Soil Springs in ULS and SLS load combinations 4

[StrEng79]

Hi,

I am making a structural FEM model of a hospital building. At the moment I am considering going with spread foundation. Footings for columns, strip footings for walls and larger bases for core walls. I wonder whether the same spring stiffness should be used in a structural FEM model for ultimate limit state (factored forces) and service limit state load combinations (not factored forces)?

On the other hand, I know that a sensitivity study should be carried out with half spring stiffness and double spring stiffness. Does this halving and doubling of spring stiffness in sensitivity study make my doubt about ULS and SLS stiffness obsolete? As this study covers all possibilities.

Thanks

 

[Settingsun]

Whether the ground response becomes significantly non-linear depends on how heavily you're stressing the ground. It might still be fairly linear under ULS loading if the footings are oversized for any reason, eg fairly light loading applied to a minimum-size pad. The first step would be to do the sensitivity analysis to see whether it's worth considering further. You can also consider whether you're being conservative in some way that would compensate for inaccuracy in the spring stiffness. For example, you might not taking advantage of permitted redistribution of moments in a concrete frame, which gives some margin against increased stresses due to lower ground stiffness at ULS.

 

[r13]

Because the soil does not know the difference between ULS (ultimate limit state) and SLS (service limit state) loads, but reacts to the intensity of the loadings, and behaves in according to its properties, which is site unique; so for your first question, you should use the spring constant derived from the soil modulus for both ULS and SLS analyses, and make sure the bearing strength and settlement criteria for each of the limit state are satisfied. Note that there is only one set of soil properties, but acceptance criteria differ.

For important project in ULS analysis, I see the need to utilizing varying stiffness method to capture structural weakness due to variation in soil stiffness. For SLS, please check with your project geotechnical engineer/consultant.

 

[StrEng79]

@steveh49
Thanks. Good point. I will investigate what happens with the soil in ULS range. Whether linearity still holds.

@r13
Thanks. Noted. I intend to make a couple of iterations as follows:

1. Assign set of springs based on received geotechnical data.
2. Check bearing stress.
3. Check estimated settlement and see whether it matches with spring stiffness in step 1.
4. Adjust spring stiffness and re-run calculation.
5. Repeat steps 2-4 until convergence.
6. Double spring stiffness & check effects on superstructure. [sensitivity study]
7. Halve spring stiffness & check effects on superstructure. [sensitivity study]

Does this make sense?

 

[r13]

I would only change step 4 to adjust footing type/size to bring bearing and settlement within allowable. Then carry out steps 6 and 7 to expose structural weakness under extreme soil conditions.

Note that there is another approach you may want to look into - provide stiffer spring under concentrate load, and weaker spring on other areas/elements. But always discuss with the geotechnical engineer to see whether the results match site conditions, and make sense.

 

[StrEng79]

@r13

I would only change step 4 to adjust footing type/size to bring bearing and settlement within allowable.

Thanks for pointing that out. That should also help keep pressure under foundations uniform as much as possible to reduce differential settlement.

 

[r13]

Yes, that's the goal.

 

[HTURKAK]

7. Halve spring stiffness & check effects on superstructure. [sensitivity study]

Does this make sense?]

Yes..It makes sense.. The use of terms SLS and ULS implies EC -7. My suggestion will be , if you are using linear springs , you may define two parametric equation for springs. The first for SLS , tangent to the P-Y curve, and second for ULS , secant which the linear K function from origin to the point of intersection of ULS load Y curve..

If the software you are using SAP 2000 ( I am familiar ) , you may define non linear spring and perform the analysis for the whole system..

Good luck..

 

[ATSE]

Another parameter to consider:
For partially and fully saturated silts and clays, the modulus (your vertical spring) increases with loading rate. In the absence of soil instability (liquefaction, cyclic softening, bearing failure), you might increase your spring k values by a factor of 2 or more (especially for soft clays).
This may be more influential and meaningful for your results, as compared to a detailed non-linear spring based on long term loading. Hopefully your GEOR has the experience and sophistication to provide long term and short term stiffnesses.

 

[StrEng79]

Yes..It makes sense.. The use of terms SLS and ULS implies EC -7. My suggestion will be , if you are using linear springs , you may define two parametric equation for springs. The first for SLS , tangent to the P-Y curve, and second for ULS , secant which the linear K function from origin to the point of intersection of ULS load Y curve..

If the software you are using SAP 2000 ( I am familiar ) , you may define non linear spring and perform the analysis for the whole system..

Thanks, HTURKAK.
I will check this out.

Another parameter to consider:
For partially and fully saturated silts and clays, the modulus (your vertical spring) increases with loading rate. In the absence of soil instability (liquefaction, cyclic softening, bearing failure), you might increase your spring k values by a factor of 2 or more (especially for soft clays).
This may be more influential and meaningful for your results, as compared to a detailed non-linear spring based on long term loading. Hopefully your GEOR has the experience and sophistication to provide long term and short term stiffnesses.

Thanks, ATSE.
I will keep your post in mind while sorting out my foundation.