Download presentation
Presentation is loading. Please wait.
Published byLauren Parsons Modified over 9 years ago
1
J. Diaz, D. Kolukhin, V. Lisitsa, V. Tcheverda Coupling of the Discontinuous Galerkin and Finite Difference techniques to simulate seismic waves in the presence of sharp interfaces J. Diaz, D. Kolukhin, V. Lisitsa, V. Tcheverda 1
2
Motivation for mathematicians 2 σ = 1.38, I = 44.9 м Free-surface perturbation
3
Motivation for mathematicians 3 σ = 1.38, I = 44.9 м Free-surface perturbation 30%
4
Motivation for geophysicists 4
5
5
6
6 Original source
7
Motivation for geophysicists 7 Diffraction of Rayleigh wave, secondary sources
8
Motivation 8
9
Standard staggered grid scheme 9
10
10 Easy to implement Able to handle complex models High computational efficiency Suitable accuracy Poor approximation of sharp interfaces
11
Discontinuous Galerkin method Elastic wave equation in Cartesian coordinates:
12
Discontinuous Galerkin method
14
Use of polyhedral meshes Accurate description of sharp interfaces Hard to implement for complex models Computationally intense Strong stability restrictions (low Courant numbers)
15
Dispersion analysis (P1) Courant ratio 0.25
16
Dispersion analysis (P2) Courant ratio 0.144
17
Dispersion analysis (P3) Courant ratio 0.09
18
DG + FD 18 Finite differences: Easy to implement Able to handle complex models High computational efficiency Suitable accuracy Poor approximation of sharp interfaces Discontinuous Galerkin method: Use of polyhedral meshes Accurate description of sharp interfaces Hard to implement for complex models Computationally intense Strong stability restrictions (low Courant numbers)
19
A sketch P1-P3 DG on irregular triangular grid to match free-surface topography P0 DG on regular rectangular grid = conventional (non-staggered grid scheme) – transition zone Standard staggered grid scheme 19
20
Experiments 20 PPWReflection 15~3 % 30~0.5 % 60~0.1 % 120??? % DG
21
FD+DG on rectangular grid P0 DG on regular rectangular grid Standard staggered grid scheme 21
22
Spurious Modes 2D example in Cartesian coordinates
23
Spurious Modes 2D example in Cartesian coordinates
24
Interface Incident waves Reflected waves Transmitted artificial waves Transmitted true waves
25
Conjugation conditions Incident waves Reflected waves Transmitted artificial waves Transmitted true waves
26
Experiments 26 PPWReflection 151.6 % 300.5 % 600.1 % 1200.03 %
27
Numerical experiments 27 P S Surface Xs=4000, Zs=110 (10 meters below free surface), volumetric source, freq=30Hz Zr=5 meters below free surface Vertical component is presented Source
28
Comparison with FD 28 DG P1 h=2.5 m. FD h=2.5 m. The same amplitude normalization Numerical diffraction
29
Comparison with FD 29 DG P1 h=2.5 m. FD h=1m. The same amplitude normalization Numerical diffraction
30
Numerical experiments 30 Xs=4500, Zs= 5 meters below free surface, volumetric source, freq=20Hz Zr=5 meters below free surface
31
Numerical Experiments 31
32
Numerical Experiment – Sea Bed Source position x=12,500 m, z=5 m Ricker pulse with central frequency of 10 Hz Receivers were placed at the seabed.
33
Numerical Experiments
34
Conclusions Discontinuous Galerkin method allows properly handling wave interaction with sharp interfaces, but it is computationally intense Finite differences are computationally efficient but cause high diffractions because of stair-step approximation of the interfaces. The algorithm based on the use of the DG in the upper part of the model and FD in the deeper part allows properly treating the free surface topography but preserves the efficiency of FD simulation.
35
Thank you for attention 35
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.