🔬 physics

Exploring Particle Clogging Mechanisms through Deterministic Pore Network Modeling

This study develops and validates a deterministic Pore Network Modeling framework to simulate particle clogging dynamics in porous media, revealing how particle size and conductance ratios influence permeability decline while identifying current limitations and proposing future enhancements for applications in groundwater remediation and filtration.

Original authors: Qianjing Tang, Amir Raoof

Published 2026-07-07

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Qianjing Tang, Amir Raoof

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a porous material, like a sponge or a layer of sand, not as a solid block, but as a giant, intricate city of tiny tunnels. Some tunnels are wide boulevards, while others are narrow alleyways. This is what scientists call a "porous medium."

This paper is about building a digital twin of this city to watch what happens when tiny travelers (particles) try to move through it. Sometimes, these travelers get stuck, blocking the tunnels. This is called "clogging." The authors, Qianjing Tang and Amir Raoof from Utrecht University, created a computer model to predict exactly how and when these blockages happen and how they change the flow of water through the system.

Here is a breakdown of their findings using simple analogies:

1. The Digital City (The Model)

Instead of trying to simulate every single drop of water and every single particle in a massive, real-world sponge (which would take a supercomputer forever), the authors simplified the sponge into a network of connected pipes.

The Pipes: These represent the "throats" (narrow connections) and "pores" (wider chambers) in the material.
The Traffic: Water flows through these pipes, carrying tiny particles along for the ride.
The Goal: They wanted to see how the traffic jams (clogs) form and how they slow down the entire city's traffic flow (permeability).

2. The "Traffic Jam" Rule (Deterministic Clogging)

In previous models, scientists often guessed when a traffic jam would happen using a roll of the dice (randomness). This paper introduces a strict rulebook (a deterministic criterion).

Think of it like a toll booth. The model calculates exactly how many cars (particles) need to pass through a specific narrow tunnel before it gets completely blocked. It's not a guess; it's a math-based count. Once that number is reached, the tunnel is considered "clogged."

3. The "Filter Cake" Effect

Here is a clever part of their discovery: Even when a tunnel is clogged, it doesn't necessarily stop all traffic.

The Analogy: Imagine a highway lane blocked by a pile of sand. Cars can't drive through the sand, but they can still drive around it if there is space, or the sand itself might be porous enough to let a few cars squeeze through slowly.
The Finding: The model shows that when a tunnel gets clogged, the particles form a "filter cake." Water can still flow through this cake, but the particles get stuck. This means the tunnel isn't completely dead; it just becomes a slow, filtered lane.

4. The Size Matters Game

The researchers tested what happens when the "travelers" (particles) are different sizes.

The Small Travelers (1-2 microns): These are like tiny ants. They can slip through many narrow alleyways. When they get stuck, they clog the small paths first, but the big highways stay open. The traffic slows down gradually, like a slow leak.
The Large Travelers (4-5 microns): These are like delivery trucks. They get stuck very quickly in the narrow alleyways. Because they block the main routes so fast, the entire city grid can get cut off into isolated islands. The traffic stops much faster, and the "clog" stays right at the entrance (the inlet) rather than spreading deep into the city.

Key Insight: Bigger particles clog the system faster, but they actually cause a less dramatic drop in total flow compared to smaller particles in some scenarios because they block the entrance so quickly that the rest of the system doesn't get a chance to get messy. It's a complex dance between how fast they block and how much they reduce the flow.

5. The "Traffic Flow" Changes

As the clogging happens, the water doesn't just slow down evenly. It gets chaotic.

The Analogy: Imagine a river. If you put a few rocks in it, the water rushes faster around the rocks. If you put a big dam in, the water has to find new paths.
The Finding: As the tunnels clog, the water speeds up in the few remaining open tunnels and slows down in the clogged ones. The difference between the fastest and slowest paths (called the "Coefficient of Variation") gets huge. The flow becomes very uneven and unpredictable.

6. What the Model Can't Do Yet (Limitations)

The authors are honest about what their digital city is missing:

The "Sticky" Factor: The model assumes particles only get stuck when they physically block the pipe. In reality, particles might stick to the walls of the pipe before they block it, like dust on a shelf. The model doesn't account for this "pre-clogging" stickiness yet.
The "Mixed Bag" Problem: The model currently only handles one size of particle at a time (like only red cars). Real life often has a mix of big trucks and small cars (bimodal mixtures), which the model cannot simulate yet.

Summary

This paper presents a new, rule-based computer model that acts like a traffic simulator for microscopic sponges. It shows that:

Clogging starts small (in narrow tunnels) and spreads.
Bigger particles clog faster but create different flow patterns than smaller ones.
Clogged tunnels aren't dead; they act as filters, letting water through but trapping particles.
The flow becomes chaotic as the system clogs, with water rushing through the few remaining open paths.

The authors hope this tool will help engineers design better filters, clean up groundwater, or improve oil recovery by predicting exactly when and how their systems will get clogged.

Technical Summary: Exploring Particle Clogging Mechanisms through Deterministic Pore Network Modeling

Problem Statement
Particle clogging in porous media is a critical process that alters permeability and fluid flow, impacting applications ranging from groundwater remediation and filtration system design to geo-energy operations. While direct numerical simulations (DNS) and lattice Boltzmann methods (LBM) offer high-resolution insights, they are computationally restricted to small domains that may not represent a Representative Elementary Volume (REV). Pore Network Modeling (PNM) provides a computationally efficient alternative for meso-scale analysis but often relies on stochastic, probability-based criteria for determining when a pore throat clogs. These stochastic approaches introduce uncertainty into predictions. Furthermore, existing models often struggle to accurately capture the dynamic interplay between fluid flow and particle transport once clogging begins, particularly regarding the separation of fluid flux from particle flux in partially clogged regions.

Methodology
This study utilizes the PoreFlow framework, a comprehensive PNM tool developed by Raoof et al. (2013), to simulate particle clogging dynamics. The methodology is built on three core components:

Deterministic Clogging Criterion: Instead of relying on random probability distributions, the model employs a deterministic criterion based on the critical particle count, $N^*$ , proposed by Wyss et al. (2006) and refined by Dersoir (2015). $N^*$ defines the number of particles required to clog a specific throat. The model calculates $N^*$ dynamically for each throat based on:
- Attachment potential: Estimated via DLVO theory and influenced by ionic strength (e.g., NaCl concentration).
- Geometric factors: Specifically the ratio of throat width ( $W$ ) to particle diameter ( $D$ ), following a power-law relationship.
- Flow rate: A threshold effect is incorporated where flow rates above a critical value ( $Q_c$ ) influence $N^*$ .
- The model assumes that once the cumulative particle count passing through a throat exceeds $N^*$ , the throat transitions to a "clogged state."
Fluid-Particle Decoupling (Self-Filtration): Upon clogging, the model separates fluid flow from particle flux. While fluid continues to flow through the clogged throat (albeit with reduced conductance), the particle flux is intercepted, effectively reducing the particle concentration in the fluid downstream to zero. This is modeled as a "filter cake" or plug growing within the throat. The length of this plug ( $L_{cl}$ ) is calculated based on the volume of retained particles, and the throat's hydraulic conductance is updated using a series resistance model (combining the open and clogged sections).
Simulation Setup: A 3D pore network (1656 pores, 3082 throats) was generated within a $1000 \times 1000 \times 2000$ $\mu$ m domain. Simulations were run under single-phase flow conditions with a constant inlet pressure (2.3 kPa) and a constant particle concentration at the inlet. The study systematically varied particle sizes (1–5 $\mu$ m) and the clogging conductance ratio (the ratio of conductance in a clogged state vs. an open state) to perform sensitivity analyses.

Key Contributions

Deterministic Framework: The study establishes a transparent, modular, and deterministic criterion for throat clogging, reducing the uncertainty associated with stochastic models while maintaining physical interpretability.
Dynamic Conductance Update: It implements a mechanism where clogged throats remain partially permeable, allowing for the simulation of "filter cake" effects where fluid flow persists even as particle transport is halted.
Sensitivity Analysis: The work provides a systematic investigation into how particle size and the conductance ratio of clogged throats influence the temporal evolution of clogging and the resulting changes in network permeability.

Results

Permeability Evolution: The model successfully captures an exponential decline in permeability consistent with experimental observations. Clogging initiates in narrower throats and propagates to larger ones over time.
Particle Size Effects:
- Larger Particles: Accelerate the clogging process, leading to rapid conductance loss and the formation of "fully clogged" states where network connectivity is significantly disrupted. These scenarios result in clogging concentrated near the inlet (shallower mean clog depth) and a subsequent decline in velocity heterogeneity (Coefficient of Variation, $CV$) as flow redistributes.
- Smaller Particles: Result in "partially clogged" states where conductance declines more gradually. These scenarios maintain higher network connectivity for longer periods and exhibit sustained increases in velocity heterogeneity ($CV$) due to the formation of preferential flow paths.
Conductance Ratio Impact: Higher conductance ratios (indicating that clogged throats still allow significant flow) delay the progression of clogging. This effect is more pronounced in scenarios with rapid clogging potential (e.g., larger particles), suggesting a critical interplay between particle size and the permeability of the clogged plug.
Spatial Distribution: As clogging progresses, the mean depth of clogging decreases (moving closer to the inlet), and the variance in clog depth reduces, indicating a concentration of clogging events near the entry point.

Significance and Limitations
The paper posits that this PNM framework offers a robust tool for predicting clogging behavior in porous media, with broad applicability in groundwater remediation, filtration design, and geo-energy. The deterministic approach enhances the interpretability of pore-scale phenomena and their collective impact on macro-scale properties.

However, the authors modestly acknowledge specific limitations:

The current model does not account for particle attachment to throat walls prior to the full clogging event, potentially underestimating total particle retention.
The model is currently limited to monodispersed particles and single-phase flow, lacking the capability to handle bimodal particle mixtures or multiphase flow conditions (e.g., in the vadose zone).
The study suggests that future improvements should incorporate particle attachment dynamics and probabilistic scenarios for multi-dispersed systems to better reflect real-world complexities.