Radu Popescu -
Jean H. Prevost -
George Deodatis
Department of Civil Engineering and Operations Research
Princeton University, Princeton, New Jersey 08544
Experimental evidence shows that there is significant spatial variability of material properties within so called ``homogeneous'' soil layers, in both natural and artificial soil deposits. It has been proved using Monte Carlo simulations of soil liquefaction that the soil system response is strongly affected by this variability, and therefore it is important to account for it in analysis and design. For the same average values of soil parameters, more pore water pressure build-up was predicted by Monte Carlo simulations, accounting for spatial variability of the soil properties than by deterministic analyses using the average values of in-situ recorded soil properties. A possible explanation is that soil liquefaction is triggered by the presence of loose pockets in the soil deposit. Deterministic design cannot account for such variability, and therefore, to avoid non-conservative results, geotechnical engineering practice relies on sometimes large factors of safety.
The objective of this study is to provide the geotechnical design with a characteristic value of soil strength (expressed as a percentile exceedence criterion) which, when used in deterministic soil liquefaction assessment, will provide an equivalent response (in terms of predicted excess pore pressure build-up and horizontal displacements) with that predicted by more expensive Monte Carlo simulations. For convenience, the soil strength is expressed in terms of the stress normalized cone tip resistance, for which several correlations with observed field behavior have been developed. The soil properties over the analysis domain are modeled as the components of a non-Gaussian stochastic field. Parametric studies are performed, involving various probability characteristics of this stochastic field, which are most likely to influence the dynamic behavior: probability distribution, degree of variability, correlation distances. A conservative characteristic value of soil strength is estimated by comparing the the results of a series of deterministic analyses using various percentiles of soil strength with the upper bound of Monte Carlo simulation predictions, resulted from a large number of generated sample functions. The dynamic analyses are performed using a multi-yield plasticity soil constitutive model implemented in the finite element code DYNAFLOW, which has been repeatedly validated in the past for soil liquefaction potential assessment.| Back to: |
|
|