Analysis of heat transfer and water flow with phase change in saturated porous media by bond-based peridynamics

This paper presents and validates a bond-based peridynamics framework for accurately modeling coupled heat transfer and pressure-driven water flow with phase change in saturated porous media, offering a robust non-local approach to predict phase interfaces and thermodynamic distributions in complex scenarios like soil freezing and thawing.

The Gist

Imagine a sponge that is completely soaked with water. Now, imagine that this sponge is sitting in a freezer, and the water inside is slowly turning into ice. As it freezes, the water expands, the flow of liquid water changes, and the temperature shifts in complex ways. This is the kind of problem scientists face when studying things like frozen ground (permafrost) or how ice forms in soil.

This paper introduces a new way to simulate and predict exactly what happens inside that "sponge" when water freezes and thaws, while also flowing through it. Here is a simple breakdown of their work:

The Problem: The "Tangled Knot" of Math

Traditionally, scientists use math that looks at a material point-by-point, like looking at a single pixel on a screen. This works well for smooth things. But when water turns to ice, things get messy:

• The Boundary Problem: The line between liquid water and solid ice is a moving target. It's like trying to draw a line on a piece of paper that keeps moving and changing shape.
• The "Snap": When water freezes, its properties change instantly. Traditional math struggles with these sudden "jumps" or sharp edges, often causing the computer simulation to crash or give wrong answers.

The Solution: The "Neighborhood Watch" (Peridynamics)

The authors propose using a method called Bond-Based Peridynamics. Instead of looking at a single point in isolation, imagine every tiny particle in the sponge is a person in a neighborhood.

• The Horizon: Each person has a "horizon" (a circle around them). They can only talk to and interact with their neighbors within that circle.
• The Bonds: If two neighbors are close, they are connected by a "bond."
• The Magic: In this model, if a bond breaks (like when ice forms and blocks water flow), the math doesn't crash. The system just stops sending messages across that broken bond. This makes it incredibly good at handling cracks, moving ice fronts, and sudden changes without getting confused.

What They Did: The "Sponge" Experiments

The team built a computer model based on this "neighborhood" idea to track three things happening at once:

1. Heat moving: How cold spreads.
2. Water moving: How liquid flows through the sponge.
3. Phase change: How water turns to ice and back.

They tested their new model in three ways:

1. The 1D Test (The Long Hallway): They simulated a long, thin strip of frozen ground. They compared their results to a known mathematical "gold standard" (an exact solution). Their model matched perfectly, proving it could handle freezing correctly.
2. The Flow Test (The River): They simulated water flowing through the material without freezing. Again, their results matched the known math perfectly.
3. The Complex Test (The Frozen Island): This was the big challenge. They created a 2D simulation of a frozen "island" of ice inside a warmer, water-filled sponge. They compared their results to a very popular, standard method called Finite Element Method (FEM).
   • The Result: Their model matched the standard method when things were calm.
   • The Superpower: When they increased the water pressure to make the water flow very fast, the standard method (FEM) got confused and failed. Their new "neighborhood" model kept working perfectly, handling the high-speed flow and melting ice without breaking a sweat.

Why It Matters (According to the Paper)

The authors explain that this successful simulation is a crucial first step. By accurately tracking how heat and water move together while ice forms and melts, they are building the foundation for a more complex model. This future model could help us understand:

• How permafrost (permanently frozen soil) behaves.
• The phenomenon of frost heave, where freezing ground pushes up and damages roads, buildings, and mines.

In short, the paper presents a new, robust "neighborhood watch" system for math that can handle the messy, moving boundaries of freezing water in soil better than the old methods, especially when the water is moving fast.

Technical Summary

Technical Summary: Analysis of Heat Transfer and Water Flow with Phase Change in Saturated Porous Media by Bond-Based Peridynamics

Problem Statement
The paper addresses the complex physical and mathematical challenges associated with modeling heat and mass transport in saturated porous media where phase change occurs (e.g., freezing and thawing of soils). These processes involve strong physical non-linearities, such as rapid changes in physical and mechanical properties between phases, and geometric discontinuities at inter-phase boundaries. Traditional numerical methods based on local differential formulations, such as the Finite Element Method (FEM), struggle to handle these discontinuities and strong non-linearities robustly, particularly when dealing with evolving interfaces and high gradients. The authors aim to develop a non-local approach capable of accurately predicting the location of phase interfaces and calculating temperature and pressure distributions under pressure-driven water flow conditions.

Methodology
The authors develop a non-local model based on bond-based peridynamics (PD). In this framework, the classical partial differential equations are replaced by integral-differential equations, allowing for the treatment of discontinuities without singularities.

1. Theoretical Formulation:

   • Governing Equations: The study formulates the conservation of mass for water (considering liquid and ice phases) and the conservation of energy (accounting for heat conduction, convection, and latent heat). These are derived from classical local equations (Darcy's law, Fourier's law, and energy balance) and reformulated into a non-local PD context.
   • Phase Change Modeling: The model incorporates a soil freezing characteristic curve (Weibull-type relation) to define liquid water saturation as a function of temperature. Latent heat is treated as a temperature-dependent specific heat capacity term.
   • Bond Classification: The porous medium is discretized into particles connected by "transport bonds" (t-bonds). Bonds are classified based on the phases of the connected particles:
     • Liquid t-bonds: Transfer both heat and water.
     • Interfacial t-bonds: Connect different regions (liquid/solid) and transfer both heat and water.
     • Solid (Impermeable) t-bonds: Connect particles within the solid (ice) region. These are defined as impermeable, transferring only heat, effectively simulating the negligible permeability of frozen soil.
   • Coupling: The model couples pressure-driven water flow with transient heat transfer. The water flux is calculated based on flow potential differences across bonds, and this flux drives convective heat transport.

2. Numerical Implementation:

   • The domain is discretized into a uniform grid of particles.
   • The mass conservation equation is solved to determine the pressure field and water flux vectors.
   • The energy conservation equation is solved to update the temperature field, incorporating both conductive and convective terms derived from the bond interactions.
   • Special attention is paid to the treatment of self-interaction terms (where a particle interacts with itself in the integral formulation) using averaging techniques for nearest neighbors.

Key Contributions

• Novel Coupling: This work presents the first bond-based peridynamic formulation that couples transient heat transfer with phase change and pressure-driven water flow in saturated porous media. Previous PD developments had addressed heat conduction or fluid flow separately, or coupled them without phase change.
• Interface Tracking: The non-local formulation naturally handles the moving interface between liquid and solid phases without requiring explicit mesh reconstruction or interface tracking algorithms typical of local methods.
• Handling High Gradients: The formulation is designed to remain stable under conditions of high pressure gradients and high water velocities, scenarios where local differential methods often suffer from numerical instability or divergence.

Results and Verification
The proposed methodology was verified through three distinct test cases:

1. 1D Heat Transfer with Phase Change: The PD results for a freezing front in a semi-infinite domain were compared against an existing analytical solution. The PD model showed very close agreement in both the position of the phase front and the temperature distribution over time.
2. 1D Heat Transfer with Water Flow: A convective-conductive problem was solved and compared against an analytical solution. The results demonstrated accurate implementation of convective heat transport.
3. 2D Coupled Problem (Freezing Inclusion): A 2D problem involving the thawing of a frozen inclusion in a porous medium under pressure-driven flow was simulated. The PD results were compared against Finite Element Method (FEM) simulations (using Comsol Multiphysics).
   • Temperature: The PD and FEM results showed good agreement for temperature distributions.
   • Pressure: The PD model successfully tracked pressure changes as the frozen inclusion melted. Notably, the PD model captured extreme negative pressure values within the frozen inclusion (due to density changes under nearly constant volume conditions) which the FEM model failed to track, leading to divergence in the FEM solution at higher pressure gradients.
   • High Velocity Flow: A fourth test case involving a "crack" in a frozen body subjected to high-velocity warm water flow demonstrated that the PD model could handle convection-dominated heat transfer and rapid phase change where FEM solutions became unstable or divergent.

Significance and Claims
The paper claims that the developed bond-based peridynamic approach provides an accurate and robust alternative to local methods for modeling coupled thermo-hydraulic processes with phase change. The verification exercises demonstrate that the methodology correctly captures the physics of heat and water transport in the presence of evolving discontinuities.

The authors position this work as a critical step toward a full thermo-hydro-mechanical (THM) model. Such a model would enable the description of complex hydrological behaviors in permafrost soils and phenomena like frost heave, which are significant challenges in civil and mining engineering. The paper modestly notes that while the current model handles the hydro-thermal coupling effectively, future work will be required to incorporate mechanical deformation and the specific physical implications of the extreme pressure gradients observed in the frozen zones.