Search

International audience; This paper aims to establish an intelligent procedure that combines the observational method with the existing deep learning technique for updating deformation of braced excavation in clay. The gated recurrent unit (GRU) neural network is adopted to formulate the forecast model and learn the potential rules in the field observations using the Nesterov-accelerated Adam (Nadam) algorithm. In the proposed procedure, the GRU-based forecast model is first trained based on the field data of previous and current stages. Then, the field data of the current stage are used as input to predict the deformation response of the next stage via the previously trained GRU-based forecast model. This updating process will loop up till the end of the excavation. This procedure has the advantage of directly predicting the deformation response of unexcavated stages based on the monitoring data. The proposed intelligent procedure is verified on two well-documented cases in terms of accuracy and reliability. The results indicate that both wall deflection and ground settlement are accurately predicted as the excavation proceeds. Furthermore, the advantages of the proposed intelligent procedure compared with the Bayesian/optimization updating are illustrated.

International audience; The aim of this paper is to investigate the effect of soil grading on the stress-strain and the critical-state behavior of granular materials from idealized spheres to natural soils as well as from discrete to continuous modeling. The three-dimensional discrete-element method has been applied to study the mechanical behavior of idealized granular materials. The results confirm that the critical-state line (CSL) shifts down-ward as grading broadens with an increase of the coefficient of uniformity Cu. An exponential relationship between the critical-state parameters and the coefficient of uniformity Cu can then be established. Furthermore, experimental investigations on an artificial material (glass beads) and a natural material (Hostun sand) were carried out. The experimental results confirm the grading dependency of the critical-state behavior, and a similar relationship between CSL and Cu, for both glass beads and natural sand, extending previous literature results for high Cu values up to 20. Moreover, the experimental results show that the CSL tends to stabilize when the coefficient of uniformity Cu becomes higher than 10. Based on all test results, a simple elastoplastic model accounting for the influence of the grain size distribution (GSD) on the mechanical behavior of granular materials has been developed within the framework of critical-state soil mechanics. Comparisons between experimental results and simulations show that the model can accurately reproduce the mechanical behavior of granular materials with different GSDs.

International audience; This paper aims at investigating the influence of grading on the undrained behavior of granular materials. Series of undrained triaxial tests were carried out with two different materials, glass balls and Hostun sand. For each material, samples with different gradings and similar relative densities were prepared. Experimental results show that the undrained shear strength decreases when the coefficient of uniformity Cu=d60/d10Cu=d60/d10 increases from 1.1 to 20. The conditions of instability for the selected granular materials were also analyzed, based on the sign of the second-order work during undrained triaxial loading. The results demonstrate a significant influence of the grading: increasing the coefficient of uniformity heightens the potential of static liquefaction and the materials become more unstable. Furthermore, numerical tests using the three-dimensional discrete element method (DEM) were conducted on assemblies of spheres. The DEM inter-particle parameters were derived from the experimental test results on glass balls. The DEM simulations showed similar behaviors compared to experimental results and confirmed the influence of the grain size distribution on the stress–strain relationship and instability phenomena.; Cet article vise à étudier l'influence de l'étalement granulométrique sur le comportement non drainé des matériaux granulaires. Des séries d'essais triaxiaux non drainés ont été réalisés sur deux matériaux différents, des billes de verre et du sable Hostun. Pour chaque matériau, des échantillons de différentes granulométries et de densités relatives similaires ont été préparés. Les résultats expérimentaux montrent que la résistance au cisaillement diminue lorsque le coefficient d'uniformité Cu=d60/d10Cu=d60/d10 augmente de 1,1 à 20. Les conditions d'instabilité pour les matériaux granulaires sélectionnés ont également été analysées, sur la base du signe du travail du second ordre au cours d'un chargement triaxial non drainé. Les résultats montrent une influence significative de l'étalement granulométrique : l'augmentation du coefficient d'uniformité accroît le potentiel de liquéfaction statique et les matériaux deviennent plus instables. En outre, des essais numériques en utilisant la méthode des éléments discrets en trois dimensions (DEM) ont été effectués sur des assemblages de sphères. Les paramètres inter-particules DEM ont été calculés à partir des résultats d'essais expérimentaux sur des billes de verre. Les simulations DEM ont montré des comportements similaires comparés aux résultats expérimentaux et confirment l'influence de l'étalement granulométrique sur la relation contrainte–déformation et les phénomènes d'instabilité.