Reconstruction of particle positions and orientations from 3D MRI images of non-spherical particle packings

Packed bed reactors of non-spherical particles are widely used in chemical industry with the aim to obtain a high active surface area and achieve a homogeneous flow. Despite this, little is known about the arrangement of particles within the bed and the influence of this arrangement on the fluid flow distribution. Magnetic Resonance Imaging (MRI) is a non-invasive tomographic imaging technique that allows 3D visualisation of the packing and flow structure. However, the individual particle information is not obtained using MRI. In this work we investigate different particle detection methods to retrieve the particle position and orientation from MRI images. Results show the successful reconstruction of random packing structures of various non-spherical particle shapes: ellipsoid, spherocylinder, cylinder and cube. The applicability of each method in relation to the particle shape, as well as strengths and drawbacks of each particle detection method are discussed. This paper shows the ability to reconstruct real packed beds of non-spherical particle shapes from MRI images, which opens several research opportunities in the field of chemical engineering.