Soil–structure interaction: when and how should it be modelled?

Rafts, buried structures, hydraulic box culverts: when soil–structure interaction becomes decisive and which methods to use, from springs to finite elements.

In many calculations, the structure is assumed to rest on rigid supports. This assumption is convenient, but it becomes wrong as soon as the deformability of the ground significantly changes the distribution of forces. This is the field of soil–structure interaction: the ground and the structure deform together, and each influences the behaviour of the other.

When is soil–structure interaction decisive?

• Rafts and strip footings: bending moments depend directly on the relative stiffness of the raft and the ground.
• Buried structures (box sections, cut-and-cover tunnels, underground stations): earth pressures and ground reactions govern the design.
• Buried hydraulic structures (box culverts, pipe culverts, chambers): fill loads, uplift and ground deformability interact.
• Pile groups: the distribution of loads between piles depends on the ground and the stiffness of the pile cap.
• Structures sensitive to differential settlement: long buildings, multi-span structures.

Modelling approaches

Elastic supports (Winkler model)

The ground is represented by independent springs characterised by a modulus of subgrade reaction. This simple approach is easy to include in a structural model. Keep in mind, however, that the modulus of subgrade reaction is not an intrinsic property of the soil: it also depends on the size of the foundation and the load level.

Coupled models and finite elements

For more complex cases, the ground is modelled as a continuum, in 2D or 3D, with an appropriate constitutive law. This type of model (for example in Plaxis, or with the geotechnical features of SOFiSTiK) accounts for group effects, the construction sequence and groundwater. In return, it requires better-defined soil parameters and more analysis time.

Choosing soil parameters

Stiffnesses are derived from the available tests, for example Ménard pressuremeter tests, widely used in France, or laboratory tests. As these parameters are uncertain, good practice is to bracket the analysis with an upper and a lower stiffness, then design for the envelope of the forces.

Common pitfalls

• Using a “standard” modulus of subgrade reaction unrelated to the actual foundation geometry.
• Ignoring long-term effects: consolidation of fine soils, creep of concrete.
• Forgetting groundwater and uplift, or considering only one water level.
• Applying only active earth pressure to a stiff buried structure, when at-rest pressure is often more representative.
• Not checking calculated settlements against the settlements the structure can tolerate.

In short

Modelling soil–structure interaction means accepting that a structure and its ground form a single system. The level of modelling should be proportionate to what is at stake. ROKENG carries out soil–structure interaction studies for foundations, retaining structures and buried structures.