A HOLISTIC FRAMEWORK FOR HYBRID MODELLING OF SOLID-LIQUID FLOWS

固液悬浮液在管道和容器中的运动是一个普遍的复杂问题，在商业上具有挑战性，在技术上也很重要。工业应用众多，例如化学品、消费品、食品、药品、石油、采矿、河流工程、建筑、发电、生物技术和生物医学。尽管市场如此之大，但由于严重缺乏对这些流动的基本理解，工业实践和过程既不有效也不理想。这种流动涉及范围广泛的复杂现象，因为流动管道通常从微米尺度到厘米尺度不等，容器从毫升尺度到立方米尺度不等。流动可能是紊流或粘性的，而载体流体可能表现出复杂的非牛顿流变性。颗粒以各种形状、大小、密度、体积和表面性质出现，这加剧了问题的复杂性。输送或处理固液悬浮液的工艺设计需要有关颗粒行为的信息，如颗粒轨迹、流线径向迁移、颗粒速度分布和固体分布。然而，在成像固液流动和测量局部流体和固体速度方面存在巨大的实际困难，因为可用的仪器很少适用。实际用途的混合物通常是浓缩的，不透明的，因此不可能可视化流动，颗粒可能变形，易碎或容易聚集。这种复杂的现象目前很难预测。它们阻碍了基础研究和严格的整体建模策略的发展，因此，工作通常遵循零碎的经验方法。本研究将采用多尺度方法来研究固液悬浮液的流动，包括复杂的非牛顿流变性流体和具有复杂性质的颗粒：(i)通过独特而精确的拉格朗日正电子发射粒子跟踪技术进行实验，该技术可以测量不透明体系中的局部三维相速度和相分布；（ii）通过开发和验证新的建模方法来预测这种流动，包括颗粒、流体和壁面之间的详细相互作用。将使用许多先进的建模技术，主要包括离散元法（DEM），计算流体动力学（CFD），光滑粒子流体动力学（SPH），晶格玻尔兹曼方法（LBM）和粗粒分子动力学（CGMD）。然而，这些方法本身都不能有效地为这些复杂的流程建模，因为它们都有优缺点。因此，我们将利用每种技术的优势，以一种有效的混合方式组装这些方法，以产生一个集成的多尺度模块化框架，该框架将在独特且知名的开源代码DL_MESO中免费提供。因此，我们将评估最好的混合方法，并通过将模型混合映射到流动特征中，为这些复杂的流动建模开发一个范例。使用混合建模方法和多尺度方法来包括集中的湍流、非牛顿流变性流体、复杂形状和性质的颗粒，将在这些复杂流动的建模方面产生巨大的飞跃。从中长期来看，这项工作的发现应该会提高英国固液处理技术的竞争力。然而，我们的工业和学术合作伙伴将能够通过参与该项目获得直接利益。

The movement of solid-liquid suspensions in pipes and vessels is a generic complex problem which is commercially challenging and technically important. Industrial applications are numerous, e.g. chemicals, consumer goods, food, pharmaceuticals, oil, mining, river engineering, construction, power generation, biotechnology and biomedical. Despite such large markets, industrial practice and processes are neither efficient nor optimal because of a severe lack of fundamental understanding of these flows. Such flows involve complex phenomena on a wide range of scales as flow conduits generally vary from the micron scale to the centimetre scale, and vessels vary from the millilitre scale to the cubic metre scale. Flows may be turbulent or viscous and the carrier fluid may exhibit complex non-Newtonian rheology. Particles occur in various shapes, sizes, densities, bulk and surface properties which exacerbates the complexity of the problem.The design of processes for conveying or processing solid-liquid suspensions requires information about particle behaviour such as particle trajectory, radial migration across streamlines, particle velocity distribution, and solids distribution. There are, however, huge practical difficulties in imaging solid-liquid flows and measuring local fluid and solid velocities, since little of the available instrumentation is applicable. Mixtures of practical interest are often concentrated and opaque so that flow visualisation is impossible, and particles may be deformable, breakable or prone to aggregation. Such complex phenomena are presently difficult to predict. They have hampered fundamental research and the development of rigorous holistic modelling strategies and, as a result, work has generally followed a piecemeal empirical approach.This proposal will use a multiscale approach to study the flow of solid-liquid suspensions including fluids of complex non-Newtonian rheology and particles with complex properties: (i) experimentally via a unique and accurate Lagrangian technique of positron emission particle tracking, which can measure local 3-D phase velocities as well as phase distribution in opaque systems; and (ii) by developing and validating novel modelling approaches to predict such flows including detailed interactions between particles, fluid and walls. A number of advanced modelling techniques will be used including principally the Discrete Element Method (DEM), Computational Fluid Dynamics (CFD), Smooth Particle Hydrodynamics (SPH), Lattice Boltzmann Method (LBM) and Coarse-Grained Molecular Dynamics (CGMD).None of these methodologies on its own, however, is able to effectively model these complex flows as they all enjoy strengths as well as weaknesses. We will, therefore, exploit the strengths of each technique by assembling these methods in an efficient hybrid fashion to produce an integrated multiscale modular framework to be made available free of charge within the unique and well-known open source code DL_MESO. Thus, we will evaluate the best hybrid approaches and develop a paradigm for modelling these complex flows by mapping the model hybrids against flow characteristics.The use of a hybrid modelling methodology and a multiscale approach to include concentrated turbulent flows, fluids of non-Newtonian rheology, particles of complex shapes and properties will produce a quantum leap advance in the modelling of these complex flows. In the medium to long-term, the findings from this work should improve the competitiveness of the UK solid-liquid processing technologies. Our industrial and academic partners, however, will be able to draw immediate benefits through engagement with the project.