Modeling spatial variability in offshore geotechnical properties for reliability-based foundation design

• Modeling spatial variability in geotechnical data for foundation design is described.
• A random field model that describes spatial variations is formulated and calibrated.
• Depth-averaged strength are correlated over longer distances than point values.
• The model is incorporated into a reliability-based framework for foundation design.
• The required increased conservatism due to spatial variability is generally small.

Design of foundations for offshore energy production typically requires soil characterization over large areas. Often, in uniform geological settings, it is neither practical nor economical to acquire geotechnical data at every foundation location. Additionally, the zone of interest for the foundation may extend deeper than the available geotechnical data. This paper describes a model of spatial variability in geotechnical properties for foundation design in deep water Gulf of Mexico. The geology consists of normally to slightly over-consolidated marine clays. Data are available for about 100 locations with soil borings, jumbo piston cores and cone penetration tests. A random field model that describes spatial variations in the design undrained shear strength is formulated and calibrated. This model is incorporated into a reliability-based framework to account for uncertainty due to spatial variability in foundation design. In this setting, depth-averaged values of design undrained shear strength are correlated over longer distances than point values due to stratigraphic features. There is less variation and greater spatial correlation in the design undrained shear strength for deeper versus shallower deposits and along the continental shelf versus off from the shelf. The increased conservatism required in foundation design due to spatial variability when site-specific strength data are not available is generally small.