• Journal of computational fluids engineering
• /
• v.4 no.1
• /
• pp.8-18
• /
• 1999

In devising a numerical approximation for the convective spatial transport of a fluid mechanical quantity, it is noted that the convective motion of a scalar quantity occurs in one-way, or from upstream to downstream. This consideration leads to a new scheme termed a pure upwind difference scheme (PUDS) in which an estimated value for a fluid mechanical quantity at a control surface is not influenced from downstream values. The formal accuracy of the proposed scheme is third order accurate. Two typical benchmark problems of a wall-driven fluid flow in a square cavity and a buoyancy-driven natural convection in a tall cavity are computed to evaluate performance of the proposed method. for comparison, the widely used simple upwind scheme, power-law scheme, and QUICK methods are also considered. Computation results are encouraging: the proposed PUDS sensitized to the convection direction produces the least numerical diffusion among tested convection schemes, and, notable improvements in representing recirculation of fluid stream and spatial change of a scalar. Although the formal accuracy of PUDS and QUICK are the same, the accuracy difference of approximately a single order is observed from the revealed results.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.24 no.9
• /
• pp.1227-1238
• /
• 2000

This study examines the effects of the discretization schemes on numerical solutions of viscoelastic fluid flows. For this purpose, a temporally evolving mixing layer, a two-dimensional vortex pair interacting with a wall, and a turbulent channel flow are selected as the test cases. We adopt a fourth-order compact scheme (COM4) for polymeric stress derivatives in the momentum equations. For convective derivatives in the constitutive equations, the first-order upwind difference scheme (UD) and artificial diffusion scheme (AD), which are commonly used in the literature, show most stable and smooth solutions even for highly extensional flows. However, the stress fields are smeared too much and the flow fields are quite different from those obtained by higher-order upwind difference schemes for the same flow parameters. Among higher-order upwind difference schemes, a third-order compact upwind difference scheme (CUD3) shows most stable and accurate solutions. Therefore, a combination of CUD3 for the convective derivatives in the constitutive equations and COM4 for the polymeric stress derivatives in the momentum equations is recommended to be used for numerical simulation of highly extensional flows.

• Journal of Hydrospace Technology
• /
• v.2 no.2
• /
• pp.57-67
• /
• 1996

The formulation for hybrid-QUICK scheme of convective transport terms in finite-volume calculation procedure is presented. Source terms are modified to apply the hybrid-QUICK scheme. Test calculations are performed for wall-driven cavity flow at Re=$10_2$, $10_3$, and $10_4$. These include the evaluation of boundary conditions approximated by third-order finite difference scheme. The stable and converged solutions are obtained without unsteady terms in the momentum equations. The results using hybrid-QUICK scheme show no difference with those using hybrid scheme at low Re ($=10_2$) and are better at higher Re ($10_3$, and $10_4$).

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.7 no.4
• /
• pp.147-157
• /
• 1987

Five finite difference schemes (explicit, Bresler, implicit, upstream, and Chaudhari scheme) for the convective dispersion model are analyzed numerically to investigate their characteristics and applicabilities. Camparative study results show that the conditionally stable Chaudhari scheme has the smallest numerical dispersion and that the unconditionally stable Bresler scheme has the overshooting in regions of oscillation. Explicit scheme is the most accurate for a dispersion-dominated flow whereas Chaudhari scheme is for a convection-dominated flow. The computation time (CPU) is the shortest for the explicit or Chaudhari scheme with the same order of magnitude and is always the longest for the Bresler scheme.

• Journal of Advanced Marine Engineering and Technology
• /
• v.20 no.3
• /
• pp.101-109
• /
• 1996

Two-dimensional lid-driven closed flows within square cavity were studied numerically for four Reynolds numbers : $10^4$, 3$\times10^4$, 5$\times10^4$ and 7.5$\times10^4$. A convective difference scheme to maintain the same spatial accurary by irregular grid correction is adopted by applying the interior division principle. Grid number is $80\times80$and its minimum size is about 1/400 of the cavity height. At Re=$10^4$, periodic migration of small eddies appearing in corner separation region and its temporal sinusoidal fluctuation are represented. At three higher Reynolds numbers(3$\times10^4$, 5$\times10^4$ and 7.5$\times10^4$), an organizing structure of four consecutive vorticles at two lower corners is revealed from time-mean flow patterns. But, instantaneous flow characteristics show very random unsteady fluctuation mainly due to the interaction between rotating shed vortices and stationary eddies within the corners.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.17 no.2
• /
• pp.467-474
• /
• 1993

Viscous flow and heat transfer phenomena in an axisymmetric cylinder which models a diesel engine have been numerically studied. In order to search for a way to minimize numerical diffusion, the effectiveness and the appropriateness of two selected numerical schemes for convective terms in the governing equations have been tested. They are Linear Upwind Difference Scheme and Hybrid Scheme. Using a standard k-.epsilon. turbulence model, the calculation has been carried out basically up to 180.deg. of crank angle. As a result, it was shown from comparison with previous experimental data that Linear Upwind Difference Scheme is less influenced than Hybrid Scheme by the numerical diffusion and it was suggested that these effects of numerical diffusion can be more significant than those due to turbulence modeling.

• 한국전산유체공학회:학술대회논문집
• /
• 1995.10a
• /
• pp.55-63
• /
• 1995

The flow field around a high-speed train including cross-wind effects has been simulated. This study solves 3-D unsteady incompressible Navier-Stokes equations in the inertial frame using the iterative time marching scheme. The governing equations are differenced with 1st-order accurate backward difference scheme for the time derivatives, 3th-order accurate QUICK scheme for the convective terms and 2nd-order accurate central difference scheme for the viscous terms. The Marker-and-Cell concept was applied to efficiently solve continuity equation, which is differenced with 2nd-order accurate central difference scheme. The 4th-order artificial damping is added to the continuity equation for numerical stability. A C-H type of elliptic grid system is generated around a high-speed train including ground. The Baldwin-Lomax turbulent model was implemented to simulate the turbulent flows. To validate the present procedure, the flow around a high speed train at constant yaw angle of $45^{\circ}\;and\;90^{\circ}$ has been simulated. The simulation shows 3-D vortex generation in the lee corner. The flow separation is also observed around the rear of the train. It has concluded that the results of present study properly agree with physical flow phenomena.

• Transactions of the Korean Society of Automotive Engineers
• /
• v.6 no.5
• /
• pp.86-96
• /
• 1998

A numerical study has been carried out of three-dimensional turbulent flows around a MIRA reference vehicle model both with and without wheels in computation. Two convective difference schemes with two k-$\varepsilon$ turbulence models are evaluated for the performance such as drag coefficient, velocity and pressure fields. Pressure coefficients along the surfaces of the model are compared with experimental data. The drag coefficient, the velocity and pressure fields are found to change considerably with the adopted finite difference schemes. Drag forces computed in the various regions of the model indicate that design change decisions should not rely just on the total drag and that local flow structures are important. The results also indicate that the RNG model with the QUICK scheme predicts fairly well the tendency of velocity and pressure fields and gives more reliable drag coefficient rather than the other cases.

• Transactions of the Korean Society of Mechanical Engineers
• /
• v.12 no.5
• /
• pp.1158-1174
• /
• 1988

A numerical technique is developed for the solution of fully developed turbulent recirculating flow in the passage of variable area using the non-orthogonal coordinate transformation. In the numerical analysis, primitive pressure-velocity finite difference equations were solved by SIMPLER algorithm with 2-equation turbulence model and algebraic stress model (ASM). QUICK scheme on the differencing of convective terms which is free from the inaccuracies of numerical diffusion has been applied to the variable grids and the results compared with those from HYBRID scheme. In order to test the effect of streamline curvatures on turbulent diffusion Lee and Choi streamline curvature correction model which has been obtained by modifying the Leschziner and Rodi's model is testes. The ASM was also employed and the results are compared to those from another turbulence model. The results show that difference of convective differencing schemes and turbulence models give significant differences in the prediction of velocity fields in the expansion region and outlet region of the combustor, however show little differences in the parallel flow region.

• Journal of computational fluids engineering
• /
• v.3 no.2
• /
• pp.17-26
• /
• 1998

The three-dimensional incompressible Navier-Stokes equations have been solved by a node-centered finite volume method with unstructured hybrid grids. The pressure-velocity coupling is handled by the artificial compressibility algorithm and convective fluxes are obtained by Roe's flux difference splitting scheme with linear reconstruction of the solutions. Euler implicit method with Jacobi matrix solver is used for the time-integration. The viscous terms are discretised in a manner to handle any kind of grids such as tetragedra, prisms, pyramids, hexahedra, or mixed-element grid. Inviscid bump flow is solved to check the accuracy of high order convective flux discretisation. And viscous flows around a circular cylinder and a sphere are studied to show the efficiency and accuracy of the solver.