董友扣副教授在国际著名科学期刊《Computers and Geotechnics》发表论文

    Lagrangian particles in the material point method (MPM) are free to flow through the background Eulerian mesh to represent material deformation. Excessive compression or tension in the kinematic field tends to entangle the particles from the initial uniform configuration, causing artificial voids or aggregations that lead to loss of the continuity of the stress field and the singular deformation gradient of the particles, undermining computational robustness and accuracy. In this paper, the reseeding operation is performed in elements depending on the numbers of the accommodated particles and their accumulated deformations. The state variables of the reseeded particles are recovered from their surrounding old counterparts. The stresses of the reseeded particles around the structure are adjusted to mitigate the fluctuations of contact force from errors in the state recovery. Benchmark problems of deep penetration of a wished-in-place T-bar and the dynamic impact of a submarine landslide on a partially buried mudmat are analysed. The results of the MPM analyses are validated by comparison with the exact solutions by theoretical analyses and the numerical predictions using computational fluid dynamics (CFD) simulations.