Direct numerical simulation of dynamic aggregate migration in fresh concrete using multi-scale Smoothed-Particle Hydrodynamics

在SPP 2005中提出的研究项目的科学目标旨在通过微结构解析SPH模拟来提高我们对受限制几何形状的分级新鲜混凝土流动的理解。在这方面，将特别注意剪切引起的颗粒迁移和润滑层形成过程，因为它们对混凝土泵送具有技术意义。由于这些过程发生在比工程长度尺度小得多的长度尺度上，因此分解表面耦合直接数值模拟（DNS）被认为是合适的方法。为了解决DNS的重要计算需求，将使用高性能计算方法（大规模并行分布式内存计算）。我们的模拟方法应明确考虑到新混凝土的多尺度性质，因为载体流体的流变性以及数值分解的骨料含量可以自适应地进行修改，以在适当的长度和时间尺度上充分代表主导机制。考虑到我们对粒子迁移过程的兴趣，我们试图在模拟中表示的特征时间尺度将按照粒子自扩散时间的顺序排列。换句话说，模拟的时间尺度必须足以研究级配混凝土悬浮液中剪切诱导的动态骨料偏析过程。这意味着我们的模拟将与水化水泥悬浮液的化学反应性相关的时间尺度相分离。因此，本项目关注的是流动过程中新鲜混凝土微观结构的演变，即由于水动力和摩擦学效应而产生的离析效应，如颗粒大小和浓度分布的不均匀。我们的科学工作方案应该量身定制，使我们的模拟结果最终能够为开发泵送过程中新鲜混凝土流动的粗粒度宏观模型提供可靠的数据和知识基础。该数据基础应涵盖由混合设计和边界条件组成的特征空间。在这方面，本项目的预期产出被认为是对2005年SPP模块3的宝贵投入。给出了一个计算优化的SPH仿真框架。这个仿真框架是基于HOOMD-blue（高度优化的面向对象的多粒子动力学），并且在我们访问的计算机集群上已经证明了良好的并行性能。核心算法相关的模拟任意形状的刚性颗粒悬浮在连续载体液体（如数据结构，时间积分例程）已经实现。在提议的项目的持续时间内，相对较少的时间将用于软件实现。

The scientific objectives of the proposed research project within the SPP 2005 aim towards improving our understanding of graded fresh concrete flow in confined geometries using microstructure-resolved SPH simulations. In this regard, special attention will be devoted to shear-induced particle migration and lubrication layer formation processes motivated by their technological significance for concrete pumping. As these processes take place at length scales which are significantly smaller than the engineering length scale, resolved surface-coupled Direct Numerical Simulations (DNS) are considered appropriate approaches. In order to address the significant computational requirements of DNS, high-performance computing approaches will be used (massively parallel distributed memory computing). Our simulation approach shall explicitly take into account the multi-scale nature of fresh concrete in that the rheology of the carrier fluid as well as the numerically resolved aggregate content can be adaptively modified to adequately represent the dominating mechanisms at a suitable length and time scale. Given our interest in particle migration processes, characteristic time scales we attempt to represent in our simulations will be in the order of particle self-diffusion times. In other words, simulated time scales must be sufficient to study shear-induced dynamic aggregate segregation processes in graded concrete suspensions. This implies that our simulations will strongly be separated from time-scales associated with the chemical reactivity of hydrating cement suspensions. As a result, this project focuses on the evolution of fresh concrete microstructure during flow, i.e. resulting in segregation effects such as heterogeneous particle size and concentration distributions, due to hydrodynamic and tribological effects. Our scientific work program shall be tailored in such a way that our simulation results can ultimately provide a reliable data and knowledge basis to develop coarse-grained macroscopic models of fresh concrete flow during pumping. This data basis shall cover a feature space composed of mix design and boundary conditions. In that respect, the anticipated output of the present project is considered a valuable input to module 3 of the SPP 2005. A computationally optimized simulation framework for SPH simulations is available. This simulation framework is based on HOOMD-blue (Highly Optimized Object-oriented Many-particle Dynamics) and has previously proven good parallel performance on computer clusters we have access to. Core algorithms related to the simulation of arbitrary-shaped rigid particles suspended in continuous carrier liquids (e.g. data structures, time integration routines) have been implemented. Within the duration of the proposed project, comparatively little time will thus be devoted to software implementation.