A newly developed line-style sand pluviator has been calibrated to prepare repeatable sand specimens of specific statuses of compactness and homogeneity for laboratory tests. Sand is falling via a bottom slot of a fixed hopper, and by moving the sample container under the slot, the container is evenly filled with sand. The pluviator is designed with high flexibility: The falling height of sand, the hopper’s opening width and the relative moving speed between the hopper and the sample box can be easily adjusted. By changing these control factors, sand specimens of a wide range of densities can be prepared. A series of specimen preparation was performed using the coarse Merwede River sand. Performance of the pluviator was systematically evaluated by exploring the alteration of achievable density, as well as checking the homogeneity and fabric of the prepared samples by CT scanning. It was found that the density of prepared coarse sand samples has monotonic correlations with none of the three control factors. Furthermore, CT scanning results suggested that the prepared samples exhibited excellent homogeneity in the horizontal direction but periodical alteration of density in the vertical direction. Based on these calibration test results, a preliminary hypothesis is proposed to describe the general working principles of this type of pluviators a priori, illustrating the mechanisms dominating the non-monotonic correlations between control factors and the relative density as well as the vertically prevalent heterogeneity of specimens. Accordingly, practical recommendations are made in a unified framework in order to lessen the load of similar calibration work.

@en

The authors present a comprehensive numerical study on the lateral response of pile foundations in sands with a constant relative density. The influence of the pile configuration (length (L) diameter (D) and flexural rigidity), load eccentricity (h), sand type and relative density (Dr) was investigated. This discussion provides some additional insights on the lateral response of large diameter monopiles in uniform sand by combining the authors work with the observations in Wang et al. (2021) and Richards et al. (2021).

@en

This paper presents a synthesis of recent and new research conducted by the authors on laterally loaded monopiles in drained sand. The research involved reduced-scale field tests, centrifuge model tests, finite element (FE) simulations and comparisons of design approaches with published experimental data. The influence of the monopile base on lateral response is first discussed by drawing on field tests and numerical simulations and it is shown that the base generally provides a negligible contribution. The applicability of the API p-y formulation is then investigated through systematic FE analyses. The results show that this formulation leads to inaccurate predictions largely due to the assumption of a high initial stiffness varying linearly with depth and an unrealistic hyperbolic tangent back-bone function. Based on new insights into pile-soil interaction together with elastic simulations of laterally loaded rigid piles and new observations based on 26 pile tests, a simple rotational spring model is proposed to allow rapid quantification of the non-linear response of rigid monopiles in uniform sand. The effect of monopile flexibility is then added through a new straightforward correction factor based on 80 extra FE simulations. Finally, an example application of the proposed approach for a typical monopile design is presented.

@en

Piles have been widely used as foundations to resist lateral loads. For the design of a laterally loaded pile, one of the most important inputs is the ultimate soil resistance (p_ult = K_ultDσv^′,whereKult is the ultimate lateral soil resistance coefficient, D is the pile diameter, and σv^′ is the vertical effective stress). However, great discrepancy can be found in the existing design equations for piles in sand. To provide new insights and clarify the discrepancy in previous studies, in this study, a series of numerical simulations were performed on piles of different configurations using the finite element model validated by centrifuge pile tests. The computed results suggest that K_ult is a function of depth ratios z/D and z/L for the flexible and rigid piles, respectively (where z is the absolute depth and L is the embedded pile length), and all existing design equations failed to reproduce the magnitude and distribution of K_ult . Additionally, the K_ult of horizontally translated fixed-head rigid piles exhibits the same pattern as that of free-head flexible piles, suggesting that the difference between free-head flexible piles and rigid piles is caused by the change of failure modes.

@en

Predicting the non-linear loading response is the key to the design of suction caissons. This paper presents a systematic study to explore the applicability of deep learning techniques in foundation design. Firstly, a series of three-dimensional finite element simulations was performed, covering a wide range of embedment ratios and different loading directions, to provide training data for the deep neural network (DNN) model. Then, hyper-parameter tuning was performed and it is found that the basic Fully-Connected (FC) neural network model is sufficient to capture the non-linear response of suction caissons with excellent accuracy and robustness. Furthermore, the optimized FC neural network model was also successfully applied to a database of suction caissons in sand, demonstrating its broad applicability. By comparing three typical DNNs, i.e., FC, Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM), it was observed that the FC neural network model excels over others in terms of simplicity, efficiency and accuracy. More importantly, by looking into the model's generalization performance, the FC neural network model can also identify the change in foundation failure mechanisms. This study demonstrates the DNN's powerful mapping ability and its potential for future use in offshore foundation design.

@en

Predicting the nonlinear load response of caisson foundations is critical to the foundation design. Despite extensive studies aimed at developing models for predicting the combined V-H-M bearing capacity of suction caissons in clay, accurately predicting the three-dimensional (3D) deflection response of the foundation remains a significant challenge. In this paper, we present a novel solution by developing a fully connected (FC) neural network model that enables load-deflection prediction of suction caissons on clay. To train and evaluate the FC model, a series of 3D finite element simulations were performed covering caissons responses with an embedment ratio of up to 1. The effect of various model hyperparameters on the model's prediction accuracy and generalisation ability was systematically investigated. The results show that the proposed model achieves load-deflection response prediction with simplicity, efficiency and accuracy, demonstrating the significant potential of deep learning technology in the geotechnical design of foundations.

@en

The response of monopiles to lateral loading has attracted considerable research interest in recent years. As monopile foundations are exposed to ever-harsher environmental conditions, the engineering tools used for their simulation should continually update and improve. Recently, the challenge of simulating the behaviour of monopiles under lateral loads has been addressed to a significant extent through a combination of numerical modelling and experimental data. Although monotonic response calculations are still relevant to monopile design, it should be acknowledged that offshore environmental loads are inherently cyclic. To improve the engineering tools for the simulation of cyclic monopile behaviour and our understanding of the relevant geotechnical mechanisms, this study presents and discusses the outcome of advanced 1D cyclic soil reaction modelling of monopile-soil interactions employed to simulate centrifuge data conducted as part of the MIDAS research project. The memory-enhanced p-y model proves capable of simulating cyclic ratcheting behaviour in complex loading histories, which promotes the discussion for the evolution of relevant soil reaction mechanisms during cyclic loads. Finally, preliminary calibration strategies for the employed cyclic soil reaction models are presented.

@en

Monopiles are the most popular foundation for offshore wind turbines. These foundations typically have a low length to diameter ratio and undergo a rigid body rotation when subjected to lateral load. This paper presents results from an extensive numerical investigation involving 3D finite element analyses to demonstrate that the lateral moment-rotation response of a monopile in sand can be represented using a single non-linear rotational spring located at a depth of about 0.75 times the pile embedment. Expressions for the elastic rotational stiffness of a monopile under very low rotations are developed and these combined with observations from measured non-linear variations of rotational stiffness, that are supported by the numerical analyses, are used to develop a simple approximate expression that can be used to determine the response of a monopile to a monotonic lateral load in sand.

@en

The scour protection layers are usually installed around the monopiles to prevent the formation of scour hole. While extensive studies have been performed on the effectiveness of scour protection layer as mitigation measure, no study to date has been found to quantify the influence of scour protection layer on the lateral response of monopiles. In this study, centrifuge tests and numerical analyses were performed to examine explicitly the influence of the scour protection layer on the monopile response under lateral load in sand. It was found that a scour protection layer with a diameter of 5 times the diameter of the monopile (5D) and equivalent surcharge pressure of 15 kPa can increase the foundation lateral capacity by more than 30% and significantly decrease the accumulation of deflection by more than 100% in dense sand. Numerical analyses suggested that the beneficial contribution is attributed to the densification of sand and the increase of effective stress around the pile, which increased the stiffness and ultimate resistance of the soil reaction curve (i.e., p-y curve). Numerical parametric studies suggested that the beneficial effect of scour protection layer is also applicable to monopile in loose sand and by considering the contribution of the scour protection layer, the embedment length of the monopile can be reduced by 10%.

@en

The large-diameter monopiles are the most preferred foundation used in offshore wind farms. However, the influence of pile diameter and aspect ratio on the lateral bearing behavior of monopiles in sand with different relative densities has not been systematically studied. This study presents a series of well-calibrated finite-element (FE) analyses using an advanced state dependent constitutive model. The FE model was first validated against the centrifuge tests on the large-diameter monopiles. Parametric studies were performed on rigid piles with different diameters (D = 4–10 m) and aspect ratios (L/D = 3–7.5) under a wide range of loading heights (e = 5–100 m) in sands with different relative densities (Dr = 40%, 65%, 80%). The API and PISA p-y models were systematically compared and evaluated against the FE simulation results. The numerical results revealed a rigid rotation failure mechanism of the rigid pile, which is independent of pile diameter and aspect ratio. The computed soil pressure coefficient (K = p/Dσ^′ v) of different diameter piles at same rotation is a function of z/L (z is depth) rather than z/D. The force–moment diagrams at different deflections were quantified in sands of different relative density. Based on the observed pile–soil interaction mechanism, a simple design model was proposed to calculate the combined capacity of rigid piles.

@en