It may be the case that at the higher percent filled value the mode of failure is changing. What’s happening is the pore pressure exceeds the weight of the wedge and the contact is lost between the wedge and the slope. What happens in *Swedge* is the following:

The increase in water pressure is associated with a drop in effective normal stress on the failure plane. In the image below, normal stress versus percent filled is plotted. At about 90% filled, there is no effective normal stress on the failure plane. When effective normal force drops to zero on a failure plane, *Swedge* assumes that contact is lost and that all shear strength, including cohesion, drops to zero. This is why you see the sudden drop in Factor of Safety; basically the cohesion is lost as soon as the normal force hits zero. If you look at the distribution in Factor of Safety without cohesion (set it to zero), you’ll see a much smoother transition in Factor of Safety with increase in water pressure.

Read over Chapter 7, “A Slope Stability Problem in Hong Kong”, of Hoek’s practical Rock Engineering Text. It talks a little about floating wedges on page 14.

*Percent Filled vs Effective Normal Stress in Joint 1*