NIA CFD Seminar, Season 6 (20162017)

#91: 07252017, Duc T. Nguyen
Finite Element Domain Decomposition Algorithms & Software With Applications in
Civil/Structural/Mechanical/Aerospace/ElectroMagnetic/Acoustics/Transportation Engineering
#90: 07182017, Nathaniel Hildebrand
Stability and Sensitivity Analysis for the Control of ShockDriven Flows
#89: 06272017, Asitav Mishra
Timedependent Adjointbased Optimization for Coupled FluidStructure Problems
#88: 06202017, Graeme Kennedy
A Robust and Flexible Coupling Framework for Aeroelastic Analysis and Optimization
#87: 06142017, Angxiu Ni
Sensitivity analysis on chaotic dynamical systems by NonIntrusive Least Squares Shadowing (NILSS)
#86: 04252017, Daming Feng
Scalable Parallel Delaunay ImagetoMesh Conversion for Shared and Distributed Memory Architectures
#85: 04112017, David Del Rey Fernández
A brief introduction to summationbyparts methods and their various flavors
#84: 03212017, Ali Uzun
WallResolved Large Eddy Simulations of Separated Flows: Part Two
#83: 03072017, Hiro Nishikawa
Hyperbolic NavierStokes Method for HighReynoldsNumber BoundaryLayer Flows
#82: 02212017, Yi Liu
ThirdOrder Inviscid and SecondOrder Hyperbolic NavierStokes Solvers for ThreeDimensional Unsteady Inviscid and Viscous Flows
#81: 02142017, Hiro Nishikawa
AccuracyPreserving Source Term Quadrature for ThirdOrder EdgeBased Discretization
#80: 12132016, Venkat Raman
Understanding the Role of Thermal Nonequilibrium in Scramjet Flows
#79: 10172016, Yi Liu
Implementation of ThirdOrder EdgeBased Scheme into FUN3D for Steady and Unsteady
Flows, Part I: Thirdorder Inviscid and Secondorder Hyperbolic NavierStokes Schemes
#78: 09272016, Juliette Pardue
Parallel TwoDimensional Unstructured Anisotropic Delaunay Mesh Generation of Complex Domains for Aerospace Applications
#77: 09202016, Sameer B. Mulani
Polynomial Chaos Decomposition with Differentiation and Applications

91st NIA CFD Seminar:
07252017
11:00amnoon (EST)
NIA Room 137
video
Finite Element Domain Decomposition Algorithms & Software
With Applications in Civil/Structural/Mechanical/Aerospace/ElectroMagnetic/Acoustics/Transportation Engineering
Serial and parallel computing methods for use in largescale structural analysis, and other engineering/science applications (such as Computational Fluid Dynamics, Acoustics, Electromagnetic, Transportation Engineering Modeling, Automotive Engineering, Mechanical/Aerospace/Civil/Electrical Engineering disciplines) are highlighted.
The seminar speaker will briefly explain important numerical topics, such as parallel sparse/dense matrix operations, Domain Decomposition (DD) partitioned algorithms, mixed directiterative (linear) solvers with DD formulation, solution algorithms for solving systems of Simultaneous Linear Equations (SLE) where the system (stiffness) matrix can be symmetrical (or unsymmetrical), positive (or negative) definite, or indefinite, how to handle SLE with MultiPoint Constraints (MPC), efficient algorithms for solving system of nonlinear equations, Sensitivity Analysis with DD formulation, Genetic and Differential Evolution (DE) Algorithms for size, shape and topology optimization.
Medium to largescale, practical NASA finite element models [such as the High Speed Civil Transport (HSCT) Aircraft, space shuttle Solid Rocket Booster (SRB), NASA LaRC finite element acoustic 3D model], Automobile Finite Element model, Shortest Paths in reallife Transportation Networks, sizeshapetopology optimization for 2D bridge structures, etc... will be presented to demonstrate the effectiveness of the developed numerical algorithms.
Speaker Bio:

Dr. Duc T. Nguyen obtained his B.S. (Northeastern University, 1974), M.S. (U.C. Berkeley, 1976), and Ph.D. (University of Iowa, 1982) degrees in Civil/Structural Engineering. He has been a Structural Engineering faculty at Old Dominion University since 1985. His teaching activities and research in LargeScale Parallel (Finite Element) Computational Mechanics, Equation & EigenSolvers, Sensitivity Analysis & Engineering Optimization, and in STEM Related Educational Topics have led to several international/national/regional awards.



90th NIA CFD Seminar:
07182017
11:00amnoon (EST)
NIA Room 141
video
Stability and Sensitivity Analysis for the Control of ShockDriven Flows
A linear global mode approach using direct and adjoint equations has been used in the past to
study the dynamics of mixing layers, free jets, wakes, boundary layers, and other open flows.
This talk presents the results from applying a similar analysis to a transitional, hypersonic
Oblique Shock Wave/Boundary Layer Interaction (OSWBLI) and a supersonic impinging jet.
These complex flows are connected because they are both compressible and have multiple
influential shock waves. They also happen to be globally unstable at certain flow conditions.
The OSWBLI at Mach 5.92 bifurcates from its original laminar state at a critical shock wave
angle of 12.9 degrees according to Direct Numerical Simulation (DNS) and Global Stability
Analysis (GSA). At bifurcation, the least stable global mode is nonoscillatory, and it selects a
spanwise wavenumber of 0.25. This value agrees well with the DNS results. Examination of the
critical global mode reveals it to be the result of interactions between small spanwise
corrugations at the base of the incident shock, streamwise vortices inside the recirculation
bubble, and spanwise modulation of the bubble strength. Furthermore, we compute the
wavemaker, which is defined as the sensitivity of an eigenvalue to base flow modification. The
wavemaker indicates that streamwise vortices inside the recirculation bubble are crucial to
instability further verifying our physical interpretation.
Multiblock GSA about an ideally expanded round impinging jet at Mach 1.5 revealed several
unstable modes with frequencies that matched experimentally measured impingement tones from
the Florida State University. Both Large Eddy Simulations (LES) and the ReynoldsAveraged
NavierStokes (RANS) equations were used to create the base flows. Using an adjoint stability
solver, we show that the supersonic impinging jet is most sensitive to base flow modification in
the downstream shear layer.
Speaker Bio:

Nathaniel Hildebrand is a Ph.D. candidate in the Aerospace Engineering and Mechanics
Department at the University of Minnesota  Twin Cities. He earned his Bachelor of Science in
Physics and Applied Mathematics from the University of Wisconsin  La Crosse. After his time
as an undergraduate student, he decided to pursue graduate level research at the University of
Minnesota under Joseph Nichols. He obtained his Masters of Science in Aerospace Engineering
and Mechanics near the end of 2016. Recently, he started working as a Pathways Intern in the
Computational AeroSciences Branch at NASA LaRC. His research interests include
aeroacoustics, transition to turbulence, hydrodynamic stability, and DNS/LES of highspeed
flows. More specifically, he is applying multiblock global stability analysis to better understand
the physics of supersonic impinging jets. He is also using stability and sensitivity analysis to
study the interaction of an oblique shock wave impinging on a Mach 5.92 laminar boundary
layer at a transitional Reynolds number.



89th NIA CFD Seminar:
06272017
11:00amnoon (EST)
NIA Room 101
video
Timedependent Adjointbased Optimization for Coupled FluidStructure Problems
In the recent past, the use of adjoint equations has become a popular approach for solving aerodynamic design optimization problems based on computational fluid dynamics (CFD). While the use of adjoint equations is now fairly well established in steadystate shape optimization, only recently have inroads been made into extending them to unsteady flow problems. This talk presents a formulation for sensitivity analysis of fully coupled timedependent aeroelastic problems in both hover and forward flight conditions. Sensitivity analysis for forward flight is considered with trim constraints. It includes the effect of blade shape parameters as well as blade cyclic pitch control parameters to enable analysis and optimization of rotors in a trimmed condition. Both forward sensitivity and adjoint sensitivity formulations are derived that correspond to analogues of the fully coupled nonlinear aeroelastic analysis problem. Both the sensitivity analysis formulations make use of the same iterative disciplinary solution techniques used for analysis, and make use of an analogous coupling strategy. The information passed between fluid and structural solvers is dimensionally equivalent in all cases, enabling the use of the same data structures for analysis, forward and adjoint problems. Upon successful verification of the fully coupled adjoint formulation, it is used to perform trim and aerodynamic shape design optimization for helicopter rotors in both hover and forwardflight conditions.
Speaker Bio:

Asitav Mishra is a postdoctoral scholar in the Department of Aerospace Engineering at the University of Michigan since 2015. Dr. Mishra obtained his Ph.D in Aerospace Engineering from the University of Maryland in 2012. Upon graduation, he held postdoctoral position at the University of Wyoming until 2015. His research interests include adjoint based coupled multidisciplinary fixed and rotarywing design optimization, vortex wakelifting surface interactions as well as performance predictions in rotary wing flows, and datadriven turbulence modeling of vortex dominated flows.



88th NIA CFD Seminar:
06202017
11:00amnoon (EST)
NIA Room 141
video
A Robust and Flexible Coupling Framework for Aeroelastic Analysis and Optimization
Novel aircraft configurations and advanced materials are enabling the use of slender, flexible wings and lifting surfaces that improve the performance of nextgeneration aircraft. These slender structures, however, are more susceptible to adverse aeroelastic phenomena. To avoid excessive weight penalties, aeroelastic effects must be considered early in the design process. This paper presents a coupled highfidelity aeroelastic framework for analysis and design optimization that can be used to address this design challenge. The framework implements two novel features that increase its flexibility and robustness compared with previous work. First, the governing equations are coupled using a load and displacement transfer scheme that is independent of the underlying finiteelement mesh and is accurate and robust for large deflections and rotations. Second, the analysis and adjoint method are formulated in a way that simplifies the introduction of new disciplines in the optimization problem. The implementation of the discrete adjoint method is verified for aerodynamic, geometric, and structural design variables using the complexstep method. The results from a preliminary design optimization of the Common Research Model wing geometry are used to demonstrate the flexibility of the proposed framework.
Speaker Bio:

Graeme J. Kennedy is an Assistant Professor of Aerospace Engineering at the Georgia Institute of Technology. Dr. Kennedy obtained his Bachelor of Applied Science in Aerospace Engineering from the University of Toronto in 2005. After graduation, he accepted an opportunity to pursue graduate studies at the University of Toronto Institute for Aerospace Studies (UTIAS). He subsequently obtained his Masters of Applied Science and Ph.D in aerospace engineering in 2007 and 2012, respectively. Upon graduation, Dr. Kennedy accepted a postdoctoral fellowship at the University of Michigan in 2012. In November, 2013, Dr. Kennedy joined the Georgia Institute of Technology as an Assistant Professor in the School of Aerospace Engineering. Dr. Kennedy's research interests lie in the area of structural and multidisciplinary design optimization of aerospace vehicles. The goal of his work is to develop and apply novel methods and algorithms to enable the design of aerospace vehicles with improved structural and systemlevel performance such as fuel burn reduction, leading to both economic and environmental benefits. Dr. Kennedy is a member of the AIAA, AHS International and SIAM. He is a recipient of the AIAA Best Multidisciplinary Design Optimization Paper award in 2015.



87th NIA CFD Seminar:
06142017
11:00amnoon (EST)
NIA Room 141
video
Sensitivity analysis on chaotic dynamical systems by NonIntrusive Least Squares Shadowing (NILSS)
This talk presents the NonIntrusive Least Squares Shadowing (NILSS) method, which computes the sensitivity for longtime averaged objectives in chaotic dynamical systems.We will show that perturbing the parameters of a simulation has a transient effect and a longtime effect. The transient effect causes similar simulations to diverge, whereas the longtime effect that shifts the chaotic attractor, and thereby perturbs the longtime averages of the simulation. Although both effects are contained in the solution of a conventional tangent equation, the transient effect is usually orders of magnitude larger than the longtime effect.
Computing the sensitivity for longtime averages requires finding a tangent solution that contains the longterm effect but tiny transient effect. NILSS achieves this by determining a linear combination of homogeneous tangent solutions and subtracting it from the conventional inhomogeneous tangent. The resulting inhomogeneous tangent solution, now with a nonzero condition, can be used to compute the sensitivity of interest.
NILSS has two advantages. The first is that it is relatively easy to implement with existing solvers. The second is that, for chaotic systems with many degrees of freedom but a few unstable modes, NILSS has a low computational cost, sometimes even comparable to that of the primal simulation.
At the end of the talk, we show the application of NILSS onto several chaotic PDE systems: 1) the Lorenz 63 system, 2) a twodimensional CFD simulation of flow over a backwardfacing step, and 3) a threedimensional CFD simulation of flow over a cylinder. With reasonable cost, NILSS gives sensitivities that reflect the trends in the longtime averaged objectives of these dynamical systems.
Speaker Bio:

Angxiu Ni is a new graduate student in the Department of Math at UC Berkeley. He received his master of science in aerospace engineering from MIT, under the advising of Qiqi Wang. His research interest is in computational methods for ODE/PDE. He developed the NonIntrusive Least Squares Shadowing (NILSS) method with Qiqi Wang, a fast method for sensitivity analysis of chaotic dynamical systems.



86th NIA CFD Seminar:
04252017
11:00amnoon (EST)
NIA Room 137
video
Scalable Parallel Delaunay ImagetoMesh Conversion for Shared and Distributed Memory Architectures
Scalable, stable and portable parallel ImagetoMesh (I2M) conversion algorithms with quality and fidelity guarantees are important for the real world bioengineering and medical applications, such as the image guided therapy, patientspecific interactive surgery simulation and so on. Supercomputers, either shared memory or distributed memory, are more and more popular to be used for high performance computing. However, most current parallel mesh generation algorithms are desktopbased. Such mesh generation algorithms, when run on supercomputers, are either conservative in leveraging available concurrency or depend on the solution of the domain decomposition problem which makes the scalability of these algorithms very limited. In addition, the implementation of parallel mesh generation algorithms on supercomputers brings new challenges because of their special memory architecture. Therefore, implementing an efficient parallel mesh generation algorithm targeting the shared and distributed supercomputers is still an open problem. I will describe several parallel mesh generation algorithms we proposed that are scalable on shared and distributed memory supercomputers. These algorithms will also contribute to the understanding of the challenging characteristics of adaptive and irregular applications on supercomputers consisting of thousands of cores.
Speaker Bio:

Daming Feng is a PhD candidate in Computer Science Department of Old Dominion University. He works since 2012 as a research assistant in the Center for RealTime Computing (CRTC) in Old Doninion University, Norfolk, VA. He advised by Dr. Nikos Chrisochoides and Dr. Andrey Chernikov. His current research focuses on quality and fidelity guaranteed mesh generation and high scalable parallel mesh generation for shared and distributed memory architectures. His other research interests mesh visualization, hybrid programming and high performance computing. Especially, he is interested in applying the integration of mesh generation and hybrid programming techniques to analyze and solve the challenging real word applications such as interactive surgery simulation and gene model expression. He published more than 10 papers and some of them are published on the journals with high cited rate such as Parallel Computing and ComputerAided Design. He earned his B.S. in Computer Science Department from Harbin University of Science of Technology and M.S. in Computer Science Department from Soochow University in China.



85th NIA CFD Seminar:
04112017
11:00amnoon (EST)
NIA Room 137
video
A brief introduction to summationbyparts methods and their various flavors
The concept of matrix difference operators having the summationbyparts (SBP) property has its origin in the finitedifference (FD) community with the goal of mimicking finiteelement energy analysis techniques for proving linear stability. The essential feature of these operators is that they are equipped with a highorder approximation to integration by parts. When combined with appropriate procedures for interelement coupling and imposition of boundary conditions, the resulting SBP framework allows for a onetoone correspondence between discrete and continuous stability proofs and in this way naturally guides the construction of robust algorithms.
In recent years, there has been a veritable explosion in the SBP concept. The SBP framework has been applied to nodal discontinuous and continuous Galerkin methods, the fluxreconstruction method, and has been shown to have a subcell finitevolume interpretation. The SBP concept has been extended to nontensor nodal distributions thereby introducing the ability to construct SBP schemes on unstructured tetrahedral meshes. Nonlinearly robust schemes have been constructed by enforcing discrete entropy stability. SBP schemes on nonconforming meshes that remain conservative and stable have been developed. Dualconsistent schemes have been developed that lead to superconvergent functional estimates, etc. In summary, the SBP concept enables the construction of flexible and robust numerical methods that have advantageous properties within a rigorous mathematical framework.
In this talk I will give a brief introduction to the SBP concept, starting with the tensorproduct variety and its various flavors. I will then show how these ideas extend to nontensor nodal distributions and will finish with some remarks on nonlinear stability.
Speaker Bio:

Dr. David Del Rey Fernández obtained his PhD in aerospace engineering from the University of Toronto Institute for Aerospace Studies (UTIAS) in 2015. He remained at UTIAS until December 2016 as a Postdoctoral Fellow and lecturer. Currently he is a Postdoctoral Fellow at NIA and NASA LaRC in the Computational AeroScience Branch. His research interests revolve around the development of flexible and robust highorder numerical methods for the solution of partial differential equations. Currently he is working on developing architectureaware nonlinearly stable summationbyparts methods applicable to structured and unstructured meshes that retain their conservation and nonlinear stability properties under various adaptive strategies such as AMR.



84th NIA CFD Seminar:
03212017
11:00amnoon (EST)
NIA Room 137
video
WallResolved Large Eddy Simulations of Separated Flows: Part Two
This talk will discuss our ongoing work on the wallresolved large eddy simulations (WRLES) of complex high Reynolds number separated flows. Such flows are very challenging to predict accurately due to the significant computational requirements of eddyresolving simulations for high Reynolds number turbulence. Recent improvements in computing capability and numerical algorithms are now making it more feasible to simulate such flows using billions of grid points. This talk will present an overview of the work completed since our previous related presentation given in the NIA CFD Seminar Series in February 2016. Several problems encountered during the course of this work will be discussed and viable solutions will be presented. Representative results from the benchmark NASA wallmounted hump and the BachaloJohnson transonic shockinduced flow separation problems will be shown. Comparisons with available experimental measurements will be made to assess the predictive capability of the simulations.
Speaker Bio:

Dr. Ali Uzun joined NIA as a Senior Research Scientist in July 2015. He received his Ph.D. in Aeronautics & Astronautics from Purdue University in 2003. He joined the Florida State University as a postdoctoral research associate immediately after completing his Ph.D. and later became a Research Scientist at the Florida Center for Advanced AeroPropulsion, Florida State University. His current research interests include computational fluid dynamics using highorder numerical methods, turbulence simulations, computational aeroacoustics and parallel computing.



83rd NIA CFD Seminar:
03072017
11:00amnoon (EST)
NIA Room 137
video
Hyperbolic NavierStokes Method for HighReynoldsNumber BoundaryLayer Flows
This talk will discuss serious accuracy and stability problems in hyperbolic schemes for highReynoldsnumber boundarylayer calculations. For a model onedimensional problem, a simple fix is derived by minimizing a firstorder discretization error, which suggests that the relaxation length scale must be inversely proportional to the Reynolds number. Accurate and efficient computations are demonstrated for model equations in one and two dimensions. Extensions to 3D NavierStokes computations are also discussed.
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, rotatedhybrid Riemann solvers, highorder 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.



82nd NIA CFD Seminar:
02212016
11:00amnoon (EST)
NIA Room 137
video
ThirdOrder Inviscid and SecondOrder Hyperbolic NavierStokes Solvers for ThreeDimensional Unsteady Inviscid and Viscous Flows
We present thirdorderinviscid implicit edgebased solvers for unsteady inviscid and viscous flows on unstructured tetrahedral grids. Steady thirdorderinviscid solvers recently developed in NASA's FUN3D code are extended to unsteady computations with implicit timestepping schemes. The physical time derivative is discretized by a backwarddifference formula, and incorporated into the thirdorder edgebase scheme as a source term. In the thirdorder edgebased scheme, the source term needs to be discretized in space by a special formula to preserve thirdorder accuracy. A very economical source discretization formula is derived, and the resulting unsteady thirdorder unstructuredgrid scheme is made completely free from second derivative computations. Developed unsteady schemes are investigated and compared for some representative test cases for unsteady inviscid and viscous flows. Preliminary results are also presented for a basic study of exploring a more stable thirdorder time integration scheme.
Speaker Bio:

Dr. Yi Liu graduated from Georgia Institute of Technology with a Ph.D degree in aerospace engineering in 2003. He also holds a M.E. from Beijing University of Aeronautics and Astronautics in Beijing, China. In 2004, he joined the National Institute of Aerospace after a oneyear postdoctoral fellowship at Georgia Tech. He has previously served as a senior research engineer at NIA in the area of computational fluid dynamics (CFD) and multidisciplinary analysis of rotorcraft configurations. He has conducted various research projects, including work in the areas of rotorcraft aerodynamic analysis and acoustic prediction; microair vehicle and flapping wing aerodynamics sponsored by ARL and NASA. Currently, he is conducting the research project of implementation of thirdorder edgebased scheme in NASA CFD solver FUN3D with collaboration of researchers at NASA LaRCComputational AeroSciences Branch.



81st NIA CFD Seminar:
02142017
11:00amnoon (EST)
NIA Room 137
video
AccuracyPreserving Source Term Quadrature for ThirdOrder EdgeBased Discretization
This talk will present new source term quadrature formulas for preserving thirdorder accuracy of the edgebased discretization for conservation laws with source terms on arbitrary simplex grids. A family of economical formulas are derived, which do not require computations nor storage of second derivatives, by eliminating a firstorder truncation error and satisfying a compatibility condition. With these formulas, the edgebased scheme can achieve thirdorder accuracy with no secondderivative computations at all. Thirdorder accuracy is demonstrated for equations with source terms on onedimensional grids and linear triangular/tetrahedral grids over straight and curved geometries.
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, rotatedhybrid Riemann solvers, highorder 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.


Relevant Publications:

AccuracyPreserving Source Term Quadrature for ThirdOrder EdgeBased Discretization, Journal of Computational Physics, Volume 344, September 2017, Pages 595622
[ bib 
pdf  journal
]


80th NIA CFD Seminar:
12132016
11:00amnoon (EST)
NIA Room 137
No Video
Understanding the Role of Thermal Nonequilibrium in Scramjet Flows
In most application of continuum fluid mechanics, the molecular motion is assumed to be at or near equilibrium, allowing the definition of a temperature of the fluid. In highspeed flows, the presence of shocks always causes nonequilibrium. More importantly, the short flow time scales imply that equilibrium may not be reached in the domain of interest. Such nonequilibrium has been widely studied in the context of hypersonic flows with application in reentry dynamics. Here, we study such nonequilibrium in internal flows of relevance of scramjets. Thermal nonequilibrium most directly affects chemical reactions. Since flame stabilization is a critical factor in the successful operation of a scramjet engine, the role of a specific form of nonequilibrium, that of vibrational motions, is studied in detail here. Novel abinitio derived rates are used to describe the ignition process. In particular, techniques to determine rates at relatively low enthalpy conditions (as compared to external flows) were developed. Direct numerical simulations of scramjettype flows (performed on the NASA HECC machines) show that nonequilibrium affects ignition counterintuitively, and is highly geometry dependent.
Speaker Bio:

Venkat Raman received his PhD from Iowa State University in 2003 in the department of chemical engineering. He was a NASA/Center for Turbulence Research Postdoctoral Fellow at Stanford University from 20032004, and a research associate in the Center for Integrated Turbulence Simulations from 20042005. From 20052014, he was on the faculty of Aerospace Engineering and Engineering Mechanics Department at The University of Texas at Austin, initially as an assistant professor (20052011) and later as tenured associate professor (20112014). Raman received an NSF CAREER award, a distinguished paper award at the International Combustion Symposium, and the Moncrief Grand Challenge Award in 2013. He held the Eli. H and Ramona Thornton Centennial Fellowship in Engineering at UT Austin. Raman focuses on the development of computational models for turbulent reacting flows with application to aircraft and scramjet engines, stationary power generation, and synthesis of novel materials. His research group uses high performance supercomputers and detailed numerical simulations to study the performance of combustion devices. His recent focus has been in the areas of numerical error analysis, uncertainty quantification, and failure predictions, aimed towards modeling catastrophic and rare events in complex devices and natural systems.



79th NIA CFD Seminar:
10172016
11:00amnoon (EST)
NIA Room 137
video
Implementation of ThirdOrder EdgeBased Scheme into FUN3D for Steady and Unsteady Flows, Part I: Thirdorder Inviscid and Secondorder Hyperbolic NavierStokes Schemes
This talk is to present the implementation of thirdorderinviscid implicit edgebased solvers for threedimensional steady and unsteady flows on unstructured tetrahedral grids. Thirdorder edgebased scheme has been implemented into NASA's FUN3D code for inviscid terms. Secondorder edgebased hyperbolic NavierStokes schemes, which achieve thirdorder accuracy in the inviscid terms, have also been implemented. For sources terms coming from manufactured solutions or unsteady BDF time derivatives, a divergence form of the source term discretization scheme has also been implemented into FUN3D, where thirdorder accuracy can be achieved with source terms presented by unsteady flow. Some key improvements are reported for the hyperbolic NavierStokes schemes. Thirdorder accuracy is verified by the method of manufactured solutions for steady flows and by inviscid moving vortex for unsteady flows for unstructured tetrahedral grids. Developed schemes are compared for some representative test cases for threedimensional inviscid and viscous flows.
Speaker Bio:

Dr. Yi Liu graduated from Georgia Institute of Technology with a Ph.D degree in aerospace engineering in 2003. He also holds a M.E. from Beijing University of Aeronautics and Astronautics in Beijing, China. In 2004, he joined the National Institute of Aerospace after a oneyear postdoctoral fellowship at Georgia Tech. He has previously served as a senior research engineer at NIA in the area of computational fluid dynamics (CFD) and multidisciplinary analysis of rotorcraft configurations. He has conducted various research projects, including work in the areas of rotorcraft aerodynamic analysis and acoustic prediction; microair vehicle and flapping wing aerodynamics sponsored by ARL and NASA. Currently, he is conducting the research project of implementation of thirdorder edgebased scheme in NASA CFD solver FUN3D with collaboration of researchers at NASA LaRCComputational AeroSciences Branch.



78th NIA CFD Seminar:
09272016
11:00amnoon (EST)
NIA Room 137
video
Parallel TwoDimensional Unstructured Anisotropic Delaunay Mesh Generation of Complex Domains for Aerospace Applications
In this paper, we present a bottomup approach to parallel anisotropic mesh generation by building a mesh generator from principles. Applications focusing on highlift design or dynamic stall, or numerical methods and modeling test cases still focus on twodimensional domains. Our pushbutton parallel mesh generation approach can generate highfidelity unstructured meshes with anisotropic boundary layers for use in the computational fluid dynamics field. The anisotropy requirement adds a level of complexity to a parallel meshing algorithm by making computation depend on the local alignment of elements, which in turn is dictated by geometric boundaries and the density functions. Our experimental results show 70% parallel efficiency over the fastest sequential isotropic mesh generator on 256 distributed memory nodes.
Speaker Bio:

Juliette Pardue is a PhD student in Computer Science at Old Dominion University under Dr. Andrey Chernikov and Dr. Nikos Chrisochoides. Her research interests include mesh generation, parallel algorithms, and computational geometry. She published a research note on a 2D sharedmemory parallel mesh generator in the 24th International Meshing Roundtable. Later she published a paper on a 2D distributedmemory parallel mesh generator in the 45th International Conference on Parallel Processing (21% acceptance rate) where her paper won the best paper award. She is currently concluding the 2D distributedmemory parallel mesh generation work with plans to submit a journal article to IEEE Transactions on Parallel and Distributed Systems, before moving on to work on a 3D distributedmemory parallel mesh generator.



77th NIA CFD Seminar:
09202016
10:00am11:00am (EST)
NIA Room 137
video
Polynomial Chaos Decomposition with Differentiation and Applications
A new nonintrusive polynomial chaos (PC) method is proposed where the response and input random variables are expanded in a similar way to the traditional PC that is then followed by differentiation of basis polynomials as well as sensitivity calculation of the response. Here, two analytical problems are studied with three different techniques for sensitivity calculation: 1) analytical differentiation, 2) higher order forward finite difference, and 3) first order forward finite difference with three different stepsizes. This "higher order finite difference" is also a new technique obtained by using Taylor series expansion and it has been observed that the results obtained by implementing this higher order finite difference have a higher order of accuracy than first order finite difference, which increases with increase in the order of chaos expansion. The number of samples required for finding expansion coefficients is equal to the number of polynomials used in the expansion and less than required for existing nonintrusive methods. In some cases, similar results to that of analytical differentiation for output are obtained by using higher order finite difference. The PCDD is applied to the modeling of an eightlayered fiber reinforced plastic (FRP) composite laminate with uncertainties in material and geometric properties. The PCDD was capable of providing similar results to 5e04 Latin Hypercube Sampling with similar accuracy and substantial computational savings.
Speaker Bio:

Dr. Sameer B. Mulani graduated from Indian Institute of Technology Bombay (IIT Bombay), Mumbai, India, with Masters in Aerospace Engineering. During his Maste's, he was awarded DAAD fellowship to carry out Master's thesis work at the Institut fur Statik und Dynamik der Luft und Raumfahrtkonstruktionen (ISD), Universitat Stuttgart, Germany. He completed his Ph.D. in July 2006 from Aerospace and Ocean Engineering at Virginia Tech. He was Postdoctoral Associate and Research Scientist until the end of 2013 December in the Department of Aerospace and Ocean Engineering at Virginia Tech and carried research in the area of MultiDisciplinary Optimization and Uncertainty Quantification. Currently, he is an Assistant Professor in the Department of Aerospace Engineering and Mechanics at the University of Alabama.



