Driven by the availability of vast computational power, due to multi-core processors, even in desktop systems, new possibilities for multiphysics simulations arise, thus resulting in increased requirements on mesh generation. Size reduction, more accurate geometrical features, and multi-segmented input structures are needed to suffice these demands. Furthermore, since meshing is the first step of a simulation flow, failing to properly control this process can jeopardize or even completely prevent a simulation. To target these issues a multi-segment non-manifold remeshing approach for three dimensions has been developed, providing a toolchain for mesh analysis, structural and non-manifold edge detection, and parallel remeshing.
Using an advancing front method to produce high quality meshes is not possible without being able to rigorously remesh the input structure as a whole. Therefore the focus has been set on developing a remeshing algorithm also for the surface of multi-segmented structures using an advancing front approach. During the surface remeshing step, charts, which are shared by two segments, are reduced to their boundary edges, thus creating non-manifold edges between them. Well known meshing algorithms cannot be applied to solve this issue, since these algorithms operate only on a manifold structure having distinctly defined transitions between charts. As a result non-manifold configurations have to be broken down to establish a manifold topology again. This issue has been solved by introducing metadata to the elements, which keeps track of their topological connections and their orientation. The developed surface remeshing step is the basis for parallel surface meshing, which will be developed as an extension.
Nowadays, various open source meshing libraries, which target different aspects of meshing, are available and using these well tested resources is an additional bonus. The design creation process was especially aimed in developing a modular framework to incorporate existing and well tested software libraries. Such a modular design can only be implemented when using the appropriate programming paradigms. Therefore the approach is based on various paradigms and their appropriate combination can alleviate the shortcoming of an individual paradigm, while making their strength available to the whole. Besides the programming paradigms, the right choice of the underlying datastructure is of utmost importance. Topological operations, such as the boundary and the co-boundary, are applied to identify irreducible edges. Therefore a solid topological framework, the Generic Scientific Simulation Environment (GSSE), is used as foundation of the meshing process.