Measurement of constriction size distributions using three grain-scale methods

T Shire and C O'Sullivan and HF Taylor and WW Sim, SCOUR AND EROSION, 1067-1073 (2016).

The grain-scale justification for empirical rules for granular filters has largely been based on simplified models of sphere packings. The development of Discrete Element Modelling (DEM) and micro Computed Tomography (mu CT) enables a more scientific appraisal of the void space within granular filter materials. The constrictions or pore throats that govern the filter's performance can now be directly measured. However, definitive partitioning of the void space is not possible for realistic grain packings and so multiple methods, with differing theoretical bases, have been proposed to identify constrictions. This contribution compares three such methods, each of which results in a Constriction Size Distribution (CSD). The methods considered are the triangulation based weighted Delaunay method (Reboul et al. 2010), a contact based method (O'Sullivan et al. 2015) and an image analysis method based on watershed segmentation (Taylor et al. 2015). Each model, along with its relative advantages is introduced. Then CSDs resulting from applying each model to the same virtual filter samples created using DEM are presented. It is shown that there is reasonable agreement despite the different basis of each approach. A consideration of empirical filter rules is carried out by normalising the full CSDs by the characteristic diameters typi-cally used to represent the retention capacity of granular filters in design. It is shown that similar CSD curves are ob- tained for different Particle Size Distributions (PSDs) when curves are normalised by characteristic diameters, irrespec-tive of the method used to identify the constrictions. This gives fundamental support to filter rules using characteristic particle diameters to represent the filtration capability of granular filters.

Return to Publications page