• 한국해안·해양공학회논문집
• /
• 제30권4호
• /
• pp.143-152
• /
• 2018

최근 개발된 FUNWAVE-TVD 파랑모형을 이용하여 적용되어 온 TVD 기법들의 수렴도와 수치적인 안정성을 비교하였다. Yamamoto and Daiguji(1993)의 minmod limiter를 사용하는 4차 정확도의 MUSCL-TVD 기법과 Erduran et al.(2005)의 van-Leer limiter를 사용하는 4차 정확도의 MUSCL-TVD 기법, Zhou et al.(2001)의 van-Leer limiter를 사용하는 2차 정확도의 MUSCL-TVD 기법을 비교하였으며, 수리실험 관측치가 제시되어 있는 Vincent and Briggs(1989)의 불규칙 파랑실험에 적용하였다. 불규칙 파랑의 비쇄파 실험 결과에서 minmod limiter를 사용하는 4차 정확도의 기법은 van-Leer limiter를 사용하는 기법이 요구하는 격자의 크기만큼 세밀한 격자를 요구하지는 않지만, 더 낮은 CFL을 사용해야 안정적인 모의가 가능하였다. 반면에 van-Leer limiter를 사용하는 기법에서는 numerical dissipation을 줄이기 위하여 보다 세밀한 격자를 필요로 하지만 비교적 높은 CFL을 사용할 수 있는 것으로 나타났다. 각 기법의 numerical dissipation의 영향을 최대한 줄이기 위하여 공간격자를 충분히 줄인 쇄파 모의 실험에서는 비쇄파 실험에 비하여 각 기법의 특성이 명확히 나타났다. Numerical dissipation이 상대적으로 작은 minmod limiter를 사용하는 기법으로 모의할 때는 격자를 충분히 줄이면 수치적인 불안정성이 나타나며 수치해가 발산하는 결과를 보였지만, van-Leer limiter를 사용하는 기법에서는 비교적 낮은 CFL을 사용하여 쇄파 모의가 완료되었으며, 관측치를 잘 재현하는 결과를 보였다.

• 한국전산유체공학회지
• /
• 제10권1호
• /
• pp.99-106
• /
• 2005

Performance of first, second and third order accurate methods for calculation of in viscid fluxes in fluid flow governing equations are investigated here. For the purpose, an upwind method based on Roe's scheme is used to solve 2-dimensional Euler equations. To increase the accuracy of the method two different schemes are applied. The first one is a second and third order upwind-based algorithm with the MUSCL extrapolation Van Leer (1979), based on primitive variables. The other one is an upwind-based algorithm with the Chakravarthy extrapolation to the fluxes of mass, momentum and energy. The results show that the thickness of shock layer in the third order accuracy is less than its value in second order. Moreover, applying limiter eliminates the oscillations near the shock while increases the thickness of shock layer especially in MUSCL method using Van Albada limiter.

• 한국전산유체공학회지
• /
• 제12권1호
• /
• pp.43-52
• /
• 2007

A comparative study on flux functions for the 2-dimensional Euler equations has been conducted. Explicit 4-stage Runge-Kutta method is used to integrate the equations. Flux functions used in the study are Steger-Warming's, van Leer's, Godunov's, Osher's(physical order and natural order), Roe's, HLLE, AUSM, AUSM+, AUSMPW+ and M-AUSMPW+. The performance of MUSCL limiters and MLP limiters in conjunction with flux functions are compared extensively for steady and unsteady problems.

• 한국수자원학회:학술대회논문집
• /
• 한국수자원학회 2019년도 학술발표회
• /
• pp.109-109
• /
• 2019

본 연구에서는 강우-유출 과정 모의를 위한 GPU 기반 확산파 모형을 개발하였다. 확산파 방정식을 풀기위한 수치기법으로는 유한체적법을 이용하였으며, van Leer TVD limiter를 적용한 MUSCL 기법을 이용하여 각 셀의 인터페이스의 물리적 성질을 재구성하여 구하였다. 또한, 침투를 고려하기 위하여 Horton 침투 모형을 이용하였다. 개발된 모형을 이용하여 1D single overland plane과 2D V-shaped overland에서 강우-유출 과정을 모의실험을 하였으며, 각각 해석해와 dynamic wave model을 이용하여 계산된 수치 결과와 비교하여 본 모형의 정확성을 검증하였다. 또한, 1D와 2D의 기복이 심한 지형에 적용하여 강우-유출과정이 본 모형을 통하여 물리적으로 타당한 해석이 가능함을 검증하였다. 마지막으로 복잡한 실제 지형에 적용하였으며, 측정값과의 비교를 통하여 실제 유역에서의 확산파 모형의 적정성을 검증하였다. 또한, 본 연구에서는 NVIDIA사의 GPU인 Geforce GTX 1050과 GPU의 병렬 연산 처리 능력을 활용할 수 있는 NVIDIA사의 CUDA-Fortran을 이용하여 GPU 기반 확산파 모형을 개발하였다. PC windows에서 CPU(Intel i7, 4.70 GHz) 기반 모형 대비 GPU 기반 모형의 계산속도 성능을 비교한 결과, 격자 간격이 증가할수록 CPU 기반 모형 대비 GPU 기반 모형의 연산 효율이 증가하였으며, 격자 간격이 $3200{\times}3200$일 때, CPU 기반 모형 대비 GPU 기반 모형의 연산 효율이 최대 약 150배 증가하였다.

• International Journal of Aeronautical and Space Sciences
• /
• 제13권3호
• /
• pp.307-316
• /
• 2012

Numerical simulations of 3D aircraft configurations are performed in order to understand the effects of turbulence models on the prediction of aircraft's aerodynamic characteristics. An in-house CFD code that solves 3D RANS equations and two-equation turbulence model equations are used. The code applies Roe's approximated Riemann solver and an AF-ADI scheme. Van Leer's MUSCL extrapolation with van Albada's limiter is also adopted. Various versions of Menter's $k-{\omega}$ SST turbulence models as well as Coakley's $q-{\omega}$ model are incorporated into the CFD code. Menter's $k-{\omega}$ SST models include the standard model, the 2003 model, the model incorporating the vorticity source term, and the model containing controlled decay. Turbulent flows over a wing are simulated in order to validate the turbulence models contained in the CFD code. The results from these simulations are then compared with computational results from the $3^{rd}$ AIAA CFD Drag Prediction Workshop. Numerical simulations of the DLR-F6 wing-body and wing-body-nacelle-pylon configurations are conducted and compared with computational results of the $2^{nd}$ AIAA CFD Drag Prediction Workshop. Aerodynamic characteristics as well as flow features are scrutinized with respect to the turbulence models. The results obtained from each simulation incorporating Menter's $k-{\omega}$ SST turbulence model variations are compared with one another.

• 한국전산유체공학회:학술대회논문집
• /
• 한국전산유체공학회 2006년도 PARALLEL CFD 2006
• /
• pp.311-314
• /
• 2006

The supersonic flow around tandem cavities was investigated by three- dimensional numerical simulations using the Reynolds-Averaged Navier-Stokes(RANS) equation with the $\kappa-\omega$ thrbulence model. The flow around a cavity is characterized as unsteady flow because of the formation and dissipation of vortices due to the interaction between the freestream shear layer and cavity internal flow, the generation of shock and expansion waves, and the acoustic effect transmitted from wake flow to upstream. The upwind TVD scheme based on the flux vector split using van Leer's limiter was used as the numerical method. Numerical calculations were performed by the parallel processing with time discretizations carried out by the 4th-order Runge-Kutta method. The aspect ratio of cavities are 3 for the first cavity and 1 for the second cavity. The ratio of cavity interval to depth is 1. The ratio of cavity width to depth is 1 in the case of three dimensional flow. The Mach number and the Reynolds number were 1.5 and $4.5{\times}10^5$, respectively. The characteristics of the dominant frequency between two-dimensional and three-dimensional flows were compared, and the characteristics of the second cavity flow due to the fire cavity flow cavity flow was analyzed. Both two dimensional and three dimensional flow oscillations were in the 'shear layer mode', which is based on the feedback mechanism of Rossiter's formula. However, three dimensional flow was much less turbulent than two dimensional flow, depending on whether it could inflow and outflow laterally. The dominant frequencies of the two dimensional flow and three dimensional flows coincided with Rossiter's 2nd mode frequency. The another dominant frequency of the three dimensional flow corresponded to Rossiter's 1st mode frequency.

• Journal of Mechanical Science and Technology
• /
• 제20권8호
• /
• pp.1256-1265
• /
• 2006

The supersonic flows around tandem cavities were investigated by two-dimensional and three-dimensional numerical simulations using the Reynolds-Averaged Navier-Stokes (RANS) equation with the k- ω turbulence model. The flow around a cavity is characterized as unsteady flow because of the formation and dissipation of vortices due to the interaction between the freestream shear layer and cavity internal flow, the generation of shock and expansion waves, and the acoustic effect transmitted from wake flow to upstream. The upwind TVD scheme based on the flux vector split with van Leer's limiter was used as the numerical method. Numerical calculations were performed by the parallel processing with time discretizations carried out by the 4th-order Runge- Kutta method. The aspect ratios of cavities are 3 for the first cavity and 1 for the second cavity. The ratio of cavity interval to depth is 1. The ratio of cavity width to depth is 1 in the case of three dimensional flow. The Mach number and the Reynolds number were 1.5 and $4.5{\times}10^5$, respectively. The characteristics of the dominant frequency between two- dimensional and three-dimensional flows were compared, and the characteristics of the second cavity flow due to the first cavity flow was analyzed. Both two dimensional and three dimensional flow oscillations were in the 'shear layer mode', which is based on the feedback mechanism of Rossiter's formula. However, three dimensional flow was much less turbulent than two dimensional flow, depending on whether it could inflow and outflow laterally. The dominant frequencies of the two dimensional flow and three dimensional flows coincided with Rossiter's 2nd mode frequency. The another dominant frequency of the three dimensional flow corresponded to Rossiter's 1st mode frequency.