Global inversion technique for geotechnical engineering applications

Standard methods using seismic waves, which are routinely used for shallow subsurface investigation in engineering, have limitations in characterizing challenging profiles that include low-velocity layers and embedded cavities. The limitations are often due to insensitivity of data used for inversion and pitfalls of local inversion schemes employed in these methods. This research focuses on overcoming these limitations by developing two new methods using both sensitive data and a global inversion scheme, simulated annealing.;The first method is an inversion technique to invert travel times for a wave velocity profile. The technique is based on an extremely fast finite-difference solution of the Eikonal equation to compute the first-arrival time through the velocity models by the multi-stencils fast marching method. The core of the simulated annealing, the Metropolis sampler, is applied in cascade with respect to shots to significantly reduce computer time. In addition, simulated annealing provides a suite of final models clustering around the global solution and having comparable least-squared error to allow determining uncertainties associated with inversion results. The capability of this inversion technique is tested with both synthetic and real experimental surface data sets. The inversion results show that this technique successfully maps 2-D velocity profiles with high variation. The inverted wave velocity from the real data appears to be consistent with cone penetration test (CPT), geotechnical borings, and standard penetration test (SPT) results.;Employed for site characterization of deep foundation design, the developed technique is applied to combined surface and borehole data. Using the combined travel time data, the technique enables to provide credible information of material at the socket and partially detect anomalies near the socket. This becomes very important because the material at and near the socket often carries a majority of load from foundations. The inversion results of the combined data, including inverted profiles and associated uncertainties, enable to characterize spatial variability in geotechnical engineering physical parameters of subsurface formations useful in the design of deep foundations. This will be particularly useful in implementing the new LRFD design methodology that can explicitly account for spatial variability and uncertainty in design parameters.;The second method is an inversion technique to invert full waveforms for a wave velocity profile. The full waveform inversion scheme is based on a finite-difference solution of 2-D elastic wave equation in the time distance domain. The strength of this approach is the ability to generate all possible wave types (body waves and surface waves, etc.) and thus to simulate and accurately model complex seismic wave fields that are then compared with observed data to infer complex subsurface properties. The capability of this inversion technique is also tested with both synthetic and real experimental data sets. The inversion results from synthetic data show the ability of detecting reverse models that are hardly detected by traditional inversion methods that use only dispersion property of Rayleigh waves. The inversion result from the real data is generally consistent with the cross-hole, SPT N-value, and material log results.