Mat foundation on expansive soils - preferred software? 1

[1835Eng]

I'm trying to determine if any commonly available FEM foundation software design packages such as RAM Concept, STAAD Foundation (or other) are capable of modelling large mat foundations subject to edge lift and center lift conditions? I have access to both RAM and STAAD here at the office, but I don't see where either explicitly accepts edge moisture variation distance or soil heave input values. I'm curious how others are modelling these soil loading conditions; any suggestions would be appreciated.

 

[humanengr]

Based on the information you provided, I would model the mat foundation and boundary conditions.
I would then apply loads (upward loading) to account for edge lift / center lift.
Is there any reason why you could not apply upward loading to simulate the lift conditions and combined those loads in the appropriate load combinations?

 

[1835Eng]

I considered the above approach for modelling the center lift ("edge drop") condition, whereby I would apply the soil springs to the mat, inset by the edge moisture variation distance, so that the perimeter of the foundation cantilevered beyond the soil support area. Haven't tried it yet, but intuitively it seems to make sense.

I don't think the edge lift condition is that straight-forward. The soil bearing pressure should be parabolic, highest at the perimeter and decreasing to a more uniform value near the center. It seems like it would be more complex than simply applying a uniform upward-acting load to some area of the slab perimeter. That's why I'm curious how others are tackling the problem.

 

[JoshPlumSE]

I work for a company that has a Foundation program (RISAFoundation), but have to admit that I never really dealt with expansive soils in my previous life as a design engineer. I understand the concept of soil heave. But, I've never calculated what those forces would be on one of my structures.

Would you be interested in your program doing this type of calculation automatically.... Or, are you just interested in being able to apply this type of uplift load manually into the program. The 2nd option seems pretty easy, I would think every foundation program out there would allow you to do something like this. It's just another type of load.

The first option, however, is interesting to me conceptually. What would this type of calculation involve? From an engineering calc standpoint I can look that up. But, I'm wondering from a program interface standpoint. For the computer to run these numbers automatically what would it need to know about the foundation and soil. My tendency is to think that it would too complex of an issue to handle easily. Involving not just soil properties, but also geometry of the foundation (and above grade structure), maybe even drainage issues for the soil and site.

 

[1835Eng]

In a perfect world, I would enter the active soil parameters from the geotech report (edge moisture distance, heave, etc.) and the program would perform the calcs, similar to a spreadsheet or design program based on the PTI procedure. I ran several STAAD Foundation Advanced models considering the loading approaches from the original post above, and the center lift case seemed to run accurately. With the soil springs applied to an inset control region, the moment diagrams followed the profile shown in the USACE ribmat design criteria for center lift, and no errors were reported in the analysis. As expected, the soil bearing pressures were highest at the corners where the load cantilevers in both directions but tapered back quickly.

For edge lift, I had a lengthy conversation with a geotech before developing the model. While this might appear to be a simple load condition, it is far from it. The swell pressure is not uniform, but is tied to the ability of the soil to expand with changes in moisture content. A stiff foundation would confine or "push back" against the soil, limiting its expansion and resulting in high exerted soil pressure. A flexible foundation would bend upward, allowing the soil to reach its full volume-change potential which results in low swell pressure. So the swell pressure and foundation stiffness are related (deflection inversely related to swell pressure), and this is not captured in the simplified loading described in the OP. It represents a theoretical "worst case" assuming the soil expansion is limited to a value resulting in maximum swell pressure, but does not capture the real-world behavior or bearing pressure. The model just sees it as an uplift load that is trying to pick the mat up, instead of the soil pressure increasing below the foundation. Therefore the service load bearing pressure and the edge deflections are fictitious numbers. The model value is limited to providing worst-case design values for concrete reinforcement.

It's not a perfect solution, so I'm still interested to know if there's a better way to approach the problem.

 

[MotorCity]

I am not sure what your particular soil properties are of the expansive soil, but my experience with expansive soil is that it is nearly impossible to try and resist those forces and deal with the excessive movement that can result. We always used piles with a void form below the mat for areas with expansive soils. Have you considered a similar approach?

 

[1835Eng]

With several feet of select fill replacing the near-surface expansive soil, the geotech is estimating heave values of 1". This is within the parameters of an appropriately stiffened mat foundation, and is the construction method preferred by the owner.

 

[MotorCity]

So if the expansive soil effect is not a factor (or negligible), I would design it for uniform uplift assuming a rigid (infinitely stiff) mat that "spans" between walls, columns, or the like above. You can size your mat thickness such that the upward deflection is small. This will result in a high, but uniform swell pressure. This was probably the assumption of the geotechs I worked with (which is why we ended up using piles.) At the end of the day, predicting actual swell pressure is an absolute "best guess" anyway.