SEARCH · Engineering Papers
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Visualization is key to the work we do at Idaho National Laboratory (INL). We support research computing, nuclear science, and educational outreach using both traditional and emerging visualization tools and technologies. In this presentation, we will introduce and summarize INL and its mission, describe some of the visualization resources INL provides, and present example applications from recent years.
Visualization is key to the work we do at Idaho National Laboratory (INL). We support research computing, nuclear science, and educational outreach using both traditional and emerging visualization tools and technologies. In this presentation, we will introduce and summarize INL and its mission, describe some of the visualization resources INL provides, and present example applications from recent years.
Utilizing an atomistic computational model, which handles both translational and spin degrees of freedom, combined molecular and spin dynamics simulations have been performed to investigate the effect of vacancy defects on spin wave excitations in ferromagnetic iron. Fourier transforms of space- and time-displaced correlation functions yield the dynamic structure factor, providing characteristic frequencies and lifetimes of the spin wave modes. A comparison of the system with a 5% vacancy concentration with pure lattice data shows a decrease in frequency and a decrease in lifetime for all transverse spin wave excitations observed. In addition, the clearly defined transverse spin wave excitations are distorted with the introduction of vacancy defects, and we observe reduced excitation lifetimes due to increased magnon–magnon scattering. We observe further evidence of increased magnon–magnon scattering, as the peaks in the longitudinal spin wave spectrum become less distinct. Finally, similar impacts are observed in the vibrational subsystem, with a decrease in characteristic phonon frequency and flattening of lattice excitation signals due to vacancy defects.
Abstract The primitive Wigner-Seitz cell and corresponding first Brillouin zone (FBZ) are typically used in calculations of lattice vibrational and transport properties as they contain the smallest number of degrees of freedom and thus have the cheapest computational cost. However, in complex materials, the FBZ can take on irregular shapes where lattice symmetries are not apparent. Thus, conventional cells (with more atoms and regular shapes) are often used to describe materials, though dynamical and transport calculations are more expensive. Here we discuss an efficient anharmonic lattice dynamic method that maps conventional cell dynamics to primitive cell dynamics based on translational symmetries. Such symmetries have not been utilized in typical lattice dynamical calculations. This leads to phase interference conditions that act like conserved quantum numbers and a conservation rule for phonon scattering that is hidden in conventional dynamics which significantly reduces the computational cost. We demonstrate this method for phonon transport in a variety of materials with inputs from first-principles calculations and attribute its efficiency to reduced scattering phase space and fewer summations in scattering matrix element calculations.
High-entropy alloys, with N elements and compositions {$c_{ν = 1,N}$} in competing crystal structures, have large design spaces for unique chemical and mechanical properties. In this work, to enable computational design, we use a metaheuristic hybrid Cuckoo search (CS) to construct alloy configurational models on the fly that have targeted atomic site and pair probabilities on arbitrary crystal lattices, given by supercell random approximates (SCRAPs) with S sites. Our Hybrid CS permits efficient global solutions for large, discrete combinatorial optimization that scale linearly in a number of parallel processors, and linearly in sites S for SCRAPs. For example, a four-element, 128-site SCRAP is found in seconds—a more than 13,000-fold reduction over current strategies. Our method thus enables computational alloy design that is currently impractical. We qualify the models and showcase application to real alloys with targeted atomic short-range order. Being problem-agnostic, our Hybrid CS offers potential applications in diverse fields.
Computational modeling and simulation have become indispensable scientific tools in virtually all areas of chemical, biomolecular, and materials systems research. Computation can provide unique and detailed atomic level information that is difficult or impossible to obtain through analytical theories and experimental investigations. In addition, recent advances in micro-electronics have resulted in computer architectures with unprecedented computational capabilities, from the largest supercomputers to common desktop computers. In conclusion, combined with the development of new computational domain science methodologies and novel programming models and techniques, this has resulted in modeling and simulation resources capable of providing results at or better than experimental chemical accuracy and for systems in increasingly realistic chemical environments.
Abstract We design a deep-learning algorithm for the discovery and identification of the continuous group of symmetries present in a labeled dataset. We use fully connected neural networks to model the symmetry transformations and the corresponding generators. The constructed loss functions ensure that the applied transformations are symmetries and the corresponding set of generators forms a closed (sub)algebra. Our procedure is validated with several examples illustrating different types of conserved quantities preserved by symmetry. In the process of deriving the full set of symmetries, we analyze the complete subgroup structure of the rotation groups SO (2), SO (3), and SO (4), and of the Lorentz group S O ( 1 , 3 ) . Other examples include squeeze mapping, piecewise discontinuous labels, and SO (10), demonstrating that our method is completely general, with many possible applications in physics and data science. Our study also opens the door for using a machine learning approach in the mathematical study of Lie groups and their properties.
Energy efficiency is motivating the search for new high-temperature (high-T) metals. Some new body-centered-cubic (BCC) random multicomponent “high-entropy alloys (HEAs)” based on refractory elements (Cr-Mo-Nb-Ta-V-W-Hf-Ti-Zr) possess exceptional strengths at high temperatures but the physical origins of this outstanding behavior are not known. Here we show, using integrated in-situ neutron-diffraction (ND), high-resolution transmission electron microscopy (HRTEM), and recent theory, that the high strength and strength retention of a NbTaTiV alloy and a high-strength/low-density CrMoNbV alloy are attributable to edge dislocations. This finding is surprising because plastic flows in BCC elemental metals and dilute alloys are generally controlled by screw dislocations. We use the insight and theory to perform a computationally-guided search over 10 7 BCC HEAs and identify over 10 6 possible ultra-strong high-T alloy compositions for future exploration.
X-ray free-electron lasers offer unique capabilities for measuring the structure and dynamics of biomolecules, helping us understand the basic building blocks of life. Notably, high-repetition-rate free-electron lasers enable single particle imaging, where individual, weakly scattering biomolecules are imaged under near-physiological conditions with the opportunity to access fleeting states that cannot be captured in cryogenic or crystallized conditions. Existing X-ray single particle reconstruction algorithms, which estimate the particle orientation for each image independently, are slow and memory-intensive when handling the massive datasets generated by emerging free-electron lasers. Here, we introduce X-RAI (X-Ray single particle imaging with Amortized Inference), an online reconstruction framework that estimates the structure of 3D macromolecules from large X-ray single particle datasets. X-RAI consists of a convolutional encoder, which amortizes pose estimation over large datasets, as well as a physics-based decoder, which employs an implicit neural representation to enable high-quality 3D reconstruction in an end-to-end, self-supervised manner. We demonstrate that X-RAI achieves state-of-the-art performance for small-scale datasets in simulation and challenging experimental settings and demonstrate its unprecedented ability to process large datasets containing millions of diffraction images in an online fashion. These abilities signify a paradigm shift in X-ray single particle imaging towards real-time reconstruction.
For a considerable time, large datasets containing problem instances have proven valuable for analyzing computer hardware, software, and algorithms. One notable example of the value of large datasets is ImageNet [1], a vast repository of images that has been instrumental in testing numerous deep learning packages. Similarly, in the domain of computational chemistry and materials science, the availability of extensive datasets such as the Protein Data Bank [2], the Materials Project [3], and QM9 [4] has greatly facilitated the evaluation of new algorithms and software approaches, while also promoting standardization within the field. These well-defined datasets and problem instances, in turn, serve as the foundation for creating benchmarking suites like MLPerf [5] and LINPACK [6], [7]. These suites enable fair and rigorous comparisons of different methodologies and solutions, fostering continuous advancements in various areas of computer science and beyond.
Plans to introduce pilot lines or fabs for 300 mm processing are in progress. The IC technology is simultaneously moving towards 0.25/0.18 micron. The convergence of these two trends places unprecedented stringent demands on processes and equipments. More than ever, computational modeling is called upon to play a complementary role in equipment and process design. The pace in hardware/process development needs a matching pace in software development: an aggressive move towards developing "virtual reactors" is desirable and essential to reduce design cycle and costs. This goal has three elements: reactor scale model, feature level model, and database of physical/chemical properties. With these elements coupled, the complete model should function as a design aid in a CAD environment. This talk would aim at the description of various elements. At the reactor level, continuum, DSMC(or particle) and hybrid models will be discussed and compared using examples of plasma and thermal process simulations. In microtopography evolution, approaches such as level set methods compete with conventional geometric models. Regardless of the approach, the reliance on empricism is to be eliminated through coupling to reactor model and computational surface science. This coupling poses challenging issues of orders of magnitude variation in length and time scales. Finally, database development has fallen behind; current situation is rapidly aggravated by the ever newer chemistries emerging to meet process metrics. The virtual reactor would be a useless concept without an accompanying reliable database that consists of: thermal reaction pathways and rate constants, electron-molecule cross sections, thermochemical properties, transport properties, and finally, surface data on the interaction of radicals, atoms and ions with various surfaces. Large scale computational chemistry efforts are critical as experiments alone cannot meet database needs due to the difficulties associated with such controlled experiments and costs.
As anyone even vaguely aware of current technology can tell you, machine learning (ML) and artificial intelligence (AI) have made exceptional breakthroughs in recent years. Generative artificial intelligence (GAI) emerged circa 2022 dominated by Large Language Models (LLMs) and generative tools for images emerged at about the same time.
While recent advances in deep learning have shown promising efficiency gains in solving time-dependent partial differential equations (PDEs), matching the accuracy of conventional numerical solvers still remains a challenge. One strategy to improve the accuracy of deep learning-based solutions for time-dependent PDEs is to use the learned model as the coarse propagator in the Parareal method and a traditional numerical method as the fine solver. However, successful integration of deep learning into the Parareal method requires consistency between the coarse and fine solvers, particularly for PDEs exhibiting rapid changes such as sharp transitions. Here, to ensure this consistency, we propose using convolutional neural networks (CNNs) to learn the fully discrete time-stepping operator defined by the same numerical scheme employed as the fine solver. We demonstrate the effectiveness of the proposed method in solving the classical and mass-conservative Allen–Cahn (AC) equations. Through iterative updates in the Parareal algorithm, our approach achieves a significant computational speedup compared to traditional fine solvers while converging to high-accuracy solutions. Our results highlight that the proposed hybrid Parareal algorithm effectively accelerates simulations, particularly when implemented on multiple GPUs, and converges to the desired accuracy in only a few iterations. Another advantage of our method is that the CNN model is trained on trajectory-based data generated from random initial conditions, such that the trained model can be used to solve the AC equations with various initial conditions without retraining. This work demonstrates the potential of integrating neural network methods into parallel-in-time frameworks for efficient and accurate simulations of time-dependent PDEs.
Directed energy deposition (DED) is an additive manufacturing process that is being rapidly adopted by industry and is well suited for the fabrication of complex components in a variety of metal alloys. In laser cladding systems such as DED, powder is blown in a stream to a metal substrate coincident with a laser necessary to deposit molten metal with 3D spatial control. The focus of both the laser and the powder stream are crucial, and best deposition occurs at a predetermined standoff height between the build surface and the print head. Generally, no monitoring of this distance is implemented in commercial DED systems. Due to potential over or under building, the standoff height often changes over time but tends to self-correct. However, inexpensive and minimally intrusive methods to identify optimal standoff are required to provide real-time control to maintain the optimal distance. The present work explores the quantification of the focus of the three-color channels of a coaxial camera to determine the standoff height. An experiment was performed in which a 254 mm wall is built and the standoff height, initially 5.0 mm below the optimal position, was then intentionally increased every 25.4 mm of wall length by an amount of 1.0 mm to a final position 7.0 mm above optimal. Computer vision is demonstrated to monitor the amount of focus in each color band and estimate standoff distance. Finally, a response can be calculated in under 40 ms using simple hardware and can work in most laser-based DED systems.
First-principles computational spectroscopy is a critical tool for interpreting experiment, per- forming structure refinement, and developing new physical understanding. Systematically setting up input files for different simulation codes and a diverse class of materials is a challeng- ing task with a very high barrier-to-entry, given the complexities and nuances of each individual simulation package. This task is non-trivial even for experts in the electronic structure field and nearly formidable for non-expert researchers. Lightshow solves this problem by providing a uniform abstraction for writing computational x-ray spectroscopy input files for multiple popular codes, including FEFF, VASP, OCEAN, exciting and XSpectra. Its extendable framework will also allow the community to easily add new functions and to incorporate new simulation codes.
Currently, high-dimensional data is ubiquitous in data science, which necessitates the development of techniques to decompose and interpret such multidimensional (aka tensor) datasets. Finding a low dimensional representation of the data, that is, its inherent structure, is one of the approaches that can serve to understand the dynamics of low dimensional latent features hidden in the data. Moreover, decomposition methods with non-negative constraints are shown to extract more insightful factors. Nonnegative RESCAL is one such technique, particularly well suited to analyze self-relational data, such as dynamic networks found in international trade flows. Particularly, non-negative RESCAL computes a low dimensional tensor representation by finding the latent space containing multiple modalities. Furthermore, estimating the dimensionality of this latent space is crucial for extracting meaningful latent features. Here, to determine the dimensionality of the latent space with non-negative RESCAL, we propose a latent dimension determination method which is based on clustering of the solutions of multiple realizations of non-negative RESCAL decompositions. We demonstrate the performance of our model selection method on synthetic data. We then apply our method to decompose a network of international trade flows data from International Monetary Fund and shows that with a correct latent dimension determination, the resulting features are able to capture relevant empirical facts from economic literature.
Explore the source record for details and available documents.
Abstract Catastrophic failure in brittle materials is often due to the rapid growth and coalescence of cracks aided by high internal stresses. Hence, accurate prediction of maximum internal stress is critical to predicting time to failure and improving the fracture resistance and reliability of materials. Existing high-fidelity methods, such as the Finite-Discrete Element Model (FDEM), are limited by their high computational cost. Therefore, to reduce computational cost while preserving accuracy, a deep learning model, StressNet, is proposed to predict the entire sequence of maximum internal stress based on fracture propagation and the initial stress data. More specifically, the Temporal Independent Convolutional Neural Network (TI-CNN) is designed to capture the spatial features of fractures like fracture path and spall regions, and the Bidirectional Long Short-term Memory (Bi-LSTM) Network is adapted to capture the temporal features. By fusing these features, the evolution in time of the maximum internal stress can be accurately predicted. Moreover, an adaptive loss function is designed by dynamically integrating the Mean Squared Error (MSE) and the Mean Absolute Percentage Error (MAPE), to reflect the fluctuations in maximum internal stress. After training, the proposed model is able to compute accurate multi-step predictions of maximum internal stress in approximately 20 seconds, as compared to the FDEM run time of 4 h, with an average MAPE of 2% relative to test data.