Post Simulation Processing for Overlapping Grids
Résumé
Nowadays, many engineering processes make intensive use of Computational Fluid Dynamics (CFD) simulations. Aeronautics has been a pioneer sector because numerical simulations not only limit security issues, as compared to "real" simulations, but it allows to drastically decrease both cost and time-to-market. CFD equations, typically Navier-Stokes ones, can only be solved using discrete approximation schemes. Hence the main task that needs to be achieved in order to set up the simulation consists in defining appropriate partitioning of the simulation domain using one or more 3D grids. This set of grids must perfectly fill the 3D area in which the flow is moving, which means that the boundary of the set of grids must be coincident with the boundary of the CAD model. For complex CAD geometries, finding the right set of of grids may be very time-consuming and even sometimes not possible. The chimera technique consists in allowing the grids to overlap in space, both with other grids and with the CAD geometry. The simulation itself is tuned for this particular configuration but produces results as the set of grids with their respective values. This raises new problems when the results have to be post-processed or for their visualization because the tools that are used for this purpose are not adapted to overlapping grids. In this paper we present a post-processing technique for chimera simulation results, which aims at generating a multi-block grid featuring no overlap. Our method starts by hollowing out one of the overlapping grids and then uses constrained Delaunay tetrahedralization for filling the gap between grids, so that no hole remains in the resulting multi-block solution.