This thesis aims to find the best way to construct a surrogate model for the inner slope stability of a dike and combine this surrogate model with other (machine learning) models to generate conceptual flood defenses, making it possible to optimize the dike design using an interdisciplinary MCA.

The stability of rock engineering projects is tightly related to the mechanical behaviour of rock discontinuities. Although the mechanical behaviour of a discontinuity is often associated with shearing (sliding), experimental data have shown that it can be accompanied by more complex phenomena such as dilation and post-peak strength reduction. Numerous researchers have contributed to the understanding and modelling of discontinuities behaviour. However, many of the constitutive models available in the literature are questionable when applied in practice either because they are highly empirical with parameters that are hardly determined or because they oversimplify the examined behaviour. Moreover, the models are often proposed for a certain range of stresses and specific stress paths, which makes numerical implementation difficult for large-scale engineering applications. This dissertation aims to evaluate the capabilities and the limitations of different constitutive models for rock discontinuity in the context of both numerical implementation and simulation of the mechanical behaviour of rock discontinuity. The first part of the research project investigates the main features of the existing models in the literature. From the investigated models, the two models highly adopted in research and engineering practice, namely the Coulomb’s model and Barton-Bandis’s model, are extensively investigated. Some enhancements and modifications are made to these two models to improve their modelling capabilities and ensure the numerical stability of numerical implementation. Regarding the Coulomb model, the adopted modifications include the reformulation of the model within the framework of strain softening providing a rigorous implemented version that describes the post-peak behaviour of a discontinuity adopting a linear reduction of the strength. Additionally, the employed modifications to the Barton-Bandis model provide a robust version of the model applying reformulations to the original yield surface that increase its validity in the whole range of the τ-σn space. Furthermore, a simplified definition of the post-peak behaviour, which aligns with the original formulation of the model but at the same time allows for a straightforward numerical implementation, is proposed. To validate the implementation of these models in PLAXIS (implementation done by the PLAXIS research team), the models are implemented in Python scripts for Constant Normal Load (CNL) shear test configuration. Concretely the implemented models are calibrated with experimental data to simulate CNL tests using a PLAXIS 2D Finite Element (FE) model and the obtained results are compared with both Python theoretical simulation and experimental results to verify the FE implementation. The results of these simulations validate the numerical implementation in PLAXIS and prove the applicability of the enhanced models to reproduce with adequate accuracy the mechanical behaviour of a rock discontinuity on a lab scale. Finally, the implemented constitutive laws are employed to perform a FE analysis of a large-scale application of a deep underground excavation in a discontinuous rock layer using PLAXIS 2D. To facilitate the creation of this complex geometry, an automatic discontinuity network generator is developed and improved using PLAXIS Python scripting API. The implemented discontinuity laws are then applied to the randomly generated discontinuity sets to simulate the behaviour of the rock mass. Stress and failure analyses are performed for the most critical discontinuities and wedges formed around the excavation to validate the numerical implementation and analyze the applicability of the constitutive models. The analysis of the boundary value problem confirms both the reliability of the numerical implementation and the applicability of the enhanced constitutive laws to simulate the analyzed large-scale problem.

Tailings dam failures are one of the most destructive phenomena in terms of the number of victims and the environmental impact generated. Over the years, different causes have been identified, with flow liquefaction being a prominent factor to consider when assessing the stability of tailings deposits. Due to the complex nature of these events, numerical models are considered an important alternative for the analysis and design phases of tailings dams. These models require the application of appropriate advanced soil constitutive models to reproduce the relevant features of the soil behaviour and, thus, obtain results that are a valid representation of the observations in reality. In this work, the capabilities and limitations of the Clay And Sand Model (CASM), originally proposed by Yu (1995, 1998) and NorSand model (Jefferies, 1993) are assessed using the finite element software PLAXIS in the analysis of flow liquefaction. The criteria considered in the assessment consist of the accuracy of the models to predict the response of the soil under different monotonic loading conditions, the efficiency of the models for their application and the consistency of results obtained from boundary value problems compared with reality.

Both CASM and NorSand are models developed within the critical state soil mechanics framework and use the concept of the state parameter (Been and Jefferies, 1985). However, they present differences in their formulations and assumptions. The performance of the model is explored in this study at three different levels: the material point level, simple boundary value problem level and complex boundary value problem. Different laboratory tests are simulated for the material point level, analysing a single stress point. For the boundary value problems, the construction of an embankment and the well-known Tar Island Dyke flow liquefaction event are simulated. Before applying the models, the implementation of CASM into PLAXIS is widely verified, and additional verification exercises for the NorSand implementation are also presented. Parametric analyses are performed to have a better insight into the effect of the parameters in the response of the models. Moreover, CASM and NorSand calibrated for the same tailings materials are used for the simulations.

Given that most of the model parameters can be estimated through very well established procedures using experimental data, in principle, the calibration of both models is relatively straightforward if the required experimental data is available and when the data is limited, correlations can be used to estimate the parameters of CASM assuming normally consolidated conditions, which is the state at which tailings are often found. The results of the thesis, at the material point level, show that CASM presents limitations to quantitatively predicting the response of dense materials. This is associated with the formulation and assumptions of the model. In the analysis of the boundary value problems, lower strengths and higher stiffness in terms of stress-strain response are obtained with CASM compared with NorSand. Despite these differences, the results of this thesis demonstrate that both CASM and NorSand can successfully reproduce flow liquefaction at the different study levels.

Numerical modelling in Geo-Engineering is used to solve complex problems by simulating, analysing, or predicting soil behaviour under certain loading and boundary conditions. The soil behaviour is simulated by constitutive models that describe the relationship between stresses and strains through a mathematical formulation. Model parameters are used to calibrate model behaviour to physical soil behaviour measured during in-situ testing (e.g. CPT) or laboratory testing (e.g. triaxial testing). The selection of model parameters is challenging as it needs to cope with aspects as, constitutive model limitations, laboratory test limitations, sample disturbance, soil heterogeneity and many other. This study shows how these model parameters can be determined, optimised and selected by using over 3000 triaxial test results performed on dutch soils (stored in text files) and machine learning tools.

In the low-lying parts of The Netherlands challenging ground conditions are often encountered. Thick layers of soft material such as peat and clay are present and pile foundations into deeper sand layers are required when objects are constructed that have to transfer significant loads to the subsurface without inducing settlement (differences). This also counts for pile foundations of wind turbines which, apart from compression and tension loads also are subjected to horizontal loading. The horizontal translational stiffness is an important and depending on the wind turbine manufacturer governing aspect in the required dimensions of the pile foundation (number and type of piles). In order to improve this horizontal translational stiffness, the foundation designer usually opts for the use of battered (raked) piles. The basic principle is that the axial stiffness of the pile will, in this way, contribute to the horizontal stiffness of the foundation as a whole. The verification of wind turbine requirements usually involve the use of generally accepted analytical 2D methods, simplifying the more complex 3D effect, to model the interaction between soil-pile and the group effect for closely spaced piles. Although the analytical model generally provides good results, investments can reduce significantly when only a small amount of piles can be saved upon. Also the effort in the field is reduced when using vertical piles...

As a consequence of the difficulty to obtain complete direct information about the subsurface, and the importance of heterogeneity of soil properties in geotechnical design, a desire exists in the geotechnical community for a smarter use of the available data in common practice. The objective of this thesis is to provide a potential solution for this, by implementing the Random Finite Element Method (RFEM) in PLAXIS, a well-known geotechnical calculation software package, allowing users to stochastically model engineering problems while accounting for the spatial variability of soil properties.

As a result of this project, practitioners can use the upside of stochastic analysis using RFEM to derive new insights about the behavior of their system, such as potential soil-related asymmetries, all while obtaining a better picture of the risks associated with a certain design. As with any technological advancement, responsible use of the framework is required to ensure the upside is attained and computational concerns do not hinder the use of the model, feature which still requires further research and improvement.

As the construction of sub-aerial and submarine geotechnical structures increase in amount, rate and size, so do their associated risks. Often, with use constitutive models, finite element analyses (FEAs) are performed in order to identify and mitigate these risks. NorSand, which is a consti- tutive model based on critical state soil mechanics for particulate materials (e.g., sand), is one of the first models to integrate the state parameter ψ into its constitutive framework to model dense and loose sands material with the same parameter set. Importantly, it is able to identify the liquefaction potential as it can simulate softening behaviour due to pore pressure increase of loose soils in undrained conditions. Since NorSand has recently been implement into PLAXIS, a geotechnical analysis software capable of performing FEAs, it must be verified, validated and applied with the software, which is done in this report. First, stress-path and parametric analyses were conducted at single stress points. The stress- path analyses show how the state variables evolve for different triaxial conditions. Systematically changing the input parameters to extremes found in literature helped determine their influence on the evolution of stresses and strains. The resulting figures can be used to help future calibrations to experimental lab test data. The PLAXIS implemented NorSand (PLAXIS NorSand) was verified by comparing it with an implementation written in Visual Basic for Applications (VBA NorSand) by the authors of the model Jefferies and Been. Verification in this context means determining if PLAXIS NorSand is able to produce outputs as intended by the authors. Various testing conditions, both triaxial and direct simple shear, showed overlap and agreement between the outputs of both implementations, verifying PLAXIS NorSand. Then, the model was compared to, albeit not in a traditional sense, an ’analytical solution’, which is the relationship between the mobilized friction ratio Mi and state parameter ψ in its simplest form. The mobilized friction ratio and stress ratio at peak strength of PLAXIS NorSand and the ’analytical solution’ were compared. The values between both showed less than 3% difference, further verifying PLAXIS NorSand. The constitutive model was then validated - i.e., established that PLAXIS NorSand is able to approximate soil behaviour as intended. First, by using the soil parameter set that had been derived from lab tests of Erksak sand as a baseline, the input parameters were varied until PLAXIS NorSand was calibrated to individual triaxial tests as best as possible. Then, triaxial tests of Erksak, Nerlerk and Ticino sand were approximated with PLAXIS NorSand without changing the soil parameters determined from lab test data. PLAXIS NorSand is able to follow lab test data decently well with one parameter set. And if one decides to take the time and calibrate individual lab tests, and deviate from soil parameters determined from a set of lab tests, they can be matched even better. Additionally, it showed a consistent need for activation of the softening flag (S = 1) in order to appropriately model loose soils in undrained conditions. Furthermore, NorSand exhibits indefinite hardening in dense soils during undrained loading, which can be avoided by employing a ’cavitation cut-off’. The last part of this report tested PLAXIS NorSand by applying it in FEAs of a simplified submerged landslide, which was subjected to 20 centimeters of displacement at the crest through a rigid slab in undrained conditions. First, the difference in slope behaviour due to change in soil density within NorSand was determined: dense soil resulted in the slope to be able to bear the full 20 centimeter displacement, whereas increasing the void ratio (i.e., increasing the positive value for the state parameter) gave the effect of even quicker slope collapse and a lower bearing capacity. In other words, when using NorSand, the looser soil the further the failure surface moves up and the quicker the structure fails to maintain equilibrium. Lastly, NorSand was compared to Modified Cam-Clay and Mohr-Coulomb to highlight the differences in their ability to model static liquefaction, while being triggered by unrealistic loading conditions. Even though none of the FEAs showed actual liquefaction, since it is accompanied with the fluidization and loss of structure, they still gave in indication of the liquefaction potential. NorSand, contrary to the other constitutive models, showed the expected high sensitivity to forced displacement resulting in clear shear bands resembling Prandtl-type failure mechanism and early onset soil body collapse.

The Automated Parameter Determination project aims to provide advanced geotechnical models parameters from in-situ tests in a transparent and flexible process. The framework was initially developed for coarse-grained soils, and it was needed to be expanded to all types of common soils. This report presents the expansion of the APD database to deal with fine-grained soils, claylike soils specifically. The APD variability assessment framework is customised to deal with log-transformed correlations’ uncertainty. Moreover, it is found a large dependency on Atterberg limits for claylike soils parametrisation. Only one set of correlations from CPT database is found in literature to obtain these limits, for which a new set is developed based on critical state soil mechanics and the assumption that the CPT friction sleeve is similar to the remoulded undrained shear strength. The proposed correlation is validated with a published database, showing acceptable results with similar variability when compared to the existing equation. A first validation of the complete model parameters for fine-grained soils using Plaxis Hardening Soil with small strain stiffness model is achieved. The Plaxis Soil Test facility is used and the results are compared with triaxial and oedometer tests, showing good results in compressibility characterisation and low estimation of friction angle, possibly attributed to organic and silt content. The friction angle characterisation is discussed, and it is concluded that a better estimation for fine-grained soils is needed to be studied. Further studies on soils in between claylike and sandlike behaviour is needed, as well for organics.

The current plan of bottom to top bottom sublevel stoping for the Kearney gold vein at the Cavanacaw mine, Omagh, Northern Ireland may not be the most effective in terms of stability. The Kearney ore body is a narrow vein gold deposit, which has been previously exploited through an open pit and is currently being developed as an underground operation using sublevel stoping (modified Avoca mining method). Stability within the mine is one of the key factors to be considered when it comes to hard rock mining. It should be considered equally as important from a safety and economic point of view. The extraction sequence plays an important role when considering the stability of a designed mine. This thesis aims to establish if the current planned sequence of extraction of bottom to top sublevel stoping is the most effective in terms of overall rock stability, or whether an alternative plan would be better? In the context of this thesis, the modified Avoca mining method is a form of sublevel stoping where material is extracted (stoped) between two drives (blind tunnels) and then backfilled.
The project addressed, a conceptual study, field testing and laboratory testing in order to yield the information required to build several numerical models. The numerical modelling was carried out on several different stoping orders which met the constraints set out by Galantas, using the Hoek-Brown model within Plaxis2D. The analysis was conducted on the total displacements, phase displacements, predicted failure points and safety factors. The analysis of the different models showed that an alter- native stoping method of middle to top bottom to middle sublevel stoping peformed better in terms of stability. This improvement in stability was shown by an increase in the minimum safety factor from 3.20 to 3.50, over the current plan. There is further evidence in the reduction of the total number of predicted failure point by 25%.

Currently, the settlements of high-rise buildings are calculated based on two methods. The single pile settlements (s1), are generated due to mobilised shaft friction and pile tip resistance. These settlements are determined with analytical methods (i.e. NEN 9997-1, 2017). Settlements related to pile group behaviour in soil layers below pile tip level (s2) are calculated with the equivalent raft approach i.e. Tomlinson (1977). However, an essential aspect for differential settlement calculations is the redistribution of loads by soil-structure interaction mechanisms, which are not currently incorporated in the calculation approach. A missing factor for a soil-structure interaction calculation is the lack of soil- displacement installation effects in the numerical methods for single pile behaviour. Therefore, this research is focussed on the implementation of a soil displacement single pile behaviour in PLAXIS 3D and thereby towards integration of the structural and geotechnical design in one calculation method for appropriate soil subgrade reactions and load redistribution. In this research numerical methods are used in combination with the embedded beam approach applied in PLAXIS 3D. To approximate the NEN9997 load-displacement behaviour of soil displacement single piles, the behaviour of the embedded beam in combination with the multi-linear shaft friction is fitted in PLAXIS 3D. The fit is conducted by increasing the stiffness of a soil volume below pile tip level. Which does account for the residual stresses after pile installation. The solution of this fit consists of two parts: improvements on behalf of the new embedded beam formulation after Smulders (2018) and a soil stiffness fit below pile tip level. Regarding the new embedded beam formulation, improvements have been determined on mesh-dependency and a stiffer load-displacement response. The fit method performs well in combination with the embedded beams while approaching the load-displacement response of soil-displacement piles according to NEN9997. However, the ultimate bearing capacity of the embedded pile should still be calculated with D-Foundations (analytical method based on NEN9997). In the pile group calculation, the behaviour of single piles is extrapolated to all piles in the PLAXIS 3D model. A plate element is modelled on top of the pile heads to achieve a redistribution of loads due to the overlying superstructure. In this way, the single pile behaviour and bending stiffness of the superstructure is implemented in the PLAXIS 3D model, and the soil-structure interaction calculation is possible. However, group effects due to the soil displacement installation method are only partly covered in this approach. The reduction or increase in skin resistance in pile groups is not covered since, the predefined maximum skin friction capacity is manually inserted as an input parameter and it is not an outcome of the PLAXIS 3D calculation. Afterwards, the proposed framework is applied to a practical case study including measurement data of a high-rise project in Amsterdam. The proposed framework turned out to capture soil-structure interaction properly by taking into account a part of the superstructure bending stiffness and single pile behaviour. Compared to current calculation approaches, complex soil behaviour is captured in redistribution of the loads. However, the calculated settlements are still deviating from the measurement data. A probable explanation for the discrepancy is the shift in applied load over time in PLAXIS 3D relative to the real construction process. In conclusion, an improvement in terms of modelling soil-structure interaction has been achieved by incorporating single pile behaviour and superstructure bending stiffness in PLAXIS 3D. Therefore, the possibilities of linking the structural design to the geotechnical design in an early stage of the design process will be feasible. Combining the designs in an early stage, the structural dimensions of columns and beams are calculated with appropriate soil subgrade reactions and load redistributions by the soil. However, it is recommended to apply more investigation on the behaviour of soil- displacement pile (groups) in combination with embedded beams because, group effects are still not completely incorporated.

In situ tests, such as the cone penetration test (CPT), are a popular tool for geotechnical engineers because they are an efficient and economical method for routine site characterisation, soil profiling and estimation of constitutive properties of soil. The main problem with interpreting in situ test results is the large amount of empiricism that engineers have to rely on. Furthermore, the rise in complexity of constitutive soil models have made soil interpretation from experimental data ever more challenging. Constitutive models in numerical analyses are used to simulate the stress-strain response of soils and each model is governed by a set of parameters that quantifies the mechanical behaviour of soil. A consequence for the increased complexity of a constitutive model is the increased number of parameters to be defined from a larger number of experimental tests. This is why parameter determination is still one of the most challenging tasks faced by geotechnical engineers. The challenges in soil interpretation from experimental data have led to an increasing demand for a more efficient parameter determination system for performing more reliable numerical simulations in geotechnical engineering. This study explores the applicability of graphs to develop a generic system for the determination of constitutive model parameters from in situ test results. Graphs are mathematical structures used to model pairwise relations between objects in a network and benefit from their ability to visualise complex problems. In the same manner, the use of graphs in a parameter determination system could generate valid relations, or paths, between parameters in a network. This should give the user of the system, i.e., the geotechnical engineer, both insight in and control over the system. The proposed strategy aims to increase the confidence of derived parameters from in situ test results. It should eventually provide guidance for the user in selecting the right constitutive model and corresponding parameters for the considered application, in order to use the full potential of numerical analysis. This study presents a proof of concept for an automated system to determine constitutive model parameters from in situ test results. Key aspects of the system are: transparency and adaptability. The system is kept transparent, since users are able to verify how available information (i.e., expertise by the engineer) is used by the system to arrive at a solution. The system is kept adaptable, since users can add their expertise into the system without having to make modifications to the system and since developers can easily expand the system in the future. This study illustrates how a system can automatically generate paths between parameters in a network (i.e., a graph),using the external database (e.g., a spreadsheet) as input by the geotechnical engineer. The focus of this system is on determining engineering parameters based on CPT data for coarse-grained soils. However, the universality of the system allows the system to be extended to a wider range of soils and in situ tests. Separate ongoing research efforts are devoted to further validating and tweaking of the system.

The determination of geotechnical parameters from in-situ tests heavily relies on the use of empirical correlations relating field measurements to soil properties. A large number of these correlations can be found across the literature, each of them yielding a different outcome for the parameter value. This results in a high variability of the potential results, which is not necessarily properly accounted for in the current geotechnical engineering practice. A new approach using graphs has recently been proposed by van Berkom [1] to automatically generate multiple parameter values based on a given set of correlations and equations. The present thesis assesses the quality of the estimates obtained from this new graph system and investigates how multiples outcomes can be combined into a unique result. The quality of the parameter’s estimates is appraised both in terms of accuracy or uncertainty, and in terms of validity. The combination of results from multiple methods showed that it is possible to either build confidence or distrust in the combined outcome depending of the consistency of the contributing results. As a result, this framework enabled a quantitative description of the inter-correlations variability. The conceptual framework has been then applied to geotechnical parameters and CPT-based correlations for coarse-grained soils. The range of applicability of the correlations (e.g. soil type, state of consolidation) is a critical factor that actively influences the final outcome of the system. This automatic system demonstrated its ability to produce parameter values relatively close to
the values obtained with the current practices. Further validation is required to assess the overall performance of the system on a broader scale.

Recent development have caused that the policy regarding the installation of onshore wind turbines on embankments has changed to 'Yes, provided that ...' in the Netherlands. (Simplified) approaches are available within professional practice to analyse the risk of liquefaction, but are considered to be conservative. In this thesis, a methodology is formulated to model the key aspects of liquefaction of a soil layer underneath a dike system subjected to cyclic loading by an onshore wind turbine. The key aspects are defined as: cyclic loading, soil-structure interaction and consolidation behaviour. A modification to the model of Seed and Rahman (1978) is implemented in this thesis.
The applicable modelling conditions (quasi-static or dynamic) are investigated for the current situation. The analysis is performed using analytic and numerical models (Plaxis 2D). Both models conclude that the loads induced by the wind turbine can be modelled as quasi-static as the loads are sufficiently slow to neglect inertia effects. The influence of the cyclic loads by the wind turbine on the liquefiable soil layer is determined by the Cyclic Stress Ratio (CSR). The influence due to the soil-structure interaction is determined using a Finite Element Analysis (Plaxis 2D). The Hardening Soil small strain model gives a more accurate prediction of the CSR in the soil, as the soil response is stiffer during un/re-loading. The presence of the embankment results in various modes of shearing in the soil, e.g. triaxial compression, triaxial extension and direct simple shear. The CSR does not include the effect of static shear stresses and various modes of shearing in the soil. This effect should rather be accounted by the Cyclic Resistance Ratio (CRR). The consolidation behaviour modelled using the model of Seed and Rahman (1978) which applies the consolidation equation with generation term. The model is able to model partial consolidation, which is defined as a state where excess pore pressures can be generated and dissipate simultaneously. The generation term is based on an empirical relationship of the development of pore pressures. The model is implemented using a Finite Difference Method in a cylindrical coordinate system to represent the dissipation behaviour of a granular soil. The method allows to model the consolidation characteristics in radial and vertical direction, a layered soil, the load intensity and the loading frequency. As a result, the maximum pore pressure ratio in the soil reduces. A limitation of the model is its uncoupled nature. Therefore, no strains are determined. The constitutive behaviour inherent to the liquefaction phenomena, such as plastic deformations, is lacking. Oostpolderdijk is used as a case study to compare the modified method of Seed and Rahman (1978) to the reference engineering method and the method of Boulanger and Idriss (2014). The comparison with Boulanger and Idriss (2014) suggests that the modified model gives reasonable results. The application of the modified method results in a more favourable result compared to the reference engineering method. However, a sensitivity analysis shows that liquefaction can’t be ruled out because the sensitivity to permeability is high.

Displacement control is of utmost importance in deep excavation design and is usually based on numerical modelling, e.g. Finite Element Method (FEM). Numerical methods tend to be more conservative when analysing soil behaviour during deep excavation, whereas for practical and economic reasons this is not favoured. The inverse analysis allows for the identification of the soil parameter set that can provide the measurements observed in the monitoring when it is applied in the model. When performed in a probabilistic concept, it reduces parameter uncertainty and enables the stochastic prediction of future soil behaviour. In this thesis, capabilities and limitations of difference advanced constitutive models are investigated. The Generalized Hardening Soil Small strain model (GHS) presents a positive aspect in modelling soil behaviour during deep excavation with its various stress/strain dependency settings. Because of the uncertainties originating from the size of the domain and limitations of site investigation, the soil parameters can only be shown as probability distributions. In order to make that distribution more accurate, comparative selection of several inverse analysis optimization algorithms is performed. Thereafter, choice of the relevant parameters is done based on the conducted sensitivity analysis and engineering judgement. Having the most competitive optimization approach selected, remote scripting with Python is used to utilise Finite Element (FE) modelling in the 2D Plaxis software. The input parameters are iteratively updated with response observation (diaphragm wall deflections) using the Ensemble Kalman filter optimisation algorithm based on a chosen excavation stage. The re-calibrated parameters are checked with the data, which was used to create synthetic measurements made using the same FE, to perform reliability assessment of the developed Python-based algorithm and investigate its capabilities and limitations. The further development of the presented optimisation method is expected to increase certainty in setting alarm thresholds in the applications of the Observational Method.

The earthquake-induced liquefaction is a high-risk phenomenon for dredging industries, which need to set strict requirements in order to avoid potential disastrous effects for the project. Different types of liquefaction exist which can be triggered over a wide range of soil types and for different loading conditions. The liquefaction triggering due to an earthquake event is dependent on the soil behaviour under undrained cyclic loading. The assessment of the liquefaction hazard during an earthquake is mainly based so far on empirical procedures. The most common used in practise is the NCEER method (Youd & Idriss, 2001) which is established according to empirical evaluation of field observations and in-situ testing. However, the NCEER method can be inaccurate for the design primarily due to its empirical nature as it is capturing different soil types and loading conditions. For that purpose, advanced constitutive models can provide more precise assessments as they can be calibrated for specific site conditions. Such a model is the PM4Sand, which is very attractive for practical applications because there are only a few model parameters to be determined in the calibration process.The first part of the current thesis project includes the validation of the PM4Sand model for both earthquake-induced and static liquefaction according to undrained Cyclic Direct Simple Shear (CDSS) tests and undrained Direct Simple Shear (DSS) respectively, performed on Ottawa F-65 Sand. The influence of the model parameters is examined throughout a parametric assessment analysis. It was observed, that the model approximates well the general features of both cyclic and static loading. Regarding cyclic loading, it produced similar responses in terms of excess pore pressures generation and stress paths even though it slightly overpredicts the cyclic resistance for small number of loading cycles and underpredicts the cyclic resistance for large number of loading cycles. Regarding static liquefaction, even if the model had initially overestimated the response, it was able to simulate successfully the static liquefaction behaviour after a recalibration process was established.The next part of the project includes the performance of the PM4Sand model for the prediction of earthquake-induced liquefaction in hydraulic fills, which are analysed for several different seismic motions. The fill is placed over a different range of relative densities and it is modelled in Plaxis software as a 1-D soil column. The fill layers that are prone to liquefy, are modelled with the PM4Sand model while the layers that are not susceptible to liquefaction are modelled with Hardening Soil Small (HSS) model. The PM4Sand layer is calibrated according to factors that are accounting for the in-situ state of the fill and the magnitude of the earthquake motions. The dynamic analyses are performed with and without consolidation and the lateral boundaries used are tied degrees of freedom. The results in terms of excess pore pressures generation are examined throughout the whole earthquake motion. Moreover, the onset of liquefaction in the hydraulic fill is captured when the excess pore pressure ratio has reached a value of around 1.0 (ru≈1). It is shown, that the PM4Sand model is indeed applicable for the prediction of earthquake-induced and static liquefaction in hydraulic fills. The effect of the in-situ state of the fill, in particular the relative density, has a critical role on the liquefaction susceptibility, which is a lot representative to what has been observed in reality. According to PM4Sand model, the loosely-packed fills DR=30% and DR=40%) are indeed more susceptible to liquefaction compared to the densely-packed fills (DR=50% and DR=60%) which showed less or even no liquefaction potential due to the earthquake events. On the other hand, the largest drawback of the NCEER method it its empirical nature, as for the current project it is proved to be conservative for the design. More specifically, it predicted liquefaction for almost all the hydraulic fills (DR=30% to DR=60%) analyzed for all different earthquake motions. Regarding the dynamic analyses with consolidation, the results related to the earthquake-induced liquefaction of the fills are more representative to realistic conditions as there is a better distribution of excess pore pressures along the soil column with respect to the dynamic analyses without consolidation. For the latter type of analysis, in the loosely-packed fills (DR=30% and DR=40%) there is a better diffusion of excess pore pressures more for the signals of low dominant frequencies regardless the peak ground acceleration values of the input signal. In the densely-packed fills (DR=50% and DR=60%) the same phenomenon takes pace more for the signals of high dominant frequencies. However, a localization of liquefied zones is observed in distinct parts along the fill layer for the rest of the signals.

Since the exploitation of wind as a renewable energy resource, jack-up vessels equipped with more than three independent legs are increasingly employed to transport and install the components of offshore wind turbines. By lowering the movable legs the vessel is able to elevate the hull from sea water level. In elevated position the vessel provides a stable platform to perform installation activities. The legs are equipped with spudcans which serve as foundation of the vessel. The elevating process consists of a preload phase to ensure sufficient capacity to withstand operational and possible storm conditions. The preloading of four-legged jack-ups is performed by alternately applying vertical loads on diagonally opposite leg pairs, up to achieving a stable condition in which nearly constant load levels can be held by each leg. The aim of this research is to develop a 3D model to asses the preload duration of the jack-up vessel Aeolus in cohesive soil. The viscous behaviour of cohesive soil, like clay, influence the acting leg load during preload of the jack-up vessel. The shear strength of clay is a function of strain rate meaning that resistance increases due to viscous effects with increasing penetration rate. During spudcan penetration the shearing resistance is high but will reduce significantly when penetration is stopped as the viscoplastic resistance diminishes. Together with the onset of isotach soil behaviour this causes the loads to redistribute between the legs occurs. In this study it is assumed that sufficient preloading is achieved when the leg load reduction is limited to 400 ton / 15 min. To satisfy this criterionmultiple load cycles of each leg pair are performed. Site specific geotechnical data and information on the structural stiffness of the Aeolus have been available for this research and allowed for an accurate analysis of the processes during the preload procedure. The Soft Soil Creep (SSC) model is used as constitutive model and accounts for viscous effects by formulating irreversible strains by means of viscoplasticity. The soil at the project site is classified and the constitutive model is calibrated based upon the available soil test results. The structural behaviour of the vessel is captured via a simplified beam configuration representing the deck structure and legs, the stiffness of the beams is verified using the results from a so-called predrive analysis. The extension of the legs is established bymeans of negatively pre-stressed node-to-node anchors. Simulations of a single spudcan penetrating at various depths and penetration rates are performed to identify the extent of viscous strain rate effects fromthe results. With the developed 3D model Small Deformation Finite Element analyses of the preload procedure are performed. The leg loads and penetrations are monitored and compared to jacking data fromthe actual project site. The processes in the soil and structure are analysed and the influence on the preload procedure and preload duration is identified. For both type of simulations six different case-calculations are performed addressing variation in the initial spudcan depth, the OCR, the penetration rate, the permeability and the type of preload procedure. The simulations indicated that the penetration of a spudcan influences the penetration of an adjacent spudcan, this reciprocal influence of the spudcans emphasizes the importance of onemodel comprising all spudcans in the same 3D soil domain. The developed model slightly overestimates the spudcan penetration and underestimates the total preload duration. Simulations of the overshooting preload procedure and an alternative preload procedure are performed with the FE model. For the soil conditions used in this research, both the overshooting and the alternative procedure are effective in reducing the number of preload cycles to satisfy the preload criterion. Compared to the normal preload procedure, it is expected the overshooting procedure improves the preload duration. For the alternative procedure however, the duration of a preload cycle increases significantly and consequently the procedure does not improve the preload duration. Using a lower spudcan penetration rate during the normal preload procedure is also effective in reducing the number of preload cycles but significantly increases the elapsed time to complete the preload procedure. The above conclusions have been made on the basis of the model results, which is calibrated for the soil conditions at the specific project site.

In this study, the assessment of the KIJK dike follows the new guidelines given by POV-M 2018 using PLAXIS. The macrostability assessment and design of the dike is examined for two cases. In the first case the dike is assessed based on the in-situ conditions (Green dike) whereas in the second case a cantilever sheet pile wall is used in the design (Blue dike) to strengthen the dike. Prior to the analysis the interpretation of the available laboratory data that were executed according to the new protocol (WBI 2017) and the parameter determination of the necessary input parameters of the considered constitutive models (SHANSEP NGI-ADP, Hardening Soil, Soft Soil models) is established. The undrained shear strength (su) of the SHANSEP NGI-ADP model is based on the SHANSEP concept and the normalised su (S (NC, OC)) and the strength increase exponent (m) parameters are needed. For the HS and the SS models the effective strength parameters φ and c are required.
The FoS is determined with the use of the strength reduction method. Moreover, the CSSM framework is coupled with the SHANSEP (Stress History and Normalized Soil Engineering Properties) concept enabling the incorporation of the effect of stress history and the stress path characterizing the undrained shear strength in the design. The CSSM framework is applied in the design by determining the strength parameters of clays from an axial strain level equal to 25% under triaxial compression whereas for peats at 40% shear strain level under direct simple shearing conditions. In this study the strength parameters were additionally determined from the service conditions strain levels which are the 2% and 5% strain levels for clays and peats, respectively. The necessary strength parameters for the constitutive models are determined in the considered strain levels exposing their influence for both clays and peats. Moreover, the study includes the examination of the constitutive models at a single element level conducted in the Soil Test facility available in PLAXIS. When it comes to the design analysis, it was found that the SHANSEP NGI-ADP model fits properly the su determined from a cone penetration test. The HS and SS models based on the effective strength parameters may result in an undrained shear strength profile which deviates from the su that the soils exhibits in the field.
The study points out that both the strain level dependency of the strength parameters and the selection of the constitutive model at critical loading influence the results in terms of the developed failure plane and the FoS, especially in the case of Green dikes. The study also elaborates the response of the models in regard to the design requirements for the calculation of displacements and the structural forces. Finally, this study answers the knowledge gap regarding how the strain level influences the strength of the examined soils and how this should be translated in the design along with the explanation on the effects of the selected constitutive model on the results.

This research has shed light on the capability of the HASP model in reproducing the dilatant behaviour of OC clays in drained and undrained conditions. The model is formulated by employing a combined hardening rule and uses the void ratio as a state variable while maintaining the simplicity of the MCC model. However, a sensitivity analysis has revealed that the model is sensitive to some input parameters which when varied slightly can largely affect the outcome of an analysis.

This has led to the formulation of the PLAXIS OC clay model while maintaining the framework of the HASP model but replacing the void ratio with volumetric strain as the state variable. Thus resulting in the use of the modified compression (λ*) and swelling (κ*) indexes which are used in obtaining the soil stiffness parameters used as model inputs. The PLAXIS OC clay model features the use of real soil stiffness parameters instead of soil indexes, the addition of small stain stiffness by T. Benz to improve model prediction in the small strain region and the elimination of the sensitivity issues noticed when using the HASP model.

The PLAXIS OC clay model is validated for boom clay (BC) at single stress points by simulating CU test and comparing with the available experimental data for the BC. Good agreement is found with experimental data as shown in the stress strain, pore water pressure and stress path plots obtained from the analysis.

Furthermore, the model is used to simulate boom clay in a trial excavation. Piezometers and extensometers are installed into the BC layer prior to the excavation to monitor the changes in porewater pressure and vertical displacement (heave) on the BC during the excavation. A comparison of the numerical and experimental data shows that good agreement is observed in porewater pressures and vertical displacement in the BC layer.

In seismically active areas, liquefaction hazards have always been a complicated aspect to evaluate as part of the seismic design of a project. In the case of the design of critical facilities, this becomes crucial, as beyond design basis conditions may elevate the seismic loads significantly and create a considerable liquefaction risk in areas of deep alluvial deposits. Furthermore, traditional semi-empirical methods lose their applicability at depths larger than 15 m, which becomes problematic if one wishes to analyse the liquefaction hazard of deep Holocene deposits. Given this shortcoming and the rapid growth of numerical tools available for geotechnical earthquake engineering, the use of liquefaction-predicting constitutive models, like PM4Sand, provides the opportunity to obtain more accurate and physically-consistent results. For this purpose, this research is divided in three main parts. The first part covers the study of the onset of liquefaction with the use of two cyclic undrained direct simple shear test databases and the identification of liquefaction-triggering criteria, in terms of pore pressures and shear strains which can consistently define a liquefied state in sands. The second part includes the thorough analysis of the capabilities of the PM4Sand model, in Plaxis, through a benchmark calibration study using one of the previously mentioned laboratory test databases, concluding in the proposal of a modified calibration methodology based on pore pressure ratio (ru) and shear strain (γ) liquefaction-triggering criteria. The third and last part covered a practical case study oriented towards the design of a critical facility, where a beyond design liquefaction hazard analysis of a hypothetical site was evaluated incorporating the findings from the previous parts. A one-dimensional liquefaction hazard analysis was performed using a single earthquake signal and soil profile, where the consistency of the PM4Sand model in terms of liquefaction-triggering was evaluated and the numerically-obtained results were compared to those calculated through one semi-empirical method. Additionally, the one-dimensional model was extended and a two-dimensional liquefaction hazard analysis, including the presence of a simplified structure, was performed with the aim of evaluating the effects of soil-structure interaction and structural load variation on the liquefaction hazard of the soil profile over distance.

The design and reassessment of quays is subject to uncertainties. Examples of these uncertainties are the soil parameters, soil behavior and the current state of the structural parts. Especially in the reassessment of existing structures it is hard to prove a quay to be safe. In this thesis Bayesian updating is used as a method to reduce this uncertainty. Bayesian updating is a technique to probabilistically update the model prediction based on measurement data. One of the solutions to obtaining this data is test loading a quay wall. How to perform a test load and possible other solutions to obtaining the measurement data are not part of this research. The research focusses on how to use the data one obtains from a test loading. The effectiveness of test loading is reviewed by performing Bayesian updates with several fictitious measurement cases. Both Blum and a finite element model are updated based on fictitious strain and displacement measurements. The results show that Bayesian updating successfully reduces the standard deviations in all model predictions. This has a significant impact on the calculated reliability indices. The key benefit of performing a Bayesian update is that a statistically most likely combination of stochastic input parameters is determined. The thesis gives recommendations towards an application of Bayesian updating in practice and as fictitious measurement cases are used, an overall strategy is given for obtaining the required measurement data in practice.