Geotechnical Database Manipulation to Effect Stochastic Analysis

by James M. Keane, D.O.T. Geotechnical Unit, Raleigh, United States,

Document Type: Proceeding Paper

Part of: Probabilistic Mechanics and Structural and Geotechnical Reliability


Existing or easily-developed databases of regional geotechnical information can be used to estimate prior probability distributions and parameters for use in stochastic analyses. Prior distribution parameters based on extensive regional data are updated using Bayesian methods with any available site-specific information. The ensuing stochastic analysis makes optimum use of bulk and specificity of all available information. Two databases of the engineering properties of North Carolina soils have been compiled for consolidation and strength characteristics. A simple Bayesian-like updating scheme is proposed for use with these data which recognizes that the coefficients of variation of many soils properties are often approximately constant. A third database contains quantitative information pertaining to over three hundred landslide events on North Carolina highway slopes. Empirical conditional probabilities of factors contributing to instability derived from this database are used to update the probability of failure for a slope using a direct application of Bayes' Theorem.

Subject Headings: Stochastic processes | Databases | Soil properties | Geotechnical investigation | Data analysis | Probability | Bayesian analysis | Soil analysis | North Carolina

Services: Buy this book/Buy this article


Return to search