Recommendations on the quantification of spatial correlation in ground properties and its effects on geotechnical design

Abstract

Geotechnical design must address unknown, complex, and heterogeneous ground conditions, which are typically investigated through direct methods such as core drilling combined with indirect methods like standard penetration tests. However, such ground investigations often provide only limited, point-specific information. To make full use of the available information for the optimization of foundation structures, the upcoming second generation of Eurocode 7 will explicitly allow the application of probabilistic and statistical methods in geotechnical design. In this paper, the well-known method of moments (MoM), also referred to as the geostatistical variogram approach, is explained and compared with the maximum likelihood method (ML) to estimate the spatial correlation of ground properties. Both methods are applied in two case studies that demonstrate the applicability, strengths, and limitations in a practical, non-academic site investigation. Additionally, the concept of spatial averaging is explained as a means of incorporating the derived (auto-)correlation length into practical geotechnical design within the framework of the semi-probabilistic design approach.