Numerical analysis of hydroelastic fluid-structure interaction problems by smoothed particle hydrodynamics method

In this study, it is aimed to develop a GPU-accelerated numerical code based on multi-resolution Riemann-SPH-TLSPH coupling for use in the analysis of complex hydroelastic FSI problems. In the present numerical coupling, while the fluid domain is modeled by Riemann-SPH formulation based on Roe's approximated Riemann solver, the solid one is analyzed by TLSPH formulation stabilized by artificial viscosity and hourglass control schemes. The present multi-resolution algorithm is based on the use of different initial particle spacing for fluid and solid domains. In addition to multiple time stepping, a hybrid time integration approach, based on the Symplectic and one-step Semi-Implicit Euler time integration formulations for fluid and solid domains, respectively, is used to improve numerical efficiency. The accuracy and efficiency of the present numerical coupling are investigated using several benchmark cases, including the free oscillating of a cantilever beam, hydrostatic water column on an elastic plate, and interaction of dam-break flows in various conditions with elastic solids. In addition to analytical solutions, experimental measurements, and other numerical computations in literature, the present model computations are also compared with the numerical results obtained from the reference analyses developed based on WCSPH-TLSPH and Rieman-SPH-TLSPH couplings with conventional time integration scheme. The obtained results show that, in all used resolutions, the developed multi-resolution Riemann-SPH-TLSPH reproduced smooth pressure and stress fields for fluid and solid domains without any numerical stability issue around the fluid-structure coupling and provided reasonable accuracy in hydroelastic deformations. Additionally, it is observed that the present hybrid time integration scheme with multiple time stepping increased computational efficiency at different rates depending on problem characteristics by comparison with reference analyses.