Elasto-inertial transitions in viscoelastic flows through cylinder arrays

This study uses numerical simulations to demonstrate that viscoelastic flow through cylinder arrays transitions to elasto-inertial turbulence via a Ruelle-Takens-Newhouse route to chaos, driven by the interaction between vortex shedding and bulk flow, while showing no direct connection to purely elastic instabilities.

The Gist

Imagine you are trying to mix sugar into a cup of hot tea. If you just stir it gently, the sugar dissolves slowly. But if you stir it vigorously, the water swirls, creates chaos, and the sugar mixes instantly. In the world of fluids, this "vigorous stirring" is called turbulence.

For a long time, scientists thought you needed a lot of energy (fast stirring) to create this chaos, especially for thick, sticky fluids like polymer solutions (think honey or melted plastic). But recently, researchers discovered something strange: these sticky fluids can become chaotic and mix incredibly fast even when they are barely moving. This is called Elasto-Inertial Turbulence (EIT).

This paper is like a detective story where scientists try to figure out how and why this happens in a very specific setting: a forest of tiny cylinders (like a bundle of straws or a filter).

Here is the breakdown of their discovery, using simple analogies:

1. The Setting: The "Cylinder Forest"

Imagine a river flowing through a dense forest of tree trunks (the cylinders).

• Normal Water (Newtonian): If the water flows slowly, it moves smoothly around the trees. If it flows fast, it starts to spin off the trees in little whirlpools (vortex shedding), creating a chaotic mess.
• Sticky Polymer Water (Viscoelastic): This fluid is like water mixed with long, tangled spaghetti strands. These strands stretch and snap back like rubber bands.

2. The Mystery: Two Different Ways to Chaos

The researchers wanted to know: Does the "spaghetti" chaos start because the fluid is too stretchy (elastic), or because it's moving too fast (inertia)?

They found that in their "cylinder forest," the chaos doesn't start the way they expected.

The "Arrowhead" That Disappears

At very low speeds, the sticky fluid forms strange, sharp shapes in the gaps between the cylinders. The researchers call these "Arrowheads."

• Analogy: Imagine a group of people walking through a narrow hallway. If they are very stiff and slow, they might form a rigid, arrow-shaped line.
• The Twist: The scientists found that as they increased the speed (even just a little bit), these "Arrowheads" vanished. They were suppressed by the flow. So, the Arrowheads are not the key to the big chaotic mix-up (EIT) in this specific setup.

The Real Culprit: The "Wake" and the "Channel"

Instead of Arrowheads, the chaos starts from a battle between two different behaviors:

1. The Wake (The Lazy Zone): Behind every cylinder, there is a "wake" where the fluid swirls slowly.
2. The Channel (The Fast Lane): Between the cylinders, the fluid rushes through fast.

The Transition Story:

1. The Snap: As they increased the "stretchiness" (elasticity) of the fluid, the flow didn't change gradually. It suddenly "snapped" from a smooth, predictable state to a slightly different, faster state. Think of it like a rubber band that suddenly jumps to a new tension.
2. The Dance: Once that snap happened, the fluid started doing a complex dance. The slow swirling behind the cylinders (the wake) began to interact with the fast rushing in the channels.
3. The Explosion: This interaction created a feedback loop. The slow swirls pushed the fast lanes, and the fast lanes pushed the swirls, until the whole system went chaotic.

3. The "Two-Speed" Music

When they listened to the "music" of the flow (using energy spectra), they heard two distinct rhythms, like a song with a fast drum beat and a slow bass line.

• The Fast Beat: This comes from the fluid rushing through the channels between the cylinders.
• The Slow Bass: This comes from the lazy, swirling wakes behind the cylinders.
• The Result: The chaos is a mix of these two speeds. The fast part creates the "turbulence," while the slow part keeps it going.

4. Why This Matters

Why do we care about sticky fluids mixing in a cylinder forest?

• Real World: This happens in oil recovery (getting oil out of rock), in medical devices, and in industrial filters.
• The Benefit: If we can trigger this "Elasto-Inertial Turbulence," we can mix things much better and faster without needing to pump the fluid at high speeds. It's like getting a high-speed blender effect using a slow, gentle stir.

The Big Takeaway

The paper concludes that in this specific "cylinder forest," the chaotic mixing does not come from the "Arrowhead" shapes (which are purely elastic). Instead, it comes from a hybrid of inertia (speed) and elasticity (stretchiness).

It's a bit like a car crash: You don't need the car to be made of rubber or moving fast to crash; you need the rubber car to be moving fast enough to hit a wall and bounce back in a chaotic way. The researchers found that the "wall" in this case is the cylinder, and the "bounce" creates the perfect storm for mixing.

In short: They mapped out the exact steps to turn a slow, sticky flow into a super-efficient mixer, proving that sometimes you need a little bit of speed and a lot of stretch to get the job done.

Technical Summary

1. Problem Statement

The paper investigates the transition to Elasto-Inertial Turbulence (EIT) in viscoelastic polymer solutions flowing through cylinder arrays, which serve as a prototypical model for porous media.

• Context: While Elastic Turbulence (ET) occurs at vanishing inertia and EIT occurs when both elastic and inertial effects are significant, the specific transition pathways and mechanisms in confined, porous-like geometries remain poorly understood.
• Gap: Previous studies in porous media have been limited to creeping flow regimes or disordered geometries where pore-scale dynamics are difficult to isolate. There is a need to understand how EIT emerges in ordered arrays, specifically whether it stems from purely elastic instabilities (like arrowhead structures) or inertial mechanisms (like vortex shedding).
• Challenge: Simulating these flows is numerically difficult due to the need for high-fidelity resolution of thin, highly stretched polymer sheets and the risk of triggering spurious Polymer-Diffusive Instabilities (PDI) if numerical diffusivity is not handled correctly.

2. Methodology

The authors employed high-fidelity Direct Numerical Simulations (DNS) using a unique numerical framework.

• Governing Equations: The flow is modeled using the incompressible Navier-Stokes equations coupled with a simplified Phan-Thien-Tanner (sPTT) constitutive model for the polymer conformation tensor. The system is solved in a weakly compressible regime (low Mach number, $Ma=0.01$) to ensure stability while maintaining negligible density variations (<0.1%).
• Geometry: A doubly-periodic unit cell of a hexagonal cylinder array with a porosity of $\phi \approx 0.773$ (cylinder spacing $S=2$).
• Numerical Scheme:
  • Method: Local Anisotropic Basis Function Method (LABFM), a mesh-free, high-order (6th-order internal, 4th-order at boundaries) collocated method. This approach offers low numerical dissipation, crucial for resolving ET/EIT dynamics, unlike standard low-order finite volume methods.
  • Stabilization: To prevent PDI (a linear instability arising from polymer diffusion), the authors used a very low dimensionless diffusivity ($\kappa = 10^{-5}$) and verified that the observed dynamics were not driven by this numerical artifact.
  • Resolution: Adaptive node spacing with a minimum resolution of $s_{min} = D/600$ at cylinder walls.
• Analysis Techniques:
  • Bifurcation Analysis: Tracking flow states (steady, periodic, chaotic) across the Reynolds ($Re$) and Weissenberg ($Wi$) parameter space.
  • Koopman Analysis: Using Residual Dynamic Mode Decomposition (ResDMD) to extract eigenvalues and eigenmodes from time-series data, allowing for the linearization of nonlinear dynamics and identification of dominant physical mechanisms.

3. Key Results

A. Flow Regimes and Transition Pathway

The study mapped the flow regimes in the $Re-Wi$ space, identifying distinct states:

1. Steady Flow: At low $Re$ and $Wi$.
2. Arrowhead Structures: At low $Re$($Re \le 25$) and moderate $Wi$, "arrowhead" structures (characteristic of ET) form in the channels between cylinders. However, these do not lead to chaos in this specific geometry; they remain quasi-periodic.
3. Elasto-Inertial Turbulence (EIT): At higher $Re$($Re \ge 100$) and increasing $Wi$, the flow transitions to a chaotic state.

The Transition Mechanism (at $Re=100$):
The transition to EIT follows a specific route as elasticity ($Wi$) increases:

1. Sub-critical Saddle-Node Bifurcation: At $Wi \approx 1.4$, the system jumps from a stable periodic orbit (upper branch, dominated by vortex shedding) to a lower stable branch with higher kinetic energy and lower elastic energy.
2. Ruelle-Takens-Newhouse (RTN) Route to Chaos: From the lower branch, the flow undergoes a series of supercritical bifurcations:
   • Periodic Orbit ($Wi=1.35$)$\to$ $T_2$-Torus ($Wi=1.75$)$\to$ Chaos ($Wi \ge 1.9$).
3. Driving Force: The transition is driven by the interaction between vortex shedding in the cylinder wakes and the bulk flow in the channels between cylinders.

B. Energy Spectra and Dynamics

• Two-Slope Power Law: Within the EIT regime, the energy spectra exhibit two distinct slopes:
  • High Frequencies ($f > f_0$): Slope $\approx f^{-3}$, driven by EIT-like dynamics in the channels between cylinders.
  • Low Frequencies ($f < f_0$): Slope $\approx f^{-1.2}$, driven by slow dynamics (oscillations) in the cylinder wakes.
• Arrowhead Suppression: Arrowhead structures, present at low $Re$, are suppressed as inertia ($Re$) increases. At $Re=50$, the flow is dominated by wake oscillations but does not become chaotic. At $Re=100$, vortex shedding triggers EIT.

C. Koopman Analysis Insights

• Mode Structure: The analysis confirmed that the fundamental oscillatory modes of the periodic orbits (pre-EIT) share structural similarities with the high-frequency modes in the EIT state.
• Wake Interaction: A distinct low-frequency mode emerges in the chaotic state, characterized by highly stretched polymers in the channels and significant wake oscillations, which are absent in the periodic states.
• Independence from PDI: The analysis confirmed that the observed chaotic dynamics are not artifacts of polymer diffusivity (PDI), as PDI modes appeared only at much higher frequencies with negligible amplitude.

4. Significance and Conclusions

• Mechanism Clarification: The study establishes that in cylinder arrays, EIT is not a direct continuation of purely elastic instabilities (arrowheads). Instead, it is triggered by an inertial vortex-shedding instability that interacts with the elastic bulk flow.
• Transition Route: The identification of the RTN route to chaos via a sub-critical saddle-node bifurcation provides a precise mathematical description of how viscoelastic flows become turbulent in porous media.
• Practical Implications: Understanding that EIT can be triggered at lower Reynolds numbers than Newtonian turbulence suggests potential for enhanced mixing and heat transfer in industrial porous media applications (e.g., catalytic reactors, oil recovery) without requiring high flow rates.
• Numerical Contribution: The successful application of the mesh-free, high-order LABFM method demonstrates a robust strategy for simulating complex viscoelastic flows without the numerical dissipation or spurious instabilities that plague traditional methods.

In summary, the paper decouples the roles of elasticity and inertia in porous media flows, showing that while elasticity is necessary for the chaotic state, the trigger for the transition in this geometry is an inertial wake instability, distinct from the purely elastic mechanisms observed in other canonical flows.