A Coupled FEM-MPM GPU-based algorithm and applications in geomechanics

Qian Zhou,Wen-Jie Xu,Ze-Kang Feng

DOI: https://doi.org/10.1016/j.compgeo.2022.104982

IF: 5.218

2022-01-01

Computers and Geotechnics

Abstract:To simulate the failure process of geomechanics, a coupled FEM-MPM algorithm is developed and accelerated by using graphics processing unit (GPU). In this algorithm, during the simulation, the distorted finite elements will be replaced by material points and the information including strain, stress, etc. are inherited from relative integration points. To verify the accuracy of the provided algorithm in the simulation of shearing failure, a biaxial test was performed using FEM, MPM and coupling method respectively. The results show that the coupled FEM-MPM algorithm is beneficial in the high accuracy of FEM at small strain, and the translation process from finite elements to material points is also very smooth. Then a three-point bending test is used and compared with laboratory test to validate feasibility of the coupling algorithm in tensile failure. As a complex application, the stability analysis of a slope based on strength reduction method (SRM) is used. After the slope loses stability, the sliding process show that FEM-MPM coupling algorithm can better overcome the computational non -convergence issues caused by the distortion of elements when only using FEM. The algorithm is also efficient by limiting the coupling region at the breakage area. As a result, the GPU-based FEM-MPM coupling algorithm provided in this study can better be used in the high-performance simulation of geomechanics.