The identification of microscale oil structures formed from deposited oil droplets on the filter front face of a coalescence filter medium is essential to understand the initial state of the coalescence filtration process. Using µ-CT imaging and a deep learning tool for segmentation, this work presents a novel approach to visualize and identify deposited oil structures as oil droplets on fibers or oil sails between adjacent fibers of different sizes, shapes and orientations. Furthermore, the local and global porosity, saturation and fiber ratios of different fiber material of the oleophilic filter medium was compared and evaluated. Especially the local and global porosity of the filter material showed great accordance. Local and global saturation as well as the fiber ratios on local and global scale had noticeable differences which can mainly be attributed to the small field of view of the µ-CT scan (350 µm on 250 µm) or the minimal resolution of approximately 1 µm. Finally, fiber diameters of the investigated filter material were analyzed, showing a good agreement with the manufacturer’s specifications. The analytical approach to visualize and analyze the deposited oil structures was the main emphasis of this work.

The exhaust from combustion engines contains particulate matter (PM), which poses potential health risks to human lungs. Current emission laws place increasingly strict limitations on both PM and particle number, leading to the necessity of using wall-flow filters to separate out a significant amount of the introduced PM. As this leads to an increase in the filter's loading, it is regenerated continuously or periodically, leading to the rearrangement of individual particulate structures inside the filter channels. Such rearrangement events cause the formation of specific deposition patterns, which affect the filter's pressure drop, its loading capacity and the separation efficiency. In order to derive predictions on the formation of specific deposition patterns, the transient behaviour of individual particle structures needs to be examined. The present work investigates the detachment and transport of particle structures during filter regeneration with three-dimensional surface-resolved simulations using a lattice Boltzmann method. The goal of this work is the determination of relevant key quantities and their interpretation with respect to predictions regarding the resulting deposition patterns. In this context, it is shown that lift forces are not the predominant detachment forces for non-spherical particle structures, and that the stopping distance of such structures is too long to avoid back-end deposition.