Abstract
Mixed formulation of FEM is frequently used for the modeling of incompressible or nearly incompressible material such as liquid, to avoid volumetric locking. In this paper, Convolutional PML is developed, which is consistent with mixed formulation of FEM for liquid. Resulting coefficient matrices are symmetric just as corresponding FE. Boundary conditions considering surface waves and solid-liquid interfaces are also formulated for convolution PML. The performance of developed Convolutional PML is demonstrated by the comparison with other boundary conditions and with a rigorous solution using reservoir models. The performance of the proposed method including solid-liquid interface elements is demonstrated using dam-reservoir-foundation rock models.
