Exploring Submarine Mass Failures with Physical Experiments and Numerical Models

Abstract:

We use physical experiments to explore the onset and the morphology of submarine mass failure. Previous experiments found that cohesive sediment alters foreset mass-flow dynamics away from the regular, small avalanches characteristic of cohesionless grain flows to produce a broad range of mass failure events. Prefailure morphology is convex from the knick-point to the delta top. Failure initiates at the knick-point, erodes headward, and creates a scarp steeper than the initial profile. Normalized size of event-type failures does not correlate with water or sediment supply. Experiments also showed increased background creep with increased sediment supply. Building on these experiments, we examine how overpressure and focused fluid flow affect failure. Our initial system is hydrostatically pressured. Pore pressure is then increased in a confined, permeable layer to initiate failure. We hypothesize that the rate of overpressure increase correlates with the size of the initial failure. A rapid pressure increase will decrease the effective stress over a larger slope area before it dissipates. We also control clay content to evaluate the impact of cohesion on failure morphology and runout. We estimate that increased cohesion produces failure blocks that shear on discrete surfaces whereas decreased cohesion produces granular flows and longer runout. The experiments are designed to allow us to control sediment properties and to isolate individual parameters. Numerical models at the same scale of the physical experiments allow us to predict timing and location of failure based on hydrologic and strength properties and the rate of pressure increase in the permeable layer, in turn allowing us to test models that predict the onset of failure. Another suite of experiments aims to investigate pore pressure and slope stability in response to ground accelerations, a mechanism that is often invoked as triggering large submarine mass failure.