A stabilized TL–WC SPH approach with GPU acceleration for three-dimensional fluid–structure interaction

Abstract

A coupled total Lagrangian (TL) and weakly compressible (WC) smoothed particle hydrodynamics (SPH) method is presented to model three-dimensional fluid–structure interactions (FSI) with deformable structures. In the coupled TL–WC SPH, the fluid phase is simulated using WCSPH, while the structure solver is based on TLSPH. The three main deficiencies of solid simulation using conventional SPH, i.e. inconsistency, tensile instability and hourglass mode are circumvented in the stabilized TLSPH by means of corrected kernel gradient, Lagrangian kernel function and hourglass control technique, respectively. The resulted stabilized TLSPH is stable, accurate and has almost quadratic convergence rate in solid modeling. To increase the accuracy in FSI modeling, the δ−SPH technique is employed to improve the pressure results in the fluid phase. Based on the Adami boundary condition (Adami et al., 2012), a unified framework for modeling solid boundaries and the fluid–structure interfaces is presented. Furthermore, the GPU parallelization is employed to accelerate the proposed TL–WC SPH method for higher efficiency. The coupled method is employed to simulate problems of pure fluid flow, elastic solids with large deformation and fluid–structure interaction with deformable structures. The numerical results are compared with analytical solutions and results from literature. The GPU efficiency and speed-up compared with CPU implementations are analyzed. The novelty of this work consists: (1) three-dimensional SPH modeling of FSI problems with deformable structures, (2) stabilized structure simulation free of hourglass mode, (3) a unified framework for FSI problems taking advantages of δ-SPH, TLSPH, hourglass control, and GPU acceleration. Importantly, with the hourglass control technique proposed by Ganzenmüller (2015), stresses can be captured accurately in the TL–WC SPH-based FSI simulations.