Automatic Block-Structured Grid Generation Progress and Challenges

One of the most labor-intensive aspects of performing computational simulations of the flow over aerospace configurations is the design and generation of appropriate computational grids, especially of the block-structured variety. A major difficulty with current systems is the design of a suitable blocking plan (or flowfield decomposition) for an arbitrary configuration. This paper presents aa integrated approach to the blocking design problem which is comprised of three key technologies: ¯ a set of procedures which automatically convert an abstract topological specification into a real block-structured grid; ¯ a rule-based expert system which controls the blocking process, based upon expertise garnered from a variety of block-structuring experts; and ¯ a nonlinear, integer optimization technique which is used to "fine-tune" the blocking plan and the resulting computational grid. It is shown that the combination of these three technologies makesit possible to efficiently generate near-optimal block-structured grids for previously-studied classes of multi-body configurations. The issue of extending the knowledge base to arbitrary multi-body configurations is explored, with special emphasis placed on the challenges posed by three-dimensional geometries.