A large share of global shipped tonnage concerns dry bulk materials, which are typically unloaded from ships using large mechanical grabs, a slow process with long waiting times and high emissions. This process can be modelled and its performance improved using the Discrete Element Method (DEM). While such research has been performed for materials like iron ore or coal, no accurate and efficient material model exists for coarse limestone, a coarse material that is difficult to penetrate and for which large improvements in handling efficiency are expected, reducing waiting times and emissions. Previous research using a material model with two-spherical particles focused on the penetration resistance of limestone and experimentally determined various characteristics of limestone, but with a computational time of 18 days did not result in a feasible full-scale material model . Particle shape-related characteristics like interlocking and dilatancy are expected to influence the bulk material behaviour of limestone, to be taken into account when making a numerical model.

A numerical model is to be created in order to evaluate and improve the grab performance in limestone. Using the state of the art experimental results as a basis, additional experimental setups were designed to determine additional material and particle characteristics, like coefficients of sliding friction, the Angle of Repose (AoR). A method was set up to classify the particle shape. Degrading particles showed that the particle roundness (angularity) influences a particle’s frictional behaviour (and thus AoR). Full-scale experiments were performed in-situ using a full-sized grab.

A DEM material model of limestone is created based on a smaller-scale lifting cylinder (AoR) setup and a full-scale grab setup. Ten different particle shapes are modelled using 5-spherical particles fitted to particle templates, while discarding the PSD. The model is first calibrated for the small-scale setup. These settings are then verified and optimised for the full-scale setup, with the optimised model coarse-grained with a factor of ×1.5, resulting in a calculation time of 3 hours and a standard deviation of 1.8% of the mean. The model accurately represents payload, knife penetration, and knife path.

Design improvements are modelled. Analysis showed that small design improvements are possible, increasing the grab performance significantly. The thesis showed that it is possible to make an efficient material model which accurately represents the coarse material by using multi-spherical particles reflecting the material’s particle shape distribution and calibrated based on a lifting cylinder (AoR) and full-scale (payload, knife path, penetration) experiment. The selection of particle shapes allows for accurate modelling of interlocking and dilatancy in a computationally efficient material model.

The dredging world is continuously growing, improving and adjusting to new challenges. As the demand for sustainable dredging practices grows and the availability of sand declines, the industry is increasingly exploring alternative materials, including fine-grained soils. This transition presents new challenges, particularly in understanding the behavior of these soils during dredging and reclamation. When stiff cohesive soils are dredged, they often form large lumps or balls, which are subsequently transported and deposited within a reclamation fill alongside a slurry.

The deposited clay lumps create a matrix with significant inter-lump voids, forming a double porosity system, consisting of these inter-lump voids and the voids within the lump, that significantly affects settlement predictability. Current settlement calculation methods typically assume a uniform layer, with soil parameters derived from the soil parameters of dredged material, which are later adjusted based on field feedback. This approach introduces substantial uncertainty, potentially leading to greater risks and increased costs.

This thesis investigates the complex deformation behavior of stiff, lumpy double porosity fills by performing both field data analysis and controlled experimental research. Field observations, including Cone Penetration Tests (CPTs), borehole data, and settlement plate measurements from a real-world reclamation project, are used to evaluate existing settlement models and the limitations of current design assumptions. In parallel, an experimental campaign involving oedometer testing and micro-CT scanning was conducted on clay lumps of varying undrained shear strengths($S_u$) to quantify deformation of macro- and, micro-voids and softening effects.

Findings reveal that traditional 1D-consolidation models, based on homogeneous assumptions, underestimate settlement magnitudes and rates during early loading phases dominated by macro-void collapse. Macro-void closure was observed to occur primarily under vertical effective stresses between 0.5 - 2 $S_u$, after which the fill behaves more homogeneously but remains structurally heterogeneous at the micro-scale. Experimental results highlighted the influence of lump strength and softening processes on overall deformation, with weaker lumps exhibiting void collapse at lower stress levels.

The study concludes that accurate settlement prediction of lumpy fills requires explicit consideration of initial macro-structure collapse, evolving lump strength, self-weight consolidation, and heterogeneous stress transfer. While conventional isotache-based models remain suitable after macro-void closure, early-stage design should incorporate adaptive, probabilistic approaches. Recommendations for future research include refining laboratory testing protocols, improving imaging resolution for micro-structural tracking, and expanding long-term monitoring to better understand the time-dependent evolution of lumpy fills under field conditions.

Debris flows are extremely rapid gravity-driven mass movements of saturated sediment in concentrations between 60% to 80% by volume that move along steep channels, eroding and entraining material, typically terminating in a fan-shaped deposit. Debris flows are responsible for causing innumerable deaths and extensive damage across the world. The mobilisation of rainfall-triggered landslides is the primary cause of such flows. Estimating the debris flow travel distance or runout is essential for managing this hazard.

The conventional approach to debris flow runout analysis idealises the triggering of landslides across the source area into an instantaneous event. When this idealisation is done, a key characteristic of debris flows, their tendency to propagate in surges or waves, is overlooked. This study analyses the effect of accounting for surging on runout estimation by spatially and time-resolving debris flow events. A prime candidate for such a study is the Mocoa Debris Flow of 2017, a tragedy that involved 273 shallow landslides mobilising into a debris flow, resulting in the death of more than 300 people and the devastation of local infrastructure, Sarmiento et al. (2019). First, the landslide inventory was analysed to assess the scale of the disaster. Then, a novel method to spatially resolve events based on stream orders is implemented, after which a runout analysis is performed for different spatially resolved scenarios using a depth-averaged numerical model. Next, the timing and distribution of landslides are assessed based on a four-day storm period using a process-based landslide susceptibility model. This assessment determines the relative volume of each debris flow surge. The surges are then incorporated into a time-resolved runout analysis. The results of both the spatially and time-resolved runout analyses are compared. We find a marginal difference in the estimate of runout based on critical performance criteria, such as area coverage ratio, in favour of the spatially resolved analysis.

This study concludes that incorporating the phenomenon of surging caused by non-simultaneous landslide events does not improve the forensic analysis of the Mocoa Debris Flow runout. One of the main limitations of this study is the absence of data measured during the event to confirm the extent of surging due to non-simultaneous landslides. A possible avenue for future research would be varying the period of rainfall that is considered by reevaluating initial groundwater conditions.

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 past decade, suction caissons have emerged as a preferred offshore foundation solution for wind turbines due to their silent installation process and potential for recyclability. However, there has been growing speculation regarding the necessity of under base filling, which involves filling the gap between the top plate of the suction caisson and the seabed. Some experts have suggested that under certain conditions, this under base filling may not be required at all. Furthermore, concerns have been raised about the efficacy of under base filling in achieving full contact between the top plate and the seabed, as it has been observed that gaps may persist even after the filling is applied. Consequently, doubts have been cast on the overall need for under base filling. However, there is limited research focused on understanding the behavior of water plugs in the absence of under base filling, at different loading conditions ( Compression , tension , cyclic etc.). This knowledge gap motivates this thesis study, which aims to investigate the behavior of water plugs specifically in dense sand samples, as sand is considered more critical compared to clay in terms of its variability in drainage conditions that can influence the foundation’s performance. To achieve this, a series of centrifuge tests were conducted on suction caissons that were partially installed and some fully installed. The results of the experiments shed light on the role of under base filling in different loading scenarios. Under monotonic compressive loading at higher rates, it was observed that under base filling played no significant role in the load transfer . Both the caissons with and without under base filling exhibited similar load transfer mechanisms, indicating that filling the gap may not be necessary in such loading conditions. Additionally, under tension loading, it was found that under base filling had little to no effect on the development of tensile capacity. By expanding our understanding of the necessity and effectiveness of under base filling, this study contributes to the ongoing discussion surrounding suction caisson design and installation practices for offshore wind turbine foundations.