Strategies Toward Automation of Overset Structured Surface Grid Generation

In recent years, structured overset grid technology has been successfully applied to computational fluid dynamics analysis on a wide variety of complex aerospace applications.1–11 While structured overset grid flow solvers such as OVERFLOW12 and LAVA13 are highly e cient relative to typical flow solvers utilizing unstructured grids, structured overset grid generation remains a labor intensive and time consuming step relative to unstructured methods. Grid generation for overset structured grids usually consists of three steps: surface grid generation, volume grid generation, and domain connectivity. The surface grid generation step consists of three main tasks. The first involves decomposition of the surface geometry into four-sided overlapping domains while capturing the surface features. The second task is to determine the grid point distribution on the domain bounding curves such that both geometry and flow features are appropriately resolved, and the third task is to select a combination of algebraic and hyperbolic methods to fill the domain interiors with grid points. For algebraic meshes, grid generation input involves specification of the four appropriately redistributed bounding curves. For the hyperbolic meshes, only one initial boundary curve needs to be prescribed, but further specifications are needed for an appropriate marching distance (constant or spatially variable) from the initial curve, the grid point distribution in the marching direction, and boundary conditions at each end of the initial curve. For complex configurations, the surface grid generation step requires significant user expertise, is highly labor intensive, and typically consumes at least 80% of the total grid generation time.14 Since the beginning of computational analysis of complex configurations using overset grids, two main advances have contributed to increasing the level of geometric complexity that can be handled. The first is the introduction of a graphical user interface15,16 that is connected to standalone overset grid generation software modules. This allows the user to visualize the results while interactively specifying and iterating on the inputs for the grid generation process. The grid generation modules provide the ability to generate both hyperbolic and algebraic grids from one, two, three or four domain bounding curves, and are thus highly suited for the overset mesh approach. The second advance was the introduction of a best practice philosophy to write preprocessing scripts as the grids are generated. Such scripts, typically written in a scripting language such as Tcl, contain all instructions for generating the surface and volume meshes for each geometric component, as well as inputs for domain connectivity, aerodynamic loads integration, and flow solver boundary conditions.