Speaker
Description
In order to model the scrape-off layer of a fusion device,
traditionally field aligned codes have been used. For stellarators, this is not feasible, and the flux coordinate independent approach (FCI) is used.
Aligning the grid with the boundary is challenging, and generating
body fitted, structured grids restricts the homogeneity and
orthogonality of the grid, making numerical integration challenging.
Unstructured grids are well established and allow high quality, body fitted meshes and there should not be any road blocks to enable FCI codes to use unstructured meshes.
For the FCI approach, the mesh is not field or flux surfaces aligned, thus removing the motivation why tokamak grids use the structured grid approach.
It is sufficient to interpolate the field for the parallel slices, but there is no fundamental reason why this should not be possible for unstructured meshes.
Currently a significant amount of time is spent on generating meshes. Closed divertors grids are especially challenging, restricting the grid quality significantly, as well as requiring a significant amount of work to generate body fitted, structured meshes. Generating unstructured meshes is trivially done, using state of the art meshing codes such as gmsh.
Unstructured meshes allow to a much larger degree to refine the mesh, where steps in the plasma pressure are to be resolved. Especially if griding is easy and fast, as in the case for unstructured meshes, iterating with different meshes might be a viable approach for running steady state transport simulations.
It might also become more interesting to generate different meshes for neutrals and plasma quantities, as the neutral mesh could be much coarser in the core, which is not possible using structured meshes.
This contribution will discuss the advantages, some alternatives and steps to a Hermes-3 version using unstructured meshes.