Bulletin of the American Physical Society
67th Annual Meeting of the APS Division of Fluid Dynamics
Volume 59, Number 20
Sunday–Tuesday, November 23–25, 2014; San Francisco, California
Session D33: Computational Methods and Modeling of Particle Laden Flows |
Hide Abstracts |
Chair: Marc Massot, Ecole Centrale Paris Room: 2022 |
Sunday, November 23, 2014 2:15PM - 2:28PM |
D33.00001: Kinetic-Based Moment Methods for DNS and LES of particle-laden flows: the Anisotropic Gaussian Closure Sabat Macole, Aymeric Vi\'e, Adam Larat, Francois Doisneau, Christophe Chalons, Marc Massot The simulation of particle-laden flows is a challenging topic due to their multiscale character. Lagrangian particle tracking methods are classically used. However, for high performance computing, such approaches deteriorate with the disperse phase inhomogeneities. Moment methods bypass this issue through an Eulerian framework allowing to use the same parallelization paradigm as the gas phase. We present recent developments for DNS and LES based on a Kinetic-Based Moment Method. The moment system is closed by assuming a presumed shape for the NDF. The selected NDF is an Anisotropic Gaussian giving the following properties: 1/ hyperbolicity; 2/ realizability of the moments; 3/maximization of entropy; 4/ H-theorem. The method is evaluated on configurations of increasing complexity that exhibit its potential and drawbacks. This method extends towards LES by means of a full kinetic-based filtering technique instead of filtering the moment equations. Thus realizability conditions are easily derived, and the main properties of the DNS system are preserved. The subgrid terms are closed following the work of Zaichik et al. 2009. The resulting LES strategy is evaluated based on filtered DNS results. [Preview Abstract] |
Sunday, November 23, 2014 2:28PM - 2:41PM |
D33.00002: A Dual-Scale Approach for Modeling Turbulent Two-Phase Interface Dynamics Marcus Herrmann Turbulent liquid/gas phase interface dynamics are at the core of many applicatons. For example, in atomizing flows, the properties of the resulting liquid spray are determined by the interplay of fluid and surface tension forces. The resulting dynamics typically span 4-6 orders of magnitude in length scales, making DNS exceedingly expensive. This motivates the need for modeling approaches based on spatial filtering or ensemble averaging. In this talk, a dual-scale modeling approach is presented to describe turbulent two-phase interface dynamics in a LES-type spatial filtering context. To close the unclosed terms related to the phase interface arising from filtering the Navier-Stokes equation, a resolved realization of the phase interface dynamics is explicitly filtered. This resolved realization is maintained on a high resolution over-set mesh using a Refined Local Surface Grid approach employing an un-split, geometric, bounded, and conservative Volume of Fluid method. The required model for the resolved realization of the interface advection velocity includes the effects of sub-filter surface tension, dissipation, and turbulent eddies. Results of the dual-scale model will be compared to recent DNS by McCaslin \& Desjardins of an interface in homogeneous isotropic turbulence. [Preview Abstract] |
Sunday, November 23, 2014 2:41PM - 2:54PM |
D33.00003: ABSTRACT WITHDRAWN |
Sunday, November 23, 2014 2:54PM - 3:07PM |
D33.00004: Immersed Boundary Methods on Non-Uniform Grids for Simulation of a Fully Resolved Bed of Particles in a Near-Wall Turbulent Flow Georges Akiki, S. Balachandar This study presents a dynamic distribution of the Lagrangian markers on a sphere when using the immersed boundary method with a non-uniform Eulerian mesh. The points are distributed in accordance with the surrounding Eulerian mesh to keep it optimized as the sphere moves in the channel. Also, a method is proposed to assign weights to the Lagrangian markers, both in the case of uniform and non-uniform distribution of the points. The newly proposed method of weight assignments uses vector spherical harmonics expansion of the weights. The error due to uneven distribution of the Lagrangian points is significantly reduced. These methods are then validated and applied in the simulation of a fully resolved bed of particles in a wall-bounded turbulent flow with periodic boundary conditions along the streamwise and spanwise directions. Results are analyzed for understanding of both effect of wall turbulence on particle motion and interaction, and the back effect of particles on the carrier-phase turbulence. [Preview Abstract] |
Sunday, November 23, 2014 3:07PM - 3:20PM |
D33.00005: An immersed boundary method for the interaction of turbulence with particles of arbitrary shape Shizhao Wang, Marcos Vanella, Elias Balaras In this work we present a computational scheme applicable to turbulence/particle interactions, targeting applications involving millions of particles of arbitrary shape. Immersed boundary methods have been frequently applied in simulating such problems, but are usually confined to spherical particles. Extension to rigid/deformable particles of arbitrary shape introduces significant challenges in achieving parallel efficiency. The proposed method is based on the moving least squares immersed boundary approach (Vanella \& Balaras, J. Comput. Physics, 228(18), 6617-6628, 2009) on uniform and adaptive block-structured grids. We will present a novel parallelization strategy based on a master/slave model: the processor on which a body/structure resides is designated the master processor, while all the processors that contain at least one block overlapping with the body are designated the slaves. As the particle moves through the fluid, its blocks association and therefore the participating processors change. Effective ways of replicating the mesh metadata on all processors will be discussed. Results for homogeneous turbulence interacting with spherical and ellipsoidal particles and comparisons with experimental results will be given. [Preview Abstract] |
Sunday, November 23, 2014 3:20PM - 3:33PM |
D33.00006: An improved numerical method for the kernel density functional estimation of disperse flow Timothy Smith, Reetesh Ranjan, Carlos Pantano We present an improved numerical method to solve the transport equation for the one-point particle density function (pdf), which can be used to model disperse flows. The transport equation, a hyperbolic partial differential equation (PDE) with a source term, is derived from the Lagrangian equations for a dilute particle system by treating position and velocity as state-space variables. The method approximates the pdf by a discrete mixture of kernel density functions (KDFs) with space and time varying parameters and performs a global Rayleigh-Ritz like least-square minimization on the state-space of velocity. Such an approximation leads to a hyperbolic system of PDEs for the KDF parameters that cannot be written completely in conservation form. This system is solved using a numerical method that is path-consistent, according to the theory of non-conservative hyperbolic equations. The resulting formulation is a Roe-like update that utilizes the local eigensystem information of the linearized system of PDEs. We will present the formulation of the base method, its higher-order extension and further regularization to demonstrate that the method can predict statistics of disperse flows in an accurate, consistent and efficient manner. [Preview Abstract] |
Sunday, November 23, 2014 3:33PM - 3:46PM |
D33.00007: An accurate and efficient Lagrangian sub-grid model for multi-particle dispersion Federico Toschi, Irene Mazzitelli, Alessandra S. Lanotte Many natural and industrial processes involve the dispersion of particle in turbulent flows. Despite recent theoretical progresses in the understanding of particle dynamics in simple turbulent flows, complex geometries often call for numerical approaches based on eulerian Large Eddy Simulation (LES). One important issue related to the Lagrangian integration of tracers in under-resolved velocity fields is connected to the lack of spatial correlations at unresolved scales. Here we propose a computationally efficient Lagrangian model for the sub-grid velocity of tracers dispersed in statistically homogeneous and isotropic turbulent flows. The model incorporates the multi-scale nature of turbulent temporal and spatial correlations that are essential to correctly reproduce the dynamics of multi-particle dispersion. The new model is able to describe the Lagrangian temporal and spatial correlations in clouds of particles. In particular we show that pairs and tetrads dispersion compare well with results from Direct Numerical Simulations of statistically isotropic and homogeneous 3d turbulence. This model may offer an accurate and efficient way to describe multi-particle dispersion in under resolved turbulent velocity fields such as the one employed in eulerian LES. [Preview Abstract] |
Sunday, November 23, 2014 3:46PM - 3:59PM |
D33.00008: An unstructured overset method for particle-resolved simulation of particle-laden flows Wyatt Horne, Krishnan Mahesh Particle-laden flows involve a large range of length scales, ranging from the larger convective length scales down to length scales smaller than particle size. We develop a particle-resolved direct-numerical simulation (PR-DNS) method to enable the accurate study of the physics of particle-laden flow at particle length scales. Unstructured meshes are attached directly to particle surfaces and to the background flow field. The different meshes are allowed to arbitrarily overlap with each other to create a single cohesive solution. A dynamic connectivity procedure is used that cuts solid bodies out of each mesh and establishes interpolation pairs between overlapping meshes. The flow is incompressible, and the numerical method is based on that developed by Mahesh et al. [J. Comput. Phys. (2004) 197:215-240]. The overall discrete conservation properties for mass, momentum and kinetic energy are analyzed. Several cases are presented showing the method's efficacy for studying particle-laden flow including single particle results and particle-to-particle interaction. [Preview Abstract] |
Sunday, November 23, 2014 3:59PM - 4:12PM |
D33.00009: LES of box turbulence with particles: SGS modeling of the particle acceleration Remi Zamansky, Mikhael Gorokhovski When the Reynolds number is high, the turbulent flow on small length scales is characterized by strong velocity gradients. If such a flow is laden by inertial particles, those gradients, or specifically the turbulent time-scales shorter than the Stokes time, induce fluctuations in the particle motion. In LES, this motivates to simulate the interaction of particle with SGS flow. In our LES of box turbulence with particles, the particle acceleration was decomposed on its resolved and residual parts. The latter was assumed resulting from interactions in the inertial range, and was simulated stochastically along the particle trajectory. It was done by two processes, one for its norm, and another for its direction. Results showed that by introducing the stochastic model for the particle residual acceleration, the particle acceleration statistics from DNS was predicted fairly well. We also proposed the stochastic model for particles bigger than the Kolmogorov size. To this end, the fluctuating drag was derived, and simulated by lognormal process. This model predicted experimental observation: stretched tails in the particle acceleration distribution invariantly to the density and the size of particle. [Preview Abstract] |
Sunday, November 23, 2014 4:12PM - 4:25PM |
D33.00010: Filter length scale for continuum modeling of subgrid physics Julian Simeonov, Joseph Calantoni Modeling the wide range of scales of geophysical processes with direct numerical simulations (DNS) is currently not feasible. It is therefore typical to explicitly resolve only the large energy-containing scales and to parameterize the unresolved small scales. One approach to separate the scales is by means of spatial filters and here we discuss practical considerations regarding the choice of a volume averaging scale $L$. We use a macroscopically homogeneous scalar field and quantify the smoothness of the filtered field using a noise metric, $\nu$, defined by the standard deviation of the filtered field normalized by the domain-averaged value of the field. For illustration, we consider the continuum modeling of the particle phase in discrete element method (DEM) simulations and the salt fingers in DNS of double-diffusive convection. We find that $\nu^2$ follows an inverse power law dependence on $L$ with an exponent and coefficients proportional to the domain-averaged field value. The empirical power law relation can aid in the development of continuum models from fully resolved simulations while also providing uncertainty estimates of the modeled continuum fields. [Preview Abstract] |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700