Transition to the ultimate regime of turbulent convection in stratified inclined duct flow

Through high-resolution direct numerical simulations up to a Reynolds number of 8000, this study demonstrates that stratified inclined duct flow transitions to an ultimate turbulent regime characterized by enhanced transport scaling ($\mathrm{Nu} \sim \mathrm{Ra}^{1/2}$) and logarithmic velocity profiles via a subcritical, hysteretic mechanism, thereby bridging the gap between laboratory limitations and realistic geophysical mixing conditions.

The Gist

The Big Picture: A River in a Tilted Tube

Imagine you have a long, rectangular tube tilted at an angle. You fill the top half with heavy, salty water (like the deep ocean) and the bottom half with light, fresh water (like a river). When you open the gate between them, the heavy water wants to slide down, and the light water wants to float up.

This creates a stratified exchange flow: two rivers flowing in opposite directions inside the same tube, separated by a thin, invisible layer. This happens in nature (like water flowing through the Strait of Gibraltar) and in buildings (air moving between rooms).

For a long time, scientists knew how this flow behaved when it was calm or slightly wavy. But they didn't know what happened when the flow got extremely violent and turbulent. That's what this paper investigates.

The Discovery: Finding the "Ultimate" Speed Limit

The researchers used supercomputers to simulate this flow at speeds and turbulence levels far higher than any previous experiment could measure. They were looking for a specific "tipping point."

The Analogy: The Highway Traffic Jam
Think of the flow like cars on a highway.

• Low Speed (Laminar): Cars are driving in neat lanes. If you add a little more gas (more tilt or density difference), the cars speed up smoothly.
• Medium Speed (Intermittent): Traffic gets bumpy. Some cars speed up, others slow down. It's chaotic but manageable.
• The "Ultimate" Regime: Suddenly, the traffic breaks free from the rules. The cars aren't just driving; they are weaving, merging, and creating a massive, efficient swarm.

The team discovered that once the flow hits a certain speed (a specific "Reynolds number"), it undergoes a phase change. It enters what they call the "Ultimate Regime."

In this regime, the efficiency of mixing the two fluids skyrockets. It's like the difference between stirring a cup of coffee with a spoon (slow, inefficient) and using a high-powered blender (fast, chaotic, but incredibly effective at mixing).

How They Knew It Happened: The "Logarithmic" Clue

How do you know the flow has hit this "Ultimate" state? The researchers looked at the walls of the tube.

The Analogy: The Riverbank
Imagine a river flowing next to a muddy bank.

• Before the transition: The water right next to the mud is very calm and smooth. It's like a quiet sidewalk next to a busy street.
• After the transition: The water right next to the mud becomes wild and turbulent. It starts following a specific mathematical pattern called a "logarithmic profile."

The researchers found that when the turbulence near the walls became strong enough (reaching a specific threshold), the entire flow switched gears. The "sidewalk" turned into a "mosh pit," and this chaos allowed the two fluids to mix much faster than physics predicted for normal turbulence.

The "Hysteresis" Surprise: The Sticky Switch

One of the most fascinating findings is that this transition isn't a simple on/off switch. It's more like a sticky door.

The Analogy: The Stubborn Door
Imagine you are trying to push a heavy door open.

1. You push harder and harder until it finally bursts open (the flow becomes "Ultimate").
2. Now, you try to close it. You have to push it much harder in the opposite direction before it actually shuts back to the "calm" state.

The paper shows that the flow has hysteresis. Once it gets into the super-turbulent "Ultimate" mode, it wants to stay there even if you slow it down a bit. It only snaps back to the calm state when you slow it down way below the speed where it originally became turbulent. This suggests the transition is "subcritical," meaning it's a sudden, dramatic jump rather than a gradual slide.

Why Does This Matter?

This isn't just about tubes in a lab. This discovery helps us understand:

1. The Ocean: How heat and salt mix in the deep ocean, which affects global climate patterns.
2. Industry: How to design better ventilation systems or mix chemicals in factories.
3. Physics: It connects this specific type of flow to a broader class of "ultimate turbulence" that scientists have been chasing for decades.

The Takeaway

The researchers found that when buoyancy-driven flows get fast enough, they don't just get "more" turbulent; they fundamentally change their behavior. They develop wild, chaotic layers near the walls that act like a super-efficient blender, mixing fluids at a rate that scales perfectly with the energy input. It's a new "ultimate" state of fluid motion that was previously out of reach for both experiments and computers.

Technical Summary

1. Problem Statement and Motivation

The Stratified Inclined Duct (SID) is a canonical experimental setup used to study buoyancy-driven exchange flows between two reservoirs of different densities connected by a tilted channel. These flows are fundamental to geophysical processes (e.g., oceanic straits, estuaries) and industrial applications (e.g., ventilation).

While the laminar, wave, and intermittently turbulent regimes of SID flow are well-documented, the highly turbulent regime remains poorly understood due to experimental and numerical limitations:

• Experimental Limitations: Laboratory experiments can reach high Reynolds numbers ($Re$) but lack the spatial and temporal resolution to resolve fine-scale boundary layers and scalar fields, particularly at realistic Prandtl numbers ($Pr \approx 7$ for heat in water).
• Numerical Limitations: Previous Direct Numerical Simulations (DNS) were restricted to $Re \sim O(10^3)$, failing to capture the transition to fully developed turbulence and the "ultimate regime" of convection.

The central question addressed is: Does SID flow exhibit a transition to an "ultimate regime" of turbulent convection (characterized by enhanced transport scaling) at high $Re$, and what are the physical mechanisms driving this transition?

2. Methodology

The authors conducted three-dimensional Direct Numerical Simulations (DNS) using the highly parallel solver AFiD.

• Governing Equations: The incompressible Navier-Stokes equations under the Boussinesq approximation were solved, coupled with a scalar transport equation for density.
• Parameters:
  • Prandtl Number: Fixed at $Pr = 7$ (simulating heat transfer in water).
  • Reynolds Number: Simulations extended up to $Re = 8000$ (corresponding to Rayleigh number $Ra \approx 1.8 \times 10^9$).
  • Geometry: A rectangular duct with aspect ratios $L/H = 30$ (streamwise) and $W/H = 1$ (spanwise), inclined at angles $\theta = 4^\circ, 7^\circ, 10^\circ$.
• Numerical Techniques:
  • Immersed Boundary Method (IBM): Used for lower $Ra$ simulations to replicate the duct-reservoir geometry.
  • Wall-Clustered Mesh: For high $Ra$($>10^8$), IBM was removed in favor of a refined, wall-clustered mesh in the vertical direction to fully resolve the turbulent kinetic boundary layers (BLs).
  • Multiple-Resolution Strategy: A Hermite interpolation method was employed to resolve the scalar field on a refined mesh while maintaining computational efficiency.
• Validation: The simulations successfully reproduced the four known qualitative flow regimes (Laminar, Wave, Intermittent, Turbulent) observed in previous experiments and lower-$Re$ DNS.

3. Key Contributions

1. First Observation of the Ultimate Regime in SID: The study provides the first 3D numerical evidence of the transition to the ultimate regime in a purely buoyancy-driven, stratified shear flow.
2. Identification of Scaling Laws: The authors establish the transition from a weakly turbulent scaling ($Nu \sim Ra^{1/3}$) to the ultimate scaling ($Nu \sim Ra^{1/2}$), indicating a fundamental change in transport physics.
3. Mechanism Elucidation: The paper identifies the transition of the kinetic boundary layers (BLs) from laminar to turbulent as the primary trigger for the ultimate regime, characterized by the emergence of logarithmic velocity profiles.
4. Hysteresis and Subcriticality: The study demonstrates that the transition to the ultimate regime is non-normal-nonlinear (subcritical and hysteretic), distinguishing it from linear instability routes.
5. Analogy to Vertical Convection: The work connects SID flow to the broader class of wall-bounded turbulent convection (specifically Convection in a Vertical Channel, CVC), showing that a streamwise linear density gradient drives the turbulence in the ultimate regime.

4. Key Results

A. Global Transport and Scaling

• Nusselt Number ($Nu$): The mass transfer efficiency ($Nu$) exhibits two distinct regimes:
  • **Low $Ra$($Ra \lesssim 10^8$):**$Nu \sim Ra^{1/3}$, consistent with weakly driven thermal convection where turbulence is not fully developed.
  • Ultimate Regime ($Ra \gtrsim 10^8$): A sharp transition occurs where $Nu \sim Ra^{1/2}$. This scaling implies that transport becomes independent of viscosity ($\nu$) and diffusivity ($\kappa$), a hallmark of the ultimate regime proposed by Kraichnan.
• Hydraulic Mass Flux ($Q_m$): The dimensionless mass flux decreases as $Ra$ increases (due to enhanced mixing) and plateaus at a value significantly below the inviscid limit of 0.5 in the current $Ra$ range, though a slight logarithmic trend suggests it may approach 0.5 at infinite $Ra$.

B. Energy Budget and Mixing

• Dissipation: The dimensionless kinetic energy dissipation ($\epsilon'_u$) scales as $Ra^{1/3}$ in the laminar/transitional regime and shifts to $Ra^{1/2}$ in the ultimate regime, mirroring the $Nu$ scaling.
• Mixing Efficiency ($\Gamma$): The ratio of buoyancy flux to dissipation ($\Gamma = N_{ub}/\epsilon'_u$) increases from $\approx 0.01$ and asymptotes to $\approx 0.1$ in the ultimate regime. This is notably lower than the widely adopted oceanographic value of $\Gamma \approx 0.2$.
• Buoyancy Flux: Unlike Rayleigh-Bénard convection, SID exhibits a finite buoyancy flux ($N_{ub} \sim Ra^{0.70}$) due to the stable stratification, which consumes a portion of the kinetic energy generated by buoyancy.

C. Boundary Layer Dynamics and Transition Mechanism

• Shear Reynolds Number ($Re_s$): The transition to the ultimate regime coincides with the shear Reynolds number at the top and bottom walls exceeding a critical threshold of $Re_s^* \approx 420$.
• Velocity Profiles:
  • Laminar BL: Smooth velocity profiles.
  • Turbulent BL: Emergence of logarithmic velocity profiles ($u^+ = \frac{1}{\kappa} \ln z^+ + B$) and spanwise streaks with spacings of 100–200 viscous units.
• Hysteresis: Time-dependent simulations ramping $Ra$ up and down reveal a hysteresis loop. The transition to the ultimate regime occurs at a higher $Ra$ during the ramp-up than the return to intermittency during the ramp-down. This confirms the subcritical nature of the transition, mediated by nonlinear instabilities within the boundary layers.

D. Analogy to Convection in a Vertical Channel (CVC)

• In the ultimate regime, the streamwise density profile becomes linear with a constant gradient ($\beta \approx 0.01$).
• When plotted against a Grashof number based on this gradient ($Gr = \beta Ra / Pr$), SID data collapses with CVC data, confirming that the ultimate regime in SID is sustained by a linear density gradient driving turbulent convection, similar to CVC.

5. Significance and Outlook

• Theoretical Impact: This work bridges the gap between stratified shear flows and wall-bounded turbulent convection. It confirms that the "ultimate regime" is a universal feature of buoyancy-driven flows, even when the shear is generated solely by buoyancy rather than mechanical forcing.
• Geophysical Relevance: The findings provide critical insights into mixing efficiency and transport rates in oceanographic settings (e.g., straits, estuaries) where flows are highly turbulent. The lower mixing efficiency ($\Gamma \approx 0.1$) compared to standard oceanographic models suggests a need to re-evaluate parameterizations in large-scale climate models.
• Future Directions: The authors note that the asymptotic behavior of $Q_m$ at infinite $Ra$ and the behavior at very small inclination angles (approaching horizontal ducts) remain open questions. Additionally, the influence of varying Prandtl numbers (e.g., salt water with $Pr \approx 700$) requires further investigation.

In summary, this paper establishes the transition to the ultimate regime in Stratified Inclined Duct flow as a subcritical, hysteretic process driven by the turbulentization of kinetic boundary layers, leading to a $Nu \sim Ra^{1/2}$ scaling and fundamentally altered mixing properties.