Pile differential settlements

[ToohTaah]

I am wondering how I can incorporate long term differential settlements into my pile foundation.

I have a FEM of concrete base slab containing about 200 piles (some are vertical and some are battered). The piles are modeled as spring supports. The analysis was done. Now, the geotechnical engineers came up with long term settlements for each pile from sustained dead load.

My question is how I can incorporate them to my model since the values I am given did not include the stiffness of the concrete base slab.

 

[kellnerp]

Change the spring constants to reflect the settlement. In other words have different spring constants for each pile.

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

"Node news is good news."

 

[ToohTaah]

Do you mean for only differential settlements load case or for all load cases?

 

[gwolf2]

If you use two-noded springs you could use a prescribed displacement of the earthed node to simulate settlement.

A non-linear spring might also be helpful.

 

[kellnerp]

Now you are asking a question the nobody can answer without additional information. We have no knowledge of what load cases you are running or what the requirements for the analysis are. We are just suggesting ways to model differential settlements.

It seems also that you are doing an iterative solution. Your geotechnical engineers are assuming that the pile will move freely with the load you gave them (I am assuming). But the pile will not move because it is fixed to a relatively rigid slab. So the pile will unload and the adjacent piles will pick up the load causing them to settle at a greater rate.

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

"Node news is good news."

 

[ishvaaag]

I remember from an ACI special publication on mats that for piled mats the measured midterm of the installed system at the Messeturm (a 300 m tall building) and another important building approached a 50% sharing of loads between piles and mats, i.e., the piles would be taking themselves only half of the column loads.

However, I will repeat: except for VERY stiff foundations or foundation plus buildings, the magnitude of the settlement dependes mainly on the magnitude and position of the column loads, i.e., for most cases, I would say, except where it is worth to prove otherwise, it may be useful to think of the foundation be flexible for settlement account. However, that -properly- can't be said without a soil-structure interaction analysis, and likely that's what the geotechnical evaluation can be giving if it has not been counting with the stiffness of the foundation plus building above. So you may first investigate what difference would make in settlements the loads be applied directly on the soil without consideration of the foundation (more or less equating to the geotechnical assumption) and when made with your actual building imparting the loads to the soil. If there is significant difference the estimate of the gotechnical report needs be corrected to account for the standing stiffness of the building. For most cases it won't make a difference but since a piled mat yours might still be one of the cases where foundation plus building stiffness has a say.

You may make a simplified first try by putting in a notional mat the added stiffness of all floors and then support it on stiffer springs at pile points and weaker springs elsewhere, then apply column loads to that piled mat. If, after proper tuning of the springs, your settlements differ in more than say 20 or 30% from those given by the geotechnical report for a significant protion of the area, i.e., you have settlements about 80% or less of those given by the geotech estimate, you may use your standing stiffness and then use your (or other better estimated values) settlement values to account for distortion or deflection issues. For most other cases it may be simpler to add directly an imposed settlement as specified by the geotechnical report at column points, and combine with every service level to see the effects on deflection control.

This same component settlement hypothesis equally without any factor, i.e., with factor 1, should be added to any of the structural strength hypothesis checks.

 

[ishvaaag]

Now to add that in all this process the more important thing is that you understand that the settlements furnished by the geotechnical studies are logical and not severely overestimated. To judge this, the main resource available is the settlement in neighbor buildings, that show real scale the behaviour.

In my view, and I understand this is the traditional practice, the settlements to be used in both the service level and strength checks should be the probabilistic ones, i.e., the actually expected to happen most of the times (95% of the times, average?) under the real loads (this may allow even some typical reduction on the service level live load per floor, as it is made for strength checks). To argue the point of safety factor 1 for settlement in strength checks must be noted that settlement in this long-term view is a rheological phenomenon, and so common treatment to the given to temperature, shrinkage and creep is reasonable.

Others might argue the use of settlements evaluated on the probability of never be surpassed 95% of the times, but this surely would modify in excess the current practice for buildings. In any case (it is my view) overestimating settlements in an order of magnitude of those likely to be expected seeing the surrounding environment (as I have seen to be presented some times) must be forfeited.

Respect the use of springs to model the response of foundations, even if the more common of the informed methods, it fails in some accounts to give significant information about the behaviour. Directly adjudicating springs to mimick on a per-point basis the expected settlement may be a useful approach but many things can be said about, starting by the normally linear assumption to model a longterm and likely significantly non-linear behaviour. Other most common consideration is that depending on your springs, the influence of any ground outside the plan in the settlement is likely to be forfeited. Elastic half-spaces can show better (within the limits of an elastic behaviour) the actual influence of surface (or footing) loads on the ground, but in the end one would need some complete program where the building structure and the solid (layered?) nonlinear soil could be put in interaction. This model once perfected would show how much the existence of an actual foundation of some stiffness is influencing settlement.