Session A8: Multiphase Flows: Turbulence I

A Direct Numerical Simulation of a Temporally Evolving Liquid-Gas Turbulent Mixing Layer

Air-blast atomization occurs when streams of co-flowing high speed gas and low speed liquid shear to form drops. Air-blast atomization has numerous industrial applications from combustion engines in jets to sprays used for medical coatings. The high Reynolds number and dynamic pressure ratio of a realistic air-blast atomization case requires large eddy simulation and the use of multiphase sub-grid scale (SGS) models. A direct numerical simulations (DNS) of a temporally evolving mixing layer is presented to be used as a base case from which future multiphase SGS models can be developed. To construct the liquid-gas mixing layer, half of a channel flow from Kim et al. (JFM, 1987) is placed on top of a static liquid layer that then evolves over time. The DNS is performed using a conservative finite volume incompressible multiphase flow solver where phase tracking is handled with a discretely conservative volume of fluid method. This study presents statistics on velocity and volume fraction at different Reynolds and Weber numbers.   

DNS of multifluid flows in a vertical channel undergoing topology changes

Multifluid flows in a vertical channel are examined by direct numerical simulations, for situations where the topology of the interface separating the different fluids changes. Several bubbles are initially placed in a turbulent channel flow at a sufficiently high void fraction so that the bubbles collide and the liquid film between them becomes very thin. This film is ruptured at a predetermined thickness and the bubbles are allowed to coalesce. For low Weber numbers the bubbles continue to coalesce, eventually forming one large bubble. At high Weber numbers, on the other hand, the large bubbles break up again, sometimes undergoing repeated coalescence and breakup. The evolution of various integral quantities, such as the average flow rate, wall-shear, and interface area are monitored and compared for different governing parameters. Various averages of the flow field and the phase distribution, over planes parallel to the walls, are examined and compared, and the microstructure of bubbles, at statistically steady state, is examined using low order probability functions.   

A DNS study on bubble-induced turbulence

Incompressible Eulerian-Eulerian two-fluid models, applied to gas-liquid flows, may experience the onset of non-physical instabilities. Recent works have shown that a dispersion term, which depends on the drag coefficient and the gradient of the gas volume fraction, is required to ensure the hyberbolicity of the PDEs. In the present work, a study on bubble-induced turbulence has been done to characterize this term. Three dimensional numerical simulations of nearly spheroidal bubbles deforming and rising in a quiescent liquid have been performed. Bubbles are initialized in a compact layer at the bottom of a tank in order to recreate a steep gradient of the volume fraction, which is locally moderate ($0.05 \le \alpha \le 0.15$), for the estimate of the dispersion term. Simulations are done with the one-fluid formulation together with a geometrical Volume-Of-Fluid (VOF) interface representation. Interfacial forces have been evaluated to propose closures for the dispersion coefficient. Moreover, the agitation in the liquid phase has been quantified, both inside and past the bubble swarm, evaluating the main scaling laws, expressed as functions of the characteristics of a single rising bubble. Results are compared with experimental works of similar configurations.   

Abstract Withdrawn

Our objective is to determine the effects of linear mean shear on the interaction of droplets with homogeneous turbulence. We performed DNS of 3130 finite-size droplets of diameter approximately equal to the Taylor lengthscale and with 5\% droplet volume fraction for droplet laden homogeneous shear turbulence at initial Taylor-scale Reynolds number Re$_\lambda$=75. We studied the effects of varying the Weber number and the shear number on the droplet/turbulence interaction. Following the derivation of the turbulence kinetic energy (TKE) equation for two-fluid flow by Dodd, Ferrante [J. Fluid Mechanics, Vol. 806, pp. 356-412], we derived the TKE equations for two-fluid flow for droplet laden homogeneous shear turbulence. We will present the numerical methods we used to simulate two-fluid homogeneous shear turbulence, as well as the effects on droplet/turbulence interaction caused by the mean shear.   

Simultaneous PLIF and PIV measurement of a near field turbulent immiscible buoyant oil jet fragmentation in water using liquid-liquid refractive index matching

Very little experimental data exits on the flow structure in the near field of a crude oil jet fragmenting in water because of inability to probe dense droplet cloud. Refractive index-matching is applied to overcome this challenge by using silicone oil and sugar water as a surrogate liquid pair. Their density ratio, viscosity ratio, and interfacial tension are closely matched with those of crude oil and seawater. Simultaneous PLIF and PIV measurements are conducted by fluorescently tagging the oil and seeding both phases with particles. With increasing jet Reynolds and Weber numbers, the oil plume breakup occurs closer to the nozzle, the spreading angle of the jet increases, and the droplet sizes decrease. The varying spread rate is attributed to differences in droplet size distributions. The location of primary oil breakup is consistent with the region of high strain rate fluctuations. What one may perceive as oil droplets in opaque fluids actually consists of multi-layers containing water droplets, which sometimes encapsulate smaller oil droplets, creating a ``Russian Doll'' like phenomenon. This system forms as ligaments of oil and water wrap around each other during entrainment. Results include profiles of mean velocity and turbulence parameters along with energy spectra.   

Turbulence intensity's effect on liquid jet breakup from long circular pipes

Long pipes which produce fully developed flow are frequently used as a nozzle in jet breakup research. We compiled experimental data from over 20 pipe jet studies for many breakup quantities and developed correlations for these quantities based on existing theories and our own theories. Previous experimental studies often had confounding between some variables (e.g., the Reynolds and Weber numbers), neglected important quantities (e.g., the turbulence intensity), or made apples to oranges comparisons (e.g., different nozzles). By independently tracking the Reynolds number, Weber number, density ratio, and turbulence intensity, and focusing only on pipe jets to keep other variables nearly constant, we minimize these issues. Turbulence is a cause of jet breakup, yet there is little quantitative research on this due to the difficulty of turbulence measurements in free surface flows. To avoid those difficulties, we exploited the fact that adjusting the roughness of a long pipe allows one to quantifiably control the turbulence intensity. We correlated turbulence intensity as a function of the friction factor. Data for rough pipes was used to include turbulence intensity in our study. Comparisons were made with theories for the effect of turbulence intensity on breakup.   

A Dual Scale Approach for Modeling Turbulent Liquid/Gas Phase Interfaces

Advances to a dual-scale modeling approach are presented to describe turbulent phase interface dynamics in a large-eddy-simulation-type spatial filtering context. Spatial filtering of the governing equations introduces several sub-filter terms that require modeling. Instead of developing individual closure-models for the terms associated with the interface, the dual-scale approach uses an exact closure by explicitly filtering a fully resolved realization of the phase interface. This resolved realization is maintained on a high-resolution over-set mesh using a Refined Local Surface Grid approach. The advection equation for the phase interface on this DNS scale requires a model for the fully resolved interface advection velocity. This velocity is the sum of the filter scale LES velocity, available from the LES flow solver, and the sub-filter velocity fluctuation that has two contributions. The first is due to sub-filter turbulent eddies, reconstructed using a local fractal interpolation technique (Scotti & Meneveau,1999), and the second is due to sub-filter surface tension forces, reconstructed using a local Taylor analogy approach. Results of the dual-scale model are compared to recent DNS of interfaces in homogeneous isotropic turbulence (Chiodi and Desjardins, 2017).