Radu Popescu -
Jean H. Prevost -
George Deodatis
Department of Civil Engineering and Operations Research
Princeton University, Princeton, New Jersey 08544
Many physical systems in general and soil materials in particular exhibit relatively large variability in their properties, even within so called homogeneous zones. Deterministic descriptions of this spatial variability are not feasible due to prohibitive cost of sampling and to uncertainties induced by measurement errors. A more rational approach to geotechnical design is made possible by use of stochastic field based techniques of data analysis, which rely more on analytical methods when dealing with various uncertainties related to soil properties.
The probabilistic characteristics of spatial variability of soil properties are studied based on two sets of in-situ measurement results. The first case study uses the results of a two dimensional measurement array consisting of 24 standard penetration test profiles, performed in a natural soil deposit in the Tokyo Bay area, Japan. The soil deposit is formed of three distinct layers (fine sand with silt inclusions, silty clay, and dense clean sand), and large spatial variations of recorded penetration resistance are observed in both vertical and horizontal directions within each soil layer. The second case is based on the results of a series of cone penetration tests performed at an artificial island in the Canadian Beaufort Sea. Though measured in a supposedly homogeneous man-made soil deposit, the recorded cone tip resistance shows significant spatial variations. The soil properties are modeled as the components of a multi-dimensional, multi-variate, non-Gaussian stochastic field, and the probabilistic characteristics of the stochastic field are estimated based on the in-situ soil test results, using the method of moments and a nonlinear regression procedure. The probability distributions, coefficients of variation, and correlation distances exhibited by the soil properties in the two cases analyzed (a natural and a man-made soil deposit), can be used as guidelines for stochastic analysis of similar soil deposits.| Back to: |
|
|