Micro-Xray tomography based pore-scale simulation of additively manufactured wicks

Abstract

An evaporator wick plays a crucial role in high heat flux cooling systems utilized in miniaturized electronic devices, refrigeration, thermoelectric power, and space systems, among others. To design an optimal wick, it is necessary to have a fundamental understanding of the heat and mass transfer at the pore scale, including effective thermal conductivity, porosity, and permeability. An emerging technique, additive manufacturing (AM), provides an innovative manufacturing approach for creating the desired non-uniform wick structures. While these non- uniform pore geometries can tailor the local heat and mass transfer of the wick, traditional volume-average study approaches are challenging to accurately predict key characteristic properties of the wick, such as thermal conductivity and permeability. Conversely, experimental procedures can be expensive and time-consuming. A micro-Xray computed tomography (μXCT) offers an effective solution by enabling accurate measurements of the micro-scale pore structures and generation of highly detailed volume meshes. These meshes can be utilized for accurate prediction of pore-scale heat and mass transfer of wicks using computational fluid dynamics (CFD). In this study, the AM wicks were prepared using powder bed fusion with carefully controlled process parameters. To develop the pore-scale simulation approach, 2,000 high quality tomographic images were generated using a voxel size of approximately 0.6 μm. Open-source tools were used to examine the optimal workflow from μXCT data to pore-scale simulation. The high-quality tomographic images of AM wicks underwent a series of pre-processing procedures, including image filtering, segmentation, and mesh generation before pore-scale CFD simulation using OpenFOAM. For parameter sensitivity study, a 400x400x400 μm3 volume was cropped from the original tomographic images. A bilateral noise reduction filter was utilized to reduce noise while preserving edges, with the best spatial and range parameters of 30 and 110 respectively. Ostu's method of automatic image thresholding produced the best results, displaying a porosity of 0.29 and an average pore diameter of 70 μm, in line with the experimental findings. OpenFOAM native mesh generator, snappyHexMesh was employed for volume meshing, and extensive parametric studies were conducted to select the optimal castellating and snapping parameters for generating high-quality mesh. A steady-state laminar simulation under a small pressure gradient was performed in x, y, and z directions using the simpleFOAM solver with a single outlet at ambient temperature. Darcy's law was applied for permeability calculation to obtain anisotropic permeability. The predicted permeability of 1.23x10-12 m2 agreed with predicted result obtained using Carman-Kozeny relation. These results provide valuable insights into the tailored heat and mass transfer of the AM wicks, facilitating optimal wick designs and AM process map.