The engineering response of fibrous soft soils, such as peats, is significantly influenced by the presence of fibres. Interpretation of laboratory tests performed on such soils is complicated due to the reinforcement provided by the fibres, which results in the material behaving as a composite, with properties different from traditional inorganic soils. The fibre matrix interaction complicates the estimation of geotechnical parameters such as shear strength and frictional properties from standard laboratory tests. Some empirical criteria to account for the contribution of fibres have been proposed, but there is no general consensus in practice. In this work, a numerical approach using finite element method and accounting for the kinematics of fibres is used to gain physical insights into the role of fibres in the mechanical behaviour of soft soils. Numerical replicas of common geotechnical laboratory tests were simulated as boundary value problems by superimposition of a soil matrix with a fibrous network. The contribution of fibres to the mechanical response of the soil sample is quantified for various stress and strain paths relevant to the engineering practice. The influence of two constitutive stress-strain relationships for the fibres upon stress reversal are compared and results suggest that fibres engage before they return to their original configuration, thus implying an influence of their compression history that could be linked to local tensioning of the fibres after compression. Further, laboratory tests on peat are used to validate the numerical approach, modelling the big fibres explicitly in the model and using an advanced model for the matrix, capable of reproducing behaviours such as distrotional hardening expected from the microscopic fibres in the matrix. The comparison highlights the effectiveness of the proposed numerical approach in accurately capturing the mechanical behaviour of natural fibrous peat as observed in laboratory tests. This approach provides valuable physical insights into the composite behaviour of peat. Furthermore, it offers a novel tool for interpreting extensive experimental data on peat samples, enabling the derivation of critical geotechnical parameters essential for the design and assessment of infrastructure founded on this unique soil.

The Netherlands, known for its extensive coastal regions and significant portions of land below sea level, faces escalating flood risks due to climate change and sea level rise. To address these threats, the country relies on the construction of dykes founded on soft soils like peat. Traditionally, peat has been regarded as the critical soil layer for infrastructure stability; however, recent studies suggest that the interface between peat and clay is crucial for assessing failure mechanisms. This research focuses on the mechanical response of peat-clay interface samples to better understand the interaction between these two materials.

The study integrates comprehensive laboratory experiments with advanced numerical modelling to investigate the peat-clay interface. The findings underscore the critical role of the clay layer in the mobilised shear strength, which is essential in dyke stability assessment. The study highlights that accurate modelling of the interface requires considering the volumetric fractions of peat and clay and the reinforcing effects of fibres. This research enhances the understanding of peat-clay interfaces in dyke systems and contributes to improving geo-infrastructure assessment.

Long-term settlement is unavoidable but by good prediction accompanied risks can be reduced. Nowadays linear isotach models belong to the state of the art but result in very small strain rates when small load increments or unloading is applied. For that reason this study has focused on the validation of those predictions using InSAR. Different causes of settlement have been studied and it was found that fluctuations in groundwater do not affect the final settlement. Another influencing factor is the presence of organic matter and its degradation. This reduction of organic matter is caused by many factors and is yet too complicated to quantify and include in predicting models. In spite of the idea that predicted strain rates by linear isotach models are too small, InSAR results show that the measured displacement rates were actually smaller than predicted some years after construction for road segments of the A2 and that a new proposed model could be used to match these displacement rates better. The C+S model by Vergote et al. (2021) is an isotach model which uses non-linear isotachs which would reduce the strain rates faster than linear isotach models and viscoplastic swell is included which also reduces the total strain rate. Results on incremental loading tests show that this model captures the behaviour in unloading stages better than the linear isotach models but for conventional incremental loading tests the two models do not differ much. For field scale embankment scenarios the results show that the C+S model with non-linear isotachs and viscoplastic swell included will predict a longer swell period with more swell, lower total strain rates after the swell period with a faster decay and in the and less residual settlement.

The term peat is commonly used to describe organic soils, composed of fragmented plant remains and fibers. Despite the extensive research on the behaviour of peat during the last 50 years, the vast majority of the attention was devoted to the volumetric behaviour. Only few attempts are reported to model the deviatoric behavior of peat, which is fundamental in many geotechnical applications as dykes founded on peat layers. This Thesis represents an attempt to understand the current knowledge on the deviatoric behaviour of peat with the aid of an hierarchical simple elasto-plastic model.

The constitutive model was conceptually developed aiming for the use of a limited number of parameters with an easy way of determination. This choice aligns with the needs of engineering practice, which demands the use of models with few and comprehensive parameters. The constitutive ingredients of the model have been derived from direct experimental evidences. The model requires 10 parameters, 5 out of them are the classical parameters of the well-known Modified Cam-clay model.\\
\\
The high compressibility of the peat along with its high friction angle intensify the end restraints occurring due to the platens of triaxial apparatus. In order to account for the effects of the boundaries, the constitutive model has been implemented in the Finite Element Programme LAGAMINE and the tests have been simulated as Boundary Value Problems. The model is in qualitative agreement with the experimental data and, in particular, with quantitative agreement until 20\% of strain for the case of consolidated drained triaxial tests. To be able to reproduce properly the case of consolidated undrained triaxial tests which impose an isochoric response, one parameter associated with the flow rule of the model should be changed. This change of a parameter is related to the fact that the fibers in the peat specimen are not considered explicitly in the constitutive equations, although they can alter the deformational response of peat under some kinematic constraints and influence the mechanical response in a predominant way at high strain levels.

Overall, the constitutive model developed in this Thesis gives good quantitative results until 20\% of axial strain. Further research on the influence of fibers on the response of peat under different strain paths and on the consequent matrix-fiber interaction should be done in order to have a model with predictive capabilities until high strain levels.