Optimization of site exploration program based on coupled characterization of stratigraphic and geo-properties uncertainties.

Due to the inherent spatial variability of the geological bodies, limited boreholes (constrained by budget), and data interpretation errors, the subsurface conditions (in terms of the stratigraphic configuration and associated geo-properties) at a site are hard to ascertain, and considerable stratigraphic and geo-properties uncertainties are induced. The stratigraphic and geo-properties uncertainties could complicate the engineering design. One way to reduce the uncertainties is to adopt an optimal site exploration program under constraints. This paper presents a novel framework for optimizing the site exploration program based on a coupled characterization of the stratigraphic and geo-properties uncertainties. Within the context of the proposed framework, a conditional random field approach, recently advanced by the authors, is employed to characterize the stratigraphic and geo-properties uncertainties simultaneously. Here, the spatial correlation of stratum existence (depicting stratigraphic configuration) and that of geo-properties are estimated with a Bayesian updating scheme. Then, the 'influence zone' at the concerned site, which is often problem-specific, is selected based on a geotechnical performance evaluation. The derived influence zone can be taken as the candidate zone for boreholes. Finally, the optimal locations and depths of boreholes are determined based on the stratigraphic and geo-properties uncertainties characterized within the influence zone. To depict the effectiveness and versatility of the presented framework, two problems (i.e., a shallow foundation problem and a slope stability problem) are analyzed. • A novel framework for optimizing the site exploration program is proposed. • Site characterization and geotechnical performance evaluation are integrated. • Characterizations of stratigraphic and geo-properties uncertainties are coupled. • The spatial correlation structure is estimated with a Bayesian updating scheme. • Geological model characterization is updated in a step-by-step manner. [ABSTRACT FROM AUTHOR]