The aim of this work is to describe a numerical framework for reliably and robustly simulating the different kinematic conditions exhibited by granular materials while spreading ---from a stagnant condition, when the material is at rest, to a transition to granular flow, and back to a deposit profile. The gist of the employed modeling approach was already presented by the authors in a recent work (Cante et al., 2014), but no proper description of the underlying numerical techniques was provided therein. The present paper focuses precisely on the detailed discussion of such numerical techniques, as well as on its rigorous validation with the experimental results obtained by Lajeunesse, et al. in Ref. ( Lajeunesse et al., 2004).\ud\udThe constitutive model is based on the concepts of large strains plasticity. The yield surface is defined in terms of the Drucker Prager yield function, endowed with a deviatoric plastic flow and the elastic part by a hypoelastic model. The plastic flow condition is assumed nearly incompressible, so a u - p mixed formulation, with a stabilization of the pressure term via the Polynomial Pressure Projection (PPP), is employed. The numerical scheme takes as starting point the Particle Finite Element Method (PFEM) in which the spatial domain is continuously redefined by a different nodal reconnection, generated by a Delaunay triangulation. In contrast to classical PFEM approximations ( Idelsohn et al., 2004), in which the free boundary is obtained by a geometrical technique (a-shape method), in this work the boundary is treated as a material surface, and the boundary nodes are removed or inserted by means of an error function. One of the novelties of this work is the use of the so-called Impl-Ex hybrid integration technique to enhance the spectral properties of the algorithmic tangent moduli and thus reduce the number of iterations and robustness of the accompanying Newton-Raphson solution algorithm (compared with fully implicit schemes respectively). The new set of numerical tools implemented in the PFEM algorithm – including new discretization techniques, the use of a projection of the variables between meshes, and the constraint of the free-surface instead using classic a-shape – allows us to eliminate the negative Jacobians present during large deformation problems, which is one of the drawbacks in the simulation of granular flows.\ud\udFinally, numerical results are compared with the experiments developed in Ref. (Lajeunesse et al., 2004), where a granular mass, initially confined in a cylindrical container, is suddenly allowed to spread by the sudden removal of the container. The study is carried out using different geometries with varying initial aspect ratios. The excellent agreement between computed and experimental results convincingly demonstrates the reliability of the model to reproduce different kinematic conditions in transient and stationary regimes.
展开▼
机译:这项工作的目的是描述一个数值框架,用于可靠而鲁棒地模拟粒状物料表现出的不同运动学条件,同时从物料静止时的停滞状态扩散到粒状流再返回到存款资料。作者在最近的工作中已经提出了采用的建模方法的要点(Cante等,2014),但是其中没有提供对底层数值技术的适当描述。本文恰好专注于对这种数值技术的详细讨论,以及对它的严格验证,以及Lajeunesse等人获得的实验结果。在参考文献中(Lajeunesse et al。,2004)。\ ud \ ud本构模型基于大应变可塑性的概念。屈服面是根据Drucker Prager屈服函数定义的,该屈服函数具有偏塑性流和次弹性模型赋予的弹性部分。假定塑性流动条件几乎不可压缩,因此采用u-p混合配方,通过多项式压力投影(PPP)稳定压力项。数值方案以粒子有限元方法(PFEM)为起点,在该方法中,通过不同的节点重新连接(由Delaunay三角剖分法生成)连续地重新定义空间域。与经典的PFEM近似(Idelsohn等,2004)不同,在自由近似中,自由边界是通过几何技术(a形方法)获得的,在这项工作中,边界被视为材料表面,边界节点为通过错误功能删除或插入。这项工作的新颖性之一是使用所谓的Impl-Ex混合积分技术来增强算法切线模的频谱特性,从而减少伴随的Newton-Raphson解算法的迭代次数和鲁棒性(与分别使用完全隐式方案)。 PFEM算法中实现的一组新的数值工具(包括新的离散化技术,网格之间变量的投影的使用以及自由表面的约束而非经典a形的约束)使我们可以消除负雅可比矩阵最后,将数值结果与参考文献中开发的实验进行了比较。 (Lajeunesse et al。,2004),最初被限制在圆柱形容器中的粒状物质由于容器的突然移开而突然散开。这项研究是使用具有不同初始纵横比的不同几何形状进行的。计算结果与实验结果之间的出色一致性令人信服地证明了该模型在瞬态和平稳状态下重现不同运动学条件的可靠性。
展开▼