• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.31 no.7
• /
• pp.9-14
• /
• 2003

In this work, a multi-block/multi-partitioning method is suggested for a multi-block parallelization. It has an advantage of uniform load balance via subdividing of each block on each processor. To make a comparison of parallel efficiency according to domain decomposition method, a multi-block/single-partitioning and a multi-block/ multi-partitioning methods are applied to the flow analysis solver. The multi-block/ multi-partitioning method has more satisfactory parallel efficiency because of optimized load balancing. Finally, it has applied to the CFDS code. As a result, the computing speed with sixteen processors is over twelve times faster than that of sequential solver.

• 한국전산유체공학회:학술대회논문집
• /
• 1995.10a
• /
• pp.110-110
• /
• 1995

• Proceedings of the Korean Society of Propulsion Engineers Conference
• /
• 2001.11a
• /
• pp.70-72
• /
• 2001

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.21 no.7
• /
• pp.900-910
• /
• 1997

Multiblock structured grids are, to a large extent, capable of filling up topologically complex flow domains in an efficient way. The proposed approach enables to use different flow models in each different block and the easy incorporation of different grid refinement strategies for different blocks. Furthermore, it may be expected that this multiblock structured approach will naturally lead to the parallel executions of calculations per block on different vector processors. In this paper, the hypercube + + structure is proposed for topological informations on multiblock grids and the B-spline volume for geometrical informations. Three samples of the-three dimensional results are presented to demonstrate the capabilities of the present approach.

• Journal of computational fluids engineering
• /
• v.1 no.1
• /
• pp.19-25
• /
• 1996

A multiblock grid generation has been developed to be reliably used for a Navier-Stokes simulation of the turbomachinery flow-fields A multiblock structure simplifies the creation of structured H-grids about complex turbomachinery geometries and facilitate the creation of a grid in the tip flow region. The numerical algorithm adopts the combination of the algebraic and elliptic method to create the internal grids efficiently and quickly. The grid refinement process is enhanced by developing strategies to utilized Bezier curves and splines along with weighted transfinite interpolation technique and by formulating the grid-imbedding method for the viscous boundary-layer meshes. For purposes of illustration, the grid generator is applied to the high turning turbine rotor blades. Two different types of computational grids are provided to be compared with respect to the grid adaptation to the flow simulations. Extension to three-dimensions was done to show the possibility of its application to the tip-flow simulations. The grid quality of the multiblock structure is good in the passages, with gloval orthogonality and adequate smoothness.

• Journal of computational fluids engineering
• /
• v.11 no.1 s.32
• /
• pp.1-7
• /
• 2006

Numerical investigation of a centrifugal compressor volute having a modified straight conical duct hill been made. Three-dimensional Reynolds-Averaged Navier-Stokes equations with $k-{\varepsilon}$ turbulence equation are solved To avoid coordinate singularity at the central axis of the duct, multi-block H-type grid is generated on the circular cross-sections of the volute and stretched toward the solid wall boundary. We obtained numerical results with three different mass flow rates at the volute inlet, namely, with the inlet conditions that give small, medium and large mass flow rates at the outlet of the conical duct. Agreement with the experimental results is observed.

• Proceedings of the KSME Conference
• /
• 2005.11a
• /
• pp.85.1-85.1
• /
• 2005

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.35 no.8
• /
• pp.661-669
• /
• 2007

Study on the feasibility assessment for applying the overlap grid method to numerical calculations on the ground effect has been performed. The objective of the present study is to settle the problem in the grid generation process. A low Mach number preconditioned turbulent flow solver using the overlap grid and the multi-block grid methods has been developed and applied to the ground effect simulation around the RAE 101 airfoil. It has been verified that the overlap grid method not only can provide sufficiently accurate solutions but also work out the grid generation problem in the ground effect simulations.

• Journal of computational fluids engineering
• /
• v.21 no.1
• /
• pp.36-42
• /
• 2016

One of the obstacles on the grid generation for complex geometries with multi-block structured grids is the domain decomposition. In this paper, the domain decomposition for two-dimensional flow is studied using the flow characteristics. The potential flow equation with the source distribution on the panel surface is solved to extract the information of the flow. The current approach is applied to a two-dimensional cylinder and Bi-NACA0012 problems. The generated grids are applied to generic flow solvers and reasonable results are obtained. It can be concluded that the current methods is useful in the domain decomposition for the multi-block structured grid.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.36 no.11
• /
• pp.1049-1055
• /
• 2008

Modern multidisciplinary computational fluid dynamics often incorporates moving boundaries, as would be required in the applications such as design optimization, aeroelasticity, or forced boundary motion. It is challenging to develop robust, efficient grid deformation algorithms when large displacement of the moving boundaries is required. In this paper, a volume grid deformation code is developed based on the finite macro-element and the transfinite Interpolation, and then interfaces to a structured multi-block Navier-Stokes in-house code. As demonstrated by an airfoil with pitching motion, the hysteresis loops of lift, drag and moment coefficients of the developed method are shown to be in good agreement with those of experimental data.