No abstract

This work presents a new high performance open-source numerical code, namely SPectral Elements in Elastodynamics with Discontinuous Galerkin, to approach seismic wave propagation analysis in viscoelastic heterogeneous three-dimensional media on both local and regional scale. Based on non-conforming high-order techniques, such as the discontinuous Galerkin spectral approximation, along with efficient and scalable algorithms, the code allows one to deal with a non-uniform polynomial degree distribution as well as a locally varying mesh size. Validation benchmarks are illustrated to check the accuracy, stability, and performance features of the parallel kernel, whereas illustrative examples are discussed to highlight the engineering applications of the method. The proposed method turns out to be particularly useful for a variety of earthquake engineering problems, such as modeling of dynamic soil structure and site-city interaction effects, where accounting for multiscale wave propagation phenomena as well as sharp discontinuities in mechanical properties of the media is crucial.whereas the matrix A associated to the bilinear form A. , / defined in (12) is such that for i, j D 1, : : : , D it holds A`k ij WD A.ˆj ,ˆk i / NI , for k,`D 1, .., 3.Now, we define V WD P U the vector of nodal velocities, we prescribe initial conditions U.0/ D u 0 and V.0/ D u 1 and we consider the system (13). Let us now subdivide the interval .0, T into N subintervals of amplitude t D T =N and set t n D nt , for n D 1, : : : , N .

International audienceThis article documents a comparative exercise for numerical simulation of ground motion, addressing the seismic response of the Grenoble site, a typical Alpine valley with complex 3D geometry and large velocity contrasts. Predictions up to 2 Hz were asked for four different structure wave-field configurations (point source and extended source, with and without surface topography). This effort is part of a larger exercise organized for the third international symposium on the effects of surface geology (ESG 2006), the complete results of which are reported elsewhere (Tsuno et al., 2009). While initial, blind computations significantly differed from one another, a remarkable fit was obtained after correcting for some nonmethodological errors for four 3D methods: the arbitrary high-order derivative discontinuous Galerkin method (ADER-DGM), the velocity-stress finite-difference scheme on an arbitrary discontinuous staggered grid (FDM), and two implementations of the spectral-element method (SEM1 and SEM2). Their basic formulation is briefly recalled, and their implementation for the Grenoble Valley and the corresponding requirements in terms of computer resources are detailed. Besides a visual inspection of PGV maps, more refined, quantitative comparisons based on time-frequency analysis greatly help in understanding the origin of differences, with a special emphasis on phase misfit. The match is found excellent below 1 Hz, and gradually deteriorates for increasing frequency, reflecting differences in meshing strategy, numerical dispersion, and implementation of damping properties. While the numerical prediction of ground motion cannot yet be considered a mature, push-button approach, the good agreement reached by four participants indicates that, when used properly, numerical simulation is actually able to handle correctly wave radiation from extended sources in complex 3D media. The main recommendation to obtain reliable numerical predictions of earthquake ground motion is to use at least two different but comparably accurate methods, for instance the present formulations and implementations of the FDM, SEM, and ADER-DGM

In this paper, a novel strategy to generate broad-band earthquake ground motions from the results of 3D physics-based numerical simulations (PBS) is presented. Physics-based simulated ground motions embody a rigorous seismic wave propagation model (i.e., including source-, path-and site-effects), which is however reliable only in the long period range (typically above 0.75-1 s), owing to the limitations posed both by computational constraints and by insufficient knowledge of the medium at short wavelengths. To cope with these limitations, the proposed approach makes use of Artificial Neural Networks (ANN), trained on a set of strong motion records, to predict the response spectral ordinates at short periods. The essence of the procedure is, first, to use the trained ANN to estimate the short period response spectral ordinates using as input the long period ones obtained by the PBS, and, then, to enrich the PBS time-histories at short periods by scaling iteratively their Fourier spectrum, with no phase change, until their response spectrum matches the ANN target spectrum. After several validation checks of the accuracy of the ANN predictions, the case study of the M6.0 Po Plain earthquake of May 29, 2012 is illustrated as a comprehensive example of application of the proposed procedure. The capability of the proposed approach to reproduce in a realistic way the engineering features of earthquake ground motion, including the peak values and their spatial correlation structure, is successfully proved.

No abstract