Coupled clogging and colloid retention mechanisms in porous media: Insights from Pore-Network modeling

Authors: LIN, DANTONG - Lanzhou University; TANG, MINPENG - Tsinghua University; ZHANG, BAOQING - Lanzhou University; ZHANG, XINGHAO - Tsinghua University; Bradford, Scott; HU, LIMING - Tsinghua University

Submitted to: Separation and Purification Technology
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 2/10/2025
Publication Date: 2/11/2025
Citation: Lin, D., Tang, M., Zhang, B., Zhang, X., Bradford, S.A., Hu, L. 2025. Coupled clogging and colloid retention mechanisms in porous media: Insights from Pore-Network modeling. Separation and Purification Technology. 363(1). Article 132055. https://doi.org/10.1016/j.seppur.2025.132055.
DOI: https://doi.org/10.1016/j.seppur.2025.132055

Interpretive Summary: Attachment of colloids such as pathogenic microorganisms, clays, organic matter, and air bubbles in soils and aquifer materials depends on the rate of mass transfer to solid surfaces. Previous approaches to predict mass transfer have not considered the full range of colloid densities and the complexities of the soil pore space geometry. This paper overcomes these limitations by using a combination of neutral network and pore-network models. Improved mass transfer predictions are shown for air bubbles that are used in the remediation of contaminants. Results of this study will be of interest to scientists, engineers, and public health professions that are concerned with the fate of pathogens, colloid associated contaminants, and the use of air bubbles in remediation.

Technical Abstract: Colloid transport and retention play a crucial role in environmental processes, particularly in groundwater contamination and remediation technologies. The collector efficiency is a key parameter for predicting colloid retention. However, traditional models like colloid filtration theory (CFT) are not suitable for colloids with densities lighter than water. In this study, we propose a method to predict colloid collector efficiency across density ranges using pore network models (PNMs) combined neural network models (NNMs) at different scales. A comprehensive database for pore throat collector efficiency (.t) is created, and a neural network model (NNM) is trained to predict .t for colloids with densities from 0 to 2.5 g/cm³. PNMs are then built to upscale colloid collection from pore-scale to macroscale by connecting them to real porous media parameters using the discrete element method (DEM) as a bridge. Using micro-nano bubbles (MNBs) as an example, we derive a formula to calculate bubble density and analyze its impact on macroscopic collector efficiency ('M). Finally, an upscaled NNM is trained to predict macroscale 'M from macroscopic parameters for boarder application. The results demonstrate the effectiveness of our NNM approach, with high R² values (>0.97) from training, for both 't and 'M. There is strong agreement between the predicted macroscopic 'M from the proposed method and traditional CFT for colloids with a density greater than 1 g/cm³, while also effectively mitigating the unrealistic 'M predictions that can occasionally arise with CFT. The analysis reveals that bubble density significantly impacts collector efficiency, particularly for larger bubbles and low-flow conditions, where bubble buoyancy becomes more influential. The proposed method provides a valuable tool for accurately predicting the collector efficiency of colloids with densities smaller than water, offering reliable insights into colloid behavior in various environmental conditions. This approach improves the design of processes such as filtration, water treatment, and pollutant removal, making it highly applicable to environmental engineering challenges.