Recovery of liquid aerosols (mists) from industrial processes is typically accomplished through coalescence filtration, employing highly porous nonwoven (fibrous), knitted or foam media which are regarded to potentially provide high collection efficiencies. Highly resolved pore-scale computational fluid dynamics (CFD) analysis of mist filtration processes is increasingly becoming an important tool for design and optimization of such filter media.
A key to efficient application-specific optimization of filter media is the ability to generate CFD-suitable virtual filter geometries with controllable geometric parameters including solidity, fibre diameters, morphology, etc. - yet, a review of the literature suggests that the current designs are heavily reliant on computed tomography (CT) scans of available filter media for accurate representation of the pore-scale structures in a computational simulation. In the present study, a novel methodology is presented for generating realistic virtual nonwoven (fibrous) and foam filter geometries with parametric customizability, using open-source tools including Python, OpenFOAM libraries, Gmsh and Blender.
Further, a methodology for the generation of a computational mesh suitable for multiphase CFD at the pore-scale is delineated for the two types of filter media generated using the present technique, viz. nonwoven and foam, using open-source tools available within the OpenFOAM framework. The proposed methodology for the generation of virtual filter media and computational mesh is validated by qualitative comparison against with images from electron-microscopy (SEM) scans of real filters as well as comparison of the single-phase pressure drops predicted from CFD simulations using the generated fibrous and foam media with different solidities, fibre (or strand of foam) diameters, filter thicknesses, against the literature.
The excellent agreement between the predicted pressure drops and the literature and its consistency over the several different geometric conditions considered for comparison reaffirms the validity of the proposed methodology for efficient virtual filter media development, which can eventually lead to enhanced parametric optimization capabilities and reduced design costs and lead times.....
Session: G10 - Short Oral + Poster Presentations
Day: 12 October 2016
Time: 14:45 - 16:45 h