In the center of Amsterdam there are historic quay walls that are more than 100 years old. Over the years, the load-bearing capacity of the quay walls decreased, and it is currently an important topic on how to strengthen these quay walls, but it is important as well to know which quay walls should be given preference for rehabilitation.

The engineering department of the municipality of Amsterdam, in collaboration with other parties, has drawn up a document Toetskader Amsterdamse Kademuren (TAK 3.2), which includes various aspects that contribute to the assessment of historic quay walls. In TAK 3.2, attention has of course been paid to the geotechnical aspects, whereby material model parameters have been determined for the Hardening Soil small strain model and the Soft Soil model. This study examines how the current parameter set in TAK 3.2 can be improved by selection of material model parameters and taking the material models into consideration as well in which the anisotropic Sekiguchi-Ohta model and the anisotropic S-Clay1 model are examined additionally.

Material model parameter sets have been compiled (with expected values) for the most influencing Holocene soil layers, such as Geulopvulling, Hollandveen and Oude zeeklei. These parameter sets have been validated and further optimized with the use of Plaxis SoilTest. The available laboratory results come from isotropically consolidated triaxial tests on Geulopvulling and Oude zeeklei, Direct simple shear (DSS) tests on Hollandveen and oedometer tests. Validation of the anisotropic models has therefore not been entirely possible for Geulopvulling and Oude zeeklei, but in the case of Hollandveen the assumption is made on K0-consolidation in the DSS test simulations, where the anisotropic models yield promising results.

Furthermore, according to the approach of TAK 3.2 in the staged construction of historic quay walls, various simulations are conducted for the assessment of historical quay walls with the Hardening Soil small strain model, Soft Soil model and S-Clay1 model, with different parameter sets. From these simulations, some good prospects are noted, such as the improvement in output results with the optimized parameter set compared to the parameter sets gathered from all laboratory data and TAK 3.2. In simulations in which the S-Clay1 model is considered in combination with the Hardening Soil small strain model and Soft Soil model, a significant decrease was noticed as well in the output results compared to simulations with exclusively the Hardening Soil small strain and/or Soft Soil model.

This study shows that the Hardening Soil small strain model as an isotropic model gives better results and fewer/no difficulties in the finite element method calculations compared to the Soft Soil model. It is further confirmed that calculating the initial phase with the K0-procedure with horizontal soil layers and surface is the best method in the case of the assessment of historic quay walls.

By understanding which differences are observed with different material models and parameter sets, this study contributes to sharpening future research to adequately analyze and assess historic quay walls in the center of Amsterdam.

Finite Element Analysis is an application used in many engineering fields, including geotechnical engineering. With increasing complexities of structures and material behavior, classical approach for non-linear finite element analysis has become computationally expensive (Gulikers, 2018). In order to improve the efficiency and accuracy of the model, Machine Learning (ML) is introduced. The ever-increasing accessibility and magnitude of computational power has made machine learning gained popularity since 1990s. Artificial Neural Networks have especially been proven to be a versatile machine learning framework, capable of solving complex tasks such as voice recognition to solving failure analysis of structures. This thesis aims at implementing soil constitutive models such as Linear Elastic and Elasto-plastic models and integrating them into a Finite Element Engine. Two different approaches have been attempted to implement the ANN into the constitutive models. The first approach involves the implementation of a "Generic" Neural network to simulate the natural anisotropic behavior of soils. Datasets have been generated with the help of theoretical soil tests classified as ‘Standard’ and ‘Non-standard’ tests. The standard tests depict an isotropic behavior, where the strains and stresses in different directions are proportionate. Non-standard tests consist of data representing anisotropic behavior where the strain increments in every direction are randomized. The neural network has first been implemented for a linear elastic soil model. Itis observed that when the network is trained and tested with similar kind of data, predictions are accurate. However, when the neural network is trained with a certain type of data (such as standard test), it is unable to predict the response for the other data (non-standard test). When the neural network (trained with standard data) is subsequently implemented in Finite Element analysis by deriving and substituting the material stiffness matrix from the neural network in a constitutive model, the difference between the isotropic linear elastic matrix (for linear elastic model) and the matrix derived from the neural network lead to significant errors, despite the accurate predictions of the neural network. The second approach is a component based approach in which the neural network is modeled to learn the typical behaviors of soil such as linear elasticity and plasticity. The data for training and testing the neural network are generated using standardized laboratory tests (CRS, DSS, Biaxial etc.). The neural network is modeled for linear elastic soil, by applying certain weight constraints. It is observed that the neural network has accurate predictions. Further implementation in constitutive model shows that the neural network is able to mimic the behavior of Linear elastic soil model. The neural network model, is subsequently implemented to predict Mohr-Coulomb behavior, and the predictions are significantly accurate. Further the generalization ability of the network is analyzed to evaluate the practical applicability of the network, followed by a sensitivity analysis to get an understanding about the influencing parameters in the network. The devised neural network, when implemented in a constitutive model displays a behavior similar to the theoretical constitutive model of Mohr-Coulomb soil. Finally, this thesis concludes with the implementation of the neural network constitutive model in Finite Element Analysis where it leads to convergence errors in the model.

The aim of this research is to find the influence of rapid drawdown on dike stability with the Delta21 project as case study. The Delta21 project is a proposed plan for a waterdefence system in front of the Haringvliet, The Netherlands. Its main element is a water retaining lake with an extensive pumping and turbine system. Whenever other storm surge barriers close, water from inland rivers can be pumped to the sea. The water retaining lake can act as an energy storage system as well by using the turbines to generate electricity at moments of high energy prices. The pumping system has the capacity to reduce the water level with 17.5m in 12hours. The dike surrounding the lake has a gradual slope as the height/width ratio below the water line is 1/20. It consists fully of a homogeneous and permeable fine sand. The estimation of material properties were based on limited CPT data, samples from the adjacent Maasvlakte II and empirical relationships. The rapid decrease of the water level in the lake causes rapid draw down conditions in the dike. Hereby the external hydrostatic pressure is reduced due to unloading effect of removing water, while a delay in the dissipation of pore pressures inside the dike is produced. This causes stress relaxation and slope instability. The research was performed using a numerical model verified with physical modeling in the form of centrifuge tests. For the numerical model the finite element model PLAXIS2D was used. A fully coupled flow deformation analysis was chosen, which can solve the full interaction between deformation, consolidation and groundwater flow simultanously in the same phase. It considers a reduced permeability and degree of saturation in the unsaturated zone. The Hardening Soil small strainmodel was selected as constitutive model due to the potential of capturing hysteresis in the stress-strain regime. Hysteresis could occur as a result of multiple cycles of water level fluctuations. To gain better understanding of the capabilities and weaknesses of the numerical model, it was verified with several physical tests from literature. The first test is a centrifuge test with failure due to increasing hydrostatic pressures within the dike. The moment of occurrence and shape of the failure contour were compared with the numerical model. The second test was a centrifuge test where a dike experiences rapid draw down conditions with the purpose to validate the pore pressure and saturation level. Thirdly, a full scale test of a dike with a changing water level was added, because full scale tests offer the closest resemblance to situations in the field. The field test was modeled to double check pore pressures and saturation. The research aimed to further investigate the development of a failure during rapid draw down, a centrifuge test was performed at 100g. Rapid draw down conditions were simulated on a sand dike with slope 1:1.6 to generate failure. By tracking grains with Particle Image Velocimetry (PIV), correctness of estimating displacement by the numerical model was verified. The addition of pore pressure sensors could give further information on pore pressure development. The numerical model proved to accurately predict positive pore pressures and the moment at which macroscopic failures occur due to drawdown. Negative pore pressures, saturation and development of the failure, in terms of size and deformation could not be accurately approximated. With understanding of the capabilities and weaknesses of the numerical model, the dike of the Delta21 project was modeled. The original design proved to be stable with a FoS of 6.20. Increasing slope angles were modeled resulting in a decreasing FoS. The influence of hysteresis was found to be very small after modelling multiple cycles of changing water level. The positive effect on stability the addition of a high permeable filter layer was modeled. Finally the risk of erosion due to outflow was analytically calculated and proved to be closely related to the numerical results.

In 2017 new guidelines concerning macro stability calculations were implemented by the Dutch Ministry of Infrastructure and Environment. These guidelines are formulated in “Wettelijk Beoordelingsinstrumentarium” (WBI). The largest difference with the previous version of the guidelines concerns the material model that is prescribed to determine shear strength parameters. In triaxial tests shear strength parameters are to be determined at ultimate state (25% axial strain), which is assumed to be a good representation of critical state. Critical state is a concept from Critical State Soil Mechanics (CSSM) and assumed to be a good representation of the state reached after large deformations induced by macro instability. Another fundamental assumption in the WBI is the use of the SHANSEP method. This method encompasses a laboratory procedure and a normalisation method. CSSM was originally defined and elaborated under isotropic stress conditions while in engineering practice anisotropic conditions are mostly used.
The goal of this thesis is to investigate the undrained soil behaviour of organic clay in triaxial tests following the SHANSEP procedure and compare the results to the CSSM framework. In order to do so a series of eight K0-consolidated triaxial tests is executed using silty organic Oostvaarders plassen clay, which is assumed to be representative for a typical Dutch soil. A large range of over consolidation ratios is applied (1-20). Both compression and extension tests are executed. The triaxial tests are complemented by two K0-CRS tests and an isotropic compression test to determine relevant soil properties. The results are compared to the CSSM. The qualitative soil behaviour is analysed as well as the actual predicted undrained shear strength Su. Parameter relationships as described by CSSM are tried to be established from the data. The undrained shear strength is predicted by using numerical and analytical formulations of the Modified Cam Clay model, which is the most basic implementation
of CSSM. From the data a clear failure line could be determined in p'-q space. In the triaxial compression tests failure at ultimate state gave very consistent result, a failure line could precisely be determined. In extension failure at peak strength showed the most consistent results. Ultimate state could not be reached under representative stresses because the formation of large shear bands and necking during shearing. A unique p'-q-e relation was much harder to establish. The general trend as described by the CSSM was clearly visible but the uncertainty was rather large. Several factors contributed to the uncertainty. Among which the void ratio determination method that was used to determine the void ratio after the triaxial test, which is prone to errors. The MCC model was not able to model the stress path correctly and p'_0 at failure was not correctly predicted resulting in incorrect Su prediction. The MCC model overestimated Su in compression tests. In extension, however, Su is well predicted by the MCC model. The SHANSEP method turned out to be a very convenient way to normalise the undrained shear strength of the triaxial tests, both in compression and extension. Only at very large OCR Su/σ'_v0 is slightly overestimated.

This thesis tried to quantify the strength reduction of the soft soil layers of the experiment through a back analysis of the failure of the dyke without sheet pile wall. Residual strength hypothesis were formulated based on a literature study and the current design norm of dyke design on soft soil layers in the Netherlands. A Limit Equilibrium Method analysis of the pre and post failure geometry served as a basis to determine the peak and residual strength of the soft soil layers following the residual strength hypotheses. The 3D effect was taken into account in these analyses, and determined to be around 20\%. A Material Point Method model of the experiment without sheet pile wall was then created to test the different residual strength hypotheses. Factors were applied on the strength properties calculated from the LEM model to match the MPM model. A factor of 1.23 was applied to account for 3D-effect and 1.16 to correct for water on passive side being absent from MPM discretization.

The Material Point Method model showed that during failure the behaviour of the clay layers could be expressed with an Undrained SHANSEP formulation. In this formulation the residual strength of clays was found to be independent of the Over-Consolidation Ratio. In a Mohr-Coulomb formulation this results in a complete loss of cohesion. Leaving the OCR out of the strength formulation of clayley layers resulted in horizontal displacement of 4.5 m, which is close to the 6 to 8 m found during the experiment. Further decrease in S-ratio of 30\% resulted in horizontal displacement going up to 7.5m in the MPM model. A reduction of 0 to 30\% of the S-ratio could therefore be concluded to be a range of friction softening. This was concluded to be in accordance with what was found in literature. Laboratory testing and correlations based on index properties effectuated prior to the experiment showed residual friction angle around 30\textdegree. The residual strength backcalculated are much lower, and therefore in contradiction with the laboratory testing results effectuated. The use of cyDSS and LDSS tests was therefore deemed inappropriate for the determination of residual strength.

The Limit Equilibrium Method analysis of the dyke with sheet pile wall was deemed inappropriate for the back analysis of the soft soil layers. The horizontal forces induced by the soil-structure interaction cannot be disregarded. It is recommended to back calculate the peak and residual strength of the peat layer using a Finite Element Method analysis. The displacement measurements of the failing dyke with sheetpile wall in the peat layer showed similarities with strain localisation in a DSS test.

Around 55% of The Netherlands is protected by flood defences like dikes. Flood defences must meet requirements which are based on an allowed flood probability. Safety assessments are performed to check whether flood defences meet these requirements. Uncertainties are explicitly included in safety assessments. Uncertainty in soil behaviour is one of the major contributors to failure probability of dikes. Therefore reduction of this uncertainty in soil behaviour leads to potential savings in reinforcement costs. One of the uncertainties in soil behaviour is the effect of aging on the behaviour of soft clays in compression. Compression behaviour of clay is strongly dependent on the transition from overconsolidated to normal consolidated soil behaviour. This transition phase in plain-strain conditions is indicated by the yield stress. The development of the yield stress due to aging is described by isotach models in terms of creep strains and stiffness parameters. Examples of isotach models are the abc-model and PLAXIS’ Soft Soil Creep model (SSC-model). According to those isotach models, a yield stress induced by aging is similar to a yield stress induced by unloading. This study verifies the development of the yield stress predicted by abc-models based on K0-CRS tests. The test material is remoulded, highly organic, silty clay with a liquid limit of 165% and a plastic limit of 56% (OVP-clay). The determination of the abc-parameters is not straightforward. The value for the creep parameter is easily overestimated due to a slow decrease in excess pore water pressure during a constant stress phase, resulting in stress-dependent strains instead of creep strains. The b-parameter is affected by the gradual transition from overconsolidated to normal consolidated behaviour and by the development of structure during aging. The development of structure is not included in the isotach models and could result in a higher yield stress than predicted by the isotach models. The possible increase of the yield stress due to the development of soil structure is not visible in the test results, but the performed tests indicate softening behaviour after aging phases. The SSC-model simulates no increase in lateral stress during aging. Some studies present data showing that the lateral stresses increase under a constant effective stress in time. Unfortunately, the development of lateral stresses cannot be quantified based on the performed tests due to stick and slip behaviour in the K0-ring during laboratory testing. Based on the test can be concluded that a minimum yield stress in time can be predicted based on isotach models. An important note is that a yield stress is always obtained with respect to a certain isotach which is dependent on the loading rate and the creep parameter. The shift of the yield stress from the imposed isotach to the reference isotach is often significant in laboratory tests, but not taken into account. Effects of development of structure during aging should be further investigated: it influences the behaviour of the soil in compression and affects the yield stress determination.