Accepted Abstracts to the 2014 Overset Grid Symposium
Oral Presentations
Abstract ID: OGS2014-001
J. W. Banks
Center for Applied Scientific Computing
Lawrence Livermore National Laboratory, Livermore, CA 94551
In this talk we discuss a newly developed class of robust and high-order accurate upwind schemes of wave equations posed in second-order form. Overlapping grids are used to represent geometric complexity. The schemes are based on embedding d'Alembert's exact solution of a local Riemann-type problem directly into the discretization. High-order accuracy is obtained using a single-step space-time scheme, and the resultant method is highly efficient in terms of both memory and speed. We will discuss the implementation on curvilinear and overlapping grids, and provide evidence that the natural dissipation inherent to these upwind schemes provides stabilization against perturbations introduced through overlapping grid interpolation. The efficacy of the approach is demonstrated on a series of tests including results from acoustics and electromagnetics.
Abstract ID: OGS2014-002
Bill Henshaw
Dept. of Mathematical Sciences
Rensselaer Polytechnic Institute, Troy, NY USA
We discuss current capabilities and some recent developments in solving fluid structure interaction problems on overlapping grids. The Overture framework and associated PDE solvers have capabilities for coupling compressible and incompressible fluids with moving rigid bodies as well as with deforming bulk solids and structural beams. For deforming bodies, overlapping grids near the interface are regenerated at each time-step using a hyperbolic grid generator. In recent work we have developed some new AMP (added-mass partitioned) algorithms that over-come the added mass instability for coupling fluids and light solids. We have developed new algorithms for a number of regimes including compressible fluids and rigid or elastic solids as well as for incompressible fluids coupled to elastic solids and structural beams/shells.
Abstract ID: OGS2014-003
Eliot Quon and Marilyn Smith
School of Aerospace Engineering
Georgia Institute of Technology, Atlanta, GA 30332
This talk addresses a class of scattered data interpolation techniques based on radial basis functions (RBF) and discusses their applicability to overset data transfer. Since the approach depends on arbitrary clouds of source points rather than a stencil, it has no dependence on connectivity information and may be applied to interface non-contiguous grids with any topology. This new overset paradigm naturally precludes the existence of orphan points and presents a care-free solution to complex, dynamic overset configurations. The RBF approach has been demonstrated for a two-dimensional inviscid vortex convection problem on grids both with and without orphan points. Over a magnitude reduction in overset errors were observed in orphan-free cases. For cases with orphan points, errors were reduced by up to 30% while transient conservation errors were reduced by up to an order of magnitude. Results from applying the same methodology to a three-dimensional turbulent ship airwake are also presented.
Abstract ID: OGS2014-004
Ralph Noack
Celeritas Simulation Technology, LLC
Many hole cutting algorithms will only out line the hole cut geometry in the grid being cut. A flood fill algorithm is required to mark all the points inside the geometry. The flood fill will be successful if the geometry is "effectively" water tight. This means small gaps can be tolerated as long as the grid being cut does not have grid lines or cells that pass through the gap. A new flood fill approach will be presented that is able to close small gaps and stop many flood fill leaks. The algorithm will be demonstrated for real geometries along with some timing comparisons to illustrate the extra cost associated with the new algorithm.
Abstract ID: OGS2014-005
Nick J. Wyman And John P. Steinbrenner
Pointwise, Inc., Fort Worth, TX 76104 USA
The generation of a connected, ready-for-simulation composite overset mesh has traditionally required the use of multiple software tools created by multiple authors with support for varying computing platforms. Managing the interaction of these tools introduces significant friction into the design-analysis workflow. The presentation will describe Pointwise’s efforts to streamline overset mesh generation and assembly through the integration of its general purpose meshing software with two popular overset grid assembly tools. In addition to integration methodology, examples of integrated overset mesh generation and assembly will be presented. A powerful additional benefit of integrated overset mesh generation and grid assembly is the ability to optimize composite grid connectivity. Pointwise has developed domain connectivity remediation tools which can reduce or eliminate orphans in both structured and unstructured grids. An example of domain connectivity improvement will be shown.
Abstract ID: OGS2014-006
Scott T. Imlay, Durrell K. Rittenberg, and Craig A. Mackey
Tecplot, Inc.
3535 Factoria Blvd SE, Suite 550
Bellevue, WA 98006
The size of datasets analyzed by post-processing and visualization tools is growing with Moore’s law. Conversely, the disk-read data transfer rate is only doubling every 36 months and is the bottleneck for traditional post-processing architectures. To eliminate this bottleneck, a sub-zone load-on-demand visualization architecture has been developed which only loads the data needed to create the desired plot. Loading starts with an query of a min/max tree to determine which sub-zones are needed. For calculated variables (for example, pressure computed from the conservative variables), the min/max values of the computed variable are estimated using interval arithmetic. For most analysis, the actual loading of the data is O(n^2/3) or less, where n is the number of nodes. The resulting visualization tool is faster and uses far less memory than traditional visualization packages. The benefits are demonstrated for large unsteady OVERFLOW datasets.
Abstract ID: OGS2014-007
Hansong Tang and Ke Qu
Dept. of Civil Eng., City College, City Univ. of New York
New York, NY 10031, USA
It is now important to predict coastal ocean flow phenomena that are distinct from each other and span a vast range of scales. We propose a framework that couples the Finite Volume Coastal Ocean Model with the Solver for Incompressible Flow on Overset Meshes using Chimera grids; the former is used to simulate large-scale background coastal flows, and the latter is employed to capture small-scale local processes. The coupling involves distinct governing equations, different numerical algorithms, and dissimilar grids, and it is two-way and realized using the Schwartz alternative iteration. The proposed framework is able to capture flow phenomena with spatial scales ranging from O (1) m to O (10,000) km and deal with multiply-connected domains, both of which cannot be handled by existing coastal ocean models. In this talk, the methodology of the framework will be discussed, and its applications we recently finished will be presented.
Abstract ID: OGS2014-009
Dominic Chandar
Institute of High Performance Computing(IHPC), Singapore
In this work, an overset grid algorithm on unstructured moving grids and its implementation methodology in OpenFOAM is presented. While existing approaches may couple Suggar++, DiRTlib with OpenFOAM, the current implementation is built upon an in-house object-oriented framework, written in C++ using the stencil-walk algorithm to search for donors from a given partitioned mesh. Following extensive testing, the library is planned to be included as an add-on to the OpenFOAM distribution. Based on the work of Roget and Sitaraman, this paper uses a variant of the Exact Inverse Map(EIM) method for the donor search procedure. Maximization of available computational resources is achieved by a dual parallelism approach where partitioned meshes search for donors in parallel using pthreads or using Intel CILK. The Parallel Domain Connectivity Function (parDCF) is bundled into a dynamic library and can be built with OpenFOAM or any other open-source CFD code with minimal effort. The parDCF routines and the integration wrapper with OpenFOAM is collectively called - Overset Parallel Engine for Aerodynamics Applications (Opera). A detailed description of the OpenFOAM directory structure for overset cases is also presented.
Abstract ID: OGS2014-0011
Orhan Shiblieyev and Jayanarayanan Sitaraman
University of Wyoming
One of the problems of the overset grid techniques is the lack of conservation. The reasons are the geometrical inconsistencies at the boundaries of the grids relative to each other and the interpolation of variables within the grids especially in the regions where a discontinuity exists. It is possible to interpolate fluxes instead of conservative or primitive variables however, this approach is cumbersome and limited to certain problems. Another way, which will be discussed in the talk, is to remesh the grids by removing the intergrid elements to obtain a single mesh. This method has been successfully applied on the node-centred unstructured meshes in the past while in the present work cell-centred meshes were used. The main difference is that for node-centred grids after remeshing the number of variables remain the same while for the cell-centred grids new variables are introduced at each time step.
Abstract ID: OGS2014-0012
J. Ezequiel Martin
Assistant Research Scientist, IIHR—Hydroscience & Engineering, University of Iowa, Iowa City, IA 52242
Ralph W. Noack
Celeritas Simulation Technology, LLC
Pablo M. Carrica
Professor, Dept. of Mechanical Engineering and Research Engineer, IIHR—Hydroscience & Engineering, University of Iowa, Iowa City, IA 52242
We propose an approach to overset simulation of multiple disconnected bodies, by partitioning the domain into sub-domains that can be solved independently. The method requires each sub-domain to be isolated from the others, ensuring that no interaction occurs between grid groups. Independent groups are linked using a static overset connectivity interpolation of wrapper grids around the sub-domains and a single background. The proposed method results in an important reduction of computational time of the overset connectivity, and major savings of computational resources. The implementation is done in our newest CFD solver, MagnusFlow, using multiple instances of Suggar++. Calculations of wind turbine farms will be presented, as well as implementation ideas for naval hydrodynamics. It is shown that for wind farms, where all turbines are topologically disconnected, ideal strong overset scalability can be achieved by running one turbine per set of processors and performing posterior manipulation of the domain connectivity information.
Abstract ID: OGS2014-015
Pieter G. Buning
NASA Langley Research Center,Hampton, VA
Thomas H. Pulliam
NASA Ames Research Center,Moffett Field, CA
This presentation will describe recent work to add "user-friendly" controls to the solution adaption capability in the OVERFLOW overset grid CFD flow solver. These controls allow the limiting of growth and maximum size of the grid system resulting from successive adapt cycles, either for steady-state or time-accurate simulations. A sample application to generate an airfoil polar will be shown, with results from a coarse baseline grid, a fully-refined grid, and an automatic solution adaption strategy used for each flow condition.
The addition of a temporal error controller and new high-order time-advance schemes will also be presented. Examples will show the intent of the controller for early termination of dual time-step sub-iterations, and possible application for adjusting the physical time-step. A vortex convection problem and a pitching airfoil in dynamic stall will be used to illustrate this work.
Abstract ID: OGS2014-016
Stuart E. Rogers
NASA Ames Research Center,Moffett Field, CA
This talk will present recent improvements to the Pegasus5 overset software. Version 5.2 of the code was just released. New features include the addition of a cell-centered overset-grid capability, improvements to the automatic hole-cutting procedure, and improvements to the parallel performance of the code. The hole-cutting procedure was enhanced through automatic multiple hole-cutting domains created through automatic, recursive splitting of the domain. The use of multiple domains improves hole-cutting resolution while increasing the parallel enables efficiency. The parallel performance and load-balancing of the code was improved using a fine-scale management of individual processes. This improves parallel scaling to take advantage of larger numbers of processors. The improved parallel performance opens the possibility to consider the use of Pegasus5 for the oversetting of time-accurate moving-body problems, and for oversetting during a grid adaptation process. The possibilities for fully integrating the Pegasus5 process with the Overflow code will be discussed.
Abstract ID: OGS2014-017
James Baeder and Tarandeep Kalra
University of Maryland, College Park, MD
compressible, structured, overset Reynolds-Averaged Navier-Stokes (RANS) based solver (OverTURNS) is used to simulate hovering rotors in ground effect (IGE). The computations are performed for a micro-scale rotor (0.086 m radius, aspect ratio of 4.387 operating at a tip Mach number of 0.08 and Reynolds number of 32,500) and a sub-scale rotor (0.408 m radius, aspect ratio of 9.132 operating at a tip Mach number of 0.24 and Reynolds number of 250,000). The micro-scale rotor is simulated for three rotor heights: 1.5R, 1.0R and 0.5R above ground (R = Rotor radius). The sub-scale rotor is simulated at one particular rotor height (i.e. 1R) but with four different tip shapes: rectangular, swept, BERP-like and slotted tip. Various mesh placement strategies are used to efficiently capture the path of the tip vortices. The qualitative and quantitative results agree well with the available experiments.
Abstract ID: OGS2014-018
S. Péron, C. Benoit, S. Landier, P. Raud
ONERA - France
In 2008, ONERA started a project to gather pre- and post-processing research tools dedicated to CFD applications in one piece of software. This software, called Cassiopee, is a set of Python modules, based on the CGNS/Python standard, integrating numerous simple functions using a CGNS/Python tree as the common data for input and output of all the functions. This package is mainly open source and contains mesh generation functions (TFI, extrusion, refinement, stretching, smoothing, ...), mesh modifications (rotation, block merging, block splitting, ...). Overset grid technology is integrated in Connector module. Blanking, overlap optimization, Chimera connectivity (donor cell search and interpolation coefficients) are accessible as elementary functions. A specific post-processing module, which takes into account the nature of points (blanked out, interpolated, computed) in overlapping regions, enables to extract slices, isosurfaces during a CFD computation. In our presentation, we would like to introduce Cassiopee package and focus on a few workflows which can be easily set up with these modules, such as computing the Chimera connectivity, generating and adapting off-body Cartesian mesh and preprocessing Immersed Boundary Method. Finally, some helicopter and aircraft applications achieved at ONERA will be presented, in order to illustrate the capabilities of Cassiopee modules to perform pre-processing of complex CFD computations.
Abstract ID: OGS2014-019
Michael J. Brazell, Jay Sitaraman, Dimitri J. Mavriplis
University of Wyoming
A parallel high-order Discontinuous Galerkin (DG) method is used to solve the Navier-Stokes equations on an overset mesh framework. The DG solver has many capabilities including: hp-adaption, curved cells, and support for hybrid, mixed element meshes. The overset grid connectivities are handled through TIOGA (Topology Independent Overset Grid Assembler). To interface with TIOGA support for high-order call back functions are needed. These include high order interpolation which is inherently built into DG. As well as an inclusion test for curved cells. Combining these capabilities with overset grids allows the DG solver to be used in problems with bodies moving in relative motion and can be used in a near-body off-body solver strategy similar to Helios. To validate the accuracy and performance of the overset DG solver multiple simulations are carried out. These simulations demonstrate the capability of the high-order DG solver to handle complex geometry and large scale parallel simulations in an overset framework.
Abstract ID: OGS2014-020
William M. Chan and Shishir A. Pandya
NASA Ames Research Center
New modules have been developed for the assessment of grid connectivity quality in the OVERGRID graphical user interface in the Chimera Grid Tools package. These include visualization of Cartesian cut planes over neighboring grids for analysis of grid overlap and relative cell sizes, automatic determination and display of grid planes that may enclose a given orphan point, visualization of fringe points that have been converted to field points due to insufficient grid overlap, and display of fringe points where the cell volume ratios between fringe points and their donor stencils are within a specified range. In a related topic, the adaptive X‐rays approach to hole‐cutting uses heuristic rules to create a spatially-variable offset from the minimum hole. Enhancements have recently been developed to these rules by comparing the distance of a grid point to the nearest wall from an unassociated component and a local grid outer boundary distance from the wall. By comparing local distances, it is anticipated that the hole‐boundaries obtained will result in less orphan points than before. Subsequent iterations to remove the orphan points would then become cheaper. Preliminary results from implementing the new rules on several test cases will be presented.
Abstract ID: OGS2014-021
Marshall C. Galbraith
Massachusetts Institute of Technology, Cambridge, MA, 02139
John A. Benek
U.S. Air Force Research Laboratory, Wright-Patterson Air Force Base, OH, 45433
Paul D. Orkwis
University of Cincinnati, Cincinnati, OH, 45221
The Chimera overset method is a powerful technique for numerically solving partial differential equations using arbitrarily overlapping grids. The solution to the partial differential equations is obtained by interpolating the solution from neighboring grids to artificial grid boundaries, i.e. grid boundaries that overlap with neighboring grids. Traditionally, implicit Chimera solvers that rely on a Newton-Raphson scheme to advance the solution do not include the inter-grid communication in the linearization. Hence, the stable time step is reduced as the grids are partitioned for parallel execution due to the explicit update of the artificial boundaries. This can lead to a significant reduction in parallel efficiency. This abstract presents a method that only requires a matrix solver based on matrix-vector multiplications to incorporate the linearization of the inter-grid communication. The method is independent of discretization and interpolation schemes. Results are presented for a Chimera solver that relies on a Discontinuous Galerkin discretization.
Abstract ID: OGS2014-023
Chao Liu, James C. Newman III, and W. Kyle Anderson
SimCenter: National Center for Computational Engineering, University of Tennessee at Chattanooga, Chattanooga, TN
In terms of mesh resolution requirements, higher-order finite-element discretization methods offer a more economic means of obtaining accurate simulations and/or to resolve physics at scales not possible with lower-order schemes. For simulations that may have large relative motion between multiple bodies, overset grid methods have demonstrated distinct advantages over mesh movement strategies. Combining these approaches offer the ability to accurately resolve the flow phenomena and interaction that may occur during unsteady moving boundary simulations. Additionally, overset grid techniques when utilized within a finite-element setting mitigate many of the difficulties encountered in finite-volume implementations. This presentation discusses the development of an overset grid methodology for use within a streamline/upwind Petrov Galerkin formulation for unsteady, viscous, moving boundary simulations. Furthermore, a novel hole cutting procedure based on solutions to Poisson equations (Elliptic Hole Cutting) is introduced and compared to existing techniques. Order of accuracy is examined via the method of manufactured solutions and the extent of the overlapped region is studied. Overset grid results are presented for several steady-state and time-dependent, moving boundary simulations with linear, quadratic, and cubic elements.
Abstract ID: OGS2014-024
Shishir A. Pandya, Arthur Huang, Dogus Hakaydin, Shayan Moini-Yekta
A preliminary study of the D8 aircraft is presented to show that there is an aerodynamics benefit for using an integrated nacelle that ingests a boundary layer if a distortion tolerant fan is used. The overset mesh development and the adequacy of the mesh are discussed with a preliminary study of mesh parameter sensitivity. Results of two D8 variants are presented. The first configuration is a traditional engine mounted in a podded nacelle. The second configuration consists of a nacelle that has been integrated into the top, rear of the fuselage. This second configuration ingests a portion of the fuselage boundary layer. The effect of the boundary layer ingestion is assessed as a comparison to the podded configuration to quantify the resulting benefit. The podded nacelle in the above study ingests free-stream flow. However, it is possible to place the nacelle so that it ingests the wing wake. The variations in performance of the podded nacelle with respect to the ingestion of the wing wake are also discussed.
Abstract ID: OGS2014-025
Dmitry Pinaev, Frank Schaefer, Eberhard Schreck
CD-adapco
This talk will present the overset grid technology which has been implemented in the commercial CFD package STAR-CCM+. The focus of this development was a robust overset grid capability on unstructured meshes. As this overset grid methodology is integrated in a widely used commercial package, it has to be adapted to a large variety of physical models including Lagrangian multiphase and fluid film modeling. In the Lagrangian framework, particle-like elements known as parcels are followed through mesh cells. The parcel tracking algorithm has been adapted to take the overset grid methodology into account. Another physical model which required special attention is the fluid film model. In this model, shell regions are attached to the three dimensional volume mesh in order to simulate problems where a thin film of fluid exists on solid boundaries. In the presentation, both the methodology and its usage in the software will be described, with emphasis on the extensions for the non-standard applications mentioned above. Some example applications will be presented to demonstrate the use of this feature in STAR-CCM+, and directions of future extensions will also be outlined.
Abstract ID: OGS2014-026
Robert Haimes
Department of Aeronautics & Astronautics, Massachusetts Institute of Technology
This presentation describes an overset approach that is comprised of a "virtual
boundary layer"-like near-body grid coupled with an off-body Adaptive Mesh
Refinement (AMR) far-field mesh for viscous fluids simulations. Unlike most
a priori grid generation systems for the Reynolds-Averaged Navier-Stokes
equations, the strand meshing paradigm is automatic, fast and requires little
memory in order to provide complete boundary-layer coverage. In addition, the
stacks of elements implied by the strands can be used to the simulation's
advantage, where they naturally provide a line direction for semi-implicit
solving.
The procedure starts from the tessellation of the geometry of interest where
the vertices coming from the BRep Edges and Nodes are marked. This allows for
a classification of these locations that drives the construction. Laplacian
smoothing and then an optimization scheme moves and untangles the strand stacks
in concave regions.
Abstract ID: OGS2014-028
Christopher M. Langel, Raymond Chow, and C.P. van Dam
University of California, Davis
An overset compatible surface roughness model extending the Langtry-Menter transition model has been implemented in OVERFLOW. The model introduces an additional scalar field roughness amplification quantity. This value is explicitly set at rough wall boundaries using surface roughness parameters and local flow quantities. An additional transport equation allows non-local effects of surface roughness to be accounted for downstream of rough sections. This roughness amplification variable is coupled with the Langtry-Menter model and used to modify the criteria for transition. An initial calibration effort was performed on flat plate configurations and a NACA 0012 airfoil with leading edge roughness. To further validate and develop the model a large number of wind tunnel experiments were conducted at Texas A&M University. These new results were analyzed and the formulation of the model adjusted accordingly. In addition to the development of the roughness model, the talk will address the use of overset grids to improve the prediction of freestream turbulence intensity decay in relation to the accuracy of correlation based transition models.
Abstract ID: OGS2014-029
John F. Dannenhoffer, III
Syracuse University, Syracuse NY 13244
Jeffrey Slotnick
Boeing Research & Technology, Huntington Beach CA 92647
Darby Vicker and Reynaldo Gomez
NASA Johnson Research Center, Houston TX 77058
Scott Scherer
Air Force Research Laboratory, Wright-Patterson AFB OH 45433
Many organizations now use the the Chimera Grid Tools (CGT) and configuration grid scripts to manage the creation of complex overset grid systems. However, generation of high-quality surface grids that define the configuration component geometry is still extremely time-consuming, tedious, and very dependent on the skill of the CFD analyst. In the past several years, an innovative, and potentially disruptive, technology was introduced (called OvrCad) to automatically generate high quality overset grid systems for arbitrarily complex configurations using geometric knowledge contained in the solid model feature tree as part of the Computer Aided Design (CAD) definition of the configuration. This new technology promises to make the generation of the entire overset grid system
essentially push-button.
This presentation describes the results of a series of cooperative agreements in which the automated grid technology was evaluated in anticipation of maturing this technology for transition to production use. The configurations that were evaluated included a wing/body high-lift configuration, a complete fixed-wing aircraft, and a MPCV launch abort vehicle. Each configuration stressed the software in unique ways, such that the overall maturity of the tool was assessed. Success metrics included overall grid generation wall clock time compared to hand gridding methods, and the accuracy of OVERFLOW flow solutions using the new grid system. The assessment also resulted in an understanding of how the software
would interoperate with the CGT grid scripts.
Abstract ID: OGS2014-030
Jayanarayanan Sitaraman
University of Wyoming
Beatrice Roget
Science and Technology Corporation
This paper presents a method to perform enables efficient and automated Overset Grid Assembly (OGA) on a system of overlapping unstructured meshes and overlapping unstructured/Cartesian meshes in a parallel computing environment where all meshes are partitioned into multiple mesh-blocks and processed on multiple cores. The main task of the overset grid assembler is to identify, in parallel, among all points in the overlapping mesh system, at which points the flow solution should be computed (field points), interpolated (receptor points), or ignored (hole points). Point containment search or donor search, an algorithm to enables efficiently determine the cell that contains a given point, is the core procedure necessary for accomplishing this task. We use an active load balancing techniques to improve the efficiency and scalability of the point containment search.
Abstract ID: OGS2014-031
Andrew M. Wissink
US Army Aviation Development Directorate - AFDD (AMRDEC), Moffett Field, CA
Jayanarayanan Sitaraman
University of Wyoming
Aaron J. Katz
Utah State University, Logan, UT
This paper will present results from the application of a strand-based CFD solver to rotorcraft simulations using the HPCMP CREATETM-AV Helios solver. Surface discretization of the rotor and fuselage are done using traditional triangles, then strand volume meshes are automatically grown around the rotor blades and traditional unstructured volume meshes are grown around the fuselage. A specialized near-body strand solver, that exploits the inherent enables efficiency of the strand data structures, is then applied to the blades. Results will be shown for rigid blade calculations with comparisons to the traditional unstructured/Cartesian dual mesh paradigm.
Abstract ID: OGS2014-032
Rajiv Shenoy
CRAFT-Tech, Inc.
Marilyn J. Smith
Georgia Institute of Technology
Michael Park
NASA Langley Research Center
An approach for feature-based grid adaptation for dynamic overset simulations is presented to enable adaptation over time intervals in a dynamic flow field so that an accurate evolution of the unsteady wake may be obtained. This approach permits grid adaptation to occur seamlessly across any number of grids that are overset, excluding only the boundary layer to avoid surface manipulations. Both inviscid and turbulent applications are reported, including their ability to improve predictions. In addition, a robust parallel localization scheme is introduced, which expedites unsteady adaptation by enabling efficient solution transfers across adapted overset grids. The complexity of the search algorithm is verified over parallel decompositions and is independent of the number of partitions.
Abstract ID: OGS2014-033
Michael Lucas
Principal Engineer for the Industrial Technology Group, Ingersoll-Rand, Davidson NC, 28036
One of the most common types of industrial air compressors is the rotary twin screw compressor. A screw compressor is essentially a pair of intermeshing helical rotor gears that have a profile optimized for the compression process. CFD modeling the air flow between the passages formed by the twin rotors offers many challenges to develop a suitable mesh. These challenges include complex shaped rotor profiles, the tight clearance between the rotors, and the counter rotating helix sweep angles. A majority of the compressor companies are currently using a structure moving mesh approach. We have found that a structure mesh contains many problems with the mesh quality and has limitations on the types of screw profiles. Presented in this paper are predictions made using Overgrid/Overflow on a screw compressor. These results will be compared to a structure moving mesh approach and measured results from a dynamometer.
Abstract ID: OGS2014-034
Yoram Yadlin and Arvin Shmilovich
The Boeing Company,Huntington Beach, CA 92647
Various flow control techniques are applied to airplane configurations that utilize Upper Surface Blowing (USB) in order to enhance their takeoff and landing capabilities. These techniques include geometrical changes to the lifting surfaces and the engine nozzle. In addition, several flow control methods are applied to further enhance the aircraft performance. Both passive and active flow control techniques are applied at representative flight conditions. Numerical results demonstrate the effectiveness of the methods in lift enhancement for USB powered configurations.
Abstract ID: OGS2014-035
Jennifer N. Abras
NAVAIR Applied Aerodynamics and Store Separation Branch, Patuxent River, MD, 20670
Nathan Hariharan
CREATE-AV, Lorton, VA, 22079
The new dual mesh solver option now available in KESTREL v5 is the focus of this presentation. This dual mesh methodology employs a Cartesian based solver in the off-body and an unstructured mesh in the near-body. The advantage of employing a Cartesian solver for a majority of the computational region include faster per node computations and higher spatial accuracy. The new off-body solver in KESTREL also provides the option of employing time-accurate adaptive mesh refinement. The predictions made with this solver will be compared with predictions made with the KESTREL v4 single mesh method as well as available wind tunnel data. The wind tunnel data is extracted from a test performed on the ROtor Body INteraction (ROBIN) mod 7 fuselage. Comparisons of integrated viscous and pressure drag, flow separation point, and centerline pressure distributions are analyzed. Parametric studies of the available options are performed.
Abstract ID: OGS2014-036
Konstantinos Vogiatzis and Nathan C. Prewitt
Engility Corp, HPCMPO PETTT
Roy P. Koomullil
University of Alabama at Birmingham
William G. Szymczak
Naval Research Lab, DC
BUB3D-RB is the only U.S. Navy owned and U.S. developed CFD software package specifically tailored for Underwater Launch (UWL) analysis. Its modeling capabilities include air-water interface hydrodynamic applications that include free surfaces, wave dynamics, cavitation, gas bleed, bubble migration, and shallow depth explosion plumes. It also includes rigid body dynamics simulations for 3 degrees-of-freedom (3DOF) – x translation, z translation, and y rotation or pitch.
The code was extended to use multiple, overlapping grids in the solution of hydrodynamics problems that include rigid, moving bodies. The communication between overset meshes is established using DiRTlib (Donor Receptor Transaction Library) and a domain connectivity information (DCI) file that is generated in a pre-processing step using SUGGAR (Structured, Unstructured, and Generalized Grid AssembleR).
This newly developed capability was tested using several model problems and was shown to produce accurate results, validating the applicability of overset methods for many hydrodynamics applications.
Abstract ID: OGS2014-037
Norman Foster
Computational Mechanics Division,Applied Research Laboratory, The Pennsylvania State University, University Park, PA 16804
It has been shown that for overset CFD calculations that utilize high order accurate inviscid flux discretization schemes, solution accuracy can be increased by using high-order overset interpolation. This can be critically relevant in many cases, such as those where it is important to accurately track vortical structures through the domain because of fluid-structure interactions, for example. Many block structured overset CFD solutions employ variants of Lagrangian interpolation to determine overset donor weights and, of these, many use 2nd order accurate interpolation stencils. This talk demonstrates the pitfalls of using lower-order (i.e. 2nd) in conjunction with high-order solver numerics. Familiar canonical problems are used to demonstrate that in certain types of problems, failure to use high-order interpolation can quickly lead to errors that undermine the reasons for using an advanced numerical scheme in first place. Results also include prediction comparisons to rotor hub wake test data.
Abstract ID: OGS2014-038
Joshua Leffell
US Army Aeroflightdynamics Directorate (AMRDEC), Moffett Field, CA & Stanford University, Stanford CA
Thomas H. Pulliam & Scott M. Murman
NASA Ames Research Center, Moffett Field, CA
The Time-Spectral methods has demonstrated success in accelerating calculations involving time-periodic flows. This talk highlights the extension of the Time-Spectral methods for use on overset grids. The standard Time-Spectral method proves unsuitable on overset grids due to mesh points that are dynamically removed from the computational domain as a function of relative motion between components. Data is therefore unavailable to uniquely determine the Fourier series expansion at such dynamically-blanked nodes. In the hybrid approach, the solution at dynamically-blanked nodes are expanded in an alternative manner which does not rely on blanked data. Implementation of the proposed approach within OVERFLOW is provided followed by a series of two- and three-dimensional test cases including high-frequency plunging airfoils and hover and edgewise-flight simulation of the quarter-scale V-22 Tilt Rotor Aeroacoustic Model (TRAM).
Abstract ID: OGS2014-040
Dr. Earl P.N. Duque
Manager of the Applied Research Group, Intelligent Light, Rutherford, NJ
Writing, storing, moving and post-processing vast unsteady datasets can interfere with an engineer’s interpretation and reporting of results. This paper will present ongoing research to develop new methods designed to extract and reduce large unsteady CFD derived volumetric data. In-Situ data extraction whereby sub-setting and segmenting the volume data using data extraction and analysis libraries directly integrated within the solver codes themselves is the first step. To further reduce the amount of unsteady CFD extract data written to disk, methods such as Proper Orthogonal Decomposition may be used for a given error band. This paper will present preliminary research and how the Overset Grid Community could use these techniques to analyze their large scale CFD solutions.
Student Posters
Abstract ID: OGS2014-008
Zhirong Shen (1,2), Decheng Wan(1), Pablo M. Carrica (2)
1. School of Naval Architecture, Ocean and Civil Engineering, State Key Laboratory of Ocean Engineering, Shanghai Jiao Tong University, 800 Dongchuan Road, Minhang District, Shanghai, 200240, China
2. IIHR-Hydroscience and Engineering, The University of Iowa, 100 C. Maxwell Stanley Hydraulics Laboratory, Iowa City, IA 52242, USA.
We present an implementation of the dynamic overset grid technique in the open source code OpenFOAM, including a hierarchy of objects that enables computation of 6DoF motions with control surfaces such as rudder and propeller. The dynamic computation of the domain connectivity information (DCI) at runtime is performed using SUGGAR. The exchange of data between OpenFOAM and SUGGAR is based on MPI and optimized using simultaneous execution of SUGGAR groups in multiple lagged mode. Several computation cases are used to validate the implementation and demonstrate the flexibility of the overset grid technique applied in ship hydrodynamics. The cases include self-propulsion of a ship in calm water, heave and pitch motions of a ship in waves and finally free body maneuvers (zigzag and turning circle) with the rotating propeller and moving rudder providing propulsion and steering. All computation cases have been compared against experimental data for validation, showing good agreement.
Abstract ID: OGS2014-013
Camli Badrya and Dr. James Baeder
University of Maryland, College Park, MD
The aerodynamic mechanism of insect flight at low Reynolds numbers presents significant computational modeling challenges. These challenges originate due to the physical nature of the flow at low Reynolds numbers that includes massive flow separation and complicated vortical structures along with the complexity of the insect kinematics. The computational model used to simulate insect-like kinematics is based on a compressible, structured and overset RANS based solver. This requires mesh systems that can model two moving wings and a fixed body. Therefore, in this work, a body-fitted mesh (O-O) is used for the flapping wings and the fixed body. A structured Cartesian background mesh is used to preserve the complex flow field features such as tip vortices for longer periods of time. This framework is then used to investigate the effects of kinematic parameters (such as flapping frequency, pitching angle, flapping amplitude, flapping stroke angle) on the insect flight performance.
Abstract ID: OGS2014-014
Daniel Prosser and Dr. Marilyn Smith
Georgia Institute of Technology, Atlanta, GA
The aerodynamics and dynamics of bluff bodies are important in a vast range of applications, including helicopter sling loads, air drops, aerial firefighting, humanitarian missions, search-and-rescue operations, and in military theaters. The aerodynamics and dynamics are complex, making the problem particularly challenging. In the case of helicopter sling loads, extensive flight testing is required for each configuration to ensure safe operation. The current work aims to improve the state of the art in coupled aerodynamic-dynamic predictions for bluff bodies via a computational first-principles approach. NASA’s FUN3D solver, which is capable of handling dynamic overset meshes and incorporates advanced turbulence approaches on unstructured grids, is employed to study the fundamental aerodynamics of bluff bodies and to perform tightly-coupled dynamic simulations. The insight gained from this first-principles approach is applied to develop physics-based reduced-order models for rapid dynamics and stability analysis of sling loads.
Abstract ID: OGS2014-022
Michael Stearns and Dr. Marilyn Smith
Georgia Institute of Technology, Atlanta, GA
Flow separation is one of the most unsteady and difficult to control regimes in aircraft aerodynamics. The resulting formation of a steep adverse pressure gradient in the boundary layer causes a marked increase in drag and turbulence. Many approaches have been studied to attempt to control the boundary layer in these conditions. These have been shown to have the potential for large increases of efficiency up to 15%, but most are active control systems that are impractical for general applications. Thus, the development of passive control systems for flow separation is highly sought after. One such passive control device and a promising finding from the field of biomimetics is the bristling scale of the Mako shark Pectoral fin. Using a parametric study using Overflow,correlated to existing experimental results, we are developing a physical understanding of how these scales control separated flow.
Abstract ID: OGS2014-027
Ethan Corle and Dr. Sven Schmitz
The Pennsylvania State University, University Park, PA
An accurate prediction of helicopter performance, noise, and vibrations is vital for future vehicle configurations and technologies. One such advanced technology is active rotor concepts capable of targeting simultaneously the challenges of performance enhancement along with noise and vibration reduction. This requires prediction tools that can accurately capture peak and unsteady load responses due to active devices. In an effort to work towards accurate predictions, this work concentrates on the development of CFD gridding and simulation techniques of active devices on a notional research rotor. In these simulations, moving-overset grids are implemented through the use of NASA's overset grid solver, OVERFLOW-D, and grid generation toolbox, Chimera Grid Tools.
Abstract ID: OGS2014-0010
Akshay Kanoria
Indian Insitute of Technology Gandhinagar, India
Dominic Chandar
Institute of High Performance Computing(IHPC), Singapore
Stanford University Unstructured (SU2) code is an open-source software suite integrated with analysis and design tools for solving multi-disciplinary problems governed by the partial differential equations (PDEs) on general, unstructured meshes. The Overset algorithm “Overset Parallel Engine for AeRodynamic Application” (OPERA) developed at Institute of High Performance Computing, Singapore allows dynamic overset mesh capabilities with any number of bodies using parallel processing. This talk presents the coupling of OPERA with SU2 by modifying the core SU2 algorithm to enhance the code capability to solve complex multi-body aerospace problems. The modified software suite has been currently tested for the flow past a sphere. The results compared with analytical results supports the potential and working of the algorithm compared to a non-overlapping mesh. More test cases using the Opera library are planned and will be made available for the final presentation. A similar coupling has been carried out with OpenFOAM (Open Source Field Operation And Manipulation) and will be discussed concurrently at this symposium. A brief comparison of results for example problems is being carried between both the solvers (SU2 and OpenFOAM) coupled with OPERA.
Abstract ID: OGS2014-0039
Brett Peters, Alex Curley, Max Henry and Mesbah Uddin
North Carolina Motorsports and Automotive Research Center, UNC Charlotte, Charlotte, NC, 28223
Numerical predictions of flow surrounding road vehicles often suffer from inaccurate assumptions while modeling the tire to road interaction. A current practice is to use stationary boundaries with rotating wall conditions or moving reference frames. Due to the complexity of the geometry and computational demand of moving mesh, the rotating wheel is over looked or over simplified during road vehicle simulations. At the tire contact patch, special grid treatment and grid blanking are required to simulate the conditions in a rolling road wind tunnel. The overset grids and numerical model are generated using NASA's Chimera Grid Tools and solved using NASA's overset grid solver, OVERFLOW. The current results will be compared to traditional methods implemented for road vehicle simulations.
- Last Updated: July 14, 2014