Abstract
Modeling seismic waves and vibrations propagating into geological structures has always been a challenge since large scale models as well as detailed basin geometry and soil layering are both needed. The optimal accuracy to model the propagation process depends on the frequency range but also on the excitation level.
Various numerical methods have been significantly improved along the years to achieve accurate and cost-effective strategies: FDM, FEM, FVM, SEM, BEM, DGM. The first key issue deals with numerical dispersion (link between wavelength and propagation features). Numerical damping may also be a (controllable) issue.
Another important question is related to spurious reflected waves at the model boundaries. To reduce such errors, absorbing boundary conditions or absorbing layer methods (PML, CALM, ALID) have gained interest in the recent years.
To model seismic waves propagating from the fault to the structure, a coupling strategy (DRM, FEM/BEM, strong vs weak) may be optimized to accurately model the wave radiation at infinity or accommodate large velocity contrast (i.e. large mesh refinement discrepancies).
Future research challenges are also discussed: physics based fault rupture, nonlinear soil/rock behavior and characterization, loading history, pore fluids, uncertainties vs spatial variability, interaction with various structures at different scales.
