Automatic Near-Body Domain Decomposition Using the Eikonal Equation

The purpose of this research is to generate high-aspect-ratioc ellsf or structured grids in the vicinity of boundaries for wall dominated phenomenas ucha sv iscous layers in aerodynamic applications. When suchd omains are discretized for complex geometries, it is importantt hat the grid fits the boundaries well, andt his becomes even more difficultt oa chievew ith strongly curved boundaries. Domain decomposition simplifies this problem by subdividing the domain into multiple blocksa nd makes the task of meshing complex geometries more manageable. There are manyb enefits to using am ulti-block strategy.I ti sp ossiblet oc on-trol the orthogonalitya nd grid quality more preciselyw ithin smallerb locks, and it allows differentm esh typesa nd generation techniques in eachi ndividual block. Also, parallel algorithms can be embedded in the mesh generation algorithm, where differentb locksc an be assigned to differentp rocessors, thus greatly improving the efficiency of the process. The presentp aper introduces an approacht oa utomatically decompose an arbitrary complex domain into face-matching multi-blockse manating from the boundaries. The subdivision procedure is based on the creation of offset surfaces which closelyfi tt heir geometries instead of arbitrary planes or surfaces. Thisfi tting is achieved through aw eak solution of the Offset DistanceF unction,w hichi sav aria-tion of the Eikonal equation. Computing the normal directions in this manner insures that they do not cross, and the resulting propagation avoids self-intersections which arise from direct construction methods. Thus thism ethodc an be applied on an arbitrary complex domain, i.e.,acon-cave,o raconvexs hape, it mayh ave sharp corners, or even be multi-connected. In addition,t he proposed methodt ransports the original parameterization to the propagated surface whicha llows rigorousm atching of block faces. The geometric and topological configurations are thus well defined whicha llows to increaset he automation levelw ithoutu ser intervention.