Scalability of a Large Eddy Simulation Immersed Boundary solver on EuroHPC JU clusters (scalability tests on MeluXina CPU)

The present project is aimed at demonstrating the scalability of an in-house, high-fidelity Large-Eddy Simulation, Immersed-Boundary solver in MPI Fortran language on the clusters MeluXina CPU, Karolina CPU and Discoverer in order to provide evidence of its suitability and portability on different architectures for future production runs in the framework of EuroHPC JU Regular Access. The results of strong and weak scaling tests will serve for preparing future proposals, selecting the most appropriate machine to conduct our future studies and choosing the best setup of our future computations. Our target is studying innovative propeller geometries, ranging from bio-inspired designs, contra-rotating propellers and ducted propellers, by means of high-fidelity, fluid dynamic simulations. All above non-conventional propeller geometries aim at improved performance, compared to typical design solutions. The purpose of high-fidelity simulations is shedding light on the physics of the flow and building a database of reference solutions for lower-fidelity approaches, widely utilized in both academia and industry. In our solver, finite-differences are utilized to discretize the filtered Navier-Stokes equations. An immersed-boundary methodology enables the use of regular grids, as Cartesian or cylindrical, making the decomposition of the overall flow problem into subdomains very straightforward, efficient and suitable to parallel computing. Communications across subdomains are handled via calls to MPI libraries. I/O operations are performed using calls to parallel HDF5 libraries. The solver is not I/O intensive, with I/O operations taking only about 5% of the overall computational cost of a typical simulation. Although the scalability of the present solver was already tested on several, different architectures, also part of the PRACE and EuroHPC JU infrastructures (Marconi KNL, Joliot-Curie KNL, Joliot-Curie SKL, Joliot-Curie Rome, MareNostrum 4, Vega CPU, LUMI-C), the test-case that will be considered in this project will be specifically designed to be representative of the computational effort of the problems we aim to tackle in the field of marine propulsion through our next proposal for EuroHPC JU Regular Access.