No way to tell from the information you have posted. We would need to know the geometry, the other soil strengths, piezometric line(s), location of critical failure surfaces, etc.

I suggest you look at the convergence data and the interslice forces for both runs.

I expect that you will find errors in one or both of the runs. Make sure you look for negative side forces.

Also, you do not provide the units for you unit weight so I do not know if 21 is reasonable or not.

Mike Lambert

The judicious use of the Bedrock material truncates the modeled failure surface downward (or outward) extension, resulting in a shorter failure surface, which results in a lower FOS. It is a geometry problem. Carefully observe the pictures of the computed failure surfaces, which are generated by the program.

I have made use of this when the observed slope failures were 'circular' until reaching a specific horizon, at which point the observed failure was along the limiting horizon. I have used the 'bedrock material' to accomplish a modeled surface which matches that observed in the field. As a side note, I have also used the 'tension crack' (without water filling) to also achieve modeled geometry which nearly matches the field observations.

I must point out that the actual limiting soil/rock horizon is sometimes only somewhat 'stronger' than the upper failed soils and my 1st inclination was to assign a 'Mohr-Coulomb' soil.

Thanks emmgjld. Very insightful.
Thanks Mike Lambert too, for the response.

What's your overlying material strengths? would help to know more about your model

Emmgjld - I have also used a similar method that you have discussed for modelling a slip similar to field observations. Instead of using a bedrock material I just forced the slip to fail above the denser material which is what we believed happened. in Slide6 there is an option to force your failure surface through selected points.