Pile foundations are used in more challenging site conditions, such as soft soils, to ensure both sufficient capacities and desired performance. As construction moves to more problematic locations and demands increase in magnitude and complexity, pile foundations have been required to accommodate these additional requirements. While the behaviour of pile foundations under vertical and lateral load combinations has been the subject of numerous studies, the response under combined horizontal and torsional loading has received considerably less attention. The emphasis of this research is on using numerical simulation to investigate the response of pile foundations in soft clay under different loading conditions. The procedure is based on nonlinear Finite Element Analysis (FEA) and advanced modelling of geomaterials to accurately describe the realistic response of the soil. In general, natural clay exhibits particularly complex behaviour, which is challenging to address in a single constitutive model while retaining some ease of use. Thus, it is crucial to identify the essential aspects of soil features which may cause a particular problem behaviour and to opt for a suitable constitutive model that captures those aspects.

Here, attention is given to the implementation of an advanced clay constitutive model that accounts for initial stress anisotropy features and its continuous evolution with plastic deformation. The effective stress elastoplastic Simple Anisotropic Clay model, SANICLAY, which has been developed by (Dafalias et al., 2006) as an extension to the Modified Cam Clay model with a minimal number of parameters, is used in this study. The SANICLAY model is implemented into the general-purpose finite element program ABAQUS via the user-defined material subroutine UMAT. The integration of the stress-strain relationship was based on an explicit method with automatic substepping and error control algorithms. The numerical implementation performance was verified and validated against the published model simulation results based on data from experimental tests. The implemented model was further applied to describe clay behaviour in solving bearing capacity boundary value problems in which different geometries and loading complexity levels are offered. In particular, the implemented model performance was first tested with the well-known undrained vertical bearing capacity of shallow strip footings and deep pile foundations problems. The model was then used to explore the effect of anisotropy on the undrained lateral bearing capacity of pile group and piled raft foundations, and the undrained combined lateral-torsional response of pile group foundations. The finite element results demonstrated that accounting for some soil anisotropy results in significantly lower vertical bearing capacity of shallow and deep foundations, compared to the Modified Cam Clay, Tresca, and plasticity-based solutions.

SANICLAY also provided quite lower lateral bearing capacities for pile groups and piled raft problems, compared to those obtained by an isotropic modified Cam Clay, although the two analyses displayed the same trend. The results from both models confirmed that the lateral resistance of a pile group is generally smaller than the sum of the analytical lateral capacities of individual piles. Based on the FEA results, it is concluded that ignoring soil anisotropy may lead to a lower factor of safety in design procedures.

The FE results of a single pile under pure torsion were in satisfactory agreement with analytical predictions for both constitutive behaviours. Whereas the individual piles in the group under pure torsion showed a more complicated response as they translate and rotate as well. This deflection-torsion coupling effect increased the overall torsional resistance of the group, especially for larger pile group sizes. The effect of anisotropy was significant in affecting the lateral resistance rather than the torsional response of the individual piles in the group, thus the overall torsional capacity of the group was lower using SANICLAY compared to Modified Cam Clay. The results further demonstrated the substantial interaction between lateral and torsional capacities, and the failure envelopes indicated that the lateral resistance of the single pile and the pile group foundations are significantly reduced by torsional moments. Anisotropy has led to further reduction of the lateral resistance due to combined lateral-torsional loading of pile group foundations. The corresponding interaction-failure envelopes, with mathematical expressions and graphical representations, were developed for both isotropic and anisotropic clay responses. Additionally, the eccentric lateral loading results illustrated the significant effect of high eccentricity levels in reducing the lateral capacity of pile group foundations.

This research offers a practical tool to handle advanced numerical modelling and finite element analysis and to solve complex geotechnical boundary value problems. Further, the outcomes provide clear evidence of the necessity of accounting for soil anisotropy during pile foundation analysis and design procedures.