Self-similar rupture of thin films of power-law fluid

Imagine a very thin layer of liquid, like a tear film on your eye or a coating of paint on a wall. Sometimes, this film isn't perfect; it has tiny weak spots. Over time, these spots get thinner and thinner until the film suddenly snaps, creating a dry hole. This is called "rupture."

This paper is a mathematical investigation into exactly how that snap happens, specifically when the liquid isn't just water (which flows easily) but something thicker or stickier, like honey, ketchup, or polymer solutions. These special liquids are called "power-law fluids."

Here is the breakdown of what the researchers found, using simple analogies:

1. The "Snap" is Predictable (Self-Similarity)

When a thin film breaks, it doesn't just collapse randomly. The researchers found that as the moment of rupture approaches, the shape of the thinning film follows a very specific, repeating pattern.

Think of it like a video of a balloon popping played in slow motion. No matter how big the balloon was to start with, the way the rubber stretches right before it pops looks the same if you zoom in and slow it down. The researchers call this "self-similarity." They figured out the mathematical "recipe" for this shape, but the recipe changes depending on how "thick" or "thin" the liquid is.

2. The "Thick vs. Thin" Liquid Spectrum

The paper focuses on a parameter called $n$ (the power-law exponent), which acts like a dial for the liquid's behavior:

$n = 1$ : The liquid is normal (like water).
$n < 1$ : The liquid is "shear-thinning." It gets thinner and flows easier when you push it (like ketchup or paint).
$n > 1$ : The liquid is "shear-thickening." It gets harder to move when you push it (like a mixture of cornstarch and water).

The researchers wanted to know: Does the "recipe" for the snap change as you turn this dial from ketchup to cornstarch?

3. The "Snake" in the Graph

The team created a giant map (a bifurcation diagram) showing all the possible ways the film could snap for every value of $n$ .

The Main Path: There is one "primary" path that is stable. If you actually run a computer simulation of the film breaking, it always follows this one path. It's like the main highway that all traffic naturally takes.
The Side Paths: There are many other theoretical paths (branches) where the film could snap, but they are unstable.
The Snake: As the researchers turned the dial for $n$ (changing the liquid type), these side paths didn't just disappear. Instead, they wove in and out of the main path, merging and splitting in a complex, snake-like pattern right around the "normal" liquid setting ( $n=1$ ). It's a very tangled knot of possibilities that only the main highway survives through.

4. The "Ghost Zone" Problem

The most difficult part of the study happened when they looked at extreme "shear-thinning" liquids (where $n$ is very close to 0, like a very runny gel).

They discovered that for these liquids, the math creates a "ghost zone."

The Analogy: Imagine trying to draw a map of a coastline. For normal liquids, the coast is smooth. But for these extreme liquids, there is a tiny, invisible strip of land (an "inner region") that is so small it's almost non-existent (exponentially small).
The Problem: Standard computer simulations are like a low-resolution camera; they miss this tiny strip entirely. Because they miss it, the math breaks down, and the computer can't find the solution.
The Fix: The researchers had to invent a new way of looking at the problem. They essentially "zoomed in" mathematically on that tiny ghost zone, stretching it out so their computers could see it. This allowed them to find a countably infinite number of new solutions that were previously hidden.

5. What Actually Happens in Reality?

Even though the math showed thousands of different theoretical ways the film could break (the "snake" branches), the computer simulations of the actual physical process always chose the single primary branch.

It's like a maze with thousands of dead-end paths and one main exit. Theoretically, you could try to walk down any path, but in reality, the physics of the fluid naturally guides it to the one stable exit.

Summary

The paper proves that while thin films of weird, non-Newtonian liquids can theoretically break in a dizzying number of complex ways (forming a "snake" of solutions), nature is picky. It almost always chooses one specific, stable way to break. The researchers also solved a major puzzle: they figured out how to mathematically see the "invisible" tiny zones that appear when the liquid is extremely runny, allowing them to map the entire process accurately.

Technical Summary: Self-similar rupture of thin films of power-law fluid

Problem Formulation
This study investigates the finite-time rupture of thin liquid films composed of power-law fluids, driven by the competition between surface tension and attractive intermolecular (van der Waals) forces. While the self-similar rupture of Newtonian films is well-established, the behavior of non-Newtonian fluids, characterized by a power-law exponent $n$ (where $n < 1$ indicates shear-thinning and $n > 1$ shear-thickening), presents significant analytical and numerical challenges. The governing equation is a fourth-order nonlinear partial differential equation (PDE) derived from lubrication theory, incorporating a disjoining pressure term to model the van der Waals attraction. The primary objective is to compute similarity solutions that describe the film profile as it approaches the rupture time $t_0$ , and to map the existence and stability of these solutions across the parameter space of $n$ .

Methodology
The authors employ a combination of direct numerical simulation, numerical continuation, and asymptotic analysis:

Numerical Simulation of the PDE: The time-dependent thin-film equation is solved using a finite-difference method with an implicit time-stepping scheme (MATLAB's ode15s). To avoid issues with non-integer derivatives inherent in power-law models, the flux is computed directly from the thickness profile using a five-point stencil for derivatives, rather than calculating the fourth spatial derivative of the thickness directly.
Similarity Solution Construction: The PDE is reduced to a system of ordinary differential equations (ODEs) via self-similar scaling. The resulting boundary-value problem (BVP) is solved using the numerical continuation package Auto07p. The authors track solution branches by varying the power-law exponent $n$ , starting from known Newtonian solutions ( $n=1$ ).
Handling Small $n$ Limits: Standard numerical continuation fails for small $n$ ( $n \lesssim 0.2$ ) due to the emergence of an exponentially small inner region where the flux vanishes. To overcome this, the authors introduce a coordinate rescaling based on the flux variable, transforming the domain to capture the rapid changes in the inner layer.
Asymptotic Analysis: For the limit $n \to 0$ , the authors perform a matched asymptotic expansion. This involves an outer region, an inner layer where the flux decays exponentially, and an innermost region. This analysis reveals the structure of the solution and motivates the rescaled numerical approach.

Key Results

Bifurcation Structure: The bifurcation diagram of similarity solutions, indexed by $n$ $n$ , exhibits a highly nontrivial "snaking" structure around $n=1$ $n = 1$ .
- For $n > 1$ , only the primary branch (the one with the largest central thickness $H(0)$ at $n=1$ ) can be continued to arbitrarily large $n$ . All other branches annihilate via fold bifurcations as $n$ increases.
- For $n < 1$ , only the first four branches (those with the largest $H(0)$ at $n=1$ ) continue toward $n \to 0$ . The remaining branches annihilate via fold bifurcations as $n$ decreases, creating a complex merging pattern near $n=1$ .
Attractor Selection: Numerical simulations of the time-dependent PDE consistently converge to the single primary branch of similarity solutions. This branch is the only one that persists in both the extreme shear-thickening ( $n \to \infty$ ) and extreme shear-thinning ( $n \to 0$ ) limits.
Small $n$ Asymptotics: In the limit $n \to 0$ , the similarity solutions possess an exponentially small inner region where the flux becomes negligible. The asymptotic analysis identifies a countably infinite number of similarity solutions in this limit. The leading-order outer problem is a third-order nonlinear system that requires numerical solution, and the matching conditions reveal a specific structure involving an innermost region governed by a third-order nonlinear ODE with no closed-form solution.
Profile Characteristics: Higher-order similarity branches exhibit oscillatory behavior in the similarity variable $\eta$ . These oscillations are more pronounced for smaller $n$ and are linked to the discrete selection of solutions required to satisfy boundary conditions, a phenomenon related to Stokes' phenomenon in the Newtonian limit.

Significance and Claims
The paper claims to extend the computation of similarity solutions from Newtonian to non-Newtonian power-law fluids, revealing a complex bifurcation structure that was previously unexplored. A major contribution is the identification of the "snaking" bifurcation behavior near $n=1$ and the demonstration that only the primary branch acts as the physical attractor for rupture dynamics.

The authors highlight two significant technical challenges addressed:

Derivative Non-existence: The non-existence of higher spatial derivatives in power-law lubrication flows at points of zero pressure gradient is managed by formulating the problem in terms of flux rather than higher-order thickness derivatives.
Exponentially Small Regions: The study uncovers severe numerical difficulties in the extreme shear-thinning limit ( $n \to 0$ ) due to exponentially small inner regions. By deriving the asymptotic structure, the authors provide a rescaled numerical scheme capable of computing solutions in this regime, a capability they suggest is necessary for simulating extreme shear-thinning fluids in general interfacial problems.

The work does not claim to resolve axisymmetric rupture or stability in two dimensions, noting these as open questions, nor does it incorporate inertial effects, which would alter the dominant balance of forces. The study remains focused on the mathematical characterization of self-similar rupture profiles under lubrication theory assumptions.