JOURNAL OF COMPUTATIONAL PHYSICS | 卷:360 |
A high-order multiscale finite-element method fortime-domain acoustic-wave modeling | |
Article | |
Gao, Kai1  Fu, Shubin2  Chung, Eric T.2  | |
[1] Los Alamos Natl Lab, Geophys Grp, Los Alamos, NM 87545 USA | |
[2] Chinese Univ Hong Kong, Dept Math, Hong Kong, Hong Kong, Peoples R China | |
关键词: Acoustic-wave equation; Multiscale method; Finite-element method; Spectral-element method; Heterogeneous medium; | |
DOI : 10.1016/j.jcp.2018.01.032 | |
来源: Elsevier | |
【 摘 要 】
Accurate and efficient wave equation modeling is vital for many applications in such as acoustics, electromagnetics, and seismology. However, solving the wave equation in large-scale and highly heterogeneous models is usually computationally expensive because the computational cost is directly proportional to the number of grids in the model. We develop a novel high-order multiscale finite-element method to reduce the computational cost of time-domain acoustic-wave equation numerical modeling by solving the wave equation on a coarse mesh based on the multiscale finite-element theory. In contrast to existing multiscale finite-element methods that use only first-order multiscale basis functions, our new method constructs high-order multiscale basis functions from local elliptic problems which are closely related to the Gauss-Lobatto-Legendre quadrature points in a coarse element. Essentially, these basis functions are not only determined by the order of Legendre polynomials, but also by local medium properties, and therefore can effectively convey the fine-scale information to the coarse-scale solution with high-order accuracy. Numerical tests show that our method can significantly reduce the computation time while maintain high accuracy for wave equation modeling in highly heterogeneous media by solving the corresponding discrete system only on the coarse mesh with the new high-order multiscale basis functions. (c) 2018 Elsevier Inc. All rights reserved.
【 授权许可】
Free
【 预 览 】
Files | Size | Format | View |
---|---|---|---|
10_1016_j_jcp_2018_01_032.pdf | 4498KB | download |