CFD-Based Quasi-3D Reduced-Order-Model for Flow Field Simulation in Microchannel Reactors

Flow fields play a crucial role in microchannel reactor performance. They consist of distribution zones, which distribute the reactants from the inlet manifold to the reaction channels and collect them again for the outlet manifold, as well as the reaction channels themselves. To accelerate the design of such flow fields, Computational Fluid Dynamics (CFD) is used to predict quantities such as flow distribution between channels and global pressure drop. However, full 3D simulations usually require tens of millions of computational cells, resulting in significant computational costs. To address this, we propose a quasi-3D reduced-order model that leverages the geometric thinness of the height of such structures relative to the in-plane dimensions. By integrating the incompressible Navier-Stokes equations across the reactor's height and applying appropriate closure models, our approach incorporates the local height information while requiring only a 2D mesh. The proposed model is implemented within the OpenFOAM framework and validated against full 3D simulations of representative geometric features. Results show a global pressure drop error below 10%, with predicted velocity profiles in height direction matching the full 3D benchmarks with an RMS error between 0.01 m/s and 0.2 m/s. ... mehrTo demonstrate the practical application of this framework, it is applied to a geometry of a PEM fuel cell. An automated meshing workflow is developed to generate the 2D mesh with associated height function for the quasi-3D model directly from the provided STL files using ray tracing. This reduced-order model lowers total computation effort by a factor of approximately 100 compared to its 3D counterpart while maintaining comparable accuracy. Figure 1 illustrates the complete workflow. In the future, the framework can be extended further by other physical effects like temperature or species transport. Moreover, its computational efficiency makes it well-suited for integration into a workflow for geometry optimization. This makes this framework a scalable tool for the design of flow fields.