A stable and accurate projection method on a locally refined staggered mesh

Abstract

SUMMARYIn this paper, a robust projection method on a locally refined mesh is proposed for two-and threedimensional viscous incompressible flows. The proposed method is robust not only when the interface between two meshes is located in a smooth flow region but also when the interface is located in a flow region with large gradients and/or strong unsteadiness. In numerical simulations, a locally refined mesh saves many grid points in regions of relatively small gradients compared with a uniform mesh. For efficiency and ease of implementation, we consider a two-level blocked structure, for which both of the coarse and fine meshes are uniform Cartesian ones individually. Unfortunately, the introduction of the two-level blocked mesh results in an important but difficult issue: coupling of the coarse and fine meshes. In this paper, by properly addressing the issue of the coupling, we propose a stable and accurate projection method on a locally refined staggered mesh for both two-and three-dimensional viscous incompressible flows. The proposed projection method is based on two principles: the linear interpolation technique and the consistent discretization of both sides of the pressure Poisson equation. The proposed algorithm is straightforward owing to the linear interpolation technique, is stable and accurate, is easy to extend from two-to three-dimensional flows, and is valid even when flows with large gradients cross the interface between the two meshes. The resulting pressure Poisson equation is non-symmetric on a locally refined mesh. The numerical results for a series of exact solutions for 2D and 3D viscous incompressible flows verify the stability and accuracy of the proposed projection method. The method is also applied to some challenging problems, including turbulent flows around particles, flows induced by impulsively started/stopped particles, and flows induced by particles near solid walls, to test the stability and accuracy.

Adaptive Grid Refinement for Free-Surface Flow Simulations in Offshore Applications

“…Most authors use a non-overlapping interface and apply linear (or even higher-order) interpolation for missing variables on the other side of the interface [4]. Another approach is to apply linear interpolation inside an overlapping interface [5]. In all cases the discretization results in a non-symmetric system of equations.…”

Turbulence Modeling, Local Grid Refinement and Absorbing Boundary Conditions for Free-Surface Flow Simulations in Offshore Applications

“…Most authors use a non-overlapping interface and apply linear (or even higher-order) interpolation for missing variables on the other side of the interface [12]. Another approach is to apply linear interpolation inside an overlapping interface [13]. In all of these cases the discretization results in a non-symmetric system of equations.…”

“…For the numerical approximation this changes into Figure 9 demonstrates the amount of reflection as a function of the angle of incidence θ and dimensionless wave number kh to compare the effectiveness of the two boundary conditions: the Higdon operator (9) and the first-order ABC (13). For the coefficients in Eq.…”

“…However, this range is between 0 to 8 with the ABC in Eq. (13). The reason for this behavior lies in the way the dispersion relation is approximated.…”

Efficient Computation and Modeling of Viscous Flow Effects in ComFLOW

“…Most authors use a non-overlapping interface and apply linear (or even higher-order) interpolation for missing variables on the other side of the interface [10]. Another approach is to apply linear interpolation inside an overlapping interface [11]. In all cases the discretization results in a non-symmetric system of equations.…”