Piled raft design - S&T or flexure

[omegaeng]

In every company I've worked for, in any country, complicated piled-rafts (complex pile, column over and wall over arrangements) have been modelled in FE software. (This too goes for complex/large/irregular pile caps.)

The piled-rafts I've designed have all been very deep, ranging from 2m to 6m. The piles under the towers are always relatively closely space. This creates a small span-to-depth ratio for the raft. Many codes explicitly state (I'm currently designing to AS3600) that when designing concrete elements below a certain span-to-depth ratio, they should be designed with strut and tie, rather than as a flexural element.

Does the entire industry (internationally), including 'certifiers' or 'building control' (depending on the country), allow this seemingly blatant overlook of the codes? Am I misunderstanding something in the codes that actually allows us to design these deep rafts as flexural elements?

I've been debating this with people in my office for the last week or so, and here are the notes I've gathered on the topic. You'll see that many people have opposing opinions. I'm interested to hear other people's thoughts on these:

1) "AS3600 states that the deep elements have to be designing using strut and tie, so that's what we have to do."
2) "Designing something for flexure will always cover you for strut and tie - strut and tie gives a conservative answer."
3) "You should model the entire raft in an FE to understand the global behaviour and to capture the 3D effects. Design the slab based on the moments given by the FE package, but then check if you can make the heaviest loads acting on the slab work using a strut and tie model."
4) "The only difference between designing strut and tie, and flexure is that the FE packages (flexure) didn't historically allow for shear deformations. In most packages these days, you can make your plate elements allow for shear deformations (for example, model "thick" plates, instead of "thin"), so it's okay to use the FE software for your design."

 

[Retrograde]

What is the overall size of the raft? In this situation I do not think that the "span" is the distance between the piles. As long as the "plane sections remain plane" assumption holds true (which it would for a large piled raft) then you can design it as a flexural element.

 

[omegaeng]

The one I'm doing now is for a fairly slender tower - approximately 30m x 30m of raft under the tower. But the last one I did was something like 90m x 120m - this was for two towers plus a common basement car-park between the towers. Both were rafts about 2000mm deep.

Why wouldn't the span be the distance between the piles? I would've thought it was.

Why can we still assume that plane sections remain plane in a large piled-raft? And how can this be justified in the code? I may be wrong, but isn't the fundamental premise of s&t that below a certain span-to-depth, the element behaves differently (creating "B" and "D" regions - where the plane sections remain plane assumption doesn't apply in the "D" regions)?

Thanks for the response! I'm hoping to get a meaningful discussion going on this!

 

[Retrograde]

If you cut a long section through the raft and draw the bending moment diagram I think you will find that you do not get shear or moment reversal in between each adjacent pile.

Think of it like this: flip the raft over and apply the pile forces as loads (and the columns/walls are the supports), what would you take as the span?

 

[omegaeng]

You're right - in the permanent condition (full building weight on piled-raft), you'll get some sort of "dishing" effect of the raft (deformed shape), so you won't get any moment or shear reversal. But this is only true if you've modelled your piles as springs and not supports.

You might get some reversals in the temporary condition (raft only, no building weight), if say, the raft is also withstanding hydrostatic uplift forces - but I think that's besides your point, right?

If I flipped the raft over, I'd be just be inclined to take my spans as the distances between my supports (which are now my columns and walls). But it sounds like you'd take the entire length or width of the raft as the span, because (correct me if I'm wrong) I think you're saying that a "span" is where we get a reversal of deflected shape (or moments and shears)?

 

[Guest262653]

I agree that the raft should be looked at as a discrete springs supported slab loaded by columns and cores or as reversed slab loaded by point loads at the pile location. In this way, it can be designed for flexure and shear. That is also the practice in my area.

Moreover, and that for me is the deciding factor, it really isn't feasible to perform an S&T design for every load combination for a large complex piled raft, at least with a global model. You need a different model for every major load combination and the 3D analysis of the nodes, IMHO, isn't still properly solved. It would be a nightmare.

However, I do agree that highly stressed locations adjacent to highly loaded core walls and columns should be checked individually with simplified local S&T models, using the maximum forces on the piles or vertical elements. So, option number 3 in your original post would seem to me the best option.

 

[omegaeng]

Good point about all those load cases!

I'm still trying to establish a technical and full-proof reason why we can design deep rafts using flexure, in spite of the code clearly stating that we should be using a strut and tie approach.

Regarding checking the highly stressed walls for s&t - I don't fully agree with this because I feel that it will be unlikely that you will have enough localised tensile reinforcement for the s&t model. Then it will become an issue of deciding where to draw the line - do you keep adding reinforcement for your ties? How many walls/combinations do you check?

 

[Guest262653]

I wasn't thinking much about the tensile reinforcement, which I agree that can be adequately designed with a plate model, but more on strut and node localised resistance next to wall corners or columns, especially when these elements are a bit far from the piles, the concrete classes are different or the columns have a high rebar percentage. For each of these highly stressed locations (or the worst...) I'd consider having an S&T model and plug in the maximum vertical force on the vertical element or the pile.

Regarding the code, I don't really know how much leeway AS3600 provides you as I'm familiar essentially with the Eurocodes. Nevertheless, if it is that rigid probably you'll already have colleagues down there with a well established and practical method for S&T analysis for these large rafts. I'd love to get a look at that!

 

[Retrograde]

But this is only true if you've modelled your piles as springs and not supports.

I would have thought for a piled raft you need to be modelling your piles as springs, as well as modelling the stiffness of the ground.

 

[JoshPlumSE]

Strut and Tie method is probably best for situations like you describe. However, my impression is that designing to flexural minimum as well as temperature / shrinkage minimums is probably enough to ensure these mats perform well.

Strut and tie is a bit of a pain, so I would probably design it using an FEM plate element model. Then when I'm essentially done, I'd do a few strut & tie calculations in the most heavily stressed areas just to verify that this leads to similar or conservative results.