Characterization of Flow Domains and Fluid Flows by Model- and Simulation-based Flow MRI (CFD-MRI)

在超滤过程中，有机或无机物质的沉积会导致膜上或膜内的污垢，从而导致生产力的损失，并增加维护和操作成本。这阻碍了这项有前途的水处理技术的广泛应用。目前还没有描述膜孔隙形态变化和膜上膜层形成过程的模型。此外，目前还没有对多孔介质中的过滤过程进行测量。磁共振成像（MRI）和计算机断层扫描（CT）等成像技术在空间分辨率和所需的图像对比度方面受到限制。本提案的主要目标是通过开发一种全新的方法来缩小这一差距，即基于模型和仿真的流动mri （CFD-MRI）。为此，使用流动模型来排除非物理测量结果，消除测量伪影并提高空间分辨率，从而更准确地描述膜壁上的流动和生物膜形态。CFD-MRI是通过制定和有效地解决拓扑优化问题来实现的，该拓扑优化问题最大限度地减少了测量与仿真结果的误差。为此，采用了梯度算法，核算法是对最近发现的伴随晶格玻尔兹曼方法（ALBM）的进一步发展。该方法的优点一方面是对流动的底层介观建模，从而可以直接计算颗粒分布并与测量的MRI值相关联，另一方面是并行算法的效率。最后，将该方法应用于过滤过程中，计算出膜内的渗透率分布和流体流动动力学。在项目结束时，获得了复杂几何形状流体流动动力学的表征工具，这也有望为其他应用提供新的发现。

In ultra filtration processes the deposition of organic or inorganic substances leads to fouling on or in a membrane and therefore to a loss in productivity and increase in maintenance and operating costs. This prevents a wider application of this promising technology for water treatment. There are still models missing for the description of the fouling process which describe morphological changes of the porosity in the membrane as well as the forming of a layer on the membrane. Furthermore, at the moment there are no measurements available from the filtration process in the porous media. Imaging techniques such as magnetic resonance imaging (MRI) and computer tomography (CT) are limited in terms of spatial resolution and required image contrast. The main objective of this pro-posal is to close this gap by developing a fundamentally new method, the model- and simulation-based flow-MRI (CFD-MRI). For this purpose, a flow model is used to exclude unphysical meas-urement results, to eliminate measurement artifacts and to increase the spatial resolution, which finally leads to a more accurate description of flow and biofilm morphology on the membrane walls.The CFD-MRI is realized by formulating and efficiently solving a topology optimization problem, which minimizes the measurement to simulation result error. For it, gradient-based algorithms are used those kernel is a further development of the recently discovered Adjoint Lattice Boltzmann Methods (ALBM). Advantages of this approach are on the one hand, the underlying mesoscopic modeling of the flow, so that the particle distributions can be directly calculated and correlated with the measured MRI values, and, on the other hand, the efficiency of the parallel algorithm. Finally, the method is applied to the filtration process, which enables to compute the distribution of the per-meability and the fluid flow dynamics inside the membrane. At the end of the project, a characteri-zation tool for fluid flow dynamics in complex geometries is obtained, which also promises to pro-vide new findings for other applications.