Differential soil settlement

[Juhiimi]

Hello,


Our design software recently introduced the option to define soil parameters at the foundation level. As a result, the entire building now experiences additional stresses, especially in concrete beams and slabs — and the effect is quite significant.


Should we consider this additional stress in the superstructure, or is it sufficient to account for it only in the substructure?
Also, if we do take it into account, should we apply the standard safety factors for loads, or is it acceptable to use reduced values in this case?

 

[HTURKAK]

the option to define soil parameters at the foundation level. As a result, the entire building now experiences additional stresses, especially in concrete beams and slabs

Lets make clear in this context , I imagine you provide WINKLER spring constants and the software calculates the settlement and differential settlement as per the load experienced by the footing.
The settlement calculation ( say predicition) is not so simple.

Your internet country code implies that the Eurocodes are applicable. The differential settlements should be considered for the design of foundation and superstructure.
Pls look Basis of structural design EN 1990 for detail and combinations.

 

[Cleilson Bernardino]

Greetings!
It is likely that your design software has introduced soil-structure interaction through winkler springs or similar. However, it is important to interpret these additional stresses with caution.

If the beams and slabs showing significant additional stresses are those located at the ground level, in direct contact with the soil, this may be a consequence of an unrealistic modeling effect — something like a parachute effect. In this situation, the slab-beam system is artificially restraining the settlement of the columns supported by the foundations. In practice, this restraint is unlikely to occur because during construction, the soil layer beneath these slabs and beams will naturally be mobilized by construction teamwork. There will be a natural accommodation and the rigid restraint implied by the model will not materialize. Therefore, the significant additional stresses predicted in the superstructure due to this soil interaction may not be realistic. Moreover, this modeling approach will also tend to underestimate the reactions transferred from the columns to the foundations.

For this reason, I would generally recommend not modeling a direct connection (i.e., springs) between ground-level beams and slabs with the soil, unless this is explicitly desired and you are ensuring that the soil layer below the slab is properly compacted and prepared to act as a structural support element.

 

[Juhiimi]

Thank you for your responses.
We are using AxisVM and this is the module that i'm talking about: AxisVM Soil. It's a little bit more complex than Winkler springs, because it uses Soil Finitie Elements (as they call it). What's your oppinion about it?

 

[HTURKAK]

It's a little bit more complex than Winkler springs, because it uses Soil Finitie Elements (as they call it). What's your oppinion about it?

I screened the doc ( AxisVM Soil..) I guess the soil is treated as an elastic medium and 3D finite element model used for the solution. This could be realistic for uniform soils however for the complex soil conditions with time dependent changing water content etc the solution is likely invalid.
I must say that still Linear elastic FEM models would give better results than the Winkler model.
The doc. does not state the validation of results for different soil profiles.

Soil - structure interaction is an interesting complex topic and still needs more digging.

 

[Cleilson Bernardino]

Finite element modeling is better than using the Winkler model, which simplifies multiple soil layers — no matter how homogeneous they may be — into a single value. Additionally, in the Winkler approach, the springs do not interact directly with each other. A point load applied at one node only affects adjacent nodes through the connection of the structure above the ground, not through stress propagation within the medium. As a result, the soil between nearby but unconnected structures is not mobilized.

PS: Saying that a FEM-based model is better does not mean that the Winkler model is bad. It is useful and serves as a good first approximation. What I usually do is assign different value ranges (0.5x and 2x the vertical subgrade reaction coefficient, kv) and assess how sensitive my structure is to differential settlements by comparing the models.
PSS: Although the FEM-based model is technically more appropriate, it will only be as good as the accuracy of the input properties provided.

 

[Greenalleycat]

Soil-structure interaction is real and will affect the stresses in your building, we know that much
Actually assessing the proper soil properties so that your model is accurate...now that's a tricky one
I think Cleilson's response is pretty bang on the money

 

[human909]

If you are getting "extra" stresses on your structure due to soil settlement then you must be getting differential settlement. Check the magnitude of the deflections and their differences and ensure they are realistic. A soil modelling error could readily introduce unrealistic differences in settlement leading to significant stresses in the foundations and superstructure.

Or the soil could be being modelled correctly and you'll need to account for that.