Springs V Pinned Restraints

[hannis]

Hi All,

Just a general query in respect of springs used to model piles for a raft slab. For initial results I tend to use pinned restraints to model piles simply to get an idea of what the likely pile loads will be; but once detailed site investigation information and/or a pile design information is available I sometimes recheck using spring coefficients very simply taken as the ultimate pile capacity divided by the calculated settlement. This can produce quite stiff spring coefficients but when compared with reactions resulting from pinned restraints are as expected lower, since the pinned restraints will be infinitely stiff for the translation directions.

My question is how stiff do spring coefficients need to be before they start returning values similar to pinned restraints, if at all?

Regards

Hannis (UK)

 

[oneintheeye]

i would imagine to get the same result as a pinned you would need to have no deflection at a given load. i.e. rigidly restrained in the load direction (back to your pin support). Why would you want to model a spring to act like a pin? Just use pin.
As an aside I use similar method to you to design pile loads. Just make sure your conservative on the spring stiffness, (stiffer). Pin supports give hughly conservative results and no load share between piles.

 

[GBor]

Ah! An analysts dream...

...how stiff do spring coefficients need to be before they start returning values similar to pinned restraints...

Well, I suppose that depends on your tolerance. What is "similar"? Is it off by 5%, 10%...? I can also envision the geometry affecting this result, so you may have to do a sensitivity analysis to determine the impact of varying these parameters.

 

[hannis]

Thanks herewegothen, your correct about the very conservative results than pinned restraints return for piles, hence the reason to check once I've got pile data. The usual case is that the contractor needs to know a pile load before he can get his pile designer to work something out. So it is a bit of a 'Catch 22' no load no pile design, no pile design no load!! What I would like to be able to do is presume a pile capacity with a likely settlement and feed this back in to the model and as you rightly point generate some load sharing. However, I've input ridiculously high spring coeff but still don't come near pinned results. Obviously as GBor points out one cannot have any deflection, since a spring cannot be indefinately stiff.