Raft foundation design

A raft spreads the whole building onto the whole of its footprint. Instead of a separate footing under each column, one slab carries the lot — which is what you reach for when the ground is too weak to carry the columns individually, or when the pads have grown so large that they are nearly a raft already.

Structurally it is a flat slab turned upside down. Everything you know about slab design still applies; the loading arrives from underneath instead of on top, and that one change moves the reinforcement, changes what governs the thickness, and makes the commonest site error on a raft a very serious one.

When a raft, and when not

Work the pads first. They are cheaper when they work. The raft becomes the answer when:

• The pads start to touch. The usual trigger is pads covering more than roughly half the footprint. Past that, the concrete you save by leaving gaps costs more in formwork, excavation and edge trimming than it saves.
• The ground is soft or variable. A raft is stiff across the whole building, so it redistributes load away from a soft spot instead of letting one column sink into it. Separate pads cannot do that.
• There is fill, or an old excavation, or made ground. Same reason — the raft bridges what you cannot see.
• The water table is high. A raft can be detailed to resist uplift and to keep water out as one continuous surface. Pads and a separate ground slab cannot.

And when a raft does not work either, that is the end of shallow foundations — the next answer is piles, which need a soil model and a site investigation, not a spreadsheet. Structura stops at the raft for exactly that reason and says so rather than producing a plausible number.

The example

A three-storey building on a regular column grid, on ground with an allowable bearing pressure of 60 kN/m² — soft enough that pads would have covered most of the plot.

Those building loads come from a load takedown — every column's share, added up at foundation level. Get them wrong and nothing below is worth anything.

Step 1 — plan size and bearing pressure

The raft extends past the outer columns so that the edge columns are not sitting on the very corner of it. Modest projections are usual; a large one is doing structural work and needs designing as a cantilever.

Try 400 mm thick. Its own weight matters here, and it is not small:

The raft's own weight is nearly a fifth of what the ground is being asked to carry. On a thick raft it can be a third. That is the first thing that makes a raft different from a pad, where self weight is a 10% allowance nobody thinks about — and it is also why a failing raft cannot always be fixed by making it thicker.

Step 2 — the pressure that actually bends it

This is the step that separates a correct raft design from a wasteful one.

The ground pushes up under the whole raft. But the part of that upward push which balances the raft's own weight acts directly underneath the concrete producing it — every square metre of slab is held up by the soil immediately below it. Nothing spans, nothing bends. Self weight cancels.

So the raft is designed for the net upward pressure from the building load only, and at ultimate:

Include self weight here and you design for about 72 kN/m² instead of 60 — roughly 20% more steel across the entire footprint, for a load that produces no bending at all. It is the single most expensive mistake available on this page, and it never fails a check, so nothing tells you.

Note also that d takes off the full bar diameter rather than half of it. There are two layers of steel at right angles in each face, and this is the average effective depth of the pair — the outer layer is deeper than the inner one, and taking the average is the usual simplification.

Step 3 — bending, as an inverted flat slab

Between the columns the raft spans in both directions under that upward pressure. It is continuous over several bays, so the moment sits between the simply supported nL²/8 and the fully fixed nL²/12. Using nL²/10 is the ordinary hand figure for roughly equal spans:

Both directions clear their minimum, which is worth noticing — on a thinner raft or a shorter span, the 0.13% minimum takes over and the bending calculation stops being the thing that sizes the steel. It applies each way, in each face, so a raft carries four layers of bars whatever the arithmetic says.

And now the part that gets built wrong

Mark it on the drawing in words, not just in section. A raft built with the mats swapped has its main steel in the compression face at every critical section, and there is no way to tell from the surface once the concrete is in.

Step 4 — punching shear, which sets the thickness

Each column is trying to punch a plug through the slab. This is normally what decides a raft's depth, so it is worth checking before you commit to a thickness rather than after.

Take the heaviest column at 720 kN ultimate. Two checks: the stress right at the column face, and the stress on a perimeter 1.5d out from it.

It passes with about 9% to spare, while bending passed with a K of 0.029 against a limit of 0.156 — nowhere near its capacity. That gap is the normal state of affairs on a raft, and it tells you what to do when one fails.

Deducting the upward pressure inside the perimeter is not a refinement, it is part of the check. The soil under that 1.7 m² is pushing up on the very plug that is trying to punch through, so it never reaches the critical perimeter. Here it is 102 of 720 kN — a seventh of the load, and enough to be the difference between a pass and a fail.

If punching fails, the options in order of cost are: a local thickening under the heaviest columns, a column head or drop panel, shear reinforcement, then thickening the whole raft. Thickening everything is the last resort — it adds self weight across 98 m² to solve a problem at one column, and that self weight goes straight back into the bearing pressure check.

Step 5 — beam shear

One more check, on a one-metre strip, at d from a column face:

Beam shear rarely governs a raft. If it does, the raft is thin for its span and punching will already have told you so.

What wind does to a raft

On a building tall enough for wind to matter, the overturning moment tilts the pressure block: one edge presses harder and the other relaxes. Treat the raft's plan as a section and add the swing:

The second check is the interesting one. If the minimum pressure goes negative, the wind is lifting one edge of the raft off the ground — the raft is not in contact any more, and the calculation above, which assumed uniform ground support, no longer describes the structure. The fixes are to lengthen the raft in the wind direction, add mass, or hold it down. See wind loads and stability for where that moment comes from.

What is not calculated here, and it is a lot

Also outside this example:

• Differential settlement across the raft, which is what actually cracks the building above.
• Heave in expansive clay, and ground water pressure under a raft below the water table — uplift can exceed the building weight on a light structure.
• Soil–structure interaction. The uniform pressure assumed above is a simplification; real contact pressure concentrates at the edges on stiff soil and at the centre on soft soil.
• Ground improvement, and whether it is cheaper than the raft it would replace.
• Beam-and-slab and cellular rafts. This is a solid flat raft. Deep ground beams change the analysis completely.

A raft is as much a geotechnical decision as a structural one. Get a soils report, and get the settlement estimate from it, before the arithmetic on this page is worth trusting.

Next

• Which foundation?
• Ground beam design
• Pad footing design
• Strip foundation design
• Soil bearing capacity
• Slab design
• Wind loads and stability
• All worked examples