Numerical simulation of dilute polymeric fluids with memory effects in the turbulent flow regime

This paper presents a novel numerical framework combining Hermite spectral methods and second-order time integration to efficiently simulate turbulent dilute polymeric fluids with memory effects, revealing that such memory phenomena weaken the drag-reducing capabilities of polymer additives.

The Gist

Imagine you are trying to predict how a river flows. Usually, water flows smoothly or gets choppy (turbulent) in a way we can calculate with standard math. But what if you add a tiny amount of long, stretchy strings (like microscopic rubber bands) into the water? These are "polymeric fluids." In the real world, adding just a pinch of these strings can make the water flow much more smoothly, reducing the friction (drag) against the pipes or banks. This is called the "drag-reducing effect."

However, these strings aren't just simple rubber bands; they have a memory. They don't just react to what's happening right now; they remember how they were stretched in the past. This "memory" makes the math incredibly difficult, like trying to solve a puzzle where every piece changes shape based on the entire history of the game.

Here is what the researchers in this paper did to solve that puzzle:

1. The Problem: A Math Monster with Too Many Dimensions

To understand these fluids, scientists usually have to track two things at once:

• The Macro View: Where the water is going (the river).
• The Micro View: How every single tiny string is stretching and twisting (the rubber bands).

Doing this for a turbulent river is like trying to track the position and mood of every single grain of sand in a beach storm simultaneously. The math becomes so huge and complex (high-dimensional) that even supercomputers struggle to solve it, especially when you add the "memory" factor (time-fractional derivatives).

2. The Solution: A Smart Shortcut (The "Hermite" Trick)

Instead of tracking every single string individually, the authors used a clever mathematical trick called the Hermite spectral method.

Think of it like this: Instead of trying to describe the exact shape of a wobbly jellyfish by drawing every tentacle, you describe the jellyfish using a few key "shapes" (like a circle, an oval, a squiggle) that you know how to combine.

• They proved that for these specific stretchy strings, you only need to track seven of these key shapes (or "modes") to get the exact same result as tracking millions of individual strings.
• They also found the perfect "zoom level" (a scaling parameter) to make this shortcut work perfectly. Before, people just guessed the zoom level; this paper proved the exact setting needed to make the shortcut mathematically identical to the full, complicated version.

3. Handling the "Memory"

The "memory" part of the fluid is tricky because it requires looking back at every moment in time. The authors used a technique called kernel compression.

• Analogy: Imagine you have a long, heavy history book you need to read to solve a problem. Instead of reading every page from start to finish every time, they found a way to summarize the whole book into a few short, weighted notes. This allowed them to turn the "memory" problem into a set of simpler, faster equations that computers can handle.

4. What They Discovered

Using this new, efficient method, they ran simulations to see how these "memory-keeping" fluids behave in turbulent flows.

• The Surprise: They found that the "memory" actually weakens the benefit of the polymers.
• The Metaphor: Imagine the polymer strings are like a team of rowers trying to smooth out a choppy boat ride. If the rowers react instantly to the waves, they do a great job. But if the rowers have a "delay" or "memory" (they react to waves that happened a second ago), they end up rowing out of sync.
• The Result: The simulations showed that because of this memory effect, the polymers were less effective at reducing drag (friction) than previously thought. In fact, in some turbulent scenarios, the memory effects made the fluid behave more chaotically, reducing the "smoothing" power of the additives.

Summary

The paper didn't invent new polymers or test them in real pipes. Instead, they built a super-efficient computer simulation that finally allows scientists to model these complex, "memory-having" fluids in turbulent conditions. Their main takeaway is a cautionary one: Memory matters. If you ignore the fact that these fluid additives remember their past, you might overestimate how well they will work to smooth out turbulent flows.

Technical Summary

Technical Summary: Numerical Simulation of Dilute Polymeric Fluids with Memory Effects in the Turbulent Flow Regime

Problem Statement
The paper addresses the numerical challenge of solving the time-fractional Navier–Stokes–Fokker–Planck (NSFP) system for dilute polymeric fluids in the turbulent regime. This system models fluids where polymer molecules exhibit memory effects, represented by a Riemann–Liouville time-fractional derivative. The problem is characterized by three primary difficulties:

1. High Dimensionality: The Fokker–Planck (FP) equation is defined on the Cartesian product of the macroscopic spatial domain ($\Omega \subset \mathbb{R}^d$) and the microscopic configuration space ($D \subset \mathbb{R}^d$), resulting in a $2d$-dimensional problem.
2. Non-locality in Time: The time-fractional derivative introduces a history dependence, requiring the storage and processing of the entire solution history, which is computationally expensive.
3. Turbulent Resolution: Resolving turbulent flow requires high spatial and temporal resolution, compounding the computational cost of the high-dimensional, non-local system.

Existing approaches, such as stochastic methods (e.g., Brownian configuration fields) or macroscopic closures for complex spring forces (e.g., FENE), often suffer from high computational costs or lack rigorous macroscopic formulations for time-fractional models.

Methodology
The authors propose a purely macroscopic numerical scheme that reduces the high-dimensional micro-macro system to a system of coupled partial differential equations (PDEs) defined only on the spatial domain. The methodology consists of the following key components:

1. Hermite Spectral Method: To handle the unbounded configuration space $D = \mathbb{R}^d$ associated with the Hookean spring model, the authors discretize the FP equation using Hermite spectral functions. This transforms the problem into a system of equations for spectral modes $\phi_{i_1, \dots, i_d}$.

   • Optimal Scaling: A critical theoretical contribution is the proof that choosing the Hermite scaling parameter $a = 1/\sqrt{2}$ ensures that the macroscopic extra-stress tensor derived from the spectral approximation exactly matches the analytical extra-stress tensor for the underlying micro-macro system. This eliminates the arbitrary parameter choices found in previous works.
   • Truncation: It is demonstrated that only spectral modes of degree 0 and 2 are required to compute the polymer-induced extra-stress tensor, reducing the system to just 7 equations in 3D (4 in 2D).

2. Kernel Compression Method: To address the non-locality of the time-fractional derivative, the Riemann–Liouville integral kernel is approximated by a weighted sum of exponentials using a rational approximation (based on the AAA algorithm). This transforms the fractional differential equation into a system of ordinary differential equations (ODEs) for auxiliary "fractional modes," avoiding the need to store the full history.

3. Time Integration and Coupling:

   • The Navier–Stokes equations are solved using a projection method (Chorin-type) with second-order time integration.
   • The macroscopic FP (MFP) equations and fractional mode ODEs are integrated using a second-order Backward Differentiation Formula (BDF).
   • Coupling terms are extrapolated to achieve a fully coupled scheme. The authors prove that this approach yields convergence rates independent of the order of the time-fractional derivative $\alpha$.

4. Spatial Discretization: The resulting macroscopic PDE system is discretized using continuous second-order finite elements on unstructured meshes.

Key Results
The authors implemented the scheme using the MFEM library and conducted extensive numerical experiments in 2D and 3D:

• Convergence: The scheme achieves optimal convergence orders. For the time-fractional FP equation, the error decays as $O(\Delta t^{1+\alpha})$. For the fully coupled NSFP system, the extra-stress tensor exhibits first-order convergence, consistent with the dependence on the time-fractional derivative.
• Flow Around a Cylinder: Simulations of flow past an obstacle show that the order of the time-fractional derivative ($\alpha$) significantly influences flow stability. While integer-order ($\alpha=1$) flows remain laminar under the tested conditions, fractional cases ($\alpha=0.5, 0.8$) transition to turbulence. Lower $\alpha$ values lead to faster turbulence development and a weaker polymer-induced force.
• Rough Wall Channel Flow: In a geometry mimicking a corroded pipeline, the study identifies three flow regimes based on the center-of-mass diffusivity ($\epsilon$) and polymer viscosity contribution: stable laminar, sustained turbulence, and an intermediate regime where turbulence is initially generated but subsequently damped.
  • Memory Effects: The simulations reveal that memory effects (fractional derivatives) weaken the drag-reducing capability of the polymers. Specifically, the magnitude of the polymer-induced force is lower in fractional cases compared to integer-order cases, leading to less effective turbulence suppression.
• Pipeline Strain Relief (3D): A 3D simulation of flow through a pipe with right-angle turns demonstrates the scalability of the method (using up to 256 cores). The results confirm that while pure solvent flow becomes turbulent after bends, the dilute polymeric fluid with memory effects maintains a laminar state, though the drag-reduction effect is modulated by the memory parameter.

Significance and Claims
The paper claims to provide the first rigorous numerical study of drag-reducing agents with memory effects in the turbulent regime. Its primary significance lies in:

1. Efficiency: By deriving a purely macroscopic model equivalent to the full micro-macro system (under the optimal scaling parameter), the method bypasses the curse of dimensionality inherent in direct numerical simulation of the FP equation.
2. Theoretical Rigor: The proof of the optimal scaling parameter $a = 1/\sqrt{2}$ ensures the macroscopic model is not an approximation in the constitutive sense but an exact representation of the stress coupling for the Hookean model.
3. Physical Insight: The numerical results challenge the assumption that memory effects always enhance drag reduction; instead, they show that memory effects can weaken the polymer-induced force and reduce the drag-reducing efficiency in turbulent flows.

The authors conclude that while their approach is successful for Hookean models, the lack of a macroscopic closure for more realistic FENE-type models remains a significant challenge for future research in time-fractional polymeric fluid simulation.