Infrastructure construction all over the world involves installation of structural elements, such as piles and ground improvement into soils and rock. The installation process itself can be quasi-static (for example jacked piles) or dynamic (vibratory methods, such as stone columns and driven piles). The installation involves very large deformations and causes changes in the pore pressures. Natural soils are complex geomaterials: they exhibit structure and their behaviour is also rate-dependent. Modelling natural soils was a key focus in the two EC-funded Research Training Networks (SCMEP under FP5, see http://scmep.civil.gla.ac.uk/ and AMGISS under FP6, see http://www.ce.strath.ac.uk/amgiss/) which form the basis for the current project.
The influence of installation on key design parameters, such as the mobilised strength at the soil-pile interface and the soil stiffness, is difficult to quantify, and as yet impossible to model. Standard Finite Element Method (FEM) is unable to produce accurate descriptions of large strain problems, due to the large distortions. Although remeshing can be used to deal with the associated numerical problems, the state variables of the soil model need to be mapped from the distorted mesh to the newly defined mesh. This introduces errors, which can be most severe when dealing with highly non-linear materials, such as natural soils. The project involves further development of the so-called Material Point Method (MPM) – a meshfree numerical method based on FEM in parallel with FEM and development of advanced material models. Ultimately, the new modelling tools will be used to model practical demonstration problems, in conjunction with appropriate soil models.