Session H22: Multiphase Flows: Particle Laden Flows I

Deposition of inertial particles in turbulent channel flows

From gas turbine fouling to the transport of aerosolized pollutants in human airways, deposition of particulate matter in a turbulent channel has a wide range of applications. Although the 1D deposition model gives reasonably accurate result to the deposition velocity, the process is sensitive to many effects, including humidity, particle charges, and particle inertia. We have designed and constructed a vertical channel to investigate the deposition process of inertial particles. This system enables us to independently control the turbulence, humidity, and temperature of the air. Several different diagnostic methods have also been implemented to cover a wide range of length and time scales that are relevant to different aspects of the problem. Taken together, these measurements pave the way for a deeper understanding of the underlying physical mechanisms driving particle deposition.   

Two-phase flow measurements using PIV/PTV on an under expanded particle-laden jet

The objective of this study is to understand the dynamics of a high-speed particle-laden under-expanded jet motivated by landings on extraterrestrial bodies. In this setup, inertial particles are entrained and accelerated by an under-expanded jet. But, due to their inertia, the particle velocity is significantly lower than that of the surrounding gas, so the two phases are coupled through aerodynamic drag. Sub-micron oil droplets are entrained upstream before the inertial particles are introduced, whose velocity is determined through a PIV system; inertial particles, after image segmenting is performed to separate them from PIV data, are tracked over time using a PTV system. The results will help to understand particle-laden flow in a new regime where the background flow is compressible and the Mach number based on the slip velocity is not negligible, which may help to pave a foundation for future studies in compressible multiphase flows.   

On the effects of subgrid dispersion in large-eddy simulation of particle-laden turbulent channel flow bounded by rough walls

Particle-laden turbulent flows observed in many engineering and environmental applications are usually bounded by rough walls. In such flows, the wall roughness affects the particle-turbulence interactions by altering the dynamics of the wall-bounded turbulence. The Eulerian-Lagrangian (EL) point-particle approach is a popular computational strategy for the study of such flows, where the dispersed particles are tracked in a Lagrangian manner, and the continuous carrier phase is evolved in the Eulerian frame. In this study, the EL framework is used to perform a large-eddy simulation (LES) of particle-laden turbulent flow bounded by rough walls under dilute suspension conditions. A particular emphasis of this study is on examining the effects of subgrid dispersion modeling on the statistics of the dispersed and carrier phases. The assessment is carried out by considering periodic turbulent channel flow at frictional Reynolds number of 180 with particles of different inertia under the one-way coupled regime and comparing LES results with results from reference direct numerical simulations. To assess the effects of subgrid dispersion while performing LES, three different models are considered, namely, a no-model approach, a random walk model, and an approximate deconvolution model.   

Intercomparison of Model Simulations of Cloudy Rayleigh-Bénard Convection in a Laboratory Chamber

The Pi Chamber at the Michigan Technological University is a cloud chamber (1 m high x 2 m x 2 m) that can support cloudy Rayleigh-Bénard convection. Supersaturation is produced by turbulent mixing of air saturated at the (cold) top and (warm) bottom wall temperatures. Droplets grow on continuously injected aerosol particles, and eventually fall out due to sedimentation. A statistically steady state is obtained for steady boundary conditions. By varying the aerosol injection rate, the injected aerosol sizes, and the temperature gradient in different experiments, a range of droplet size distributions (DSDs) is obtained. Theoretical analyses demonstrate that the DSD shape (i.e., the droplet size PDF) depends on both the mean and the variance of the supersaturation (SS). However, the PDFs for the extreme cases of no mean SS or no variance of SS do not differ much, which makes experimental discrimination between cases difficult based on the PDF alone. To help elucidate the relative roles of mean SS and variance of SS, a model intercomparison study has been performed as part of the 10th International Cloud Modelling Workshop. Seven models were involved: Three large-eddy simulation models, two direct numerical simulation models, and a 1D linear eddy model.   

Modeling of Particle Layer Deposition, Shear, Saltation and Heat Transfer in a High Mass Loading Gas-Particle Flow

We extend our previously published Eulerian model for the heat transfer of a highly mass loaded high Reynolds number particle-laden pipe flow. In high pressure co-flowing nitrogen-copper powder experiments that preceded this effort, it was observed that copper particles accumulated on the bottom of flow channel. Accordingly, we extend the two-field (gas+disperse particles) model to a three-field model (gas+disperse particles+wall bounded particle layer). This requires incorporation of deposition and saltation mass transfer models between the disperse and near-continuous layer fields. A drag model to account for the drag between this particle layer and the carrier gas is introduced, as is an attendant interfacial heat transfer model. In this presentation we provide details of the complete three-field model with emphasis on the particle layer mass, momentum and energy inter-field transfer. We apply the 3-field model and compare its Nusselt number performance to a two-field model that neglects the copper powder buildup, and experimental measurements carried out by our group.   

Detailed Validation of Ejecta Transport Models

Recent experiments performed at Los Alamos National Laboratory have shown that metal particles ejected from a shocked surface show vastly different behavior when the medium they are ejected into is inert or reactive. For particles ejected into an inert medium, the particles appear to breakup in a standard liquid droplet breakup mechanism. However, if the metal particles are ejected into a reactive medium, they exhibit what appears to be delayed break-up behavior where they almost breakup in batches. The physical processes which dictate this phenomenon are mainly unknown and are of interest to those who wish to accurately model the trajectories of the ejecta after they form. This work shows preliminary simulations of the non-reacting ejecta experiment configurations. The simulation results are compared to experimental data to validate the ejecta transport models, such as the mass sourcing and multiphase momentum and energy transfer, which are currently implemented in the hydrocode. The goal is to demonstrate that these models are capable of capturing the details of the inert ejecta experiments and are a suitable starting point for the future modeling of the reactive ejecta phenomena.   

Numerical Simulations of Granular bed Erosion Under Martian Conditions

We present high-resolution Eulerian–Lagrangian simulations of sheared granular beds under Martian conditions. Erosion of the granular material on the surface of the Mars is important to understand the risks associated with plume-surface interactions (PSI) during spacecraft landing, in addition to understanding observed aeolian sediment transport. While much work has been done to characterize saltation and erosion, giving rise to phenomenological scaling laws, previous studies are almost entirely centered around the sub-aqueous regime (i.e. particle-to-fluid densityratios of O(1)) and in the incompressible limit. Unlike the more classically studied sub-aqueous sediment transport on Earth, Martian conditions exhibit density ratios of O(10⁵), and tangential flow induced by rocket plumes during landing events can reach supersonic speeds. In addition, recent experiments have found lower pressure environments yield a lower than expected saltation threshold. The present work focuses on erosion regimes under PSI-relevant Martian conditions with an emphasis on Mach number effects.   

Accounting for Pseudoturbulence in Eulerian-Lagrangian Simulations of Liquid-Solid Beds

An Eulerian-Lagrangian (EL) model is used for the simulation of liquid-solid flows to capture important features of the flow when particle trajectory crossing plays an important role. The basic EL model does not resolve pseudoturbulence in fluid, partially resolves coupling with particle velocity fluctuations, and thus requires closures for fluid velocity fluctuations and particle velocity fluctuations. A closure for fluid velocity fluctuations extracted from particle-resolved direct-numerical simulation is implemented in an EL code. A velocity Langevin model is then proposed for the effect of the fluid-phase pseudoturbulence on the particle velocity. When the fluid density is significant relative to the particle density, particle-particle interactions mediated by the fluid give rise to a non-negligible force in the particle equation of motion, which is called fluid-mediated particle-particle (PFP) force. The simulation results exhibit the significant contribution of pseudoturbulence and PFP terms.   

Inertial focusing of spherical and non-spherical particle in curved channel

We report an experimental investigation on migration and focusing of spherical and non-spherical particles in a curved channel. Experiments were performed for a range of Reynolds number 267 to 1571 and corresponding Dean number 36 to 216. Three types of particles, spherical, cylindrical and cubical, with volume equivalent diameter to channel dimension ratio of 0.05 to 0.1, were used for experiments. The suspension was very dilute so that inter-particle hydrodynamic interaction could be neglected. Particles are found to be focused in two elliptical annuls at the upper and lower half of the channel cross-section. The probability distribution function of the particle distribution has been calculated and the spread of the focusing zone has been characterized as a function of the Dean number and shape of particles.   

Optimising graphene exfoliation via numerical studies of particle-laden Taylor-Couette flow

In this study, a novel experimental rig is used to perform liquid phase exfoliation of graphite in order to produce graphene. This predominantly makes use of Taylor-Couette flow, with exfoliation taking place in the narrow gap between a stationary outer cylinder and rotating inner one. Numerical simulations of the flow with particles are performed, utilising a point particle approach, at a range of rotational speeds and particle sizes. This allows one to identify regions of high particle concentration and shear rates; regions where these coexist are where exfoliation is maximised. Our results suggest that larger particles migrate towards the outer cylinder wall preferentially, where shear rates are high due to the large velocity gradient in the near wall region. These results complement experimental studies, where utilisation of larger graphite flakes has generated higher graphene concentrations.