FEM modeling of basement/retaining walls
FEM modeling of basement/retaining walls
(OP)
Other threads on this site have discussed modeling basement walls as pinned-pinned, pinned-fixed, or somewhere in between, as well as transfer of lateral earth pressures into diaphragms and abutting walls. My question is a bit more generic, and I'm interested in hearing how others approach this. Let's say there's a cast-in-place basement wall system with full unbalanced soil on one side. It's a walk-out basement (i.e., concrete walls wrap only part of the building footprint). The geometry of the system in plan has multiple jogs, angles and returns, such that there's obviously some inherent ability for the entire system to self-stabilize, without relying on diaphragms. In fact this would be a relevant condition during construction, if the walls are backfilled prior to the construction of the diaphragm. Designing the wall as a simple 1' long cantilevered retaining wall (not taking into account wall returns) would probably be overly conservative. Designing it as pinned-pinned is not realistic if there's no diaphragm and the contractor doesn't want to temporarily brace the top of the wall, or if the returning walls are minimal or widely spaced. If the wall returns are small or modest in comparison the overall geometry, it seems necessary to model the entire geometry within a generic FEM package, and to approximate the contribution of the returning walls, the partial fixity of the wall to the footing, the two-way spanning of the wall elements, etc. Is it reasonable to go to this level of effort on something like a house, or does it just not get done? One particular question is how people model the fixity of the base of the wall to the footing. Standalone retaining wall packages know how to do this for design of basic wall sections designed in isolation, but within a generic 3d FEM package, it seems like some fake (dummy) buttresses are needed to actually transfer the out-of-plane bending due to lateral earth pressures from the base of the wall to the footing.





RE: FEM modeling of basement/retaining walls
Building codes pretend that basement walls span vertically but the anchorage at the top of a wall (that WOULD be required if one way action alone was at play) is often not considered properly or is not readily achieved.
I think two way bending is usually what makes a basement wall work as well as it does and based on the forensic work that I have done over the years, THAT is where it's at.
My experience has lead me to consider horizontal bending more likely than vertical bending UNLESS the wall is too long.
I have used of Visual Analysis FEM to try to get to the actual bending behavior HOWEVER - I THOUGHT I read somewhere that VA doe not performa "block stability check". HAVE YOU SEEN THAT ?
What I HAVE done is eliminate vertical restraint for nodes that are in tension, under a foundation wall. Usually this is an trial and error process.
Thoughts?
RE: FEM modeling of basement/retaining walls
I don't know how they establish the spring constants (on their model they use springs for the soil under the footing) and I think one would need to fiddle around a bit to get confidence that the footing "fixity" is well modeled but...looks like a good start.
HAVE YOU SEEN THAT and if so, what do you think?
RE: FEM modeling of basement/retaining walls
A common way to model soil-foundation interaction is to use compression only springs. If the footing or wall is modeled with plate elements, you can use the Soil Spring Generator found in the Model Menu by first selecting the plates you want to be supported by springs. From there, you will need to enter a subgrade modulus, which is usually provided from a geotechnical engineer for the specific soil conditions. The soil spring generator figures out the tributary area to each spring and adjusts the spring constant "k" accordingly.