Ventus: An Overset Adaptive Cartesian Simulation Framework for Moving Boundary Problems

Abstract

A novel object-oriented solver framework named Ventus is being developed as a simulation capability and technology demonstrator for algorithms and concepts relevant to the aerodynamics of compressible flows. The Ventus framework is a multiple grid, multiple solver paradigm, that supports completely general polyhedral grids that interface with each other through overset grid communication. Automatic hole-cutting and overset grid assembly capabilities are provided to facilitate simulatation of multiple bodies in close proximity, and automatic adaptive Cartesian octree grids are generated to efficiently resolve the off-body region and match the resolution of the near-body grid system. Ventus has been developed with three core ideas in mind: (1) Minimalism with respect to the required user input, while providing extreme flexibility with respect to optional user input options; (2) The unique character of each grid type should be respected. Whether the grid is structured, unstructured, or tree-based, the basic grid operations should be overloaded to provide implementations that are heavily optimized for each specific grid type, thus minimizing the memory and CPU requirements; and (3) The desire for a modular and extensible solver framework that will ultimately allow users and developers to implement new solvers and new physical models separately and plug them in using simple API calls. Ventus offers a thoroughly envisaged, polished simulation framework design that seeks to provide a streamlined user experience at the front end, with robust, efficient and powerful numerics at the back end. The high-level organization of the Ventus simulation framework is introduced, and the methodologies of the automatic grid generation as well as the hole-cutting and automatic overset grid assembly approaches are described in detail. A compressible Navier-Stokes solver has been implemented in the Ventus framework, and a series of case studies are presented to verify the numerical order of accuracy, provide solution validation for moving body problems utilizing overset and adaptive Cartesian octree grids, and demonstrate the effectiveness of the grid adaptation procedures for low and high speed flow problems.