Identification of the mechanical behavior and coupling with improved internal structural analysis of frozen particle-fluid-systems

冷冻颗粒-流体系统（PFS）可以被认为是颗粒增强复合材料，在自然和技术系统中经常遇到。当PFS中的水基流体在低温下经受机械载荷时，会发生相变，导致其机械行为和相关性能发生变化。例如，在北极地区，土壤稳定性对建筑项目至关重要，人工地面冻结（AGF）通常用于稳定采矿和隧道工程的土壤。同样，在低温下储存在筒仓中的材料或经受低温研磨的材料暴露于其机械性能的变化。因此，重要的是使用原位方法研究冷冻PFS，以了解其在现实条件下的特性和行为。目前，不同的方法被用来模拟冻结PFS，包括有限元法（FEM）模拟和离散元法（DEM）模拟与交替接触模型。然而，这些方法都有局限性，例如较高的计算成本，离散化的问题，以及无法精确描述复杂材料的行为。为了克服这些限制，本项目提出使用键合颗粒模型（BPM），它通过用桥连接颗粒来扩展DEM，形成团聚体。一个新开发的固体债券模型考虑蠕变行为来描述冻结PFS在不同的温度，应变速率和体积冰含量的力学。还将进行微型计算机断层扫描（Micro-CT）测量，以根据不同体积的冰含量和颗粒生成冷冻PFS的内部结构。将使用这种使用X射线的具有高空间分辨率的切片图像的内部结构的非破坏性三维成像。利用原位Micro-CT装置，将开发小型化单轴压缩试验装置，以真实的实时捕获耦合机械载荷的冻结过程。人工神经网络（ANN）将被用来校准的DEM-BPM材料参数，并预测在冻结和加载过程中的团聚体的机械性能。通过所提出的方法可以显着减少预测团聚体的力学性能所需的计算时间。为此，将创建和扩展一个由数千个模拟组成的数据库，以与其他场景和现实实验数据相关联，这将进一步增强该工具的预测能力。

Frozen particle-fluid systems (PFS) can be considered as particle-reinforced composite materials and are frequently encountered in both natural and technical systems. When subjected to mechanical loading at low temperatures for water-based fluids within the PFS, a phase transition occurs, leading to changes in their mechanical behavior and associated properties. For instance, soil stability is critical in the Arctic for construction projects, and artificial ground freezing (AGF) is commonly used to stabilize soil for mining and tunneling projects. Similarly, materials stored in silos at low temperatures or subjected to cryogenic grinding are exposed to changes in their mechanical properties. Therefore, it is important to investigate frozen PFS using in-situ methods to understand their characteristic properties and behavior under realistic conditions. Currently, different approaches are used to simulate frozen PFS, including finite element method (FEM) simulations and discrete element method (DEM) simulations with an alternate contact model. However, these approaches have limitations, such as higher computational costs, problems with discretization, and an inability to precisely describe the behavior of complex materials. To overcome these limitations, this project proposes using the Bonded Particle Model (BPM), which extends the DEM by connecting particles with bridges, forming agglomerates. A newly developed solid bond model considers creep behavior to describe the mechanics of frozen PFS in relation to different temperatures, strain rates, and volumetric ice contents. Micro-computed tomography (Micro-CT) measurements will also be carried out to generate the internal structure of frozen PFS based on different volumetric ice contents and particles. This non-destructive three-dimensional imaging of the internal structure with a high spatial resolution of the slice images using X-rays will be used. Using an in-situ Micro-CT device, a miniaturized uniaxial compression test apparatus will be developed to capture the freezing process with coupled mechanical loading in real time. Artificial neural networks (ANNs) will be used to calibrate the DEM-BPM material parameters and predict the mechanical properties of the agglomerate during freezing and loading. The computational time necessary for predicting the mechanical properties of agglomerates can be significantly reduced through the proposed method. For this purpose, a database consisting of thousands of simulations will be created and expanded to correlate with additional scenarios and realistic experimental data, which will further enhance the predictive capabilities of this tool.