Tower raft - 100% bearing, no piles 1

[omegaeng]

Hi,

So I'm reviewing a raft design for another firm and I don't see the logic in their design approach. I was hoping that someone could tell me what they thought of the design approach and maybe, how they would have designed it.

Design brief:
30 story tower, 100% bearing raft (no piles)
Core walls of tower land on raft
Raft was sized as 1400 deep, and they've left 1400 overhangs from the core walls to the edge of the raft

Rough sketch:

https://imgur.com/a/8nYCjhx

My initial thoughts:
This is just like a giant pad footing. The purpose of the raft is the spread the loads so that it can bear on the rock without causing settlement (SLS failure, not ULS).

The designers methodology:
1) Get total dead, live and worst case wind loads (reactions) and divide it by the total area of the raft (to get an area load)
2) Ensure bearing capacity of rock is okay
3) Check overturning (as no tension piles, make sure it doesn't flip over from wind)
4) To design reinforcement, flip raft upside down, apply the worst case wall loads as a UDL and use the perpendicular loads as rigid supports (this is confusing, see image above - green pen)

I don't understand this approach. It looks like it was sized so it can work as some sort of strut and tie (1400 deep raft and 1400 overhands = 45 degrees). I tried to just take the most loaded wall and to strutted that load out, and got my tension reinforcement from that - but it was too much reinforcement.

As above, I just want to know:
1) Can you understand the logic to the designers approach?
2) How would you have designed the reinforcement?

 

[Guest262653]

1) Nope. It doesn't makes any sense to me either.

2)
Model 1 - Simplified: determine maximum bearing pressure, create a FEM plate model of the raft only, consider all walls as supports (taking into account wall openings where appropriate), apply maximum bearing pressure and get internal forces;
Model 2 - Global model: Include the raft in the global building analysis model with Winkler springs and determine design forces directly;
Model 3 - Continuum model: As model 2 but substitute Winkler springs by 3d elastic soil elements to get appropriate spring coupling;
Model 4 - Non-linear continuum model: As model 3 but with a soil model that adequately represents the soil non-linear behaviour.

For a 30 story tower, I'd be sure to go at least for model 2. Depending on local geotechnical conditions and stiffness of loading elements, models 3 or 4 could be needed.

 

[omegaeng]

Thanks for your response.

He did a variation of your Model 1, but only did one "strip" (a strip about the heaviest loaded wall). But I still don't get the logic. Can you think of any other ways to simplify it - say, using just the heaviest wall?

Someone else just suggested that it can be simplified by designing it as a massive single column on a massive pad footing. As in, turn the core into a single block and uniformly apply the loads from the global model on it.

Basically like this:

https://i1.wp.com/civilblog.org/wp-content/uploads/2014/10/pad-footing.png?resize=222,290&ssl=1

What do you think? To me, we'd be missing points where we can potentially be underestimating the vertical loads.

They have a global model, I agree that they should just use that, but I want to also try to provide simpler alternatives.

On another note, do you see this as a flexural problem or a strut and tie problem?

 

[WinelandV]

I've reviewed some bulk building foundation designs that took a similar approach - Get total loaded weight of the building and divide by the area of the matt/raft slab. If it were bearing on soil, I can see the argument for why it would work (any localized overstress of the soil will just redistribute to the surrounding soil). One shortcoming was no punching shear check, but was otherwise correct (bulk building tend to be supported by columns, not walls). I'm not sure how it would work on rock, though. Maybe fatdad or oldestguy will see this and weigh in.

As to how I would model it, I'd go with avscorreia's Model 2 option.

 

[omegaeng]

winelandv - for clarify, which was the similar approach you saw? I think you were talking to my second post (designing it as a massive column)?
And then how would you get your reinforcement - strut and tie?

I'm no geotech engineer, but I didn't know that soil can redistribute forces. Concrete can obviously - it cracks and the steel is engaged, and then forces are redistributed. Does something similar happen with soil?

How should the punching shear check have been done? Just the total footprint of the core on the raft?

Also, I know I said it's on rock, but it's actually on soil - 4000kPa capacity. Now you've got me thinking about a bunch of other things! I'm wondering how it would've been different if it was on rock!

Thanks for taking the time to write back.

 

[WinelandV]

Omegaeng,

No, the approach I've seen is similar to your first post (you managed to squeeze in a second post before I could fire off my post). The designer essentially took the DL + LL from the structure (all the structure and bins weigh X, all the bulk material being stored weighs Y, thus a total weight Z = X + Y), and divided the total load by the total area of the matt. Since the resulting pressure was less than the allowable bearing pressure, it was deemed OK. To design the rebar, they set column lines as "supports" and designed against moments generated by the uniform bearing pressure (using the continuous beam moment factors). However, since the structure was supported by columns, the designer neglected to checking punching/2-way shear, which was a fairly big deal because some of the columns had over 700 kips of factored load. It resulted in a thicker matt, so not a huge deal.

As far as rock and soil allowable bearing pressures, with soil, it tends to be the case that settlement is the limiting bearing pressure constraint (that is, the soil could hold more load, but you limit it because you don't want doors to be 3" below grade). Now, with rock, I don't have as much experience. But in the few number of times that I had rock to bear on, settlement of the rock wasn't listed as an issue in the geotech reports. The strength of said rock is what was limiting.

So, as to my comment about using the "sum all the loads and divide by the area" technique being better suited to soil than rock, it's because on soil, the matt tends to be much more rigid than the soil AND the soil can (USUALLY) take more load than we're giving it credit for. So, say you overstress a local area under the matt. A couple things happen - 1) the soil will compress more and 2) this will cause the more rigid matt to engage more soil (locally) - thus causing the local overstress to get smoothed out.

Please note that I am speaking in broad, general strokes. I'd recommend getting ahold of the Geotech report (at a minimum) and then also giving a call to the geotechnical engineer and talking directly to them about this.

Also, I'm not saying that it's correct to model a matt slab this way, just that I've seen it done like this and this explanation is what I came up with as to why it might be a legitimate technique.

Again, I would have modeled your situation using Model 2 in avs' post.