Lagrangian modelling of porous media filtration flow

Much research works, which include both experimental and numerical methods, have been carried out in the attempt of garnering a deeper understanding of the hydrodynamics in a granular filter bed, or in systems of porous media (packed beds) mainly. However, it has appeared that the numerical models d...

Full description

Bibliographic Details
Main Author: Chew, Alvin Wei Ze
Other Authors: Law Wing-Keung, Adrian
Format: Final Year Project (FYP)
Language:English
Published: 2015
Subjects:
Online Access:http://hdl.handle.net/10356/64150
Description
Summary:Much research works, which include both experimental and numerical methods, have been carried out in the attempt of garnering a deeper understanding of the hydrodynamics in a granular filter bed, or in systems of porous media (packed beds) mainly. However, it has appeared that the numerical models designed, by far, were not emulating the actual granular filter bed effectively due to the employed tube to particle diameter ratio. It was also concurred that there were little or no studies made on the investigations of the impurity particle interactions in the domain of multiple spherical grains, which would be a more exact representation of the impurities’ attachment to the filter grains in the actual bed. As such, there were 2 objectives in this study: (a) to design a robust numerical domain which could emulate the actual filter bed to a good extent, and (b) to attempt to insert a lagrangian particle into the designed numerical model, which comprised of multiple spherical objects, for investigating the particle’s behaviour. For the first objective, there were 4 numerical domains created with the employment of the OpenFOAM CFD software, for which domains 1 and 2 effectively represented an actual granular clean filter bed due to the 5-12% deviation between the simulated and theoretical head loss values. For domains 3 and 4, there were inaccuracies in the simulated head loss values which could be ascribed to the low quality meshes. For the second objective, there was a need to modify the underlying C++ coding in order to hybridize both the icoFoam solver and solidParticleCloud class, which was not achievable within the timeframe of this project.