Partitioned Simulation of Fluid-Structure Interaction on Cartesian Grids

This contribution describes recent developments and enhancements of the coupling tool preCICE and the flow solver Peano used for our partitioned simualtions of fluid-structure interaction scenarios. Peano brings together hardware efficiency and numerical efficiency exploiting advantages of tree-structured adaptive Cartesian computational grids that, in particular, allow for a very memory-efficient implementation of parallel adaptive multilevel solvers – an efficiency which is crucial facing the large computational requirements of multi-physics applications and the recent trend in computer architectures towards multi- and many-core systems. preCICE is the successor of our coupling tool FSIsce and offers a solver-independent implementation of coupling strategies and data mapping functionalities for general multi-physics problems. The underlying client-server-like concept maintains the full flexibilty of the partitioned approach with respect to exchangeability of solvers. The data mapping relies on fast spacepartitioning tree algorithms for the detection of geometric neighbourhood relations between components of non-matching grids.