Seminar - Computational Nuclear Science and Engineering
Schedule Fall 2008:
- Sep 28, 2008 (Computational Nuclear Science
and Engineering Seminar) Todd Palmer, OSU NERHP,
Seminar Overview and Intro to Computational Science
Abstract: This term, my seminar will bring a large number of nationally renowned computational scientists to Corvallis to talk about their research applying high performance computing to the solution of energy-related multiphysics problems. In today's presentation, I will talk about the course expectations (attendance to presentations, active participation and read/summarize an article from Computing in Science and Engineering), the schedule of presentations, and describe the pressing need for education in and scope of computational science.
- Oct 7, 2008(Computational Nuclear Science
and Engineering Seminar) Jonathan Barr, Pacific Northwest National Laboratory,
The Nonproliferation Graduate Program (NGP) at PNL
Abstract:The NGP is a year-long fellowship program structured to promote awareness of professional opportunities in nonproliferation and to develop an exceptional talent pool that could aid the National Nuclear Security Administration in its national and international security work. NGP fellows work within NNSA's Office of Defense Nuclear Nonproliferation on programs in support of detecting, preventing, and reversing the proliferation of weapons of mass destruction, while mitigating the risks of nuclear operations. The full-time program provides students with specialized, hands on training and practical experience on projects and initiatives over the course of the year, enabling participants to gain valuable experience with government agencies, national laboratories, and non-governmental organizations. Fellows work alongside senior NNSA officials and program managers, and opportunities within the USG are a possibility upon completion of their assignments. Applications for the 2009 program are currently being accepted and the application deadline is October 30. For more information about the program, students can go to http://ngp.pnl.gov.
Bio sketch:Mr. Barr is a research engineer who specializes in the design and development of sensor systems. He has worked on sensor systems with combinations of various technologies including photoacoustic spectroscopy, fluorescence spectroscopy, LWIR cameras, light sources (UV, VIS, IR lasers and Xe flash lamps), instrument control systems, signal acquisition/processing, and functionalized preconcentrating materials. Mr. Barr also has extensive experience with synthesis of micro-/nano-materials and associated processing and analysis/characterization. He received a BS in Mechanical Engineering from Kansas State and a MS in Mechanical Engineering/Materials Science from Washington University in St. Louis. Mr. Barr is also a member of both the American Society of Mechanical Engineers and the Materials Research Society.
- Oct 9, 2008(Computational Nuclear Science
and Engineering Seminar) Oleg Roderick, Portland State University Dept. of Mathematics,
Uncertainty Quantification: Improved Stochastic Finite Element Approach [Presentation (.pdf)]
Abstract: In our work, we introduce a stochastic finite element-based approach to describing the uncertainty of a complex system of differential-algebraic equations with random inputs. For our test system, we take a 3-dimensional steady-state model of heat distribution in the core of a nuclear reactor. The heat exchange between the fuel and the coolant depends on thermo-dynamical properties of the involved materials; the estimation of temperature dependence of these properties includes experimental error. The problem of uncertainty quantification may be solved by sampling methods; or through the creation of a surrogate model (a valid simplified version of the system). The choices for representing the output of the system include linear approximation, or a projection onto a set of interpolating functions. A generic stochastic finite element method (SFEM) approach uses a complete set of orthogonal polynomials (a polynomial chaos), of degrees up to 3-5. Given statistical information on the inputs, SFEM model provides explicit description of the distribution of the output. If input uncertainty structure is not provided, the surrogate model can still be used to estimate the range and variance of the output. One disadvantage of the approach is the large dimension of the model, requiring many full integrations of the system for interpolation. We construct the surrogate model as a goal-oriented projection onto an incomplete space of interpolating polynomials; find the coordinates of the projection by collocation; and use derivative information to significantly reduce the number of the required collocation sample points. The basis may be trimmed to linear functions in some variables, and extended to high order polynomials in the others. Derivatives of the output with respect to random parameters are obtained using an efficient adjoint method with elements of automatic differentiation, the relative magnitudes of the derivatives are also used to decide on the importance of the variables. The resulting model is more computationally efficient that random sampling, or generic SFEM; and has significantly greater precision than linear models. Currently, we work on applying the analysis to an extended model of the reactor core, with additional uncertainties coming from the description of neutron interaction, non-uniform flow of the coolant, and structural deformation of the core elements. We will investigate the possibilities to further improve the approach through the optimal choice of the collocation points; and a more sophisticated sensitivity analysis resulting in an optimal polynomial basis. We are open to suggestions of future collaboration on the mathematical and applied aspects of the study.
- Oct 14, 2008(Computational Nuclear Science
and Engineering Seminar) Dmitriy Anistratov, North Carolina State University Dept. of Nuclear Engineering, Nonlinear Computational Methods for Simulating Interactions of Radiation with
Matter in Physical Systems [Presentation (.pdf)]
Abstract: In some physical systems, processes of energy redistribution are driven significantly by fluxes of particles (radiation). Such phenomena take place, for example, in nuclear reactors, stars, laser fusion targets etc. To simulate such physical systems, one needs to formulate and solve multiphysical models that include the transport equation which is a background for mathematical models of neutron transport in nuclear reactors, radiative transfer in plasmas etc. The dimensionality of transport problems is large. In this talk, deterministic methods for solving the multidimensional transport equation on regular and unstructured grids by means of nonlinear projective transport methods are presented. These computational methods are based on formulation of special low-order problems coupled with the original high-order transport equation. The structure of low-order equations are particularly attractive for solving multiphysics problems in which the transport equation is coupled with equations of matter.
Bio sketch:Dmitriy Anistratov graduated with MS in Physics from Moscow Institute of Physics and Technology in 1985. He got his Ph.D. in Physical and Mathematical Sciences from Institute for Mathematical Modeling of Russian Academy of Sciences in 1993. In 1985, D. Anistratov joined the radiative hydrodynamics group at Keldysh Institute of Applied Mathematics of USSR Academy of Sciences. In 1990, he moved with this research group to Institute for Mathematical Modeling of Russian Academy of Sciences. During period of 1995-2000, he was Visiting Assistant Professor of Nuclear Engineering at Texas A&M University. In 2000, Dr. Anistratov joined the Department of Nuclear Engineering at NC State University. D. Anistratov works in the field of computational physics, numerical transport theory and numerical analysis. His research interest is the development of computational methods for solving various particle transport problems. His work involves developing methods for problems of neutron transport for reactor physics applications (e.g. full- core calculations), radiative transfer in physical systems (e.g. radiative hydrodynamics), methods for multiphysics coupling, numerical methods for solving the multidimensional transport equation on unstructured grids, and transport iterative methods.
- Oct 21, 2008 (Computational Nuclear Science
and Engineering Seminar)
Dana Knoll, Idaho National Laboratory, Parallel Algorithms and Software for Multiphysics Computational Nuclear Engineering [Presentation (.pdf)]
Abstract: There is a growing trend in nuclear reactor simulation to consider multiphysics problems. This can be seen in reactor analysis where analysts are interested in coupled flow, heat transfer and neutronics, and in fuel performance simulation where analysts are interested in thermomechanics with contact coupled to species transport and chemistry. These more ambitious simulations usually motivate some level of parallel computing. Many of the coupling efforts to date have been simple "code coupling" or first-order operator splitting, often referred to as loose coupling. While these approaches can produce answers, they usually leave questions of accuracy and stability unanswered. Additionally, the different physics often reside on separate grids which are coupled via simple interpolation, again leaving open questions of stability and accuracy. We are developing a capability to evolved tightly coupled multiphysics tools for nuclear engineering applications. We are utilizing the Jacobian-free Newton-Krylov method along with physics-based preconditioning. We are also leveraging a significant level of previously developed software in order to build the Multiphysics Object-Oriented Simulation Environment (MOOSE). MOOSE is then used to rapidly develop other multiphysics application codes. In this presentation we will provide some simple analysis on operator-splitting errors. Next we will discuss some algorithmic issues related to JFNK and physics-based preconditioning. Finally, we will discuss MOOSE, including examples from BISION, our 3-D fuel performance code, and PRONGHORN, our 3-D coupled flow, heat transfer, and neutronics code for pebble bed gas cooled reactors.
Bio Sketch: Dana Knoll's education includes BS Chem Eng, Univ. of Minnesota, 1983; MS Nuc Eng, Univ. of Washington, 1985; PhD Nuc Eng, Univ of New Mexico, 1991. He has held professional positions at Martin Marietta Aerospace (2 yrs) , INEL (5 yrs), LANL (10 yrs), and INL (2 yrs). He is currently the technical leader of the INL multi-physics group inside the Nuclear Engineering and Science Division. Dana's research interests include computational fluid dynamics and multi-physics methods development, plasma physics and general computational nuclear engineering. Dana has authored or co-authored over 70 peer-reviewed publications and he is an associate editor for the Journal of Computational Physics.
- Oct 28, 2008 (Computational Nuclear Science
and Engineering Seminar) Kevin Clarno,
Oak Ridge National Laboratory
Multiphysics and Numerical Complexities of Nuclear Reactor Simulation [Presentation (.ppt)]
Abstract: Traditional nuclear reactor simulation has consisted of isolating all of the physics in fine-mesh models (or experiments) to define coarse-mesh data that approximates the multi-physics effects (few-group cross sections, power-to-flow maps, etc.). This data is then used in a coarse-mesh multi-physics solver to determine the power, temperature, density distributions in the system. However, this approach has introduced severe limitations as we attempt to increase the predictive capability of the software. This seminar will discuss several specific effects of the fine-mesh physics on full-core neutronics solutions as well as the challenges of doing this properly in the current multi-level approach to radiation transport. Specifically addressed issues are associated with nuclear fuels and include the effects on reactivity and end-of-life isotopics of the radial temperature profile, the radial depletion profile, and the thermo-chemical-mechanical changes of the fuel and cladding during irradiation. A brief introduction to reactor simulation with SCALE (TRITON) and NESTLE will be provided as an overview.
Bio Sketch: Kevin T. Clarno earned a PhD and M.S. from Texas A&M University, as well as a B.S from the Massachusetts Institute of Technology in Nuclear Engineering. He has been employed as a Computational Nuclear Engineer in the Reactor Analysis Group of the Nuclear Science and Technology Division of Oak Ridge National Laboratory since 2004, after serving as a Naval Nuclear Propulsion Fellow at Bettis Atomic Power Laboratory from 2002-2004. Dr. Clarno is actively involved in both nuclear reactor analysis and the development and improvement of ORNL nuclear analysis tools. Dr. Clarno is the Principal Investigator for the development of a high-performance computing Boltzmann transport solver for reactor simulation. He has been actively involved in the software development of developmental coupled-physics solvers as well as production codes in SCALE. Dr. Clarno has published more than 25 papers in journals, conference proceedings, and technical reports, spanning diverse areas of nuclear science and engineering. His reactor analysis portfolio includes the ACR, ESBWR, AHTR, PBMR, and VHTR. He has made contributions in the development and implementation of linear and non-linear methods of accelerating 1-, 2-, and 3-D transport solvers, including CENTRM, NEWT, and NEWTRNX.
- Nov 11, 2008 (Computational Nuclear Science
and Engineering Seminar) Sourabh Apte, OSU Mechanical Engineering
Multiscale Simulations and Modeling of Particulate Flows in Oxycoal Reactors [Presentation (.ppt)]
Abstract: Energy production in today's oil-constrained resources demands development of alternative power generation concepts as well as retrofitting existing fossil-fuel based power plants. New approaches utilizing oxygen-rich combustion for reduction in pollutant formation together with carbon capture and sequestration have received considerable interest. As the first step toward predicting multiphase, multiphysics, turbulent reacting flows in oxycoal reactors, we are developing multiscale models for large-scale simulations of coal-particle laden turbulent flows. In this talk, we will review simulation strategies typically employed in Particle-In-Cell methods and limitations of models used for particle motion and dispersion in turbulent flows. A multiscale approach for finite volume simulations of particle-laden turbulent flows involves integration of numerical strategies for different particle sizes relative to the computational grid resolution. We will review numerical schemes for (i) particles that are fully resolved on a background grid (particle size greater than the grid resolution) and (ii) under-resolved particles with size smaller than the background mesh. We will discuss how these schemes may be combined into a unified framework for simulations of particulate flows together with possible solutions for parallelization and load balancing issues.
Bio Sketch: Dr. Apte is assistant professor of Mechanical Engineering at the Oregon State University. He grew up in the city of Pune (India) and received his BS (1994) in Mechanical Engineering from the University of Pune (CoEP), MS (1996) from the Indian Institute of Science, Bangalore (IISC) and the doctoral degree in Mechanical and Nuclear Engineering from Pennsylvania State University (PennState) (2000). Following his Ph.D., Dr. Apte joined the Center for Turbulence Research (CTR) and the Center for Integrated Turbulence Simulations (CITS) at Stanford University as an Engineering Research Associate. He joined Oregon State University in October 2005. He currently has a part-time appoinment with DoE's National Energy Technology Laboratory (NETL-ARC) in Albany (OR) under the ORISE program. Dr. Apte's research interests are in the areas of computational modeling and analysis of two-phase turbulent flows. He is interested in predictive simulations of turbulent flows in the presence of droplets, bubbles, or solid particles of arbitrary shape. His current research efforts focus on the development of high-fidelity numerical algorithms on structured and unstructured grids for fully resolved simulations of rigid particulates as well as deforming interfaces in turbulent flows. He is also working on reduced-order, subgrid modeling of the disperse-phase turbulent flows in practical configurations. Specific topics of his interest are: internal flow and combustion dynamics in propulsion systems, droplets and spray systems, bubble dynamics and cavitating flows, fictitious domain methods for flow structure-boundary interaction problems, particle-based interface tracking schemes, particle-laden flows in oxy-coal reactors, internal gravity waves, sediment transport in river and coastal regions, and alternative energy sources such solar energy based biofuel reforming and biomass gasification. Dr. Apte is a member of the American Physical Society (APS-DFD), the Combustion Institute (CI), and the Institute for Liquid Atomization and Spray Systems (ILASS).
- Nov 13, 2008 (Computational Nuclear Science
and Engineering Seminar) K. Nelson and Mark Galvin, Graduate Students, OSU NERHP,
LaTeX and Beamer - Open Source, Cross Platform Alternatives to Microsoft Office [Presentation (.pdf)]
Abstract: Preparing simple WYSIWYG (what you see is what you get) documents and presentations using Microsoft Office is relatively quick and easy. However complex jobs, like the papers, reports, and thesis routinely encountered by students, frequently require numerous typesetting and formatting adjustments to achieve professional looking results. The layout of equations, tables, and figures are particularly problematic for all but the most experienced Word/Powerpoint users, and even then it often "doesn't look quite right." The open source software system LaTeX is an extensive set of macros that utilize the TeX typesetting system, originally developed to typeset math-intensive technical manuscripts, to automatically handle formatting and layout according to (customizable) typesetting rules thereby allowing the writer to focus on the content. Special environments, numbering, figure/table placement, cross-referencing, and acronym definition and usage are handled by LaTeX and are automatically updated when new document content is added or the document is restructured. Furthermore, if you use BibTeX in combination with LaTeX, generating citations and a bibliography for your document is as simple as specifying a citation style and pointing LaTeX to your bibliographic database. User-contributed packages extend LaTeX even further, beyond just beautiful documents to presentations and webpage design as well. This presentation will introduce LaTeX and Beamer (a presentation package), and identify popular LaTeX resources for beginning users. In addition, a recommended strategy for conducting a literature search and organizing the results using JabRef, an open source bibliography reference manager, will be presented.
Bio Sketches: Mark Galvin is a doctoral candidate in the Department of Nuclear Engineering and Radiation Health Physics. He holds a BS in Nuclear Engineering ( Oregon State University, 1992), MS in Nuclear Engineering (Massachusetts Institute of Technology, 2001), a Naval Engineer's degree (MIT, 2001), as well as several Program Management and Systems Engineering certifications from the Department of Defense. He retired in 2007 from the U.S. Navy after a career that began in submarine nuclear propulsion operations, training, and maintenance, and ended in strategic weapons systems acquisition, program management, and technical oversight. He learned LaTeX while writing his thesis at MIT, since there wasn't a copy of Microsoft Word to be found anywhere on campus. He is currently investigating implementing reactor-representative behavior in the OSU MASLWR test facility for evaluation of advanced natural circulation reactor plant anticipatory control schemes.
K Nelson is a masters student in the Department of Nuclear Engineering and Radiation Health Physics. He holds a BS in Mechanical Engineering (Brigham Young University Idaho, 2006). He did an internship with the International Atomic Energy Agency in Vienna Austria. I learned LaTeX while completing my graduate work at OSU. At the first of the year I will start working for Wolf Creek Nuclear Power plant doing safety analysis work.
- Nov 18, 2008 (Computational Nuclear Science
and Engineering Seminar) Thomas Brunner, Sandia National Laboratory,
Scalable Z-Pinch Radiation Algorithms -- or -- Why my code users like me playing with really big computers
Abstract: Simulation of Z-pinch experiments at Sandia present many computational challenges. Z-pinches convert stored electrical energy into large X-ray pulses in a few hundred nanoseconds. Many tightly-coupled physical processes, including magnetohydrodynamics, thermal conduction, and thermal radiation transport, need to be simulated in order to optimize experimental designs. The time and length scales of the system require large parallel computations, where the physical domain is decomposed onto each processor. An improved parallel algorithm for the Implicit Monte Carlo has been developed. The previous algorithm did not support cases where the number of particles at the beginning of the time step is unknown, for example when particle splitting is used for variance reduction. Additionally, several race conditions existed in the old algorithm that caused periodic code hangs (upsetting users, as you might expect). This new algorithm is believed to be robust against all race conditions. This algorithm retains excellent scaling of over 80%, even above ten thousand processors.
Bio Sketch: Tom Brunner has been a member of the Z-Pinch Theory group at Sandia National Laboratories since 2000. His work has included developing new physics methods to support users and developing automated code verification suites. One specific area of Tom's technical expertise is in developing massively parallel Monte Carlo schemes for radiation transport. Before Sandia, Tom received his PhD. in Nuclear Engineering from the University of Michigan in 2000, and his B.S. from Purdue in 1996. He has also been serving a term as a member of the American Nuclear Society's Mathematics and Computation Executive committee since 2006.
- Nov 20, 2008(Computational Nuclear Science
and Engineering Seminar) Alexei Soldatov, PhD candidate, OSU NERHP,
Design of an 8.0%-Enriched Nuclear Core for MASLWR
Abstract: In order to address the energy needs of developing countries and remote communities, Oregon State University has proposed a design of the Multi-Application Small Light Water Reactor (MASLWR). The passive safety systems, natural circulation, long core lifetime, off-site refueling and use of standard equipment are the basic design features of the MASLWR reactor design. In order to achieve 5 years of operation without refueling, use of 8% enriched fuel is necessary. The issues related with increased enrichment fuel properties was discussed in a series of publications, however evaluation of the fuel characteristics within MASLWR operation parameters range is necessary for a further design work and decision-making. This presentation will cover such fuel characteristics as 1) Multiplication factor, 2) Depletion and Isotopic composition of the irradiated fuel, 3) Effective fraction of the delayed neutrons and neutron lifetime, 4) Neutron spectrum changes, 5) Reactivity coefficients and 6) Reactivity control mechanisms. All these parameters were evaluated and compared for the MASLWR and PWR operational conditions for standard 4.5% and increased up to 8.0% fuel enrichment. The results of the evaluation will be interesting as for MASLWR designers, but also for the those who plan to work on innovative LWR design (such as IRIS, CAREM, SMART) an on fuel campaign extension for the conventional PWR and BWR.
Bio Sketch: Alexey Soldatov was born in Moscow, USSR in 1980. The first research was performed in the high school and was dedicated to the environmental aspects and thermal pollutions of the NPP. In the 1999 Alexey Soldatov won the Russian President Yeltsin Scholarship for the one year of training abroad, and come to OSU for first time. In 2002 he get degree of B.S. in Nuclear Engineering with a diploma work: The core design evaluation for WWER-1000 for utilization of the weapon grade plutonium. In 2004 Alexey Soldatov has received his master degree in Technical Physics with a minor in political science, with a thesis “the evaluation of the innovative fuel cycle technologies from the non-proliferation prospective. After the graduation Alexey Soldatov work as a consultant for concern Rosenergoatom (Russian utility), British Energy (UK Utility), Tractebelle (Belgium Utility), European Commission Nuclear Safety Division and IAEA. In 2006 Alexey decided to continue his education towards a Ph.D. Degree at OSU. His PhD research focuses on MASLWR Core design with fuel enrichment up to 8.0%