Deformable Overset Grid for Multibody Unsteady Flow Simulation

Abstract

A deformable overset grid method is proposed to simulate the unsteady aerodynamic problems with multiple flexible moving bodies. This method uses an unstructured overset grid coupled with local mesh deformation to achieve both robustness and efficiency. The overset grid hierarchically organizes the sub-grids into CLUSTERs and LAYERs, allowing for overlapping/embedding of different type meshes, in which the mesh quality and resolution can be independently controlled. At each time step, mesh deformation is locally applied to the sub-grids associated with deforming bodies by an improved Delaunay graph mapping method that uses a very coarse Delaunay mesh as the background graph. The graph is moved and deformed by the spring analogy method according to the specified motion and then the computational meshes are relocated by a simple one-to-one mapping. An efficient implicit hole-cutting and inter-grid boundary definition procedure is implemented fully automatically for both cell-centered and cell-vertex schemes based on the wall distance and an alternative digital tree (ADT) data search algorithm. This method is successfully applied to several complex multi-body unsteady aerodynamic simulations and the results demonstrate the robustness and efficiency of the proposed method for complex unsteady flow problems, particularly for those involve simultaneous large relative motion and self-deformation. --------------------

Enhancement on parallel wall distance calculation methodology for partitioned unstructured grid

“…The hierarchical overset grid system containing a main foil as the background mesh and two sub‐grids of the leading‐edge slat and trailing‐edge flap. The implementation of the overset system is achieved by a fully implicit automatic parallel chimera hole cutting method for the hierarchically organized grids based on ADT technique, 1 which excludes the grid nodes or cells that do not participate in the computation and identifies the inter‐grid boundary of overlapping regions (as shown in white borders in Figure 14B). The minimum wall distance field of each sub‐grid (Figure 15A) is computed by the parallel advancing front method at first as an indicator to determine the inter‐grid boundary, which insures that the grid elements closer to the wall boundary are retained for the computation.…”

“…High‐fidelity unsteady flow simulation involving multiple bodies in relative motion has been routinely performed recently using overset grids or sometimes called the Chimera grids, which allows modeling the multi‐component systems with an optimum body‐fitted grid in each. In principle, minimum wall distance, that is, the minimum distance to a solid wall, plays a pivotal role in overset grid assembly to determine the intergrid boundaries and the interpolation fringe points 1–4 . Additionally, minimum wall distance is a primary parameter in computational fluid dynamics (CFD) when considering near‐wall turbulence behaviors by means of Reynolds–Average Navier–Stokes (RANS) methods or Large Eddy Simulations (LES) 5–10 .…”

“…In principle, minimum wall distance, that is, the minimum distance to a solid wall, plays a pivotal role in overset grid assembly to determine the intergrid boundaries and the interpolation fringe points. [1][2][3][4] Additionally, minimum wall distance is a primary parameter in computational fluid dynamics (CFD) when considering near-wall turbulence behaviors by means of Reynolds-Average Navier-Stokes (RANS) methods or Large Eddy Simulations (LES). [5][6][7][8][9][10] It is no doubt that the wall distance has to be re-computed for each time step in dynamic unsteady simulations, occupying a fraction of time expenditure.…”

Improvements on overlapping unstructured grids

“…In the field of computational fluid dynamics, overlapping grids have been used widely to simulate complex flows. [1][2][3][4][5] The structured overlapping grid technique was first proposed by Steger 6 and has been well developed. However, with no structured constraints on grid nodes and better flexibility, unstructured grids are more adaptable to complex configurations.…”

Parallel algorithms for moving boundary problems by local remeshing

“…The grids move rigidly according the movements of bodies, requiring little processing on itself. If adopting unstructured grids (Gong et al , 2018; Horne and Mahesh, 2019; Karman and Sahasrabudhe, 2007; Löhner et al , 2001; Togashi et al , 2001; Xiao et al , 2016) instead of block structured grids, the flexibility of the overset grid approach can be enhanced even more. Nevertheless, the grid assembly step of this approach involves the time-consuming process of cutting holes, searching for donor cells and interpolating solutions.…”