Standard and goal-oriented adaptive mesh refinement applied to radiation transport on 2D unstructured triangular meshes (Q617481)

In the case of applied mathematics and engineering the problem of numerical simulations of multi-dimensional deterministic particle transport process remains a challenging issue. An approach based on an uniformly distributed fine mesh can be too costly in computer memory and CPU time. In this paper the authors describe a cell-based spatial adaptive mesh refinement (AMR) technique and goal-oriented AMR technique for the radiation transport equation. The AMR approach has been pioneered since the 1980's for finite volume and finite element techniques. A goal-oriented AMR is more recent and is based on the fact that an accurate solution is not always needed. However, whereas the standard and goal-oriented AMR are used one should be aware on some important aspects that are connected with their usage. There are three important tools, namely: (i) reliable local error estimate, (ii) flexible geometrical data structure and (iii) projection/restriction operators -- they are described carefully in the case of AMR. The paper consists of 7 Sections. The second one is a presentation of the Discontinuous Galerkin Finite Element Method (DGFEM) for the spatial discretization of the \(S_N\) transport equation and also shows the standard iterative techniques to solve such equations. The next Section concerns on three error estimators -- they are based on difference variant, projection variant and jump indication. But in the case of goal-oriented AMR these error estimators need to be redefined -- this was done in Section 4. Section 5 shows the problem of mesh adaptation; selection of mesh cells for refinement and coupling between mesh cells of various refinement levels. The penultimate Section shows 5 examples: the first one is a simple problem showing the performance of AMR, the second is a searchlight problem, the third, fourth and fifth are problems with different material opacities, but in the case of last two examples we have the comparison of standard and goal-oriented AMR approaches. In the case of all examples we have their detailed description and presentation of 2D diagrams with results. The last Section presents an outlook on open questions.