Check your slope stability program on how to model concrete. I have seen both the compression strength used and also 1/2 where it was asssumed a concrete facing would fail in shear. I think a tension failure was more likely, but unlike soil, once you have a failure in concrete it does not regain strength. I do not believe either were accurate for the situation and believe concrete should be modeled as a traditional structural item with reinforcement designed on expected movement.
The strength of the concrete should not really matter since it is many many times the strength of the surrounding soil. Therefore the critical slip surface is not going to pass through the concrete structure. If you are using the search routines included in the software to find the critical surface, just use a very high value for both cohesion and internal friction and make sure that the critical surface does not intersect the structure. Note: you may need to adjust the search routine since some have problems if there is a "chunk" of very high strength material next to some lower strength material. Usually you can adjust the search grid to a finner pattern and still find the critical surface.
If for some reason, the critical surface is passing through the structure, you will need to run a different analysis either by hand or using a different software package since no slope stablility software is going to give acurate and/or meaningfull results in a situation like that.
GPT for the win. That is the way I do it as well. The structural analysis for the wall should make it so that it is resisting the forces put on it. That is more of an internal stability check, not he global that you are alluding to.
I like this method as it forces all failure surfaces around your concrete member, but only if, as you and TDAA mention in your posts, you check for a critical surace through your structure. With most programs inorder to see this you have to *FIRST* model the dike without the concrete. After you do this and design your structure you can shoot the cohesion up.
My concern is that I have seen models where a thin concrete core or surface treatment was used to show factor of safeties 2 or 3 times higher than just earth with no analysis of the internal forces on the concrete facing. Sort of like showing shotcrete as the important part of an MSE wall.
I agree that any structure must be designed for the forces that act on it. However, I do not in general believe that a slope stability program can be used to determine those forces. The exeception is when you have an active slide, i.e. F.S.=1.0. Under those conditions the stability program can be used to determine the forces acting on a stucture being constructed to stabilize the slide.
With respect to your comments about seeing models with a thin layer of concrete and high factors of safety, I have no doubt that you have seen them and that engineers have reported the results. It is hard to just not laugh when someone does that kind of analysis and then REALLY believes the answer that the computer program spits out.
Following up on Blueoak's comment: Sometimes you need to look at the concrete as a structural engineer would. For example, you can't always assume that you can mobilize the shear strength of the concrete or a soil-cement facing for erosion control. The bending resistance may govern, or the facing might lift off. Soil-cement columns for foundation improvement need to be checked for bending, and not just shear - not a rare error.
On the other hand, it may be that the strength just doesn't matter, as GPT said. In your case, the I-wall may be so strong that the supporting soil would fail and not the wall, regardless of whether the concrete is 2 ksi or 6 ksi.
Depending on the geometry of the concrete portion of the wall and the soil properties in the dike, the concrete can concentrate loads at the base thus increasing the soil loads near the failure surface. The net result may be a lower factor of safety than anticipated via simple modeling.
