NIA CFD Seminar, Season 4 (2014-2015)
|
#60:   08-06-2015, Joan G. Moore
M4D, an Open Source Research CFD Code for the Calculation of Classical and Turbulent/Transitional Flows
#59:   07-15-2014, Graeme Kennedy
Large-Scale Topology Optimization Methods for Additively Manufactured Structures
#58:   07-14-2015, Li Wang
Flow Analysis and Adjoint-Based Design Optimization Using a High-Order CFD Method
#57:   01-29-2015, Qiqi Wang
Towards Aerospace design in the Age of Extreme-Scale Supercomputing
#56:   12-16-2014, Hiro Nishikawa
Accuracy-Preserving Boundary Quadrature for Edge-Based Finite-Volume
Scheme: Third-order accuracy without curved elements
#55:   12-02-2014, Matteo Parsani
Entropy stable wall boundary conditions for the compressible Navier-Stokes equations
#54:   10-06-2014, Tomohiro Irie
Recent Applications of Overset Mesh Technology in SC/Tetra
#53:   09-09-2014, Chau-Lyan Chang
Numerical Simulations of General Conservation Laws Using the Space-Time Conservative CESE Method
#52:   08-27-2014, Graeme Kennedy
Structural and Multidisciplinary Design Optimization of Aircraft with Next-Generation Lightweight Materials
#51:   08-26-2014, Hiro Nishikawa
First-, Second-, and Third-Order Hyperbolic Navier-Stokes Solver
|
08-04-2015  
11:00am - noon (EST)  
NIA Room 137 
   
video
M4D, an Open Source Research CFD Code for the Calculation of Classical and Turbulent/Transitional Flows
M4D features unsteady convection adapted control volumes and the MARV/MARVS Reynolds stress models. Convection adapted control volumes are a paradigm shift in CFD. Used with tri-linear discretization of convected properties in space over a fixed grid (formally 3-d 2nd-order accurate), they provide a balance between accuracy and stability not found in fixed volume methods. The MARV/MARVS Reynolds stress models are advanced turbulence models which calculate transition naturally based on an understanding of homogeneous shear flow at high dimensionless strain rates.
The presentation will concentrate on the convection adapted control volumes - the method, the combination of stability and accuracy, with the examples of an inviscid Kelvin-Helmholtz shear layer instability and near-DNS of flow in a square channel. The steady flow Reynolds stress model examples of transitional flow and heat transfer in a turbine cascade (Butler et al.) and of a backward facing step (Kasagi) also use the convection adapted control volumes.
Speaker Bio:
|
It is fifty years since Joan and John Moore met in M.I.T.'s Gas Turbine Laboratory. John had come with a B.Sc. (Eng.) in Mechanical Engineering from Imperial College, London to obtain an S.M. and then an Sc.D from M.I.T. Joan with a B.S. in Applied Mathematics from M.I.T., had the job of writing computer codes and helping Graduate students with theirs. Thus began a life-long CFD and turbulence modeling collaboration. John is currently a Professor Emeritus of Mechanical Engineering at Virginia Tech. Their first 'retirement' project, their book "Functional Reynolds Stress Modeling" was published in 2006. And now Joan has written her 4th CFD code, M4D, but her first one unfettered by external sponsorship.
Further details can be found at Homepage of Joan and John Moore .
|
|
Relevant Publications:
|
Using Multi-Dimensional Linear Discretization Over Unsteady Convection Adapted Control Volumes,
AIAA-2014-2780.
[ Paper (pdf) ]
TRANSITION CALCULATIONS WITH THE MARVS REYNOLDS STRESS MODEL, European Turbomachinery Conference, 2015,
ETC2015-113,
[ Paper (pdf) ]
An RSM/EVM Flow/Heat Transfer Model Applied to Pre-Transitional and Turbulent Boundary Layers, AIAA -2010-4314,
[ Paper (pdf) ]
|
|
07-15-2015  
11:00am - 12:00pm (EST)  
NIA Room 142     
video
Large-Scale Topology Optimization Methods for Additively Manufactured Structures
Additive manufacturing methods give engineers greater freedom to design structures with fewer geometric and processing constraints than conventional manufacturing methods. Additive manufacturing, therefore, has the potential to enable the design and production of low-weight high-performance structures. However, optimization of additively-manufactured structures using conventional optimization techniques, such as topology optimization, is challenging due to the demanding mesh requirements and large size of the design problem. In this presentation, these difficulties will be addressed through scalable methods for large-scale analysis and optimization. The proposed approach utilizes a multigrid-preconditioned Krylov method for solving large structural finite-element problems coupled with a parallel interior-point method for large-scale constrained optimization. The proposed method will be demonstrated on a large-scale mass-constrained compliance minimization problem for a structure discretized using a 64 × 64 × 256 element mesh, resulting in 3.26 million structural degrees of freedom, 5.24 million design variables and 1.05 million linear constraints.
Speaker Bio:
|
Graeme Kennedy is an Assistant Professor in the School of Aerospace Engineering at the Georgia Institute of Technology where he leads his research group focused on developing novel design optimization methods for structural and multidisciplinary aerospace systems. Before joining the Georgia Tech faculty, he worked as a Postdoctoral Research Fellow at the University of Michigan in the Multidisciplinary Design Optimization lab. He received his Ph.D. from the University of Toronto Institute for Aerospace Studies (UTIAS) under the supervision of Prof. Joaquim R.R.A. Martins in 2012 and his M.A.Sc. from UTIAS under the supervision of Prof. Jorn Hansen in 2007. He received his undergraduate degree in Aerospace Engineering from the University of Toronto in 2005. A complete list of papers and ongoing projects is available on Dr. Kennedy's website: http://gkennedy.gatech.edu/..
|
|
|
07-14-2015  
11:00am - 12:00pm (EST)  
NIA Room 142     
video
Flow Analysis and Adjoint-Based Design Optimization Using a High-Order CFD Method
The computational analysis and design for multiscale physics and complex systems demand
computationally accurate and efficient discretization methods. High-order discretization methods
have gained increasing popularity in a wide range of application scenarios in sciences and
engineering. In this presentation, the focuses will be two-fold. First, the techniques of a highorder
discontinuous Galerkin finite element method for aerodynamic turbulent flow simulation
will be discussed. In particular, the spatial discretization procedures and the implementation for
Reynolds Averaged Navier-Stokes as well as large-eddy simulation will be covered in detail. The
second focus will be on the integration of the high-order CFD method with sensitivity analysis
capabilities for aerodynamic shape optimization. The discrete adjoint method is derived in a
mathematically rigorous framework and provides important search directions towards an
optimum design solution. Simulation of turbulence using the developed high-order discontinuous
Galerkin method will be presented with an emphasis on examining the accuracy of different
orders of discretization schemes. Several optimization examples will be described as well to
demonstrate the effectiveness of the discrete adjoint algorithm in steady and unsteady design
optimization.
Speaker Bio:
|
Li Wang earned her PhD degree in Mechanical Engineering from the University of Wyoming in
2009 and she is currently a Research Assistant Professor in the SimCenter at the University of
Tennessee, Chattanooga. She has successfully developed computational software to perform
high-fidelity aerodynamic simulation and design. Her research centers on turbulent flow
simulation and modeling, high-order finite element CFD methods, adjoint-based sensitivity
analysis and design optimization, adaptive meshing, multigrid acceleration strategies, and highperformance
computing. Since 2009 she has extended the software capabilities to high-frequency
electromagnetic applications. She is currently serving as Associate Member of the American
Institute of Aeronautics and Astronautics (AIAA) Applied Aerodynamics Technical Committee
and has been an active participant in multiple research grants and contracts from NASA, DoD
and U.S. Army.
|
|
|
01-29-2015  
11:00am - noon (EST)  
NIA Room 137 
   
video
Towards Aerospace Design in the Age of Extreme-Scale Supercomputing
Extreme scale supercomputing will soon offer a million times the computing power of a desktop - an as drastic upgrade as that from a slide rule to a desktop computer in the 1990s. I believe this will revolutionize how we aerospace engineers work. It will enable us to rapidly and confidently refine and optimize our designs. But this revolution can only happen through innovating our computational algorithms.
In Computational Fluid Dynamics, high-fidelity simulations such as Large Eddy Simulation can often reliably predict the performance of aerospace vehicles and engines. But with today's algorithms, these simulations take days if not weeks. With today's optimization algorithms, it may take months if not years for us to reach a good design. Can we shorten each high fidelity CFD simulation to minutes, by innovating how we solve PDEs, better utilizing the skyrocketing concurrency in supercomputers? Better, can we shorten an entire high fidelity optimization to minutes by innovating how we do optimization, again utilizing more concurrency than we currently can? Even better, can we shorten a high fidelity optimization with hundreds of design parameters to minutes, by computing high fidelity design gradients, even when the simulations are turbulent and chaotic, and gradients in the traditional sense would diverge?
I believe that the answers are yes, yes and yes. In this talk, I will show you why I believe so, and discuss how we all can advance aerospace design into the age of extreme-scale supercomputing.
Speaker Bio:
|
Qiqi Wang is an assistant professor of aeronautics and astronautics at
MIT. He obtained his PhD from Stanford in 2009, and B.S. from University of Science and Technology of China in 2004.
|
|
|
12-16-2014  
11:00am - noon (EST)  
NIA Room 137 
   
video
Accuracy-Preserving Boundary Quadrature for Edge-Based Finite-Volume
Scheme: Third-order accuracy without curved elements
This talk will discuss a third-order edge-based finite-volume scheme on unstructured grids.
It will be shown why the edge-based scheme can be third-order and also why it cannot be
third-order if the numerical flux is exact for quadratic fluxes.
A general boundary flux quadrature formula is presented that preserves third-order
accuracy at boundary nodes with linear elements.
Numerical results show that the general formula as well as acccurate boundary normals
are essential to achieve third-order accuracy for a curved boundary problem with linear elements.
Speaker Bio:
|
Dr. Hiro Nishikawa is Associate Research Fellow, NIA.
He earned Ph.D. in Aerospace Engineering and Scientific
Computing at the University of Michigan in 2001. He then worked as a postdoctoral fellow at the University
of Michigan on adaptive grid methods, local preconditioning methods, multigrid methods,
rotated-hybrid Riemann solvers, high-order upwind and viscous schemes, etc., and joined NIA in 2007.
His area of expertise is the algorithm development for CFD, focusing on
hyperbolic methods for robust, efficient, highly accurate viscous discretization schemes.
[ Homepage |
CFD book |
Free CFD codes |
CFD Notes
]
|
|
Relevant Publication:
|
H. Nishikawa, " Accuracy-preserving boundary flux quadrature for finite-volume discretization on unstructured grids",
Journal of Computational Physics, Volume 281, January 2015, Pages 518-555, 2015.
[ bib |
pdf | journal
]
First, Second, and Third-Order Finite-Volume Schemes for Navier-Stokes Equations, AIAA Paper, 2014-2091.
[ AIAA Paper 2014-2091 (pdf) ]
|
|
12-02-2014  
11:00am - Noon (EST)  
NIA Room 137 
   
webcast
Entropy stable wall boundary conditions for the compressible Navier-Stokes equations
Non-linear entropy stability and a summation-by-parts framework are used to derive entropy
stable wall boundary conditions for the compressible Navier-Stokes equations. A semi-
discrete entropy estimate for the entire domain is achieved when the new boundary condi-
tions are coupled with an entropy stable discrete interior operator. The data at the boundary
are weakly imposed using a penalty flux approach and a simultaneous-approximation-term
penalty technique. Although discontinuous spectral collocation operators are used herein
for the purpose of demonstrating their robustness and efficacy, the new boundary conditions
are compatible with any diagonal norm summation-by-parts spatial operator, including fi-
nite element, finite volume, finite difference, discontinuous Galerkin, and flux reconstruction
schemes. The proposed boundary treatment is tested for three-dimensional subsonic and
supersonic flows. The numerical computations corroborate the non-linear stability (entropy
stability) and accuracy of the boundary conditions.
Speaker Bio:
|
Dr. Matteo Parsani is currently a Post-doctoral Fellow at NASA Langley Research Center, working with Dr. Mark H. Carpenter. Prior to this he was a Post-doctoral Researcher at KAUST, working in the group of Professor David Ketcheson. He received his Ph.D. in Mechanical and Aerospace Engineering from Free University of Brussels in December 2010. His research interests include high-order accurate methods for large-eddy simulation and aeroacoustics, efficient explicit and implicit time integrators and acceleration techniques for compressible flows. In 2011 his PhD thesis was selected among the best National Ph.D. theses to be presented at the ECCOMAS Olympiad workshop in Athens.
|
|
Relevant Publication:
|
Matteo Parsani, Mark H. Carpenter, and Eric J. Nielsen, Entropy Stable Wall Boundary Conditions for the Compressible Navier-Stokes Equations, NASA-TM--2014-218282.
[ NASA-TM--2014-218282 (pdf) ]
|
|
10-06-2014  
1:00pm - 2:00pm (EDT)  
NIA Room 101 
   
video not available (realtime webcast only)
Recent Applications of Overset Mesh Technology in SC/Tetra
Software Cradle has introduced overset mesh technology to SC/Tetra since 2007, and we are dealing with a lot of demands from industries using the technology. This is a necessary function for flow simulation with moving components in complex geometries, which is not uncommon in real products. SC/Tetra's overset technology has been improved continuously, and is now able to be coupled with free-surface analysis, flow-structure interaction and dynamical moving components (6DOF). In the latest version, a special treatment for analyses of the flow in piston engines was introduced. The treatment achieves practical accuracy for mass conservation in the cylinder.
|
|
09-09-2014  
11:00am - noon (EDT)  
NIA Room 137 
   
video
Numerical Simulations of General Conservation Laws Using the Space-Time Conservative CESE Method
Governing equations of most engineering disciplines can be written as general conservation laws by enforcing mass, momentum, and energy balances. Modern computational methods are devised to provide accurate solutions to these conservation laws in the discretized space. The space-time conservation element solution element (CESE) method introduced in 1990s is a numerical framework for general conservation laws designed to provide discretized solutions in the space-time domain with considerations to ensure accuracy and robustness. The CESE method is constructed based on a non-dissipative, space time inversion invariant core scheme. Numerical dissipations are added as required. Discretized equations for dependent variables and high derivatives are formulated by enforcing both local and global conservations. It can be shown that fundamental quantities such as mass, momentum, and energy are strictly conserved both in the local conservation elements as well as the entire computational domain. To handle solutions with discontinuities, the integration volumes have interfaces that only encompass the smooth regions where solution polynomials are valid. With these constructs, the CESE numerical framework is free of ad-hoc reconstructions of physical quantities associated with interfacial discontinuity or approximations of kinetic energy. This talk discusses the fundamental concepts and development of the CESE framework with primary focus on 3D Navier-Stokes computations. High fidelity simulations of problems with multiple temporal/spatial-scales and physics are tackled with time accurate local time-stepping and high-order frameworks for unstructured meshes. Applications of the CESE method in other disciplines outside of NASA will be briefly discussed.
Speaker Bio:
|
Dr. Chau-Lyan Chang is a research scientist from the Computational AeroSciences Branch at NASA Langley Research Center. His primary research interest is in unstructured mesh CFD methods and code development. He also works on numerical computations of boundary layer stability and transitions. He actively maintains LASTRAC software and interacts with users from academia and industry.
.
|
|
|
08-27-2014  
11:00am - noon (EDT)  
NIA Room 101 
   
video
Structural and Multidisciplinary Design Optimization of Aircraft with Next-Generation Lightweight Materials
The use of advanced lightweight structures has enabled significant performance improvements for the present generation of transport aircraft. New structural materials, manufacturing techniques and multi-functional structural technologies will lead to even greater improvements for future aircraft. These new technologies give engineers greater flexibility to tailor aircraft structures to meet stringent design requirements. However, the large design space associated with this flexibility can be difficult to navigate since there is a limited knowledge base to help guide design decisions. Advanced computational design methods that employ high-fidelity structural and multidisciplinary analysis are key tools to help engineers understand the complex trade-offs inherent in aircraft design, especially in the context of advanced structural technologies. In this seminar, I will present our work on structural and aerostructural optimization that begins to address these challenges. To meet the computational demands of high-fidelity simulation and design, we use gradient-based design optimization techniques in conjunction with parallel computational methods and efficient adjoint-based derivative evaluation. To illustrate our efforts in these areas, I will describe the development of our in-house parallel finite-element code designed for multidisciplinary analysis and gradient-based optimization of composite structures called the Toolkit for the Analysis of Composite Structures (TACS). To demonstrate the capabilities of our structural and aerostructural design optimization framework, I will present the results of a study comparing the design of metallic and composite wings for a large transport aircraft. These results will show the benefits of using an integrated, gradient-based aerostructural analysis and design optimization framework.
Speaker Bio:
|
Dr. Graeme Kennedy is an Assistant Professor in the School of Aerospace Engineering at the Georgia Institute of Technology where he leads his research group focused on developing novel design optimization methods for structural and multidisciplinary aerospace systems. Before joining the Georgia Tech faculty, he worked as a Postdoctoral Research Fellow at the University of Michigan in the Multidisciplinary Design Optimization lab. He received his Ph.D. from the University of Toronto Institute for Aerospace Studies (UTIAS) under the supervision of Prof. Joaquim R.R.A. Martins in 2012 and hisM.A.Sc. from UTIAS under the supervision of Prof. Jorn Hansen in 2007. He received his undergraduate degree in Aerospace Engineering from the University of Toronto in 2005. A complete list of papers and ongoing projects is available on Dr. Kennedy's website: http://gkennedy.gatech.edu/.
|
|
|
08-26-2014  
11:00am - noon (EDT)  
NIA Room 137 
   
video
First-, Second-, and Third-Order Hyperbolic Navier-Stokes Solver
Is it possible that a third-order CFD solver is less expensive on a given grid than a conventional second-order solver? No, it is impossible because a higher-order scheme requires more work on the same grid. However, as history demonstrates, it only takes a radical idea to turn the impossible into the possible. This talk will investigate whether the hyperbolization of the viscous terms is radical enough to make it happen. The Navier-Stokes equations are made hyperbolic, discretized by first, second, and third-order finite-volume schemes with upwind fluxes, and solved by a fully implicit solver: Newton's method for the first-order scheme, and a defect correction method for others. The developed solver will be compared with a conventional second-order solver for some simple but realistic viscous flow problems, focusing on computation time and accuracy especially in the viscous stresses and heat fluxes on fully unstructured viscous grids.
Speaker Bio:
|
Dr. Hiro Nishikawa is Associate Research Fellow, NIA.
He earned Ph.D. in Aerospace Engineering and Scientific
Computing at the University of Michigan in 2001. He then worked as a postdoctoral fellow at the University
of Michigan on adaptive grid methods, local preconditioning methods, multigrid methods,
rotated-hybrid Riemann solvers, high-order upwind and viscous schemes, etc., and joined NIA in 2007.
His area of expertise is the algorithm development for CFD, focusing on
hyperbolic methods for robust, efficient, highly accurate viscous discretization schemes.
[ Homepage |
CFD book |
Free CFD codes |
CFD Notes
]
|
|
Relevant Publications:
|
First, Second, and Third-Order Finite-Volume Schemes for Navier-Stokes Equations, AIAA Paper, 2014-2091.
[ AIAA Paper 2014-2091 (pdf) ]
First, Second, and Third-Order Finite-Volume Schemes for Advection-Diffusion, JCP, Volume 273, 2014.
[ Preprint (pdf) |
Journal ]
First, Second, and Third-Order Finite-Volume Schemes for Diffusion, JCP, Volume 256, 2014
[ Preprint (pdf) |
Journal ]
|
|
|