Bulletin of the American Physical Society
66th Annual Meeting of the APS Division of Fluid Dynamics
Volume 58, Number 18
Sunday–Tuesday, November 24–26, 2013; Pittsburgh, Pennsylvania
Session G5: CFD IV |
Hide Abstracts |
Chair: S. Levent Yilmaz, MathWorks Room: 327 |
Monday, November 25, 2013 8:00AM - 8:13AM |
G5.00001: 2D Unstructured Finite Volume Lattice Boltzmann Model for Flow with Complex Geometric Boundaries Leitao Chen, Laura Schaefer Many of the numerical issues of LBM (lattice Boltzmann method) are not yet fully solved. One of the issues is its inability of handling complex geometric boundaries. Some published work, which is based on collision-streaming discretization of the LBE and corresponding lattice-like mesh, introduced successful treatments for curved boundaries. However, those schemes are not applicable to the boundaries with large curvature like porous media since the lattice-like mesh is not able to recognize it. In order to solve this issue, a 2D FVM (finite volume method)-based numerical framework is proposed, which completely uncouples the lattice structure and the spatial discretization and therefore brings the freedom of using any type of lattice structure while keeping the basic framework unchanged. The model is solved on an unstructured triangular mesh and triangular control volume. Boundary schemes of isothermal and thermal flow for the new numerical framework are also studied. Finally, a variety of isothermal and thermal flow problems are simulated and compared with other work. The proposed model can simulate the flow with a complex geometry to the desired accuracy in addition to complementing the simple geometry of the existing LB model. [Preview Abstract] |
Monday, November 25, 2013 8:13AM - 8:26AM |
G5.00002: A Second-Order Finite-Difference Scheme for the Lattice Boltzmann Method Parthib Rao, Laura Schaefer The lattice Boltzmann method (LBM) is being increasingly used as an alternative solver for the isothermal Navier-Stokes equation, as well as for other complex flows. However, due to an innate coupling between the velocity and the configuration space, LBM is restricted to uniform grids. This is a serious impediment for simulating flows with large gradients, flow around objects, etc. The discrete Boltzmann-BGK equation, which forms the basis of LBM, can be viewed as a set of hyperbolic equations with constant coefficients and a source term. We, therefore, use the finite difference method to discretize the Boltzmann-BGK equation (FDLBM). In FDLBM, the velocity-lattice is uncoupled from the spatial lattice allowing us to choose discrete velocities and space-time steps independently. The currently available FDLBM models have either narrow a stability range, or have large computational costs. To overcome these constraints, we employ the Lax-Wendroff scheme for the advection part, and central-difference for the spatial gradients, resulting in a scheme that is both explicit and second-order in both space and time. The proposed scheme is validated for an isothermal incompressible lid-driven cavity flow. The results indicate improved stability (in terms of CFL conditions) compared to the current explicit FDLB models, due to the addition of the second-order temporal terms. The maximum Reynolds number that can be simulated stably is also much higher. The relationship between the discrete time-step and the relaxation parameter, and extension of the FDLBM to a non-uniform mesh are also discussed. [Preview Abstract] |
Monday, November 25, 2013 8:26AM - 8:39AM |
G5.00003: A Block-Structured Adaptive Mesh Refinement Technique with a Finite-Difference-Based Lattice Boltzmann Method Abbas Fakhari, Taehun Lee A novel adaptive mesh refinement (AMR) algorithm for the numerical solution of fluid flow problems is presented in this study. The proposed AMR algorithm can be used to solve partial differential equations including, but not limited to, the Navier-Stokes equations using an AMR technique. Here, the lattice Boltzmann method (LBM) is employed as a substitute of the nearly incompressible Navier-Stokes equations. Besides its simplicity, the proposed AMR algorithm is straightforward and yet efficient. The idea is to remove the need for a tree-type data structure by using the pointer attributes in a unique way, along with an appropriate adjustment of the child block's IDs, to determine the neighbors of a certain block. Thanks to the unique way of invoking pointers, there is no need to construct a quad-tree (in 2D) or oct-tree (in 3D) data structure for maintaining the connectivity data between different blocks. As a result, the memory and time required for tree traversal are completely eliminated, leaving us with a clean and efficient algorithm that is easier to implement and use on parallel machines. Several benchmark studies are carried out to assess the accuracy and efficiency of the proposed AMR-LBM, including lid-driven cavity flow, vortex shedding past a square cylinder, and Kelvin-Helmholtz instability for single-phase and multiphase fluids. [Preview Abstract] |
Monday, November 25, 2013 8:39AM - 8:52AM |
G5.00004: Numerical investigations on the vortex-induced vibration of moving rigid body by using the Lattice Boltzmann Method Xiaohai Jiang, Taehun Lee, Yiannis Andreopoulos, Zhexuan Wang Vortex-induced vibrations (VIV) phenomena related to self-excited energy harvesters consisting of circular or square cylinders have been investigated numerically by using the BGK or MRT Lattice Boltzmann Method. In the present work such a harvester is placed inside a channel flow and is allowed to oscillate without a structural restoring force in a direction normal to the flow. Currently the half-way bounce-back boundary scheme and interpolations are being used to model the moving boundary. The numerical results were compared to the ones by classical CFD methods and experiments. A good agreement was obtained. The vortex dynamics and the development of the flow patterns for different flow parameters such as Reynolds number, blockage and aspect ratios will be presented. Particular emphasis is given to the dynamics of vortex pairing observed in several of the simulations. The present approach will be extended to simulate the flexible beam with the Immersed Boundary Method. [Preview Abstract] |
Monday, November 25, 2013 8:52AM - 9:05AM |
G5.00005: Isothermal Multiphase Flow using a Multi-domain Lattice Boltzmann Method Christopher J. Forster, Marc K. Smith In an effort to increase the useful property range of Lattice Boltzmann multiphase flow simulations, a multiple fluid domain approach has been developed. Specifically, the purpose of this approach is to allow higher density and viscosity ratios across fluid interfaces with minimal spurious currents and instability. The multiple domains are coupled through interpolated boundary conditions. A level-set method on a collocated grid is used to track the interface location, which provides the necessary information for implementing moving, interpolated boundary conditions on each of the domains. The multi-domain Lattice Boltzmann method coupled with a level-set method allows for a sharp interface to apply the interfacial conditions and surface tension forces, while implicitly handling topological changes. To demonstrate the capabilities of this method, a test case of buoyancy driven bubble train flow will be presented with several increasing density and viscosity ratios. [Preview Abstract] |
Monday, November 25, 2013 9:05AM - 9:18AM |
G5.00006: A Highly-Parallelized Perfectly Stirred Reactor (PSR) Model Using GPU Acceleration Sudip Adhikari, Abhilash J. Chandy Perfectly stirred reactors (PSR), which are idealized systems, where species undergoing chemical reactions have high rate of mixing, have been found to be very useful in testing and developing chemical reaction mechanisms for combustion research. The PSR model requires solving systems of nonlinear algebraic equations governing the chemical reactions, which typically are of the order of hundreds for realistic engineering systems and also involve multiple time scales ranging over a few orders of magnitude. As a result, the equations are stiff and the solution is highly compute-intensive. In spite of dramatic improvements in central processing units (CPUs) made during the past several decades, PSR solutions, while they remain feasible are computationally very expensive. An alternative approach is the application of accelerator technologies, such as graphics processing units (GPUs) that can improve the performance of such algorithms. A highly parallelized GPU implementation is presented for the PSR model, using a robust and efficient non-linear solver. Parallel performance metrics are presented to demonstrate the capability of GPUs to accelerate chemical kinetics calculations. [Preview Abstract] |
Monday, November 25, 2013 9:18AM - 9:31AM |
G5.00007: Asynchronous schemes for CFD at extreme scales Aditya Konduri, Diego Donzis Recent advances in computing hardware and software have made simulations an indispensable research tool in understanding fluid flow phenomena in complex conditions at great detail. Due to the nonlinear nature of the governing NS equations, simulations of high Re turbulent flows are computationally very expensive and demand for extreme levels of parallelism. Current large simulations are being done on hundreds of thousands of processing elements (PEs). Benchmarks from these simulations show that communication between PEs take a substantial amount of time, overwhelming the compute time, resulting in substantial waste in compute cycles as PEs remain idle. We investigate a novel approach based on widely used finite-difference schemes in which computations are carried out asynchronously, i.e. synchronization of data among PEs is not enforced and computations proceed regardless of the status of messages. This drastically reduces PE idle time and results in much larger computation rates. We show that while these schemes remain stable, their accuracy is significantly affected. We present new schemes that maintain accuracy under asynchronous conditions and provide a viable path towards exascale computing. Performance of these schemes will be shown for simple models like Burgers' equation. [Preview Abstract] |
Monday, November 25, 2013 9:31AM - 9:44AM |
G5.00008: A Parallel Hexahedral Unstructured Adaptive Mesh Refinement Library Carlos Ballesteros, Marcus Herrmann Adaptive mesh refinement (AMR) libraries can simplify the generation of meshes surrounding complex or moving boundaries, as well as focus computational resources only in the areas of the solution domain that are of interest through the use of recursive cell refinement. By applying AMR within an unstructured hexahedral mesh framework, the resulting mesh retains the favorable numerical properties of hexahedral elements, while possessing characteristics advantageous for usage in high-performance computing. These properties include straightforward refinement and coarsening operations; as well as explicit connectivity between solution cells, which make neighbor-cell lookups, domain decomposition and load balancing simple, especially when compared with tree AMR approaches. The parallel scalability of a unstructured hexahedral AMR library, FARCOM, will be presented, with its ability to generate meshes illustrated with several test cases. Additionally, extensions to convection-diffusion, incompressible flow, and immersed-boundary problems will be discussed. [Preview Abstract] |
Monday, November 25, 2013 9:44AM - 9:57AM |
G5.00009: Parallel Cartesian grid refinement for 3D complex flow simulations Dionysios Angelidis, Fotis Sotiropoulos A second order accurate method for discretizing the Navier-Stokes equations on 3D unstructured Cartesian grids is presented. Although the grid generator is based on the oct-tree hierarchical method, fully unstructured data-structure is adopted enabling robust calculations for incompressible flows, avoiding both the need of synchronization of the solution between different levels of refinement and usage of prolongation/restriction operators. The current solver implements a hybrid staggered/non-staggered grid layout, employing the implicit fractional step method to satisfy the continuity equation. The pressure-Poisson equation is discretized by using a novel second order fully implicit scheme for unstructured Cartesian grids and solved using an efficient Krylov subspace solver. The momentum equation is also discretized with second order accuracy and the high performance Newton-Krylov method is used for integrating them in time. Neumann and Dirichlet conditions are used to validate the Poisson solver against analytical functions and grid refinement results to a significant reduction of the solution error. The effectiveness of the fractional step method results in the stability of the overall algorithm and enables the performance of accurate multi-resolution real life simulations. [Preview Abstract] |
Monday, November 25, 2013 9:57AM - 10:10AM |
G5.00010: Domain decomposition for coupled Stokes and Darcy flows with floating Stokes domains ChangQing Wang, Ivan Yotov A non-overlapping domain decomposition method is presented to solve a coupled Stokes-Darcy flow problem in parallel by partitioning the computational domain into multiple subdomains. Specifically, in the case where floating Stokes subdomain occurs, an approach based on the FETI methods is introduced and tested. [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