A Parallel Adaptive Mesh Refinement Software for Complex Geometry Flow Simulations

We present a parallel adaptive mesh refinement method with unstructured connectivity. The AMR package is being developed as an open-source project to support immersed boundary flow simulations around complex geometry. In the current work, we explain the parallel implementation of 2:1 balanced implicit-octree generation. A Cartesian processor topology is used to generate the initial mesh for a given geometry. Each processor then constructs its own local tree. One-sided communication features of the MPI-II library is used to carry out the information exchange with other processors. Several complex geometries are used to demonstrate the versatility of the AMR package