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 L33: Computational Methods and Modeling of Multiphase Flows I |
Hide Abstracts |
Chair: Marcus Herrmann, Arizona State University Room: 2022 |
Monday, November 24, 2014 3:35PM - 3:48PM |
L33.00001: Numerical Modelling of Three-Fluid Flow Using The Level-set Method Hongying Li, Jing Lou, Zhi Shang This work presents a numerical model for simulation of three-fluid flow involving two different moving interfaces. These interfaces are captured using the level-set method via two different level-set functions. A combined formulation with only one set of conservation equations for the whole physical domain, consisting of the three different immiscible fluids, is employed. Numerical solution is performed on a fixed mesh using the finite volume method. Surface tension effect is incorporated using the Continuum Surface Force model. Validation of the present model is made against available results for stratified flow and rising bubble in a container with a free surface. Applications of the present model are demonstrated by a variety of three-fluid flow systems including (1) three-fluid stratified flow, (2) two-fluid stratified flow carrying the third fluid in the form of drops and (3) simultaneous rising and settling of two drops in a stationary third fluid. [Preview Abstract] |
Monday, November 24, 2014 3:48PM - 4:01PM |
L33.00002: ABSTRACT WITHDRAWN |
Monday, November 24, 2014 4:01PM - 4:14PM |
L33.00003: DNS and modeling of bubbly flows in vertical channels Ming Ma, Jiacai Lu, Gretar Tryggvason The transient motion of bubbly flow, in a vertical channel is studied, using direct numerical simulations (DNS) where every continuum length and time scale is resolved. Nearly spherical bubbles of the same size, injected into laminar upflow, are quickly pushed to the walls due to lift. The velocity then slows down, eventually resulting in some of the bubbles returning to the core forming a mixture where the weight matches the imposed pressure gradient and the void fraction is easily predicted. Unlike the statistically steady state, where the flow structure is relatively simple and in some cases depends only on the sign of the lift coefficient, the transient evolution is more sensitive to the governing parameters. The DNS results are used to provide values for the unresolved closure terms in a simple average model for the flow, found by mining the data, using various techniques such as regression and neural networks. Results for a large number of bubbles of several different sizes in turbulent upflow are also presented and the prospects of using a similar approach for LES-like simulations of more complex flows are discussed, including the simplification of the interface structure resulting from filtering. [Preview Abstract] |
Monday, November 24, 2014 4:14PM - 4:27PM |
L33.00004: ABSTRACT WITHDRAWN |
Monday, November 24, 2014 4:27PM - 4:40PM |
L33.00005: Comparisons and Limitations of Gradient Augmented Level Set and Algebraic Volume of Fluid Methods Lakshman Anumolu, Douglas Ryddner, Mario Trujillo Recent numerical methods for implicit interface transport are generally presented as enjoying higher order of spatial-temporal convergence when compared to classical methods or less sophisticated approaches. However, when applied to test cases, which are designed to simulate practical industrial conditions, significant reduction in convergence is observed in higher-order methods, whereas for the less sophisticated approaches same convergence is achieved but a growth in the error norms occurs. This provides an opportunity to understand the underlying issues which causes this decrease in accuracy in both types of methods. As an example we consider the Gradient Augmented Level Set method (GALS) and a variant of the Volume of Fluid (VoF) method in our study. Results show that while both methods do suffer from a loss of accuracy, it is the higher order method that suffers more. The implication is a significant reduction in the performance advantage of the GALS method over the VoF scheme. Reasons for this lie in the behavior of the higher order derivatives, particular in situations where the level set field is highly distorted. For the VoF approach, serious spurious deformations of the interface are observed, albeit with a deceptive zero loss of mass. [Preview Abstract] |
Monday, November 24, 2014 4:40PM - 4:53PM |
L33.00006: Development of high order numerical methods for particle-laden flows on unstructured grids: A realizability-preserving Discontinuous Galerkin method for moderate Stokes number flows Adam Larat, Macole Sabat, Aymeric Vi\'e, Christophe Chalons, Marc Massot The simulation of particle-laden flows is of primary importance for several industrial applications, like sprays in aeronautical combustors or particles in fluidized beds. Our focus is on Moment methods that describes the disperse phase as a continuum. The accuracy and performance of such approaches highly depends on the number of controlled moments for correctly describing the physics of the flow, but also on the numerics that are used to solve the continuous system of equations at a discrete level. In the present work, we investigate the use of Discontinuous Galerkin methods to solve for the convective part of the moment equations. By deriving realizability conditions on the moment system that are associated to a convex space, a projection strategy is used to maintain the solution in the realizable space. This method is applied to the resolution of the pressure less gas dynamics and the Anisotropic Gaussian moment approach, the former solving for low Stokes number flows where no Particle Trajectory Crossing occurs, while the latter is solving for moderate Stokes number flows and can handle PTC through a pressure tensor in the convective term. The strategy is assessed on turbulent flows through comparisons with Lagrangian results. [Preview Abstract] |
Monday, November 24, 2014 4:53PM - 5:06PM |
L33.00007: Horizontal annular flow modelling using a compositional based interface capturing approach Dimitrios Pavlidis, Zhizhua Xie, James Percival, Jefferson Gomes, Chris Pain, Omar Matar Progress on a consistent approach for interface-capturing in which each component represents a different phase/fluid is described. The aim is to develop a general multi-phase modelling approach based on fully-unstructured meshes that can exploit the latest mesh adaptivity methods, and in which each fluid phase may have a number of components. The method is compared against experimental results for a collapsing water column test case and a convergence study is performed. A number of numerical test cases are undertaken to demonstrate the method's ability to model arbitrary numbers of phases with arbitrary equations of state. The method is then used to simulate horizontal annular flows. [Preview Abstract] |
Monday, November 24, 2014 5:06PM - 5:19PM |
L33.00008: Accurate VoF based curvature evaluation method for low-resolution interface geometries Mark Owkes, Marcus Herrmann, Olivier Desjardins The height function method is a common approach to compute the curvature of a gas-liquid interface in the context of the volume-of-fluid method. While the approach has been shown to produce second-order curvature estimates for many interfaces, the height function method deteriorates when the curvature becomes large and the interface becomes under-resolved by the computational mesh. In this work, we propose a modification to the height function method that improves the curvature calculation for under-resolved structures. The proposed scheme computes heights within columns that are not aligned with the underlying computational mesh but rather the interface normal vector which are found to be more robust for under-resolved interfaces. A computational geometry toolbox is used to compute the heights in the complex geometry that is formed at the intersection of the computational mesh and the columns. The resulting scheme has significantly reduced curvature errors for under-resolved interfaces and recovers the second-order convergence of the standard height function method for well-resolved interfaces. [Preview Abstract] |
Monday, November 24, 2014 5:19PM - 5:32PM |
L33.00009: Accurate curvature estimates from volume fractions on unstructured meshes using embedded height functions Christopher Ivey, Parviz Moin A novel methodology for extracting curvatures from volume fractions on non-convex, unstructured grids is presented. Estimating curvature in volume of fluid methods is difficult due the discontinuous nature of the volume fraction field. On simple structured meshes, height functions can be used to map the volume fraction field to a surface height field that is smooth along the surface. Our algorithm utilizes a local cartesian mesh and a suitable interpolation strategy to harness the height function technique developed for uniform meshes. Accuracy of the algorithm is demonstrated through comparison with the with the reconstructed distance function method on unstructured meshes and with the traditional height function method on uniform meshes of similar grid density. [Preview Abstract] |
Monday, November 24, 2014 5:32PM - 5:45PM |
L33.00010: Development of Numerical Method for Two-phase Flows on Three-dimensional Arbitrarily-shaped Polyhedral Meshes Kohei Suzuki, Takesi Omori, Takeo Kajishima Although the advantage of using arbitrarily-shaped polyhedral meshes for the industrial flow applications is clear, their employment to two-phase flows is rather limited due to the poor prediction accuracy of the existing numerical methods on such meshes. We present a numerical method based on VOF (Volume of Fluid) method which works on arbitrarily-shaped three-dimensional polyhedral meshes with little volume/shape error for the interface advection and with little curvature estimation error. To make the implementation in three-dimensional geometry feasible, we extend THINC (Tangent of Hyperbola Interface Capturing) method for polyhedral meshes which does not require laborious geometric arithmetics. In the oral presentation we will also show that the combination of RDF (Reconstructed Distance Function) algorithm and the carefully selected discretization procedure gives good performance in the interface curvature estimation. [Preview Abstract] |
Monday, November 24, 2014 5:45PM - 5:58PM |
L33.00011: A Numerical Study of Mesh Adaptivity in Multiphase Flows with Non-Newtonian Fluids James Percival, Dimitrios Pavlidis, Zhihua Xie, Federico Alberini, Mark Simmons, Christopher Pain, Omar Matar We present an investigation into the computational efficiency benefits of dynamic mesh adaptivity in the numerical simulation of transient multiphase fluid flow problems involving Non-Newtonian fluids. Such fluids appear in a range of industrial applications, from printing inks to toothpastes and introduce new challenges for mesh adaptivity due to the additional ``memory'' of viscoelastic fluids. Nevertheless, the multiscale nature of these flows implies huge potential benefits for a successful implementation. The study is performed using the open source package Fluidity, which couples an unstructured mesh control volume finite element solver for the multiphase Navier-Stokes equations to a dynamic anisotropic mesh adaptivity algorithm, based on estimated solution interpolation error criteria, and conservative mesh-to-mesh interpolation routine. The code is applied to problems involving rheologies ranging from simple Newtonian to shear-thinning to viscoelastic materials and verified against experimental data for various industrial and microfluidic flows. [Preview Abstract] |
Monday, November 24, 2014 5:58PM - 6:11PM |
L33.00012: A Multiscale FSI Analysis of Flow past a Cylinder James Chen, Maurin Lopez Micropolar fluid theory generalizes classical continuum mechanics by incorporating the microscale spinning effects of fluid molecules. It has been seen that Micropolar fluid theory shows strong promises of predicting the microscale fluid behaviours in the continuum level. With the two-level motion in Micropolar theory, the interaction between macromotion and micromotion in the fluid flows can be utilized to interpret flow phenomena. It is understood that the classical fluid theory is not fully capable of explaining fluids phenomena involving energy dissipation across multiple length scales from theoretical perspective. Such phenomena include vortex formation, boundary layer development and etc. Flow past a cylinder is studied as an example. An in-house developed solver based in a high order spectral difference method to solve the Micropolar equations with moving and deformable grids for fluid-solid interaction (FSI) is used. By studying how the translational velocity (macromotion) dissipates into gyration (micromotion) it is possible to understand how the energy cascade into smaller scales for vortex formation, this mechanism explains how vortices form and how the coherent structures of vortices and eddies construct. [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