Author:
Publisher:
ISBN:
Category :
Languages : en
Pages : 0
Get Book Here
Book Description
NREL researchers use advanced machine learning techniques to define improved methods using deep learning models to resolve reacting flows in turbulent combustion flows, reducing the computational burden, increasing computational speed, and improving accuracy. These advancements reduce cost and improve fidelity of rapid-turn-around engineering calculations.
Author:
Publisher:
ISBN:
Category :
Languages : en
Pages : 0
Get Book Here
Book Description
Of the challenge, approach, results, and impact of co-optimized machine-learned manifold models for research of large eddy simulation of turbulent combustion.
Author: Ashish Shireeshkumar Newale
Publisher:
ISBN:
Category :
Languages : en
Pages : 191
Get Book Here
Book Description
There is a time critical need for design of fossil fuel based energy conversion devices that attain the dual and usually competing objectives of high efficiency and low pollutant emissions. The design of such devices can be informed by, and in certain instances derived from predictive computations. A crucial component of reacting flow simulations that are predictive is the turbulent combustion model. Probability density function (PDF) methods have been shown to accurately capture flames with strong turbulence chemistry interactions. However, PDF methods are known to be more computationally intensive than simpler topology based approaches such as steady laminar flamelet models. The recently proposed pre-partitioned adaptive chemistry (PPAC) methodology mitigates the cost of using particle PDF methods while maintaining their accuracy. PPAC generates a set of reduced models in an offline preprocessing stage, which are then dynamically utilized at runtime for integrating particle compositions. In the first part of this work, PPAC is augmented by combining it with complementary dimension reduction (rate-controlled constrained equilibrium (RCCE)) and storage retrieval methods (in-situ adaptive tabulation (ISAT)). The combined PPAC-RCCE-ISAT method is shown to outperform standalone PPAC by avoiding redundant direct integrations leading to a significant reduction in the CPU cost, and achieving a sizable reduction in the memory requirement by retaining fewer variables at runtime. Though PPAC has been developed for reducing the computational cost of particle PDF computations, it had previously been tested only in a partially stirred reactor (PaSR). Consequently, an integrated PPAC (-ISAT) particle PDF solver is developed as part of the current work. A detailed assessment of PPAC and PPAC-ISAT in LES/PDF simulations of turbulent combustion is completed using the developed solver. For a large-scale simulation of Sandia flame D, the coupled PPAC-ISAT particle PDF solver is shown to reduce the average wall clock time of a standalone ISAT implementation using the detailed mechanism by 39%, with a minimal loss of accuracy. A key assumption made in the PPAC framework is that the compositions used in the offline preprocessing stage are representative of those encountered at runtime. Hence, the efficient generation of a representative database is crucial to the success of PPAC. The suitability of existing canonical 0D-1D reactors is examined for this purpose. Specifically, compositions obtained from these canonical reactors are compared to the compositions extracted from a variety of direct numerical simulations using an ISAT based approach. We show that the compositions obtained from 1D counterflow flames and PaSR are representative of a significant fraction of the compositions encountered in turbulent combustion simulations. To directly quantify the impact of using databases generated from canonical 0D-1D reactors, we use the coupled PPAC-ISAT particle PDF solver for performing LES/PDF simulations of Sandia flame D. We explore two databases: a first one generated using compositions extracted from 1D counterflow flames, and a second one using compositions from a PaSR. We show that the use of these efficiently generated databases leads to results that are comparable to the case where the database is comprised of compositions extracted from the LES/PDF simulation itself. Finally, avenues for further research that can significantly improve the utility of PPAC for enabling computationally-efficient and accurate particle PDF computations are identified.
Author: Michael K. Stöllinger
Publisher:
ISBN: 9781124890227
Category : Coal, Pulverized
Languages : en
Pages : 157
Get Book Here
Book Description
The use of probability density function (PDF) methods for turbulent combustion simulations is very attractive because arbitrary finite-rate chemistry can be exactly taken into account. PDF methods are well developed for non-premixed turbulent combustion. However, many real flames involve a variety of mixing regimes (non-premixed, partially-premixed and premixed turbulent combustion), and the development of PDF methods for partially-premixed and premixed turbulent combustion has turned out to be a challenging task. This thesis demonstrates a promising way to overcome this problem by extending existing PDF methods to cover a variety of mixing regimes. This extension of PDF methods is done by a generalization of the standard scalar mixing time scale model to account for the fast chemical reactions that are present in premixed combustion. The suitability of the new mixing time scale model is shown by applications to several premixed turbulent Bunsen flames that cover various regimes ranging from flamelet to distributed combustion. Moreover, the combined performance of several mixing models and time scale models is investigated for a turbulent premixed flame. Motivated by the success of the PDF method in turbulent gaseous combustion, the method is extended to describe dilute dispersed turbulent gas-solid reacting flows. Such flows are found in pulverized coal combustion and entrained flow coal gasification. The solid particles are modeled by a stochastic Lagrangian method which is combined with a joint velocity-composition PDF method for the gas phase. Coal specific devolatilization and char reaction models are included in the Lagrangian particle method and models for the mass, momentum and heat exchange between the phases are proposed. To account for the radiative heat transfer, a computationally efficient radiation model is coupled to the gas and particle phase equations. The proposed approach is applied in simulations of pulverized coal combustion in a semi-industrial scale furnace and validated by comparison of simulation results with available measurements.
Author: National Academies of Sciences, Engineering, and Medicine
Publisher: National Academies Press
ISBN: 0309388805
Category : Science
Languages : en
Pages : 351
Get Book Here
Book Description
As the nation's economic activities, security concerns, and stewardship of natural resources become increasingly complex and globally interrelated, they become ever more sensitive to adverse impacts from weather, climate, and other natural phenomena. For several decades, forecasts with lead times of a few days for weather and other environmental phenomena have yielded valuable information to improve decision-making across all sectors of society. Developing the capability to forecast environmental conditions and disruptive events several weeks and months in advance could dramatically increase the value and benefit of environmental predictions, saving lives, protecting property, increasing economic vitality, protecting the environment, and informing policy choices. Over the past decade, the ability to forecast weather and climate conditions on subseasonal to seasonal (S2S) timescales, i.e., two to fifty-two weeks in advance, has improved substantially. Although significant progress has been made, much work remains to make S2S predictions skillful enough, as well as optimally tailored and communicated, to enable widespread use. Next Generation Earth System Predictions presents a ten-year U.S. research agenda that increases the nation's S2S research and modeling capability, advances S2S forecasting, and aids in decision making at medium and extended lead times.
Author: Pavlo O. Dral
Publisher: Elsevier
ISBN: 0323886043
Category : Science
Languages : en
Pages : 702
Get Book Here
Book Description
Quantum chemistry is simulating atomistic systems according to the laws of quantum mechanics, and such simulations are essential for our understanding of the world and for technological progress. Machine learning revolutionizes quantum chemistry by increasing simulation speed and accuracy and obtaining new insights. However, for nonspecialists, learning about this vast field is a formidable challenge. Quantum Chemistry in the Age of Machine Learning covers this exciting field in detail, ranging from basic concepts to comprehensive methodological details to providing detailed codes and hands-on tutorials. Such an approach helps readers get a quick overview of existing techniques and provides an opportunity to learn the intricacies and inner workings of state-of-the-art methods. The book describes the underlying concepts of machine learning and quantum chemistry, machine learning potentials and learning of other quantum chemical properties, machine learning-improved quantum chemical methods, analysis of Big Data from simulations, and materials design with machine learning. Drawing on the expertise of a team of specialist contributors, this book serves as a valuable guide for both aspiring beginners and specialists in this exciting field. Compiles advances of machine learning in quantum chemistry across different areas into a single resource Provides insights into the underlying concepts of machine learning techniques that are relevant to quantum chemistry Describes, in detail, the current state-of-the-art machine learning-based methods in quantum chemistry
Author: National Academies of Sciences Engineering and Medicine
Publisher:
ISBN: 9780309685382
Category :
Languages : en
Pages :
Get Book Here
Book Description
Fusion energy offers the prospect of addressing the nation's energy needs and contributing to the transition to a low-carbon emission electrical generation infrastructure. Technology and research results from U.S. investments in the major fusion burning plasma experiment known as ITER, coupled with a strong foundation of research funded by the Department of Energy (DOE), position the United States to begin planning for its first fusion pilot plant. Strong interest from the private sector is an additional motivating factor, as the process of decarbonizing and modernizing the nation's electric infrastructure accelerates and companies seek to lead the way. At the request of DOE, Bringing Fusion to the U.S. Grid builds upon the work of the 2019 report Final Report of the Committee on a Strategic Plan for U.S. Burning Plasma Research to identify the key goals and innovations - independent of confinement concept - that are needed to support the development of a U.S. fusion pilot plant that can serve as a model for producing electricity at the lowest possible capital cost.
Author: Hugh M. Cartwright
Publisher: Royal Society of Chemistry
ISBN: 1788017897
Category : Science
Languages : en
Pages : 564
Get Book Here
Book Description
Progress in the application of machine learning (ML) to the physical and life sciences has been rapid. A decade ago, the method was mainly of interest to those in computer science departments, but more recently ML tools have been developed that show significant potential across wide areas of science. There is a growing consensus that ML software, and related areas of artificial intelligence, may, in due course, become as fundamental to scientific research as computers themselves. Yet a perception remains that ML is obscure or esoteric, that only computer scientists can really understand it, and that few meaningful applications in scientific research exist. This book challenges that view. With contributions from leading research groups, it presents in-depth examples to illustrate how ML can be applied to real chemical problems. Through these examples, the reader can both gain a feel for what ML can and cannot (so far) achieve, and also identify characteristics that might make a problem in physical science amenable to a ML approach. This text is a valuable resource for scientists who are intrigued by the power of machine learning and want to learn more about how it can be applied in their own field.
Author: Heike Jagode
Publisher: Springer Nature
ISBN: 3030598519
Category : Computers
Languages : en
Pages : 382
Get Book Here
Book Description
This book constitutes the refereed post-conference proceedings of 10 workshops held at the 35th International ISC High Performance 2020 Conference, in Frankfurt, Germany, in June 2020: First Workshop on Compiler-assisted Correctness Checking and Performance Optimization for HPC (C3PO); First International Workshop on the Application of Machine Learning Techniques to Computational Fluid Dynamics Simulations and Analysis (CFDML); HPC I/O in the Data Center Workshop (HPC-IODC); First Workshop \Machine Learning on HPC Systems" (MLHPCS); First International Workshop on Monitoring and Data Analytics (MODA); 15th Workshop on Virtualization in High-Performance Cloud Computing (VHPC). The 25 full papers included in this volume were carefully reviewed and selected. They cover all aspects of research, development, and application of large-scale, high performance experimental and commercial systems. Topics include high-performance computing (HPC), computer architecture and hardware, programming models, system software, performance analysis and modeling, compiler analysis and optimization techniques, software sustainability, scientific applications, deep learning.
Author: Rubin Landau
Publisher: Princeton University Press
ISBN: 1400841186
Category : Science
Languages : en
Pages : 685
Get Book Here
Book Description
Computational physics is a rapidly growing subfield of computational science, in large part because computers can solve previously intractable problems or simulate natural processes that do not have analytic solutions. The next step beyond Landau's First Course in Scientific Computing and a follow-up to Landau and Páez's Computational Physics, this text presents a broad survey of key topics in computational physics for advanced undergraduates and beginning graduate students, including new discussions of visualization tools, wavelet analysis, molecular dynamics, and computational fluid dynamics. By treating science, applied mathematics, and computer science together, the book reveals how this knowledge base can be applied to a wider range of real-world problems than computational physics texts normally address. Designed for a one- or two-semester course, A Survey of Computational Physics will also interest anyone who wants a reference on or practical experience in the basics of computational physics. Accessible to advanced undergraduates Real-world problem-solving approach Java codes and applets integrated with text Companion Web site includes videos of lectures
