Numerical acceleration method for static and dynamic analysis of deepwater laying pipelines

Abstract

The deepwater pipelay finite element model often uses relatively small time steps for numerous iterative calculations to ensure sufficient precision, resulting in significant computational time consumption. In this paper, a novel numerical acceleration method is proposed, combining the vector form intrinsic finite element (VFIFE) method with GPU parallel techniques to address this issue. Efficient GPU computational solvers for static and dynamic analyses of deepwater J-lay and S-lay pipelines are developed by utilizing CUDA-based algorithms to handle key mechanical processes, including the calculation of pipeline internal forces and moments, top excitation, pipe-stinger roller interaction, hydrodynamic forces, and pipe-seabed soil interaction, etc. The method is applied to shell element, beam element, and refined beam element models, with coding strategies optimized for GPU parallel execution. Subsequently, localized pipeline solvers and global pipelay solvers are established to showcase the method's potential to significantly reduce computational time while maintaining accuracy. This study emphasizes the advantages of combining the VFIFE method with GPU parallel techniques and contributes efficient computational solvers for the mechanical analysis of deepwater pipeline laying.