Unveiling colloidal transport and deposition: Exploring pore network and random forest models for insights

Abstract

The transport and deposition of colloids in porous media are common phenomena in nature and industry. In this study, a pore network model (PNM) was used to model the geometry, flow field, and colloidal mass transfer of porous media. The finite element method (FEM) was utilized to establish a database of the contact efficiency ηt of the pore-scale collectors, and a random forest model (RFM) was trained. Reasonable prediction of ηt shows that the developed RFM greatly improves the computational efficiency and frees users from repeated numerical simulations. The effects of the colloidal size and seepage velocity on colloidal transport and the surface deposition behavior were studied, the ηt distribution in the PNM was determined, and breakthrough and retention curves were calculated. The results show that the ηt distribution width and mean value increase with the increase in the colloidal particle size and decrease with the increase in the seepage velocity in the range of colloidal particle sizes and seepage velocities considered in this paper. The peak concentration of the breakthrough curve decreases with increasing colloidal particle size and increases with increasing seepage velocity. Surface deposition produces exponential or uniformly shaped retention curves, with retention concentrations increasing with increasing colloidal particle size and decreasing with increasing seepage velocity. The evolution analysis of the retention curve revealed notable detachment and reattachment effects of the colloids in porous media. The results of this study have potential applications in various fields such as environmental remediation, groundwater management, and filtration technology optimization.