Phase-field investigation of non-isothermal solidification coupled with melt flow dynamics

This paper presents a thermodynamically consistent, non-isothermal phase-field model that incorporates Korteweg stress to accurately simulate the coupled dynamics of melt flow and solidification, revealing how thermal capillary effects and viscosity interpolation schemes influence dendritic growth morphology and velocity.

The Gist

Imagine you are watching a pot of chocolate melt and then cool down to form a solid bar. As it cools, it doesn't just turn into a smooth block; it often grows into intricate, tree-like shapes called dendrites. These tiny "trees" determine how strong the final chocolate bar (or a metal engine part) will be.

This paper is about building a better computer simulation to predict exactly how these "chocolate trees" grow when the liquid is moving around.

Here is the breakdown of what the authors did, using simple analogies:

1. The Problem: Missing the "Surface Tension" Push

For years, scientists have used computer models to simulate how liquids freeze. They usually treat the liquid like water flowing in a pipe and the freezing part like a wall.

However, the authors realized these old models were missing a crucial ingredient: Thermal Capillary Stress.

• The Analogy: Imagine a crowd of people (the liquid molecules) trying to move past a growing ice sculpture (the solid).
  • Old Models: They assumed the ice sculpture was just a static wall. The crowd would just flow around it.
  • The Reality: The ice sculpture isn't just a wall; it has a "skin" (surface tension). If one side of the ice is hotter than the other, that "skin" pulls on the crowd, creating a tiny current that pushes the liquid around the ice.
  • The Mistake: Previous computer models ignored this "skin pull." They forgot that temperature differences at the edge of the ice actually push the liquid, which in turn changes how the ice grows.

2. The Solution: A "Thermodynamically Consistent" Model

The authors built a new, more honest computer model. They added the missing "skin pull" (which they call Korteweg stress) into the equations.

• The Analogy: Think of it like upgrading a video game physics engine.
  • Old Game: If you drop a ball, it falls straight down.
  • New Game: If you drop a ball near a fan, the air pushes it sideways. The new model accounts for the "wind" created by the temperature differences at the freezing edge.

3. What They Discovered

When they ran their new, more accurate simulations, they found some surprising things:

• The "Self-Generated" Current: Even if you don't stir the liquid, the freezing process itself creates tiny currents near the edge of the ice because of those temperature differences.
• Slower Growth: These tiny currents actually slow down the growth of the "ice trees" slightly. It's like trying to run up a hill while a gentle breeze is blowing against your face; you don't get as far as you would in still air.
• The "Wind Tunnel" Effect: When they added an external force (like a fan blowing on the chocolate), the trees grew unevenly.
  • The side facing the wind (upstream) grew faster because the wind brought fresh, cold liquid to it.
  • The side with the wind at its back (downstream) grew slower because the wind carried away the cold, leaving warmer liquid behind.
  • Result: The "tree" became lopsided, leaning away from the wind.

4. The "Viscosity" Trick: How to Stop the Liquid

One of the biggest headaches in these simulations is telling the computer: "Hey, the liquid stops moving once it turns into solid."

• The Problem: In a computer, the boundary between liquid and solid isn't a sharp line; it's a fuzzy zone. If you just tell the computer "make the solid very thick (viscous)," the liquid sometimes still slips through the fuzziness, like water leaking through a wet sponge.
• The Fix: The authors tested two different ways to tell the computer how to thicken the material.
  • Method A (Direct): "Add up the thickness." This failed; the liquid slipped.
  • Method B (Inverse): "Divide by the thickness." This worked perfectly. It's like saying, "If the material is 100% solid, the resistance to flow is infinite." This method successfully stopped the liquid dead in its tracks at the solid boundary, just like real life.

Why Does This Matter?

This might sound like abstract math, but it has real-world consequences.

• Better Metals: When we cast car parts or airplane engines, we want the metal to cool evenly. If the "ice trees" grow unevenly (due to currents we didn't account for), the metal can have weak spots that might crack later.
• 3D Printing: In 3D printing metals, the material melts and freezes in milliseconds. Understanding these tiny currents helps engineers print stronger, more reliable parts.

In a nutshell: The authors fixed a hole in the computer models of freezing liquids. They added the missing "push" caused by temperature differences, which allowed them to predict exactly how freezing shapes will look and grow, especially when the liquid is moving. It's like upgrading from a black-and-white sketch to a high-definition, physics-accurate movie of how nature freezes.

Technical Summary

Here is a detailed technical summary of the paper "Phase-field investigation of non-isothermal solidification coupled with melt flow dynamics" by Oyedeji et al.

1. Problem Statement

Solidification processes, such as casting and additive manufacturing, involve complex interactions between phase transitions, heat/solute transfer, and melt flow. These interactions dictate the resulting microstructure (e.g., dendritic growth) and material properties.

• The Gap: While Phase-Field (PF) models coupled with Navier-Stokes equations are standard for simulating these dynamics, most existing models suffer from a critical thermodynamic inconsistency. They frequently neglect the Korteweg stress (capillary stress) in the momentum equation.
• The Consequence: Neglecting this stress term, which arises from the coupling between the phase field and the velocity field, leads to models that violate non-equilibrium thermodynamics. Specifically, it prevents the model from capturing thermal capillary effects (flow driven by temperature gradients at the interface) and capillarity-driven melt flow, which are crucial under non-isothermal conditions.

2. Methodology

The authors developed a thermodynamically consistent, non-isothermal phase-field model that explicitly couples the phase-field order parameter with the incompressible Navier-Stokes equations.

• Thermodynamic Framework:
  • The model is derived using non-equilibrium thermodynamics (Onsager relations) based on entropy and free energy functionals.
  • It introduces cross-coupling terms between the phase-field order parameter ($\phi$) and the internal energy density ($e$).
  • Key Innovation: The momentum equation explicitly includes the divergence of the Korteweg stress tensor ($\nabla \cdot \sigma_K$). This term accounts for interface tension effects and thermal capillary effects induced by temperature gradients.
• Governing Equations:
  • Phase-Field: An Allen-Cahn type equation governing the evolution of $\phi$.
  • Energy: A conservation equation for internal energy including heat conduction and viscous dissipation.
  • Momentum: The Navier-Stokes equation augmented with the Korteweg stress term ($\nabla \cdot \sigma_K$) and body forces.
  • Viscosity: A variable viscosity model is used to enforce the no-slip boundary condition at the solid-liquid interface, where viscosity diverges in the solid phase.
• Numerical Implementation:
  • Simulations were performed for 2D dendritic growth.
  • The study compared simulations with and without the thermal capillary term ($\nabla \cdot \sigma_K$).
  • The study also investigated forced convection (imposed inlet flow) and the impact of different viscosity interpolation schemes (direct vs. inverse) on boundary condition enforcement.

3. Key Contributions

1. Thermodynamic Consistency: The paper provides a rigorous derivation of a phase-field model that includes the Korteweg stress, ensuring consistency with non-equilibrium thermodynamics.
2. Explicit Inclusion of Thermal Capillarity: It demonstrates that the divergence of the Korteweg stress acts as a driving force for melt flow near the interface, a mechanism often omitted in previous literature.
3. Viscosity Interpolation Analysis: The authors provide a critical numerical comparison of viscosity interpolation schemes, proving that inverse interpolation is superior to direct interpolation for enforcing the no-slip boundary condition in high-viscosity-ratio solidification problems.

4. Key Results

A. Model Verification

• The model was validated against analytical solutions for planar interface evolution and the classical Ivantsov solution for dendritic growth.
• Results showed excellent agreement (< 2% error) with theoretical predictions, confirming the model's accuracy in capturing diffusion-controlled solidification dynamics.

B. Thermal Capillary Effects (Natural Convection)

• Induced Flow: When the thermal capillary term ($\nabla \cdot \sigma_K$) was included, melt flow was spontaneously generated around the solid-liquid interface due to temperature gradients, even without external forcing.
• Impact on Growth: This induced flow slightly reduced the dendrite tip velocity compared to simulations without the term.
• Morphology: The reduced velocity resulted in shorter dendrite arms. In scenarios with imposed temperature gradients, the flow further influenced the asymmetry of growth rates between different dendrite tips.

C. Forced Convection

• Simulations with an imposed inlet flow ($u_{in} = 0.4$) demonstrated significant asymmetry in dendritic growth.
• Upstream vs. Downstream: The dendrite tip on the upstream side (west) experienced accelerated growth due to a steeper temperature gradient, while the downstream tip (east) was suppressed.
• Directionality: The north and south tips remained largely symmetric and unaffected, highlighting the directional nature of forced flow influence.

D. Viscosity Interpolation

• Direct Interpolation: Using linear interpolation for viscosity ($\nu(\phi)$) led to significant deviations from analytical sharp-interface solutions, particularly at high viscosity ratios ($\nu_s/\nu_l = 1000$). It failed to strictly enforce the no-slip condition.
• Inverse Interpolation: Using inverse interpolation ($1/\nu(\phi)$) yielded velocity profiles that closely matched analytical sharp-interface solutions across both solid and liquid domains, regardless of the interface thickness or viscosity ratio. This confirms its necessity for accurate melt flow modeling.

5. Significance

This work addresses a fundamental flaw in many existing phase-field simulations of solidification. By incorporating the Korteweg stress, the model:

• Corrects Thermodynamic Inconsistencies: It aligns the simulation framework with the laws of non-equilibrium thermodynamics.
• Reveals Hidden Physics: It uncovers the role of thermal capillarity in driving melt flow, which can alter dendrite morphology and growth kinetics in ways previously unaccounted for.
• Improves Numerical Robustness: The findings on viscosity interpolation provide a clear guideline for researchers to avoid numerical artifacts when simulating high-viscosity-ratio systems (typical of metal solidification).

Overall, the paper establishes a more rigorous and accurate framework for predicting microstructural evolution in manufacturing processes involving coupled solidification and fluid flow.