Experimental investigation into Lagrangian statistics of droplets in homogeneous isotropic turbulence

This study experimentally demonstrates that neutrally buoyant droplets in homogeneous isotropic turbulence, despite undergoing size-dependent breakup and exhibiting longer integral times and extended ballistic regimes, behave dynamically similarly to finite-size rigid particles.

The Gist

The Big Picture: A Stormy Ocean of Oil and Water

Imagine a giant, transparent, soccer-ball-shaped tank filled with water. Inside this tank, scientists are spinning twelve giant propellers to create a chaotic, swirling storm. This is turbulence—the same kind of chaotic motion you see in a rushing river or a stormy ocean.

Now, imagine injecting a small amount of oil into this storm. Because the oil and water are mixed, the violent swirling motion breaks the oil into thousands of tiny droplets. These droplets are the "actors" in this experiment. The scientists wanted to answer a simple question: How do these oil droplets move through the storm, and does their size matter?

The Setup: A High-Speed Camera Trap

To watch the droplets, the scientists built a high-tech "cage" with clear walls and 12 spinning propellers.

• The Actors: They used silicone oil (which is roughly the same weight as water, so it doesn't sink or float) dyed with a glowing red color.
• The Directors: Four high-speed cameras acted like a team of referees, filming the droplets from every angle simultaneously.
• The Goal: They used special software to track the path of individual droplets in 3D, like following a single fish in a school, but doing it for thousands of fish at once.

Discovery 1: The "Shattering" Effect

First, they looked at the size of the droplets.

• The Analogy: Think of a large rock being thrown into a waterfall. The water hits the rock, and it shatters into smaller and smaller pebbles.
• The Finding: When the propellers spin faster (creating a stronger storm), the oil droplets get smashed into smaller pieces. The faster the turbulence, the smaller the droplets.
• The Pattern: The sizes of the droplets followed a predictable pattern (a "log-normal distribution"). Interestingly, as the storm got stronger, the droplets didn't just get smaller; they also became more uniform in size, like a factory sorting pebbles by size.

Discovery 2: The "Ghost" vs. The "Boulder"

Next, they looked at how the droplets moved. They compared the oil droplets to "tracers" (tiny, invisible particles that move exactly with the water, like a ghost).

• Speed and Jitters: Surprisingly, the oil droplets moved almost exactly like the ghost tracers. Whether the droplet was small or medium-sized, its speed and how much it "jittered" (accelerated) were very similar to the water itself.
• The Filter Effect: However, the largest droplets acted like a sieve or a noise-canceling headphone. Because they are bigger, they can't feel the tiniest, fastest jitters of the water. They smooth out the chaos. If the water vibrates wildly on a microscopic scale, a big droplet just ignores those tiny shakes and keeps moving smoothly.

Discovery 3: The "Momentum" Difference (The Real Surprise)

This is where things got interesting. While the speed was similar, the memory of the droplets was different.

• The Analogy: Imagine two cars driving on a bumpy road.
  • Car A (Small Droplet): It's a tiny go-kart. When the road bumps, it bounces instantly. It changes direction immediately.
  • Car B (Large Droplet): It's a heavy truck. When the road bumps, it wobbles a bit, but its heavy weight keeps it going in the same direction for longer. It has more inertia.
• The Finding: The larger droplets held onto their speed longer. If they were moving left, they kept moving left for a longer time before the water forced them to turn.
• The "Ballistic" Phase: In physics, "ballistic" means moving in a straight line without being stopped. The study found that big droplets spent more time in this "straight-line" mode before the turbulence forced them to wander randomly. They were "stubborn" about their direction.

Why Does This Matter?

This research is like learning the rules of a game that happens everywhere in nature and industry:

1. Rain: How raindrops form and fall through storm clouds.
2. Engines: How fuel sprays break up in a car engine to burn efficiently.
3. Pollution: How oil spills or chemical clouds spread in the ocean or air.

The Bottom Line:
Even though these oil droplets are squishy and can wiggle inside, in this specific type of storm, they behave very much like solid, rigid balls. The bigger they are, the more they ignore the tiny, chaotic jitters of the water and the more they stick to their own path.

The scientists proved that you can predict how these droplets move by treating them like solid particles, which makes modeling complex systems (like weather or engines) much easier for computers to handle.

Technical Summary

1. Problem Statement

Droplet-laden turbulent flows are ubiquitous in natural and industrial processes (e.g., combustion, rain formation, chemical mixing). A major challenge in modeling these systems is the intrinsic polydispersity of droplets, which arise from continuous deformation, coalescence, and breakup under turbulent strain.

While the dynamics of small, passive tracers and finite-size rigid particles are well-studied, there is a significant gap in understanding finite-sized, neutrally buoyant liquid droplets. Specifically:

• Most existing studies focus on rigid spheres or bubbles with tightly controlled, monodisperse sizes.
• Real-world droplets are deformable, possess internal circulation, and exist in naturally polydisperse populations generated by the turbulence itself.
• It remains unclear how the interplay of inertia, finite size, and deformability affects Lagrangian statistics (velocity/acceleration correlations and dispersion) compared to rigid particles.

2. Methodology

The authors conducted high-fidelity experiments using a custom-built facility and advanced tracking techniques.

• Experimental Facility:
  • A closed, "soccer-ball-like" chamber (approx. 40 cm diameter, 30 L volume) driven by 12 independently controlled propellers.
  • Generates highly Homogeneous Isotropic Turbulence (HIT).
  • Five Reynolds numbers ($Re_\lambda$) were tested, ranging from 179 to 287.
• Droplet Generation:
  • Silicone oil (neutrally buoyant, $\rho_d \approx \rho_c$) was injected into water at a low volume fraction ($\sim 0.1\%$) to ensure one-way coupling (droplets do not significantly alter the carrier flow).
  • Turbulence naturally breaks the injected oil into a polydisperse population of droplets.
• Measurement Technique:
  • 3D Particle Tracking Velocimetry (PTV): Four high-speed cameras captured droplet motion from multiple angles.
  • Ray-traversal Algorithm: Used to reconstruct 3D trajectories and determine the sphere-equivalent diameter ($D$) from voxel reconstructions of droplet silhouettes.
  • Resolution: Capable of tracking droplets with diameters ranging from $D \approx 2.5\eta$ to $D \approx 8.9\eta$ (where $\eta$ is the Kolmogorov length scale).
  • Data Volume: Trajectories spanned $\sim 87$ Kolmogorov timescales ($\tau_\eta$), ensuring converged statistics.

3. Key Contributions

1. Natural Polydispersity: Unlike previous studies using pre-selected rigid particles, this work utilizes turbulence to generate a naturally evolving, polydisperse droplet population, allowing for simultaneous measurement of size distribution and size-conditioned dynamics.
2. Deformable vs. Rigid Comparison: It provides the first experimental evidence comparing the Lagrangian statistics of deformable, neutrally buoyant droplets against theoretical predictions for rigid particles (incorporating Faxén corrections).
3. Scaling Laws: It establishes empirical scaling laws for droplet size distribution and dispersion timescales as a function of the Taylor-Reynolds number ($Re_\lambda$).

4. Key Results

A. Droplet Size Distribution

• Log-Normal Distribution: The droplet sizes follow a log-normal distribution.
• Reynolds Number Dependence: As $Re_\lambda$ increases, the mean droplet diameter ($\langle D \rangle$) decreases, and the distribution becomes narrower (more monodisperse).
  • Scaling: $\langle D \rangle \sim Re_\lambda^{-1.61}$ and $D_{peak} \sim Re_\lambda^{-1.41}$.
  • These exponents are close to the theoretical Kolmogorov-Hinze prediction ($Re_\lambda^{-1.2}$), though the observed mean diameter is significantly smaller than the theoretical Hinze scale ($d_H$), suggesting breakup occurs primarily in high-strain regions near the propellers rather than the bulk.
• Variance: The standard deviation of the size distribution ($\sigma_0$) decreases as $Re_\lambda^{-1.84}$, indicating stronger turbulence leads to more uniform droplet sizes.

B. Lagrangian Statistics (Velocity & Acceleration)

• Velocity PDFs: The probability density functions (PDFs) of velocity for finite-size droplets are nearly identical to those of passive tracers (Gaussian-like), showing only a very weak dependence on size.
• Acceleration PDFs: Acceleration PDFs also collapse onto a single curve for both tracers and droplets, well-described by a stretched-exponential function.
  • Finite-Size Effect: The tails of the droplet acceleration PDFs are slightly less pronounced than tracers, indicating that finite-size droplets act as spatial filters, smoothing out the most intense small-scale turbulent fluctuations.
  • Variance: The acceleration variance decreases slightly with increasing droplet size, consistent with Faxén-corrected numerical predictions for rigid particles.

C. Lagrangian Dynamics (Temporal Statistics)

• Velocity Autocorrelation: Larger droplets exhibit significantly longer velocity integral times ($T_v$) compared to tracers. The correlation decays more slowly as droplet size increases.
• Acceleration Autocorrelation: In contrast, acceleration autocorrelation times show negligible dependence on droplet size.
• Mean Squared Displacement (MSD):
  • MSD transitions from a ballistic regime ($\sim \tau^2$) to a diffusive regime ($\sim \tau$).
  • Extended Ballistic Regime: Larger droplets maintain the ballistic regime for a longer duration ($T_b$) before transitioning to diffusion.
  • Conclusion: Larger droplets retain their velocity memory longer due to increased inertia, behaving dynamically like rigid finite-size particles despite being deformable.

5. Significance and Conclusion

• Behavioral Equivalence: The study concludes that in the investigated size range ($D < 10\eta$), mildly deformable, neutrally buoyant droplets behave similarly to rigid finite-size particles regarding their Lagrangian dynamics. The primary finite-size effect is the spatial filtering of small-scale velocity gradients and the extension of the inertial (ballistic) regime.
• Limitations: The study notes that for much larger droplets where deformation and internal circulation become dominant, deviations from rigid-particle behavior are expected.
• Impact: This work validates the use of rigid-particle models (with Faxén corrections) for neutrally buoyant droplets in moderate turbulence. It also establishes a robust experimental platform for future investigations into droplet breakup, deformation, and interactions in dense emulsions, bridging the gap between theoretical models and complex industrial applications.