Bulletin of the American Physical Society
65th Annual Meeting of the APS Division of Fluid Dynamics
Volume 57, Number 17
Sunday–Tuesday, November 18–20, 2012; San Diego, California
Session H3: Multiphase Flows: Numerical Methods II |
Hide Abstracts |
Chair: Marcus Herrmann, Arizona State University Room: 23B |
Monday, November 19, 2012 10:30AM - 10:43AM |
H3.00001: Computation of dendritic crystal growth in supercooled water using a level-set method Antonio Criscione, Daniel Kintea, Ilia V. Roisman, Suad Jakirlic, Zeljko Tukovic, Cameron Tropea Over the last decades various computational approaches have been developed to simulate solidification and interfacial pattern formation phenomena. For the solution of the time-dependent moving boundary problem, which governs the dendritic crystal growth, the phase-field method has been usually used in simulations to avoid the difficulties of tracking a sharp boundary. In the present work a computational model is developed by using a level-set method. The relevant numerical algorithm is implemented into the open source software OpenFOAM. The heat transfer equations are solved in both the liquid and solid phase independently from each other. At the interface a Dirichlet boundary condition for the temperature field is imposed and a ghost-face method is applied to ensure accurate calculation of the normal derivative needed for the jump condition, i.e. for the interface-velocity in the normal direction. For the sake of updating the level set function a narrow-band around the interface is introduced. Within this band, whose width is temporally adjusted to the maximum curvature of the interface, the normal-to-interface velocity is appropriately expanded. The steady-state results are in agreement with the microscopic solvability theory. [Preview Abstract] |
Monday, November 19, 2012 10:43AM - 10:56AM |
H3.00002: Direct numerical simulation of the Leidenfrost Effect Lucia Rueda Villegas, S\'ebastien Tanguy We present direct numerical simulations of the impact of a single droplet on a heated flat surface in the Leidenfrost regime. To that end, we solve the Navier-Stokes equations, the energy equation, and the species mass fraction equation. The Level Set method is used to track the liquid-gas interface motion and the Ghost Fluid Method is implemented to treat the jump conditions. To get rid of the temporal stability condition due to viscosity, an implicit temporal discretization is used. Some specific numerical methods have been developed to deal with droplet vaporization interface jump conditions. Since the vapor layer is very thin compared to the droplet size, a non-uniform structured grid strongly refined near the wall is used to capture the droplet bounce. We present numerical simulations that enable us to study accurately the bouncing dynamics by analyzing the momentum balance during the droplet bounce. Moreover, we determine from such computation the ratio of the droplet heat transfer flux by comparing the energy used for the phase change (latent heat) to the energy used for droplet heating (specific heat). We then compare the shape of the droplet during the impact with some experimental results. [Preview Abstract] |
Monday, November 19, 2012 10:56AM - 11:09AM |
H3.00003: Gradient Augmented Level Set Reinitialization Approach Lakshman Anumolu, Mario Trujillo A new reinitialization technique in the framework of augmented level set methods is proposed. The reinitialization PDE introduced by Sussman et al. to reconstruct signed distance function is reformulated into gradient augmented framework and solved using semi-Lagrangian approach. In order to address the issue of interfacial drift we introduce a hybrid strategy, in which this erroneous drift is handled well by anchoring the interface. In this approach, two regions, interfacial and non-interfacial regions are identified in the computational domain, where the level set and its gradient values are updated explicitly by locating the interface for the nodes belonging to the interfacial region. Two approaches are followed to identify the interface, of which, one uses the underlying Hermite polynomial evaluated along the characteristic curve, and the other uses the variant of Newton's method proposed by Chopp. Results show 4$^{th}$ and 3$^{rd}$ order spatial convergence rate for the level set function and its gradient respectively. Effect of temporal schemes is also studied with two temporal schemes, first order Euler and third order RK schemes. Unlike the numerical oscillations noted by Min in their work with Euler scheme, we have not noticed them in the presented hybrid scheme. [Preview Abstract] |
Monday, November 19, 2012 11:09AM - 11:22AM |
H3.00004: A Quadrature-Free Conservative Level Set RKDG for Simulating Atomization Zechariah Jibben, Marcus Herrmann We present an arbitrary high-order, quadrature-free, Runge-Kutta discontinuous Galerkin (RKDG) method for the solution of the conservative level set equation (Olsson et al., 2007), used for capturing phase interfaces in atomizing multiphase flows. Special care is taken to maintain high-order accuracy in the reinitialization equation, using appropriate slope limiters when necessary and a shared basis across cell interfaces for the diffusive flux. For efficiency, we implement the method in the context of the dual narrow band overset mesh approach of the Refined Level Set Grid method (Herrmann, 2008). The accuracy, consistency, and convergence of the resulting method is demonstrated using the method of manufactured solutions (MMS) and several standard test cases, including Zalesak's disk and columns and spheres in prescribed deformation fields. Using MMS, we demonstrate $k+1$ order spatial convergence for $k$-th order orthonormal Legendre polynomial basis functions. We furthermore show several orders of magnitude improvement in shape and volume errors over traditional WENO based distance function level set methods, and $k-1$ order spatial convergence of interfacial curvature using direct neighbor cells only. [Preview Abstract] |
Monday, November 19, 2012 11:22AM - 11:35AM |
H3.00005: A mass-conserving volume of fluid method for DNS of droplet-laden isotropic turbulence Antonino Ferrante, Michael Dodd We developed a mass-conserving wisps-free volume of fluid (VoF) method for direct numerical simulation (DNS) of droplet-laden turbulent flows. We used the continuous surface force (CSF) model to include the surface tension within a split-advection and mass-conserving VoF. The liquid-gas interface curvature is computed accurately using a variable-stencil height-function technique. We modified the sequence of the advection sweeps, and our results show that, in the case of non-zero Weber number, the algorithm is accurate and stable. We present DNS results of fully-resolved droplet-laden incompressible decaying isotropic turbulence at initial $Re_{\lambda}=190$ using a computational mesh of $1024^3$ grid points, droplet volume fraction 0.1 tracking the volumes of 7000 droplets of Weber number We$\ =0.5$ based on the r.m.s. velocity fluctuation, droplet-to-fluid density ratio 10, and initial droplet diameter equal to the Taylor length-scale of turbulence. [Preview Abstract] |
Monday, November 19, 2012 11:35AM - 11:48AM |
H3.00006: Effect of bubble-bubble interaction on mass transfer in bubbly flow using a multi-scale approach Bahman Aboulhasanzadeh, Gretar Tryggvason Mass transfer in the liquid phase of gas-liquid multiphase flows generally takes place at a much shorter time scale than the momentum transfer and so leads to thin mass boundary layers around the bubble which is difficult to capture using direct numerical simulation (DNS). We have developed a subscale analytical approach using a Front Tracking method in which we use solution of a boundary layer equation for mass transfer on the bubble interface to transfer mass from bubble onto a regular grid and then we follow the mass in the domain using a reasonably coarse gird. This way we are able to considerably reduce the computational cost. Here we implement the method in a three-dimensional code and perform direct comparison of its results with experimental data. We show that this approach gives accurate results compared to experimental data and semi-empirical correlations. Then we use our subscale approach to study the effect of Reynolds number and void fraction on the effect of bubble-bubble interactions on the mass transfer in buoyant bubbly flows. [Preview Abstract] |
Monday, November 19, 2012 11:48AM - 12:01PM |
H3.00007: Homogeneous and isotropic turbulence laden with particles of different Stokes numbers George Mallouppas, Berend van Wachem, William George The interactions of homogeneous isotropic turbulence with particles of various Stokes numbers are examined. The talk focuses on a series of DNS of forced turbulence laden with discrete particles performed on a $128^3$ periodic box. Several parameters are varied such as the Stokes number and the Taylor Reynolds number. The modification of one-point statistics due to the presence of particles is investigated. Moreover, the relation of forcing with light and heavy particles is investigated by evaluating the correlation coefficients between the forcing and the particles. It is shown that our newly proposed forcing scheme has a limited direct effect on the particles. An important consequence of the presence of the particles is the modification of the dissipation spectrum of the fluid, which depends on the particle Stokes number, and more surprisingly, on the Taylor Reynolds number. This is examined in view of the two-way coupling spectrum, which acts as a dissipative-transfer term. The dispersion of fluid and discrete particles are compared with analytical solutions by assuming the velocity autocorrelation function is of exponential form. Finally, the talk will address the importance of preferential concentration of the particles and its effect on the two-way coupling. [Preview Abstract] |
Monday, November 19, 2012 12:01PM - 12:14PM |
H3.00008: On the direct numerical simulation of moderate-Stokes-number turbulent particulate flows using algebraic-closure-based and kinetic-based moments methods Aymeric Vie, Enrica Masi, Olivier Simonin, Marc Massot To simulate particulate flows, a convenient formalism for HPC is to use Eulerian moment methods, which describe the evolution of velocity moments instead of tracking directly the number density function (NDF) of the droplets. By using a conditional PDF approach, the Mesoscopic Eulerian Formalism (MEF) of F\'evrier et al. 2005 offers a solution for the direct numerical simulation of turbulent particulate flows, even at relatively high Stokes number. Here, we propose to compare to existing approaches used to solved for this formalism: the Algebraic-Closure-Based Moment method (Kaufmann et al. 2008, Masi et al. 2011), and the Kinetic-Based Moment Method (Yuan et al. 2010, Chalons et al. 2010, Vi\'e et al. 2012). Therefore, the goal of the current work is to evaluate both strategies in turbulent test cases. For the ACBMM, viscosity-type and non-linear closures are envisaged, whereas for the KBMM, isotropic and anisotropic closures are investigated. A main aspect of the current methodology for the comparison is that the same numerical methods are used for both approaches. Results show that the new non-linear closure and the Anisotropic Gaussian closures are both accurate in shear flows, whereas viscosity-type and isotropic closures lead to wrong results. [Preview Abstract] |
Monday, November 19, 2012 12:14PM - 12:27PM |
H3.00009: Simulating Primary Atomization at Arbitrary Density Ratios: a Stable and Conservative Framework Vincent Le Chenadec, Heinz Pitsch The present work focuses on two recent developments for Direct Numerical Simulation of two-phase flows, and their application to computations of turbulent primary atomization of liquid jets at large density ratios. Mass conservation properties of the algorithm are improved by means of a second-order unsplit Volume-of-Fluid method coupled to the Level Set approach. The three-dimensional volume fraction transport scheme is shown to reduce numerical artifacts known to pollute the interface representation in under-resolved regions of the flow. In the interface vicinity, the momentum conservation as well as stability of the flow solver are guaranteed by a monotonicity preserving geometric transport of the momentum, defined consistently with the volume fraction transport. Away from the interface, the flux computation is switched to a centered discretization in order to avoid excessive numerical dissipation. This framework is assessed in a set of validation cases, and applied to simulate the primary atomization of a turbulent round jet in quiescent gas at air/water density ratio and moderate Reynolds and Weber numbers. [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