• 한국전산구조공학회논문집
• /
• 제26권4호
• /
• pp.301-308
• /
• 2013

본 논문에서는 새로운 비중첩 영역 분할 기법을 바탕으로 한 병렬해석의 정확도 분석이 수행된다. Tak 등(2013)에 의해 제안된 이 방법에서 분할된 하위도메인들은 서로 중첩되지 않으며 계면요소(interfacial element)라 불리는 가상연결유한요소를 통해 서로 간의 관계가 결정된다. 이 접근법의 주요 장점은 영역 분할시 floating 도메인에서 발생할 수 있는 특이강성행렬(singular stiffness matrix)을 계면요소의 결합을 이용하여 가역행렬(invertible matrix)로 변환할 수 있다는 것과 기존의 FETI법에 비하여 해석시간과 스토리지(storage) 사용을 줄일 수 있다는 것이다. 반면에 3개 이상의 하위도메인들이 한 점에서 연결되는 경우를 의미하는 cross point에서는 해석의 정확도가 저하되는 경향이 나타났다. 따라서 본 논문에서는 새로운 비중첩 영역 분할기법에 대해 다양한 영역분할의 경우에 따라 발생하는 하나의 cross point에 접촉하는 하위도메인의 개수에 따른 정확도 분석이 수행되고 정확도가 저하되는 원인분석 및 대책이 논의된다.

• 한국전산유체공학회지
• /
• 제15권2호
• /
• pp.55-61
• /
• 2010

A dependence of weighting parameter in preconditioning method for solving low Mach number flow with incompressible flow nature is investigated. The present preconditioning method employs a finite-difference method applied Roe‘s flux difference splitting approximation with the MUSCL-TVD scheme and 4th-order Runge-Kutta method in curvilinear coordinates. From the computational results of benchmark flows through a 2-D backward-facing step duct it is confirmed that there exists a suitable value of the weighting parameter for accurate and stable computation. A useful method to determine the weighting parameter is introduced. With this method, high accuracy and stable computational results were obtained for the flow with low Mach number in the range of Mach number less than 0.3.

• 한국유체기계학회 논문집
• /
• 제1권1호
• /
• pp.72-80
• /
• 1998

Computational analysis of incompressible flows by numerically solving Navier-Stokes equations using multi-block finite volume method is conducted on a parallel computing system. Numerical algorithms adopted in this study $include^{(1)}$ QUICK upwinding scheme for convective $terms,^{(2)}$ central differencing for other terms $and^{(3)}$ the second-order Euler differencing for time-marching procedure. Structured grids are used on the body-fitted coordinate with multi-block concept which uses overlaid grids on the block-interfacing boundaries. Computational code is parallelized on the MPI environment. Numerical accuracy of the computational method is verified by solving a benchmark test case of the flow inside two-dimensional rectangular cavity. Computation in the axial compressor cascade is conducted by using 4 PE's md, as results, no numerical instabilities are observed and it is expected that the present computational method can be applied to the turbomachinery flow problems without major difficulties.

• 한국전산유체공학회:학술대회논문집
• /
• 한국전산유체공학회 2008년도 학술대회
• /
• pp.84-91
• /
• 2008

Generalized hydrodynamic (GH) theory for multi-species gas and the computational models are developed for the numerical simulation of hypersonic rarefied gas flow on the basis of Eu's GH theory. The rotational non-equilibrium effect of diatomic molecules is taken into account by introducing excess normal stress associated with the bulk viscosity. The numerical model for the diatomic GH theory is developed and tested. Moreover, with the experience of developing the dia-tomic GH computational model, the GH theory is extended to a multi-species gas including 5 species; O$_2$, N$_2$, NO, O, N. The multi-species GH model includes diffusion relation due to the molecular collision and thermal phenomena. Two kinds of GH models are developed for an axisymmetric flow solver. By compar-ing the computed results of diatomic and multi-species GH theories with those of the Navier-Stokes equations and the DSMC results, the accuracy and physical consistency of the GH computational models are examined.

• 한국전산유체공학회:학술대회논문집
• /
• 한국전산유체공학회 2004년도 추계 학술대회논문집
• /
• pp.187-191
• /
• 2004

The conventional reliability based design optimization(RBDO) methods require high computational cost compared with the deterministic design optimization(DO) methods. To overcome the computational inefficiency of RBDO, single loop methods have been proposed. These need less function calls than that of RBDO but much more than that of DO. In this study, the approximate reliability method is proposed that the computational requirement is nearly the same as DO and the reliability accuracy is good compared with that of RBDO. Using this method, the 3-D viscous aerodynamic shape design optimization with uncertainty is performed very efficiently.

• 한국전산유체공학회:학술대회논문집
• /
• 한국전산유체공학회 2008년 추계학술대회논문집
• /
• pp.84-91
• /
• 2008

Generalized hydrodynamic (GH) theory for multi-species gas and the computational models are developed for the numerical simulation of hypersonic rarefied gas flow on the basis of Eu's GH theory. The rotational non-equilibrium effect of diatomic molecules is taken into account by introducing excess normal stress associated with the bulk viscosity. The numerical model for the diatomic GH theory is developed and tested. Moreover, with the experience of developing the dia-tomic GH computational model, the GH theory is extended to a multi-species gas including 5 species; $O_2,\;N_2$, NO, O, N. The multi-species GH model includes diffusion relation due to the molecular collision and thermal phenomena. Two kinds of GH models are developed for an axisymmetric flow solver. By compar-ing the computed results of diatomic and multi-species GH theories with those of the Navier-Stokes equations and the DSMC results, the accuracy and physical consistency of the GH computational models are examined.

• Journal of Advanced Marine Engineering and Technology
• /
• 제30권7호
• /
• pp.771-781
• /
• 2006

Car acceleration or deceleration induce the surface slope of liquid fuel in the LNG tank. Slope changes the surface area wetted by liquid fuel in the tank and consequently heat leak to the tank. The Fortran program, 'Pro-Heatleak', is developed to evaluate heat leak on LNG tank. The verification test proves the high accuracy of the developed program. The difference between MathCad and computational results is less than 0.07 %. Computational analyses of heat leak are carried out for 10 gallons and 20 gallons of fuel vapor in the tank. With the increasing of fuel vapor volume by 10 percent the wetted surface area and heat leak respectively decrease by 13 percent. The difference between maximum and minimum heat leak is about 10 percent for both 10 gallons and 20 gallons of fuel vapor in the tank.

• 한국전산구조공학회:학술대회논문집
• /
• 한국전산구조공학회 1998년도 봄 학술발표회 논문집
• /
• pp.68-74
• /
• 1998

A finite element analysis for physical phenomenon which are governed by parabolic equation, has some inefficiencies caused by much computational time and large storage space. In this paper, a comparative study is performed to suggest the best efficient transient analysis algorithms for parabolic equations. First, the general finite element analysis techniques are summarized in views of formulation procedures, treatments of convection terms. and time stepping methods. Results of several combinations applied to one dimensional convection-diffusion equation and Burger equation are represented and compared using some criteria such as accuracy, stability, and computational time. Through the results, some guidelines to select a algorithm for solving parabolic equations are proposed for diffusion dominant and convection dominant cases. Finally applicability of two dimensional extension of the result is also discussed.

• 한국전산구조공학회:학술대회논문집
• /
• 한국전산구조공학회 1992년도 가을 학술발표회 논문집
• /
• pp.133-138
• /
• 1992

The three dimensional analysis of Framed-Tube structures is neither easy nor efficient because of longer computational time, large memory requirement, tedious input preparation and bulky output. An efficient analysis model for framed-tube structure is proposed in this study. The proposed model can save the computational effort by using the assumption of the rigid floor diaphragm effect and matrix condensation technique. Moreover, it is develpoed by assembling two dimensional frames using the link degrees of freedom which are temporary used to satisfy the vertical displacement compatibility at the corners of a framed-tube. The accuracy and the efficiency of this analytical model is established by comparing with the results using the computer code SAPIV which is based on the three dimensional finite element model.

• 한국전산유체공학회:학술대회논문집
• /
• 한국전산유체공학회 1999년도 춘계 학술대회논문집
• /
• pp.11-22
• /
• 1999

A numerical prediction method has been proposed to predict non-linear free surface oscillation in a three-dimensional container. The fluid motions are numerically predicted with Navier-Stokes equations discretized in a Lagrangian scheme with sufficient numerical accuracy. The profile of a free surface is precisely represented with three-dimensional body-fitted coordinates (BFC), which are regenerated in each computational step on the basis of the arbitrary Lagrangian-Eulerian (ALE) formulation. In order to confirm the reliability of the computational method, it was firstly applied to three-dimensional flows within complicated-shaped rigid boundaries, such as curved pipes and ducts. Than it was applied to benchmark computations related to free surface oscillations. Following these basic verifications, non-linear sloshings in a cylindrical tank and transitions from sloshing to swirling motions were numerically predicted. Throughout these computations, the applicability of the present computational method has been confirmed and some of the predicted free surface motions were visualized as sequential images and animations to understand their dynamic futures.