Modeling SOG in SAFE - Convert to Nonlinear Cases

[EngDM]

I've got a slab that has some discrete tension/shear points that I've been running in SAFE. To get the design footprint I inputted my subgrade modulus and ran the slab, and increase the slab thickness/footprint until I no longer was getting positive bearing values, I.E. slab doesn't uplift.

I was watching a video on creating an envelope for the soil pressure, and the presenter converted all of the load cases to nonlinear load cases, and now when I run my slab I don't have uplift pressures, it only displays as 0; it is only showing where the slab pushes down on the soil with no indicator (other than it being zero) that the slab may be in uplift.

I'm just curious as to what this non-linear case actually performs, and if I should have converted my cases to non-linear at all. I was never taught to do that by my co-workers.

Also, after converting to non-linear, my load combinations are no longer defined by my load cases, but have been "flattened" into a single one, so to convert back to linear I apparently have to re-type all of my load cases...

 

[Celt83]

For compression only springs my understanding is that the solution process iteratively either deactivates or significantly reduces the spring stiffness at locations where the deformations would result in tension reactions, I believe on each successive iteration is also checks that the new deformation doesn't result in compression and if so reactivates or increases the stiffness factor.

Since this a non-linear solution process the principle of superposition no longer applies to each load combination needs to be solved rather than solving each load case and then combining the results.

 

[Celt83]

Difference between a linear run where the spring is allowed tension vs non-linear compression only run:



your method of adding thickness to the left model to remove the uplift would in this case be restoring the resultant load point to within the kern.

the negative value on the right model is related to the Finite Element model used by the program, it allows for a linear variation in the spring constant from the i to j end of the member. To remove the negative number I would need more element subdivisions to get a better approximation of the solution.
subdivided into 8 segments it converges to all compression:

 

[PMR06]

Regarding the load combo's being "flattened", SAFE converted that combo into a non-linear case. It should show up in the list of load cases with -NL at the end.

I am not aware of an automated way to undo the "create non-linear combo" process, but the information is still there in the cases and you can use the database editor and Excel to pretty quickly copy/paste that data into new linear combos.

 

[EngDM]

Difference between a linear run where the spring is allowed tension vs non-linear compression only run:

View attachment 1999

your method of adding thickness to the left model to remove the uplift would in this case be restoring the resultant load point to within the kern.

the negative value on the right model is related to the Finite Element model used by the program, it allows for a linear variation in the spring constant from the i to j end of the member. To remove the negative number I would need more element subdivisions to get a better approximation of the solution.
subdivided into 8 segments it converges to all compression:
View attachment 2001

Interesting. So performing an NL solution actually amplifies my compression. So is running an NL solution where all uplifts become zero stress similar to your diagrams a correct analysis? Would this ensure no uplift?

For compression only springs my understanding is that the solution process iteratively either deactivates or significantly reduces the spring stiffness at locations where the deformations would result in tension reactions, I believe on each successive iteration is also checks that the new deformation doesn't result in compression and if so reactivates or increases the stiffness factor.

Since this a non-linear solution process the principle of superposition no longer applies to each load combination needs to be solved rather than solving each load case and then combining the results.

When you say it needs to be solved first rather than each load case and superpositioning, does the software do this once I specify the case as non-linear, with the load patterns inputted into the load case?

 

[Celt83]

In my opinion your analysis that lets the springs go into tension is not consistent with the boundary conditions created by soil, soil like concrete is usually idealized to have no tensile capacity.

For the nonlinear run I’d take a look at the software manual on that topic or reach out to their tech support.

 

[EngDM]

In my opinion your analysis that lets the springs go into tension is not consistent with the boundary conditions created by soil, soil like concrete is usually idealized to have no tensile capacity.

For the nonlinear run I’d take a look at the software manual on that topic or reach out to their tech support.

If I were to increase slab size to not get tensile in the springs, should be okay no?

 

[sticksandtriangles]

If I were to increase slab size to not get tensile in the springs, should be okay no?

Yes, that would be ok and consistent with the assumption that soil is a compression only element as Celt mentioned.

Increasing slab size to remove all tension in your springs will likely yield a very large and conservative footing that will likely get you a few blank stares from contractors.

I generally make a few nonlinear load cases in SAFE (with compression only soil springs), reviewing soil load pressures under these specific cases and make sure that ~50% of your footing is still in compression to prevent overturning.

 

[EngDM]

Yes, that would be ok and consistent with the assumption that soil is a compression only element as Celt mentioned.

Increasing slab size to remove all tension in your springs will likely yield a very large and conservative footing that will likely get you a few blank stares from contractors.

I generally make a few nonlinear load cases in SAFE (with compression only soil springs), reviewing soil load pressures under these specific cases and make sure that ~50% of your footing is still in compression to prevent overturning.

Yea this is for a SOG with equipment on it that generates moments of 1100kn-m. If the slab is large so be it. Rather a heavy slab then damaging super expensive equipment.

 

[milkshakelake]

I have done mat slabs with large and expensive equipment, and my design was to make sure 50% of it is on soil like what sticksandtriangles said. I generally take my normal load combinations and make them all nonlinear. Some people do cracking analysis, which I think is overkill for foundations on soil, and more appropriate to elevated slabs, at least for small things. But it's your prerogative and engineering decision to make it fully bear on soil in all cases, including wind and seismic.