternating direction line Jacobi(DDADI) preconditioners are studied. The performance of 2nd order upwind numerical fluxes such as 2nd order upwind TVD scheme and MUSCL-type linear reconstruction scheme are compared in the inviscid and viscous turbulent flow caculations.

Cross-Flow Fan(CFF) are widely used lot industrial equipments and household electric appliances. A design method for CFFs, however, has not been well established because of the complexity of the internal flow. Numerical analysis was performed by using STAR-CD. In this study present the internal flow of CFF, which has varies pin number, and their flowrate were compared

All modem, aerospace gas turbines must operate with hot stage gas temperature several hundreds of degrees hotter than the melting temperatures of the materials used in their construction. Complicated cooling schemes need to be employed in the combustor walls and In the high pressure turbine stages. Internal passages are cast or machined into the hot sections of aero-gas turbine engines and air from the compressor is used for cooling. In many cases, the cooling system is engineered to utilize jets of high velocity air, which impinge on the internal surfaces of the components. They are divided by Impinging cooling method and Vortex cooling method. Specially, Research of new cooling system(Vortex cooling method) that overcome inefficiency of film cooling and limitation of space. The focus of new cooling system that improve greatly cooling efficiency using quantity's cooling air which is less is set in surface heat transfer elevation. Therefore, In this study, the numerical analysis have been performed for characteristic of flow and thermal in the swirl chamber and compared with the flow field measurement by LDV. especially, for understanding of high heat transfer efficiency in vicinity of wall. we considered flow structure and mechanism of vortex and heat transfer characteristic in variation of Reynolds number.

The present work analyzed the effect of mixing vane shape on the flow structure and heat transfer downstream of mixing vane in a subchannel of fuel assembly, by obtaining velocity and pressure fields, turbulent intensity, flow-mixing factors, heat transfer coefficient and friction factor using three-dimensional RANS analysis. NJl5, NJ25, NJ35, NJ45, which were designed by the authors, were tested to evaluate the performances in enhancing the heat transfer. Standard $\kappa-\epsilon$ model is used as a turbulence closure model, and, periodic and symmetry conditions are set as boundary conditions. The flow blockage ratio is kept constant, but the twist angle of mixing vane is changed. The results with three turbulence models( $\kappa-\epsilon$, $\kappa-\omega$, RSM) were compared with experimental data.

In the present research, a parametric study of aerodynamic characteristics for multi-element airfoils is performed. The major geometric parameters of interest are the gap distance between airfoils and relative deflection angle of slat/flap. The present results are mainly obtained by using inviscid flow calculation, and the aerodynamic characteristics are focused on the surface pressure distribution and the lifts. The results of the present research may be used as not only qualitative data but also quantitative data for small angle of attack flows, where the viscous effect does not play major role in terms of surface pressure distribution and lifts. A further research in this subject including viscous calculation and more geometric parameters is to be performed in the future.

The steam turbine rotor blade is designed using the Turbine Rotor Design Package developed by the authors. It can quickly accomplish blade shape design in the power plant industry. The quasi-3d code is employed for analysis of passage flow in the blade sections. Iterative change of each blade shape is made by moving position of control points in the Bezier curve under GUI(graphic user interface) environment. The full 3-D blade shape is obtained by stacking of the section blades.

Numerical analysis has been conducted in order to simulate the flow characteristics by tip clearance of turbine cascades. A 3-D Navier-Stokes CFD code based on body-fitted coordinate system, pressure-correction and finite volume method has been used along with a commercial CFD code. The present results have showed that the development and generation of leakage vortex, vortex within tip clearance, etc. are clearly simulated, consistent with the generally known tendency. The leakage vortex occurs mainly by a separation of leakage flow that arises due to a pressure difference between two surfaces of the blade at the tip.

The effect of nose radius on aerodynamic heating are investigated by using the Wavier-Stokes code extended to thermochemical nonequilibrium airflow. A spherical blunt body, whose radius varies from 0.003048 m to 0.6096 m, flying at Mach 25 at an altitude of 53.34 km is considered. Comparison of heat flux at stagnation point with the solution of Viscous Shock Layer and Fay-Riddell are made. Obtained result reveals that the flow chemistry for very small radius is nearly frozen, and therefore the contribution of heat flux due to chemical diffusion is smaller than that of translational energy. As the radius becomes larger, the portion of diffusion heat flux becomes greater than translational heat flux and approaches to a constant value.

A two-dimensional direct numerical simulations was peformed to investigate the flame structure of $CH_4/N_2$-Air counterflow nonpremixed flame interacting with a single vortex. The detailed transport properties and a modified 16-step augmented reduced mechanism based on Miller and Bowman's detailed reaction mechanism were adopted in this calculation. To quantify the strain on flame induced by a vortex, a scalar dissipation rate (SDR) is introduced. Results show that the fuel- and air-side vortex cause an unsteady extinction. In this case, the flame interacting with a vortex is extinguished in much larger SDR than steady flame. It was also found that air- side vortex extinguishes a flame more rapidly than fuel -side vortex.

The transient flow fields in direct injection engine was analyzed by using the STAR-CD CFD code during the intake/compression processes. The grids are generated by using the IC3M. The CFD results were compared with experimental data. The results showed that the used techniques were well suited for the flow analyses on any internal combustion engines.

The heat transfer characteristics of a turbulent flow in a ribbed two-dimensional channel have been investigated numerically. The fully elliptic governing equations, coupled with a four-equation turbulence model, $\kappa-\omega-\bar{t^2}-\epsilon_t$, are solved by a finite volume method of SIMPLE type. Calculations have been carried out for three rib cross-sections : square, triangular, and semicircular, with various rib pitches and Reynolds numbers. The procedure appears to be satisfactory as the results for the square rib compare favorably with available experimental data and earlier calculation. The optimal rib pitch that yields the maximum heat transfer has been identified. It is also found that the square rib is most effective in enhancing the heat transfer. The semicircular rib, on the other hand, incurs the least amount of pressure drop but the improvement in heat transfer is substantially lower.

This study is focused on investigating the characteristics of internal flow of the polymer reactor and its effect on the polymer quality. Four types of polymer reactor which have different kind of impeller, baffle and operation condition were calculated by CFD. Fluent 6 have been used to simulate mixing phenomena of reactor. According to the comparison of computational results and SEM photographs of polymer particle, distribution of turbulent dissipation rate greatly influences on the quality of polymer. So, distribution of turbulent dissipation rate to be important criterion to predict polymer quality.

The purpose of this 3-D numerical simulation is to calculate and examine a 500 kW Horizontal Axis Wind Turbine (HAWT) power performance and compare to calculation data(BEM method) from Delft University. The experimental approach, which has been the main method of investigation, appears to be reaching its limits, the cost increasing relate with the size of wind turbines. Hence, the use of Computational Fluid Dynamics (CFD) techniques and Navier-Stokes Solvers are considered a very serious contender. We has used the CFD software package CFX-TASC flow as a modeling tool to predict the power performance of a wind turbine on the basis of its geometry and operating data. The wind turbine with 40m diameters rotor, it was scaled to compare with the calculation data from delft university. The HAWT, which has eight-rpm variations are investigated respectively. The pitch angle is $+0.5^{\circ}$ and wind speed is fixed at 5m/s. The tip speed ratio (TSR) of the HAWT ranging from 2.89 to 9.63.

Characteristics of flow rate control has been studied for a cavitating venturi adopted in a liquid rocket propellant feed system. Numerical simulation has been peformed to give about $10\%$ discrepancy of mass flow rate to the experimental data for cavitating flow regime. Mass flow rate is confirmed to be saturated for pressure difference higher than $3\times10^5$pa when the upstream pressure is fixed to $22.8\times10^5$pa and the downstream pressure is varied. The evaporation amount depends substantially to non-condensable gas concentration. However the mass flow rate characteristic is relatively insensitive to the mass fraction of non-condensable gas. So it is reduced by only $2\%$ when the non- condensable gas concentration is increased from 1.5PPM to 150PPM. From the previous comparison the expansions of the non-condensable gas and the evaporation of liquid are verified to have same effect to pressure recovery.

전산유체역학분야의 고속 연산을 위해서 병렬처리가 보편화되고 있으며 이러한 병렬해석은 주로 클러스터에서 저렴한 비용으로 수행되고 있다. 전산유체역학을 위한 클러스터 컴퓨터에서의 해석프로그램의 성능은 클러스터에 사용되는 프로세서의 성능뿐만 아니라 클러스터 내부의 통신 장비의 성능에 크게 좌우된다. 본 논문에서는 클러스터 컴퓨터의 구축에 널리 사용되고 있는 Myrinet2000, Gigabit Ethernet, Fast Ethernet 등의 네트웍 장치에 대해서 Netpipe, Linpack, NAS NPB, 그리고 MPINS2D Navier-Stokes 해석프로그램을 사용하여 성능을 비교하였다. 이를 통해서 향후 전산유체역학을 위한 클러스터 구축시 최대의 가격대 성능비를 얻을 수 있는 방법을 제시하고자 한다.

In the present study a 3D data visualization and analysis program with stereoscopic viewing is introduced. The GUI of the program is based on Qt-library, while all the graphic rendering is performed with OpenGL library. The program allocates memory dynamically according to the data size so that the problem size is only limited by the computer's hardware memory. The stereoscopic viewing is realized by carefully-calibrated projection and color-masking of red and blue color for the left and right eye, and the only hardware needed for the stereoscopic visualization of 3D data is a cheap and easily-available red/blue glasses. Further work for addition of more functions and options to the present program will be continued.

Numerical analysis of the flowfield around a 50-meter class airship is performed to determine the optimal position for the airspeed probe installation. The turbulent flow around the hull with gondola is analyzed to examine the characteristics of the data measured by the probe attached to the gondola, and they turned out to show the nonlinear relation between the freestream and measured angles of attack and be influenced by the Reynolds number. New position of the hull nose was proposed and the effect of various factors on the flowfield around the nose was also examined. The analysis with a panel method showed that the effect of empennage was negligible, and the effect of gondola and boundary layer thickness had also little impact. It was shown that the freestream angle of attack would be the only independent variable for the probe position around the hull nose in constructing the calibration matrix.