CABA: A Computational Framework for Large-Scale Parallel Adaptive Cartesian Grid Immersed Boundary Methods at High Reynolds Numbers
Please login to view abstract download link
Immersed boundary methods (IBMs) are powerful approaches for simulating flows around complex geometries. Among the various IBM implementations, adaptive Cartesian grid methods offer a compelling combination of geometric flexibility, high computational efficiency, and natural support for local mesh refinement. These characteristics have driven growing interest in their development and application over recent years. However, to fully leverage the potential of adaptive Cartesian IBMs, the entire simulation workflow—from grid generation and solution adaptation to parallel scalability—must be designed with both accuracy and performance in mind. In this work, we therefore introduce the CABA framework: a high-performance computational environment for large-scale parallel simulations using adaptive Cartesian grid immersed boundary methods at high Reynolds numbers. We present the underlying numerical methodology, including the discretization schemes, adaptive mesh refinement (AMR) strategies, and efficient parallel immersed boundary treatments. Implementation details are discussed with an emphasis on boundary treatment and scalability on HPC architectures. We conclude with several application examples that demonstrate the framework’s capability to handle industrially and scientifically relevant problems. An outlook on ongoing and future developments within the CABA project is also provided.
