Discrete element model for crack propagation in brittle materials

We propose a discrete element model for brittle rupture. The material consists of a bidimensional set of closed‐packed particles in contact. We explore the isotropic elastic behavior of this regular structure to derive a rupture criterion compatible to continuum mechanics. We introduce a classical criterion of mixed mode crack propagation based on the value of the stress intensity factors, obtained by the analysis of two adjacent contacts near a crack tip. Hence, the toughness becomes a direct parameter of the model, without any calibration procedure. We verify the consistency of the formulation as well as its convergence by comparison with theoretical solutions of tensile cracks, a pre‐cracked beam, and an inclined crack under biaxial stress. Copyright © 2015 John Wiley & Sons, Ltd.     

Modeling the 2D behavior of dry‐stone retaining walls by a fully discrete element method

The dry‐stone retaining walls (DSRW) have been tipped as a promising solution for sustainable development. However, before recently, their behavior is relatively obscure. In this study, discrete element method (DEM) approach was applied to simulate the plane strain failure of these walls. A commercial DEM package (PFC2D™) was used throughout this study. The authors used a fully discrete approach; thus, both the wall and the backfill were modeled as discrete elements. The methodology for obtaining the micromechanical parameters was discussed in detail; this includes the three mechanical sub‐systems of DSRWs: wall, backfill and interface. The models were loaded progressively until failure, and then the results were compared with the full‐scale experimental results where the walls were loaded, respectively, with hydrostatic load and backfill. Despite its complexity and its intensive calculation time, DEM model can then be used to validate a more simplified approach. Copyright © 2015 John Wiley & Sons, Ltd.     

A probabilistic approach for computing water retention of particulate systems from statistics of grain size and tessellated pore network

The paper offers an analytical determination of the hydraulic properties of an unsaturated soil with reference to its retention curve, which describes the relationship between the volumetric water content and capillarity through matric suction. The analysis combines a particulate approach focused on the physics at the pore scale, including microstructural aspects, with a probabilistic approach where the void space and grain size are considered as random variables. In the end, the soil water characteristic curve of an unsaturated granular medium along a drying path can be derived analytically based on the sole information of particle size distribution. The analysis hinges on the tessellation of a wet granular system into an assemblage of tetrahedral unit cells revealing a pore network upon which capillary physics are computed with respect to pore throat invasion by a non-wetting fluid with evolving pendular capillary bridges. The crux of the paper is to pass from particle size probability distribution to a matching void space distribution to eventually reveal key information such as void cell and solid volume statistics. Making reasonable statistically based assumptions to render calculations tractable, the water retention curve can be readily constructed. Model predictions compare quite favourably with experimental data available for actual soils, especially in the high saturation range. Having a sound scientific basis, the model can be made amenable to address a variety of soils with a wider range of particle sizes.     

Granular element method for three‐dimensional discrete element calculations

This paper endows the recently‐proposed granular element method (GEM) with the ability to perform 3D discrete element calculations. By using non‐uniform rational B‐Splines to accurately represent complex grain geometries, we proposed an alternative approach to clustering‐based and polyhedra‐based discrete element methods whereby the need for complicated and ad hoc approaches to construct 3D grain geometries is entirely bypassed. We demonstrate the ability of GEM in capturing arbitrary‐shaped 3D grains with great ease, flexibility, and without excessive geometric information. Furthermore, the applicability of GEM is enhanced by its tight integration with existing non‐uniform rational B‐Splines modeling tools and ability to provide a seamless transition from binary images of real grain shapes (e.g., from 3D X‐ray CT) to modeling and discrete mechanics computations.Copyright © 2013 John Wiley & Sons, Ltd.     

An original semi‐discrete approach to assess gas conductivity of concrete structures

For civil engineering structures with a tightness role, structural permeability is a key issue. In this context, this paper presents a new proposition of a numerical modelling of leakage rate through a cracked concrete structure undergoing mode I cracking. The mechanical state of the material, considered in the framework of continuum mechanics based on finite element modelling, is described by means of the stress‐based nonlocal damage model which takes into account the stress state and provides realistic local mechanical fields. A semi‐discrete method based on the strong discontinuity approach to estimate crack opening is then considered in the post‐treatment phase. Using a Poiseuille's like relation, the coupling between the mechanical state of the material and its dry gas conductivity is performed. For validation purposes, an original experimental campaign is conducted on a dry concrete disc loaded in a splitting setup. During the loading, gas conductivity and digital image correlation analysis are performed. The comparison with the 3D experimental mechanical global response highlights the performance of the mechanical model. The comparison between crack openings measured by digital image correlation and estimated by the strong discontinuity method shows a good agreement. Finally, the results of the semi‐discrete approach coupled with the gas conductivity compared with experimental data show a good estimation of the structural conductivity. Consequently, if the mechanical problem is well modelled at the global scale, then the proposed approach provides good estimation of gas conductivity. Copyright © 2016 John Wiley & Sons, Ltd.     

结合模平方的双树复小波变形监测数据滤波

针对变形监测数据的去噪问题,该文在分析离散小波变换去噪不足的基础上,提出了一种基于模平方的双树复小波变形监测数据滤波方法。该方法利用双树复小波变换的完全重构、近似平移不变性和较好的方向选择性等特点,通过最小尺度空间的小波系数得到噪声强度,并结合模平方处理法确定各层的阈值,经重构阈值处理后的各层小波系数即得到去噪后的信号;经算例,并与传统离散小波变换对比分析。结果表明:双树复小波变换的分解效果优于传统离散小波变换,能较好地表现出细节部分的频率信息,使变形信号的周期性变化特征更为明显。该方法去噪更彻底,进一步提高了消噪的精度和可靠性,可作为变形监测数据降噪处理的新方法。     

基于重力全张量数据的Parker-Oldenburg算法研究

相比重力数据,重力张量数据通常包含更多的异常信息。本文根据重力数据与重力张量数据的关系,利用位场转化技术,将重力张量数据应用于传统的Parker-Oldenburg密度界面反演算法中。通过模型试验,证明了在网格间距较大或者数据存在一定噪音时,使用本文算法进行反演能得到更好的效果。实验结果说明利用重力张量数据可以有效地提高密度界面反演的分辨率。     

基于改进K-SVD字典学习方法的地震数据去噪

为实现更好的地震数据去噪技术,笔者引入一种新的算法:快速迭代收缩阀值法(FISTA),通过FISTA和K-奇异值分解(K-SVD)不断迭代更新K-SVD字典,利用更新得到的K-SVD字典对地震数据进行稀疏表示,去除稀疏系数中较小的数值,使数据中的随机噪声得到压制。对层状模型合成地震记录,Marmousi模型合成地震记录以及实际地震数据进行对比实验,得出FISTA算法较OMP算法能更好地提高地震数据的信噪比,同时有效地保护了反射信号。