Orianne Jenck

Pore water pressure (PWP) response is significant for evaluating the earth dams’ stability, and PWPs are, therefore, generally monitored. However, due to the soil heterogeneity and its non-linear behavior within earths, the PWP is usually difficult to estimate and predict accurately in order to detect a pathology or anomaly in the behavior of an embankment dam. This study endeavors to tackle this challenge through the application of diverse machine learning (ML) techniques in estimating the PWP within an existing earth dam. The methods employed include random forest (RF) combined with simulated annealing (SA), multilayer perceptron (MLP), standard recurrent neural networks (RNNs), and gated recurrent unit (GRU). The prediction capability of these techniques was gauged using metrics such as the coefficient of determination (R²), mean square error (MSE), and CPU training time. It was found that all the considered ML methods could give satisfactory results for the PWP estimation. Upon comparing these methods within the case study, the findings suggest that, in this study, multilayer perceptron (MLP) gives the most accurate PWP prediction, achieving the highest coefficient of determination (R² = 0.99) and the lowest mean square error (MSE = 0.0087) metrics. A sensitivity analysis is then presented to evaluate the models’ robustness and the hyperparameter’s influence on the performance of the prediction model.

In the absence of high-quality soils, the rigid inclusion reinforcement technique has been extensively used in geotechnical engineering practice to meet the serviceability requirements of construction projects. This article aims to investigate the behavior of a single footing placed on soft soil reinforced by four rigid inclusions using three-dimensional (3D) finite difference modeling (FDM). A set of full-scale experimental tests are used as references for the development of the numerical approach. The concurrence of the numerical and experimental data allows us to assess the load transfer platform thickness influence on the system under centered and eccentric loading conditions. Some preliminary insights can thus be obtained based on this analysis for the soft soil improvement using rigid inclusions at the design stage.

La Revue Française de Géotechnique (RFG) est une publication scientifique parrainée par les comités français de mécanique des sols, de mécanique des roches, de géologie de l’ingénieur et des géosynthétiques

Foundations of offshore and nearshore wind energy production systems are subjected to multidirectional and cyclic loads, due to the combined action of wind and waves and in the particular case of mutualized anchor foundations for floating wind turbines, to the phase shift between the loads generated in the adjacent anchored turbines. This article presents a three-dimensional numerical model developed with FLAC3D to analyse the impact of the change in direction of the horizontal load during the cycles. The typical case of a 1.7 m diameter and 10 m-long pile founded in a dense homogeneous sand is considered. A specific procedure has been implemented to apply force-controlled cycles with a change in lateral load direction. The results are compared to mono-directional lateral cyclic loads with the same average and cyclic forces. The results of the parametric study highlight the effect of the average value and amplitude of the cyclic loading on the accumulation of pile head horizontal displacements during the cycles. When a multidirectional cyclic loading is applied, it also leads to an accumulation of the deviated horizontal displacements, and the resulting accumulated horizontal displacements are larger than for a mono-directional cyclic loading of the same amplitude.

A technical solution for tidal turbine foundation in granite seabed consists of grouted steel piles. The piles would be subjected to cyclic loading due to the severe service conditions. The mechanical behaviour at the interface between the pile and the surrounding media is one of the key points that determine the bearing capacity of the foundation system. Experimental research work has been carried out in the laboratory to study the grouted pile-to-rock connection (GPRC) and focused more precisely on the pile to grout connection when the pile is equipped with shear-keys that enhance the interface capacity. Monotonic and cyclic shear tests were performed using a specific direct shear test device (BCR3D), allowing application of sample confining conditions close to the in-situ conditions, namely constant normal stiffness conditions. Cyclic tests - either under one way or two-way shear loading application - were performed on several samples, under both constant volume and various constant stiffness boundary conditions. This study has shown that the strength of the interface is highly dependent on the combination of mean load, cyclic amplitudes and number of cycles applied to the interface. Cyclic failure was observed with less than 30 cycles for two-way and one-way cyclic loading.

The research work presented in this paper is a contribution to the understanding of the behaviour of soil reinforcement by rigid inclusions under a quasi-static vertical cyclic loading. While the system behaviour under static loading is well known, the response of this technique under cyclic and/or dynamic loading is not yet fully understood and cannot be predicted. The first step of the study consists in performing a series of tests on a small-scale laboratory model, with a scale factor of 1/10 on the lengths, simulating the case of rigid inclusions using different thicknesses of the load transfer platform (LTP) considering different boundary conditions at the ground surface. A Digital Image Correlation (DIC) method is also used in order to study the displacement field inside the LTP. The parametric study performed with the physical model gives an insight on the influence of LTP's thickness and on the presence of a rigid slab at its surface on the behaviour of this technique under cyclic loading. The experimental results highlighted an accumulation of the settlements during the cyclic loading in all the studied cases. They also showed that the use of a small LTP thickness decreases the settlements’ accumulation observed during the cycles. Results also showed that the presence of the rigid slab increases the settlements’ accumulation. Nonetheless, its presence results in a reduction of the total settlement during a further monotonic loading. In a second step, the experimental results will serve as a database for the validation of a 3D numerical model.

While the behaviour of rigid inclusion improvement systems under monotonic loading is well known, the response under cyclic loading is not yet fully understood and cannot be predicted due to the lack of knowledge about the load transfer mechanisms in the granular platform. This article proposes a three-dimensional numerical modelling under a quasi-static vertical cyclic loading. Numerical results are compared to experimental results obtained on a three-dimensional small scale model, to discuss the numerical procedure and the constitutive models implemented. For the load transfer platform, a shear hardening elastoplastic model is implemented; for the soft soil, the modified Cam Clay model is used, as it features well the compressive and swelling behaviour. Some aspects of the behaviour of the system could be correctly reproduced, such as the loads on inclusions and the settlements under monotonic loading, the settlement accumulation and the decrease of the load transfer onto the inclusions along the cyclic loading, as well as the qualitative influence of a rigid slab placed at the platform surface on the settlement accumulation. However, other features could not be satisfactorily taken into account, namely the behaviour under a rigid slab submitted to cyclic loading in terms of load on the inclusions.