Uncovering bistability phenomena in two-layer Couette flow experiments using nonlocal evolution equations

This paper demonstrates that a nonlinear, nonlocal asymptotic model derived from the Navier-Stokes equations accurately predicts and characterizes the bistability of unimodal and bimodal travelling waves observed in two-layer Couette flow experiments, while also revealing new symmetry-breaking branches and time-periodic orbits through comprehensive bifurcation analysis.

The Gist

The Big Picture: A Dance of Two Fluids

Imagine a long, circular racetrack (like a giant donut). Inside this track, there are two layers of liquid: a thick, heavy layer at the bottom (like honey or water) and a thin, lighter layer floating on top (like oil).

Now, imagine the top lid of this racetrack starts spinning. As it spins, it drags the top layer of liquid along with it. Because the two liquids have different thicknesses and "stickiness" (viscosity), they don't move in perfect harmony. Instead, the boundary between them starts to ripple and wave.

This paper is about understanding how these waves behave, specifically when the top layer is very thin. The researchers built a sophisticated mathematical model to predict these waves and compared it directly to real-life experiments.

The Main Discovery: The "Two-Choice" Mystery

The most exciting thing they found is a phenomenon called bistability.

Think of it like a light switch that has a weird glitch. Usually, if you flip a switch, the light is either ON or OFF. But in this fluid system, if you spin the lid at a specific speed, the fluid doesn't just pick one wave pattern. It can settle into two completely different, stable wave patterns, depending on how you started the race.

• The "Unimodal" Wave: Imagine a single, smooth hill rolling around the track.
• The "Bimodal" Wave: Imagine two hills rolling around the track at the same time.

If you start the experiment with a tiny nudge in one direction, the fluid creates the single-hill wave. If you nudge it the other way, it creates the two-hill wave. Both are stable at the exact same speed. It's as if the fluid is saying, "I can be a single-hill wave or a double-hill wave; it's up to you how I start!"

The researchers' new math model predicted this "two-choice" behavior perfectly, matching the real-world experiments better than previous models did.

The New Twist: The "Broken Symmetry" Wave

While studying the "two-hill" wave, the researchers discovered something even stranger. As they increased the speed of the spinning lid, the two hills in the "bimodal" wave suddenly stopped being identical twins.

Imagine two dancers spinning in a circle, holding hands. At first, they are perfectly balanced. But as the music speeds up, one dancer suddenly gets taller or leans more than the other. They are still dancing together, but the symmetry is broken.

The paper identifies a new type of wave (called Branch 2*) where these two peaks become unequal. This wasn't seen in the old experiments or previous math models, but the new equations predicted it would happen. It's like finding a new dance move that nobody knew existed.

The "Traffic Jam" of Waves

The researchers also looked at what happens when things get really fast. They found that the system can get very complicated, supporting waves with three or even four hills at once.

They mapped out a "traffic map" of these waves:

• Stable Zones: Areas where the waves are calm and predictable.
• Unstable Zones: Areas where the waves start to wobble, shake, or turn into chaotic, time-periodic orbits (like a wave that keeps changing its shape rhythmically instead of staying still).

They found that in certain speed ranges, the fluid could be "tristable"—meaning it could choose between three different stable states (one-hill, two-hill, or a broken two-hill) depending on the initial push.

Why Does This Matter?

You might ask, "Who cares about waves in a donut-shaped tank?"

1. Better Models: The math they used is special because it accounts for the "inertia" (the momentum) of the thick bottom layer. Previous models ignored this, which is why they failed to predict the "two-choice" behavior accurately. This new model is like upgrading from a bicycle to a sports car; it handles the physics much better.
2. Real-World Applications: This kind of fluid behavior happens in many places:
   • Oil and Gas: When oil and water flow together in pipes.
   • Coatings: When painting a car or coating a chip with a thin layer of liquid.
   • Nature: How waves form in the atmosphere or oceans where layers of air or water have different densities.

The Takeaway

This paper is a success story of math meeting reality.

• The Problem: Real experiments showed weird, unpredictable "two-choice" behavior that old math couldn't explain.
• The Solution: The team used a new, more accurate mathematical model that respects the physics of the thick bottom layer.
• The Result: The model didn't just explain the known behavior; it predicted new, hidden behaviors (like the broken-symmetry waves) that scientists can now go out and look for in their labs.

In short, they cracked the code on how these fluid waves choose their shape, revealing that the fluid world is far more complex and full of "hidden choices" than we previously thought.

Technical Summary

1. Problem Statement

The paper investigates the stability and nonlinear dynamics of interfacial long waves in two-layer plane Couette flow. Specifically, it addresses the phenomenon of bistability, where two distinct stable travelling waves (one unimodal and one bimodal) coexist at the same lid velocity. This behavior was previously observed experimentally by Barthelet et al. (1995) in a rotating annular device but lacked a comprehensive quantitative explanation using asymptotic models that retain inertial effects.

The study focuses on a configuration with a thin, more viscous upper layer lying over a thick, denser lower layer (thickness ratio $d = h_2/h_1 = 0.25$). The goal is to reproduce experimental observations, resolve discrepancies in previous quantitative comparisons, and uncover new dynamical features (such as symmetry-breaking branches and higher-wavenumber modes) that were not previously reported.

2. Methodology

The authors employ a combination of asymptotic analysis, numerical continuation, and direct comparison with experimental data.

• Mathematical Model:

  • The study utilizes a nonlinear, nonlocal asymptotic evolution equation derived from the Navier-Stokes equations.
  • The derivation assumes a thin upper film ($d \ll 1$) and matches a lubrication solution in the thin layer with a linearized Navier-Stokes (Orr–Sommerfeld) solution in the thick lower layer.
  • This coupling introduces nonlocality and retains crucial inertial effects from the thick layer, which are essential for accurate predictions at experimentally relevant Reynolds numbers.
  • The governing equation (Eq. 3.1) is a nonlocal PDE for the interface height $H(x,t)$, involving a fourth-order dispersion term and a nonlocal inertial term derived from the function $F''(1; \alpha)$.

• Numerical Approach:

  • Travelling Wave Solutions: Solved using a truncated Fourier series representation. The nonlinear algebraic system for Fourier coefficients and wave speed is solved via Newton iteration.
  • Stability Analysis: Linear stability is determined by solving an eigenvalue problem for perturbations to the travelling waves.
  • Time-Dependent Simulations: The PDE is integrated numerically using a Fourier spectral discretization in space and a fourth-order exponential-time-differencing Runge–Kutta method in time.
  • Basins of Attraction: Computed by varying initial conditions (linear combinations of unimodal and bimodal solutions) to map the threshold separating the two stable states.

• Experimental Comparison:

  • Parameters are matched to the "1d–2" fluid pair from Barthelet et al. (1995) (mineral oil/glycerine-water system).
  • The model is validated against experimental data for various lid velocities ($U/U_L$), focusing on interfacial profiles, wave speeds, and harmonic amplitudes.

3. Key Contributions

The paper makes several significant advancements over previous studies (specifically Kalogirou et al., 2016):

1. Resolution of Quantitative Discrepancies: Previous models showed poor quantitative agreement in the bistability regime. This study resolves this by using a nonlocal model equation specifically derived for a thin upper layer (as opposed to a lower thin layer), leading to remarkable agreement with experimental data.
2. Systematic Bifurcation Analysis: The authors provide explicit bifurcation and stability diagrams for unimodal (Branch 1) and bimodal (Branch 2) travelling waves, delineating their stability boundaries and coexistence regions.
3. Discovery of Symmetry-Breaking Branch: A new travelling-wave branch (Branch 2*) is identified. This branch bifurcates from the bimodal family via a pitchfork bifurcation, breaking the $\pi$-periodic symmetry of the bimodal wave.
4. Mapping of Higher-Wavenumber Modes: The study maps higher-wavenumber travelling-wave families (Branches 3 and 4) and identifies time-periodic orbits arising via Hopf bifurcations.
5. Characterization of Multistability: The paper defines the basins of attraction for coexisting stable states and identifies regimes of tristability (coexistence of Branch 2, 2*, and 3).

4. Key Results

• Bistability Window:

  • The model confirms the existence of a bistability window where unimodal (2$L$-periodic) and bimodal ($L$-periodic) waves coexist.
  • For the experimental parameters ($R=709, \nu=0.01, m=2.76$), this occurs in the range $0.016 \lesssim \Lambda \lesssim 0.036$.
  • The basins of attraction are successfully separated; the threshold $\theta^*$ between the two states shifts as $\Lambda$ increases, with the bimodal state's basin expanding until it destabilizes.

• Symmetry-Breaking (Branch 2*):

  • At $\Lambda \approx 0.036$, the stable bimodal branch (Branch 2) undergoes a pitchfork bifurcation.
  • A new symmetry-breaking branch (2*) emerges, where the two peaks of the bimodal wave become unequal (breaking $\pi$-periodicity).
  • This branch acts as a local or global attractor for a broad range of $\Lambda$ ($0.036 \lesssim \Lambda \lesssim 0.137$).
  • In the range $0.057 \lesssim \Lambda \lesssim 0.066$, the system exhibits tristability, where Branch 2, Branch 2*, and Branch 3 can coexist.

• Higher-Wavenumber and Periodic Dynamics:

  • Branches 3 ($2\pi/3$-periodic) and 4 ($\pi/2$-periodic) are identified.
  • Branch 3 exhibits complex stability windows, becoming stable, unstable (via Hopf), and restabilizing as $\Lambda$ increases.
  • Hopf Bifurcations: The loss of stability in Branch 2* ($\Lambda \approx 0.137$) and Branch 3 leads to time-periodic orbits (limit cycles).
  • In the range $0.080 \lesssim \Lambda \lesssim 0.088$, the system exhibits quasi-periodic dynamics with modulated amplitudes.

• Experimental Validation:

  • Profiles: The model reproduces the steep fronts and pronounced troughs observed in experiments for $U/U_L$ ratios up to 1.77.
  • Harmonics: The evolution of the first three harmonic amplitudes matches experimental traces well, capturing the ordering of mode appearance and the reduction in time to reach steady states as the Reynolds number increases.
  • Quantitative Match: The model successfully predicts the saturated amplitudes and periods, with minor deviations attributed to density stratification effects not fully captured in the simplified model.

5. Significance

This work provides a comprehensive theoretical framework for understanding complex interfacial dynamics in two-layer Couette flow. By resolving the quantitative mismatch between previous models and experiments, it validates the necessity of nonlocal inertial terms for thin upper layers.

The discovery of the symmetry-breaking branch and tristability significantly expands the known dynamical landscape of this system, suggesting that multistability is more common and complex than previously thought. These findings offer concrete guidance for designing future experiments to observe symmetry-breaking and higher-wavenumber modes, bridging the gap between asymptotic theory and physical observation in fluid mechanics.