Linearized instability of Couette flow in stress-power law fluids

This paper investigates the linearized stability of Couette flow in stress-power law fluids with non-monotonic stress-strain rate behavior, demonstrating that flow stability is fundamentally determined by the interplay between boundary conditions and the specific branch of the constitutive curve, where ascending branches yield stable solutions and descending branches lead to instability.

The Gist

Imagine you are trying to push a heavy box across a floor. Usually, the harder you push (the more "stress" you apply), the faster it moves (the more "strain rate" you get). This is a predictable, straight-line relationship.

But what if the floor was made of a strange, magical material? What if, at a certain point, pushing harder actually made the box move slower for a moment, before speeding up again? Or what if, depending on how you started pushing, the box could end up moving at three different speeds even though you were pushing with the exact same force?

This is the world of Stress-Power Law Fluids, the subject of the paper you shared. These are complex fluids (like certain gels, polymers, or biological fluids) that don't behave like water or honey. They have a "non-monotonic" relationship between stress and speed, meaning their behavior is wobbly and unpredictable.

Here is a simple breakdown of what the researchers discovered, using everyday analogies:

1. The Setup: The "Magic" Floor

The scientists studied a classic physics experiment called Couette Flow. Imagine two parallel plates with a fluid in between.

• Scenario A: You move the top plate at a specific speed and see what happens to the fluid.
• Scenario B: You push the top plate with a specific force and see how fast it moves.

In normal fluids (like water), both scenarios give you one clear, predictable result. But in these "magic" fluids, the math gets messy.

2. The Problem: The "Three-Path" Fork in the Road

The researchers found that when they tried to move the top plate at a specific speed (Scenario A), the fluid didn't always agree on what to do.

• The Fork: For certain speeds, the fluid could theoretically settle into three different states:
  1. A slow, calm flow.
  2. A chaotic, unstable flow.
  3. A fast, smooth flow.

It's like standing at a fork in the road where the sign says, "You can go Left, Right, or Straight," but the map doesn't tell you which one is safe to take.

3. The Discovery: The "Roller Coaster" of Stability

To figure out which of these three states is real and which is just a mathematical ghost, the researchers used Linearized Stability Analysis. Think of this as giving the fluid a tiny, gentle nudge (a "perturbation") to see if it recovers or crashes.

They discovered a pattern that looks like a roller coaster track:

• The Uphill Sides (Ascending Branches): If the fluid is in the "slow" state or the "fast" state (the parts of the curve going up), it is rock solid. If you nudge it, it wobbles for a second and then snaps back to its original path. These are Stable.
• The Downhill Slide (Descending Branch): If the fluid is in that weird middle state where pushing harder makes it slow down, it is unstable. It's like trying to balance a ball on the very top of a hill. The slightest nudge sends it tumbling down into one of the stable states. This state is Unconditionally Unstable.

The Takeaway: Nature hates the "middle" state. If the fluid is forced into that unstable zone, it will immediately try to jump to one of the stable zones.

4. The Twist: How You Control the Experiment Matters

The paper highlights a crucial difference between how you control the experiment:

• Speed Control (Velocity Boundary Conditions): If you tell the machine, "Move at 5 mph," the machine might get confused and offer you three different flow patterns. You have to check which one is stable.
• Force Control (Traction Boundary Conditions): If you tell the machine, "Push with 10 Newtons of force," the fluid has only one choice. It can't be confused.
  • If that 10 Newtons pushes the fluid onto the "uphill" part of the roller coaster, it's stable.
  • If that 10 Newtons pushes it onto the "downhill" part, it's unstable and will crash.

5. Why Does This Matter?

You might ask, "Who cares about a fluid that acts weird?"
These fluids are everywhere in the real world:

• Blood: In tiny vessels, blood can behave strangely under high stress.
• Industrial Slurries: Concrete, paint, and drilling muds often have these non-linear properties.
• Food Science: Ketchup and mayonnaise are famous for being "shear-thickening" or "shear-thinning."

Understanding which states are stable helps engineers design better machines. For example, if you are pumping a complex fluid through a pipe, you don't want the flow to suddenly jump into an unstable state and cause a clog or a burst.

Summary Analogy

Imagine you are driving a car with a broken accelerator pedal.

• Normal Car: Press the pedal 10%, you go 10 mph. Press 20%, you go 20 mph. Simple.
• This "Magic" Fluid Car: Press the pedal 10%, you might go 10 mph, 50 mph, or the car might stall.
  • The researchers found that the "10 mph" and "50 mph" settings are safe (stable).
  • The "stall" setting is dangerous (unstable); if you try to drive there, the car will immediately jerk forward or backward to a safe speed.
  • If you control the car by speed (cruise control), you might accidentally land in the dangerous zone. If you control it by force (how hard you press), you can avoid the danger zone entirely, provided you don't press too hard.

In short: This paper maps out the "safe zones" and "danger zones" for these tricky fluids, showing that their stability depends entirely on whether they are being pushed by a specific speed or a specific force, and that the "middle ground" is always a place to avoid.

Technical Summary

1. Problem Statement

The paper investigates the linearized stability of plane Couette flow for a specific class of non-Newtonian fluids known as stress-power law fluids. These fluids are characterized by non-monotonic stress–strain rate relationships, meaning that a single strain rate can correspond to multiple stress values (and vice versa). This behavior is observed in complex fluids exhibiting phenomena like discontinuous shear thickening and shear banding.

The core problem addresses the ambiguity in steady-state solutions:

• Under velocity boundary conditions, the system may admit one, two, or three distinct steady-state solutions for the same flow configuration.
• Under traction (stress) boundary conditions, the solution is unique.
• The paper seeks to determine which of these mathematically valid states are physically admissible by analyzing their linear stability. It aims to distinguish instabilities arising from the constitutive nonlinearity from those induced by geometry (hence the choice of plane Couette flow over Taylor-Couette flow).

2. Methodology

A. Constitutive Modeling

The authors derive a generalized stress-power law model within the thermodynamic framework of Rajagopal and Srinivasa.

• Framework: Instead of prescribing stress as a function of strain rate ($T = f(D)$), they use an implicit constitutive relation $g(T, D) = 0$. Specifically, they define the symmetric part of the velocity gradient $D$ explicitly in terms of the deviatoric stress $S$.
• Derivation: The model is derived by maximizing the rate of entropy production subject to the Second Law of Thermodynamics.
• Key Innovation: The authors construct a non-convex rate of dissipation potential ($\hat{\zeta}$). Unlike standard models that assume convex potentials (ensuring unique solutions), the non-convex potential allows for multiple maximizers, mathematically capturing the non-monotonic (S-shaped) stress-strain rate response.
• Governing Equations: The resulting constitutive equation relates the magnitude of the strain rate to the magnitude of the stress via parameters $a, b, \alpha, \beta,$ and $n$.

B. Base Flow Analysis

The authors analyze plane Couette flow between two parallel plates separated by distance $2h$.

• Boundary Conditions: Two cases are considered:
  1. Velocity BCs: Both plates move at prescribed velocities.
  2. Mixed BCs: Stress is prescribed on the top plate, and velocity on the bottom.
• Solutions: Exact analytical solutions for the base flow velocity and stress profiles are obtained. For velocity BCs, the stress is constant across the gap, but the velocity profile depends on the roots of a non-linear algebraic equation, leading to the possibility of multiple solutions.

C. Stability Analysis

• Linearization: The governing equations are linearized around the base states using small perturbations of the form $e^{i(kx - st)}$, where $k$ is the wavenumber and $s$ is the complex growth rate.
• Eigenvalue Problem: The linearized system reduces to an Orr–Sommerfeld-type eigenvalue problem. The stability is determined by the imaginary part of the eigenvalue $m$ (where $s = km$).
  • $Im(m) < 0$: Stable (perturbations decay).
  • $Im(m) > 0$: Unstable (perturbations grow).
• Numerical Solution: The eigenvalue problem is solved numerically using a pseudospectral collocation method based on Gauss–Lobatto points. This method ensures high accuracy and spectral convergence for smooth eigenfunctions.

3. Key Contributions

1. Thermodynamic Derivation: The paper provides a rigorous thermodynamic derivation for non-monotonic stress-power law fluids using a non-convex dissipation potential, offering a physical basis for models that exhibit multiple steady states.
2. Stability Criteria for Multiple States: It establishes definitive stability criteria for the multiple solutions arising from non-monotonic constitutive laws:
   • Solutions on ascending branches of the constitutive curve are unconditionally stable.
   • Solutions on the descending branch are unconditionally unstable.
3. Boundary Condition Dependence: The study highlights a fundamental difference in stability behavior based on boundary conditions:
   • Velocity BCs: Stability depends on the relative Reynolds number ($Re = Re_u - Re_l$). Multiple solutions can coexist; the "most viscous" (highest slope) ascending branch is the most stable.
   • Traction BCs: The base state is unique. Stability is determined solely by whether the prescribed traction lies on an ascending (stable) or descending (unstable) branch.
4. Absence of Marginal Stability: The authors found no neutrally stable states at the inflection points where the solution branches switch. The transition between stable and unstable regimes is abrupt, with the growth rate $m$ not vanishing at the boundaries of the existence regions.

4. Results

• Velocity Boundary Conditions:

  • In the region where three solutions coexist (defined by a specific range of Reynolds numbers), the two solutions on the ascending branches are stable, while the one on the descending branch is unstable.
  • Among the two stable ascending branches, the solution corresponding to the higher slope (higher effective viscosity) exhibits faster decay of perturbations (is "more stable").
  • Stability is independent of the absolute velocities of the plates, depending only on their difference.

• Mixed (Traction-Velocity) Boundary Conditions:

  • The solution is unique for a given traction.
  • If the applied stress places the system on the descending branch of the constitutive curve, the flow is unstable regardless of the lower plate's velocity.
  • If the stress places the system on an ascending branch, the flow is stable.
  • The decay rate of perturbations is influenced by the velocity of the lower plate, even though the stability/instability classification is not.

• Perturbation Structure:

  • Streamline analysis of the perturbations shows that increasing the wavenumber $k$ results in stronger shearing of the fluid.
  • The spatial structure of the perturbations is qualitatively similar for both boundary condition types.

5. Significance

This work is significant for several reasons:

• Physical Admissibility: It resolves the ambiguity of non-unique solutions in non-monotonic fluids by proving that only states on the ascending branches of the constitutive curve are physically realizable in a stable flow.
• Experimental Relevance: The findings explain experimental observations of shear banding and discontinuous shear thickening, suggesting that the "unstable" descending branch is never observed in steady flow because any perturbation will drive the system toward a stable ascending branch.
• Methodological Rigor: By combining thermodynamic derivation with rigorous linear stability analysis (Orr-Sommerfeld) and high-order numerical methods, the paper sets a standard for analyzing complex fluid instabilities.
• Future Directions: The authors identify the stability of Taylor-Couette flow (which includes curvature effects) as a critical next step, as the interplay between geometric instability and constitutive nonlinearity could lead to richer dynamics.

In summary, the paper demonstrates that for stress-power law fluids with non-monotonic responses, flow stability is fundamentally governed by the slope of the constitutive curve and the type of boundary condition applied, with the descending branch being inherently unstable and thus physically inaccessible in steady state.