A PARALLEL AGGLOMERATION MULTIGRID METHOD FOR THE ACCELERATION OF COMPRESSIBLE FLOW COMPUTATIONS ON 3D HYBRID UNSTRUCTURED GRIDS

Although the unstructured grids may accurately describe a complex geometry along with reduced requirements of user interaction for their generation/adaptation, the corresponding solvers appear to be relatively inefficient comparing to the structured ones. This drawback can be significantly alleviated by the employment of the agglomeration multigrid method, based on the solution of the flow problem on successively coarser grids, derived from the initial finest one through the fusion of the adjacent finite control volumes. In this study a parallel agglomeration multigrid methodology is presented, developed to enhance an existing academic CFD code, named Galatea, which employs the RANS (Reynolds-Averaged Navier-Stokes) equations along with appropriate turbulence models, to simulate inviscid and viscous laminar or turbulent compressible fluid flows. The fusion strategy considers the neighboring control cells' merging on a topology-preserving framework, which resembles the advancing front technique, while depending on the flow type (inviscid, laminar or turbulent) and subsequently on the type of the initial finest grid (tetrahedral or hybrid), an isotropic agglomeration or a directional one can be performed, either selected by the user. The multigrid accelerated iterative solution of the flow and turbulence models is achieved via the Full Approximation Scheme (FAS) in a V-cycle process, incorporated in the Full Multigrid (FMG) Scheme, considering as such two stages, the preliminary and the main one. The proposed methodology is validated against well-documented test cases, which concern inviscid, laminar and turbulent compressible flow over a rectangular wing with NACA0012 airfoil and the DLR-F6 aircraft model, demonstrating its capability for improved computational performance.