Shear strength, cohesion, and overconsolidation in low-stress sediments and their importance for submarine slope failure

Abstract:

Factor-of-safety analyses of submarine slope failure depend critically on the shear strength of the slope material, which has two components: friction and cohesion. While evaluating friction is from laboratory testing is common, cohesion is commonly ignored despite its potential importance in resisting failure. Here, we report on laboratory experiments conducted at effective normal stresses of < 2 MPa where we measure shear strength, but and also directly measure sediment cohesion by measuring its shear strength in a direct-shear apparatus by removing the applied effective normal stress (σ_n’ = 0). We document systematic behavior demonstrating that cohesion depends positively on clay mineral content and consolidation stress. The dependence on clay content suggests that the mechanism of cohesion is hydrogen bonding between charged clay surfaces and water molecules. The stress dependence indicates that estimating cohesion from extrapolating a Coulomb-Mohr failure envelope could lead to inaccurate predictions. Furthermore, the proportion of shear strength attributable to cohesion is higher at lower stresses corresponding to shallow depths where landslide failures are expected.

Measuring shear strength and cohesion over a wide range of overconsolidation ratios (OCR) for sediments containing clays show that significantly higher peak strengths are expected to occur for OCR > 4, and the primary source of this strength increase is not friction, but rather increased cohesion which depends log-linearly on the OCR. Our data suggest that in areas which have experienced unroofing due previous mass movements, overconsolidated clays can be stronger than cohesionless sediments such as quartz silt/sand. Overconsolidated clays would exhibit increased peak strength toward the surface, therefore failure would be expected to occur deeper where the OCR is lower. In seismically active areas, this could explain why slope failure recurrence appears to be lower than expected when considering potential triggering from local earthquakes.