pith. sign in

arxiv: 2604.05892 · v1 · submitted 2026-04-07 · ⚛️ physics.flu-dyn

Elasto-inertial transitions in viscoelastic flows through cylinder arrays

Jack R. C. King , Henry M. Broadley , Miguel Beneitez This is my paper

Pith reviewed 2026-05-10 18:41 UTC · model grok-4.3

classification ⚛️ physics.flu-dyn
keywords elasto-inertial turbulenceviscoelastic flowscylinder arraysbifurcationschaosporous mediavortex sheddingenergy spectra
0
0 comments X

The pith

Viscoelastic flows through cylinder arrays reach elasto-inertial turbulence via a subcritical saddle-node bifurcation followed by supercritical bifurcations in a Ruelle-Takens-Newhouse route to chaos.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper uses numerical simulations to map the transition from steady Newtonian flow to chaotic elasto-inertial turbulence in viscoelastic fluids passing through periodic cylinder arrays. It shows that elasticity first triggers a sudden jump away from the Newtonian state through a subcritical saddle-node bifurcation, after which the flow undergoes a sequence of supercritical bifurcations that produce chaos. The mechanism rests on the coupling between vortex shedding in the wakes behind each cylinder and the faster flow through the gaps between cylinders. This route operates independently of purely elastic instabilities and produces distinct slow and fast timescales visible in the energy spectra.

Core claim

With increasing elasticity, EIT is reached via an initial sub-critical saddle-node bifurcation from the Newtonian state and then follows a series of supercritical bifurcations, in a Ruelle-Takens-Newhouse route to chaos. This transition is driven by the interaction between vortex shedding in cylinder wakes, and the bulk flow between cylinders. Within the EIT regime, slow dynamics in cylinder wakes interact with fast dynamics in channels between cylinders, leading to two distinct slopes in the energy spectra. At low Reynolds numbers arrowhead structures are present, but these are suppressed at higher inertia. In the present configuration, there is no direct connection between EIT and purely弹性

What carries the argument

The interaction between vortex shedding in cylinder wakes and the bulk flow between cylinders, which initiates the saddle-node bifurcation and sustains the subsequent supercritical cascade to chaos.

If this is right

  • EIT appears at Reynolds numbers below the threshold for inertial turbulence in Newtonian fluids.
  • Energy spectra inside the EIT regime exhibit two power-law slopes arising from slow wake dynamics and fast inter-cylinder channel dynamics.
  • Arrowhead flow structures form only at low Reynolds numbers and vanish once inertia increases.
  • The EIT state develops without requiring or connecting to purely elastic instabilities.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same wake-channel coupling may set the transition threshold in other periodic porous geometries, allowing reduced-order models to predict onset without full three-dimensional simulations.
  • Varying polymer relaxation time in experiments should shift the saddle-node bifurcation point in a measurable way that tests the elasticity dependence found here.
  • The suppression of arrowhead structures with rising inertia suggests that mixing patterns in porous media will change qualitatively once Reynolds number exceeds a modest value.

Load-bearing premise

The chosen viscoelastic constitutive model and numerical discretization faithfully reproduce the bifurcation sequence of real dilute polymer solutions without introducing artifacts that would change the observed transition path.

What would settle it

Experimental observation of a direct link between EIT and purely elastic instabilities, or of a transition sequence lacking the initial saddle-node bifurcation, would falsify the claimed route.

Figures

Figures reproduced from arXiv: 2604.05892 by Henry M. Broadley, Jack R. C. King, Miguel Beneitez.

Figure 1
Figure 1. Figure 1: Left panel: A schematic of the flow configuration. The computational domain is the minimal periodic unit of the cylinder array shown by a dashed line. Right panel: isocontours of resolution, with the inset illustrating a typical node distribution near the boundary. ∂ρui ∂t + ∂ρuiuj ∂xj = − ∂p ∂xi + β Re ∂ 2ui ∂xj∂xj + (1 − β) ReW i cji ∂xj + F0δi1, (2.2) ∂cij ∂t + uk ∂cij ∂xk − ∂ui ∂xk ckj − ∂uj ∂xk cik = … view at source ↗
Figure 2
Figure 2. Figure 2: Diagram of observed flow regimes in the Re − W i space: Black stars - Steady flow (S); blue triangles - Periodic vortex shedding (PVS); red triangles - Unsteady Arrowheads (UA); Blue circles - Periodic wake oscillations (PWO); blue stars - Quasi-periodic wake oscillations (QPWO); red stars - Elasto-inertial turbulence (EIT). Dashed black lines show isocontours of El = W i/Re. As in plane Poiseuille flow Mo… view at source ↗
Figure 3
Figure 3. Figure 3: Instantaneous isocontours of vorticity ω for a range of Re at W i = 0 (top panel), and W i = 10 (bottom panel) [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Instantaneous isocontours of conformation tensor trace cii for a range of Re at W i = 10. observed at Re = 100 for all simulated W i ∈ [1.9, 320], and for all simulated Re > 100, W i ⩾ 2 (illustrated by the red stars in figure 2). We note here that for the present geometry, the parameter space in which we observed arrowheads (red triangles in figure 2) is not connected to the region where we observe EIT (r… view at source ↗
Figure 5
Figure 5. Figure 5: Flow snapshots at Re = 10 for various W i. Panel a) shows streamlines, and isocontours of velocity magnitude at W i = 0. Panels b) to d) show streamlines and isocontours of conformation tensor trace (normalised by W i) for b) W i = 1, c) W i = 2, and d) W i = 10. f1 ≈ 1.3. There is a significant decrease in flow rate Q˙ from W i = 0 to W i = 1. Although the flows at these low W i are qualitatively similar,… view at source ↗
Figure 6
Figure 6. Figure 6: Time evolution of normalised volumetric flow rate Q˙ (left) and EE/W i (right) for Re = 100 and a range of W i. The insets show the variation with W i of the temporal averages of these quantities. Note more values of W i are used for the inset than are plotted in the main figures. ✁✂✄ ✁✂☎ ✁✆ ✁☎ ✁✂☎✝ ✁✂☎✄ ✁✂✆✞ ✁✆✆ ✁✆✞ ✟ ✠ ✡☛☞✌ ✠ ✡✍ ✎✏✑✒ ✎✏✑☛ ✎✏✑✌ ✎✏✑✓ ✎✏✑☛✒ ✎✏✑✌✒ ✎✏✑✔✒ ✎✏✑✕✒ ✎✏✑☛✖✒ gE K [PITH_FULL_IMAGE:fi… view at source ↗
Figure 7
Figure 7. Figure 7: Frequency spectrum of the spatially averaged kinetic energy EgK for Re = 100 and a range of W i. Note the lines have been vertically shifted to allow for ease of visualisation. 0 0.01 0.015 0.02 0.025 0.03 0.035 0.04 10-1 100 101 102 1 1.5 2 0.01 0.015 0.02 10-1 100 101 102 0 20 40 60 80 100 120 0 1 2 0 50 100 Au Ac W i W i [PITH_FULL_IMAGE:figures/full_fig_p010_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Bifurcation diagrams for Re = 100 with varying W i, for two measures: Au = ⟨u2u2⟩ 1/2 V (left panel) and Ac = [PITH_FULL_IMAGE:figures/full_fig_p010_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Approximation of the attractors for the lower branch PO at W i = 1.35 (left), the T 2 -torus at W i = 1.75 (middle), and the chaotic solutions at W i = 2 (right). Top: time-series of the rms of the volume-integrated kinetic energy. Bottom: attractors approximated by 20 embeddings of the time series above. The coordinates are aligned with the principal coordinates of the attractor and rescaled for illustrat… view at source ↗
Figure 10
Figure 10. Figure 10: Pseudospectra contours for the residuals of the approximation of the Koopman eigenvalues and computed eigenvalues with residuals (colour). Left: Lower branch periodic orbit at W i = 1.35, Middle: Upper branch periodic orbit at W i = 1.35. Right: EIT at W i = 10. enumerations of the first M columns, and the second to final columns respectively, such that Y = F(X). (4.5) Following Colbrook et al. (2023) whe… view at source ↗
Figure 11
Figure 11. Figure 11: Contours of the Koopman modes from the lower branch periodic orbit at Re = 100, W i = 1.35. Top: cii, symmetric logarithmic colour scale has been used except for a small region around 0. Bottom: u2. Left: Mean flow, middle: fundamental oscillatory frequency, right: first harmonic [PITH_FULL_IMAGE:figures/full_fig_p014_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Contours of the Koopman modes from the upper branch periodic orbit at Re = 100, W i = 1.35. Top: cii, symmetric logarithmic colour scale has been used except for a small region around 0. Bottom: u2. Left: Mean flow, middle: fundamental oscillatory frequency, right: first harmonic [PITH_FULL_IMAGE:figures/full_fig_p014_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Contours of the Koopman modes from EIT at Re = 100, W i = 10. Top: cii, symmetric logarithmic colour scale has been used except for a small region around 0. Bottom: u2. Left: Mean flow mode, middle: mode with frequency f < f0, right: mode with frequency f ⩾ f0, note the structural similarity between this mode with f = 0.666 and the fundamental physical mode of the lower branch solution in figure 11. The e… view at source ↗
Figure 14
Figure 14. Figure 14: Left panel: Frequency spectrum of the spatially averaged kinetic energy EgK for W i = 10 and a range of Re. Note the lines have been vertically shifted to allow for ease of visualisation. The black dotted line indicates the signature of arrowheads. Right panel: Time evolution of the volume averaged conformation tensor trace EE for several values of Re at W i = 10. The left inset shows variation the normal… view at source ↗
Figure 15
Figure 15. Figure 15: Left panel: Frequency spectrum of the spatially averaged kinetic energy EgK for Re = 50 and a range of W i. Note the lines have been vertically shifted to allow for ease of visualisation. Right panel: Time evolution of the volumetric flow rate Q˙ for several values of W i at Re = 50. The inset shows the variation with W i of ⟨EE/W i⟩ t for Re = 50 [PITH_FULL_IMAGE:figures/full_fig_p018_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Snapshots showing isocontours of the conformation tensor trace (normalised by W i) for Re = 50 and a range of W i. the spectra and these peaks are at lower frequencies. For W i = 10 and W i = 20, there is a small peak in EgK at fa ≈ 3. This is the imprint of weak arrowhead structures, but is only present for W i ∈ [10, 20], and not for smaller or larger values of W i. The right panel of figure 15 shows th… view at source ↗
Figure 17
Figure 17. Figure 17: Frequency spectrum of the spatially averaged kinetic energy EgK at Re = 100 and W i = 10 for three resolutions. same as in our simulations with smin = 1/600. Additionally, for simulations at Re = 10, W i = 160 for smin ∈ [1/400, 1/500, 1/600, 1/750], we find that for smin ⩽ 1/500 the flow dynamics (unsteady arrowheads) are independent of the resolution, and the mean kinetic energy ⟨EK⟩ t is converged to w… view at source ↗
Figure 18
Figure 18. Figure 18: The influence of Ma at W i = 10, for Re = 10 (left panel) and Re = 100 (right panel). The left panel shows the time-evolution of the volume averaged conformation tensor trace EP = ⟨cii⟩V , with the inset showing the frequency spectrum of the volume averaged kinetic energy, for Re = 10. The right panel shows the frequency spectrum of the volume averaged kinetic energy for Re = 100. Broomhead, D. S. & King,… view at source ↗
read the original abstract

For dilute solutions of polymers, chaotic flow states can occur at lower Reynolds numbers than required for inertial turbulence in Newtonian fluids, offering the potential for increased mixing efficiency. These states may be promoted by the flow geometry, and in recent years, porous media have gained attention as a promising setting in which viscoelastic instabilities may be exploited, although studies have primarily been in the creeping flow regime. Cylinder arrays serve as a prototypical porous media, giving a controlled setting in which to investigate flow dynamics. Here we explore the transition to elasto-inertial turbulence (EIT) in cylinder arrays via detailed numerical simulations. With increasing elasticity, EIT is reached via an initial sub-critical saddle-node bifurcation from the Newtonian state and then follows a series of supercritical bifurcations, in a Ruelle-Takens-Newhouse route to chaos. This transition is driven by the interaction between vortex shedding in cylinder wakes, and the bulk flow between cylinders. Within the EIT regime, we observe an interaction between slow dynamics in cylinder wakes, and fast dynamics in channels between cylinders, leading to two distinct slopes in the energy spectra. At low Reynolds numbers arrowhead structures are present, but these are suppressed at higher inertia. In the present configuration, we find no direct connection between EIT and purely elastic instabilities.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript uses time-dependent numerical simulations to study viscoelastic flows through cylinder arrays as a model porous medium. It claims that elasto-inertial turbulence (EIT) is reached from the Newtonian base state via an initial sub-critical saddle-node bifurcation as elasticity increases, followed by a sequence of supercritical bifurcations that produce a Ruelle-Takens-Newhouse route to chaos. The transition is driven by the interaction of vortex shedding in cylinder wakes with the bulk inter-cylinder flow. In the EIT regime, slow wake dynamics couple to fast channel dynamics, producing two distinct slopes in the energy spectra; arrowhead structures appear at low Reynolds number but are suppressed at higher inertia, and EIT is found to have no direct connection to purely elastic instabilities.

Significance. If the reported bifurcation sequence and driving mechanism hold, the work provides a concrete mechanistic picture of how inertia and elasticity interact to produce chaos in a controlled geometry relevant to porous-media flows. The separation of slow wake and fast channel timescales, together with the explicit distinction from purely elastic instabilities, adds useful insight for low-Re mixing applications. The use of direct numerical simulation to trace the full transition path is a clear strength.

major comments (2)
  1. [results section on bifurcation sequence] The classification of the initial transition as a sub-critical saddle-node bifurcation (abstract and results section on the bifurcation sequence) is not accompanied by explicit evidence of hysteresis or a discontinuous jump in an order parameter such as kinetic energy or drag coefficient when the Weissenberg number is varied. This evidence is load-bearing for the claimed route to chaos.
  2. [numerical methods section] The numerical methods section provides no mesh-convergence data, wake-region resolution details, or error estimates for the reported bifurcation points and frequencies. Because the identification of the sub-critical saddle-node and the subsequent frequency-locking steps in the Ruelle-Takens-Newhouse route is known to be sensitive to numerical diffusion and stress boundary-layer resolution, this information is required to substantiate the central claim.
minor comments (2)
  1. [abstract] The abstract states that 'detailed numerical simulations support' the claims but does not indicate the constitutive model (Oldroyd-B, FENE-P, etc.) or the range of Reynolds and Weissenberg numbers examined; these details should appear in the abstract or be cross-referenced to the methods section.
  2. [figure captions] Figure captions for time series and spectra should explicitly label the (Re, Wi) values and the identified dynamical regime (periodic, quasi-periodic, chaotic) to allow readers to connect the plots directly to the bifurcation diagram.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their careful reading and constructive comments on our manuscript. The points raised regarding evidence for the sub-critical bifurcation and numerical validation are well taken, and we will strengthen the paper accordingly. We address each major comment below.

read point-by-point responses

  1. Referee: [results section on bifurcation sequence] The classification of the initial transition as a sub-critical saddle-node bifurcation (abstract and results section on the bifurcation sequence) is not accompanied by explicit evidence of hysteresis or a discontinuous jump in an order parameter such as kinetic energy or drag coefficient when the Weissenberg number is varied. This evidence is load-bearing for the claimed route to chaos.

    Authors: We agree that explicit evidence of hysteresis or a discontinuous jump in an order parameter is important to substantiate the sub-critical saddle-node classification. Our identification of this bifurcation rests on the abrupt onset of finite-amplitude chaotic states directly from the Newtonian base flow as the Weissenberg number increases, without intervening stable periodic regimes. To address the referee's concern, we will add in the revised manuscript additional simulations demonstrating hysteresis loops in both kinetic energy and drag coefficient, obtained by sweeping the Weissenberg number upward and downward across the transition point. revision: yes

  2. Referee: [numerical methods section] The numerical methods section provides no mesh-convergence data, wake-region resolution details, or error estimates for the reported bifurcation points and frequencies. Because the identification of the sub-critical saddle-node and the subsequent frequency-locking steps in the Ruelle-Takens-Newhouse route is known to be sensitive to numerical diffusion and stress boundary-layer resolution, this information is required to substantiate the central claim.

    Authors: We acknowledge that the numerical methods section lacks explicit mesh-convergence data, wake-region resolution details, and error estimates, which are necessary given the sensitivity of bifurcation identification to numerical diffusion. In the revised manuscript we will expand this section to include systematic mesh-convergence studies, quantitative details on grid resolution within cylinder wakes and stress boundary layers, and error estimates for the reported critical Weissenberg numbers and frequencies. These additions will confirm that the identified sub-critical transition and subsequent Ruelle-Takens-Newhouse sequence are robust. revision: yes

Circularity Check

0 steps flagged

No circularity: bifurcation sequence and transition route obtained directly from time-dependent viscoelastic DNS

full rationale

The paper reports the elasto-inertial transition sequence (sub-critical saddle-node followed by supercritical bifurcations in a Ruelle-Takens-Newhouse route) as the observed outcome of direct numerical integration of the incompressible Navier-Stokes equations coupled to a viscoelastic constitutive model. No quantity is defined in terms of another that is later 'predicted'; no parameters are fitted to a data subset and then re-used as a forecast; no uniqueness theorem or ansatz is imported via self-citation to force the reported route; and the driving mechanism (wake vortex shedding interacting with inter-cylinder bulk flow) is identified from the computed fields rather than imposed by construction. The derivation chain therefore remains self-contained against the governing PDEs and the numerical discretization.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on standard incompressible viscoelastic flow equations and numerical integration; no free parameters, invented entities, or ad-hoc axioms are mentioned in the abstract.

axioms (2)
  • domain assumption The flow obeys the incompressible Navier-Stokes equations augmented by a viscoelastic constitutive relation.
    Standard modeling choice for dilute polymer solutions in the abstract's context.
  • domain assumption Cylinder-array geometry with appropriate boundary conditions represents prototypical porous media.
    Explicitly stated framing in the abstract.

pith-pipeline@v0.9.0 · 5532 in / 1417 out tokens · 43899 ms · 2026-05-10T18:41:18.463480+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

3 extracted references · 3 canonical work pages

  1. [1]

    M., Whalley, R

    Abed, W. M., Whalley, R. D., Dennis, D. J.C. & Poole, R. J.2016 Experimental investigation of the impact of elastic turbulence on heat transfer in a serpentine channel. Journal of Non-Newtonian Fluid Mechanics231, 68–78. Alcocer, F. J. & Singh, P.2002 Permeability of periodic arrays of cylinders for viscoelastic flows.Physics of Fluids14(7), 2578–2581. Ba...

    work page 2016

  2. [2]

    Broomhead, D. S. & King, G. P.1986 Extracting qualitative dynamics from experimental data.Physica D: Nonlinear Phenomena20(2-3), 217–236. Browne, C. A. & Datta, S. S.2021 Elastic turbulence generates anomalous flow resistance in porous media.Science Advances7(45), eabj2619. Browne, C. A. & Datta, S. S.2024 Harnessing elastic instabilities for enhanced mix...

    work page 1986

  3. [3]

    Morozov, M

    22 Galindo-Rosales, F. J., Campo-Dea ˜no, L., Sousa, P. C., Ribeiro, V. M., Oliveira, M. S.N., Alves, M. A. & Pinho, F. T.2014 Viscoelastic instabilities in micro-scale flows.Experimental Thermal and Fluid Science59, 128–139. Garg, P. & Rosti, M. E.2025 Elastic turbulence hides in the small scales of inertial polymeric turbulence.Phys. Rev. Lett.135, 0740...

    work page arXiv 2014