A raft foundation does not sit on a rigid floor; it floats on a deformable soil that pushes back unevenly. The convenient assumptions we make to size it quickly - a rigid raft, a linear pressure distribution - can be badly wrong in both directions, over-estimating moments in one place and missing them in another. Knowing when those assumptions hold, and when to abandon them, is the heart of soil-structure interaction.
The rigid-raft assumption
The simplest model treats the raft as infinitely stiff. Under that assumption the contact pressure varies linearly across the raft (uniform under concentric load, a trapezoid under eccentric load), and the bending moments follow from that tidy distribution. It is quick, hand-checkable, and entirely adequate for a small, thick, lightly loaded raft on competent ground.
The trouble is that real rafts are not rigid, and real soil does not push back linearly.
Real soil concentrates pressure under stiff, heavily loaded zones and relaxes elsewhere - the exact opposite of the smooth linear distribution the rigid model assumes.
Under a stiff core or a heavily loaded column, the soil reaction piles up; away from it, the pressure drops off. A rigid analysis can therefore put the peak bending moment in the wrong place and at the wrong size.
What coupling the soil actually changes
A coupled model represents the soil as something that deforms - a bed of springs (the Winkler model) or a continuum - so the raft and the ground find their pressure distribution together, by compatibility, rather than having a linear one imposed. The consequences are often counter-intuitive:
- The pressure redistributes toward the stiff, heavily loaded regions, matching how the ground really behaves.
- For a flexible raft, this redistribution frequently reduces the peak design moment compared with the rigid idealisation, because the load shares out instead of cantilevering off a notional rigid plate.
- The settlement profile becomes dished rather than uniform, which itself feeds back into the moments.
This is the important nuance: coupling the soil is not automatically more conservative. Sometimes it relieves moments the rigid model invented; sometimes it reveals concentrations the rigid model smeared away. You cannot know which without doing it.
When the extra effort is worth it
Not every raft needs a coupled analysis. A short decision guide:
- Small, thick, uniformly loaded raft on firm ground - the rigid assumption is fine. Do not over-model it.
- Thick raft on soft soil, or a raft with very uneven stiffness or loading (a tower core on a podium, for instance) - a coupled spring or continuum model earns its keep, and often pays for itself by reducing the design moment.
- Whenever the rigid result looks surprisingly large under a stiff zone - that is a sign the linear distribution is misplacing the peak, and a coupled check is warranted.
The input to any of this is the ground stiffness, which traces back to the site investigation - the same data the SPT bearing-capacity article deals with. The takeaway: the rigid raft is a useful shortcut, but it is a model, not the truth, and for thick rafts on soft or uneven ground the soil deserves a seat at the table.